Platform A

ILO Flagship Report
World
X Employment
and Social Outlook
The role of digital labour
platforms in transforming
the world of work
2021
2021
X World Employment
and Social Outlook
The role of digital labour

platforms in transforming
the world of work
International Labour Office • Geneva

Copyright © International Labour Organization 2021
First published 2021
Publications of the International Labour Office enjoy copyright under Protocol 2 of the Universal Copyright
Convention. Nevertheless, short excerpts from them may be reproduced without authorization, on condi-
tion that the source is indicated. For rights of reproduction or translation, application should be made to
ILO Publications (Rights and Licensing), International Labour Office, CH-1211 Geneva 22, Switzerland, or
by email: rights@ilo.org. The International Labour Office welcomes such applications.
Libraries, institutions and other users registered with a reproduction rights organization may make copies
in accordance with the licences issued to them for this purpose. Visit www.ifrro.org to find the reproduc-
tion rights organization in your country.
World Employment and Social Outlook 2021: The role of digital labour platforms
in transforming the world of work
International Labour Office – Geneva: ILO, 2021.

1 v

ISBN 978-92-2-031944-4 (print)

ISBN 978-92-2-031941-3 (web PDF)

employment / unemployment / labour policy / labour market analysis /

economic and social development / future of work / technological change / electronic network /

EDP personnel / business economics / working conditions / labour force survey / digital economy /

digital labour platforms

13.01.3
ILO Cataloguing in Publication Data
The designations employed in ILO publications, which are in conformity with United Nations practice, and
the presentation of material therein do not imply the expression of any opinion whatsoever on the part
of the International Labour Office concerning the legal status of any country, area or territory or of its
authorities, or concerning the delimitation of its frontiers.
The responsibility for opinions expressed in signed articles, studies and other contributions rests solely
with their authors, and publication does not constitute an endorsement by the International Labour Office
of the opinions expressed in them.
Reference to names of firms and commercial products and processes does not imply their endorsement
by the International Labour Office, and any failure to mention a particular firm, commercial product or
process is not a sign of disapproval.
Information on ILO publications and digital products can be found at: www.ilo.org/publns.
Photo credits
Cover: © luza studios on gettyimages.com
Back cover: (top) © ketut subiyanto on Pexels.com
(bottom) © LPETTET on iStock.com
Chapter 1: © Dean Mitchell on iStock.com
Chapter 2: © kate_sept2004 on iStock.com
Chapter 3: © Photo by Standsome Worklifestyle on Unsplash
Chapter 4: © CHANDAN KHANNA on gettyimages.com
Chapters 5 and 6: © Bloomberg/Contributor on gettyimages.com
Pleasenotethatthefiguresonrevenueandworkinghoursonpages66and166havebeenupdated
sincethedigitalversionofthisreportwasfirstpublishedon23February2021.
Produced by the Publications Production Unit (PRODOC) of the ILO.

The ILO endeavours to use paper sourced from forests managed
in an environmentally sustainable and socially responsible manner.
Code: DESIGN/WEI/PMSERV
3
Preface
Technological innovation is transforming every part of our lives. The ability to
quickly and cheaply exchange large amounts of data and information has laid the
foundations for the rise of the digital economy and digital labour platforms. In both
developed and developing countries businesses and consumers have embraced
this transformation, as services and goods are delivered in ways that are cheaper
and more convenient. Digital labour platforms are now part of our everyday lives.
This transformation extends to the world of work. Digital labour platforms offer new
markets for businesses and more income-generating opportunities for workers,
including those who were previously outside the labour market. Such platforms are
leading to changes not just to the organization of enterprises and work processes
but in many cases to the relationship between workers and businesses as well.
It is widely considered that the COVID-19 pandemic has accelerated changes that
were already under way, both in society and at work. These include the expanded use
of digital platforms and related technological innovations like cloud computing and
the use of big data and algorithms. The result has been innovative ways of working,
and flexibility for both workers and businesses. The remote working arrangements
adopted by many during the past year have brought a rise in e-commerce, e-services
and online freelance work. For many who lost their jobs, in both developing and de-
veloped countries, digital labour platforms have offered opportunities to earn some
income. Many businesses have relied on digital labour platforms to keep operating,
reach new markets and reduce costs.
But there are challenges. This new business model allows platforms to organize work
without having to invest in capital assets or to hire employees. Instead, they mediate
between the workers who perform the tasks and clients, and manage the entire
work process with algorithms. Workers on digital labour platforms often struggle to
find sufficient well-paid work to earn a decent income, creating a danger of working
poverty. Many do not have access to social protection, which is particularly con-
cerning during a pandemic. They are frequently unable to engage in the collective
bargaining that would allow them to have these and other issues addressed.
This report is the first major attempt by the ILO to capture the experiences of workers
and businesses with digital labour platforms. It is based on surveys and interviews
with 12,000 workers in 100 countries, and with 70 businesses, 16 platform companies
and 14 platform worker associations operating in multiple sectors and countries.
4 The role of digital labour platforms in transforming the world of work
To address the challenges raised by this new way of working, many governments
have taken regulatory steps to tackle issues such as the employment relationship,
health and safety standards and inadequate social protection. Private, non-state
actors and employers’ and workers’ organizations have also taken initiatives.
However, variations in these regulatory responses have created further challenges.
The matter is made more complex because many digital labour platforms operate
across multiple borders and jurisdictions. The result is regulatory uncertainty for
workers, businesses and governments alike.
Digital labour platforms have the potential to benefit both workers and businesses
– and through them, society more generally. But they will only fulfil this positive
potential, and help us achieve the Sustainable Development Goals, if the work
opportunities they provide are decent. Ensuring that all workers, irrespective of
their contractual status, are covered by key labour standards will be critical, as will
social dialogue.
A clearer understanding of the operation of digital labour platforms, and a more
effective and consistent approach to them, are therefore essential. There is a need
for international policy discussions and coordination, which could lead over time to
that clearer understanding and a more effective and consistent approach to digital
labour platforms worldwide.
Guy Ryder
ILO Director-General
5
Contents
Preface 3
Abbreviations 14
Acknowledgements 15
Executive summary 18
1 The digital transformation of industry

and the world of work 29
Introduction
31
1.1 The rise of the digital economy 34

1.1.1 Key features of the digital economy 34
1.1.2 The rise of digital platforms 36
1.1.3 Open source innovation 37
1.1.4 Concentration of market power among a few platform companies 38
1.2 Digital platforms: Pervading and penetrating

differentsectorsoftheeconomy 39
1.2.1 Digital platforms that offer services to individual users 39
1.2.2 Digital platforms facilitating and mediating exchange between users 41
1.2.3 Digital labour platforms mediating work 43
1.3 Digital labour platforms: Estimates of the number

of platforms and workers 46
1.3.1 Number of digital labour platforms 46
1.3.2 Number of workers engaged on digital labour platforms 47
1.3.3 Trends in labour demand and supply on selected
online web-based platforms 51
1.4 The data-driven economy and the rise of machine-learning algorithms 58

1.4.1 Potential use of data 58
1.4.2 Issues related to user rights over data 59
1.4.3 The rise of machine-learning algorithms 61
1.5 Financing the rise of digital labour platforms 63

1.5.1 Geography of digital labour platforms: funding and revenue 65
Conclusion 69
2 The business model and strategies
of digital labour platforms 71
Introduction
73
2.1 Types of digital labour platforms 74
2.1.1 Online web-based platforms 74

2.1.2 Location-based platforms 75

2.2 Revenue model 78
2.2.1 Freelance and contest-based platforms 79

2.2.2 Competitive programming platforms 84

2.2.3 Microtask platforms 84

2.2.4 Taxi platforms 85

2.2.5 Delivery platforms 88

2.3 Recruitment and matching of workers with clients 90
2.3.1 Work relationships on platforms 90

2.3.2 Basic requirements for opening an account on platforms 93

2.3.3 Algorithmic matching of clients and workers 93

2.4 Work processes and performance management 95
2.4.1 Work processes and communication 95

2.4.2 Algorithmic performance management 97

2.5 Digital labour platforms’ rules of governance
and workers’ freedom to work 98
Conclusion
101
3 Thediffusionofdigitallabourplatforms
in the economy: How and why are businesses
using them? 103
Introduction
105
3.1 Businesses using online web-based platforms 106
3.1.1 Recruitment 106

3.1.2 Cost reduction and efficiency 107

3.1.3 Access to knowledge for innovation 109

3.2 Businesses using location-based platforms 114
3.3 Opportunities from digital platforms for BPO companies
and digital technology start-ups 118
3.3.1 Transformations in BPO companies 118

3.3.2 Emergence of digital technology start-ups 121

3.4 Impactofdigitalplatformsontraditionalbusinesses 126
Conclusion
130
Contents 7
4 Digitallabourplatformsandtheredefinition
of work: Opportunities and challenges for workers 133
Introduction
135
4.1 Basic demographic characteristics of platform workers 136
4.1.1 Age distribution of platform workers 137

4.1.2 Participation of male and female workers on platforms 137

4.1.3 Participation of workers from rural and urban areas 138

4.1.4 Participation of migrants on platforms 139

4.1.5 Health status of workers on platforms 140

4.1.6 Education levels of platform workers 141

4.1.7 Worker motivation for engaging in platform work 143

4.1.8 Worker satisfaction with platform work 145

4.2 Worker experience and the quality of work on digital labour platforms 147
4.2.1 Access to a sufficient amount of work 147

4.2.2 Worker earnings on digital labour platforms 154

4.2.3 Working hours and work–life balance 166

4.2.4 Occupational safety and health 171

4.2.5 Access to social protection 174

4.3 Worker autonomy and control under algorithmic management 177
4.3.1 Autonomy and control over work 177

4.3.2 Ratings, evaluation and dispute resolution 179

4.4 Skills acquisition and mismatch 184
4.5 Platform design and discrimination 189
Conclusion 191
5 Ensuring decent work on digital labour platforms 195
Introduction
197
5.1 Regulation by digital labour platforms: Terms of service agreements 198
5.1.1 Platform terms of service agreements 198

5.1.2 Will the digital labour platforms improve terms of service

by themselves? 201

5.2 Regulating digital platforms for labour and social protection:
What should be the goals? 202
5.2.1 Labour standards for all working people: ILO instruments 203

5.2.2 Convention principles that could be adapted to all digital labour

platform workers, irrespective of their status 206

5.2.3 Elements of decent work closely tied to employment:

The Employment Relationship Recommendation, 2006 (No. 198) 209

5.2.4 Employment-related standards and self-employed

platform workers 210

5.3 Achieving decent work: Regulatory responses to platform work 211
5.3.1 Freedom of association, collective bargaining
and other forms of social dialogue 211
5.3.2 Non-discrimination 216
5.3.3 Forced labour and child labour 218
5.3.4 Occupational safety and health 219
5.3.5 Social security 221
5.3.6 The COVID-19 pandemic and its implications for health
and safety at work and social security 224
5.3.7 Payment systems, fair termination and clear terms
of engagement 225
5.3.8 Access to data, privacy and job mobility 227
5.3.9 Grievance and dispute resolution 228
5.3.10 The employment relationship 230
5.3.11 Remuneration and working time 234
5.3.12 Platform work and labour clauses in trade agreements 236
Conclusion 237
6 Seizing the opportunity: A way forward 239
Introduction
241
6.1 Opportunitiesandchallengesondigitallabourplatforms 242
6.1.1 Opportunities and challenges for businesses 242
6.1.2 Opportunities and challenges for workers 243
6.2 Emergingregulatoryresponses 245
6.2.1 National jurisdictions 245
6.2.2 Initiatives by social partners 246
6.2.3 Initiatives by other non-state actors 247
6.3 Overcomingthechallengestoseizethebenefits 248
6.3.1 Addressing the regulatory gaps 248
6.3.2 Relevance of other fields of law and policy for decent work
on digital labour platforms 252
6.4 Awayforward 255
Appendices 257
1. Digital labour platforms: Estimates of workers, investments and revenues

2. ILO interviews with digital platform companies and analysis of terms
of service agreements
3. ILO interviews with businesses and clients
4. ILO surveys, interviews and statistical analysis
5. ILO Interviews with unions and associations
References 259
Contents 9
Boxes
1.1 Terminology used in the report 33

1.2 Cloud infrastructure and computing services 35

1.3 COVID-19 impact on online web-based platforms 56

1.4 Collective user rights over community data 60

2.1 Private employment agencies 82

2.2 Pricing by taxi platforms and potential for litigation:

The case of Ola and Uber in India 87

2.3 Platform cooperatives 88

2.4 Monitoring work processes on digital labour platforms 96

3.1 Apache Software Foundation 110

3.2 Using the Topcoder community for technological solutions 111

3.3 Wipro’s new strategy to develop human resource capabilities

and innovate using digital labour platforms 112

3.4 Customers’ motivation for using app-based taxi and delivery services 117

3.5 Proliferation of AI start-ups 122

3.6 “Jordan”, the automated virtual assistant: A case study 124

3.7 Open source community platforms in the retail sector 129

4.1 Circumventing geographical barriers to accessing work 151

4.2 COVID-19 impact on availability of and access to work 153

4.3 Overcoming low pay and payment barriers 159

4.4 COVID-19 impact on income 162

4.5 COVID-19 impact on occupational safety and health 173

4.6 COVID-19 and social protection 175

4.7 Underutilization of skills 187

5.1 Collective bargaining: Denmark

HilfrandUnitedFederationofDanishWorkers(3F)Agreement 214

5.2 Safety and health at work: Brazil 220

5.3 Work injury insurance: China and the Republic of Korea 223

5.4 The employment relationship: California Labor Code 232

Figures
1.1 Landscape of digital platforms 40

1.2 Outsourcing of tasks on a freelance platform across countries,

inflow of work and earnings, 2019 45

1.3 Number of active digital labour platforms globally, selected categories 47

1.4 Global share of digital labour platforms, by number of employees,
January 2021 48
1.5 Estimates of workers engaged on digital platforms based on surveys 49

1.6 Online global labour supply and demand on major online

web-based platforms, 2017–21 52

1.7 Global demand for labour across occupational categories on five major

online web-based platforms, 2018 and 2020 52

1.8 Distribution of global labour supply and demand on major online

web-based platforms, by country and occupational category,

2018 and 2020 53

1.9 Gender distribution of labour supply on online web-based platforms,
by occupation, selected countries, October 2020 to January 2021 55
1.10 Online labour demand and supply, the United States and India,

2018 and 2020 56

1.11 Total investments from venture capital and other investors,

by platform category, 1998–2020 64

1.12 Total funding from venture capital and other investors, selected

categories of digital labour platforms, by region, 1998–2020 65

1.13 Estimated annual revenue of digital labour platforms,

selected categories, by region, 2019 66

1.14 Estimated annual revenue of large platforms

and selected digital labour platforms, 2019 67

2.1 Types of digital labour platforms 76

2.2 The platform business model: Business strategies 78

2.3 Number of employees directly hired by digital labour platforms,

2019–20 91

2.4 Indicators used to determine client–worker matching on freelance

and contest-based platforms 94

2.5 Upwork work diary 96

Contents 11
4.1 Age distribution, by occupation 137

4.2 Share of female respondents, by occupation and country 138

4.3 Share of migrant respondents in the taxi and delivery sectors 139

4.4 Share of respondents who consider their health to be poor

or very poor, by occupation and country 140

4.5 Educational levels of workers, by occupation and country 142

4.6 Most important reason for performing work on digital labour

platforms, by occupation and country 144

4.7 Worker satisfaction levels, by occupation and country 146

4.8 Design of a platform: The worker experience 148

4.9 Proportion of workers who would like to do more online work,

by type of platform, development status and sex 149

4.10 Most important reasons for not being able to undertake more

online work, by type of platform, development status and sex 149

4.11 Share of respondents who identify online work as their primary

source of income, by type of platform, development status and sex 154

4.12 Hourly earnings (paid and unpaid) on online web-based platforms,

by type of platform, development status and sex 156

4.13 Hourly earnings of survey respondents on microtask platforms

compared to their counterparts in the traditional labour market,

India and the United States, by sex 156

4.14 Hourly earnings in the taxi and delivery sectors, by country 160

4.15 Hourly earnings of app-based workers compared to their traditional

counterparts in the taxi and delivery sectors, by country 163

4.16 Hours worked in a typical week (paid and unpaid), by type of platform,

development status and sex 167

4.17 Hours worked in a typical week in the taxi and delivery sectors,

by country 169

4.18 Main reasons for stress in the app-based taxi and delivery sectors 171

4.19 Main concerns regarding personal and physical safety

in the app-based taxi and delivery sectors 172

4.20 Proportion of respondents in the app-based taxi and delivery sectors

that are unable to refuse or cancel work without repercussion,

by country 178

4.21 Rejection of online work, by type of platform and country 180

4.22 Knowledge and use of appeal mechanisms on freelance platforms 181

4.23 Deactivation of accounts in the app-based taxi and delivery sectors 183

4.24 Skills in relation to tasks on online web-based platforms,

by type of platform 186

4.25 Proportion of respondents who received training from app-based

taxi and delivery platforms, by country 188

4.26 Types of training provided by app-based taxi and delivery platforms 188

4.27 Proportion of respondents on freelance platforms who have

experienced discrimination, by development status and sex 189

4.28 Proportion of respondents on app-based taxi and delivery platforms

who have experienced or witnessed discrimination or harassment,

by sex and country 190

4.29 App-based taxi drivers and delivery workers having faced

or witnessed discrimination or harassment from different entities 190

5.1 Decent work elements applicable to all platform workers

irrespective of contractual status 208

5.2 Number of protest incidents related to working conditions on digital
labour platforms per month globally, January 2017 to July 2020 215
5.3 Different approaches to establishing an employment relationship

between workers and digital labour platforms 234

6.1 Policy fields to be addressed to ensure protection

for platform workers 250

6.2 Other fields of law and policy relevant for platform workers 253

Contents 13
Tables
1.1 Number of registered and active workers on selected digital labour

platforms, September 2020 50
2.1 Revenue model of selected online web-based platforms, January 2021 80

2.2 Subscription plans for online web-based platforms, January 2021 83
2.3 Revenue model of selected taxi platforms in selected countries, 2019–20 86

2.4 Criteria for receiving bonuses or incentives on Uber, selected countries 86

2.5 Revenue model of selected delivery platforms in selected countries,

2019–20 89

4.1 Number of respondents, by survey 136

4.2 Hourly earnings on online web-based platforms, by type of platform,


4.3 Fees paid by respondents on freelance platforms, by platform 158

4.4 Commission fees paid by app–based taxi drivers, by country

and platform 164

4.5 Proportion of respondents on online web-based platforms

covered by social protection benefits, by type of platform,


4.6 Proportion of respondents in the taxi and delivery sectors

covered by social protection benefits 176

4.7 Monitoring and organizing work on freelance platforms,

by development status and sex 178

4.8 Hourly earnings (paid and unpaid) with different education levels

on online web–based platforms, by type of platform,


4.9 Types of tasks performed by respondents on freelance platforms,

by field of study 186

5.1 Decent work for platform workers: Fundamental principles

and rights at work applicable to all workers, irrespective

of contractual status 204

5.2 Decent work for platform workers: Other key labour standards

applicable to all workers irrespective of contractual status 205

5.3 Further elements of decent work for platform workers:

Convention principles that could be adapted to all digital

labour platform workers, irrespective of contractual status 207

Abbreviations
AI artificial intelligence
API application programming interface
B2B business to business
B2C business to consumers
BPO business process outsourcing
CAIT Confederation of All India Traders
CEACR ILO Committee of Experts on the Application
of Conventions and Recommendations
FTAs free trade agreements
GDPR General Data Protection Regulation
GPS Global Positioning System
ICT information and communications technology
IPO initial public (stock) offering
IT information technology
MNE multinational enterprise
NSSO National Sample Survey Office (India)
OFN Open Food Network
OLI Online Labour Index (Oxford)
PCBU person conducting a business or undertaking
PPE personal protective equipment
PSTE persons in special types of employment
RDC remote desktop computer
SDGs Sustainable Development Goals
SMEs small and medium-sized enterprises
TaaS Talent as a Service
VPN virtual private network
WEC World Employment Confederation
WTO World Trade Organization
15
Acknowledgements
This report was prepared by the ILO Research

Department. The report was coordinated by
Uma Rani, who was the lead author along with
Rishabh Kumar Dhir, Marianne Furrer, Nóra Gőbel
and Angeliki Moraiti of the ILO, Sean Cooney (The
University of Melbourne) and Alberto Coddou
Mc Manus (Universidad Austral de Chile). Sean
Cooney was the lead author of Chapter 5 of the
report along with Alberto Coddou Mc Manus and
Angeliki Moraiti.
We are grateful to Andrea Renda (Centre for European Policy Studies
and European University Institute) and Abdul Muheet Chowdhary (South
Centre) for providing specific inputs related to competition and regulation,
and taxation issues discussed in the report. We are immensely grateful to Matías
Golman for collecting data on platform companies, their funding and revenues, and
for his assistance with the statistical analysis. Thanks to Alberto Coddou Mc Manus
for helping us coordinate a team of international legal experts: June Namgoong
(Korea Labour Institute), Ricardo Buendia Esteban (University of Bristol) and Jorge
Leyton Garcia (Pontificia Universidad Católica de Chile), who provided in-depth
analysis and insights about the regulatory mechanisms related to digital labour
platforms in the different regions. Thanks also to Khaoula Ettarfi and Hannah
Johnston for their research assistance in providing inputs to the questionnaires,
undertaking a preliminary literature review on business models and social dialogue,
organizing interviews with representatives of businesses and conducting interviews
with workers in the Middle East and Latin America. The report also benefited from
the research assistance of Yiren Wang who interned with the team.
We would like to thank Richard Samans (Director, Research Department) for his tech-
nical inputs and support, and for ably steering the process in the final stages of the
report. Thanks also to Maria-Luz Vega and Lawrence Jeff Johnson (Deputy Directors,
Research Department) for their management support. We are grateful to Manuela
Tomei (Director, Conditions of Work and Equality Department) for providing guidance
and inputs as a technical adviser. We would like to express our gratitude to Damian
Grimshaw (Former Director, Research Department, and Professor, King’s College
London) for his support towards research on digital labour, for providing valuable
inputs and suggestions, and for his continued engagement in the preparation of
this report.
The report benefited from a number of back Lebanon); Omar Gasca (independent researcher,
ground papers prepared by international experts Mexico); Youssef Sadik (Université Mohammed V
on this topic: Mariya Aleksynska (independent de Rabat, Morocco); and Natalia Kharchenko and
researcher), Digital work in Eastern Europe: Overview Oleksandr Pereverziev (Pollster, Ukraine). Thanks
of trends, outcomes and policy responses; Julie Yujie to Patrick Karanja (independent researcher, Kenya)
Chen (University of Toronto) and Sophie Sun Ping for conducting interviews with representatives
(Chinese Academy of Social Science), From flexible of business process outsourcing companies in
labour to “sticky labour”: A tracking study of workers Kenya and also for organizing meetings with key
on food-delivery platforms in China; Antonia Asenjo stakeholders and government representatives in
(independent researcher), Economía de plata Kenya during our visit to Nairobi in October 2019.
formas y condiciones de trabajo: caso de repartidores We would like to thank SoundRocket, a survey
en Santiago, Chile; Andrey Shevchuk and Denis research company specializing in social sciences,
Strebkov (National Research University Higher for providing assistance with the questionnaire
School of Economics), Freelance platform work in design and helping to execute the surveys on
Russia, 2009–2019; and Ioulia Bessa, Simon Joyce, microtask, freelance and competitive program
Denis Neumann, Mark Stuart, Vera Trappmann ming platforms. We would also like to thank Ruixin
and Charles Umney (University of Leeds), Worker Wang (Harbin Institute of Technology, China) and
protest in the platform economy. We would also like Natalia Kharchenko and Oleksandr Pereverziev
to thank our colleague from the ILO Country Office (Pollster, Ukraine) for implementing the online
for Argentina, Elva Lopez Mourelo, for preparing survey in China and Ukraine, respectively.
the report Work on delivery platforms in Argentina:
Analysis and policy recommendations. The report gained considerably from the sub
stantive inputs provided by the members of the
The report benefited from collaboration with the Research Department’s Research Review Group
Online Labour Observatory, and in particular Vili and colleagues from the ILO at the annual meeting
Lehdonvirta, Fabian Stephany, Otto Kässi and organized in November 2019. We would like to thank
Fabian Braesemann from the Oxford Internet Professors Jennifer Bair (University of Virginia), Iain
Institute, University of Oxford. We are very Begg (London School of Economics), Haroon Bhorat
thankful to Fabian Stephany for providing us (University of Cape Town), Jayati Ghosh (Jawaharlal
with data on online digital labour platforms on a Nehru University), Kamala Sankaran (University
regular basis for Chapter 1 of the report. of Delhi), Lord Robert Skidelsky (University of
Warwick) and Bart Van Ark (The Conference Board)
The surveys and interviews were implemented
for their constructive inputs and comments during
and coordinated by consultants in the coun
the entire process. We would also like to thank the
tries concerned. We would like to thank: Pablo
two anonymous peer reviewers who provided
Vinocur and Raúl Mercer (FLASCO, Argentina);
substantive comments and inputs.
Alberto Coddou Mc Manus and Antonia Asenjo
(Universidad Austral de Chile, Chile); Sophie Sun The team would like to express their gratitude
Ping (Chinese Academy of Social Sciences, China); to the following external peer reviewers for
Peter Narh and Pius Siakwah (University of Ghana, their substantive inputs and insights: Valerio
Ghana); Abhishek Kumar and Dushyant Chawla De Stefano (Katholieke Universiteit Leuven);
(independent researchers, India) and Preeti Enrique Fernández Macías and Annarosa Pesole
Mudaliar and Balaji Parthasarthy (International (Joint Research Centre, European Commission);
Institute of Information Technology, India); Torbjörn Fredriksson (UNCTAD); Guy Mundlak
Michael Martin and Hansen Julianto (Proxima ( Tel Aviv Universit y); María Luz Rodríguez
Research, Indonesia); Maggie Ireri and Grace M. Fernández (University of Castilla – La Mancha);
Maina (Trends and Insights for Africa Research, Anna Ilsøe (Københavns Universitet); Koen
Kenya); Redha Hamdan, Rania Nader and Lea Frenken (Utrecht University); Andrey Shevchuk
Bou Khater (Consultation and Research Institute, (National Research University Higher School of
Acknowledgements 17
Economics); M. Six Silberman (Organise Platform); Jealous Chirove (ILO Country Office for the United
Mohammed Amir Anwar (University of Edinburgh); Republic of Tanzania, Kenya, Rwanda and Uganda);
Padmini Swaminathan (ex-Director, MIDS) and Sara Elder, Christian Viegelahn (ILO Regional
J. Krishnamurthy (ex-ILO). Office for Asia and the Pacific); Bharti Birla, Xavier
Estupiñan (ILO Decent Work Technical Support
The report greatly benefited from detailed inputs
Team (DWT) for South Asia and ILO Country
and comments provided by the ILO Bureau
Office for India); Tendy Gunawan (ILO Country
for Workers’ Activities and the ILO Bureau for
Office for Indonesia); Andrés Marinakis (ILO DWT
Employers’ Activities, and we would like to thank
and Country Office for the South Cone of Latin
them for their engagement throughout the prep-
America); Elva Lopez Mourelo (ILO DWT and
aration of the report.
Country Office for Argentina); Anne Posthuma
Colleagues in the ILO provided valuable inputs (ILO, Cinterfor, Uruguay); Michael Braun, Maurizio
and comments in the preparation of the report. Bussi, David Mosler (ILO Regional Office for Europe
We are grateful to: Claire Harasty and Alim and Central Asia).
Khan (Office of the Deputy Director-General for
We are grateful to the French government for
Policy); Cecile Balima, Xavier Beaudonnet, Karen
their financial support in this research, as part of
Curtis, Tim de Meyer, Emmanuelle St-Pierre
the cooperation agreement 2015-2020 with the
Guilbault, Erica Martin, Irini Proios Torras, Lisa
International Labour Office.
Tortell, Anna Torriente, Maria Marta Travieso
(International Labour Standards Department); We would like to thank May Hofman and Nina
Ashwani Aggarwal, Paul Comyn, Patrick Daru, Vugman for editing and copy-editing the report,
Guillaume Delautre, Henri Ebelin, Christine as well as the ILO Publications Production Unit
Hofmann, Dorothea Schmidt-Klau (Employment (PRODOC) for the design, layout and production
Policy Department); Simel Esim, Emmanuel of the report. We would also like to thank the
Julien, Vic Van Vuuren (Enterprises Department); ILO Department of Communication and Public
Christina Behrendt, Kroum Markov, Quynh Anh Information (DCOMM) for their coordination of the
Nguyen, Shahrashoub Razavi (Social Protection launch of the report and related communication
Department); Mariangels Fortuny, Waltteri activities in different regions.
Katajamaki, Oliver Liang, Hitomi Nakagome,
Colleagues from the ILO Library have been of
Elisenda Puertas (Sectoral Policies Department);
tremendous support for this report and especially
Colin Fenwick, Youcef Ghellab, Susan Hayter,
during the COVID-19 lockdown, and we would
Konstantinos Papadakis (Governance and
like to sincerely thank them for their support. The
Tripartism Department); Janine Berg, Umberto
valuable secretarial assistance provided by Laura
Cattaneo, Olga Gomez, Martine Humblet, Martin
Finkelstein is greatly appreciated. We would like
Oelz, Shauna Olney, Esteban Tromel, Brigitte
to thank Judy Rafferty for managing the editing
Zug-Castillo (Conditions of Work and Equality
and translations.
Department); Marva Corley-Coulibaly, Angela
Doku, Veronica Escudero, Sabrina de Gobbi, Finally, we are immensely grateful to all the
Carla Henry, Tahmina Karimova, Stefan Kühn, 12,000 workers, the 85 business representatives
Hannah Liepmann, Bashar Marafie, Rossana and 14 representatives of worker associations
Merola, Ira Postolachi, Pelin Sekerler Richiardi, around the globe who agreed to participate in
Nikolai Rogovsky, Tzehainesh Teklè, Maria-Luz the ILO surveys and interviews, took the time to
Vega (Research Department); Coen Kompier, share their experiences, and provided valuable
Ken Chamuva Shawa, Jean-Marie Hakizimana, inputs without which this report would not have
Pamphile Sossa (ILO Regional Office for Africa); been possible.
Executive
summary
The digital economy is transforming the world

of work. Over the past decade, the expansion in
broadband connectivity and cloud computing,
along with innovations in information and com-
munications technologies, have enabled economic
transactions and the exchange of large amounts
of data and information between individuals,
businesses and devices. Data is increasingly a
key asset driving the digital economy. Related to
data analytics problems within a designated
these transformations is the proliferation of digital
time on competitive programming platforms; or
platforms in several sectors of the economy. Since
completing short-term tasks, such as annotating
March 2020, the COVID-19 pandemic has led to
images, moderating content, or transcribing a
an increase in remote-working arrangements,
video on microtask platforms. The tasks on loca-
further reinforcing the growth and impact of the
tion-based platforms are carried out in person
digital economy. While digital platforms provide
in specified physical locations by workers, and
a range of services and products, this report fo-
include taxi, delivery and home services (such
cuses on digital labour platforms, which mediate
as a plumber or electrician), domestic work and
work and have rapidly penetrated a number of
care provision.
economic sectors as a result of innovations in
digital technologies. The development of digital labour platforms has
the potential to provide workers, including women,
Digital labour platforms are a distinctive part of
people with disabilities, young people and migrant
the digital economy. They allow individuals or
workers, with income-generating opportunities. In
business clients to arrange a ride, order food or
developing countries in particular, such platforms
find a freelancer to develop a website or translate
are regarded as a promising source of work op-
a document, among many other activities and
portunities, leading many governments to invest
assignments. By connecting businesses and
in digital infrastructure and skills. Businesses are
clients to workers, they are transforming labour
also benefiting, as they can use these platforms
processes, with major implications for the future
to access a global and local workforce to improve
of work. Digital labour platforms can be classified
efficiency and enhance productivity, and enjoy
into two broad categories: online web-based
wider market reach.
and location-based platforms. On online web-
based platforms, tasks or work assignments are The opportunities provided by platforms are
performed online or remotely by workers. These accompanied by some challenges. For workers,
tasks may include carrying out translation, legal, these relate in particular to regularity of work
financial and patent services, design and software and income, working conditions, social protec-
development on freelance and contest-based tion, skills utilization, freedom of association and
platforms; solving complex programming or the right to collective bargaining. Many of these
challenges are quite pronounced for workers in the opportunities and overcome the challenges
informal and non-standard work arrangements emerging from the rise of digital labour plat-
and are increasingly affecting those engaged forms, to ensure sustainable enterprise devel-
on digital labour platforms, who are a relatively opment and decent work for all, and to advance
fast-growing share of the workforce. The conse- achievement of the United Nations Sustainable
quences of the COVID-19 pandemic are exposing Development Goals.
the risks and inequalities for workers, particularly
for those engaged on location-based platforms. The past decade has seen a fivefold increase
For traditional businesses, the challenges include in the number of digital labour platforms,
unfair competition from platforms, some of which are concentrated in a few countries.
which are not subject to conventional taxation
and other regulations, including those relating to The number of online web-based and location-
their workforces. Additional challenges for trad- based (taxi and delivery) platforms rose from 142
itional businesses include the amount of funding in 2010 to over 777 in 2020. The number of online
required to continuously adapt to digital trans- web-based platforms tripled over this period,
formation, especially for small and medium-sized while the number of taxi and delivery platforms
enterprises (SMEs), and the inadequate availability grew almost tenfold. A large proportion of these
of reliable digital infrastructure, particularly in the platforms are concentrated in just a few locations,
global South. including the United States of America (29 per
cent), India (8 per cent) and the United Kingdom
This report seeks to enhance our understanding of
of Great Britain and Northern Ireland (5 per cent).
how digital labour platforms are transforming the
world of work, and the implications of that trans- Digital labour platforms offer two types of work
formation for employers and workers. It draws relationship: workers are either directly hired by
on the findings of ILO surveys conducted among a platform or their work is mediated through a
some 12,000 workers in 100 countries around the platform. In the first case, they are categorized
world working on freelance, contest-based, com- as employees with an employment relationship
petitive programming and microtask platforms, to their employer, while in the second case they
and in the taxi and delivery sectors. It also draws are categorized as self-employed or independent
on interviews conducted with representatives contractors by the platforms. Those working
of 70 businesses of different types, 16 platform under an employment relationship tend to be re-
companies and 14 platform worker associations sponsible for the functioning of the platform and
around the world in multiple sectors. comprise a relatively small fraction of the platform
workforce. For instance, the freelance platform
This work provides a pioneering and compre-
PeoplePerHour has about 50 employees, while it
hensive international overview of the platform
mediates work for 2.4 million skilled workers.
business model and business strategies, based
on an analysis of the terms of service agreements Estimating the actual size of the platform-mediated
of 31 major online web-based and location-based workforce is a challenge owing to non-disclosure
platforms, and on the experiences of workers of data on the part of the platforms. Surveys by
and clients on these platforms. It also explores researchers and statistical agencies in Europe and
regulatory gaps with regard to platform govern- North America between 2015 and 2019 suggest
ance, and reviews multiple initiatives undertaken that the proportion of the adult population that
by governments and social partners to bridge has performed platform work ranges between
these gaps. Finally, it suggests ways to leverage 0.3 and 22 per cent.
On online web-based platforms, labour The business strategies adopted by digital

supply exceeds demand, placing labour platforms comprise four key elements.
downward pressure on earnings.
Four key elements enable platforms to establish a
Tracking labour supply and demand on major market base, leverage network effects and expand
online web-based platforms since 2017, the Online rapidly, while generating benefits for businesses
Labour Observatory reveals that there has been and workers.
an increase in both demand and supply for free-
Revenuestrategy: The revenue strategies of
X
lance and microtask work. Since the COVID-19
digital labour platforms are based on offering
outbreak, the labour supply on platforms has
subscription plans and charging various types
increased significantly, while the demand for
of fees to platform workers and/or the busi-
work has decreased and shifted towards tasks
nesses, clients or customers that use them.
related predominantly to software development
Online web-based platforms offer multiple
and technology. The demand for work on the five
subscription plans and customized services to
major online web-based platforms largely origi-
clients, with free trials to attract subscribers.
nates from developed countries, while the labour
They also offer workers subscription plans with
supply originates predominantly from developing
incremental benefits at extra cost, which tend
countries. The evidence indicates that on some
to be essential for accessing more work. Digital
digital labour platforms there is excess labour
labour platforms often charge a commission
supply, which leads to greater competition among
fee to workers and businesses; such fees tend
workers for task assignment and puts downward
to be higher for workers than clients on online
pressure on the price of the tasks to be performed.
web-based platforms. For instance, Upwork
generated 62 per cent of its 2019 revenue
The global distributions of investment
from various types of fees charged to workers,
in digital labour platforms and platform
while 38 per cent was generated through fees
revenues are geographically uneven.
charged to clients. On location-based platforms,
About 96 per cent of the investment in digital la- workers typically pay a commission fee on taxi
bour platforms is concentrated in Asia (US$56 bil- platforms whereas on delivery platforms it is
lion), North America (US$46 billion) and Europe businesses and customers that generally do so.
(US$12 billion), compared to 4 per cent in Latin
Recruitment and matching of workers with clients:
X
America, Africa and the Arab States (US$4 billion).
Digital labour platforms use algorithms for
Platforms providing taxi services have received a
the matching of tasks or clients with workers,
much larger share of venture capital funds than
which has been transforming a traditional hu-
delivery or online web-based platforms. Among
man resource process that typically involved
taxi platforms, the distribution of funding is
human interaction. While traditional human
uneven, with 75 per cent of funds concentrated
resource practices base recruitment selection
in only two platform companies.
largely on education levels and experience,
Digital labour platforms globally generated algorithmic matching is often determined by
revenue of at least US$52 billion in 2019. About indicators such as ratings, client or customer
70 per cent of the revenues generated were reviews, rates of cancellation or acceptance of
concentrated in just two countries, the United work, and worker profiles. On online web-based
States (49 per cent) and China (23 per cent), while platforms, this matching process may also take
the share was much lower in Europe (11 per cent) into consideration a worker’s subscription plans
and other regions (17 per cent). The seven largest and optional purchased packages. This practice
technology companies globally had a cumulative risks excluding some workers from accessing
revenue of over US$1,010 billion in 2019, and most tasks, particularly those from developing coun-
of these companies invest heavily in digital la- tries and those with lower incomes.
bour platforms as well.
Work processes and performance management:

X SMEs in particular have benefited
Algorithmic management of workers is central from location-based platforms.
to the platform business model. Platforms pro-
Many traditional businesses, particularly SMEs,
vide a variety of software and hardware tools
have started using location-based platforms,
to facilitate the work process, monitor workers
predominantly in the restaurant and retail sec-
and enable communication between the client
tors. Such businesses are increasingly relying
and the platform worker. These include moni-
on digital labour platforms as a way to cope with
toring of workers on location-based platforms
greater competition and the need to expand their
using the Global Positioning System, and tools
customer base, to keep pace with a transforming
that automatically capture screenshots or key-
marketplace and to respond to consumer prefer-
board strokes on online web-based platforms.
ences. Many restaurants are heavily dependent
Moreover, algorithms assess, evaluate and rate
on delivery platforms, particularly since the out-
platform worker performance and behaviour
break of the COVID-19 pandemic, to enhance their
using a number of metrics, such as client re-
visibility among consumers and expand their mar-
views and customer feedback.
kets, as well as to improve productivity, efficiency
Rulesofplatformgovernance: Digital labour plat-
X and profitability.
forms tend to unilaterally shape the governance
architecture within the platform through their Digital labour platforms have also supported
terms of service agreements, which have to be the growth of start-ups and the reorientation
accepted by workers, clients and businesses of some sectors.
for them to be able to access the platform.
Many digital start-ups have emerged around the
Besides requiring the observance of the codes
world, particularly in the field of artificial intelli-
of conduct regarding the use of the platform,
gence (AI), to meet the demands of automated
these agreements also cover aspects such as
work processes and analytics. As AI technology
acceptance or rejection of work, deactivation of
is still far from fully automating work, such
platform accounts and data usage. This form of
start-ups rely heavily on digital labour platforms
governance allows platforms to exercise consid-
and the human intelligence of platform workers,
erable control over platform workers’ freedom
who are dispersed globally, to complete tasks
to work, and can shape how and under what
and train machine-learning algorithms through
conditions clients or businesses engage with
a “human-in-the-loop” process.
platform workers, through exclusivity clauses,
for instance. Digital labour platforms have also made it possible
for some businesses to reorient their business
Diverse types of businesses, from start-ups strategies in certain sectors and access wider
to Fortune 500 companies, are increasingly markets. The business process outsourcing (BPO)
relying on online web-based platforms. industry, for example, is experiencing a transform-
ation wherein customer demands are now being
Businesses use online web-based platforms for
met through digital means instead of the provision
three broad reasons: to streamline recruitment
of voice-based services, and the customer journey
processes; to reduce costs and improve efficiency;
from beginning to end is managed using digital
and to access knowledge and seek innovation. The
tools. These include Facebook and WhatsApp
organizational performance of many companies
messages, web chats or emails, and AI bots for
has improved through innovations facilitated by
providing real-time feedback.
open source platforms, as well as through access
to a global pool of workers with diverse skills via
digital labour platforms.
BPO companies are also trying to sustain their labour platforms, they represent only four in ten
business by relying on work from online web- workers on online web-based platforms and one
based platforms, apart from directly working in ten workers on location-based platforms. In
with clients. Many technology companies are some countries, app-based delivery platforms
outsourcing tasks, such as content review, are an important source of work opportunities
transcription, annotation and image tagging, to for migrants.
workers in developing countries, often as part of
Gender-based occupational segregation of tasks is
their corporate social responsibility, with a view
common on freelance platforms. Women are more
to providing employment opportunities to young
likely than men to perform professional services
graduates and those from disadvantaged back-
(such as legal services, translation, writing and
grounds, for example. While it is often perceived
editing), and tasks related to business services or
that such tasks are done by AI, in practice they
sales and marketing. Few women mentioned that
require human value judgement, which is pro-
they performed tasks related to technology and
vided by BPO workers mainly based in developing
data analytics.
countries, or “invisible” workers on online web-
based platforms. Worker motivation to work on digital labour
platforms varies across the different types
While businesses can benefit from platforms,
of platforms and by gender.
challenges abound.
Complementing an existing income and the
Many businesses that depend on online web- preference or need to work from home or for job
based platforms struggle to strategically manage flexibility are the two main motivating factors
the various forms of work arrangements and risk for platform workers on online web-based plat-
losing internal human resource capacity. For forms. On freelance platforms, the preference or
businesses that depend on delivery platforms, need to work from home or for job flexibility is
high commission fees can reduce profits while the chief motivator, while on microtask platforms
poor digital infrastructure can have an impact on complementing pay from other income sources is
the smooth running of the business. Traditional the most important factor. In contrast, the main
businesses, particularly in the retail sector, are motivating factors for workers on competitive
facing market disruptions from large e-com- programming platforms are to improve skills and
merce platform companies and are confronted career opportunities. The preference or need to
with challenges such as competition issues, un- work from home or for job flexibility is particularly
favourable contractual terms, non-transparency important for women in developing and devel-
on the part of platforms (especially with regard oped countries alike. On location-based platforms,
to data, rankings and pricing), weak dispute reso- lack of alternative employment opportunities, job
lution mechanisms and, more broadly, an uneven flexibility and better pay compared to other avail-
playing field. able jobs are the key motivating factors.
Survey findings indicate that a majority Work on digital labour platforms is the main
of workers on digital labour platforms source of income for many workers…
are highly educated and male.
On location-based platforms, the overwhelming
The findings from the ILO surveys of workers on majority of workers indicated that this was the
online web-based and location-based (taxi and case. About one third of the workers on online
delivery) platforms show that the majority of web-based platforms stated that platform work
platform workers are below the age of 35 years was their main source of income; the propor-
and highly educated, in particular in developing tions were higher in developing countries and
countries. While women do find work on digital for women.
… but there are major differences between Working hours vary across location-based
the earnings of workers on online web-based platforms and online web-based platforms...
platforms in developed and developing countries.
Workers on online web-based platforms work
Average hourly earnings in a typical week for 27 hours on average in a typical week, including
those engaged on online web-based platforms both paid and unpaid work, with about one third
are US$3.4, while half of the workers on these of their time, or eight hours, spent on unpaid
platforms earn less than US$2.1 per hour. For work. About half of them have other paid jobs,
workers on freelance platforms, average hourly working 28 hours on average per week in these
earnings are US$7.6, while on microtask platforms jobs in addition to their platform work, which can
they amount to US$3.3. Workers in developing make for a long work week. Some workers on
countries tend to earn less than those in developed online web-based platforms face unpredictable
countries; on freelance platforms, for instance, work schedules and unsocial hours, particularly
they earn 60 per cent less, even after controlling in developing countries, as clients are often based
for basic characteristics and types of tasks per- in developed countries. This may have negative
formed. Earnings on online web-based platforms implications for their work–life balance.
are influenced by time spent on unpaid tasks (such
On location-based platforms, most workers in the
as looking for work or building up a profile), compe- taxi and delivery sectors work with high intensity
tition due to excess labour supply, high commission and for long hours, on average 65 hours per week
fees, and non-payment due to rejection of work. in the taxi sector and 59 hours per week in the
Evidence of the existence of a gender pay gap on delivery sector. On app-based taxi and delivery
freelance platforms is mixed. After controlling platforms, a high proportion of respondents
for basic characteristics, such as education level (79 and 74 per cent respectively) mentioned that
and work experience, at the global level there they had some degree of stress due to their work,
is no difference in hourly earnings, while at the often related to traffic congestion, insufficient
country level there is a significant gender pay gap pay, lack of orders or clients, long working hours,
in some cases. A gender pay gap is also found on the risk of work-related injury and pressure to
location-based platforms in some countries. drive quickly.
In developing countries, earnings … but many workers on both types of platforms

in the app-based taxi and delivery sectors would like to do more work.
tend to be higher than in the traditional sectors. Many workers on both online web-based and
Hourly earnings for app-based taxi drivers location-based platforms stated that they would
and delivery workers vary across the countries like to do more work than they do. They are unable
analysed in this report, and tend to be higher to do so mostly due to the unavailability of enough
than in the traditional sectors. In the taxi sector work or of well-paid tasks. Furthermore, platform
in particular, platforms are able to provide services design may also restrict workers from certain de-
to customers at low cost, hence expanding the veloping countries from accessing well-paid jobs
business. In addition, the bonuses and incentives on online web-based platforms.
provided to workers have attracted a large number
The relevance of skills and qualifications
of workers, thereby increasing the labour supply,
acquired through formal education varies
which can exceed the expected demand and
on digital labour platforms.
result in intense competition. This situation also
has the potential to reduce income-generating Platforms are redefining the relationship between
opportunities for those in the traditional sectors. formal education and access to work, as worker
In some of the countries surveyed, over 70 per profiles, ratings and reputation are vital for
cent of the traditional taxi drivers reported that accessing work. Varying degrees of vertical
compared to when they started to work as taxi and horizontal skills mismatch can be observed
drivers, the number of trips in a typical day, and on digital labour platforms. A high proportion
daily earnings, had decreased. of workers on freelance and competitive
programming platforms stated that their skills workers are unaware of any formal process for
were a good match with their work, and many were filing a complaint or seeking help in such cases.
undertaking tasks that were potentially related to On freelance platforms, when such a process
their field of study. However, skills mismatch is is known and used by workers the outcomes
quite prominent for those engaged on microtask are favourable to them in many cases. On loca-
platforms, where a highly educated workforce tion-based platforms, where workers sometimes
performs tasks that tend to require few or no face account deactivation, about half of the
specific skills. Similarly, a sizeable proportion appeals against deactivation are successful.
of workers on platforms in the taxi and delivery
sectors are highly educated. Platform workers are often unable
to engage in collective bargaining.
Working conditions on digital labour platforms In many jurisdictions, competition law prohibits
are largely regulated by terms of service self-employed workers from engaging in collective
agreements. bargaining, on the basis that they constitute a
Terms of service agreements are contracts of cartel. However, the ILO Right to Organise and
adhesion and are unilaterally determined by the Collective Bargaining Convention, 1949 (No. 98),
platforms. They define aspects related to working and the Freedom of Association and Protec-
time, pay, customer service etiquette, applicable tion of the Right to Organise Convention, 1948
law and data ownership, among others. They (No. 87), provide that freedom of association
tend to characterize the contractual relationship and collective bargaining shall be available to all
between the platform and the platform worker as workers. Some countries, such as Canada, Ireland,
other than employment, regardless of the actual Japan and Spain, have introduced exceptions for
nature of the relationship. As a result, platform certain categories of dependent self-employed
workers cannot access many of the workplace workers, which allow them to engage in collective
protections and entitlements that apply to bargaining. Another challenge to the collective
employees. organization of digital labour platform work-
ers is that they are geographically dispersed.
Platform design and algorithmic management Nevertheless, some workers based in different
are defining the everyday experiences of workers regions have been able to organize, including
on digital labour platforms. through digital means, while on location-based
platforms in particular they have also under-
Platforms use algorithms to match workers with taken strike action, initiated litigation and a drive
clients or customers, a process in which worker towards unionization. Some workers have also
ratings are decisive. The ratings are themselves al- established platform cooperatives.
gorithmically determined, according to a number
of metrics, which include acceptance and rejec- The majority of workers on digital labour
tion rates. This in effect limits workers’ ability and platforms do not have social security coverage.
freedom to reject work. A considerable number
There are large gaps with regard to health
of workers surveyed in the app-based taxi and
insurance and work-related injury provision,
delivery sectors indicated that they were unable to
unemployment and disability insurance, and
refuse or cancel work on account of the negative
old-age pension or retirement benefits. While
impact this would have on their ratings, which
access to social protection is limited, workers in
could lead to reduced access to work, lost bonuses,
the app-based taxi and delivery sectors, particu-
financial penalties and even account deactivation.
larly women, face various occupational safety and
Rejection of work or low ratings are common health risks. Not having social security coverage
on digital labour platforms, although many has created significant challenges for all platform
workers believe that the reasons for such rejec- workers during the COVID-19 pandemic, especially
tions are not always justifiable. Most platform those on location-based platforms.
A considerable number of workers on digital the surveyed workers who were provided with
labour platforms have experienced or witnessed PPE stated that the quantity or quality of PPE
discrimination or harassment. provided was inadequate. Moreover, eight out
of ten workers had incurred additional financial
Discrimination on online web-based platforms is
expenditure as they had been obliged to purchase
associated with exclusion from work opportunities
PPE themselves.
or low pay, on the basis of nationality and gender,
which was mentioned particularly by women
Regulatory responses from many countries have
respondents and workers residing in developing
started to address some of the issues related to
countries. Workers on location-based platforms
working conditions on digital labour platforms.
also indicated having faced or witnessed dis-
crimination or harassment. App-based taxi drivers Countries have taken various approaches to ex-
reported facing aggressive or rude behaviour, tending labour protections to platform workers.
mainly by clients, traditional taxi drivers and police These include:
officers, in the course of their work. App-based
Occupational safety and health: Laws in Australia
X
delivery workers mentioned instances of discrimin-
and New Zealand have adopted broader statu-
ation based on the grounds of their occupation by
tory language and extended occupational
customers, restaurants as well as the police.
safety and health coverage to all workers. In
Brazil, a judicial decision has extended existing
The COVID-19 pandemic has exposed many
safety and health legal standards to platform
of the risks confronting workers on digital
workers.
labour platforms.
Social security: Several countries have intro-
X
The ILO rapid-assessment survey in four countries
duced innovations to extend social security to
captured the implications of the pandemic for
platform workers. These include requiring that
workers on location-based platforms. The majority
platforms cover the accident insurance costs
of the workers in both the taxi and delivery sectors
of self-employed workers (France); extending
indicated declining demand, which had reduced
social security for self-employed workers (many
the earnings for nine out of ten taxi drivers and
Latin American countries); and providing work
seven out of ten delivery workers. To compensate
injury and death benefits to workers on par-
for the loss of income, some workers reported that
ticular platforms (Indonesia and Malaysia). In
they had started to engage in additional work activ-
response to the COVID-19 pandemic, some
ities, or provided taxi and delivery services outside
countries have extended sickness benefits to all
the platforms through their private contacts; many
workers (Ireland) and unemployment benefits
had also reduced unnecessary expenditure, used
to uninsured self-employed workers (Finland
savings, deferred payment of bills, or taken a loan.
and the United States).
Some workers on location-based platforms
Employment relationship: Employee status
X
reported working throughout the crisis due to
remains important, as most labour and social
economic necessity, despite feeling anxiety about
protections are associated with it. Countries
contracting COVID-19 while at work. Seven out of
have adopted various approaches to the classi-
ten workers indicated not being able to take paid
fication of platform workers, often arising from
sick leave, or to receive compensation, in the event
litigation, which fall along a spectrum between
they were to test positive for the virus, thus risking
very broad and very narrow approaches to
the health of others in addition to their own health.
employment status. These include: (i) classi-
Some location-based platforms have undertaken fying them as employees, often based on the
specific measures to mitigate occupational safety amount of control exercised by the platform;
and health risks among workers, including the (ii) adopting an intermediate category in order
provision of safety training and personal pro- to extend labour protection; (iii) creating a
tective equipment (PPE). However, about half de facto intermediate category to ensure that
they obtain certain benefits; and (iv) classifying Given that digital labour platforms operate
them as independent contractors, often based across multiple jurisdictions, there is a need
on the degree of their flexibility and autonomy. for some form of international policy dialogue
Working time and remuneration: Some new
X
and coordination.
approaches to labour standards have been Governments and non-state actors are in many
specifically adapted to digitally based work. For cases regulating digital labour platforms, but
instance, French law provides that a platform’s these initiatives vary considerably. Countries face
voluntary social charter should include the challenges in enforcing regulations, particularly
“right to disconnect” and methods of enabling with regard to online web-based platforms,
self-employed platform workers to obtain a where the platforms, clients and workers are
“decent price” for their work. located in different jurisdictions. In this regard,
Dispute resolution: Some platforms may restrict
X the ILO Maritime Labour Convention, 2006, sets
dispute resolution to a particular jurisdiction an important precedent as it concerns an industry
through arbitration clauses, which can be with multiple parties operating across different
limiting for workers. This has been successfully jurisdictions. Such an approach could also be
challenged in some jurisdictions; the Supreme considered for digital labour platforms. Another
Court of Canada, for example, invalidated a important point of departure is the ILO Tripartite
platform’s arbitration clause on the ground Declaration of Principles concerning Multinational
that it “makes the substantive rights given by Enterprises and Social Policy, 2017, which provides
the contract unenforceable”. guidance to multinational enterprises on social
policy and inclusive, responsible and sustainable
Accesstodataandprivacy: Governments are
X
workplace practices.
increasingly adopting measures regarding data
protection and privacy, including in Brazil, India, International policy dialogue and coordination
Nigeria and the European Union. In France, a are also vital to ensure regulatory certainty and
recent amendment to the Labour Code gives the applicability of universal labour standards,
self-employed platform workers in the trans- given the diversity of responses by countries and
portation industry the right to access data platform companies. It is important that the ILO
related to their platform activities. fundamental principles and rights at work are im-
plemented for all platform workers, irrespective
With growing regulatory concerns,
of their status. In addition, principles rooted in
platform companies and worker organizations
other ILO Conventions, such as those related to
have also been addressing the issues raised.
fair payment systems, fair termination and access
to dispute resolution, should also be extended to
In Denmark, a collective bargaining agreement
platform workers.
between a trade union and a cleaning platform
has allowed some platform workers to transition
A way forward…
to employee status. Platform companies have also
been developing codes of conduct either unilat- A way forward would be to engage in a process of
erally or in collaboration with other platforms global social dialogue aimed at ensuring that the
to address some of the challenges confronting opportunities arising from digital labour platforms
workers. Six digital labour platforms have signed are leveraged, and the challenges addressed, so
the World Economic Forum Charter of Principles that digital labour platforms are best positioned
for Good Platform Work, which covers issues such to provide decent work opportunities, foster the
as safety and well-being, flexibility, fair conditions, growth of sustainable enterprises and con-
social protection, voice and participation, and tribute towards achievement of the Sustainable
data management. Development Goals. The ILO’s independent Global
Commission on the Future of Work recommended
the development of an international governance ensuring that workers’ employment status is

X
system that sets certain minimum rights and pro- correctly classified and is in accordance with
tections and requires platforms and their clients to national classification systems;
respect them. It also called for a “human-in-com-
ensuring transparency in ratings or rankings of
X
mand” approach to algorithmic management,
workers and businesses using digital platforms
surveillance and control in order to ensure that
such as online web-based, location-based and
“final decisions affecting work are taken by
e-commerce platforms;
human beings”.
ensuring transparency and accountability of
X
The ILO’s Centenary Declaration for the Future of
algorithms for workers and businesses;
Work calls for “policies and measures that ensure
appropriate privacy and personal data protection, protecting workers’ personal and work data,
X
and respond to challenges and opportunities in as well as data relating to businesses and their
the world of work relating to the digital transform- activities on platforms;
ation of work, including platform work” in order to
working towards ensuring that self-employed
X
promote inclusive and sustainable development,
platform workers enjoy the right to bargain
full and productive employment and decent work
collectively, for example through greater har-
for all.
monization of competition law with labour law;
These objectives can best be achieved through
reaffirming that anti-discrimination and occu-
X
social dialogue among the relevant stakeholders,
pational safety and health laws apply to digital
most particularly the digital labour platforms, the
labour platforms and their workers;
platform workers, and their representatives and
governments. A concerted effort across multiple ensuring adequate social security benefits
X
international forums and organizations will be for all workers, including platform workers,
critical to ensuring that digital labour platforms by extending and adapting policy and legal
develop further in a manner that strongly contrib- frameworks where necessary;
utes to inclusive and sustainable development.
ensuring fair termination processes for plat-
X
Such a process of regulatory dialogue and coord-
form workers;
ination should have at its core an effort to ensure
that domestic laws implementing the fundamental ensuring access to independent dispute reso-
X
principles and rights at work as well as other lution mechanisms;
key legal provisions, such as those in respect of
ensuring that platform workers are able to ac-
X
occupational safety and health and social security,
cess the courts of the jurisdiction in which they
apply to all workers, including digital labour plat-
are located if they so choose;
form workers. With the right engagement and
preparation, this process could lead over time to providing for wage protection, fair payments
X
a clearer understanding and a more effective and and working time standards;
consistent approach at the enterprise, national
allowing platform workers to move freely
X
and international levels, with a view to:
between platforms, including by facilitating
ensuring fair competition and creating an en-
X portability of workers’ data, for example regard-
abling environment for sustainable enterprises; ing ratings; and
requiring and promoting clear and transparent
X aiming at effectively taxing the digital economy,
X
terms of engagement and contractual arrange- including platforms, clients and workers, as well
ments for workers and businesses, including as their transactions.
as reflected in labour and consumer laws;
1 The digital
transformation
of industry
and the world
of work
The rise of Platforms rely on two distinct types of workers
digital labour
platforms Workers directly
employed by
the platform
99designs 139 1,200,000
Appen 800 1,000,000
HackerRank 200 11,000,000
5
Concentration of
Meituan 54,580 3,987,000

platforms in
rise
PeoplePerHour 50 2,400,000
since 2010
Rappi 1,500 25,000

USA
29% Uber 26,900 5,000,000
Workers mediated
India by the platform
Hybrid
Delivery
8%
Taxi
Online
web-based
United
ICT-enabled
Kingdom
5%
2000 2010 2020 Data-driven
Exponential growth
in the number of platforms
119
Algorithmically
managed
US$ billion
Global investment in platforms 49 % 11 % 23 % 17 %
United States Europe China Other
Taxi regions
US$ 62 billion
96%
invested in Asia,
North America
Delivery and Europe
US$ 37 billion
4%
52
Hybrid
US$ 17 billion invested in
Latin America,
US$
Africa and
the Arab States billion
Global revenue generated by platforms
Online web-based
US$ 3 billion
1. The digital transformation of industry and the world of work 31
Digital labour platforms

Introduction have emerged as a distinctive
The pace at which technological advances and
feature of the digital economy.
innovations are taking place is unprecedented.
The information and communications tech-
nology (ICT) revolution of the early 1990s led to a clients to workers, and provide new opportunities
rapid diffusion and adoption of the internet that for both workers and businesses. In addition,
transformed a number of economic sectors and technological advances have facilitated new ways
reshaped regional, national and international of organizing work, thereby transforming work
markets. It led to a geographical fragmentation processes and how people work. The COVID-19
of industry as firms could subcontract, outsource pandemic has further reinforced their role in
and offshore through global supply chains at a the economy. The expansion of such platforms
relatively low cost. The expansion of broadband has occurred alongside the increased relevance
connectivity and the availability of high-speed of data – particularly big data – that can now be
internet enabled the rapid development of digital stored and analysed through cloud computing.
infrastructure from the early 2000s. Widespread These developments have been supported by
use of the internet and ICT devices by both the availability of venture capital funds, which
businesses and individuals paved the way for web- have played a fundamental role in financing the
based economic transactions (on platforms such diffusion of digital platforms.
as Amazon and eBay), and laid the foundation for
Digital labour platforms offer income-generating
the digital economy (Castells 2010).
opportunities to workers and their flexible work
Over the past decade, the availability of cloud arrangements may be more convenient for
infrastructure and computing services has facili- certain workers, such as women, persons with
tated the growth of digital platforms that have disabilities and young people. They also provide
gradually penetrated almost all sectors of the opportunities for those marginalized in traditional
economy. One can identify three broad categories labour markets, such as refugees and migrant
of such platforms: those that provide digital workers. In addition, they provide an avenue for
services and products to individual users, such workers to complement their earnings from low-
as social media; those that mediate exchange of paying or seasonal jobs (Surie and Sharma 2019).
goods and services, such as e-commerce or busi- Because digital labour platforms are emerging
ness-to-business (B2B) platforms; and those that as an important source of income-generating
mediate and facilitate labour exchange between opportunities, many governments in developing
different users, such as businesses, workers and countries are investing in digital infrastructure
consumers, including digital labour platforms such and supporting training programmes developed
as Upwork or Uber. These platforms are redefining by the private sector to equip the workforce with
the means of economic exchange and increasingly digital skills (Graham, Hjorth and Lehdonvirta
shaping the world of work. 2017; Heeks 2017).
This report focuses on two main types of digital Digital labour platforms bring significant benefits
labour platforms: online web-based platforms, to businesses. For instance, online web-based
where tasks are performed online and remotely by platforms have enabled businesses to access
workers and are allocated to a crowd (on microtask workers and to source talent globally, allowing
and competitive programming platforms) or to them to reduce costs and improve productivity
individuals (on freelance and contest-based plat- (Corporaal and Lehdonvirta 2017). With regard to
forms); and location-based platforms, where tasks location-based platforms, businesses are able to
are performed at a specified physical location benefit from access to a wider market, a broader
by individuals such as taxi drivers and delivery customer base and labour supply, and improved
workers (see figure 1.1). These platforms have revenues and productivity. Other forms of digital
emerged as a distinctive feature of the digital platforms, such as e-commerce platforms, enable
economy in the way they connect businesses and businesses to sell their products to a wider market
(for example, small and medium-sized enterprises 2016 was about US$4.8 billion. The total revenue
(SMEs) selling on Amazon). of one of the biggest online web-based plat-
forms – Upwork – increased from US$164 million
Along with these opportunities, several challenges
in 2016 to US$301 million in 2019 (Upwork 2020).
have also emerged for both businesses and
As digital labour platforms continue to rapidly
workers. Many SMEs face potential competition
proliferate and increasingly shape the world
issues as a result of aggressive pricing by digital
of work, addressing the challenges arising for
labour platforms. For workers, the challenges
workers and businesses will be critical to fully
relate to regularity of work and income, working
leveraging the income-generating potential of
conditions, social protection and access to their
digital labour platforms and meeting the United
fundamental rights of freedom of association
Nations Sustainable Development Goals (SDGs). In
and collective bargaining. As such the digital
this regard, it is beneficial to explore the core func-
transformation has the potential to increase in-
tioning of digital labour platforms’ business model
formal and non-standard work, which can result
and their interactions with other businesses, with
in income and job insecurity (OECD 2020a). The
a view to gaining a better understanding of the ex-
COVID-19 pandemic has further revealed the enor-
periences of businesses and workers that engage
mous risks in these areas for workers engaged on
with these platforms. This report thus reviews the
digital labour platforms (ILO 2020a and 2020b).
opportunities and challenges that digital labour
This report focuses on the rapid changes that platforms present to businesses and workers, and
digital labour platforms bring to work, work prac- the nature of regulations and public policies that
tices and the business landscape. Digital labour might be required to ensure that both workers
platforms are blurring the previously clear distinc- and businesses are protected and able to sustain
tion between employees and the self-employed. and thrive in the process.
Global economic competition is further resulting
in the growth of atypical work arrangements and Chapter 1 traces the rise of the digital economy
a corresponding decline in employment-related and digital platforms, focusing on digital labour
and other benefits, as well as a polarization of platforms in particular. It assesses the impact of
the workforce (Berg 2019). In addition, innovative such platforms on different economic sectors and
digital technologies are changing human resource labour markets, highlighting the distinctive aspects
management practices, as algorithms increasingly of digital labour platforms that are transforming
replace humans in allocating, evaluating, and ad- the world of work. It gives some estimates of the
ministering rewards for work mediated through number of platform companies and the number
these platforms. These profound and rapid of workers whose work is mediated through these
changes have major implications for workers’ well- platforms based on the available literature. It also
being and working conditions around the world, discusses the roles of data and finance in the rapid
especially in middle- and low-income countries. rise of these platforms, and the challenges the
The regulation of digital labour platforms has platforms pose to both businesses and workers.
hence been under discussion in several countries,
Chapter 2 discusses the business strategies and
with debates under way particularly on the role
key elements of the platform business model of
of regulatory frameworks for ensuring decent
both online web-based and location-based plat-
work on these platforms and fair competition
forms based on an analysis of the terms of service
for businesses.
agreements of 31 major platforms and interviews
Although digital labour platforms are at a rela- with representatives of 16 online web-based and
tively early stage of development, they have location-based platforms. The key elements of
been growing rapidly over the past decade. Kuek the platform business model which are discussed
et al. (2015), on the basis of interviews with rep- include revenue models and pricing strategies,
resentatives of online microtask and freelance recruitment practices, algorithmic management
platform companies and data disclosed by them, of work processes and evaluation of workers, and
estimated that their global annual market size in rules of platform governance.
Chapter 3 examines the diffusion of digital labour 12,000 respondents globally. It presents a first
platforms across various sectors of the economy, major overview of the worker experience on
and explores how and why businesses use them, digital labour platforms in multiple sectors and
based on interviews with representatives of 70 countries, particularly in developing countries.
SMEs and large enterprises. Using case studies, Chapter 5 takes a broad approach to regulation
it explores the opportunities arising from digital to describe the forms of governance and initiatives
labour platforms for new digital technology undertaken by platforms, governments and social
start-up companies and business process out- partners to address the emerging challenges.
sourcing (BPO) companies. It also analyses the Chapter 6 suggests policies that may be required
implications of such platforms for traditional at the national, international and multilateral
enterprises, particularly SMEs in the retail sector. levels to ensure decent work for workers and fair
competition for enterprises on digital labour plat-
Chapter 4 explores the oppor tunities and
forms. Box 1.1 provides definitions of key terms
challenges for workers on digital labour plat-
and concepts used in this report.
forms, based on surveys conducted with some
Box 1.1 Terminology used in the report

X
Information and communications technology (ICT) covers a range of technological aspects and
includes internet access, data, cloud computing, software, and hardware, among others. ICT is
used in areas ranging from telecommunications, broadcast media and audio-visual processing
to finance, medicine, social media, and digital labour platforms. ICT incorporates both the
internet-enabled sphere as well as the mobile one powered by wireless networks, although it
also includes older technologies, such as landline telephones, radio and broadcast television.
Information technology (IT) is a subset of ICT and is more specifically the use of computer sys-
tems, including all hardware and software, as well as peripheral equipment and infrastructure.
Digital economy “incorporates all economic activity reliant on, or significantly enhanced by the
use of digital inputs, including digital technologies, digital infrastructure, digital services and
data. It refers to all producers and consumers, including government, that are utilising these
digital inputs in their economic activities” (OECD 2020b, 5).
Digital platforms are online entities providing digital services and products. These digital services
facilitate “interactions between two or more distinct but interdependent sets of users (whether
firms or individuals) who interact through the service via the Internet” (OECD 2019a, 21). These
interactions can include exchange of labour, goods (e-commerce) or software.
Digital labour platforms facilitate work using “digital technologies to ‘intermediate’ between
individual suppliers” (platform workers and other businesses) and clients (EU 2020, 1), or directly
engage workers to provide labour services. The work undertaken on these platforms is also
commonly referred to as “platform work” or “gig work”.
Algorithmic management refers to giving the responsibility of assigning tasks and making deci-
sions to an algorithmic system of control, with limited human involvement. The algorithmic
management system improves through self-learning algorithms based on data.
Worker is defined in accordance with the ILO’s international labour standards, which include
both employees and the self-employed (or independent contractors). Workers on digital labour
platforms are also called “gig workers”, “crowdworkers” or “platform workers” in the literature.
A taxonomy of how these workers are described by different platforms in their terms of service
agreements is presented in Appendix 2, table A2.3.
Client refers to users of digital platforms, whether businesses, firms or consumers.
1.1 The rise of the digital economy
The rise of the digital economy is associated risks exacerbating inequalities, particularly
with the development of a number of software between countries. Addressing this divide re-
technologies, such as artificial intelligence (AI), quires concerted policy action. Even developing
cloud computing and blockchain, among others. countries that have a stronghold in IT-enabled
In the past decade, cloud computing, high-speed and software services, such as India, lag “behind
connectivity and data storage capacities have in terms of internet bandwidth, connection speed
expanded significantly, enabling economic trans- and network readiness” (UNCTAD 2018, VIII).
actions and the exchange of large amounts of data
Consequently, the growth of platform companies
and information between individuals, businesses
is concentrated in certain parts of the world,
and devices. At the same time, innovations by
while developing countries have emerged pre-
technology companies, such as Google and Apple,
dominantly as users of such platforms. Developing
along with the availability of software as an open
countries continue to face challenges, mostly due
source and as a service, have led to widespread
to the above-mentioned shortcomings in digital
use and adoption of smart mobile phones,
infrastructure, as well as inadequate financial in-
computers and servers by both individuals and
frastructure, human resources and institutional
businesses (Evans and Schmalensee 2016). These
capacities. Examples include insufficient access
platforms are driving innovations and have
to capital and unavailability of a technologically
generated the development of a wide range of
skilled labour force, as well as the lack of a fa-
applications across many sectors of the economy,
vourable regulatory environment for technology
which has the potential to bring about industrial
entrepreneurship. In this context, it becomes
transformation. In the process platforms have
imperative to probe some of the key features that
also provided new entrepreneurial opportunities
have facilitated the rise of the digital economy,
in the digital realm that are not only enabling the
and to better understand the opportunities and
creation of new products, services and processes,
challenges it has created for economic processes
but are also transforming erstwhile offline labour
that are transforming today’s world of work.
processes and business models. The rise of the
digital economy can hence be situated at an inter-
section at which ICTs and their users, both people 1.1.1 Key features
and businesses, increasingly rely on digital modes of the digital economy
of exchange, both socially and economically.
An increasingly prominent aspect of the digital
The rise of the digital economy at the country
economy is its ability to provide access to a variety
level is contingent upon the availability of digital
of solutions “as a service” due to the widespread
infrastructure, which remains uneven around the
availability of cloud infrastructure and cloud
globe. Digital infrastructure in developing coun-
computing (see box 1.2). The availability of cloud
tries still lags behind that in developed countries.
infrastructure services has allowed the digital
This is largely because “high costs of additional
economy to evolve into a much more diverse
international bandwidth to access overseas
environment, and these services are playing a
servers and data centres still limit the uptake of
critical role in shaping the global economy. Some
cloud services” in many developing countries
of the key features of the digital economy are:
(UNCTAD 2019, 8). As a result, in 2019 most cloud
traffic was generated in North America, followed Asset-lightness. The availability of cloud service
by Asia and the Pacific and Western Europe, which infrastructure allows firms to reduce costs related
together accounted for about 90 per cent of that to leasing or renting hardware and
traffic (UNCTAD 2019). The uneven growth of the downloading software and applica-
digital economy perpetuates a digital divide and tions, and to manage on-demand
access to applications or storage through a cloud Mobility. Cloud infrastruc-

provider (see box 1.2). This makes businesses agile ture services allow platform
and allows them to focus on core services. businesses to conduct their re-
gional or global operations virtually
The availability of software application programs
from any location, irrespective of where their
and “tools as a service” on platforms reduces
clients, suppliers or consumers are based. It is one
duplication costs and improves productivity, as
of the distinct features of platform businesses that
these programs can be used for similar tasks or
they are able to leverage the intangibles – that is,
be customized for new tasks, meaning that pro-
software, applications and tools – that are at the
gramming code does not have to be written from
core of their business (OECD 2014).
scratch (Boudreau, Jesuthasan and Creelman
2015; Lakhani, Garvin and Lonstein 2012). This
reduces developers’ costs in terms of time and Box 1.2 Cloud infrastructure
X
money and improves their productivity. Over and computing services
time, with increased use of programming code,

the number of applications and tools available There are three main cloud infrastruc-
to platform users increases. The availability of ture and computing services:
software application programs and tools as a Infrastructure as a Service consists of
service also offers an environment for rapid de- cloud computing infrastructure, such
velopment and improved productivity. as hardware, virtual machines, servers,
Network effects. The success of a platform de- cloud storage and networks, that firms
pends on its ability to attract a sufficient number can rent or lease. The services are pro-
of users from all sides of the market (clients and vided by platform companies such as
workers). Platforms adopt both pricing and non- Amazon Web Services (AWS), Microsoft
pricing strategies – such as providing free access Azure and Google Compute Engine, and
or rewards – to attract users from the different are also available on open source plat-
sides of the market. These strategies create more forms such as OpenStack, CloudStack
value for users and attract even more users in and Nimbus.
order to create a critical mass, thereby creating Platform as a Service is a cloud comput-
network effects (Evans and Schmalensee 2008). ing service that provides components
Platforms also attract and retain third-party such as operating systems, program-
developers to innovate and add value to the ming language and development tools,
platform by providing them with access to appli- database management and web servers.
cations and tools at low or zero cost (Boudreau These services are offered by AWS Elastic
and Jeppesen 2015). In these ways, platforms Beanstalk, Microsoft Azure, Google App
create network effects. Engine, and other platform companies.
Datafication. The increase in computing power They are also available on open source
and the availability of cloud storage have enabled such as Dokku, Flynn and Apache Stratos.
data collection, storage and analysis on a mas- Software as a Service offers users soft-
sive scale and at a far more rapid pace than ever ware or applications over the internet
before. Data has become integral to platform through a client interface. This includes
businesses, as it can be monetized, for example various statistical programmes, software
through targeted advertising. Data can be used packages, Dropbox, Slack, and Google
for myriad purposes, such as predicting consumer Apps, among others. They are also
behaviour, improving products and services, and available on open source such as Apache
managing workers via algorithms. Hadoop software library.
Source: Adapted from OECD (2014).
Cloud infrastructure
1.1.2 The rise of digital platforms has facilitated the growth
Digital platforms have been able to build on some of digital platforms.
of the distinct features of the digital economy, and
have penetrated diverse sectors of the economy
(see section 1.2). In addition, increasing reliance
on ICTs, from smartphones to computers, has This distinct feature also creates challenges from
created multiple opportunities for platform a regulatory point of view for two reasons. First, it
businesses to emerge and thrive. Moreover, the is always possible for users to disguise their loca-
nature and organization of the digital economy tion using virtual personal networks (OECD 2014).
has further facilitated the rapid rise of platform Second, when users and platform companies are
businesses. For instance, the availability of cloud based in different countries the application of
infrastructure services at reduced costs, along labour and tax laws becomes complicated, as such
with the availability of venture capital funding, laws differ across jurisdictions (see sections 5.3.9
has reduced entry barriers and enabled the rapid and 6.3).
growth of digital platforms over the past decade
Finally, the rise of digital platforms has created
(Cusumano, Gawer and Yoffie 2019).
entrepreneurial opportunities for digital tech-
Cloud infrastructure has facilitated the growth of nology start-ups and third-party developers
digital platforms in many countries and regions, to innovate and develop new products, tools,
as it makes them asset light. Investment by plat- application programs and services on platforms
forms in traditional capital assets, such as cars, (Miric, Boudreau and Jeppesen 2019), which drives
hotels or warehouses, is often minimal; platforms further digital transformation (see section 3.3.2).
tend to invest instead in digital infrastructure
Digitalization allows datafication through the
and are overwhelmingly dependent on data,
collection of massive amounts of data. User
skills, ideas and physical assets provided by their
data has emerged as one of the most valuable
users (both clients and workers). For example,
assets for platforms, as it provides a basis on
Uber does not heavily invest in cars, but it has
which new products can be built and serves
been able to expand and scale in 69 countries
to improve efficiency and productivity. Earlier
at an unprecedented pace (within 11 years of its
data-processing methods and software were not
creation) (Uber 2020a). It has 26,900 employees
adequate to handle large amounts of data, and
and 5 million drivers, who either own or lease
innovative methods were required for processing
cars, with the majority of them being labelled as
such data (Sheriff 2018). Some data-processing
self-employed or “driver-partners” (Uber 2020a
methods carried out by humans, such as tagging,
and 2020b; Appendix 2). Uber orches-
classifying, categorizing, cleaning, structuring
trates its services through its app,
and organizing, remain relevant, as, despite
which is its “linchpin” (algorithmic
developments in AI, they cannot be fully
management), by matching customers
automated. Digital labour platforms, such
with drivers: its key assets are the
as microtask platforms, emerged due to
network of users (drivers and
the failure of AI to classify images, sounds and
consumers), data and the brand
texts, as human intelligence is required to process
(Teece 2018a, 43).
such data (Irani 2015). For instance, when
Furthermore, cloud infrastruc- Amazon was developing its product cata-
ture ser vices allow platform logues with a view to making it easy
businesses to be virtual and for buyers to access them through
mobile as they operate with in - the search function (a process com-
tangibles, and their users (clients plicated in particular by duplicate
and workers) can be based product entries), the solution was to
anywhere in the world. create an internal website that enabled
employees to go through catalogue entries and 1.1.3 Open source innovation

mark any duplicates when they had some time
(Silberman 2015). One major contributor to the rise of the digital
The availability of this innovative tool (the internal economy has been the availability of open source
website) through which tasks could be performed platforms for software applications (such as
in a quick and efficient manner led Amazon to Apache Hadoop, GitHub), which can be accessed
start Amazon Mechanical Turk (AMT) in 2005, by both digital and non-digital firms at zero cost.
whereby a wide variety of simple data-processing Open source software platforms are used by a
tasks could be done by workers from around number of digital labour platforms, including Uber
the globe in a cost-effective manner (Silberman and Upwork. For example, GitHub, an open source
2015). For instance, data can be processed in repository of software, application programs and
two days by engaging 60,000 workers from the tools, allows users (firms or developers) to access
crowd on platforms, instead of hiring hundreds and customize these programs and tools without
of workers to finish the task in few weeks (Irani having to make any substantial investment of
2015). Recognizing the power of the platform to time and money. The availability of tools and pro-
complete tasks at such a rapid pace using a global grams through open source also provides many
pool of workers led to the rise in microtask plat- platforms with an opportunity to diversify across
forms. Such platforms have been instrumental in a range of services or sectors, depending on the
the processing of data needed by many digital and demands of the users, in a short time span and
non-digital companies (see sections 1.4 and 3.3.2). with low investment.
The outsourcing of work through digital labour Open source platforms and software are in-
platforms has resulted in the creation of an invis- creasingly being recognized as working tools for
ible workforce tasked with cleaning, processing innovation. This is partly because open source
and organizing vast amounts of data, often in software is free to acquire and thus offers an al-
precarious working conditions (see Chapter 4) to ternative to the spread of software that has either
meet the needs of a data-driven digital economy. a general-purpose or special licence. Platform
This outsourcing has allowed firms to benefit from companies and large firms work with open source
the double advantage of reducing costs and at the platforms instead of competing with them, as this
same time building data archives which can be gives them access to a pool of diverse knowledge
used for machine learning and training algorithms and capabilities, which in turn speeds up the in-
for future automation (Rani and Singh 2019). Such novation process at zero cost (see section 3.1.3;
invisible and – for all intents and purposes – unpaid Gawer 2014). For instance, Google opened up its
work is even more prevalent on taxi platforms, Android patents,1 and Tesla opened up the com-
where the drivers, apart from transporting people, pany’s patent portfolio to external developers
are in the process also feeding data into the com- for free so that they could innovate and develop
pany database to be used in training algorithms, tools, programs or software. 2 Open source col-
which then automate the management of the laboration is also gaining ground among public
company’s operations, such as dispatching drivers sector agencies that are opening up their patents
or surge pricing (Chen and Qiu 2019). The workers to developers for free: the National Aeronautics
are often unaware that they are doing this “data and Space Administration of the United States
work”, and they are not compensated for it. of America, for example, has made hundreds of
1 For more information, see: http://techcrunch.com/2015/07/23/google-offers-to-sell-patents-to-startups-to-boost-its-wider-cross-

licensing-initiative.
2 For more information, see: https://www.digitaltrends.com/cars/good-guy-elon-musk-opens-teslas-patents-gives-free-access-
technology/. This was later followed by other car manufacturers such as Ford, see: http://www.digitaltrends.com/business/ford-
to-open-electric-vehicle-patents-news-pictures/.
patents available to developers. 3 The existence economy and who should participate in it. They
of open source platforms and software has also use various mechanisms, such as licensing of their
encouraged these agencies to facilitate innov- intellectual property rights, and technical frame-
ation and development by offering open access works,4 to provide access to cloud services (Teece
to their intellectual property to external firms or 2017; Parker, Van Alstyne and Choudary 2016).
third-party developers through their application They also regulate access to the use of platform
programming interface. infrastructure by determining whether they will be
more “open” or “closed” (to attract a select group
of participants) (Zhao et al. 2019).
1.1.4 Concentration of market
The concentration of power among platform com-
power among a few platform panies raises challenges for countries as they shape
companies their economies, particularly when governments
and businesses are seeking to establish secure and
Cloud services and computing providers tend decent employment for their workers, a situation
to be concentrated among a few large multi- which primarily affects developing countries. For
national enterprises such as Alibaba, Alphabet example, online microtask platforms such as AMT
(includes Google), Amazon, Apple, Facebook, outsource data-processing, clerical and low-end
Microsoft and Tencent. While some of these tasks, which are used for training AI, to workers
companies also manufacture products, they are dispersed around the world. Consequently, while
predominantly platform businesses which are these platforms create opportunities to earn an
able to employ some of the distinct features of income, the quality of the work created raises some
the digital economy to create new and extremely important considerations. The question of quality
efficient ways of facilitating the interaction of large of work arises not only in terms of remuneration,
numbers of users, applications and businesses regularity of work and social protection, but also
or service providers. They are geographically in terms of the content of the work, as such tasks
concentrated in just two countries, China and the can be repetitive, low end and mind-numbing,
United States, and the estimated annual revenue and they are often performed by highly educated
of these seven companies combined for 2019 workers (see section 4.1.6). As a large proportion
was about US$1,010 billion (see figure 1.14 in of workers in developing countries continue to
section 1.5). The concentration of wealth among work in the informal economy, this development
a few platform companies may in some instances trajectory of the digital economy might push highly
allow them not only to coordinate, steer and educated and skilled workers in these countries to
manage innovation and development, but also to pursue work under precarious or informal working
shape infrastructure development in the digital conditions and uncertain work arrangements
economy. Their dominant position allows them to (World Bank 2020), and therefore requires careful
determine the boundaries governing the digital policy considerations.
3 For more information, see: http://www.nasa.gov/press-release/nasa-offers-licenses-of-patented-technologies-to-start-up-com-

panies.
4 The term “technical framework” refers to defining the boundaries where innovation can take place.
1.2 Digital platforms: Pervading and penetrating

different sectors of the economy
Digitalization is permeating different sectors of Some platforms, such as Google and Facebook,
the economy, thereby improving efficiency and also offer search functions and advertising to
creating new sources of value. Figure 1.1 provides users, which reduces their search costs.
a landscape of digital platforms, showing that
In addition, social media platforms such as
almost all major economic sectors are experi-
Facebook are disrupting the advertising sector
encing penetration. The use of digital platforms
as they draw on the large amounts of data re-
in the various sectors of the economy is quite
lating to their 2.8 billion users5 to enable clients
diverse. Platforms can be classified into three
to reach audiences across geographical locations
broad categories: those that offer digital services
(Fumagalli et al. 2018). The availability of cloud in-
or products to individual users, such as search
frastructure is also dramatically transforming the
engines or social media; those that facilitate
news and media industry. Online news and media
and mediate between different users, such as
platforms are competing fiercely with traditional
business-to-business (B2B); and digital labour
news outlets, with a significant impact on the
platforms. While most platforms can be allocated
latter’s businesses and workers.
to one of these categories, some “hybrid” plat-
forms provide services across multiple categories.
The penetration of digital platforms is having a
profound impact on firms and sectors, as they Digitalization is
reorganize markets and work arrangements,
affecting competition and challenging regulatory permeating different
models, thereby altering the rules of the game
(Kenney and Zysman 2016). This section provides sectors of the economy.
a glimpse into some of these impacts across the
various sectors of the economy.
There are also some platforms that provide video
streaming services to individuals and businesses,
1.2.1 Digital platforms as well as serve as social media platforms, such
that offer services as YouTube, which have not only created oppor-
tunities for content creators to earn incomes by
to individual users posting videos, but are also disrupting the adver-
tising industry. For instance, YouTube generated
Digital platforms are shaping social and economic
more than US$34 billion in advertising revenue
exchanges, from social media platforms such as
over three years (Alexander 2020). However, on
Facebook or TikTok that occupy an increasing role
social media platforms such as Facebook, YouTube
in the socio-economic lives of people around the
and TikTok, the labour of users posting content
world, to communication platforms such as Skype,
who are often not paid or paid very little is vital
WhatsApp, Viber or Zoom, which are playing a
for generating network effects and revenues. This
rapidly growing role in maintaining business con-
report does not examine this type of labour.
tinuity and remote working during the COVID-19
pandemic and also in people’s personal lives.
5 As of December 2020, Facebook had 2.8 billion monthly active users. For details, see: https://investor.fb.com/investor-news/
press-release-details/2021/Facebook-Reports-Fourth-Quarter-and-Full-Year-2020-Results/default.aspx.
40
Figure 1.1 Landscape of digital platforms
Services provided Types of digital platforms
The role of digital labour platforms in transforming the world of work

• Social media platforms • Facebook • TikTok • Twitter
• Electronic payment platforms • PayPal • Paystack • Paytm
• Crowdfunding platforms • Catarse • Ketto • Kickstarter
Provide services News, media and entertainment • Apple TV+ • BuzzFeed • Netﬂix
to individual
users
Advertising • Gumtree • Kenhoo • OLX
• Other digital services platforms

Search, information and reviews • Feedly • Google Search • Yelp
Rental goods and assets • Airbnb • Homestay • Makemytrip
Communication • Skype • Viber • Zoom
Applications marketplace • Apple App Store • Aptoide • Google Play Store
Freelance and contest-based • 99designs • Kabanchik • Upwork

Online web-based platforms
Microtask • AMT • Clickworker • Microworkers
Competitive programming • Codeforces • HackerRank • Topcoder
Medical consultation • 1Doc3 • DocOnline • MDLive
Mediate work • Digital labour platforms Taxi • Bolt • Ola • Uber
Delivery • Meituan • Rappi • Uber Eats
Location-based platforms Home services • Doit4u • Task Rabbit • Urban Company
Domestic work • Batmaid • BookMyBai • SweepSouth
Care services • Care24 • CareLinx • Greymate Care
Retail and wholesale • Alibaba • Amazon • Mercado Libre

Facilitate Manufacturing marketplace and analytics • AnyFactory • Laserhub • Xometry
and mediate • Business to business (B2B) platforms
exchange Agriculture marketplace and analytics • Agri Marketplace • FarmCrowdy • Ninjacart
Financial lending and analytics • Ant Group • Avant • Nummo
Mediating work Services provided include:

and providing • Hybrid digital platforms delivery, taxi, retail, entertainment, • Jumia • Gojek • Grab
other services electronic payment
Digital platforms have facilitated access to a for both businesses and workers are not yet as
number of products, such as software programs profound as in the retail sector. The past few years
or streamed music, which can be delivered have also observed a rise in hybrid platforms
digitally and remotely to consumers and busi- which offer both labour and other services such
nesses. This has led to a shift from tangible to as e-commerce and payments.
intangible products, and can potentially lead to
Retail platforms. The most successful B2B and
unfair competition, as streaming music on Spotify,
B2C models are in the online retail sector. The
for example, might not be taxed while an imported
most successful examples of these models include
CD would be taxed. This can have implications for
Alibaba, Amazon and Flipkart. The global retail
revenue generated through customs and tariffs,
e-commerce market size in 2019 was valued at
especially for developing countries, as there is cur-
US$4.25 trillion (Grand View Research 2020). The
rently a moratorium on customs duties regarding
e-commerce platforms compete with small retail
electronic transfer of products and services. In
stores and offer better pricing, as they are able
this context, there is an ongoing discussion on
to reduce transaction costs and costs related to
the World Trade Organization (WTO) digital trade
renting stores and hiring retail personnel. The
rules: there is no consensus among countries,
decline in retail businesses, exacerbated by the
including developing countries regarding the con-
shutdown associated with the COVID-19 pan-
tinuation of the moratorium, with some preferring
demic, has the potential to displace thousands of
to end it in order to access revenue that could be
jobs in physical retail stores. A study conducted
used to finance digital infrastructure or other
in Nordic countries shows that e-commerce rev-
public goods (Rani and Singh 2019; UNCTAD 2018).
enues tripled over the period 2008 to 2018, and
For instance, countries could use the funds to allo-
there was a 27 per cent increase in revenue in the
cate resources for social protection expenditures
first quarter of 2020 (Rolandsson 2020).
which have been squeezed during the COVID-19
crisis, with major implications for the welfare of The rise of these platforms and new digital tech-
workers (see section 4.2.5; Behrendt, Nguyen and nologies is also affecting retail employment; in
Rani 2019). Nordic countries its share in total employment
declined between 2009 and 2019 (Rolandsson
1.2.2 Digital platforms 2020). It has also created challenges regarding the
quality of the new jobs that have been generated,
facilitating and mediating particularly those associated with logistics (such
exchange between users as warehouse and delivery workers for Amazon),

where much of the employment growth has been
The availability of cloud infrastructure has led to in last-mile delivery and as low-paid jobs in ware-
a rise in B2B, business-to-consumers (B2C) and houses (MIT 2020). Many of the delivery workers
digital labour platform business models (see are being classified as independent contractors
figure 1.1). The rise of such platforms is reshaping and as such find themselves outside the scope of
the business landscape and changing the bound- employment protection, with irregular and low
aries of the firm, while also creating competition pay, and no protection in case of accidents at work
and opportunities for traditional businesses. (De Stefano 2019).
In the B2B and B2C domain, the online retail Several retail platforms have enabled SMEs and
sector has seen a prominent rise, which has led individual entrepreneurs to access a larger cus-
to a disruption of the traditional retail sector with tomer base by selling their products through
a significant impact on both retail stores and the the platforms. For example, 60 per cent of the
workers they employ. Penetration of platforms products sold on Amazon are from third-party
into other sectors, such as manufacturing, agri- sellers (1.7 million SMEs) (Bezos 2020). While
culture and finance, is at a comparatively early enabling access to a larger customer base, retail
stage and the implications of these platforms platforms also tend to charge different types of
fees to third-party sellers for every unit sold. In individual customer needs in a more flexible and
2018, it was estimated that third-party sellers paid timely manner, and also provide factory-related
Amazon US$39.7 billion in fees;6 its referral fee analytics rapidly for improved efficiency. This
ranges from 6 per cent (personal computers) to might also have implications for working condi-
45 per cent (Amazon device accessories).7 The high tions as workers may have to work longer hours
fees charged by the platform have had a major to meet the delivery deadlines.
impact on the earnings of these small businesses
Agriculture platforms. The agricultural sector
during the COVID-19 pandemic, many of which
has seen a rise in the number of farm manage-
had already been struggling due to the slowdown
ment software tools and technologies, such as
of the economy.
Agricolus, for providing market solutions and
Many platforms have their own range of products, improving productivity. Use of the Internet of
which compete with those from third-party sellers. Things with sensors to collect real-time data and
However, the scale of the data that platforms are integrated monitoring systems to create optimal
able to gather and analyse in making decisions conditions for sowing, watering, fertilizing and
about products to be sold, or their price, or about harvesting is becoming increasingly widespread
attracting users or customers, has allowed them ( Jayne, Yeboah and Henry 2017). These tools,
to consolidate their position in the market. The along with big data analytics, help in optimizing
pricing decisions based on such data can there- agricultural operations through precision
fore have a large impact on third-party sellers on farming, or in improving crop yields and environ-
the platform as well as traditional retail stores, mental management, among other applications.
due to information asymmetry. The competition Other digital platforms, such as Agri Marketplace,
faced by enterprises, particularly SMEs, from are connecting farmers with markets. Although
platforms both within and outside the platform the opportunities and challenges arising from
marketplace, has started to come under scrutiny the use of these digital tools and platforms in the
(see section 3.4). agricultural sector are not yet well understood,
they are considered to have the potential to bring
Manufacturing platforms. The manufacturing
benefits to smallholders, particularly as a result of
sector is progressively undergoing a digital
improved planting and crop rotation and through
transformation, whereby the supplier relationship
access to wider markets in the years to come.
is mediated through digital business platforms
such as Laserhub, Tao-factory or Xometry. Some Financial platforms. Another sector where sig-
platforms, such as Tao-factory, which operates nificant transformations are taking place is the
largely in garment and light industries, connect financial sector, which has become increasingly
enterprises with consumers or customers on diverse and is competing with the traditional
e-commerce platforms, such as Taobao. Once a banking sector. For instance, the entry of com-
buyer on the e-commerce platform places the panies such as Apple, Alphabet (includes Google)
order, the value chain is set in motion, and the and Ant Group into the financial services sector,
order is manufactured and delivered within seven and the rise of payment platforms such as PayPal,
to nine days (Butollo 2020). Other platforms, such Paytm, Venmo or TransferWise, are having a
as Laserhub and Xometry, connect suppliers notable impact on the traditional banking sector
with material processing industries (Butollo and (N.L. Johnson 2020). Many of the large companies
Schneidemesser, forthcoming). The platform net- in particular are able to leverage their existing
works are based on geographical proximity, and relationship with customers and their data and
such developments can compete with traditional to cross-subsidize their new offerings. Further in-
business and supply chains as they can respond to novation and expansion in new financial services
6 For more information, see: https://www.marketplacepulse.com/marketplaces-year-in-review-2019#google.

