POVERTY AND EQUITY GLOBAL PRACTICE Pinpointing Poverty in Europe New Evidence for Policy Making Kenneth Simler Pinpointing Poverty in Europe New Evidence for Policy Making Kenneth Simler © 2016 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 1 2 3 4 19 18 17 16 This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the ­following conditions: Attribution—Please cite the work as follows: Simler, Kenneth. 2016. “Pinpointing Poverty in Europe: New Evidence for Policy Making.” World Bank, Washington, DC. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered content or error in this an official World Bank translation. The World Bank shall not be liable for any ­ translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by The World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third- party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to the Publishing and Knowledge Division, The World Bank, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights​ @­worldbank.org. Cover photo: Banet/Shutterstock.com. Used with permission; further permission required for reuse. Cover design: Bill Pragluski, Critical Stages, LLC. Contents Acknowledgments v Abbreviations vii Pinpointing Poverty in Europe 1 Introduction 1 The Project 2 The Results: The First EC–World Bank Poverty Maps 7 Using the Fresh Information 24 The Next Steps 29 Final Remarks 30 Notes31 Appendix A  PovMap and the Validation Study 35 Appendix B  Poverty Mapping Briefs 39 Bibliography 41 Boxes 1 Looking behind the Averages: The Power of Poverty Mapping 3 2 Validation Study of Small Area Poverty Estimation Methods 8 3 Two Examples of Non-EU Poverty Maps: Albania and Morocco 25 Maps 1 At-Risk-of-Poverty Rates, Estonia 12 2 Population Living below the Poverty Threshold, Estonia 13 3 At-Risk-of-Poverty Rates, Hungary 14 4 Population Living below the Poverty Threshold, Hungary 15 5 At-Risk-of-Poverty Rates, Latvia 17 6 At-Risk-of-Poverty Rates, Poland 18 7 At-Risk-of-Poverty Rates, Romania 20 8 At-Risk-of-Poverty Rates, the Slovak Republic 21 9 At-Risk-of-Poverty Rates, Slovenia 23 Table 1 Level of Aggregation and Data Sources for Poverty Maps, Phase 1, 2012–14 9 Pinpointing Poverty in Europe iii Acknowledgments This report summarizes the results of a large study of poverty in the European Union carried out by a core team comprising Kenneth Simler (Task Team Leader), Robin Audy, Kimberly Bolch, Alexandru Cojocaru, Céline Ferré, Indiana Taylor, Azhar Hussain, Maureen Itepu, Sandor Karacsony, Kadeem Khan, Joost de Laat, Peter Lanjouw, Lei Pan, Ericka Rascon, Thomas Sohnesen, Roy van der Weide, and Qinghua Zhao. The team worked under the overall guidance of Mamta Murthi, Arup Banerji, Benu Bidani, Carolina Sánchez, Katarina Mathernova, and Christian Bodewig. The team thanks Robert Zimmermann for doing a masterful job of synthesizing the project’s outputs and editing the final document. The project was financed by the European Commission through the Europe 2020 Programmatic Trust Fund. The team thanks its European Commission counterparts for guidance and support, particularly Georg Fischer, Rudolf Niessler, Lewis Dijkstra, Enrica Chiozza, and Bartek Lessaer. The team also thanks the technical experts on the project’s Scientific Steering Committee— Bart Buelens, Chris Elbers, Stefano Falorsi, Denisa Florescu, Javier Gallego, Martin Karlberg, and Isabel Molina—for their detailed and constructive critiques of the analysis. ­ The team also thanks colleagues in national statistical institutes who were partners in this study, including Julia Aru, Kaja Sõstra (Statistics Estonia); Judit Dobszayné Hennel, Éva Ménesi, Ildikó Merkl (Hungarian Central Statistical Office); Viktors Veretjanovs (Central Statistical Bureau of Latvia); Maciej Beresewicz, Tomasz Józefowski, Tomasz Klimanek, Jacek Kowalewski, Anna Małasiewicz, Andrzej Młodak, Marcin Szymkowiak, Łukasz Wawrowski (Central Statistical Office of Poland); Andreea Cambir, Nicoleta Caregea, Silvia Pisica (Romania National Institute of Statistics); Viera Doktoríková, Róbert Vlačuha (Statistical Office of the Slovak Republic), and Danilo Dolenc, Tomaž Smrekar (Statistical Office of the Republic of Slovenia). Pinpointing Poverty in Europe v Abbreviations ARoP at risk of poverty EB empirical best EC European Commission ELL Elbers-Lanjouw-Lanjouw EU European Union EU-SILC EU-Statistics on Income and Living Conditions LAU local administrative unit NSI National Statistical Institute NUTS Nomenclature of Territorial Units for Statistics TIPSE Territorial Dimension of Poverty and Social Exclusion in Europe Pinpointing Poverty in Europe vii Pinpointing Poverty in Europe Introduction More than 122 million people, almost 25 percent of European Union (EU) citizens, are at risk of poverty or social exclusion. The funding and targeting of basic services and of  programs to reduce poverty and social exclusion depend on the availability of detailed knowledge of the disparities in socio- economic ­well-being across and within EU member states. However, the poor are spread about unevenly in the EU across municipalities, districts, and regions in ways that are not entirely known. Because the World Bank and the EU have the common objective of build- ing competitive and sustainable economies and reducing poverty and social exclusion and because of the World Bank’s extensive experience, the EC and the Bank entered into an agreement to assist EU member states in undertak- ing expert analysis of census and survey data to create accurate, detailed geo- graphical presentations of poverty. This note highlights the context and main findings of the ongoing European Commission (EC)–World Bank project, Poverty Mapping in the European Union, including the EU goals of reducing poverty and social exclusion and the difficulties of precisely identifying the poor in the EU. •• Outputs: The main project output has been monetary poverty maps on selected countries that acceded to the EU in or after 2004. In the first phase of the project, small area poverty maps were produced on Estonia, Hungary, Latvia, Poland, Romania, the Slovak Republic, and Slovenia. The maps portray sub-national geographical areas, such as municipalities, counties, and districts, and highlight the small areas most likely to exhibit the highest risk of poverty in each of these EU member states. This note describes the technical aspects of this effort and offers brief case studies of the development of the poverty maps in these seven EU member states. •• Insights: The new maps demonstrate clearly that the risk of poverty is het- erogeneous across small geographical areas. Moreover, there may be large numbers of poor people concentrated in small discrete areas within larger regions that show low overall poverty rates. For example, urban areas may have large numbers of poor, but low poverty rates, while rural areas may have few poor, but high poverty rates. Or a region with moderate rates of poverty may be composed of smaller districts that have a mix of very high and very low poverty rates. Thus, less detailed poverty maps may give evidence that is misleading and even contradicts the evidence revealed by more finely grained poverty maps. •• Impact: The new maps complement other information on the correlates of poverty that enhances regional policy and program design. They have effectively guided decision making and policy making at the sub-­ national and national levels in EU member states and assisted the EC in allocating national and regional development and financing program funds efficiently Pinpointing Poverty in Europe 1 to areas with the greatest need. As the national partners of the project, gov- ernment authorities, mainly experts at national statistical institutes and ministries, have benefited from substantial knowledge and skill spillovers about poverty mapping through the project. The spillovers have gained prominence through academic research by catalyzing greater communica- tion and collaboration. The Bank, meanwhile, has developed new expertise on poverty mapping in more highly developed countries. Nordregio, a con- sortium of Nordic research centers, has been using the methodology to create maps elsewhere in the EU under the EU's Territorial Dimension of ­ Poverty and Social Exclusion (TiPSE) project. The EU has also begun con- ducting mapping exercises on other issues, including social exclusion. •• Knowledge and innovation: This note describes other applications of the maps in various countries, including countries outside the EU with which the World Bank has worked on similar initiatives. The project has advanced thinking about the methods and uses of poverty mapping, helped improve the mapping tools, and generated capacity building. Thus, a validation study carried out in Denmark and Slovenia through the project to test various alternative poverty mapping methods has fine- tuned understanding of these methods. Likewise, based on the experi- ence gained through the project, the World Bank has upgraded its PovMap interactive poverty mapping software. This note illustrates fresh goals in the ongoing project, initiatives envisioned on the horizon, and expectations of new outcomes and impacts. This includes new estimates of poverty in small sub-national geographical areas in Bulgaria, Croatia, the Czech Republic, and Lithuania; fresh maps on Hungary and Latvia based on more recent data; and exploration of refinements to the risk of poverty methodology that take into account sub-national variations in the cost of living across these areas. The note also offers examples of how poverty mapping has been used to inform policy, selected from among the dozens of non-EU countries in which poverty mapping has been carried out with the assistance of the World Bank. The Project As long as poverty and social exclusion exist, social progress may only advance haltingly because certain population groups are unable to participate fully in economic and social development. Thus, poverty and social exclusion are barriers to the achievement of overall territorial cohesion within countries and across regions. One of the five headline targets of Europe 2020—the strat- egy for the economic advancement of the EU in 2010–20—is therefore to reduce the number of people at risk of poverty or social exclusion by 20 ­million by the year 2020.1 The recent economic slowdown and the decline in the pace of the reduction of poverty and inequality are posing additional obstacles to rapid progress toward the realization of the targets. According to data of Eurostat, the EU statistics agency, more than 122  million people in the EU—almost 25  percent of EU 2 Pinpointing Poverty in Europe citizens—are at risk of poverty or social exclusion.2 Moreover, poverty and social exclusion are distributed unevenly across municipalities, districts, and regions. Indeed, it is spread about in ways that were clear to no one until quite recently. In the 2014–20 multiannual financial framework, the EU has budgeted €1 ­ rillion to boost growth and job creation, which are instrumental for reducing t­ poverty and social exclusion. Success depends on developing the correct policies and programs and targeting them effectively. It also depends on the availability of detailed knowledge of the disparities across and within member states, but espe- cially in those member states with high levels of poverty and social exclusion. Identifying the poor accurately helps in delivering antipoverty relief and basic services and guiding programs aimed at eliminating poverty and shar- ing prosperity more equitably. To help EU member states take advantage of the EU funds available for efforts to reduce poverty and social inclusion most efficiently, the EC decided to develop small area poverty maps on each of the EU member states so that the countries could address problems in their need- iest regions. It was known that poverty maps allow a focus in greater detail on the spatial distribution of poverty at the local level. However, the sub-national data avail- able to the EC has typically had only limited value for this purpose. Moreover, the EC lacked the in-house expertise and specialized resources to produce useful maps of poverty at a scale meaningful for more effective targeting. It therefore approached the World Bank because of the Bank’s extensive expe- rience in poverty mapping and because the Bank and the EU have the com- mon objective of building competitive and sustainable economies and reducing poverty and social exclusion. Originally, the EC requested the World Bank to help create maps of poverty in all EU member states, but, for practical and institutional reasons, the task was narrowed to drawing up maps on the more recent EU members emerging among the former transition countries of Central and Eastern Europe (see box 1). The EC Directorate-General for Employment, Social Affairs, and Inclusion and the EC Directorate-General for Regional and Urban Policy entered into a formal agreement with the World Bank in 2011. The main output the EC requested from Box 1  Looking behind the Averages: The Power of Poverty Mapping These maps of the risk of poverty rates in the Slovak Republic illustrate the information gains provided by small area estimation. The top panel shows the sub-national risk of poverty rates at the level of the four oblasts. Oblasts in the Slovak Republic correspond to the “NUTS 2” territorial division in the EU, and is the sub-national level for application of regional policies, including financial support from the EU.3 The oblast map shows a clear east-west gradient in poverty rates, with the poorest oblast in the east near the border with Ukraine and the least poor in the west, home of the national capital Bratislava and the borders with Austria and the Czech Republic. This east-west pattern of poverty risk appears in most central and eastern European countries. box continues next page Pinpointing Poverty in Europe 3 Box 1  Looking behind the Averages: The Power of Poverty Mapping (continued) Oblasts (NUTS 2) 6.3 6.3–13.2 13.2–14.1 14.1–16.7 Regions (NUTS 3) 6.2 6.2–10.6 10.6–11 11–14.2 14.2–16.3 16.3–18.3 18.3–19.8 19.8–21 box continues next page 4 Pinpointing Poverty in Europe Box 1  Looking behind the Averages: The Power of Poverty Mapping (continued) As one disaggregates to the eight regions of the Slovak Republic (NUTS 3) in the middle panel, regional differences within the oblasts are revealed. The east-west pattern is still maintained, but differences in poverty rates within oblasts appear, especially in the eastern part of the country. Districts (LAU 1) 3.4%–6% 6.1%–9.4% 9.4%–11.7% 11.71%–14.1% 14.11%–18% 18.1%–21.97% 21.98%–26.2% 26.21%–31.56% Source: Estimates using data from the 2011 EU-SILC and the 2011 Population and Housing Census collected by the Statistical Office of the Slovak Republic. Boundary map courtesy of Geodesy, Cartography and Cadastre Authority of the Slovak Republic. Drilling down deeper to the 79 districts of the Slovak Republic (LAU 1, bottom panel) reveals considerable heterogeneity in risk of poverty rates that was obscured when looking at the more aggregate regions. The more granular information on poverty opens up possibilities for more accurately targeted interventions and more cost-effective policies to reduce poverty. the Bank was technical assistance in the construction of monetary poverty maps on countries that had acceded to the EU in or after 2004. The maps were to portray sub-national geographical areas, such as municipalities, counties, and districts, and highlight the small areas most likely to exhibit the highest risk of poverty in each member state and, thereby, provide a comprehensive picture of poverty in this part of Europe (World Bank 2014a). The objective was to help guide decision making and policy making at the sub-national and national levels in EU member states and assist the EC to allocate European Structural and Investment Funds efficiently to programs serving the areas with the greatest need. In cooperation ­ with the EC, especially the Directorate-General for Regional and Urban Policy, Pinpointing Poverty in Europe 5 a consortium of Nordic research centers (Nordregio) was to perform similar work in the rest of the EU, using the same small area methodology as the World Bank in some countries and using tax registry data in others. Government authorities in the individual European countries, mainly national statistical institutes (NSIs), but also ministries, are the national partners of the project. Despite common EU goals (such as income convergence across and within member states) and technical standards (such as the standards associated with EU household surveys), the engagement with each EU member state has  been unique. Working arrangements, analytical approaches, and skill transfers have been tailored to the context of each country, taking into account the considerable variation in staff expertise, the time available to work on the project, data readiness, and procedures for accessing sensitive and confiden- tial data. The customized approach to each country also extended to the dis- semination of the results through official publications and web sites and also outreach to local media outlets. What Is Poverty Measurement in the EU? Low income is not a failsafe indicator of poverty experienced across ­geographical areas because, similar to the rates of poverty and social exclusion, which vary widely across EU member states, there is also considerable vari- ability in living standards across