Feature

Local Economy
Regional poverty mapping 0(0) 1–23
! The Author(s) 2015

in Europe – Challenges, Reprints and permissions:

sagepub.co.uk/journalsPermissions.nav
DOI: 10.1177/0269094215601958
advances, benefits and lec.sagepub.com

limitations

Andrew Copus and Patricia C Melo

The James Hutton Institute, Aberdeen, UK

Stefan Kaup
ILS – Research Institute for Regional and Urban Development, Dortmund, Germany

Gergely Tagai
Institute for Regional Studies, MTA KRTK, Budapest, Hungary

Panagiotis Artelaris
Harokopio University, Athens, Greece

Abstract
The ‘at-risk-of-poverty’ rate is the most widely recognised indicator of income poverty. Its prin-
cipal advantage is that it is relatively straightforward to define and (given appropriate data) to
calculate. National at-risk-of-poverty rates play a key role in monitoring EU2020 objectives
relating to combating poverty. Regional patterns of poverty have the potential to deepen our
understanding of processes of impoverishment and differentiation, and how they can be more
effectively addressed by policy. Estimating regional poverty rates, and especially producing a
European map, is a challenging task, given current data resources. This paper begins by placing
the at-risk-of-poverty rate within the wider conceptual context relating to poverty, social exclu-
sion and deprivation. It then provides an account of an exercise to map at-risk-of-poverty rates at
NUTS 3 across 20 European countries. Together with data derived from national registers (where
available) and more direct apportionment methods, coverage of most of Western Europe is
achieved. The patterns revealed are described, and generalisations, which serve as pointers to

Corresponding author:
Andrew Copus, Social, Economic and Geographical
Sciences Group (and seconded half-time to Nordregio,
Stockholm), James Hutton Institute, Craigiebuckler,
Aberdeen AB15 8QH, Scotland, UK.
Email: Andrew.Copus@hutton.ac.uk

Downloaded from lec.sagepub.com by guest on September 22, 2015

2 Local Economy 0(0)

further research on the processes responsible, are derived. The paper concludes with some
reflections on the value of regional at-risk-of-poverty rates in advancing our understanding of
the distribution and causes of poverty, and hence appropriate interventions to ameliorate it.

Keywords
at-risk-of-poverty rate, ESPON, EU2020, poverty

Annual income twenty pounds, annual 2001, 2002). Some households are poorer,
expenditure nineteen pounds nineteen or more deprived than their income would
and six, result happiness. Annual income lead us to expect, and vice versa. For this
twenty pounds, annual expenditure twenty reason indicators of income poverty are
pounds nought and six, result misery. – sometimes described as ‘indirect’ measures,
Mr Mcawber in David Copperfield by
whilst those which measure outcomes in
Charles Dickens
terms of deprivation are termed ‘direct’
(Ringen, 1988).
Introduction So why are indirect indicators of income
poverty, such as the at-risk-of-poverty
This paper is concerned with indicators and (ARoP) rate, or child poverty rates, so com-
mapping of income poverty, their capacity monly used as a basis for policy decisions?
to clarify our understanding of spatial pat- The answer is simply that, at a national
terns and processes of disadvantage, and level, they are relatively easy to estimate
their potential role in policy targeting and from existing data sources, either income
monitoring. registers covering 100% of the population,
It is important, at the outset, to distinguish or, more commonly, national or European
income poverty from several closely related sample surveys, such as the EU Survey of
phenomena with which it may sometimes be Income and Living Conditions (EU-SILC).1
confused. Income poverty is measured solely Poverty is a spatially heterogeneous phe-
in terms of the disposable income available to nomenon, and poverty rates can vary widely
the household or individual. It does not con- over space. The study of geographical
sider non-pecuniary issues which are com- inequalities therefore is a very important
monly associated with the broader concepts dimension of poverty analysis. Regional or
of poverty, deprivation (Townsend, 1987) local policies require a better understanding
and social exclusion (Talbot et al., 2012). of patterns of poverty at an appropriate
Neither is expenditure, or the various differ- scale. Regional differentiations of poverty
ent calls on income which vary with location, risk should be taken into serious consider-
personal circumstances and characteristics, ation in policy making and policy imple-
taken into account. mentation. Poverty might have strong
Analysis has shown that households and local characteristics and this is something
individuals affected by income poverty do that should be carefully examined by
not always coincide with those who experi- policy makers because, sometimes, targeted
ence material deprivation (inability to pur- policies might be more effective than general
chase basic goods and services) or other less interventions. National level indicators are
tangible forms of poverty (Whelan et al., useful to monitor the global trends but

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 3

disaggregated information for lower geo- key regional patterns and processes arising
graphical (administrative) areas is probably from the data. The final section summarises
more useful for policy makers. From this the main conclusions and recommendations
perspective, poverty maps that provide a for European wide research in the topic.
description of the spatial distribution of wel-
fare and poverty within a country at different Policy context for the analysis
spatial levels and can be used to investigate
the relationship between poverty and other The EU has no specific, dedicated,
economic, social and geographic factors, are Community policy to address poverty and
a useful tool for analysis and policy making. social exclusion (for an extensive discussion
As Crow et al. (2009) suggest visualisation about poverty issues in the EU see Betti
matters in the study of inequality and pov- et al., 2012; Fahey, 2007; Marlier and
erty since maps (and graphs) might present Atkinson, 2010; Whelan and Maıtre,
powerful stories about progress, social 2010). Whilst a number of community poli-
change and development. cies, especially Cohesion policy, undoubt-
Where the source is register data it should edly have some impact, poverty and social
be a relatively simple matter to disaggregate exclusion are mainly tackled through inter-
national data and generate ARoP rate maps. ventions organised at the Member State
More commonly, where the data source is a level. Since 2000 these have been ‘orche-
national survey, sampling regimes are unli- strated’ through a procedure known as the
kely to be robust enough at a local or regio- Open Method of Coordination, within the
nal level, and some form of modelling or structures provided first by the (2000–10)
apportionment will be required. This paper Lisbon Objectives (CEC, 2004), and more
provides an account of an attempt to com- recently (2010–20) by EU2020 (CEC,
pile a map of ARoP rates for more than 1200 2010). At the Lisbon meeting in 2000, the
NUTS 3 regions2 across 20 European European Council set the strategic goal of
Countries. We believe this to be the first suc- ‘greater social cohesion’ and committed to
cessful attempt to provide a European-wide take steps ‘to make a decisive impact on the
map of income poverty at the NUTS 3 eradication of poverty’. This strategy put
region level.3 Our work hopefully provides poverty and social exclusion at the heart
better support to the design of regional pov- of EU social policy and led to the adoption
erty alleviation strategies than previous, in 2001 of the Laeken social indicators
more aggregate, maps for NUTS 2 or including, among others, poverty indica-
NUTS 1 regions (CEC, 2005; Verma et al., tors. Tackling poverty is also one of the
2006), which can hide considerable local objectives of the Europe 2020 Strategy.
variation in levels of income poverty. The key poverty and social exclusion
The paper has the following structure: the target in the context of EU2020 is to lift
next section provides the policy context for 20 million people out of poverty and
the analysis undertaken, while ‘The ARoP social exclusion by the year 2020
rate as an indicator’ section offers a brief (Eurostat, 2004, 2005, 2007, 2012).
discussion of the validity of the ARoP rate The EU2020 target has been operationa-
as an indicator of poverty. The following sec- lised in terms of three indicators:
tion discusses the empirical methodologies
used and their main strengths and disadvan- . The number of persons at risk of pov-
tages. The subsequent section presents the erty, i.e. the number of persons in house-
composite poverty map for European holds whose disposable income is less
NUTS 3 regions and discusses some of the than 60% of the national median;