7 For more information, see: https://sell.amazon.com/pricing.html#referral-fees.
have the potential for financial inclusion of those individual workers providing labour services at
engaged in the informal economy in developing the homes of individual customers. Digital labour
countries. This trend could result in innovations platforms are enabling the reorganization of activ-
in traditional banking operations in response to ities that have conventionally relied on traditional
the competition due to digital transformation, employment relationships into work performed
and could lead to scaling down of employment. by independent contractors or the self-employed.
Work is often performed on an on-demand basis,
wherein the logic of the “just-in-time” inven-
1.2.3 Digital labour platforms tory system is applied to the labour process
mediating work (Vallas 2018, 49). Compensation is on a
piece-rate basis, and workers, who are de-
Digital labour platforms are the predominant form fined as self-employed, are required
of platform connecting workers with businesses to provide their own capital
and clients, and have significant implications for equipment (Stanford
the world of work. It is these platforms that are the 2017; Drahokoupil and
main focus of this report. Currently, there are two Fabo 2016).
main types of digital labour platform: online web-
based platforms, where tasks are performed online Although platform workers
and remotely by workers; and location-based plat- are usually classified as inde-
forms, where tasks are performed at a specified pendent contractors, they often
physical location by individuals (see figure 1.1). do not have the freedom and autonomy
Online web-based platforms include microtask, to organize their work. Moreover, innovative
freelance, contest-based, competitive program- platform practices such as algorithmic manage-
ming and medical consultation platforms, while ment are used to allocate work and manage,
location-based platforms include those offering supervise and reward workers (see section 4.3.1).
taxi, delivery, domestic, care and home services. Digital labour platforms have tremendous con-
Much attention has been given in recent years trol over the organization of work and workers’
to location-based platforms such as Deliveroo, compensation, while “still claiming to be only an
Glovo and Uber, especially in developed countries. intermediary” (Kenney and Zysman 2016, 62). Such
Online web-based platforms are also gaining outcomes of technological advances represent a
popularity among businesses. Many freelance return to the past as the workers are engaged as
and competitive programming platforms, such as casual labour and paid on a piece-rate basis, which
Upwork and Topcoder, though less well-known, adds to the growing informal or non-standard
have been operating for over two decades. workforce in developing and developed countries
alike. This situation presents new challenges to
Online web-based labour platforms offer varied
traditional work arrangements and the standard
ser vices to both individual customers and
employment relationship (see section 5.3.10), as
business clients. For instance, freelance and
well as exacerbating existing challenges, notably
contest-based platforms enable workers to
the use of non-standard forms of work.
connect with clients for specific tasks ranging
from translation to graphic design, while medical A number of digital labour platforms compete
consultation platforms allow individuals to access with businesses in traditional sectors, relying
medical advice from doctors online. Through the on data and competitive pricing. Location-based
use of technology, location-based platforms me- platforms, such as taxi platforms, have disrupted
diate various services, such as taxi and delivery established transportation business models by
services, which often continue to operate in harnessing data and algorithms to match
parallel with traditional labour passengers with drivers in real
markets. Other such mediated time (Clewlow and Mishra 2017).
services include domestic, Uber, for example, entered a
care and home services, with tightly regulated taxi market in
Work is outsourced
on these platforms by
the United States and challenged traditional taxi
drivers by offering low-cost subsidized fares and businesses in the global
allowing individuals who were not licensed taxi
drivers to offer rides (Horan 2019). Furthermore,
North, and performed by
Uber’s entry into and surge in the market have workers in the global South.
been funded extensively by venture capital funds
despite the losses incurred since Uber’s inception
(see section 1.5). Similarly, freelance platforms,
market, followed by the Philippines (US$16 million)
such as Freelancer or Zhubajie (ZBJ), are competing
and Ukraine (US$13 million). Overall, the picture
with traditional employment agencies by matching
of outsourcing work through digital platforms
tasks with workers for businesses at a lower cost
has not changed compared to 2013 (Graham
and without the need to abide by the
et al. 2017), while the volume of transactions has
protections associated with an
increased and almost all countries now have a
employment relationship.
higher share of domestic employers outsourcing
M o r e o v e r, a t r e n d h a s tasks on these platforms. Hence, online labour
developed towards out- markets are more diffused around the globe.
sourcing work, both
By engaging with platform workers in locations
low-skilled and high-
with lower price and wage levels, businesses
skilled, especially as
can further reduce their costs, while providing
traditional businesses
employment opportunities. The median hourly
look to digital labour plat-
wages on the platform, which is the inner circle
forms and digital tools to
illustrated in figure 1.2, are clearly higher in de-
meet their needs for human resources. These
veloped countries than in developing countries.
platforms host workers from around the world,
However, the geographical location where the
enabling businesses to complete their tasks at a
tasks are completed is related not only to the
faster pace and lower price than if the tasks were
price level but also to skill requirements, both
performed on site. In many instances, the work
technical and language, as well as the availability
is outsourced on these platforms by businesses
of IT infrastructure. For instance, among de-
in the global North, and performed by workers in
veloping countries a much larger share of tasks
the global South. This is illustrated by data from
is completed by workers in South Asia and East
200,000 projects collected on a major freelance
Asia compared to Central Asia and sub-Saharan
platform for the period January to December
Africa. This is despite low wage levels in the latter
2019.8 Figure 1.2 displays whether the demand
regions and can be attributed to the availability of
for work comes from within the country or from
the requisite IT and other skills and infrastructure
abroad, and the size of the bubble shows the
in the former regions. As a result, a number of
inflow of trade (volume of work) in millions of US
developing countries are investing in the devel-
dollars coming into the country. The data collected
opment of IT infrastructure in order to be able to
shows that the demand for work largely originates
benefit from this outsourcing model. In such a
from Australia, Canada, Germany, New Zealand,
context, it is of critical importance to analyse the
the United Kingdom of Great Britain and Northern
opportunities and challenges arising from digital
Ireland, and the United States. A large proportion
labour platforms.
of this work is performed by workers in developing
countries, particularly in India (US$26 million), This report focuses on the rise of digital labour plat-
which accounts for almost 20 per cent of the total forms to gain a more nuanced and comprehensive
8 In order to map the countries where online work is performed, transaction data for 2019 was obtained from one of the largest
online freelance platforms. This platform offers services across a range of occupations. The anonymized transaction data was
obtained for the period January to December 2019 using the application programming interface; this is a sample of all projects on
the platform, with a total volume of US$135 million.
Figure 1.2 Outsourcing of tasks on a freelance platform across countries, inflow of work and earnings, 2019
SWE UKR
SVK
NOR
FIN BLR
CAN GBR
DNK ROU
IRL POL MDA RUS
NLD DEU SRB
HRV GEO
BIH FRA MKD KAZ
USA PRT ITA BGR ARM CHN
JPN
ESP ALB TUR
GRC NPL
MAR CYP ISR PAK
TUN ARE BGD
DZA EGY
JAM THA
MEX DOM
VNM
LCA SEN BEN
SLV NGA ETH
COL PHL
NIC VEN
CRI KEN MYS
GHA UGA IND SGP
ECU CMR
IDN
PER
BRA
BOL
MDG MUS AUS
LKA
Foreign employer CHL
URY ZAF
ARG NZL
Local employer
Total inflow of remaining countries (in million US$)

Total inflow
in million US$ RUS Northern, Southern and Western Europe 1.2 (16 countries)
1 Latin America and the Caribbean 0.7 (23 countries)

Eastern Europe 0.5 (2 countries)
Median hourly earnings in US$ Arab States 0.3 (8 countries)

East Asia 0.3 (5 countries)
10
3 10 15 20 25 35 South-East Asia and the Pacific 0.1 (8 countries)
Central and Western Asia 0.1 (4 countries)
Note: For country codes see Appendix 1.
Source: Data collected by Fabian Braesemann, Oxford Internet Institute, iLabour Project.
understanding of the implications of their rapid taxi and delivery services, which not only comprise
growth for both businesses and the workers some of the largest and most well-funded labour
whose work is mediated through such platforms. platform companies globally, but also mediate
The various types of digital labour platforms work for a large number of workers. Some of these
examined in this report include: freelance, platforms have had a far-reaching social and eco-
contest-based, competitive programming and nomic impact in many countries, at times severely
microtask platforms, which are among the leading disrupting long-standing traditional sectors. The
online web-based platforms facilitating labour ex- report makes an attempt to understand the
changes between workers and clients (including nuances of these diverse types of digital labour
businesses), covering multiple forms of activities, platform, and the emerging opportunities and
skills and tasks. Location-based platforms include challenges for the world of work today.
1.3 Digital labour platforms:
Estimates of the number of platforms and workers
The traditional statistical methods used in en- services. Among the online web-based platforms,
terprise and worker surveys do not fully capture the majority are freelance platforms (181), with a
these types of digital labour platforms or the lower number of microtask (46), contest-based
number of people whose work is mediated by (37) and competitive programming (19) platforms.
them and their working conditions. This raises
a huge challenge with regard to estimating the
number of workers involved and the number of
digital platforms in operation as well as the extent The number of
of their penetration. This section presents some
estimates of and trends in relation to the number digital labour platforms
of active digital labour platforms, using new online
databases, and some estimates of the numbers
has grown rapidly
of workers engaged or mediated through these
platforms, using various sources. Finally, based
over the past decade.
on data from major English-speaking online web-
based platforms, certain trends in demand for
Online web-based platforms, such as Elance (today
work and the supply of labour are also presented.
Upwork, after its merger with oDesk) (1999) and
Topcoder (2001), were pioneers in setting up
1.3.1 Number of digital crowd-based digital labour platforms as a business
model. Whereas Elance tried to build a global mar-
labour platforms ketplace to connect freelancers with employers,
Topcoder tried to build a “community of program-
The number of digital labour platforms, both
mers” who could re-use basic computer program
online web-based and location-based, has grown
components and find innovative solutions to
rapidly over the past decade. Focusing on online
software problems, thus saving clients time and
web-based platforms (microtask, freelance and
money (Lakhani, Garvin and Lonstein 2012, 2).
competitive programming) and location-based
The notion of using “crowdworkers”, which gained
platforms in the taxi and delivery sector, glo-
momentum from the beginning of the 2000s, led
bally, there were at least 777 active platforms9
to the growth of online web-based platforms. The
operating in January 2021 (based on data from
global recession of 2008–09 fostered the develop-
the Crunchbase database;10 see figure 1.3). The
ment of online labour platforms across different
number of platforms in the delivery sector is
regions, as businesses came to rely on them for
the highest (383), followed by online web-based
outsourcing various tasks (see figure 1.3).
platforms (283), taxi sector (106) and there are
five hybrid platforms which provide varied types The global recession of 2008–09 also saw the rise
of services such as taxi, delivery and e-commerce of taxi and delivery platforms as an alternative
9 This figure would be much higher if all types of digital labour platforms were included, as in figure 1.1.
10 Crunchbase is a database that contains business information about private and public companies and start-ups. It obtains its
data in four ways: the venture program, machine learning, an in-house data team, and the Crunchbase community. The venture
program allows investors to keep their firm’s Crunchbase profile up to date and provides members with free access to company
data on Crunchbase and other discounts. Members of the public can submit information to the Crunchbase database. The list of
companies and start-ups in the database provides data on their location, funding history, investment activities, acquisition trends
and number of employees. It covers platforms from 98 countries around the globe. As it is self-reporting, it is likely that some
active platforms, especially from developing countries, are not listed in the database.
Figure 1.3 Number of active digital labour platforms globally, selected categories
400 383
350
Online
Number of platforms
web-based
300 283
250 Taxi
200
Delivery
150
106
100 Hybrid
50
5
0
2005
1999
2000
2001
2002
2003
2004
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
Note: Only currently active platforms are included.
Source: Crunchbase database.
to traditional taxi and delivery services: by using workers who undertake platform work. Despite
technology, clients could access these services at the absence of such transparency, an attempt is
a competitive price, with the platforms also pro- made here to provide estimates based on surveys
viding work opportunities. These platforms gained conducted by researchers and statistical offices.
popularity among many different users and grew This section also examines the issue of excess
rapidly between 2012 and 2018 (see figure 1.3). supply of labour on online web-based platforms.
During this period, activities on delivery platforms
expanded from food delivery services to grocery, Workers directly hired
courier services and more. The past five years have
seen a growth in hybrid platforms such as Grab by digital labour platforms
and Jumia which offer a wide range of labour and Data on the number of employees directly hired
other services, and some of the taxi and delivery by digital labour platforms is available for 749 plat-
platforms are also shifting towards a hybrid model. forms (96 per cent) of the 777 platforms, either
from annual reports or databases (Crunchbase
1.3.2 Number of workers and Owler). These employees perform tasks
related to the creation, maintenance and overall
engaged on digital labour functioning of the platform, and are hired on a
platforms full-time or part-time basis, or on a fixed-term
basis (Kenney and Zysman 2018a). Platforms also
Digital labour platforms offer two types of work hire freelancers for the development and main-
relationships: workers who are directly hired by tenance of the platform. For example, Upwork,
the platforms (internal employment), and workers apart from employing 570 employees globally,
whose engagement and work are mediated also “engaged over 1,200 freelancers to provide
through the platforms (external employment) services … on a variety of internal projects” in 2019
(ILO, EU and OECD, forthcoming). Numbers in (Upwork 2019, 4). Information on the engagement
the latter category are particularly challenging of freelancers by platforms is difficult to capture
to estimate due to the paucity of data, as most unless platform companies declare it in their
platforms do not disclose the number of active annual reports.
Figure 1.4 Global share of digital labour platforms, by number of employees, January 2021
40
35 Online
Percentage of platforms
web-based
30
25 Taxi
20
Delivery
15
10 Hybrid
5
0
1–10 11–50 51–100 101–250 251–500 501–1000 1001–5000 5001–10 000 ≥10 001
Number of employees
The analysis of the available data shows that in Workers whose engagement
terms of employment, many online web-based
and location-based platforms are micro and small
and work are mediated through
enterprises, directly employing either fewer than digital labour platforms:
10 employees or 11–50 employees (see figure 1.4). Estimates based on surveys
Only a few delivery and taxi platforms have more
than 1,000 employees. Uber is the largest em- The lack of transparency on the part of platforms
ployer among taxi platforms (26,900 employees; in sharing data has led researchers and statistical
mainly highly skilled professionals such as lawyers, offices to use surveys to estimate the number
marketing experts, software engineers and other of workers whose work is mediated by the plat-
professionals) (Uber 2020a). It also has taxi drivers forms. The research shows large variations in the
who are full-time employees (Kenney and Zysman estimates due to definitional and methodological
2018a). In the delivery sector, a number of plat- differences. On the definitional differences, sur-
forms, including Meituan, Delivery Hero, Swiggy veys have used broad or narrow definitions of
and Ele.me, have more than 10,000 employees. the types of platform covered and the reference
These large entities, apart from hiring employees period in question. In terms of types of platform,
for managing and running the platforms, also hire the broad definitions used cover digital labour
delivery workers on a full-time or part-time basis. platforms, e-commerce, rental and payment
This strategy allows them to ensure a reliable platforms, while narrower definitions are re-
service as they capture market share, and they stricted to digital labour platforms (both online
change their labour practices once they achieve web-based and location-based). Concerning
their objective (see section 2.3.1). the reference period, broad definitions include
individuals who have performed tasks or have
ever worked or earned money on a platform, or
did so during the previous year, while narrow
definitions are restricted to whether they have
worked on a platform in the previous month or
week, or do so on a monthly or weekly basis. On
Figure 1.5 Estimates of workers engaged on digital platforms based on surveys (percentage of adult population)
Types of platforms covered
Digital labour platforms Digital labour platforms, e-commerce and rental platforms
Reference period
Ever 14 EU Member States, 20171 9.7 United States, 201513 22

7 European countries, 2016–172 9–22
Last year Canada, 2015–163 0.3

Switzerland, 20194 0.4 Switzerland, 20194 1.6
Norway, 2016–175 1 Denmark, 20176 2.4
Denmark, 20176 1 United States, 201610 4.5
Sweden, 20167 2.5 Finland, 201714 7
United Kingdom, 20168 4
16 EU Member States, 20189 11
Last month/ United States, 201610 1.1

monthly 14 EU Member States, 20171 7.7
16 EU Member States, 20189 8.6
Last week/ United States, 201511 0.5

weekly United States, 201712 1
1
Pesole et al. (2018); 2 Huws et al. (2017); 3 Canada, Statcan (2017); 4 Switzerland FSO (2020); 5 Alsos et al. (2017); 6 Ilsøe and Madsen (2017);
7 Sweden SOU (2017); 8 CIPD (2017); 9 Urzì Brancati, Pesole and Fernández Macías (2020); 10 Farrell, Greig and Hamoudi (2018); 11 Katz and
Krueger (2016); 12 United States BLS (2018); 13 Burson-Marsteller, Aspen Institute and Time (2016); 14 Statistics Finland (SF) (2018).
Source: ILO compilation based on the above sources.
the methodological differences, the surveys follow States). When the time period is further narrowed
either an income-based or a job-based approach. down to the past month, then the estimate of
An additional definitional difficulty relates to workers engaged on digital labour platforms in
having a clear understanding of the definition of these 16 EU Member States declines to 8.6 per
“platform” among the respondents. cent of the adult population. Narrowing down to
the previous week, the estimates show that the
Using a very broad definition, estimates indicate
proportion of workers who are engaged on digital
that 22 per cent of the working-age population in
labour platforms varies between 0.5 per cent in
the United States have offered some kind of good
the United States and 12 per cent in selected
or service using a digital platform, and about one
European countries.
third of them also reported earning at least 40 per
cent of their monthly income from platform work Some surveys have also captured the propor-
(see figure 1.5; see also Appendix 1, table A1.2). tion of the population that uses digital services,
However, the estimates in other countries range covering digital labour, e-commerce and rental
between 1.6 per cent (Switzerland) and 7 per cent platforms in Canada and the United States (see
(Finland) if the past year is used as the reference Appendix 1, table A1.2). The findings in the United
period. Focusing more narrowly on having ever States show that about 42 per cent of the adult
worked or earned income only on digital labour population has purchased or used one of the ser-
platforms, the estimates vary between 9 and vices (Burson-Marsteller, Aspen Institute and Time
22 per cent for selected European countries. If the 2016). The Canadian labour force survey also cap-
time period is narrowed down to the past year, tured the proportion of the adult population that
estimates range between 0.3 per cent (Canada) has used taxi or accommodation services, which
and 11 per cent (16 European Union (EU) Member amounted to 9.5 per cent (Canada, Statcan 2017).
Workers whose work is mediated The proportion of active workers on these

platforms is measured either by the number of
through online web-based projects completed or by the income earned
platforms: Estimates based on since their registration on the platform, as a
data available on platforms proxy. About one third of registered workers
have completed at least one project successfully
Information on the number of workers registered on these platforms (PeoplePerHour, 99designs
on various platforms can be found on the web- and Workana). If the threshold is increased to ten
sites of the platforms themselves (see table 1.1). projects, considering workers having earned a
Not all registered workers are active and able to reasonable amount of income from the platform,
access tasks and work on a regular basis, however, then the share of active workers drops to 10 per
which results in an overestimation of those num- cent or less.
bers. Workers may also be registered on multiple
On Freelancer and Guru, the number of active
platforms and thus possibly be counted twice,
workers was captured using incomes earned from
which makes it difficult to estimate the number of
these platforms. According to data retrieved on
workers who depend on platforms to earn a living.
Freelancer, 95,813 workers were registered on
A recent study has attempted to collect and an- the platform, and while a large proportion of
notate publicly accessible data on freelance and them (73 per cent) had earned some income, only
contest-based platforms (Pesole and Rani, forth- 27 per cent had earned more than US$1,000 (see
coming). Data was retrieved, whenever technically table 1.1). On Guru, meanwhile, out of 1.05 million
feasible, from the online interface (website or registered workers, only 0.5 per cent had earned
application) of five freelance and contest-based any income and 0.1 per cent had earned more than
platforms (see table 1.1). The data obtained re- US$1,000. The large differences in the proportion
lates to September 2020 and shows that Guru has of active users on various platforms could be partly
about 1 million registered workers, while on the due to some platforms, for instance Freelancer,
other four platforms the number ranges between charging workers a fee for maintaining their in-
42,000 and 126,000 workers. active account (see table 2.1), while platforms such
Table 1.1 Number of registered and active workers on selected digital labour platforms,
X
September 2020
Active or successful workers

Number Oversupply
of registered of workers
At least one project/ More than 10 projects/
workers (%)
more than US$1 more than US$1000
PeoplePerHour* 126475 29143 (23%) 10798 (9%) 91.0
99designs* 42781 15794 (37%) 4271 (10%) 90.0
Workana* 95600 26312 (28%) 4820 (5%) 95.0
Freelancer** 95813 69993 (73%) 26195 (27%) 73.0
Guru** 1048575 4862 (0.5%) 1385 (0.1%) 99.9
* Refers to active workers based on completed projects. **Refers to active workers based on income earned.
Note: “Oversupply” is defined as the difference between registered and active workers (more than 10 projects/more
than US$1000 earned) on the platform. Figures in parentheses are percentages of total registered workers.
Source: Pesole and Rani (forthcoming).
as Guru do not charge membership or basic fees There has been an increase in both labour demand
and may allow workers to have accounts even if and supply for such work on online web-based
they are inactive. The different approaches and platforms between 2017 and 2020. The onset of
strategies used by the platforms make it difficult the COVID-19 pandemic and the resultant switch
to provide any reliable estimates of workers whose to remote work and teleworking have led to an in-
work is mediated through these platforms on crease in demand for such work from mid-April to
the sole basis of information available on regis- June 2020, unlike previous years. The data shows
tered workers. that supply has been rising faster than demand, in-
cluding during the COVID-19 crisis (see figure 1.6).
Overall, the lack of common definition and
This indicates that while it may be easy to register
methodological approaches, as well as lack of
as a worker on a platform, being able to receive
transparency on the part of the platforms are an
work and earn a substantial amount of income can
obstacle to estimating the number of workers
be quite difficult, especially since workers have to
whose work is mediated through digital labour
compete with one another globally to obtain the
platforms. This calls for digital labour platforms to
tasks posted on platforms. The excess supply of
be transparent and disclose the number of active
labour on platforms can also be deduced from
workers whose work is mediated through them.
table 1.1, which shows that more than 90 per cent
of the workers on some platforms are unable to
1.3.3 Trends in labour demand find projects to work on or earn an income. This
is not specific to freelance and contest-based
and supply on selected online platforms; it can also be observed on microtask
web-based platforms platforms where the number of registered users
is far higher than the number of tasks posted,
Data tracked on the four largest English-language which results in competition for tasks even when
online web-based platforms shows that the the remuneration for performing the tasks is low
number of registered workers on these platforms (Dube et al. 2020).
has been increasing since 2017 (see figure 1.6).
Some researchers have analysed AMT, a microtask
This data has been collected by researchers at
platform, to show that the excess supply of labour
the Oxford Internet Institute since 2016 and rep-
and the monopsony among platforms do not en-
resents at least 70 per cent of the market traffic
courage businesses to price their tasks at a higher
for work mediated through online web-based
rate and allow them to fix rates convenient to them
platforms and involving workers and requesters
(Dube et al. 2020; Kingsley, Gray and Suri 2015).
from 105 countries (Kässi and Lehdonvirta 2018).
In fact, this trend impacts on the distributional
The data on labour supply captures the number
gains on these platforms since it has a consid-
of workers registered on these platforms (though
erable effect on wages, “with workers paid less
not necessarily active), and labour demand cap-
than 13 per cent of their productivity” (Dube et al.
tures the number of public projects and tasks
2020, 44). Some platforms have recently changed
that are posted by clients. This data has been
their strategies to address the excess supply of
used to construct the Online Labour Index (OLI),
labour by offering membership or subscription
which measures the use of online labour platforms
plans and charging additional fees so that workers
“over time and across countries and occupations”
have better opportunities to access tasks on the
(Kässi and Lehdonvirta 2018, 241).11
platform (see section 2.2 for details).
11 The index is based on tracking all projects and tasks posted on five platforms (Freelancer, Guru, AMT, PeoplePerHour and
Upwork). For details about the methodology used to construct the index, see Kässi and Lehdonvirta (2018).
Figure 1.6 Online global labour supply and demand on major online web-based platforms, 2017–21
350
300
Index (1 Aug. 2017 = 100)
250
200
150
100
50
0
Aug. 17
Oct. 17
Jan. 18
July 18
Apr. 18
Oct. 18
Jan. 19
Apr. 19
July 19
Oct. 19
Jan. 20
Apr. 20
July 20
Oct. 20
Jan. 21
Demand Supply Linear (supply)
Note: Labour supply is captured from four platforms (Fiverr, Freelancer, Guru and PeoplePerHour). Labour demand is captured from
five platforms (Freelancer, Guru, AMT, PeoplePerHour and Upwork). The data is retrieved every 24 hours from each platform.
Source: Online Labour Observatory (iLabour Project, Oxford Internet Institute and ILO).
Demand for and supply
Figure 1.7 Global demand for labour across of labour across occupations
occupational categories on five major online
web-based platforms, 2018 and 2020 The tasks performed on these platforms
can be classified into the following occu-
Professional 2 2018 pational categories: software development

services 3
2020 and technology; creative and multimedia;
Sales and 10 writing and translation; clerical and data
marketing support 11 entry; sales and marketing support; and
professional services. Globally, a large pro-
Clerical 12
and data entry 10
portion of tasks are completed in the field
of software development and technology,
Writing 14 whose share increased from 39 per cent
and translation 12
to 45 per cent between 2018 and 2020 (see
Creative 23
figure 1.7). Professional, and sales and mar-
and multimedia 20 keting services also gained in importance,
Software
whereas occupations such as creative and
39
development multimedia, writing and translation, and
45
and technology clerical and data entry declined between
0 5 10 15 20 25 30 35 40 45
Percentage share
2018 and 2020.
Source: As for figure 1.6.

Figure 1.8 Distribution of global labour supply and demand on major online web-based platforms,
by country and occupational category, 2018 and 2020
(a) Online labour demand
50
40
Percentage share
30
20
10
0
2018 2020
2018 2020
2018 2020
2018 2020
2018 2020
2018 2020
2018 2020
2018 2020
United States United Australia Canada India Germany Singapore Israel

Kingdom
(b) Online labour supply
35
30
25
Percentage share
20
15
10
2018 2020
2018 2020
2018 2020
2018 2020
2018 2020
2018 2020
2018 2020
India Bangladesh Pakistan United States Philippines United Ukraine

Kingdom
Clerical and data entry Professional services Software development and technology
Creative and multimedia Sales and marketing support Writing and translation

The clients who demand such work are largely from the United States has declined over the
based in developed countries, with four of the same period, while that from the United Kingdom
top five countries belonging to this group (see has increased.
figure 1.8a). Globally, in 2020 about 40 per cent of Given the large, highly educated English-speaking
the demand for such work was from clients based workforce in India, it is not surprising that the
in the United States. Compared to 2018, however, share of platform work completed by workers from
the share of demand from the United States for that country is quite substantial. The high global
such work has declined while that from Australia, demand for software development and technology
Canada, Germany, India and the United Kingdom has also led to an increase in the supply of labour
has increased. The share of tasks or projects for these tasks. The rise in the share of total supply
posted by clients in these countries remains com- coming from India was driven by an increase in the
paratively small. The demand for such work from share of labour supply in software-related tasks,
clients in Europe, excluding the United Kingdom, which is consistent with the extensive offshoring of
represents only about 16 per cent of the total. IT, BPO and software services to India (see box 1.3
Among the countries in Asia, about 8 per cent of for more details). The other occupational category
the global demand for such work comes from cli- where the share of labour supply from India
ents based in India, while the share of other Asian increased was that of creative and multimedia
countries is very small (1–2 per cent). The presence services (3 percentage points).
of clients from Africa and the Middle East on these
platforms is even smaller. Online web-based platforms often do not provide
information on gender, and it is therefore difficult
The disaggregation of demand for work by oc- to disaggregate the distribution of workers by
cupation and by country shows that software sex. To resolve this issue, researchers have used
development and technology are the most an algorithm that allows them to infer the sex of
sought-after occupations on these platforms the worker from first names, country of origin and
across countries (see figure 1.8a). The share of date of birth (as certain names were quite popular
demand in this field has increased worldwide at a certain point in time), using historical data
between 2018 and 2020, with higher demand from (Blevins and Mullen 2015). Based on this algorithm,
clients in India compared to other countries. The a small random sample of workers from the Online
share of demand for creative and multimedia, Labour Observatory were disaggregated by sex
clerical and data entry, and writing and transla- across different occupations for India, Ukraine and
tion has declined in most countries, the largest the United States (see figure 1.9).
decline being observed in the United States. As
The distribution shows that the participation of
these recent trends relate to the period when the
women on online web-based platforms is lowest
global economy is experiencing the effects of the
in India (21 per cent), while it is higher in Ukraine
COVID-19 crisis, the decline in the demand for
(39 per cent) and the United States (41 per cent).
such tasks may be due to the uncertainty caused
The distributions by sex at the country level are
by the pandemic.
very similar to the findings from the online sur-
In contrast to the demand for work, the supply of veys conducted by the ILO (Berg et al. 2018; see
labour on these platforms originates mainly from section 4.1.2). Across occupations, in all three
a number of developing countries, in particular countries the proportion of women is quite high in
Bangladesh, India, Pakistan, the Philippines and writing and translation. A higher share of women
Ukraine, apart from the United Kingdom and the in the United States is engaged in clerical and
United States (see figure 1.8b). Workers from India data entry, creative and multimedia, and sales
are the largest suppliers of global labour; India’s and marketing compared to other countries. In
share of total supply rose by about 8 percentage India, the share of women across all occupations is
points between 2018 and 2020, while it declined lower than in other countries, even in occupations
in other developing countries, except Ukraine. such as writing and translation, which are female-
Meanwhile, the share of the supply of labour dominated in the other two countries.
Figure 1.9 Gender distribution of labour supply

on online web-based platforms, by occupation,
selected countries, October 2020 to January 2021
Clerical and data entry
All occupations
49
100
80 78
80
51
Percentage of workers
61 59 20 22
60 79
India Ukraine United
States
40
Creative and multimedia
20 39 41
51 53
21
79
0
India Ukraine United
States
49 47
21
Female Male India Ukraine United
States
Professional services Sales and marketing
64 63
68 69
81 84
32 36 37
31
19 16
India Ukraine United India Ukraine United

States States
Software and technology Writing and translation
30 39
53
84 83 81
70 61
47
16 17 19
India Ukraine United India Ukraine United

States States

Box 1.3 COVID-19 impact on online web-based platforms

X
After the widespread outbreak of COVID-19, there was a decline in both the demand for work
and the supply of labour in March 2020, after which activity picked up gradually from early April
2020 (see figure 1.6). On the demand side, there was a rise between April and May, after which
demand declined gradually and then stagnated until October, when it picked up again. The
impact of COVID-19 seems to affect clients and workers differently across countries. To under-
stand these impacts, two countries are analysed: the United States and India. These countries
have the largest presence in both posting of tasks and projects (labour demand) and registered
workers (labour supply) on platforms.
In the United States, the demand for labour declined soon after the outbreak of the
pandemic in mid-March (see figure 1.10a), and it picked up in April and continued to
rise until May. This decline was observed across all occupational categories until
late October. The declining trend could be due to firms or clients being cautious
and reducing their expenditures, including by outsourcing non-essential tasks
as a result of a fall in their revenues, and postponing expenditures for the future
(Stephany et al. 2020). In October 2020 there was an increase in demand across
all occupational categories, with the largest increase in tasks related to clerical and
data entry and professional services, and the levels were higher than those observed
in February 2020.
The labour supply has increased substantially compared to labour demand (see figure 1.10a).
There was a steep increase in the number of registered users on these platforms originating
from the United States in April and May 2020, particularly in software development and tech-
nology, and in creative and multimedia services, followed by a small decline during the next
few months. The increase observed in these two categories may have been prompted by the
expectation of higher demand for such tasks.
Figure 1.10 Online labour demand and supply,
the United States and India, 2018 and 2020
(a) United States
140
120
Index (1 Jan. = 100)
100
80
60
Jan. Feb. Mar. Apr. May June July Aug. Sep. Oct. Nov. Dec.
Demand 2018 Demand 2020 Supply 2018 Supply 2020
Figure 1.10 (cont’d)
(b) India
140
120
1 Jan. = 100)
100
Figure 1.10 Online labour demand and supply,

the United States and India, 2018 and 2020
Box 1.3 (cont’d)
X
(a) United States
In India, by contrast, both labour demand and labour supply increased from
140
mid-March 2020 onwards (see figure 1.10b). The increase in demand was
largely driven by clerical and data entry, professional services, and software
development and technology, and demand was 50 per cent higher than at
120
the beginning of 2020. The increased demand for software development and
technology could be due to the need for software solutions that enable a smooth
functioning
100 of a remote working environment. The increase in demand for work
across other occupations could be attributed to the declining revenues of companies, and
it is possible that many firms or clients were considering these platforms as a substitute for
on-site 80
work (Stephany et al. 2020).
At the same time, there was also a steep increase in the number of registered workers across
all occupations,
60 except for professional services. The general increase in labour supply was
unaffected byJan.the seasonal
Feb. patterns,
Mar. Apr. in May
contrast to what
June July wasAug.
observed
Sep. in the
Oct. global
Nov.trends,
Dec.
indicating a steady demand of such work locally and regionally.
(b) India
140
120
100
80
60
Jan. Feb. Mar. Apr. May June July Aug. Sep. Oct. Nov. Dec.

1.4 The data-driven economy and the rise

of machine-learning algorithms
With the growth of digital platforms, data has While personal data can be sensitive, aggregated
become a valuable strategic economic resource data can be used for a number of purposes by
across various sectors of the economy. The im- multiple stakeholders. The use of structured
portance of data has been gaining momentum data can lead to significant changes in the value
since the beginning of the 2000s, and digitalization chain of almost every economic sector, from
has facilitated the collection, processing, storage, retail to healthcare, insurance or agriculture, as
use and transfer of data for different purposes the economy moves towards access-based ser-
(Rani and Singh 2019). The advances in cloud vices. For instance, data collected by e-commerce
infrastructure such as cloud storage and cloud platforms on consumer preferences – their
computing have enabled businesses to not only consumption patterns and tastes, and so on –
collect data at a speed and scale that was not provides rich insights that can serve businesses
possible at the end of the twentieth century, but in making economic decisions about product
also to store, structure and analyse data (Sheriff listing, designs, prices, inventory and logistics
2018). This section focuses on some of the con- (Rani and Singh 2019). Such insights can also be
cerns related to how data is being used, who owns sold to other companies for advertising purposes
data and how it impacts different users. or used for developing new products and ser-
vices, which in turn can help platforms generate
Data can be collected from a vast array of sources
considerable revenue. This not only gives them a
(websites, internet-based devices such as mobile
competitive edge over traditional businesses (such
phones, and so on), and digital platforms have
as small retail stores) but could also lead to unfair
emerged as spaces where data, such as driver
competition where companies such as Amazon
and customer data on Uber, or worker or client
or Google use the data they collect to promote
data on Upwork, can be gathered using trackers
their own products and services in search listings
and other digital tools. The data collected can be
(see section 3.4).
either structured, semi-structured or unstruc-
tured. Unstructured data is estimated to account
for more than 90 per cent of the data available to
organizations globally (Sheriff 2018). This unstruc-
tured data contains a bundle of information which,
when structured, can be aggregated to analyse With the growth
important trends and relationships.
of digital platforms, data
1.4.1 Potential use of data has become a valuable
Structured data, both aggregated and personal, is strategic economic resource.
valuable and can be used by multiple stakeholders
such as workers, businesses, communities and
governments for various purposes. Data col-
lected at the workplace can potentially be used On digital labour platforms, the vast amounts of
by companies to plan, to enhance operations, to data gathered from users are used for business
accelerate decision-making or to maximize per- purposes, including to improve work organization,
formance with a view to improving organizational to match users, for machine learning and training
goals (Sheriff 2018). Such data can also be used algorithms, and to improve automated deci-
to monitor worker performance, which may affect sion-making processes (Choudary 2018; see also
workers negatively (Ball 2010). Chapter 2). For instance, Upwork’s annual report
for 2019 states: “[d]uring a large concentration of market power

the search process, we (UNC TAD 2019). The network
leverage our pro - ef fec ts, along with data
prietary data to lock-in and aggregation
help freelancers (more data leads to ex-
and clients ponential increase in its
efficiently con- value), allow companies
ne c t ” ( Upwor k or plat forms to become
2019, 6). Similarl y, data monopolies (“data-opolies”)
on taxi platforms workers (Stucke 2018), thereby raising concerns
generate large amounts of data with regard to privacy, transfer of wealth
which are partly captured through the from consumers and workers to companies or
navigation technologies (GPS) that these platforms, and disruption to markets.
platforms invariably deploy. The data is then used
The accumulation of data among a few players can
by the platforms’ matching and pricing algorithms
lead to excessive market power and competition
for various purposes, including to propose to the
issues. For instance, Uber12 (including Uber Eats)
driver the best route to reach a given destination
have acquired a number of their competitors, such
(Choudary 2018).
as Careem, Cornershop and Postmates, and one
Although most of the data generated on digital of the assets shared or acquired as part of these
labour platforms is used by the platforms them- transactions, is data. This often allows platforms
selves for internal business purposes, if such data to amass an extensive amount of data, which is
is shared locally and globally, and used judiciously, also observed in other sectors such as delivery,
it can benefit society as a whole. Aggregated data e-commerce and social media and gives them a
in the fields of health, agriculture or environment, competitive advantage over other players in the
among others, could also be useful for policy- market (Cusumano, Gawer and Yoffie 2019).
makers to progress towards achieving the SDGs
(UN 2019). Similarly, real-time traffic information Although data is generated by users (workers,
collected through app companies such as DiDi, Ola businesses or consumers) on digital labour
or Uber could be used to relieve traffic conges- platforms, in practice it is considered to be the
tion and redirect traffic, especially in developing property of platforms. As a result, in recent years,
countries where there are challenges in relation various initiatives have been put forward to ad-
to infrastructure (Rani and Singh 2019). dress this misappropriation of data and ensure
more equitable forms of user rights over data. For
example, data protection frameworks such as the
1.4.2 Issues related to EU’s General Data Protection Regulation (GDPR),
user rights over data provide data subjects (including workers on digital
labour platforms) with a range of rights over their
Although there has been considerable emphasis data that allow them to exercise considerable
on data as a new form of capital that can be lever- control over it (rights of access, rectification,
aged and monetized to create revenue (Sadowski portability and more) (see section 5.3.8). Such
2016), issues around its value and user rights rights could empower workers (including those
have only recently gained attention. While it can on platforms) and ensure greater transparency,
be used to serve individual, economic and societal so as to enable them to effectively engage in
interests, the data collected tends to be owned collective bargaining with platforms to improve
by a few companies or digital platforms that have their working conditions (Rani and Singh 2019).
12 Uber’s privacy policy states that Uber may share “personal data with others in connection with, or during negotiations of, any
merger, sale of company assets, consolidation or restructuring, financing, or acquisition of all or a portion of our business by or
into another company”.
Moreover, the accumulation of data by platforms As data is generated by different users and is
has led some to consider whether data could be useful for economic decision-making and societal
treated as “labour” instead of “capital”. This notion development, it could be a primarily common
would allow data to be perceived and treated as or public asset, that is, there could be collective
the property of those who generate it and not as user rights over community data (P.J. Singh 2020;
an end product of consumption that is collected Rani and Singh 2019). A framework regulating
by the company or platform. Workers could then collective user rights over data could require plat-
collectively organize as a “data labour union” and forms and companies to share community data
bargain for fees for their data (Arrieta-Ibarra and be subject to a licence for using it (P.J. Singh
et al. 2018). 2020; see box 1.4). This type of user right would
allow countries to exercise legal and regulatory
While there is much to be gained by treating data
power over platforms and companies to ensure
as labour, practical questions arise about how
fairness vis-à-vis all economic actors, including
to assess its value and what criteria ought to de-
platform workers. It could also help traditional
termine data fees. A related question is whether
companies to compete on a more level playing
such fees will constitute one-off payments or be
field and strengthen national digital industries.
charged on a recurring basis. Moreover, mon-
This could potentially lead to the development
etizing data might even be counterproductive,
of appropriate public data infrastructures, espe-
because in the digital economy “the marginal value
cially in developing countries, which would in turn
of any one person’s data contribution is very low”,
contribute to empowering platform workers and
since aggregated or grouped data has more value
improving their lives, and to meeting the SDGs
than individual personal data (P.J. Singh 2020, 8).
(Rani and Singh 2019).
Box 1.4 Collective user rights over community data

X
What does the concept of collective user rights over community data mean? It represents the
idea that communities should have economic rights over the data they generate. In the case
of workers, such rights could take the form of a collective stake in the company, for instance,
in the form of co-determination rights in the business. In the context of “data as labour” and
the monetization of data more generally, economic rights to data can be complicated when
employment relationships are taken into consideration, as any remuneration of data could be
considered to be already included in the remuneration, as part of the overall work. Therefore,
workers’ data needs to be distinguished from their labour. In addition, data should be viewed
as having a permanent value, as it can be used in multiple contexts. In light of this, collective
economic rights over community data cannot and should not translate into a monetary sum;
rather, the data should amount to a collective stake in the resulting products or services of a
company, or, at the very least, the resulting products or services should not be used in a way
that is harmful to platform workers.
In India, the Committee of Experts on Non-Personal Data Governance Framework has adopted
a similar approach whereby “the rights over community Non-Personal Data collected in India
should vest with the trustee of that community, with the community being the beneficial owner,
and such data should be utilized in the best interest of that community” (2020, 23). The rationale
behind this approach has been to maximize welfare, as India has a large consumer market, and
the entry of data monopolies might lead to imbalances in the bargaining power of the various
stakeholders, with just a few companies having access to large data sets that are accumulated
in a predominantly unregulated environment, and with consequences for citizens, workers,
businesses including start-ups, SMEs and the Government.
Source: P.J. Singh (2020); India, Ministry of Electronics & Information Technology (2020).
Digital labour platforms

The excessive power and control over data by a
continuously use the vast
few companies needs to be counterbalanced by amount of data collected
policies preventing anticompetitive behaviour and
misuse of data; in other words, effective compe- for improving their machine-
tition and antitrust policies must be developed to
prevent such firms from abusing their dominance learning algorithms.
by leveraging the data they accumulate. Given the
asymmetries of economic power within the digital
economy, for developing countries to benefit from
reputation (De Stefano 2019; see section 2.4). If
the digital revolution they must build their digital
platform workers do not perform well or if the
infrastructure (broadband, cloud computing and
quality of their work falls short of the standards
data infrastructure) and digital policies in order
set by the algorithm, this can result in their not re-
to “ensure equitable distribution of gains arising
ceiving any tasks or at times even being dismissed
from data which are generated within national
(deactivation of their account) from the respective
boundaries” (UNCTAD 2018, VII).
platform (see sections 2.5 and 4.3.2).
Furthermore, the use of algorithms may in some
1.4.3 The rise of machine- cases exacerbate or amplify pre-existing biases
learning algorithms or create new ones. The algorithms are coded by
human programmers based on a set of norms and
The availability of data on a massive and unpre- instructions; if bias is fed into the system, it can
cedented scale, coupled with enhanced computing result in discriminatory practices. In addition, it
capacities, has led to major breakthroughs in is important to note that the algorithms are only
AI technologies. These are already being used ex- as good as the data that is fed into them; if there
tensively in a number of fields, such as search and are gaps or errors in the data then the algorithms
product recommendation engines, speech rec- might automate existing patterns of discrimin-
ognition, fraud detection, image understanding, ation (UN 2019). The use of AI-enabled algorithms
robotics and natural language processing. AI also can hence disrupt many existing regulatory ap-
facilitates new human resource practices, such proaches, leading to potential gaps in terms of
as management by algorithms, which are not liability, consumer protection and the protection
restricted to digital labour platforms but are also of fundamental rights (see Chapters 5 and 6).
increasingly used in traditional sectors such as
Alongside the increasing use of machine-learning
retail warehouses or white-collar occupations
algorithms in both the digital and non-digital
to assess worker productivity and their capacity
sphere, there is as yet no transparency with regard
to perform certain tasks (Akhtar, Moore and
to the source code13 of algorithms, which is not at
Upchurch 2018).
all accessible to the platform workers. If a worker’s
Digital labour platforms continuously use the task is rejected or their account deactivated, or if
vast amount of data collected for improving their they receive a low rating by the algorithm, they
machine-learning algorithms in order to match are often unable to find out the reason(s) for said
workers with clients or customers, allocate tasks, actions or sanction nor how they can improve their
set prices, monitor and evaluate tasks, and award performance. Accessing the underlying source
payments and rankings. The algorithms are code of an algorithm is the only way to inspect
designed to measure workers’ speed and atten- whether such an algorithm is producing anticom-
tiveness in completing the assigned task, apart petitive or discriminatory outcomes. However, it is
from taking into consideration their ratings and difficult to access the source code of an algorithm,
13 Source code refers to “a collection of computer instructions which are processed and executed, and whose human-readable
version (called source code) is usually protected by copyright and often kept confidential to protect proprietary information”
(UNCTAD 2018, 91).
as it is protected by trade secrecy laws and by in- These restrictions could further deepen global
tellectual property rules at the WTO level (Smith North–South inequalities by aggravating the
2017).14 There have been instances, however, dependence of developing countries on software
where access to the source code has been granted; monopolies which are usually concentrated in de-
for example, the US District Court for the Northern veloped countries, and by depriving them of the
District of California granted access to Uber’s opportunity to adapt software to their own reality
source code to Waymo’s15 counsel and an expert and use it for local development (Neeraj 2017).
to ascertain whether there had been a case of
The rise of data as capital and an asset, and its
trade secret misappropriation.16 Uber committed
relevance to AI, has also led venture capitalists
to not using any of Waymo’s intellectual property
and private investors to invest in digital plat-
(whether hardware or software) in its self-driving
forms (see section 1.5) and digital technology
technology and paid Waymo 0.34 per cent of its
start-ups (see section 3.3.2). For instance, the
equity as part of the settlement.17
recommendation engine of Netflix reportedly
To ensure fairness for workers and businesses on saves US$1 billion every year for the company as
digital platforms, both labour and e-commerce, it reduces the subscriber monthly churn and is
it is crucial for governments to have access to the able to recommend based on previous choices
source codes of the algorithms in appropriate cir- (Gomez-Uribe and Hunt 2015). The potential of
cumstances and under appropriate conditions. For machine-learning algorithms to raise such reve-
instance, without accessing Google’s, Amazon’s nues has also led venture capitalists to invest in
or Uber’s source code, it is impossible to inspect AI start-ups, which raised a record US$26.6 billion
whether a company’s ranking or pricing algorithm in 2019 (compared to US$16.8 billion in 2017)
produces anticompetitive outcomes, or whether (K. Johnson 2020). Given the rise of business
its rating algorithms lead to account deactivation models supported by data and AI, and the po-
that amounts to unfair dismissal. In this regard, the tential for enhanced profitability, venture capital
proposals on e-commerce rules agreed by WTO investments are further supporting the growth of
member countries at the WTO level that prohibit digital platforms, which are seen as fundamental
the transfer of or access to source code could pose to taking forward such a profound data-based
a major threat to ensuring decent work and fair transformation in the economy.
competition on digital platforms (Smith 2017).18
14 See Article 39 of the WTO Agreement on Trade-Related Aspects of Intellectual Property Rights: https://www.wto.org/english/
docs_e/legal_e/27-trips.pdf.
15 Waymo is an autonomous driving technology development company, subsidiary of Alphabet (includes Google).
16 For more information, see: https://cdn.arstechnica.net/wp-content/uploads/2017/05/Uber.Waymo_.Order_.pdf.
17 For more information, see: https://www.wired.com/story/uber-waymo-lawsuit-settlement/.
18 See, for example, the following WTO documents: JOB/GC/94; JOB/GC/100; INF/ECOM/22.
1.5 Financing the rise of digital labour platforms
Venture capital has played a key role in the rise of digital labour platforms has been limited, and
of digital platforms (including digital labour plat- fraught with data limitations. This report takes
forms) over the past decade. The stock market recourse to databases such as Crunchbase and
value of the major technology companies or Owler to extract available information on these
“superstar firms”, and of digital labour platforms, aspects, while also drawing on the annual reports
has also continued to rise. These companies or filings by platform companies to the Securities
attract investment even though some of them and Exchange Commission of the United States
continue to have operating losses (Kenney and where information is available. For funding, the
Zysman 2019). This section looks at the rise report uses data only from Crunchbase, and this
of venture capital investment in digital labour data is available for only 47 per cent (367 plat-
platforms, and their concentration in particular forms) of the 777 digital labour platforms listed
sectors and geographical regions. It also considers on the Crunchbase database. These platforms
the concentration of market power in the hands have together received a total funding of
of a few platform companies and digital labour US$119 billion (as of 30 January 2021). There are
platforms, and their implications for businesses substantial differences in investment between
and platform workers. platforms offering taxi or delivery services and
The rise of venture capital investment in digital those providing online web-based services. The
platforms is rooted in the belief that start-ups highest investments are in taxi service platforms,
offer large capital gains, given that many sectors with 61 platforms having received US$62 billion
and industries can be disrupted with the advances between 2007 and 2020. This is followed by de-
in ICT, ranging from smartphones and big data livery platforms where US$37 billion has been
to machine learning and the Internet of Things invested in 164 platforms, while investments
(Kenney and Zysman 2019). Globally, venture in online web-based platforms are the lowest,
capital investments in digital start-ups have at about US$3 billion for 142 platforms (see
grown sixfold between 2010 (US$52 billion) and figure 1.11). Five hybrid platforms were identified
2019 (US$295 billion) (Rowley 2020; Florida and which provide a range of services from payment
Hathaway 2018). A significant proportion of these to taxi or delivery services and e-commerce; these
investments were made in companies based in platforms have received US$17 billion between
the United States (US$136.5 billion), followed by 2010 and 2020.
companies in China (US$36.5 billion for January to Based on the funding information available on
mid-November 2019, which was a major drop from platform companies, the distribution of funding
US$93.4 billion in 2018), Europe (US$36 billion) and
is considerably skewed among taxi platforms, with
India (US$14.5 billion) (PitchBook 2020; Teare and
75 per cent concentrated in just two companies
Kunthara 2020; Kunthara 2019; M. Singh 2019).
(Uber and DiDi), while the remaining 25 per cent
In comparison, investments in Latin America
went to 59 companies. The distribution of funding
(US$4.6 billion) and Africa (US$1.3 billion) were
is slightly less skewed for delivery platforms, with
relatively low (Azevedo 2020; WeeTracker 2020).
the top five platforms (DoorDash, Delivery Hero,
Data on funding or investment and revenue of Ele.me, Lalamove and Instacart) accounting for
digital platforms is not easily available, particularly 49 per cent. In the case of online web-based
for platform companies that are yet to release an platforms, about 33 per cent of funding is con-
Initial Public Offering (IPO). In this regard, re- centrated among the top three platforms (ZBJ,
search on the flow of investment in and revenue Scale AI and Upwork).
other major platform companies such as Alphabet

(includes Google) and DiDi, and other investors
Figure 1.11 Total investments from venture like SoftBank (one of Uber’s largest shareholders)
capital and other investors, by platform
(Uber 2020a, 12). The rapid growth in revenues
category, 1998–2020
and the valuation of Uber is explained by these
70
venture capital investments, which have served to
heavily subsidize consumers and drivers through
60
62 various incentives, and what some have argued to
US$ billion
50
allow for “artificial market power to subvert normal
40
market dynamics” (Horan 2019). This situation has
30 37
led to a disruption of the traditional taxi industry in
20
that it has allowed platforms, irrespective of their
10 17
revenues, both to establish their market power
3
0 and to gain a dominant market position.
Online Taxi Delivery Hybrid
web-based
Note: Number of platforms and period
for which data on total funding was available:
online web-based: 142 (1998–2020);
The access to venture

taxi: 61 (2007–20); delivery: 164 (1999–2020);
and hybrid: 5 (2010–20).
capital funding has allowed

Source: Crunchbase database. many platforms to operate at
a loss for long periods of time.
The concentration of funding on just few com- Given the availability of venture capital funding,
panies by many venture capitalists, who are many platform companies tend to remain private
betting on these platforms to dominate the for long periods of time, as opposed to making
market, is based on the high rate of return from an IPO; this situation has led to the growth of so-
their network effects or “winner-take-all” effects called unicorns, which are privately held start-up
(Kenney and Zysman 2018b, 6). The access to companies valued at over US$1 billion (Kenney and
venture capital funding has also allowed many Zysman 2018b). These companies can continue to
platforms to operate at a loss for particularly function for long periods even when incurring
long periods of time, which has exacerbated their losses by raising private funds and avoiding the
disruptive effects on the traditional sectors. For scrutiny of public markets or traditional investors
instance, both Uber (US$25.2 billion in 28 rounds) (Kenney and Zysman 2019; Schleifer 2019). The
and Grab (US$10.1 billion in 31 rounds and an trend of large valuations despite unprofitability
additional US$2 billion in 2021) have continued is not unique to companies that are not publicly
to receive funding despite incurring substantial traded; it is estimated that 64 per cent of platform
losses. Grab, which is valued at US$14 billion, con- companies valued at more than US$1 billion that
tinues to remain private (as of 2019), while hoping have completed a venture capital-backed IPO since
to make profits so that it can go public (Soon and 2010 were unprofitable (Clark 2019). While several
Choudhury 2019). By contrast, Uber, which has platforms are profitable, the fact that many con-
incurred “significant losses since its inception” tinue to operate and receive funding from venture
and has an accumulated deficit of US$16.4 billion, capitalists despite losses over long periods raises
was able to go public in 2019 when it was valued questions about both their economic and social
at US$82.4 billion (Uber 2020a, 12; de la Merced impact as well as the welfare-generating aspects
and Conger 2019). Despite its continuous losses, of this innovative business model (Kenney and
the company is able to attract investment from Zysman 2019).
Figure 1.12 Total funding from venture capital and other investors, selected categories
of digital labour platforms, by region, 1998–2020 (US$ million)
Northern, Southern
North America East Asia
and Western Europe
Via CloudFactory Malt Eastern Europe
587 78 36
Delivery Bolt Yandex.Taxi*
Hero Wolt 535 1 000
Lyft 4 400 814 Cabify DiDi
Dada-JD
Uber Deliveroo Foodpanda* 477
7 100 Central and Western Asia 21 200 Daojia
25 200 1 700 750 950
Glovo Yemeksepeti*
Just Eat Fiverr Gett
635 44
Upwork Takeaway* 111 928
Scale AI 1 052 Lalamove
169 Arab States South Asia
278 2 500
Careem* Zhubajie Ele.me
Instacart Africa 772 446 3 330
Ola
2 400 Stack
DoorDash Halan Jahez 3 800 Meituan
5 800 goPuff Overflow 26 37 Woowa Bros* 1 070
153 Jumia
1 300 Zomato 483
Group
908 1 900
Postmates Swiggy South-East Asia and the Pacific
901 Latin America and the Caribbean 1 600
PedidosYa* Rappi Gojek
329 1 700 5 000
iFood Grab
592 10 100
Easy Taxi*
77 99*
241
Online web-based Taxi Delivery Hybrid
* Platform has been acquired or merged, see Appendix 1.

Note: Number of platforms and period for which data on total funding was available: online web-based: 142 (1998–2020);
taxi: 61 (2007–20); delivery: 164 (1999–2020); and hybrid: 5 (2010–20).
1.5.1 Geography of digital sectors, as well as online web-based platforms

(Upwork and ZBJ), are located in the United States,
labour platforms: China or Europe. In terms of funding, taxi plat-
funding and revenue forms have received a significantly larger share
of venture capital financing than online web-based
The global distribution of investment in digital platforms (see figure 1.12). Uber’s total funding
labour platforms is quite skewed, with the triad (US$25.2 billion) is nine times greater than that
composed of Asia (US$57 billion), North America of all the online web-based platforms analysed
(US$46 billion) and Europe (US$12 billion) bene- put together (US$2.6 billion for 142 online web-
fiting from 96 per cent of investments compared based platforms).
to 4 per cent going to Latin America, Africa and the
Arab States, indicating a digital divide. Although
some important players are emerging in these
regions, such as Gett and Fiverr (Israel), Jumia
The global distribution
Group (Nigeria) and Rappi (Colombia), the most of investment in digital labour
well-funded platforms in the taxi (Uber and DiDi)
and delivery (DoorDash, Delivery Hero and Ele.me) platforms is quite skewed.
Figure 1.13 Estimated annual revenue of digital labour platforms, selected categories,
by region, 2019 (US$ million)
Northern, Southern
North America and Western Europe
Eastern Europe East Asia
Just Eat Ele.me
Takeaway* Yandex. Dáme Jídlo*
157
Lyft 1 700 Wolt Cabify Taxi* 32
Deliveroo 104 501 Woowa Bros*
Uber 3 616 345 Codeforces
624 Delivery 270
10 745 20 Meituan Lalamove
Glovo Hero Bolt Central and Western Asia 8 532
Upwork 300 1 502 115 DiDi 100
301 Fiverr Gett Yemeksepeti* 401
Roamler CloudFactory 107 Quhuo
Applause* 1 000 231
22 20 Tech
Arab States South Asia
Instacart Toptal 136 2 055
2 900 300
Careem* Zomato* Swiggy CrowdWorks Dada-JD
119 Talabat* 394 183 84
Uber Eats Daojia
DoorDash Africa 67 PickMe 438
2 510
Grubhub 885 Ola 88
1 312 Postmates GoDropping Playment 365
640 4 13 South-East Asia and the Pacific
Jumia Gojek
Latin America and the Caribbean Group 1 300
Cornershop 180 Grab
200 2 300 GoCatch
PedidosYa* iFood 7
250
Rappi 208 Menulog*
99* 220 Appen
51
7 Easy Taxi* 409
10
Online web-based Taxi Delivery Hybrid
* Platform has been acquired or merged, see Appendix 1.
Note: Number of platforms for which data on revenue was available: online web-based: 106; taxi: 31; delivery: 101; and hybrid: 5.
Source: Owler database, annual reports and filings by platform companies to the Securities and Exchange Commission of the United States.
With regard to revenue, this report relies on data together account for 17 per cent of the revenue.
collected from the Owler database, annual reports Uber, located in the United States, has the highest
and filings by platform companies to the Securities revenue (US$10.7 billion) among taxi platforms,
and Exchange Commission of the United States. while Meituan, located in China, has the highest
The data on revenue is available for only about revenue (US$8.5 billion) among delivery platforms
31 per cent (243) of the platform companies. The (see figure 1.13).
revenue generated through digital platforms is
Among online web-based platforms, Appen,
further evidence of the geographical concen-
Upwork, Toptal and Fiverr, which are based in
tration of wealth, as about 70 per cent of global
Australia, Israel and the United States, respect-
revenues are concentrated in just two countries,
ively, generate the highest revenues. The revenue
the United States (49 per cent) and China (23 per
generated by online web-based platforms is
cent).19 About 11 per cent of global revenue is
smaller than that of location-based platforms.
concentrated in Europe, while all the other regions
19 It is possible that if the information on revenues was available for a larger number of platforms, then the concentration of
revenue might be less skewed.
For example, in 2019, Uber generated a revenue companies that are dominating the global digital
of US$10.7 billion, which is about 36 times that economy. The estimated market value of the
generated by Upwork (US$301 million). Uber digital economy was US$7 trillion in 2017, based
received funding of US$25.2 billion, compared on the top 242 companies. However, seven “super
to US$169 million received by Upwork, which is platforms” based in China and the United States
about 150 times more. Furthermore, the valuation represented 69 per cent of the total market value
of Uber at the IPO was US$82.4 billion, while that of the digital economy (KPMG 2018, 9). The seven
of Upwork was US$1.5 billion (de la Merced and largest technology companies (Amazon, Apple,
Conger 2019; Belvedere 2018). The key element in Alphabet (includes Google), Microsoft, Alibaba,
this difference could be that the taxi sector allows Facebook and Tencent) based in the United States
these companies to gather vast amounts of data or China had a cumulative revenue of US$1,010 bil-
on users (workers, clients and customers), which lion in 2019 (see figure 1.14).
has intrinsic commercial value as it is linked to spe-
In comparison to these major technology com-
cific localities and infrastructure, and it also allows
panies, the largest digital labour platforms (both
these companies to expand their services. This,
location-based and online web-based) are small
in addition to using such data to train algorithms
in terms of revenue generation (see figure 1.14).
for pricing, allocating tasks, or for predicting
Amazon and Apple generated over US$280 billion
and mitigating traffic congestion (Chen and
and US$260 billion in revenue in 2019 respect-
Qiu 2019), could be potential reasons for such a
ively, while some of the largest location-based
high valuation.
and online-web based platforms such as Uber,
While digital labour platforms are disrupting both Meituan, Instacart, Appen, and Upwork generated
traditional business models and employment rela- a combined revenue of only about US$31.2 billion
tionships, they are small compared to the platform in 2019. Moreover, some of the major technology
Figure 1.14 Estimated annual revenue of large platforms and selected digital labour platforms, 2019 (US$ million)
Appen Toptal
409 300
Facebook Upwork
Microsoft 70 967 301
125 843
Alibaba Biggest companies

Uber
56 152
10 745 Lyft
3 616 Online web-based
Gett
1000 Taxi
Amazon Apple Tencent
280 522 260 174 54 594 Delivery
Meituan
8 532 Hybrid
Instacart
2 900
Uber Eats
2 510
Alphabet
(incl. Google) Grab
161 857 Quhuo Tech 2 300
2 055
Gojek
1 300
Note: For each of the digital labour platform categories only the seven companies with the highest revenue are included.
For the taxi sector, these are Uber, Lyft, Gett, Careem, Yandex.Taxi, DiDi and Ola; for the delivery sector Meituan, Instacart, Uber Eats,
Just Eat Takeway, Delivery Hero, GrubHub and DoorDash; for online web-based platforms Appen, Upwork, Toptal, Fiverr, Applause,
Guru and Justanswer; and the hybrid platforms Grab, Quhuo Tech, Gojek, Dada-JD Daojia and Jumia Group.
Sources: Owler database, annual reports and filings by platform companies to the Securities and Exchange Commission of the United States.
The rise of large

technology companies has
companies are also investing in digital labour plat- resulted in a concentration
forms. Google Ventures (now Alphabet) invested in
Uber in 2013 and owned a 5.2 per cent stake in the of market power.
company in 2019 (Levy 2019); Facebook, Alphabet
(includes Google) and Tencent have invested in
Gojek (Gupta 2020); and Apple, Alibaba, Booking,
Similar trends can be observed among other
Softbank and Tencent have invested in DiDi (Chen
companies, such as Gojek and Grab in South-East
and Qiu 2019).
Asia or Jumia in Africa, which are expanding into
The rise of such large technology companies more countries and diversifying into a wide range
has also resulted in a concentration of market of services: online retail, travel marketplace, trans-
power, as these companies are diversifying and portation and logistics, food and grocery delivery,
offering an increasing range of services, often home and maintenance, entertainment, and
through acquisitions or mergers with other plat- payment, among others. While investments have
forms. Amazon is a case in point as it offers a wide been rising and new investors are increasingly
range of services including online retail, delivery, playing an enabling role in financing the digital
cloud computing, a crowdsourcing marketplace, platforms, the current model of venture capital in-
and entertainment. The greater market concen- vestment that focuses on a few companies despite
tration could help companies achieve monopoly their large losses raises concerns with regard to
power, and could lead to potential issues related to the sustainability of this model, and particularly
pricing, as well as having an impact on influencing to the over-valuation of companies.
regulation and even innovation.
The competitive advantage and market power
Such concentration of market power among a exercised by these companies is not necessarily
few companies is increasingly the case for digital based on inherent competitive advantage, as they
labour platforms, where easy access to venture are often loss-making and propped up by venture
capital financing enables these companies to capital funds rather than profits in the short to
reach new markets and enhance their competi- medium term. This distorts competition, chal-
tiveness. For instance, DiDi in China merged lenges the traditional understanding of monopoly
with Kuaidi in 2015 and acquired Uber China in or oligopoly power, and blurs the boundaries
2016 (Chen and Qiu 2019), triggering an antitrust of the organization, not just in the sense of
investigation by the Chinese government in 2018.20 employment relationship but also in terms of
Globally, it acquired the company 99 in Brazil, and finance, which is obviously fundamental for the
is developing strategic partnerships with platform survival of a company. The dominance of such
companies in a number of countries in Asia, Africa, companies might also lead to sustainability issues
Latin America, the Middle East and Europe (Chen for traditional companies, smaller businesses and
and Qiu 2019). third-party sellers (see section 3.4).
20 For more information on antitrust investigation, see: http://www.xinhuanet.com/english/2018-11/16/c_137611764.htm.

Conclusion
This chapter has shown that with the ICT revolu- by platform companies. This data is being used
tion and the pervasive spread of the internet, the for machine learning, developing new products,
world of work is experiencing many fundamental enhancing efficiency and productivity, and
transformations. There is clear evidence of the shaping pricing structures and the organization
diffusion and penetration of digital platforms into of work; while the users, and in particular platform
various sectors of the economy. The availability of workers, often have no economic rights over
cloud computing and technological innovations such data.
has enabled the development of a distinct form of
At the same time, the rise of the digital economy
business model which has created opportunities
more broadly, and of digital labour platforms
and challenges for both workers and businesses.
more specifically, has been financed through
Digital labour platforms in particular, as mediators venture capital funding, which has allowed
of work, have grown at a rapid pace, changing the companies to expand rapidly and compete with
way in which work is organized. They are impacting traditional sectors despite often being unprofit-
several economic sectors, so that businesses need able. This financing model has also enabled large
to adapt both to the changes being introduced by platform companies to acquire global market
digital technologies and to new forms of compe- dominance even though they are based in only a
tition arising from these platforms. handful of countries. This could further widen the
digital divide and increase economic inequality,
The growth of digital labour platforms has indeed and also pose challenges for companies based
created additional income-generating opportun- in developing countries to compete in the global
ities for workers around the world but has also digital economy.
given rise to a number of challenges which need
Furthermore, the challenges related to the rise of
to be addressed. Although estimates regarding
the digital economy could complicate efforts by
the number of workers whose work is mediated
governments of developing countries to adopt
through such platforms continue to face data-
appropriate regulations to ensure fair compe-
related and methodological challenges, in
tition for businesses and adequate protections
instances where data is available there are clear
for workers. Engaging with and addressing such
signs that labour supply is exceeding demand.
challenges will be decisive in leveraging the po-
The rise of the digital economy and the prolifer- tential opportunities emerging from the digital
ation of digital labour platforms are occurring economy and labour platforms to promote decent
alongside the increased relevance of user data, work and advance progress towards achieving the
which is largely owned, controlled and managed Sustainable Development Goals.
2
The business
model and
strategies
of digital
labour
platforms
The business model Despite the diversity
of digital labour in types of platforms...
platforms
Freelance
Contest-based
Online Microtask
web-based
platforms
Competitive programming
Taxi
Delivery
Location-based
platforms
... their business model has common elements
Commission fees and subscription

plans are integral to the platform
revenue model
Indicators for matching Recruitment

clients and workers Revenue
and model
matching
Client reviews
60%
Ratings
Worker profile
50%
62%
fees from workers
38%
fees from clients
46%
Project history or portfolio Business
27% model Upwork’s revenue in 2019
was about US$300 million
Rate proposed by the worker
21%
Monitoring, tracking and

evaluating workers through
Work digital tools and algorithms
Unilaterally determined Rules of processes and
by platforms: governance performance
• Exclusivity clauses management
• Deactivation of accounts
• Dispute resolution
2. The business model and strategies of digital labour platforms 73
Introduction
The current evolution of the digital economy is resource management practices and may have
transforming business and society, and is also implications for the future of work. For instance,
leading to the “platformization” of traditional on taxi platforms “algorithmic management
business practices. The availability of digital tools allows a few human managers in each city to
and cloud infrastructure has enabled the devel- oversee hundreds or thousands of drivers on a
opment of innovative business models, such as global scale” (Lee et al. 2015, 1603).
digital labour platforms, of which there are two
Second, the organization of work, which allows
broad types: online web-based and location-based
platform companies to provide services without
platforms. Online web-based platforms offer the
having to invest in capital equipment or bear the
flexibility of undertaking work from any location,
operational costs (Stanford 2017). For instance, on
at any time. While some of the tasks available
online web-based and location-based platforms,
on these platforms are new, such as image and
capital equipment such as computers or vehicles is
data annotation, labelling and data processing, a
provided by the workers, who also bear the costs
number of others, such as translation, transcrip-
related to fuel, maintenance, purchase of licences,
tion and software development, were previously
or internet charges.
performed and continue to be performed in the
traditional labour market. The distinguishing The third feature is the creation of a highly seg-
features of such platforms are that technology mented dual labour market, which consists of
enables work to be outsourced globally across two categories: a small core workforce directly
borders and that work can be performed remotely employed by the platform (internal employment)
from any location. and a large outsourced workforce whose work
is mediated through the platform (external
On location-based platforms, work is performed employment) (ILO, EU and OECD, forthcoming;
in a specified physical location, with taxi and de- Rahman and Thelen 2019). Workers in the first
livery services being among the most prevalent category have an employment relationship, while
examples of such platforms. Like the activities on those in the latter are typically categorized as
online web-based platforms mentioned above, “self-employed” or “independent contractors”
taxi and delivery services are not inherently new by the platform and are without an employment
and continue to be conducted in traditional labour relationship but often have to pay various types
markets. What is new in the digital economy is of fees for accessing tasks (Webster 2020). This
that these services are mediated through a digital model allows digital labour platform companies
application. Platform-based taxi and delivery to raise revenue and provide services by shifting
services have created employment opportunities the risks and costs related to capital equipment
owing to changing consumer preferences, and and operations to workers.
workers in these sectors are increasingly relying
This chapter explores some of the features of the
on app-based services for their incomes, particu-
digital labour platform business model, including
larly in developing countries.
algorithmic management of work, the revenue
Three distinct features can be identified in the model and business strategies. It also reviews the
digital labour platform business model. First, rules of governance, which are unilaterally set by
the introduction of algorithmic management of the platforms, and are hence market-driven to
work processes and performance (Moore and some degree. The analysis for this chapter draws
Joyce 2020; Griesbach et al. 2019; Lee et al. 2015). on the terms of service agreements of 31 online
Allocation and evaluation of work performance web-based and location-based platforms, their
are based on metrics and ratings integrated online websites, and semi-structured interviews
into an algorithmically determined performance conducted by the ILO with 16 digital labour
management system, while work is monitored platform companies (both online web-based
using digital tools. This mode of management is and location-based platforms) based in different
a fundamental departure from traditional human countries (see Appendix 2).
The chapter comprises five sections. The various labour platforms, and the algorithmic matching
types of platforms that are analysed in this of clients and platform workers. The management
chapter and the report as a whole are described of work processes and evaluation of work on
in section 2.1. Section 2.2 discusses the revenue platforms are explored in section 2.4. Section 2.5
model and the pricing strategies that platforms reviews the rules of governance on platforms and
use to appeal to workers and clients. Section 2.3 client–worker engagement, as well as the collec-
describes the recruitment practices on digital tion and use of data.
2.1 Types of digital labour platforms
As discussed in Chapter 1, digital labour plat- specific task, based on a proprietary database
forms provide a variety of services, drawing on that consists of indicators such as ratings
diverse skill sets of workers, and operate in two and reviews, and facilitate the client–worker
broad categories, online web-based platforms and relationship in all its dimensions. The nature
location-based platforms. They can be further dis- of services provided differs across these plat-
tinguished based on the type of tasks performed, forms, from a wide range of activities1 and skills
their duration and complexity. This report reviews (e.g. Freelancer, PeoplePerHour and Upwork) to
four types of online web-based platforms and two service offerings of specialized or targeted skills
types of location-based platforms (see figure 2.1). sets 2 (Toptal). This business strategy allows
workers with multiple skills to access various
tasks on the same platform, and businesses
2.1.1 Online web-based to access a wide range of skilled workers at a
platforms single place. There are other types of freelance
platforms as well, where the platform matches
Online web-based platforms are gaining in popu- the freelancer directly with the client or busi-
larity among businesses as they enable them not ness for specific services, rather than through
only to outsource tasks to a global workforce at a marketplace. For instance, some translation
reduced cost but also to complete assignments platforms maintain a “network” of freelance
at a faster pace than is possible in the traditional translators, who are assigned translation
outsourcing model (see section 3.1.2). Among tasks by the platform when a client puts in a
online web-based platforms, this report focuses request. Such platforms do not have an open
on freelance and contest-based, competitive pro- marketplace visible to all the users and are not
gramming and microtask platforms, which are analysed this report.
some of the leading platforms facilitating labour
Contest-based platforms specialize in organizing
X
exchange between workers and clients.
competitive design contests within their pool
Freelance platforms function like a marketplace,
X of talent to provide creative or artistic services
enabling clients to have work performed in and products, such as graphic design, to clients
fields such as translation, financial services, (e.g. 99designs,
legal services, patent services, design and data Designhill and
analytics. They match clients with workers for a Hatchwise).
1 Activities range from computer programming and analytics to design, translation, and
legal and accounting services.
2 Toptal advertises service offerings through its exclusive community of developers,
designers, finance experts, and project and product managers.
Digital labour platforms the crowd, then rebundling and delivering them
provide a variety of services, back to the clients. Some of these platforms
also provide clients with access to their applica-
drawing on diverse skill sets tion programming interface (API), which allows
clients to directly crowdsource the tasks on the
of workers. platform. In addition, there are other types of
microtask platforms that have emerged, such
as Scale AI or Mighty AI, that provide data and
The services3 offered are similar across plat- image annotation services; they crowdsource
forms, which compete through their pricing the tasks to their “crowd”, which is maintained
strategy and by attracting a pool of the “best” by the platform on a website that is only acces-
or outstanding designers through various sub- sible to the workers, and is different from the
scription and other plans. website which is meant for marketing purposes
and for the clients. Such platforms are not part
Competitiveprogrammingplatforms are spaces
X of the analysis in this chapter but are discussed
where a community of software developers and in section 3.3.2.
programmers can compete to provide business
and research solutions related to artificial intel-
ligence, data analytics, software development 2.1.2 Location-based
and other technical fields, within a designated
time, with the winner(s) chosen by the clients.
platforms
These platforms provide wide-ranging services The activity of location-based platforms centres
to companies, from software solutions and data on taxi and delivery services, which have been the
analytics (Kaggle and Topcoder) to recruitment subject of discussion and scrutiny in recent
services for hiring programmers, developers or years because of the way in which platform
data scientists (HackerEarth and HackerRank), companies are mediating the work op-
among others, through their community of tar- portunities of a growing number of
geted talent. Some of these platforms, such as workers, with potential implica-
CodeChef and Kaggle, also link up with academic tions for the future of work.
institutions and offer online practice sessions Digital labour platforms
and contests for students and young software in these t wo sec tors
professionals to hone their programming skills. have grown rapidly with
the help of venture capital
Microtask platforms specialize in tasks of short
X
funding (see section 1.5).
duration, such as transcribing a short video,
checking data entries, adding keywords to Taxiplatforms such as Bolt,
X
classify a product for artificial intelligence Careem, Grab, Gojek, Little, Ola
and machine learning purposes, or tasks and Uber facilitate ride-hailing services
related to accessing content (such as visiting by connecting customers seeking a ride with
websites to increase traffic) or checking for workers offering their services through the
sensitive content. Platforms such as Amazon platform. Customers are updated at every step,
Mechanical Turk (AMT), Appen, Clickworker provided with an approximate waiting time, an
and Microworkers provide a range of services4 estimated fare and ride duration, and have the
to clients and support them in unbundling tasks ability to track their driver and their ride in real
into smaller segments and dispersing them to time through their mobile application.
3 Including logo and identity design, web and app design, business and advertising, clothing, arts and illustration, packaging,
book and magazine design, among others.
4 Including data cleaning, categorization, tagging, sentiment analysis, creating and moderating content, video and audio tran-
scription, among others.
76
platforms
platforms
Location-based
Source: ILO elaboration.

