Understanding Poverty and Wellbeing: A Note With Implications For Research and Policy
Understanding Poverty and Wellbeing: A Note With Implications For Research and Policy
Understanding Poverty and Wellbeing: A Note With Implications For Research and Policy
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
PREFACE
CONTEXT OF THIS PAPER
In November 2010, the UK Department for International Development invited a small group of academics to discuss and take stock of what we think we know about poverty. This led to conversations about Where are the gaps in our knowledge and learning? and, What is the future direction for research on poverty? DFID Research and Evidence Division convened two roundtable discussions with a group of senior academic researchers in November 2010 and March 2011 to reect on new poverty concepts and indicators, to discuss their implications for policy and practice and, looking forward, to identify gaps that should be addressed through future research and synthesis. This nal poverty and well-being note is an output from a further stage in the process where a drafting group came together to reect on these questions in more detail and drafted this note to outline a position on key learning and future directions for research, which was then circulated back to the wider group for comment and feedback. The purpose of publishing the nal paper now is to gather comment and generate debate over the issues raised across a broader range of voices. It is an opportunity both to test the level of consensus that can be built on dening priority research areas on poverty and wellbeing whilst identifying the remaining points of debate and divergence. The purpose of this debate is to stimulate further thinking and high quality research on poverty, and help inform the decisions and thinking of those working and engaged in poverty research.
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
1. INTRODUCTION
This note is the product of sessions convened by DFID with poverty researchers in late 2010 and early 2011. DFID invited a group of researchers to set out their views on the key innovations in the meaning and measurement of poverty1. The evidence base for the note is the extensive range of poverty research carried out by the institutions and individuals concerned researchers who have been working on these issues in depth and over a long period of time and doing so very much in partnership with research institutes around the world - plus a broader literature published in books and academic journals. The objective of this note is to take stock of poverty research to date, to propose a set of building blocks or lessons learnt in terms of poverty policy analysis and research, and identify challenges and future directions. In doing so the note is intended to contribute to policy debates relating to the assessment of poverty and wellbeing (such as the question of what framework of goals and measures might replace the UN Millennium Development Goals beyond 2015) as well as consideration of future directions for poverty research. The audience for this note and those invited to respond to the contents, includes scholars, bilateral and multi-lateral donors engaged in poverty analysis and research, and national policy communities in developing countries. While this note highlights some limitations in contemporary research on poverty it must be observed that there have been substantial advances over the last ten years. Human development concepts and measures have been institutionalised; analysis has increasingly moved beyond correlations to examine causality; dynamic analyses of changes in individual and household conditions are much more common; new multi-dimensional poverty measures have emerged and approaches to combining quantitative and qualitative methods have advanced. In addition, ethical guidance, data availability and access to analytical software have improved. The problem this note addresses is as follows: Poverty is a political phenomenon power relations determine the distribution of opportunities and benets - and poverty is multi-dimensional. However, the policy discourse evident in the last decade and in the MDGs lags behind our understanding of realities on the ground in signicant ways. There has been considerable progress in 3
understanding poverty as a multi-dimensional phenomenon but with an emphasis on universally agreed and easily quantiable measures highlighting (as the MDGs do) health, education and nutrition. This sometimes gives an unbalanced view which underplays the signicance of socio-cultural difference and more qualitative elements (powerlessness, stigma, discrimination and isolation, for example). The more reductionist approaches also can underplay the signicance of the interconnections among deprivations and how progress on different poverty dimensions is linked. At the country level, poverty analysis and donor strategy often fails to disaggregate between social groups and remains weak in its understanding of poverty dynamics and the signicance of context. This in turn may lead to a lack of innovation in the development and testing of new policy responses. In addition we note that at this point the lived experience of poverty is changing in signicant ways, as is the global political economy of poverty reduction policy. To pick out a few signicant issues: The two countries which contain the largest absolute numbers of the global poor based on last surveys (India and China) have moved from low to middle-income status in the last decade and a half and are increasingly emerging as major global powers and foreign aid donors in themselves. A wider group of 30 countries have also experienced drastic increases in average per capita incomes - enough to move from low to middle-income country status. These trends, combined with the fact that absolute poverty in global aggregate numbers has not fallen signicantly, mean that most of the worlds poor people now live in middle-income countries. Increasingly urbanisation challenges our traditional ways of quantifying the experience of poverty. The changed context demands greater precision in money-metric measures (requiring price adjustments between urban and rural areas, accurate comparative adjustments for all marketed goods and services to take account of the potentially high cost of transport, rent, safe sanitation and energy in urban areas), and the analysis, through effective use of multidimensional methods, of less tangible elements (such as the potential health consequences of poor environmental quality, crime and social tensions, adverse social and psychological consequences of inequality and inequity and the low social status associated with informal livelihoods and settlements).