countries and across regions within countries. Relative poverty is therefore more generally used in the European context and is normally specified as income below a minimum acceptable level. The at-risk- of-poverty (ARoP) rate is the key indicator of such relative income poverty in relation to the Europe 2020 targets. It is the population share of people whose yearly disposable income is below the ARoP threshold, which is 60 percent of the national median yearly disposable income of household members after social transfers.4 This means that each of the countries calibrates their own ARoP rates according to the national median income. Mapping these rates illustrates patterns of within-country relative poverty more effectively. In previous years, the EC had to rely on less detailed data and maps for program planning and the allocation of EU funds. The ARoP rate for large aggregated areas likewise fails to capture the interaction in finer detail such as within cities, between small urban and rural areas, or in dispersed regions. To understand the role of poverty in wider processes of territorial cohesion, it  is important to establish patterns and explore processes at an even finer geographical scale. More precise information about this within-country dis- tribution of poverty was therefore sought through the new mapping methods that have figured in the EC–World Bank poverty mapping project. Constructing the Poverty Maps The maps are constructed using a variation of the small area estimation method- ology that was developed to allow accurate estimates of monetary poverty and inequality at lower levels of disaggregation (see Elbers, Lanjouw, and Lanjouw 2002, 2003). At least two datasets are required to implement the method. One is a detailed household survey that includes a welfare measure, typically income or 6 Pinpointing Poverty in Europe consumption per capita or per adult equivalent. The other data source is a national census or, alternatively, a large national survey that includes a significant share of the country’s population. The method typically consists of three steps. The first step involves the close scrutiny of a population census and a household living standards survey or a household expenditure survey to ­ identify a set of candidate socioeconomic variables (such as educational attainment, housing characteristics, or access to basic services) that are avail- able in both the census and the survey. The analysts examine the data sources to ensure that the variables are comparable, that is, that they are worded in a similar way, including in the response options. In the second step, the household survey data and the variables that have been identified in step 1 are used to develop a regression model of disposable household income as a function of individual, household, and community characteristics.5 The standard EU measure of disposable household income is used, which includes income after tax payments and the receipt of social transfers. The disposable income is equivalized by adjusting it according to the size and age composition of households.6 The independent or explanatory variables commonly include the demographical, educational, and employ- ment characteristics of households, as well as community characteristics such as population density and physical access to infrastructure. In the third step, the set of estimated parameters obtained from the ­regression models is applied to the set of matching variables in the population census data to yield the predicted income of each household in the census. Once such information is available, summary measures of poverty and inequality are estimated for sets of census households across any geographical unit with a sufficient number of households to obtain reliable estimates, such as provinces, counties, districts, subdistricts, or municipalities. Statistical tests are then performed to assess the reliability and precision—the standard errors and confidence intervals—of the various measures that have been produced. After the exercise has been completed for all sub-national units or groups of units, the resulting estimates of poverty and inequality and the related stan- dard errors at a variety of levels of geographical disaggregation are projected onto geographical maps using geographic information system mapping tech- niques to facilitate the presentation and visual analysis of patterns. The Results: The First EC–World Bank Poverty Maps One of the initial steps in the project, from February through November 2012, was a methodological validation study proposed by the EC to test slightly different approaches to small area poverty estimation and identify a preferred method for estimating the EU poverty maps (box 2). The validation study was guided by a steering committee, which was composed of leading experts in small area estimation at European universities and NSIs and senior staff at Eurostat, the EC Directorate-General of Employment, Social Affairs, and Inclusion, and the EC Directorate-General of Regional and Urban Policy. Pinpointing Poverty in Europe 7 Box 2  Validation Study of Small Area Poverty Estimation Methods The study used validation methods to compare slightly different approaches to the estimation of poverty in small areas. The principal comparison was between the Elbers-Lanjouw-Lanjouw (ELL) method that had been the main approach used by the World Bank and its partners for the past 15 years or so and an extension of the ELL method proposed in the statistical literature as an enhancement and widely known as the empirical best (EB) method (Elbers, Lanjouw, and Lanjouw 2002, 2003; Molina and Rao 2010). The advantage of the EB appeared of marginal conse- quence in low-income countries such as those in which the Bank had been applying the method- ology for years already and where sample surveys typically cover only a small share of the areas in which poverty is being estimated, but it seemed potentially significant in the context of the survey sampling schemes in European countries. The validation study was carried out in Denmark and Slovenia because, in addition to the EU Statistics on Income and Living Conditions database (EU-SILC) and population census data, complete sets of household income data were available through the tax registries in these coun- tries and could be linked to survey and census data on individuals and households. The meth- odologies were validated by using them to estimate the poverty rates in small areas as if only the EU-SILC and census data were available and then comparing the resulting rates with the poverty rates calculated from the income tax data, which served as the benchmark reference. By testing selected variations of the ELL and EB on areas in which income data were available for the entire population, it was possible to determine which approaches came closer to the poverty rates as measured directly by the income registry data. Aided by the guidance of experts on the steering committee, the study was able to quantify the impact of adding the EB to the ELL framework. The systematic testing also furthered the under- standing of other elements of poverty mapping. The validation study found that the six estimation methods that had been tested correctly iden- tified 80–90 percent of the municipalities ranked among the poorest based on the income tax registry data. (Municipalities are the first and second local administrative units [LAUs]—LAU 1 and LAU 2—in Denmark and Slovenia, respectively.) This thus served to support a critical ­ self-examination of the poverty mapping methods used within the World Bank and their appli- cability in a European setting. The  steering committee was chaired by the directors of these Directorates- General. The exercise thus benefited not only from the knowledge of technical experts, but also from users of this type of information in the EC. Informed by the results of the validation study, production of the actual small area poverty maps was carried out in EU member states during 2012–2014. Constructing the maps required access to data not available to the World Bank through Eurostat. This typically involved conducting analyses on-site at NSIs in individual EU member states, with the widely varying involvement of NSI analysts. This also meant that the preferred approach of relying on survey and census microdata, that is, data at the household and individual levels, was sometimes not possible. In such cases it became necessary to resort to using 8 Pinpointing Poverty in Europe aggregate, or area-level, data to estimate poverty rates in small areas. After the initial results had been  obtained, the draft poverty maps were shared with local experts as an  informal validation and to open discussions about the likely underlying causes of the observed spatial patterns in poverty and about potential policy remedies. The analysis was then revisited and refined, taking into consideration the comments of the local experts. During the first project phase, small area poverty maps were produced for seven EU member states: Estonia, Hungary, Latvia, Poland, the Slovak Republic, and Slovenia, which joined the EU in 2004, and Romania, which joined in 2007.7 Table 1 summarizes the level of spatial disaggregation achieved in each of the poverty maps, using both the Nomenclature of Territorial Units for Statistics (NUTS)–LAU classification system and the common English names for the respective regions in each country. The table also indicates the main data sources and whether the maps are based on microdata or aggregate data.8 It was widely assumed that the spatial variation in poverty was considerable within the highly aggregated NUTS 2 regions that are used for EU regional policies and programs. However, concrete and sufficiently precise informa- tion about the nature of these regional disparities was lacking. The new higher-resolution poverty maps based on the small area poverty mapping methodology provide more finely grained information on ­sub-­national variations in poverty than was previously available (also see a ­ ppendix A). The greater geographical disaggregation of the new maps permits more precise identification of the parts of these larger regions that exhibit particularly high rates of poverty and require greater attention in development assistance and poverty reduction programs, whether through EU funds or other resources. Combined with data on population size, they also provide information on where most of the poor are located. The maps likewise complement other information on the correlates of poverty that enhances regional policy and Table 1  Level of Aggregation and Data Sources for Poverty Maps, Phase 1, 2012–14 Country Level of poverty map Data sources and comments Estonia Municipalities, with some grouping of small Microdata from 2012 EU-SILC and 2011 municipalities; total of 132 areas, corresponding population census approximately to LAU 2 Hungary General electoral districts; total of 174 areas, Microdata from 2005 EU-SILC and 2005 corresponding approximately to LAU 1 microcensus Latvia Statistical regions; total of 6 areas, Aggregate area-level data from 2012 EU-SILC corresponding to NUTS 3 and 2011 population census Poland Statistical subregions; total of 66 areas, Aggregate area-level data from 2011 EU-SILC corresponding to NUTS 3 and 2011 population census Romania Counties and Bucharest; total of 42 areas, Microdata from 2011 EU-SILC and 2011 corresponding to NUTS 3 population census Slovak Districts; total of 79 areas, corresponding to Aggregate area-level data from 2011 EU-SILC Republic LAU 1 and 2011 population census Slovenia Municipalities, with some grouping of small Microdata from 2010 EU-SILC and 2011 municipalities; total of 195 areas corresponding population census approximately to LAU 2 Source: World Bank 2014a. Pinpointing Poverty in Europe 9 program design. The contrast between maps resulting from lower and higher resolution in poverty data is evident in the EU poverty maps produced under this project. Highlight: Producing and Using Poverty Maps in Estonia The poverty mapping exercise in Estonia is an outstanding example of how the poverty mapping methodology—combined with exceptional institutional cooperation and commitment by national officials—made it possible to pro- duce extremely detailed poverty estimates. To help achieve the EU’s Europe 2020 target to reduce the number of people living at risk of poverty or social exclusion by 20 million by the year 2020, the government of Estonia has set a national goal of reducing the ARoP rate from 17.5  percent in 2010 to percent by 2020, which is equivalent to an absolute reduction in the 15.0  ­ number of the poor and socially excluded of 36,248 people (Estonia, Government Office 2014).9 To inform the steps needed to reach this goal, Statistics Estonia, with the assistance of the World Bank, decided to construct detailed poverty maps of the country. With a population of around 1.3 million, Estonia is a small country, and the EU NUTS 2 classification covers the entire national territory (as is also the case in Cyprus, Latvia, Lithuania, Luxembourg, and Malta). Clearly, this is inadequate for identifying sub-national variations in poverty. Tallinn, the capital, enjoys high average living standards, but isolated islands and the northeast are relatively much poorer. Estonia is divided as follows: five groups of counties (NUTS 3), 15 counties (LAU 1), 226 local government units (LAU 2), and circa 5,000 settlement units. The EU-SILC survey in Estonia is based on a sufficiently large sample to produce reasonably precise direct estimates of the risk of poverty within the counties (LAU 1), but not within the local government units (LAU 2). Some of the local government units have such small populations—fewer than 500 households, for example—that precise poverty estimates are not possible even if one uses the poverty mapping methodology based on combining sur- vey and census data. Precise poverty estimates are crucial for ranking the pov- erty rates of local government units reliably, an important step in the practical targeting of antipoverty initiatives. After consultations and some experimentation with preliminary find- ings, representatives of Statistics Estonia and the World Bank agreed on a system for grouping sparsely populated local government units with other local government units to produce trustworthy poverty estimates. The cri- teria for the grouping were based on the following principles: (1) each area should have at least 1,000 households, (2) local government units that belong to the same group should be located within the same county so that the estimates would remain additive, (3) local government units that belong to the same area have similar socioeconomic characteristics, and (4) the local government units should have common borders. The munic- ipality of Tallinn is sufficiently large—a population of over 400,000—that it was possible to disaggregate it on the basis of the municipality’s eight 10 Pinpointing Poverty in Europe urban administrative districts. In addition to these eight districts of Tallinn, the division resulted in 124 groups, each consisting of 1–6 local government units. The project made use of the data from the Estonian Social Survey of 2012 (Estonia’s implementation of the EU-SILC), in which income variables refer to 2011, and data from the housing, household, and personal questionnaires of the Population and Housing Census as of December 31, 2011. Estonian privacy laws allow no access to census microdata by outside analysts before ­ the publication of official reports. This could have blocked or delayed the efforts to produce the poverty maps. However, Statistics Estonia was so com- mitted to completing the maps that it adjusted its existing work program and assigned two of its experts to work on the project. In addition to contributing their technical proficiency and local knowledge, the Statistics Estonia experts had sole responsibility for handling the census microdata, thereby ensuring compliance with the privacy laws. The poverty maps confirm existing knowledge about poverty in Estonia, but also reveal new insights. Thus, for example, panel a, map 1 shows graph- ically that direct estimates from the Estonian Social Survey suggest there is some poverty heterogeneity across counties, but the estimates for groups of local government units illustrated in panel b, map 1 reveal considerably more variation in poverty incidence within counties. Previous surveys indi- cated that counties in the northeast and southeast had the highest rates of poverty, but the more disaggregated estimates reveal contrasts between the lower poverty estimates in large cities and surrounding areas relative to other parts of the counties. There are also local government units with high poverty incidence in counties that otherwise have low to moderate overall poverty rates. The new poverty maps demonstrate that poverty rates in the poorest areas and poverty rates in the least poor areas differ by a factor of almost seven, ranging from 6 percent to 40 percent. The EU poverty rate for the NUTS 2 area, that is, the entire country, was 17.6 percent. As Estonia is a single NUTS 2 region, finer geographic disaggregation is required to show sub-national variations in poverty. Estonia's EU-SILC is representative at the LAU 1 level (15 counties). The small area estimation methodology allows further disaggregation to the LAU 2 level (local government units), revealing smaller pockets of poverty that require more well-targeted policy attention. Targeting poor areas alone can have limitations, however. Policy makers have an interest both in areas where poverty is high and in areas that have the highest number of poor people. These two are not always the same: areas that are poor may also be sparsely populated, whereas large cities tend to have low poverty rates, but large numbers of poor people because of large populations. This contrast is readily seen in the large concentration of poor people in the cities of Tallinn and Tartu and the surrounding areas (map 2). Even though the ARoP rates are low, these cities are home to a large share of Estonia’s poor. On the other hand, Ida-Viru county in the northeast has both high poverty rates and a large number of poor people. This is most pronounced in the for- mer military-industrial city of Narva, Estonia’s third largest city, which has Pinpointing Poverty in Europe 11 Map 1  At-Risk-of-Poverty Rates, Estonia a. Countries (LAU 1) 12.6 12.6 – 15.8 15.8 – 17.63 17.630 – 18.3 18.3 – 21.7 21.7 – 24.7 24.7 – 25.1 25.1 – 26.9 b. Groups of local government units (LAU 2) 5.8 – 9.4 9.4 – 14.4 14.4 – 17.2 17.2 – 18.3 18.3 – 21.7 21.7 – 25.7 25.7 – 29.9 29.9 – 34.3 Source: Estimates using data from the 2012 Estonian Social Survey and the 2011 Population and Housing Census collected by Statistics Estonia. 