Online web-based
Taxi
Delivery
Microtask
Freelance
Contest-based
Be
Figure 2.1 Types of digital labour platforms
at
Co
rne
Zo rsh AM EP
ma
to op T WK
Bo
lt
De De
The role of digital labour platforms in transforming the world of work
live Ap sig Fre

pe nh ela
Sw
roo n ill nc
igg Ca er
bif
y y Ha
tch
DiD Cli wis Fre
i Fo ckw e ela
nc
od Co ork eh
Ca de er un
Ub ree Ch
ef
99 t
er
Ea m de
sig Ka
ts Glo Mi
cro ns ba
vo nc
Ha wo hik
cke rke
DiD rs
To i rEa Pe
ter rth op
s Go leP
erH
foo Ha
d cke ou
r
Go rRa To
jek nk pta
l
Sin Gr
De ab Ka
Fo gg
lan od le Up
tal Gr wo
ab rk
Jum
Ra ia To
pp Fo pc
od ZB
i od
Lit er J
tle
Me
itu
an
Pe Ola
did
os
Ya
Ub
er
Deliveryplatforms such as Deliveroo, Glovo,

X agreements (see also Aleksynska 2021; Moore
Jumia Food, Rappi, Swiggy and Zomato facili- and Joyce 2020). These different elements play an
tate transactions between customers, workers, important role in shaping working conditions on
and business clients (such as restaurants, digital labour platforms.
supermarkets and pharmacies). They provide
customers with a range of products at a com- This chapter reviews the business strategies of
petitive price without the customers having 31 selected platforms that were covered by the
to leave their physical location, and business ILO worker surveys (see Chapter 4); it also includes
clients with a wider customer network (see sec- some other prominent platforms in order to better
tion 3.2). A different type of delivery platform understand the functioning of the digital labour
is also emerging, which has its own grocery platform business model (see figure 2.1). Some
warehouse or ghost kitchens (also called virtual of the digital labour platforms were established
or cloud kitchens), which can only be accessed at the turn of the century, while others have
by consumers through the app (Lee 2020). emerged in the past decade, and are emulating
This model draws on the principles of retail the existing platform business model.
e-commerce platforms, such as Amazon, where Platform business strategies are based on some
a bricks-and-mortar store is absent. It enables of the key elements described below, and some
delivery platforms to reduce costs and expand of the location-based platforms also adapt their
their businesses while also delivering food strategies to their national or legal contexts
and groceries, and has been growing during (Aleksynska 2021). The business strategies
the COVID-19 pandemic. Some of these ghost adopted by the platforms reviewed in this chapter
kitchens also link up with delivery platforms can be encapsulated in four interlinked key
and provide food delivery services. elements: revenue model (commission fees and
While a wide range of tasks are mediated subscription plans); recruitment and matching
through online web-based and location-based of workers with clients; work processes and
platforms, it is possible to identify some common performance management; and rules of plat-
elements or practices in the business model form governance (see figure 2.2). The analysis
across these different types of platforms. These of these four elements is based on the terms of
include price-setting and remuneration-setting service agreements of the respective platforms
mechanisms, charging of commission fees to and on information from their websites (see
workers and clients, matching of workers with Appendix 2B), as well as on interviews conducted
clients, allocation and evaluation of work with 16 online web-based and loca -
through algorithms, monitoring tion-based platform companies
of work using different digital (see Appendix 2A for the
tools, use of rating systems list of platforms). The
and engagement with the different elements are
workforce through the plat- discussed in turn in
forms’ terms of service the next four sections.
Figure 2.2 The platform business model: Business strategies
Work processes and

performance management Recruitment and
matching of workers
 Tools for communication
 Providing access to accounts
 Monitoring of work
Revenue model on platforms
 Performance management
 Matching algorithm
 Commission fees  Ratings, feedback and reviews  Work assignment
 Subscription/membership plans  Intermediate payments  Determination of price
 Monetary subsidies for the task
 Bonus schemes  Refund policies
 Priced features and other fees
Rules of platform
governance
 Exclusivity clauses
 Acceptance/rejection of work
Network effects Algorithmic management
 Deactivation of accounts
 Dispute resolution
 Data collection and usage
Source: ILO, based on the concepts outlined in Moazed and Johnson (2016).
2.2 Revenue model

A key element in the success of a platform is subsidized (Cusumano, Gawer and Yoffie 2019;
whether it can attract a sufficient number of users Horan 2019). Platforms become potentially at-
(clients or customers and workers) and create net- tractive to clients only when the available number
work effects. The pricing strategy of a platform is of workers actively participating on them reaches
an important instrument for leveraging network a certain limit, or critical mass (Liu et al. 2019).
effects and also limiting multi-homing,5 as this The pricing on digital labour platforms is thus
can have an impact on its potential revenues and dependent on the available pool of workers on
profits (Cusumano, Gawer and Yoffie 2019; Rochet the supply side and the number of clients on the
and Tirole 2003). As part of their pricing strategies, demand side.
platforms sometimes incentivize one side of the
platform through subsidies, which can motivate
the other side to join (asymmetric); alternatively, The pricing strategy
they sometimes provide incentives to both sides
(symmetric) to attract users. For instance, on
of a platform is an important
taxi platforms both the customers (low cost of instrument for leveraging
rides) and taxi drivers (bonuses or other financial
incentives besides per-ride compensation) are network effects.
5 Multi-homing refers to users signing up on multiple platforms. For instance, when a delivery worker signs up on two or more
platforms such as Cornershop, Rappi and Uber Eats to access work, then the worker is said to be multi-homing.
The commission fee

The pricing strategy adopted by platforms to
charged by freelance and
appeal to clients or customers and workers contest-based platforms
includes setting the price for the task, charging
different types of fees, and providing subscription to workers is higher than
plans. The different fees charged and the subscrip-
tion plans offered across the various platforms that charged to clients.
are presented in tables 2.1 and 2.2 for online
web-based platforms, and tables 2.3 and 2.5 for
location-based platforms. it provided a large volume of services (Cusumano,
Gawer and Yoffie 2019; Pofeldt 2016). Furthermore,
2.2.1 Freelance and in 2019 it made some more changes to its pricing
model by introducing new paid memberships for
contest-based platforms clients and new “connects” pricing for workers
to bid for projects (Upwork 2019). In 2020, more
The price setting on freelance platforms varies de-
changes were introduced to the “connects” system:
pending on the projects or tasks. Workers usually
this allowed workers, including new workers, to
display their hourly rates in their profiles, and the
have free “connects”, and additional “connects”
rates are then negotiated with the client. On some
to be allocated to workers depending on their
platforms, such as Freelancer, PeoplePerHour and
subscription plans. Similarly, in China, to expand
Upwork, the price can be determined on an hourly
its market share and attract new workers, the
basis or fixed price based on the tasks involved.
platform Zhubajie (ZBJ) moved from a “pure com-
On contest-based platforms, such as 99designs,
mission model”, whereby it charged a 20 per cent
Designhill and Hatchwise, the price that clients
commission between 2005 and 2012, to removing
pay for a particular contest is set by the platform
all service charges for projects, except for design
through its subscription plans. The price varies
competitions and piece-rate projects, in 2015. The
depending on the contest category (for example,
platform was able to adopt the strategy of subsi-
labelling, logo, app design) and the subscription
dizing workers and clients thanks to the availability
plan chosen. The 99designs platform allows cli-
of large venture capital funds (2.6 billion Chinese
ents to set the price for both one-to-one projects
yuan or US$402 million) (Chen, forthcoming).
and contests, but for the latter it specifies that
their price has to be above a minimum threshold The commission fee charged by freelance and
corresponding to the price of the least expensive contest-based platforms to workers is higher than
subscription plan. that charged to clients on most platforms being
analysed; it is a percentage of the negotiated price
Freelance and contest-based platforms charge
for the task or service and varies between 20 and
commission fees to the platform worker, while the
35 per cent. This leaves workers with between
client is often subsidized and either pays a lower
80 and 65 per cent of their negotiated price as
fee for its account to be processed or no fee at all.
earnings and has implications for their income
Some exceptions exist, such as Toptal, which does
security (see section 4.2.2). Some freelance plat-
not seem to charge workers commission fees.
forms, such as Upwork and PeoplePerHour, reduce
Platforms compete with each other mainly the worker’s commission fee to 5.0 or 3.5 per cent
through their pricing strategies, which, as a result, if the worker provides regular services to the same
change constantly. For instance, Upwork made sig- client and has earnings in excess of US$10,000 or
nificant changes to its pricing model in May 2016: US$7,000, respectively. This in effect locks workers
from charging workers a flat rate of a 10 per cent into the platform, requiring them to build up
commission fee it moved to a tiered structure (5 to their reputation and work relationship with the
20 per cent) based on the amount earned with a client in order to obtain repeated contracts and
particular client (see table 2.1). The pricing model reduce their commission fees. This practice is also
for business clients was also changed to stimulate adopted by some contest-based platforms, for
more business by charging less to clients to whom example 99designs.
Table 2.1 Revenue model of selected online web-based platforms, January 2021
X
Clients Workers
Maintenance
Maintenance
Commission
Commission
Transaction
Transaction
Additional
Additional
Deposit
fee5
fee5
fee
fee
fee
fee
fee
fee
Freelance platforms
500: 20%
Upwork – – 3%1 – – 501–10 000: 10% – $0–30 $0.15–12
> 10 000: 5%
350: 20%
£0–0.6 $0–29.99
PeoplePerHour $9.952 2.5%3 – $15 351–7000: 7.5% $9.952 $8.95–29.95
+ 10% or 2.5%3
>7000: 3.5%
$5–35;
0.75% of bid
$0–0.30 50% of
Freelancer $3 or 3% $102 – 0–20% $102 $0–25 amount
+ 2.3%; $15 contest
$0.50–50
prize
Toptal – – – $500 – – – – –
Contest–based platforms
Designhill 5% – 5% 1
– – 25–35% – ✓ –
Hatchwise – – – – $39 – – ✓ –
5–15%;
99designs 5% – – – – 20% of the first – ✓ –
$500 earned4
Microtask platforms
AMT 20–40% – – – – – – 2.9–3.9% –
Clickworker 20–40% – – – – – – – –
Appen 20% – – – – – – – –
Microworkers 7.5% $5 2
– – – – $5 2
3–7.50% –
1
Payment processing fee. 2 For inactive accounts. 3 Currency conversion fee. 4 Client introduction fee. 5 Additional fee for
clients includes fees for prioritizing or highlighting their projects or tasks; and for workers includes fees for making their profiles
more visible, and also for purchasing connects or credits to bid for projects.
Source: ILO compilation based on respective platform websites, terms of service agreements, field surveys and interviews.
The practice of charging commission fees is workers to make a security deposit for software
prevalent among online freelance platforms that projects of about 30 to 50 per cent of the project
operate globally as well as those that operate reward to the platform until the completion of
regionally, such as Kabanchik and FreelanceHunt work (Chen, forthcoming). This practice is unique
in Ukraine and 680 and ZBJ in China. Some plat- for workers and puts an additional burden on
forms in China (such as 680), however, also require workers to raise the amount in order to access
specific work on these platforms. The practice of process are set up in such a way that workers who
requiring a security deposit is targeted towards the have subscription plans or have purchased “con-
clients in other cases, and is much more prevalent nects” or paid an additional fee are more likely to
on platforms established in the United States or get projects and tasks. This strategy helps the plat-
Europe which often provide escrow services (such forms to improve their intermediation service and
as PeoplePerHour and Upwork). This ensures that attract clients, while transferring the costs of the
the worker does not bear the risk of financial loss matching process to the workers. In this system,
if the client disappears, or if an order is cancelled, since the workers depend on the platforms for
or if the fees are not paid or only partially paid, their income, they often have little choice but to
which could also threaten the smooth functioning incur costs to increase their chances of finding
of the platform (Shevchuk and Strebkov 2017). The work. This system could potentially present an
escrow services also work to the client’s advantage obstacle for some workers from the global South,
as they ensure that if the client is not satisfied with as they might not be able to access certain tasks
the services, then no payment is due. for lack of adequate financial means; this could,
consequently, negatively affect their earnings (see
The revenue model of freelance and contest-based
sections 4.2.1 and 4.2.2).
platforms is based on different types of fees and
subscription plans that are charged to workers and The clients, on the other hand, on some freelance
clients. To improve their intermediation services platforms are invited to try the platform services
and to manage the workforce on the platforms, free of cost initially, before they choose a “sub-
some freelance and contest-based platforms offer scription plan” (see table 2.2). On the basis of the
workers the option of buying a “subscription plan” plan chosen, they are offered various support
or of bidding for projects that are posted by the services and benefits. In addition to the subscrip-
clients. Platforms such as Freelancer offer various tion plans, freelance platforms offer large clients
subscription plans that are priced from US$0.99 to customized pricing and services based on their
US$69.95 per month, providing the workers with demand and budget. Toptal’s revenue model is
various benefits and services, which include giving based only on customized pricing and the plat-
their profile greater visibility, providing access to form offers clients the option to hire workers
a certain number of bids per month, and being on an hourly, part-time or full-time basis with a
able to follow employers, among others. Design minimum requirement of services for 80 hours,6 at
platforms such as Designhill also provide annual prices ranging from US$60 per hour (developers)
subscriptions ranging from US$100 to US$200 to more than US$8000 per week (finance experts)
(see table 2.2). Upwork and PeoplePerHour have depending on the skills requirements.7 The client
introduced “connects” or “proposal credits” that is required to deposit an initial amount of US$500
workers have to purchase to be able to bid for as security. The platform offers a “no-risk” trial
projects. Platforms also offer workers a range of of three experts for a position, and if the client
other services such as “highlighting” or “featuring” is not satisfied, they are guaranteed to have
their projects or proposals, for a fee, to enhance their deposit back. This allows Toptal not only to
their visibility so that they stand out when cli- ensure clients’ satisfaction but also to establish a
ents search for workers on the platform. These good reputation for the services it provides in the
fees are in addition to the commission fees that multi-sided market.
workers pay to the platform, which vary across the
Contest-based design platforms offer two kinds
different platforms.
of subscription plans to clients. Designhill and
Workers are often encouraged to subscribe to paid 99designs offer guaranteed contests, which are
services, as the algorithms used for the matching non-refundable. If no winner is selected the prize
6 This information is based on an ILO interview with a “Supervisor” at Toptal.

7 This information was obtained from the frequently asked questions (FAQs) section of the Toptal website, August 2020.

The fees charged Box 2.1 Private employment agencies

X
to workers significantly Temporary agency work as practised in recent

decades is a regulated form of work. It involves
contribute to platform a triangular employment relationship wherein a
worker is employed by an employment agency that
revenue. matches them with an employer. The ILO Private
Employment Agencies Convention, 1997 (No. 181),
defines a private employment agency as a natural
or legal person engaged in “matching offers of and
amount is equitably distributed among the par- applications for employment” and/or “employing
ticipating designers. For other contests there is workers with a view to making them available to a
a 100 per cent “money-back guarantee”, which third party which assigns their tasks and supervises
enables the platforms to attract clients. These the execution of these tasks” (Art. 1).
platforms offer clients varying pricing plans for
each contest and the platforms often compete The World Employment Confederation (WEC),
with one another on the pricing of the plans a global representative of private employment
and services offered, as in traditional markets. services, welcomes the “online talent platform
Among such offerings are access to top designers, technology” and embraces platforms, emphasizing
access to a greater number of contest entries, and the value that these bring to jobseekers. It asserts,
prioritized support. however, that in order to ensure a level playing field
platforms must comply with global standards for
The fees charged to workers significantly con- private employment services, mainly “the ban to
tribute to platform revenue, particularly among charge recruitment fees to workers” and the “com-
freelance platforms. For instance, about 90 per pliant and confidential use of personal data” (WEC
cent of Upwork’s revenue for 2019 came from 2020, 2). This is in accordance with ILO Convention
the “marketplace”, and it earned 62 per cent of its No. 181, which provides that agencies “shall not
US$300 million revenue from different types of charge directly or indirectly, in whole or in part, any
fees charged to workers, compared to 38 per cent fees or costs to workers” (Art. 7). The Convention
from the clients (Upwork 2019, 107). This is despite also regulates the processing of workers’ personal
the fact that Upwork provides “payroll services” via data to ensure that their privacy is protected and
a third party, and customized services for 30 per respected.
cent of Fortune 500 companies (Upwork 2019).
The practice of charging fees to workers may be The WEC maintains that platform work is in essence
contrary to international labour standards,8 which a new way of organizing work, and that given its
prohibit agencies, employers and intermediaries diverse nature it cannot be adequately regulated
from charging fees (see box 2.1; see also Chapter 5 by a one-size-fits-all regulation. Rather, platform
for further discussion). Despite the practice of work calls for the redesign of existing labour market
charging fees to raise revenues, most of the institutions to accommodate a more dynamic world
platforms have a history of making net losses, of work, together with a minimum floor of rights
which brings into question the sustainability of which includes respect for the ILO Fundamental
the business model. Upwork, for instance, had Principles and Rights at Work and which promotes,
an “accumulated deficit of US$172 million” as among others, portability and transferability of
of December 2019, and the platform is uncer- benefits across jobs and sectors, as well as access
tain about achieving or sustaining profitability to training and lifelong learning (WEC 2020).
(Upwork 2019, 11).
8 The ILO Protection of Wages Convention, 1949 (No. 95), and the Private Employment Agencies Convention, 1997 (No. 181).
Table 2.2 Subscription plans for online web-based platforms, January 2021
X
Clients Workers
Subscription Customized Subscription Customized

Free trials Free trials
plan service contract plan service contract
Freelance platforms
Upwork ✓ $49.99/ month ✓ – $14.99/ month –
Based on points
PeoplePerHour – ✓ – – –
system
$0.99–69.95/
Freelancer ✓ ✓ ✓ ✓ –
month
Toptal ✓ – ✓ – – –
Content–based platforms 1
99designs – $299–1299 – – – –
Designhill – $249–999 ✓ – $100–2002 –
Hatchwise – $89–399 – – – –
Competitive programming platforms3
Topcoder – – ✓ – – –
HackerRank ✓ $249–599 ✓ – – –
HackerEarth ✓ $119–279 ✓ – – –
Kaggle ✓ ✓ ✓ – – –
CodeChef – – ✓ – – –
Microtask platforms
AMT – – ✓ – – –
Clickworker – – ✓ – – –
Appen – – ✓ – – –
Microworkers – – ✓ – – –
1
Subscription plans for a logo design contest; plans vary across different contest types. 2 Designhill offers its designers
annual designer membership subscription plans. 3 Subscription plans for recruitment purposes. These charges are
monthly, to be billed annually.
Source: ILO compilation based on platform websites and terms of service agreements.
2.2.2 Competitive 2.2.3 Microtask platforms

programming platforms On microtask platforms the prices are usually
determined unilaterally, either by the platform
On competitive programming platforms, the
or by the client. On AMT, for instance, clients
prices for subscription plans and for competitions
determine the price for tasks and decide whether
are fixed by the platforms themselves. The rev-
to accept the completed task and pay workers,
enue model of these platforms is largely based
while Clickworker specifies on its platform that for
on charging clients and includes two types of
participants from Germany, the price should be
revenue streams (see table 2.2). First, platforms
equivalent to the German minimum wage. Appen
provide clients with recruitment services to which
and Microworkers have a basic formula to esti-
they can subscribe through various plans pro-
mate the cost of a job, taking into consideration
posing a range of services and benefits. Second,
any specifications indicated by the client and all
they charge fees to clients wherein they provide
related costs.
customized services and develop a range of pro-
jects, from prototypes to the development of new
algorithms, based on specific client requirements.
Both recruitment and customized services are pro-
vided by means of competitions or “hackathons”
On microtask
in which the platform community of developers, platforms the prices
programmers or data scientists takes part. The
Topcoder platform also offers “Talent as a Service” are usually determined
(TaaS) programmes to clients and recommends
workers from the Topcoder community of pro-
unilaterally, either by the
grammers to meet specific skills requirements. platform or by the client.
Competitive programming platforms do not
charge fees to developers and programmers; they
Workers on microtask platforms are not charged a
build communities of programmers and devel-
commission fee; instead, clients are charged a fee
opers who can provide top-quality services while
that is determined in relation to the amount paid
at the same time honing their skills. Workers on
to the platform workers. The commission fee is
these platforms are rewarded through monetary
typically assessed and charged at the time of pay-
prizes and non-monetary benefits (Boudreau
ment for the work performed, and varies between
and Hagiu 2009), which include the opportunity
7.5 and 40 per cent.9 Some platforms, such as AMT
to participate in regular contests and competi-
and Microworkers, offer additional services to
tions, access to software libraries, rankings and
their clients if they want to target specific groups
skills ratings, peer reviews, and for highly rated
of workers based on age, sex, experience or
or ranked programmers, sharing of their profiles
nationality, for which the platforms charge an add-
with companies for hiring purposes.
itional fee in terms of either a percentage of the
task or a fixed amount per assignment (ranging
from US$0.05 to US$1.00 on AMT).10 Microtask
platforms also offer custom-tailored services
based on client requirements.
9 Based on information provided by the platforms covered by the microtask survey.

10 This information is based on surveys conducted on these two platforms in 2017.
2.2.4 Taxi platforms Such bonus schemes usually depend on the

number of rides accomplished in a day or a week;
The ride fare on taxi platforms is determined by the drivers are incentivized to meet targets, which
the platform using algorithms that are based on can result in working long hours to earn the extra
factors such as distance, time taken to reach the money promised (Surie and Koduganti 2016; see
destination, fuel cost, type of vehicle and financial section 4.2.3). Over time, however, the targets are
capacity of the customers to spend in a particular increased and the rewards reduced, which also af-
geographical area of the city.11 During periods of fects the incomes of the taxi drivers. The pricing
high demand, platforms also use surge pricing mechanisms followed by taxi platforms can also
algorithms that allow them to determine the ride lead to extensive litigation (see box 2.2). Moreover,
fare based on demand and supply. drivers often find it hard to meet the final target,
The revenue model of taxi platforms is based on as the algorithm often does not assign enough
charging commission fees to the taxi driver. The rides when drivers are getting close to their target
commission fee, which is a percentage of the (Rosenblat and Stark 2016). This situation could
ride fare, varies within and between platform also be due to oversupply of workers competing
companies. For instance, the commission fee for rides on these platforms (van Doorn 2017). To
charged by Uber is 25 per cent in most countries encourage clients or customers to use their plat-
under consideration (see table 2.3; section 4.2.2 forms, the companies provide rewards or coupons
provides additional details), but in some countries or subsidize the costs of rides, keeping them low
where there is intense competition, a lower fee is compared to traditional taxis or other companies.
charged (20 per cent in India; 5 per cent in Kenya).
Companies also vary the commission fee based Many taxi platforms are able to provide subsidies,
on the income earned by the drivers12 and raise bonuses and other incentives because of funding
revenues through their surge pricing algorithms made available by venture capital and other funds
(Lee et al. 2015). (see section 1.5). This strategy allows these plat-
forms to have network effects, enter new markets
Taxi platforms also try to motivate and retain (countries) and expand their customer base there.
workers and clients or customers through gami- Uber, which is a dominant player in the taxi sector,
fication and rewards. Gamification for taxi drivers,
has raised US$25.2 billion from 28 funding rounds
which takes the form of offering incentives or
of venture capital (to January 2021),13 expanded its
bonuses to stimulate their engagement, was re-
services in 69 countries and had an accumulated
ported by three quarters of app-based taxi drivers
deficit of US$16.4 billion in December 2019 (Uber
(see section 4.2.2). The strategy adopted to attract
2020a). Uber is able to sustain its business and
taxi drivers differs across countries depending on
market share largely as a result of the availability
local demand, cultural context and the presence
of funds from venture capital, which allow it to
of business competitors. For example, Uber’s
offerings of bonuses and incentives vary consid- subsidize both sides of the market and also to
erably between countries (see table 2.4). Among penetrate a number of new markets (Cusumano,
Uber drivers who reported being offered bonuses Gawer and Yoffie 2019; Horan 2019). The investors
and incentives, in most countries a high propor- are betting on a winner-takes-all outcome, wherein
tion indicated being rewarded for completing a Uber would emerge as a market leader and then
certain number of rides. Another way in which reduce the subsidies or even increase the commis-
platforms incentivize drivers is by offering them sion fees charged to the drivers, or raise the price
bonuses during specific times (peak demand), or of the rides (Cusumano, Gawer and Yoffie 2019).
for working asocial hours, a practice that is quite Along with the rise in digital labour platforms, an
popular among all taxi platforms. A significant alternative platform structure, the “platform co-
proportion of Uber drivers in Chile, Lebanon and operative”, which is collectively owned and funded
Ukraine reported receiving similar offers. (see box 2.3), is increasingly gaining ground.
11 These indicators are based on ILO interviews with taxi platform companies.

12 Based on ILO interviews with taxi platform companies.

13 Based on information collected from Crunchbase database.

Table 2.3 Revenue model of selected taxi platforms in selected countries, 2019–20
X
Clients Workers
Maintenance Transaction Commission Maintenance Transaction

fee fee fee fee fee
Uber
Chile ✓ ✓ 25% (18–35) ✓ ✓

Ghana ✓ ✓ 25% (15–25) ✓ ✓
India ✓ ✓ 20% (15–44) ✓ ✓
Kenya ✓ ✓ 5% (5–25) ✓ ✓
Lebanon ✓ ✓ 25% ✓ ✓
Mexico ✓ ✓ 25% (10–37) ✓ ✓
Ukraine ✓ ✓ 25% (10–35) ✓ ✓
Careem
Lebanon ✓ ✓ 20% (15–25) ✓ ✓

Morocco ✓ ✓ 25% (10–40) ✓ ✓
Bolt
Ghana ✓ ✓ 20% (10–25) ✓ ✓

Kenya ✓ ✓ 20% ✓ ✓
Ukraine ✓ ✓ 15% (10–40) ✓ ✓
Ola (India) ✓ ✓ 20% (15–40) ✓ ✓
Little (Kenya) ✓ ✓ 5% (5–20) ✓ ✓
Grab (Indonesia) ✓ ✓ 20% (5–40) ✓ ✓
Gojek (Indonesia) ✓ ✓ 20% (10–33) ✓ ✓
Notes: The data on commission fees for taxi platforms is based on the ILO selected country surveys of taxi drivers (see
Appendix 4A). The figures shown are the commission fees (2019–20) that were mentioned most often by respondents per
country and platform. Figures in parentheses are the range of commission rates mentioned by taxi drivers.
Source: ILO compilation based on respective platform websites, terms of service agreements, field surveys and
interviews.
Table 2.4 Criteria for receiving bonuses or incentives on Uber, selected countries
X
(percentage of respondents)
Reaching Reaching
New Working asocial hours Working during
or exceeding or exceeding a certain
drivers (night or holiday) high-demand hours
an hourly threshold number of rides
Chile 1 25 28 74 28
Ghana 4 4 27 92 3
India 0 0 8 98 12
Kenya 11 27 33 78 0
Lebanon 3 41 8 58 65
Mexico 0 4 11 88 38
Ukraine 4 20 33 85 42
Note: Figures refer to workers who reported being offered bonuses or incentives by Uber.
Source: ILO selected country surveys of app-based taxi drivers (2019–20).

Box 2.2 Pricing by taxi platforms and potential for litigation:

X
The case of Ola and Uber in India
Litigation in India illustrates the complexity and uncertainty of applying competition law to
platform work. Uber entered the Indian market in 2013, by which time another local platform
company, Ola, already had a three-year head start. MERU (a radio taxi company) alleged that
both Ola and Uber subsidized the cost of rides to attract customers,1 and to compete with
traditional taxi drivers and taxi companies.
Both Ola and Uber aggressively recruited drivers by providing them with financing to purchase
or lease vehicles, and various other incentives (Surie 2018). Uber gave incentives of 2,000 rupees
(US$31.2) for completing 12 rides per day in early 2016 to drivers in New Delhi; although by
December 2016 it had changed its incentive model to offering such incentives just once a week
for completing 40 to 50 rides, and also increased the commission rate from 20 to 25 per cent
(Dhillon 2018). Similarly, an Ola driver noted that he was earning as much as 75,000 rupees
(US$1028.7) to 100,000 rupees (US$1371.6) a month working 12–13 hours a day in 2016, but by
2017 the amount had dropped to 40,000–45,000 rupees (US$548.6–617.2) a month working
15–16 hours a day, due to the changes in the trip incentive model (Ayyar 2017).
Furthermore, platform drivers were also incentivized to recommend other drivers and were paid
a one-off sum per successful referral, which varied across the cities. They were also offered free
insurance, free registration for vehicles, cash discounts and lucky draws for domestic appliances.
The measures helped Uber to create network effects in the Indian market and to challenge its
competitor Ola and the traditional taxi sector. The latter has since then dwindled in numbers
in many Indian cities. In response, Ola introduced the minimum guarantee scheme to attract
workers and assured them of a minimum amount after meeting a particular target.2
MERU filed a series of complaints before the Competition Commission of India3 alleging that Ola
and Uber were engaging in practices contrary to Sections 3 (anti-competitive agreements) and 4
(abuse of dominant position) of the Competition Act 2002. On the one hand, the Competition
Commission decided in Ola and Uber’s favour and found that given the nature of competition
within the radio taxi markets of Chennai, Hyderabad, Kolkata and Mumbai, prima facie dom-
inance of Uber and Ola individually could not be made out (para. 41), and with regard to Section 3
the allegation did not hold merit (para. 37). On the other hand, MERU successfully appealed to
the Competition Appellate Tribunal on an earlier case of 2015 that it had lost,4 with regard to
alleged predatory pricing and the Tribunal reversed the Commission’s decision and ordered an
investigation into MERU’s allegations.5 Uber subsequently filed an appeal before the Supreme
Court of India against the Tribunal’s reversal, which was rejected by the court in September
2019.6 The experience in India is not necessarily reflective of other jurisdictions, where both the
relevant competition legislation and the business practice might differ substantially.
1
From Case No. 96 of 2015: Rates for Uber Black: November 2013, 20 rupees/km; June 2014, 18 rupees/km;
November 2014, 18 rupees/km; February 2015, 12 rupees/km. While the traditional taxi rates were 23 rupees/km
in November 2013. 2 Information based on ILO interviews with workers. 3 Competition Commission
of India, MeruTravelSolutionsPvt.Ltd.andANITechnologiesPvt.Ltd.andUberIndiaSystemsPvt.Ltd. & Ors.
Case No. 25–28 of 2017. 4 Competition Commission of India, MeruTravelSolutionsPvt.Ltd.andUberIndia
SystemsPvt.Ltd. & Ors. Case No. 81 and No. 96 of 2015. 5 Competition Appellate Tribunal, MeruTravelsSolutions
Pvt.Ltd.v CompetitionCommissionofIndia& Ors., Appeal No. 31 of 2016. 6 Supreme Court of India, Uber India
SystemsPvt.Ltd.vCompetitionCommissionofIndia& Ors., Civil Appeal No. 641 of 2017.
2.2.5 Delivery platforms

Box 2.3 Platform cooperatives
X
On delivery platforms, the delivery fare for the
Platform cooperatives are collectively
workers is determined by the platform using algo-
owned and have been gaining in popu-
rithms that are based on a number of factors, such
larity over the past decade. Platform
as demand and distance, among others, and it is
cooperatives are designed and owned by
only once the delivery workers have accepted the
their members, who usually pay a small
delivery that the fares are made available to them.
contribution from their earnings towards
the maintenance and development of Delivery platforms charge restaurants, shops
the platform.1 Given that work on these and supermarkets a commission fee and charge
platforms is co-determined and decisions customers a delivery fee. The commission fee
are taken based on participatory demo- charged to restaurants or supermarkets ranges
cratic processes, platform cooperatives between 12 and 35 per cent depending on the
are likely to be more transparent and platform and country (see table 2.5). Delivery
accountable to their members than platforms also charge customers delivery fees: for
digital labour platforms in which many instance, Cornershop, Jumia Food and Uber Eats
functions are algorithmically managed. charge a minimum delivery fee to the customer,
while on other platforms delivery fees vary based
There are currently various platform
on factors such as distance (Deliveroo and Glovo)
cooperatives operating in a number of
or a percentage of the purchase price (Jumia Food
sectors, from taxi (such as Green Taxi
and PedidosYa). As reported by many restaurants,
Cooperative and ATX co-op Taxi, in the
platforms also charge business clients higher com-
United States and Eva in Canada) and
mission fees if they offer their products through
delivery (such as Coopcycle2) services
multiple platforms. Delivery platforms often state
to house-cleaning (such as Up&Go,
in the exclusivity clauses of contracts that they will
New York City) and e-commerce (such
charge lower commission fees for clients working
as Fairmondo, Germany). Their vision is
exclusively with them.
to create a genuine “sharing” economy,
committed to fair labour practices. For Some of the delivery platforms also provide
instance, Eva is a cooperative that allows discounts to customers as a strategy to expand
driver members, rider members and their business in the specific region or area. For
worker members to be part of the co- instance, Toters in Lebanon gave a 50 per cent
operative. The drivers earn about 15 per discount to customers for their purchase from
cent more than on other available taxi certain restaurants and shops, and these costs
platforms in the region.3 The cooperative were borne either by the platform or at times
structure of many of the platforms has jointly with the restaurants or shops. In the event
also allowed their members to self-or- of cancellation, customers are often charged a
ganize efficiently during the COVID-19 cancellation fee that comprises the price of the
pandemic by equitably distributing tasks products ordered and the delivery fee, if a delivery
among themselves. worker has already been assigned the task. Some
platforms also offer premium memberships to
1
This contribution tends to be much
lower than the commission charged by
customers, whereby the delivery fee is waived if
the digital labour platforms studied in this the orders exceed a certain amount.
report. 2 Coopcycle is a network of bike
delivery cooperatives that operates in Belgium,
Denmark, France, Poland, Spain, the United
Kingdom and the United States. 3 For more
details, see: https://eva.coop/#/driver; http://
cities-ess.org/topics/eva-coop/?lang=en.
Table 2.5 Revenue model of selected delivery platforms in selected countries, 2019–20
X
Clients (restaurants, shops and supermarkets) Customers
Commission fee Delivery fee

Commission fee
per order per order
(%)
(US$) (US$)
Chile
Rappi 19–28 1.95–5.47 1.40–5.61
Uber Eats 15–33 1.68–2.67 1.68–3.09
PedidosYa 25–28 1.25–4.91 1.25–5.61
Cornershop 15 5.47–6.87 5.47–6.87
India
Swiggy 22–24 – –
Zomato 12–25 – –
Lebanon
Toters 20–25 – –
Zomato 10–20 – –
Kenya
Uber Eats 15–25 – –
Jumia Food 16–20 – 1.37
Glovo 15–20 – –
Ukraine
Glovo 28–35 – –
Mexico
Uber Eats 26–35 – –
DiDi Food 20–30 – –
SinDelantal 22–30 – –
Source: ILO compilation based on respective platform websites, terms of service agreements, field surveys
and interviews with restaurants, shops or supermarkets in the respective country.
2.3 Recruitment and matching of workers with clients
Digital labour platforms are transforming human

resource practices and the employment relation-
ship, which has major implications for the future
There are two types
of work. This section discusses recruitment of work relationship on
practices, matching of workers with clients and
assignment of tasks. digital labour platforms:
internal employment or
2.3.1 Work relationships
external employment.
on platforms
There are two types of work relationship on digital drivers in 69 countries around the world for whom
labour platforms: workers are either directly Uber mediates work (Uber 2020b). A number of
hired by the platforms (internal employment) delivery platforms also have a higher number
or their work is mediated through the platforms of directly hired workers (more than 5,000) than
(external employment). Figure 2.3a shows the other types of platforms; Meituan, for example,
number of employees directly hired by online has 54,580 full-time employees. This is largely
web-based platforms, which varies between because many of these companies hire delivery
50 (PeoplePerHour) and 800 (Appen). In contrast, workers as employees to establish a market base.
about 2.4 million skilled workers were registered Once their objectives are achieved, however, some
globally on PeoplePerHour as of January 2021. of them change their labour practices and hire
The number of employees hired directly by workers on a part-time or a piece-rate basis. For
location-based platforms is far higher than on example, Delivery Hero (Germany), PedidosYa
online web-based platforms (see figure 2.3b). (Argentina) and Swiggy (India) initially hired
On location-based taxi platforms, this number workers on a full-time basis, but once they had
varies between roughly 1,200 (Careem) and established their market
26,900 (Uber), although this rep- position, they terminated
resents only a fraction of the many of the full-time
approximately 5 million cont rac t s and hire d
workers on a per-task basis, and have been

The number of
progressively reducing the number of workers employees directly hired
directly employed. In the case of Meituan (China),
the platform has been hiring workers through (internal employment) by
third-party staffing agencies (Sun, Chen and Rani,
forthcoming). Among the platforms surveyed, platforms is a mere fraction
the number of employees directly hired (internal
employment) by platforms is a mere fraction of
of the number of workers
the number of workers whose work is mediated whose work is mediated.
(external employment).
Figure 2.3 Number of employees directly hired by digital labour platforms, 2019–20
(a) Online web-based platforms
800
800
700
600
600
Number of employees
570
500
478
400
300
200
200
100 139
50 43
0
Toptal Upwork Freelancer PeoplePerHour 99designs HackerRank Kaggle Appen
Freelance Contest-based Competitive programming Microtask
(b) Taxi and delivery platforms
60 000
50 000
40 000
Number of employees
30 000
20 000
10 000
0
Uber Ola Grab Lyft Gojek Bolt Careem Meituan Swiggy Zomato Deliveroo Glovo Rappi
Taxi Delivery
Sources: Owler database, annual reports, filings by platform companies to the Securities and Exchange Commission of the United States
and platform websites.
Workers directly hired by platforms have an competitive wages based on hours worked, and
employment relationship, while those whose paid time off.14
work is mediated by platforms are typically
Related to the ongoing discussion on misclassifi-
considered by the platforms as “self-employed”,
cation of platform workers, some location-based
“independent contractors”, “third party service
platforms offer insurance coverage for accidents
providers”, “designers”, “freelancers” and so on,
and hospitalization at no extra cost to workers.
and consequently do not have an employment re-
Deliveroo’s insurance policy, for instance, covers
lationship (see Appendix 2B for the different terms
riders from the moment they are online and for
used by platforms for workers). These platforms
one hour after going offline, and provides sup-
justify their approach to their relationship with
porting income when they are unable to work
their workers on the basis that workers have the
following injury. In France, notably, Deliveroo
flexibility to choose their own work schedules (see
riders also benefit from paid sick leave – €30 per
Chapter 5 for more details). Furthermore, some
day for 15 days – provided they have completed
platforms, such as AMT, Clickworker and Upwork,
at least 30 rides in the previous eight weeks. In-
even specify that users of the platforms will not
ride insurance and social protection benefits are
be offered employment related benefits such as
offered to varying degrees by Uber depending
sick leave, health insurance or retirement benefits.
on the country,15 and in India all taxi platforms
Platform companies are able to devolve their re- are obligated to provide health and life insurance
sponsibility for providing the requisite employment to taxi drivers. Some of the delivery platforms
or social protection benefits to their workers and (such as Swiggy) also provide medical and acci-
to save on labour costs. This also provides plat- dent insurance coverage to workers and their
forms with greater employment flexibility than family members.
traditional employment agencies, which rely on
With the spread of the COVID-19 pandemic,
dependent employees (Schwellnus et al. 2019).
some delivery platforms are looking to improve
Some industry executives have estimated that
working conditions and protections for those
classifying platform workers as employees instead
whose work they mediate. For instance, the CEO of
of independent contractors would cost platform
JustEatTakeaway, one of the largest delivery plat-
companies 20 to 30 per cent more (Scheiber 2018).
forms globally, recently emphasized: “We’re a large
Uber mentions in its annual report that if drivers
multinational company with quite a lot of money
were to be classified as employees then it would
and we want to insure our people […] We want to
have to “fundamentally change” its business
be certain they do have benefits, that we do pay
model, which would “have an adverse effect on
taxes on those workers” ( Josephs 2020). Good
[its] business and financial condition” (Uber 2020a,
practices are also followed by some other delivery
13). Similar consequences are also mentioned
platform companies. These include BOX8, which
by online web-based platforms such as Upwork
has been providing food and grocery delivery in
(Upwork 2019, 15). However, some companies
Indian cities since 2012, and which offers full-time
such as Alto in the United States have come up
contracts to its employees, and provides social
with an alternative model and hire drivers as em-
protection benefits and incentives for upskilling.16
ployees providing, for example, health benefits,
14 For more details, see: https://www.ridealto.com/driver-application.

15 Uber provides a range of protections, including accident, injury, illness, and paternity benefits for drivers and delivery workers
in partnership with AXA in European markets and in partnership with Chubb in Australia and South Africa. For more details, see:
https://www.uber.com/en-GB/blog/supporting-drivers-with-partner-protection-from-axa/; https://www.uber.com/za/en/drive/
insurance/; https://www.uber.com/en-AU/newsroom/partnersupportaustralia/.
16 For more details, see: https://www.taciturban.net.in/companies/box8/.
2.3.2 Basic requirements 2.3.3 Algorithmic matching

for opening an account of clients and workers
on platforms Platforms are introducing a paradigm shift in the
conventional human resource process of how cli-
Online web-based platforms adopt various stra-
ents (demand) and workers (supply) are matched.
tegies to build their talent pool, so as to attract
Instead of assigning workers and tasks through
clients. For this purpose, they verify the skill levels
human interaction, some platforms use fully
of workers before a platform account can be
automated matching processes for assignment of
opened. At one end of the spectrum are freelance
work. Workers are automatically matched to client
platforms, which conduct rigorous screening
requirements and assigned a task on the basis of
processes that can last from one to three weeks
a number of platform-specific indicators. These
(e.g. Toptal), or have online skill tests17 (e.g.
include a combination of worker ratings, worker
Upwork) or a designer curation team that reviews
profiles (such as expertise level and skills), client
applications by potential workers (e.g. 99designs).
reviews, availability, time zones and hourly rates,
At the other end of the spectrum are competitive
among other factors. An analysis of 117 freelance
programming and microtask platforms, which
and contest-based platforms shows that ratings
anyone can join without their skills being vetted.
(50 per cent) and client reviews (60 per cent) are
Some platforms stipulate in their terms of agree-
the two major factors used in assigning tasks to
ments that they do not permit registration of
workers (see figure 2.4). Other factors taken into
people from countries that are subject to sanc-
consideration include worker profiles (46 per cent),
tions. Many platforms also reserve the power
project history or portfolio (27 per cent) and the
in their terms of service agreements to refuse
rate proposed by the worker (21 per cent).
registration of a “user” at their own discretion.
On location-based platforms, registration or
onboarding is fairly straightforward, though in
order to actually access and complete tasks on
the platforms, workers have to meet certain add- Some platforms use
itional requirements. In most cases, taxi drivers
and delivery workers are required to provide
fully automated matching
themselves with the necessary equipment, such processes for assignment
as a smartphone, vehicle (car, scooter or bicycle)
and thermal backpack (in the case of delivery of work.
platforms). In some countries, taxi platforms
lease cars to drivers. Platforms usually require
formal identification, such as a driving licence, Some freelance platforms rely exclusively on
social security or identity card, and vehicle-related algorithmic matching (based on targeted indi-
information, such as vehicle registration and insur- cators) of clients with workers (e.g. Freelancer,
ance. In some cases vetting (such as criminal or PeoplePerHour), while others use a mix of algo-
other background checks) is imposed by regula- rithmic matching and human interaction to assign
tions and can lead to more rigorous onboarding the task to the worker (e.g. Toptal, Upwork).18
processes. For example, after incidents of sexual On these latter platforms, algorithmic matching
assault of passengers in India and China, Uber and provides the client with a shortlist of the top
DiDi introduced background checks (Uber 2020c; three to five workers who could perform the task.
Yuan 2018). The client is then assigned a design specialist or
17 Upwork has recently discontinued its online skill tests.

18 This information is based on ILO interviews with the platform companies.
Figure 2.4 Indicators used to determine client–worker matching on freelance and contest-based platforms
60
60
50
50
40
Percentage
46
30
20 27
21
10
0
Client reviews Ratings Worker profile Project history Rate proposed
or portfolio by the worker
Note: Information was available and considered for analysis for 117 freelance and contest-based platforms.
Source: ILO estimates based on online labour platforms listed on Crunchbase database and the platform websites.
supervisor to discuss the task requirements and

the specific skills needed, and is provided with chat Platforms also
and video-conference tools for scheduling inter-
views with one or two workers from the shortlist.
allow workers to bid on
This enables the client and the worker to finalize specific tasks through
the contract agreement and to negotiate the price,
working schedule and deadline. the payment of a fee.
While ratings and client reviews are an important
part of the matching process, platforms also
Most challenges or hackathons on competitive
allow workers to bid on specific tasks through the
programming platforms are open to the com-
payment of a fee which gives them more visibility
munity of developers, coders and programmers,
(see section 2.2). These practices carry the risk
except some to which the platforms invite only
of excluding some workers with better worker
highly rated or ranked programmers. Eligibility
ratings who have not paid the fee or those with
to perform the various tasks on microtask plat-
low purchasing power from participating in a
forms is determined by worker ratings, which are
fair matching process (see section 4.2.1). On con-
algorithmically determined. In addition, on some
test-based platforms, the clients, based on the
platforms clients can specify further criteria for in-
subscription plan for which they have opted, often
cluding or excluding workers, such as nationality,
set the price and the requirements of the project,
gender or age (see section 2.2.3). Tasks are then
and workers can then submit their portfolio and
automatically made available to eligible workers
proposals within a limited time. The contests are
on a first-come, first-served basis.
either open to all designers or are restricted to
top-level designers based on such factors as rat- Task assignment on both taxi and delivery plat-
ings, client reviews, work histories and repeated forms is generated by algorithms and based on
assignments with clients, and the client’s require- worker ratings, which are calculated through indi-
ments. Some platforms, such as 99designs, also cators such as ratings by clients, cancellation rate
restrict the number of contests that designers can and acceptance rate. Workers are often given a
enter on the platform per month, based on their limited timeframe (usually a few seconds) to decide
skill level. whether to accept or reject a ride or delivery. In
addition, taxi platforms use “surge pricing” based their skills, profiles and thereby opportunities.
on demand, which can strongly influence drivers This is most common on freelance platforms,
to make themselves available in areas where there which offer workers online training and tests free
is a peak in demand (Duggan et al. 2020; Rosenblat of charge to help them improve their chances of
and Stark 2016). Some of these practices are incon- obtaining tasks. PeoplePerHour, for example, has
sistent with the platforms’ assertions that workers an “academy” where workers can take courses,
are free to set their own working schedules and gain skills, access training programmes and earn a
accept or reject work, because acceptance or re- PeoplePerHour academy diploma, which can then
jection of work assignments can have significant be displayed on their profile. These training tools
implications for workers’ ratings and thus the and skills help workers, particularly new entrants,
amount of work they will be assigned in future to access work or improve earnings. Upwork and
(see section 4.3.1).
Kaggle allow workers to take tests at no cost and
Platforms also incentivize workers to build their then provide feedback, so that they can assess
profiles by using online training tools to enhance their own abilities and learning needs.
2.4 Work processes and performance management
The use of digital tools and algorithmic manage- provide clients with an escrow account, to which a
ment are radically transforming work processes specified amount is transferred when the contract
and performance management on digital labour is approved, and from which the payment is re-
platforms. Platforms provide a variety of tools leased to the worker’s account only once the client
to organize the work processes and communi- is satisfied with the completed work. Some plat-
cation between the client and the worker, so as forms such as Designhill allow clients to request
to ensure that the worker follows the job instruc- unlimited revisions of work by designers at no
tions carefully. extra cost. Competitive programming platforms
provide contestants with software tools and have
clear codes of conduct for those who participate
2.4.1 Work processes in challenges and competitions.
and communication In contrast to freelance and competitive program-
Workers are often required to install software ming platforms, on microtask platforms there is
and hardware tools, to deliver work within a no communication between the client or platform
prescribed period of time and to be available at and the workers. The entire work process of al-
a specified time (see section 4.3.1), as laid down location, evaluation and remuneration for a task
in platforms’ terms of service agreements. These is algorithmically managed. Workers on these
tools also allow clients to track the progress of platforms are prohibited from using any auto-
their projects and monitor worker performance mated methods to perform tasks. For example,
(see box 2.4). These practices are prevalent among AMT specifies that automated methods must not
freelance platforms and the degree of monitoring be used as a substitute for human intelligence
using digital tools often resembles that found in and independent judgement. Some of these plat-
traditional employment relationships (Rogers forms also prohibit workers from subcontracting
2018). Furthermore, in order to optimize the their work. Microtask platforms do not use any
client experience, some platforms also refund work-monitoring tools but they allow clients to
clients if the work is not up to their expectations check how much attention a worker is paying to
or if the delivery is not executed according to the a task by adding test questions. If a worker gives
terms agreed. Both Upwork and PeoplePerHour too many incorrect responses, he/she loses access
to that task and forgoes payment for it. Another communication, agreement, transfer of assets,
common strategy consists in allowing clients to de- sharing of contact details, transaction or payment
termine the time limit (minutes or seconds) within between users (clients and workers) from taking
which the task should be completed, which allows place outside the platform (see Appendix 2B).
them to exercise some control over the worker. This allows the platforms to maintain their pos-
ition as intermediaries and prevents workers
Platforms often provide strict guidelines on the
from accessing clients through other means
nature of the content that can be shared through
(see section 4.2.1).
official platform communication channels, a prac-
tice which is most common among freelance and Taxi and delivery platforms define various aspects
competitive programming platforms. The guide- of the work process, such as behaviour and cus-
lines analysed for this report also prohibit any tomer service etiquette, instructions for handling
Box 2.4 Monitoring work processes on digital labour platforms

X
Upwork provides workers on an hourly contract with a “work diary” which, once enabled, records
the number of hours worked and the number of keystrokes made, and takes random screen-
shots (six times an hour) while they work on a project (see figure 2.5). The client can access this
information to monitor the worker’s activity and progress.
For fixed-price tasks, Upwork and Freelancer suggest that clients organize projects by mile-
stones, whereby payment is contingent on achieving the agreed milestone and clients have
access to ongoing status reports. As workers have to report to clients and enter data recording
their work activity on a regular basis, the flexibility, autonomy and control they exercise over their
work is constrained.
Figure 2.5 Upwork work diary
This Week Last Week Since Start

Overview
4:30 hrs
50
32:00 hrs Dispute
167:00 hrs
Work Diary $225.00 $1,600.00 $8,350.00
Contract History
Time tracking
Payments History
Last Worked 1 hour ago Activities
Memo: Forums check Track how much time Valeria
View Profite
Active Window: Work Diary - Google Chrome Kochubei spends towards your
projects, products or customers
View Work Diary
Add or Remove Manage All
Source: Upwork work diary, from https://www.youtube.com/watch?v=qAXbzLUcjic.

Online web-based
platforms often prohibit On most platforms, such ratings determine the

any communication, nature and amount of the work assigned and
thereby the level of earnings to which the worker is
sharing of contact details, entitled. On all digital labour platforms, any delay
in or non-completion of work negatively affects
transaction or payment ratings. A lower rating can result in reduced work
between users (clients and opportunities or even deactivation of a worker’s
account. Ratings, which serve to quantify a client’s
workers) from taking place satisfaction with a designated service, are also
becoming a significant managerial practice for
outside the platform. organizations in service industries beyond digital
labour platforms (Wu et al. 2019).
deliveries and determination of working time.

Most platforms provide guidelines on non-discrim-
ination, anti-harassment, use of safety equipment
such as helmets and vests, and the importance of The use of algorithms
abiding by traffic laws and regulations. Drivers on
some platforms are instructed to take the least
to evaluate performance
costly route and refrain from making unauthor- is yet another way of
ized stops. Workers on these platforms are tracked
through the Global Positioning System (GPS), often digitalizing human
in real time, by both the platform company and the
customers, and data is collected on the number resource management.
of rides and deliveries accepted or rejected, on
earnings, and on driving metrics such as speed.
This data is then used for training the platform’s
machine-learning algorithms, which can influence All platforms use algorithms to calculate ratings,
worker ratings, access to work, fare-setting for but the indicators that are considered for the
rides or surge pricing (see section 4.3.1). calculations differ across platforms. On freelance
platforms, to take two examples, Upwork has a rat-
ings system which includes a “job success score”19
2.4.2 Algorithmic and client feedback, while on Freelancer ratings
performance management are based on the number of reviews received from
previous clients, the workers’ earnings scores,
The use of algorithms to evaluate performance their success rate in completing jobs within the
is yet another way of digitalizing human resource agreed deadline and within the price or budget,
management, replacing human supervision and whether they have been hired repeatedly
and redefining working relationships. Work is by the same client, among other factors. The
evaluated based on a number of metrics such as variations in the metrics adopted by the platforms
ratings, client reviews and evaluations, which allow and their relative weight in the algorithms used to
workers to build a reputation on these platforms. evaluate workers make the portability of ratings
There is little transparency about how worker rat- across platforms difficult, which in turn dissuades
ings determined by the algorithms are calculated. workers from moving across platforms, owing
19 The platform specifies that the job success score is calculated as the difference between successful and negative contract out-
comes, divided by total outcomes. However, an ILO interview with a manager from Upwork revealed that the job success score is
actually calculated using more complicated metrics.
to the high costs in terms of time and monetary below a particular threshold (95 per cent in the
resources required to build their reputation and case of AMT). AMT provides a “Masters” qualifica-
ratings again from scratch: workers are thus in tion to some workers who have completed at least
effect locked into a specific platform, instead of 1,000 tasks and who maintain a high approval
being able to multi-home on several platforms rating, which gives them access to varied work
(see section 4.2.1). opportunities.20 However, there is no transparency
with regard to the set of parameters or criteria
Performance is evaluated on many competitive
used for defining the “Masters” qualification
programming platforms using the Elo rating
(Kingsley, Gray and Suri 2015).
system, which calculates a worker’s expected
rank in a contest; if the actual rank is better than Taxi platforms evaluate worker performance
the expected rank then the rating will increase, using customer feedback and ratings, which are
otherwise it will decrease. On these platforms, the based on service quality and drivers’ acceptance
ratings are also dependent on the performance and cancellation of rides,21 among other factors
of other participants in the competition and the (such as speeding or damaging the vehicle). These
number of competitions in which the worker has are taken into consideration for calculating a con-
participated, among other factors. solidated rating. Workers on delivery platforms
are evaluated through feedback provided by other
Workers on microtask platforms are evaluated
platform users (clients and business partners), and
according to their ability to consistently submit
factors such as cancellation rates, participation
high-quality results and maintain a high approval
during peak periods, seniority, number of deliv-
rate, which in turn determines the kind of work to
eries and speed of delivery.
which they have access. Once tasks are completed
by workers, they are evaluated by an algorithm, The algorithmic assignment, evaluation and
which in turn accepts or rejects the tasks and management of tasks have major implications for
makes the payment or not to the worker. Rejection workers, who may not have access to a fair dispute
of work has a considerable impact on workers’ resolution mechanism to contest or appeal what
ratings, and on some platforms, such as AMT, they consider unfair rejection of work or poor
workers might not receive tasks if their ratings are ratings (see section 2.5).
2.5 Digital labour platforms’ rules of governance

and workers’ freedom to work
Digital labour platforms are adapting busi- to the ability of enterprises, particularly small and
ness practices to a digital environment. These medium-sized enterprises, to operate freely, and
practices are laid down in the terms of service are examined below.
agreements, which are unilaterally determined
by each platform and govern how users (both
Exclusivity clauses
workers and clients) interact with the platform
and among themselves. They include exclusivity Some platforms impose an exclusivity clause of
clauses, and cover acceptance or rejection of 24 months whereby, if a worker and a client meet
work, deactivation, dispute resolution, and data on the platform, both are required to use the
collection and usage. These practices pose new platform as their sole work channel for 24 months
challenges to workers’ freedom to work as well as (e.g. Upwork and 99designs). If either of the two
20 For more details, see: https://www.mturkcrowd.com/threads/masters-qualification-info-everything-you-need-to-know.1453/.

21 The cancellation rate represents the percentage of journeys cancelled after accepting a request.
The terms of service

agreements are unilaterally standards. Workers are often not notified that their
determined by the platforms. accounts will be deactivated and they realize that
their accounts have been deactivated only when
they log in, thus adversely affecting their access
to work.
parties chooses to opt out within that period,
On some contest-based platforms, accounts
they are required to pay a percentage of the esti-
can be deactivated if designers do not meet the
mated earnings over the following 12 months. In
platform’s quality standards or if the work is not
the case of Upwork, this payment is 12 per cent
original. On competitive programming platforms,
of the anticipated earnings, calculated by multi-
accounts are often deactivated for plagiarism. For
plying the worker’s hourly rate by 2,080; in the
instance, on Topcoder, if a developer is found to be
case of 99designs, the payment is either 15 per
cheating the platform initiates an investi-
cent of the anticipated earnings or a payment of
gation to decide on his/her continued
US$2,500. Some delivery platforms also dissuade
access. On microtask platforms,
business clients from using multiple platforms by
accounts can be terminated
specifying in the exclusivity clauses of their con-
if workers’ ratings fall
tracts that commission charges will be lower for
below a certain
clients working exclusively with them.
threshold, if they
are found guilty
Acceptance or rejection of work of using au to -
mated methods,
Platforms often define the situations in which
plagiarizing or
work can be accepted or rejected. On microtask
infringing intel-
platforms, clients only pay for completed work
lectual property
that they have approved, so that workers are not
rights, or failing to
paid if their work does not meet the client’s, or in
reply to attention
some cases the platform’s, standards. Both taxi
questions correctly.
and delivery platforms often provide workers with
the freedom to accept work at their own discre- Location-based platforms
tion. A closer look at the business model of such can terminate accounts, particularly if workers
platforms shows, however, that such freedom breach the relevant terms of service. Other rea-
is unattainable in practice, as non-acceptance sons for deactivation include low ratings, poor
of work and rejection of work have implications performance, prolonged periods of inactivity, and
for worker ratings and future work assignments breaches of codes of conduct, which often include
(see section 4.3.1). anti-discrimination and harassment clauses.
Deactivation Dispute resolution

Platforms reserve the right to put on hold or de- Terms of service agreements usually contain entire
activate worker accounts at their own discretion, sections dedicated to dispute resolution, in which
and in particular when a worker is considered to the governing law and jurisdiction are clearly
have breached the terms of service. Such terms specified. Such sections tend to be lengthier in
often include prohibitions on payments and com- the case of online web-based platforms, given
munications outside the platform, prohibitions on that their dispute resolution procedures usually
the use of subcontractors or automated methods, take the form of arbitration proceedings, the
and prohibitions on having multiple accounts on conditions of which are defined in detail by the
a platform. Deactivation can also occur when platforms. In addition, online web-based plat-
workers have low ratings or have failed to verify forms often include different dispute resolution
their identity or to keep up with a platform’s policies depending on the issue in question.
Some freelance platforms, such as PeoplePerHour

Data collection allows
and Upwork, provide dispute resolution services; platforms to monitor what
these often have a cost and are provided to
workers based in the country where the platforms is happening in real time.
are registered, and therefore do not provide much
support or assurance to workers based elsewhere.
On most microtask platforms, workers have little
to gain in practice by resorting to dispute reso- happening in real time and to improve algorithmic
lution when clients do not pay for work, as the management and automated decision-making
pay per task is often so meagre that the worker for matching and other purposes. This real-time
cannot afford to waste time fighting for such pay. intelligence is a valuable competitive advantage
Moreover, clients are typically not required to give for digital labour platforms with regard to pricing
a reason for non-payment (Berg et al. 2018). On and matching decisions. It also enables them to
taxi and delivery platforms, workers are frequently increase the effectiveness of targeted advertising
subject to the jurisdiction of the courts of the place (Cusumano, Gawer and Yoffie 2019) and to attract
where the services are being provided, although users to the platform. For example, Careem has
there are some exceptions. For example, in the developed an AI platform called Yoda which pre-
cases of Bolt and Glovo the disputes are referred dicts what the demand in a certain place will be in
to specific courts in Estonia and Spain, respectively. two weeks’ time and where drivers will be needed.
Similarly, disputes in the case of Uber are subject to This helps reduce waiting times and secure more
arbitration proceedings in the Netherlands, except fares for drivers.22
for those concerning India and the United States
(see Appendix 2B and Chapter 5 for a discussion The privacy policies of platforms generally stipu-
on dispute resolution mechanisms). late that they use the data collected to communicate
with, notify, support and verify users, to provide
and improve or personalize their services, and to
Data collection and usage ensure security and compliance with legal obliga-
tions. However, some of the platforms analysed,
All of the online web-based and location-based
such as Uber and Deliveroo, specifically mention
platforms under analysis engage in extensive data
that they engage in automated decision-making.
collection. Personal information on users (workers
Uber uses data for automated decision-making
and clients/customers) is collected either directly
to enable dynamic pricing, to match drivers with
or indirectly. Indirect data collection takes place
passengers, to determine ratings and to deacti-
through cookies, web beacons, or embedded
vate users with low ratings, while Deliveroo uses
scripts, or through third parties such as Google
data to confirm payments to riders and to detect
Analytics, social networking services or business
fraudulent transactions. Among online web-based
partners. For example, on taxi platforms, this
platforms, Freelancer and Upwork use data for
covers data related to the worker’s location, which
automated decision-making to match users to jobs
is tracked using GPS, as well as ratings, accelera-
and to determine workers’ rankings. Meanwhile,
tion and braking data, communications between
Topcoder’s privacy policy states that the platform
users and even data stored on users’ personal de-
does not rely on automated decision-making. Data
vices, such as address book information or names
collection strengthens platforms’ screening and
of applications installed.
monitoring powers, which can have significant
Data collection allows online web-based and implications for workers’ access to platforms and
location-based platforms to monitor what is to work.
22 For more details, see: https://blog.careem.com/en/careems-destination-prediction-service/.

Conclusion
This chapter has shown how digital labour plat- A distinct feature of digital labour platforms is
forms have used some of the key features of algorithmic management, which is fundamen-
the digital economy to develop a distinct busi- tally shaping work processes and performance
ness model. In-depth analysis of the business management on the platforms. The algorithmic
model across various online web-based and matching of workers with tasks, clients or cus-
location-based platforms reveals that there are a tomers often factors in characteristics such as
number of common elements among the different ratings, client or customer reviews, cancellation or
types of platforms. A combination of interde- acceptance rates, and skill levels. At the same time,
pendent elements, such as pricing, recruitment, particularly on some online web-based platforms,
matching, work organization and rules of platform some of these factors can be sidestepped through
governance, are shaping the ways in which these the payment of additional fees, thereby creating
platforms compete among themselves, while barriers to accessing work for those workers who
transforming the world of work. may lack adequate financial means to pay such
fees, notably in developing countries.
Some aspects of these elements have implications
for the future of work. By mediating work, plat- In addition, monitoring tools and software that
forms are matching clients and customers with a trace keyboard inputs or capture screenshots at
range of workers who have different skill levels random intervals on many online web-based plat-
and who perform various types of tasks, from forms can curtail workers’ freedom and autonomy.
high-skilled work such as software programming Similarly, on taxi platforms, GPS monitoring, and
to low-skilled work such as delivering food or car- acceptance and cancellation rates can lead to low
rying out microtasks. In doing so, platforms have ratings, which in turn affect access to work and in
developed a revenue model that in some cases some cases can lead to deactivation of the worker’s
places a financial burden on workers, through account. Furthermore, the governance of plat-
the commission fees or subscription plans and forms through terms of service agreements, which
other fees required if they are to access work. are unilaterally determined, enables platforms
These fees can at times be volatile and reduce to exercise considerable control over workers’
workers’ earnings, particularly in a context of freedom to work, and in some instances also
excess labour supply. In other cases, fees may also restricts clients’ or businesses’ ability to engage
be borne by businesses, such as restaurants or with workers, such as through exclusivity clauses.
shops on delivery platforms, which has an impact
A nuanced engagement with such elements of the
on their revenue.
digital labour platform business model underlines
Moreover, the digital labour platform business the fact that it is important to move beyond the
model relies heavily on workers whose work is me- discourse of flexibility, as often publicized by
diated through the platforms and are categorized many platforms. It is critical to further explore
as “self-employed” or “independent contractors”, these issues and to build a deeper understanding
rather than employees. This is one of the funda- of the impact of such a business model on both
mental shifts of this business model and as such traditional businesses and workers. These aspects
has serious implications for the future of work. are discussed in the following chapters.
3
The diffusion
of digital labour
platforms
in the economy
How and why are businesses
using them?
Opportunities Why do businesses use
and challenges online web-based platforms?
for businesses • Recruitment • Innovation
• Cost reduction and efficiency
Why do businesses and consumers use location-based platforms?
Delivery sector
• Improved visibility
• Enhanced productivity
• Increased demand
• Expansion of customer base Taxi sector
• Safety
• Convenience
• Competitive price
Challenges faced by Opportunities for business process

traditional businesses outsourcing companies and start-ups
• Increased competition • Transformation and expansion

of BPO companies
• High commission charges
• Lack of transparency in ratings • Proliferation of AI start-ups
• Lack of digital infrastructure
3. The diffusion of digital labour platforms in the economy 105
Introduction
The spread of information and communications and facilitated algorithmic management of work
technologies (ICTs) in the 1990s led to the de- processes (Moore and Joyce 2020). Platforms have
verticalization of large businesses and allowed thus introduced new work arrangements, often
businesses of varying sizes to relocate their challenging the traditional business models.
services and production processes to different Many of these platforms have clients ranging
regions of the world. This process brought about from start-up companies to some of the Fortune
a change in work organization, as businesses 500 companies and multinational corporations
started working more and more with sub-con- (Wood et al. 2019a; Corporaal and Lehdonvirta
tractors, subsidiaries and business process 2017). In its Global Human Capital Trends report,
outsourcing (BPO) companies (Rani and Furrer, Deloitte (2018) observed that a diverse “workforce
forthcoming). It also spurred the emergence of ecosystem” is gradually replacing the employment
networked organizations, linking outsourcing, relationship. Such an ecosystem includes a diversi-
franchising and temporary agency work, which fied portfolio comprising workers, talent networks,
has led to fragmentation of work and blurring of service providers and gig workers, offering em-
organizational boundaries (Grimshaw et al. 2017). ployers a combination of flexibility, capability and
a different economic model of sourcing talent.
The current wave of technological advances, such
as cloud computing, has opened up a new means This chapter examines the diffusion of digital
of outsourcing work, namely online web-based labour platforms in the different sectors of the
digital labour platforms, which enable businesses economy, exploring how and why businesses
to access workers with a wide range of skills and use online web-based and location-based digital
expertise from around the globe. Platform work labour platforms and how these platforms are
is indeed the latest manifestation of outsourcing challenging and transforming the established
services enabling businesses to adjust their practices of traditional businesses. The analysis is
workforce, in addition to adopting non-standard based on semi-structured interviews conducted
work arrangements (short-term, fixed-term, by the ILO with representatives of different types
temporary and hourly contracts, among others) of businesses (70 enterprises), which include
for core and non-core tasks within an organization information technology (IT) companies, digital
in order to meet its demands (Hyman 2018; ILO technology start-up companies, business clients
2016; Weil 2014). Digital labour platforms create who use delivery and taxi platforms, and BPO
unprecedented possibilities for outsourcing companies that provide digital services (see
services to workers globally, in the case of online Appendix 3). The interviews provide insights into
web-based platforms (Wood et al. 2019a; Santos the businesses’ use of these platforms, and their
and Eisenhardt 2005), and for accessing labour experience in doing so.
available in local markets, in the case of loca-
The chapter is divided into four sections.
tion-based platforms.
Sections 3.1 and 3.2 explore how and why certain
Digital labour platforms are not only fissuring the businesses use digital labour platforms, and the
workplace but are also reorganizing work activ- benefits and challenges involved. Section 3.3
ities; they can therefore be considered as being considers the opportunities presented by digital
new players in the temporary staffing industry platforms, focusing on BPO companies to under-
(van Doorn 2017). While casualization or gig work stand how they adapt to the digital economy. It
is not new, the use of technology to manage a also examines the new digital technology start-up
contingent workforce and offer their services companies that have proliferated in order to
to businesses, customers or individuals is a new understand their motivations and the services
phenomenon. These platforms use search algo- they offer to businesses and digital platforms,
rithms to match workers with businesses, allowing illustrating some insights through case studies.
companies to find talent more rapidly than ever Section 3.4 discusses some of the implications of
before, thereby reducing search costs. In addition, digital platforms for traditional businesses, with a
digital tools have enabled remote collaboration focus on the retail sector.
3.1 Businesses using online web-based platforms
Businesses are finding innovative ways of provided by both freelance and competitive pro-
outsourcing work through alternative work gramming platforms. Freelance platforms, such as
arrangements involving the use of independent Toptal, exclusively specialize in matching workers
contractors, freelancers, gig workers and crowd- from their talent community to businesses; the
workers. New talent networks or digital labour workers can be contracted on an hourly, part-
platforms such as InnoCentive, Toptal, Upwork time or full-time basis. These platforms provide
and 99designs are increasingly being used as a companies with a choice of workers with whom
means of outsourcing work. It is estimated that they can engage before the decision to hire or not
“these types of talent networks now manage over is taken. Similarly, Upwork offers “Payroll” service,
US$2 billion in outsourced activity, employing a premium service offered through third-party
hundreds of millions of people in every geography providers to hire workers under an employment
of the world” (Deloitte 2019, 23). These platforms relationship. It also collaborates with large tech
are considered to be very important for an organ- companies such as Microsoft and offers them
ization’s competitive advantage in the future, these services so that they can have better access
according to a survey of 700 business leaders in to a skilled workforce. Such matching services
the United States (Fuller et al. 2020). This section are also provided by competitive programming
explores the purposes for which businesses are platforms, such as HackerRank, HackerEarth,
using online web-based digital labour platforms. Kaggle and Topcoder. These platforms offer
The literature on this subject is still limited, though “talent as a service” to businesses, and based on
growing, and the analysis is supplemented with their needs recommend workers with specific
interviews conducted for this report with IT, plat- technical skills from their respective communities
form and digital technology start-up companies. of programmers, developers and data scientists
Based on the analysis, three broad purposes can (see Chapter 2).
be identified as to why online web-based platforms Second, competitive programming platforms,
are being used by businesses: for recruitment pur- such as HackerRank, HackerEarth and Kaggle, help
poses; for reducing costs and improving efficiency; companies to organize the recruitment process.
and for accessing knowledge for innovation. As discussed in Chapter 2, these platforms offer
different types of subscription plans or customized
services to businesses for recruitment services in
3.1.1 Recruitment the fields of data science, AI and other techno-
logical domains. The recruitment services provided
Digital transformation has brought about an
include screening and short-listing workers with
unprecedented change in recruitment practices
specific skills and competencies, who can then be
around the globe. Companies are increasingly
interviewed by the companies. This speeds up the
changing their human resource practices (Deloitte
screening process, thereby making recruitment
2017) and using artificial intelligence (AI) and auto-
more efficient for businesses and at the same time
mation to assess and interview candidates. Online
reducing the efforts and costs of hiring. To assess
web-based platforms, such as freelance and com-
workers’ technical skills, these platforms organize
petitive programming platforms, are also gaining
hackathons, competitions and other challenges,
in popularity for recruiting workers in two ways.
which are often algorithmically programmed and
First, online web-based platforms are a growing conducted either online with the participation of
means of hiring workers with specific skills, as they developers from around the world, or in specific
algorithmically match workers to the vacancies locations, such as university campuses. The ser-
and tasks of business entities and offer them vices provided by platforms such as HackerRank
customized services. The matching services are reduce the time lags for businesses in generating
Online web-based
a shortlist of qualified candidates for a job, apart
platforms are a growing
from assisting in removing bias in the selection means of hiring workers
process (Grooms 2017). In addition, such plat-
forms assist businesses to hire talented individuals with specific skills.
who demonstrate advanced design thinking and
capabilities and can provide solutions across a
range of sectors. A number of companies
such as Adobe, Altimetrik, and others A survey conducted by Deloitte in 2019 showed
use these recruitment services that businesses outsourced work for multiple
offered by competitive activities such as IT (33 per cent), operations
programming platforms (25 per cent) and marketing (15 per cent), as well
(Babu 2015). as research and development (R&D) (15 per cent)
(Deloitte 2019). It was also observed that “most or-
The demand for such
ganizations look at alternative work arrangements
services by companies
as a transactional solution, not as a strategically
has been growing over
important source of talent” (Deloitte 2019, 23).
the past decade. For in-
Researchers at the Oxford Internet Institute con-
stance, HackerEarth has
ducted a survey of nine Fortune 500 companies
more than 750 corporate
and asked about their motivation for using digital
customers worldwide
labour platforms compared to traditional staffing
across various sec-
agencies (Corporaal and Lehdonvirta 2017). The
tors of the economy,
findings show that these companies outsource
such as Amazon, L&T
work to workers on digital labour platforms to
Infotech, Wipro and UBS, which use their
address staffing needs related to content mar-
platform for recruitment (Bhalla 2017; Babu 2015).
keting, translation, administrative support and
These platforms thus seem to be altering trad-
customer service, design, IT and data, for the
itional recruitment practices in some companies.
following reasons:
easier and faster access to a specialized, global

X
3.1.2 Cost reduction and flexible labour force;
and efficiency low cost of hiring workers, and reducing over-
X
head costs by some 25 to 30 per cent;
Digital labour platforms provide businesses with
an internet-mediated marketplace. Businesses set quicker outsourcing of work (2–4 days) com-
X
up the tasks and requirements and the platforms pared to traditional employment agencies
match these to a global pool of workers who can (6–8 weeks);
complete the tasks within the specified time. This shorter time schedules as tasks are completed
X
process ostensibly helps businesses to adopt an faster;
extremely agile and lean structure for their core
tasks. In principle, rather than hiring additional greater flexibility achieved by outsourcing short
X
staff or subcontracting through established and small tasks;
firms, organizations can more easily outsource
reduced administrative procedures and con-
X
a diverse range of activities to a geographically
tractual arrangements as the work can be easily
dispersed crowd, in various sectors such as fi-
contracted out; and
nancial services, legal services, patent services,
logistics and healthcare. These platforms are access to highly qualified professionals and
X
increasingly used by large businesses and small expertise, providing new opportunities for
and medium-sized enterprises (SMEs), as well as knowledge creation and delivery of quality
early-stage start-ups. work.
Digital labour
platforms help companies Buxmann and Lambinus 2020; Boudreau,
to reduce costs and Jesuthasan and Creelman 2015). This is also the
case among some big technology companies.
shorten time schedules. For example, “Apple has turned to large numbers
of users and developers distributed around the
world to propel its growth by creating apps and
Another study by Harvard Business School and
podcasts that enhance its products” (Boudreau
Boston Consulting Group surveyed 700 busi-
and Lakhani 2013, 62). Lakhani, Garvin and
nesses in the United States to understand
Lonstein (2012, 8) also show that clients are able to
the usage of digital labour platforms such as
substantially reduce the cost of building their com-
InnoCentive, Freelancer, Toptal and Upwork. The
pany website by using competitive programming
survey revealed that some 30 per cent of the
platforms (for instance US$35,000 was paid) in-
companies used these platforms extensively while
stead of paying “$350,000 to a large IT consulting
for another 30 per cent the usage was moderate.
firm, or $200,000 to a small IT consulting firm,
Accessing workers through these platforms also
or $80,000 to individual contractors”. Similarly,
helped about 40 per cent of the companies to
a software development project that took six
boost productivity and increase innovation (Fuller
months to complete through a platform would
et al. 2020).
have taken twice as long had it been undertaken
The CEO of a major microtask platform men- within the company (Corporaal and Lehdonvirta,
tioned in an ILO interview that large businesses 2017). According to Fuller et al. (2020, 7), com-
were their biggest clients and the source of about panies are moving beyond experimentation and
80 per cent of their revenues. One such client using platforms on an ad hoc basis with the aim
processes 100 million lines of data every year of “developing an integrated strategy … that uses
through the platform, for annotating, classifying … platforms not just to tap the best talent … but
and categorizing to make them machine-readable also to get the most out of the latent capabilities
and train machine-learning algorithms. This work of their full-time employees”.
is integrated through an application programme
As their reliance on digital labour platforms grows,
interface, which allows the clients to outsource
businesses also face challenges in strategically
work directly to crowdworkers on the platform.
managing the workforce engaged under multiple
Apart from training machine algorithms, the data
work arrangements (Deloitte 2018). A majority
also provides insights into consumption patterns
of business respondents (54 per cent) in a 2019
and can be useful to companies when making
survey conducted by Deloitte underscored that
business decisions and for targeted marketing.
“they either managed alternative workers incon-
Studies in the automotive industry show that com-
sistently or had few or no processes for managing
panies use these platforms for data and image
them at all” and that this was largely because they
processing, which support the development of
used these workers to “fill slots” (Deloitte 2019, 23
AI for autonomous and connected vehicles, en-
and 24). Yet, despite the challenges they present
hanced speech interfaces and virtual assistants for
to businesses, about 30 per cent and 17 per cent
drivers, as well as for training algorithms on the
of the business respondents perceive that gig
basis of various traffic scenarios and geographical
workers and crowdworkers improve organizational
mapping without the need for human supervision
performance respectively (Deloitte 2019). During
(Tubaro and Cassili 2019; Schmidt 2019).
the COVID-19 pandemic, with rising demand, plat-
The strategy of using a crowd to be cost-effective forms were offering more value-added services
is quite widespread across a variety of industries, to companies and they are “gearing up to play a
including the automotive, chemical, financial, more significant role in closing the skills gap” in
research and technology industries (Tauchert, the future (Fuller et al. 2020, 8).
Open source and

competitive programming
3.1.3 Access to knowledge
platforms facilitate
for innovation
Digital platforms, such as open source and com-
and provide opportunities
petitive programming platforms, facilitate and for innovation.
provide opportunities for innovation that are
beneficial to both businesses and workers. The
rise of the internet and the rapid expansion of ICTs
have made it easier for businesses to access know- Many leading IT companies collaborate closely
ledge through multiple means. Over the past two with open source platforms on innovation, re-
decades, two strategies in particular have been search and development, seeking solutions to
gaining prominence among businesses in terms of specific problems. Microsoft collaborates with
innovation, ideas and expanding their knowledge the Apache Software Foundation (an open source
boundaries: first, collaboration and co-creation on volunteer community of developers) and makes
open source platforms; and, second, collaboration products and innovations available through such
with competitive programming platforms that platforms. Collaboration and engagement of
organize open competitions or challenges for in- businesses with open source platforms is not ne-
novation and development. This section focuses cessarily cost-related; it enables them to improve
on open source and competitive programming their public relations and gain legitimacy, and to
platforms to explore how they potentially help learn from and align with the latest innovations
businesses’ efforts to innovate and develop. in their field (Lerner and Tirole 2005; see box 3.1).
There are also challenges associated with en-
Open source platforms gaging in open source platforms for businesses,
as they have to make a decision about the extent
Open source platforms are growing in popularity to which they would like to share the intellectual
because their underlying software is not proprie- property (IP) in exchange for the benefits of innov-
tary and can consequently be accessed, modified ation (Henkel,
or even developed by anyone. Large IT, Schöberl and
financial and retail companies use A le x y 2014) .
such platforms for purposes of This is largely
development and innovation, due to ineffective
instead of pursuing in-house IP protection mech-
development or outsourcing to anisms and the threat of
other IT companies (Thakker, imitation from competitors
Schireson and Ngu yen - Huu (Teece 2018b). However, govern-
2017). Most of the digital labour ments and businesses are facilitating
platforms analysed in this report, and encouraging open access to
including Bolt, Ola, Rappi, IP for innovation and development
S w i g g y, To p c o d e r a n d (see Chapter 1).
Upwork, use the two most
common open source
web servers – Apache and
Nginx.1 They also use
open source tools and
sof t ware to develop
their technologies.
1 This information is based on an analysis of platform websites using a website profiling tool (Builtwith).
Box 3.1 Apache Software Foundation

X
The Apache Software Foundation (ASF) is an open source volunteer community of developers
that was set up in 1999. It has over 350 open source projects such as Hadoop, Spark, Cassandra,
CloudStack and Flink. A high proportion of websites on the internet today and most of the digital
labour platforms discussed in this report are powered by the Apache HTTP Web Server, which
led to the ASF’s formation in 1999. The software developed through open source projects in
the Foundation is distributed under the Apache licence and is a free and open source software,
which can be further developed and innovated by software programmers or coders. It is a
business-friendly licence and allows entrepreneurs to leverage and create all types of businesses
around it.
Businesses can post questions and computational problems and access services from the online
community of experts willing to provide solutions at zero cost. A large number of volunteers
(developers and programmers) put in time and effort to work at the ASF, while others are paid
by their employers to contribute. They find the experience rewarding as they are able to acquire
new skills working with their peers in the community, and they can establish relationships with
experts in the domain with whom they can interact in the future. In addition, “programming in
these communities requires a high degree of motivation, as programmers and developers have
to invest a lot of time before they can see concrete results and most programmers are interested
in the art of creating it [the code] rather than the money” (ILO interview with a representative
of the ASF).
Many major technological companies regularly send their in-house programmers and devel-
opers to the ASF to work on complex problems with the community. This helps workers not only
to acquire complex programming skills in a short time and at almost zero cost, but also to come
up with innovative ideas for their business activities. It also provides them with an opportunity
to enhance their reputation, achieve recognition in their company and develop their career.
About half the developers on Apache projects are paid by technological companies such as
Facebook, Google, IBM and Microsoft. Some of these companies also have dedicated open
source departments responsible for building their relationship with the Foundation.
Apache does not charge anyone for using the platform and it costs the ASF approximately
US$5,000 to provide infrastructure support for each project. With more than 350 projects
and initiatives, the infrastructure support alone costs them US$1.75 million. The Foundation
is funded through sponsorship or the donation of funds from some of the big technology com-
panies to support infrastructure services and conferences. The companies benefit from being
associated with the ASF brand as this makes it easier for them to attract customers, and they
gain enormously from the knowledge and expertise of the open source communities. Some
companies also share their software as open source under the Apache Foundation umbrella,
which allows for the development of the software through the community and is beneficial to all
users in the ecosystem. For instance, Cassandra, an SQL (Structured Query Language) database,
was originally donated by Facebook to the Foundation and attracted developers from other
companies such as DataStax, Google and Microsoft, who further developed it by contributing
additional features or enhancing its services.
Sources: ILO interview, 2019; Apache Software Foundation website and documentary feature.
Innovating by using competitive

Box 3.2 Using the Topcoder
programming platforms X
community for technological
Businesses also access knowledge for innovation solutions
and build new capabilities by using competitive
Topcoder, a competitive programming
programming platforms that provide solutions
platform, offers companies access to
by organizing challenges or competitions2 (see
talented digital workers from around
box 3.2). These platforms provide businesses
the globe who can provide a range of
with access to their community of programmers,
potential solutions to their projects at
developers and coders in various technology do-
a low cost and in a short period. For
mains. There is growing reliance on these digital
instance, the Topcoder community was
labour platforms for innovation because they
invited to help an ambitious crowd-
attract a community of programmers to solve a
sourcing healthcare initiative focused on
wide variety of problems related to AI, machine
cancer, for a prize of US$55,000 over ten
learning, data science, security and so on, based
weeks. The initiative focused on tumour
on the innovation needs of businesses, in return
delineation in lung cancer, which claims
for prize money (as stated by representatives
over 150,000 lives annually in the United
of companies in ILO interviews). For example,
States alone, and the “challenge” was to
Netflix’s filtering algorithm for predicting user
produce an AI solution to treat a critical
matchings for films was initially based on a user
lung tumour. Topcoder joined forces with
rating on a scale from 1 to 5; to improve the accu-
Harvard Medical School and the Dana-
racy of the rating predictions, an open competition
Farber Cancer Institute to create and
was organized (Gomez-Uribe and Hunt 2015).
test automatic delineation algorithms
Communities of programmers, coders and de- to help improve treatments of cancerous
velopers on competitive programming platforms tumours in patients’ lungs.
potentially help companies access creative ideas
Over the ten-week, three-phase crowd
and diverse solutions in a way that was previously
innovation challenge, 564 contestants
not possible (Lakhani, Garvin and Lonstein 2012;
from 62 countries registered and 34
Terwiesch and Xu 2008). ILO company interviews
contestants submitted 45 algorithms,
revealed that the real value of these platforms lies
resulting in multiple AI solutions capable
in the quality and speed with which solutions are
of targeting lung tumours with an accu-
provided, which would be difficult to achieve solely
racy equal to that of an expert radiation
with internal resources. This process also provides
oncologist, but more rapidly.
flexibility and easy access to highly skilled talent
around the globe. Competitive programming Sources: ILO interview, 2019; see also https://
www.topcoder.com/case-studies/harvard-
platforms such as Topcoder have leveraged the tumor-hunt/; https://jamanetwork.com/
crowdsourcing model to find solutions to some journals/jamaoncology/fullarticle/2730638.
of the world’s most complex and sophisticated
problems (improvements in cancer treatment,
faster DNA sequencing and improved energy
pipeline security, among others) by pairing their
community of IT talent with businesses (see also
Lakhani et al. 2013; see box 3.2).
2 The idea of introducing competition for innovations or solutions is not new and can be traced back to “the Longitude competition
in 1714, when the British government announced an open call (with monetary prizes), for developing a method to measure a ship’s
longitude precisely” (Mao et al. 2017, 59). While internet-based innovation competition can be traced to 2001 with the InnoCentive
platform, which tried to attract a crowd for drug development, a number of other platforms later emerged for software develop-
ment and data analytics (Mao et al. 2017).
Box 3.3 Wipro’s new strategy to develop human resource capabilities and innovate
X
using digital labour platforms
Wipro Limited (hereafter Wipro), founded in 1982, is today one of the leading Indian companies
providing high-quality IT-enabled services globally. Since the early 2000s it has been offering
a range of services to clients, including data analytics, AI and cloud computing. The shift from
traditional IT services to integrated services in specific industries meant that Wipro had to build
and/or acquire a completely new skill set, especially in business strategy and design skills. To
this end, Wipro introduced a strategy based on four key components:
(i) Aligning business strategy with talent strategy

Wipro radically shifted its approach to recruitment. Instead of hiring workers with “I-shaped”
profiles (involving in-depth knowledge and expertise in a particular technology, such as Java)
or “T-shaped” or “pie-shaped” profiles (in-depth knowledge and expertise that can be applied
to different industries), Wipro hires workers with “X-shaped” profiles (software and design
expertise, along with detailed knowledge of business strategy and implementation). Wipro
managers also rotate workers every two years among different industry lines to increase their
exposure to various industries, as well as to transfer knowledge among industry clients while
continuously learning new skills.
(ii) Leveraging in-depth technology expertise to yield innovative client solutions

Wipro explored a variety of crowdsourcing initiatives to meet this strategic goal. In 2016 it
acquired Topcoder, a platform marketplace bringing together 1.5 million developers, designers
and data scientists. In 2017, with the help of Topcoder, Wipro developed an internal crowd-
sourcing platform – TopGear – to bridge the technology skills gap and create a project-deployable
workforce. This demonstrates how structured, internal crowdsourcing efforts can increase
individual and organizational adaptability. The platform acts as an opportunity for both teams
and individuals to:
support employees in learning and applying skills to a range of projects;

X
encourage flexibility and value-driven outcomes by giving individuals more agency;

X
provide a channel for employees to reap benefits that go beyond work;

X
develop their design, coding, testing and data science tools and expertise by crowd-
X
sourcing tasks or projects to internal talent;
provide multiple innovative solutions to their clients by posting complex problems on the
X
platform as a “challenge” for prize money; and
access platform workers for specific projects for a short time span, enabling flexible
X
resourcing.
The TopGear team supported an internal project team in a large-scale workforce transform-
ation that involved everyone from manual testers to automation engineers. Their development
and implementation of a learning plan upskilled 80 per cent of the account team, resulting in
a 20 per cent increase in annual productivity for the department concerned. Building on the
success of TopGear, Wipro launched the new Hybrid Crowd Platform, aiming to make func-
tional enhancements to create a flexible workforce for the future and to revolutionize talent
resourcing internally and for its enterprise clients.
Hybrid Crowd provides a way for all businesses (in addition to Wipro itself) to connect their
internal talent teams with the more than 1.5 million members of Topcoder’s global community.
Box 3.3 (cont’d)

X
Integrating these talent pools, the platform enables enterprises to supplement their teams,
on demand, with experts from the crowd. Through Hybrid Crowd, organizations can engage
three different types of crowds: private, certified and public. According to K.R. Sanjiv, Chief
Technological Officer of Wipro Limited:
Hybrid Crowd platform is the cornerstone of Wipro’s ongoing digital transformation and
it enables the team to provide an even broader spectrum of digital services and meet just-
in-time requirements. It also gives our digital transformation experts [Wipro employees]
increased opportunities to learn new skills, earn, and gain recognition by competing in
crowdsourcing competitions.
(iii) Encouraging collaboration and innovation

Wipro organizes internal and external hackathons and ideathons on company premises and on
the Topcoder platform to develop skills and expertise among Wipro’s employees, and to find
innovative solutions. Employees can compete either individually or in teams for the challenges
posted by clients. Multiple winners are awarded prize money and their achievements are widely
publicized within the company. The contest model allows employees to evaluate their skills
against their peers, but the open, informal nature of the gamified training encourages com-
munication and support. Senior managers claim that this strategy boosts commitment among
employees and has a positive impact on their performance and productivity.
(iv) Aligning and collaborating with key stakeholders in the platform ecosystem

