52882 v1 Report No. 37992 Croatia: Living Standards Assessment Volume 1: Promoting Social Inclusion and Regional Equity November 2006 Poverty Reduction and Economic Management Unit Europe and Central Asia Region Document of the World Bank CURRENCY AND EQUIVALENT UNITS (as of May 30, 2006) Currency Unit = HRK 1 US$ = 5.67 HRK ABBREVIATIONS ALMM Active Labor Market Measures ALMP Active Labor Market Program ASSC Areas of Special State Concern BEEPS Business Environment and Enterprise Performance Survey CBS Central Bureau of Statistics CSW Centers of Social Welfare CARDS Community Assistance for Reconstruction, Development, and Stabilization CBN Cost of Basic Needs CES Croatian Employment Services HRK Croatian kuna DDP Development Data Platform ECA Europe and Central Asia EBRD European Bank for Reconstruction and Development EC European Commission FINA Financial Agency GDP Gross Domestic Product GNP Gross National Product GVA Gross Value Added NBS Household Budget Survey IT Information Technology IBRD International Bank for Reconstruction and Development IMF International Monetary Fund LFS Labor Force Survey NMS New Member States NUTS Nomenclature des Unites Territoriales Statistiques NGO Non-governmental Organization OECD Organization for Economic Co-operation and Development PEP Pre-Accession Economic Program PPP Purchasing Power Parity PPS Purchasing Power Standards ROP Regional Operations Program SME Small and Medium Enterprise UNESCO United Nations Educational, Scientific, and Cultural Organization USD US dollar Vice President Shigeo Katsu Country Director Anand K. Seth Sector Director Cheryl W. Gray Sector Manager AsadAlam Task Leader Salman Zaidi iii PREFACE AND ACKNOWLEDGEMENTS This report is the product of a collaborative process involving staff from the Croatian Government (Central Bureau of Statistics, Ministry of Economy, Labor, and Entre-preneurship, Ministry of Health and Social Welfare, Ministry of Sea, Transport, Tourism, and Development and Ministry of Finance), Croatian and other consultants, and staff from the World Bank. The scope of the work was developed in consultation with the Government, Croatian academics and researchers, and staff from the European Commission in Zagreb. Background papers for this report were prepared by Danijel Nestic, Giovanni Vecchi, Juan Munoz, Zeljko Lovrincevic, Davor Mikulic, Vedran Sosic, and Xubei Luo, and are included as volume 2 of this report. Early findings of the background work were presented at several workshops held in Zagreb attended by government and non-government officials, including key donor community representatives (UNDP, EC, etc) and the team is grateful for all the suggestions and feedback received from the various participants of these workshops. The team would like to thank Ms. Vera Babic (State Secretary, Ministry of Economy, Labor, and Entrepreneurship), Ms. Dorica Nikolic (State Secretary, Ministry of Health and Social Welfare), Ms. Vesna Mastela Buzan (Senior Advisor, Ministry of Health and Social Welfare), Ms. Franka Vojnovic (Head of Unit, Ministry of Sea, Tourism, Transport and Development) and Mr. Darko Jukic (Acting Director, Central Bureau of Statistics) and his staff at the Central Bureau of Statistics for their comments, advice, and help at various stages in the preparation of this report. Appreciation is also due to Mr. Zeljko Bacic and the State Geodetic Directorate, who provided the county-level maps used in this report. The World Bank task team included Salman Zaidi (task team leader), Xubei Luo, Sanja Madzarevic-Sujster, Necmeddin Bilal Erdogan, Tomislava Ujevic , Anton Marcincin, Ljiljana Tarade, Dubravka Jerman, and Helena Makarenko. In addition the team also benefited from useful discussions and contributions from Paula Lytle, Vera Dugandzic, Gloria La Cava, Inguna Dobraja, and Ivan Drabek. Rachel Weaving helped edit the report. The report was undertaken under the guidance of Anand Seth, Country Director, Cheryl Gray, Sector Director, and Asad Alam, Sector Manager. Vera Babic, Enrique Aguado Asenjo and Pierella Pad were the peer reviewers of the report. In addition, the team would like to thank Satu Kahkonen, Bernard Funck, Hermann von Gersdorff, Myla Taylor Williams, Albert Martinez, and all the staff of the Zagreb office for their comments, advice, and help throughout the course of this work. v CONTENTS 1. POVERTY LEVELS AND TRENDS .............................................................................................................. 1 1.1 RECENT ECONOMIC DEVELOPMENTS AND CHALLENGES ..................................................................... 2 1.2 SECTORAL AND REGIONAL TRENDS IN OCTPUT, INCOME, AND EMPLOYMENT ...................................... 6 1.3 CHANGES IN POVERTY AND INEQUALITY: 2002-04.......................................................................... ,9 1.4 POVERTY IN CROATIA: AN INTERNATIONAL PERSPECTIVE ................................................................ 12 1.5 THE CmlVERGENCE CHALLENGE: REDUCING THE INCOME GAP WITH EUROPE .................................. 14 2. PROFILE OF POVERTY AND INEQUALITY IN EARNINGS .......................................................................... 17 2.1 PROFILE OF THE POOR .................................................................................................................. 17 2.2 POVERTY AND REGION OF RESIDENCE ........................................................................................... 21 2.3 ECONOMIC ACTIVITIES AND INCOME SOURCES OF THE PooR............................................................ 23 2.4 EMPLOYMENT, EARNINGS, AND REGIONAL DISPARITIES .................................................................. 25 3. THE TwIN CHALLENGES OF EXTERNAL AND INTERNAL INCOME CONVERGENCE ................................... 30 3.1 POOR PEOPLE OR POOR REGIONS? ................................................................................................ 32 3.2 How SIGNIFICANT ARE REGIONAL DISPARITIES IN LIVING CONDITIONS? .......................................... 35 3.3 SCSTAINING HIGH GROWTH THROUGH FASTER JOB CREATION ....................................................... .40 3.4 BUILDING ON LoCAL COMPARATIVE ADVANTAGES ........................................................................ .43 3.5 LABOR MOBILITY AND GREATER LABOR MARKET FLEXIBILITY ..................................................... .46 3.6 IMPROVING THE ADEQUACY AND TARGETING OF SOCIAL SAFETY NETS ........................................... .47 APPENDIX 8 FIGURES AND TABLES Figure 1-1: Good economic performance over the past decade ................................................................ 2 Figure 1-2: Total employment rate in Croatia falls well short of the Lisbon target... ............................... 4 Figure 1-3: High Long-term Unemployment. .......................................................................................... .4 Figure 1-4: Growth in total employment in Croatia has lagged behind GDP growth and wages ............. 5 Figure 1-5: Total employment by sector, 2003 ......................................................................................... 6 Figure 1-6: Rising regional inequality in GDP per capita ........................................................................ 7 Figure 1-7: Small firms in Croatia have been quite dynamic in job creation ........................................... 8 Figure 1-8: High county-level market SME shares were associated with higher growth in 2001-03 ...... 8 Figure 1-9: Survey-based estimates show that income grew much faster than consumption ................. 11 Figure 1-10: Cross-country poverty comparisons ................................................................................... 13 Figure 1-11: Growth in per capita GDP .................................................................................................. 14 Figure 1-12: GDP per capita: Croatia and selected EU countries .......................................................... 15 Figure 1-13: Convergence and demographics ......................................................................................... 15 Figure 1 14: Convergence, labor market participation, and productivity ............................................... 16 Figure 2-1: Poverty incidence and breakdown by size of household ...................................................... 17 Figure 2-2: High poverty incidence among the elderly .......................................................................... 18 Figure 2-3: Pension receipts and poverty incidence among the elderly ................................................. 19 Figure 2-4: Poverty incidence and breakdown by educational attainment of household head ............... 19 Figure 2-5: Regional variation in poverty incidence across Croatia ....................................................... 21 Figure 2-6: Regional variation in selected welfare indicators ................................................................ 22 Figure 2-7: Sources of income for poor and non-poor households ........................................................ 23 Figure 2-8: Poverty incidence by employment status of the head of household .................................... 