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
It is increasingly recognized that high and growing levels of inequality are a signicant factor in individual and collective ill-being as the impacts of vertical and horizontal inequality on conict, growth and poverty are researched. The last decade has exposed poor people to increasing volatility and compound risk on multiple levels from the price of food, to increasing incidence of natural disasters due to our changing climate. Looking ahead such shifting vulnerabilities are likely to form the context for poverty reduction over the next 10-20 years as multiple shocks and stressors interact and the effects are compounded.
So, while we recognise that the last decade has seen great progress in some dimensions of poverty reduction (notably in terms of improved human development indicators in most non-conict affected countries, and great reductions in income poverty in East Asia) a better understanding of how poverty should be assessed is important to ensure continued progress in the eradication of absolute poverty and reduction of disparities in the next ten years.
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
2. BUILDING BLOCKS
Seeking consensus on basic questions such as the nature of poverty, well-being, and how they can be measured runs the risk of producing results which are either bland, or somehow misrepresent or caricature different views and positions. Nonetheless mapping the areas and issues where there is a reasonable consensus and noting areas of disagreement is an important exercise at this point, which is the objective of this section of the paper. As public policy makers (and politicians) become increasingly interested in understanding the nature of deprivation and well-being, useful lessons can be drawn from reviewing research. Poverty research and analysis should be multidimensional, dynamic, and relational in character - and pluralistic in its use of methods and its theoretical base. A finding of much recent research is that the understanding of poverty requires an analysis of inequality. Further, consumption inequality is growing in many of the countries where the worlds poor are concentrated. The relationship between poverty and inequality, however, is complex. Increasingly inequality as measured by the traditional quantitative tools (e.g. the gini coefficient) is driven by growing separation of the very wealthiest people from the rest of the society. Therefore reducing, or even eradicating, poverty may in practice have little direct quantitative impact on the measured level of inequality. Inequality matters therefore not because of a direct mechanical linkage but because of its impacts on society and politics. There is extensive evidence that inequality can contribute to conflict and fragility which are huge impediments to sustainable poverty reduction. Horizontal inequality or inequality between social groups is especially associated with outbreaks of violent conflict and presents a major obstacle to eradicating poverty among marginalised peoples. Even beyond this, high levels of inequality may impede effective poverty reduction by undermining the institutional fabric of a competitive economy and a cohesive society, leading to a political economy which is inimical to the formulation and implementation of effective anti-poverty policies and programmes. High levels of inequality tend in particular to undermine the long-run sustainability of economic growth, through negative impacts on human capital, institutional legitimacy and social cohesion. 5
Poverty research and policy analysis to date suggests the following building blocks or lessons learnt: i. Poverty analysis/research should certainly be multi-dimensional and this is now widely accepted. Money-metric measures of poverty (based on assessments of either income or consumption) are useful but do not give us the whole picture because: a) poverty is multi-dimensional and encompasses decits in material aspects of life, but also in a range of less tangible areas for example, voice, human security, isolation, dignity, lack of time, and subjective wellbeing; b) the correlation of straightforward material well-being with levels of cash consumption or income is complex (e.g. high population densities and urban planning may mean that the urban poor have to spend money on rent, transport, and energy at a higher level than rural populations to secure the same level of physical and social well-being) and c) income poverty and deprivations in human development although closely correlated at low levels of average per capita income can have weaker associations at higher per capita incomes. Money metric measures remain useful, but need rening especially to take account of the shifting and nely graded realities produced by the urbanisation of deprivation. They may also need to be complemented by measures which reect the joint distribution of deprivations in other dimensions directly, as well as effective qualitative research. Continued investment in approaches for better capturing the multi-dimensionality of poverty which are appropriate to different policy environments is therefore necessary including combining qualitative and quantitative measures, listening to poor peoples own views on what indicators are important, developing new measures and metrics, and combining indicators. There was some debate within the drafting group on the relative importance of dashboards meaning a number of indicators that are tracked simultaneously but that are not brought together into a single index and multi-dimensional composite measures. Dashboards take policy makers away from an over-focus on one construct (the money-metric poverty line) and enable the different dimensions of poverty to be tracked simultaneously but separately. In contrast, an alternative view