Wipro also invests in an ecosystem of start-ups and in collaboration with Microsoft accelerators
taps into innovations. It establishes long-term partnerships with clients to identify solutions in
emerging technologies, such as blockchain or AI, in their respective industries, and collaborates
with open source software communities such as GitHub, SourceForge and others.
Source: ILO interviews, 2019 and 2020.
Traditional IT outsourcing firms are facing Google acquired Kaggle, a data-science platform,
increased competition from competitive pro- in 2017, enabling it to use Kaggle’s community
gramming platforms, with companies turning of data scientists to analyse data at the speed
to platform communities to solve their problems required to be competitive in the AI space.3 These
and outsource their work. To overcome this chal- developments raise critical questions about future
lenge, traditional companies are starting to build career opportunities for highly skilled IT workers
or buy emerging or well-established platforms if companies of such calibre are increasingly using
that can provide the skills and technology that and relying on digital labour platforms, a trend
they lack (Cusumano, Gawer and Yoffie 2019). For observed during the COVID-19 pandemic (see
instance, the IT outsourcing firm Wipro acquired Chapter 1). There is also a broader question with
the Topcoder platform in 2016, and with it the skills regard to building capabilities within companies
and expertise to provide technical services in a and whether this practice is sustainable in the long
range of sectors, resulting in a change in Wipro’s term if firms are increasingly going to leverage
strategy and delivery model (see box 3.3). Similarly, expertise through crowdsourcing.
3 See https://techcrunch.com/2017/03/08/google-confirms-its-acquisition-of-data-science-community-kaggle/.
3.2 Businesses using location-based platforms
Location-based platforms, such as taxi and delivery The motive for using apps is to sell products,
platforms, have created easy access to services for to increase the customer base as well as to
individual consumers (see box 3.4) and are being increase demand – Restaurant using app-
increasingly used in many countries by SMEs, based delivery services (Morocco)
restaurants and individual entrepreneurs. The
When it is raining, the demand for delivery
growing reliance on such platforms stems from
increases because people do not want to
competition, the need to expand the customer
come to the restaurant and they prefer to
base and to cope with a transforming marketplace
order through the delivery apps. This is true
as well as consumer preferences. Some traditional
even during weekends – Restaurant using app-
businesses that have started using delivery plat-
based delivery services (Kenya)
forms include those in the restaurant and retail
sectors. This section examines the opportunities Many restaurants use multiple platforms to
and challenges that restaurants and small busi- provide services to customers, for three main
nesses encounter with location-based platforms. reasons. First, each platform has its own customer
The analysis and conclusions in this section are database, which allows restaurants to reach more
based on semi-structured interviews conducted customers. Second, having a presence on multiple
by the ILO with representatives of 47 businesses platforms helps retain customers who often
and their clients in selected developing countries switch across apps to get the best deals. Third,
(Ghana, Indonesia, Kenya, Lebanon, Morocco and doing so helps smaller restaurants to compete
Ukraine) between October 2019 and March 2020 with bigger restaurants or chains and to benefit
(see Appendix 3, table A3.1 for details). from various kinds of promotions and advertise-
ments offered by the platforms.
The motivation behind joining multiple plat-

forms is to get more visibility, so that we do
The growing reliance not lose customers to coffee chains who have
on location-based platforms a presence on these apps – Restaurant using

app-based delivery services (Lebanon)
stems from competition. By using multiple platforms we are able to
target as many people as possible because
each plat form has its own customer
Restaurants base – Restaurant using app-based delivery
services (Kenya)
The restaurant business in particular has wit-
The constant advertising of our food
nessed enhanced consumer demand for deliveries
items through platforms leads to high
through platforms that customers often consider
demand – Restaurant using app-based delivery
easy and convenient to use. ILO interviews
services (Morocco)
with 27 restaurant owners in six countries (see
Appendix 3, table A3.1) reveal that their markets Delivery platforms are also helping restaurants
have expanded and that they are able to reach new to improve their productivity through multiple
customers thanks to increased visibility through means. First, platform companies offer restaur-
app-based delivery platforms. In addition, the ants web analytics and monitoring tools that help
ease of ordering through the platforms has led to them track their customers’ preferences; this in
increased demand from offices during the week turn allows for greater insight into the best ways
and households during weekends or during poor to develop their business strategies and pricing
weather conditions. structures. Second, platforms provide periodic
reviews and training on digital integration, as well While app-based platforms have prompted a rise
as advice on business strategy and advertising. in demand for food from restaurants, they have
Third, digital tools for tracking orders, preparing also led to greater reliance on the digital economy,
products for dispatch and managing accounts which in turn requires a well-functioning digital
and payments also help increase restaurants’ infrastructure. Many restaurants reported that
productivity. Finally, the rating systems on unstable internet connectivity had an impact on
platforms create an incentive for restaurants to their business, particularly relating to dispatching
increase the speed of delivery and improve their orders. Restaurants also faced challenges with
packaging, which not only enhances their com- regard to delays on the part of delivery workers,
petitiveness but also improves their ranking and leading to cancellations, and some complained of
gives them greater visibility among customers. poor service on the part of platform companies,
which provoked complaints from customers.
We are making sure that all our staff are
In addition, platforms charge a commission fee
aware and are trained and have the know-
of about 15 to 25 per cent, which affects the
ledge on how to do the packaging at a
restaurants’ profit margins and sustainability.
rapid speed, so that the orders are ready
Some restaurants also mentioned that they were
for the delivery worker to pick up. The
penalized with high commission fees if they used
quality and the quantity are important for
multiple platforms.
rating – Restaurant using app-based delivery
services (Kenya) The current deal with Toters is 25 per cent
commission for each order. We think this is
The platform company provides recommen-
extremely high. We recently tried to nego-
dations through email about how to increase
tiate a better deal with Toters but did not
profitability – Restaurant using app-based
succeed, so we decided to work with other
delivery services (Ukraine)
companies until we have developed our own
Several restaurants reported that the increase in application as an exit strategy – Restaurant
demand for deliveries led to a greater workload, using app-based delivery services (Lebanon)
especially for kitchen staff. Some restaurants
Poor internet infrastructure is the main
hired temporary or short-term workers to meet
issue in Lebanon, which often leads to inter-
the increasing demand.
ruptions – Restaurant using
We have a list of temporary workers, we call app - based deliver y
them from time to time, especially on week- services (Lebanon)
ends. We pay them according to the hours
worked – Restaurant using app-based delivery
services (Morocco)
We used to have permanent employees
who delivered products, but now we use
the platforms and have fired those
employees – Restaurant using app-
based delivery services (Ukraine)
Small retail businesses Small retail businesses using e-commerce plat-

forms are also able to access data about their
The ILO interviews with representatives of 16 orders and sales using web analytics, which allows
small retail businesses and independent sellers them to respond to changing demand. To deliver
in Ghana, Indonesia and Kenya show that the goods or products to customers, small retail
small businesses are increasingly using social businesses often use multiple delivery platforms,
media platforms such as Twitter, Facebook and take advantage of various promotions and offers,
Instagram, as well as e-commerce platforms, to and have preferences for platforms based on the
sell their products. Some of them have no physical services offered.
stores and operate exclusively from home. All the
We are able to compare how many orders
small retail businesses surveyed are increasingly
were delivered, which helps to analyse and
relying on delivery platforms to deliver goods
monitor the sales – Small retail business on an
to customers who place orders online, and they
e-commerce platform using app-based delivery
are able to sell products from anywhere and to a
services (Kenya)
wider customer base, resulting in higher incomes.
Delivery services have also enabled independent However, small retail businesses and independent
sellers to focus on the production and manage- sellers also face challenges similar to those of
ment of goods rather than on deliveries. restaurants with regard to unstable internet
access, delays in receiving orders from delivery
The delivery platforms have helped us to
workers, and the capacity to adapt to an online
increase orders compared to the previous
business model. Businesses that use e-commerce
months, which has led to an increase in rev-
platforms also face the challenge of changes in
enue and profits – Small retail business on an
commission fees without any notice, which affects
e-commerce platform using app-based delivery
their revenue. For independent sellers, delays by
services (Kenya)
delivery workers also have serious implications for
Delivery platforms help to deliver the customer relationships as they are often reliant on
products on time and reduce the delivery a smaller customer base.
workload, which is a relief. So, I monitor my
The delivery worker is delayed by two hours,
social media feeds regularly and as soon as
then the customer is frustrated and gets
I get an order, I call the delivery guys and
angry with me – Independent seller using app-
inform them about the pick-up and delivery
based delivery services (Ghana)
time – Independent seller using app-based de-
livery services (Ghana)
Box 3.4 Customers’ motivation for using app-based taxi and delivery services
X
The rise of location-based platforms provides individuals with more choices to access services than
are available through traditional means. To understand the motivations of customers using these
platforms, the ILO conducted in-person interviews with a small sample of 33 customers in Chile,
Ghana, Kenya and India between October 2019 and March 2020. The option of accessing taxi services
or ordering a product through a digital app or at the click of a button has added to the popularity of
these platforms among customers. For most of them, convenience, ease of use, low prices, trans-
parency and reliability were some of the reasons for using app-based services.
The main motivation for the customers to use app-based taxi services was the lower price compared
to traditional taxi services, as well as various offers and discounts. They also pointed out that they feel
safe as GPS enables them to track the driver and to share their location with family and friends. In
addition, in some countries, app-based taxi services are available in certain locations where it is often
difficult to find traditional taxis. Customers of delivery services emphasized that delivery apps provide
them with a variety of products to choose from and help them save transportation costs and time.
Convenience, comfort, privacy, security, flexibility and also knowing that there will be no
need for any negotiation – Customer of app-based taxi services (Ghana)
I don’t have to wait on the road or street and can book the taxi anywhere. I can buy all the
products in one click instead of going to the shop and it saves my time – Customer of app-based
taxi and delivery services (India)
Many customers also use multiple platforms simultaneously for taxi and delivery services as this
enables them to choose the cheapest and most convenient option. For instance, on taxi platforms,
customers can compare offers across multiple apps to find the best deal in terms of fare, driver rating
and location. Delivery platforms enable customers to compare the price of a product across different
apps and to choose the product offered at the most favourable terms and the shortest delivery time.
All customers emphasized the importance of ratings, as this feature allows them to provide feedback
about the quality of services and to see other customers’ opinions about products and services.
I can say that sometimes it helps you get a better price for delivery to the same location
because you can check both apps and get to know which one is cheaper – Customer of app-
based taxi services (Ghana)
Ratings provide a better perspective based on others’ experiences and help me assess safety
issues – Customer of app-based taxi services (India)
Alongside the benefits of app-based taxi and delivery services, some challenges were also identified
by customers, the main ones being internet connectivity and technical glitches on platforms. Other
concerns in the taxi sector included the rise in app-based taxi fares, instances of disagreement with
app-based taxi drivers, cancellations or rudeness on the part of the drivers, lack of transparency of
waiting charges, and surge pricing. In the delivery sector, the challenges included mix-up of food
and other items, delays in orders, cancellations and instances of extra charges, as well as excessive
advertisements on the platforms and a platform design that promotes more consumption.
The delivery apps make the interface more interesting and more appealing so that even if you
are not willing to buy anything, by just clicking and swiping, you may end up buying some-
thing that you didn’t even think you needed – Customer of app-based delivery services (Ghana)
Prices, especially for app-based taxi services, have increased a lot – Customer of app-based
taxi services (India)
Some customers said that taxi and delivery platforms provide job opportunities, especially for
migrant workers, and raised concerns over working conditions and insurance for app-based taxi
drivers and delivery workers.
Source: ILO interviews.
Corporate clients (taxi services) App-based taxi platforms compete with one an-
other to attract corporate clients. For instance,
The ILO interviews with four corporate clients Maramoja specifically targets corporate clients,
in Kenya reveal that they tend to use app-based offering them far lower prices than other platform
services as they are considered to be convenient, companies, while Bolt proposes services at a rate
readily available and reliable. Safety features avail- similar to that charged to individual customers.
able in taxi apps (for example an SOS button and However, corporate clients also face challenges
driver tracking) and the ease and convenience of relating mainly to customer service, non-trans-
making payments via a digital tool were among parency in cancellation charges and poor internet
other important factors reported by the clients connection. The period when the interviews were
for preferring app-based taxi services. conducted was also marked by strikes called by
app-based taxi drivers. This was reported as one
We normally sensitize our staff to use spe-
of the key challenges by the clients in Kenya, as
cific taxi platforms when they meet with a
their business was affected by the temporary halt
client, as it is more reliable – Corporate client
of platform-facilitated taxi services.
of app-based taxi services (Kenya)
When app-based taxi drivers were on strike,
You can track the driver, wherever you are
there were no services available. This was
and one can feel safe – Corporate client of app-
quite different when compared to con-
based taxi services (Kenya)
tracting with a taxi company – Corporate client
of app-based taxi services (Kenya)
3.3 Opportunities from digital platforms for BPO

companies and digital technology start-ups
Digital platforms create oppor tunities for
innovation and entrepreneurship for start-up com-
3.3.1 Transformations
panies, BPO companies, software developers and in BPO companies
programmers, among others. Low IT infrastruc-
ture costs and access to open source platforms The rapid advances in, and adoption of ICTs
have reduced the costs of setting up a business, since the 1990s have led to the outsourcing or
and provide an opportunity to experiment with relocation of services to developing countries,
innovative ideas. This section focuses on two creating new markets and employment op-
trends: the transformations in BPO companies portunities for IT-enabled services, call centres
in response to the needs of organizations in the and for BPO companies (Rani and Furrer, forth-
digital era; and the growth of digital technology coming; Parthasarathy 2010). This has helped
start-ups that provide new technological products large companies to reduce their operating
and AI-enabled services. costs – by accessing labour pools for software and
R&D services and for customer support centres
at a relatively low cost – and to enhance their
productivity (Graf and Mudambi 2005). Developing
countries, such as Brazil, India and the Philippines,
have integrated ICT development into their na-
tional development policies, which has allowed
them to dominate the BPO market (Parayil 2005).
Over the past decade, some African countries,
BPO companies are
including Ghana, Kenya and South Africa, have

also become a viable location for BPO companies
adopting new strategies
due to their cost competitiveness (Anwar and in order to adapt to and
Graham 2019).
The rise of the digital economy and the prolif-
sustain their businesses
eration of digital platforms are leading some in the digital economy.
traditional BPO companies to adopt new strategies
in order to adapt to and sustain their businesses
in the digital economy and to provide the services
needed by large companies. Based on semi-
structured in-depth interviews with managers or
The interviews with Kenyan companies show
representatives of 11 companies4 in two countries
that digital tools and technologies have enabled
(India and Kenya) between April 2019 and January
them to provide what they consider to be im-
2020, this section focuses on these strategies.
proved, on-demand customer-friendly services,
All the BPO companies that participated in the as well as technical support and management
ILO survey are SMEs, which have adopted various of social media. BPO companies such as HN, IN
strategies to adapt to the changing needs of and CCI provide clients in the insurance, banking,
their clients. The BPO companies in Kenya are telecom and retail sectors in the domestic and
largely reliant on work outsourced from large international markets with a range of services,
international companies. Since 2014, they have including market research, customer care,
been transitioning from voice-based services to tracking of consumer preferences, digital mar-
digital services. The nature of the tasks however, keting, pricing strategies and communications
such as handling clients and customer complaints, strategies, using various digital channels. These
has remained the same. The digital channels that services help their client companies improve
the BPO companies have started using to pro- customer experience and operational efficiency,
vide these services include various social media so that they can remain competitive in the digital
channels, email and AI bots.5 In addition, digital business environment.
tools such as web analytics have allowed them
With the use of AI across a range of sectors, from
to track the entire journey of the customer from
automobiles to social media and e-commerce, data
start to endpoint, enabling them to engage with
labelling and content moderation have become a
customers, cater to their specific needs and pro-
key requirement for many companies. A number
vide the required services.
of “big tech” companies, such as Facebook, Google
Anwar and Graham (2019, 214) made a similar ob- and Microsoft, have also started outsourcing
servation in their survey of seven BPO companies content review and moderation, data annotation,
in Johannesburg, South Africa, where they found image tagging, object labelling and other tasks to
that most of the BPO companies were making a BPO companies. The company interviews revealed
digital transition by providing customer services that these tasks are being outsourced by the “big
through “multiple digital channels such as voice tech” companies as part of their corporate social
calls, automated interactive voice responses, responsibility programmes. The objective of out-
webchat and WhatsApp”. In one of the com- sourcing is to have a social impact in developing
panies they surveyed, the number of voice calls countries by providing employment opportun-
decreased by more than 50 per cent between 2012 ities to young graduates or school-leavers, and to
and 2016, with voice calls being replaced by non- support people from disadvantaged backgrounds.
voice digital channels. This strategy has also led to the growth of new
4 The names of the BPO companies cited have been changed to preserve their anonymity.
5 A bot is a simple automated tool or a computer program that can complete an action using artificial intelligence or natural
language processing.
BPO companies and call centres, which perform employment opportunities by hiring women
these tasks in a number of developing countries, and young people from poor households with
including India and Kenya. Some of the data la- basic computer and numerical skills and English
belling companies, such as Infolks and iMerit in literacy, they have also destabilized many small
India, which operate in smaller towns, do so to BPO companies, which now face a reduction in
create employment opportunities among under- outsourced work.
privileged communities, while creating annotation
Some companies, such as AT, which rely on large
tools (Murali 2019).
companies for outsourcing tasks, have struggled
Some new BPO companies, such as FS and CO, to operate in the market due to this decline in
India, stated in the ILO interviews that content outsourced work. To sustain its business, in
moderation not only provides a business op- addition to working directly with its clients, AT
portunity but also allows them to perform a very has established good relations with an online
important task for society as they “act as a firewall web-based platform, which outsources work
or gatekeeper or a watchdog for the internet”. to them. This strategy by small BPO companies
Both these companies also provide services such of accessing work through online web-based
as flagging counterfeit products and fraudulent platforms such as eLance, oDesk (now Upwork)
practices in advertisements and product reviews; and Guru was also observed during the period
safeguarding copyright material and ensuring 2010–14 (Foster et al. 2018). However, they found
that there is no copyright violation on e-commerce that small companies were struggling to survive
websites; and identifying fraud profiles and scam- on the basis of such work alone as the tasks were
mers on dating websites, among others, for large of short duration and low value, and such com-
international and local companies. Interviews with panies had to turn towards domestic markets to
workers and the CEO of FS revealed that about sustain their business.
90 per cent of the workers who perform content
Labour cost competitiveness has also led to the
moderation and other tasks are graduates or
emergence of new types of companies, such as
postgraduates with engineering and computer
CF, which has set up its delivery centres in India
science skills. Some of the companies offering IT-
and Nepal and uses both local and crowdsourced
enabled services, such as Accenture, Genpact and
labour through its platform to provide services to
Cognizant, have also diversified and entered into
large companies in Europe and the United States.
the content moderation business, hiring univer-
The main service provided by the CF delivery
sity graduates to perform these tasks (Mendonca
centres relates to image
and Christopher 2018).
annotation and data label -
Due to labour cost competitiveness in Kenya, ling of still video shots of
many large companies have also estab - road signs, traffic lights
lished their own subsidiaries. For and pedestrians, to train
instance, SS, an international autonomous vehicles to
company that is one of the recognize these objects
largest outsourcers of and navigate real-life
varied tasks (such as si t u a t i o n s w i t h l i t t l e
data entry, annotation human super vision. They
and transcription) also provide services such as
to small BPO firms transcription, categorization,
and crowdworker s tagging and content moder-
in Kenya, set up its ation. The company uses a
own delivery centre hybrid workforce of online
in Nairobi. While workers and locals, which
such subsidiary com - enables it to train the local
panies create local workforce in these tasks, and
Advances in AI
repetition of tasks allows them to ensure quality,
and natural language
precision and efficiency, while at the same time processing have made
maintaining competitiveness in the market.
Tasks such as data labelling and content mod-
it possible for start-up
eration have not had much traction among companies to advertise
traditional BPO companies in Kenya. Some of
these firms, including HN, IN and CCI, stopped and sell their services to
performing them after a year or so as they con-
sidered them to be low-end and low-value tasks. businesses as AI-enabled.
Furthermore, this work did not offer any oppor-
tunity for upward mobility in terms of either skills
upgrading or learning for the company and at the
same time profit margins were low and difficult to applications and AI-enabled services. The growth
sustain in the long run. of these start-ups has been driven by three factors:
Ease of entry, with low investment in physical
X
3.3.2 Emergence of digital assets compared to traditional start-ups and

availability of IT infrastructure at a low cost.
technology start-ups The availability of open source platforms and
software allows for experimentation with new
The digital economy and the expansion of digital
ideas and innovations to improve efficiency or
platforms have led to the emergence of new
productivity.
players: digital technology start-ups that provide
new tools, products and services that enhance Advances in AI and natural language process-
X
efficiency and functioning of the digital eco- ing, which have made it possible for start-up
system. Moreover, the heightened expectation companies to advertise and sell their services
around automating specific tasks (Nedelkoska to businesses as AI-enabled, with the resultant
and Quintini 2018; Frey and Osborne 2017; lowering of costs through the replacement of
Arntz, Gregory and Zierahn 2016) has created workers with AI.
new demand and opportunities for AI-enabled
Availability of venture capital and accelerator
X
services. In 2020, the global start-up economy
funds to start-ups, which has played a crucial
generated US$3 trillion in value and provided
role by providing opportunities to entrepre-
many entrepreneurial opportunities; although
neurs in developed and developing countries
only 14 per cent of the start-up founders were
alike.
female (Startup Genome 2020).
This section examines the motivations behind
Creation of products
the rise of digital technology start-ups and how
the products or services they provide benefit and services
companies, including digital labour platforms.
Most start-up companies try to find niche areas
The analysis is based on semi-structured inter-
where they can provide innovative services to plat-
views conducted with ten digital technology
forms or traditional companies that improve their
start-ups based in San Francisco (United States),
productivity. Developments in AI and, specifically,
Bengaluru (India), Cherkasy (Ukraine) and Warsaw
advances in data analytics and tracking tech-
(Poland) between July 2019 and March 2020
niques, have had major implications for pricing
(see Appendix 3).
and marketing strategies, customer service man-
Two types of digital technology start-ups can be agement and risk assessment; hence the growth
distinguished, based on the responses to the of start-ups that provide products and services,
ILO interviews: those that create technological including web analytics and tracking, to traditional
products and services, and those that provide AI companies as well as to digital labour platforms.
Companies such as Crazyegg and Rytangle pro-

vide digital platforms or traditional companies Box 3.5 Proliferation of AI start-ups
X
with real-time data about the users accessing their
platforms. Most digital platforms and traditional Venture capitalists and other investors
companies today have web analytics and tracking have been interested in investing in the
tools installed to track customer behaviour, which automation of wide-ranging tasks, from
helps target their customers and improve their secretarial to legal services, causing
pricing and marketing strategies. many start-ups to market themselves
as AI companies in order to access such
Companies like Cloudinary offer advanced soft- funding (Schmidt 2017; ILO interviews
ware application solutions for digital platforms with AI start-ups). For example, Scale AI,
or traditional companies that allow for image and Playment and Mighty AI explicitly market
video processing, management of image and themselves as AI companies seeking to
storage facility. Start-ups such as NoticeBoard appeal to the automotive industry in
have developed communication applications that preparing for and designing the next
help to improve communications between large generation of driverless cars (Schmidt
fleets of ground staff or delivery workers – 1,000 2019).
or more workers requiring supervision and
management – and their managers. E-commerce Similarly, there has been a proliferation
platforms and trucking companies in various re- of start-ups that provide companies
gions have been using these applications to track with virtual assistant services, such as
and manage their workers. A number of other appointment scheduling, note taking at
start-up companies provide customized software meetings, or AI-managed mail. Some of
applications to traditional companies and digital the leading start-up companies in terms
labour platforms according to their requirements, of venture capital investment in these
often using open source tools and applications. areas include x.ai (US$44.3 million) and
Clara Labs (US$11.4 million) (information
based on Crunchbase database).
Provision of AI applications
Legal services, considered to be one of
The past decade has witnessed the growth of AI the largest markets in the world, have
start-ups due to the availability of vast financial also seen a phenomenal rise in start-ups
resources from governments, the private sector (Toews 2019). Most legal start-ups,
and venture capitalists (Nitzberg, Seppälä and such as LawGeex, Klarity, Clearlaw
Zysman 2019). These start-ups offer a range of AI or LexCheck, market themselves as
applications to companies, either fully automated providing automated AI legal services,
or human-powered. Most such start-ups have including contract drafting, review
two profiles: one for clients, with a website and a and negotiation, thereby reducing the
company name, exclusively focused on providing tedium of certain aspects of legal work.
services using AI; and another for crowdworkers, Furthermore, they emphasize that AI can
with a website and company name offering work automatically absorb written documents,
opportunities and the chance to earn an income “analyse them in full using natural lan-
(Schmidt 2019; ILO interviews with AI start-ups). guage processing (NLP) technology, and
Many of these companies have emerged in fields determine which portions of the
such as virtual assistance (secretarial tasks), legal contract are acceptable and
services, microtasks (image and data annotation) which are problematic”
and others which use crowdworkers to provide (Toews 2019).
the services (see box 3.5). An investment review
of 2,800 AI start-ups across Europe in 2019 found
that about 40 per cent of them did not have any-
thing to do with AI (Ram 2019).
The AI start-ups interviewed by the ILO are

Start-ups offer
human-powered. However, AI start-ups often a range of AI applications
do not mention to their clients that their tasks
are completed by a globally dispersed human to companies,
workforce through digital labour platforms.
Tubaro, Casilli and Coville (2020, 7) argue that
either fully automated
the reason why most AI start-ups are not auto- or human-powered.
mating these tasks is that “machine learning is
expensive, as it requires powerful hardware, the
brainpower of highly qualified computer scien-
tists, and top-quality data”, while it is easier and
cheaper to “fragment the work into microtasks to entirely novel situations is still an enormous
and sub-contract them to low-paid workers challenge for AI and robotics” and AI applications
through platforms”. Most AI start-ups differen- are in their infancy in a number of sectors (MIT
tiate themselves from crowdsourcing platforms, 2020, 34). Even within the narrow applications of
such as AMT, Clickworker or CrowdFlower (now AI that are used for hiring practices, obtaining a
Appen), and market their crowd workforce as bank loan or face recognition, AI is revealing limi-
qualified or trained workers, or as experts in the tations; AI decisions risk being discriminatory as
field (Schmidt 2019; ILO interviews with start-ups; they can exhibit historical biases and their logic
see box 3.6). Many of these AI-enabled services cannot be explained (Bodie et al. 2016).
and the development of AI are in fact subsidized
Therefore, while one would have expected that
by crowdworkers, as they are needed initially to
tasks such as automating a virtual assistant would
train the AI models to correctly infer patterns that
be relatively easy given the purported advances
can be automated over time. As a result, many of
in technology, the fact that AI still operates with
them are inadvertently helping large established
human-in-the-loop assistance shows that natural
companies to become “data-opolies” and control
language processing is still in the development
the market (Stucke 2018, 275).
phase (see box 3.6). Though natural language
Currently, these systems are designed to operate processing capabilities are advancing, there is
as human-in-the-loop, with a worker reviewing still a long way to go before the entire workflow
the AI analysis and making the final decisions process of a particular task is powered by AI end
(Armour and Sako 2020). Advances in AI and to end and AI completely replaces workers. Thus,
machine learning are ostensibly not eliminating while a business might adopt “virtual assistant”
humans from the performance of tasks, but are technology, persuaded that AI is processing its
transforming their role and “integrating humans requests and thereby replacing its workforce, in
and computers more tightly” (Tubaro, Casilli and practice the tasks are outsourced to crowdworkers
Coville 2020, 6). Furthermore, the AI applications through digital labour platforms. An ILO survey
available today are suited for limited usage; a gen- of about 300 online home-based workers in the
eral AI that can perform cognitive tasks as workers Philippines found that about 14 per cent of the re-
do remains far beyond the reach of current tech- spondents were working as “virtual assistants” for
nology. A Massachusetts Institute of Technology clients based in Australia, Canada, the Philippines
taskforce, which looked at the implications of AI and the United States (ILO 2021; King-Dejardin,
on jobs in a number of sectors, such as insurance, forthcoming). There might be some jobs lost or
healthcare, autonomous vehicles and manufac- generated due to AI, but most importantly AI is
turing, found that much of the AI systems that are leading to a shift in the nature of the employment
deployed today can solve a limited set of specific relationship, as tasks are performed by invisible
problems, based on large amounts of data and by workers on digital labour platforms, raising ques-
extracting patterns. However, “the ability to adapt tions about the quality of jobs.
Box 3.6 “Jordan”, the automated virtual assistant: A case study

X
Jordan.inc,1 founded in San Francisco, United States, in 2014, aims to provide business clients
with an automated service for the scheduling of meetings. The company raised US$120,000 as
seed capital and a further US$11 million from venture capital funds. It provides virtual assistant
services and sells monthly subscription packages ranging from US$99 to US$399. By 2019,
Jordan.inc had around 350 clients and a workforce of 18 in its San Francisco office (technical
and engineering staff who develop the AI), together with some 200 workers around the world
who perform microtasks through digital labour platforms.
The product, sold as “Jordan”, is a virtual assistant that coordinates and schedules meetings.
Instead of sending multiple emails back and forth, a client can simply copy Jordan into all emails
that refer to meeting requests, and Jordan then schedules the meetings and enters them into the
business calendar in less than 45 minutes. The company claims that it is continuously improving
Jordan with the help of smart and self-motivated “Jordan Remote Assistants”. Clients have
praised Jordan for its efficiency and accuracy, which the company attributes to the combination
of precise machine intelligence and the judgement of an expert team of workers. But what does
this mean in practice?
The challenge of automating the “virtual” assistant

The challenge of automating a meeting schedule is that it requires the ability to understand the
often idiosyncratic requirements of clients expressed in an email. For humans, this is a function
of our natural language processing intelligence, but for AI this requires an additional large-scale
input of data about customer preferences and behaviour for the AI to be trained to recognize
patterns and make the correct decisions.
For example, a virtual assistant such as Jordan is not yet able to understand or process email
content such as, “Hey, I can do a call next week”. According to AI developers at Jordan.inc, the
wording of this message makes it difficult for AI to understand that: (i) the sender is requesting
a meeting; (ii) the type of meeting request is a call; and (iii) the meeting is to be scheduled next
week.
It appears that human expertise is still required for a simple message such as this, so that the
information can be decomposed into a structure that AI can process.
Implementation strategy for building and perfecting AI

Jordan.inc implemented its strategy to develop and automate the virtual assistant service in
two phases:
(i) Phase 1: Exclusively human-driven

The goal of the first phase was to build a client base so that technicians could collect data
and develop AI for organizing meeting schedules. Initially, the company founders manually
connected different calendars, messaged people and scheduled the meetings. They learned that
the key qualities of a virtual assistant are good communication skills, intuition and a pleasant
communication style. They then hired workers from Upwork, one of the largest online web-
based freelance platform, and trained them to schedule meetings manually. As the client base
expanded, Jordan.inc designed its own digital labour platform called “Workplace Jordan Remote
Assistant” (JRA) instead of hiring workers through Upwork.
Box 3.6 (cont’d)

X
(ii) Phase 2: Hybrid (human–machine interaction)

In the second, “hybrid” phase, AI developers at Jordan.inc attempted to automate the workflow
process and build algorithms so that over time it would become cheaper to schedule tasks by
reducing reliance on the growing JRA platform workforce. This phase involved a combination
of human–machine interactions (a human-in-the-loop system), whereby workers on the JRA
platform would extract parameters relevant for scheduling the meeting – availability of the
participant, location, date and time – from emails, and on this basis train the AI, then check
whether the parameters were being correctly used by the AI, and correct the decisions taken
by the AI if necessary, thereby improving its future performance.
Final outcome
In 2020, Jordan.inc continued to combine the virtual assistant service with the human-in-the-
loop system, despite its original ambition of developing a fully automated service. At this stage,
human–machine interaction is integrated throughout the entire workflow and human judge-
ment remains critical for reviewing final decisions. Administrative scheduling tasks have thus
been only partially replaced by AI. In fact, work has been dispersed in the form of thousands of
microtasks around the world to an invisible online crowd of workers. The JRA platform workers
are based in around ten different countries, including the Philippines and the United States.
On the company website, Jordan.inc now explicitly mentions that scheduling workflows are
efficient and accurate because they combine machine learning and expert human support. The
development of a virtual assistant that can deliver 90 per cent precision through AI language
processing alone would not be sufficient to attract and sustain a viable client base. The CEO of
Jordan.inc has admitted that “AI has a long way to go before it can completely replace humans”.
1
This case is based on an interview with representatives of a start-up company whose name has been changed

to Jordan.inc (and the product to Jordan) to preserve anonymity.

Source: ILO interview, 2020.

Current AI advances in certain fields are demon- et al. 2020, 27). AI applications can be deployed by
strating medium-term implications for work, digital platforms to analyse such data and deliver
workers and businesses through algorithmic personalized recommendations to customers in
matching, rating and pricing on e-commerce, real time. For instance, Amazon is said to change
business-to-business and digital labour platforms. the price of its listed products every 10 minutes,
In addition, AI seems to have radically altered which is more often than any retail shop can ever
marketing and sales activities in the consumer do (Mehta, Detroja and Agashe 2018). This is made
goods, retail and banking sectors through the possible due to the availability of large amounts
use of data analytics and tracking tools that of data that are collected on their consumers
produce vast amounts of customer transaction using various analytical tools on the platform.
and attribute data (Chui et al. 2018). This data The implications of such developments in AI for
is used in taking marketing decisions such as traditional businesses and workers are discussed
“pricing, promotions, product recommendations, in section 3.4 and Chapter 4, respectively.
enhanced customer engagement” (Davenport
3.4 Impact of digital platforms

on traditional businesses
The rise of digital platforms is resulting in com- dominant market position. The degree of market
petition between platforms and traditional power concentration can be discouraging not
businesses, with some platforms establishing a only for traditional businesses, but also for new
dominant position in the market, such as Amazon platform entrants.
in the online retail sector or Uber in the taxi sector.
At the same time, certain dynamics within e-com-
These developments are presenting traditional
merce markets have raised concerns regarding
enterprises, particularly SMEs, with opportunities
“anticompetitive collusive and unilateral conduct
and challenges. This section examines the implica-
by economic operators” (OECD 2019b, 5). Amazon,
tions of the rise of digital platforms for traditional
for instance, has been criticized for its competition
businesses, with a focus on the retail sector.
practices and their implications, particularly for
There has been increasing consolidation in SMEs, and is facing antitrust claims in a court in
the digital economy, with about 5 per cent of the United States (Bloomberg Law 2020). Large
platform companies (21 companies) making technology companies, such as Amazon, Apple,
20 per cent of the total net income among com- Facebook and Google, are increasingly being
panies on Standard & Poor’s 500 Index in 2019 investigated by competition authorities around the
(Moazed 2019; UNCTAD 2019). Consolidation world (see also Stucke 2018).6 The Confederation
is also occurring at the country and regional of All India Traders, an organization representing
levels. For instance, in India, two platforms small businesses in India, has been alleging,
(Amazon and Flipkart) controlled about 63 per including through street protests, that unfair com-
cent of the market share in online retail in 2018 petition practised by Amazon is a threat to small
(S&P Global Market Intelligence 2019). Similarly, businesses in the country (Sonnemaker 2020).
in the European Union (EU), where there were
While some large traditional businesses may
over 10,000 platform start-ups in 2018, these
be able to acquire platforms to improve their
accounted for only 2 per cent of the total value of
competitiveness, most SMEs are unlikely to have
all platforms, while the seven largest platforms
adequate resources for such undertakings. Many
accounted for 69 per cent of the estimated value of
SMEs therefore use digital platforms, such as
the digital economy (European Commission 2019;
Alibaba, Amazon or Flipkart, to gain access to a
KPMG 2018). The consolidation is due to some of
wider customer base and to build and sustain their
the major platforms acquiring both smaller plat-
business. However, traditional businesses, par-
forms and traditional businesses. For example,
ticularly SMEs, encounter a number of challenges
Amazon and Alibaba, the two biggest e-commerce
in conducting their activities through digital
platforms, have acquired businesses in a range
platforms (Crémer, de Montjoye and Schweitzer
of sectors, from entertainment and finance to
2019; OECD 2019b; UNCTAD 2019; Duch-Brown
news and fresh food. In 2018, the US-based
2017a; Martens 2016). Some of these challenges
retail chain Walmart acquired Flipkart, one of the
are described below.
largest online retail platforms based in India, for
an unprecedented US$16 billion, in a move to take The contractual terms between platforms and
on Amazon in the online retail market (Economic business users, many of which are SMEs, are
Times 2018). Economies of scale, network effects unilaterally determined by the platforms and
and data collection enable platforms to achieve a are generally complex and unclear (European
6 See, for instance, for Amazon: https://ec.europa.eu/commission/presscorner/detail/pl/ip_19_4291; for Apple: https://ec.europa.eu/

commission/presscorner/detail/en/ip_20_1073; for Facebook: https://www.nytimes.com/2020/12/09/technology/facebook-antitrust-
monopoly.html; and for Google: https://www.bbc.com/news/business-54619148.
There has been
Commission 2016a and 2018). For instance, the

increasing consolidation
criteria for blocking a user’s account are not clearly in the digital economy,
defined, which can have severe implications for
the continuation of their business operations with about 5 per cent
(European Commission 2016a). Contracts are uni-
laterally determined, the commission fees charged of platform companies
by platforms to business users can vary consider-
ably, and platforms can increase rates arbitrarily
(21 companies) making
without any negotiation (this was particularly 20 per cent of the
observed for location-based platforms).7 During
the first months of the COVID-19 pandemic, total net income.
when many restaurants were largely dependent
on delivery platforms to continue their business
operations, the commission fees charged ranged
between 15 and 35 per cent in the United States, Competition between platforms and traditional
while discounts were being offered to consumers businesses is also increasingly shaped by data.
(Cagle 2020). This is especially so when platforms rely on data
they collect from their business users to promote
Another factor shaping competition between
their own goods and services in the marketplace.
platforms and traditional businesses is the clas-
Google, for example, was fined in 2017 by the
sification of platform services. Digital labour
European Commission for abusing its dominant
platforms often avoid sector-specific regulations,
position as a search engine by prominently
such as those governing the taxi sector, by insisting
placing its comparison shopping service “Google
that they are merely technology companies pro-
Shopping” in its search results.9 More recently,
viding intermediation services. In the EU, a number
the European Commission has launched an
of businesses have emphasized “that there is a
investigation into Amazon based on preliminary
problem stemming from the fact that entire sectors
findings that the platform is using the data of
that are subject to sector-specific rules now in fact
sellers trading on the platform to directly compete
compete with online platforms in these same sec-
with them.10
tors, yet those online platforms are not subjected
to the same regulations” (European Commission Furthermore, competition issues can occur not
2016b, 17). These rules cover areas such as con- only when platforms promote their own goods
sumer protection, social security, labour market and services over those of competitors, but also
regulation, and taxation of and technical standards when they favour certain business users on the
relating to goods and services markets (Martens platform. In India, a number of antitrust cases
2016). Nevertheless, judicial decisions such as that have been filed by associations of businesses
by the Court of Justice of the EU, which held that against retail platforms like Amazon and Flipkart,
Uber’s services must be classified as services in the alleging preferential seller treatment through an-
field of transport,8 can help create a level playing ti-competitive practices such as deep discounting
field with the traditional taxi sector. (Kalra 2020).11 This alleged preferential treatment
7 Based on ILO interviews with restaurant owners.

8 Case C-434/15 Asociación Profesional Elite Taxi v Uber Systems Spain, SL [2017], available at: http://curia.europa.eu/juris/liste.
jsf?num=C-434/15.
9 European Commission, Antitrust Case 39740 – Google Search (Shopping). For a summary of the decision, see the press release:
https://ec.europa.eu/commission/presscorner/detail/en/IP_17_1784.
10 For more details, see: https://ec.europa.eu/commission/presscorner/detail/en/ip_20_2077; see also https://ec.europa.eu/
commission/presscorner/detail/en/IP_19_4291.
11 See, for example: Competition Commission of India, Case No. 09 of 2020, Case No.40 of 2019 and Case No. 20 of 2018.
becomes even more problematic considering that agreements (Competition Commission of India
the decision to promote one business user over 2020) that can also lead to anticompetitive
another is often based on algorithmic ranking practices. In 2019, Google was fined by the
which is non-transparent (European Commission European Commission for abusing its dominant
2017a). It is estimated that in the EU, the aggre- position in the market for online search adver-
gated financial impact of the uncertainty derived tising intermediation by including exclusivity
from opaque practices on online platforms is clauses in its agreements with third-party web-
between €2 billion and €19.5 billion per year sites that prevented other online advertising
(Duch-Brown 2017b). As a result, there have been brokers from placing their search adverts on
a number of alternative platform initiatives, such these websites.12
as open source community platforms or platform
Another challenge for many business users relates
cooperatives, that have tried to bring about more
to copyright or intellectual property right infringe-
transparency by building fairer distribution sys-
ments enabled through digital platforms, which
tems (see box 3.7).
have implications for their profits and reputation.
Opaque practices on e-commerce platforms However, regulatory frameworks are unclear
are also observed in algorithmic pricing. More about the responsibility of digital platforms in
specifically, data collected on these platforms instances where the intellectual property rights
allows them to analyse the demand for goods of business users are infringed. In a recent case
and services, and to adapt prices accordingly before the Competition Commission of India, a
via algorithms. Furthermore, data collection business alleged, among other issues, that coun-
allows platforms to target the preferences of terfeit products with its branding were appearing
consumers and businesses, including through on Amazon at “unfair and discriminatory prices”,
rebates, incentives and loyalty programmes. to which the Commission replied that the issue,
Many SMEs, however, lack such data or the finan- though of concern, was not one of antitrust.13
cial means to be able to compete with platforms
The rise of digital labour platforms also poses
and their pricing systems. As a result, access
challenges that have not yet been adequately
to data, combined with their pricing strategies,
addressed for both domestic and international
offers platforms a competitive advantage over
taxation regimes. Challenges in relation to
traditional businesses (Mehta, Detroja and
taxation have also arisen with regard to data,
Agashe 2018). This potentially threatens the sus-
especially given the fundamental role of data in
tainability of traditional businesses, and in turn
the creation of value (OECD 2014). Traditional,
the income stability of the workers engaged in
formal enterprises have more clearly defined
these enterprises. Such pricing strategies are not
obligations regarding taxation and may end up
specific to the retail sector but are also quite wide-
paying higher taxes than many platforms, which
spread in the taxi sector, which raises important
therefore have a competitive advantage. On this
questions from a competition law perspective
issue, the European Commission announced in
(Fountoukakos, Pretorius and Geary 2018).
2020 that it will move forward with a digital tax
The competition and business operations on should the negotiations at OECD level not produce
some platforms are also shaped by exclusivity immediate and satisfactory results.14
12 European Commission, Antitrust Case 40411 – Google Search (AdSense). For a summary of the decision, see the press release:
https://ec.europa.eu/commission/presscorner/detail/en/IP_19_1770.
13 For more details, see: Competition Commission of India, Case No. 09 of 2020, Paras 8 and 28 https://www.medianama.com/
wp-content/uploads/CCI-Amazon.pdf.
14 For more details, see: https://www.politico.eu/article/gentiloni-eu-ready-to-launch-new-digital-tax-if-us-stalls-global-talks/.
Box 3.7 Open source community platforms in the retail sector

X
The Open Food Network (OFN), a global open source software platform operating in the retail
sector, is a virtual space in which farmers, wholesalers and communities can set up their own
online stores and collaborate in selling their produce. It operates in a number of developing and
developed countries, including Australia, Belgium, Brazil, Canada, Colombia, Costa Rica, France,
India, Norway, South Africa, the United Kingdom and the United States. The aim is to create
fairer and more transparent food supply chains, and to move towards regenerative forms of
agriculture so as to build resilient natural systems.
The OFN platform offers subscription packages to shops or business users; for example, on
the United Kingdom OFN platform, shops are offered four subscription packages depending
on their size and scaling needs. These are: Basic (£1 minimum donation per month), Starter
(2.4 per cent of monthly sales (including VAT)), Scale (£60 per month plus 0.6 per cent of monthly
sales (including VAT)) and Enterprise (custom pricing). Depending on the plan, shops can benefit
from additional digital tools and assistance, yet they all get full voting rights irrespective of
the plan selected.
Sources: https://www.openfoodnetwork.org/find-your-local-open-food-network/; https://about.

openfoodnetwork.org.uk/pricing-and-plans/.

Finally, a key challenge that many businesses face business in equitable circumstances, but also
relates to dispute resolution. The need to ensure to ensuring business continuity when they are
fair dispute resolution with platforms has been confronted with unjustified delisting or freezing
invoked by business users in the EU, especially of assets (European Commission 2017c). All these
with regard to sudden delisting of accounts challenges are increasingly being subject to thor-
(European Commission 2017b). For business ough consideration in a number of countries and
users such as SMEs, fast and easy redress mech- Chapters 5 and 6 discuss some of the measures
anisms are not only crucial to ensuring fairness that have been taken.
and safeguarding their fundamental right to do
Conclusion
This chapter has shown that a wide variety While the proliferation of platforms and their use
of businesses are increasingly using digital by businesses have provided opportunities, a
labour platforms, both online web-based and multitude of challenges have also emerged. BPO
location-based, in their efforts to achieve greater companies face competition from large companies
efficiency and expand their customer base, among and the prevalence of low-end and low-value
other factors. Several benefits for businesses in tasks reduces their margins, particularly in the
using online web-based platforms have emerged: case of SMEs. For businesses that are dependent
the platforms allow them to streamline recruit- on delivery platforms, poor digital infrastructure
ment processes and to better match talent with as well as platform glitches or delays caused
needs, to reduce costs and to enhance their access by the delivery couriers can have a significant
to knowledge and innovate faster. Having access impact on the smooth running of the business,
to a large global pool of workers with diverse while high commission fees can reduce profits.
skill sets accessible through online web-based Traditional businesses, particularly in the retail
platforms can be seen to be contributing towards sector, are facing market disruptions from large
improved organizational performance for many e-commerce platform companies. While some
firms. For several of these businesses, SMEs in businesses have resorted to joining platforms to
particular, the use of location-based delivery reach a wider customer base, they face challenges
platforms has opened up opportunities to expand in terms of unfair competition, unfavourable con-
their markets as well as increase productivity and tractual terms, non-transparency on the part of
profitability, while taxi platforms have enhanced the platforms (with regard to data and pricing),
the convenience and accessibility of transporta- weak dispute resolution mechanisms, and, more
tion for many businesses and consumers. broadly, an uneven playing field. Many of these
issues are also increasingly receiving regulatory
Furthermore, the rise of digital platforms has
attention, particularly from competition author-
created opportunities for entrepreneurship and in-
ities in many countries.
novation for BPO companies and digital start-ups.
BPO companies have been able to transition from Despite such challenges, digital platforms have
providing voice-based to digital services in order become pervasive in today’s society and economy,
to cater to the demands of their clients. Many especially since the outbreak of the COVID-19 pan-
digital start-ups have also sprung up to meet the demic. Given the increasing reliance of businesses
demands of automated and AI-enabled services, on digital labour platforms and the fact that these
for example in analytics and tracking. However, as platforms are gradually shaping the world of work,
AI technology is far from mature and a completely it becomes all the more relevant and urgent to
autonomous AI remains a distant prospect, many better understand the implications of these
such start-ups draw on human intelligence to un- developments for the worker experience in the
dertake tasks and support machine learning by digital economy. In this regard, the next chapter
engaging a globally dispersed workforce that is captures the diverse experience of workers on
available every day and round the clock (24/7) on both online web-based and location-based digital
digital labour platforms at a relatively low cost. At labour platforms.
the same time, digital platforms in sectors such as
retail have also benefited many businesses, espe-
cially SMEs, which can increasingly sell products
globally through e-commerce platforms.
4
Digital labour
platforms and
the redefinition
of work
Opportunities and
challenges for workers
Design of a platform
The worker experience Accessing and performing work
69%
of app-based
70% of them
have taken a loan
taxi drivers
65 hours own their vehicle
is the average working Resources

week of an app-based
taxi driver
required
Terms of service
agreements
Access
to the
Workers' ratings are
platform
decisive for accessing work
Freelance platforms
Commission fees
82% Demonstrate
Taxi capability Online web-based
72% 3.5% to 20%

Delivery Taxi
65% Costs 5% to 25%
incurred
Working time
of every hour
Algorithmic
matching
1/3 is unpaid on
online web-based
platforms
process
Workers on freelance platforms
Autonomy 47% are monitored by their

and
clients for hours worked…
control 46% are required to take
screenshots of their work…
43% are required to be available

during a specific time…
Quality On a regular basis

Workers whose account
was deactivated
assurance
of work After finishing work
Taxi Delivery
Lack of awareness of a
19% 15% dispute resolution mechanism
Dispute
resolution Freelance platforms
mechanism 52%
Taxi
42%
Delivery
32%
Workers whose work was rejected
Microtask platforms Hourly earnings (US$)

Non-
86% payment Payment
Microtask
Freelance platforms 3.3
Delivery
35% 0.9 to 3.5
Freelance
7.6
Social protection (access to pension) Taxi
1.1 to 8.2
Online web-based Taxi Delivery 0 10
20 % 18% 17%
4. Digital labour platforms and the redefinition of work 135
The online world is

complicated and full of
opportunities and hopes,
and of course is also full of
various traps and pitfalls.
X Male respondent on freelance platform
EPWK (China)
Introduction
The previous chapters have presented the This chapter presents findings from ILO surveys
emergence of digital labour platforms, their conducted among workers engaged on online
business model, and how they are changing the web-based and location-based platforms. It docu-
organization of work. Platforms are increasingly ments worker experience on online web-based
redefining, through the use of technology, how platforms such as microtask, freelance or
economic relationships are established between contest-based and competitive programming
workers and clients or customers, many of whom platforms through surveys conducted at the
are geographically dispersed around the world. global level, and at the country level in China and
Ukraine. Through extensive field-based surveys
Simultaneously, digital labour platforms are
it also presents new insights into the situation of
creating opportunities for work and gaining
workers in taxi and delivery services in developing
popularity globally among policymakers and
countries, which so far has remained inadequately
governments as a means of boosting economic
explored. By drawing on the findings of surveys
development, along with enhanced information
conducted among some 12,000 respondents, the
and communications technology (ICT) pene-
chapter provides a first major comprehensive
tration in many countries (AfDB et al. 2018; Roy,
picture of the worker experience on digital labour
Balamurugan and Gujar 2013; Narula et al. 2011).
platforms in multiple sectors and countries.
Moreover, digital labour platforms are attracting
workers across multiple sectors and countries The chapter begins by providing the basic demo-
as they provide flexibility in work schedules, the graphic characteristics of the platform workers
option to undertake work from any place and at surveyed and their motivations for undertaking
any time, and the ability to choose the tasks to platform work in section 4.1. Section 4.2 explores the
be performed (Berg et al. 2018; AfDB et al. 2018). heterogeneity of worker experience in navigating
complex platform designs to obtain work, perform
Despite the opportunities emerging through
tasks and receive income, thereby bringing to the
digital labour platforms, concerns are being
fore the opportunities and challenges encountered
raised about the worker experience on such
with regard to access to work, earnings, working
platforms, particularly with regard to working
time, social protection, and occupational safety and
conditions – from limited access to work and social
health. Section 4.3 focuses on how digital labour
protection to low earnings and income volatility
platforms use algorithms to manage and evaluate
(Rani and Furrer, forthcoming; Federal Reserve
workers and how that practice is impacting the
Board 2019; Berg et al. 2018; Farrell and Greig
extent of autonomy and control that workers can
2016; United Kingdom, Department for Business,
exercise over their work. Section 4.4 investigates
Energy and Industrial Strategy 2018a). Ensuring
the worker experience with regard to skills acqui-
decent work opportunities for all calls for a better
sition and development, and skills mismatch as
understanding of the platform worker experience,
digital labour platforms increasingly redefine the
and of worker motivations, opportunities and
relationship between formal education and tasks
challenges across multiple sectors, countries and
performed. Section 4.5 discusses the role of plat-
contexts.
form design in shaping the worker experience in
the context of non-discrimination issues.
4.1 Basic demographic characteristics
of platform workers
The ILO conducted several sur veys across countries, and the app-based delivery sector in 11
countries and sectors between 2017 and 2020 countries, comprising about 5,000 respondents
(see table 4.1). In the global surveys conducted spanning the Arab States, Africa, Asia and the
on microtask (2017), freelance and competitive Pacific, Eastern Europe, and Latin America and the
programming platforms (2019–20), about 2,900 Caribbean. This was complemented by a survey
respondents from 100 countries took part. In of over 2,200 respondents in traditional taxi (nine
addition, two country-specific surveys of workers countries) and delivery (four countries) sectors.
on online web-based platforms were conducted
All the surveys contained both quantitative and
in China (1,107 respondents) and Ukraine (761 re-
qualitative questions, including open-ended text
spondents) in 2019. In this chapter, the term “online
questions aiming to obtain insights into the ex-
work” includes the combined data from the global
perience of workers engaged in these sectors (see
and country-specific surveys to provide a broad
Appendix 4A). Given the lack of official statistical
overview of the worker experience on online
information on the numbers and characteristics
web-based platforms. When referring to “devel-
of platform workers (see section 1.3), including
oped” or “developing” countries with regard to
those using online web-based and location-based
these platforms, for methodological reasons (see
platforms, there was no sampling base from
Appendix 4A) only the global surveys are taken into
which a random sample could be drawn. The
account; the country-specific surveys are excluded.
statistics presented in this chapter therefore re-
Surveys were also conducted among workers on flect the findings of the ILO surveys, and are not
location-based platforms during 2019 and 2020 necessarily representative of a global or coun-
with a focus on the app-based taxi sector in nine try-level population.
Table 4.1 Number of respondents, by survey

X
Number of
Online web-based platforms Main platforms covered
respondents
Freelance and contest-based Freelancer, Upwork 449

CodeChef, Codeforces, HackerRank, Iceberg,
Competitive programming 62
Global surveys Topcoder
AMT, Clickworker, CrowdFlower (now Appen),
Microtask 2 350
Microworkers, Prolific
China 680, EPWK, ZBJ, k68 1107
Country-specific
surveys Advego, Freelance, Freelancehunt, Freelancer,
Ukraine 761
Kabanchik, Upwork
Number of
Location-based sectors Countries surveyed
respondents
Chile, Ghana, India, Indonesia, Kenya, Lebanon,

App-based 2 077
Mexico, Morocco, Ukraine
Taxi
Chile, Ghana, India, Indonesia, Kenya, Lebanon,
Traditional 1 864
Mexico, Morocco, Ukraine
Argentina, Chile, China, Ghana, India, Indonesia,
App-based 2 965
Delivery Kenya, Lebanon, Mexico, Morocco, Ukraine
Traditional Chile, India, Kenya, Lebanon 347
Sources: ILO global surveys of crowdworkers (2017) and workers on freelance and competitive programming platforms
(2019–20); ILO surveys of platform workers in China (2019) and Ukraine (2019); and ILO selected country surveys of taxi
drivers and delivery workers (2019–20).
4.1.1 Age distribution 4.1.2 Participation of male

of platform workers and female workers
Across the sectors surveyed, the majority of on platforms
workers engaged on online web-based and loca-
About four in ten workers on online web-based
tion-based platforms are below 35 years of age.
platforms are women, while in developing
The average age of workers on online web-based
countries only about two in ten are women (see
platforms is about 31 years and is higher among
figure 4.2). These figures underline the fact that,
workers from developed countries (35 years)
in a similar way to the offline labour market,
than in developing countries (30 years). Workers
the online labour market poses challenges for
engaged in competitive programming tend to be
women in accessing work. Among competitive
the youngest (22 years) (see figure 4.1.), indicating
programmers, only 1 out of 62 respondents
that many are using these platforms to hone
was female, which reflects the occupational
their skills. In the taxi and delivery sectors, app-
segregation in the IT sector (see also Aleksynska,
based taxi drivers (36 years) and delivery workers
Bastrakova and Kharchenko 2018; Shevchuk and
(29 years) tend to be younger than those engaged
Strebkov, forthcoming).
in traditional settings (taxi drivers: 44 years;
delivery workers: 31 years). The app-based taxi and delivery sectors are
largely male-dominated. Women comprise fewer
than 10 per cent of workers in these sectors, and
this proportion is even lower in the traditional sec-
tors (below 5 per cent), as shown in figure 4.2. The
share of women is considerably higher in some
countries, for instance in Indonesia in the app-
based taxi sector (13 per cent), where female-only
Figure 4.1 Age distribution, by occupation
Online work Taxi and delivery sectors

Microtask
0.20 0.08 App-based taxi

33 Freelance 29 31 36 44
22 27 32 33
0.15 0.06 Traditional taxi
Competitive
Density
Density
programming
App-based delivery
0.10 0.04
China
Traditional delivery
0.05 0.02
Ukraine
0 0
20 30 40 50 60 70 20 30 40 50 60 70 80
Age (years) Age (years)
Note: Vertical dashed lines indicate mean values.
Sources: As for table 4.1.

taxis are preferred by some female clients

Figure 4.2 Share of female respondents, to mitigate risks of violence and harassment
by occupation and country (Straits Times 2015). In Kenya, where only 5 per
cent of app-based taxi drivers are women, some
Online work platforms are undertaking special measures to
60 encourage their greater participation, such as
57 priority access to vehicle financing (Taxify). A plat-
Percentage of respondents
50 form with female-only taxis has also emerged

47
40 (An-Nisa Taxi) (Osman 2019).
41
37 37 38
30
20 24
30
4.1.3 Participation of workers
10
from rural and urban areas
2
0 This section focuses on workers on online web-
based platforms, not on taxi or delivery services as
Freelance
Competitive
programming
Developed
countries
China
China, Ukraine)
Developing
countries
Ukraine
Total
Microtask
Total (without
the surveys for the latter were conducted in urban

areas only. There is limited penetration of online
web-based platforms in rural areas, particularly
in developing countries. The vast majority of re-
Taxi sector spondents (84 per cent) on such platforms reside
in urban or suburban areas. The share of those
20
performing online work who live in rural areas
or small towns is higher in developed countries
15 (23 per cent) than in developing countries (16 per

cent). With increased ICT connectivity and its
spread to rural areas, there is income-generating
10
potential for online work in these areas, whereby
skilled workers would be able to access jobs in the
5 global labour market (Kalleberg and Dunn 2016).
I live in an area where there are few oppor-
0
tunities for this type of work. My only other
Chile
Ghana
India
Indonesia
Kenya
Lebanon
Mexico
Morocco
Ukraine
Total
option to work in this field would be to move

to a big city, pay high rent and reduce the
time I spend with my family and friends –
App-based Traditional Female respondent on freelance platform
Upwork (Ireland)
Delivery sector
25
20
15
10
0
Argentina
Chile
China
Ghana
India
Indonesia
Kenya
Lebanon
Mexico
Morocco
Ukraine
Total
App-based Traditional

4.1.4 Participation of migrants In some countries, many migrant workers engage

in the app-based delivery sector. The proportion
on platforms of migrant workers is higher in this sector (15 per
cent) than in the app-based taxi sector (1 per
I signed up to Upwork after emigrating.
cent), and similar differences exist in the trad-
I used it to get started in a new country as
itional delivery and taxi sectors. However, there
a freelancer. I got work online very quickly
are considerable variations across countries (see
and it provided me with an income to get
figure 4.3). Argentina and Chile, for instance, each
started – Female respondent on freelance plat-
with a high proportion of migrant workers in the
form Upwork (Canada)
app-based delivery sector (over 70 per cent),
Online web-based platforms offer some oppor- have seen a large influx of Venezuelan refugees
tunities to migrant workers1 in accessing work, and migrants into their national labour markets,
particularly in developed countries. The ILO sur- who face uncertain employment prospects even
veys reveal that of those engaged on freelance though many have high education levels (ILO
platforms, 17 per cent are migrant workers. Their 2020c): in Argentina and Chile, 43 and 47 per cent
share is higher in developed countries (38 per cent) respectively of migrant respondents had attained
than in developing countries (7 per cent), and is a university degree. Working in the app-based de-
higher among women (39 per cent) than men livery sector emerges as an option for many due
(36 per cent) in developed countries, while it is to the lack of other available jobs corresponding
similar across the sexes in developing countries. to their education, low entry barriers and ease of
This could be indicative of the intersectional access to this sector, as well as discrimination in
barriers (such as those based on gender, migrant accessing jobs elsewhere.
status, indigenous or tribal identity, among
In the period immediately before I started
others) to accessing offline work faced particularly
working as a courier, I was a salaried em-
by many migrant women (King-Dejardin 2019).
ployee. I quit because I suffered discrimination
and exploitation against Venezuelans – Male
respondent on app-based delivery platform
Uber Eats (Chile)
Figure 4.3 Share of migrant respondents in the taxi and delivery sectors
Taxi sector Delivery sector
40 80
70 App-based
30 60
Traditional
50
20 40
App-based
30
10 20 Traditional
10
0 0
Chile
Ghana
India
Total
Indonesia
Kenya
Lebanon
Mexico
Morocco
Argentina
China
Ghana
India
Indonesia
Kenya
Lebanon
Mexico
Morocco
Ukraine
Ukraine
Total
Chile
Source: ILO selected country surveys of taxi drivers and delivery workers (2019–20).
1 In this chapter, “migrants” refers to workers born in a country that is different from where they were residing at the time of the survey.
Figure 4.4 Share of respondents who consider

their health to be poor or very poor,
by occupation and country
Online work
4.1.5 Health status of workers
on platforms
4
3 I use a wheelchair and experience severe

chronic pain as a result of a congenital ortho-
paedic condition. My days often depend upon
2
my pain. Freelancing gives me the flexibility
to set my own schedule and
1
work where and when
I n eed to. I can
0 China) work in 10-minute
Total
Male
Female
bursts if that is
Freelance
Competitive
programming
Developed
countries
Developing
countries
China
Microtask
Total (without
what I need to do.

And I often do
that – Female
respondent on
Taxi sector freelance plat-
form Upwork
8
(United States)
6 I became a driver because

my health is not good enough to work some-
where else – Male respondent on app-based taxi
4
platform DiDi (Mexico)
2
Some people in poor health or with disabilities
are able to find work on online web-based and
location-based platforms. About 2 per cent of
0
respondents on online web-based platforms
Morocco
Chile
India
Indonesia
Kenya
Lebanon
Mexico
Ukraine
Total
Ghana
reported poor or very poor health status, with no

major differences by sex (see figure 4.4.). Online
work can also provide opportunities for persons
with disabilities, given the additional barriers
they encounter in labour markets (Fundación
Delivery sector ONCE and the ILO Global Business and Disability
Network 2019). In particular, some respondents
15 in poor health or with disabilities identified the
possibility to work from home as being beneficial
12 in finding and carrying out work. The proportion

of respondents reporting poor or very poor health
9
in the app-based taxi and delivery sectors varied
across countries. In the app-based taxi sector it
6
ranged between 0 and 4 per cent, while in the
traditional sector it was slightly higher. The pro -
3
portion of delivery workers with poor or very poor
0
health ranged between 0 and 2 per cent in the
app-based delivery sector (see figure 4.4).
Mexico
Argentina
Chile
China
Ghana
India
Indonesia
Kenya
Lebanon
Morocco
Ukraine
Total
Sources: ILO global surveys of crowdworkers (2017)

and workers on freelance and competitive programming
platforms (2019–20); ILO survey of platform workers
in China (2019); and ILO selected country surveys of taxi
drivers and delivery workers (2019–20).
4.1.6 Education levels A sizeable proportion of workers engaged in the

app-based taxi and delivery sectors have high
of platform workers educational levels, including women and young
people. Even though these sectors are often
Workers on online web-based platforms are gen-
considered to be low-skilled, 24 and 21 per cent
erally highly educated, especially in developing
of app-based taxi drivers and delivery workers
countries. Over 60 per cent of respondents
respectively are highly educated (see figure 4.5).
engaged in online work, women and men alike,
These proportions are lower in the traditional sec-
are highly educated (having attained a university
tors. In some countries, such as Chile and India, a
degree) (see figure 4.5). A higher proportion of
considerably higher proportion of app-based taxi
workers engaged on freelance platforms (83 per
drivers and delivery workers are highly educated
cent) are highly educated compared to those
compared to those in the traditional sectors.
on microtask (64 per cent) and competitive pro-
gramming (50 per cent) platforms. A larger share Furthermore, even though there are fewer
of respondents on competitive programming women engaged in the app-based taxi and de-
platforms (73 per cent) are pursuing a degree livery sectors, a higher proportion of them are
compared to those on freelance (25 per cent) and highly educated (42 and 29 per cent respect-
microtask (21 per cent) platforms. ively) compared to men (24 and 20 per cent
respectively). Younger app-based taxi drivers
A larger proportion of workers on online web-
and delivery workers (18–24 years) tend to be
based platforms in developing countries (73 per
highly educated (24 and 17 per cent respectively)
cent) are highly educated compared to those in
compared to workers in the traditional sectors
developed countries (61 per cent). This propor-
(12 and 4 per cent respectively). This reflects the
tion is even higher among women in developing
challenges in the context of youth employment,
countries (80 per cent). This could be due to factors
where young people are often confronted with
such as the lack of opportunities in the local offline
poor employment opportunities (ILO 2020d and
labour markets, as well as additional barriers to
2020e) and look for any alternative possibilities
women in particular that prevent them from ac-
to earn an income (Aleksynska 2021; Anwar and
cessing work outside their homes, including care
Graham 2020; Surie and Koduganti 2016).
responsibilities and prevailing gender norms.
I took a training programme in the mech-
I started freelancing a couple of weeks after
anical field for operating machines. The
I graduated from college. I think I had gone
training is now over, and until I find a job
to a couple of interviews beforehand but
in that field, I am working as a delivery
none of them called back so I decided to try
boy – Male respondent on app-based delivery
freelancing – Female respondent on freelance
platform Uber Eats (India)
platform Upwork (Philippines)
Figure 4.5 Educational levels of workers, by occupation and country
Online work
Microtask
Freelance
Below primary/primary
Developed countries
Secondary
Developing countries
Higher secondary
China
Ukraine Bachelor's degree
Male
Postgraduate degree
and above
Female
Total (without China, Ukraine)
Total
0 20 40 60 80 100

App-based (A) Traditional (T) App-based (A) Traditional (T)
A Total A
T T
Argentina A
A Chile A
T T
China A
A Ghana A
T
A India A
T T
A Indonesia A
T
A Kenya A
T T
A Lebanon A
T T
A Mexico A
T
A Morocco A
T
A Ukraine A
T
0 20 40 60 80 100 0 20 40 60 80 100
Percentage of respondents Percentage of respondents

4.1.7 Worker motivation for I live in an overpopulated country where it

is very tough to get a good job. The pay is
engaging in platform work better than usual jobs, I am my own boss
and I like the freedom – Male respondent on
Complementing pay from other income sources
freelance platform Upwork (Bangladesh)
is the main motivation for performing tasks on
online web-based platforms (39 per cent), followed Working from home or job flexibility are particu-
by the preference or need to work from home or larly important for women. A higher proportion
for job flexibility (29 per cent), and as a form of of women (35 per cent) than men (25 per cent) on
leisure or because it is enjoyable (18 per cent) (see online web-based platforms are motivated by the
figure 4.6). Complementing pay is a major motiv- preference or need to work from home or for job
ating factor among younger workers in particular flexibility, and this is the case in developed and
(48 per cent for those aged 18–24 years) compared developing countries alike. About 23 per cent of
to older workers. women who perform online work have children
under the age of six years. As women with young
I also wanted to earn extra income to sup-
children tend to face a “motherhood employment
port some financial obligations for my family.
penalty” and globally account for the lowest
The salary I earn from my current job is not
employment rates (ILO 2018a; Grimshaw and
enough to cover the growing need of my
Rubery 2015), online work is providing opportun-
family – Male respondent on freelance platform
ities to work while managing care responsibilities.
Upwork (Philippines)
As a woman, I prefer to work from home.
I wanted a side income and had a try. And
I earn better than others. I have a child.
I was surprised I could earn some money –
I can maintain my family instead of doing a
Female respondent on freelance platform
regular job. That’s the reason I prefer to work
Upwork (Canada)
from home – Female respondent on freelance
In developing countries, the key motivating factors platform Upwork (Bangladesh)
are the preference or need to work from home or
Competitive programmers are motivated to work
for job flexibility (36 per cent) and complementing
on platforms to improve their skills, establish
pay (26 per cent), whereas in developed countries
networks and enhance their future career pro-
it is mostly complementing pay (43 per cent).
spects. About 85 per cent of respondents were
Furthermore, although not being able to find
motivated by this factor, which is a considerably
traditional work is also a motivation for some in
higher proportion than that of respondents on
both developing and developed countries (7 and
freelance platforms (12 per cent). While some
8 per cent respectively), better pay than in other
respondents participated mainly on CodeChef
available jobs is particularly relevant for those in
and Codeforces, which are platforms primarily
developing countries (11 per cent).
used for improving skills, others participated on
HackerRank and Topcoder, with the prospect of
earning prizes, apart from enhancing skills and
employment prospects.
Figure 4.6 Most important reason for performing work on digital labour platforms,
by occupation and country
Online work
Microtask
Freelance
Could not find other employment
Developed countries
Pay is better than other available jobs
Developing countries
To complement pay
China
Ukraine To improve skills/network/

career perspectives
Male
Because of the job flexibility
Female
As a form of leisure/I enjoy it
Total (without China, Ukraine)
Other
Total
0 20 40 60 80 100

App-based (A) Traditional (T) App-based (A) Traditional (T)
A Total A
T T
Argentina A
A Chile A
T T
China A
A Ghana A
T
A India A
T T
A Indonesia A
T
A Kenya A
T T
A Lebanon A
T T
A Mexico A
T
A Morocco A
T
A Ukraine A
T
0 20 40 60 80 100 0 20 40 60 80 100
Percentage of respondents Percentage of respondents

I like to do competitive programming I participate in freelance work because I was

because it challenges us to push our limits never this available to my children when I
and think out of the box. It also helps to get a worked in the corporate set-up. This allowed
job in big companies as their tests are similar me to become a mom yet still provide like a
to competitive programming competitions – breadwinner – Female respondent on freelance
Male respondent on competitive programming platform Upwork (Philippines)
platform HackerRank (India)
Online platforms are very good because
The lack of alternative employment opportunities there is free time for other activities and
is a prime motivating factor for many workers on no pressure from the employer as opposed
location-based platforms. This is the case for to working in an office – Male respondent on
40 and 28 per cent of the respondents in the app- freelance platform Kabanchik (Ukraine)
based taxi and delivery sectors respectively, and
In the app-based taxi and delivery sectors, the
also a major motivating factor in the traditional
majority of workers surveyed are satisfied with
sectors. Other key motivating factors among
their work. A higher proportion of app-based taxi
app-based workers include job flexibility, as well
drivers are satisfied or very satisfied compared to
as better pay (see figure 4.6). At the same time,
traditional taxi drivers, while the opposite is the
however, there are some differences across coun-
case in the delivery sector (see figure 4.7). In the ab-
tries and also across population sub-groups in
sence of traditional work opportunities in the local
some countries. For instance, in Chile, while those
labour market, app-based work provides workers
app-based delivery workers born in the country
with an income. This context might influence their
are motivated by flexibility (42 per cent), migrant
satisfaction levels despite negative aspects of the
workers in particular are motivated by a lack of al-
work, such as long working hours and high work
ternative employment opportunities (38 per cent).
intensity (see section 4.2.3; Prabhat, Nanavati and
App-based taxi driving was the only job that Rangaswamy 2019; Griesbach et al. 2019).
was available – Male respondent on app-based
taxi platform Safe Boda (Kenya) When there is no other source of earning and
all the daily expenses are paid for through
I started working as an app-based taxi this income only, then I have to be satisfied
driver to get instant money as I had an with it – Male respondent on app-based taxi
economic emergency situation caused by platform Uber (India)
unemployment – Male respondent on app-
based taxi platform Beat (Chile) App-based taxi drivers in Morocco stand out as
an outlier with high dissatisfaction levels (68 per
cent), which could be associated with a strike
4.1.8 Worker satisfaction during the period of data collection that is likely to
with platform work have created higher awareness levels among the
drivers about working conditions and pay. At the
A majority of workers on online web-based same time, while a single-measure job satisfaction
platforms are either satisfied or very satisfied with indicator may provide some insights, concerns
their work, and these figures are similar across the have been raised regarding its reliability and prev-
sexes (see figure 4.7). This is more likely to be the alent anomalies (Brown, Charlwood and Spencer
case for workers in developing countries (80 per 2012; Rose 2003; Oshagbemi 1999). Responses
cent) than for those in developed countries (71 per to single-measure job satisfaction questions
cent), and particularly so for women in developing have been observed to overestimate satisfaction
countries (84 per cent). levels in comparison with multiple-item measures
Figure 4.7 Worker satisfaction levels, by occupation and country
Online work
80 41 Very satisfied
32
60 23 34 20
21 21 20
25 21
8
40 Satisfied
47
20 50 50 43 51 48 47 38 48 46 50
Dissatisfied
0 or very dissatisfied
7 1 0 7 4 2 9 6 6 6 6
20
Ukraine
Total
Freelance
Competitive
programming
Developed
Developing
countries
China
Male
Female
Microtask
China, Ukraine)
Total (without
countries
Taxi sector
Very satisfied and satisfied
80 App-based
60
40 Very satisfied and satisfied
Traditional
20
0 Dissatisfied and very dissatisfied
20 App-based
40
Dissatisfied and very dissatisfied
60 Traditional
Indonesia
Morocco
Chile
Ghana
India
Kenya
Mexico
Ukraine
Total
Lebanon
Delivery sector
80
Very satisfied and satisfied

60 App-based
40
20 Very satisfied and satisfied
Traditional
0
20 Dissatisfied and very dissatisfied
App-based
India
Mexico
Chile
China
Ghana
Indonesia
Kenya
Lebanon
Morocco
Ukraine
Total
Argentina
Dissatisfied and very dissatisfied

Traditional

(Oshagbemi 1999). These responses are deter- The work is very stressful and I think we
mined more by “intrinsic” characteristics (such as deserve better pay – Male respondent on app-
flexibility, among others), rather than “extrinsic” based taxi platform Yango (Ghana)
characteristics (such as pay, contractual status or
Given such challenges with the single-measure in-
prospects for promotion, among others) (Rose
dicator, research has emphasized the importance
2003, 526–527). In Kenya, app-based taxi drivers
of complementing it with other dimensions or
have reported high levels of satisfaction (see
multiple-item measures (Brown, Charlwood and
figure 4.7) yet have undertaken strikes regarding
Spencer 2012; Oshagbemi 1999). In this regard,
pay and lack of consultation (Ochieng 2019;
understanding the granularity of the working
Nyawira 2019). Moreover, when asked about pay or
conditions and how work is being organized on
regularity of work, respondent dissatisfaction was
platforms is fundamental to a better appreciation
evident across both location-based and online web-
of both the worker experience and the opportun-
based platforms. Notably, 44 per cent of app-based
ities and challenges that are emerging.
taxi drivers and 38 per cent of app-based delivery
workers felt that they were paid unfairly.
4.2 Worker experience and the quality of work
on digital labour platforms

The working conditions and worker experience
on digital labour platforms can vary considerably.
4.2.1 Access to a sufficient
They are linked to and shaped by the relationship amount of work
of the worker with the platform, and the way in
which work is organized and managed by the A major challenge in the labour market is matching
platform. Initially, the experience may appear jobs and workers with corresponding skills. The
seamless, where tasks or clients are connected rise of digital labour platforms has been seen as a
to workers via the platform for pay. Navigating way to connect workers directly to work opportun-
through a digital labour platform and ultimately ities. However, the experience of many workers
receiving payment for work done can, how- on digital labour platforms is nevertheless marred
ever, be fraught with barriers and challenges. by several challenges to accessing a sufficient
Figure 4.8 captures the worker experience on amount of work.
digital labour platforms, both online web-based The majority of workers on online web-based
and location-based, from obtaining access to and platforms, particularly in developing countries,
performing work, to receiving feedback and pay- would like to undertake more online work. Of
ment. It also demonstrates the degree to which the respondents engaged in online work, 86 per
a worker is responsible for both the resources cent expressed this desire (see figure 4.9), with
required (equipment, vehicle, vehicle insurance, very small differences between male and female
software and hardware), and the costs incurred respondents. A higher proportion of respondents
(subscription and membership plans, additional from developing countries (92 per cent) reported
fees to access tasks, working time, fuel and main- the desire to do more online work compared to
tenance costs and internet costs). The rest of the those in developed countries (85 per cent). This is
chapter relates the experience of workers and the case despite the fact that many respondents
describes their working conditions, based on the have another paid job, in developed (56 per cent)
new data collected for selected sectors. and developing countries (41 per cent), and the
country-level surveys show these proportions to
be quite high in Ukraine (68 per cent).
Figure 4.8 Design of a platform: The worker experience
Accessing and performing work

Equipment
Resources
Vehicle
required
Vehicle insurance
Software/hardware
Background information
Licensing requirements
Access to Access restrictions
the platform Accepting the terms of
service agreements
Multi-homing
Induction and training

Building up profile Demonstrate
Ratings capability
Subscription/membership plans
Additional fees to access tasks
Costs Unpaid working time
incurred (search costs)
Fuel/maintenance costs
Internet costs
Ratings
Speed of delivery/
performing tasks
Algorithmic
Acceptance rate
matching
Physical proximity
process
Client/customer
requirements
Work instructions
Monitoring/tracking
Choice over task/
Autonomy location/time/tools/
and control equipment
Exclusivity clauses
Feedback mechanisms
Acceptance/rejection of work Quality
Account deactivation assurance After finishing work
Withholding payment of work
Dispute
resolution Communication channels
mechanism Awareness
Payment for tasks

Rejection of work + Bonuses/tips
Cancelled orders/rides Non-payment Payment – Commission fees
– Withdrawal/transaction fees
= Earnings
Source: ILO elaboration based on United Kingdom, Department for Business, Energy and Industrial Strategy (2018a).
Figure 4.9 Proportion of workers who would like to do more online work,
by type of platform, development status and sex
100
90 91
88 89 88 88 88
85 85 85 86
80 82
70
60
Total (without
China, Ukraine)
Freelance
Competitive
programming
Developed
countries
Developing
countries
China
Ukraine
Female
Microtask
Total
Male
Sources: ILO global surveys of crowdworkers (2017) and workers on freelance and competitive programming
platforms (2019–20); ILO surveys of platform workers in China (2019) and Ukraine (2019).
Figure 4.10 Most important reasons for not being able to undertake more online work,
100
I am not qualified
80
There is not enough
available work
60
It is too difficult
40
to find clients
20
I do not have the time
0
Total (without
China, Ukraine)
The pay is not

Freelance
Competitive
programming
Developing
countries
Ukraine
Microtask
Developed
countries
Male
Total
China
Female
good enough
Other
Note: Figures presented refer to workers who reported that they would like to undertake more online work.
Sources: As for figure 4.9.
One key factor preventing people from doing translate into having a greater amount of online
more work on online web-based platforms is lack work. Irrespective of experience (less than one
of availability of sufficient work. About 45 per year to more than three years), over 40 per cent of
cent of respondents reported this to be the case the workers on online web-based platforms who
(see figure 4.10). A similar trend was visible across would like to undertake more work find it difficult
developed and developing countries, and in both to access a sufficient amount.
sexes. Other reasons for not doing more work
Experience does not matter, but if you have
include not finding well-paid tasks (18 per cent)
fulfilled orders, this is the only way to get a
on microtask platforms and difficulty in finding
client – Female respondent on freelance plat-
clients (41 per cent) on freelance platforms.
form Freelancehunt (Ukraine)
Furthermore, the amount of experience a worker
has on platforms often does not necessarily
Competition on platforms can be strong, and one I have never paid the fee to move away
task or project can attract 100 to 200 or even more from the platform. I generally don’t move
workers, as observed in China (Chen, forthcoming). off the platform unless the client requests
This is also rooted in increasing labour supply (see it, as Upwork are very strict about it –
section 1.3), partly due to initiatives undertaken by Male respondent on freelance platform Upwork
governments to train people and promote digital (Malaysia)
labour platforms as a source of income generation
The ability of workers on online web-based plat-
(Galpaya and Senanayake 2018; Graham, Hjorth
forms to access a sufficient amount of work is also
and Lehdonvirta 2017; Graham et al. 2017).
shaped by the platform design. Platforms often
More and more people are working on the provide subscription plans or charge an additional
platform, in all industries, and the competi- fee to enable workers to have enhanced access
tion is fierce – Male respondent on freelance to work. This can allow workers to apply for tasks
platform ZBJ (China) faster than non-paying members, to have their
proposals reviewed quickly and to ensure that
Workers on online web-based platforms may use
their profiles appear at the top of the list when po-
multiple platforms in an attempt to find a sufficient
tential clients search for workers (United Kingdom,
amount of work. Workers in developed countries
Department for Business, Energy and Industrial
are more likely to use multiple platforms (52 per
Strategy 2018a; see section 2.3). However, these
cent) compared to those in developing countries
practices can also prove to be a serious entry
(44 per cent). This may be due to limited financial
barrier, particularly for workers from developing
means among workers in developing countries to
countries who have limited financial capacity for
pay platform fees or subscriptions. In the case of
such investments, and illustrate a context wherein
freelance platforms, the majority of respondents
skills do not necessarily define access to work.
use only one platform (59 per cent), which may be
due to the cost of building their profile, reputation There is the option to upgrade and become
or qualifications and establishing a client base a premium member, which provides more
across multiple platforms. benefits to freelancers – Female respondent
on freelance platform Upwork (Albania)
Working on different sites has made it pos-
sible to earn an income that is greater than Workers from certain developing countries in
what traditional companies offer – Female particular face barriers to accessing tasks due to
respondent on freelance platform Text (Ukraine) exclusion by the platform or by the clients based
on nationality or language spoken (Rani and Furrer,
Freelance platforms often do not allow workers to
forthcoming; Graham, Hjorth and Lehdonvirta
accept work off-platform from clients they meet
2017; Beerepoot and Lambregts 2015; see sec-
on platforms, thus limiting their ability to create
tion 4.5). There is also a perception that associates
a client base (Green et al. 2018). For instance, a
low quality of work with workers from developing
majority of respondents reported that platforms
countries, which leads to them being less likely to
attempted to restrict them from working with
receive well-paid tasks (Galperin and Greppi 2017;
clients off-platform (69 per cent), or that they had
Lehdonvirta et al. 2014). To circumvent these bar-
never worked off-platform with the clients met on
riers, workers have adopted mechanisms such as
the platform (74 per cent). While some platforms
using virtual private networks (VPNs) or remote
offer the option of paying an additional fee to
desktop computers (RDCs) to mask their location
work with clients off-platform (see section 2.5),
(see box 4.1). “Glitches” or inefficiencies on plat-
workers may face repercussions (for example
forms, low or inconsistent internet speeds and
being blocked from the platform) for undertaking
high costs of internet access in some countries are
off-platform work without informing the platform
other barriers for workers, especially when clients
and paying the corresponding fees.
request large files (such as videos) or when tasks
have to be performed in a short time (Galpaya and
Senanayake 2018; Berg et al. 2018).
Box 4.1 Circumventing geographical barriers to accessing work

X
There is a widely held belief that a worker with the requisite skills can perform tasks from any
place and at any time around the world on online web-based platforms. However, several plat-
forms do not allow workers to open an account if they reside in certain developing countries. Even
when accounts are created, challenges persist in accessing work, as these workers are not able to
perform the better-paid assignments that are often reserved for workers in developed countries.
To circumvent the geographical barriers introduced by the platforms, some workers stated that
they resorted to “innovative” means, such as using VPNs and RDCs to mask their real location
and instead provide the most favourable one. Respondents reported that by using a VPN they
could open accounts on the platforms, clear the qualification tests, which they would otherwise
fail, and also access better-quality and higher-paying assignments. For example, one respondent
who had an undergraduate degree in economics and statistics noted that using Connecticut,
United States, as a location instead of her own country allowed her access to academic writing
assignments, and to an average income of about US$490 per month. The respondent said,
“I think they [clients] believe that I am in the United States because otherwise I don’t think they
would give me work.”
Some workers also used VPNs to create or open accounts to sell to other workers through
informal online groups such as Facebook or WhatsApp. The price of such accounts was deter-
mined by their quality, which included ratings, client reviews and so on. Faced with exclusion in
accessing work due to platform design, many workers in developing countries have no option
but to purchase accounts through such informal groups. Several such accounts were created
using false documentation and information, raising legal and ethical concerns. They could some-
times be deactivated by the platform without notice. In this situation, the workers who bought
those accounts had to bear the entire loss and also forgo their access to work.
Source: ILO interviews with workers in developing countries.
Sometimes I may not be able to complete developing countries, where the platform algo-
work because of a power blackout. So now- rithms did not prioritize them owing to their
adays some clients prefer to hire people low ratings (Galpaya and Senanayake 2018). On
from places that don’t experience black- microtask platforms, new entrants may have to
outs – Female respondent on freelance platform complete unpaid tasks to demonstrate skills or
Upwork (Kenya) earn qualification and face competition from ex-
perienced workers who may use tools and scripts
Accessing a task often requires demonstrating
on some platforms that notify them about the
skills, particularly through ratings from pre-
availability of tasks so that they can take them
vious customers, test scores and work history,
rapidly (Hanrahan et al. 2019).
and algorithmic matching can play a crucial
role here. About 79 per cent of respondents on On freelance platforms, experienced workers or
freelance platforms reported that reviews and those with higher spending capacities may have
feedback from previous clients were key factors an advantage in accessing work. On Upwork, for
in obtaining new work; other factors included example, some experienced workers or those
previous work portfolios (54 per cent), and the with high ratings can remove or hide a certain
number of completed jobs (53 per cent). This can number of low ratings, while others, including new
create challenges, particularly for new entrants, entrants, may have to pay for arbitration if they do
as was observed on freelance platforms in several not agree with their ratings.
Upwork allows “Top Rated” freelancers to Platform design features affect location-based
remove/hide one review/rating from their workers’ ability to access a sufficient amount
profile. I simply requested to remove a of work through ratings and acceptance rates.
particular low rating – Male respondent on Most app-based workers engaged in the taxi and
freelance platform Upwork (Pakistan) delivery sectors use only one platform to access
work (about 85 per cent each), and the rate is
Lack of sufficient work was already a concern in
almost 100 per cent in Indonesia and Morocco.
2019 for many workers on location-based plat-
A significant reason for this could be the need to
forms, a situation which has been exacerbated
maintain their ratings, which allow them to access
for many during the COVID-19 pandemic (see
work and obtain bonuses (see sections 4.2.2 and
box 4.2). In 2019, about 69 per cent of respondents
4.3.2). A high proportion of respondents in the
in the app-based delivery sector reported that
app-based taxi and delivery sectors (over 70
they would like to complete a greater number of
and 60 per cent respectively) in most countries
deliveries; these workers indicated that the key
reported that both their ratings and acceptance
reason for not being able to do so was the lack
rates had an impact on the amount of work they
of availability of work (83 per cent). This could
received. Similar proportions reported that their
be due to increased competition and enhanced
acceptance rate had an impact on the amount
labour supply; a majority of respondents reported
of work they received. Maintaining ratings and a
an increase in the number of platform companies
high acceptance rate limits the capacity of workers
(56 per cent) and delivery workers in the area
on location-based platforms to use multiple plat-
(62 per cent).
forms (“multi-homing”), as they can only fulfil one
There are many couriers and therefore order at a time.
I don’t receive many orders – Male respondent
When I decline a request for a ride, my
on app-based delivery platform Rappi (Mexico)
acceptance rate decreases and the app sends
In the traditional delivery sector, workers who fewer trips in the future – Male respondent on
wanted to increase the number of deliveries app-based taxi platform DiDi (Mexico)
(58 per cent) similarly indicated that not enough
Platforms can also have strict rules regarding
work was available (90 per cent), partly due to the
multi-homing. For instance, in China some delivery
competition from app-based delivery workers.
platforms make it mandatory for workers to share
Many app-based and traditional taxi drivers also
their location, take a selfie in their work clothes,
reported an increase in the number of platform
and upload the picture on social media groups to
companies in this sector (about 72 per cent),
demonstrate they are on duty, and workers can
resulting in greater competition for the work
be fined one day’s earnings if they do not respond
available, and some drivers were therefore also
within 15 minutes. Similarly, in India and Lebanon
engaging in both app-based and traditional work
for example, more than 90 per cent of app-based
from time to time.
delivery workers noted that they were either
I used to work for a traditional taxi company given or expected to buy the uniforms and bags
but stopped because demand for traditional necessary for working with the platforms.
companies has dropped – Male These tend to have insignias or
respondent on app-based logos that indicate to cus-
taxi plat form Uber tomers the association of
(Lebanon) the worker with a specific
Box 4.2 COVID-19 impact on availability of and access to work

X
To assess the impact of the COVID-19 pandemic on workers in the taxi and delivery sectors (both
app-based and traditional), rapid-assessment surveys were conducted by the ILO in Chile, India,
Kenya and Mexico. The interviews were conducted by telephone in August 2020 with 182 delivery
workers and 222 taxi drivers who had participated in the 2019 survey (see Appendix 4A). Of these
workers, 14 per cent (56 respondents) had permanently stopped working in their respective
sector. Among these, 32 per cent (18 respondents) had left the sector for reasons related to
COVID-19 (such as work restrictions during lockdown, or fear of the virus), and the remainder had
left for reasons not related to COVID-19 (found another job; dissatisfaction with pay or working
conditions; or, in the case of Chile, low demand and heightened risk related to the social unrest
that started in October 2019).
I stopped working for Ola because the income was very low and they charged a very
high commission and taxes. I was working for 12 hours and still wasn’t able to make my
monthly loan payments – Male respondent previously working on app-based taxi platform
Ola (India)
Boxes 4.4, 4.5 and 4.6 on the impact of COVID-19 on income, social protection and occupa-
tional safety and health draw on findings from the surveys conducted with the remaining 348
respondents who were still working, or planned to return to work, in the taxi (197 respondents)
and delivery (151 respondents) sectors in August 2020. Of these, 68 per cent of respondents
had to take a break from working, while 32 per cent worked throughout the pandemic because
of economic necessity, and often despite feeling very concerned that they were at risk of con-
tracting COVID-19 while at work. For some, the break lasted less than one month (24 per cent),
while for the majority it lasted two or more months (59 per cent). The reasons that led to work
interruption in the taxi and delivery sectors (both app-based and traditional) included lack of
demand, restrictions on movement and fear of contracting the virus, or, in a few cases, the
fact that respondents or their family members had contracted it. At the time of the interviews,
26 per cent (89 respondents) had not yet returned to work but were planning to do so once the
situation would allow it.
Because I’m a foreigner, if I don’t work, I don’t eat, and my family needs me – Male
respondent on app-based delivery platform PedidosYa (Chile)
It is my main source of income and I support my family – Male respondent on app-based
taxi platform Uber (Kenya)
The majority of workers who continued working throughout the pandemic were in the app-
based delivery sector where demand levels were mixed: a few of them reported an increase
(13 per cent) or no change in demand (15 per cent), while many reported a decrease (72 per cent).
Some app-based delivery workers were transporting parcels and purchases from supermarkets
or pharmacies, whereas before the pandemic it was mainly food from restaurants.
Several app-based taxi companies, such as DiDi, Beat and Uber (in Chile and Mexico), have added
delivery of parcels or goods to their services, in addition to passengers, since the pandemic
started. However, both app-based and traditional taxi drivers (76 and 83 per cent respectively)
were still more likely than delivery workers (52 and 65 per cent respectively) to have stopped
working (or to have been forced to stop) at some point. Among the app-based taxi drivers, a vast
majority (89 per cent) reported a decline in demand, which may have been due to restrictions
on movement, limitations on providing their services during the lockdown and the shock to the
tourism industry.
Source: ILO rapid-assessment surveys (2020).
platform, and can also have relevance for ac- engaged on freelance platforms (59 per cent). It
cessing insurance benefits, making it more difficult is also the primary source of income for many
for the worker to multi-home. This often leads to a workers in developing countries (44 per cent),
situation where the worker is accessing the market and especially among women in these countries
solely through one platform or is directly tied to (52 per cent). However, low and unstable incomes
a platform, thereby raising concerns with regard are a concern for some of these workers.
to employment relationships (see section 5.3.10).
I think it’s a great thing that people can make
Sometimes, the team leader makes visits an income in this way, I just wish I had a bit
in the field to check delivery bags and uni- more financial stability – Female respondent
form – Male respondent on app-based delivery on freelance platform Upwork (Greece)
platform Zomato (India)
The vast majority of respondents on competitive
Someone hit me with a motorcycle and I got programming platforms (97 per cent) do not
injured. Toters company did nothing to help rely on competitive programming as a source
me only because I wasn’t wearing their uni- of income. Only 12 per cent of the respondents
form – Male respondent on app-based delivery reported having won any financial prize from
platform Toters (Lebanon) competitions in the past year, and of these, it was
mostly a case of one or two competitions. The
4.2.2 Worker earnings prize amount varies depending on the competi-
tion, and respondents reported participating in
on digital labour platforms competitions where they were offered anything
between a few US dollars and US$10,000.
The heterogeneity of work undertaken through
digital labour platforms – from software devel- Earnings on online web-based platforms are
opment to delivery services – is also reflected in impacted by time spent doing unpaid tasks, and
worker earnings, which vary considerably from vary across different types of platforms. Average
high- to low-skilled work.2 Work on online web- hourly earnings (paid) in a typical week for those
based platforms is the primary source of income engaged in online work is US$4.9 (see table 4.2),
for many, particularly in developing countries, and with the majority of workers (66 per cent) earning
especially for many women. About one third of less than the average. This does not take into
respondents reported that online work was their account the fact that workers spend a lot of time
main source of income (see figure 4.11), and this accessing online work, especially given the over-
proportion was particularly high among those supply of workers on some platforms (Graham et
Figure 4.11 Share of respondents who identify online work as their primary source of income,
60
59
50
40 44
30 36
32 32
29 29 30
20
22 22
10
3
0
Freelance
Competitive
programming
Developed
countries
Developing
countries
China
Ukraine
Male
Female
Total (without
China, Ukraine)
Total
Microtask
2 The earnings analysis in this section excludes competitive programmers who earn prizes.
Table 4.2 Hourly earnings on online web-based platforms, by type of platform,

X
development status and sex (in US$)
Total hourly earnings

Paid hourly earnings
(paid and unpaid)
Number of Number of
Mean Median Mean Median
observations observations
Freelance 11.2 7.2 296 7.6 5.3 297
Microtask 4.4 3.0 2026 3.3 2.2 2021
Developed countries 6.1 4.5 1413 4.5 3.3 1405
Developing countries 4.1 2.0 908 2.8 1.4 913
China 4.2 2.9 1 036 2.7 1.8 1 039
Ukraine 4.7 2.3 651 3.2 1.4 652
Male 5.0 3.0 2 451 3.5 2.2 2 452
Female 4.8 2.9 1 557 3.4 1.9 1 556
Total (without China

5.3 3.3 2 322 3.9 2.5 2 318
and Ukraine)
Total 4.9 3.0 4 009 3.4 2.1 4 009
Note: Data is trimmed at 1 and 99 per cent by sector.