24 Figure 2-9: Poverty shares by employment status of the head of household.......................................... 24 Figure 2-10: Low monthly earnings for women and youth .................................................................... 26 vii Figure 2-11: Distribution of monthly earnings, by region ...................................................................... 27 Figure 2-12: People in the eastern and central regions have lower education attainment ......................28 Figure 3-1: Poverty incidence at the county leveL ................................................................................ 31 Figure 3-2: Average per capita incomes at the county level generally seem to be correlated with thecducational attainment of the population ................................................................................. 32 Figure 3-3: Simple vs. partial correlations .............................................................................................. 33 Figure 3-4: The "gateway effect" rising regional inequalities ............................................................... 37 Figure 3-5: Redistributive role of taxes and transfers ............................................................................. 38 Figure 3-6: Social transfers reduce regional inequality .......................................................................... 39 Figure 3-7: Social assistance and unemployment benefits are generally well targeted geographically.40 Figure 3-8: Robust regional rankings based on various development indicators ...................................44 Figure 3-9: Social assistance is the best targeted program in Croatia ................................................... .48 Figure 3-10: All programs have relatively low coverage rates, even among the poor ........................... .48 Figure 3-11: Importance of transfers for beneficiaries of the social assistance program ...................... .49 Figure 3-12: Who benefits from social transfer programs? ................................................................... .49 Table 1-1: Croatia: Key economic indicators ............................................................................................ 3 Table 1-2: Key labor force indicators for Croatia: 2000 - 04 .................................................................... 5 Table 1-3: Regional classification of counties in Croatia ......................................................................... 6 Table 1-4: Poverty in Croatia is quite limited and shallow ..................................................................... 10 Table 1-5: About 1 in 20 Croatians consume well below the national poverty line ............................... 10 Table 1-6: Poverty incidence in Croatia .................................................................................................. 11 Table 1-7: Subjective welfare measures indicate improved living conditions ........................................ 12 Table 1-8: Income inequality remained fairly stable between 2002 and 2004 ....................................... 12 Table 1-9: International comparisons of poverty and inequality ............................................................ 13 Table 1-10: Key social indicators: cross-country comparison ................................................................ 14 Table 2-1: Employment, unemployment, and monthly earnings by educational attainment.. ................ 25 Table 2-2: Labor market disadvantages of women ................................................................................. 26 Table 2-3: Selected labor market indicators, by region .......................................................................... 26 Table 2-4: Employment. unemployment, and monthly earnings, by educational attainment.. ............... 27 Table 2-5: Oaxaca-Blinder Decomposition: Effect of individual vs. other characteristics on earnings .29 Table 3-1: Relative poverty risk by educational attainment of the household head ............................... 34 Table 3-2: Relative poverty risk by region of residence ......................................................................... 35 Table 3-3: Regional Variation in GRP per capita, selected countries ..................................................... 36 Table 3-4: Alternative lists of five poorest counties in Croatia .............................................................. .43 BOXES Box I. 1: Main Topics Covered in the Background Papers ....................................................................... 1 Box 1.2: Tourism in Croatia ...................................................................................................................... 3 Box 1.3: Deriving an Absolute Poverty Line for Croatia.......................................................................... 9 Box 2.1: Poverty and the elderly in Croatia: Some key insights from the 2002-04 RBS data ............... 20 Box 2.2: Understanding Differences across Groups through the Oaxaca-Blinder Decomposition ........ 29 Box 3.1: Regional Inequalities in Croatia: Legacy of the Past ............................................................... 35 Box 3.2: Polices, Institutions, and Job Creation: Some Lessons from Cross-country Experience ........ .41 Box 3.3: Doing Business: Objective Measures of Business Regulations and Their Enforcement ........ .42 Box 3.4: Regional Operational Programs in Croatia ..............................................................................45 Box 3.5: Reforming the Social Welfare System in Croatia ....................................................................50 EXECUTIVE SUMMARY The Croatian economy has petformed moderately well in the past decade, enabling a gradual narrowing of the income gap with the European Union. Using a cost-of-basic-needs poverty line, poverty in Croatia is found to be low, with only a small proportion of the poor facing hard-core deprivation. Looking ahead, the task offaster external income convergence with the EU will be challenging, and will require both faster job creation as well as flexibility in the allocation ofjobs and workers in the economy. These will also help with more rapid improvement in living conditions in lagging regions. To these ends, the report highlights three sets of interrelated policy challenges and priorities: (1) sustaining high rates of growth to permit continued income convergence with Europe; (2) promoting greater labor mobility, including measures aimed at building human capital to improve workers' opportunities; and (3) improving the adequacy and effectiveness of social safety nets within a responsiblefiscalframework. In examining regional disparities, several development indicators show that regional disparities in living conditions are significant (though on average no higher than in EU countries), and only partially explained by human capital and other such individual attributes. Building on local comparative advantages offers the best way forward to improve living conditions in lagging regions. ]. As Croatia prepares to join the European Union, the government is working on (]) a joint inclusion memorandum, and (2) a national regional development strategy outlining plans for reducing internal disparities. This report offers data and analysis relevant for both these tasks, shedding light on the strategic priorities facing the government and some of the main trade-offs involved. POVERTY IN CROATIA IS RELATIVELY Low ... 2. The Croatian economy has performed quite well over the past decade (Figure 1). Real GDP per capita increased by more than 40 percent between 1996 and 2005. Inequality in Figure 1: GDP per capita growth both consumption and income on the other hand has been quite stable and 8 low between 2002 and 2004. During 6 this period, the poverty headcount rate I appears to have either stagnated (based 4 on consumption per adult-equivalent), or 2 fallen at about 1 percentage point a year O-!-'--.,.......................,..........---.......--.,..................'--..........---...........--, (based on income per adult-equivalent) -2 1997 1998 1999 2000 2001 2002 2003 2004 2005 between 2002 and 2004. The latter trend -4 CJ EU-25 · Croatia is more plausible, given the economy's good growth performance over this period. In addition, subjective measures of well being, recorded by successive household surveys, point to some improvement in living standards over time. 3. Poverty in Croatia is found to be quite limited. Some 1] percent of the population is poor, and another 10 percent is at risk of poverty in the sense that their average consumption level is less than 1 Using the latest available data (2004 Household Budget Survey) and a poverty line of 22,145 kunas (PPP$4.343) per equivalent adult per year derived using the cost of basic needs methodology. 25 percent above the poverty line. Some groups, numbering about 1 percent of the population, face severe deprivation. 