12 Pinpointing Poverty in Europe Map 2  Population Living below the Poverty Threshold, Estonia 326 – 1,562 1,562 – 4,934 4,934 – 9,841 9,841 – 17,826 Source: Estimates using data from the 2012 Estonian Social Survey and the 2011 Population and Housing Census collected by Statistics Estonia. among the highest poverty rates, despite an urbanized and well-educated population. This is an outcome of the disappearance of the manufacturing base in these places. Six Other Poverty Mapping Exercises The World Bank and NSIs carried out initial poverty mapping exercises in six other countries besides Estonia. Summaries of these experiences are provided below. More detail is supplied in appendix B. Hungary In 2013, 3.3 million people, or 33.5 percent of the population, were at risk of poverty or social exclusion in Hungary.10 To help achieve the Europe 2020 target to reduce the number of people living at risk of poverty or social exclu- sion by 20  million by the year 2020, the government of Hungary has set a national goal of reducing the number of the poor or socially excluded by 450,000 people (Hungary, Ministry for National Economy 2015). To inform the steps needed to reach this goal, the Hungarian Central Statistical Office, with the assistance of the World Bank, constructed detailed poverty maps of the country (map 3).11 The poverty maps corroborate existing knowledge about poverty, but also reveal much more. For example, previous surveys showed that the highest rates of poverty occur in the northeast (map 3, panel a). The statistical Pinpointing Poverty in Europe 13 Map 3  At-Risk-of-Poverty Rates, Hungary a. Seven planning and statistical regions (NUTS 2) 7 7–8.82 8.82–13.2 13.2–16.7 16.7–17.9 17.9–18.5 18.5–20.6 b. Subregions (LAU 1) 4.72–7.7 7.7–10.46 10.46–12.7 12.7–13.2 13.2–17.98 17.98–22.4 22.4–26.98 26.98–31.62 Source: Estimates using data from the 2005 EU-SILC and 2005 Microcensus collected by the Hungarian Central Statistical Office. 14 Pinpointing Poverty in Europe subregion (LAU 1) poverty map shows that many of the subregions in the northeast face an elevated risk of poverty, although Eger, Miskolc, and Nyíregyháza stand out as areas with only moderate poverty incidence (map 3, panel b). In contrast, Southern Transdanubia is heterogeneous, comprising low poverty incidence in subregions such as Pécs and relatively high poverty incidence in nearby Siklós, Sellye, Szigetvár, or Szentlőrinc. More generally, there is a much higher degree of heterogeneity in poverty incidence in the statistical subregions relative to the estimates available directly from the EU-SILC survey for the statistical regions. The density of the population below the poverty threshold is, on average, also higher in districts with higher poverty incidence (map 4). Nonetheless, the rankings of subregions by poverty incidence and by poverty density are quite divergent; districts such as Budapest, Pécs, Gyõr, and Székesfehérvári have a low poverty incidence, but, given their large populations, they account for a large share of the total population below the ARoP threshold. Latvia To help reach the EU’s Europe 2020 target, the government of Latvia has set a national goal of reducing by 121,000 the number of people at risk of poverty or living in households with low work intensity (Latvia, Ministry of ­ Economics 2015).12 The EC has relied on the NUTS 2 territorial classification to determine eligibility for aid from European Structural and Investment Funds and for ­ Map 4  Population Living below the Poverty Threshold, Hungary 894–6,878 6,878–15,905 15,905–35,587 35,587–93,100 Source: Estimates using data from the 2005 EU-SILC and 2005 Microcensus collected by the Hungarian Central Statistical Office. Pinpointing Poverty in Europe 15 program planning. In smaller countries such as Latvia, the NUTS 2 classifica- sub-national tion corresponds to the entire national territory, that is, with no ­ divisions (map 5, panel a). The EU-SILC in Latvia is representative at the statistical region level (NUTS 3), and the Central Bureau of Statistics reports ARoP estimates at that level. Using small area estimation techniques, it was possible to improve the pre- cision of these poverty estimates for the six regions of Latvia. According to these estimates, there is quite a bit of heterogeneity across regions (map 5, panel b). Pieriga and Riga in north central Latvia are the only regions where the incidence of poverty is below the national average.13 Estimates of poverty in other regions range from 21 percent to as high as to 28 percent in Latgale in southeastern Latvia. Riga Region has the lowest estimated risk of poverty, 12.9 percent, which is less than one-half the risk in the poorest region, Latgale (28.0 percent). However, because Riga Region is much more populous, each of the two regions has approximately 85,000 people living at risk of poverty. While the remaining four regions have a somewhat heterogeneous poverty incidence, they all have a similar number of poor people, in the range of 50,000–60,000. Poland About 10 million people, or 26.7 percent of the population, were at risk of poverty or social exclusion in Poland in 2012.14 The government has set a national goal of reducing the number of the poor and socially excluded by 1.5  million people (Poland, Ministry of Economy 2015). To further the steps  needed to achieve this goal, the Central Statistical Office of Poland and the Statistical Office of Poznań, with the assistance of the World Bank, constructed detailed poverty maps of the country (map 6). ­ Previous surveys have shown higher poverty rates in the southeastern voivodships (in particular Lubelskie and Swietokrzynskie) and Lubuskie in the west (map 6, panel a). The more detailed poverty map shows that most of the statistical subregions with the highest ARoP rates are in the east and southeast of the country (map 6, panel b). However, the subregional map also reveals visible differences in poverty incidence between the low pov- erty cities (Krakow, Poznań, Warsaw, and Wroclaw) and surrounding lower population-density areas with higher poverty incidence. The subregional ­ map also highlights heterogeneity in poverty estimates within voivodships; the voivodship of Pomorskie includes Trojmiejski, with an estimated pov- erty incidence of 7 percent, alongside Slupski and Starogardzki, both with poverty rates in excess of 20 percent. There is an overall positive and rela- tively strong relationship across subregions between poverty incidence and the total number of individuals living below the poverty threshold. The highest absolute numbers of the poor are located in the east and southeast. At the same time, Warsaw stands out because it has the lowest estimated poverty incidence among statistical subregions, but, on account of its size, it ranks 27th among 66 subregions in the absolute number of individuals living below the poverty threshold. This is not observed in other large cities 16 Pinpointing Poverty in Europe Map 5  At-Risk-of-Poverty Rates, Latvia a. NUTS 2 classification 19.2 b. Statistical subregions (NUTS 3) 12.93 12.93 – 15.93 15.93 – 19.1 19.1 – 23 23 – 24.9 24.9 – 28.1 Source: Estimates using data from the 2012 EU-SILC and 2011 Population and Housing Census collected by the Latvia Central Statistical Bureau. Pinpointing Poverty in Europe 17 Map 6  At-Risk-of-Poverty Rates, Poland a. Voivodships (NUTS 2) 12.3 12.3 – 13.5 13.5– 14.8 14.8– 17.1 17.1– 18.9 18.9– 19.3 19.3– 21.7 21.7– 26 b. Statistical subregions (NUTS 3) 6.3–8.7 8.7– 12.6 12.6– 15.1 15.1– 16.9 16.9– 19.8 19.8– 22.9 22.9– 26.1 26.1– 30.2 Source: Estimates using data from the 2011 EU-SILC and the 2011 Census of Population and Housing collected by the Central Statistical Office of Poland. 18 Pinpointing Poverty in Europe such as Krakow, Poznań, or Wroclaw, where poverty incidence and the absolute number of poor both appear to be low. Romania The government has set a national goal of reducing the number of poor and socially excluded by 580,000 people (Romania, Ministry of Economy, Trade, and Tourism 2015).15 To achieve the progress needed to realize this goal, the National Institute of Statistics, with the assistance of the World Bank, con- structed detailed poverty maps of the country. Previous surveys have shown the highest rates of poverty in the Northeast Region (map 7, panel a). The county-level poverty map (map 7, panel b) shows that all the counties in that region, with the exception of Bacău, have elevated ARoP rates. In contrast, the South Region is heterogeneous, comprising coun- ties with high poverty rates, such as Călărași and Teleorman, and counties with relatively low poverty rates, such as Prahova. Similarly, Cluj County has the sec- ond lowest poverty rate in Romania (after Bucharest), but its neighboring coun- ties in the Northwest Region (Bistrița-Năsăud, Maramureș, Sălaj, and Satu Mare) have higher than average poverty rates. Despite its lower poverty rate, Cluj has more people at risk of poverty than Sălaj, and Bucharest has more people at risk of poverty than six other counties. The maps suggest a complementary approach to the allocation of resources for poverty reduction: Bucharest and other urban areas might be given greater attention because many of the poor live in relatively rich counties. The maps clearly illustrate that the Northeast, especially Botoșani, Iași, and Suceava, have high poverty rates and large numbers of poor people and should be given high priority by either criterion. The Slovak Republic In 2012, 1.1 million people, or 20.5 percent of the population, were at risk of poverty or social exclusion in the Slovak Republic.16 The government has set a national goal of reducing the number of the poor and socially excluded by 170,000 people (the Slovak Republic, Ministry of Finance 2015). As a step toward this goal, the Statistical Office of the Slovak Republic, with the assis- tance of the World Bank, constructed detailed poverty maps of the country. Previous surveys have shown the highest poverty rates in the eastern oblasts (map 8, panel a). Yet the district map (map 8, panel b) reveals considerably more heterogeneity in the incidence of poverty. In the east of the country, the highest poverty incidence appears to be concentrated primarily among the districts along the border with the center (Kežmarok, Poprád, and Rožňava) and in the districts along the border with Ukraine (Snina and Sobrance), while the incidence of poverty is relatively low in Košice. At the same time, districts such as Poltár, Revúca, and Rimavská Sobota in the center also show high incidence, although incidence is moderate in the center overall. In 23 of 79 districts, the district poverty estimates are statistically different from the estimates for the oblasts in which the districts are located. Even though the poverty headcount is generally correlated with the absolute size of the poor population, this is not always the case. Districts such as Nitra, Trenčín, Trnava, Pinpointing Poverty in Europe 19 Map 7  At-Risk-of-Poverty Rates, Romania a. Development Regions (NUTS 2) 2.7 2.7– 15.9 15.9– 18.5 18.5– 22.6 22.6– 26.5 26.5– 29.7 29.7– 30.9 30.9– 33.7 b. Countries (NUTS 3) 3.9 3.9– 17.46 17.46– 22 22– 22.9 22.9– 27.75 27.75– 29.7 29.7– 32.17 32.17– 43.66 Source: Estimates using data from the 2011 EU-SILC and 2011 Population and Housing Census collected by the Romania National Institute of Statistics. 20 Pinpointing Poverty in Europe Map 8  At-Risk-of-Poverty Rates, the Slovak Republic a. Oblasts (NUTS 2) 6.3 6.3–13.2 13.2–14.1 14.1–16.7 b. Districts (LAU 1) 3.4–6 6–9.4 9.4–11.7 11.7–14.2 14.2–18 18–22 22–26.2 26.2–31.6 Source: Estimates using data from the 2011 EU-SILC and the 2011 Population and Housing Census collected by the Statistical Office of the Slovak Republic. Boundary map courtesy of Geodesy, Cartography and Cadastre Authority of the Slovak Republic. Pinpointing Poverty in Europe 21 and Žilina have low poverty rates, but rank relatively high among districts in the absolute size of the poor population. Meanwhile, districts such as Krupina, Poltár, Sobrance, and Stropkov have higher poverty headcounts, but represent only a small share of the total population living below the ARoP threshold. Slovenia In 2012, 392,000 people, or 19.6  percent of the population, were at risk of poverty or social exclusion in Slovenia.17 The government has set a national goal of reducing the number of poor and socially excluded by 40,000 people (Slovenia, Ministry of Finance 2014). To achieve progress toward this goal, the Statistical Office of the Republic of Slovenia, with the assistance of the World Bank, constructed detailed poverty maps of the country. Previous surveys have shown higher poverty rates, albeit only marginally, in the east and southeast (map 9, panel a). Most of the municipalities with the highest ARoP rates are in the east. However, some municipalities in western Slovenia (Kobarid and Tolmin in Goriška) have a high incidence of poverty (map 9, panel b). There is considerable heterogeneity in incidence across the east. For instance, in the Savinja region, incidence ranges from 12 percent in the municipalities of Celje, Valenje, and Vojnik to about 25 percent in Luče, Rogatec, and Solčava. Similarly, the Drava region combines municipalities at low incidence, such as Lenart and Starše, and municipalities at high inci- dence, such as Cerkvenjak and Juršinci. There is an overall negative relation- ship between municipal poverty incidence and the total number of individuals living below the poverty threshold. Thus, municipalities such as Luče, Rogatec, and Solcava account for few poor individuals, despite relatively high poverty rates. Meanwhile, municipalities such as Koper, Kranj, and Ljubljana show low incidence, but rank high in the absolute number of the poor on account of their relatively large populations. Dissemination and Outreach Activities The guidance provided by the World Bank has generated substantial knowl- edge exchange and skill spillovers to the project steering committee and the NSI counterparts. The spillovers have likewise gained prominence through academic research by catalyzing greater communication and collaboration. After the production of the first draft of the poverty maps in a country, the Bank has shared the initial findings in seminars among local experts and govern- ment authorities. This has served as an informal validation of the maps, and also as a catalyst for dialogue on the causes for the observed patterns of poverty. After finalizing the poverty maps the World Bank team prepared short technical reports on the results, often co-authored with NSI counterparts. These have been specifically requested by Eurostat as a deliverable to support the evaluation and documentation of the technical soundness of the analyti- cal work. The technical reports18 each describe (1) the spatial divisions in the country and the target level of the poverty map, (2) the modeling approach, (3) the specific data sources, and (4) the results of the small area estimations, including regression results, point estimates and confidence intervals or 22 Pinpointing Poverty in Europe Map 9  At-Risk-of-Poverty Rates, Slovenia a. Macroregions (NUTS 2) 10.7 16.1 b. Municipalities (LAU 2) 12.3–15.9 15.9–18.56 18.56–20.69 20.69–21.7 21.7–25.1 25.1–27.47 27.47–31.51 31.51–39.23 Source: Estimates using data from the 2010 EU-SILC and 2011 Census of Population, Households, and Dwellings collected by the Statistical Office of the Republic of Slovenia. Pinpointing Poverty in Europe 23 ­tandard errors for poverty rates, and regression diagnostics. The reports s demonstrate the knowledge gains possible from the small area estimation approach. They also demonstrate the results achievable in less than ideal ­ circumstances, such as if census microdata are not available. In Estonia and Poland the reports were released as official publications of the respective national statistical institutes. To complement the technical reports, two-page summaries of the poverty map results were prepared for each country. These poverty mapping briefs, which are included in the appendix, provide information about the motivations for producing the poverty maps, summarize some of the main results, and out- line how the poverty maps may help in the design of policies and programs. Using the Fresh Information How Similar Information Has Been Used: The Experience of Non-EU Countries Understanding which areas have higher poverty rates can potentially allow the more accurate and efficient targeting of resources for development and poverty reduction. Indeed, poverty maps are to be used to help guide the allo- cation of EU funds. The maps may also force more thinking in sub-national and national deci- sion making and policy making on how best to balance the targeting of poor areas and poor people to combat poverty and social exclusion, and how to improve standards of living. For this purpose, it is important to understand why these areas are relatively poorer and address the associated issues. The reasons are likely to vary from place to place and may include inadequate infra- structure, lack of economic activity, and an insufficiently skilled work force. Led by cutting-edge research and widespread empirical application of the poverty mapping methodology, more than 60 countries have, since the late 1990s, gained experience in the small area estimation of poverty illustrated by poverty maps. This experience illustrates innovative ways to use the informa- tion provided by the maps. For example, poverty maps were combined with maps of market accessibility in Bangladesh to provide an ex ante estimation of the poverty reducing impact of building a new bridge across the Padma River to link the isolated southwest of the country with the more prosperous northern and eastern regions. In Vietnam, policy makers used the detailed poverty map to assess the targeting accuracy of Program 135, which aims to reduce poverty in ethnic minority areas. The assessment helped identify errors of inclusion and exclusion in program coverage, leading to a refine- ment in targeting criteria (Swinkels and Turk 2007). In Mexico, the govern- ment used poverty map results to guide the allocation of funds under the Habitat Program, which makes a series of integrated poverty reducing inter- ventions in urban areas with the highest concentrations of poor people (Lopez-Calva et al. 2007). Box 3 highlights the ways that poverty maps were used to inform development policies in Albania and Morocco. 