Sources: ILO global surveys of crowdworkers (2017) and workers on freelance platforms (2019–20); ILO surveys of
platform workers in China (2019) and Ukraine (2019).
al. 2017). If the unpaid time (see section 4.2.3) is similar activities in the traditional labour market,
factored in, total average hourly earnings (paid after controlling for basic characteristics (see
and unpaid) decline to US$3.4, with the majority Appendix 4B). Microtask workers earn 64 per cent
of workers (66 per cent) earning less than the less in India, and 81 per cent less in the United
average (see figure 4.12). States than in the traditional labour market.
In both countries, this gap is wider for female
The average hourly earnings (paid) in a typical
workers than for male workers (see figure 4.13).
week are higher for workers on freelance plat-
forms (US$11.2) than those on microtask platforms Major differences exist between the earnings of
(US$4.4). On freelance platforms, when unpaid workers on online web-based platforms in de-
time is accounted for, the total average hourly veloped and developing countries. For workers
earnings (paid and unpaid) drop to US$7.6, in developing countries, as well as those in the
with 64 per cent earning below the average. country-level surveys in China and Ukraine,
On microtask platforms, when unpaid time is average hourly earnings (both paid, as well as paid
accounted for, the total average hourly earnings and unpaid) are lower than for those in developed
(paid and unpaid) decline to US$3.3, with 63 per countries (see table 4.2).
cent earning below the average.
It has been very great so far but it’s sad
Furthermore, a statistical analysis shows that Nigerians don’t get high paying jobs – Male
workers on microtask platforms earn significantly respondent on microtask platform Microworkers
less than their counterparts who undertake (Nigeria)
Figure 4.12 Hourly earnings (paid and unpaid) on online web-based platforms,
by type of platform, development status and sex (in US$)
Total Freelance Microtask

Including freelance, microtask, China, Ukraine
80 80 80
3.4 7.6 3.3
% of respondents
% of respondents
% of respondents
60 60 60
40 40 40
20 20 20
0 0 0
0 5 10 15 20 ≥25 0 5 10 15 20 ≥25 0 5 10 15 20 ≥25
Hourly earnings Hourly earnings Hourly earnings
China Freelance by development status Microtask by development status
80 80 80
2.7 5.5 12.6 2.1 4.0
% of respondents
% of respondents
% of respondents
60 60 60
Developed Developed
40 40 40
Developing Developing
20 20 20
0 0 0
0 5 10 15 20 ≥25 0 5 10 15 20 ≥25
0 5 10 15 20 ≥25
Ukraine Freelance by sex Microtask by sex
80 80 80
3.2 7.4 8.0 3.3 3.3
% of respondents
% of respondents
% of respondents
60 60 60
Male Male
40 40 40
Female Female
20 20 20
0 0 0
0 5 10 15 20 ≥25 0 5 10 15 20 ≥25 0 5 10 15 20 ≥25
Note: Data is trimmed at 1 and 99 per cent by sector. Vertical dashed lines indicate mean values.
Figure 4.13 Hourly earnings of survey respondents on microtask platforms compared to their counterparts
in the traditional labour market, India and the United States, by sex (estimated percentage difference)
Estimated % difference in hourly earnings
India United States

0
Total
–20 Male
–40 Female
–64 –62
–60 –69
–79
–81
–83
–80
–100
Sources: ILO global survey of crowdworkers (2017); India, NSSO Periodic Labor Force Survey, 2017–18;
United States, Census Bureau, Current Population Survey, 2017.
On microtask and freelance platforms the average platforms shows a significant gender pay gap,
hourly earnings (paid and unpaid) for those in de- with women earning 26 per cent less in Ukraine,
veloped countries are also much higher (US$4 and while in China they earn 32 per cent less on plat-
US$12.6 respectively) than for those in developing form 680 (it is not significant on other platforms
countries (US$2.1 and US$5.5 respectively). This in China) (see Appendix 4B).
disparity is quite high on freelance platforms
Competition among workers on online web-based
even after controlling for basic characteristics
platforms, high commission fees and non-pay-
and type of task performed; workers in de-
ment for tasks have implications for earnings. On
veloping countries tend to earn 60 per cent less
freelance platforms there is intense competition
than their counterparts in developed countries
between workers. To increase their exposure and
(see Appendix 4B). The analysis also shows that
build their profiles many respondents reported
workers in developed countries tend to earn more
accepting low-paying work (62 per cent); lowering
when they have a regular client than those with
their bids to get work (60 per cent); or performing
no regular clients. This could indicate that the
tasks for free (13 per cent). Instances of underbid-
higher-paying tasks are being performed by those
ding were also reported in China and Ukraine (44
in developed countries. The lower pay associated
and 20 per cent respectively). Moreover, during
with workers from developing countries may be
the COVID-19 pandemic, there has been an in-
due to the perceptions of clients and platform
crease in the number of workers registered on
design, which prevents these workers from ac-
online web-based platforms (see section 1.3), and
cessing higher-paying tasks (see section 4.5).
many have had to bid below their target rates so
Findings on the gender pay gap on online web- as to secure work (Stephany et al. 2020).
based platforms are mixed. Previous studies have
The problem is competition through
shown that women continue to earn less than men
underbidding, which generates a price war
even in the virtual world (Aleksynska, Shevchuk
culture – Female respondent on freelance plat-
and Strebkov 2021; Liang et al. 2018; Foong et al.
form Upwork (France)
2018; United Kingdom, Department for Business,
Energy and Industrial Strategy 2018b; Adams- Earnings are also affected by the different types
Prassl and Berg 2017). The ILO surveys reveal that of fees (commission or service fees, subscription
when looking at averages, a gender pay gap can fees, bidding charges and so on) charged by plat-
often be observed, but this may not always be sig- forms (see table 4.3). A considerable proportion
nificant when a statistical analysis is undertaken. of respondents on freelance platforms reported
The average hourly earnings (paid and unpaid) paying a per-task service fee, varying between
for female online workers (US$3.4) are slightly 36 per cent (Russian and Ukrainian platforms other
lower than for male (US$3.5), and they are also than Kabanchik) and 65 per cent (Upwork) of re-
lower for women in developed countries (US$4.2) spondents. The amount of commission or service
than for men (US$4.8). In developing countries, fees charged differs across platforms. For instance
however, the average hourly earnings (paid and on Upwork, Freelancer and PeoplePerHour com-
unpaid) for women are higher (US$3.4) than for mission fees vary between 3.5 and 20 per cent (see
men (US$2.6). These higher earnings among also section 2.2.1). Furthermore, a substantial pro-
women in developing countries could be due to portion of respondents (69 per cent) also reported
their higher education levels compared to those of paying a fee to submit and receive proposals on
men (see section 4.1.6), allowing them to perform Upwork, where some are required to buy “con-
better-paid tasks. nects” to bid for projects.
However, controlling for all basic characteristics, I tried paying monthly to get more bids and
the statistical analysis shows that there is no sig- now pay to buy connects to place bids. Ten
nificant difference between the hourly earnings connects cost US$1.50 but some bids need you
(paid and unpaid) on freelance platforms for male to pay six connects, so those don’t last long.
and female respondents, and this holds true in Sometimes I’ve paid US$20 within a week or
both developed and developing countries. The two and not gotten jobs – Female respondent on
analysis of the country-level surveys on freelance freelance platform Upwork (United States)
Table 4.3 Fees paid by respondents on freelance platforms, by platform

X
Other Russian/
Fees paid Freelancer Upwork Kabanchik
Ukrainian platforms
One-time registration 22 4 16 15
Annual registration 24 5 2 8
Premium membership 0 5 – –
Fees to submit/receive proposals 22 69 29 27
Fees to appear first in search 7 10 6 9
Per-task service fees 48 65 42 36
Transaction/payment fees 24 41 14 21
Withdrawal fees 24 61 21 29
Foreign currency exchange 15 24 4 8
Other 0 2 0 0
None 7 0 15 23
Sources: ILO global surveys of workers on freelance platforms (2019–20); and ILO survey of platform workers
in Ukraine (2019).
I am charged by Upwork when I apply for thereby limiting their ability to access more work.
a job, when I get paid for a job, and when Moreover, on freelance platforms where creative
I withdraw my payments – Female respondent contests are common, competitions are designed
on freelance platform Upwork (Philippines) to enable clients to select one of several designs
developed by a number of professionals. As a
In addition, workers also pay fees for withdrawing result, the worker who submits the winning design
money and for changing currency (see table 4.3), is paid, while the work undertaken and submitted
which also impacts earnings. The high transaction by the “losing” individuals is unpaid. A similar situ-
costs have led some workers from developing ation may also be the case for freelancers in the
countries to circumvent the platform itself and in- offline world, although in the case of an agency
stead use social media channels such as LinkedIn the risk would not fall on the worker.
or Facebook to access work and improve their
earnings (see box 4.3). App-based taxi and delivery platforms are the
main source of income for 84 and 90 per cent of
Non-payment for tasks undertaken, including respondents respectively in these sectors. These
unfair rejections, also has an impact on earnings. proportions are slightly higher in the traditional
On microtask platforms for instance, where the sectors (around 92 per cent each). An over-
entire process of allocating and evaluating work whelming majority of women in the app-based taxi
is done through algorithms, unfair rejections are (76 per cent) and delivery sectors (86 per cent) are
common, often without any feedback (see sec- also dependent on this work as their main source
tion 4.3.2). This not only leads to lost earnings for of income.
completed work but also reduces workers’ ratings,
Box 4.3 Overcoming low pay and payment barriers

X
A number of workers from developing countries on online web-based platforms who were interviewed were
concerned about low pay due to high commission fees, withdrawal charges and instances of non-payment
for completed tasks. Workers also faced payment barriers on some platforms due to embargos on online
payment gateways in some countries.
I receive emails that there are opportunities for translation but I can’t accept because there are no
ways to receive payments – Femaleinterviewee(SyrianArabRepublic)
To overcome such situations, many respondents reported that they resort to finding clients directly through
LinkedIn, Facebook and other social media platforms, as well as informal channels. Some were also using
labour platforms that cater to their specific region and that are built in the regional language, thereby
reducing competition from other regions of the world. Directly interacting with clients enabled some workers
to negotiate better pay, use payment options that were not embargoed and develop trust with clients. Yet
this also created challenges, as it can be time-consuming to build connections in order to find work on a
regular basis.
I thought of directly choosing clients. I started on LinkedIn and started by adding people. It takes
more effort but it’s convenient for the clients and we get to personally know each other – Female
interviewee(OccupiedPalestinianTerritory)
Source: ILO interviews.
Hourly earnings (including waiting times) for Platform design and the business model have im-
workers in these sectors vary across countries (see plications for earnings in both the app-based and
figure 4.14). They range between US$1.1 (India) traditional sectors. In the app-based sectors, the
and US$8.2 (Lebanon) for app-based taxi drivers, earnings include bonuses and incentives. The ma-
and between US$0.9 (Ghana) and US$3.5 (Ukraine) jority of respondents across app-based taxi and
for app-based delivery workers. Furthermore, in delivery sectors reported being offered bonuses
all countries the majority of workers earn less (over 76 per cent), as well as receiving them (over
than the average in these sectors. Moreover, 60 per cent). For over 85 per cent of respondents,
incomes of workers on location-based platforms these bonuses form an important part of their
have been severely impacted by the COVID-19 income. Therefore, higher earnings in the app-
pandemic (see box 4.4). based sectors may be related to higher economic
incentives or bonuses provided by platforms.
Earnings in the app-based sectors tend to be
higher than in the traditional sectors, although Offers of bonuses tend to depend on the plat-
the differences vary considerably across countries. form companies and the countries where they
A statistical analysis (see Appendix 4B) controlling operate. For instance, in Morocco only 15 per cent
for basic characteristics shows that app-based taxi of app-based taxi drivers reported being offered
drivers earn between 22 per cent (Ukraine) and bonuses, a majority of whom were engaged with
86 per cent (Ghana) more than their traditional the platform Careem; while in Indonesia nearly
counterparts (see figure 4.15). In the delivery all the app-based taxi drivers (99 per cent) who
sector however, whereas app-based workers earn were engaged with Gojek or Grab were offered
more in Kenya (39 per cent more) and Lebanon bonuses. Where they are available, bonuses have
(25 per cent more) than their traditional counter- created a strong incentive structure through
parts, they earn less in Chile (24 per cent less).3 gamification that encourages workers to work
long hours and with high intensity.
3 India was excluded from the regression analysis for delivery workers as the survey of traditional delivery workers was con-
ducted among dabbawalas (those making traditional lunchbox deliveries) in Mumbai, while the app-based survey was conducted
in Bengaluru and Delhi, which limits the comparability of income figures.
Figure 4.14 Hourly earnings in the taxi and delivery sectors, by country (in US$)
Chile Ghana India

50 50
50
3.7 7.2 1.7 3.4 0.6 1.1
% of respondents
% of respondents
% of respondents
40 40
40
30 30
30
20 20
20
10 10
10
0 0 0
0 5 10 15 0 2 4 6 8 10 0 1 2 3 4
Indonesia Kenya Lebanon
50
50
50
1.4 1.7 1.6 2.0 3.7 8.2
% of respondents
% of respondents
% of respondents
40
40
40
30
30
30
20
20
20
10
10
10
0 0 0
0 2 4 6 0 2 4 6 0 5 10 15
Mexico Morocco Ukraine
50
50
50
2.9 5.3 1.9 2.9 4.9 5.8

% of respondents
% of respondents
% of respondents
40
40
40
30
30
30
20
20
20
10
10
10
0 0 0
0 5 10 15
0 2 4 6 8 10
0 5 10 15
App-based taxi Traditional taxi

Argentina Chile China

60 60
60
2.3 3.4 5.1 3.2

% of respondents
% of respondents
% of respondents
40 40
40
20
20
20
0 0 0
0 2 4 6 0 5 10 15
0 2 4 6 8
Ghana India Indonesia
60
60
60
0.9 1.0 1.1 1.2

% of respondents
% of respondents
% of respondents
40
40
40
20
20
20
0 0 0
0 1 2 3 0 5 1 1.5 2 2.5 0 1 2 3
4
Kenya Lebanon Mexico
60 60
60
1.1 1.5 2.4 3.4 2.8

% of respondents
% of respondents
% of respondents
40
40
40
20
20
20
0 0 0
0 1 2 3 4 0 2 4 6 8 0 2 4 6 8
Morocco Ukraine
60
60
1.6 3.5
% of respondents
% of respondents
App-based delivery
40
40
20
20
0 0
0 1 2 3 4 0 2 4 6 8 10
Hourly earnings Hourly earnings
Notes: Data is trimmed at 1 and 99 per cent by sector and country. Vertical dashed lines indicate mean values.

Box4.4 COVID‑19impactonincome
X
In the countries considered in the COVID-19

rapid-assessment survey, and among workers
who were active at the time of the survey, nine
out of ten app-based and traditional taxi drivers
and a varying number of delivery workers (45 per cent
in Chile and 85 per cent in Kenya) reported a decrease in
income since the start of the pandemic. In most instances this was due to decelerating
demand, although in the delivery sector some reported an increase in demand (see
box 4.2).
I work fewer hours than before the pandemic because there are no
trips available. As I didn’t work for two months I finished my savings –
Male respondent on app-based taxi platform InDriver (Mexico)
To compensate for the loss of income, some app-based respondents also reported
providing additional taxi (31 per cent) or delivery (14 per cent) services through private
contacts in addition to undertaking app-based work. For most, this was a small supplement
to their income earned through app-based work, while in some instances it exceeded their
usual earnings. For example, one app-based driver in India reported that while he was not
driving for Uber during the lockdown, he was able to drive passengers from Delhi to their
home villages, thereby earning more than five times the usual monthly amount via the app.
Some app-based workers also reported doing other jobs. Some had started a new job
(7 per cent) or had carried out small tasks (4 per cent) since the beginning of the pan-
demic, while others had continued working in an existing job (8 per cent) or continued
to carry out small tasks to earn extra money (3 per cent). In addition, 13 per cent of the
workers reported that they used to have another job that they had lost since the start of
the pandemic.
For over 90 per cent of respondents in the taxi sector and over 70 per cent of respondents
in the delivery sector (both app-based and traditional), the pandemic has had consequences
for the financial situation of their household. To manage their reduced financial capacity,
79 per cent of workers reported that they had reduced unnecessary expenditure;
65 per cent had used their savings; 48 per cent had deferred payment of bills; while
29 per cent had taken a loan from friends, family or neighbours, 13 per cent from
a bank and 4 per cent from the app company. Some had also moved apartments
or migrated back to their home village (11 per cent); sold possessions (10 per cent);
started farming or keeping animals (7 per cent); or had taken other measures to
reduce expenditures or earn extra income (7 per cent). In particular, for 43 per cent
of app-based taxi drivers, loan repayments for their vehicles were also pending. While
about half were able to reach an agreement to defer the payments in full or in part, the
other half were still obliged to pay the same premiums as before the pandemic.
About one third of respondents across sectors received some form of aid from the
government or their community. In Chile and India, a majority of respondents received
in-kind support from the government and some also received financial support. Only a few
respondents (9 per cent) received any financial or in-kind support from the app or company
that they worked with, with some exceptions such as traditional delivery workers in Kenya
(53 per cent) and app-based drivers in India (26 per cent).
Figure 4.15 Hourly earnings of app-based workers compared to their traditional counterparts
in the taxi and delivery sectors, by country (estimated percentage difference)


100 100
80 86 80
79 78
73 72
60 60
40 48 40
34 39
20 26 20 25
22
0 0
Ukraine
Chile
Ghana
India
Mexico
Morocco
Indonesia
Kenya
Lebanon
-24
-20
Chile Kenya Lebanon
Location-based platforms, particularly in the taxi expected demand and lead to intense competition.
sector, are able to provide services to customers According to 43 per cent of app-based taxi drivers,
at low cost by providing bonuses and incentives it is becoming harder to qualify for bonuses over
to the workers (see section 2.3). This could po- time due to changes introduced by the plat-
tentially distort local labour markets and reduce forms. The number of workers affected is
income-generating opportunities for those in the particularly high in certain countries:
traditional sectors. Across countries, about half India is one case in point, where
the traditional taxi drivers reported that since 84 per cent reported that
they had started working the number of trips qualifying for bonuses
made in a typical day and their daily earnings had has become increas-
decreased, while about one third said that these ingly difficult.
had remained stable. In some countries (Chile,
The changes to
India and Mexico), as many as 70 per cent of the

the bonus incen-
drivers reported a decrease in the number of trips
tive scheme are
and daily earnings.

burdensome for
In addition, traditional taxi drivers often reported drivers and make us
longer waiting periods between rides compared more tired if we
to app-based taxi drivers. For instance, in India work all day – Male

traditional taxi drivers reported that they had to respondent on app-
wait on average of 93 minutes between two rides, based taxi platform

while for app-based taxi drivers the wait was only Gojek (Indonesia)

16 minutes.
Initially it was good to join Ola but now the
The bonuses offered by platform companies have bonuses are reduced, as are the earnings –
attracted a large number of workers, thereby Male respondent on app-based taxi platform
increasing labour supply, which can exceed the Ola (India)
Apart from bonuses, the incomes of app-based (69 per cent), a large majority of them (70 per cent)
workers are also affected by commission fees have taken out a loan for the purchase. In some
charged by the platforms, especially for taxi instances, loans have been acquired from plat-
drivers, or additional expenses for transferring form companies. This may lead to a lock-in of
money. The commission fees can vary within a workers with the platform that sponsors the loan,
platform and across countries (see table 4.4). For and can cause financial problems as rates might
instance, Uber’s commission fees vary between be reduced over time.
5 per cent (Kenya) and 25 per cent in a number of
I took a loan for this car because at the start
countries. Even within a country, the commission
we used to earn a lot. But then Ola and Uber
fee can vary depending upon the vehicle and the
reduced their rates and now I can’t make my
distance, ranging between 5 and 40 per cent.
monthly loan repayments – Male respondent on
Uber’s commission is high so I don’t get app-based taxi platform Uber (India)
to earn much after deductions – Male
Furthermore, limited access to insurance in the
respondent on app-based taxi plat form
event of an accident (see section 4.2.4) or damage
Uber (Ghana)
to the vehicle, or faults in the vehicle or equipment
Other factors impacting worker earnings include hosting the application can also lead to additional
payment for vehicle loans or rent. While many financial burdens for workers and loss of access
of the app-based taxi drivers own their vehicle to work.
Table 4.4 Commission fees paid by app–based taxi drivers,

X
by country and platform (percentages)
Country Uber Beat Bolt Cabify Careem DiDi Gojek Grab Ola
25 25 25
Chile – – – – – –
(18–35) (20–30) (15–25)
25 15
Ghana – – – – – – –
(15–25) (10–25)
20 20
India – – – – – – –
(15–44) (15–40)
20 20
Indonesia – – – – – – –
(10–33) (5–40)
5 20
Kenya – – – – – – –
(5–25) (5–22)
20
Lebanon 25 – – – – – – –
(15–25)
25 25 15 10
Mexico – – – – –
(10–37) (15–28) (12–20) (9–30)
25
Morocco – – – – – – – –
(10–40)
25 15
Ukraine – – – – – – –
(5–35) (10–40)
Note: The figures shown are the commission fees (at the time of research) that were mentioned most often by
respondents per country and app, followed by the range of answers in parentheses.
I find that there are costs associated with bonus rates compared to part-time workers on
the maintenance of the car that are not con- these platforms. About 19 per cent of respond-
sidered and the rates are very low, so it does ents on Glovo in Ukraine reported that they were
not allow us to cover them – Male respondent regular employees, 14 per cent reported that they
on app-based taxi platform Beat (Chile) were temporary workers, and 67 per cent could
be classified under self-employed or independent
In addition, if clients cancel or return orders, the
contractors. In Argentina, PedidosYa used to have
workers may have to pay out of pocket, or cover
full-time employees but now hires independent
long distances without additional pay. About
contractors, especially with increasing competi-
70 per cent of the app-based delivery workers
tion from other platforms (López Mourelo 2020).
reported at least one cancellation in a typical
week. In the event of last-minute cancellations, The minimum income guarantee will be
workers reported that they might have to return paid if I log in for 12 hours, six days a week
the order to the platform office (50 per cent), or to and complete 60 orders, without holidays
the restaurant or firm (42 per cent), or pay out of on weekends. Then I receive 6,000 rupees
pocket (7 per cent), or that they might be able to (about US$85) per week – Male respondent on
keep it free of charge (3 per cent). app-based delivery platform Zomato (India)
Sometimes when an order is cancelled you A gender pay gap exists in a few countries on
have to return to give back the products location-based platforms. While few women
and the time and expense of gasoline are are engaging in the app-based taxi and delivery
not returned – Male respondent on app-based sectors (see section 4.1.2), gender pay gaps can
delivery platform PedidosYa (Chile) be observed in some countries where adequate
data is available. In the app-based delivery sector,
For workers in the app-based delivery sector,
a significant gender pay gap can be observed in
the nature of the contracts offered by the plat-
Argentina and Chile, where women earn about
forms also has implications for their earnings.
17 per cent less than men, while there is no
Depending on the platform, some may receive
significant gender gap in earnings in Ukraine
a regular income while others may be more
(see Appendix 4B). Gender pay gaps are also
dependent on bonuses or the number of orders,
observed in some developed countries. A study
making their incomes more unpredictable and
in the app-based taxi sector in the United States
volatile. For instance, in India delivery platforms
based on data collected from over a million Uber
may have workers on a full-time (working 10 to
drivers finds that men earn around 7 per cent
12 hours a day) or part-time basis (working
more than women. This gender gap is attributed
4 hours a day), temporary shifts (2 or 3 days
to differences in experience, preferences for loca-
of work per week) or a per delivery basis. On
tion of work and driving speed (Cook et al. 2018).
Swiggy and Zomato, a substantial proportion of
Women’s ability to access more work might also
respondents were working on a full-time basis
be restricted as they are cautious and less likely
(74 and 96 per cent respectively). These full-time
to work during night hours for fear of experi-
workers have a “minimum income guarantee”,
encing discrimination, harassment or violence
which means that they receive a guaranteed
(see section 4.5).
income if they complete a minimum number of
hours and orders. They may also have higher
For every hour of paid

tasks, workers spend about
4.2.3 Working hours
and work–life balance 23 minutes on freelance
Flexibility in setting one’s work schedules in an
platforms and 20 minutes
effort to attain work–life balance is among the on microtask platforms
major reasons for choosing to work on platforms.
ILO survey findings reveal that working hours doing unpaid work.
can vary considerably across different types of
platforms, with serious implications for work–life
platforms (42 per cent) and competitive program-
balance.
ming (23 per cent), where they work on average
Workers on online web-based platforms spend a 22 and 27 hours a week, respectively.
lot of time doing unpaid work. On average, they
I used to work 40 hours a week in a business
work 27 hours in a typical week, including both
and 20 hours a week in freelance – Male inter-
paid and unpaid work (see figure 4.16), with about
viewee (Colombia)
one third of their time, or eight hours, spent on
unpaid work. For every hour of paid tasks, workers Workers in developing countries work longer
spend about 23 minutes on freelance platforms hours on online web-based platforms. On average,
and 20 minutes on microtask platforms doing respondents in developing countries work much
unpaid work. While half the workers on online more (32 hours) in a typical week (including paid
web-based platforms work for 20 hours a week and unpaid work) than their counterparts in devel-
or less, one out of five respondents reported oped countries (20 hours). This may be ascribed to
working over 40 hours per week. On freelance their greater reliance on online work as a primary
platforms, the average number of hours worked source of income (see section 4.2.2). In addition,
was 30 hours, with almost a quarter working for workers in developing countries spend more time
more than 40 hours. Average working hours are a week doing unpaid work (9 hours) than those in
comparatively low on competitive programming developed countries (5 hours). This discrepancy
(18 hours) and microtask platforms (24 hours). could be due to the restrictions that workers from
However, while half of the workers on microtask developing countries can encounter on the plat-
platforms work 16 hours or less, 20 per cent of forms, such as exclusion from certain tasks (see
those surveyed were working over 40 hours, indi- section 4.5).
cating considerable variations.
Some workers on online web-based platforms
Workers on online web-based platforms spend have unpredictable work schedules and unsocial
their time doing both online work and other paid hours, particularly in developing countries. While
jobs. About half of online web-based platform the platforms promote flexibility and freedom
workers have other paid jobs, which primarily to work at any time, the ILO surveys show that
includes salaried employees (45 per cent), em- workers in fact face limitations in choosing their
ployees working for an hourly or daily wage work schedules. On freelance platforms, about
(23 per cent), and freelancers (21 per cent). They 82 per cent of respondents indicated that clients
spend 28 hours on average in a typical week in request them to be available outside normal
those jobs. In addition, they work on average for hours either sometimes or regularly (see also
26 hours on online platforms, which can make section 4.3.1). Such requests from clients are more
their working week long. The need for such long common in developing (85 per cent) than in devel-
working hours is indicative of low pay in both the oped countries (76 per cent). About 63 per cent of
online and offline labour markets. On microtask respondents in China and 31 per cent in Ukraine
platforms, the majority of workers are engaged in reported that they receive such requests. This
other paid jobs (52 per cent) and on average work may be due to the fact that most clients tend to
30 hours in that job. The proportion of workers be based in developed countries (see section 1.3),
engaged in other paid jobs is lower on freelance with associated time differences.
Figure 4.16 Hours worked in a typical week (paid and unpaid),

Total Competitive programming

40 40
27 18
% of respondents
% of respondents
30 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80≥90 0 10 20 30 40 50 60 70 80≥90
Working hours Working hours
China Ukraine
40 40
26 36
% of respondents
% of respondents
30 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80≥90 0 10 20 30 40 50 60 70 80≥90
Freelance Microtask
40 40
30 24
% of respondents
% of respondents
30 30
20 20
10 10
0 0
0 10 20 30 40 50 60 70 80≥90 0 10 20 30 40 50 60 70 80≥90
Freelance by development Microtask by development
40 40
23 33 20 33
% of respondents
% of respondents
30 30
Developed Developed
20 20
Developing Developing
10 10
0 0
0 10 20 30 40 50 60 70 80≥90 0 10 20 30 40 50 60 70 80≥90
Freelance by sex Microtask by sex
40 40
% of respondents
29 30
% of respondents
22 26
30 30
Male Male
20 20
Female Female
10 10
0 0
0 10 20 30 40 50 60 70 80≥90 0 10 20 30 40 50 60 70 80≥90
Note: Vertical dashed lines indicate mean values.

I am a full-time mother during the day In some countries, the weekly average is as high
and full-time freelancer during the night. as 82 hours (India, app-based taxi) and 63 hours
My “night” is the US “day” so, besides that per week (Kenya and Lebanon, app-based de-
I need some sleep, everything works well for livery) (see figure 4.17).
me – Female respondent on freelance platform
We have to work for 12 hours and sometimes
Upwork (Romania)
more but we don’t get paid accordingly –
Similarly, on microtask platforms the majority Male respondent on app-based taxi platform
of workers in developing countries (53 per cent) Ola (India)
work during the night (10 p.m. to 5 a.m.); they have
A sizeable proportion of workers in both the app-
to adapt to the temporal distribution of jobs, as
based taxi (41 per cent) and delivery (38 per cent)
work is often posted during US business hours –
sectors work seven days a week. About 28 per cent
evening or night time for workers in Africa and
of respondents in the app-based taxi sector re-
Asia (Rani and Furrer, forthcoming; O’Neill 2018).
ported working for over 12 hours, while half the
Most workers in the taxi and delivery sectors respondents in the app-based delivery sector
work long hours and have high work intensity. reported working for over 10 hours on three or
Long working hours in the transpor tation more days per week. Through gamification, plat-
sector, particularly in traditional taxi services forms have created opportunities to access higher
(Gwilliam 2005), have been an enduring concern earnings or bonuses, which incentivize workers to
in developing countries. This situation has also work long hours. This may also lead to high work
penetrated app-based platforms. While in the intensity, as workers often do not take breaks in
traditional taxi and delivery sectors average order to be able to meet their targets or due to the
working hours are 70 and 57 hours per week fear of losing a client or an order. On some plat-
respectively, they are 65 and 59 hours respect- forms, workers’ break times are controlled by the
ively in the app-based sectors, with over half of algorithm, sometimes to the second, and workers
respondents working more than these averages. can be fined for staying offline for too long.
Figure 4.17 Hours worked in a typical week in the taxi and delivery sectors, by country
Chile Ghana India

40
40
40
46 67 64 78 81 82
% of respondents
% of respondents
% of respondents
30
30
30
20
20
20
10
10
10
0 0 0
0 20 40 60 80 ≥100 0 20 40 60 80 ≥100 0 20 40 60 80 ≥100
Working hours Working hours Working hours
Indonesia Kenya Lebanon
40
40
40
67 69
65 81 66 69
% of respondents
% of respondents
% of respondents
30
30
30
20
20
20
10
10
10
0 0 0
0 20 40 60 80 ≥100 0 20 40 60 80 ≥100 0 20 40 60 80 ≥100
Mexico Morocco Ukraine

40
40
40
60 75 60 67 41 49
% of respondents
% of respondents
% of respondents
30
30
30
20
20
20
10
10
10
0 0 0
0 20 40 60 80 ≥100 0 20 40 60 80 ≥100 0 20 40 60 80 ≥100
App-based taxi Traditional taxi

Argentina Chile China
60 52 60
40 61 60
61
% of respondents
% of respondents
% of respondents
40 40
40
20
20
20
0 0 0
0 20 40 60 80 ≥100 0 20 40 60 80 ≥100 0 20 40 60 80 ≥100
Ghana India Indonesia
60
61 60
50 72 60
60
% of respondents
% of respondents
% of respondents
40
40
40
20
20
20
0 0 0
0 20 40 60 80 ≥100
0 20 40 60 80 ≥100 0 20 40 60 80 ≥100
Working hours
Kenya Lebanon Mexico
60
56 63 60
63 71 60
51
% of respondents
% of respondents
% of respondents
40
40
40
20
20
20
0 0 0
0 20 40 60 80 ≥100
0 20 40 60 80 ≥100 0 20 40 60 80 ≥100
Working hours
Morocco Ukraine
60
55 60
40 App-based delivery
% of respondents
% of respondents
40
40
20 20
0 0
0 20 40 60 80 ≥100
0 20 40 60 80 ≥100
Working hours
Working hours
Note: Vertical lines indicate mean values.

4.2.4 Occupational safety working hours, the risk of work-related injury and
pressure to drive quickly (see figure 4.18). Worker
and health movement is closely monitored by the platforms
and can be tracked by the clients in real time,
On online web-based platforms the need to work
which further increases the pressure to reach
unsocial hours not only limits platform workers’
destinations faster so as to ensure future orders
ability to be flexible in choosing their own working
or rides, which can be cancelled for even slight
times but also has an impact on their work–life
delays. This can also have severe implications for
balance, and at times can also lead to social
workers’ occupational safety and health, notably
isolation (Anwar and Graham 2020; Dedeoglu,
since workers often do not have access to social
forthcoming). The survey of freelance platforms
protection coverage (see section 4.2.5) and can
from Ukraine shows that for many respondents,
put the customer’s safety at risk as well.
work–life balance did not improve, or stayed the
same, compared to their previous job situation I crashed once when I worked for 48 hours
(61 per cent). In addition, 23 per cent reported straight – Male respondent on app-based taxi
that they were either often or always stressed, platform Cabify (Chile)
and the majority (58 per cent) were worried about
Clients refuse to take the order if there are
having enough work in the future. Such situations
delays – Male respondent on app-based delivery
have implications for the health of these workers,
platform Deliveroo (Kenya)
many of whom are already confronted with long
working hours in front of a screen. Workers in the app-based taxi and delivery
sectors, particularly women, also face several oc-
Freelancing work is done at night so this is
cupational safety and health risks. About 83 per
a big problem. I am a hard worker and work
cent of workers engaged in the app-based taxi
over 16 hours per day. This has an impact
sector and 89 per cent in the app-based delivery
on my health and mental stress. My family
sector reported having safety concerns about
depends on my income – Male respondent on
their work, often relating to road safety, theft
freelance platform Upwork (Bangladesh)
and physical assault (see figure 4.19). In countries
On app-based taxi and delivery platforms, a high with women in the sample, a larger proportion of
proportion of respondents (79 and 74 per cent re- women than men were concerned about physical
spectively) reported feeling stressed by their work assault in the app-based taxi sector, indicating
and working conditions. This is often due to traffic that safety concerns, along with violence and har-
congestion, insufficient payment, lack of jobs, long assment at work, remain issues to be addressed.
Figure 4.18 Main reasons for stress in the app-based taxi and delivery sectors
70
68
60
50 52 App-based taxi
40 44 43
30 36 34 App-based delivery
32 31 33
29 27 29 29 29
20 26
10 15
5 5
0
Insufficient
Working shifts
are too long
Traffic
Long waiting
periods
payment
congestion
Other
Risk of work-
related injury
Time pressure
to drive quickly
Insufficient number
of rides/orders
Risk of crime
Note: Figures presented refer to workers who reported their work-related stress level to be 5 or more on a scale from 0 to 10.

Figure 4.19 Main concerns regarding personal and physical safety

in the app-based taxi and delivery sectors
100
80 88
App-based
74 taxi
60
53 50 50 App-based
40 47
41 delivery
37
20
5 3
Road safety Weather conditions Theft Physical assault Other
Note: Figures presented refer to workers who reported their work-related personal and physical safety to be 8 or lower
on a scale from 0 to 10, where 0 refers to not safe at all and 10 refers to being completely safe.
I was involved in a serious harassment case: platforms did not take any measures to prevent
A man started to watch a porn movie on his workplace risks, which affected their ability to
cell phone and then he invited me to see work and earn an income.
the film together, while offering money –
I had an accident. I want Swiggy to take
Female respondent on app-based taxi platform
responsibility and give me support. No one
Beat (Chile)
picks up calls in the call centre. They should
I was hit by another vehicle. I was admitted help me in emergency situations – Male
to the hospital and hence was not able to respondent on app-based delivery platform
work for two weeks – Male respondent on app- Swiggy (India)
based taxi platform Bolt (Ghana)
In addition, the occupational safety and health
About 10 per cent of workers engaged in the app- risks associated with COVID-19 are further ex-
based taxi sector and 21 per cent in the app-based acerbating the threats to workers’ well-being,
delivery sector reported having experienced a particularly for those in the taxi and delivery
work-related injury or an accident. These pro- sectors (see box 4.5). In the United Kingdom, the
portions were high in Morocco (34 per cent of occupation of taxi or cab driver, or chauffeur, was
app-based taxi drivers), and Mexico (47 per cent among those with the highest rates of death due
of app-based delivery workers). A majority of to COVID-19 (United Kingdom, Office for National
respondents (over 80 per cent) reported that the Statistics 2021).
Box 4.5 COVID-19 impact on occupational safety and health

X
A majority of app-based workers (71 per cent) who were working at the time of
the survey (see box 4.2) reported that the platforms had introduced measures
to reduce health-related risks during the COVID-19 pandemic, although the
proportions varied depending on the country and the platform. Among app-
based taxi drivers, they ranged from 24 per cent (Kenya) to 81 per cent (India),
and among delivery workers, from 48 per cent (Chile) to 92 per cent (Kenya). Such
measures included compulsory mask wearing, contactless and cashless delivery, limiting
the number of passengers, and sanitizing hands, equipment and vehicles, among others. Some
of the taxi platforms also required that the car be equipped with a separation between driver
and passengers, or offered to make the necessary modifications. While in certain cases this was
provided free of cost, some companies charged the worker for the material.
I have to log in and upload a photo two or three times a day, while wearing a mask.
I also have to have the contact tracing app running – Male respondent on app-based taxi
platform Uber (India)
Wearing masks, always carrying sanitizer, wearing gloves and keeping distance while
delivering – Male respondent on app-based delivery platform Jumia (Kenya)
About half the app-based taxi drivers and 81 per cent of the app-based delivery workers further
reported that the app shared information or provided safety training in response to COVID-19,
which was considered useful by the large majority (89 per cent) of the workers who received it.
Platform companies also provided personal protective equipment (PPE) such as masks, gloves or
hand sanitizer to workers, as reported by 31 per cent and 67 per cent of app-based taxi drivers
and delivery workers respectively. Of these, about half the workers reported that the quantity
of PPE was insufficient, and one third reported that the quality was inadequate. Furthermore,
14 per cent of those who were offered this equipment did not manage to use it, and another
11 per cent reported that it was difficult to access the material as it often had to be picked up at
certain locations and during certain times that could be inconvenient for workers.
In order to avoid the crowds at the office where they distributed PPE, I started to buy it
myself – Male respondent on app-based taxi platform DiDi (Mexico)
Every month I get one disposable mask, one 50ml bottle of sanitizer and one pair of
gloves. For one month, one pair of gloves and one mask is not sufficient as they easily
get torn – Male respondent on app-based delivery platform Dunzo (India)
Almost all respondents in both the app-based and traditional taxi and delivery sectors (94 per
cent) also took personal measures to reduce work-related risks of becoming infected with
COVID-19. For 83 per cent of the app-based workers, this implied additional financial expend-
iture, especially when they had to buy face masks, gloves or sanitizer as they were not provided
with any, or received insufficient amounts of, PPE from the platform company.
Furthermore, while a large majority of app-based taxi (88 per cent) and delivery
(96 per cent) workers working at the time of the survey considered them-
selves to be essential workers during the pandemic, many of them

also reported being dissatisfied (24 per cent somewhat dissatisfied,

33 per cent very dissatisfied) with what the platform was doing

to protect them from COVID-19.

4.2.5 Access to social Only a small proportion of workers on online web-

based platforms have social security coverage.
protection Around 40 per cent of respondents reported that
they had health insurance 4 (see table 4.5), with
Social protection, or social security, is a human
small differences by sex. A higher proportion of
right and includes benefits for unemployment,
respondents on microtask platforms (61 per cent)
employment injury, sickness, old age, disability,
reported that they had health insurance, which
survivors and health protection, as well as for
could mean that they were covered through their
maternity, children and families (ILO 2017a). The
main job5 or through their spouse; however, the
organization of work on digital labour platforms
proportions were quite low among respondents
has raised considerable concerns regarding in-
on freelance (16 per cent) and competitive pro-
adequate social protection coverage for workers
gramming platforms (9 per cent). A very small
engaged on such platforms (Behrendt, Nguyen
proportion of respondents (less than 20 per cent)
and Rani 2019; Wood et al. 2019a; OECD 2018).
on online web-based platforms reported being
These concerns have been magnified with the
covered for employment injury, unemployment
COVID-19 pandemic, as many platform workers
and disability insurance, or for old-age pen-
have limited or no access to paid sick leave and
sions or retirement benefits (both public and
sickness benefits or to unemployment benefits
private pension plans), and the coverage is low
(ILO 2020a and 2020b; Ustek-Spilda et al. 2020;
across different types of platforms. There are
McGee 2020; see also box 4.6).
Table 4.5 Proportion of respondents on online web-based platforms covered

X
by social protection benefits, by type of platform, development status and sex
Health Employment Unemployment Disability

Pension
insurance injury insurance insurance
Freelance 16 1 2 2 6
Competitive
9 6 4 2 6
programming
Microtask 61 21 16 13 35
Developed countries 61 17 17 15 35
Developing countries 43 18 9 7 23
China 30 18 6
Ukraine 12 5 5 4 4
Male 42 18 13 12 21
Female 39 11 10 11 18
Has other job 45 17 14 13 26
No other job 37 12 10 10 14
Total (without China,

53 17 14 11 30
Ukraine)
Total 41 15 12 12 20
4 The survey question relates to health insurance, which could be public social health insurance or a private insurance. It should be
noted that ILO social security standards promote social security mechanisms to ensure effective health coverage without financial
hardship through collectively financed mechanisms based on the principle of solidarity, in particular social health insurance and
tax-financed provision.
5 Where social protection coverage is related to a job in the “traditional” economy, concerns arise with regard to app-based plat-
forms free-riding with regard to the financing of social security at the expense of the “traditional” economy, with implications for fair
competition as well as for the equitable and sustainable financing of social protection systems (Behrendt, Nguyen and Rani 2019).
Box 4.6 COVID-19 and social protection

X
Access to paid sick leave and sickness benefits, or unemployment benefits, is quite
limited for online and location-based workers, thereby creating unique risks during
the COVID-19 pandemic. As many workers depend entirely on task-based work for
their earnings, without paid sick leave and sickness benefits (ILO 2020b) they could
not afford to self-quarantine even if COVID-19 symptoms were to appear, posing
risks both to themselves and to others. At the same time, given the healthcare costs
in some countries and the lack of health insurance coverage for platform workers,
being tested and treated for COVID-19 may be challenging.
The ILO COVID-19 rapid-assessment surveys reveal that, of respondents who stayed
in the taxi and delivery sectors (both app-based and traditional; see box 4.2), 8 per
cent reported that either they or a household member had tested positive or had
COVID-19-related symptoms. Although most of them took a break from working,
and some app-based workers informed the app companies about their situation,
only one app-based worker received a one-time financial compensation (US$70
or the equivalent of two days of app-based work). In another case, an app-
based worker who did not have symptoms but lived with someone who
tested positive was sent on unpaid leave and had his account deactivated
one month after informing the app company about his situation.
Some app-based taxi and delivery platforms have undertaken
specific measures to mitigate financial and occupational
safety and health risks among workers (see box 4.6),
for instance, the provision of financial assistance or
support for up to two weeks in the event a worker is
diagnosed with COVID-19. However, about 70 per cent
of app-based workers reported that they would be
unable to take paid sick leave, or receive compensation,
in the event they were to test positive for COVID-19. This
could be due to lack of or limited awareness about such provisions, or poor
implementation, as was also observed in other studies (Fairwork Project 2020).
While the risks of contracting COVID-19 may be lower for workers on online web-
based platforms than for those on location-based platforms, limited access to
health protection nonetheless creates challenges in this instance as well. Workers
on online web-based platforms who show symptoms often find it difficult to access
adequate healthcare because of lack of health coverage, or they need to bear the
healthcare costs themselves. In addition, the lack of sickness benefits discourages
them from taking time off work. Such a situation in turn creates vulnerabilities for
both workers and those dependent on them, while also potentially undermining
public health efforts to curb the virus.
Some governments have introduced temporary measures to cushion the adverse
impact of the pandemic on workers, such as guaranteeing their access to quality
healthcare, enhancing their income security by increasing benefit levels, and
extending coverage to previously uncovered groups of workers through ongoing
or new programmes. While many countries have extended coverage of their ongoing
programmes or schemes to self-employed workers, some countries have specifically
highlighted platform workers as a target group for such measures. For example, in
Ireland sickness benefits have been extended to all workers, and in Finland and the
United States temporary unemployment benefits have been extended to include
workers not covered by unemployment insurance, including self-employed workers
in the platform economy (ILO 2020a; ISSA 2020).
Despite their exposure

no differences with regard to social protection
to high occupational safety
coverage across different age groups, leaving and health risks, only about
both younger and older workers vulnerable to
socio-economic and health-related shocks. 30 per cent of respondents
Workers on online web-based platforms are less in the app-based taxi and
likely to have social protection coverage in de-
veloping countries than in developed countries. delivery sectors reported
A low proportion of workers from developing
countries undertaking online work reported that they were covered
having health insurance (43 per cent), old-age
pension/retirement benefits (23 per cent), un-
for employment injury.
employment protection (9 per cent), disability
insurance (7 per cent) or employment injury
of serious health problems (94 and 80 per cent
protection (18 per cent). Fewer than 10 per cent
respectively), only about half were covered by
of respondents in the country-level surveys in
health insurance.
China and Ukraine reported that they were cov-
ered for old age pension or retirement benefits. We should have an accident insurance and
Such proportions are comparatively higher in social benefits – Male respondent on app-based
developed countries, for instance for health insur- delivery platform iVoy (Mexico)
ance coverage (61 per cent) or old-age pension or
Despite their exposure to high occupational safety
retirement benefits (35 per cent) (see table 4.5).
and health risks, only about 30 per cent of respond-
The findings clearly underscore that inadequate
ents in the app-based taxi and delivery sectors
social protection coverage for workers on online
reported that they were covered for employment
web-based platforms is a concern across both
injury (see table 4.6). Workers in a number of
developing and developed countries, although
developing countries reported having taken out
workers in developed countries have slightly
the relevant private insurance in the absence of
better coverage due to certain well-established
support from the company or adequate protection
institutional structures (ILO 2017a).
through public schemes. For instance, in Mexico
In the countries surveyed, a majority of respond- 70 per cent of app-based taxi drivers reported
ents engaged in the app-based taxi and delivery being covered for employment injury, of whom
sectors do not have social protection coverage. over 91 per cent had private insurance. While some
Only a small proportion of respondents in the of the platform companies, such as Swiggy in India
app-based taxi and delivery sectors are covered or Grab in Indonesia, provide accident insurance
by unemployment protection, disability insurance coverage (SwiggyBytes 2017; Grab 2017), several
(less than 10 per cent) and old-age pensions or respondents who had suffered an accident re-
retirement benefits (both public and private pen- ported that they did not receive any support from
sion plans) (less than 20 per cent) (see table 4.6). these companies. In addition, such measures may
Although a majority of these workers reported also create challenges in terms of portability of
that they had access to healthcare in the event social protection entitlements between jobs.
Table 4.6 Proportion of respondents in the taxi and delivery sectors

X
covered by social protection benefits
Health Employment Unemployment Disability

Pension
insurance injury insurance insurance
App-based taxi 51 27 5 4 18
Traditional taxi 52 23 3 3 14
App-based delivery 53 31 7 6 17
Traditional delivery 40 31 16 4 23

4.3 Worker autonomy and control

under algorithmic management
Platform work is often associated by its pro- to install specific software, or meet certain hard-
ponents with greater worker autonomy and ware and software requirements;6 their working
control over how work is performed (Mulcahy hours are monitored by clients; they are requested
2016, MGI 2016). Concerns are being raised, to be available during specific times by clients;
however, regarding new forms of worker control and they are required to use a monitoring system
resulting in loss of autonomy, facilitated by the for submitting screenshots of the work done
design of platforms and their algorithms (Pichault (see table 4.7). Such mechanisms for monitoring
and McKeown 2019; Wood et al. 2019b; Schorpf, and control tend to be more frequent for platform
Flecker and Schonauer 2017). These algorithms workers from developing countries and women.
rely on data generated by workers on various In some cases platform workers also reported
aspects of work undertaken, and often workers signing non-disclosure agreements preventing
lack any access to or control over their data (see them from working with others off the platform,
section 1.4). This results in information asymmetry which include detailed instructions for tools to
wherein the platform has large amounts of data be used for the work as well as requirements for
on the workers, and the work being undertaken forgoing any claim to intellectual property for
by the worker, while the worker has little infor- the work they performed (Darkwah and Tsikata,
mation about how that data is being utilized by forthcoming).
the platform.
I had to install the desktop app for time
Algorithmic management is defining the everyday tracking pretty intrusive app. Takes screen-
work experience, performance and achievement shots, photos from web cam, mouse clicks
for workers using the data generated by workers and keyboard usage – Male respondent on
while working on the platform (Duggan et al. freelance platform Upwork (Argentina)
2020; Jarrahi et al. 2019; Rosenblat and Stark
2016). It also has implications for how workers Autonomy and control over work is limited for
receive feedback and ratings, resolve disputes, workers in the app-based taxi and delivery sectors.
and navigate payments or non-payments for the A key facet of autonomy and control over work is
work undertaken. related to their ability to choose working hours and
break times, as well as to decline certain orders,
for reasons such as exhaustion or safety concerns.
4.3.1 Autonomy Their schedules and destinations can, however, be
and control over work shaped by their ratings and “levels”, as well as by
other incentive structures of the platforms such
Monitoring of work and determining working as surge pricing. A sizeable proportion of workers
methods are common on online web-based in the app-based taxi (37 per cent) and delivery
platforms. Platforms provide clients and workers (48 per cent) sectors are unable to refuse or cancel
with various tools that are used to communicate, work, as such refusal or cancellation is likely to
manage and monitor work in progress, especially have negative implications for their ratings. This
on online freelance platforms (see section 2.5). The may result in reduced access to work, lost bonuses,
ILO surveys show that a substantial proportion of financial penalties and even deactivation of the
respondents are required by platforms or clients platform worker’s account (see figure 4.20).
6 Hardware requirements relate to the speed of the central processing unit, the speed of the internet connection, webcam, micro-
phone and so on. Software requirements relate to operating systems, software for specific tasks (such as transcription or photo
editing) and time tracking software, among others.
Table 4.7 Monitoring and organizing work on freelance platforms,

X
by development status and sex (percentage of respondents)
Monitoring Availability
Required Required by clients Screenshot required
by platform by client of hours of the work during
worked specific times
Sometimes
Sometimes
Sometimes
Hardware Installation Hardware Installation
Regularly
Regularly
Regularly
or software of specific or software of specific
requirements software requirements software
Freelance 22 41 46 47 47 38 46 37 43 47
Total 17 27 41 48 34 42 34 40 31 51
Developed
Male 17 27 35 44 26 46 30 36 34 48
Female 18 27 47 52 40 39 38 43 29 53
Total 24 48 48 47 53 36 51 36 49 46
Developing
Male 23 47 43 46 52 37 53 34 50 44
Female 28 48 55 49 55 34 47 39 46 50
China – – – – 4 48 – – – –
Ukraine – – – – 7 17 12 28 13 32
Total 22 41 46 47 13 36 24 31 24 38
Sources: ILO global survey of workers on freelance platforms (2019–20); and ILO surveys of platform workers in China (2019)
and Ukraine (2019).
Figure 4.20 Proportion of respondents in the app-based taxi and delivery sectors
that are unable to refuse or cancel work without repercussion, by country
App-based taxi sector App-based delivery sector
80 80
77
70
70
60 60
56 58 57
50 53 50
48 49 48
40 40 45
36 37 39
30 34 30 36
31
27 26
20 20
20
18
10 13 10 15
0 0
Chile
Ghana
India
Indonesia
Kenya
Lebanon
Mexico
Ukraine
Total
Morocco
Argentina
Chile
Ghana
Lebanon
Mexico
Morocco
Ukraine
Total
Indonesia
Kenya
India

If I reject orders I will not be able to choose to decide whether to accept or decline an order;
a shift time that I want – Male respondent on on Uber, drivers receive a request and are given
app-based delivery platform Glovo (Ukraine) between 15 and 40 seconds to decide, based on
limited information.
If I refuse work the acceptance rate de-
creases, therefore the amount of orders
decreases – Female respondent on app-based 4.3.2 Ratings, evaluation
delivery platform Cornershop (Chile)
and dispute resolution
I cancelled three trips and I was deactivated
for one hour – Non-binary respondent on app- Key uses of algorithms on platforms include
based taxi platform DiDi (Mexico) matching workers and clients, evaluating the
work performed and providing ratings (Duggan
In Chile, PedidosYa allows delivery workers to pre-
et al. 2020; Wood et al. 2019b; see also Chapter 2).
book their shifts in advance, and the time slots
Platforms also reject work or “deactivate” workers
for work are selected on the basis of the “level”
if their ratings fall below a certain threshold, and
assigned to each worker, which in turn depends on
such rejections and deactivations are often algo-
the number of hours worked and orders accepted
rithmically managed.
during specific days or time slots. If workers work
fewer hours or reject orders, they may not obtain Ratings play a decisive role in accessing work on
the slot of their choice and they also receive fewer online web-based platforms. A high rating was
orders. On taxi platforms, in order to ensure the reported as a key factor in obtaining new work
required numbers of drivers for their clients during by 83 per cent of the respondents on freelance
peak hours, platform companies use surge pricing, platforms. A high proportion of respondents
which allows them to address supply and demand (87 per cent) in the country-level survey of Ukraine
problems (Liu et al. 2019; Rosenblat and Stark 2016). also reported that their rating was either very
Many workers indicated that they relied on surge or somewhat important to them. Almost half of
prices to increase their incomes. As the fares are the respondents on competitive programming
higher during peak hours in certain zones, they are platforms reported that a high ranking was
incentivized to drive to these zones during specific advantageous in their job search. While higher
times and accept a minimum number of rides to ratings play a role in facilitating access to work,
avail themselves of the benefits. lower ratings can sometimes lead to deactivation
of worker accounts. Some workers on web-based
To increase my income, I go to work to areas
platforms in China (6 per cent) and Ukraine (11 per
with dynamic pricing, work during peak
cent) reported instances of accounts being
hours, avoid heavily-travelled roadways –
deactivated. This can occur without any
Male respondent on app-based taxi platform
explanation provided to
DiDi (Mexico)
the worker, and with little
You cannot refuse more than 10 per cent of opportunity for communi-
the daily orders, otherwise they will down- cation or contestation.
grade you from the list – Male respondent on
app-based delivery platform PedidosYa (Chile)
In both the app-based taxi and delivery sectors
workers are often also closely monitored by the
platform and the clients with the help of GPS sys-
tems, and workers can be contacted at any time
once an order is placed. Such tracking also enables
platforms to define the routes that workers take
to complete orders, and they carefully monitor
the time spent. Moreover, respondents also
reported that they tend to have very little time
Upwork relies heavily on feedback ratings rating or rejection of work by a client, which may
from clients. If you do not receive feedback be unfair or fraudulent but will nevertheless be
from a client or can no longer contact a factored into the algorithms can affect a worker’s
client, Upwork thinks you did not complete overall ratings. In China, 62 per cent of respond-
the job. Therefore your success rating suf- ents on freelance platforms reported exposure to
fers. And unless you are already top rated deception or fraudulent treatment at least a few
you cannot dispute it – Male respondent on times by clients or platforms.
freelance platform Upwork (Philippines)
Rejection of work is common on online web-based
Over 60 per cent of respondents on freelance platforms although not all rejections are justifiable,
platforms, as well as in the country-level survey particularly on microtask platforms. Respondents
of Ukraine, reported that they did not receive any on microtask (86 per cent) and freelance (34 per
form of evaluation other than their rating. This cent) platforms reported having had work rejected
is the case even if they were poorly rated, which by clients, and only a minority reported that the re-
limits their capacity to learn and perform better jections were justifiable (see figure 4.21). The high
in future. A higher proportion of workers from rates of unfair rejections, particularly on microtask
developed countries (68 per cent) and especially platforms, reflect the fact that work tends to be
women (71 per cent) did not receive any form of supervised by algorithms rather than by humans.
work evaluation. These algorithms can be designed in such a way
that they approve tasks done by multiple workers
Worker ratings are influenced by both the clients based on the majority of responses, independently
and the platform’s algorithms. For instance, a low of the correct response, a practice that can lead to
Figure 4.21 Rejection of online work, by type of platform and country

Freelance Microtask
15
Have you ever had work rejected? 22 Have you ever had work rejected?
Yes No 20 Yes No 31
32 47
27 14
66 34 7
86
China 3 Ukraine
Have you ever had work rejected? 16 15
Have you ever had work rejected?
17
Yes No Yes No
48 37
32 31
47 65 35
53
Were the rejections justifiable?

They were all justifiable Most of them were justifiable Some/a few of them were justifiable None of them were justifiable

unfair rejections of work. Moreover, such unfair re- Ratings are critical for most workers in the
jections and the consequent denial of payment for app-based delivery and taxi sectors. A majority
the work can result in lower ratings for the worker, of respondents in the app-based taxi and de-
with implications for future work opportunities, livery sectors reported that their ratings had an
and can also lead to deactivation of the worker’s impact on the amount of work (72 and 65 per cent
account (Berg et al. 2018). respectively) and the type of work (for instance in
terms of earnings or distance: 58 and 47 per cent
The clients didn’t accept the work for reasons
respectively) they received. About one in four
that were not in the requirements initially. In
workers engaged in these sectors believed that
my opinion, they just wanted the work to be
their current rating was not an accurate reflection
done for free – Female respondent on freelance
of how well they performed. This was particularly
platform Upwork (Belarus)
high in Lebanon among app-based taxi drivers
Some requesters reject work randomly (47 per cent) and in India among app-based
without convincing reasons, maybe to get delivery workers (43 per cent). Sometimes these
work done without paying the compensa- ratings were influenced by factors beyond the
tion – Male respondent on microtask platform worker’s control, such as delays in receiving a food
AMT (India) order from a restaurant, or traffic congestion.
About half of the respondents on freelance They lowered my rating and it wasn’t my
platforms reported that they were not aware of fault: the user didn’t want to pay for the
a formal process available to file a complaint or
trip, and they didn’t pay me for the trip. My
seek help (see figure 4.22). Among respondents
account was then suspended for three days –
who were aware of such processes, 31 per cent
Male respondent on app-based taxi platform
reported that they had contested or appealed
Bolt (Mexico)
a rating or evaluation. Of these, 77 per cent re-
ported a favourable outcome, 18 per cent reported A sizeable proportion of workers in the app-based
that their appeal was denied and no change was taxi and delivery sectors are also unaware of any
made to their rating or evaluation, and 5 per cent formal process for filing a complaint or seeking
reported that their rating or evaluation worsened help, which was reported by 42 and 32 per cent
or that they were faced with some form of retribu- of app-based taxi drivers and delivery workers,
tion. A higher proportion of men (79 per cent) than respectively. Moreover, this was also reported
women (73 per cent) had a favourable outcome. by the majority of respondents in both sectors in
Figure 4.22 Knowledge and use of appeal mechanisms on freelance platforms

Outcome of appeal
Favourable
Is there a mechanism to appeal
a rating or evaluation?
77
Denied/no change
48 Negative repercussions
52 Have you ever appealed 18
a rating or evaluation? 5
31
69
Yes No
Source: ILO global survey of workers on freelance platforms (2019–20).

About 19 and 15 per cent

Indonesia and Morocco. As discussed in Chapter 2,
of workers in app-based
dispute resolution mechanisms are often specified taxi and delivery sectors
in the platform’s terms of service agreement.
However, many workers in the app-based taxi respectively reported
(58 per cent) and delivery (49 per cent) sectors
reported that they had not seen their platform’s having experienced
terms and conditions. Of those who had seen the
terms of service agreements applicable to them,
account deactivation.
almost one third reported not having read, not
remembering or not having understood them.
who reported occurrences, deactivation lasted
Some workers in both sectors had complained for up to seven days for the majority in both the
or requested assistance from the platform taxi (73 per cent) and delivery (69 per cent) sectors,
company: 28 per cent in the taxi and 36 per cent while it was permanent for some (7 and 15 per
in the delivery sectors. These complaints were cent respectively).7 Thus workers were effectively
mainly related to payment issues (48 and 41 per blocked from accessing any work through the
cent respectively); conflict with the customer (35 platform, sometimes permanently.
and 24 per cent respectively); technical problems
Uber Eats deactivated my account for three
with the app (23 and 31 per cent respectively);
days because I cancelled an order that was
and cancelled rides or orders (12 and 36 per cent
far away – Male respondent on app-based
respectively). When workers do seek assistance or
delivery platform Uber Eats (Mexico)
file complaints, a sizeable proportion in both the
app-based taxi (49 per cent) and delivery (37 per My account was deactivated permanently
cent) sectors reported not being satisfied with the because I went back to my home village –
outcome, with some even reporting being pun- Male respondent on app-based taxi platform
ished, through the imposition of fines, for example. Gojek (Indonesia)
They answer to you but at the end, you About 65 per cent of workers in the app-based taxi
always get adversely affected. They charge and delivery sectors who reported deactivation
everything to you, even the shipping rates – considered that it was unjustified, and a substan-
Male respondent on app-based delivery platform tial proportion (69 and 83 per cent respectively)
SinDelantal (Mexico) had appealed against it, with 48 and 59 per cent
respectively reporting dissatisfaction with the
I was not satisfied with the platform’s re-
outcome. This proportion was quite high (over
sponse to my complaint. They punished me
60 per cent) among app-based taxi drivers in
and I had to pay the fine – Female respondent
Mexico and Morocco.
on app-based taxi platform Beat (Chile)
I was deactivated for one year. When I
Some workers in the app-based taxi and delivery
went to the app offices they ignored me.
sectors have experienced deactivation of their
Because the app only believes the client
accounts: this was the case for about 19 and
when there’s a problem and ignores the
15 per cent of workers in the app-based taxi and
driver – Male respondent on app-based taxi
delivery sectors respectively (see figure 4.23). The
platform Uber (Mexico)
proportions were particularly high in Ghana and
Morocco (37 per cent of app-based taxi drivers) I don’t know why the account was deacti-
and Mexico (45 per cent of app-based delivery vated. I appealed and there were no results.
workers). The reasons for deactivation included My problems were solved only after our
low ratings, non-acceptance of work, taking leave association intervened – Male respondent on
and complaints from customers. Among those app-based taxi platform Careem (Morocco)
7 Workers whose accounts had been permanently deactivated on one platform were using another platform at the time of the survey.
Figure 4.23 Deactivation of accounts in the app-based taxi and delivery sectors
App-based taxi sector

Satisfied with outcome
of appeal
48
Account has been deactivated
Appealed deactivation
81
52
69
19
31
Agree with deactivation
35 65
Yes No
App-based delivery sector

Satisfied with outcome
of appeal
Account has been deactivated 59
85 Appealed deactivation
41
15 83
Agree with deactivation

17
35
65
Yes No

4.4 Skills acquisition and mismatch
Digital labour platforms provide work oppor- a university degree, when controlling for basic
tunities ranging from low-skilled, short-term, characteristics. In developing countries there is
repetitive tasks or delivering goods to high-skilled no significant difference between workers with
website development or data analytics. Despite the different education levels. The statistical analysis
diversity of tasks available, there are challenges also shows that respondents holding a bachelor’s
with regard to skills mismatch, training require- degree are not associated with significantly
ments or accumulation of skills on platforms. different hourly earnings than those without a
university degree, and this is true in both devel-
Platforms are redefining the relationship between
oped and developing countries. Other studies
formal education and access to work (Teubner,
have also indicated that workers’ education is not
Adam and Hawlitschek 2019; see section 4.3.2).
necessarily correlated with their income levels on
The relevance of skills and qualifications acquired
platforms; rather, it is previous work experience
through formal education varies on online web-
and reviews obtained from clients that influence
based platforms. Workers often do not have to
their earnings (Herrmann et al. 2019a and 2019b).
provide their educational qualifications on online
platforms; instead it is their profile, ratings or repu- Moreover, workers in developing countries with a
tation which are vital for accessing well-paid tasks. postgraduate degree and above tend to have lower
average hourly earnings than workers from devel-
That said, a statistical analysis shows that there is
oped countries with only secondary education or
some correlation between educational qualifica-
below (see table 4.8). This difference is more pro-
tions and earnings (see Appendix 4B). Workers on
nounced when looking at median incomes.
freelance platforms with a postgraduate degree
and above in developed countries tend to earn In addition, on some plat forms, such as
70 per cent more than their counterparts without PeoplePerHour or Freelancer, workers can improve
Table 4.8 Hourly earnings (paid and unpaid) with different education levels on online web–based platforms,
X
by type of platform, development status and sex (in US$)
Mean Median
Postgraduate Postgraduate
Secondary Higher Bachelor’s Secondary Higher Bachelor’s
degree degree
and below secondary degree and below secondary degree
and above and above
Freelance – 7.8 7.3 8.1 – 5.8 4.7 6.5
Microtask 3.4 3.1 3.6 2.9 1.7 2.2 2.5 1.9
Developed 3.8 3.8 5.1 4.5 2.3 2.9 3.9 2.7
Developing 2.7 2.1 2.8 3.5 1.0 1.1 1.6 1.9
China – 2.6 2.7 3.4 – 1.6 1.8 3.0
Ukraine – 2.2 3.0 3.7 – 1.2 1.5 1.5
Male – 3.0 3.7 4.0 – 2.0 2.3 2.3
Female 2.6 2.8 3.6 3.7 2.1 1.7 2.3 1.7
Total
(without
3.5 3.4 4.1 4.0 1.7 2.3 2.6 2.3
China and
Ukraine)
Total 3.5 2.9 3.6 3.9 1.7 1.9 2.3 2.0
– indicates there were not enough observations to yield any meaningful analysis.

Note: Data is trimmed at 1 and 99 per cent by sector.


their access to work by paying for additional ser- I participate in competitive programming to
vices and increased visibility (see sections 2.3 and work on interesting problems while learning
4.2.1), rather than through the skills and qualifi- new skills along the way – Male respondent on
cations they have acquired via formal education. competitive programming platform HackerRank
(United States)
To enable workers to improve their skills and
enhance their profiles, several platforms are Varying degrees of vertical and horizontal skills
increasingly providing opportunities through mismatch can be observed on online web-based
skills certification schemes (see section 2.3.3). platforms. Workers with higher educational
About 20 per cent of respondents on freelance achievements are not necessarily finding work
platforms reported that they had completed commensurate with their skills. Many respond-
classes or training to obtain such certification. ents on online web-based platforms reported
These have demonstrated some positive impact that they have more skills than demanded for the
on earnings, primarily for relatively new entrants, tasks (see figure 4.24), with small differences by
although verified work history and ratings tend sex. The skills mismatch is quite pronounced for
to have greater impact for experienced workers those engaged on microtask platforms, where
(Kässi and Lehdonvirta 2019). many (57 per cent) have a university degree, of
whom a majority have a specialization in STEM
Competitive programming platforms in particular
(science, technology, engineering and medicine),
provide an opportunity for workers to learn new
but undertake tasks such as responding to surveys
skills or upgrade their current skill sets, estab-
and experiments, content access, data collection
lish networks and improve their employability.
and so on, which tend to require few or no specific
About 76 per cent of respondents on competitive
skills. On freelance platforms, a sizeable propor-
programming platforms reported that they
tion of respondents (29 per cent) reported that
competed regularly (more than once a month) in
they had more skills than were demanded for the
programming contests. The contests also help
task. A higher proportion of women (68 per cent)
these workers to solve interesting challenges
than men (59 per cent) reported that their skills
that the platforms are confronting. Furthermore,
were a good match. On competitive programming
they often complement and help to overcome
platforms, an overwhelming majority of respond-
limitations in formal academic training, and assist
ents (77 per cent) reported that their skills were
workers to adapt to the changing needs of the
adequate for doing well on the platform. The need
market (Brito and Gonçalves 2019). Some plat-
for more technical training was prominent among
forms, such as CodeChef, Kaggle, HackerEarth and
respondents on web-based platforms in China
HackerRank, also offer real-time problem-solving
(54 per cent) and Ukraine (33 per cent).
contests or challenges for universities as part
of their academic curricula. Workers are further I have the necessary skills for the jobs I do
motivated to participate in these platforms to but additional skills will definitely allow me
improve their ratings or rankings, which reflect to be more successful – Female respondent on
their ability to program, with many platforms such freelance platform Upwork (Ukraine)
as HackerRank, Topcoder and others providing
On some online web-based platforms, horizontal
recruitment services to large companies (Grooms
skills mismatch is not prevalent. On freelance
2017; see section 3.1.1).
platforms many respondents undertook tasks
Competitive programming helps me prepare potentially related to their field of study (see
for challenging questions and these are often table 4.9). For example, 60 per cent of respondents
asked in interviews by big companies, hence with a university degree in the arts completed cre-
it serves as a valuable preparation tool – Male ative work, and 61 per cent of those with an IT and
respondent on competitive programming plat- computer degree performed technology-related
form HackerRank (India) tasks. However, several instances of horizontal
Figure 4.24 Skills in relation to tasks on online web-based platforms, by type of platform
I have more skills than demanded for the tasks 29

Freelance
My skills are a good match with the tasks that I complete 62

I have some skills, but I do require additional training 9
I do not have sufficient skills 0
programming
I have more skills than demanded for the tasks

Competitive
2
I have adequate skills to complete the tasks 77
I would need further training 22
I would need further language training 5
I have the skills to do more demanding tasks 84

Microtask
I have sufficient skills to complete the tasks 96

I would need further job-related training 32
I would need further language training 33

My present skills correspond well with all the tasks
China
28
I would need further technical training 54
I would need further English-language training 9

Ukraine
My present skills correspond well with all the tasks 32

I would need further technical training 33
I would need further English-language training 29
0 20 40 60 80 100
Note: The responses relating to skill requirements for tasks vary across the different platforms, as the questions were formulated
differently. In addition, respondents on microtask and competitive programming platforms were asked to multi-select their response,
while in the cases of freelance platforms and country surveys of China and Ukraine the respondents were asked to select
only one response.
Table 4.9 Types of tasks performed by respondents on freelance platforms, by field of study
X
Type of task performed (%)

Number of
Field of study
respondents Business Technology- Data Sales and Professional
Creative Other
services related analytics marketing services
Arts 25 40 16 20 60 20 68 4
Communication and media 15 20 7 13 47 13 80 27
Economics, finance
85 64 14 36 26 25 59 20
and accounting
Engineering 35 34 31 20 51 17 74 26
Formal and natural sciences 30 47 23 43 17 33 77 17
IT and computers 61 31 61 28 41 21 36 10
Law 11 55 9 9 0 18 100 9
Medicine and health 18 50 11 28 22 28 83 33
Social sciences 84 57 7 21 25 21 77 33
Total 364 48 22 27 32 23 66 21
Note: Figures presented refer to workers with a university degree. Types of tasks include: Business services: accounting, administrative
support, customer service, business consulting; Technology-related: IT and network administration, mobile development, general
programming, software development, web development, website development; Data analytics: analytics, data sciences, database
management; Creative: design, graphics, music and audio, video and animation, other creative activities; Sales and marketing: general sales
and marketing, digital marketing; Professional services: architecture, legal, translation, transcription, writing, editing, general consulting.
skills mismatch can also be observed where Gender-based occupational segregation of tasks
workers performed tasks that were unrelated to was common on freelance platforms. The propor-
their field of expertise. For instance, 22 per cent of tion of female respondents on freelance platforms
respondents with a university degree in medicine performing tasks related to technology and data
and health performed creative tasks. A survey of analytics (8 and 22 per cent respectively) was
workers on Russian-language freelance platforms lower than for men (32 and 29 per cent respect-
also shows that about one third of the workers ively). A higher proportion of women performed
(36 per cent) were engaged in tasks that were ab- tasks related to business services, particularly in
solutely unrelated to their field of study (Shevchuk developing countries (70 per cent). Women were
and Strebkov, forthcoming). Horizontal skills mis- also more engaged than men in professional
match was particularly prevalent in developing services such as in the legal field, translation,
countries (see also box 4.7). writing and editing or sales and marketing. In
China and Ukraine, 50 and 52 per cent of female
There are very few tasks that match my skills,
respondents respectively undertook work in pro-
especially tasks related to civil engineering
fessional services, which was considerably higher
work and civil 3D work – Male respondent on
than for male respondents (34 and 29 per cent
freelance platform Upwork (Pakistan)
respectively). The proportion of women engaged
in technology-related tasks was far lower in both
Box 4.7 Underutilization of skills

X
Many platform workers in developing countries are unable to access opportunities aligned with
their skill sets. As a result, a number of highly educated platform workers in these countries are
performing academic writing tasks on online web-based platforms to assist students with their
assignments and theses, a situation which also in fact leads to legal and ethical concerns. One
respondent with a degree in accounting, for example, was completing student assignments for
US$50, while sometimes also being involved in the writing of Master’s and PhD theses. Often
posted by anonymous students based in developed countries, the topics for such theses ranged
from social sciences to biological sciences, and could also be extremely specific. Platform workers
needed to perform such assignments rapidly and struggled to gain expertise on the issues, as
seen in one comment: “You don’t master one subject. So you adapt as a journalist.”
Workers also spend many hours completing these writing tasks. One respondent highlighted
that ten pages for a Master’s thesis can take about ten hours, while seven pages for a PhD thesis
can take about nine hours. In this regard, there were concerns about rejections, with reported
instances of work being rejected without grounds provided and no revision requested, which
results in lost time and effort for the worker without any earnings for work accomplished.
Quite a number of times work has been rejected. After you are done with the work,
they reject the work and then they send you the rejection notice and then they don’t
repost the work – Male interviewee (Kenya)
In the case of content moderation, interviews with workers in call centres revealed that over
95 per cent of those hired are IT professionals with a university degree in engineering or
computer science. These workers are primarily tasked with monitoring and removing offen-
sive, obscene, false or illegal content from online platforms. These tasks had no relation to
their qualifications, and the workers did not see any learning or career advancement in these
jobs. The major reason for pursuing this kind of work was the lack of alternative employment
opportunities. While IT professionals are recruited to perform these tasks, research has shown
that a number of sectors, including the ICT sector, are facing a shortage of such workers (ILO,
2020g). This clearly demonstrates the need to address skills mismatches and ensure that the
expertise of IT professionals is used efficiently.
Source: ILO interviews with workers in developing countries.
China (5 per cent) and Ukraine (2 per cent), with app, followed by customer service and safety
almost a quarter of male respondents reporting guidelines (see figure 4.26). Some respondents
that they performed such tasks. were critical about the quality of training offered
Most workers in the app-based taxi and delivery by the platforms, which they did not consider to
sectors have access to some sector-specific be very helpful for their work. With the spread of
training, although the quality has been questioned the COVID-19 pandemic, many platforms have also
by some. High educational levels among workers in started to raise awareness and deliver training in
these sectors (see section 4.1.6) could be indicative occupational safety and health, including through
of a skills mismatch, given that they are considered digital means (Fairwork Project 2020; see box 4.6).
low-skilled. Nevertheless, specific training can be Training is useless and most delivery workers
beneficial to good work performance in these that finish the training don’t know what to
sectors. In this regard, a large majority of respond- do – Male respondent on app-based delivery
ents in both the app-based taxi (75 per cent) and platform Rappi (Chile)
delivery (85 per cent) sectors reported receiving
some degree of training, with significant variations For training, I receive videos on the app.
at the country level (see figure 4.25). Training they I learn only through these – Male respondent
received included information on how to use the on app-based taxi platform Uber (India)
Figure 4.25 Proportion of respondents who received training from app-based taxi
and delivery platforms, by country
100
97
94 95
90 93 93
89 89 90
80 86 87 87 85
81 81
75 75
71
50 51
20 25
0
Argentina Chile China Ghana India Indonesia Kenya Lebanon Mexico Morocco Ukraine Total
Figure 4.26 Types of training provided by app-based taxi and delivery platforms
100
80 86
81
75
71
55
40 47
App-based delivery
39 38
34
31 29
20
16 4 3 6
0
Safety Customer Etiquette Info about Info about How to use Licensing Other
service bonuses deactivation the app information
Note: Figures presented refer to workers who reported having received training from location-based platforms.

4.5 Platform design and discrimination
A considerable proportion of workers on online While being interviewed for the job some
web-based platforms have experienced dis- clients plainly state a preference for some
crimination, particularly women and workers localities more than others. These are
in developing countries. About 19 per cent of mostly jobs you could have openly competed
respondents on freelance platforms reported on – Female respondent on freelance platform
that they had experienced discrimination. This Upwork (Malawi)
proportion was lower in developed countries
A considerable proportion of workers in the
(12 per cent) than in developing countries (22 per
app-based taxi and delivery sectors have also
cent), where it was also particularly high among
experienced discrimination or harassment.
female respondents (25 per cent) (see figure 4.27).
About 20 and 29 per cent of respond-
The qualitative responses from respondents in ents in the app-based taxi and delivery
developing countries further reflect the fact that sectors respectively reported being
discrimination often takes the form of exclusion subject to discrimination or harass-
from work opportunities or low pay. This was ment, or were aware of such
observed on several platforms where certain instances faced by their
higher-paying tasks were allocated to workers peers while performing
from developed countries through the platform work (see f igure 4.28).
design. Discrimination based on other factors, A higher propor tion of
such as gender, ethnicity, age or language spoken, women (where they were
was also reported. present in the sample) than
men reported this to be the
Some job postings specify that you cannot
case in the app-based taxi sector.
apply if you are a non-native speaker, even
In the app-based delivery sector,
if you are qualified – Female respondent on
a particularly high proportion of
freelance platform Upwork (Nicaragua)
respondents from India, Mexico
I have had a male client imply that a certain and Morocco (predominantly male,
task may go over my head because I am a as few women are present in the
woman – Female respondent on freelance plat- sector in these countries) reported facing
form Upwork (South Africa) discrimination or harassment.
Figure 4.27 Proportion of respondents on freelance platforms who have experienced

discrimination, by development status and sex
25
25
20
22
20 21 Total
19
15 16 17 Male
10 12 Female
8
5
Developed countries Developing countries Total

Figure 4.28 Proportion of respondents on app-based taxi and delivery platforms who have
experienced or witnessed discrimination or harassment, by sex and country
App-based taxi App-based delivery
50 50
Female
40 42 40 44
38 40
38
Male
30 32 30 33
31
29
30
Total
20 20 23
App-based taxi
20
17 18 18
10 13 4 Total
13 10 4 App-based delivery
2
5
0 0
Chile
Ghana
India
Indonesia
Kenya
Lebanon
Mexico
Morocco
Ukraine
Total
Argentina
Chile
Lebanon
Mexico
Morocco
Ukraine
Total
Ghana
Indonesia
Kenya
India
Note: Disaggregation by sex is displayed only for countries where the sample contained at least ten female respondents.
Figure 4.29 App-based taxi drivers and delivery workers having faced
or witnessed discrimination or harassment from different entities
60
60
56
50
30 33
20 25
17 16
10 13
11 4 9 6
8 7 3
0
The app Clients or Restaurants Shopping The police Taxi Other Other
company customers malls drivers drivers
Note: Figures presented refer to workers who reported that they or one of their colleagues had suffered discrimination or harassment
while performing work.