Downloaded from lec.sagepub.com by guest on September 22, 2015

4 Local Economy 0(0)

. The number of persons not able to afford conventional to use the national median as
four of out of nine items indicative of the benchmark. This means that regional
material deprivation; ARoP rates can be quite closely correlated
. The number of persons living in house- with their median disposable income levels.
holds where adults (together) work less If each region had its own ARoP bench-
than 20% of a full time year. mark, based upon its regional median
income, variation in the ARoP would be
The three indicators above, added entirely a function of the local income dis-
together but avoiding ‘double counting’ of tribution – or degree of inequity (Eurostat,
individuals constitute the means of moni- 2004). To express it another way the geog-
toring the EU2020 goal in aggregate raphy of ARoP rates is a complex combin-
terms, though differences in the way in ation of variations in income levels and
which Member States define their targets distributions. In terms of Figure 1, regional
mean that it cannot be reconciled directly rates vary partly as a result of shifts in the
with the 28 national objectives.4 income distribution curve to the left or
Data requirements for the three EU2020 right, and partly due to changes in the
indicators are mainly satisfied from the EU- shape of the distribution.
SILC. Although, in the context of the As we have already mentioned, it is
EU2020 targets, monitoring is only required important to recognise the fact that meas-
at Member State level, Eurostat publishes ures related to disposable income may not
NUTS 2 data for the three constituent indi- identify all individuals and groups who are
cators. Coverage varies from country to experiencing poverty even in a narrow
country, some at NUTS 2, some NUTS 1 financial sense. Some authors favour the
and some for the whole country (NUTS 0). adjustment of ARoP rates by excluding
In most countries, the most recent data cur- housing costs (rent and mortgage interest)
rently relates to 2012 (Copus, 2014). from disposable income (e.g. Aaberg et al.,
Within the context of the 2014–20 pro- 2008; Pittau et al., 2011). The rationale for
gramme of the European Structural and this change is that housing costs are the
Investment Funds, some opportunities exist most significant component of regional dif-
for regionally targeted poverty alleviation ferences in the cost of living within coun-
measures. This explains current interest in tries, and that excluding them is a way to
estimating and mapping regional ARoP rates. ‘level the playing field’ between the regions.
Analysis by the European Commission
The ARoP rate as an indicator (CEC, no date) suggested that this adjust-
ment would (on average) increase the ARoP
The ARoP rate is defined as the percentage rate (from 17 to 21% for the EU27), affect-
of persons in households whose equivalised5 ing some Member States more than others,
disposable income is less than 60% of the and reducing the difference between urban
national median. This indicator has some and rural areas (Dijkstra, 2012, personal
rather unusual characteristics, which make communication).
it rather tricky to estimate and to interpret. However, recent research in the UK sug-
It is both an indicator of the level of income gests that housing costs are not the only
and its distribution. The relative strength of form of expenditure which varies substan-
these two sources of variation depends upon tially between regions (Hirsch et al., 2013).
the choice of ‘benchmark’ to define the A broad range of consumer goods, food
‘60%’ of median disposable income. It is and fuels all tend to be more expensive in

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 5

Figure 1. The ARoP rate. Source: Own elaboration.

remote rural or island areas. In addition, et al., 2002, 2005). This method is based
sparsity and climate may impact upon the heavily on the literature on Small Area
average expenditure profile of families in Estimation techniques. Small area estima-
these areas, increasing the travel cost of tion techniques are one methodology that
daily life, and the cost of heating the can be used to provide reliable information
home. However, this line of research has about poverty rates (Haslett and Jones,
yet to progress beyond case studies into 2010). However, it was quickly realised
the field of regional indicators. that in certain countries, notably the
Nordic states and the Netherlands, estima-
NUTS 3 estimation methodol- tion was unnecessary, since ARoP rates
ogy – A pragmatic choice deter- could simply be derived from register data.
mined by data availability Since these are derived from full population
data, the issues of sampling and statistical
The TiPSE project, upon which this paper is modelling errors do not arise.
based, was tasked with producing a NUTS 3 In a few countries national statistical
map of ARoP rates for all countries within agencies have produced their own ‘official’
the ESPON programme area,6 except the estimates of regional ARoP rates. These
countries of Central and Eastern Europe were considered a more reliable option
which acceded in 2004 and 2007. The latter than other estimation strategies available
are the subject of a preceding project, con- to the TiPSE research team, since they are
ducted by the World Bank (WB). often based on data sources not publically
The initial intention was to follow, in all available. However, careful scrutiny of
countries, the World Bank’s PovMap methodology and definitions was necessary
(WBPM) methodology,7 and associated in order to ensure an acceptable degree of
software, which generates NUTS 3 esti- comparability with the ARoP rates for
mates of the ARoP rate through a regres- other European countries.
sion modelling technique, combining Where register data were not available,
information from EU-SILC and 2011 and estimation had not been carried out
Population Census microdata (Elbers by national statistical authorities, the