4. Using an internationally comparable Figure 2: Cross-country poverty comparisons poverty line for the region, Croatia's MOLDOVA poverty rates are the lowest among the 80 World Bank's client countries in Europe ALBANIA and Central Asia. 2 This is in part because ~ 80 eTURKEY ROMANlA the Gini coefficient for Croatia is low compared to those for other countries ~ ~- ~ ~.o SERBIA . RUSSIA in the region. Croatia's low poverty rate (around 4 percent, if an international I J: 20 -8 POLAND ARIA ESTONIA.MACEDONIA .HUNGAR poverty line of $PPP 4.30 per day per person in 2004 prices is applied) 1000 2000 3000 4000 matches what one might expect from Consu(~~~OU~b) capita Croatia's consumption-per-capita level Source: World Bank (2005) and HBS 2004. (Figure 2). Social indicators for the country are broadly in line with those Figure 3: Lagging employment in Croatia (1990=100) in the upper middle-income countries '20 --mp of the region. For instance, infant and 110 - - "R«IMl wa~ --Ovilian employment child mortality rates are lower than in 100 Latvia or Hungary, but higher than in 8. the Czech Republic or Slovenia. 10 5. Job creation, however, has lagged behind GDP growth and wages (Figure ,. 3). At only 54.7 percent, Croatia's .990 .992 1994 (996 199B 2000 2002 2004 employment rate is one of the lowest in Source: Central Bureau of Statistics. Index, 1990=100. Europe, and the proportion of long-term unemployed (Le. workers without jobs for 12 months or longer) is higher than in all EU countries except Poland and Slovakia. A PROFILE OF POVERTY AND INEQUALITY IN EAAA'lNGS 6. Household size, education, age, and employment status ofthe household Figure 3: Lagging employment in Croatia (1990=100) head are important correlates of '20 --mp - .. .. Real net waFS poverty. Households that are small (1-2 110 - - QviliaJ'J employmenl .00 people) and large (more than 6 people) 90 face greater poverty risks than others. IiO More than three-quarters of the poor 10 live in households headed by individuals 60 with only primary or less education. And poverty is tightly associated '" 40 with the activity status of the head of 1990 '992 1994 199fji 1998 2{)()() 2002 2004 household (Figure 4); those household Source: Central Bureau of Statistics. Index, 1990=100. heads who are retired, unemployed, or 2 At least among those eountries for which household survey data are available. x economically inactive feature prominently among the poor. 7. The risk ofpoverty increases with age. The incidence of poverty is highest among households headed by the elderly, who face a poverty risk twice the average. Within Figure 5: Poverty and the elderly households headed by elderly people, 65 62 60 those who do not receive pensions are particularly vulnerable, facing a poverty risk more than five times the national average (Figure 5). Even those receiving pensions have almost twice the national average poverty risk. Retired people are more vulnerable to poverty in rural areas than in urban. 0-'--""""'-- 8. A high percentage ofindividuals, not 16-30 31-49 50-64 65+ pens. 65+ no pens. Age of Household Head only among the poor, have limited or no access to basic services. One fourth of Source: World Bank estimates based on 2004 HBS. the poor live in dwellings without water supply, and 8 percent live in dwellings Figure 6: Sources of income: poor and non-poor without toilets, and 8 percent without a telephone line. In general, as one might have expected, access to services in Croatia is considerably worse in rural areas compared to urban, and for the poor compared to the non-poor. 9. Poverty and labor market status: The poor in Croatia get more than half their income from transfers and relatively little from productive activities (Figure 6). This is consistent with their low rates wages pensions imputed ren1 of labor force participation and their Source: World Bank estimates based on 2004 HBS. relatively advanced age. Households headed by the unemployed are a small but very vulnerable group. 10. Women and youth face clear disadvantages in the labor market. Women have higher unemployment rates and lower wages (Table I) than other labor market participants. Employment rates among youth and average earnings are particularly low (Figure 7)-fewer than one in four people aged 15-25 years have jobs, and Table 1: Labor market disadvantages of women the unemployment rate for this age group Men Women is three times that of the age group 25 Participation rate (%) 71.9 58.6 50. Employment rate (%) 61.4 47.8 Source: World Bank estimates based on pooled Unemployment rate (%) 13.0 16.4 2002-04 LFS. Monthly earnings (kunas) 3514 2 978 1 xi Figure 7: Low monthly earnings for women and youth ~ o 10000 1--..... ----- F....... I Source: World Bank estimates based on pooled 2002-04 LFS. THE CHALLENGE OF EXTERNAL INCOME CONVERGENCE 11. Though Croatia has been narrowing the income gap with the EU25 average, sustaining this trend is likely to be challenging. Projected Figure 8: Convergence of GDP per capita declines in the country's population 18.000 and labor force suggest that per capita income is likely to decline as the century 16.000 advances (Figure 8, scenario A). To 14.000 prevent this from happening, both the 12,000 ~ / employment rate and labor productivity 10.000 / B. Employment rale rises to: will need to rise substantially. If 8.000 A. Base-c&'ie Lisbon tIItget trend Mth Croatia's total employment rate were to 6.000 shrinking labor rise to 70 percent by 2025, this would force 4,000 help overcome the adverse impact of 2.000 the decline in working age population o (Figure 8, scenario B), but even under 2005 2010 2015 2020 2025 this scenario, Croatia's per capita GDP Source: World Bank estimates based on databases of the World would barely reach two-thirds of the Bank and World Bank population projections, EU25 level by 2025. If the goal is to attain higher living standards, output per worker would have to increase substantially too (i.e. in addition to the projected increase in employment rate). 12. To these ends, the report highlights three sets of interrelated policy challenges and priorities: (I) sustaining high rates of growth to permit faster external income convergence; (2) promoting greater labor mobility, including measures aimed at building human capital to allow workers access to better opportunities, both within and outside the country; and (3) improving allocation of spending on social safety net programs within a responsible overall fiscal framework. (1) Sustaining high growth through/aster job creation 13. To improve labor utilization, Croatia needs to adopt a fairly wide menu of policy measures to hclp foster job creation: addressing the remaining discouraging features of the investment climate, including by facilitating the registration of property and by improving the efficiency of the xii legal system; helping real wages to adjust to productivity and local market conditions (particularly wages in public enterprises, where the restructuring agenda remains incomplete); and allowing lower wages to be paid to young workers or those in economically depressed regions (through employment subsidies)-in preference to their having no jobs at all. To meet the challenge of job creation would also require the government to address its fiscal problem, which in tum has been due to problems with an unreformed health sector, rising social spending, and continued state support to loss-making state-owned enterprises. Government spending in Croatia accounts for about half of GDP-a large share even by European standards. Total subsidies to the enterprise sector constituted 3.4 percent of GDP in 2003, well over half the overall government deficit of 6.2 percent during this fiscal year. Between 2000 and 2004, Croatia's external debt rose from 61 to 88 percent of GDp' 4 (2) Labor mobility and greater labor market flexibility 14. Multivariate analysis of labor force Figure 9: Average schooling by region survey data shows that as much as one third of the wage differential s between Adriatic South Region · · · · · · · · · · · leading and lagging regions in Croatia can be attributed to differences in worker Altlatic I-bth Region · · · · · · · · · · · characteristics, including education (Figure 9). This finding points to the ag~Region.1I1I1I1I1I1I1I1I1I1I1I1I importance of raising the human capital Eastern Region · · · · · · of workers in lagging regions, as well as of facilitating greater labor mobility as Central Region · · · · · · a means to reduce interregional earning 9.5 10 10.5 11 11.5 12 disparities. Public policy can play an important role in reducing barriers Source: World Bank estimates based on pooled 2002-04 LFS. to workers' interregional migration, through improving information about job opportunities in other regions, reducing housing market imperfections and distortions, and improving transport links. Conversely, social assistance programs that pay inadequate attention to work-disincentive issues may reduce the mobility of labor across regions. 