24 Pinpointing Poverty in Europe Box 3  Two Examples of Non-EU Poverty Maps: Albania and Morocco Albania By late 2000, Albania had no reliable, nationally representative survey to assess the living condi- tions of the population adequately. In 2001, however, the Institute of Statistics, with donor sup- port, carried out the country’s first census since the end of the communist era. The following year, the institute implemented a nationally representative multitopic household survey. During the same year, the World Bank began to prepare a poverty assessment, and the government com- pleted the preparation of a Poverty Reduction Strategy Paper. The convergence of these events and the related data provided the ideal opportunity to construct small area estimates of poverty and inequality, as well as poverty maps. The maps clearly show how an analysis of poverty might improve because of the detail at a more finely disaggregated level (map B3.1). The aim of the creation of the poverty maps was to estimate poverty and inequality for each of the regions, districts, and municipalities or communes, and for each submunicipality of Tirana. This was because the poor are concentrated in rural and mountainous areas, but also in regions that are more well off, because the government was seeking to decentralize the delivery of services to local governments, which might develop independent strategies and interventions, and because the targeting of government antipoverty programs needed to be improved. All these aims show useful possible applications of the maps. Map B3.1  Poverty Maps, Poverty Headcount Ratio, Albania a. Regions b. Districts c. Communes and municipalities < 24% < 24% 24% – 28.6% < 24% 24% – 28.6% 28.6% – 36% 24% – 28.6% 28.6% – 36% 36% – 43% 28.6% – 36% > 36% > 43% > 36% Source: Instat 2004. box continues next page Pinpointing Poverty in Europe 25 Box 3  Two Examples of Non-EU Poverty Maps: Albania and Morocco (continued) Following the mapping exercise, individuals involved in the preparation of the maps and poten- tial users of the maps were interviewed. The interviews revealed that there were three ways in which the poverty maps were being applied: (a) as a benchmark against existing resource allocation criteria, for example, whether social-assistance block grants allocated according to previously established criteria correlate with current poverty rates; (b) as a tool in targeting public spending; and (c) for the provision of data to monitor the progress toward achieving the Millennium Development Goals. Several nongovernmental organizations have also relied on the poverty maps in supplying advisory services to local governments and donor agencies and in designing joint intervention strategies. The poverty maps have likewise been used to improve the prioritization of investments in secondary roads. Morocco The Moroccan government requested the support of the World Bank in 2002 to learn how to use poverty mapping techniques and produce a detailed, disaggregated map of poverty throughout the country. At the time, the Bank was scheduled to undertake a poverty report on the country. This enabled the emergence of an active dialogue on how to improve targeting in social spending. Thus, there was already substantial interest in the issues. The task team made a significant effort to situate the report with appropriate counterparts, such as pivotal people at ministries. When the first poverty map and other analyses were completed, key policy makers had already been sensi- tized to the significance of poverty maps. The need for finer geographical targeting of the poor was intensified because of two factors. First, because of the threat of terrorism, renewed attention was being focused on the vulnerability of the urban poor and the phenomenon of rural-urban migration. More information was needed on pockets of poverty and vulnerability. Second, because of a growing budget deficit, there was strong pressure to make public expenditures more effective. The desire to minimize benefit leak- age to the nonpoor had gained strength. The poverty mapping program proceeded for almost a year on technical and policy levels. The technical focus permitted a transfer of the capacity to construct poverty maps to local experts, and the policy focus stimulated interest in the potential utility of the maps for policy purposes; if the maps were to have an impact on policy, they had to appeal to decision makers who would be able to apply them to realize the country’s policy goals. The Bank’s poverty report was published two months after the government’s poverty map report (Planning Commission 2004; World Bank 2004). Both publications provided analysis of the spatial aspects of poverty and inequality, but also other issues. Eventually, the government con- structed a second, updated poverty map. box continues next page 26 Pinpointing Poverty in Europe Box 3  Two Examples of Non-EU Poverty Maps: Albania and Morocco (continued) That poverty varied greatly not only among provinces, but also within provinces was an eye opener for policy makers and led to a reappraisal of targeting strategies. Whereas the major poor-area ­ development project, the Barnamaj al Aoulaouiyat al Ijtimaiya (Social Priorities Program), had relied on province-level targeting, the new information on poor communes provided the government with fresh insights into the possible design of a more finely targeted program, the National Initiative for Human Development. (Litvack 2007, 216) The National Initiative for Human Development was launched in May 2005. The government announced that $1 billion would be allocated to the program, half of which would go to efforts to target extra resources to the poorest 360 rural communes and poorest 250 urban neighborhoods. The maps were to be used for targeting purposes. The maps also began to play an interesting role in promoting local governance. Several governors had been surprised by some of the poverty map data and had undertaken site visits to confirm the data. The poverty maps thus contributed to a much broader agenda of transparency and good governance. Sources: Carletto, Dabalen, and Moubayed (2007); Litvack (2007). Other Impacts and Uses EU Cohesion Policy Cohesion policy in the EU is oriented towards reducing territorial disparities in economic and social conditions. It is financed through the European Structural and Investment Funds—especially the Regional Development Fund, the Cohesion Fund, and the Social Fund—with a budget of more than €350 ­billion for 2014–20. The bulk of cohesion policy funding is concentrated on less developed European countries and regions to help them catch up. Cohesion policy supports job creation, business competitiveness, economic growth, and sustainable development, seeks to improve the quality of life among citizens, and helps combat poverty in all EU cities and regions. To enhance the application of the funding, the EC has used poverty maps during negotiations with the EU member states to ensure that the strategies selected target poverty reduction, especially in areas with high poverty rates. Reducing poverty involves unique approaches in remote rural areas and in deprived inner-city neighborhoods. The poverty maps, in combination with other sources of information, have thus also been used in the design of a mix of investments to confront the underlying causes of poverty specific to each area. The EC has encouraged member states to allocate more funding per person from the European Social Fund to areas with high poverty rates as identified in the poverty maps. In the new 2014–20 programming period, each member state must dedicate at least 20  percent of the European Social Fund (that is, €16 billion in a total budget of €80 billion) to promote social inclusion and com- bat poverty. This guarantees a concentration on a few key priorities, each with a critical mass of funding that can have a substantial impact. Pinpointing Poverty in Europe 27 Mapping Social Exclusion While poverty and social exclusion are closely related, they are distinct in a number of ways (Copus 2014). Poverty is commonly defined relative to a minimum income. Social exclusion refers not only to income or physical well-being, but also to inclusion within various aspects of society, such as the labor market, administrative systems, communities, institutions, and demo- cratic processes. The difficulty of capturing patterns of exclusion is height- ened because social exclusion is intrinsically a set of processes rather than static characteristics. Much like the measurement of human development through the human development index or shared prosperity through the shared prosperity convergence index, social exclusion is measured indirectly by examining proxy indicators. Mapping social exclusion therefore involves separately gauging indirectly the risk of experiencing exclusion in its various dimensions, such as the labor market, education, health care, and so on, and then considering ways to synthesize the results. The EU’s Territorial Dimension of Poverty and Social Exclusion in Europe (TIPSE) project seeks to describe and understand regional patterns of poverty and social exclusion and considers how these may be monitored more effectively, including for the identification of general implications for policy design and implementation (Copus 2014). A review of the literature led the TIPSE project team to conclude that social inclusion may be repre- sented as comprising four broad domains, each of which may be disaggre- gated into individual component dimensions. These domains, each followed, in parentheses, by the component dimensions are (1) earning a living (income, employment), (2) access to services (health care, education, housing, transport and communications), (3) social environment (age, eth- nic composition, migrants, crime and safety), and (4) political participa- tion (citizenship). Taking a page from the World Bank’s approach, the researchers turned to the population census as a principal source of data on the domains at the NUTS 3 level. A substantial dataset of NUTS 3 proxy indicators, which are, in principle, harmonized and broadly comparable between 2001 and 2011, was eventually assembled. Representative Case Studies The TIPSE project selected ten case study areas to represent various European macroregions, territorial and socioeconomic typologies, and welfare regimes. The case studies, which include focus on five different thematic challenges: ethnicity-related social exclusion; age-related exclusion, both youth and elderly, and access to services of general interest in sparsely populated areas; urban education, with a focus on educational success, school performance, and segregation patterns; patterns and processes of ethnic and social segrega- tion in metropolitan regions; and unemployment. An overarching conclusion of the research is that a much stronger regional evidence base is required to enable EU and national policies to address effec- tively the challenges of poverty and social exclusion as hindrances to 28 Pinpointing Poverty in Europe territorial cohesion and balanced territorial development. Policy must become more well informed especially in three areas: geography, trends through time, and less tangible aspects beyond labor market issues. The mapping of poverty and social exclusion is expected to play a crucial role. The Next Steps The World Bank and the European Commission have agreed to carry out a second phase of the EU poverty mapping project (phase 2, 2015–16). The World Bank is continuing to supply technical assistance to the EC and EU member states for the development of small area poverty maps of member states that acceded to the EU in or after 2004 (excluding Cyprus and Malta). New estimates of poverty in small sub-national geographical areas are also to be provided, and refinements to the risk of poverty methodology that take into account sub-national variations in the cost of living across these areas and countries are to be investigated. New Maps Where They Did Not Exist Based on 2011 population censuses, new maps were to be produced on Bulgaria, the Czech Republic, and Lithuania. This was not possible in the first phase of the project because of delays in census data processing by NSIs and ongoing negotiations about the data confidentiality agreements required by the NSIs. The EC is financing a follow-on activity, in which one of the top agenda items is completion of these poverty maps. The World Bank will work with the NSI in Bulgaria to complete poverty maps at the municipality level (LAU 1). The additional EC funds will also be used to produce small area maps for Croatia, which was not included in the original EC-World Bank agree- ment because it was not a member of the EU at the time. Croatia’s Ministry of Regional Development and European Union Funds is especially interested in using the poverty maps, along with other regional data, to assess the effec- tiveness of government spending on social programs and infrastructure investments. Better Maps Where They Already Exist Work with the Hungary Central Statistical Office to update the poverty maps of the country using the 2011 population and housing census and the 2012 EU-SILC survey is ongoing. Achieving further disaggregation of the existing poverty maps in selected countries, particularly Latvia, is planned. Related Areas of Investigation Since the onset of the global financial crisis in 2008, public debt has increased dramatically; income has declined among many people across the EU; employment rates have fallen in most countries; and the unemployment rate is higher than it has been for over 20 years, while poverty and social Pinpointing Poverty in Europe 29 exclusion have become more widespread. At the same time, disparities in employment and unemployment rates and in gross domestic product per capita have widened across regions in many countries, while, in others, the rates have stopped narrowing. This means that the EU’s Europe 2020 employ- ment and poverty targets are now significantly more distant than they were when they were first established, and it will require a substantial effort in the coming years to achieve them in a context of significant budgetary con- straints (EC 2014). A greater risk of poverty and social exclusion is another legacy of the economic crisis. There are now around 9  million more people at risk of poverty or social exclusion in the EU relative to the situation before the crisis. A key issue is the variation within countries. The risk of poverty tends to be much lower in cities than in rural areas in less developed EU member states, while, in cities in the more developed member states, the reverse is the case. Accordingly, in the latter, to meet the national Europe 2020 poverty targets requires a major reduction in the number of people at risk of poverty or social exclusion in urban centers, while, in the former, the main challenge is to reduce the numbers at risk in rural areas (EC 2014). The new European Commission that took office in November 2014, led by EC President Jean-Claude Juncker, places new emphasis on invest- ments to revitalize economic growth in Europe. The detailed geographic estimates of poverty from the poverty maps may help assess the distribu- tional impacts of such investments, especially investments in physical infrastructure. The current EU approach to measuring the ARoP indicator implicitly assumes uniform costs of living within each EU member state. Thus, a sin- gle ARoP threshold is set for each member state corresponding to 60 per- cent of the national median income (see above). To the extent that a given income level supports a higher standard of living in a low-cost area, or a lower standard of living in a high-cost area, this assumption represents an inconsistency in the measurement of the risk of poverty. In line with the phase 2 agreement between the EC and the World Bank, the World Bank will draw on its global experience and on existing European data to explore effective and practical ways to enhance the methodology for gauging the risk of poverty so as to take into account sub-national variations in the cost of living, including those that have worsened in the last few years, and incorporate these variations into the ARoP measure and produce consistent indicators. Final Remarks The EC and the World Bank concur that the direct interaction through the project has been beneficial to both institutions and, through them, the coun- tries involved in the project, especially because of the provision of analytical, advisory, and knowledge services and technical assistance and the poverty mapping outputs. 