Downloaded from lec.sagepub.com by guest on September 22, 2015

6 Local Economy 0(0)

preferred estimation procedure was the models based upon aggregate (rather than
WBPM. However, PovMap is rather individual or household) data (Copus and
demanding in terms of data. There are two Coombes, 2014). These too combine EU-
principal technical constraints: SILC and Census data, but using regional
Firstly, the model requires that the EU- aggregate data rather than microdata.
SILC microdata contain regional identifiers Again it is necessary that the SILC data
which allow the sample to be divided into a contain NUTS 2 or NUTS 1 identifiers, in
number of geographic ‘clusters’, for each of order to permit modelling at one of these
which the relationship between poverty and aggregate levels. In this case, an additional
a set of socio-economic covariates is derived constraint is added; there must be a suffi-
through regression modelling. These clus- cient number of cluster-level regions to
ters must be intermediate between allow a statistically significant regression
the whole country and the level at which model to be achieved. There are unfortu-
the ARoP rate estimates are required (in nately a number of European countries in
this case NUTS 3). Effectively this means which there are less than 10 NUTS 2
that the EU-SILC microdata need to con- regions. The requirement for consistent def-
tain either NUTS 2 or NUTS 1 identifiers. initions and categories is also important
In a number of EU Member States the here, though the availability of 2011
sample size is insufficient, and the regional Census data is less of a limitation, since
identifiers are suppressed. regional data have generally become avail-
Secondly, it is necessary to have a suffi- able earlier than microdata. In practice, the
cient number of indicators which are small number of NUTS 2 regions was the
common to both the EU-SILC and the main barrier to implementation of such
Census microdata. This is required because ‘area-based’ models, which again proved
the parameters obtained from the income feasible in only a small number of countries.
regression model estimated using EU-SILC Where none of the above options were
data are applied to the more spatially disag- available recourse to univariate apportion-
gregated Census data to produce regional ment procedures was necessary. The usage
measures of income and ARoP rates (in of such an apportionment method assumes
our case for NUTS 3). A high level of com- that there is a close dependency between the
patibility, both in terms of definition, and spatial distribution of one dimension repre-
(because many variables are categorical sented by a single indicator and the distri-
rather than continuous) in terms of classes, bution of the targeted ARoP rate. The
is required. correlation between these two distributions
In addition to these technical constraints, must be high. In the case of a positive cor-
the timing of the project (2012–14), and relation, a high value of the underlying vari-
delays in the release of 2011 Census micro- able means a high number for the ARoP
data, proved serious limitations to the rate and vice versa. Therefore, the variable
research. Indeed the switch from conven- used must be carefully selected. When inter-
tional census to register-based databases in preting the output of the method it is also
many European countries may mean that always necessary to keep in mind that it
microdata samples are unlikely to become shows mainly the distribution of the one-
available at all. As a result of these com- dimensional indicator, projected to the
bined difficulties, PovMap modelling was ARoP rate. To calculate the ARoP rate
only possible in a handful of countries. over NUTS 3 regions of one country, the
The next-best option for estimating average national (NUTS 0) ARoP rate,
NUTS 3 ARoP rates would be regression extracted from EU-SILC, is multiplied by

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 7

Table 1. Overview of approaches implemented.

Country Method Year

Albania WBPM 2008

Austria WBPM 2011
Belgium APPT 2011
Croatia WBPM 2002–4
Cyprus EUSILC 2010
Denmark REG 2010
Finland REG 2010
France NSI 2011
Germany ABM 2011
Greece WBPM 2011
Iceland EUSILC 2010
Ireland EUSILC 2009
Italy ABM 2011
Luxembourg EUSILC 2011
Malta EUSILC 2011
Netherlands REG 2010
Norway REG 2010
Portugal APPT 2011
Spain APPT 2010
Sweden REG 2010
Switzerland APPT 2011
Turkey ABM 2011
United Kingdom WBPM/NSI WBPM: 2001. NSI: 2005–8 (SCO),
2007–8 (EW), 2009/10-2011/12 (NI).

ABM: area-based model; APPT: univariate apportionment; EUSILC: European Union Statistics on Income
and Living Conditions; NSI: estimates produced by the National Statistical Institute; REG: register data;
WBPM: World Bank PovMap.

the z-standardised value of the selected indi- used (in three of these by TiPSE research-
cator. This method leads at first to an incon- ers). Area-based models (ABMs) were used
sistency in comparison to probably available in three countries, and univariate appor-
EU-SILC-based ARoP rates on NUTS 1 or tionment in four. In five countries (mostly
NUTS 2. To overcome this, the calculated those which comprise a single NUTS 3
NUTS 3 rates may be adjusted to the avail- region) the AROP rate was derived directly
able higher level. This method was used in from EU-SILC, whilst in three countries
Belgium, Spain, Portugal and Switzerland, estimates made by the national statistical
where the number of NUTS 1 and NUTS institute were used.8
2 regions is too few to allow a more sophis- In addition to the data listed in Table 1,
ticated regression model and availability of the maps and analysis presented in this
proxy indicators relatively poor. paper incorporate (WBPM) ARoP rate esti-
Table 1 shows that register data were mates for Hungary, Latvia, Romania,
available in five of the countries covered Slovakia and Slovenia, generously made
by the TiPSE mapping exercise. In five available to us by DG Regio. In those coun-
more countries the PovMap model was tries for which no NUTS 3 estimates are

Downloaded from lec.sagepub.com by guest on September 22, 2015

8 Local Economy 0(0)

available, Eurostat data9 at the lowest based on statistical regression models.

NUTS level available was used to complete These weaknesses arise from small sample
the maps and to populate the dataset. size, poor model specification and the
inappropriate choice of statistical estimator.
Strengths and weaknesses of the All three issues can affect the quality of the
alternative approaches to the parameters obtained from the income
estimation of regional ARoP rates models estimated in PovMap to generate
estimates of indicators of income poverty.
As discussed above, the choice of approach The principal reason for using the ABM
used to estimate ARoP rates for NUTS 3 regression method, instead of the WBPM
regions was constrained by issues of data method, was the unavailability of appropri-
availability. Ideally, we would like to have ate microdata. Even if Census microdata do
used estimates produced by each country’s exist, the challenges involved in reconciling
respective national statistics authority, the Census microdata variables with the
either directly from register-based data or EU-SILC microdata variables may be too
indirectly from national surveys or the com- great. With the ABM approach the covari-
bination of national surveys with Census ates at NUTS 2 and NUTS 3 may both be
data. Where this approach was not possible derived from the Census, and the problem
we had to carry out some form of estima- of matching their definitions does not arise.
tion (in order of preference): On the down side, the ABM approach is
vulnerable to the ecological fallacy. In
other words, instead of estimating relation-
. World Bank PovMap (WBPM); ships between disposable income and socio-
. ABM; economic covariates at a household level, it
. Univariate apportionment (APPT). does so on the basis of NUTS 2 regional
averages. The latter may mask substantial
The main advantages of the WB poverty intra-regional differences, which could
mapping methodology (WBPM), compared potentially distort the estimates at NUTS 3.
to the simpler ABM and univariate appor- A further disadvantage (shared with the uni-
tionment (APPT) include: variate apportionment method) is that the
definition of the ARoP rate becomes fixed
. Better representation of the multi-dimen- by the rates available at NUTS 2. It is not
sion nature of poverty through the inclu- possible (as in the case of the WBPM
sion of multiple proxy variables (for approach) to carry out sensitivity analysis
welfare) simultaneously; with different poverty lines.
. Availability of measures of the statistical Where the number of regions available to
precision (i.e. confidence intervals) of the implement the ABM methodology is small,
indicators of poverty produced; we had to adopt a simpler and less data
. Ability to compute different measures demanding alternative based on direct uni-
of poverty (e.g. headcount index, pov- variate apportionment to produce ARoP
erty gap index, poverty severity index) rates. The main advantage of the APPT
and inequality (e.g. gini coefficient, gen- approach is its simplicity. The main disad-
eralised entropy measures, Atkinson’s vantage is that the weights used in the
inequality measures). apportionment may not correlate well with
the distribution of income poverty rates
The WBPM methodology also has weak- within the sub-populations of interest (in
nesses, which are common to any technique our case NUTS 3 regions), giving rise to