A lot of discrimination, especially from discrimination from the platforms as well, across
restaurants and some customers. We are both the app-based taxi (17 per cent) and delivery
discriminated against because of our clothes (11 per cent) sectors. Furthermore, in the delivery
and for the work we do – Male respondent on sector, 44 per cent of respondents reported
app-based delivery platform Rappi (Mexico) that they had faced or witnessed discrimination
or harassment from restaurants or businesses
I was involved in a sexual harassment situ-
where they would go to collect food or goods (see
ation by a passenger, where he offered
figure 4.29). In the taxi sector, 33 per cent reported
money for sex – Female respondent on app-
that they had faced or witnessed discrimination or
based taxi platform Uber (Chile)
harassment from other taxi drivers.
Among those subject to or having witnessed such
I am stressed by the harassment by trad-
instances, a majority in both sectors reported
itional taxi drivers who accuse us of driving
discrimination or harassment from the client (see
them into poverty – Male respondent on app-
figure 4.29). Many platforms provide workers’
based taxi platform Uber (Lebanon)
names and photographs to clients, which can allow
for discriminatory client behaviour. This has also We face discrimination in restaurants,
been observed during the COVID-19 pandemic, they don’t let us sit inside, can’t even use
when customers have cancelled drivers belonging their washrooms, or have any water – Male
to certain communities (Chapman and Frier respondent on app-based delivery platform
2020). Some workers also reported instances of Uber Eats (India)
Conclusion
This chapter, drawing on data from around 12,000 challenges to worker well-being. Many workers,
workers, has shown that digital labour platforms on both online web-based and location-based
(both online web-based and location-based) are platforms, reported that they would like to work
the main source of income for many workers. The more than they do but that factors such as excess
majority of workers on these platforms are young labour supply and unavailability of enough work or
(35 years and below) and highly educated. While well-paid tasks prevent their greater engagement.
women, including those with care responsibilities, In addition, workers from developing countries
also find work on digital labour platforms, fewer often face additional barriers due to exclusion by
women than men participate in such platforms, platforms or clients, or fee charges that exceed
especially in developing countries. Furthermore, their financial capacities.
location-based platforms, especially in the delivery
Earnings on online web-based platforms are influ-
sector, provide an important source of work op-
enced by time spent on unpaid tasks. Workers on
portunities for migrants in some countries. The
these platforms, on average, work 27 hours in a
motivations of workers to perform tasks on online
typical week (including both paid and unpaid work),
web-based platforms are often to complement
with one third of their time, or eight hours, spent
pay from other income sources, job flexibility or
on unpaid work. The average hourly earnings (paid
preference to work from home, while on loca-
and unpaid time) on these platforms are US$3.4,
tion-based platforms they engage due to a lack
and half of these workers earn less than US$2.1 per
of alternative employment opportunities as well as
hour. While findings on the existence of a gender
better pay when compared to other available jobs.
pay gap are mixed, there is a clear gap between the
While digital labour platforms provide opportun- earnings of workers in developing and developed
ities to workers, a granular focus on working countries. Earnings on these platforms are also
conditions and the organization of work on impacted by intense competition, high commission
the platforms reveals that they present many charges and non-payment for tasks completed.
Hourly earnings (including waiting times) vary Platforms are redefining the relationship between
across countries in the app-based taxi and formal education and access to work, as worker
delivery sectors, and tend to be higher than in profiles, ratings and reputation are now im-
the traditional sectors. Bonuses and incentives portant for accessing work. Varying degrees of
offered by platforms attract a large number of vertical and horizontal skills mismatches can be
workers. This has led to an oversupply of workers, observed on digital labour platforms. A high pro-
which has implications for earnings in both the portion of workers on freelance and competitive
app-based and traditional sectors. Earnings are programming platforms reported that their skills
also affected by order cancellations, and, particu- were a good match with the tasks they were per-
larly in the taxi sector, loan repayments, declining forming, and many were undertaking tasks that
bonuses and commission charges. Most workers were potentially related to their field of study.
in the taxi and delivery sectors work long hours Skills mismatch is, however, quite prevalent for
and endure high work intensity to meet their those engaged on microtask platforms, where a
income needs and targets for bonuses. highly educated workforce is engaged to perform
tasks that tend to require few or no specific skills.
Social protection stands out as a major concern,
Similarly, a sizeable proportion of workers on plat-
with the majority of workers on digital labour
forms in the taxi and delivery sectors are highly
platforms not having coverage. This has serious
educated. This points to a substantial challenge
implications for workers, in particular workers on
related to skills mismatches and underutilization
location-based platforms who are exposed to oc-
of worker skills, especially in developing countries.
cupational safety and health risks. This situation
has further been exacerbated since the outbreak The findings also reveal that a considerable pro-
of the COVID-19 pandemic. portion of workers on digital labour platforms
experience discrimination or harassment. In some
The everyday experience of workers is defined by
instances platform design exacerbates exclusion,
platform design and algorithmic management.
in particular of workers from certain developing
Platforms use algorithms to match workers with
countries, by preventing workers from accessing
clients or customers, for which worker ratings are
higher-paid tasks on online web-based plat-
decisive. These ratings are also algorithmically
forms, or through features on location-based
determined, using a number of metrics which in-
platforms that allow for discriminatory behaviour
clude acceptance and rejection rates. This practice
by clients. Such a situation is intensified by in-
ultimately limits workers’ freedom and ability to
herent structural problems, particularly for female
make work-related decisions, and particularly to
workers on location-based platforms, where
reject assigned work. Furthermore, client-gener-
they may be exposed to insecurity, violence and
ated ratings and evaluations are not always fair
harassment.
or transparent, and negative ratings can lead to
deactivation of worker accounts. Despite such The challenges brought to the fore on digital
serious implications for workers, they have limited labour platforms through the granular explor-
awareness about options available for dispute ation of platform worker experience underscore
negotiation or resolution. Moreover, new forms of the urgency of re-evaluating platform work while
platform and client monitoring and control limit it is still in its nascent stages. Its potential for
the flexibility and autonomy of workers. Workers creating decent work opportunities for workers,
on online web-based platforms also report that the including women, persons with disabilities,
platforms or clients require them to be available at migrant workers, and indigenous and tribal
specific times or to install software that captures peoples, among others, is significant. Addressing
their work habits and times. For workers on loca- the challenges highlighted in this report will be
tion-based platforms, algorithmic management vital for transforming digital labour platforms to
is shaping scheduling, working hours and access ensure that they become and continue to evolve
to future work, while the worker is penalized for as sustainable and inclusive avenues for decent
task cancellations. work opportunities.
5 Ensuring decent
work on digital
labour platforms
Ensuring
decent work
Diverse practices of regulation for
platform workers across the world
Denmark
Hilfr collective agreement pertaining
Canada France to certain platform workers
Unreasonable dispute
Right to disconnect for
resolution process
some platform workers Germany, Austria, Sweden
invalidated
Organizing online web-based platform workers
Republic of Korea
Work injury benefits
extended to some
platform workers
United States
Diverse
Spain
approaches to
China
Mandatory
classification of
Work injury benefits
employment
platform workers
extended to some
injury insurance
legislation platform workers
India
Peru Indonesia
Social security
Establishment
Nigeria
benefits extended Work injury benefits
of fund for
Data protection to platform workers extended to some
COVID-19 and
regulation affecting platform workers
data transparency
platform workers
Brazil
Occupational safety and health
coverage transcending
Australia
employment relationship Occupational safety and health
South Africa obligations transcending
employment relationship
Uruguay Anti-discrimination
Digital social security law applying to
contributions and all workers
tax payments for New Zealand
platform workers Occupational
Argentina safety and health
Prohibition of child labour, obligations transcending
including through platform work employment relationship
Fundamental principles Labour standards in other Labour standards in other

and rights at work ILO instruments of ILO instruments relevant
general application to platform work
 Freedom of association  Occupational safety and health  Payment systems

and collective bargaining
 Social security  Fair termination
 Elimination of discrimination
 Job creation policies  Access to data and privacy
 Elimination of forced labour
 Access to labour inspection  Clear terms and conditions
 Effective abolition of
 Job mobility
child labour
 Access to dispute
resolution
5. Ensuring decent work on digital labour platforms 197
Introduction
Digital labour platforms provide new opportun- by major ILO instruments and declarations,
ities for economic growth and job creation, and including the ILO Centenary Declaration for the
have the potential to contribute to the realization Future of Work which calls on ILO Member States
of the United Nations Sustainable Development to: “Promot[e] sustained, inclusive and sustainable
Goals, particularly Goal 8, which is to “[p]romote economic growth, full and productive employment
inclusive and sustainable economic growth, and decent work1 for all through … policies and
employment and decent work for all”. These plat- measures that … respond to challenges and op-
forms have also facilitated the inclusion of many portunities in the world of work relating to the
businesses, including small and medium-sized digital transformation of work, including platform
enterprises, into the digital economy, thereby work” (ILO 2019a, III C (v)).
improving their productivity and enhancing their
This chapter first considers how digital labour plat-
market base.
forms use service agreements to set the rules for
As previous chapters have shown, the conditions of work mediated through them (section 5.1). In sec-
work are unilaterally determined by the platforms, tion 5.2 the discussion turns to what “decent work”
which raises a number of challenges. The broad for platform workers means, grounding the an-
discretionary power over working conditions ex- alysis in ILO Conventions and Recommendations.
ercised by many platforms is apparently not yet As there are frequently gaps between the current
adequately counterbalanced by initiatives from rules for digital platform work and what they
governments and other stakeholders (Kapczynski should be from a decent work perspective, sec-
2020; Berg et al. 2018; Agrawal et al. 2013). The tion 5.3 proceeds to examine initiatives from
regulatory capacity of the platforms, mediated governments and the social partners that try to
through service agreements, is buttressed by the bridge those gaps.
ability of the platforms to rapidly collect extensive
Before turning to the substantive discussion,
data on workers and their performance at low
one important point about scope must be made.
cost. This information can be harnessed so as both
The chapter is grounded in a broad concept of
to craft rules that best benefit the platforms and
“regulation” and is not confined to legislation
to give effect to them.
and court judgments only. “Regulation” does not
The extensive power of the platforms to monitor refer here simply to a rule made by a government
and control work relationships can, at least in prin- (Black 2002, 28; see also Black 2001). The inquiry
ciple, be exercised so as to create more and better extends beyond state-based rules and processes
jobs. For example, both individual and aggregated to interventions by social partners and other
worker data might be used to reduce excessive non-state actors that “influence the course of
working hours, improve safety and health at events” (Parker and Braithwaite 2003, 119). So,
work, identify discriminatory hiring patterns, or regulation in the sense used here encompasses
maintain efficient and accurate payment systems not only legislation, court judgments, decisions of
(Rogers 2018). However, observations in the pre- administrative agencies and government policy
vious chapters have suggested that this is not yet statements, but also collective agreements, multi-
generally the case. lateral accords, codes of conduct, contracts and
even informal arrangements. They all influence,
This chapter examines the regulation of digital
or at least potentially influence, the “regulatory
platforms, focusing on labour and social protec-
space” (Hancher and Moran 1998; Scott 2001) in
tion. It covers not only regulation emanating from
which digital labour platforms operate.
the platforms themselves but also initiatives from
governments and the social partners – employer Of course, this is not to say that these different
and worker organizations. The chapter is informed forms of regulation are equivalent; they have
1 See also the ILO Declaration on Social Justice for a Fair Globalization, adopted by the International Labour Conference at its
97th Session, Geneva, 10 June 2008.
different levels of authority and different modes digital labour platforms in setting the rules
of enforcement. Regulations emanating from through terms of service agreements becomes
governmental institutions, including legislatures, apparent since such agreements are a form of
administrative agencies and courts, prevail over (contractual) regulation. This role is examined in
private regulation through contracts. Collective section 5.1 based on an analysis of terms of service
agreements, too, generally prevail over individual agreements of 31 major digital labour platforms
contracts, although the precise legal status of (see Appendix 2). Second, the analysis reveals
collective agreements varies from jurisdiction to the important part that organizations of workers
jurisdiction. And depending on the jurisdiction, all and employers have played, sometimes together
of these may have binding legal effects whereas with governments, in developing responses to
codes of conduct and informal arrangements platform work through social dialogue and tripar-
may not. tism.2 These responses have taken the form, for
There are two major advantages to examining example, of innovative collective agreements and
regulation in a broad sense, rather than state- initiatives of labour law reforms. These initiatives
based laws alone. First, the significant role of are examined in sections 5.2 and 5.3.
5.1 Regulation by digital labour platforms:

Terms of service agreements
As described in Chapter 2, the way in which This means that their content is determined
digital labour platforms draft their terms of ser- unilaterally by the platform; the other party is
vice agreements, and implement them through able only to accept or refuse. There is little or no
technology, determines the working conditions negotiation (Berg et al. 2018). The users, both
of the workers they engage and mediate (Berg workers and clients, must accept the terms of
et al. 2018; De Stefano 2016; Pasquale 2015). This service before accessing the platform or building
section examines the effect of those agreements, their profiles on it (Pasquale 2015). While adhe-
drawing on a sample of 31 major digital labour sion contracts offer great efficiencies and savings
platforms (online web-based and location-based) through the reduction of transaction costs, the
that has been compiled and analysed by the frequent inequality of bargaining power can
ILO (Appendix 2B). lead to unfair terms (Kessler 1943; Hillman and
Rachlinski 2002). Moreover, as Chapter 2 demon-
strated, the obligations set out in the terms of
5.1.1 Platform terms service are frequently monitored and assessed
of service agreements through algorithmic management and workflow
tools provided by the platforms (see also Duggan
On most digital labour platforms, terms of ser- et al. 2020; Rosenblat and Stark 2016).
vice agreements serve as the first major point of
While the wording of terms of service agreements
regulatory engagement between the workers and
often differs from platform to platform and from
the platform. The agreements invoke the law of
jurisdiction to jurisdiction, there is an underpin-
contract to give legal effect to their stipulations
ning business logic (Sanders and Pattison 2016).
(Berg et al. 2018).
The business organization aspects of the agree-
On most digital labour platforms, terms of service ments (including the ratings) have been analysed
agreements tend to be “contracts of adhesion”. in Chapter 2. This section of the report builds on
2 On the significance of tripartism in regulatory theory, see the pioneering work of Ayres and Braithwaite (1992, 54–100).
Terms of service their terms of service agreements, in regulating

agreements tend to be conditions pertaining to remuneration, hours of
work and other issues, the platforms are largely
“contracts of adhesion”. unconstrained by labour protection legislation.
And notwithstanding the denial of an employment
relationship, terms imposed on workers by
that analysis by considering how the agreements many service agreements frequently
seek to define the nature of the relationship entail restrictions on worker autonomy
between platforms and workers, to construct and flexibility. These restrictions
methods of control, and to shape the dispute are enforceable through poten-
resolution process in favour of the platforms. tial account deactivation (Lobel
2016). They were discussed
First, the characterization of the contractual re- at length in Chapter 2 and
lationship between the platform and the worker in section 4.3.1.
as other than one of employment is a striking
feature of many terms of service agreements. For instance, the ILO
The ILO analysis in Appendix 2, table A2.3, reveals analysis shows that
various expressions used to avoid the creation many online web-based
of employment relationships. The workers en- plat for ms include in
gaged through the platforms are described as their terms exclusivity
“independent contractors”, “independent third- or “non-circumvention”
party providers”, “drivers”, “captains”, “delivery clauses binding workers
partners”, “driver partners”, “freelancers”, “click- and their clients to the plat-
workers”, “hackers”, and so on (see also Malin form for up to two years
2018; Xie 2018; Rodríguez Fernández 2017; Pinsof (see also section 2.5).
2016; De Stefano 2016; Aloisi 2016). Some microtask plat-
forms also limit the
You acknowledge and agree that this use of automated
Agreement is not an employment agreement processes to perform
(under labor law, tax law and/or social security
tasks (see section 2.4). Location-based platforms
perspectives). Therefore, you understand that
stipulate matters such as route instructions,
this Agreement shall not, in any means, be
working time and GPS tracking. Both online web-
interpreted as an industrial relation between
based and location-based platforms commonly
you and Grab – Terms of Use, Grab (Indonesia),
have terms regulating the acceptance and rejec-
English version as of 30 November 2020
tion of work and the manner of communication
However, the ILO surveys of app-based taxi drivers between the worker and client or customer, as
and delivery workers show that some platform well as customer service etiquette (see Chapter 2,
workers do not understand their work in these Appendix 2B). These rules are monitored through
terms. The consequence of this characterization data received by the platforms and processed
is that, assuming it is upheld as valid in legal by algorithms. For example, in delivery service
proceedings, platform workers are denied many agreements, the worker may be required to be
employee entitlements. These entitlements may reachable by the client at any time during delivery
include minimum pay, maximum working hours, and may be subject to real-time tracking.
leave and some social security entitlements, all
Some of these terms are quite explicable (such
of which may be (depending on the jurisdiction)
as customer service etiquette and limitations on
related to the establishment of an employment
excessive working hours) or indeed essential (as
relationship (Berg et al. 2018; De Stefano 2016).
in terms designed to ensure customer safety or
Second, on the basis that they have excluded prevent discriminatory action by the worker). But
the application of these entitlements through others may be unduly restrictive.
The platforms are largely

unconstrained by labour of the terms of service agreements in the ILO an-
alysis stipulate that the platforms reserve the right
protection legislation. to amend terms at any time, the amendments
becoming effective upon posting online. Keeping
Third, many agreements inhibit the capacity of track of these amendments is often not an easy
workers to contest the decisions of platforms, task for platform workers.
including automated decisions. Many terms of Furthermore, a prominent feature of the contracts
service agreements are not fully accessible or analysed by the ILO is that the methods of dis-
readily comprehensible to workers. This means pute resolution are, to the extent legally possible,
that they may not be aware of their rights and chosen by the platforms. These are methods
obligations. For instance, some platforms do not which are cost-effective to the platforms and/or
systematically link terms of service documenta- maximize their prospects of success. As the terms
tion to their homepage or the FAQ (frequently of service are contracts of adhesion, it is of no avail
asked questions) sections, and link to them only for the platform worker to seek legal advice in a
mid-way through the sign-up process (Berg et al. bid to negotiate a different arrangement which
2018). The ILO analysis shows that terms of service might better reflect both parties’ preferences.
agreements tend to be long, sometimes exceeding
What are these dispute resolution methods? One
10,000 words, and that they are complicated and
is that, where the jurisdiction permits, arbitration
legalistic (see also Venturini et al. 2016; Bygrave
is generally preferred over court proceedings for
2015; Bakos, Marotta-Wurgler and Trossen 2014).
matters involving contractual disputes, statutory
The extent to which these complexities give rise to entitlement claims, discrimination complaints and
a lack of comprehension among platform workers safety and health at work incidents. For example,
varies according to the platform, the sector and the Upwork terms of service examined in the ILO
the workers’ education level. Some terms of analysis refer disputes to Judicial Arbitration and
service are relatively abstruse; others are set Mediation Services (JAMS), a private arbitration
out in relatively accessible and clear language. provider, with respect to users located within the
Accordingly, empirical research can produce United States. The terms of service of Amazon
widely divergent findings. In the ILO global and Mechanical Turk and HackerEarth commonly refer
country surveys of workers conducted on online disputes to arbitrators selected by the American
web-based platforms, 80 per cent of workers Arbitration Association, while Uber commonly
read the terms of service, and of those, around refers them to the International Chamber
79 per cent indicated that they understood of Commerce.
the content. About 48 per cent of the
There is nothing inherently untoward about
location-based workers (42 per cent
referring disputes to alternative dispute
of the app-based taxi drivers and
resolution processes, since mutually agreed
51 per cent of the app-based delivery
arbitration is often more efficient and less
workers) reported that they had seen
costly than immediate access to judicial
a copy of the terms of service. Out of
proceedings. Indeed, for workers on some
those, 70 per cent said that the terms
platforms, internal grievance processes
and conditions were clear to them,
frequently provide satisfactory outcomes
while others said they were unclear,
(see section 4.3.2). In several of the service
or that they did not remember or had
agreements, the platforms agree to pay
not read them.
part or all of the arbitration fees.
The terms of service agreements This may be because mandatory
are not static; their content can vary employment arbitration must sat-
over time as the platforms make isfy certain criteria so as to permit
unilateral changes within the the enforcement of statutory rights
scope of the contract. A majority (Halegua 2016).
However, some terms of service agreements

purport to require a worker to use an institution
(arbitration or court) based in another country. And
in some jurisdictions, notably the United States,
judicial proceedings examining substantive claims have a plausible
can often be excluded altogether through clauses case. On the other
in agreements, provided certain procedures are hand, some terms of
complied with. This is highly problematic: Estlund service agreements permit
(2018) shows that workers in the United States are access to the courts where this
much more likely to be deterred from pursuing would favour the platforms. This is
their claims by mandatory arbitration clauses than the case with actions involving intellec-
they are if the ordinary court system is available tual property rights. Access to the courts in
to them (see also Colvin 2019; Halegua 2016). intellectual property cases enables platforms
Reviewing the empirical evidence, Estlund (2018, to access powerful judicial remedies (such as
707) comments: “It now appears that, by imposing injunctions) which are beyond the powers of pri-
mandatory arbitration on its employees, an em- vate arbitrators.
ployer can ensure that it will face only a miniscule
The extent to which the digital labour platforms
chance of ever having to answer for future legal
shape rules and processes therefore depends
misconduct against employees”.3
on how much regulatory space governments
Moreover, private arbitration does not contribute and judges are prepared to yield to them. The
to the development of the law the way judicial de- responses of legislatures, courts and stakeholders
cisions generally do (through binding precedents, to the impact of these terms of settlement clauses
guiding cases or other influential judgments). This are considered further in section 5.3.9 below.
development is particularly important given that
the state-based regulation of digital platforms is
inchoate. Furthermore, private arbitration can lack 5.1.2 Will the digital labour
transparency; it is generally not open to the public platforms improve terms
the way that court processes normally are.
of service by themselves?
Another method used in service agreements to
shape dispute resolution is to prevent “class” or Before considering regulation external to the
“representative” actions where these are pro- platforms, it is important to first consider ways
vided for in national legislation. These are actions in which the platforms themselves have sought
in which many plaintiffs (for example, platform to address many of the problems currently em-
workers) with a similar interest in the subject anating from their business practices, including
matter of a legal proceeding are represented their service agreements. In January 2020,
collectively against a common defendant (for many of the prominent digital labour platforms,
example, a platform). This creates economies of including Uber, Deliveroo and Grab, signed the
scale for the plaintiffs by enabling multiple claim- World Economic Forum Charter of Principles for
ants to pool resources and launch proceedings Good Platform Work (WEF 2020). The Charter of
which might be too complex or expensive for any Principles commits the platforms to diversity and
one individual. Prohibiting class actions, although inclusion; safety and well-being; flexibility and
economically rational from the perspective of the fair conditions; reasonable pay and fees; social
platforms, may have the effect of suppressing protection; learning and development; voice and
disputes because individuals give up, even if they participation; and data management.
3 While this comment applies to employees, the reasoning applies to any worker, including self-employed, since they are similarly
denied access to the courts through the mandatory arbitration clauses.
The Charter also provides that “[t]erms and condi- Many of these Charter commitments could be
tions should be transparent, clearly stated, easily further elaborated or adapted. For instance, to
understandable, and provided to workers in an regulate through social dialogue rather than uni-
accessible form” (Clause 3.1) and that “[p]latforms laterally, other affected parties (such as worker
should promote a culture of transparency and organizations) could also be represented among
human accountability across use of algorithms, the drafting signatories. The Charter calls for
and ensure that fairness and non-discrimination “multi-stakeholder cooperation”, but at the time
are a priority in the design of algorithms” of writing other affected parties were not yet
(Clause 3.4). These statements reflect a significant represented among the drafting signatories. The
recognition of the problems and suggest a pre- analysis in the next section suggests that it is not
paredness to address them. enough to leave the platforms alone in the regu-
latory space. Other actors need to shape it as well.
5.2 Regulating digital platforms for labour

and social protection: What should be the goals?
If the rules set by platforms need to be comple- data and, increasingly, the law of the internet and
mented and counterbalanced by other regulatory of algorithms. It is not possible within the bounds
initiatives, what sort of interventions are appro- of this report to provide a comprehensive account
priate? The remainder of this chapter explores of these matters; an overview of their application
this question from both normative and descrip- to digital platforms is provided in section 6.3.2.
tive perspectives. It examines the principles and
A broad range of jurisdictions is surveyed,
standards that should shape interventions and
identifying initiatives from various economic, insti-
then looks at real world examples, drawing on
tutional and social contexts. Given this diversity, it
innovations from governments, courts and social
is not claimed that any one initiative – even where
partners in many different jurisdictions.
it is particularly successful – can be “transplanted”
The discussion concentrates on the regulation of into a different context and operate in a similar
work. There are, of course, many other relevant way. Such transplantations often take unexpected
fields – for example, competition (antitrust), tax, in- turns (Teubner 1998), at least without careful
tellectual property, corporate governance, privacy, adaptation to local circumstances (Berkowitz,
Pistor and Richard 2003).
Nor is it possible to explain at length the particular
settings in which each initiative has taken effect;
Worker
X
that would bog down the discussion in the details
The term “worker” has different legal of many national peculiarities. The purpose of
meanings in different jurisdictions. As the survey is to point to developments as poten-
mentioned in Chapter 1, the term is tial stimuli for future actions, not as models to
used in the broad sense contemplated replicate or impose. These future actions should
by the ILO instruments discussed, in be shaped by their domestic and multilateral
light of comments of the ILO supervisory contexts and by the involvement of the relevant
bodies. Unless otherwise specified, the social partners. They will also need to address the
term includes both the employed and the limitations in many of the initiatives that will be
self-employed (including independent identified – for, as will be seen, some are yet to
contractors). have a clear or lasting impact.
“Decent work” principles

Much of the literature concerning labour pro-
apply not only to employment
tection and digital labour platforms deals with relationships but to all work
whether a worker is employed by a platform, is
self-employed, or, in some jurisdictions, falls into arrangements, including
an intermediate category (Cherry and Aloisi 2017;
Waas et al. 2017). This is because the terms of
work mediated through
service agreements generally deny that platform digital labour platforms.
workers are employees, preventing access to
statutory entitlements such as minimum wage
and some forms of social security. This gives rise
to what is commonly described as the “misclassi-
fication issue”.
5.2.1 Labour standards
The misclassification issue is a critical one and is
often discussed early on in publications about for all working people:
the regulation of platform labour. However, the ILO instruments
analysis in this chapter will defer that discussion
because it will first focus on standards that are, The decent work objectives in the key ILO dec-
from the perspective of the ILO Constitution larations connect to the ILO Conventions and
and ILO instruments, applicable to all workers Recommendations as well as its Constitution.
irrespective of their contractual status. These They make clear that certain fundamental prin-
are principles for which the touchstone is simply ciples and rights apply to all working people. In
working (Countouris 2019). In other words, as the particular, it is well established that the principles
expression “decent work” suggests, the principles and rights set out in tables 5.1 and 5.2 apply to all
apply not only to employment relationships but to workers. This would include workers howsoever
all work arrangements, including work mediated described engaged through digital labour plat-
through digital labour platforms undertaken in a forms (De Stefano and Aloisi 2019).
self-employed capacity.
Table 5.1 lists the ILO fundamental principles and
There have been many scholarly and stakeholder rights at work, consisting of eight fundamental
efforts to identify which labour and social se- Conventions. To cite a recent statement by the
curity protections should apply to all workers ILO Committee of Experts on the Application of
(see, for example, Risak and Lutz 2020; Rodríguez Conventions and Recommendations (CEACR): “the
Fernández 2019; Xie 2018; Davidov 2014; Fudge, full range of fundamental principles and rights at
McCrystal and Sankaran 2012; Freedland and work are applicable to platform workers in the
Countouris 2011; Supiot 2001; and the Frankfurt same way as to all other workers, irrespective of
Declaration on Platform-Based Work4). Those their employment status” (ILO 2020h, Para. 327).
efforts inform this chapter. However, the analysis
The labour standards in table 5.2, while not in-
is primarily grounded in the “decent work” concept
cluded in the fundamental rights and principles
as stated in the ILO’s Centenary Declaration for
at work, are also critical to the provision of decent
the Future of Work and the Declaration on Social
work; the ILO supervisory bodies have stated that
Justice for a Fair Globalization, as well as Goal 8 of
they, too, apply to all workers.
the Sustainable Development Goals. Other inter-
national instruments are also directly relevant but
a proper examination of them is beyond the scope
of this chapter.
4 http://faircrowd.work/unions-for-crowdworkers/frankfurt-declaration/.
Table 5.1 Decent work for platform workers: Fundamental principles and rights at work
X
applicable to all workers, irrespective of contractual status
Fundamental principles
Scope of application of Conventions and Recommendations
and rights
The Freedom of Association and Protection of the Right to Organise Convention, 1948
(No. 87), provides that “[w]orkers […] without distinction whatsoever, shall have the
right to establish and, subject only to the rules of the organisation concerned, to join
Freedom of association
organisations of their own choosing” (Art. 2). This includes the self-employed (see ILO
and effective recognition
2012, Para. 53; ILO 2018b, Paras 328–330, 387–389). The Right to Organise and Collective
of the right to
Bargaining Convention, 1949 (No. 98), provides that all workers should enjoy protection
collective bargaining
against discrimination (including victimization and retaliation) on the basis of their union
activities and employer interference in their organizations. All workers, including platform
workers, should also enjoy the right to bargain collectively.1
Both the Equal Remuneration Convention, 1951 (No. 100), and the Discrimination
Non-discrimination (Employment and Occupation) Convention, 1958 (No. 111) (which applies to “employment
and equal remuneration and occupation”) apply to “all workers, both nationals and non-nationals, in all sectors of
activity, in the public and the private sectors, and in the formal and informal economy”
(see ILO 2012, Paras 658, 733).
The Forced Labour Convention, 1930 (No. 29), and its Protocol of 2014, as well as the
Elimination Abolition of Forced Labour Convention, 1957 (No. 105), are applicable to “all workers in
of forced labour the public and private sectors, migrant workers, domestic workers and workers in the
informal economy” (see ILO 2012, Para. 262).
The Minimum Age Convention, 1973 (No. 138), and the Worst Forms of Child Labour
Convention, 1999 (No. 182), cover all branches of economic activity and all types of
Elimination
employment or work regardless of employment status (which includes self-employment)
of child labour
as well as informal employment in both the formal and informal economy (see ILO 2012,
Para. 339).
1
Articles 5 and 6 contain qualifications concerning the armed forces and the police, as well as public servants engaged in
the administration of the State. See also ILO 2012, Para. 168; ILO 2018b, Para. 1285; and Direct Request (CEACR) – adopted
2019, published 109th ILC Session (2021) Right to Organise and Collective Bargaining Convention, 1949 (No. 98) – Belgium
(Ratification: 1953) (which specifically refers to platform workers). The CEACR had a diverse debate and discussion on
collective bargaining rights of self-employed workers in 2016 with respect to Ireland. In its consensual conclusion the
“committee suggested that the Government and the social partners identify the types of contractual arrangements
that would have a bearing on collective bargaining mechanisms” Individual Case (CAS) – Discussion: 2016, Publication:
105th ILC Session (2016).
The ILO Constitution5 and Conventions impose also contain provisions relevant to employers,
binding obligations on Member States that ratify workers and their organizations. Member States,
them. They do not bind individual enterprises working with organizations of employers and
or workers directly, although many of them still workers, incorporate the principles and obliga-
illuminate good practice. Recommendations pro- tions into their domestic law, adapting them to
vide non-binding guidelines. They are primarily national conditions. For example, they may pass
directed at Member States, although they may a law on anti-discrimination consistent with the
5 In the 1998 Declaration on Fundamental Principles and Rights at Work, the International Labour Conference declared that “all
Members, even if they have not ratified the Conventions in question, have an obligation arising from the very fact of membership
in the Organization to respect, to promote and to realize, in good faith and in accordance with the Constitution the principles
concerning the fundamental rights which are the subject of those Conventions”.
Table 5.2 Decent work for platform workers: Other key labour standards
X
applicable to all workers irrespective of contractual status
Labour standards Scope of application of ILO instruments
The Occupational Safety and Health Convention, 1981 (No. 155), “applies to all branches of
economic activity” and to “all workers” in these branches (Arts 1 and 2). Other key occupational
safety and health instruments include the Protocol of 2002 to the Occupational Safety and
Health Convention, 1981; and the Promotional Framework for Occupational Safety and
Health Convention, 2006 (No. 187) (see also ILO 2009, Para. 33; ILO 2017b). The Violence and
Occupational Harassment Convention, 2019 (No. 190), also applies to all workers (including jobseekers and
safety and health ex-workers) (Art. 2) and addresses violence and harassment involving third parties (Art. 4). These
Conventions deal not only with physical harm but with psychological health, which may be
particularly at risk as a result of online activities or isolation. The standards establish a defined
set of responsibilities with regard to the creation and maintenance of a safe and healthy working
environment. The ILO Centenary Declaration for the Future of Work also recognizes that safe
and healthy working conditions are fundamental to decent work.
The Social Security (Minimum Standards) Convention, 1952 (No. 102), and other ILO social
security instruments set minimum benchmarks for the protection of economically active
persons, including the self-employed (see ILO 2019b). These benchmarks pertain to sickness,
medical care, disability, maternity, employment injury, unemployment and old age, and include
Social security
support for family members. The Social Protection Floors Recommendation, 2012 (No. 202), calls
for a set of basic social security guarantees pertaining to essential healthcare and basic income
security. These should ensure “universality of protection, based on social solidarity” throughout
the life course (Para. 3).
The Employment Policy Convention, 1964 (No. 122); the Employment Policy (Supplementary
Employment
Provisions) Recommendation, 1984 (No. 169); and the Transition from the Informal to the Formal
and job creation
Economy Recommendation, 2015 (No. 204), are relevant to all workers, including the self-
policy
employed and those in informal economies (see ILO 2020h, Para. 29).
The principal relevant instruments which apply to the protection of workers and conditions of
work are the Labour Inspection Convention, 1947 (No. 81); the Protocol of 1995 to the Labour
Labour inspection Inspection Convention, 1947; and the Labour Inspection Recommendation, 1947 (No. 81). Certain
limited exceptions apply: for example, the armed forces may be excluded (see also ILO 2006,
Paras 44–49).
constitutional arrangements in their countries. to act (or refrain from acting) in a particular way.
Further regulatory forms of incorporation, de- Thus, digital labour platforms should, for example,
pending on the topic covered by the instrument, be prohibited from impairing freedom of asso-
include collective agreements, arbitration awards ciation (such as through adverse treatment of
and court decisions. union organizers). They should be prevented from
In some cases (such as unemployment benefits, discriminating between workers on protected
employment policy and labour inspection) the grounds such as sex or race (including by relying
responsibility for implementing the standards in on an algorithm they have generated). They should
ILO instruments falls on governments, rather than not be permitted to engage workers involuntarily
digital labour platforms (although the latter could or below minimum working age. And they should
be required to contribute to a particular benefit be required, so far as is reasonably practicable,
scheme). In most cases, though, giving effect to to ensure that workers are not exposed to unsafe
the Conventions and Recommendations entails and unhealthy working conditions, or to violence
regulation which requires digital labour platforms and harassment.
While these obligations can be stated simply in the 5.2.2 Convention principles
abstract, the practical implications can be more
difficult to specify. One major problem is that, that could be adapted to
as is the case with many contemporary business all digital labour platform
models (Goldman and Weil 2020; ILO 2020i; Weil
2014), digital labour platforms often interpose a workers, irrespective of their
party – such as a client, passenger or customer – status
between themselves and the workers (a form of
“fissuring”). This provides the basis for an argu- The fundamental principles and rights, and stand-
ment that the platforms are not responsible for ards set out in tables 5.1 and 5.2 do not exhaust
the acts of that third party. Thus, where a pas- the elements of decent work that regulation of
senger refuses to accept a driver engaged through digital labour platforms could aim to protect.
a digital platform because of their race, or where a There are further other standards that, while
driver is injured through the acts of the passenger, not as unequivocally rooted in ILO instruments,
the platform can allege that it has breached no are applicable to all working people, and are im-
obligation; it is the third party who has. portant requirements of just working conditions
One answer to this problem is to impose respon- for platform workers (Berg et al. 2018; Johnstone
sibility by linking it not to a particular kind of et al. 2012). The standards set out in table 5.3 are
contractual relationship, but to the capacity to ma- derived from ILO instruments which, while some-
terially influence outcomes (Johnstone and Stewart times limited to specific categories of workers
2015; compare Goldman and Weil (2020) on the (such as waged employees, domestic workers,
analogous concept of a “controlling employer” in homeworkers or workers engaged through pri-
the United States). Other legal formulations which vate employment agencies), give expression to
extend responsibility beyond a direct contractual principles that address the problems identified
relationship include “permit to work”,6 “involved in earlier chapters (compare in relation to the
in a contravention”,7 and “ensure the safety and European Union (EU), Kilhoffer et al. 2020; Cherry
health of persons … who may be affected by any and Poster 2016).
undertaking”.8 Many jurisdictions also provide for Job mobility is relevant here because of the use
shared liability between a principal and a subcon- of exclusivity or “non-circumvention” clauses and
tractor, particularly at a sectoral level, in relation because of the lack of portability of ratings sys-
to wages and sometimes social security; China tems – the incompatibility of metrics used by the
and several Latin American countries provide major platforms tends to lock workers into a single
examples (Zou 2017a; ILO 2016; Cooney, Biddulph platform (Prassl 2018; De Stefano 2016). Both of
and Zhu 2013). The appropriate formulations vary these aspects were discussed in Chapter 2. A par-
from jurisdiction to jurisdiction. tial response to this second problem would be to
The underlying point is to use language which en- develop interoperability mechanisms between
sures that a digital platform upholds labour rights platforms to enable platform workers to carry
where it is “in charge”.9 Such language need not their ratings, work and financial histories across
impose excessive responsibility – it can be qual- platforms in order to access work, and to also
ified by well-known expressions such as “so far as use them to access social security (Schmidt 2017).
is reasonably practicable”. This directs attention There is therefore a close link between portability
to the practical work arrangements that can be of data and job mobility.
shaped by the platform.
6 See, for example, The Child and Adolescent Labour (Prohibition and Regulation) Act, 1986 (India), section 3.

7 See, for instance, Fair Work Act, 2009 (Australia), section 550 (see also Johnstone and Stewart 2015).

8 See, for instance, Workplace Safety and Health Act, 2009 (Singapore), section 14 (in relation to “principals”, defined in section 4).

9 See the analysis of Prassl (2018, 101–102), who advocates a “functional” approach to assigning responsibility. See also Fudge

(2006, especially 622–625); Davies and Freedland (2006); Prassl (2015); Prassl and Risak (2016).

Table 5.3 Further elements of decent work for platform workers: Convention principles that
X
could be adapted to all digital labour platform workers, irrespective of contractual status
Labour standards Comment on the application of ILO instruments
The principles in the Protection of Wages Convention, 1949 (No. 95), pertaining to:
payments in legal tender; direct payments to workers; free disposition of wages;
prohibition of improper deductions and bonds; regularity and timeliness of payment;
full payment upon termination of the work contract; payment records; and information
as to applicable pay rates, are all relevant to work organized through digital labour
Payment systems
platforms. So are prohibitions preventing private employment agencies from charging
fees, set out in Article 7 of the Private Employment Agencies Convention, 1997 (No. 181),
a Convention of broad application. These principles need to be adapted to deal with
specific characteristics of platform work, such as automatic monitoring, evaluation and
rejection of work by the platforms’ algorithms.
While the Termination of Employment Convention, 1982 (No. 158), is limited to the
employment relationship, the core principle that a work relationship should not be
Fair termination terminated arbitrarily or unjustly is pertinent to arrangements between a digital labour
platform and a worker, at least where arrangements have been ongoing and are likely
to continue.1
The processing of personal data by digital labour platforms should respect

workers’ privacy, a principle recognized in Article 6 of the Private Employment
Agencies Convention, 1997 (No. 181).2 The related Private Employment Agencies
Recommendation, 1997 (No. 188), provides that “[m]easures should be taken to
Access to data and privacy
ensure that workers have access to all their personal data as processed by automated
or electronic systems, or kept in a manual file” (Para. 12(2)). These measures should
include the right of workers to obtain and examine a copy of any such data and the right
to demand that incorrect or incomplete data be deleted or corrected.
The principle that workers should be informed of the terms and conditions of their
engagement “in an appropriate, verifiable and easily understandable manner”
and preferably in written contracts appears in Article 7 of the Domestic Workers
Clear terms of engagement
Convention, 2011 (No. 189), and in Paragraph 5 of the Home Work Recommendation,
1996 (No. 184).3 Again, that principle can be adapted to apply to contractual
arrangements with digital labour platforms.
Job mobility here refers to the capacity of workers to leave their engagement with one
platform and commence work with another, or to work independently of a platform.
Job mobility
This principle is reflected in Article 1 of the Employment Policy Convention, 1964
(No. 122), which refers to “freely chosen employment”.4
While there is no specific ILO Convention dealing comprehensively with grievance

and dispute resolution processes,5 such processes are essential to give effect to the
standards in ILO instruments. Guidance on grievance processes can be found in the
Grievance and dispute Examination of Grievances Recommendation, 1967 (No. 130). Adapting these processes
resolution to digital labour platforms would include enabling workers to contest ratings and
evaluations. Grievance and dispute resolution processes should be efficient, fairly
constituted, and accessible to the parties (Budd and Colvin 2008; Ebisui, Cooney and
Fenwick 2016).
1
Note the concept of “valid reason” in the Convention. 2 See also the ILO Code of practice on the protection of workers’
personal data. 3 See also the Private Employment Agencies Recommendation, 1997 (No. 188), Para. 5. 4 See also the
Employment Policy (Supplementary Provisions) Recommendation, 1984 (No. 169). The issue may be thought of as a “right
to change jobs”, an aspect of the right to work found in other international instruments such as the International Covenant
on Economic, Social and Cultural Rights (ICESCR). 5 Note the Voluntary Conciliation and Arbitration Recommendation,
1951 (No. 92).
Figure 5.1 Decent work elements applicable to all platform workers

irrespective of contractual status
Labour standards in other Labour standards in other

Fundamental principles
ILO instruments of ILO instruments relevant
and rights at work
general application to platform work
 Freedom of association  Occupational safety and health  Payment systems

and collective bargaining
 Social security  Fair termination
 Elimination of discrimination
 Job creation policies  Access to data and privacy
 Elimination of forced labour
 Access to labour inspection  Clear terms and conditions
 Effective abolition of
 Job mobility
child labour
 Access to dispute
resolution
The contention that the standards in table 5.3 Figure 5.1 draws the discussion in this sec-
constitute elements of decent work for all workers, tion together; it sets out the principles and rights
notwithstanding their varied scope of application necessary to provide decent work for all platform
in the Conventions, is reinforced by the fact that workers. The nature of platform work means that
they are present in international human rights several relatively novel questions arise about
treaties10 and in other regulatory contexts where how to implement the principles and rights.
there is information asymmetry and inequality of For example, how should the right to collective
bargaining power. For example, consumer protec- bargaining be conceived in the context of online
tion legislation, which also deals with contracts web-based platforms? How do principles relating
of adhesion imposed by a firm on individuals, to fair termination and data access operate in the
likewise contains provisions dealing with fair context of ratings and deactivation? How do prin-
pricing, fair termination rules, privacy, access to ciples concerning health and safety at work apply
data, transparency of terms and dispute resolu- in the context of contest-based and competitive
tion.11 In many jurisdictions, consumer protection programming platforms? These are issues that
law – which is being regularly amended to respond need to be discussed in future.
to transactions via digital platforms – will thus be a
useful additional source for designing appropriate
regulatory frameworks for platform work.12
10 See the general comment on Article 7 of the International Covenant on Economic, Social and Cultural Rights (ICESCR) https://
tbinternet.ohchr.org/_layouts/15/treatybodyexternal/Download.aspx?symbolno=E%2fC.12%2fGC%2f23&Lang=en, in particular
Para. 4 on scope, and general comment on Article 6 by the Committee on ESCR (E/C.12/GC/18). See also Para. 6 of the general
comment on Article 6 of ICESCR, which extends “decent work” to a somewhat broader notion and extends the rights to independent
workers https://tbinternet.ohchr.org/_layouts/15/treatybodyexternal/Download.aspx?symbolno=E%2fC.12%2fGC%2f18&Lang=en).
11 See, for instance, the numerous EU directives dealing with unfair contract terms, unfair pricing, consumer rights and digital trans-
actions, including: Council Directive 93/13/EEC; Directive 98/6/EC; Directive 2005/29/EC; Directive 2011/83/EU; Directive 2019/2161.
12 See, for example, Superintendencia Industria y Comercio, Colombia, Order of 5 September 2019 concerning the digital labour
platform Rappi S.A.S.
In many jurisdictions,
rights which should, according
5.2.3 Elements of decent work to the ILO instruments,
closely tied to employment: apply to all workers are
The Employment Relationship often in practice limited
Recommendation, 2006 (No. 198)
only to employees in laws
There are other issues that are elements of decent
work but are not, as far as ILO instruments are and judicial decisions.
concerned, applicable to all workers; they are tied
to the employment relationship. Examples include
This principle is known as the “primacy of facts”
working time,13 certain forms of leave (such as ma
(ILO 2020h, Para. 230). Recommendation No. 198
ternity leave)14 and rates of remuneration. Platform
provides systematic guidance concerning the
workers who are genuinely self-employed, with
concepts (such as subordination and dependence)
the capacity to generate their own income through
and indicators (such as control and integration)
their entrepreneurial activities, generally will not
that legal authorities can use to establish an
fall within the scope of these instruments.
employment relationship (Paras 12 and 13). It also
It is therefore crucial for legal systems to have refers to the possibility of presuming or deeming
sophisticated principles for determining whether certain workers to be employees (Para. 11).15
a worker is in fact an employee, however they are
In a recent review of the application of the Recom
described in terms of service agreements. This
mendation to digital labour platforms, CEACR
is especially so where employment is disguised
has highlighted the very diverse approaches
as self-employment in order to avoid the obliga
taken in jurisdictions around the world toward
tions which flow from employment protection
determining the employment status of platform
legislation. Moreover, in many jurisdictions, rights
workers, emphasizing that the “primacy of facts”
which should, according to the ILO instruments,
principle should be invoked when making this
apply to all workers are often in practice limited
determination: “this new form of work calls for
only to employees in laws and judicial decisions.
a thorough examination of the real conditions
Thus, the practical impact of the rules determining
of such workers, which is not always readily ap
employment status often extends well beyond their
parent” (ILO 2020h, Para. 326).
implications for employment-related ILO standards.
It follows that, from the perspective of Rec
The ILO Employment Relationship Recommen
ommendation No. 198, the clauses in the terms
dation, 2006 (No. 198) (see discussion in ILO 2016;
of service agreements stating that the relationship
ILO 2020h), provides that the determination of the
between the worker and the platform is not one
existence of an employment relationship,
of employment are not definitive. While judges
should be guided primarily by the facts relating may, depending on the jurisdiction, give weight
to the performance of work and the remuner to a contractual term defining the relationship as
ation of the worker, notwithstanding how the purely commercial, where the primacy of facts
relationship is characterized in any contrary principle is applicable it is the actual operation of
arrangement, contractual or otherwise, that the arrangements which will frequently determine
may have been agreed between the par the true nature of the contractual relationship
ties. (Para. 9) (see also Waas et al. 2017). This issue is further
discussed in detail in section 5.3.10.
13 There is a short discussion on working time and the gig economy in ILO 2018c, Paras 750–757.
14 Article 2 of the Maternity Protection Convention, 2000 (No. 183), provides that this Convention applies to all employed women,
including those in atypical forms of dependent work.
15 See, for example, section 2750.3 Labor Code (California); Code du travail (France) Arts 7311-1 and 7311-3; Dockès 2019.
There are many aspects

of decent work which
5.2.4 Employment-related
standards and self-employed should extend to platform
platform workers workers, irrespective of
Even if it is the case that platform workers are their contractual status.
truly self-employed, does this mean that principles
relating to working time, leave and remuneration
should have no application to them? Certainly, The precise mode of setting remuneration
genuine entrepreneurs engaged in commercial standards for self-employed platform work is
relationships are understood to control their own controversial and complex, and the problem is
hours and to bear the financial risk of failure or compounded by the diversity of arrangements
success. Nonetheless, there are cogent arguments under which platform work can be carried out.
for maintaining that some degree of regulatory One possibility would be to enable platform
intervention on remuneration and working workers who are genuinely self-employed to
time may be appropriate for the self-employed organize and negotiate base rates with the plat-
engaged through platforms, even if it were of a forms through collective bargaining. However, as
different nature to that governing employment discussed below, competition regulation in many
relationships (Goldman and Weil 2020). jurisdictions is a major obstacle to this approach,
so the specification of applicable pay rates by gov-
First, in relation to working time and leave, exces-
ernments is increasingly being considered.
sive hours are a risk to health and safety (see ILO
2018c). It was made clear above that occupational Summing up the discussion so far in this part of
safety and health regulation should be applicable the chapter, there are many aspects of decent
to all workers, not just employees. work which should extend to platform workers,
irrespective of their contractual status. The
Second, in relation to remuneration (see Cherry
question of whether conditions which have
and Poster 2016), inadequate pay naturally un-
historically been closely tied to employment con-
dermines effective payment systems (which, as
tracts, such as working time and remuneration,
mentioned in table 5.3, should apply to all working
should be extended to genuine self-employed
people). A clear base pay rate is a prerequisite for
platform workers is more vexed. The discussion
identifying non-compliance with payment obliga-
has suggested reasons for addressing these
tions, such as where a bond has been imposed
matters while recognizing the distinctiveness of
or there has been an improper deduction (Prassl
genuine self-employment.
2018).Furthermore, inadequate pay arrangements
may impose social security costs on the wider The next section moves from questions of prin-
society, since social security systems should be ciple to examining developments in practice
universal. For example, they may mean that a around the world.
worker needs to supplement income through un-
employment benefits. In addition, they may mean
that there is insufficient contribution to schemes
such as retirement pensions, where these are
linked to work income. And inadequate pay can
drive excessive working hours.
5.3 Achieving decent work:
Regulatory responses to platform work
In most of the jurisdictions examined here, the ILO Member States that ratify Convention No. 98
regulatory response to platform work is in flux. are required to establish machinery to protect
Draft legislation is being keenly debated. Court workers against anti-union discrimination and to
judgments conflict or are under appeal. The take measures to encourage and promote volun-
impact of newly formed associations of platform tary negotiation between employer and worker
workers remains uncertain. It may take several associations. Conventions Nos 87 and 98 establish
more years before the contours of labour and one particular form of social dialogue – collective
social protection for these workers take shape. bargaining – as a fundamental right of all workers.
Nonetheless, there have already been some land- This could be used to enable workers to negotiate
mark developments. matters such as minimum rates of pay, the design
of fair ratings systems, dispute resolution systems
The discussion here follows the structure of
and reasonable working hours, irrespective of
the previous section. It begins by considering
how the workers are classified contractually.
initiatives linked to the principles set out in
Consensual agreements between platforms and
figure 5.1, then examines the various approaches
representative organizations of workers could
to determining the employment relationship and
also ease the regulatory burden on States, which
the issues of working time and remuneration. The
would not need to legislate and enforce to the
final section discusses the potential applicability of
same degree.
labour standards in trade agreements to platform
work. For considerations of space, not all issues The approach reflected in Conventions Nos 87 and
are examined here in depth, such as employment 98 contrasts markedly with that used by the digital
creation and labour inspection. labour platforms, as platforms tend to regulate
unilaterally. The Conventions, on the other hand,
are directed at enabling regulatory collaboration
5.3.1 Freedom of association, through social dialogue, including, when gov-
collective bargaining and ernments are involved, tripartism. Further, the
concept of collective bargaining set out in the
other forms of social dialogue Collective Bargaining Convention, 1981 (No. 154),
involves “negotiations … for (a) determining
As discussed above, all workers, including platform
working conditions and terms of employment;
workers, enjoy the rights to organize and engage
and/or (b) regulating relations between employers
in collective bargaining under Conventions Nos 87
and workers …” (Art. 2) (emphasis added).
and 98. ILO Member States that ratify Convention
No. 87 are required to permit “[w]orkers and If the platforms seem hesitant to engage in social
employers, without distinction whatsoever, … dialogue, what has been the response of workers
to establish and, subject only to the rules of the and governments? In many jurisdictions, platform
organisation concerned, to join organisations of workers have been associating in order to improve
their own choosing without previous authorisa- their conditions, whether through negotiations
tion” (Art. 2). This Convention enables workers
to associate in a broad variety of organizational
forms (ILO 2018b, Para. 502).
The regulatory response
to platform work is in flux.
with the platforms or by lobbying governments to legal. Many jurisdictions do not permit workers
adopt new regulatory initiatives. Sometimes plat- who are not in an employment relationship to
form workers join established unions;16 sometimes bargain collectively. This is sometimes because
they form new organizations.17 collective bargaining laws do not extend to the
self-employed (Beaudonnet, 2020).
From the union point of view, what we
demand as a union is protection for this But a more fundamental issue is that many com-
work. A minimum assurance both in terms petition (antitrust) laws prohibit the self-employed
of income and in terms of labour law: occupa- from engaging in coordinated negotiation on the
tional safety, work accidents, security. Well, if basis that this would constitute a cartel (see, for ex-
the driver crashes, that would be a work acci- ample, Kilhoffer et al. 2020; OECD 2019c; Johnston
dent. – App-baseddriver,PresidentoftheUber V and Land-Kazlauskas 2019; Lianos, Countouris and
RegiónTradeUnion,thefirstformallyregistered De Stefano 2019; Aloisi and Gramano 2018). This
tradeunionofapp-baseddrivers(Chile) is problematic since some platform workers are
genuinely self-employed but exhibit a comparable
Whatever their structure, worker associations
degree of economic dependence to employees
need to engage in social dialogue with the plat-
(McCrystal 2014).18
forms and regulatory authorities if they are to
achieve their objective of protecting the inter- The EU provides a prominent example of how
ests of their members. In principle, collective competition law may inhibit collective bargaining
bargaining offers a way to reshape the terms by self-employed platform workers. Jurisprudence
of service agreements in a more equitable way, of the Court of Justice of the EU generally prohibits
since it usually involves the conclusion of a binding bargaining by genuinely self-employed workers
legal instrument that has been negotiated rather who are considered as “undertakings” (Schiek
than unilaterally designed (see, for example, and Gideon 2018).19 However, the precise scope of
Rogers 2018). the prohibition is uncertain and is currently being
discussed by EU institutions.20 Some EU Member
However, there are numerous difficulties in
States appear to allow some degree of collective
applying collective bargaining to platform
bargaining by dependent self-employed workers;
workers under existing legal frameworks. The
these include Italy, Germany and Spain (Kilhoffer
difficulties are especially acute for online web-
et al. 2020). Two further recent examples of ap-
based platforms because the workers are
parent inroads into this restriction have occurred
physically dispersed. A first set of obstacles is
in France and Ireland. In France, amendments
16 Such as IG Metall in Germany; Unionen in Sweden; Canadian Union of Postal Workers and Uber Drivers United (part of the United
Food and Commercial Workers), both in Canada; and Independent Drivers Guild in New York (part of the International Association
of Machinists & Aerospace Workers, United States).
17 Such as Asosiasi Driver Online (ADO) in Indonesia; Rider Union in Seoul; National Union of Professional e-hailing Driver partners
(NUPEDP) in Nigeria; the Sindicato Independiente Repartidores por Aplicaciones (SIRA) in Mexico; the Asociación de Conductores
Unidos de Aplicaciones in Chile; the Asociación de Conductores de Aplicaciones de Uruguay in Uruguay; Digital Taxi Forum in Kenya;
United Private Hire Drivers in the United Kingdom; Rideshare Drivers United in California; Philadelphia Drivers in Pennsylvania;
and NYC Taxi Workers Alliance in New York. Some of these associations participate in the International Alliance of App-Based
Transport Workers.
18 While competition regulation has generally focused on product markets, leaving work regulation to govern labour markets,
there is mounting evidence that the two are intertwined. Labour market concentration and other sources of labour power give
rise to “monopsony” or “oligopsony”; one or a small number of firms are able to use a dominant labour market position and their
power to dictate contractual terms, to reduce remuneration and other working conditions and to impede job mobility (Naidu,
Posner and Weyl 2018).
19 See, for example, the decision of the European Court of Justice in Case C-413/13, FNVKunstenInformatieenMedia, ECLI: EU:
C:2014:2411 (holding that collective bargaining by self-employed workers violated EU competition law). On the concept of “worker”
under European law, see Countouris (2018).
20 See https://ec.europa.eu/info/law/better-regulation/have-your-say/initiatives/12483-Collective-bargaining-agree-
ments-for-self-employed-scope-of-application-EU-competition-rules for details.
to the Labour Code (Codedutravail), which in- Kim 2020). This does not mean that the collective
cluded the insertion of a specific title pertaining bargaining rights of the self-employed should be
to self-employed platform workers, 21 accords regulated in the same way as those of employees
those workers a right to collective action which is (Stewart and McCrystal 2019). The CEACR has
protected from retaliatory actions on the part of highlighted tripartite consultations as a vehicle for
the platforms (such as contract termination). The adjusting collective bargaining processes so that
amendments also guarantee the right to organize they can apply to self-employed workers.23
and to “assert collective interests” through unions
But even if the competition law concerning
(fairevaloirparsonintermédiaireleursintérêtscol
self-employed workers can be addressed, national
lectifs)although collective bargaining as such is
frameworks for collective bargaining can give rise
not specifically mentioned (Kilhoffer et al. 2020).
to further difficulties. For example, many countries
Ireland has passed the Competition (Amendment)
require the identification of a “bargaining unit”
Act 2017 which excludes certain categories of
or one or more “representative unions” in order
self-employed workers from its Competition Act
for collective bargaining machinery to operate.24
in order to enable them to bargain collectively.
These concepts are difficult to apply in the context
Outside the EU, some countries permit certain of online web-based platforms, since the work-
categories of self-employed workers to bargain force is diffused across geographical and industry
collectively. In Canada, dependent contractors boundaries. The problem is especially acute in
such as food couriers have been held to have those systems where the law requires collective
this right.22 Other examples include Japan (Waas bargaining to occur predominantly at the level of
et al. 2017) and Australia (McCrystal 2014; the an enterprise.
Australian provisions have been strengthened
with effect from 2021). Moreover, in countries A second set of problems with the viability of col-
such as Argentina, there is no regulation on the lective bargaining concerns practical rather than
issue, and this is widely understood to mean that legal obstacles. Platform workers often do not
there is no prohibition on self-employed workers share a common workplace where they can phys-
organizing themselves for the improvement of ically interact and organize. Where they are able
their working conditions (Goldín 2020). to unite, their organizations may be ephemeral,
lack resources, structure and a clear focus, or out-
Despite these exceptions, competition law in many
comes which retain their members’ enthusiasm.
countries is thus either an actual or a potential im-
pediment to enabling those platform workers who In countries with strong union movements and
are self-employed to exercise their right to bargain robust and flexible bargaining systems (Mundlak
collectively. This impedes solutions to issues such 2020), legal and practical obstacles to negotiating
as pay, working time, evaluation and safety from agreements for platform workers are being in-
being developed through this strong form of creasingly overcome, often with the assistance
social dialogue. International work on reconfig- of established unions. In Denmark, an ingenious
uring the application of competition law so that and groundbreaking collective agreement allowed
it does not undermine bargaining by vulnerable freelance domestic or cleaning workers to take the
self-employed workers is needed (for a sophis- option to transition to the status of employees (see
ticated proposal in the American context, see box 5.1; Kilhoffer et al. 2020).
21 Travailleursutilisantuneplateformedemiseenrelationparvoieélectronique (Codedutravail, Arts L.7341-1 – 7342-11 introduced

by Act No. 2016-1088 of 8 August 2016).
22 CanadianUnionofPostalWorkersvFoodoraInc.d.b.a.Foodora, 2020 CanLII 16750 (ON LRB) Para. 171.

23 See, for example, Observation (CEACR) – adopted 2017, published 107th ILC session (2018) Right to Organise and Collective

Bargaining Convention 1949 (No. 98) – Netherlands (Ratification: 1993).

24 See, for example, Canada Labour Code Part I, Division 3.

Box 5.1 Collective bargaining: Denmark

X
Hilfr and United Federation of Danish Workers (3F) Agreement
In April 2018, the United Federation of Danish Workers (3F) signed a collective agreement with
Hilfr, a Danish-owned digital labour platform which facilitates cleaning in private households.
It has more than 200 active “freelance” workers, most of whom are young, with many being
migrants.
The agreement concerns issues such as optional transition from freelance to employee status,
insurance coverage for all workers, processes to deal with profiles and ratings, and dispute
resolution (which controversially involves arbitration rather than access to the labour courts). The
agreement initially operated successfully (Ilsøe 2020), with more than one third of the cleaning
workers converted to employee status and thus enjoying higher wages and better leave entitle-
ments, although most workers have not joined the union. However, as discussed in Chapter 6,
this agreement has been criticized by the Danish Competition Council.
In situations where collective bargaining is not According to the Leeds Index of Platform Labour
feasible (and often even when it is), platform Protest (Bessa et al., forthcoming; see also
workers often direct their efforts toward obtaining Joyce et al. 2020), since 2015 there has been an
other regulatory interventions ( Johnston and increasing number of such actions around the
Land-Kazlauskas 2019; Wood, Lehdonvirta and world, with at least 1,253 occurring in 57 coun-
Graham 2018; Rodríguez Fernández 2018). One tries between January 2017 and July 2020 (see
approach involves working collaboratively with figure 5.2). Argentina, China, India, the United
governments and platforms to improve working Kingdom and the United States each had over
conditions (albeit without a binding collective 100 protests. These actions continued even during
agreement as an outcome). This may lead to vol- the COVID-19 pandemic.
untary measures.
According to the Index, pay was by far the most
For example, in the Republic of Korea, the prominent cause of dispute actions prior to the
Economic, Social and Labour Council, a presiden- pandemic (64 per cent), followed by employment
tial advisory body, has been providing a forum status (20 per cent), health and safety (19 per
where relevant stakeholders (representatives of cent) and regulatory issues (17 per cent). Health
workers, employers and the government) have en- and safety disputes constituted more than half the
gaged in dialogue. The Council has set up multiple number of disputes since the pandemic started,
committees dealing with issues regarding digital with Latin America being particularly affected. The
platforms, including the Committee on the Digital Index suggests that strikes are associated with pay
Transformation and Future of Work. On 27 May issues, whereas litigation focuses on employment
2020, the Committee announced a code of con- status and regulatory issues.
duct that formulates guidelines for fair contract
Around 80 per cent of dispute actions involve in-
terms between workers and platform companies
formal groups of workers. The involvement of trade
on matters such as payment method, fees, tax,
unions (new or established) varied significantly ac-
non-discrimination, performance assessment
cording to the region, being much more common
programmes and dispute settlement.
in Europe as well as Australia and New Zealand. In
In cases where the platforms do not wish to these regions, together with North America, legal
engage in dialogue, at least initially, platform actions against platforms (frequently relating to
workers increasingly engage in dispute actions employment status) are much more common than
such as strikes, demonstrations and litigation. in other parts of the world.
Figure 5.2 Number of protest incidents related to working conditions on digital labour platforms
per month globally, January 2017 to July 2020
70
Number of protest incidents
60
50
40
30
20
10
0
Jan. 17
Feb. 17
Mar. 17
Apr. 17
May 17
June 17
July 17
Aug. 17
Sep. 17
Oct. 17
Nov. 17
Dec. 17
Jan. 18
Feb. 18
Mar. 18
Apr. 18
May 18
June 18
July 18
Aug. 18
Sep. 18
Oct. 18
Nov. 18
Dec. 18
Jan. 19
Feb. 19
Mar. 19
Apr. 19
May 19
June 19
July 19
Aug. 19
Sep. 19
Oct. 19
Nov. 19
Dec. 19
Jan. 20
Feb. 20
Mar. 20
Apr. 20
May 20
June 20
July 20
Source: Bessa et al. (forthcoming, figure 1).
The ILO country and global surveys also show Drivers Association in the United Kingdom or-
that the level of unionization is quite low among ganized a digital strike against Uber through its
workers on both online web-based and loca- Twitter account, based on the idea of a “digital
tion-based platforms, with 5 and 1 per cent of picket line” which workers would cross if they
microtask and freelance workers respectively, and opened the app.25 The ILO country surveys show
less than 3 per cent of app-based taxi drivers re- that about 9 per cent of app-based taxi drivers
porting being a member of a union, while almost and 3 per cent of app-based delivery workers have
none of the app-based delivery workers reported participated in a coordinated group action such as
that they belonged to a union. a protest, or demonstration, or logging out of the
app, with noticeable differences across countries.
Protest actions such as strikes and demonstra-
tions are often undermined by practical difficulties In the ILO country surveys, about 28 per cent
faced particularly by online web-based platform of app-based taxi drivers and 33 per cent of
workers, essentially due to the physical dispersion app-based delivery workers were members of a
of the platform workforce. Platform workers on social media group related to their work. Virtual
location-based platforms have sought novel groups have also been crucial to mobilizing de-
ways, mainly through “mass self-communication livery workers in Latin America (Hidalgo Cordero
networks”, to increase their associational power and Valencia Castro 2019; Ottaviano, O’Farrell
(Woodcock and Graham 2020; Wood 2015). In the and Maito 2019).26 In Argentina, several couriers
ILO interviews the representatives of unions and organized the first Latin American delivery
worker associations (see Appendix 5, table A5.1) platform workers’ strike through a social media
revealed that they adopt different strategies to group, calling on platform workers to position
organize workers and to bring about legislative themselves in hotspots and accept and cancel
changes. For example, the United Private Hire requests from the platform in order to call
25 ILO interview with representative of United Private Hire Drivers Association in the United Kingdom; the strike was organized
in October 2018.
26 ILO interviews with representatives of MAREA (Mancomunal de repartidores de aplicaciones, Chile), SIRAPPs (Sindicato de
Repartidores de Aplicaciones, Mexico) and Niunrepartidormenos (Mexico).
Health and safety
disputes constituted more than as Johannesburg (South Africa) and Concepción

half the number of disputes (Chile), the story is similar: the lack of effective pro-
tection against platforms deactivating or blocking
since the pandemic started. the accounts of those who attempt to mobilize con-
stitutes a powerful threat against collective action
by app companies.32 Locating this mechanism in
labour law can be problematic where the scope of
attention to detrimental changes in the calcula-
that law is confined to employment relationships.
tion of earnings.27
Another form of action is exemplified by unions in
Austria, Germany and Sweden28 that have estab- 5.3.2 Non-discrimination
lished the Fair Crowd Work website.29 This website
As described in sections 4.2.2 and 4.5, there is
provides analyses of platforms (including reviewing
evidence of discrimination and harassment in the
terms of service and comparing rates of payment),
platform economy, including a gender pay gap in
as well as information for platform workers who
certain location-based services and gender- and
wish to join unions. One innovative feature of the
ethnicity-based exclusion from work opportunities
website is that it collects information from workers
on online web-based platforms in some countries.
and unions in order to review and rate the working
Many terms of service agreements (or related
conditions on different platforms (see also the dis-
policies) studied in the ILO analysis contain prohib-
cussion of Turkopticon in Silberman and Irani 2016).
itions on discriminatory conduct and harassment,
Finally, worker organizations can lobby govern- particularly in the taxi sector (see Appendix 2B).
ments to prompt legislative change. In Costa The Charter of Principles for Good Platform Work
Rica, an app-based drivers’ association sought (WEF 2020) also contains a provision about diver-
to respond to the problem of deactivations by sity and inclusion: “Platforms should strive to be
advocating in favour of an independent admin- inclusive and usable by a diverse population of
istrative body for dispute settlement between workers, and should encourage qualified partici-
app companies and drivers.30 This association is pants from all national, religious, gender, sexual
attempting to incorporate this suggestion into a orientation and ethnic backgrounds, including
Bill, which at the time of writing was before the persons with disabilities” (Principle 1.1).
Costa Rican legislature.31
These are important steps in combating dis-
Whatever form of social dialogue and mobil- crimination, but they do not dispense with the
ization platform workers adopt, there needs to need for other forms of regulation, particularly
be an effective remedy in the event of retaliation statutory and judicial measures that establish
against them for their union activities, as required equality rights. For example, some contractual
under Article 1 of Convention No. 98 (De Stefano anti-discrimination clauses may apply only to the
and Aloisi 2018). According to representatives users, not the platforms. Or they may cover cer-
of platform workers’ associations in cities such tain grounds of discrimination, such as race or sex,
27 ILO interview with legal adviser of App Personal, a platform workers’ association in Argentina.
28 IG Metall (the German Metalworkers’ Union), the Austrian Chamber of Labour, the Austrian Trade Union Confederation, and
the Swedish white-collar union Unionen.
29 See http://faircrowd.work/ for more details.
30 ILO interview with a representative of ACOPLATEC (Asociación de Conductores de Plataformas Tecnológicas y Afines), Costa
Rica, 24 April 2020.
31 Proyecto de Ley Reguladora de los Servicios de Transporte de Personas por medio de Plataformas Tecnológicas, expediente
legislativo No. 21587, presented on 11 September 2019.
32 ILO interview with representative of The Movement (app-based drivers), Johannesburg, South Africa, and with representative
of Agrupación de Repartidores Penquistas (delivery workers), Concepción, Chile.
In some jurisdictions,
anti-discrimination rights
but not others, such as political opinion or union
have been more thoroughly
activities. They may not address the position of developed in the context
workers with family responsibilities,33 an issue that
has become more salient with the impact of the of employment relationships
COVID-19 pandemic (see ILO 2020j).
than in other forms
Many jurisdictions do have well-developed anti-
discrimination laws which are binding on both the of work relationship.
platforms and the platform workers. Nonetheless,
platform work poses difficult questions for equality
regulation. Some of these are novel variations of the implementation of general principles. As we
long-standing and general questions relating to have seen, the Conventions require comprehensive
discrimination law, including: which grounds of dis- application of anti-discrimination law; it should not
crimination should be prohibited? How should apply only to discrimination in employment.
discrimination be determined? Which distinctions
The EU provides an example of this broader appli-
should not be considered discrimination? How
cation in its treaties and directives (Kilhoffer et al.
should causation be understood (and who should
2020; Countouris and Ratti 2018). Both platforms
prove what)? What sorts of relationships should
and workers are required to abide by non-discrim-
be covered by discrimination law? Who should be
ination norms in the context of both employment
responsible where discriminatory conduct is estab-
and self-employment, although different directives
lished? And, what remedies should be available?
apply to employees and analogous “workers” on
To these, new questions can be added that the one hand, and the self-employed on the other.
relate specifically to the digital labour platforms. The prohibited grounds of discrimination are ex-
Barzilay and Ben-David suggest that major shifts tensive, including all the grounds set out in the
in thinking about discrimination are needed in the ILO Discrimination (Employment and Occupation)
context of platform work; for example, they pro- Convention, 1958 (No. 111), as well as age, sexual
pose shifting the goal of anti-discrimination law orientation and disability. 34 Additionally, both
“from aiming to determine who is doing the dis- direct and indirect discrimination are covered,
crimination to answering how the discrimination as is the case with Convention No. 111. Direct
is being effectuated” (2017, 428). The role of algo- discrimination refers to less favourable treatment
rithms merits particular attention: appropriately because of a characteristic (such as sex) whereas
designed algorithms could be less discriminatory indirect discrimination refers to an apparently
than humans – prone to stereotyping, prejudice neutral provision, criterion or practice which re-
and other cognitive errors as we are – but as they sults in unjustifiable unequal treatment because
are designed by humans they can nevertheless of a characteristic.
“reproduce or even exacerbate structural biases”
However, the application of these provisions to
(Bornstein 2018, 570; see also Ajunwa 2020). This
digital labour platforms is problematic in several
issue is considered further in section 6.3.2.
ways. First, in some jurisdictions (such as the
Jurisdictions vary greatly in the manner in which United Kingdom), the implementation of anti-
they address (or not) all these questions in their discrimination rights have been more thoroughly
anti-discrimination law, so that giving effect to developed in the context of employment relation-
the anti-discrimination Conventions in the con- ships than in other forms of work relationship (see,
text of digital labour platforms will take different for example, Fredman 2015; Bamforth 2004). This
regulatory forms. Nonetheless, while no template inhibits independent contractors from resorting
can be provided, there are useful illustrations of to anti-discrimination law.
33 See the Workers with Family Responsibilities Convention, 1981 (No. 156).
34 See, for example, the EU Charter of Fundamental Rights, Art. 21.
Moreover, the application of EU anti-discrimination constitutional provisions comprehensively and

regulation is made more challenging by the frag- include additional grounds, such as family respon-
mented nature of work on digital labour platforms sibility and HIV status. Again, the application of
(Countouris and Ratti 2018; see also Barnard and these laws to digital labour platforms is potentially
Blackham 2017; Blackham 2018). Countouris and far-reaching, but as yet untested.
Ratti (2018) suggest that the way judges approach
Another approach, which instead of focusing on
legal equality protections (such as whether they
the applicability of existing general discrimination
give broad or narrow interpretations) can be cru-
law to platform work aims at sector-specific regu-
cial in determining how adequately they address
lation, is reflected in a bill presented in 2020 by
the complexity of platform work.
the executive power to the Argentine Congress,
International human rights treaties35 also con- which states: “Companies must respect, in imple-
tain broad prohibitions on discrimination, which menting their algorithms, the principle of equality
could capture discriminatory practices in work and non-discrimination.”39
undertaken through digital labour platforms. A final point concerns the application of anti-
For example, in Latin America the Inter-American discrimination law to workers engaged outside
system of human rights has been developing a the jurisdiction in which the platform is based.
strong conception of the right to equality and As discussed in section 4.2.1, workers located
non-discrimination. According to the Inter- in developing countries can find themselves ex-
American Court of Human Rights, “the principle cluded from work on online web-based platforms.
of equality before the law, equal protection If a platform chooses to operate across several
before the law and non-discrimination belongs to jurisdictions, it may be discriminatory to system-
jus cogens [non-derogable international norms], atically disadvantage persons from a particular
because the whole legal structure of national and ethnicity or countries, among other factors, in
international public order rests on it and it is a accessing work.
fundamental principle that permeates all laws”.36
In this scenario, equality and anti-discrimination
rights prevail over domestic laws, including the 5.3.3 Forced labour
law of contracts. A platform worker could ground and child labour
a complaint against a platform on these norms,
irrespective of the contractual form. While forced labour and child labour have not
attracted the same attention as other matters
In South Africa, anti-discrimination law has an
connected with digital labour platforms, there
explicit constitutional support that binds private
is potential for them to occur. For example, De
actors.37 Grounds of discrimination include race,
Stefano (2016) points to the potential for prison de-
sex, pregnancy, marital status, ethnic or social
tainees and children to be engaged in crowdwork.
origin, colour, sexual orientation, age, disability,
religion, conscience, belief, culture, language Many countries now couch their prohibitions of
and birth.38 The Employment Equality Act, 1998, forced labour and (other than in limited circum-
and the Promotion of Equality and Prevention of stances) child labour in comprehensive terms such
Unfair Discrimination Act, 2000, implement these that platforms would be covered if they were used
35 See, for example, the Universal Declaration of Human Rights, the International Covenant on Civil and Political Rights, the
International Convention on the Elimination of All Forms of Racial Discrimination, and the Convention on the Elimination of all
Forms of Discrimination against Women.
36 Advisory Opinion OC-18/03 of 17 September 2003, Juridical Condition and Rights of Undocumented Migrants, Para. 101.

37 The Constitution of South Africa, Section 9(4).

38 The Constitution of South Africa, Section 9(3).

39 Translated from the Spanish text by the ILO: Proyecto de ley presentado por el Poder Ejecutivo Nacional, “Estatuto del Trabajador

de Plataformas Digitales Bajo Demanda”, 6 May 2020.

The regulation of
to facilitate these abuses. An example of a broad
occupational safety and
prohibition is provided by Argentina’s Prohibition health as it is conceived
of Child Labour and Protection of Adolescents Law:
This law applies to the work of persons under
in ILO Conventions and in
eighteen (18) years in all its forms. … The work most national jurisdictions
of persons under the age of sixteen (16) years is
prohibited in all its forms, whether or not there mandates collaborative
is a contractual employment relationship, and
whether or not the work is remunerated. The
workplace arrangements,
labour inspectorate must exercise its functions
to achieve compliance with this prohibition.40
rather than the unilateral
development of policies.
5.3.4 Occupational safety
and health
section 19(1); see also New Zealand, 2015 Act,
This is an area in which many jurisdictions have
section 36(1)). A PCBU must also ensure, “so far
transcended the employment relationship, fo-
as is reasonably practicable, that the health and
cusing on the capacity to influence worker health
safety of other persons is not put at risk from work
and safety (Garben 2019). This thwarts blame-
carried out as part of the conduct of the business
shifting and denial of responsibility based on a
or undertaking” (Australia, 2011 Act, section 19(2);
particular contractual form. The approach puts
New Zealand, 2015 Act, section 36 (2)).
jurisdictions in a good position to craft obligations
with respect to platforms, the workers engaged Workers, too, must take reasonable care for their
through them, and any other individual or entity own health and safety and “take reasonable care
involved in platform-mediated work processes. that [their] acts or omissions do not adversely
The responsibility of these actors is not absolute; affect the health and safety of other persons”
it is bounded in many jurisdictions by the well- (Australia, 2011 Act, section 28; New Zealand, 2015
known phrase – stated in the Occupational Safety Act, section 45). Thus, a driver would be required
and Health Convention, 1981 (No. 155) – “so far as to take reasonable care of passengers allocated
is reasonably practicable”. through a ride-hailing app. Another advantage of
this approach is that it enables all workers, not just
For example, in Australia (Work Health and Safety
employees, to cease or refuse to carry out work
Act, 2011) and New Zealand (Health and Safety at
where there is an imminent and serious danger to
Work Act, 2015), the central concepts of legislation
life or health (Australia, 2011 Act, section 84; New
on safety and health at work are not “employer”
Zealand, 2015 Act, section 83; see also Convention
and “employee” but “a person conducting a
No. 155, Art. 13).
business or undertaking” (PCBU), a “worker” and
“workplace”, all broadly defined (Johnstone and This framework appears well placed to capture
Stewart 2015). “A PCBU must ensure, so far as is the various permutations of digital platform
reasonably practicable, the health and safety of: labour, either as supplemented with specific regu-
(a) workers engaged, or caused to be engaged latory material (which the legislation permits) or
by the person, and (b) workers whose activities with fairly minor amendments (see Stewart and
in carrying out work are influenced or directed Stanford 2017). It is able to address location-based
by the person, while the workers are at work in aspects of platform work, as it is relatively straight-
the business or undertaking” (Australia, 2011 Act, forward to build on existing standards such as
40 Translated from the Spanish text by the ILO: Prohibición del Trabajo Infantil y Protección del Trabajo Adolescente, Ley 26.390,
Art. 2. This prohibition of work to persons below 18 years of age is also explicitly incorporated in the Argentinian Bill referred to above.
Box 5.2 Safety and health at work: Brazil

X
“There is no doubt that [the defendant] centralizes and organizes, via a digital platform, the
connection between workers and third parties (i.e. companies that supply food products and
consumers).
It is the right of workers, in a broad sense, to have the risks inherent to their work reduced,
through health, hygiene and safety standards (Federal Constitution, article 7, caput and XXII).
Part of the responsibility for achieving this right lies with companies (ILO Convention No. 155,
Arts 16/21; Decree No. 1254/94 of the Presidency of the Republic). Law 8080/90 … states that
health is a fundamental human right, whose full exercise must be promoted by the State, without
excluding the responsibility of all, including companies (article 2, caput and paragraph 2). [In]
a broad sense, the nature of the defendant’s activities imposes strict liability for any damages
caused to service providers (Civil Code 927).”
Source: Translated from the Portuguese text by the ILO: Poder Judiciário ||| Justiça do Trabalho Tribunal
Regional do Trabalho da 2ª Região 82ª Vara so Trabalho de São Paulo ||| TutAntAnt 1000396-28.2020.5.02.0082.
those applicable to transportation. For example, Australia, has produced extensive materials for
an undue emphasis on high acceptance rates home-based work pertaining to location, hours of
(discussed in section 4.2.1) could be considered work, equipment, communication methods, and
a safety risk if it led to drivers travelling at exces- work performance and expectations.41 These are
sive speeds or carelessly. Similarly, gamification readily adaptable to platform work.
schemes (discussed in section 4.2.2) which push
Another form of regulatory intervention in the
workers towards excessively long hours and
health and safety space comes from judicial inter-
high-intensity work could be considered injurious
pretations which adapt existing legal principles
to health. And certainly, platforms need to take
to the platform labour market. Thus, in a recent
such measures, so far as is reasonably practicable,
judicial decision from São Paolo, Brazil, the court
to protect drivers, particularly women, from vio-
took a broad view of responsibility for health and
lence and harassment.
safety, again transcending the employment rela-
At first glance, online web-based platforms pose tionship (see box 5.2).
more of a challenge, since the workplace is often
The most influential platforms have indicated their
the home. However, in the legislative model con-
preparedness to accept some responsibility in
sidered here, “workplace” is defined as “a place
this field. Thus, the Charter of Principles for Good
where work is carried out for a business or under-
Platform Work (WEF 2020) provides that:
taking and includes any place where a worker
goes, or is likely to be, while at work” (Australia, Platforms should have policies or guidelines in
2011 Act, section 8(1); New Zealand, 2015 Act, sec- place, appropriate to the locations and modes
tion 20(1)); this clearly covers working at home. The of work, to help protect workers from health
lockdowns prompted by the COVID-19 pandemic and safety risks, and should endeavour to
have required many workers to work from home protect and promote the physical and mental
(telework) and this has led to workplace health and wellbeing of workers. Users/clients should
safety authorities developing clearer standards for acknowledge and adhere to the policies and
such work. For example, the State of Queensland, guidelines (Principle 2.2).
41 See https://www.worksafe.qld.gov.au/laws-and-compliance/workplace-health-and-safety-laws/specific-obligations/health-and-
safety-for-working-from-home.
While this is an important inclusion in the Charter, must also share

the regulation of occupational safety and health data with the tax
as it is conceived in ILO Conventions and in most authorities on
national jurisdictions mandates collaborative worker s ’ incomes
workplace arrangements, rather than the unilat- and activities; those authorities
eral development of policies.42 then transmit relevant details
to the social security agencies.
Some plat form workers can
5.3.5 Social security opt to allow the platform to
Strengthening social protection systems requires deduct contributions and pay
a combination of contributory (mainly social them directly to the authorities
insurance) and non‐contributory, tax‐financed (Ogembo and Lehdonvirta 2020).
social protection mechanisms. While there is no In Spain, mandatory employment injury
“one-size-fits-all” solution, social security can insurance legislation includes workers in
be extended to platform workers by adapting dependent self-employment (Behrendt and
policy, legal and administrative frameworks. Nguyen 2018).
Several countries have introduced innovations A number of countries in Latin America
to enhance coverage of diverse forms of work, have introduced “monot ax ” mech -
including where there are complex and unclear anisms to extend insurance coverage to
contractual relationships. self-employed workers and micro and
One area which illustrates the issues at stake small enterprises. This promotes their
is work injury. Where a platform worker is an transition to the formal economy. For ex-
employee of the platform (or at least a regular ample, in Uruguay, monotax participants pay
employee), the platform will in many jurisdictions a flat rate covering tax and social security con-
be required to pay social insurance contributions tributions, which entitles them (or their workers)
or, where such coverage does not exist, insur- to the same benefits as employees (other than
ance premiums for private injury compensation unemployment benefits); they can also choose
insurance covering that worker. However, where to voluntarily contribute to social health insur-
workers cannot establish an employment relation- ance. The Government has introduced specific
ship they may be required to self-insure. As this is measures to extend coverage to workers on taxi
a significant financial burden for low-paid workers, platforms (Freudenberg 2019). To obtain their
many may fail to do so – with catastrophic con- licence to operate, drivers using taxi apps must
sequences in the event of a major injury. Road be registered with social insurance and tax author-
traffic accidents and other work-related incidents ities under the same conditions as employees. The
are among the most prominent issues for loca- apps allow drivers to register while automatically
tion-based platform work.43 adding a social security contribution to the price
Many countries are developing solutions to

address this issue, as well as adopting broader
measures to extend social protection to plat- Social security can be
form workers. For example, in France, platforms
are liable for the accident insurance fees of
extended to platform workers
self-employed workers, depending on whether by adapting policy, legal and
a threshold for platform usage is reached (Code
dutravail, Arts L7342-2 and 7342-4). Platforms administrative frameworks.
42 See, in particular, the Occupational Safety and Health Convention, 1981 (No. 155), Arts 19 and 20; the Violence and Harassment
Convention, 2019 (No. 190), Art. 9(a). See also the LEGOSH database maintained by the ILO.
43 These concerns constitute one of the priorities of associations of platform delivery workers in Chile and Mexico.
Information technology
also opens up a wider range However, to the extent that an arrangement
of potential solutions for is a purely private scheme, it risks being less
equitable and effective than public schemes.45
extending social protection. This is because particularly vulnerable low-income
earners and workers with non-linear working
careers are unlikely to enjoy adequate levels of
of each ride and transferring it to the Uruguayan
protection, which may in turn aggravate inequal-
social security institution (Behrendt, Nguyen and
ities, including gender inequalities (Behrendt,
Rani 2019; Behrendt and Nguyen 2018). A similar
Nguyen and Rani 2019).
approach is currently being adopted in Brazil,
where the Government plans to extend coverage In the absence of a legal framework clearly
of its monotax mechanism to drivers working on detailing how social protection should apply
digital platforms, granting them access to sick- to platform workers, wider coverage of these
ness, maternity and disability benefits as well as workers can to some extent be achieved through
old-age pensions (La Salle and Cartoceti 2019). case law, as has been occurring in China and the
Republic of Korea (see box 5.3). This approach
Similarly, in Indonesia the government agency
has its shortcomings. The expense and duration
responsible for social securit y (the Badan
of court proceedings can deter workers, and gov-
Penyelenggara Jaminan Sosial Ketenagakerjaan44)
ernments are generally better placed than courts
works in partnership with the financial sector to
to design the type, level, eligibility and financing of
facilitate the making of registration and contri-
social security for platform workers. Nonetheless,
bution payments so as to extend the coverage of
disputes about the application of social protection
work injury and death benefits to Gojek drivers
legislation regularly come before the courts, and
(Indonesia’s largest ride-hailing on-demand plat-
a broad purposive interpretation as in the cases
form). This encourages Gojek drivers to register
illustrated here can help to fill gaps in coverage.
online with the agency, while their social security
contributions are drawn directly from their driver Developments in the countries mentioned here, as
accounts (Nguyen and Cunha 2019). A similar well as in many other jurisdictions, establish that
arrangement exists in Malaysia between the platform workers can effectively be brought under
national social security agency, Perseko, and the the umbrella of social security. In India, the Code
platform company Grabcar (La Salle and Cartoceti on Social Security was introduced in September
2019). In China, the dominant ride-share platform, 2020 to extend protection to all workers, including
DiDi Chuxing (滴滴出行) has set up its own medical platform workers, irrespective of the existence
insurance plan (点滴医保) with contributions from of an employment relationship.46 Guaranteeing
the platform and/or the workers, depending on universal social protection throughout the life
the particular scheme. Some location-based plat- cycle for all, including workers in all forms of work,
forms, like Deliveroo, Glovo, Ola, Swiggy and Uber, based on sustainable financing, solidarity and risk
also provide both drivers and passengers in-ride sharing, is not only a matter of realizing the human
insurance to varying degrees (see section 2.3). right to social security, but is also important in
Deliveroo’s insurance, for instance, covers riders establishing a level playing field between different
against injuries and third-party liability while they types of engagement and ensuring fair competi-
are online and for one hour after they have gone tion between platforms and traditional companies
offline, while Swiggy’s insurance coverage includes (Behrendt, Nguyen and Rani 2019).
compensation of family members in case of illness.
44 See https://www.bpjsketenagakerjaan.go.id/.
45 See http://finance.sina.com/bg/tech/technews/sinacn/2019-07-09/doc-ifzkvvxn2512524.shtml; see also Behrendt and
Nguyen (2018).
46 See https://labour.gov.in/sites/default/files/SS_Code_Gazette.pdf.
Box 5.3 Work injury insurance: China and the Republic of Korea
X
A 2018 judgment of the Haidian District People’s Court in Beijing concerned a courier
who was injured while working through the FlashEx app in Beijing. He sought work
injury insurance benefits from the company operating FlashEx (the platform). The plat-
form denied liability on the basis that the courier was engaged under a “cooperation
contract” rather than a contract of employment and that the work injury insurance
regulations did not apply. The platform pointed to the written agreement and the
courier’s control over work hours and delivery quotas.
The Court rejected these arguments. It held that in reality there were many factors that
pointed to a labour relationship between the platform and the courier. For example,
the courier was dependent on the platform for his income and so worked long hours
each day exclusively for it. The platform also exercised a high degree of control over
the courier.
More fundamentally, the Court considered the social and economic consequences of
finding liability. It noted the important relationship between accident compensation
and health and safety at work. If a platform entity does not bear the financial conse-
quences of accidents suffered by workers using its platform (in terms of insurance
arrangements), it will have little incentive to consider improved safety measures. The
Court further emphasized that “internet companies cannot fail to undertake legal and
social responsibility because they have adopted new technologies and new business
methods”.
Finally, the Court noted that the uncertainty about contractual status should not
deprive the courier of his entitlement to work injury insurance, as this was a basic
labour right.
(LIXiangguovBeijingTongchengBiyingTechnologyCompanyLtd. Labour Dispute. First
instance civil judgment, Beijing Haidian District People’s Court, Minshi Panjueshu
(2017) Jing 0108, Minchu 53634 Hao)1
In the same year, the Supreme Court of Korea (Republic of Korea) adopted a similar pur-
posive approach to the application of the Industrial Accident Compensation Insurance
Act in two cases also involving food delivery couriers. The Court overturned decisions
from lower courts which had relied on contract wording rather than actual practice.
The Court found that the couriers were to be deemed “employee-like” – persons in
special types of employment (PSTE) – for the purposes of the Insurance Act: Supreme
Court Decision, 2016Du49372 Decided 26 April 2018; and Supreme Court Decision,
2017Du74719 Decided 26 April 2018. On remittal, the Seoul High Court determined
that the couriers were indeed PSTEs: Seoul High Court Decision 2018Nu43523 decided
16 January 2019; Seoul High Court Decision 2018Nu44496 decided 17 January 2019.2
Subsequent amendments to the Republic of Korea’s Occupational Health and Safety
Law have brought PSTE within that law’s scope as well (see art. 78).
1
Translated from the Chinese text by the ILO: 李相國與北京同城必應科技有限公司勞動爭議一審民事
判決書, 北京市海淀區人民法院 , 民事判決書, (2017)京0108民初53634號 .
2
Translated from the Korean text by the ILO: 대법원 2018. 4. 26. 선고 2016두49372; and 대법원 2018.
4. 26. 선고 2017두74719. On remittal, the Seoul High Court determined that the couriers were indeed
PSTEs: 서울고등법원 2019. 1. 16. 선고 2018누43523; and 서울고등법원 2019. 1. 17. 선고 2018누44496.
Social protection acts

To be sure, complex questions arise when a person
as an automatic stabilizer
works through a platform only on a sporadic basis for the economy.
or works through multiple platforms. How should
social security mechanisms be adapted so that
even infrequently used platforms are required
to contribute a fair share? How should multiple there legally mandated channels through which
platforms share the costs? These and other ques- workers can participate in health and safety gov-
tions await comprehensive resolution, but such ernance to devise appropriate safety procedures?
questions are not as novel or complex as they
Second, do workers have effective access to health-
may first appear. Intermittent and casual work,
care provided through social health insurance or
including for multiple employers, have a long
national health services without encountering
history and sophisticated regulatory and policy
financial hardship?
responses have been devised. For example, coun-
tries have successfully extended social protection Third, are workers entitled to take leave from
to workers with multiple employers through a work on the grounds of illness or quarantine?
combination of contributory and non-contrib- And if so, are they provided with adequate income
utory social protection mechanisms (ILO 2016; security through paid sick leave or sickness
ILO 2019b). Information technology not only benefits during their absence? Government- and
gives rise to new complexities, it also opens up a employer-provided sick leave varies greatly among
wider range of potential solutions for extending jurisdictions (see OECD data, for example, in OECD
social protection. 2020c, figures 1 and 2).
During a pandemic, the need to prevent loss of
5.3.6 The COVID-19 pandemic income creates a problematic incentive to attend
work despite illness, thereby potentially leading
and its implications for health to infection of others (see box 4.6; ILO 2020b;
and safety at work and social Adams-Prassl et al. 2020). For example, 80 per
cent of virus transmission leading to a major out-
security break in Melbourne, Australia was attributed to
workplaces, “much of it spreading among casual
The COVID-19 pandemic has thrown into sharp
workers without access to sick leave, who worked
relief the interrelationship between platform
while displaying symptoms” (Sakkal and Ilanbey
work, occupational safety and health law, social
2020). Australia has now introduced paid pan-
security and measures to protect the public from
demic leave. As the OECD has pointed out:
the virus. There have been outbreaks in many
work contexts. For instance, if a single delivery Paid sick leave is a crucial tool for addressing
worker has coronavirus, and if that person does the economic impact of the COVID-19 crisis for
not quarantine due to the pressures of earning workers and their families. It can provide some
their livelihood, this can lead to the infection of income continuity for workers who are unable
large numbers of people, including fellow workers to work because they have been diagnosed
and customers. with COVID-19 or have to self-isolate. By en-
suring that sick workers can afford to remain at
This raises multiple regulatory issues. First, are
home until they are no longer contagious, paid
the workers engaged through platforms covered
sick leave also helps to slow the transmission of
by safety and health legislation, irrespective of
the virus (OECD 2020c, 2; see also ILO 2020b).
their contractual status? If so, how can regulators
ensure that delivery and other forms of platform The COVID-19 crisis has highlighted the im-
work conform to legal obligations? How can both portance of both health and safety regulations
platforms and their workers take reasonable and social protection for persons engaged
measures to maintain health, not only of the in all forms of work. They protect individuals
workforce but also of the general public? Are during the crisis by preventing contamination,
providing access to healthcare and responding 5.3.7 Payment systems,

to the massive income losses resulting from the
deep economic downturn. Social protection also fair termination and clear
acts as an automatic stabilizer for the economy, terms of engagement
including by improving consumption and partially
offsetting crisis-induced volatility in aggregate As noted above, laws regulating payment
demand (ILO 2020b). procedures, terminations and transparency of
obligations and entitlements are increasingly
In response to the crisis, some platforms have
common, not just in relation to employment re-
introduced measures to support infected workers
lationships but also in other contexts where one
who need to remain in quarantine. The adequacy
party requires another to enter into a contract of
of these provisions has been contested (Marshall
adhesion, notably consumer contracts. This trend
2020; Fairwork Project 2020; see box 4.5) and
reflects the fact that where one party unilaterally
government measures have therefore been
determines content it may give insufficient regard
required to provide more comprehensive re-
to the interests of the other. For example, the
sponses. For example, Ireland has extended
party in the stronger position may include terms
sickness benefits to all workers currently excluded,
which unreasonably impose an excessive fee, fi-
while the Governments of Finland and the United
nancial penalty or deduction, or enable unilateral
States have extended unemployment benefits to
termination at any time for any reason.
workers not covered by unemployment insurance,
including self-employed workers in the platform Platform work often occupies a “grey zone”
economy (ILO 2020a). between employment and commercial or con-
sumer regulation, but in many jurisdictions there
A Peruvian COVID-19-related protocol47 applicable
is significant convergence between the two fields
to all workers engaged on delivery platforms es-
in relation to fair contract terms. In principle, this
tablishes numerous health and safety obligations
makes it less likely that platform workers will “fall
for platforms, including the sharing of data to
between the cracks”; in practice, the breadth of
prevent an agglomeration of delivery workers in
the scope of application of clauses in relevant
pick-up locations. It requires platforms to create
legislation will have a significant impact on which
mandatory checkpoints for monitoring health con-
workers are in fact covered.
ditions and implementing sanitization measures.
Furthermore, it establishes a common fund to pay An illustration not only of the similarity between
sick leave and related health expenses for delivery the commercial or consumer and the employment
workers (repartidores independientes) who are approaches to payment systems, termination and
either infected with the virus or come into close transparency, but also of how platform workers
contact with infected people. can also “fall between two stools”, can be found
in two recent instruments of the EU. These are the
These are just a few of the many measures taken
Transparent and Predictable Working Conditions
by countries as they scramble to deal with the
Directive 48 (TPWC), which applies to workers
rapidly changing consequences of the COVID-19
(Art. 1(2)) (persons other than the genuinely
pandemic. What is clear is that integrating com-
self-employed); and the Regulation on Promoting
prehensive health and safety and social protection
Fairness and Transparency for Business users
measures into an overall pandemic response is
of Online Intermediation Services49 (P2B), which
critical to combating the virus.
applies to online intermediation services and
47 Resolución Ministerial, Nº 00163-2020-Produce, Lima, 21 de Mayo De 2020; Resolución Ministerial N° 239-2020-MINSA.