15. Similar analysis of the earnings differentials between public and private sector workers shows that the average monthly earnings of workers in state firms are about 30 percent higher than those of workers in the private sector. Furthermore, differences in employee characteristics-such as age, years of education, and gender-at best explain only about one-third of the earning differential between these two sectors, while the remainder is unexplained. This in tum points to the problem of incomplete restructuring of public sector enterprises in Croatia. The wage differential between these enterprises and other firms may hinder the reallocation of workers across sectors and regions. Policies aimed at correcting these market imperfections can have an important role to play in 3 Croatia's overall investment climate and business environment are perceived as relatively favorable and improving. In 2003, the government introduced a responsible wage policy, which reversed the trend of rising unit labor costs, and reduced non-wage labor costs. In addition, a new Labor Code was introduced, which is better aligned with EC guidelines and practices in EU member countries. 4 In response to the country's rising external debt and vulnerability, the government has begun to reduce the public sector deficit from 6.2 percent of GDP in 2003 to 4.1 percent in 2005 through slower wage bill growth, lower transfers and public investment, and structural reforms. xiii facilitating a freer flow of workers. When implementing such policies, it would be important to ensure the presence of effective social safety nets to protect the incomes of workers displaced in the wake of enterprise restructuring. (3) Improving allocation ofspending on social safety net programs 16. A key feature of the government's program to reduce poverty and social exclusion is to improve the effectiveness of the social protection system, by (1) rationalizing spending and improving targeting, and (2) strengthening system administration. Social spending in Croatia is high by international standards, at 31 percent of GDP, but not well targeted to needy people. Only a small share of social spending-O.7 percent of GDP-goes to means-targeted programs. Data for 2003 show that pensions and health insurance compensation were the two largest categories of transfers, accounting for 51 percent and 5 percent respectively of total incomes from transfers, while child allowances (about 4 percent), social assistance (about 3 percent), and unemployment benefits (about 2 percent) were the next most important. 17. Social assistance is the best-targeted program in Croatia (Figure 10). However, total transfers under this program are small in relation to other programs. As a result, the social assistance program does not reach the great majority of poor people: it covers only about 13 percent of the poorest fifth of the population. While cash social benefits total about 4 percent of GDP (i.e. far higher than the average level of 2.5 percent in Austria, Italy, Germany, Slovakia, Poland and Hungary), only about 0.26 percent of GDP is allocated to the means-tested social assistance program (EU countries spend 1.4 percent of GDP on average on similar programs). Yet for its poor beneficiaries, the social assistance program is an important income source, supplying 15 percent of their average consumption per adult-equivalent. Figure 10: Social assistance is tbe besMargeted program in Croatia··· Sbare ofspendlngllC.'C'l"Wlng to the poorest OIle~ftftb of the 70 population 60 SO 70 ~O 60 50 )40 34 44 fO 30 · QveraH population. Poorest one~firth 3029 30 20 20 10 10 O1ikiAUo'MU1ct 0I1'1er family Soc... Unem:ployme'H Any of these 0" AIIoWft.l!lt;e!l Ass3solance beneflls benefits Otitd Allowance Other Family Unemployment Any of thez .2002 .2004 AUOMnce5 benefJls benefits ... but bas very low coverage rates, even among tbe poorest one-fiftb of tbe population (rigbt panel) Source: World Bank estimates based on 2002 and 2004 HBS. 18. The report recommends a four-pronged strategy to improve the system of social safety nets: · Spend relatively less on child allowances and child tax benefits, by improving their targeting to needy families, and spend relatively more on social assistance, which given that the overall depth of poverty is not that high, can go a long way towards eliminating poverty in the country; xiv · Increase program coverage rates in poorer regions by directing a larger share of consolidated central and local government social expenditures towards these regions; · Improve system administration (e.g. through better integration of centers for social welfare and employment services, greater clarity of their respective roles, better use of information technology; better integration of national and local social programs, exchange of information, etc); and · Use more active measures-such as better counseling and job-placement services-to help reduce long-term joblessness and promote labor market participation. How SIGNIFICANT ARE REGIONAL DISPARITIES IN LIVING STANDARDS? 19. Output, employment patterns, and living conditions obviously differ across Croatia's regions, though not significantly more than in the EU on average (Table 2). Table 2: Regional Variation in GDP per capita, selected countries Poorest Richest Coefficient Country # of units Year (Bum) (Bum) Ratio of variation Bulgaria 6 2002 1 701 3054 l.79 0.26 France 22 2002 19111 38854 2.03 0.l8 Spain 19 2002 11 214 23077 2.05 0.20 Poland 16 2002 3708 8067 2.17 0.22 Croatia 5 2003 3,983 8799 2.21 0.35 Italy 21 2002 13 697 32279 2.35 0.26 Romania 8 2002 1 751 4603 2.59 0.42 UK* 37 2002 17268 45028 2.60 0.23 Germanv 41 2002 15638 44151 2.82 0.26 Turkey 26 2001 730 3063 4.19 0.42 Russia 7 2003 1 129 5743 5.11 0.58 *NUTS2 regions of inner and outer London are merged. Except in Russia and Croatia, regions are defined at the NUTS2 level. Russian regions are defined on the basis of the seven official administrative regions. Croatian regions are the analytic regions described in the report. 20. Differences across regions (Table 3)5 reflect diverse factors such as the impact of the recent war and varied geographic terrain, and as noted earlier, differences in the educational attainment and human capital of their residents. The economies of the most developed counties-Zagreb city, Istria, and Primorsko-Goranska-have large services sectors, active small and medium size enterprises, and large-scale entrepreneurs, and appear well positioned to continue growing rapidly. Counties in Adriatic Croatia appear to be similarly well positioned, and are likely to benefit from 5 Following the same analytic regional classification as used in the World Bank's earlier Economic Vulnerability and Welfare Study, this report uses a breakdown of five main geographic regions to examine regional variation in living conditions. While representing only one of several possible regional classification schemes, this grouping provides a useful summary of regional variation, but at the same time also ensuring that the sample size in each group is large enough to permit reasonable statistical accuracy. This grouping also follows logical geographical, administrative, social and economical convergence characteristics. xv continued rapid growth in tourism. By contrast, counties in Eastern and Central Croatia have a less favorable economic structure, with a relatively large share of agriculture, and appear to face a significant risk of lagging further behind. Table 3: Selected indicators, by region HBS 2004 Labor Force Surve s Consumption Analytical capita (2003, Employment Unemployment per capita per Region Croatia= I 00) rate (%) rate (%) ear Central 42602 60.8 11.5 Eastern 37593 47.9 19.9 Za reb 58584 55.9 11.8 Adriatic North 52703 58.5 9.9 Adriatic South 81 48.9 20.2 Croatia 47,326 54.5 14.5 21. County and region-level data paint a much finer picture of these disparities. Figure 11 shows, for example, that while the Eastern and Central regions are both poor in monetary terms and human capital endowments, unemployment rates are quite high in the former but much lower in the latter. Similarly, while the two Adriatic analytic regions have similar levels of earnings per worker and human capital, unemployment rates are considerably higher in the south. Figure 11: Variation in selected indicators by county and region Poverty Headcount Rate (2002-04 HBS) Ave. income per-capita (Croatia=l00, 2002-04 HBS) GDP per capita (2003) Post-Secondary Completion Rate (2002-04 LFS) xvi Unemployment Rate (2002·04 LFS) Average Monthly Earnings (2002-04 LFS) Source: World Bank estimates based on pooled 2002-04 HBS and LFS. 22. Headcount poverty rates vary across regions, from less than 3 percent of the population in urban Zagreb to more than 20 percent in the rural Central and Eastern regions. Ranking regions by level of poverty incidence, the following broad typology emerges: · Low poverty: Zagreb Region, Adriatic North, and urban Adriatic South (taken together, comprising about half Croatia's population, but only one-sixth of the poor); · Moderate poverty: Rural Adriatic South, urban Central, urban East (roughly one-fourth of Croatia's population and about 30 percent of the poor); · High poverty: Rural Central and Rural Eastern region (roughly one-fourth of the population, but more than half of the poor). 