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 9

measurement error in the estimates of the Although due regard should be paid to
ARoP rate. As with the ABM approach, it the above cautions the composite map pre-
is not possible to carry out sensitivity ana- sented below is nevertheless of considerable
lysis with different poverty lines. interest as the best approximation to
A consequence of the necessary adoption Europe-wide NUTS 3 regional variation in
of different methodologies to estimate the ARoP rate with currently available
NUTS 3 ARoP rates is a limited ability to data.
make the rates more comparable between
countries. This requires careful consider- The resulting poverty maps
ation and interpretation of the composite
European map of NUTS 3 ARoP rates, pre- Figure 2 shows all the NUTS 3 ARoP rates
sented in the following section. There are estimated or collected by the TiPSE
two elements to the comparability issue: research team. As already explained, each
country has a different poverty threshold,
(1) The incorporation of national poverty depending upon the distribution of house-
lines. Clearly the median household hold disposable income across its popula-
income of some countries in the south tion. These range from E20,362 in
and east of Europe is very much lower Switzerland to E5,520 in Greece. From
than those of the centre and north-west. one perspective this could be said to be jus-
Sadly, there is no way to implement a tified by differences in the cost of living, and
common poverty line. However, some by different expectations or perceptions of
simple adjustments can be made, poverty. Nevertheless, it seems problematic
which to some extent at least shed that such differences take place abruptly
light upon the implications. One per- along national borders, and either the map
spective on this would be to view the must be carefully interpreted with this in
map of ARoP rates (Figure 2) as pri- mind, or some form of adjustment must
marily showing patterns of the disper- be attempted.
sion of income, since differences in Taking the first of these options, the pat-
level are to some extent subsumed by tern revealed by Figure 2 is mostly quite
the national poverty lines. In effect it reassuring. The highest rates of poverty
is closer to a map of degrees of regional are in Romania, Bulgaria, Eastern Poland,
inequality, rather than of differences in Eastern Germany, the Baltic States,
levels of income. Careful consideration Southern Spain, Southern Italy, Greece
should be given to the implications of and Turkey, whilst the lowest rates are gen-
this for the potential use of the map in erally found in Northern Italy, Austria,
policy targeting. Southern Germany, Netherlands the South
(2) Differences in estimation error, due to of England, Norway, Southern Sweden and
the different approaches used. Only the Iceland.
WBPM approach allows for variance In Figure 3 the ARoP rates are shown as
estimation. These are available for within-country-quintiles. The darkest greys
United Kingdom, Greece and Austria pick out those regions within the highest
(where the WBPM approach was imple- 20% in each country, whilst the regions
mented) on request from the authors. with the lightest shading are those in the
With respect to the other approaches 20% of regions with the lowest ARoP
it is still important to recognise estima- rates. In this map broad macro-regional dis-
tion error as a constraint to compar- parities are ‘downplayed’ and more loca-
ability between countries. lised variation is emphasised. The pattern

Downloaded from lec.sagepub.com by guest on September 22, 2015

10 Local Economy 0(0)

Figure 2. NUTS 3 at risk of poverty rates: unadjusted.

reveals a tendency for lower ARoP rates in Turkey, the Southern parts of Italy,
the vicinity of capitals and other large cities Greece, France and Spain, South-West
(but not necessarily in the cities themselves, Ireland, West Wales, Western Scotland,
if tightly bounded), and relatively high rates Eastern Germany, Northern Sweden and
of income poverty in remoter regions (such Eastern Finland). The area along the
as Eastern Romania and Poland, Eastern Franco-Belgian border and the North-East

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 11

Figure 3. NUTS 3 at-risk-of-poverty rates: national quintiles.

coast of the Netherlands also show up as this approach and the previous map is
having relatively high rates of income that the index reflects the scale/degree of
poverty. the disparity between each region and the
Figure 4 shows the same ARoP data, national mean, a metric which is to some
but this time expressed as an index of extent lost in the quintile approach.
the national mean. The difference between Figure 4 therefore enables us to pick out

Downloaded from lec.sagepub.com by guest on September 22, 2015

12 Local Economy 0(0)

Figure 4. NUTS 3 at risk of poverty rates: national average ¼ 100.

the more extreme values, both positive in Southern Italy and Spain). Others are
and negative. Some of these reinforce the less expected, such as the low rates of
generalisations derived from Figure 3 (e.g. poverty along the border between Spain
low rates around capital cities, high rates and France.

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 13

What can the regional ARoP transition are included in the TiPSE project
rates tell us about patterns of Final Report (Copus, 2014), but will not be
poverty in relation to different reported in detail here, since the interpret-
kinds of region? ation of the graphs is less clear cut.
The first typology (Figure 5) is the clas-
The above maps and commentary provide sification of NUTS 3 regions by rurality and
some initial first impressions of the spatial accessibility (Dijkstra and Poelman, 2008,
variation of income poverty at the NUTS 3 2011). The five categories are predomin-
level. However, they do not take us very far antly urban, intermediate close to a city,
in terms of developing an explanation of the intermediate remote, predominantly rural
causes of regional differentiation in income close to a city and predominantly rural
poverty. One simple way to begin to shed remote. There are very few regions in the
light upon this is to use regional typologies third category, and for this reason we do
to explore how ARoP rates vary in different not distinguish accessible and remote inter-
kinds of region, using a series of simple bar mediate regions.
charts, which present ARoP rates averaged Figure 5 shows that there are some quite
across each type of region within each coun- substantial differences between ARoP rates
try. This approach both avoids including across this typology. Three broad groups of
data from countries in which a typology is countries emerge. In four central countries
not relevant (such as island regions in (Belgium, Germany, Denmark and
Austria), and means that data from differ- Netherlands) income poverty rates are
ent countries (with different poverty lines) higher in urban areas than in intermediate
are kept separate. or rural areas. In a further 10 countries,
In this section, we describe how poverty income poverty rates are higher in rural
rates vary between rural and urban regions, and/or intermediate regions. The strongest
and between island and mainland regions. associations with rurality are in the
Typologies (all derived from the ESPON Mediterranean countries (Spain, Portugal,
programme10) relating to metropolitan Greece and Italy), and in former socialist
regions, border regions, mountain regions, countries (Romania, Hungary). In Turkey
coastal regions and regions in industrial accessible rural areas (mostly along the

Figure 5. At risk of poverty by urban–rural type, selected countries.

Downloaded from lec.sagepub.com by guest on September 22, 2015

14 Local Economy 0(0)

Figure 6. At risk of poverty rate, island and mainland regions.