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
is that large dashboards can be confusing and do not make tradeoffs between dimensions explicit. Single multi-dimensional indicators such as those featured in the 2010 Human Development Report can provide politically useful comparative headline analysis, that can be adapted locally, and that show the joint deprivations poor people experience at the same time. However, it is clear that indicators can move in different directions, and tracking only composite indices may mask this. ii. Poverty analysis/research should be dynamic Mobility underlies any static picture of deprivation. The transient poor move in and out of poverty; the chronic poor remain trapped in situations and relationships which produce deprivation and capability losses over long periods. Being a member of a poor group or social category (e.g. an indigenous group) may present a particularly powerful poverty trap, and makes escape from poverty especially problematic. The combination and interaction of multiple risks causes many populations to be trapped in chronic poverty. Longitudinal information and analysis is critically important to understanding poverty dynamics and causality. Whereas this has typically been focused on movements over time in income/consumption measures of living standards, multi-dimensional dynamic analysis is also revealing particularly of the factors which determine the life-chances of the children of poor households. As inter-generational equity becomes an increasingly important global and national political issue, systematically understanding how social mobility between generations occurs will also become an important objective for research, as it has been in OECD countries. Signicant improvements in longitudinal data at national level are needed in order to track social and economic mobility. Such improvements also create new possibilities for evaluating policies, as well as monitoring the effects of crises. Most current nationally representative data sets are a series of cross sections which do not revisit the same households at different points in time. This works well for establishing the incidence of poverty but provides a much weaker base for understanding causality than panel data sets which revisit the same households at different points in time. A 6
better understanding of causality is essential for effective design of anti-poverty interventions. There are a slowly expanding number of national panel household surveys though still far fewer than we need. The rate of expansion could be greatly enhanced. These need to (a) be undertaken over a sufciently long span, and with sufciently long periods between survey dates to observe signicant mobility; (b) include information on different social groups and categories; (c) be partnered with qualitative research which enhances the richness of understanding and addresses issues which are difcult to address through surveys. iii. Poverty analysis/research should incorporate an understanding of relationships. The last decade of research on the causes of chronic poverty has shown that a key factor in keeping certain categories or social groups in a state of chronic poverty (transmitted through the generations) is the operation of power relations which stigmatise the people involved, undermine their condence, and systematically close off options for individual or collective advancement. This state of affairs can be captured by concepts such as social exclusion, social capital, disempowerment, isolation, adverse incorporation, discrimination, horizontal inequalities and socially constructed gender, ethnic, social and spatial differences. These elements capture the relational character of poverty including the engagement of all social groups in sustaining the state of poverty in which certain groups are trapped. A stronger focus on structural inequalities, on systematic obstacles to redistributive processes and on social groups and interactions between them is part of this: we know, for example, how powerful interactions between culturally dened groups are in the generation and resolution of violent conict. Relations between employers and employees, between genders, and generations are also drivers of changes in wellbeing as well as political change. And it is values, norms, incentives and laws and regulations which shape such important political and economic relationships. In an increasingly inter-connected world important relationships stretch across boundaries, so the analysis needs to be multi-scalar as well as multi-disciplinary.