23. Trends in per capita GDP show a small rise in interregional inequality in Croatia between 2001 Figure 12: GDP per capita by region 170 .2001.2003 and 2003 (Figure 12). The ratio between GDP 150 per capita in the highest and lowest ranked region rose marginally from 2.1 to 2.2 during these two r 130 years. This should not be surprising. Most of the new member states of the EU experienced I 110 '!l 90 the so-called "gateway effect" during the period surrounding their accession to the European '" 70 Union, whereby their capital cities attracted much 50 Zagreb Central Adriatic Adriatic II Eastern interest from foreign investors, and grew much North Sovlh more rapidly than their other regions. Indeed, Croatia's regional disparities may not have widened by as much as those of the new member states. Seen in this light, the fact that some parts of Croatia have been growing faster than others is not a cause for concern, particularly because some of the benetlts of high growth in these regions have been and can be shared by other regions through transfers. xvii 24. While regional disparities in GDP per capita may appear stark at first sight, further analysis reveals that primary incomes and gross disposable incomes in Croatia are in fact much more equally distributed, because taxes, social security contributions, and social and other transfers help to equalize incomes Figure 13: Redistributive role of taxes and transfers in Croatia across regions (Figure 160 13). This in turn translates 140 into smaller regional vari 120 ations in consumption per National A v~ capita. 100 80 25. Worse-than-average III 60 living standards in the 40 Central and Eastern re 20 gions are linked to the poorer labor market out o -s i I ~ ~ j i I ~ j ~ 1 ~ j '" -6 comes in these regions, i B with lower participation j j « ! j « ! I rates, lower employment GDP Primary income Disposable income rates,6 and lower wages. Source: World Bank staff estimates. All estimates are for 2003, and are Multivariate analysis ba expressed in per capita terms. sed on data from labor force surveys shows that only about one-third of the total variation in earning across Croatia's regions can be explained by differences in human capital and other such individual characteristics of workers. The remainder of the earnings differential must be explained by other factors, induding geographical ones (in most EU countries, the remainder is typically much lower than what was observed in Croatia). County-level estimates of poverty incidence and its correlates, derived from household budget survey data, allow a more detailed exploration of the causes of poverty-whether specific to individuals or specific to regions. The results are consistent with those of the earlier exploration based on earnings data from labor force surveys: key correlates of poverty--employment status, age of household head, and household size--explain some but not all of the observed differences in poverty levels across regions. Thus other region-specific factors, such as variations in the quality of local infrastructure, likely play an important role in differences in regional living standards. Building on local comparative advantage 26. Unexplained regional differences in earnings and living standards suggest that a focus on regional development makes good sense for Croatia. In the context of preparing national regional development analysis and planning, an important question is how best to define regional disparities and at what level of territorial aggregation. 7 It appears from our analysis that counties in Croatia are probably too small a unit for this purpose, if only because their relative rankings vary with the indicator of living standards that is chosen (Table 4). For example, not one of the five counties with the highest poverty rates as calculated using the household survey data features in the list of five poorest counties as ranked by per capita GDP. 6 Though not in the Central region-see Table 3. 7 The government proposal of the National Regional Development Strategy aims to introduce a consistent evaluation and categorization system for all local and regional government units. xviii Table 4: Alternative lists of five poorest counties in Croatia Ranking GDPper Per capita Per capita Ave. monthly Headcount (I =poorest) ca.Qita {2003} income consumQtion eami~ Poverty Rate {% 1 Vukovar- Virovitica- Virovitica- Karlovac Karlovac Sirmium Podravina Podravina SI. Brod- Virovitica- 2 Posavina Osijek-Baranja Podravina Varazdin Sisak-Moslavina 3 Sibenik-Knin Karlovac Sisak-Moslavina Bje\ovar-B ilogora Bjelovar-Bilogora Koprivnica- 4 Pozega-Slavonia Sisak-Moslavina Krapina-Zagorje Koprivnica-Krizevci Krizevci 5 Kral2ina-Za~orie Pozega-Slavonia Osijek-Baranja SI. Brod-Posavina Osijek-Baranja 27. By contrast, conclusions are more Figure 14: Robust regional poverty rankings consistent and robust at the regional level: 25 Poverty Rate (2002104) the Eastern and Central parts of the country 20 consistently appear disadvantaged. These 15 II regions' poverty rates are above the national I) average, those in the Adriatic South are at ·· par with, and those in the Adriatic North and Zagreb regions are lower than, the national average (Figure 14). Central Eastern Acltalic South Zagreb Adriatic North 28. Initiatives to boost growth in lagging Source: World Bank estimates based on 2002-04 HBS. regions are most likely to succeed if they work with the forces of economic geography, through building on local comparative advantages and drawing on local knowledge and initiative. In Croatia, most of the regional operational programs (ROP) are being developed at the county level over the period 2004-10 (some started in earlier years); local projects are planned through a highly consultative and collaborative process, and steps are taken throughout the process to help ensure that the projects are well aligned with the county's comparative and competitive advantages. In this respect, the ROP exercise is fully consistent with these strategic principles. However, the comparative advantage of a region may lie across several different counties. Looking ahead, as the Croatian government begins to prepare higher-level regional development projects, it will be important to give high priority to initiatives that build upon commonalities across contiguous counties (that is, in terms of areas of comparative advantage identified), so as to avoid bogging down the overall regional development effort in an undue proliferation of mini-projects. xix 1. POVERTY LEVELS AND TRENDS Croatia's economy has peiformed moderately well in recent years, enabling a gradual narrowing of the income gap with the European Union. But significant structural challenges remain, in particular that of raising employment. Peiformance has varied among regions, reflecting variations in the growth of different sectors of the economy and, possibly, employment creation patterns. Using a cost-of-basic needs poverty line specially derived for Croatia, we find poverty in Croatia to be quite limited and shallow. Estimates ofpoverty based on consumption per adult equivalent have remained virtually unchanged in recent years, but estimates based on income per adult-equivalent suggest that poverty has declined, and the latter trend is more plausible. Using an international poverty line to compare countries in the Europe and Central Asia region, we find poverty in Croatia to be lower than in most of the other middle-income countries of the region. But demographic trends suggest that sustaining the convergence with average income in the 25 member states of the enlarged European Union will be challenging, requiring substantial improvements in both employment rates and labor productivity. 1.1 As Croatia prepares to join the European Union, the Government is working on (l) a joint inclusion memorandum8, and (2) a national regional development strategy outlining plans for reducing internal disparities, for which substantial structural funds will be available from the EU. This report offers data and analysis relevant for both these tasks, shedding light on the strategic priorities facing the government and some of the main trade-offs involved. Six background papers, issued as Volume 2 of the report, analyze selected issues related to these topics (Box 1-1). Box 1-1: Main Topics Covered in the Background Papers 1. Poverty Estimationjor Croatia: Methods and Measurement Issues: Discusses the methodological steps involved in estimating poverty rates for Croatia on the basis of the 2004 Household Budget Survey (HBS). 2. Poverty Comparisonsjor Croatia: 2002-2004: Presents poverty comparisons for Croatia over time, across regions, and across counties using data for 2002-04 HBS series. 3. A Poverty Profile jor Croatia: Develops a profile of the poor and the micro-determinants of poverty using data from the 2004 HBS. 4. Regional Development and Social Indicators in Croatia: Provides a comprehensive profile of social and economic characteristics of Croatia's regions at the NUTS III level (and also the five main analytic regions used in other background papers). The paper includes two appendices: (1) presents regional GDP by counties for 200 1-03, and (2) provides preliminary data on gross disposable income of the household sector in Croatia. 5. Regional Disparities in Labor Market Peiformance in Croatia: Reviews the labor market performance in Croatia in 2002-04 using data from the Labor Force Survey series. 