Eastern border) exhibit the highest poverty often contain areas with high rates of
rates. In the final group (UK, France, income poverty, whilst in the South
Sweden, Austria and Norway), poverty and East cities tend to have relatively
rates seem less affected by rural–urban lower rates.
differences. (3) Accessible rural areas, especially those
In some national contexts, NUTS 3 is close to larger cities and capitals, tend
too large a scale to pick up the role of insu- to have relatively low rates of income
larity. Most islands are subsumed within poverty.
larger regions which are predominantly (4) Remote rural regions often exhibit rela-
‘mainland’. Nevertheless, in Italy, Spain, tively high ARoP rates.
Greece and France (Figure 6) island regions (5) In some countries island regions have
have significantly higher poverty rates. In higher ARoP rates than mainland
Sweden, UK and Denmark island and regions.
mainland rates are almost identical. The (6) The relationship between mountain
remaining two countries, Portugal and regions, border regions and industrial
Finland can be seen as special cases, due regions and poverty rates is variable,
to the relatively low rates in Madeira (the depending upon national and macro-
only Portuguese island region for which region context.
there is data), and in Finnish Åland.
Taken together, the poverty maps, and
the graphical analysis based on all the typol- Associations between poverty
ogies considered by the TiPSE project, sug- rates and some potential
gest the following observations about the socio-economic drivers of
geography of income poverty: poverty

(1) At a macro-scale the highest rates of The majority of the ESPON typologies
poverty tend to be in the relate to geographical features, rather than
Mediterranean countries and Turkey, socio-economic characteristics. The latter
the lowest in the Northern and may be explored through correlation ana-
Western countries. lysis with a selection of key indicators
(2) The relationship between capital cities, from the Eurostat Regio database, and
and secondary cities, and ARoP rates is from the Population Census variables col-
complex. Broadly speaking large cities lected by the TiPSE project as part of the
in the North and West of Europe efforts to map aspects of social exclusion.11

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 15

In choosing indicators the underlying aim is Table 2 describes the 32 indicators

to identify socio-economic characteristics included in the correlation analysis which
which seem likely to be associated with spa- follows. It shows that the majority are avail-
tial variations in poverty rates. The selec- able at NUTS 3 level, but that six NUTS 2
tion of indicators will necessarily be indicators (derived from the Labour Force
constrained by data availability at NUTS 3. Survey (LFS) are also included. Most of the
It is of course acknowledged that correlations indicators relate to the year 2011, though in
are not evidence of causality, but they may a few cases recourse to earlier data has
nevertheless provide valuable pointers to been necessary. For some indicators, gaps
policy approaches and targeting. in the NUTS 3 data availability have been
The selected indicators fall into five filled with data for NUTS 2, and in a small
broad groups, according to assumptions number of cases NUTS 1 or NUTS 0. As a
about the way in which they may be asso- result data are available for at least 1100
ciated with regional variations in the inci- NUTS 3 for each of the indicators. At
dence of poverty: best the number of regions rises to almost
1500.
(1) Agglomeration: The first group of indi- The methodology used to explore rela-
cators represent the assumption that tionships between ARoP rates and the indi-
low rates of poverty may be associated cators listed in Table 2 was simple. Pearson
with agglomerative advantages. Access product moment correlation coefficients
to larger centres of economic activity were calculated between the ARoP rate
may increase earning potential, while and each of the 31 indicators, first for the
peripherality may increase vulnerability full set of regions within the ESPON space
to poverty. for which data were available, and then for
(2) Productivity: The second category of subsets of regions, EU15 and EU12, and a
indicator is derived from the assump- grouping based upon the Esping-Andersen
tion that higher productivity in a (1994) classification of welfare regimes. The
region may be associated with lower results of these analyses are presented in
poverty rates. Tables 3 and 4. In each case coefficients
(3) Labour market: Poverty rates are often which are not significant at the 0.01 level
associated with labour market charac- are identified by grey shading, whilst those
teristics, high levels of inactivity or in excess of  0.5 are highlighted in bold.
unemployment is assumed to lead to a The ‘ESPON space’ column of Table 3 is
higher incidence of poverty. based upon the full database of up to 1448
(4) Sectoral structure: Poverty can be caused regions. At first sight this may lead
by lags in adjustment of the economy to us to expect the strongest associations.
a changing economic environment or the However, this is not the case, since it is
impact of globalisation. It is reasonable clear from the EU15 and EU12 columns,
to assume that regions with a high inci- and from Table 4, that different relation-
dence of poverty may have particular ships pertain in different parts of Europe.
structural characteristics. This perhaps explains the fact that only
(5) Human capital: In addition to sectoral four indicators have coefficients in excess
characteristics, it is conceivable that of  0.5. The strongest association, which
high rates of poverty may be associated is sustained across the ESPON space is
with social characteristics, such as edu- between poverty rates and unemployment
cation and skills, age, household/family rates (both those derived from the LFS
structures, ethnicity or citizenship. and the Census). Employment rates are

Downloaded from lec.sagepub.com by guest on September 22, 2015

 16
Table 2. The indicators used in the correlation analysis.

Indicator NUTS
Theme name Description Source Year level n

Agglomeration Density Population density Eurostat database table 2011 3 1422

demo_r_d3dens
Access MM Multimodal accessibility index ESPON Territorial Observation 2009 3 1331
No. 2 (2009)
Access RD Road accessibility index ESPON Territorial Observation 2009 3 1123
No. 2 (2009)
Productivity GDP PPS HAB GDP per capita Eurostat database table 2011 3 1436
Downloaded from lec.sagepub.com by guest on September 22, 2015

nama_r_e3gdp
Labour Unemp(Reg) Unemployment rate Eurostat database table 2009 3/2 1430
market lfst_r_lfu3rt
Unemp(Cen) Unemployment rate 2011 Census (TiPSE database) 2011 3 1355
Yunemp(Reg) Youth unemployment rate Eurostat database table 2009 3/2 1433
lfst_r_lfu3rt
Yunemp(Cen) Youth unemployment rate 2011 Census (TiPSE database) 2011 3 1355
LTUnemp(Reg) Long-term unemployment rate Eurostat database table 2011 2 1429
lfst_r_lfu2ltu
EARate Economic activity rate Eurostat database table 2011 2 1436
lfst_r_lfp2actrt
Inact Economic inactivity rate 2011 Census (TiPSE database) 2011 3 1355
EmpRate Employment rate Eurostat database table 2011 2 1436
lfst_r_lfe2emprt
Sectoral A Per cent employed in NACE Eurostat database table 2010 3 1311
structure sector A nama_r_e3em95r2
B-E Per cent employed in NACE Ditto ditto ditto 1327

Local Economy 0(0)

sectors B-E
C Per cent employed in NACE Ditto ditto ditto 1231
sector C
F Per cent employed in NACE Ditto ditto ditto 1327
sector F
(continued)
 Table 2. Continued.