48 Directive (EU) 2019/1152 of the European Parliament and of the Council of 20 June 2019 on transparent and predictable working
conditions in the EU.
49 Regulation (EU) 2019/1150 of the European Parliament and of the Council of 20 June 2019 on promoting fairness and transparency
for business users of online intermediation services.
A preferable approach
online search engines provided, or offered to
may be to require platforms
be provided, to business and corporate website to charge commissions to
users (Art. 1(2)).
Kilhoffer et al. write in their study for the European
clients rather than to workers.
Commission that despite their different spheres
of application, “both legal instruments contain […] 2020, 185). In layperson’s terms, this gap would
very similar approaches and material provisions arguably include location-based platform workers
when it comes to the conditions that have to be in the taxi sector. Nonetheless, the similarity
respected by platform businesses in their con- between the two instruments suggests that the
tractual relationship with professional individuals core common features could be readily extended
who are making use of the digital apps when de- to cover these excluded workers.
livering their services” (2020, 173; the analysis in
this section draws heavily on this study). Broadly Another issue which does not yet appear to have
speaking, these two instruments each regulate been addressed comprehensively relates to com-
the following issues: missions and related fees, especially those set by
online web-based platforms and taxi platforms.
timely information as to the “essential aspects”
X
As discussed in Chapters 2 and 4, many platforms
of the employment relationship (TPWC) and charge commission fees, which can be as high as
“plain and intelligible” terms and conditions 25 per cent. There are two conflicting consider-
(P2B) of the relationship; ations here. One is that the platforms need to have
circumstances in which contract modifications
X a revenue stream in order for their business model
can be made; to operate; a commission is a well-recognized
mode of generating one. On the other hand, there
rules as to whether work for other businesses
X
has long been a policy concern that third-party in-
can be restricted;
termediaries should not impose costs on workers.
remuneration rules and, in the case of the
X This is reflected in the ILO Private Employment
TPWC, rules about the components, frequency Agencies Convention, 1997 (No. 181) (Art. 7; see
and methods of payment; also De Stefano and Wouters 2019), and also in
domestic legislation. For example, the Labour
rules as to termination, including specification
X
Standards Act of the Republic of Korea contains
of reasons which, in the case of the P2B, must
the following provision:
be related to grounds in the contract; and
Elimination of Intermediary Exploitation: No
methods of redress.
X
person shall intervene in the employment of
In order to prevent abusive practices related another person for making a profit or gain
to atypical contracts, such as on-demand or benefit as an intermediary, unless otherwise
zero-hour contracts, the TPWC urges Member prescribed by any Act (Art. 9).
States to apply a rebuttable presumption of the
There are ways of reconciling these two objectives.
existence of an employment relationship (Art. 11).
One is to maintain that the policy concern applies
Meanwhile, the P2B (Art. 5) requires that providers
only to employees of online web-based platforms,
of online intermediation services clearly outline
not the self-employed. However, it is not clear
the parameters that determine the ranking of
that employees are the only category of worker
users and their relative importance.
that should be protected against excessive com-
Unfortunately, the two instruments leave one missions, especially if the boundary between
gap which requires quite convoluted language to employees and the self-employed is blurred. A
specify: genuine “self-employed platform workers preferable approach may be to require platforms
who are not relying on platforms that are purely to charge commissions to clients rather than
providing digital intermediation information workers and/or to limit commissions by criteria
society services to consumers” (Kilhoffer et al. such as proportionality.
The right to data

portability is particularly
5.3.8 Access to data,
important for platform
privacy and job mobility
Recent years have seen an increasing regulatory
workers.
focus on privacy and data protection. Many
developments in this field are highly relevant to
are introducing similar laws. For instance, both
platform work, not only because data collection
Nigeria’s Data Protection Regulation, 2019, and
and transfers are central to the business model
India’s Personal Data Protection Bill, 2019, include
of digital labour platforms but also because the
the qualification of technical feasibility. The latter,
new laws apply to data subjects irrespective of
currently before the Indian Parliament, provides
employment status.
a further qualification whereby the right to data
One such development is the emergence of data portability is further restricted in relation to trade
protection regimes such as the EU’s General Data secrets (Art. 19).
Protection Regulation (GDPR),50 in force since May
The right not to be subject to decisions based
2018. The Regulation establishes several individual
solely on automated processing could also have
rights, such as the right to be informed, to access
a major impact. Automated decision-making is
data, to data portability, to data erasure, and
central to key aspects of platform operations,
not to be subject to a decision based solely on
from price-setting and the matching of users to
automated processing. Meanwhile, the California
determination of users’ reputational status and
Consumer Privacy Act in the United States, which
deactivation. This right could address the opacity
came into force in January 2020, establishes a
of the algorithms used by platforms, which lies at
specific right to opt out of the sale of personal
the root of the major concerns shared by many
information (section 1798.120). Platform workers
platform workers, such as wrongful deactivations
can benefit from these rights. For instance, several
and changes in platform pricing mechanisms.
app-based drivers in the United Kingdom have
filed a lawsuit against Uber for withholding their The EU’s GDPR and similar laws in other jurisdic-
data contrary to the GDPR, while others have set tions do permit the right not to be subject to fully
up a cooperative which pools driving data and automated decisions to be overridden where the
uses it not only to help workers optimize their processing of personal data is necessary for the
income capacity, but also to assist city agencies in performance of a contract, is authorized by law or
making more informed and effective transporta- is based on consent. However, workers still have
tion planning decisions.51 the right to obtain human intervention, express
their point of view and contest the decision in
The right to data portability is particularly im-
question. To understand the scope of this right, it
portant. According to Article 20 of the GDPR,
should be read together with other clauses of the
platform workers can obtain a copy of their data “in
GDPR, such as Articles 13(2)(f) and 14(2)(g), which
a structured, commonly used and machine-read-
stipulate that, with regard to automated decisions,
able format”. They also enjoy the right to have the
the data controller should provide “meaningful
data transmitted directly from one controller to
information about the logic involved, as well as
another, where “technically feasible”. While this
the significance and the envisaged consequences
broad provision might seem to largely address the
of such processing for the data subject”.
data portability problem, the condition of technical
feasibility is a potential obstacle. This is apparent Moreover, data protection laws outline certain
not only in the EU, but in other jurisdictions that principles and the legal basis for processing
50 Regulation (EU) 2016/679 of the European Parliament and of the Council of 27 April 2016 on the protection of natural persons
with regard to the processing of personal data and on the free movement of such data, OJ L119.
51 Driver’s Seat Cooperative: https://www.driversseat.co.
A strong case can be

personal data, which generally include transpar-
made for platforms to be
ency and consent. Some regulations, however, subject to the courts and
refer to more specific principles. For instance, the
Brazilian General Data Protection Law (Lei Geral tribunals of the jurisdiction
de Proteção de Dados 2018) requires that the
processing of personal data be non-discriminatory
in which the worker is based.
and protect credit (arts 6(IX) and 7(X)). While data
protection laws will have an impact on the worker–
that workers have access to transparent and
platform relationship, it is too early to determine
accountable mechanisms, where applicable, for
how extensive this will be. Their application may
resolving disputes with users/clients and with
be hindered in practice by other areas of law such
other workers within a reasonable timeframe”
as contract and trade secrecy, both of which,
(WEF 2020, Principle 7.2).
separately and in combination, contribute to the
opacity of platform algorithms and their insulation Fair internal review processes are crucial here.
from scrutiny (De Stefano 2019; Kapczynski 2020). They lead to efficiencies in terms of reduced
Moreover, the concept of “consent” as a means of costs and time, and, as discussed in section 4.3.2,
authorizing data processing may be problematic often produce favourable outcomes for workers.
for platform workers, particularly given that there However, internal processes do not obviate the
is often a serious imbalance in bargaining power need for access to external dispute resolution
(Todolí-Signes 2019). mechanisms, such as courts. This is particularly
so where it is a business practice (such as the
Finally, in addition to these general data protection
characterization of the worker) that is being chal-
laws, national laws specifically relevant to platform
lenged, rather than an individual decision (such as
worker data are emerging. For example, in France,
an erroneous rating).
one of the jurisdictions where the GDPR directly
applies, the Codedutravail was amended to in- The discussion in section 5.1.1 showed that in
clude the following provision: some jurisdictions, digital labour platforms can
unilaterally choose the kind of external dispute
[Self-employed platform-based workers] have
resolution system to be specified in their terms
the right to access all platform data concerning
of service agreements. The danger of such uni-
their own activities that enable them to be iden-
lateral systems is that access to the general court
tified. They have the right to receive this data in
system can be blocked. This means that where, for
a structured format and the right to transmit it.
example, there is a consistent misclassification of
The precise scope of this data as well as their
workers, a structural flaw in the internal review
access, extraction and transmission methods
system, or a systemic problem with algorithmic
will be defined by decree (Art. L7342-7).
decision-making, it is not subject to judicial over-
sight even where this leads to deactivation.
5.3.9 Grievance and A strong case can be made, therefore, for plat-
dispute resolution forms to be subject to the courts and tribunals
of the jurisdiction in which the worker is based,
Section 4.3.2 described several situations in including general procedural rules such as class
which platform workers need to access grievance actions. In some jurisdictions, notably in civil law
and dispute resolution processes, such as for countries such as Argentina, Chile and Mexico,
contesting poor performance evaluations, low the law requires that disputes be referred to the
ratings and work rejection (especially on microtask ordinary court system, and terms of service agree-
platforms), and, for many workers, temporary or ments often reflect that requirement.
permanent deactivation.
In those jurisdictions where access to the courts
The Charter of Principles for Good Platform can only be restricted on limited grounds (many
Work provides that “platforms should ensure common law jurisdictions other than the United
States), platform workers are bringing litigation rights given by the contract unenforceable by
that challenges the dispute resolution mech- a driver against Uber. No reasonable person
anisms in platforms’ terms of service agreements. who had understood and appreciated the im-
Challenges to clauses referring disputes to arbitra- plications of the arbitration clause would have
tion outside the jurisdiction where the services are agreed to it (Heller litigation, Para. 95).
being provided have been particularly successful.
The argument that the local court was not the
A prominent recent example is the Canadian appropriate forum (the platform being based in
Heller litigation,52 which resulted in a major judg- Europe) was also put forward by the platform
ment of the Supreme Court of Canada in 2020 company, but again rejected in the Aslamcase in
(see Coiquaud and Martin 2019, noting that their the United Kingdom (discussed in section 5.3.10).
article was written before the final decision in the Despite these adverse decisions, it does not follow
litigation). Heller was an Uber Eats courier who to- that dispute resolution clauses in terms of service
gether with co-workers commenced a class action agreements have no practical effect – even if they
against Uber arguing that they were employees are invalid, they may deter potential litigants who,
with entitlements under Ontario’s Employment not being legal experts, believe they are effective.
Standards Act 2000. Uber applied for a stay – an
Effective dispute resolution has been hindered
order that the court cease hearing the matter.
in some developing countries owing to a lack of
This was on the basis that Heller was bound by a
clarity with regard to the appropriate legal entity
mandatory arbitration clause that required him to
against which to bring a claim. For instance, a
submit his dispute to mediation and then arbitra-
claim concerning unfair dismissal lodged by Uber
tion in the Netherlands.
drivers in South Africa failed because the drivers
The Supreme Court of Canada held that the sued a member of the Uber group incorporated in
arbitration clause was invalid on the ground of South Africa, rather than Uber BV, a private limited
unconscionability.53 The majority wrote: company registered in the Netherlands, which was
held to be the relevant contracting party.54
The arbitration clause, in effect, modifies every
other substantive right in the contract such Attempts by location-based platforms to subject
that all rights that Mr. Heller enjoys are subject the work relationship to the law of an external
to the apparent precondition that he travels to jurisdiction are also likely to fail. For example, pri-
Amsterdam, initiate an arbitration by paying vate international law in China, the EU, India and
the required fees and receive an arbitral award the State of California in the United States entails
that establishes a violation of this right. It is that, at least in simple cases not involving cross-
only once these preconditions are met that Mr. border work, the relevant law will usually be that
Heller can get a court order to enforce his sub- of the jurisdiction in which the worker is located,
stantive rights under the contract. Effectively, especially if an employment relationship may be
the arbitration clause makes the substantive involved or labour statutes are being applied.55
52 UberTechnologiesInc.v.Heller, 2020 SCC 16.

53 The Canadian approach to unconscionability has two elements: an inequality of bargaining power and an improvident trans-
action. Other common law jurisdictions have different tests.
54 UberSouthAfricaTechnologyServices(Pty)LtdvNationalUnionofPublicServiceandAlliedWorkers(NUPSAW)andOthers (C449/17)
[2018] ZALCCT 1; [2018] 4 BLLR 399 (LC); (2018) 39 ILJ 903 (LC). A similar case was observed in Kenya, KanuriLtdandothersvUber
KenyaLtd, High Court of Kenya at Nairobi, Civil Suit No. 356 of 2016, available at: http://kenyalaw.org/caselaw/cases/view/134444.
55 As Cherry shows in her study of California, the EU and India, courts will not automatically give effect to choice of law clauses in
terms of service agreements, preferring to place considerable weight on the physical location of the worker (Cherry 2020, 204–228).
See also Grušić (2012); Law of the People’s Republic of China on Choice of Law for Foreign-related Civil Relationships, adopted at
the 17th session of the Standing Committee of the 11th National People’s Congress on 28 October 2010; Interpretation of the
Supreme People’s Court on Several Issues Concerning Application of Law of the People’s Republic of China on Choice of Law for
Foreign-related Civil Relationships (I), as adopted at the 1563rd Session of the Judicial Committee of the Supreme People’s Court
on 10 December 2012, article 10; Council Regulation (EC) 44/2001 of 22 December 2000 on jurisdiction and the recognition and
enforcement of judgments in civil and commercial matters, Section 5.
This is just as well since, as Cherry (2020, 226) 5.3.10 The employment
comments: “if multinational platform operators
can choose the law that they impose through relationship
an adhesive contract, they might decide to pick
While this chapter has emphasized that many
jurisdictions in which there either is no law; ones in
principles and rights cover workers irrespective of
which there seems to be a favorable precedent or
their contractual status, the reality is that in many
the likelihood of one; or jurisdictions where labor
jurisdictions, laws giving effect to those principles
standards are quite low”.
and rights apply only to employees. This reality
The regulatory position of online web-based has been repeatedly revealed in the foregoing
work – which takes place in the “virtual world” – is discussion, which has pointed to the limited scope
potentially more complicated and little explored of much labour protection regulation in areas such
(Cherry 2020).56 In contrast to location-based plat- as collective bargaining and social security.
forms, the platform, the worker and the client can
One reason for the close connection between
be in three different countries (Cherry 2020; see
labour protection and the employment rela-
also Arthurs 2010; Mundlak 2009). For example,
tionship is that many labour laws were initially
many Amazon Mechanical Turk workers are based
drafted in a period in which the boundar y
in India (Difallah, Filatova and Ipeirotis 2018;
between employment and entrepreneurship
Berg et al. 2018), as are many Upwork workers
seemed to be easier to identify, at least in indus-
(Horton, Kerr and Stanton 2017). The platforms
trialized countries (Deakin 2007). That era of
themselves are based in the United States, and
vertically integrated industrial firms has passed,
users (clients) can be located in many different
fundamentally unsettling the binary employee–
countries, although commonly in the more devel-
entrepreneur divide (Weil 2014; Fudge, McCrystal
oped economies (see section 1.3).
and Sankaran 2012; Davidov and Langille 2011;
Of itself, this geographic dispersion is not novel. Freedland and Countouris 2011; Fudge 2006; Stone
International secondments of workers have been 2004; Freedland 2003; Supiot 2001; Collins 1990).
common for many years, and legal principles However, legislatures and courts have often failed
have developed to determine the governing law to thoroughly recast labour protection in a more
to apply in the event of disputes (Morgenstern comprehensive direction (safety and health at
1985; Morgenstern and Knapp 1978). But with work, forced and child labour and non-discrimin-
online work, the workers do not in fact leave their ation law being prominent exceptions). In short,
domestic legal system. It is the terms of service while in theory employment status should not be
agreements which travel, virtually, to meet the relevant to principles and rights which apply to all
workers where they are based. These agreements workers, in practice it very much is.
may attempt to import foreign restrictions on
But appropriate classification of workers is also
worker entitlements. There is a paucity of cases,
important because certain legislated labour rights
statutes and other governmental regulation ap-
are, insofar as ILO instruments are concerned, not
plying these principles to purely online work and
applicable to all workers. They continue to depend
so the legal position remains uncertain, even in
on an employment relationship. Certainly, the
the major jurisdictions. Cherry (2020) considers a
distinction may not always be clear-cut. Working
number of options that would fill the regulatory
time rules can both link to comprehensive stand-
gap, including regulation similar to the Maritime
ards (where they have an impact on health and
Labour Convention, 2006, and the GDPR, as well as
safety) and be specific to employment (as in the
corporate social responsibility initiatives.
case of paid annual leave).57 Yet, the latter cannot
always be absorbed into the former, and for such
56 This discussion has benefited considerably from Cherry’s analysis of the issues.
57 See the Holidays with Pay Convention (Revised), 1970 (No. 132); and ILO (2018c, Para. 280).
Appropriate classification
of workers is also important is commercial). Nor is inconsistency within a single

because certain legislated jurisdiction unknown; one judge may decide to
give greater weight to a contractual statement
labour rights are, insofar denying employment than another. What consti-
tutes, for one judge, control sufficient to establish
as ILO instruments are employment, may be insufficient for another (see,
concerned, not applicable for example, the review of EU cases in Kilhoffer
et al. 2020; see also, with respect to China, Ban
to all workers. 2020; Zou 2017b).
Furthermore, since there is a diverse
employment-specific rights it is essential to estab- range of platform-based labour
lish an employment relationship. forms (with dif ferent terms of
service), it is hard to determine to
All of this means that employee status remains
what extent a decision in respect
highly important. It is not surprising then that
of one form (say, drivers) will be
much of the litigation and many of the legislative
applicable to another form (say,
debates concerning work through digital labour
freelancers). For all that,
platforms turn on this issue. For platform workers,
it is possible to iden-
employment is a gateway to accessing a panoply
tify certain trends
of rights. For many platforms, it can be a major
that fall along a
threat to their business model.
spectrum between
Not only are the stakes high, but the context-spe- very broad and very
cific nature of the inquiry and the different narrow approaches
national approaches to answering it produce great to finding employment.
diversity and great uncertainty around the world.
The first approach is to em-
To be sure, many national systems apply tests that
phasize the practical control that the
in general terms correspond to those set out in
platform has over the transactions taking
the Employment Relationship Recommendation,
place through the platform. For example,
2006 (No. 198) (see section 5.2.3). As Prassl (2018,
in 2020, France’s highest court, the Cour de
100) comments: “Depending on each jurisdiction
Cassation, found that an Uber driver was an em-
and claim, the ordinary tests of employment
ployee of that platform. 58 The Court held that
and labour law apply to work in the on-demand
several key factors pointed to a labour contract
economy – and, given platforms’ tight control over
(contratdetravail). The driver was integrated into
many aspects of service delivery, they will often
a system that existed only because it was created
point towards employment status, leaving only
and maintained by the platform, a system in which
the genuine entrepreneurs outside the scope of
the platform controlled the rates and the terms
protective laws and regulations”.
upon which the transport was offered.59 The driver
But the unavoidably porous nature of the tests, in the litigation did not have his own clients. He
and their sensitivity to factual variations, mean could not choose his own itineraries. Uber was
that similar cases can yield one finding in one jur- able to temporarily deactivate the driver’s account
isdiction (the relationship is one of employment) after three refusals to drive and the driver could
and an opposite one in another (the relationship lose access to his account if his rate of cancellation,
58 Arrêt no 374 du 4 mars 2020 (19-13.316) – Cour de cassation – Chambre sociale. See also cases in Spain including Judgment STS
2924/2020 of 25 September 2020, Tribunal Supremo (sala de lo social), which found that there was an employment relationship
between a platform (Glovo) and its couriers.
59 Rogers (2018) notes that the degree of monitoring of platforms (including in the case of web-based platforms, recording key-
strokes and taking regular screenshots) could lead to a presumption of employment.
Box 5.4 The employment relationship: California Labor Code

X
Tests for misclassification vary around the world. An increasingly influential approach is the “ABC”
test which has been developed in particular in California: see DynamexOperationsWest,Inc.v.
SuperiorCourtofLosAngeles(2018)4Cal.5th903(Dynamex), a case involving delivery drivers for
a same-day delivery company (Dynamex). The Dynamex test now forms part of the California
Labor Code.
2750.3. California Labor Code:
(a) (1) For purposes of the provisions of this code and the Unemployment Insurance Code,
and for the wage orders of the Industrial Welfare Commission, a person providing labor
or services for remuneration shall be considered an employee rather than an independent
contractor unless the hiring entity demonstrates that all of the following conditions are
satisfied:
(A) The person is free from the control and direction of the hiring entity in connection with
the performance of the work, both under the contract for the performance of the work
and in fact.
(B) The person performs work that is outside the usual course of the hiring entity’s business.
(C) The person is customarily engaged in an independently established trade, occupation,
or business of the same nature as that involved in the work performed.
There are a number of exceptions, including the liberal professions.
In November 2020, a majority of Californian voters supported “Proposition 22”, formulated
by Uber, Lyft and other taxi platforms, which inserted Chapter 10.5 (App-based Drivers and
Services) into the Business and Professions Code. That chapter prevails over the California Labor
Code. It defines app-based drivers as independent contractors but specifies a number of benefits
pertaining to minimum earnings, health benefits, accident insurance, anti-discrimination, public
safety and rest periods. There is no specified right to organize or to bargain collectively.
or number of “problematic behaviour” reports, Recent judicial analyses in the State of California in
was excessive. The Court therefore determined the United States, now codified in state legislation,
that: “work was carried out under the authority reflect a similar take on classification although
of an employer which had the power to issue their application to digital labour platforms
orders and directives, to supervise their imple- has been altered by amendments to the State’s
mentation, and to sanction breaches. Therefore, Business and Professions Code effected through
the driver’s self-employed status was fictitious.”60 a successful 2020 ballot measure (see box 5.4).61
60 See https://www.courdecassation.fr/IMG/20200304_arret_uber_note_%20ENGLISH.pdf. This approach to control appears more

likely to find that a platform worker is an employee than that adopted by the European Court of Justice in, for example, BvYodel
DeliveryNetworkLtd (2020) C-692/19, a case involving a parcel courier decided the following month, although there were certain
factual and legal differences between the Uber drivers and Yodel couriers.
61 Note divergent approaches federally and in other US states: compare, for example, National Labor Relations Board, Office of
the General Counsel, Advice Memorandum (Cases 13-CA-163062, 14-CA- 158833 and 29-CA-177483), 16 April 2019 (finding that
Uber drivers were independent contractors) and LuisVegav.PostmatesInc,decision of New York State Court of Appeals (26 March
2020) (courier found to be an employee).
A second approach is apparent in countries but not in litigation concerning other labour
which have an intermediate category between standards (discrepancies between regulatory
employment and self-employment. An illustrative definitions permit this approach). This creates a de
case is theAslamlitigation, a decision of the United facto intermediate category. Some court decisions
Kingdom Court of Appeal,62 where the majority of in China and the Republic of Korea (such as the
the Court held that while the claimant drivers were one discussed in box 5.3) provide an example (see
not employees, they were “workers”, a category the discussion in Zou 2017a; see also the diverse
that entitled them to the minimum wage and paid views of Chinese scholars in Ban 2020; Xie 2018;
leave. This is because they were found “to do or Yan 2018; Peng and Cao 2016).
perform personally … work or services for another
A fourth approach, which is in some respects
party to the contract whose status [was] not by
the opposite of the first, is to emphasize the
virtue of the contract that of a client or customer
ways in which platforms do not exercise control.
of any profession or business undertaking carried
In contrast to the analysis of the French Cour de
on by the individual.”63
Cassation discussed above, courts and tribunals
The Court remarked that there was “a high in countries such as Australia have pointed to
degree of fiction in the wording … of the standard the fact that platform workers such as drivers
form agreement between [Uber] and each of have control over “whether, when, and for how
the drivers” (para. 90).64 It affirmed the original long they perform work”; they are “not subject to
judgment by the Employment Tribunal, agreeing any formal or operational obligation to perform
that: “it is not real to regard Uber as working “for” work”.65 The fact that a worker does not operate
the drivers and that the only sensible interpret- a business of their own and works solely for the
ation is that the relationship is the other way platform is not determinative.66 Similar reasoning
round. Uber runs a transportation business. The has been adopted by Brazilian courts in relation
drivers provide the skilled labour through which to Uber drivers.67
the organisation delivers its services and earns
To reiterate, whatever the approach adopted (see
its profits” (Para. 95). This case is at the time of
figure 5.3), there will be some workers who are
writing under appeal to the Supreme Court of the
“truly” self-employed, especially in the fourth cat-
United Kingdom, which should lead to a definitive
egory. The fundamental Conventions mentioned
outcome for that jurisdiction.
in table 5.1 require that they, too, enjoy those
A third approach is where courts give weight to the labour rights which apply irrespective of contrac-
purpose for which employment status is sought in tual status. Remedying misclassification, while
the particular litigation. This means that a platform extremely important, does not address the case
worker could, at least in principle, be found to be of these workers and further regulatory measures
an employee in accident compensation litigation may be needed (Xie 2018).
62 [2018] EWCA Civ 2748.

63 Employment Rights Act 1996 (United Kingdom) section 230(3).
64 [2018] EWCA Civ 2748. See also AutoclenzLtdvBelcher[2011] UKSC 41.
65 See https://www.fairwork.gov.au/about-us/news-and-media-releases/2019-media-releases/june-2019/20190607-uber-media-
release. See also KaserisvRaiserPacific[2017] FWC 6610; AmitaGuptavPortierPacificPtyLtd;UberAustraliaPtyLtdt/aUberEats
[2020] FWCFB 1698.
66 UberAustraliaPtyLtdt/aUberEats[2020] FWCFB 1698 at Paras 71–72; see also ACEInsuranceLimitedvTrifunovski [2013] FCAFC 3.
67 See Superior Tribunal de Justiça, 28 de agosto de 2019, Case No. 164.544 – MG (2019/0079952-0) and Tribunal Superior do
Trabalho, 5 de fevereiro de 2020, Processo No TST-ED-RR-1000123-89.2017.5.02.0038.
Figure 5.3 Different approaches to establishing an employment relationship between workers

and digital labour platforms
Tends towards Intermediate categories or

Tends towards
self-employment interpretation varies depending
employment
on entitlement at issue
 What does the platform not control?  What does the platform control?
 Emphasis on written documentation  Focus on practice more than

on contractual documentation
 Workers can be self-employed
even if they do not have  If workers do not have their own business,
a substantial business they are likely to be employees
5.3.11 Remuneration freelancers sometimes underbid and workers on

microtask platforms earn substantially less than
and working time their counterparts in the traditional labour market.
If this remuneration is the worker’s sole source of
Principles concerning remuneration and working
income, it may necessitate income supplemen-
time are closely tied to the employment re-
tation (including with respect to pensions and
lationship in the ILO Conventions, so robust
other forms of social insurance). This may have
classification principles are essential to ensure
an impact on the social security system.
that all employees benefit from them. However, it
does not follow that no provision at all should be Third, Chapters 2 and 4 have illustrated that plat-
made on these matters for self-employed workers. form workers are vulnerable to predatory pricing
tactics, and pay commission and transaction fees,
Some standards on remuneration and working
which erode their payments.
time are necessary to give effect to universal
rights and principles, such as occupational safety What are the possibilities for developing norms
and health, social security and fair payment sys- about remuneration and working time that ad-
tems. First, if a self-employed platform worker dress these matters and are appropriate for the
works long hours on a regular basis to earn their self-employed? From the perspective of social
income, this might compromise their health and dialogue and collaborative regulation, it would
safety. As workers find it difficult to access suffi- be ideal if they could emerge from collective
cient work, they have a powerful incentive to work bargaining between the parties themselves.
excessive hours (paid and unpaid), which can However, we have seen that antitrust law in many
lead to poor mental and physical health, among countries frustrates this approach; for example,
other conditions. an hourly rate may be vulnerable to challenge on
the grounds of price-fixing (Kilhoffer et al. 2020).
Second, for many workers who are dependent
on platform work, such as freelance work and Another response to this conundrum is to extend
microtasks, as their main source of income, the existing labour standards to “non-employee”
income may not be sufficient to meet their basic platform workers. A recent bill presented by
living expenses. As discussed in section 4.2.2, Chilean Senators in 2020 attempts to provide basic
Standards on
remuneration and working any direct retaliation for disconnecting, workers
time are necessary to who do so may place themselves at a significant
disadvantage. Several jurisdictions are moving
give effect to universal to specify minimum pay rates for self-employed
location-based platform workers. French law
rights and principles. (Codedutravail Art. L7342-9(2)) provides that a
platform’s social charter should include methods
of enabling a self-employed worker to obtain a
“decent price” (prixdécent).
guarantees to self-employed platform workers, in-
The Indian Motor Vehicle Aggregators Guidelines
cluding the right to receive an hourly pay rate that
of November 2020 specify that aggregators –
complies with the domestic legal framework on
digital intermediaries or marketplaces where
minimum wage.68 This approach can be also seen
passengers connect with drivers for the purpose
in the Aslam litigation in the United Kingdom (see
of transportation – must comply with certain
section 5.3.10), which, pending the final appeal
criteria in order to obtain a licence, which include
outcome, has enabled Uber drivers to be covered
obligations pertaining to working time and remu-
by minimum pay and working time regulations
neration. For instance, aggregators must ensure
(because they are “workers” for the purposes
that drivers are not logged in for more than
of those regulations) even though they are not
12 hours a day, even when drivers are engaged
employees. However, regulatory initiatives which
with multiple aggregators; once the connection
have created an intermediate category between
limit is reached, a mandatory break of 10 hours
employment and self-employment have attracted
will apply (Art. 7(2)(d)). Furthermore, the guidelines
criticism; some argue that they undermine the
provide that the relevant city taxi fares apply and
employment category (Stewart and McCrystal
that the base fares must correspond to a minimum
2019; Cherry and Aloisi 2017; De Stefano 2016; for
of 3 kilometres to account for dead mileage and
a defence, see Davidov 2014).
the distance and fuel needed
A further approach is to conceive of new labour to reach customers
standards that are specifically adapted to digitally (Art. 13(1) and (2)). It
based work. One such standard is the “right to is also specified that a
disconnect” (droitàladéconnexion), which was in- maximum surge pricing
troduced in 2017 for salaried employees in France of 1.5 times the base fare
(Codedutravail Art. L7342-9(1)). This standard was applies and that drivers must
extended to platform workers in the transporta- receive at least 80 per cent of
tion industry in 2019, which enabled self-employed the fare (Art. 13(3) and (4)).
platform workers in the taxi sector to “switch off”
In the city of Seattle in the
from the platforms without retaliation provided
United States, legal challenges
this standard constitutes part of the platform’s
to the validity of the
voluntary social charter.
cit y ’s ordinance
However, of itself, this measure could be permit ting Uber
ineffective in many situations. As discussed in and Lyft drivers to
section 4.2.2, many platforms use algorithms that bargain collec tively
reward longer connection periods or otherwise have led the city to set up
create incentives to work long hours in order for a minimum compensation
workers to receive bonuses (particularly loca- system for those drivers instead
tion-based platforms). While there may not be (Seattle Municipal Code, Ch. 14.31).
68 Senado de Chile, Proyecto de Ley que establece garantías básicas a las personas que prestan servicios a través de plataformas
digitales, BoletínNº 13.496-13, article 3.
A similar approach appears to have been adopted 5.3.12 Platform work

in the State of California following the successful
“Proposition 22” ballot measure (see box 5.4). and labour clauses
The relationship between minimum remuneration in trade agreements
standards for the self-employed and the wages
The discussion so far has focused on domestic and
system for employees is complex. In countries
regional legal systems. Another important form of
with a universal minimum wage, that will be a
regulation is transnational. The most prominent
relevant reference point for negotiations among
example of transnational regulations, in terms of
the relevant stakeholders. On the other hand, in
legal impact, are free trade agreements (FTAs).
countries with industry-level minimum pay clas-
Considerations of space preclude an extensive
sifications (set, for example, through collective
discussion of the relevance of FTAs to platform
agreements or awards) levels set for workers with
work, but some general comments can be made.
a comparable degree of skill may be pertinent.
A steadily increasing number of FTAs contain labour
One important objection to the development of
clauses that reference the 1998 ILO Declaration on
standards pertaining to self-employed platform
Fundamental Principles and Rights at Work (see
workers is that in some contexts there may be
ILO 2019c; ILO 2017c; Agustí-Panareda, Ebert and
multiple parties involved in determining rates of
LeClercq 2014) as well as other matters, such as
pay for platform workers and this may render the
health and safety at work. More recent agree-
application of remuneration regulations more
ments also refer to the ILO Decent Work Agenda
difficult, such as where a customer determines
and the 2008 Declaration on Social Justice for a
the pay rates. However, as Prassl (2018, 104) has
Fair Globalization (for example, FTAs negotiated
suggested, it is possible to formulate a rule to deal
with the EU) as well as an expanded list of pro-
with this: the platform could be responsible for
tected rights (for example, the Comprehensive
setting up its systems such that “for any given
and Progressive Agreement for Trans-Pacific
amount of working time, no value below the
Partnership and the United States–Mexico–Canada
relevant proportion of an hourly minimum wage
Agreement) (Compa 2019; Namgoong 2019).
can be entered by the customer”.
Some of these clauses are problematic because
A further point can be made about enforcement.
they may not correctly reference the fundamental
As discussed in sections 2.4 and 4.3.1, many plat-
Conventions (Agustí-Panareda, Ebert and LeClercq
forms provide “coordination tools” to workers for
2014) and because they may be so loosely worded
managing workflows, including tracking routes
that it is difficult to put them into effect (Santos
(in the case of location-based platforms) and
2018; Tham and Ewing 2016). Still, as FTAs com-
recording keystroke activity and taking screen-
monly impose binding legal obligations on their
shots (in the case of web-based platforms) (see
signatories, they are an important site of regu-
also Appendix 2B). This means that platforms
latory intervention pertaining to the transnational
are in a position to determine working hours
activities of digital labour platforms. The impact
quite accurately, so they should be able to avoid
of trade agreements, and especially their labour
underpayments. Furthermore, as Rogers (2018)
clauses, on platform workers is a matter that could
argues, the data generated in this way could
receive more attention in trade negotiations than
be made available to government authorities,
has hitherto been the case.
and possibly worker organizations, to check
for non-compliance. On the other hand, some FTAs may contain
provisions that curtail the ability of States to
regulate the conditions of platform workers;
such provisions may be found in chapters dealing
with investment, e-commerce and cross-border
FTAs are an important

trade in services. For example, certain FTAs
site of regulatory
protect digital platform firms’ cross-border data intervention pertaining to
flows and prevent governments from localizing
their presence or requiring them to transfer or the transnational activities
disclose their source code and algorithms. In this
context, a crucial question is whether platform
of digital labour platforms.
companies merely provide technological services.
If so, they may be permitted to provide services
from abroad, without the need to engage with
domestic sectorial regulations such as those cov- arbitration panels but also because the mere
ering transport, data protection, tax and labour threat of invoking them may have a chilling effect
law. Such stipulations have an impact not only on initiatives to enhance the conditions of plat-
because of actual litigation before international form workers.
Conclusion
This chapter has illustrated that jurisdictions In many cases, these obligations are arguably
regulating platform work occupy different pos- inferior to those of an employer to an employee,
itions in the space between two poles. At one leading to a number of legislatures and courts
pole, the platform worker is conceived of as an to construe the platform/worker relationship
independent agent for whom the platform bears as one of employment. However, this trend to-
no responsibility. At the other pole, the platform wards considering the relationship to be one of
worker is considered to be an employee of the employment is not, as matters presently stand,
platform, which becomes responsible for com- universal or inexorable. The de facto creation of
plying with all obligations under labour and social an “intermediate” category of work relationship
protection legislation. seems just as likely to be entrenched in many
other jurisdictions.
From the perspective of decent work, particu-
larly as articulated through ILO Conventions and However contractual arrangements between plat-
Recommendations, platform workers should form and worker are construed, ensuring decent
benefit from many labour and social protection work for platform workers requires social dialogue;
rights, irrespective of their contractual status. platforms, workers, their representatives and gov-
In many contexts, this means that the platforms ernments should all be involved in creating better
need to take some responsibility for ensuring working conditions. In particular, platform workers
decent work for platform workers because they should be able to associate so as to negotiate with
can materially influence their working conditions. the platforms. Collective bargaining should also be
They cannot remain at the “no responsibility” pole. available, whether in its traditional forms centred
on the employment relationship, or in new forms
And indeed, while arguments as to the appropriate
emerging in many countries which are suited to
characterization of platform workers continue to
the self-employed. It is through social dialogue
play out around the world, it is possible to detect
that innovative and consensual approaches to
considerable movement away from the “no re-
regulating platform work for the benefit of all
sponsibility” pole in many countries. Increasingly,
stakeholders are likely to emerge.
platforms – whether of their own volition, through
social dialogue or as a result of regulatory action
by governments – are undertaking obligations
relating to social security, health and safety, data
protection, minimum earnings, the prevention of
Platform workers
discrimination and so forth. should benefit from many
labour and social protection
rights, irrespective of their
contractual status.
6
Seizing the
opportunity
A way forward
Need for international
policy dialogue and
coordination
With growing regulatory

concerns there have been...
• Initiatives from
national jurisdictions
• Initiatives by social partners
• Initiatives by other non-state actors
Policy measures to ensure

protection for workers
Freedom
of association
and collective
bargaining
Data
Other fields of law and policy protection Employment
relationship
relevant for platform workers
Dispute
resolution Occupational
mechanism safety and
health
Non-
Artificial discrimination
Competition Taxation
intelligence Social
Fair security
remuneration
and working
time
6. Seizing the opportunity: A way forward 241
Introduction
Digital platforms have grown exponentially over Although still in their early stage, digital labour
the past decade, facilitated by the availability of platforms are increasingly shaping the future
cloud infrastructure, cloud computing services of work. They have created opportunities for
and venture capital financing. The information workers and businesses alike, but their rise has
and communications technology revolution in also disrupted some sectors of the economy (such
particular has resulted in widespread adoption as the taxi sector) and has created challenges
by businesses and individuals of technological for the future of work, which are summarized
solutions and devices, spurring demand for both in section 6.1.
digital products and services, as well as creating
Section 6.2 discusses some of the emerging
an environment for platforms to grow rapidly.
hard- and soft-law responses or initiatives taken
The role of the digital economy has been further
by governments, social partners, platforms and
reinforced by the consequences of the COVID-19
other stakeholders (such as cooperatives) to ad-
pandemic, as it has enabled the continuity of busi-
dress the challenges faced by workers on digital
ness and remote working.
labour platforms, in particular by drawing on
Among digital platforms, digital labour platforms national practices.
have distinctive features that, coupled with their
Section 6.3 explores key pathways for leveraging
rapid pace of growth, are transforming the world
the opportunities and overcoming the challenges
of work. These platforms can be categorized into
by drawing on guidance from international labour
online web-based and location-based platforms.
standards, as well as national law and practice,
They cut across multiple economic sectors that
which are relevant for ensuring decent work for
require the application of diverse skills, from
workers on digital labour platforms.
undertaking deliveries to data analytics.
Finally, the chapter describes how digital labour
This report has provided a pioneering and com-
platforms can realize their potential for generating
prehensive international overview of the business
decent work opportunities for workers and sup-
strategies of digital labour platforms, as well as
porting efforts by businesses to thrive, thereby
insights into the experience of workers and busi-
advancing progress towards the Sustainable
nesses on online web-based and location-based
Development Goals.
platforms, drawing on surveys and interviews
conducted with some 12,000 workers and repre-
sentatives of 85 businesses around the world in
multiple sectors.
6.1 Opportunities and challenges

on digital labour platforms
The pervasive development of the digital economy start-ups and traditional companies in such coun-
and the growth of digital labour platforms have tries have struggled to keep pace and compete on
been accompanied by a digital divide across a level playing field. Furthermore, the available
and within countries. There are key digital infra- data on online web-based platforms shows that
structure gaps and many developing countries in the majority of workers performing tasks on these
particular face challenges with regard to ensuring platforms reside in developing countries, while the
adequate access to reliable digital infrastructure demand is being generated primarily in developed
for both their populations and businesses. This countries. This trend has been magnified since the
results in constrained benefits from the digital outbreak of the COVID-19 pandemic in March 2020.
economy and limited growth of the employment
and entrepreneurial opportunities that it can po-
tentially offer. 6.1.1 Opportunities and
Weak digital infrastructure poses a major chal- challenges for businesses
lenge for businesses that rely on platforms, as
The reliance of many businesses, particularly
poor connectivity can have implications for their
SMEs, on platforms has been growing, especially
smooth running. It could also limit workers on
with increasing consumer preference for the use
online web-based platforms from performing
of app-based platforms. Depending on the type
tasks in an efficient manner, thereby making it
of business concerned, enterprises are looking to
more difficult to compete with workers in coun-
digital labour platforms for a range of services,
tries with better infrastructure. Such challenges
from accessing a global pool of talent through
create a serious threat in developing countries
online web-based platforms to expanding their
to the sustainable creation and growth of en-
customer base through delivery platforms. This
terprises, particularly SMEs, which are critical
situation in turn is also creating additional demand
for employment generation and achieving the
for platform work in activities ranging from infor-
Sustainable Development Goals.
mation technology (IT), operations, marketing,
In addition, the report has shown that the growth and research and development, to delivering
of digital labour platforms is geographically prepared food and groceries. Such a rise in
uneven, and concentrated in North America, demand is playing an important role in influencing
Europe and Asia. The global distribution of in- employment opportunities in both the offline and
vestment in digital labour platforms is similarly online labour markets.
skewed, as 96 per cent of investment is concen-
Through online web-based platforms, businesses
trated in those regions, compared to 4 per cent
are able to access talent globally, which is enabling
going to Latin America, Africa and the Arab States.
them not only to innovate but also to use these
About 70 per cent of the global revenues gener-
platforms for their recruitment processes, reduce
ated through these platforms are concentrated in
costs and improve efficiency (see section 3.1).
just two countries: the United States and China.
Platforms facilitate the use by businesses of a
There is further concentration of market power
diverse workforce ecosystem that consists of
among a few location-based platforms, facilitated
workers with a multitude of contractual arrange-
through large-scale venture capital funding that
ments. Many businesses, ranging from SMEs and
has enabled them to diversify and rapidly expand
start-up companies to Fortune 500 companies,
in new markets despite often being unprofitable.
have hence come to rely on such platforms not
Many of these platforms have entered markets only to tap the best talent but also to improve
in developing countries, while both domestic organizational performance (see section 3.1).
Through online
web-based platforms, 6.1.2 Opportunities
businesses are able to recruit and challenges for workers
workers, innovate, reduce The rise of digital labour platforms is creating new
income-generating opportunities for workers.
costs and improve efficiency. Online web-based platforms provide opportun-
ities to workers, including those with disabilities,
to perform various types of tasks, and facilitate
exchanges between clients (businesses) and
workers. Location-based platforms, such as
With regard to location-based platforms, delivery those for taxi and delivery services, provide op-
platforms in particular are enabling businesses portunities, including for migrants, the low-skilled,
such as restaurants and retail stores to expand underemployed or unemployed, although capital
their customer base and markets by adapting assets are required to access these jobs. Workers
more quickly to changing customer preferences, on online web-based platforms have the flexibility
and to enhance productivity and profitability. to perform the tasks from any location, allowing
Delivery platforms are also playing a key role some women and men to combine work with child-
during the COVID-19 crisis in ensuring business care and elder care responsibilities. Competitive
continuity. Similarly, taxi platforms are allowing programming platforms are enabling workers
businesses and consumers to benefit from the to develop and hone their skills in coding, data
greater convenience and accessibility they provide analytics and programming, among others, and
at lower cost. to develop a community of peers in programming
and coding.
The benefits afforded by digital labour platforms
at the same time present several challenges for
businesses. The reliance of digital labour plat-
forms on large-scale venture capital investment
can create an uneven playing field that can result
The rise of digital
in competition issues for traditional businesses. labour platforms is creating
Such businesses further lack access to adequate
finance to upgrade and to respond more rapidly to new income-generating
changing market dynamics. Large platforms have
also become dominant in many sectors and are opportunities for workers.
putting at risk the sustainability of both traditional
businesses and platform start-ups.
Survey findings show that location-based and
Platforms unilaterally determine the terms of ser- online web-based platforms have become a
vice agreements, which can also have implications valuable source of work and income for many
for businesses. For instance, in the delivery sector workers, particularly in developing countries. They
platforms may charge high commission fees which also show that workers are motivated to perform
affect the profit margins of restaurants. In the tasks on online web-based platforms in order to
retail sector, while traditional businesses have complement income, or because of the job flexi-
expanded their customer base by using e-com- bility offered or a preference to work from home.
merce platforms, they often face unfavourable Lack of alternative employment opportunities,
contractual terms, non-transparency with regard job flexibility and better pay compared to other
to data and pricing, and weak dispute resolution available jobs are the main motivating factors on
mechanisms, among other issues. location-based platforms (see section 4.1.7).
There is an urgent need

Work on digital labour platforms is also accom- for governments to address
panied by numerous challenges which relate to
regularity of work and income, decent working the challenges with regard
conditions, social protection, skills utilization
and the freedom of association and right to to the working conditions
collective bargaining. Many of these challenges
are also prevalent for workers in informal and non-
of platform workers.
standard work arrangements and are increasingly
affecting those engaged on digital labour plat-
forms. The working conditions on digital labour substantial amount of time doing unpaid work,
platforms are largely regulated by terms of service and workers on location-based platforms often
agreements, which are unilaterally determined by spend a lot of time waiting for work – and this time
the platforms and constitute contracts of adhesion is not compensated.
that define various aspects, including remuner- There are also differences on online web-based
ation and working time. These agreements tend platforms in earnings between platform workers
to categorize workers as independent contractors, from developed countries and those from de-
which makes it either impossible or expensive for veloping countries, with the latter earning less
them to access many of the workplace protections as they are often excluded from accessing high-
and entitlements that apply to employees recog- er-paid tasks. Furthermore, workers on online
nized as such. In many countries self-employed web-based platforms often face unpredictable
workers are either not covered by social security work schedules and unsocial hours, particularly
systems or, if covered (voluntary or mandatory in developing countries due to the temporal dis-
affiliation), bear the entire cost of their protection tribution of tasks, which are often posted during
without any cost-sharing by platforms. US business hours. Workers on location-based
The findings from this report also show that platforms work long hours to meet their targets
while online web-based platforms offer new op- so that they can obtain their bonuses and maintain
portunities to workers to perform tasks, workers access to work (see section 4.2.3).
often struggle to find a sufficient amount of work The lack of social security coverage is a major
due to the unavailability of enough well-paid concern for workers on digital labour platforms,
work. The reasons for this include excess labour wherever they are located. The conditions created
supply, which generates increased competition by the COVID-19 pandemic are exacerbating their
among workers, and a platform design that may vulnerability and creating additional risks for
discriminate against certain workers and that workers who interact with the public as a regular
often charges them various fees to access work. part of their work. Occupational safety and health
The consequences of the COVID-19 pandemic are risks are especially significant on taxi and delivery
exacerbating the lack of availability of work and platforms, and the lack of sickness benefits and
resulting in a decline in workers’ incomes, while paid sick leave is compelling workers to continue
exposing them to greater risk because of the lack working even when infected, thus putting at risk
of social protection. their own health as well as that of their clients and
On location-based platforms, apart from the lack the wider public (ILO 2020b). Notably, the lack of
of a sufficient amount of work, declining remu- unemployment protection and other income sup-
neration rates and high commission charges also port measures leaves them in a highly vulnerable
affect incomes. On online web-based platforms, position (ILO 2020a).
workers’ incomes are affected by high levels of Algorithmic management of workers is ubiquitous
competition and commission fees, as well as on digital labour platforms. Algorithms determine
sometimes through unjustified rejection of, or the allocation of tasks, performance evaluation,
non-payment for, completed tasks. Workers ratings and acceptance or rejection of work. They
on online web-based platforms often spend a also determine schedules and working hours
as well as access to future work opportunities, Furthermore, the algorithms used by both online
on both online web-based and location-based web-based and location-based platforms are
platforms. Such issues have serious implications trained using data that often carries existing
for the notion of flexibility, as well as autonomy biases and thus may lead to human discrimination
and control over work on digital labour platforms being embedded in the very architecture of the
(see section 4.3). algorithms. Platform workers, especially in the taxi
and delivery sectors, also reported experiencing
Platform design can also play a role in exacer-
or witnessing discrimination or harassment,
bating discrimination, especially on online
mainly on the part of clients and customers, but
web-based platforms, and there is evidence that a
also by the police in certain instances, on the basis
considerable number of workers have experienced
of the work they perform (see section 4.5).
discrimination in accessing work or high-paying
tasks, particularly women and workers in de- There is an urgent need for governments to ad-
veloping countries. On location-based platforms, dress the challenges with regard to the working
the apps are sometimes designed in such a way conditions of platform workers, including access
that they allow for human biases in the code of to social security, so that the income and work op-
the algorithms, which can then lead to inadvertent portunities generated by these platforms can be
discrimination against some workers. leveraged to promote decent work.
6.2 Emerging regulatory responses
A growing number of countries have started to ad- specific to platform-based work, and classi-
dress the challenges related to working conditions fying these workers as employees, to prevent
on digital labour platforms. Chapter 5 identifies their misclassification.
various regulatory developments, both hard law
Several developed and developing countries have
and soft law, that could serve as potential stimuli
extended or adapted existing laws to platform
for further action. Such developments include
workers, especially in the areas of occupational
initiatives by national jurisdictions as well as by
safety and health, and social security. For instance,
social partners and other non-state actors. This
occupational safety and health standards have
section summarizes both the hard- and soft-law
been extended to platform workers through a
initiatives in order to emphasize their importance,
judicial decision in Brazil, while in India a new code
and at the same time highlights the prevalence of
on social security has extended social security to
regulatory uncertainty and the need for coherent
all workers irrespective of their employment rela-
regulatory frameworks and public policies at the
tionship, including platform workers. Similarly, a
national and international levels.
number of Latin American and Asian countries have
been leveraging technology and the IT infrastruc-
6.2.1 National jurisdictions ture to provide social security to platform workers.
As described in Chapter 5, a number of countries

have adopted various regulatory approaches
to apply existing labour protection and social Several developed and
security legal frameworks to platform workers,
especially focusing on location-based plat-
developing countries have
forms such as taxi and delivery services. These extended or adapted existing
include, adapting existing legislation to plat-
form workers where needed, developing rules laws to platform workers.
Some countries have also developed new ap- 6.2.2 Initiatives

proaches or rules specific to platform-based work
in the areas of working time and remuneration, as by social partners
well as access to data and privacy. The labour code
In addition to the measures referred to in
in France, which was amended in 2019 to extend
Chapter 5, a number of soft-law initiatives have
certain working time provisions to self-employed
been undertaken in national jurisdictions by
platform workers in the transportation industry,1
governments and social partners. Some of these,
is one case in point. The law provides that a plat-
such as codes of conduct, have been developed
form’s voluntary social charter should include
by public bodies in collaboration with unions and
the “right to disconnect” and enables a method
platform companies. For instance, the munici-
to obtain a “decent price” for self-employed plat-
pality of Bologna, in Italy, adopted a Charter of
form workers. Similarly, in Brazil, India, Nigeria,
Fundamental Rights of Digital Labour in the urban
the European Union and the State of California
context in 2018. The charter provides guidance
in the United States, new legislative rules and
on fair wages, health and safety, protection of
measures regarding data protection and privacy
personal data and the right to disconnect; the
are being established which are also relevant for
platforms that sign the charter are encouraged
digital labour platforms and workers irrespective
to observe it.2
of their employment status.
Similarly, in the Republic of Korea, the Economic,
Finally, countries have also adopted various
Social and Labour Council, in cooperation with
approaches to the classification of platform
representatives of workers, platform companies
workers in the case of location-based platforms,
and the Government, has adopted a code of
often arising from litigation. There are four dis-
conduct that provides guidelines for fair contract
tinct approaches. The first is to classify platform
terms between workers and platform companies
workers as employees based on the degree of
on matters such as payment methods, fees, tax,
control exercised over them by the platform, as
non-discrimination, performance assessment
was observed in the case of Uber taxi drivers in
programmes and dispute settlement.
France and Glovo delivery workers in Spain. The
second approach is to classify platform workers
as an intermediate category partially covered by
labour protection and social security, as some Soft-law initiatives have
courts have done in the United Kingdom. The third
approach is to adopt a de facto intermediate cate- been undertaken in national
gory, wherein certain benefits such as workplace
injury compensation are provided to workers, as jurisdictions by governments
was observed in the case of China. The fourth
approach is to consider platform workers as
and social partners.
self-employed, because of the control they have
over setting their own schedules, as in the case With growing regulatory concerns and innovations
of Australia and Brazil. The employment status of by governments, platform companies themselves
platform workers has been and continues to be a have also been addressing the challenges faced
controversial issue, with national courts adopting by workers. For instance, in Denmark a collective
different approaches to recognizing these workers bargaining agreement between a trade union
as employees.
1 The transportation industry includes transportation of passengers by car and delivery of goods using two- or three-wheeled vehicles.
2 For more details, see: https://digitalplatformobservatory.org/initiative/charter-of-fundamental-rights-of-digital-labour-in-the-
urban-context/.
and a cleaning platform (Hilfr) 3 was reached Trade unions have also been helping associations
in 2018 ( Jesnes and Oppegaard 2020). This of platform workers in the taxi and delivery sec-
enabled Superhilfr workers on the Hilfr platform tors with legal challenges. For instance, an unfair
to transition to employee status and thereby dismissal case against Uber in South Africa was
be covered by a collective agreement with the filed by the National Union of Public Service and
labour union 3F. However, the Danish Competition Allied Workers. Similarly, in Canada, the case
Council noted in its assessment in August 2020 that recognized the right of Foodora workers to
that “Freelancehilfrs/Superhilfrs, most likely, are unionize and bargain collectively was brought
not employees of Hilfr from a competition law to the Ontario Labour Relations Board by the
point of view”, and “that the minimum hourly fee Canadian Union of Postal Workers.
may create a ‘price floor’, which may limit the com-
petition between the Freelancehilfrs”. Hilfr has in
its response to the assessment committed to en- 6.2.3 Initiatives by other
suring that Superhilfrs are employees and that the non-state actors
company bears the financial risk for their cleaning
work, which is in accordance with their intention The importance of addressing the challenges that
when they entered the collective agreement with confront platform workers is also being increas-
the union 3F. In addition, it has offered to remove ingly recognized by other non-state actors. These
“the minimum hourly fee for Freelancerhilfrs from have developed soft-law instruments, such as
the platform”.4 codes of conduct, principles of good platform work
and platform certifications. A prominent example
Other codes of conduct have been adopted and
of such an instrument is the World Economic
initiatives taken by trade unions, such as the
Forum Charter of Principles for Good Platform
“FairCrowdWork” initiative and the “Ombuds
Work (2020). The Charter covers issues such as
office of the Crowdsourcing Code of Conduct”
safety and well-being, flexibility, fair working con-
in Germany. The latter sets out a basic set of
ditions, social protection, voice and participation,
guidelines with a view to promoting trust and
and data management. Six major digital labour
cooperation among platforms, clients and crowd-
platforms (Cabify, Deliveroo, Grab, MBO Partners,
workers. The mandate of the Ombuds office is to
Postmates and Uber Technologies) signed and
seek compliance with the code of conduct and
committed to adhering to the Principles at the
resolve disputes between workers and signatory
2020 World Economic Forum Annual Meeting held
platforms, regardless of the location of the worker.
in Davos, Switzerland.
The Ombuds office is composed of a board of
five – one worker, one trade union representative, Similarly, the Fairwork Foundation,6 a consortium
one platform representative, one Crowdsourcing of university researchers, provides a code of
Association representative, and a neutral good practices and principles for the regulation
chair – and resolves disputes by consensus, with of platform work, to ensure decent work stand-
IG Metall, a German trade union, handling the ards on digital labour platforms. The researchers
administration. As of December 2019, a total of 44 at the foundation have translated the principles
cases had been submitted to the Ombuds office into measurable thresholds, and they evaluate
by workers via its online form.5 platforms against those thresholds by providing
ratings and certifications to platforms.
3 There are two types of service providers on Hilfr’s platform: “Freelancehilfrs” and “Superhilfrs”. The latter is covered by a collective
agreement with the labour union “3F”. A “Freelancehilfr” can become a “Superhilfr” automatically after working for 100 hours,
however a worker can choose to remain a “Freelancehilfr”.
4 For more details, see: https://www.en.kfst.dk/nyheder/kfst/english/decisions/20200826-commitment-decision-on-the-use-of-
a-minimum-hourly-fee-hilfr/.
5 For more details, see: https://ombudsstelle.crowdwork-igmetall.de/en.html.
6 For more details, see: https://fair.work/en/fw/homepage/.
The importance of
Moreover, in recent years a number of platform
cooperatives have been established with the
addressing the challenges
support of unions across a range of sectors, that confront platform
from taxi (such as Green Taxi Cooperative, Eva)
and delivery (such as Coopcycle), to healthcare workers is also being
(such as NursesCan) and e-commerce (such as
Fairmondo) to ensure fair working conditions
increasingly recognized
for platform workers (see box 2.3). SMart is one
such cooperative of autonomous workers, oper-
by other non-state actors.
ating in nine countries (Austria, Belgium, France,
Germany, Hungary, Italy, the Netherlands, Spain
and Sweden). It assumes the role of employer also provide medical and accident insurance
vis-à-vis its members to help them access social coverage to workers and their family members
security benefits. Members also benefit from legal (see section 2.3.1).
aid, safety training and insurance coverage, and as
Regulatory developments, albeit disparate,
employees they also have access to certain legal
provide a significant point of departure for recog-
entitlements under labour and social security laws.
nizing the magnitude of the challenges emerging
With growing regulator y concerns, certain from digital labour platforms for workers. They
platform companies have started to engage in also provide the preliminary building blocks for
addressing some of the issues related to working constructing the way forward. Moving ahead, it
conditions. For instance, some location-based will be essential that regulatory and public policy
platforms offer insurance coverage to workers frameworks become more cohesive and better
or paid sick leave (such as Deliveroo) or in-ride coordinated, including at the international level,
insurance and social protection benefits (such as and that they introduce regulatory certainty and
Uber). Some of the delivery platforms (e.g. Swiggy) are grounded in international labour standards.
6.3 Overcoming the challenges to seize the benefits

In order for the potential benefits of digital labour
platforms to be fully realized, it will be necessary
6.3.1 Addressing
for them to provide decent work, thereby con- the regulatory gaps
tributing to the achievement of the Sustainable
Development Goals. The meaning of “decent As described in Chapter 5, the ILO’s fundamental
work” in the context of platform-based labour principles and rights at work, and some of its
through the lens of international labour standards key Conventions and Recommendations, are
was discussed in Chapter 5. Ensuring the applica- applicable to all workers, irrespective of their
tion of some of the key labour standards to all contractual status. Thus, irrespective of whether
workers, irrespective of their contractual status, platform workers are classified as employees or
would be a pronounced step forward. The recom- as self-employed, they should enjoy the right to
mendations for policy action in this section consist associate, to bargain collectively, and to be pro-
of addressing the regulatory gaps by means of tected against discriminatory conduct and unsafe
legislative responses, and exploring other fields of workplaces. They should be provided with health
law relevant to ensuring decent work, particularly and safety protection and social security and be
on digital labour platforms. guaranteed a range of other key rights at work (see
The ILO’s fundamental

also OECD 2020a). The principles and rights artic-
ulated in international labour standards remain
principles and rights at
fully relevant to the operations of digital labour work, and some of its
platforms, though questions may arise about how
to make them operational in a particular context. key Conventions and
The ILO Centenary Declaration for the Future of Recommendations, are
Work calls for “[s]trengthening the institutions of
work to ensure adequate protection of all workers, applicable to all workers.
and reaffirming the continued relevance of the
employment relationship as a means of pro-
viding certainty and legal protection to workers,
while recognizing the extent of informality and The first recommendation for policy action re-
the need to ensure effective action to achieve lates to freedom of association and collective
transition to formality” (ILO 2019a, 5). Since bargaining. As discussed in Chapter 5, levels of
States are responsible for the implementation of unionization among workers on digital labour
ratified international labour standards, they can platforms are low. However, momentum has
through their national legislation and enforcement built in recent years concerning the organ-
mechanisms ensure that digital labour platforms ization of workers, especially on location-based
comply with laws that are in line with international platforms, spawning a number of platform
labour standards. For this reason, the national worker associations. One important challenge
regulatory framework is crucial, as it has a bearing is to devise collective bargaining structures for
on the practices of companies. In the absence of self-employed workers, where platform workers
ratification, international labour standards repre- are truly self-employed. A first policy recommen-
sent the most useful reference for national policy dation would thus be to ensure that legislative
and legislative design. frameworks guarantee that all workers have the
right to organize and bargain collectively. These
In recent years, there have been a number of
need not replicate the traditional systems con-
initiatives in countries and regions that aim to
structed with employees in mind. Social dialogue
implement these standards for platform workers
would be fundamental and could serve to address,
(see Chapter 5). It is clear that while there has
through negotiation, many of the issues identified
been considerable progress, much remains in
in this report, such as terms of engagement on
flux. Moreover, there are fundamental differ-
platforms, rules about commission fees, ratings
ences between countries. These are not simply
and deactivation, pricing, use of data, and evalu-
reflections of national differences – they call into
ation systems.
question whether standards should apply to
all or only to certain categories of workers. For The second recommendation for policy action
example, collective bargaining for self-employed relates to addressing the issue of the employment
platform workers is possible in some parts of relationship. As discussed in Chapter 5, coun-
the world (such as Australia, Canada and Japan) tries have taken dif ferent and sometimes
while in others there are significant obstacles even opposing approaches to determining the
to providing collective bargaining rights to such employment status of platform workers and to
workers (such as the European Union). If all the question of their classification as employees
workers whose work is mediated by platforms are or independent contractors. This means that a
to enjoy decent working conditions, the current taxi driver working for the same platform may be
patchwork of innovations and initiatives needs to classified as an employee in one country, as an in-
be consolidated and extended, while respecting dependent contractor in another, and as a member
the distinctive regulatory approaches of different of an intermediate category in a third. The ILO
jurisdictions. Ensuring decent work for platform Employment Relationship Recommendation, 2006
workers would require addressing regulatory gaps (No. 198), is an important reference as it can pro-
in eight crucial areas, as shown in figure 6.1. vide guidance “to guarantee effective protection
Figure 6.1 Policy fields to be addressed to ensure protection

for platform workers
Freedom
of association
and collective
bargaining
Data
protection Employment
relationship
Dispute
resolution Occupational
mechanism safety and
health
Non-
discrimination
Social
Fair security
remuneration
and working
time
for workers who perform work in the context of safety and health, including the risk of accident or
an employment relationship” (Para. 1),7 which can contagion. In this respect, regulatory practices such
enable legislatures and courts to achieve greater as those in Australia or in Brazil provide examples,
consistency in this regard. where through legislation and judicial decisions,
workplace safety and health have been extended
Two further and interrelated recommendations
to all workers, including platform workers.
for policy action relate to occupational safety and
health and social security. Addressing these mat- Likewise, access to social security coverage is
ters in the context of all forms of platform work has also key to ensuring access to medical care and
become especially pressing in light of the COVID-19 income security for platform workers. Access to
pandemic, the consequences of which are exacer- healthcare and sickness benefits during periods
bating the devastating effects of the lack of social of ill health, and income support in the event of
security coverage of platform workers. Work unemployment and loss of earnings are essential
processes need to be redesigned so that workers responses in the context of the pandemic and
and other persons using platforms are, so far as beyond. While some progress is being made,
is reasonably practicable, protected from risks to often on a voluntary basis, to address work injury
7 For more details, see: https://www.ilo.org/dyn/normlex/en/f?p=NORMLEXPUB:12100:0::NO::P12100_INSTRUMENT_ID:312535.

in some countries, other areas require attention, seek to eliminate discrimination on the basis of
such as health insurance, sickness benefits, ma- race, colour, sex, religion or social origin, and other
ternity and unemployment protection, disability grounds decided at the national level. While the
benefits and old age pensions. Greater state inter- platforms’ terms of service agreements prohibit
vention is required to guarantee the human right discriminatory conduct, and many jurisdictions
to social security to platform workers through a have well-developed anti-discrimination laws,
clearly identified legal framework. there nonetheless exist issues of discrimination
on platforms that emanate from the ways in which
Some of the innovative approaches and devel-
they and their algorithms are designed. Another
opments referred to in Chapter 5 suggest that
challenge that requires attention concerns the
platform workers can be effectively brought
application of anti-discrimination laws to platform
under the umbrella of social security regard-
workers in situations where the platform and its
less of their employment relationship. This is
clients are not in the same jurisdiction.
essential not only to guarantee adequate social
security coverage for platform workers, but also The seventh recommendation relates to providing
to ensure fair competition for enterprises. What access to efficient, equitable and participative
is clearly required is to extend social security to dispute resolution processes for all platform
platform workers by adapting existing policy, legal workers, to ensure that where a dispute over a
and administrative frameworks, including con- work entitlement occurs, it can be appropriately
tributory and non-contributory mechanisms, in resolved. One major issue with online web-based
line with the Social Security (Minimum Standards) platforms is that a platform and its clients and
Convention, 1952 (No. 102), the Social Protection workers can each be located in a different juris-
Floors Recommendation, 2012 (No. 202), and other diction and identifying and applying the law of a
relevant standards. particular jurisdiction in such a situation can be
The fifth recommendation for policy action relates challenging. Furthermore, there are risks that in
to implementing norms for fair remuneration and some jurisdictions the platform can select a gov-
working time. These issues are closely interlinked, erning law of contract from countries where the
as low remuneration leads to excessive working labour standards are weak, which would deprive
hours with implications for both physical and workers of any (potential) benefits (Cherry 2020).
mental health, as has been illustrated in Chapter 4. Even location-based platforms can purport to do
Therefore, though the ILO Conventions relating to so or even deny workers access to local courts.
minimum wages and working time are contingent Although many jurisdictions do not permit this,
upon the employment relationship, they should in those that do, social dialogue processes could
arguably be extended to all platform workers. serve to ensure that workers are able to take their
Some national jurisdictions, such as France, cases to the local jurisdiction.
have extended certain working time provisions
The final recommendation relates to data pro-
to self-employed platform workers in the trans-
tection, as the digital economy is driven by data,
portation industry, and have adopted innovative
and the ownership of and control over data can
mechanisms to calculate their hourly rates; such
have significant impacts on workers, businesses
mechanisms could be adapted to ensure fair
and countries’ development processes. Digital
remuneration and impose a maximum limit on
platforms currently assume exclusive owner-
working time.
ship over the user data they collect, and they
The sixth recommendation for policy action relates control the ways in which such data can be used
to non-discrimination and equality of treatment and shared. This default ownership of data by
with respect to gender, disability, nationality, eth- platforms inevitably leads to an imbalance of
nicity and migrant status, among others. The ILO power between capital and labour, with workers
Equal Remuneration Convention, 1951 (No. 100), being unable to leverage their data to engage in
and the Discrimination (Employment and collective action and collective bargaining. A more
Occupation) Convention, 1958 (No. 111), ensure balanced data governance framework should be
equal remuneration for men and women, and also explored that shares user rights over data with the
platform workers, individuals and communities 6.3.2 Relevance of other

that generate it. There has been some progress in
this regard through data protection regimes such fields of law and policy
as the EU’s General Data Protection Regulation for decent work on digital
(GDPR), and a number of developing countries
such as Brazil, India and Nigeria are drafting data labour platforms
protection legislation along similar lines, as illus-
The initiatives in labour protection and social
trated in Chapter 5.
security touch upon only some of the issues re-
Such data protection laws apply to workers ir- lated to work carried out through digital labour
respective of their employment status and are platforms. Other fields of regulation, such as
instrumental in giving data subjects more control competition law and the emerging law covering
over their data by according them individual algorithms and taxation, are also highly relevant
rights of access, deletion, portability and more. for ensuring decent work and shape the legal
For instance, the GDPR gives workers the right and policy space in which platforms and their
to access personal data and to request that it be workers operate (see figure 6.2). These are dis-
corrected if inaccurate. If a rating system is auto- cussed below.
mated, the worker has the right to a meaningful
explanation and to “obtain human intervention” Competition
in the decision-making process. Furthermore, as
per Article 40 of the GDPR, digital platforms can In many jurisdictions, competition law prohibits
adopt codes of conduct wherein they can commit self-employed workers from exercising their right
themselves to the fair and transparent processing to collective bargaining, on the grounds that
of data. this would constitute a cartel. This weakens the
position of workers on digital labour platforms,
In addition to ensuring individual rights over data,
preventing them from engaging in coordinated
important strides are being made towards estab-
negotiation with platform operators. However,
lishing collective rights over community data. The
some countries have been introducing exceptions
bargaining power of communities is potentially
for certain categories of dependent self-employed
greater than that of individuals, and thus collective
workers, and similar efforts are also being made
rights over community data can help workers to
at EU level. The Executive Vice-President of the
meaningfully negotiate their working conditions
European Commission, in charge of competition
with platforms (P.J. Singh 2020; De Stefano 2019).
policy, clarified in June 2020 that the Commission
The issue of data ownership and control also tran- was committed to improving the working condi-
scends national borders, and there is an intense tions of platform workers, especially in today’s
debate around data localization versus the free labour market where the concepts of “worker”
flow of data (UNCTAD 2018). Data localization is and “self-employed” have become blurred,
gaining prevalence among developing countries, stressing that “competition rules are not there to
so as to ensure a more equitable sharing of the stop workers forming a union” and that therefore
value created within the digital economy and to there was a “need to provide clarity to those who
bridge the digital divide. However, it is also argued need to negotiate collectively in order to improve
that data localization could hinder the movement their working conditions”. 8 Ensuring that all
of data and thus undermine the agility and mo- workers, including the self-employed, benefit from
bility offered by the digital economy. It therefore collective representation, bargaining and negoti-
becomes evident that there is a need for effective ation would entail reviewing existing competition
data governance to strike a balance between pri- law and making certain that workers on digital
vacy and domestic development on the one hand, labour platforms enjoy these rights. Furthermore,
and free flow of data on the other. competition law can also help in addressing
8 See https://ec.europa.eu/commission/presscorner/detail/en/ip_20_1237.
Figure 6.2 Other fields of law and policy relevant

for platform workers
Artificial
Competition Taxation
intelligence
issues related to non-competition clauses, ex- tasks, evaluating and rating performance, deacti-
clusivity agreements, high commissions, and vating users, and dynamically calculating prices.
preferential or discriminatory treatment on digital The opacity of the algorithms deployed for such
labour platforms. automated decision-making entails a number
of risks for workers and businesses, such as dis-
crimination on the basis of gender, ethnicity and
physical location of the worker, among others, as
The opacity of algorithms well as unfair competition. The ILO’s independent
entails a number of risks Global Commission on the Future of Work calls
for “adopting a ‘human-in-command’ approach
for workers and businesses. to artificial intelligence that ensures that the final
decisions affecting work are taken by human
beings” (Global Commission 2019, 13).
Competition law is also relevant for ensuring a To address issues of discrimination or unfair
level playing field for businesses. This is the case competition, and to rectify potential violations,
for both traditional businesses and technology it would be necessary to have access to and
start-ups, which may be either directly competing analyse the source code of an algorithm. Such
with digital platforms or using platforms to access access is of the utmost importance, in order
a wider customer base. Emerging scrutiny re- to ensure transparency and accountability in
garding antitrust issues in the context of platform artificial intelligence (AI). For instance, if the
companies is indicative of the growing relevance matching algorithm of a digital labour platform
of ensuring fair competition in the digital economy excludes workers on the basis of nationality
(see section 3.4). from accessing tasks, then an examination of
the algorithm’s source code is the only way to
Artificial intelligence ascertain whether it has been programmed to do
so and whether such discrimination is justified.
This report has shown that work processes on Similarly, it would be difficult to show that the
digital labour platforms are increasingly being pricing algorithm of an app-based taxi company
automated, and that digital labour platforms is producing anti-competitive outcomes without
deploy algorithms for various purposes. These inspecting its source code. At present, however,
include matching workers and clients, assigning several trade agreements prohibit access to or the
transfer of source codes of algorithms, and similar the ambit of taxation. This has implications for tax
proposals are pending in the WTO e-commerce revenues in developing countries, in particular
negotiations. A blanket prohibition on access has with regard to digital trade and taxation of both
serious potential implications for the pursuit of workers and businesses. The lack of adequate
legitimate public interest objectives such as com- public funds available for social expenditure al-
bating discrimination and protecting consumers location affects the provision of social protection
and workers. to the population, including workers (Behrendt,
Nguyen and Rani 2019). Furthermore, it can have
To mitigate the undesirable consequences of AI
a detrimental effect both on the recovery from the
deployment and practices, a number of ad hoc
COVID-19 crisis and on the economic outcomes
policies are emerging that highlight the need for
of countries looking to the digital economy as a
transparency and accountability. Some jurisdic-
vehicle for sustainable development.
tions (Australia, China, the EU, Japan, Singapore,
and the United States) have already started to Two proposals to effectively tax the digital
develop regulatory frameworks for AI,9 which may economy are under consideration at the United
in the future impose transparency and account- Nations and the OECD. The OECD/G20 Inclusive
ability obligations for deployers of algorithms to Framework has put forward a “two-pillar ap-
safeguard against the adverse effects of AI. In proach”.10 The first pillar seeks to update existing
addition, governments could consider adopting nexus rules so that a multinational enterprise
public policies that favour the use of open (MNE) can be taxable on its global profits, while
source technologies and allow for the auditing the second pillar seeks to put in place a global
of algorithms’ source codes by public regulatory minimum corporate tax rate so that an MNE would
authorities or specialized agencies. pay this minimum tax in each jurisdiction where it
operates, effectively ending the era of tax havens.
Taxation The developing country members of the UN Com-

mittee of Experts on International Cooperation
Taxation is another area that indirectly affects in Tax Matters have put forward an alternative
platform workers and has direct implications for proposal: first, to tax software payments as royal-
developed and developing countries alike. Digital ties,11 by amending Article 12 (Royalties) of the UN
labour platforms rely heavily on intangibles such Model Double Taxation Convention between De-
as software and algorithms (OECD 2014). This veloped and Developing Countries; and, second,
allows them to access global markets but can be to introduce a new Article 12B12 that would tax
challenging from the perspective of taxation. In “income from automated digital services”, defined
addition, when the platforms, clients or businesses as “any payment in consideration for any service
and workers are from different jurisdictions, it be- provided on the internet or an electronic network
comes even more challenging to bring them within requiring minimal human involvement from the
9 See relevant discussions at the EU: https://ec.europa.eu/digital-single-market/en/artificial-intelligence; the OECD principles of

responsible AI: https://www.oecd.org/going-digital/ai/principles/; Beijing AI principles: https://www.baai.ac.cn/news/beijing-ai-prin-
ciples-en.html; Australian Government AI Ethics Principles: https://www.industry.gov.au/data-and-publications/building-austral-
ias-artificial-intelligence-capability/ai-ethics-framework/ai-ethics-principles; Japan, Social Principles of Human-Centric AI: https://
www.cas.go.jp/jp/seisaku/jinkouchinou/pdf/humancentricai.pdf; Singapore’s National AI Strategy: https://www.smartnation.
gov.sg/why-Smart-Nation/NationalAIStrategy; United States Memorandum for Guidance for Regulation of Artificial Intelligence
Applications: https://www.whitehouse.gov/wp-content/uploads/2020/01/Draft-OMB-Memo-on-Regulation-of-AI-1-7-19.pdf and
among the G20: https://www.mofa.go.jp/files/000486596.pdf. See also the formation of a Global Partnership on AI: https://oecd.
ai/wonk/oecd-and-g7-artificial-intelligence-initiatives-side-by-side-for-responsible-ai.
10 For more details, see: https://www.southcentre.int/tax-cooperation-policy-brief-10-november-2019/.
11 For more details, see: https://www.un.org/development/desa/financing/sites/www.un.org.development.desa.financing/
files/2020-10/CRP38%20Inclusion%20of%20software%20payments%20in%20royalties%20%207OCT20.pdf.
12 For more details, see: https://us5.campaign-archive.com/?u=fa9cf38799136b5660f367ba6&id=f330625ffc.
service provider”.13 This would include online seeks to tax this income on either a gross or net
intermediation platform services, social media basis. If accepted, this proposal would update the
services and the sale of user data, among others. existing UN Model Tax Convention and would be
The proposal gives the taxing right to the juris- a move towards ensuring fiscal space, especially
diction where the payments have been made and in developing countries.
6.4 A way forward