6. Assessing the Flexibility oj the Croatian Labor Market: The paper review key labor market institutions in Croatia, and uses data from the Financial Agency to prepare a profile of labor market dynamics in the country in recent years. 8 As required by the EC, the joint inclusion memorandum outlines the principle challenges facing a prospective EU member state in terms of poverty and social exclusion, presents the major policy measures taken by the prospective member to start translating the EU's common objectives on poverty and social exclusion into national policies, and identifies the key policy issues for future monitoring and policy review. As a contribution to the improvement of data in Croatia, the report presents disaggregated statistics collated from a variety of sources. In particular, data from several rounds of the Household Budget Survey (HBS) and Labor Force Survey (LFS), including the latest, are pooled together to obtain new relatively precise estimates of key indicators of living standards at the regional and county 1evels.9 A comprehensive list of the indicators prepared, tables presenting these indicators at the county and regional level, and maps illustrating the spatial variation in these indicators are presented in the Appendix. 1.2 Using the latest available data we derive a poverty line for Croatia based on the cost of basic needs. We hope that both the poverty line and the methodology for deriving regional and county level estimates of key indicators of living standards will prove of continuing usefulness to the Croatian authorities. 1.1 RECENT EcmlOMIC DEVELOPMENTS AND CHALLENGES 1.3 The Croatian economy has performed quite well over the past decade (Figure 1-1), driven by generally strong domestic demand. Real GDP per capita increased by more than 40 percent between 1996 and 2005. Exports of goods and services recorded healthy growth over this period, and rose from 47 to 49 percent of GDP between 2000 and 2005 (Table 1-]). Figure 1·1: Good economic performance over the past decade Nominal GDP. in HRK million Nominal GDP. in USD million 250,000 40,000 35,000 200,000 30,000 150,000 25,000 20,000 100,000 IS,OOO 10,000 50,000 0 11III ~ .... ~ ~ ~ ~ ~ ~ '" ~ ~ ~ 5,000 0 1111111 ~ ~ 00 ~ ~ ~ ~ ~ ~ ~ ~ '" Nominal GDP growth, % Real GDP growth, % 15 7 \3 11 5 9 11.111111 7 3 I , I 1IIII 5 3 -I ..... oo a- .,., .,., 'l:> S N !"'\ 8 8 -1 'l:> S 00 8 S 0 § 8 .... 8 '" '" ::; '" ::; '" ::; '" ::; N 8 .... 8 .... 8 .... .... .... '" ::; '" '" ::; - a- .... 0 .... .... .... .... Source: Central Bureau of Statistics, Croatia. 9 For a more detailed description and rationale of the pooling procedure, see Background Paper No.5. 2 1.4 Croatia's economic recovery since the 1999 recession has been accompanied by a high external account deficit. While exports of services (in particular tourism and transport - see Box 1-2) recorded strong growth, this was not enough to outweigh the rise in merchandise imports that was spurred by high public investments. The high external account deficit in tum has pushed up the external debt. Despite significant inflows of foreign direct investment, Croatia's external debt, presented in Euro terms, rose from 60.6 percent of GDP in 2000 to 82.5 percent in 2005 (or in USD terms, rose from 61.2 percent to 78.5 percent of GDP during the same period). Box 1-2: Tourism in Croatia Given Croatia's favorable natural endowment, of an extensive coastline with warm Mediterranean climate and numerous scenic islands, tourism is one of the most important sectors of the economy, providing an estimated 317,000 jobs in 2004, or around 14 percent of total employment, and contributing indirectly to about one-quarter of GDP. In 2003, four-fifths of Croatia's deficit in goods exchange in the current account was covered by tourism receipts. Investments in this sector have intensified over the past three years, primarily towards the hotel industry (e.g. refurbishing and modernizing existing hotels and tourist settlements, construction of new small 'family' hotels). Over the next ten years, tourism is expected to grow at about 6.6 percent annually in real terms. Key challenges in this regard will be to continue to increase the total number of arrivals while attracting a more up-market clientele. The Central Bureau of Statistics estimates that the number of foreign overnight stays in Croatia increased from around 16.5 in 1996 to 46 million in 2005; howeverthis is still well short of the 1989 level of 54.5 million. Sources: Government Pre-Accession Economic Program 2005-2007, IMF Selected Issues Table 1-1: Croatia: Key economic indicators r- 2000 2001 2002 2003 2004 2005 Real Sector percentage chan !e RealGDP 2.9 4.4 5.6 5.3 3.8 4.3 Exports of goods and services 12.0 8.1 1.2 11.4 5.4 4.6 Imports of goods and services 3.7 9.8 13.4 12.1 3.5 3.5 , Fixed investment -3.8 7.1 13.9 24.7 4.4 4.8 I Private consumption 4.2 4.5 7.7 4.6 3.9 3.4 I Average CPI inflation 4.6 ' 3.8 1.7 1.8 2.1 3.3 Gross waR,es 7.0 3.9 6.0 4.8 6.4 4.4 Unemployment rate OLO) 16.1 15.8 14.8 14.3 13.8 12.7 - as % ofGDP ~e of goods and services Exports of goods and services (in US$) 46.9 48.5 45.9 50.2 49.9 49.0 Imports of goods and services (in US$) 52.1 54.4 56.7 58.1 57.1 56.4 General Government Fiscal deficit* -6.5 -6.8 -4.9 -6.2 -4.9 ! -4.1 Ouasi fiscal deficit (HBOR) ... ... ... -0.6 -0.4 -0.1 Overall deficit -6.5 -6.8 -4.9 -6.8 -5.3 -4.2 Overall public sector debt, including 49.5 51.1 50.7 51.1 52.1 52.8 _ contingent liabilities ...E&ernal accounts (in US$) Current account balance -2.5 -3.7 -8.3 -7.2 -5.2 -6.7 FDI net 5.9 6.0 2.5 6.5 2.5 4.0 Reserves (months of imports) 4.4 5.2 5.4 5.7 5.2 4.9 *General Government deficit includes only 53 local government units. Sources: CBS, Croatian National Bank and Ministry of Finance, WB staff calculations. 3 1.5 The Central Bank has pursued a relatively tight monetary policy to partly compensate for the government's loose fiscal stance, and inflation has therefore remained generally low. In addition, the government has responded to the rising external debt and vulnerability by reducing the deficit from 6.2 percent in 2003 to 4.1 percent in 2005 through slower wage bill growth, lower transfers and public investment, and structural reforms (Table I-I ).10 1.6 One of Croatia's biggest challenges is to boost employment. Croatia has one of the lowest employment rates in Europe, at only 54.7 percent, well below the so-called Lisbon employment rate target of 70 percent set by the European Commission (EC) for the year 2010 (Figure 1-2).11 Croatia's unemployment rate is higher than the EU25 average (around 14 percent versus 9 percent respectively in 2004) and its labor force participation rate is lower. Figure 1-2: Total employment rate in Croatia falls well short of the Lisbon target Iceland Denmark ~den United Austria Slovenia Germany Czech France Estonia Latvia Lithuani Spain Lisbon Target Romania Italy Slovakia HWlgary Croatia Poland o 10 20 30 40 50 60 70 80 90 Source: Eurostat. Data for all countries are for 2004. and rates shown are for ages 15-64 yrs. 1.7 The proportion of long-term unem Figure 1-3: Higb Long-term Unemployment ployed-workers without jobs for 12 14 months or longer-is higher than in all EU 12 countries except Poland and Slovakia (Figure 1-3). Long-term unemployment is generally 10 much more difficult to address than frictional 8 unemployment. It is often concentrated in I 6 specific groups such as those with fewer - 4 skills and little education, and among young people, and often can lead to profound economic and social problems. 2 0 United Kingdom I Slovenia I EU25 Average Croatia Poland Slovakia Source: Eurostat. Data are for 2004. to IMF Country Report No. 06/128, Washington DC. March, 2006. II The employment rate in this instance is defined as the share of the country's total population aged 15 - 64 that was working when the survey was fielded. 4 1.8 As in other countries in Central and Eastern Europe, employment in Croatia fell quite rapidly during the early years of transition in the early 1990s, when the total output of the economy fell substantially in real terms (Figure 1-4). The decline in employment slowed down somewhat after an economic upturn in 1994, but job creation in the expanding economy has generally remained slow; CBS data show an expansion only from 1.35 million in 1999 to 1.42 million in 2005. Figure 1-4: Growth in total employment in Croatia has lagged behind GDP growth and wages 120 110 ---GOP · Real net wages .. .. ---Civilian employment 100 90 , 80 70 60 ~ I 50 · 40 1990 1992 1994 1996 1998 2000 2002 2004 Source: Central Bureau of Statistics. Index, 1990=100. Data from various labor force surveys show virtually no change in the participation rate in recent years (Table 1-2). Table 1-2: Key labor force indicators for Croatia: 2000 - 04 Percent I 2000 2001 2002 I 2003 2004 Participation Rate 50.8 49.7 50.9 I 50.3 I 50.5 Employment Rate 42.6 41.8 43.3 I 43.1 I 43.5 Unemolovment Rate 16.1 15.8 14.8 14.3 L 13.8 Source: Published LFS data; data are for population aged 15 yrs and older. 