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
iv. Poverty analysis/research should be pluralist in its methods and theoretical approaches. The advantages of a mixed methods approach to poverty reduction using both quantitative and qualitative approaches are well established. However, the integration, sequencing and balance between quantitative and qualitative data collection, data analysis and criteria to determine rigour in mixed methods approaches is still a fruitful area of exploration to improve practice. When mixed method approaches are utilised research can be reliable, representative and gain analytical depth (on understanding causal links and relationships in particular). An approach which takes account of different theoretical and analytic perspectives (drawn from social and political theory as well as economic theory) also enriches the understanding and effectiveness of poverty analysis. An important element of the epistemological mix is to recognise that the views of poor people matter in researching poverty much can be hidden in researchers blind spots: e.g. the importance of seasonality, and the value which poor people attach to dignity and independence. Areas where more progress can be made in improving the range of methods and approaches include the following: involving the poor, policy makers and practitioners from the beginning in setting a research agenda, so that it reects peoples own questions, expertise, insights and priorities, as well as a better focus on policy concerns and tractable policy problems; building on existing experience from urban social movements in facilitating the collection, analysis and use of poverty data by poor people themselves; integrating poor peoples own priorities and experiences into measures of poverty which are used by policy communities; improving the methods and approaches available for rapidly gathering data on poverty in fragile and conict affected environments, and during periods of livelihood stress and shocks. We also need to deploy mixed methods that better integrate micro and macro level data as, at present, so much poverty research focuses on one or the other.
v. Poverty analysis should explore the political underpinnings of effective policy (and policy failures). The analysis of poverty is not a neutral technocratic exercise. Different interests, ideologies and social constructions of reality inuence the ways in which poverty is understood, and the ways in which our knowledge about poverty and its causes gets transferred into policy and practice. Power can act to bias an institution against certain kinds of actions, or against the knowledge of certain kinds of people. An effective policy analysis embraces the ways in which power works within institutions that dene, measure and act to address poverty as well as the ways in which power relationships in the private and social spheres work to create and sustain poverty.
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
information on poor peoples well-being. The ongoing global economic crisis has once again revealed the paucity of high-frequency data on poverty at a global level. Consequently, international institutions trying to track poverty impacts of the initial shock in 2009 relied heavily on simulations which gave a relatively optimistic view about changes in global poverty in response to the crisis. The simulations showed no actual increase in poverty, but an increase when compared to a downward counterfactual trend. Improved availability of high-frequency data particularly in forms which demonstrate the differential impacts on different social groups - is needed so that appropriate policy responses can be made in reaction to crises. Dedicated and intensive experimentation with mobile phone based data gathering and viral surveys is needed to break through the longstanding barriers to rapid poverty measurement, analysis and reduction. c) Contextualised knowledge versus generalised assessment how to get the balance right? The role of context in shaping poverty is now better understood. Different contexts also require different policy approaches to get to the same objective. But the political economy of pro-poor policy at the national level often demands a single, reductionist measure so that decisions can be made about allocating scarce resources which are recognised as fair across different environments. Equally although there is widespread agreement that context is critical, there are remarkably similar ndings from many studies (such as the importance of health shocks in plunging households into poverty), and we know that there seem to be some virtually unstoppable aspects of progress in human development. What is missing is systematic comparison based on sound methods. This could be within as well as between countries. For example, a comparison of poverty dynamics across Indian states has controversially argued that there should not be a single poverty policy in India because conditions are so varied. More systematic comparison will help to determine the scale at which poverty policies are best focused local, regional, national, or global. Given the strong state of knowledge on the global signicance of health shocks in impoverishment, strong global insurance mechanisms may be a relevant response.
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
d) Broader/more complex definitions of human wellbeing versus simplicity. There is a ferment of research on new ideas about wellbeing that seek in particular to capture subjective wellbeing and/or the interaction of material and non-material wellbeing. There is also a genuine area of contestation in the current debate about the value of having a single, simple measure of poverty at the national level (usually the consumption-based poverty line,). Critics point to the severe limitations of this kind of measure in many areas (for example the failure to factor access to safe water into poverty measurement in a way which adequately reflects its importance). But for national purposes a single over-riding metric is easier to feed into decision making. Multidimensional indices that can be decomposed should be explored, as should specific multidimensional analyses for example on the quality of education, or multidimensional health. These could be framed in terms of institutional services and outputs (schools, furniture, books, toilets, teachers), and/or of outcomes (learning achievements, years of schooling). The relevance of subjective and perceptual data (for example on quality of services) also requires careful study. For LDCs and LICs grappling with sustainability indicators (linked to climate) alongside existing wellbeing indicators will be enough of a challenge. The policy communities in these countries may not be able to usefully deal with and apply a more complex set of measures. For some MICs more complex definitions and dashboards may be useful and manageable. e) Analysis of the impacts of policy change. Measurement is obviously important, because it provides the basis for appropriate policy. Yet policy analysis tends to lag behind measurement research. As countries across the world adopt an increasingly large and complex range of policies, there is a need for systematic evaluation of their impact on poverty. This includes macro-policies, meso and micro-interventions. In general, most research has been devoted to micro-interventions and little to meso and macro-levels. In part the focus on micro-interventions may have been driven by the strong wave of interest in randomised control trials and the experimental method more broadly. The need for a control group to test the effectiveness of an intervention means that such methods are far easier to apply to assessing 9
the effectiveness of household level interventions than for system-wide policy changes. Longitudinal data can enable the study of natural experiments where the effect of a system wide change over time can be assessed using before and after comparisons rather than the with and without comparison of the classic RCT. Another way forward would be to undertake a rigorous set of case studies that systematically analyse multidimensional poverty over time for a set of countries and identify the distinctive macro, meso, and micro policy interventions and sequences that have led to sustained multidimensional poverty reduction.