30 Pinpointing Poverty in Europe The maps, meanwhile, are already supplying useful estimates of poverty in a readily understandable and practical manner on small sub-national geo- graphical areas such as districts or municipalities. They are helping guide allocations of European funds, antipoverty program targeting, and decision making and policy making at sub-national and the national levels in EU member states. The project is ongoing and continues to enable the EC and the World Bank to collaborate and exchange experiences and expertise on themes of mutual interest supported by the pillars of the Europe 2020 strategy within the frame- work of smart, sustainable, and inclusive economic and social growth. These pillars broadly parallel the strategy pillars established in the Europe and Central Asia Region of the World Bank. The EC has been a discerning consumer of the output of poverty m ­ apping, while remaining cognizant of the constraints imposed by institutional bar- riers. The exchange of ideas and information has been extremely construc- tive, and it is expected to become deeper as the new poverty maps are applied in EC operations and in World Bank initiatives. The EC–World Bank project has helped change the way the World Bank and other analysts approach poverty mapping, and it has come to embody new challenges and opportunities to improve efforts to reduce poverty and social exclusion in Europe and elsewhere throughout the world. Notes 1. The five EU-wide targets are (a) to raise the employment rate of the population aged 20–64 from the current 69  percent to at least 75  percent; (b) to invest 3 percent of gross domestic product in research and development by ­ improving the conditions for such investment by the private sector and to develop a new indicator to track innovation; (c) to reduce greenhouse gas emissions by at least 20 percent relative to 1990 levels or, if conditions are promising, by 30 percent, to increase the share of renewable energy in final energy consumption to 20 per- cent, and to achieve a 20 percent increase in energy efficiency; (d) to reduce the share of early school leavers to 10 percent from the current 15 percent and to increase the share of the population aged 30–34 who have completed tertiary education from 31 percent to at least 40 percent; and (e) to reduce the number of Europeans living at risk of poverty or social exclusion by 20 million people by 2020 (EC 2010). 2. Income and Living Conditions, Main Tables (database), Eurostat, Luxembourg, http://ec.europa.eu/eurostat/web/income-and-living-conditions/data /main-tables (Accessed April 3, 2016). Also see World Bank (2014a). 3. NUTS is a geocode classification standard for referencing subdivisions of coun- tries for statistical purposes. In the EU, it is a hierarchical system for the divi- sion of economic territory for the development of regional statistics and regional socioeconomic analyses and the framing of EU regional policies. NUTS 1 rep- resents major socioeconomic regions, typically large national regions or groups of regions, each with a population from 3 million to 7 ­ million people. NUTS 2 corresponds to basic national regions for the application of regional policies and is usually composed of individual regions with populations from 800,000 to 3 mil- lion. NUTS 3 is usually counties or districts, each with a population from approx- imately 150,000 to 800,000. In Poland, for instance, NUTS 1 is the country’s six Pinpointing Poverty in Europe 31 regions; NUTS 2 is the 16 voivodeships; and NUTS 3 is the 66 statistical subre- gions. To date, the NUTS 2 classification has been used to determine the eligibility for aid through European Structural and Investment Funds. Below the NUTS 3 classification, areas are defined according to local administrative units (LAUs). Most EU member states have two LAUs. (LAU 1 and LAU 2), but some have only one. See “Common Classification of Territorial Units for Statistical Purposes,” EUR Lex (database), Publications Office of the European Union, Luxembourg, http://eur-lex.europa.eu/legal-content/EN/ ​ ­TXT/?qid=1428951729674&uri=URI SERV:g24218. 4. In the reckoning of these rates, yearly disposable income is equivalized across households. This means that household income data are subjected to calcula- tions so as to render them comparable across households of different sizes and compositions. 5. In statistics, regression analysis is a process for estimating the relationships among variables. It includes various techniques to model (analyze) several vari- ables, whereby the focus is on the relationship between a single dependent vari- able and one or several independent variables. A dependent variable is a variable of which the value depends on the value of another variable (the independent variable). For instance, a household’s income (the dependent variable) might depend on the number of household members who are employed, the sectors in which they are employed, and their occupational skill levels (the independent variables). 6. A modified Organisation of Economic Co-operation and Development scale is used to calculate equivalized income. The scale assigns a weight of 1.0 to the first adult household member aged 14 years or older, 0.5 to other household members aged 14 or older, and 0.3 to household members younger than 14  years. 7. Maps were also to have been produced on Bulgaria, the Czech Republic, and Lithuania, but this has not been possible because of delays in census data process- ing by NSIs and institutional impediments. Croatia, which joined the EU in 2013, will be included in the second round of poverty map construction, which began in 2015. See the text, following, for more details. 8. In survey and census data, microdata represent information at the level of individual respondents or households, such as individual educational attain- ­ ment or the number of household members. Aggregate or area-level data are summary statistics for a given geographical area, such as the proportion of area residents who have completed secondary school or the average household size in the area. 9. The text here is based on Statistics Estonia and World Bank (2014a) and World Bank (2013). 10. The text here is based on the Poverty Mapping Brief on Hungary (HCSO and World Bank 2014). 11. The maps for Hungary are based on the 2005 EU-SILC and a 2005 microcensus. New maps using the latest complete census (2011) and the 2011 EU-SILC are being produced in 2016 under phase 2 of the EC-WB project. 12. The text here is based on the Poverty Mapping Brief on Latvia (CSB and World Bank 2014). 13. The initial mapping exercise did not have access to census microdata on Latvia. The processing of the microdata has now been completed, and the construction of new poverty maps, including data on 119 cities and municipalities, is planned for 2016. 14. The text here is based on the Poverty Mapping Brief on Poland (CSO and World Bank 2014). 15. The text here is based on the Poverty Mapping Brief on Romania (NIS and World Bank 2014). 32 Pinpointing Poverty in Europe 16. The text here is based on the Poverty Mapping Brief on the Slovak Republic (Statistical Office of the Slovak Republic and World Bank 2014). 17. The text here is based on the Poverty Mapping Brief on Slovenia (Statistical Office of the Republic of Slovenia and World Bank 2014). 18. The technical reports are Statistical Office in Poznań, Central Statistical Office and the World Bank (2014), Statistics Estonia and World Bank (2014b), World Bank (2013a), World Bank (2013b), World Bank (2013c), World Bank (2014b), and World Bank (2014c). Full citations are listed in the references. Pinpointing Poverty in Europe 33 Appendix A PovMap and the Validation Study Building Poverty Maps: The World Bank PovMap Software Several methodologies are available to produce finely disaggregated poverty maps. The small area estimation methodology developed at the World Bank provides several benefits relative to alternative poverty maps based on com- posite thematic indicators using administrative data, such as maps based on a basic needs index. First, maps based on composite indicators are rather arbitrary in both the choice of variables and the weights assigned to the ­ variables, yielding ad hoc, easily disputable outputs. Also, drawing a connec- tion between these composite indicators and income poverty is not straight- forward and often prone to criticism. In contrast, the approach employed by the World Bank in the EU and in other poverty mapping efforts is based on official poverty lines using a clear and transparent methodology, which may produce easily interpretable results, and the statistical precision of which may be properly gauged. Despite this and many other advantages, however, small area estimation has some drawbacks, too. The first concern is related to the significant data requirements; indeed, small area estimation poverty maps rely on large and sequenced survey data collection and censuses. Thus, given the low frequency of census undertakings, small area estimation maps would typically be updated only every decade or so, although there are methods for updating maps on the basis of new household surveys or other population surveys. In addition, the analytical skills and human resources needed for the production of small area estimation maps are substantial and often beyond the available skill level in many countries. This also makes the full transfer of ownership and acceptance by local policy makers more difficult to accomplish in a short time. Finally, there is a limit to the level of disaggregation that may be achieved with small area estimation given the increasing level of imprecision as one moves to successively smaller population sub-groups. In 2004, the World Bank research department developed special-purpose software, PovMap, for the estimation of poverty in small areas. The main advantage of PovMap relative to standard statistical software is the speed and efficiency it allows in performing hundreds of simulations on large census datasets containing millions of observations. The software is also effective in handling the complex error structures used in small area estimation. PovMap’s graphical user interface simplifies operations, while retaining, in the back- ground, the required rigor and complexity of the computations. From the outset, the EC–World Bank project has supported the enhance- ment of the PovMap software. In response to the methodological findings Pinpointing Poverty in Europe 35 from poverty mapping in the EU, several new features were added to the  latest release of the software, PovMap 2.5.1 This new version offers increased flexibility so users can choose between conventional ELL methods or EB enhanced methods. Initially, this was an outcome of ­ addressing the methodological questions raised during the ­ EC–World Bank project validation study (see the main text and below). However, computational methods have also been refined, and new approaches and solutions have been generated and integrated throughout the project so that new features have been regularly added to the software package. Although some of the most recent innovations are experimental, they have improved the flexibility of the software and produced useful regres- sion diagnostic output. The software is being used by analysts within and outside the World Bank. The EC–World Bank Mapping Methodology Validation Study Elbers and van der Weide (2014) suggest that the choice between EB and ELL is largely determined by two key factors: (1) how much information one stands to ignore or lose, which depends on how many of the small areas have been sampled in the income survey and the size of the area error relative to the total error; and (2) the degree of nonnormality in the data. In regard to the first decisive factor, two key observations emerged from the poverty mapping exercises. The first observation recognizes that the number of small areas ­ covered by the EU-SILC—the instrument used by EU member states to col- lect household data on income—tends to be substantial. It is typical for the EU-SILC to cover 50–100 percent of the small areas for which poverty is to be estimated, compared with a norm of 5–25 percent in low-income countries. However, the size of the location effect is relatively small in the EU member states. Thus, for example, the correlation of household incomes within a small area is less likely to affect the estimates on EU member states. This suggests that the gains from the use of the EB are modest for most areas, although there may still be a handful of areas where the EB could realize a substantial improvement. This brings us to the second decisive factor for choosing a small area esti- mation method: the degree of nonnormality in the data, that is, the issue of data redundancy. If this is notable, then ELL would be the preferred method. However, the amount of empirical evidence collected on the question of nor- mality is rather limited. Monte Carlo simulations carried out through the project show that ignoring the nonnormality of errors can lead to biased esti- mates of poverty and inequality as high as 2–3 percent on a poverty rate of 20–30 percent (Elbers and Van der Weide 2014). Under each of the ELL and EB umbrellas are a host of other assumptions and choices that are also testable, such as how to model the random error associated with location or how to handle heteroskedastic errors in modeling household income. 36 Pinpointing Poverty in Europe Likewise, the study provided other insights that helped improve the understanding of small area poverty estimation. In particular, it generated ­ knowledge about appropriate choices in the incorporation of survey sampling weights in regressions, ways to account for heteroskedasticity in household income models, and whether the distribution of model parameters should be estimated parametrically or through bootstrap methods. Overall, the study also spurred new research in small area poverty estimation, including more intensive dialogue among various practitioners. The research-focused aspects of the project were presented at numerous academic conferences and results seminars in new EU member states and at the EC. Finally, through the methodological validation, it became apparent that current approaches to small area estimation face difficulties in producing estimates of poverty accurately in the tails of the distribution, that is, those areas with especially high or low poverty rates. This will be an important avenue of further research to continue improving the way that poverty ­ mapping is practiced by the World Bank and others. ­ Note 1. PovMap is freely available from the World Bank website, at http://iresearch​ .