Copus et al.
Indicator NUTS
Theme name Description Source Year level n

G-J Per cent employed in NACE Ditto ditto ditto 1231

sectors G-J
K-N Per cent employed in NACE Ditto ditto ditto 1231
sectors K-N
O-U Per cent employed in NACE Ditto ditto ditto 1327
sectors O-U
ElemOcc Per cent employed in elementary 2011 Census (TiPSE database) 2011 3 1327
occupations
Downloaded from lec.sagepub.com by guest on September 22, 2015

Human LowEd(Reg) Per cent of persons with low Eurostat database table 2011 2 1420
capital educational attainment edat_lfse_09
LowED(Cen) Per cent of persons with low 2011 Census (TiPSE database) 2011 3 1367
educational attainment
HiEd(Reg) Per cent of persons with high Eurostat database table 2011 2 1420
educational attainment edat_lfse_13
HiEd(Cen) Per cent of persons without high 2011 Census (TiPSE database) 2011 3 1367
educational attainment
HRST Per cent of population with ter- Eurostat database table 2011 2 1422
tiary education and/or hrst_st_rcat
employed in science and
technology
CDR(Reg) Child dependency rate Eurostat database table 2011 3 1436
demo_r_pjanaggr3
CDR(Cen) Child dependency rate 2011 Census (TiPSE database) 2011 3 1448
OADR(Reg) Old age dependency rate Eurostat database table 2011 3 1422
demo_r_pjanaggr3
OADR(Cen) Old age dependency rate 2011 Census (TiPSE database) 2011 3 1448
LoneP(Cen) Lone parent households as a 2011 Census (TiPSE database) 2011 3 1366
percentage of all households
Foreign(Cen) Per cent of population born 2011 Census (TiPSE database) 2011 3 1367
abroad
NonCit(Cen) Per cent of population who are 2011 Census (TiPSE database) 2011 3 1367
not citizens of the country

17
18 Local Economy 0(0)

Table 3. Pearson correlation coefficients between ARoP rates (unadjusted) and selected socio-economic
indicators, ESPON space, EU15 and EU12.

Theme Indicator ESPON space EU15 EU12

Agglomeration Density 0.029 0.107 0.285

Access MM 0.386 0.396 0.580
Access RD 0.384 0.474 0.706
Productivity GDP PPS HAB 0.385 0.300 0.561
Labour market Unemp(Reg) 0.542 0.667 0.099
Unemp(Cen) 0.551 0.618 0.111
Yunemp(Reg) 0.471 0.603 0.150
Yunemp(Cen) 0.493 0.526 0.375
LTUnemp(Reg) 0.408 0.631 0.195
EARate 0.397 0.312 0.247
Inactivity 0.275 0.243 0.315
EmpRate 0.503 0.522 0.260
Sectoral structure A 0.387 0.279 0.714
(employment) B–E 0.261 0.336 0.301
C 0.298 0.369 0.288
F 0.096 0.205 0.430
G–J 0.103 0.054 0.582
K–N 0.231 0.185 0.496
O–U 0.007 0.180 0.372
ElemOcc 0.509 0.569 0.415
Human capital LowEd(Reg) 0.476 0.437 0.646
LowED(Cen) 0.212 0.143 0.659
HiEd(Reg) 0.476 0.437 0.642
HiEd(Cen) 0.163 0.116 0.269
HRST 0.448 0.341 0.517
CDR(Reg) 0.358 0.063 0.238
CDR(Cen) 0.351 0.045 0.361
OADR(Reg) 0.176 0.009 0.301
OADR(Cen) 0.159 0.002 0.357
LoneP(Cen) 0.276 0.401 0.193
Foreign(Cen) 0.126 0.071 0.204
NonCit(Cen) 0.050 0.009 0.111

Notes:
1
Grey shaded cells in the table indicate coefficients which are not significant at 0.01.
2
NACE Classification:
A: Agriculture, forestry and fishing
B–E: Industry (except construction)
C: Manufacturing
F: Construction
G–J: Wholesale and retail trade, transport, accommodation and food service activities, information and communication
K–N: Financial and insurance activities; real estate activities; professional, scientific and technical activities; administrative
and support service activities
O–U: Public administration and defence; compulsory social security; education; human health and social work activities;
arts, entertainment and recreation, repair of household goods and other services

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 19

Table 4. Pearson correlation coefficients between ARoP rates (unadjusted) and selected socio-economic
indicators – by welfare regime classification.

Welfare regimes

Theme Indicator Nordic Liberal State-based Family-based Post-Socialist

Agglomeration Density 0.122 0.517 0.146 0.105 0.244

Access MM 0.128 0.221 0.067 0.559 0.581
Access RD 0.288 0.244 0.027 0.664 0.716
Productivity GDP PPS HAB 0.288 0.108 0.086 0.612 0.522
Labour Market Unemp(Reg) 0.869 0.477 0.646 0.474 0.112
Unemp(Cen) 0.837 0.104 0.530 0.638 0.072
Yunemp(Reg) 0.807 0.277 0.356 0.288 0.148
Yunemp(Cen) 0.802 0.069 0.321 0.410 0.243
LTUnemp(Reg) 0.629 0.022 0.608 0.205 0.081
EARate 0.399 0.257 0.271 0.331 0.009
Inactivity 0.115 0.374 0.392 0.060 0.192
EmpRate 0.538 0.253 0.389 0.511 0.036
Sectora A 0.231 0.253 0.162 0.275 0.687
structure B-E 0.238 0.040 0.181 0.442 0.224
C 0.311 0.057 0.204 0.445 0.200
F 0.001 0.081 0.055 0.269 0.421
G-J 0.464 0.063 0.052 0.065 0.588
K-N 0.161 0.176 0.129 0.284 0.479
O-U 0.068 0.290 0.298 0.295 0.400
ElemOcc 0.010 0.506 0.165 0.511 0.427
Human LowEd(Reg) 0.180 0.269 0.141 0.465 0.610
capital LowED(Cen) 0.155 0.270 0.111 0.003 0.472
HiEd(2) 0.180 0.269 0.141 0.465 0.607
HiEd(Cen) 0.166 0.010 0.081 0.028 0.231
HRST 0.287 0.264 0.116 0.384 0.466
CDR(Reg) 0.745 0.011 0.254 0.565 0.218
CDR(Cen) 0.482 0.044 0.219 0.558 0.272
OADR(Reg) 0.526 0.143 0.317 0.352 0.217
OADR(Cen) 0.572 0.162 0.329 0.369 0.237
LoneP(Cen) 0.469 0.529 0.295 0.090 0.163
Foreign(Cen) 0.205 0.254 0.112 0.253 0.272
NonCit(Cen) 0.400 0.132 0.055 0.263 0.083

Notes:
1
Grey shaded cells in the table indicate coefficients which are not significant at 0.01.
2
NACE Classification:
A: Agriculture, forestry and fishing
B–E: Industry (except construction)
C: Manufacturing
F: Construction
G–J: Wholesale and retail trade; transport; accommodation and food service activities; information and communication
K–N: Financial and insurance activities; real estate activities; professional, scientific and technical activities; administrative
and support service activities
O–U: Public administration and defence; compulsory social security; education; human health and social work activities;
arts, entertainment and recreation, repair of household goods and other services