As we move forward, leveraging the potential of its inspiration from the ILO Maritime Labour
the digital economy and of digital labour platforms Convention, 2006 (No. 186), which sets a guiding
in particular is likely to be critical to advancing precedent as it concerns seafarers who transcend
sustainable development, creating an enabling geographical borders, and involves multiple
environment for enterprises and promoting parties operating across different jurisdictions.
decent work for all. It will, however, also be vital A similar sectoral approach could be considered
to address the challenges that have been brought for digital labour platforms.14 It also calls for
about by the rise of digital labour platforms, and a “human-in-command” approach, regulation
to shape the transformations that the world of of data use, and algorithmic accountability in
work is experiencing in a way that benefits both the world of work, as well as reformed taxation
businesses and workers. Overcoming the chal- systems (Global Commission 2019). Moreover,
lenges and seizing the opportunities before us will the ILO Centenary Declaration for the Future of
require hard-law and soft-law regulatory frame- Work calls for “policies and measures that ensure
works and public policies to adequately address appropriate privacy and personal data protection,
the unprecedented complexities of today’s digital and respond to challenges and opportunities in
economy, in which labour regulation has a decisive the world of work relating to the digital trans-
role to play. Such an approach would also require formation of work, including platform work”
a mix of responses across other fields of law and (ILO 2019a, 6) in order to promote inclusive and
policy, such as competition, artificial intelligence sustainable economic growth, full and productive
and taxation, which are relevant from the perspec- employment and decent work for all.
tive of labour. Given the diversity and inchoate
Another important point of departure could
nature of many of the regulatory responses to
be the ILO Tripartite Declaration of Principles
platform labour, some form of international
concerning Multinational Enterprises and Social
regulatory dialogue and policy coordination would
Policy (MNE Declaration) (2017). It provides
assist in clarifying some of the regulatory uncer-
guidance to multinational enterprises on social
tainties and restating the applicability of universal
policy and inclusive, responsible and sustainable
labour standards.
workplace practices. It defines MNEs as including
To this end, the ILO’s independent Global enterprises which control services outside the
Commission on the Future of Work calls for country in which they are based. Platforms could
an international governance system that re- use the guidance provided in the MNE Declaration
quires platforms and their clients to respect to develop clear codes of conduct for members,
certain minimum rights and protections (Global including published procedures for workers to
Commission 2019). The Commission draws raise their concerns.
13 For more details, see page 9 of the Co-Coordinators’ Report to the UN Committee of Experts on International Cooperation
in Tax Matters, available at: https://www.un.org/development/desa/financing/sites/www.un.org.development.desa.financing/
files/2020-10/CITCM%2021%20CRP.41_Digitalization%2010102020%20Final%20A.pdf.
14 See also Markov and Travieso (2019), who suggest the establishment of an internationally agreed online protocol through an
international standard, which could lead to international harmonization of digital platforms.
A concerted effort
A concerted effort across multiple international
forums and organizations will be critical to
across multiple international
ensuring that digital labour platforms develop forums and organizations
further in a manner that strongly contributes to
inclusive and sustainable development. Such a will be critical to ensuring
process of regulatory dialogue and coordination
should have at its core an effort to ensure that that digital labour platforms
domestic laws implementing the fundamental
principles and rights at work as well as other key
contribute to inclusive and
legal provisions, such as those in respect of occu- sustainable development.
pational safety and health and social security,
apply to all workers, including digital labour plat-
form workers.
working towards ensuring that self-employed
X
Given the range of relevant actors and policy platform workers enjoy the right to bargain
areas, progress can best be achieved through collectively, for example through greater har-
social dialogue among the relevant stakeholders, monization of competition law with labour law;
most particularly the digital labour platforms, the
reaffirming that anti-discrimination and occu-
X
platform workers, and their representatives and
pational safety and health laws apply to digital
governments. With the right engagement and
labour platforms and their workers;
preparation, their efforts could lead over time to
a clearer understanding and a more effective and ensuring adequate social security benefits
X
consistent approach at the enterprise, national for all workers, including platform workers,
and international levels, with a view to: by extending and adapting policy and legal
frameworks where necessary;
ensuring fair competition and creating an enab-
X
ling environment for sustainable enterprises; ensuring fair termination processes for plat-
X
form workers;
requiring and promoting clear and transparent
X
terms of engagement and contractual arrange- ensuring access to independent dispute reso-
X
ments for workers and businesses, including as lution mechanisms;
reflected in labour and consumer laws;
ensuring that platform workers are able to
X
ensuring that workers’ employment status is
X access the courts of the jurisdiction in which
correctly classified and is in accordance with they are located if they so choose;
national classification systems;
providing for wage protection, fair payments
X
ensuring transparency in ratings or rankings of
X and working time standards;
workers and businesses using digital platforms
allowing platform workers to move freely
X
such as online web-based, location-based and
between platforms, including by facilitating
e-commerce platforms;
portability of workers’ data, for example re-
ensuring transparency and accountability of
X garding ratings; and
algorithms for workers and businesses;
aiming at effectively taxing the digital economy,
X
protecting workers’ personal and work data,
X including platforms, clients and workers, as well
as well as data relating to businesses and their as their transactions.
activities on platforms;
Appendices
https://www.ilo.org/weso2021
1. Digital labour platforms:

Estimates of workers, investments
and revenues
2. ILO interviews with digital

platform companies and analysis
of terms of service agreements
3. ILO interviews with businesses

and clients
4. ILO surveys, interviews

and statistical analysis
5. ILO Interviews with unions

and associations
References
Adams-Prassl, Abi, and Janine Berg. 2017. “When Home Affects Pay: An Analysis of
the Gender Pay Gap Among Crowdworkers”, Social Science Research Network
(SSRN) Scholarly Paper ID 3048711.
Adams-Prassl, Abi, Teodora Boneva, Marta Golin, and Christopher Rauh. 2020. “The
Large and Unequal Impact of COVID-19 on Workers”. VoxEU, 8 April.
AfDB (African Development Bank), ADB (Asian Development Bank), EBRD (European
Bank for Reconstruction and Development), and IDB (Inter-American Development
Bank). 2018. The Future of Work: Regional Perspectives. Washington, DC.
Agrawal, Ajay, John Horton, Nicola Lacetera, and Elizabeth Lyons. 2013. “Digitization
and the Contract Labor Market: A Research Agenda”, National Bureau of Economic
Research (NBER) Working Paper 19525.
Agustí-Panareda, Jordi, Franz Christian Ebert, and Desirée LeClercq. 2014. “Labour
Provisions in Free Trade Agreements: Fostering Their Consistency with the
ILO Standards System”, Background Paper, Social Dimensions of Free Trade
Agreements. Geneva: ILO.
Ajunwa, Ifeoma. 2020. “The Paradox of Automation as Anti-Bias Intervention”. Cardozo
Law Review 41.
Akhtar, Pav, Phoebe Moore, and Martin Upchurch. 2018. “Digitalisation of Work
and Resistance”. In Humans and Machines at Work: Monitoring, Surveillance and
Automation in Contemporary Capitalism, edited by Phoebe Moore, Martin Upchurch
and Xanthe Whittaker, 17–44. Palgrave Macmillan.
Aleksynska, Mariya. 2021. “Digital Work in Eastern Europe: Overview of Trends,
Outcomes, and Policy Response”, ILO Working Paper.
Aleksynska, Mariya, Anastasia Bastrakova, and Natalia Kharchenko. 2018. Work on
Digital Labour Platforms in Ukraine: Issues and Policy Perspectives. Geneva: ILO.
Aleksynska, Mariya, Andrey Shevchuk, and Denis Strebkov. 2021. “Online Platform
Work in Russia and Ukraine: Gender Differences in Earnings and Work Satisfaction”.
In Gendering Post-Soviet Space: Demography, Labor Market and Values in Empirical
Research, edited by Tatiana Karabchuk, Kazuhiro Kumo, Kseniia Gatskova and
Ekaterina Skoglund, 277–299. Singapore: Springer.
Alexander, Julia. 2020. “Creators Finally Know How Much Money YouTube Makes, and
They Want More of it”. The Verge, 4 February.
Aloisi, Antonio. 2016. “Commoditized Workers: Case Study Research on Labor Law
Issues Arising from a Set of On-Demand/Gig Economy Platforms”. Comparative
Labor Law & Policy Journal 37 (3): 653–690.
Aloisi, Antonio, and Elena Gramano. 2018. “Non-Standard Workers and Collective
Rights”. In Industrial Relations in Europe Conference (IREC). Leuven.
Alsos, Kristin, Kristin Jesnes, Beate Sletvold Ølstad, and Torstein Nesheim. 2017. “Når
sjefen er en app”, Fafo Report No. 41. Oslo: Fafo Research Foundation.
Anwar, Mohammad Amir, and Mark Graham. 2019. “Does Economic Upgrading Lead
to Social Upgrading in Contact Centers? Evidence from South Africa”. African
Geographical Review 38 (3): 209–226.
———. 2020. “Between a Rock and a Hard Place: Freedom, Flexibility, Precarity and
Vulnerability in the Gig Economy in Africa”. Competition & Change 1–22.
Armour, John, and Mari Sako. 2020. “AI-Enabled Business Models in Legal Services:
From Traditional Law Firms to Next-Generation Law Companies?” Journal of
Professions and Organization 7 (1): 27–46.
Arntz, Melanie, Terry Gregory, and Ulrich Zierahn. 2016. “The Risk of Automation
for Jobs in OECD Countries”, OECD Social, Employment and Migration Working
Papers No. 189.
Arrieta-Ibarra, Imanol, Leonard Goff, Diego Jiménez-Hernández, Jaron Lanier, and
E. Glen Weyl. 2018. “Should We Treat Data as Labor? Moving Beyond ‘Free’”.
AEA Papers and Proceedings 108 (May): 38–42.
Arthurs, Harry. 2010. “Extraterritoriality by Other Means: How Labor Law Sneaks
Across Borders, Conquers Minds, and Controls Workplaces Abroad”. Stanford
Law and Policy Review 21 (3): 527–554.
Ayres, Ian, and John Braithwaite. 1992. Responsive Regulation: Transcending the
Deregulation Debate. Oxford Socio-Legal Studies. New York: Oxford University
Press.
Ayyar, Ranjani. 2017. “Ever-Changing Incentives Keeping Drivers on Their Toes”.
The Times of India, 31 March.
Azevedo, Mary Ann. 2020. “Report: VC Funding in Latin America More Than Doubled
to a Record $4.6B in 2019”. Crunchbase News, 6 May.
Babu, Venkatesha. 2015. “Coolest Start-Ups 2015: HackerEarth Helps Find Top-Quality
Coders for Tech Companies”. Business Today, 21 June.
Bakos, Yannis, Florencia Marotta-Wurgler, and David R. Trossen. 2014. “Does Anyone
Read the Fine Print? Consumer Attention to Standard-Form Contracts”. The Journal
of Legal Studies 43 (1): 1–35.
Ball, Kirstie. 2010. “Workplace Surveillance: An Overview”. Labor History 51 (1): 87–106.
Bamforth, Nicholas. 2004. “Conceptions of Anti-Discrimination Law”. Oxford Journal
of Legal Studies 24 (4): 693–716.
Ban, Xiaohui. 2020. “Identifying Labour Relationship in the Sharing Economy: Judicial
Practice in China”. In Regulating the Platform Economy: International Perspectives
on New Forms of Work, edited by Lourdes Mella Méndez, 48–60. Abingdon, UK:
Routledge.
Barnard, Catherine, and Alysia Blackham. 2017. “Discrimination and the Self-Employed:
The Scope of Protection in an Interconnected Age”. In European Contract Law and
the Charter of Fundamental Rights, edited by Hugh Collins, 197–218. Cambridge,
UK: Cambridge University Press.
Barzilay, Arianne Renan, and Anat Ben-David. 2017. “Platform Inequality: Gender in
the Gig-Economy”. Seton Hall Law Review 47 (2): 393–431.
References 261
Beaudonnet, Xavier. 2020. “Le droit de négociation collective des travailleurs con
sidérés comme indépendants au regard des norms de l’organisation internationale
du travail”. In Le droit de négociation collective des travailleurs indépendants: Cadrages
théoriques et études de cas, edited by Daniel Dumont, Auriane Lamine and Jean-
Benoît Maisin, 55–79. Droit Social, Larcier.
Beerepoot, Niels, and Bart Lambregts. 2015. “Competition in Online Job Marketplaces:
Towards a Global Labour Market for Outsourcing Services?” Global Networks
15 (2): 236–255.
Behrendt, Christina, and Quynh Anh Nguyen. 2018. “Innovative Approaches for
Ensuring Universal Social Protection for the Future of Work”, ILO Future of Work
Series, Research Paper 1.
Behrendt, Christina, Quynh Anh Nguyen, and Uma Rani. 2019. “Social Protection
Systems and the Future of Work: Ensuring Social Security for Digital Platform
Workers”. International Social Security Review 72 (3): 17–41.
Belvedere, Matthew J. 2018. “Upwork Shares Rocket More Than 50% Higher at the
Open on Their First Day of Trading”. CNBC, 3 October.
Berg, Janine. 2016. “Income Security in the On-Demand Economy: Findings and
Policy Lessons from a Survey of Crowdworkers”, ILO Conditions of Work and
Employment Series No. 74.
———. 2019. “Protecting Workers in the Digital Age: Technology, Outsourcing and
the Growing Precariousness of Work”. Comparative Labour Law and Policy Journal
41 (1): 66–94.
Berg, Janine, Uma Rani, Marianne Furrer, Ellie Harmon, and M. Six Silberman. 2018.
Digital Labour Platforms and the Future of Work: Towards Decent Work in the Online
World. Geneva: ILO.
Berkowitz, Daniel, Katharina Pistor, and Jean-Francois Richard. 2003. “Economic
Development, Legality, and the Transplant Effect”. European Economic Review
47 (1): 165–195.
Bessa, Ioulia, Simon Joyce, Denis Neumann, Mark Stuart, Vera Trappmann, and
Charles Umney. Forthcoming. “Worker Protest in the Platform Economy”, ILO
Working Paper.
Bezos, Jeff. 2020. “Statement by Jeff Bezos to the U.S. House Committee on the
Judiciary: Testimony before the Subcommittee on Antitrust, Commercial, and
Administrative Law”. The Amazon Blog, 28 July.
Bhalla, Tarush. 2017. “With $4.5M under Its Belt, HackerEarth Is Finding New Ways
to Engage with Developers in India”. YourStory, 12 April.
Black, Julia. 2001. “Decentring Regulation: Understanding the Role of Regulation and
Self-Regulation in a ‘Post-Regulatory’ World”. Current Legal Problems 54 (1): 103–146.
———. 2002. “Critical Reflections on Regulation”. Australian Journal of Legal Philosophy
27: 1–35.
Blackham, Alysia. 2018. “‘We Are All Entrepreneurs Now’: Options and New Approaches
for Adapting Equality Law for the ‘Gig Economy’”. International Journal of
Comparative Labour Law and Industrial Relations 34 (4): 413–434.
Blevins, Cameron, and Lincoln Mullen. 2015. “Jane, John ... Leslie? A Historical Method
for Algorithmic Gender Prediction”. Digital Humanities Quarterly 9 (3).
Bloomberg Law. 2020. “Amazon Accused of Monopolization, Sweeping Price-
Fixing Scheme”. https://news.bloomberglaw.com/mergers-and-antitrust/
amazon-accused-of-monopolization-massive-price-fixing-scheme.
Bodie, Matthew T., Miriam A. Cherry, Marcia L. McCormick, and Jintong Tang. 2016.
“The Law and Policy of People Analytics”, Social Science Research Network (SSRN)
Scholarly Paper ID 2769980.
Bornstein, Stephanie. 2018. “Antidiscriminatory Algorithms”. Alabama Law Review
70 (2): 519–572.
Boudreau, John W., Ravin Jesuthasan, and David Creelman. 2015. Lead the Work:
Navigating a World Beyond Employment. London: Wiley.
Boudreau, Kevin J., and Andrei Hagiu. 2009. “Platform Rules: Multi-Sided Platforms
as Regulators”. In Platforms, Markets and Innovation, edited by Annabelle Gawer.
Cheltenham, UK: Edward Elgar.
Boudreau, Kevin J., and Lars B. Jeppesen. 2015. “Unpaid Crowd Complementors: The
Platform Network Effect Mirage”. Strategic Management Journal 36 (12): 1761–1777.
Boudreau, Kevin J., and Karim R. Lakhani. 2013. “Using the Crowd as an Innovation
Partner”. Harvard Business Review 91(4): 61–69.
Brito, Miguel, and Celestino Gonçalves. 2019. “Codeflex: A Web-Based Platform for
Competitive Programming”. In 2019 14th Iberian Conference on Information Systems
and Technologies (CISTI), 1–6.
Brown, Andrew, Andy Charlwood, and David A. Spencer. 2012. “Not All That It Might
Seem: Why Job Satisfaction Is Worth Studying Despite It Being a Poor Summary
Measure of Job Quality”. Work, Employment and Society 26 (6): 1007–1018.
Budd, John W., and Alexander J.S. Colvin. 2008. “Improved Metrics for Workplace
Dispute Resolution Procedures: Efficiency, Equity, and Voice”. Industrial Relations:
A Journal of Economy and Society 47 (3): 460–479.
Burson-Marsteller, The Aspen Institute, and Time. 2016. “The On-Demand Economy
Survey”, Future of Work Initiative, 21 June. Washington, DC: The Aspen Institute.
Butollo, Florian. 2020. “Digitalization and the geographies of production: Towards
reshoring or global fragmentation?” Competition & Change, ISSN 1477-2221.
Butollo, Florian, and Lea Schneidemesser. Forthcoming. “Specialization Revisited:
B2B Platforms as Distribution-Driven Path towards Digitized Manufacturing”.
International Labour Review.
Bygrave, Lee A. 2015. Internet Governance by Contract. Oxford: Oxford University Press.
Cagle, Susie. 2020. “‘Fees Are Murder’: Delivery Apps Profit as Restaurants Forced to
Close Doors”. The Guardian, 3 April.
Canada, Statcan (Statistics Canada). 2017. The Sharing Economy in Canada.
Castells, Manuel. 2010. The Rise of the Network Society. New York: Wiley.
Chapman, Lizette, and Sarah Frier. 2020. “Uber and Lyft Cars Become Vector for
Racism amid Coronavirus Fears”. Los Angeles Times, 6 February.
Chen, Julie Yujie, and Jack Linchuan Qiu. 2019. “Digital Utility: Datafication, Regulation,
Labor, and DiDi’s Platformization of Urban Transport in China”. Chinese Journal of
Communication 12 (3): 274–289.
Chen, Yiu Por (Vincent). Forthcoming. “Online Digital Labour Platforms in China:
Working Conditions, Policy Issues and Prospects”, ILO Working Paper.
References 263
Cherry, Miriam A. 2020. “A Global System of Work, A Global System of Regulation?:

Crowdwork and Conflicts of Law”. Tulane Law Review 94 (2): 184–245.
Cherry, Miriam A., and Antonio Aloisi. 2017. “Dependent Contractors in the Gig
Economy: A Comparative Approach”. American University Law Review 66 (3): 635–689.
Cherry, Miriam A., and Winifred R. Poster. 2016. “Crowdwork, Corporate Social
Responsibility, and Fair Labor Practices”. In Research Handbook on Digital
Transformations, edited by F. Xavier Olleros and Majlinda Zhegu, 291–312.
Cheltenham, UK; Northampton, MA: Edward Elgar.
Choudary, Sangeet Paul. 2018. “The Architecture of Digital Labour Platforms: Policy
Recommendations on Platform Design for Worker Well-Being”, ILO Future of
Work Series, Research Paper 3.
Chui, Michael, James Manyika, Mehdi Miremadi, Nicolaus Henke, Rita Chung, Pieter
Nel, and Sankalp Malhotra. 2018. “Notes from the AI Frontier: Insights from
Hundreds of Use Cases”. McKinsey Global Institute.
CIPD (Chartered Institute of Personnel and Development). 2017. To Gig or Not to Gig?
Stories from the Modern Economy. London.
Clark, Kate. 2019. “Unicorns Aren’t Profitable, and Wall Street Doesn’t Care”.
TechCrunch (blog). 26 March. https://social.techcrunch.com/2019/03/26/
unicorns-arent-profitable-wall-street-doesnt-care/.
Clewlow, Regina R., and Gouri Shankar Mishra. 2017. “Disruptive Transportation:
The Adoption, Utilization, and Impacts of Ride-Hailing in the United States”,
Research Report UCD-ITS-RR-17-07, Institute of Transportation Studies, University
of California, Davis.
Coiquaud, Urwana, and Isabelle Martin. 2019. “Accès à la justice des travailleurs de
plateformes numériques: Réponses contrastées des tribunaux canadiens et
américains”. Relations industrielles / Industrial Relations 74 (3): 577–588.
Collins, Hugh. 1990. “Independent Contractors and the Challenge of Vertical
Disintegration to Employment Protection Laws”. Oxford Journal of Legal Studies
10 (3): 353–380.
Colvin, Alexander J.S. 2019. “The Metastasization of Mandatory Arbitration”. Chicago-
Kent Law Review 94 (1): 3–24.
Compa, Lance. 2019. “Trump, Trade, and Trabajo: Renegotiating Nafta’s Labor Accord in
a Fraught Political Climate”. Industrial Journal of Global Legal Studies 26 (1): 263–304.
Competition Commission of India. 2020. “Market Study on E-Commerce in India”.
Cook, Cody, Rebecca Diamond, Jonathan Hall, John List, and Paul Oyer. 2018. “The
Gender Earnings Gap in the Gig Economy: Evidence from over a Million Rideshare
Drivers”, National Bureau of Economic Research (NBER) Paper W24732.
Cooney, Sean, Sarah Biddulph, and Ying Zhu. 2013. Law and Fair Work in China.
Abingdon: Routledge.
Corporaal, Greetje F., and Vili Lehdonvirta. 2017. “Platform Sourcing: How Fortune
500 Firms Are Adopting Online Freelancing Platforms”. Oxford Internet Institute,
University of Oxford.
Countouris, Nicola. 2018. “The Concept of ‘Worker’ in European Labour Law:
Fragmentation, Autonomy and Scope”. Industrial Law Journal 47 (2): 192–225.
———. 2019. “Defining and Regulating Work Relations for the Future of Work”, ILO
Governance and Tripartism Department.
Countouris, Nicola, and Luca Ratti. 2018. “The Sharing Economy and EU Anti-
Discrimination Law”. In The Cambridge Handbook of the Law of the Sharing Economy,
edited by John J. Infranca, Michèle Finck and Nestor M. Davidson, 486–498.
Cambridge, UK: Cambridge University Press.
Crémer, Jacques, Yves-Alexandre de Montjoye, and Heike Schweitzer. 2019.
“Competition Policy for the Digital Era”. European Commission, Directorate-
General for Competition.
Cusumano, Michael A., Annabelle Gawer, and David B. Yoffie. 2019. The Business of
Platforms: Strategy in the Age of Digital Competition, Innovation, and Power. New
York: Harper Collins.
Darkwah, Akosua K., and Dzodzi Tsikata. Forthcoming. “Home Based Work and
Homework in Ghana: An Exploration”. Geneva: ILO.
Davenport, Thomas, Abhijit Guha, Dhruv Grewal, and Timna Bressgott. 2020. “How
artificial intelligence will change the future of marketing”. Journal of the Academy
of Marketing Science 48: 24–42.
Davidov, Guy. 2014. “Setting Labour Law’s Coverage: Between Universalism and
Selectivity”. Oxford Journal of Legal Studies 34 (3): 543–566.
Davidov, Guy, and Brian Langille. 2011. The Idea of Labour Law. Oxford: Oxford
University Press.
Davies, Paul, and Mark Freedland. 2006. “The Complexities of the Employing
Enterprise”. In Boundaries and Frontiers of Labour Law, edited by Guy Davidov
and Brian Langille. Portland: Hart.
Deakin, Simon. 2007. “Does the ‘Personal Employment Contract’ Provide a Basis
for the Reunification of Employment Law?”. Industrial Law Journal 36 (1): 68–83.
Dedeoglu, Saniye. Forthcoming. “Homeworkers in Turkey: Home Bounded-Global
Outreach”. Geneva: ILO.
de la Merced, Michael J., and Kate Conger. 2019. “Uber I.P.O. Values Ride-Hailing Giant
at $82.4 Billion”. The New York Times, 9 May.
Deloitte. 2017. 2017 Deloitte Global Human Capital Trends: Rewriting the Rules for the
Digital Age. Deloitte University Press.
———. 2018. 2018 Deloitte Global Human Capital Trends: The Rise of the Social Enterprise.
Deloitte University Press.
———. 2019. 2019 Deloitte Global Human Capital Trends: Leading the Social Enterprise
– Reinvent with a Human Focus. Deloitte University Press.
De Stefano, Valerio. 2016. “The Rise of the ‘Just-in-Time Workforce’: On-Demand
Work, Crowd Work and Labour Protection in the ‘Gig-Economy’”, ILO Conditions
of Work and Employment Series No. 71.
———. 2019. “‘Negotiating the Algorithm?’: Automation, Artificial Intelligence and
Labour Protection”. Comparative Labour Law and Policy Journal 41 (1): 15–46.
De Stefano, Valerio, and Antonio Aloisi. 2018. “European Legal Framework for ‘Digital
Labour Platforms’”. Luxembourg: European Commission Joint Research Centre.
References 265
———. 2019. “Fundamental Labour Rights, Platform Work and Human-Rights

Protection of Non-Standard Workers”. In Labour, Business and Human Rights Law,
edited by Janice R. Bellace, Samuel Blank and Beryl Haar, 359–379. Cheltenham,
UK; Northampton, MA: Edward Elgar.
De Stefano, Valerio, and Mathias Wouters. 2019. “Should Digital Labour Platforms
Be Treated as Private Employment Agencies?” European Trade Union Institute
(ETUI) Foresight Brief.
Dhillon, Amrit. 2018. “‘My Life is Spent in This Car’: Uber Drives Its Indian Workers to
Despair”. The Guardian, 4 December.
Difallah, Djellel, Elena Filatova, and Panos Ipeirotis. 2018. “Demographics and Dynamics
of Mechanical Turk Workers”. In Eleventh ACM International Conference, 135–143.
Marina Del Rey, CA: ACM Press.
Dockès, Emmanuel. 2019. “New Trade Union Strategies for New Forms of Employment”.
European Labour Law Journal 10 (3): 219–228.
Drahokoupil, Jan, and Brian Fabo. 2016. “The Platform Economy and the Disruption
of the Employment Relationship”, European Trade Union Institute (ETUI) Policy
Brief No. 5.
Dube, Arindrajit, Jeff Jacobs, Suresh Naidu, and Siddharth Suri. 2020. “Monopsony in
Online Labor Markets”. American Economic Review: Insights 2 (1): 33–46.
Duch-Brown, Néstor. 2017a. “The Competitive Landscape of Online Platforms”. JRC
Technical Reports. European Commission.
———. 2017b. “Platforms to Business Relations in Online Platform Ecosystems”.
JRC Technical Reports, JRC Digital Economy Working Paper 2017-07. European
Commission.
Duggan, James, Ultan Sherman, Ronan Carbery, and Anthony McDonnell. 2020.
“Algorithmic Management and App-Work in the Gig Economy: A Research Agenda
for Employment Relations and HRM”. Human Resource Management Journal 30 (1):
114–132.
Ebisui, Minawa, Sean Cooney, and Colin Fenwick. 2016. Resolving Individual Labour
Disputes: A Comparative Overview. Geneva: ILO.
Economic Times, The. 2018. “Walmart Acquires Flipkart for $16 Billion in the World’s
Largest Ecommerce Deal”. 10 May.
Estlund, Cynthia. 2018. “The Black Hole of Mandatory Arbitration”. North Carolina
Law Review 96 (3): 679–710.
European Commission. 2016a. “Business-to-Business Relationships in the Online
Platforms Environment: Legal Aspects and Clarity of Terms and Conditions of
Online Platforms”.
———. 2016b. “Online Platforms: Contrasting Perceptions of European Stakeholders”.
———. 2017a. “Business-to-Business Relationships in the Online Platforms
Environment: Algorithms, Ranking and Transparency”.
———. 2017b. “Unfair Platform-to-Business Trading Practices: Focus Group with
Business Users”.
———. 2017c. “Unfair Platform-to-Business Trading Practices: Effective Dispute
Resolution and the Fundamental Right to Conduct a Business”.
———. 2018. “Study on Contractual Relationships between Online Platforms and

Their Professional Users”.
———. 2019. “How Do Online Platforms Shape Our Lives and Businesses?”
———. 2020. “Competition: The European Commission Launches a Process to Address
the Issue of Collective Bargaining for the Self-employed”. Press Release, 30 June.
EU (European Union). 2020. “The Platform Economy and Precarious Work”, Policy
Department for Economic, Scientific and Quality of Life Policies, Directorate-
General for Internal Policies.
Evans, David S., and Richard Schmalensee. 2008. “Markets with Two-Sided Platforms”.
In Issues in Competition Law and Policy, Vol. 1, edited by Wayne D. Collins, Josepha
Angland and the American Bar Association. Chicago: ABA Section of Antitrust Law.
———. 2016. Matchmakers: The New Economics of Multisided Platforms. Boston, MA:
Harvard Business Review Press.
Fairwork Project, The. 2020. The Gig Economy and Covid-19: Fairwork Report on Platform
Policies. Fairwork Foundation.
Farrell, Diana, and Fiona Greig. 2016. Paychecks, Paydays, and the Online Platform
Economy: Big Data on Income Volatility. New York: JPMorgan Chase Institute.
Farrell, Diana, Fiona Greig, and Amar Hamoudi. 2018. The Online Platform Economy
in 2018: Drivers, Workers, Sellers and Lessors. New York: JPMorgan Chase Institute.
Federal Reserve Board. 2019. “Report on the Economic Well-Being of U.S. Households
in 2018”. Washington, DC.
Florida, Richard, and Ian Hathaway. 2018. Rise of the Global Startup City: The New Map
of Entrepreneurship and Venture Capital. Washington, DC: Center for American
Entrepreneurship.
Foong, Eureka, Nicholas Vincent, Brent Hecht, and Elizabeth M. Gerber. 2018.
“Women (Still) Ask For Less: Gender Differences in Hourly Rate in an Online Labor
Marketplace”. Proceedings of the ACM on Human-Computer Interaction 2 (CSCW)
53: 1–21.
Foster, Christopher, Mark Graham, Laura Mann, Timothy Waema, and Nicolas
Friederici. 2018. “Digital Control in Value Chains: Challenges of Connectivity for
East African Firms”. Economic Geography 94 (1): 68–86.
Fountoukakos, Kyriakos, André Pretorius, and Lisa Geary. 2018. “Market Definition in
a Rapidly Changing (Digital) World: The Case of Ride-Sharing”. Competition Policy
International, June 25.
Fredman, Sandra. 2015. “Equality Law: Labour Law or an Autonomous Field?” In The
Autonomy of Labour Law, edited by Alan Bogg, 257–273. Oxford and Portland,
Oregon: Hart.
Freedland, Mark. 2003. The Personal Employment Contract. Oxford: Oxford University
Press.
Freedland, Mark, and Nicola Countouris. 2011. The Legal Construction of Personal Work
Relations. Oxford: Oxford University Press.
Freudenberg, Christoph. 2019. “Rising Platform Work: Scope, Insurance Coverage
and Good Practices among ISSA Countries”. Geneva: International Social Security
Association (ISSA).
References 267
Frey, Carl Benedikt, and Michael A. Osborne. 2017. “The Future of Employment: How
Susceptible Are Jobs to Computerisation?” Technological Forecasting and Social
Change 114 (C): 254–280.
Fudge, Judy. 2006. “Fragmenting Work and Fragmenting Organizations: The Contract
of Employment and the Scope of Labour Regulation”. Osgoode Hall Law Journal
44 (4): 609–648.
Fudge, Judy, Shae McCrystal, and Kamala Sankaran. 2012. Challenging the Legal
Boundaries of Work Regulation. Oxford and Portland, Oregon: Hart.
Fuller, Joseph, Manjari Raman, Allison Bailey, Nithya Vaduganathan, James Palano,
et al. 2020. Building the On-demand Workforce. Boston: Harvard Business School
and Boston Consulting Group.
Fumagalli, Andrea, Stefano Lucarelli, Elena Musolino, and Giulia Rocchi. 2018. “Digital
Labour in the Platform Economy: The Case of Facebook”. Sustainability 10 (6): 1–16.
Fundación ONCE and the ILO Global Business and Disability Network. 2019. “Making
the Future of Work Inclusive of People with Disabilities”.
Galpaya, Helani, and Laleema Senanayake. 2018. “Online Freelancing: Potential for
Digital Gig Work in India, Sri Lanka and Myanmar”. In The 22nd Biennial Conference of
the International Telecommunications Society: “Beyond the Boundaries: Challenges for
Business, Policy and Society”, June 24th–27th, 2018, Seoul, Korea. Seoul: International
Telecommunications Society (ITS).
Galperin, Hernan, and Catrihel Greppi. 2017. “Geographical Discrimination in the Gig
Economy”. Social Science Research Network (SSRN) Scholarly Paper ID 2922874.
Garben, Sacha. 2019. “The Regulatory Challenge of Occupational Safety and Health in
the Online Platform Economy”. International Social Security Review 72 (3): 95–112.
Gawer, Annabelle. 2014. “Bridging Differing Perspectives on Technological Platforms:
Toward an Integrative Framework”. Research Policy 43 (7): 1239–1249.
Global Commission on the Future of Work (Global Commission). 2019. Work for a
brighter future. Geneva: ILO.
Goldín, Adrían. 2020. “Los trabajadores de plataforma y su regulación en la Argentina”,
documentos de Proyectos (LC/TS.2020/44), Santiago, Comisión Económica para
América Latina y el Caribe (CEPAL).
Goldman, Tanya, and David Weil. 2020. “Who’s Responsible Here? Establishing Legal
Responsibility in the Fissured Workplace”, Institute for New Economic Thinking
Working Paper No. 114.
Gomez-Uribe, Carlos A., and Neil Hunt. 2015. “The Netflix Recommender System:
Algorithms, Business Value, and Innovation”. ACM Transactions on Management
Information Systems 6 (4): 1–19.
Grab. 2017. “Overview of Grab’s Personal Accident Insurance”. Grab ID (blog). 9 May.
https://www.grab.com/id/en/blog/asuransi/.
Graf, Michael, and Susan M. Mudambi. 2005. “The Outsourcing of IT-Enabled Business
Processes: A Conceptual Model of the Location Decision”. Journal of International
Management 11 (2): 253–268.
Graham, Mark, Isis Hjorth, and Vili Lehdonvirta. 2017. “Digital Labour and Development:
Impacts of Global Digital Labour Platforms and the Gig Economy on Worker
Livelihoods”. Transfer: European Review of Labour and Research 23 (2): 135–162.
Graham, Mark, Vili Lehdonvirta, Alex Wood, Helena Barnard, Isis Hjorth, and David
Peter Simon. 2017. “The Risks and Rewards of Online Gig Work at the Global
Margins”. Oxford Internet Institute, University of Oxford and Gordon Institute
of Business Science (GIBS), University of Pretoria.
Grand View Research. 2020. “Retail E-Commerce Market Size, Share and Trends
Analysis Report, 2020–2027”. May.
Green, Daryl, Craig Walker, Abdulrahman Alabulththim, Daniel Smith, and Michele
Phillips. 2018. “Fueling the Gig Economy: A Case Study Evaluation of Upwork.
Com”. Management and Economics Research Journal 4: 104–112.
Griesbach, Kathleen, Adam Reich, Luke Elliott-Negri, and Ruth Milkman. 2019.
“Algorithmic Control in Platform Food Delivery Work”. Socius: Sociological Research
for a Dynamic World 5: 1–15.
Grimshaw, Damian, Colette Fagan, Gail Hebson, and Isabel Tavora. 2017. “A new
labour market segmentation approach for analysing inequalities: Introduction
and overview”. In D. Grimshaw, C. Fagan, G. Hebson and I. Tavora (eds): Making
work more equal: A new labour market segmentation approach, 1–32. Manchester,
Manchester University Press.
Grimshaw, Damian, and Jill Rubery. 2015. “The Motherhood Pay Gap: A Review of
the Issues, Theory and International Evidence”, ILO Conditions of Work and
Employment Series No. 57.
Grooms, Jignasha Amin. 2017. “The Art and Science of Attracting Today’s Next-Gen
Workforce Is Digital by Design”. Strategic HR Review 16 (4): 177–181.
Grušić, Uglješa. 2012. “Jurisdiction in Employment Matters under Brussels I:
A Reassessment”. International & Comparative Law Quarterly 61 (1): 91–126.
Gupta, Ruchi. 2020. “Why Facebook Joined Google and Tencent in Gojek Investment”.
Market Realist, 4 June.
Gwilliam, Kenneth M. 2005. “Regulation of Taxi Markets in Developing Countries:
Issues and Options”, Urban Transport Thematic Group Transport Note No. TRN-3.
The World Bank.
Halegua, Aaron. 2016. “United States”. In Resolving Individual Labour Disputes:
A Comparative Overview, edited by Minawa Ebisui, Sean Cooney and Colin Fenwick,
311–347. Geneva: ILO.
Hancher, Leigh, and Michael Moran. 1998. “Organizing Regulatory Space”. In A Reader
on Regulation, edited by Robert Baldwin, Colin Scott and Christopher Hood,
148–172. Oxford Readings in Socio-Legal Studies. Oxford: Oxford University Press.
Hanrahan, Benjamin V., David Martin, Jutta Willamowski, and John M. Carroll. 2019.
“Investigating the Amazon Mechanical Turk Market Through Tool Design”. Computer
Supported Cooperative Work 28: 795–814.
Heeks, Richard. 2017. “Decent Work and the Digital Economy: A Developing Country
Perspective on Employment Impacts and Standards in Online Outsourcing,
Crowdwork, etc.”, Centre for Development Informatics Working Paper No. 71,
University of Manchester.
Henkel, Joachim, Simone Schöberl, and Oliver Alexy. 2014. “The Emergence of
Openness: How and Why Firms Adopt Selective Revealing in Open Innovation”.
Research Policy 43 (5): 879–890.
References 269
Herrmann, Andrea M., Petra Zaal, Maryse M.H. Chappin, and Brita Schemmann. 2019a.
“We Don’t Need No Education! How the Gig Economy Challenges the Current
Education Paradigm of the Western World”. In 31st Annual Conference of the Society
for the Advancement of Socio-Economics (SASE): “Fathomless Futures: Algorithmic and
Imagined”. New York: Society for the Advancement of Socio-Economics.
———. 2019b. “Does Education Still Matter in Online Labor Markets?”. In Perspectives
on the Sharing Economy, edited by Dominika Wruk, Achim Oberg and Indre Maurer.
Newcastle upon Tyne: Cambridge Scholars Publishing.
Hidalgo Cordero, Kruskaya, and Belén Valencia Castro. 2019. “Entre la precarización
y el alivio cotidiano: Las plataformas Uber Eats y Glovo en Quito”. Friedrich Ebert
Stiftung Ecuador.
Hillman, Robert A., and Jeffrey J. Rachlinski. 2002. “Standard-Form Contracting in the
Electronic Age”. New York University Law Review 77 (2): 429–495.
Horan, Hubert. 2019. “Uber’s Path of Destruction”. American Affairs 3 (2).
Horton, John, William R. Kerr, and Christopher Stanton. 2017. “Digital Labor Markets
and Global Talent Flows”, National Bureau of Economic Research (NBER) Working
Paper No. 23398.
Huws, Ursula, Neil Spencer, Dag Syrdal, and Kaire Holts. 2017. Work in the European Gig
Economy: Research Results from the UK, Sweden, Germany, Austria, the Netherlands,
Switzerland and Italy. Brussels and Hatfield, UK: Foundation for European
Progressive Studies (FEPS), UNI-Europa and University of Hertfordshire.
Hyman, Louis. 2018. Temp: How American Work, American Business, and the American
Dream Became Temporary. New York: Viking.
ILO (International Labour Organization). 2006. Labour Inspection. Report of the
Committee of Experts on the Application of Conventions and Recommendations.
ILC.95/III(1B).
———. 2009. ILO Standards on Occupational Safety and Health: Promoting a Safe
and Healthy Working Environment. Report of the Committee of Experts on the
Application of Conventions and Recommendations. ILC 98/III(1B).
———. 2012. Giving Globalization a Human Face: General Survey on the Fundamental
Conventions concerning Rights at Work in Light of the ILO Declaration on Social Justice
for a Fair Globalization, 2008. Report of the Committee of Experts on the Application
of Conventions and Recommendations. ILC 101/III(1B).
———. 2016. Non-Standard Employment around the World: Understanding Challenges,
Shaping Prospects.
———. 2017a. World Social Protection Report 2017–19: Universal Social Protection to
Achieve the Sustainable Development Goals.
———. 2017b. Working Together to Promote a Safe and Healthy Working Environment:
General Survey on the Occupational Safety and Health Instruments concerning
the Promotional Framework, Construction, Mines and Agriculture. Report of the
Committee of Experts on the Application of Conventions and Recommendations.
ILC.106/III(1B) .
———. 2017c. Handbook on Assessment of Labour Provisions in Trade and Investment
Arrangements.
———. 2018a. Care Work and Care Jobs for the Future of Decent Work.
———. 2018b. Freedom of Association: Compilation of Decisions of the Committee on

Freedom of Association, 6th ed.
———. 2018c. Ensuring Decent Working Time for the Future: General Survey concerning
Working-time Instruments. Report of the Committee of Experts on the Application
of Conventions and Recommendations. ILC.107/III(B).
———. 2019a. ILO Centenary Declaration for the Future of Work. International Labour
Conference. 108th Session.
———. 2019b. Universal Social Protection for Human Dignity, Social Justice and
Sustainable Development: General Survey concerning the Social Protection Floors
Recommendation, 2012 (No. 202). Report of the Committee of Experts on the
Application of Conventions and Recommendations. ILC.108/III/B.
———. 2019c. Labour Provisions in G7 Trade Agreements: A Comparative Perspective.
———. 2020a. “Unemployment Protection in the COVID-19 Crisis: Country Responses
and Policy Considerations.”
———. 2020b. “Sickness Benefits during Sick Leave and Quarantine: Country Responses
and Policy Considerations in the Context of COVID-19.”
———. 2020c. “Appeal: Venezuelan Refugees and Migrants in Latin America and the
Caribbean”, 11 August.
———. 2020d. World Employment and Social Outlook: Trends 2020.
———. 2020e. Global Employment Trends for Youth 2020: Technology and the future of jobs.
———. 2020f. “Social Protection Monitor: Social Protection Responses to the COVID-19
Crisis around the World”.
———. 2020g. Synthesis Report: Skills Shortages and Labour Migration in the Field of
Information and Communication Technology in Canada, China, Germany, India,
Indonesia, Singapore and Thailand.
———. 2020h. Promoting Employment and Decent Work in a Changing Landscape.
Report of the Committee of Experts on the Application of Conventions and
Recommendations. ILC109/III(B).
———. 2020i. “Achieving Decent Work in Global Supply Chains”. TMDWSC/2020.
———. 2020j. “The COVID-19 Response: Getting Gender Equality Right for a Better
Future for Women at Work”. Policy Brief.
———. 2021. Working from home: From invisibility to decent work.
ILO, EU, and OECD. Forthcoming. Technical Expert Group Draft Handbook on the
Measurement of Platform Employment.
Ilsøe, Anna. 2020. “The Hilfr Agreement: Negotiating the Platform Economy in
Denmark”, FAOS Research Paper No. 176. Employment Relations Research Centre,
Department of Sociology University of Copenhagen.
Ilsøe, Anna, and Louise Weber Madsen. 2017. Digitalisering af arbejdsmarkedet:
Danskernes erfaring med digital automatisering og digitale platforme. Copenhagen:
FAOS, University of Copenhagen.
India, Ministry of Electronics & Information Technology. 2020. “Report by the
Committee of Experts on Non-Personal Data Governance Framework”.
India, National Sample Survey Office (NSSO). n.d. Periodic Labor Force Survey, 2017–2018.
References 271
Irani, Lilly. 2015. “Difference and Dependence among Digital Workers: The Case of
Amazon Mechanical Turk”. South Atlantic Quarterly 114 (1): 225–234.
ISSA (International Social Security Association). 2020. “Social Security Measures for
the Self-Employed during the COVID-19 Crisis”, 9 April.
Jarrahi, Mohammad Hossein, Will Sutherland, Sarah Beth Nelson, and Steve Sawyer.
2019. “Platformic Management, Boundary Resources for Gig Work, and Worker
Autonomy”. Computer Supported Cooperative Work (CSCW) 29: 153–189.
Jayne, Thomas, Felix Kwame Yeboah, and Carla Henry. 2017. “The Future of Work in
African Agriculture: Trends and Drivers of Change”, ILO Research Department
Working Paper No. 25.
Jesnes, Kristin, and Sigurd M. Nordli Oppegaard (eds). 2020. Platform work in the Nordic
models: Issues, Cases and Responses. Copenhagen: Nordic Council of Ministers,
Nordic Council of Ministers Secretariat.
Johnson, Nicholas L. 2020. “Financial Platforms Are Disrupting More Than Just
Lending (Infographic)”. Applico (blog). https://www.applicoinc.com/blog/
infographic-platforms-disrupting-finance/.
Johnson, Khari. 2020. “CB Insights: AI Startup Funding Hit New High of $26.6 Billion
in 2019”. VentureBeat, 22 January.
Johnston, Hannah, and Chris Land-Kazlauskas. 2019. “Organizing On-Demand:
Representation, Voice, and Collective Bargaining in the Gig Economy”, ILO
Conditions of Work and Employment Series No. 94.
Johnstone, Richard, Shae McCrystal, Igor Nossar, Michael Quinlan, Michael Rawling,
and Joellen Riley. 2012. Beyond Employment: The Legal Regulation of Work
Relationships. Sydney: The Federation Press.
Johnstone, Richard, and Andrew Stewart. 2015. “Swimming against the Tide: Australian
Labor Regulation and the Fissured Workplace”. Comparative Labor Law & Policy
Journal 37 (1): 55–90.
Josephs, Jonathan. 2020. “Just Eat to Stop Using Gig Economy Workers”. BBC News,
14 August.
Joyce, Simon, Denis Neumann, Vera Trappmann, and Charles Umney. 2020. “A Global
Struggle: Worker Protest in the Platform Economy”, European Trade Union Institute
(ETUI) Policy Brief No. 2.
Kalleberg, Arne L., and Michael Dunn. 2016. “Good Jobs, Bad Jobs in the Gig Economy”.
Perspectives on Work 20 (2): 10–14.
Kalra, Aditya. 2020. “Amazon Faces New Antitrust Challenge from Indian Online
Sellers: Legal Documents”. Reuters, 26 August.
Kapczynski, Amy. 2020. “The Law of Informational Capitalism”. Yale Law Journal
129 (5): 1460–1515.
Kässi, Otto, and Vili Lehdonvirta. 2018. “Online Labour Index: Measuring the Online
Gig Economy for Policy and Research”. Technological Forecasting and Social Change
137 (December): 241–248.
———. 2019. “Do Digital Skill Certificates Help New Workers Enter the Market? Evidence
from an Online Labour Platform”, Cesifo Working Paper No. 7810.
Katz, Lawrence F., and Alan B. Krueger. 2016. The Rise and Nature of Alternative Work
Arrangements in the United States, 1995–2015. National Bureau of Economic Research
(NBER) Working Paper No. 22667.
Kenney, Martin, and John Zysman. 2016. “The Rise of the Platform Economy”. Issues
in Science and Technology 32(3) (Spring).
———. 2018a. “Work and Value Creation in the Platform Economy”, Social Science
Research Network (SSRN) Scholarly Paper ID 3253673.
———. 2018b. “Entrepreneurial Finance in the Era of Intelligent Tools and Digital
Platforms: Implications and Consequences for Work”, Berkeley Roundtable on
the International Economy (BRIE) Working Paper 2018-8.
———. 2019. “Unicorns, Cheshire Cats, and the New Dilemmas of Entrepreneurial
Finance”. Venture Capital 21 (1): 35–50.
Kessler, Friedrich. 1943. “Contracts of Adhesion: Some Thoughts about Freedom of
Contract”. Columbia Law Review 43 (5): 629.
Kilhoffer, Zachary, Willem Pieter De Groen, Karolien Lenaerts, Ine Smits, Harald
Hauben, Willem Waeyaert, Elisa Giacumacatos, Jean-Philippe Lhernould, and
Sophie Robin-Olivier. 2020. “Study to Gather Evidence on the Working Conditions
of Platform Workers”, VT/2018/032. Brussels: European Commission.
Kim, Eugene K. 2020. “Labor’s Antitrust Problem: A Case for Worker Welfare”. Yale
Law Journal 130.
King-Dejardin, Amelita. 2019. The Social Construction of Migrant Care Work: At the
Intersection of Care, Migration and Gender. Geneva: ILO.
———. Forthcoming. “Homeworking: Bad Job? Good Job”, ILO Working Paper.
Kingsley, Sara Constance, Mary L. Gray, and Siddharth Suri. 2015. “Accounting for
Market Frictions and Power Asymmetries in Online Labor Markets: Market Friction
and Power in Online Labor Markets”. Policy & Internet 7 (4): 383–400.
KPMG. 2018. “Unlocking the Value of the Platform Economy: Mastering the Good,
the Bad and the Ugly”. Amsterdam: Dutch Transformation Forum.
Kuek, Siou Chek, Cecilia Paradi-Guilford, Toks Fayomi, Saori Imaizumi, Panos
Ipeirotis, Patricia Pina, and Manpreet Singh. 2015. The Global Opportunity in Online
Outsourcing. Washington, DC: World Bank.
Kunthara, Sophia. 2019. “VC Dollars For China Take A Dip In 2019”. Crunchbase News,
5 December.
Lakhani, Karim R., Kevin J. Boudreau, Po-Ru Loh, Lars Backstrom, Carliss Baldwin, Eric
Lonstein, Mike Lydon, Alan MacCormack, Ramy A. Arnaout, and Eva C. Guinan.
2013. “Prize-Based Contests Can Provide Solutions to Computational Biology
Problems”. Nature Biotechnology 31 (2): 108–111.
Lakhani, Karim R., David A. Garvin, and Eric Lonstein. 2012. “TopCoder (A): Developing
Software through Crowdsourcing”, Harvard Business School Case Study 610-032.
La Salle, Dominique, and Greta Cartoceti. 2019. “Social Security for the Digital Age:
Addressing the New Challenges and Opportunities for Social Security Systems”.
Geneva: ISSA.
Lee, K. Wilson. 2020. “Ghost kitchens: Reasons to Adopt this Type of Food Delivery
Model”. Forbes, 11 December.
References 273
Lee, Min Kyung, Daniel Kusbit, Evan Metsky, and Laura Dabbish. 2015. “Working with
Machines: The Impact of Algorithmic and Data-Driven Management on Human
Workers”. CHI 2015, 1603–1612.
Lehdonvirta, V., H. Barnard, M. Graham, and I. Hjorth. 2014. “Online Labour Markets:
Levelling the Playing Field for International Service Markets?” In Crowdsourcing
for Politics and Policy. Oxford: Oxford Internet Institute.
Lerner, Josh, and Jean Tirole. 2005. “The Economics of Technology Sharing: Open
Source and Beyond”. The Journal of Economic Perspectives 19 (2): 99–120.
Levy, Ari. 2019. “Alphabet’s Investment in Uber Has Multiplied by 20-Fold since 2013”.
CNBC, 11 April.
Liang, Chen, Yili Hong, Bin Gu, and Jing Peng. 2018. “Gender Wage Gap in Online
Gig Economy and Gender Differences in Job Preferences”, NET Institute Working
Paper No. 18-03.
Lianos, Ioannis, Nicola Countouris, and Valerio De Stefano. 2019. “Re-Thinking the
Competition Law/Labour Law Interaction: Promoting a Fairer Labour Market”.
European Labour Law Journal 10 (3): 291–333.
Liu, Weihua, Xiaoyu Yan, Wanying Wei, and Dong Xie. 2019. “Pricing Decisions for
Service Platform with Provider’s Threshold Participating Quantity, Value-Added
Service and Matching Ability”. Transportation Research Part E: Logistics and
Transportation Review 122: 410–432.
Lobel, Orly. 2016. “The Law of the Platform”. Minnesota Law Review 101 (1): 87–166.
López Mourelo, Elva. 2020. Work on delivery platforms in Argentina: Analysis and policy
recommendations, Buenos Aires: ILO Country Office for Argentina.
Malin, Martin H. 2018. “Protecting Platform Workers in the Gig Economy: Look to the
FTC”. Indiana Law Review 51 (2): 377–412.
Mao, Ke, Licia Capra, Mark Harman, and Yue Jia. 2017. “A Survey of the Use of
Crowdsourcing in Software Engineering”. Journal of Systems and Software 126
(April): 57–84.
Markov, Kroum, and María Marta Travieso. 2019. “Technologies, the Future of Work
and the ILO’s Normative Approach”. In ILO100 Law for Social Justice, edited by
George P. Politakis, Tomi Kohiyama and Thomas Lieby, 953–980. Geneva: ILO.
Marshall, Aarian. 2020. “Covid-19 Opens the Door for Gig Workers to Win Sick Pay”.
Wired, 7 June.
Martens, Bertin. 2016. “An Economic Policy Perspective on Online Platforms”. JRC
Technical Reports, Institute for Prospective Technological Studies Digital Economy
Working Paper 2016/05. European Commission.
McCrystal, Shae. 2014. “Collective Bargaining Beyond the Boundaries of Employment:
A Comparative Analysis”. Melbourne University Law Review 37 (3): 662–698.
McGee, Luke. 2020. “Self-Isolate or Get Paid? That’s the Choice for Gig Workers in a
Virus Outbreak, and It’s a Big Problem for the Rest of Us”. CNN, 8 March.
Mehta, Neel, Parth Detroja, and Aditya Agashe. 2018. “Amazon Changes Prices on
Its Products about Every 10 Minutes: Here’s How and Why They Do It”. Business
Insider, 10 August.
Mendonca, Jochelle, and Nilesh Christopher. 2018. “India’s Graduates Line up to Rid
Facebook of Inappropriate Content”. The Economic Times, 5 September.
MGI (McKinsey Global Institute). 2016. “Independent Work: Choice, Necessity, and
the Gig Economy”. McKinsey Global Institute.
Miric, Milan, Kevin J. Boudreau, and Lars Bo Jeppesen. 2019. “Protecting Their
Digital Assets: The Use of Formal and Informal Appropriability Strategies by
App Developers”. Research Policy 48 (8).
MIT (Massachusetts Institute of Technology). 2020. The Work of the Future: Building
Better Jobs in an Age of Intelligent Machines. Cambridge: MIT.
Moazed, Alex. 2019. “Platform Businesses Account for 20% of S&P 500
Returns”. Applico (blog), 26 August. https://www.applicoinc.com/blog/
platform-businesses-account-for-20-of-sp-500-returns/.
Moazed, Alex, and Nicholas L. Johnson. 2016. Modern Monopolies: What It Takes to
Dominate the 21st Century Economy. New York: St. Martin’s Press.
Moore, Phoebe V., and Simon Joyce. 2020. “Black box or hidden abode? The expansion
and exposure of platform work managerialism”. Review of International Political
Economy, 27 (4): 926–948.
Morgenstern, Felice. 1985. “The Importance, in Practice, of Conflicts of Labour Law”.
International Labour Review 124 (2): 119–132.
Morgenstern, Felice, and Blaise Knapp. 1978. “Multinational Enterprises and the
Extraterritorial Application of Labour Law”. International and Comparative Law
Quarterly 27 (4): 769–793.
Mulcahy, Diane. 2016. “Who Wins in the Gig Economy, and Who Loses”. Harvard
Business Review, 27 October.
Mundlak, Guy. 2009. “De-Territorializing Labor Law”. Law & Ethics of Human Rights
3 (2): 189–222.
———. 2020. Organizing Matters: Two Logics of Trade Union Representation. Cheltenham,
UK: Edward Elgar.
Murali, Anand. 2019. “How India’s Data Labellers Are Powering the Global AI Race”.
Factor Daily, 21 March.
Naidu, Suresh, Eric A. Posner, and E. Glen Weyl. 2018. “Antitrust Remedies for Labor
Market Power”. Harvard Law Review 132 (2): 536–601.
Namgoong, June. 2019. “Two Sides of One Coin: The US-Guatemala Arbitration and
the Dual Structure of Labour Provisions in the CPTPP”. International Journal of
Comparative Labour Law and Industrial Relations 35 (4): 483–510.
Narula, Prayag, Philipp Gutheim, David Rolnitzky, Anand Kulkarni, and Bjoern
Hartmann. 2011. “MobileWorks: A Mobile Crowdsourcing Platform for Workers
at the Bottom of the Pyramid”, Human Computation Conference, Papers from the
2011 AAAI (Association for the Advancement of Artificial Intelligence) Workshop,
San Francisco, California, 8 August.
Nedelkoska, Ljubica, and Glenda Quintini. 2018. “Automation, Skills Use and Training”,
OECD Social, Employment and Migration Working Papers No. 202.
References 275
Neeraj, R.S. 2017. “Trade Rules on Source-Code Deepening the Digital Inequities by
Locking up the Software Fortress”, Centre for WTO Studies, Indian Institute of
Foreign Trade Working Paper CWS/WP/200/37.
Nguyen, Quynh Anh, and Nuno Cunha. 2019. Extension of Social Security to Workers
in Informal Employment in the ASEAN Region. Bangkok: ILO.
Nitzberg, Mark, Timo Seppälä, and John Zysman. 2019. “The Hype Has Eclipsed the
Limitations of Third-Wave Artificial Intelligence”. ETLA Economic Research, 12 May.
Nyawira, Lindsay. 2019. “Digital Taxi Drivers Go on Strike, Demand Higher Rates”.
The Star, 15 July.
Ochieng, Abiud. 2019. “Online App Taxi Drivers in Kenya to Go on Strike over Pricing
Deal”. The East African, 15 July.
OECD (Organisation for Economic Co-operation and Development). 2014. Addressing
the Tax Challenges of the Digital Economy.
———. 2018. “The Future of Social Protection: What Works for Non-Standard Workers?”,
Policy Brief.
———. 2019a. An Introduction to Online Platforms and Their Role in the Digital
Transformation.
———. 2019b. “Implications of E-Commerce for Competition Policy: Background Note”.
———. 2019c. OECD Employment Outlook 2019: The Future of Work.
———. 2020a. Digital Economy Outlook 2020.
———. 2020b. A Roadmap toward a Common Framework for Measuring the Digital
Economy. Report for the G20 Digital Economy Task Force, Saudi Arabia, 2020.
———. 2020c. “Supporting People and Companies to Deal with the COVID-19 Virus:
Options for an Immediate Employment and Social-Policy Response”, ELS Policy
Brief on the Policy Response to the COVID-19 Crisis.
Ogembo, Daisy, and Vili Lehdonvirta. 2020. “Taxing Earnings from the Platform
Economy: An EU Digital Single Window for Income Data?”. British Tax Review, 82–101.
O’Neill, Jacki. 2018. “From Crowdwork to Ola Auto: Can Platform Economies Improve
Livelihoods in Emerging Markets?”. In The Future of Work in the Global South,
edited by H. Galperin and A. Alarcon, 28–31. Ottawa: International Development
Research Centre.
Oshagbemi, Titus. 1999. “Overall Job Satisfaction: How Good Are Single versus
Multiple‐item Measures?”. Journal of Managerial Psychology 14 (5): 388–403.
Osman, Osman Mohamed. 2019. “There Are More Female Cab Drivers on Nairobi
Roads than Ever before and More Are Coming”. Quartz Africa, 26 January.
Ottaviano, Juan Manuel, Juan O’Farrell, and Matías Maito. 2019. “Organización sindical
de trabajadores de plataformas digitales y criterios para el diseño de políticas
públicas”, Análisis No. 40. Friedrich Ebert Stiftung Argentina.
Online Labour Observatory (iLabour Project). n.d. Oxford Internet Institute and
International Labour Organization. Database.
Parayil, Govindan (ed.) 2005. Political Economy and Information Capitalism in
India. Technology, Globalisation and Development Series. London: Palgrave
Macmillan UK.
Parker, Christine, and John Braithwaite. 2003. “Regulation”. In The Oxford Handbook
of Legal Studies, edited by Mark Tushnet and Peter Cane, 119–145. Oxford: Oxford
University Press.
Parker, Geoffrey G., Marshall W. Van Alstyne, and Sangeet Paul Choudary. 2016.
Platform Revolution: How Networked Markets Are Transforming the Economy and
How to Make Them Work for You. New York, NY: W. W. Norton & Company.
Parthasarathy, Balaji. 2010. “The Computer Software Industry as a Vehicle of Late
Industrialization: Lessons from the Indian Case”. Journal of the Asia Pacific Economy
15 (3): 247–270.
Pasquale, Frank. 2015. The Black Box Society. Cambridge, MA; London: Harvard
University Press.
Peng, Qingwen, and Dayou Cao. 2016. “是劳动关系还是劳务关系？以滴滴出行为例解析
中国情境下互联网约组车平台 [Labour relationship or Labour service relationship?
An analysis of platform-based car booking using the example of Didi]”. 中国人力
资源开发 [Human Resources Development of China]: 93–97.
Pesole, Annarosa, Maria Cesira Urzì Brancati, Enrique Fernández-Macías, Federico
Biagi, and Ignacio González Vázquez. 2018. Platform Workers in Europe: Evidence
from the COLLEEM Survey. Luxembourg: Publications Office of the European Union.
Pesole, Annarosa, and Uma Rani. Forthcoming. “How many online workers?: Estimates
based on selected online web-based and location-based platforms”, European
Commission Working Paper.
Pichault, François, and Tui McKeown. 2019. “Autonomy at Work in the Gig Economy:
Analysing Work Status, Work Content and Working Conditions of Independent
Professionals”. New Technology, Work and Employment 34 (1): 59–72.
Pinsof, Jennifer. 2016. “A New Take on an Old Problem: Employee Misclassification in
the Modern Gig-Economy Notes”. Michigan Telecommunications and Technology
Law Review 22 (2): 341–374.
PitchBook. 2020. “US Venture Capital Investment Surpasses $130 Billion in 2019 for
Second Consecutive Year”. PRNewswire, 14 January.
Pofeldt, Elaine. 2016. “Upwork’s New Pricing Model Sparks Outcry”. Forbes, 7 May.
Prabhat, Shantanu, Sneha Nanavati, and Nimmi Rangaswamy. 2019. “India’s
‘Uberwallah’: Profiling Uber Drivers in the Gig Economy”. In Proceedings of the
Tenth International Conference on Information and Communication Technologies
and Development, 1–5. ICTD ’19. Ahmedabad, India: Association for Computing
Machinery.
Prassl, Jeremias. 2015. The Concept of the Employer. New York: Oxford University Press.
———. 2018. Humans as a Service: The Promise and Perils of Work in the Gig Economy.
New York: Oxford University Press.
Prassl, Jeremias, and Martin Risak. 2016. “Uber, Taskrabbit, and Co.: Platforms as
Employers – Rethinking the Legal Analysis of Crowdwork”. Comparative Labor
Law and Policy Journal 37 (3): 619–651.
Rahman, K. Sabeel, and Kathleen Thelen. 2019. “The Rise of the Platform Business
Model and the Transformation of Twenty-First Century Capitalism”. Politics &
Society 47 (2): 177–204.
References 277
Ram, Aliya. 2019. “Europe’s AI Start-ups Often Do Not Use AI, Study Finds”. Financial
Times, 5 March.
Rani, Uma, and Marianne Furrer. Forthcoming. “Digital Labour Platforms and New
Forms of Flexible Work in Developing Countries: Algorithmic Management of
Work and Workers”. Competition & Change.
Rani, Uma, and Parminder Jeet Singh. 2019. “Digital Platforms, Data, and Development:
Implications for Workers in Developing Economies”. Comparative Labour Law and
Policy Journal 41 (1): 263–287.
Risak, Martin, and Doris Lutz. 2020. Arbeit in der Gig-Economy: Rechtsfragen neuer
Arbeitsformen in Crowd und Cloud. Vienna: ÖGB Verlag.
Rochet, Jean-Charles, and Jean Tirole. 2003. “Platform Competition in Two-Sided
Markets.” Journal of the European Economic Association 1 (4): 990–1029.
Rodríguez Fernández, Maria Luz. 2017. “Plataformas, Microworkers y Otros Retos del
Trabajo en La Era Digital”. In El Futuro del Trabajo que Queremos, edited by Laura
Mora Cabello de Alba and Maria Luz Rodríguez Fernández, 95–113. Albacete:
Bomarzo.
———. 2018. “Sindicalismo y Negociación Colectiva 4.0”. Temas Laborales 144: 27–41.
———. 2019. “Calificación jurídica de la relación que une a los prestadores de servi-
cios con las plataformas digitales”. In Plataformas digitales y mercado de trabajo,
edited by Maria Luz Rodríguez Fernández, 57–89. Madrid: Ministerio de Trabajo,
Migraciones y Seguridad Social.
Rogers, Brishen. 2018. “Fissuring, Data-Driven Governance, and Platform Economy
Labor Standards”. In The Cambridge Handbook of Law of the Sharing Economy,
edited by Nestor M. Davidson, Michèle Finck and John J. Infranca, 303–315. Online:
Cambridge University Press.
Rolandsson, Bertil (ed.). 2020. Digital Transformations of Traditional Work in the Nordic
countries. Report from The Future of Work: Opportunities and Challenges for
the NORDIC models. Copenhagen: Nordic Council of Ministers, Nordic Council
of Ministers Secretariat.
Rose, Michael. 2003. “Good Deal, Bad Deal? Job Satisfaction in Occupations”. Work,
Employment and Society 17 (3): 503–530.
Rosenblat, Alex, and Luke Stark. 2016. “Algorithmic Labor and Information
Asymmetries: A Case Study of Uber’s Drivers”. International Journal of Communication
10: 3758–3784.
Rowley, Jason D. 2020. “The Q4/EOY 2019 Global VC Report: A Strong End to a Good,
but not Fantastic, Year”. Crunchbase News, 8 January.
Roy, Shourya, Chithralekha Balamurugan, and Sujit Gujar. 2013. “Sustainable
Employment in India by Crowdsourcing Enterprise Tasks”. In ACM DEV ’13:
Proceedings of the 3rd ACM Symposium on Computing for Development, 1–2. New
York: Association for Computing Machinery (ACM).
Sadowski, Jathan. 2016. “Companies Are Making Money from Our Personal Data – But
at What Cost?” The Guardian, 31 August.
Sakkal, Paul, and Sumeyya Ilanbey. 2020. “Masks Made Mandatory, Aged Care Visits
Limited as 363 Cases Announced”. The Age, 19 July.
Sanders, Donald E., and Patricia Pattison. 2016. “Worker Characterization in a Gig
Economy Viewed through an Uber Centric Lens”. Southern Law Journal 26 (2):
297–300.
Santos, Alvaro. 2018. “The Lessons of TPP and the Future of Labor Chapters in Trade
Agreements”, Institute for International Law and Justice (IILJ) Working Paper
2018/3, MegaReg Series.
Santos, Filipe M., and Kathleen M. Eisenhardt. 2005. “Organizational Boundaries and
Theories of Organization”. Organization Science 16 (5): 491–508.
Scheiber, Noam. 2018. “Gig Economy Business Model Dealt a Blow in California
Ruling”. The New York Times, 30 April.
Schiek, Dagmar, and Andrea Gideon. 2018. “Outsmarting the Gig-Economy through
Collective Bargaining: EU Competition Law as a Barrier to Smart Cities?”.
International Review of Law, Computers & Technology 32 (2–3): 275–294.
Schleifer, Theodore. 2019. “SoftBank, the Most Powerful – and Controversial – Tech
Investor in Silicon Valley, Explained”. Vox, 10 May.
Schmidt, Florian A. 2017. Digital Labour Markets in the Platform Economy: Mapping the
Political Challenges of Crowd Work and Gig Work. Bonn: Friedrich-Ebert-Stiftung.
———. 2019. “Crowdsourced Production of AI Training Data: How Human Workers
Teach Self-Driving Cars how to See”, Hans-Böckler-Stiftung Working Paper No. 155.
Schorpf, Philip, Jorg Flecker, and Annika Schonauer. 2017. “On Call for One’s Online
Reputation: Control and Time in Creative Crowd Work”. In The New Digital Workplace:
How New Technologies Revolutionise Work, edited by Kendra Briken, Shiona Chillas,
Martin Krzywdzinski and Abigail Marks, 89–111. Palgrave MacMillan.
Schwellnus, Cyrille, Assaf Geva, Mathilde Pak, and Rafael Veiel. 2019. “Gig Economy
Platforms: Boon or Bane?” OECD Economics Department Working Papers No. 1550.
Scott, Colin. 2001. “Analysing Regulatory Space: Fragmented Resources and
Institutional Design”. Public Law (Summer): 283–305.
Sheriff, Mohamed K. 2018. “Big Data Revolution: Is It a Business Disruption?”.
In Emerging Challenges in Business, Optimization, Technology, and Industry, edited
by Lotfi Tadj and Ajay K. Garg, 79–91. Cham, Switzerland: Springer International
Publishing.
Shevchuk, Andrey, and Denis Strebkov. 2017. “Entrepreneurial Potential in the
Digital Freelance Economy: Evidence from the Russian-Language Internet”. In
Entrepreneurship in Transition Economies: Diversity, Trends, and Perspectives, edited
by Arnis Sauka and Alexander Chepurenko, 401–419. Societies and Political Orders
in Transition Series. Cham, Switzerland: Springer International Publishing.
———. Forthcoming. “Freelance Platform Work in Russia, 2009–19”. Geneva: ILO.
Silberman, M. Six. 2015. “Human-Centered Computing and the Future of Work: Lessons
from Mechanical Turk and Turkopticon, 2008–2015”. University of California, Irvine.
Silberman, M. Six, and Lilly Irani. 2016. “Operating an Employer Reputation System:
Lessons from Turkopticon, 2008–2015”. Comparative Labor Law & Policy Journal
37 (3): 505–541.
Singh, Manish. 2019. “Indian Tech Startups Raised a Record $14.5B in 2019”. TechCrunch,
30 December.
References 279
Singh, Parminder Jeet. 2020. “Economic Rights in a Data-Based Society”. Bonn:

Friedrich-Ebert-Stiftung.
Smith, Sanya Reid. 2017. “Some Preliminary Implications of WTO Source Code
Proposal”, Third World Network Briefings.
Sonnemaker, Tyler. 2020. “Indian Business Owners Are Furious about Amazon’s
$1 Billion Expansion into Their Country and Are Calling Jeff Bezos an ‘Economic
Terrorist’”. Business Insider, 15 January.
Soon, Stella, and Saheli Roy Choudhury. 2019. “Grab CEO Says the Company Can Go
Public ‘Once We’re Profitable’”. CNBC, 14 November.
S&P Global Market Intelligence. 2019. “Flipkart Is No. 1 in India but Faces Formidable
Foe in Amazon, Say Experts”, 10 October.
Stanford, Jim. 2017. “The Resurgence of Gig Work: Historical and Theoretical
Perspectives”. The Economic and Labour Relations Review 28 (3): 382–401.
Startup Genome. 2020. “The Global Startup Ecosystem Report 2020”.
Statistics Finland (SF). 2017. Labour force survey: Platform jobs 2017. Helsinki: Statistics
Finland.
Stephany, Fabian, Michael Dunn, Steven Sawyer, and Vili Lehdonvirta. 2020. “Distancing
Bonus or Downscaling Loss: The Changing Livelihood of US Online Workers in
Times of COVID-19”. Journal of Economic and Social Geography 111 (3): 561–573.
Stewart, Andrew, and Shae McCrystal. 2019. “Labour Regulation and the Great Divide:
Does the Gig Economy Require a New Category of Worker?”. Australian Journal of
Labour Law 32 (1): 4–22.
Stewart, Andrew, and Jim Stanford. 2017. “Regulating Work in the Gig Economy:
What Are the Options?”. The Economic and Labour Relations Review 28 (3): 420–437.
Stone, Katharine. 2004. From Fidgets to Digits: Employment Regulation for the Changing
Workplace. New York: Cambridge University Press.
Straits Times, The. 2015. “Indonesia’s Muslim Women Hail Female-Only Motorbike
Taxis”, 30 December.
Stucke, Maurice E. 2018. “Should We Be Concerned about Data-Opolies?” Georgetown
Law Technology Review 2 (2): 275–324.
Sun, Sophie, Julie Yujie Chen, and Uma Rani. Forthcoming. “From Flexible Labor
to ‘Sticky Labor’: A Tracking Survey of Workers in the Food-Delivery Platform
Economy of China”. Geneva: ILO.
Supiot, Alain. 2001. Beyond Employment. Changes in Work and the Future of Labour Law
in Europe. Oxford: Oxford University Press.
Surie, Aditi. 2018. “Are Ola and Uber Drivers Entrepreneurs or Exploited workers”.
Economic and Political Weekly 53 (24).
Surie, Aditi, and Jyothi Koduganti. 2016. “The Emerging Nature of Work in Platform
Economy Companies in Bengaluru, India: The Case of Uber and Ola Cab Drivers”.
E-Journal of International and Comparative Labour Studies 5 (3).
Surie, Aditi, and Lakshmee V. Sharma. 2019. “Climate Change, Agrarian Distress,
and the Role of Digital Labour Markets: Evidence from Bengaluru, Karnataka”.
Decision 46: 127–138.
Sweden, Government Offices (SOU). 2017. Ett arbeidsliv i förändring – hur påverkas
ansvaret för arbetsmiljön.
SwiggyBytes. 2017. “Swiggy’s Response: A Note from the CEO”. 31 July.
Switzerland FSO (Federal Statistical Office). 2020. “Internet-mediated Platform Work
Is Not Very Common in Switzerland”, 19 May.
Tauchert, Christoph, Peter Buxmann and Jannis Lambinus. 2020. “Crowdsourcing Data
Science: A Qualitative Analysis of Organizations’ Usage of Kaggle Competitions”. In
Proceedings of the 53rd Hawaii International Conference on System Sciences. Manoa:
University of Hawaii.
Teare, Gene, and Sophia Kunthara. 2020. “European Venture Report: VC Dollars Rise
In 2019”. Crunchbase News, 14 January.
Teece, David J. 2017. “Dynamic Capabilities and (Digital) Platform Lifecycles”. In
Entrepreneurship, Innovation, and Platforms, edited by Jeffrey L. Furman, Annabelle
Gawer, Brian S. Silverman and Scott Stern. 211–225. Advances in Strategic
Management Series, Vol. 37. Bingley, UK: Emerald Publishing.
———. 2018a. “Business Models and Dynamic Capabilities”. Long Range Planning 51
(1): 40–49.
———. 2018b. “Profiting from innovation in the digital economy: Enabling technolo-
gies, standards, and licensing models in the wireless world”. Research Policy 47:
1367–1387.
Terwiesch, Christian, and Yi Xu. 2008. “Innovation Contests, Open Innovation, and
Multiagent Problem Solving”. Management Science 54 (9): 1529–1543.
Teubner, Gunther. 1998. “Legal Irritants: Good Faith in British Law or How Unifying
Law Ends up in New Divergences”. The Modern Law Review 61 (1): 11–32.
Teubner, Timm, Marc T.P. Adam, and Florian Hawlitschek. 2019. “Unlocking Online
Reputation”. Business & Information Systems Engineering.
Thakker, Dharmesh, Max Schireson, and Dan Nguyen-Huu. 2017. “Tracking the Explosive
Growth of Open-Source Software”. TechCrunch (blog), 7 April. https://social.tech-
crunch.com/2017/04/07/tracking-the-explosive-growth-of-open-source-software/.
Tham, Joo-Cheong, and Keith D. Ewing. 2016. “Labour Clauses in the TPP and TTIP: A
Comparison without a Difference Special Issue – The Trans-Pacific Partnership”.
Melbourne Journal of International Law 17 (2): 369–403.
Todolí-Signes, Adrián. 2019. “Algorithms, Artificial Intelligence and Automated
Decisions Concerning Workers and the Risks of Discrimination: The Necessary
Collective Governance of Data Protection”. Transfer: European Review of Labour
and Research 25 (4): 465–481.
Toews, Rob. 2019. “AI Will Transform the Field of Law”. Forbes, 19 December.
Tubaro, Paola, and Antonio A. Casilli. 2019. “Micro-Work, Artificial Intelligence and the
Automotive Industry”. Journal of Industrial and Business Economics 46 (3): 333–345.
Tubaro, Paola, Antonio A. Casilli, and Marion Coville. 2020. “The Trainer, the Verifier,
the Imitator: Three Ways in which Human Platform Workers Support Artificial
Intelligence”. Big Data and Society 7 (1).
Uber. 2020a. 2019 Annual Report.
———. 2020b. 2020 Investor presentation, 6 February.
References 281
———. 2020c. “What does the background check look for?” https://help.uber.com/
driving-and-delivering/article/what-does-the-background-check-look-for?no-
deId=ee210269-89bf-4bd9-87f6-43471300ebf2.
UN (United Nations). 2019. Report of the UN Secretary-General’s High-Level Panel on
Digital Cooperation: The Age of Digital Interdependence.
UNCTAD (United Nations Conference on Trade and Development). 2018. Trade and
Development Report 2018: Power, Platforms and the Free Trade Delusion.
———. 2019. Digital Economy Report 2019: Value Creation and Capture – Implications
for Developing Countries.
United Kingdom, Department for Business, Energy and Industrial Strategy. 2018a.
“The Experiences of Individuals in the Gig Economy”.
———. 2018b. “The Characteristics of Those in the Gig Economy”.
United Kingdom, Office for National Statistics. 2021. “Coronavirus (COVID‑19) Related
Deaths by Occupation, England and Wales: Deaths registered between 9 March
and 28 December 2020”.
United States, Bureau of Labor Statistics (BLS). 2018. “Electronically Mediated
Work: New Questions in the Contingent Worker Supplement”. BLS Monthly Labor
Review, September.
United States, Census Bureau. 2017. Current Population Survey, 2017.
Upwork. 2019. Annual Report 2019.
———. 2020. “Upwork Investor Relations: Upwork Reports Third Quarter 2020 Financial
Results”, 4 November.
Urzì Brancati, Maria Cesira, Annarosa Pesole, and Enrique Fernández Macías. 2020.
“New Evidence on Platform Workers in Europe”, EUR 29958 EEN. Luxembourg:
Publications Office of the European Union.
Ustek‑Spilda, Funda, Mark Graham, Alessio Bertolini, Srujana Katta, Fabian Ferrari,
and Kelle Howson. 2020. “From Social Distancing to Social Solidarity: Gig
Economy and the Covid-19”. OECD Development Matters (blog), 27 March. https://
oecd-development-matters.org/2020/03/27/from-social-distancing-to-social-sol-
idarity-gig-economy-and-the-covid-19/.
Vallas, Steven P. 2018. “Platform Capitalism: What’s at Stake for Workers?” New Labor
Forum 28 (1): 48–59.
van Doorn, N. 2017. “Platform Labor: On the Gendered and Racialized Exploitation
of Low‑Income Service Work in the ‘On‑Demand’’ Economy”. Information,
Communication & Society 20(6): 898–914.
Venturini, Jamila, Luiza Louzada, Marilia Maciel, Nicolo Zingales, Konstantinos
Stylianou, and Luca Belli. 2016. Terms of Service and Human Rights: An Analysis of
Online Platform Contracts. Council of Europe and FGV Direito Rio: Editora Revan.
Waas, Bernd, Wilma B. Liebman, Andrew Lyubarsky, and Katsutoshi Kezuka. 2017.
Crowdwork: A Comparative Law Perspective. Frankfurt am Main: Bund‑Verlag.
Webster, Edward. 2020. “The Uberisation of work: the challenge of regulating platform
capitalism. A commentary”. International Review of Applied Economics, 34 (4):
512–521.
WEC (World Employment Confederation). 2020. “A Decent Level Playing Field for
Platform Work: Policy Recommendations for Sustainable Growth of Platform
Work and the Provision of Quality Online Talent Platform Services”.
WeeTracker. 2020. “African Venture Capital and Tech Startups Funding Report 2019”,
4 January.
WEF (World Economic Forum). 2020. Charter of Principles for Good Platform Work.
Geneva.
Weil, David. 2014. The Fissured Workplace: Why Work Became So Bad for So Many and
What Can Be Done to Improve It. Cambridge, MA: Harvard University Press.
World Bank, The. 2020. Beaten and broken? Informality and COVID‑19. Washington, DC:
The World Bank.
Wood, Alex J. 2015. “Networks of Injustice and Worker Mobilisation at Walmart”.
Industrial Relations Journal 46 (4): 259–274.
Wood, Alex J., Mark Graham, Vili Lehdonvirta, and Isis Hjorth. 2019a. “Networked
but Commodified: The (Dis)Embeddedness of Digital Labour in the Gig Economy”.
Sociology 53 (5): 931–950.
———. 2019b. “Good Gig, Bad Gig: Autonomy and Algorithmic Control in the Global
Gig Economy”. Work, Employment and Society 33 (1): 56–75.
Wood, Alex J., Vili Lehdonvirta, and Mark Graham. 2018. “Workers of the Internet
Unite? Online Freelancer Organisation among Remote Gig Economy Workers in Six
Asian and African Countries”. New Technology, Work and Employment 33 (2): 95–112.
Woodcock, Jamie, and Mark Graham. 2020. The Gig Economy: A Critical Introduction.
New York: Polity Press.
Wu, Qingjun, Hao Zhang, Zhen Li, and Kai Liu. 2019. “Labor Control in the Gig Economy:
Evidence from Uber in China”. Journal of Industrial Relations 61 (4): 574–596.
Xie, Zengyi. 2018. “互联网平台用工劳动关系认定 [Labor Relations and Legal Status
of Online Platform Workers]”. Peking University Law Journal 30 (6): 1546–1569.
Yan, Tian. 2018. “平台经济用工与新业态劳动关系 [Employment under the Platform
Economy and New Labor Relations]”. China Law Review 24 (6): 119–120.
Yuan, Li. 2018. “Customer Died. Will That Be a Wake-up Call for China’s Tech Scene?”.
The New York Times, 29 August.
Zhao, Yang, Stephan von Delft, Anna Morgan-Thomas, and Trevor Buck. 2019. “The
Evolution of Platform Business Models: Exploring Competitive Battles in the World
of Platforms”. Long Range Planning 53 (4): 101892.
Zou, Mimi. 2017a. “The Regulatory Challenges of Uberization in China: Classifying
Ride-Hailing Drivers”. International Journal of Comparative Labour Law and Industrial
Relations 33 (2): 269–294.
———. 2017b. “Regulating the Fissured Workplace: The Notion of the Employer in
Chinese Labour Law”. Bulletin of Comparative Labour Relations, 183–203.
Advancing social justice,
promoting decent work
The International Labour

Organization is the United This ILO flagship report explores how the contemporary platform
Nations agency for the world economy is transforming the way work is organized, and analyses
of work. We bring together the impact of digital labour platforms on enterprises, workers and
governments, employers and society as a whole.
workers to drive a human-centred
The report offers a comprehensive picture of the experience
approach to the future of work
of workers and businesses on online web -based and location -
through employment creation,
based platforms, drawing on surveys and interviews with some
rights at work, social protection
12,000 workers and representatives of 85 businesses around
and social dialogue.
the world, in multiple sectors. It also provides insights into the
business model of digital labour platforms, examines regulatory
responses around the globe, and presents a way forward to ensure
that all platform work is decent work.
ilo.org
International Labour Organization

Route des Morillons 4
1211 Geneva 22
Switzerland

Platform A

Uploaded by

Copyright:

Available Formats

Platform A

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Platform A

Uploaded by

Copyright:

Available Formats

ILO Flagship Report

and Social Outlook

The role of digital labour

International Labour Office • Geneva

ISBN 978-92-2-031944-4 (print)

employment / unemployment / labour policy / labour market analysis /

ILO Cataloguing in Publication Data

Produced by the Publications Production Unit (PRODOC) of the ILO.

1 The digital transformation of industry

1.1 The rise of the digital economy 34

1.2 Digital platforms: Pervading and penetrating

1.3 Digital labour platforms: Estimates of the number

1.4 The data-driven economy and the rise of machine-learning algorithms 58

1.5 Financing the rise of digital labour platforms 63

2 The business model and strategies

of digital labour platforms 71

2.1 Types of digital labour platforms 74

2.1.1 Online web-based platforms 74

2.2 Revenue model 78

2.2.1 Freelance and contest-based platforms 79

2.3 Recruitment and matching of workers with clients 90

2.3.1 Work relationships on platforms 90

2.4 Work processes and performance management 95

2.4.1 Work processes and communication 95

2.5 Digital labour platforms’ rules of governance

and workers’ freedom to work 98

using them? 103

3.1 Businesses using online web-based platforms 106

3.1.1 Recruitment 106

3.2 Businesses using location-based platforms 114

3.3 Opportunities from digital platforms for BPO companies

and digital technology start-ups 118

3.3.1 Transformations in BPO companies 118

3.4 Impactofdigitalplatformsontraditionalbusinesses 126

4.1 Basic demographic characteristics of platform workers 136

4.1.1 Age distribution of platform workers 137

4.2.1 Access to a sufficient amount of work 147

4.3 Worker autonomy and control under algorithmic management 177

4.3.1 Autonomy and control over work 177

4.4 Skills acquisition and mismatch 184

4.5 Platform design and discrimination 189

5 Ensuring decent work on digital labour platforms 195

5.1 Regulation by digital labour platforms: Terms of service agreements 198

5.1.1 Platform terms of service agreements 198

5.2 Regulating digital platforms for labour and social protection:

What should be the goals? 202

5.3 Achieving decent work: Regulatory responses to platform work 211

5.3.1 Freedom of association, collective bargaining

and other forms of social dialogue 211

5.3.2 Non-discrimination 216

5.3.3 Forced labour and child labour 218

5.3.4 Occupational safety and health 219

5.3.5 Social security 221

5.3.6 The COVID-19 pandemic and its implications for health

and safety at work and social security 224

5.3.7 Payment systems, fair termination and clear terms

5.3.8 Access to data, privacy and job mobility 227

5.3.9 Grievance and dispute resolution 228

5.3.10 The employment relationship 230

5.3.11 Remuneration and working time 234

3.4 Impactofdigitalplatformsontraditionalbusinesses 126

6.1 Opportunitiesandchallengesondigitallabourplatforms 242

6.2 Emergingregulatoryresponses 245

6.3 Overcomingthechallengestoseizethebenefits 248

6.4 Awayforward 255