­worldbank.org. Pinpointing Poverty in Europe 37 Appendix B Poverty Mapping Briefs The following are the complete versions of the Poverty Mapping Briefs for the seven countries completed during the first phase of the EC-World Bank ­ project, Poverty Mapping in the European Union, which are summarized in the main text. Pinpointing Poverty in Europe 39 Country Policy Brief Poverty in Europe March 2016 Poverty and Equity Global Practice Pinpointing Poverty in Estonia Rates of poverty and social exclusion vary member states, have developed a set of high widely across European Union (EU) member resolution ­poverty maps. The greater geo- states and there is also a high degree of graphic disaggregation of the new poverty variability in living standards within member maps reveals which parts of these larger re- states. In its 2014–2020 multiannual finan- gions have ­ particularly high rates of poverty cial framework the EU has budgeted one and require greater attention for poverty re- trillion euros to support growth and jobs and duction programs. to reduce the ­ number of people living at risk The poverty maps for Estonia confirm ex- of poverty or social exclusion by 20 million by isting knowledge about poverty in Estonia, the year 2020. To this end, the Government but also reveal new insights.2 For example, of Estonia has set a national goal of reducing previous surveys have shown counties in the the risk of monetary poverty rate from 17.5 North-Eastern and South-Eastern regions to percent in 2010 to 15.0 percent by 2020.1 have the highest rates of poverty (map 1, Success depends on developing the right left panel), but the more disaggregated esti- policies and programs and targeting them ef- mates reveal contrasts between lower pov- fectively; however, the European Commission erty estimates in large cities and surrounding has previously had to rely on sub-national areas vis-à-vis other parts of the counties data at a relatively high level of aggregation (map 1, panel a). Such is the case in the for program planning and the allocation of EU county of Jogeva, which has a high poverty funds. The European Commission and the rate overall, but has a lower poverty incidence World Bank, in cooperation with individual EU in the western regions. There are also local Map 1 At-Risk-of-Poverty Rates, Estonia a. Counties (LAU 1) b. Grouped local government units (LAU 2) 5.8–9.4 12.6 9.4–14.4 12.6–15.8 14.4–17.2 15.8–17.63 17.2–18.3 17.630–18.3 18.3–21.7 18.3–21.7 21.7–24.7 21.7–25.7 24.7–25.1 25.7–29.9 25.1–26.9 29.9–34.3 Source: Estimates using data from the 2012 Estonian Social Survey and the 2011 Population and Housing Census collected by Statistics Estonia. Note: The risk of poverty rates are defined using the EU standard of 60 percent of median national equivalized income after social transfers. 1 government units with high poverty incidence Map 2  Population Living below the Poverty in counties that otherwise have low-to-mod- Threshold, Estonia erate overall poverty rates. For example, the county of Tartu has a moderate poverty rate at the LAU1 level,3 but more disaggregated estimates suggest a sharp divide between the low poverty incidence in the city of Tartu and its surroundings, and the higher poverty incidence in the eastern municipali- ties. Knowing which areas have higher pov- erty rates can help in more efficiently targeting resources for development and poverty reduction. 326–1,562 Targeting poor areas alone can have its 1,562–4,934 4,934–9,841 limitations. Policy makers have an interest 9,841–17,826 both in areas where poverty is high and also in areas that have the highest number of Source: Estimates using data from the 2012 Estonian Social Survey and the 2011 Population and Housing Census collected poor people. These two are not the same: by Statistics Estonia. areas that are very poor may also be sparsely populated, whereas large cities tend to have low poverty rates, but large numbers of poor poverty than was previously available and people because of large populations. This may help improve resource allocation. The contrast is readily seen in the large concen- maps also force more thinking on how best to tration of poor people in Tallinn and Tartu cit- allocate resources aimed at improving stan- ies and their surrounding areas (map 2). Even dards of living, balancing the targeting of poor though risk of poverty rates are low, these areas and poor people. While the right combi- cities are home to a large share of Estonia’s nation of approaches will vary by country, the poor population. On the other hand, Ida-Viru maps provide important information to help county in the northeast has both high pov- improve policies and programs to combat erty rates and a large number of poor peo- poverty and exclusion. ple. This is most pronounced in the former military-industrial city of Narva, Estonia’s Notes 1. Estonia, Government Office. 2014. National Reform third-largest city, which has a large ethnic Programme “Estonia 2020.” May 8. Tallinn, Estonia: Russian majority. Government Office, Stenbock House. Poverty maps do not provide all the an- 2. These maps combine microdata from the 2011 popu- swers—they must be combined with other lation census and the 2012 EU-SILC survey. information, including local expertise, to 3. The NUTS (Nomenclature des Unités Territoriales Statistiques) classification is a hierarchical system of inform decision-making. After identifying dividing up the economic territory of the European the areas or populations in greatest need it Union for the development of regional statistics, re- is necessary to understand why these places gional socioeconomic analysis, and the framing of EU are poor. The reasons are likely to vary from regional policies. To date the NUTS 2 classification place to place, and may include inadequate has been used for ­ determining eligibility for aid from European Structural Funds. Below the NUTS 3 infrastructure, lack of economic activity, classification areas are defined according to Local an insufficiently skilled work force, or other Administrative Units (LAU). Most EU member states reasons. Poverty maps provide finer grained have LAU 1 and LAU 2 divisions, but some only have information on sub-national variation in LAU 2. © 2016 International Bank for Reconstruction and Development / The World Bank. Some rights reserved. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. This work is subject to a CC BY 3.0 IGO license (https://creativecommons.org/licenses/by/3.0/ igo). The World Bank does not necessarily own each component of the content. It is your responsibility to determine whether permission is needed for reuse and to obtain permission from the copyright owner. If you have questions, email pubrights@ worldbank.org. SKU K8682 2 Country Policy Brief Poverty in Europe March 2016 Poverty and Equity Global Practice Pinpointing Poverty in Hungary Rates of poverty and social exclusion vary states, have developed a set of high-resolution widely across European Union (EU) member poverty maps.2 The greater geographical dis- states, and there is also a high degree of aggregation of the new poverty maps reveals variability in living standards within member which parts of these larger regions have partic- states. In its 2014–20 multiannual finan- ularly high rates of poverty and require greater cial framework, the EU budgeted €1 trillion attention in poverty reduction programs. to support growth and jobs and to reduce The poverty maps for Hungary confirm the number of people living at risk of pov- existing knowledge about poverty in Hungary, erty or social exclusion by 20 ­million by the but also reveal new insights. For example, year 2020. To this end, the Government of previous surveys have shown that the high- Hungary has set a national goal of reducing est rates of poverty occur in northeastern the number of the poor and socially excluded Hungary (map 1, panel a). The statistical by 450,000 people by 2020.1 subregion-level (LAU 1)3 poverty map (map 1, Success depends on developing the ap- panel b) shows that many of the subregions propriate policies and programs and target- in the northeastern corner have elevated ing them ­effectively. However, the European risk of poverty rates, although Eger, Miskolc, Commission has previously had to rely on and Nyíregyháza stand out as areas with sub-national data at a relatively high level of only moderate poverty incidence. In con- aggregation for program planning and the allo- trast, Southern Transdanubia is heteroge- cation of EU funds. The EC and the World Bank, neous, comprising low poverty incidence in in cooperation with individual EU member subregions such as Pécs, with relatively high Map 1 At-Risk-of-Poverty Rates, Hungary a. Seven planning & statistical regions (NUTS 2) b. Subregions (LAU 1) 7 4.72–7.7 7.7–10.46 7–8.82 10.46–12.7 8.82–13.2 12.7–13.2 13.2–16.7 13.2–17.98 16.7–17.9 17.98–22.4 17.9–18.5 22.4–26.98 18.5–20.6 26.98–31.62 Source: Estimates using data from the 2005 EU-SILC and 2005 Microcensus collected by the Hungarian Central Statistical Office. Note: The risk of poverty rates are defined using the EU standard of 60 percent of median national equivalized income after social transfers. 1 poverty incidence in nearby Siklós, Sellye, Map 2  Population Living below the Poverty Szigetvár, or Szentlőrinc. More generally, there Threshold, Hungary is a much higher degree of heterogeneity in poverty incidence at the statistical subregion level vis-à-vis the estimates available directly from the EU-SILC survey for the seven plan- ning and statistical regions. Knowing which subregions have higher poverty rates can help in more efficiently targeting resources for de- velopment and poverty reduction. Targeting poor areas alone can have limitations. Policy makers have an interest ­ both in areas where poverty is high and in 894–6,878 6,878–15,905 areas that have the most poor people. These 15,905–35,587 two are not the same: areas that are poor 35,587–93,100 may also be sparsely populated, whereas Source: Estimates using data from the 2005 EU-SILC and 2005 large cities tend to have low poverty rates, Microcensus collected by the Hungarian Central Statistical Office. but large numbers of poor people because of the large populations. In Hungary, the density of the population below the poverty threshold is, on average, also higher in sub-­ and poor people. While the appropriate com- regions with higher poverty incidence (map 2). bination of approaches will vary by country, Nonetheless, the rankings of subregions by the maps provide important information to poverty incidence and by poverty density help improve policies and programs to com- are quite divergent; sub-regions such as bat poverty and social exclusion. Budapest, Pécs, Gyõr, and Székesfehérvári have a low poverty incidence, but, given their Notes large populations, they account for a large 1. Hungary, Ministry for National Economy. 2015. “National share of the total population below the risk Reform Programme 2015 of Hungary.” Ministry for National Economy, Budapest. of poverty threshold. 2. At the time of preparing these poverty maps of Poverty maps do not provide all the an- Hungary, the 2011 Population Census data were still swers. They must be combined with other in- being processed. Therefore these maps are based on formation, including local expertise, to inform the 2005 EU-SILC and the 2005 Microcensus carried decision making. After identifying the areas out by the Hungarian Central Statistical Office, which covered two percent of the population. The poverty or populations in greatest need, one must maps will be updated under Phase 2 of the EC-World understand why these places are poor. The Bank project, using the 2011 Population Census and reasons are likely to vary from place to place 2011 EU-SILC. and may include inadequate infrastructure, 3. The NUTS (Nomenclature des Unités Territoriales lack of economic activity, an insufficiently Statistiques) classification is a hierarchical system of dividing up the economic territory of the European skilled workforce, or other reasons. Poverty Union for the development of regional statistics, re- maps provide more finely grained information gional socioeconomic analysis, and the framing of EU on sub-national variations in poverty than regional policies. To date the NUTS 2 classification was previously available and can potentially has been used for determining eligibility for aid improve resource allocation. The maps also from European Structural Funds. Below the NUTS 3 classification areas are defined according to Local force more thinking on how best to allocate Administrative Units (LAU). Most EU member states resources aimed at improving standards of have LAU 1 and LAU 2 divisions, but some only have living, balancing the targeting of poor areas LAU 2. © 2016 International Bank for Reconstruction and Development / The World Bank. Some rights reserved. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. This work is subject to a CC BY 3.0 IGO license (https://creativecommons.org/licenses/by/3.0/ igo). The World Bank does not necessarily own each component of the content. It is your responsibility to determine whether permission is needed for reuse and to obtain permission from the copyright owner. If you have questions, email pubrights@ worldbank.org. SKU K8683 2 Country Policy Brief Poverty in Europe March 2016 Poverty and Equity Global Practice Pinpointing Poverty in Latvia Rates of poverty and social exclusion vary individual EU member states, have developed widely across European Union (EU) member a set of high-resolution poverty maps.2 The states, and there is also a high degree of greater geographical disaggregation of the variability in living standards within member new poverty maps reveals which parts of states. In its 2014–20 multiannual financial these larger regions have particularly high framework, the EU budgeted €1 trillion to rates of poverty and require greater attention support growth and jobs, which will contrib- in poverty reduction programs. ute to the goal of reducing the number of The poverty maps for Latvia confirm exist- people living at risk of poverty or social exclu- ing knowledge about poverty in Latvia, but also sion by 20 million by the year 2020. To con- reveal new insights. The EC has relied on the tribute to this goal, the Government of Latvia NUTS 2 territorial classification3 to determine has set a national goal of reducing by 121,000 eligibility for aid from European Structural and the number of people at risk of poverty or Investment Funds and for program planning. living in households with low work intensity.1 In smaller countries such as Latvia, the NUTS Success depends on d ­ eveloping the appro- 2 classification corresponds to the entire na- priate policies and programs and targeting tional territory, that is, with no sub-national di- them effectively. However, the EC has previ- visions (map 1, panel a). The EU-SILC in Latvia ously had to rely on sub-national data at a rel- is representative at the statistical region level atively high level of aggregation for program (NUTS 3), and the Central Bureau of Statistics planning and the allocation of EU funds. The reports risk of poverty estimates at that level. EC and the World Bank, in cooperation with Using small area estimation techniques, it Map 1 At-Risk-of-Poverty Rates, Latvia a. NUTS 2 classification b. Statistical subregions (NUTS 3) 12.93 12.93–15.93 15.93–19.1 19.1–23 19.2 23–24.9 24.9–28.1 Source: Estimates using data from the 2012 EU-SILC and 2011 Population and Housing Census collected by the Latvia Central Statistical Bureau. Note: The risk of poverty rates are defined using the EU standard of 60 percent of median national equivalized income after social transfers. 1 was possible to improve the precision of these Map 2  Population Living below the Poverty poverty estimates for the six regions of Latvia. Threshold, Latvia According to these estimates, there is quite a bit of heterogeneity across regions (map 1, panel b). Riga and Pieriga in northern central Latvia are the only regions where the inci- dence of poverty is below the national average. Estimates of poverty in other regions range from 21 percent to as high as to 28 percent in Latgale in southeastern Latvia. Knowing which regions have higher poverty rates can help more efficiently target resources for de- velopment and poverty reduction. 52,613 52,613–55,365 Targeting poor areas alone can have limita- 55,365–59,169 tions. Policy makers have an interest both in 59,169–85,255 areas where poverty is high and in areas that Source: Estimates using data from the 2012 EU-SILC and 2011 have the most poor people. These two are not Population and Housing Census collected by the Latvia Central the same: areas that are poor may also be Statistical Bureau. sparsely populated, whereas large cities tend to have low poverty rates, but large numbers of poor people because of the large popula- one must understand why these places are tions. For example, Riga Region has the lowest poor. The reasons are likely to vary from estimated risk of poverty, 12.9 percent, which place to place and may include inadequate is less than one-half the risk in the poorest infrastructure, lack of economic activity, region, Latgale (28.0 percent). However, be- an insufficiently skilled workforce, or other cause Riga Region is much more populous, reasons. Poverty maps provide more finely each of the two regions has approximately grained information on sub-national varia- 85,000 people living at risk of poverty. While tions in poverty than was previously avail- the remaining four regions have somewhat able and can potentially improve resource heterogeneous ­ poverty incidence, they all have allocation. The maps also force more think- a similar number of poor people, in the range ing on how best to allocate resources aimed of 50,000–60,000 (map 2). at improving standards of living, balancing While poverty estimation at the regional the targeting of poor areas and poor people. level adds a significant nuance to national es- While the appropriate combination of ap- timates, more revealing spatial heterogeneity proaches will vary by country, the maps pro- is possible if census microdata are employed. vide important information to help improve The initial poverty maps in Latvia were lim- policies and programs to combat poverty ited to the six statistical regions because cen- and social exclusion. sus microdata were not available at that time. The processing of the microdata has now Notes been completed, and it is planned to revisit 1. Latvia, Ministry of Economics. 