Downloaded from lec.sagepub.com by guest on September 22, 2015

20 Local Economy 0(0)

also closely related to poverty rates. Youth Of the five welfare regime groups the
unemployment rates and long-term strongest and clearest relationship with
unemployment rates are close behind. labour market characteristics is in the
Although none of the broad NACE Nordic countries, where coefficients rise
employment sectors have a coefficient to > 0.8 for unemployment and youth
above 0.4 (the primary sector reaches þ unemployment. Old age dependency rates
0.39), employment in elementary occupa- are also strongly (positively) correlated
tions shows a relatively strong association with poverty rates in the Nordic states.
with ARoP rates (þ0.51). The other three Child dependency rates have an ambiguous
broad groups of indicators, (agglomeration, relationship with poverty here, the LFS and
productivity, and human capital) all have Census indicators pointing in opposite
coefficients of less than 0.5, suggesting that directions.
if there are relationships with the incidence In the liberal welfare regime group (UK
of poverty they are not consistent across the and IE) the number of regions is smaller, and
ESPON space. The closest association from it may be that this explains the relatively small
these three broad groups is with educational number of strong associations. The presence
attainment. of lone parents, and persons in elementary
We begin to see different relationships in occupations, is both associated with elevated
different parts of the ESPON space as soon poverty rates. Population density is also posi-
as we separate the EU15, from the New tively associated with poverty, reinforcing the
Member States (NMS) of the EU12.12 The findings presented above, where the UK was
difference may be summed up as follows: In one of the few countries in which metropolitan
the EU15 countries poverty is strongly asso- poverty rates were higher than those for non-
ciated with labour market characteristics and metropolitan regions.
employment in elementary occupations. By In the countries with ‘State-based’ wel-
contrast, in the NMS accessibility, primary fare regimes, labour market characteristics
sector employment, education and skills, and were again strongly associated with poverty,
productivity are all much more important though youth unemployment was less clo-
than labour market characteristics. sely correlated than in the Nordic countries.
The fact that GDP per capita is strongly Here, also unlike the Nordic countries age-
(negatively) associated with poverty rates in related dependency does not seem to play
the NMS suggests that here overall per- such a strong role. Regional productivity
formance of regional economies plays a and accessibility/agglomeration are only
key role, whereas in the EU15 the low cor- very weakly associated with poverty here.
relation coefficient suggests that regional Turning to the ‘Family-based’ welfare
performance is less influential than distribu- regimes (the Mediterranean countries and
tional effects. Turkey), accessibility and child dependency
The classification of countries according rates have distinctive associations with pov-
to welfare regime developed in the TiPSE erty. Overall regional performance (GDP
project was based on the work of Esping- per capita) is also important here. Labour
Andersen (1994), extended to cover the market indicators exhibit less strong associ-
ESPON area. It is described in more detail ations than in the Nordic countries,13 whilst
in Talbot et al. (2012). This classification employment in elementary occupations is
suggests some further interesting relation- associated as strongly with poverty as it is
ships between ARoP rates and the eco- in the Liberal welfare regime countries.
nomic and social characteristics of the Finally, the post-socialist countries share
constituent regions (Table 4). certain key characteristics with the previous

Downloaded from lec.sagepub.com by guest on September 22, 2015

Copus et al. 21

group, notably the fact that poverty is therefore recommend that Eurostat and
strongly associated with overall regional the national statistical institutes explore
economic performance, and with accessibil- the possibility of adding standard living
ity. However, here education and training, cost indicators to the EU-SILC, and
and employment in the primary sector, are using these to adjust regional AROP rates.
also strongly associated with variations in
poverty.
Conflict of interest
Recommendations and The author(s) declared no potential conflicts of
suggestions for future research interest with respect to the research, authorship,
and/or publication of this article.
Given our earlier discussions about the rela-
tion between data availability and empirical
methodology, we propose the following Funding
recommendations:
The author(s) disclosed receipt of the following
financial support for the research, authorship,
. That (in the short term) NUTS 2 identi-
and/or publication of this article:
fiers be included in the EU-SILC micro-
This publication is based on results of the
data for all countries. This would allow
ESPON project ‘The Territorial Dimension of
increasing the number of countries in
Poverty and Social Exclusion in Europe
which the PovMap methodology could
(TiPSE)’. ß ESPON 2013, TiPSE, Nordregio.
be implemented.
. That (in the medium term) consideration
be given to aligning EU-SILC covariates Notes
with the variables of the hyper-tables of 1. www.epp.eurostat.ec.europa.eu/portal/page/
the new Eurostat Census Hub. portal/microdata/eu_silc (accessed 12 June
. In the longer term, since the trend seems 2014).
to be increasingly towards register/ 2. NUTS is an abbreviation for Nomenclature des
unités territoriales statistiques. This is
survey-based systems and away from
Eurostat’s hierarchical classification of regions,
the conventional census, consideration
from Member States (NUTS 0) down to small
should be given to specifying regional areas – Local Area Unit (LAU)2.
ARoP rates as an element of the 3. CEC (2005) and Verma et al. (2006) con-
Census Hub hyper-tables. These should sidered different methodologies to calculate
be expressed both in terms of national poverty indicators on small areas such as
poverty lines and an EU-28 poverty NUTS 2 or NUTS 3 regions. They relied on
line, adjusted for Purchasing Power methodologies including area based and indi-
Parity. vidual micro data. Due to lack of available
. As noted above, poverty measures which data from European wide sources, they were
take account only of income, whilst only able to produce indicators on the NUTS
ignoring variations in living costs are 3 level for Italy. For the other countries, they
produced indicators on NUTS 2 level.
not always able to identify those most
4. www.ec.europa.eu/europe2020/pdf/targets_
affected by poverty. However to adjust en.pdf (accessed 10 April 2014).
for housing cost but to ignore the very 5. This is an adjustment for household size.
significant cost increases associated with 6. The ESPON programme covers the 28 EU
insularity and peripherality would Member State, plus Iceland, Liechtenstein,
appear to introduce an unintentional Norway and Switzerland. The ‘ESPON
urban bias in the indicator. We would space’, to be covered (as far as practicable)