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
Specifying the context for policy responses to poverty better MICs becoming donors yet having large pockets of poverty at home; fragile states and the global redistribution of poverty; competing/complementary discourses and policy priorities (climate change; growth; neomalthusianism); resource scarcity/abundance and its implications; continually growing global interdependence of national economies and the risks and opportunities this brings; the increasing separation of small super-wealthy elites from the broader population. Social differentiation, exclusion and poverty both at the intra-household level (gender and generational) - and at the level of social categories and group identities (e.g. ethnic, cultural, religious, regional, caste, etc.). Pro-poor urbanisation policies including urban policies as well as national infrastructure and relevant regional policies (taking into account different sizes of settlements).
The how.... In order to generate policy relevant poverty diagnostics in parallel to the innovations in growth diagnostics, and opportunities for stronger policy evaluation, there are two sets of issues which wont happen automatically, and where long-term investment needs to be made: Longitudinal quantitative and qualitative research is an important new basis for improved poverty diagnostics. This implies improved and extended availability of panel data sets to enable the quantitative tracking of households over time which has proved invaluable for policy analysis. In combination with this or separately longitudinal ethnographic studies provide a rich source of understanding of the factors which reduce or exacerbate poverty over time. Longitudinal studies enable natural experiments to emerge from data where comparisons can be made both between groups and over time which indicate the importance of particular poverty reduction interventions. Much of the early work which indicated the poverty reducing significance of social protection interventions was of this kind. In-country capacity building: stronger in-country networks including not only research institutes but research users, research analysts,
10
UNDERSTANDING POVERTY AND WELLBEING: A NOTE WITH IMPLICATIONS FOR RESEARCH AND POLICY
statistical ofces and citizens are needed. This means systematically and strategically developing capacity of various sorts, including ability to use different forms of evidence (e.g. administrative data as well as surveys) and work with interested urban and rural civil society. There should be a particular focus on the capacity to do longitudinal mixed methods, policy relevant, research; and an active effort to identify opportunities that information and communications technology provides to leapfrog conventional approaches to data collection, checking and analysis, with potential for gains in cost, speed, frequency and granularity of poverty research.
References
1. The rst two roundtable discussions were convened by DFID Research and Evidence Division. These discussions were not intended to be representative of any particular epistemic community. The participants had conducted policy analysis and/or research for DFID in the last decade. Participants were: Sabina Alkire (University of Oxford/OPHI), Jo Boyden (University of Oxford and Director of Young Lives), John Hammock (OPHI), David Hulme (University of Manchester), Clare Melamed (ODI), Allister McGregor (IDS), Diana Mitlin (IIED), Andrew Norton (ODI), Andrew Shepherd (ODI), Frances Stewart (University of Oxford), Andy Sumner (IDS), Gaston Yalonetzky (University of Oxford/OPHI).The process of compiling this note was led by Andrew Norton and Andy Sumner. Joanna McGowan (DFID) and Tim Conway (DFID) helped facilitate the process. Joanna Macrae, Malcolm Worboys, Paul Wafer, and Zoe Stephenson also participated from DFID. This paper reects the views of the academic group and not DFID.
11