2015. “National the poverty mapping exercise using census Reform Programme of Latvia for the Implementation of the ‘Europe 2020’ Strategy: Progress Report.” April, and EU-SILC microdata. By using micro- Ministry of Economics, Riga, Latvia. data, one will be able to estimate the risk of 2. These maps combine aggregate data from the 2011 poverty for much smaller geographical units population census and the 2012 EU-SILC survey. and provide much higher-­ resolution poverty 3. The NUTS (Nomenclature des Unités Territoriales estimates than are possible directly from Statistiques) classification is a hierarchical system of dividing up the economic territory of the the EU-SILC. It is expected that reasonably European Union for the development of regional © 2016 International Bank for Reconstruction precise poverty estimates can be obtained for statistics, regional socioeconomic analysis, and the and Development / The World Bank. Some rights reserved. The findings, interpretations, Latvia’s 119 municipalities and cities. framing of EU regional policies. To date the NUTS 2 and conclusions expressed in this work do not Poverty maps do not provide all the an- classification has been used for determining eligibil- necessarily reflect the views of The World Bank, its swers. They must be combined with other ity for aid from European Structural Funds. Below Board of Executive Directors, or the governments the NUTS 3 classification areas are defined accord- they represent. The World Bank does not guarantee information, including local expertise, to ing to Local Administrative Units (LAU). Most EU the accuracy of the data included in this work. This work is subject to a CC BY 3.0 IGO license inform decision making. After identifying member states have LAU 1 and LAU 2 divisions, but (https://creativecommons.org/licenses/by/3.0/ the areas or populations in greatest need, some only have LAU 2. igo). The World Bank does not necessarily own each component of the content. It is your responsibility to determine whether permission is needed for reuse and to obtain permission from the copyright owner. If you have questions, email pubrights@ worldbank.org. SKU K8684 2 Country Policy Brief Poverty in Europe March 2016 Poverty and Equity Global Practice Pinpointing Poverty in Poland Rates of poverty and social exclusion vary program planning and the allocation of EU widely across European Union (EU) member funds. The EC and the World Bank, in cooper- states, and there is also a high degree of ation with individual EU member states, have variability in living standards within member developed a set of high-resolution poverty states.1 In its 2014–20 multiannual financial maps.3 The greater geographical disaggrega- framework, the EU budgeted €1 trillion to tion of the new poverty maps reveals which support growth and jobs and to reduce the parts of these larger regions have ­particularly number of people living at risk of poverty or high rates of poverty and require greater at- social exclusion by 20 million by the year 2020. tention in poverty reduction programs. To this end, the Government of Poland has The poverty maps confirm existing knowl- set a national goal of reducing the number of edge about poverty in Poland, but also reveal the poor and socially excluded by 1.5 million new insights. For example, previous surveys people.2 have shown the southeastern voivodships (in Success depends on ­ developing the ap- particular Lubelskie and Swietokrzynskie) as propriate policies and programs and target- well as Lubuskie in the west to have higher ing them effectively. However, the EC has poverty rates (map 1, panel a). The more de- previously had to rely on sub-national data tailed poverty map shows that most of the relatively high level of aggregation for at a ­ statistical subregions with the highest risk of Map 1 At-Risk-of-Poverty Rates, Poland a. Voivodships (NUTS 2) b. Statistical subregions (NUTS 3) 12.3 6.3–8.7 12.3–13.5 8.7–12.6 13.5–14.8 12.6–15.1 14.8–17.1 15.1–16.9 17.1–18.9 16.9–19.8 18.9–19.3 19.8–22.9 19.3–21.7 22.9–26.1 21.7–26 26.1–30.2 Source: Estimates using data from the 2011 EU-SILC and the 2011 Census of Population and Housing collected by the Central Statistical Office of Poland. Note: The risk of poverty rates are defined using the EU standard of 60 percent of median national equivalized income after social transfers. The NUTS (Nomenclature des Unités Territoriales Statistiques) classification is a hierarchical system of dividing up the economic territory of the European Union for the development of regional statistics, regional socioeconomic analysis, and the framing of EU regional policies. To date the NUTS 2 classification has been used for determining eligibility for aid from European Structural Funds. Below the NUTS 3 classification areas are defined according to Local Administrative Units (LAU). Most EU member states have LAU 1 and LAU 2 divisions, but some only have LAU 2. 1 poverty rates are in the east and southeast Map 2  Population Living below the Poverty of the country (map 1, panel b). However, the Threshold, Poland subregional map also reveals visible differ- ­ ences in poverty incidence between the low poverty cities (Krakow, Wroclaw, Poznań, Warsaw) and lower population-density surrounding areas with higher poverty inci- ­ dence. The subregional map likewise highlights heterogeneity in poverty estimates within voivodships; the voivodship of Pomorskie in- cludes Trojmiejski, with an estimated poverty incidence of 7 percent, alongside Starogardzki and Slupski, both with poverty rates in excess 38,742–76,840 76,840–108,548 of 20 percent. Knowing which subregions 108,248–146,792 have higher ­ poverty rates can help more effi- 146,792–194,591 ciently target resources for development and Source: Estimates using data from the 2011 EU-SILC and the poverty reduction. 2011 Census of Population and Housing collected by the Central Targeting poor areas alone can have Statistical Office of Poland. limitations. Policy makers have an interest ­ both in areas where poverty is high and in areas that have the most poor people. These place to place and may include inadequate two are not the same: areas that are poor infrastructure, lack of economic activity, may also be sparsely populated, whereas an insufficiently skilled workforce, or other large cities tend to have low poverty rates, reasons. Poverty maps provide more finely but large numbers of poor people because of grained information on sub-national varia- the large populations. In Poland, there is an tions in poverty than was previously avail- overall positive and relatively strong relation- able and can potentially improve resource ship at the subregional level between the pov- allocation. The maps also force more think- erty incidence rate and the total number of ing on how best to allocate resources aimed individuals living below the poverty threshold. at improving standards of living, balancing The highest absolute numbers of the poor are the targeting of poor areas and poor people. located in the east and southeast of Poland While the appropriate combination of ap- (map 2). At the same time, Warsaw stands proaches will vary by country, the maps pro- out as having the lowest estimated poverty vide important information to help improve incidence among statistical subregions, but, policies and programs to combat poverty on account of its size, it ranks 27th out of 66 and social exclusion. subregions in terms of the absolute number of individuals below the poverty threshold. Notes This is not observed in other cities such as 1. The results presented here are from the study “Poverty Poznań, Wroclaw, or Krakow, where both maps at the Subregional Level in Poland Based on Indirect Estimation.” They are not official statistical poverty incidence and the absolute number data and are entirely the result of an experimental of poor people appear to be low. work. The estimates have been elaborated by the Poverty maps do not provide all the an- Central Statistical Office of Poland in the framework swers. They must be combined with other of its collaboration with the World Bank set out in the information, including local expertise, to letter of intent dated 26 June 2013. 2. Poland, Ministry of Economy. 2015. “National Reform inform decision making. After identifying Programme Europe 2020: Update 2015/2016.” April the areas or populations in greatest need, 28, Ministry of Economy, Warsaw. one must understand why these places are 3. These maps combine aggregate data from the 2011 © 2016 International Bank for Reconstruction poor. The reasons are likely to vary from population census and the 2011 EU-SILC survey. and Development / The World Bank. Some rights reserved. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. This work is subject to a CC BY 3.0 IGO license (https://creativecommons.org/licenses/by/3.0/ igo). The World Bank does not necessarily own each component of the content. It is your responsibility to determine whether permission is needed for reuse and to obtain permission from the copyright owner. If you have questions, email pubrights@ worldbank.org. SKU K8685 2 Country Policy Brief Poverty in Europe March 2016 Poverty and Equity Global Practice Pinpointing Poverty in Romania Rates of poverty and social exclusion vary at a relatively high level of aggregation for widely across European Union (EU) member program planning and the allocation of EU states, and there is also a high degree of funds. The EC and the World Bank, in cooper- variability in living standards within member ation with individual EU member states, have states. In its 2014–20 multiannual financial developed a set of high-resolution poverty framework, the EU budgeted €1 trillion to maps.2 The greater geographical disaggrega- support growth and jobs and to reduce the tion of the new poverty maps reveals which number of people living at risk of poverty parts of these larger regions have particularly or social exclusion by 20 ­ million by the year high rates of poverty and require greater at- 2020. To help reach this EU-wide target, the tention in poverty reduction programs. Government of Romania has set a national The poverty maps for Romania confirm goal of reducing the number of the poor and existing knowledge about poverty in Romania, socially excluded by 580,000 people.1 but also reveal new insights. For example, Success depends on developing the ap- previous surveys have shown the Northeast propriate policies and programs and target- Region to have the highest rates of poverty ing them effectively. However, the EC has (map 1, panel a), but the county-level poverty previously had to rely on sub-national data map (map 1, panel b) shows that all counties Map 1 At-Risk-of-Poverty Rates, Romania a. Development regions (NUTS 2) b. Counties (NUTS 3) 2.7 3.9 2.7–15.9 3.9–17.46 15.9–18.5 17.46–22 18.5–22.6 22–22.9 22.6–26.5 22.9–27.75 26.5–29.7 27.75–29.7 29.7–30.9 29.7–32.17 30.9–33.7 32.17–43.66 Source: Estimates using data from the 2011 EU-SILC and 2011 Population and Housing Census collected by the Romania National Institute of Statistics. Note: The risk of poverty rates are defined using the EU standard of 60 percent of median national equivalized income after social transfers. The NUTS (Nomenclature des Unités Territoriales Statistiques) classification is a hierarchical system of dividing up the economic territory of the European Union for the development of regional statistics, regional socioeconomic analysis, and the framing of EU regional policies. To date the NUTS 2 classification has been used for determining eligibility for aid from European Structural Funds. Below the NUTS 3 classification areas are defined according to Local Administrative Units (LAU). Most EU member states have LAU 1 and LAU 2 divisions, but some only have LAU 2. 1 of that region, with the exception of Bacău, Map 2  Population Living below the Poverty have elevated risk of poverty rates. In con- Threshold, Romania trast, the South Region is heterogeneous, comprising counties with high poverty rates, such as Călărași and Teleorman, and coun- ties with relatively low poverty rates, such as Prahova. Similarly, Cluj County has the second-lowest poverty rate in Romania (after Bucharest), but its neighboring counties in the Northwest Region (Bistrița-Năsăud, Maramureș, Sălaj, and Satu Mare) have higher poverty than the Romanian average. Knowing which counties have higher poverty rates can 46,000–78,355 help more efficiently target resources for de- 78,355–107,891 107,891–140,400 velopment and poverty reduction. 140,400–218,757 Targeting poor areas alone can have lim- Source: Estimates using data from the 2011 EU-SILC and 2011 itations. Policy makers have an interest both Population and Housing Census collected by the Romania in areas where poverty is high and in areas National Institute of Statistics. that have the most poor people. These two are not the same: areas that are poor may The reasons are likely to vary from place to also be sparsely populated, whereas large cit- place and may include inadequate infrastruc- ies tend to have low poverty rates, but large ture, lack of economic activity, an insuffi- numbers of poor people because of the large ciently skilled workforce, or other reasons. populations. For example, despite its lower Poverty maps provide more finely grained poverty rate, Cluj County has more people at information on sub-national variations in risk of poverty than Sălaj, and Bucharest has poverty than was previously available and can more people at risk of poverty than six other potentially improve resource allocation. The counties. Poverty map 2 suggests a comple- maps also force more thinking on how best to mentary approach to allocating resources for allocate resources aimed at improving stan- poverty reduction, with Bucharest and other dards of living, balancing the targeting of poor urban areas given greater attention as many of areas and poor people. While the appropriate the poor live in relatively rich areas. The maps combination of approaches will vary by coun- clearly illustrate that the Northeast, especially try, the maps provide important information Botoșani, Iași, and Suceava, have high poverty to help improve policies and programs to rates and large numbers of poor people and combat poverty and social exclusion. should be given high priority by either criterion. Poverty maps do not provide all the an- Notes swers. They must be combined with other 1. Romania, Ministry of Economy, Trade, and Tourism. 2015. “National Reform Programme 2015.” April, information, including local expertise, to ­ Ministry of Economy, Trade, and Tourism, Bucharest, inform decision making. After identifying the Romania. areas or populations in greatest need, one 2. These maps combine microdata from the 2011 popu- must understand why these places are poor. lation census and the 2011 EU-SILC survey. © 2016 International Bank for Reconstruction and Development / The World Bank. Some rights reserved. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. This work is subject to a CC BY 3.0 IGO license (https://creativecommons.org/licenses/by/3.0/ igo). The World Bank does not necessarily own each component of the content. It is your responsibility to determine whether permission is needed for reuse and to obtain permission from the copyright owner. If you have questions, email pubrights@ worldbank.org. SKU K8686 2 Country Policy Brief Poverty in Europe March 2016 Poverty and Equity Global Practice Pinpointing Poverty in the Slovak Republic Rates of poverty and social exclusion vary them effectively. However, the EC has previ- widely across European Union (EU) member ously had to rely on sub-national data at a states, and there is also a high degree of relatively high level of aggregation for pro- variability in living standards within member gram planning and the allocation of EU funds. states. In its 2014–20 multiannual financial The EC and the World Bank, in cooperation framework, the EU budgeted €1 trillion to with individual EU member states, have support growth and jobs and to reduce the developed a set of high-resolution poverty number of people living at risk of poverty maps.2 The greater geographical disaggrega- or social exclusion by 20 ­million by the year tion of the new poverty maps reveals which 2020. To help reach this goal, the government parts of these larger regions have particularly of the Slovak Republic has set a national goal high rates of poverty and require greater at- of reducing the number of the poor and so- tention in poverty reduction programs. cially excluded by 170,000 people.1 The poverty maps confirm existing knowl- Success depends on developing the appro- edge about poverty in the Slovak Republic, priate policies and programs and targeting but also reveal new insights. Previous surveys Map 1 At-Risk-of-Poverty Rates, the Slovak Republic a. Oblasts (NUTS 2) b. Districts (LAU 1) 3.4–6 6–9.4 9.4–11.7 6.3 11.7–14.2 14.2–18 6.3–13.2 18–22 13.2–14.1 22–26.2 14.1–16.7 26.2–31.6 Source: Estimates using data from the 2011 EU-SILC and the 2011 Population and Housing Census collected by the Statistical Office of the Slovak Republic. Boundary map courtesy of Geodesy, Cartography and Cadastre Authority of the Slovak Republic. Note: The risk of poverty rates are defined using the EU standard of 60 percent of median national equivalized income after social transfers. The NUTS (Nomenclature des Unités Territoriales Statistiques) classification is a hierarchical system of dividing up the economic territory of the European Union for the development of regional statistics, regional socioeconomic analysis, and the framing of EU regional policies. To date the NUTS 2 classification has been used for determining eligibility for aid from European Structural Funds. Below the NUTS 3 classification areas are defined according to Local Administrative Units (LAU). Most EU member states have LAU 1 and LAU 2 divisions, but some only have LAU 2. 