1.9 Higher economic output in recent years has benefited workers less through job growth than through higher wages. Labor productivity per person employed in Croatia has risen considerably in the past decade relative to the EU average and has been reflected in rising wages over the period 2001-03. 5 1.2 SECTORAL AND REGIONAL TRENDS IN OUTPUT, INCOME, AND EMPLOYMENT Overall, the services sector is the biggest and most dynamic sector ofCroatia's economy. Regional inequality has risen somewhat in recent years, reflecting differences in the growth of output and, possibly in rates afjob creation and destruction. 1.10 Output, income, and employment patterns differ significantly across the regions of Croatia. The differences reflect diverse factors such as the impact of the recent war, varied geographic terrain, and special problems faced by those regions with a higher concentration of traditional industries currently facing problems. Following the same analytical regional classification used in the World Bank's earlier poverty assessment to maintain comparability, this report uses a breakdown of five main geographic regions (Table 1-3) to examine the variations. 12 Table 1-3: Regional classification of counties in Croatia Analytical Reldon Counties included in the Reldon Zqgreb Zagreb County and Zagreb City. Central Krapina-Zagorje, Sisak-Moslavina, Karlovac, Varazdin, Koprivnica- Krizevci, Bielovar-Bilogora, and Medimurie. Eastern Virovitica-Podravina, Pozega-Slavonia, Slav. Brod-Posavina, Osijek- Baranja, and Vukovar-Sirmium. Adriatic North Primorie-Gorski Kotar Lika-Seni. and Istria. Adriatic South Zadar Sibenik-Knin. Split-Dalmatia and Dubrovnik-Neretva. 1.11 As in other developed economies, the services sector makes the largest contribution to output and employment in Croatia, though there are important differences across regions. The smallest share of employment is in agriculture (Figure 1-5). 1.12 Agriculture: The agriculture sector Figure 1-5: Total employment by sector, 2003 in Croatia is small compared to that in other countries with similar income levels in the region, and three-quarters of the workers it employs are self-employed farmers. The sector is quite varied, reflecting the country's geographic diversity, with cultivated lowland regions, mountainous regions where pastures predominate, and coastal areas where viticulture, fruit, and olive production are Services (Private) 46% prominent. Agriculture provides the largest share of total employment in the Central and Eastern regions (19 - 20 percent), while in other regions it constitutes only 4 - 5 Source: Data from CBS. percent. 12 While representing only one of several possible regional classification schemes, this grouping provides a useful summary of regional variation in Croatia, but at the same time also ensuring that the survey sample size in each group is large enough to permit reasonable statistical accuracy. This grouping also follows logical geographical, administrative. social and economical convergence characteristics. 6 1.13 Industry: Industry supplies a relatively large share of jobs in Central Croatia (30 percent), particularly in the counties of Varazdin and Sisak-Moslavina, while in other regions it contributes only about 19 23 percent. Croatia's main industries are foods and beverages. petroleum, chemicals, electrical manufacturing, paper, printing and publishing, and shipbuilding. The sector is diverse, including rapidly growing industries such as publishing and some capital goods manufacturing, alongside activities such as textiles or metal industries that face increased international competition. A sizeable share of output is still produced in state-owned companies that have not yet been fully restructured. 1.14 Services jobs in the private sector predominate in the Zagreb and Adriatic regions (particularly in the counties of Zagreb and Dubrovnik) and provide more than half of the total employment in these regions, compared to considerably smaller shares in the Central and Eastern regions (31 percent and 37 percent respectively). Public sector jobs, by contrast, are fairly evenly distributed across all regions, comprising about 18 - 23 percent of total employment, depending on the region. 1.15 The tertiary sector has achieved the fastest growth in output, while the output offarming and fisheries has declined. 13 Thus, in Zagreb as well as the Adriatic regions, where the tertiary sub-sector accounts for a larger share of the overall economy, per capita GDP grew faster than in the Central and Eastern regions. 1.16 Trends in per capita GDP show Figure 1-6: Rising regional inequality in GDP per capita a small rise in inequality at the regional 170 .2001.2003 level in Croatia between 2001 and 2003 150 (Figure 1-6). The ratio between GDP per capita in the highest and lowest ranked region rose from 2.1 to 2.2 130 f during these two years. This is a short period over which to measure a trend, 110 '5 90 I ~ · but such a rise would be consistent with 70 what has taken place in many of the EU new member states, as discussed 50 Zagreb Central Adriatic Adriatic Eastern further in Chapter 4. North South 1.17 It is possible that rising inter- Source: World Bank estimates. regional inequality may be linked to differential patterns of job creation and destruction at the regional level. Those of Croatia's regions that have achieved the best economic performance have a dynamic small and medium enterprise sector and a relatively large share of newly created private firms. Since the mid-l 990s these firms have had considerably higher rates ofjob creation and destruction than other types of firms (Figure 1-7).14 D This pattern of differential growth across sectors is in line with the experience of new member states (NMS) just before their EU accession; regions with a higher share of the tertiary sector (excluding the public sector) grew faster than others. In the case of the industrial sector, internal structure (e.g. export orientation, technology transfer, etc.) seemed to matter more than total share in the economy. By contrast, regions with a large share of agriculture and public sector activity tended to experience slower than average growth. For more details, see Background Paper No.4. 14 The firm-level data on job flows are from FINA (Croatia's financial agency). For more details of this analysis, see Background Paper No.6. 7 Figure 1-7: Small firms in Croatia have been quite dynamic in job creation 40 Job Creation (Percent) 40 Job Destruction (Percent) 35 35 30 30 25 25 20 20 15 ]0 5 o 1994 1995 1996 1997 1998 1999 2000 2001 1994 1995 1996 1997 1998 1999 2000 2001 .1-10.11-20 [] 21-50 [] 51-200 .201-500.501 81-10 8 11-20 [] 21-50 [] 51-200 8201-500.501 ··· and over time have increased their share in total employment 1-10 wOll 2 poverty line) 1682511 39.8 100.0 Overall 4.227000 100.0 100.0 Source: World Bank estimates based on 2004 Household Budget Survey. 1.26 Trends in poverty over time were derived by first converting the 2004 poverty line (i.e. 22,145 kunas) into 2002 prices using the consumer price index to account for changes in the price level between 2002 and 2004. This yielded a poverty line of 21 ,390 kunas per adult-equivalent per year in 2002 prices. Data from the 2002 Household Budget Survey were then used to derive various poverty estimates based on the distribution of consumption and incomes per-adult equivalent. 1.27 Recent trends in poverty are ambiguous. Depending on the welfare measure used, the poverty headcount rate in Croatia appears to have either stagnated (based on consumption per adult-equivalent), or else fallen at about 1 percent a year (based on income per adult-equivalent) between 2002 and 2004 (Table 1-6). Other distribution-sensitive measures of poverty (P 1, P2) also point to a stagnation or decline in poverty, depending on the welfare measure used. 10 Table 1-6: Poverty incidence in Croatia Poverty ConsumDtion Incomes Measure 2002 2004 2002 2004 Pove~ line used 22,145 21,390 145 (per a ult-equivalent per year) 21,390 Poverty headcount rate 11.2% 11.1 % 13.0% 10.4% Povertv gap - PI 2.6% 2.6% 3.2% 2.4% ertv severitv - P2 1.0% 1.0% 1.2% 0.9% Source: World Bank estimates based on 2002 and 2004 HBS. 1.28 Income grew much faster than consumption in 2002-04, as can be seen by examining the growth incidence curves-that is, the variation in growth rates over the distribution-for changes in consumption and incomes per adult -equivalent (Figure 1-9). The distribution of consumption per adult -equivalent was stagnant on average, as well as across most of the distribution except the left and right tails, which showed positive and negative growth respectively over this period. As a result, it is not surprising to see virtually no change in poverty rates over this period based on this welfare measure. By contrast, incomes per adult -equivalent grew by about 6 percent over this period, on average. While all income groups recorded positive real income growth, both the poor and the rich did quite well over this period relative to the middle class. Figure 1·9: Survey-based estimates show that income grew much faster than consumption Per-Equivalent Adutt Consumption Per-Equivalent Adult Income I I j I I I 20 40 60 60 100 a 20 40 60 80 100 Percentiles Percentiles Source: World Bank estimates based on 2002 and 2004 HBS. Straight line represents mean of growth ["dtes. 1.29 Subjective measures of well being suggest some improvements have taken place. The Household Budget Survey asked sample households their opinions regarding their own living standards. While measures based on these responses are not necessarily comparable with the absolute poverty measures presented above, the responses provide some useful indications of perceptions of changes in living conditions. The proportion of the population reporting itself as living "with great difficulty" declined from 13 percent in 2002 to 10 percent two years later, while the proportion reporting itself as living "well" or "very well" rose from around 10 percent to more than 20 percent over the same period (Table 1-7). 