Downloaded from lec.sagepub.com by guest on September 22, 2015

22 Local Economy 0(0)

in the TiPSE mapping exercise also included CEC – Commission of the European Com-
Turkey and the Balkan states. munities, DG Social Affairs, Employment
7. See www.go.worldbank.org/A16JQP9HJ0. and Inclusion (No date) Social Situation
8. Standard errors (SEs) of the small area Monitor – Research findings – Implications
ARoP estimates were estimated for the for poverty rates. Available at: www.ec.euro-
countries for which the WBPM method- pa.eu/social/main.jsp?catId¼1050&intPageId
ology was possible to implement (i.e. ¼1933&langId¼en (accessed 27 August
United Kingdom, Greece and Austria) and 2014).
are available on request from the authors. CEC – Commission of the European
Given the deterministic nature of the other Communities (2004) Facing the challenge.
two approaches to the estimation of small The Lisbon strategy for growth and employ-
area ARoP estimates (i.e. ABM, APPT), it ment. Report from the High Level Group
is not possible to estimate standard errors chaired by Wim Kok.
for these estimates. CEC – Commission of the European Com-
9. Table ilc_li41. munities (2010) Europe 2020 – A Strategy
10. These are all available in a spreadsheet for Smart, Sustainable and Inclusive Growth.
which may be downloaded from the COM 2020.
ESPON website www.espon.eu/main/ Copus AK (ed.) (2014) ESPON TiPSE main
Menu_ToolsandMaps/ESPONTypologies/ report. Final report. ESPON 2013 pro-
(accessed 1 April 2014). gramme. Luxembourg: ESPON. Available
11. Annex 6 and Annex 8 of the TiPSE Final at: www.espon.eu/main/Menu_Projects/
Report. Menu_AppliedResearch/tipse.html (accessed
12. Croatia is not included here due to the sub- 27 August 2014).
stantial proportion of missing data. Copus AK and Coombes M (2014) Small area
13. This seems to be related to data gaps in estimation of at risk of poverty rates. Second
Turkey, since in Table 4 the Mediterranean ESPON 2013 scientific report. Luxemburg:
countries (without Turkey) exhibit strong ESPON 2013 Coordination Unit.
associations between poverty rates and Crow B, Zlatunich N and Fulfrost B (2009)
labour market indicators. Mapping global inequalities: Beyond income
inequality to multi-dimensional inequalities.
References Journal of International Development 21(8):
1051–1065.
Aaberg R, Langørgen A, Mogstad M, et al.
Dijkstra L and Poelman H (2008) Remote rural
(2008) The Impact of Local Public Services
regions, how proximity to a city influences the
and Geographical Cost of Living Differences
performance of rural regions. Regional Focus
on Poverty Estimates. Discussion Paper No.
No 1, DG Regio, European Commission.
3686. IZA. Bonn: Institute for the Study of
Available at: www.ec.europa.eu/regional_po-
Labor. Available at: www.ftp.iza.org/
licy/sources/docgener/focus/2008_01_rur-
dp3686.pdf (accessed 27 August 2014).
al.pdf (accessed 21 November 2013).
Betti G, Gagliardi F, Lemmi A, et al. (2012)
Dijkstra L and Poelman H (2011) Regional
Subnational indicators of poverty and depriv-
typologies: A compilation. Regional Focus
ation in Europe: Methodology and applica-
No 1. DG Regio, European Commission.
tions. Cambridge Journal of Regions,
Available at: www.ec.europa.eu/regional_po-
Economy and Society 5: 129–147.
licy/sources/docgener/focus/2011_01_typolo-
CEC – Commission of the European
gies.pdf (accessed 21 November 2013).
Communities, DG Employment and Social
Elbers C, Lanjouw J and Lanjouw P (2002)
Affairs (2005) Regional Indicators to Reflect
Micro-level Estimation of Welfare. Policy
Social Exclusion and Poverty. Final report
Research Working Paper Series 2911.
of Project VT/2003/45 by Verma V, Betti G,
Washington, DC: The World Bank.
Lemmi A, et al.

Downloaded from lec.sagepub.com by guest on September 22, 2015

 Copus et al. 23

Elbers C, Lanjouw J and Lanjouw P (2005) Marlier E and Atkinson A (2010) Indicators of
Imputed welfare estimates in regression ana- poverty and social exclusion in a global con-
lysis. Journal of Economic Geography 5: text. Journal of Policy Analysis and
101–118. Management 29(2): 285–304.
Esping-Andersen G (1994) After the Golden Age: Pittau M, Zelli R and Massari R (2011) Do spa-
The Future of the Welfare State in the New tial price indices reshuffle the Italian income
Global Order. Occasional Paper no. 7. Geneva: distribution? Modern Economy 2: 259–265.
World Summit for Social Development. Ringen S (1988) Direct and indirect measures of
ESPON (2009) Territorial Observation No 2: poverty. Journal of Social Policy 17: 351–365.
Trends in Accessibility, Luxembourg. Talbot H, Madanipour A and Shucksmith M
Available at: www.espon.eu/main/Menu_ (2012) ESPON TiPSE Review of Concepts of
Publications/Menu_TerritorialObservations/ Poverty and Social Exclusion. Annex 1 to
trendsinaccessibility.html (accessed 23 June Final Report. ESPON 2013 programme.
2014). Luxembourg: ESPON. Available at: www.e-
Eurostat (2004) Poverty and Social Exclusion in spon.eu/main/Menu_Projects/Menu_Applied
the EU. Statistics in Focus 16. Luxembourg: Research/tipse.html (accessed 27 August
Eurostat. 2014).
Eurostat (2005) Income Poverty and Social Townsend P (1987) Deprivation. Journal of
Exclusion in the EU25. Statistics in Focus Social Policy 16: 125–146.
13. Luxembourg: Eurostat. Verma V, Betti G, Natilli M, et al. (2006)
Eurostat (2007) Measuring Progress Towards a Indicators of Social Exclusion and Poverty in
More Sustainable Europe. Luxembourg: Europe’s Regions. Working Paper n. 59.
Eurostat. Siena: Department of Quantitative Methods,
Eurostat (2012) 23 % of EU Citizens Were at University of Siena.
Risk of Poverty or Social Exclusion in 2010. Whelan C, Layte R and Maı̂tre B (2002)
Statistics in Focus 9. Luxembourg: Eurostat. Multiple deprivation and persistent poverty
Fahey T (2007) The case for an EU-wide meas- in the European Union. Journal of European
ure of poverty. European Sociological Review Social Policy 12: 91–105.
23(1): 35–47. Whelan C, Layte R, Maı̂tre B, et al. (2001)
Haslett J and Jones G (2010) Small-area estima- Income, deprivation and economic strain:
tion of poverty: The aid industry standard An analysis of the European Community
and its alternatives. Australian and New Household Panel. European Sociological
Zealand Journal of Statistics 52(4): 341–362. Review 17(4): 357–372.
Hirsch D, Bryan A, Davis A and Smith N (2013) Whelan C and Maitre B (2010) Welfare regime
Minimum Income Standard for Remote and and social class variation in poverty and eco-
Rural Scotland-Summary and Key Findings. A nomic vulnerability in Europe: an analysis of
Report for Highlands and Islands Enterprise. EU-SILC. Journal of European Social Policy
Available at: http://www.hie.co.uk/regional- 20(4): 316–332.
information/economic-reports-and-research/
archive/a-minimum-income-standard-for-
remote-rural-scotland.html (accessed 22
August 2015).

Downloaded from lec.sagepub.com by guest on September 22, 2015

View publication stats