1 have shown the eastern oblasts to have the Map 2  Population Living below the Poverty highest rates of poverty (map 1, panel a), and Threshold, the Slovak Republic this may also be seen in the ­ district-level poverty map (map 1, panel b). Yet, the dis- trict-level map also reveals considerably more heterogeneity in poverty incidence across space vis-à-vis the oblast-level map. In the east, the highest poverty incidence appears to be concentrated primarily along the border with the center (Rožňava, Poprád, Kežmarok) and along the Ukrainian border (Sobrance and Snina), while poverty incidence is relatively low in Košice. At the same time, 2,031–5,162 districts such as Revúca, Rimavská Sobota, 5,163–9,798 9,799–16,120 and Poltár in the center also have high pov- 16,121–25,383 erty incidence, even though poverty incidence Source: Estimates using data from the 2011 EU-SILC and the is moderate in the center overall. In 23 out 2011 Population and Housing Census collected by the Statistical of 27 districts, the district-level poverty Office of the Slovak Republic. Boundary map courtesy of Geodesy, Cartography and Cadastre Authority of the Slovak Republic. estimate is statistically different from the estimate for the oblast in which the district is located. Knowing which districts have higher poverty rates can help more efficiently tar- information, including local expertise, to get resources for development and poverty inform decision making. After identifying the reduction. areas or populations in greatest need, one Targeting poor areas alone can have limita- must understand why these places are poor. tions. Policy makers have an interest both in The reasons are likely to vary from place to areas where poverty is high and in areas that place and may include inadequate infrastruc- have the most poor people. These two need ture, lack of economic activity, an insuffi- not be the same: areas that are poor may ciently skilled workforce, or other reasons. also be sparsely populated, whereas large Poverty maps provide more finely grained cities tend to have low poverty rates, but information on sub-national variations in large numbers of poor people because of the poverty than was previously available and can large populations. Even though, in the Slovak potentially improve resource allocation. The Republic, the poverty headcount is generally maps also force more thinking on how best to correlated with the absolute size of the poor allocate resources aimed at improving stan- population, this is not universally the case dards of living, balancing the targeting of poor (map 2). Districts such as Žilina, Nitra, Trnava, areas and poor people. While the appropriate and Trenčín have low poverty rates, but rank combination of approaches will vary by coun- relatively high among districts in the abso- try, the maps provide important information lute size of the poor population. Meanwhile, to help improve policies and programs to districts such as Poltár, Sobrance, Stropkov, combat poverty and social exclusion. and Krupina have higher poverty headcounts, but represent only a small share of the total Notes population living below the risk of poverty 1. Slovak Republic, Ministry of Finance. 2015. “National Reform Programme of the Slovak Republic 2015.” threshold. April, Ministry of Finance, Bratislava, Slovak Republic. Poverty maps do not provide all the an- 2. These maps combine aggregate data from the 2011 swers. They must be combined with other population census and the 2011 EU-SILC survey. © 2016 International Bank for Reconstruction and Development / The World Bank. Some rights reserved. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. This work is subject to a CC BY 3.0 IGO license (https://creativecommons.org/licenses/by/3.0/ igo). The World Bank does not necessarily own each component of the content. It is your responsibility to determine whether permission is needed for reuse and to obtain permission from the copyright owner. If you have questions, email pubrights@ worldbank.org. SKU K8687 2 Country Policy Brief Poverty in Europe March 2016 Poverty and Equity Global Practice Pinpointing Poverty in Slovenia Rates of poverty and social exclusion vary program planning and the allocation of EU widely across European Union (EU) member funds. The EC and the World Bank, in cooper- states, and there is also a high degree of ation with individual EU member states, have variability in living standards within member developed a set of high-resolution poverty states. In its 2014–20 multiannual financial maps.2 The greater geographical disaggrega- framework, the EU budgeted €1 trillion to tion of the new poverty maps reveals which support growth and jobs and to reduce the parts of these larger regions have particularly number of people living at risk of poverty high rates of poverty and require greater at- or social exclusion by 20 million by the year tention in poverty reduction programs. 2020. To help reach this EU-wide target, the The poverty maps confirm existing government of Slovenia has set a national ­ knowledge about poverty in Slovenia, but goal of reducing the number of the poor and also reveal new insights. For example, pre- socially excluded by 40,000 people.1 vious surveys have shown the eastern and Success depends on developing the ap- southeastern parts of the country to have propriate policies and programs and target- higher poverty rates, albeit only marginally ing them effectively. However, the EC has (map 1, panel a). The municipality-level pov- previously had to rely on sub-national data erty map (map 1, panel b) shows that most at a relatively high level of aggregation for of the municipalities with the highest risk of Map 1 At-Risk-of-Poverty Rates, Slovenia a. Macroregions (NUTS 2) b. Municipalities (LAU 2) 12.3–15.9 15.9–18.56 18.56–20.69 20.69–21.7 21.7–25.1 25.1–27.47 10.7 27.47–31.51 16.1 31.51–39.23 Source: Estimates using data from the 2010 EU-SILC and 2011 Census of Population, Households, and Dwellings collected by the Statistical Office of the Republic of Slovenia. Note: The risk of poverty rates are defined using the EU standard of 60 percent of median national equivalized income after social transfers. The NUTS (Nomenclature des Unités Territoriales Statistiques) classification is a hierarchical system of dividing up the economic territory of the European Union for the development of regional statistics, regional socioeconomic analysis, and the framing of EU regional policies. To date the NUTS 2 classification has been used for determining eligibility for aid from European Structural Funds. Below the NUTS 3 classification areas are defined according to Local Administrative Units (LAU). Most EU member states have LAU 1 and LAU 2 divisions, but some only have LAU 2. 1 poverty rates are in the east of the country. Map 2  Population Living below the Poverty However, also visible are some municipali- Threshold, Slovenia ties with high poverty incidence in western Slovenia (Kobarid and Tolmin in Goriška), and considerable heterogeneity in poverty inci- dence across eastern Slovenia. For instance, in the Savinja region, poverty incidence ranges from 12 percent in the municipalities of Valenje, Vojnik, and Celje to about 25 per- cent in Rogatec, Luče, and Solčava. Similarly, the Drava region combines municipalities at low incidence, such as Starše and Lenart, with high-incidence municipalities, such as 355–2,434 Cerkvenjak and Juršinci. Knowing which sub- 2,435–5,767 5,768–12,556 regions have higher poverty rates can help 12,557–48,591 more efficiently target resources for develop- ment and poverty reduction. Source: Estimates using data from the 2010 EU-SILC and 2011 Census of Population, Households, and Dwellings collected by the Targeting poor areas alone can have limita- Statistical Office of the Republic of Slovenia. tions. Policy makers have an interest both in areas where poverty is high and in areas that have the most poor people. These two are not The reasons are likely to vary from place to the same: areas that are poor may also be place and may include inadequate infrastruc- sparsely populated, whereas large cities tend ture, lack of economic activity, an insuffi- to have low poverty rates, but large numbers ciently skilled workforce, or other reasons. of poor people because of the large popula- Poverty maps provide more finely grained tions. In Slovenia, there is an overall negative information on sub-national variations in relationship between municipal poverty in- poverty than was previously available and can cidence and the total number of individuals potentially improve resource allocation. The below the poverty threshold (map 1). Thus, maps also force more thinking on how best to municipalities such as Rogatec, Luče, and allocate resources aimed at improving stan- Solcava account for few poor individuals, de- dards of living, balancing the targeting of poor spite relatively high poverty rates. Meanwhile, areas and poor people. While the appropriate municipalities such as Ljubljana, Kranj, and combination of approaches will vary by coun- Koper, having low poverty incidence, rank try, the maps provide important information high in the absolute number of the poor on to help improve policies and programs to account of their relatively large populations. combat poverty and social exclusion. Poverty maps do not provide all the an- swers. They must be combined with other Notes 1. Slovenia, Ministry of Finance. 2014. “National Reform information, including local expertise, to ­ Programme 2014–2015.” April, Ministry of Finance, inform decision making. After identifying the Ljubljana, Slovenia. areas or populations in greatest need, one 2. These maps combine microdata from the 2011 popu- must understand why these places are poor. lation census and the 2010 EU-SILC survey. © 2016 International Bank for Reconstruction and Development / The World Bank. Some rights reserved. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. This work is subject to a CC BY 3.0 IGO license (https://creativecommons.org/licenses/by/3.0/ igo). The World Bank does not necessarily own each component of the content. It is your responsibility to determine whether permission is needed for reuse and to obtain permission from the copyright owner. If you have questions, email pubrights@ worldbank.org. SKU K8688 2 Bibliography Bedi, Tara, Aline Coudouel, and Kenneth Simler, eds. 2007. More Than a Pretty Picture: Using Poverty Maps to Design Better Policies and Interventions. Washington, DC: World Bank. Carletto, Calogero, Andrew Dabalen, and Alia Moubayed. 2007. “Constructing and Using Poverty Maps for Policy Making: The Experience in Albania.” In More Than a Pretty Picture: Using Poverty Maps to Design Better Policies and Interventions, edited by Tara Bedi, Aline Coudouel, and Kenneth Simler, 53–66. Washington, DC: World Bank. Copus, Andrew. 2014. TiPSE: The Territorial Dimension of Poverty and Social Exclusion in Europe, Final Report, Executive Summary. European Spatial Planning Observation Network, Nordic Centre for Spatial Development, and James Hutton Institute, Aberdeen, United Kingdom. CSB (Latvia, Central Statistical Bureau) and World Bank. 2014. Poverty Mapping in Latvia: Making Better Policies through Better Targeted Interventions. Poverty Mapping Brief. Washington, DC: World Bank. CSO (Central Statistical Office of Poland) and World Bank. 2014. Poverty Mapping in Poland: Making Better Policies through Better Targeted Interventions. Poverty Mapping Brief. Washington, DC: World Bank. EC (European Commission). 2010. Europe 2020: A Strategy for Smart, Sustainable, and Inclusive Growth. Brussels: Communication from the Commission (March 3), EC. http://eur-lex.europa.eu/LexUriServ/LexUriServ​ .do?uri=COM:2010:2020:FIN:EN:PDF. ———. 2014. Investment for Jobs and Growth: Promoting Development and Good Governance in EU Regions and Cities. Sixth Report on Economic, Social and Territorial Cohesion. Luxembourg: Publications Office of the European Union. Elbers, Chris, Jean O. Lanjouw, and Peter F. Lanjouw. 2002. “Micro-Level Estimation of Welfare.” Policy Research Working Paper 2911, World Bank, Washington, DC. ———. 2003. “Micro-Level Estimation of Poverty and Inequality.” Econometrica 71 (1): 355–64. Elbers, Chris, and Roy van der Weide. 2014. “Estimation of Normal Mixtures in a Nested Error Model with an Application to Small Area Estimation of Poverty and Inequality.” Policy Research Working Paper 6962, World Bank, Washington, DC. Estonia, Government Office. 2014. National Reform Programme “Estonia 2020.” May 8. Tallinn, Estonia: Government Office, Stenbock House. HCSO (Hungarian Central Statistical Office) and World Bank. 2014. Poverty Mapping in Hungary: Making Better Policies through Better-Targeted Interventions. Poverty Mapping Brief. Washington, DC: World Bank. Hungary, Ministry for National Economy. 2015. National Reform Programme 2015 of Hungary. Budapest: Ministry for National Economy. INSTAT (Albania, Institute of Statistics). 2004. Poverty and Inequality Mapping in Albania. Report. Tirana, Albania: INSTAT. Pinpointing Poverty in Europe 41 Latvia, Ministry of Economics. 2015. National Reform Programme of Latvia for the Implementation of the ‘Europe 2020’ Strategy: Progress Report. Riga, Latvia: Ministry of Economics. Litvack, Jennie. 2007. “The Poverty Mapping Application in Morocco.” In More Than a Pretty Picture: Using Poverty Maps to Design Better Policies and Interventions, edited by Tara Bedi, Aline Coudouel, and Kenneth Simler, 208–24. Washington, DC: World Bank. Lopez-Calva, Luis-Felipe, Lourdes Rodríguez-Chamussy, and Miguel Székely. 2007. “Poverty Maps and Public Policy in Mexico.” In More Than a Pretty Picture: Using Poverty Maps to Design Better Policies and Interventions, edited by Tara Bedi, Aline Coudouel, and Kenneth Simler, 208–24. Washington, DC: World Bank. Molina, Isabel, and J. N. K. Rao. 2010. “Small Area Estimation of Poverty Indicators.” Canadian Journal of Statistics 38 (3): 369–85. Morocco, Planning Commission. 2004. Carte de la pauvreté communale. Report. Rabat, Morocco: Planning Commission. NIS (Romania, National Institute of Statistics) and World Bank. 2014. Poverty Mapping in Romania: Making Better Policies through Better Targeted Interventions. Poverty Mapping Brief. Washington, DC: World Bank. Poland, Ministry of Economy. 2015. National Reform Programme Europe 2020: Update 2015/2016. April 28. Warsaw: Ministry of Economy. Romania, Ministry of Economy, Trade, and Tourism. 2015. National Reform Programme 2015. Bucharest, Romania: Ministry of Economy, Trade, and Tourism. Slovak Republic, Ministry of Finance. 2015. National Reform Programme of the Slovak Republic 2015. Bratislava, Slovak Republic: Ministry of Finance. Slovenia, Ministry of Finance. 2014. National Reform Programme 2014–2015. Ljubljana, Slovenia: Ministry of Finance. Statistical Office in Poznań, Central Statistical Office and the World Bank. 2014. Poverty Maps at the Subregional Level in Poland based on Indirect Estimation. Warsaw: Central Statistical Office. Statistical Office of the Republic of Slovenia and World Bank. 2014. Poverty Mapping in Slovenia: Making Better Policies through Better Targeted Interventions. Poverty Mapping Brief. Washington, DC: World Bank. Statistical Office of the Slovak Republic and World Bank. 2014. Poverty Mapping in Slovakia: Making Better Policies through Better Targeted Interventions. Poverty Mapping Brief. Washington, DC: World Bank. Statistics Estonia and World Bank. 2014a. Poverty Mapping in Estonia: Making Better Policies through Better Targeted Interventions. Poverty Mapping Brief. Washington, DC: World Bank. ———. 2014b. Small Area Estimates of Poverty in Estonia: Methodological Report. Tallinn, Estonia: World Bank. Swinkels, Rob and Carrie Turk. 2007. “Poverty Mapping: The Experience of Vietnam.” In More than a Pretty Picture: Using Poverty Maps to Design Better Policies and Interventions, edited by Tara Bedi, Aline Coudouel, and Kenneth Simler, 261–86. Washington, DC: World Bank. 42 Pinpointing Poverty in Europe World Bank. 2004. Kingdom of Morocco Poverty Report: Strengthening Policy by Identifying the Geographic Dimension of Poverty. Report 28223-MOR, Social and Economic Development Group, Middle East and North Africa Region. Washington, DC: World Bank. ———. 2013a. Small Area Estimates of Poverty in Hungary: Methodological Report. Washington, DC: World Bank. ———. 2013b. Small Area Estimates of Poverty in Latvia: Methodological Report. Washington, DC: World Bank. ———. 2013c. Small Area Estimates of Poverty in Romania: Methodological Report. Washington, DC: World Bank. ———. 2014a. EU Accession Countries: Poverty Mapping of New Members in EU, Completion Memo. Report ACS9402 (July 11). Washington, DC: Europe and Central Asia Region, World Bank. ———. 2014b. Small Area Estimates of Poverty in the Slovak Republic. Washington, DC: World Bank. ———. 2014c. Small Area Estimates of Poverty in Slovenia: Methodological Report. Washington, DC: World Bank. Pinpointing Poverty in Europe 43 SKU K8619