11 Table 1-7: Subjective welfare measures indicate improved living conditions With its dis~osable monthly Percent of Respondents income, the ousehold lives: I 2002 2004 With great difficulty 13.4 10.0 With difficulty 25.6 22.7 With some difficulty 29.5 28.6 Fairly well 21.8 17.5 Well 8.0 19.1 Very well 1.7 2.1 Total 100.0 100.0 Source: World Bank estimates based on 2002 and 2004 HBS. l.30 Inequality in both consumption and income has been quite stable in recent years. While income per adult-equivalent in both years was more unequally distributed than consumption, comparisons over time of both distributions show that inequality, as measured by the Gini coefficient and inter-decile ratios, remained fairly stable during 2002-04 (Table 1-8). Table 1-8: Income inequality remained fairly stable between 2002 and 2004 Consumption Income Inequality Measure I- 2002 2004 2002 H004 Cons. share of the bottom decile 0.038 0.038 0.034 ..036 Consumption share of the tog decile 0.210 0.204 0.211 ... 0.218 Decile ratio: (decile 9/decile 1) 3.181 3.182 3.592 3.553 Gini coefficient 0.258 0.253 0.270 0.275 Theil entropy measure 0.123 0.107 0.122 0.127 Mean log deviation 0.116 0.109 0.126 0.128 Source: World Bank estimates based on 2002 and 2004 HBS. 1.4 POVERTY IN CROATIA: AN INTERNATIONAL PERSPECTIVE Croatia's poverty rates are the lowest among the World Bank's client countries in the Europe and Central Asia region. But its social indicators are broadly in line with those in the upper middle income countries of the region. 1.31 This section compares poverty levels in Croatia with those in other middle-income countries in Europe and Central Asia, using the same PPP$4.30 per capita per day international poverty lines used in a recent World Bank study on living conditions in the region. 16 We derive a welfare measure and poverty line in local currency for Croatia using as similar a method as possible to that used for the other countries in the regional study, so as to maximize the cross country comparability of the estimates derived. 1.32 Croatia has the highest average per capita consumption among the countries of the region: some 20 percent higher than Latvia, and some 30 and 40 percent higher than Macedonia and Hungary, respectively (Table 1-9). The share of food in the consumption basket (around 40 percent) is somewhat large for a country at Croatia's income level. This is probably because food Murthi, YemtsoY, et ai., 2005: Growth, Poverty, and Inequality: Eastern Europe and the Former Soviet Union. [6 Alam, The World Bank, Washington DC. 12 prices in Croatia are relatively high compared to those of non-food items and also compared to food prices in other transition countries. Table 1-9: International comparisons of poverty and inequality I Year of Consumption PerCa~ita Food . Povert;: Rate I (~PPP .30/day I Gini Coefficient Survey (PPP ) Sbare (%) overty LineJ (Per Capita) . Croatia 2004 4156 41.6 4 0.264 Hun~ary 2002 2890 38.7 12 0.250 Latvia 2003 3401 41.0 17 0.350 . Belarus 2002 :2 704 68.1 21 0.292 Ukraine 2003 2496 72.2 22 0.268 Macedonia 2003 3 171 54.2 24 0.373 Lithuania 2003 2762 44.5 24 0.325 Estonia 2003 2753 42.2 26 0.330 Poland 2002 2,611 39.8 27 0.320 Bulaaria 2003 2248 58.7 33 0.277 Russia 2002 2179 55.8 41 0.338 Serbia 2002 1993 60.8 42 0.292 Turkey 2002 1.816 38.8 58 0.393 Romania 2003 1624 57.8 58 0.288 Albania i 2002 1388 61.7 71 0.319 Moldova 2003 1046 66.4 85 0.328 Source: World Bank (2005): World Bank estimates based on 2004 HBS for Croatia. 1.33 Croatia's poverty rates are the lowest in the region. If the international poverty line of$PPP 4.30 per day per person is applied, the poverty rate for Croatia is around 4 percent, much lower than in the other countries in Figure 1-10: Cross-country poverty comparisons the region for which data are MOLDOVA available. Similarly, the Gini 80 coefficient for Croatia is low ALBANIA compared to those for other countries in the region. ROMANlA .TURKEY 1.34 The low rate of po SERBIA· verty in Croatia compared to other countries in the region is in line with what one might expect from Croatia's income level (Figure 1-10). The · o ~-----------r----------~----------~--- dotted line in the figure shows 1000 2000 3000 Consumption per capfla 4000 (PPP USD) the estimated relationship between headcount poverty Source: World Bank (2005) and HBS 2004. rate and income levels from a cross-country regression using a quadratic fit. 1.35 While poverty rates in Croatia are lower than elsewhere, other social indicators are more or less in line with those in other countries in the region. For instance, infant and child mortality rates in Croatia are lower than in Latvia or Hungary, but higher than in the Czech Republic or Slovenia (Table 1-10). 13 Table 1-10: Key social indicators: cross-country comparison ! Secondary I Mortality Mortality Life Adult school rate, infant rate, under-5 eXl?ectancy Illiteracy (%) enrollment (per 1,000) (per 1,000) at buth, total (Net, %) (years) 1990 2004 1990 2003 1990 2004 1990 2004 1990 2004 Croatia 3 1.9 63 85.0 12 6.3 13 6.3 72.2 75.4 Bulgaria 2.8 1.4 63 88.3 14.8 12.3 16 12.3 71.3 72.4 Czech Rep. .. .. 86 90.4 10 3.9 11 3.9 71.7 75.7 Estonia 0.2 0.2 82 87.9 15 5.7 17 5.7 69.5 71.6 Hunl!arv 0.9 0.7 75 91.6 15 7.2 16 7.2 69.3 72.6 Latvia 0.2 0.3 77 87.4 16 9.8 20 9.8 69.3 71.5 Lithuania 0.7 0.4 81 94.1 17 7.5 13 7.5 71.3 71.9 Poland .. 0.3 76 91.5 16 7.1 19 7.1 70.9 74.5 Romania 2.9 2.7 73 81.1 ! 27 17.3 32 17.3 69.7 71.0 Slovak Reo. .. 0.3 .. 88.0 ! 14 6 15 6 70.9 74.0 Slovenia 0.4 0.3 89 95.3 8 4 9 4 73.3 76.6 Sources: UNESCO, World Bank database (DDP). 1.5 THE CONVERGENCE CHALLENGE: REDUCING THE INCOME GAP WITH EUROPE Croatia has been narrowing the income gap with the European Union, but sustaining this trend will be challenging. Both the employment rate and labor productivity will need to rise substantially. 1.36 Croatia's economic growth rate has exceeded the European Union average in eight ofthe past ten years (Figure 1-11). Between 1996 and 2005, Croatia's per capita Figure 1-11: Growth in per capita GDP GDP rose by 55 percent, or about 2.5 times faster than that of the European Union overall. In relation to the EU25 8 average, Croatia's per capita GDP rose 6 from 37 percent in 1995 to 46 percent in 4 2004 (Figure 1-12). 2 1.37 The revised GDP series recently released by CBS shows that per capita -2 1997 1998 1999 2000 2001 2002 2003 2004 2005 GDP in Croatia in purchasing power -4 c EU-25 · Croatia parity terms is higher than the above numbers suggest. Methodological corrections taking into account possible under-estimation of GDP indicate that the gap between per capita GDP in Croatia and other EU countries may be 16 percent less than that indicated by these statistics. 17 17 Source: Ministry of Finance correspondence. 14 Figure 1-12: GDP per capita: Croatia and selected EU countries Slovenia Czech Republc !-lmgary Slovakia EStonia Lithuania Fbland Ooatia Latvia Romania Bulgaria Turkey 0 90 Source: Eurostat. GOP per capita in Purchasing Power Standards; EU25=lOO. 1.38 Taking this cue, let us assume that per capita GDP in Croatia is now as high as 55 percent of the EU25 level. Does the past trend rate of convergence, of roughly one percentage point a year, imply that Croatia is well on track to attaining an average living standard of 75 percent of the EU25 average living standards by 2025? A likely decline in population and labor force suggests that closing the remaining income gap with the EU25 will be quite challenging. Demographic projections show that Croatia's popUlation will start to fall by 2006 and that the size of the economically active age group will start to shrink by 2010. Figure Figure 1-13: Convergence and demographics 1-13 shows how GDP per 11,800 capita in constant PPP US dollars of 2000 will evolve 11,600 if it is driven solely by the 11,400 decline in the labor force that will result from these 11.200 demographic trends. 11.000 10,800 10,600 10,400 . 2005 2010 2015 2020 2025 GDP per capita (PPP 2000 prices) trend due to smaller labor force Source: World Bank estimates based on popUlation projections. 15 1.39 Far from continuing to rise, as it has during the past decade, GDP per capita will start to decline after 2010. As noted, these projections assume that GDP per capita growth is driven only by changes in labor force population dynamics. In what follows, we relax this assumption and extend the analysis in two respects. 1.40 We first analyze how GDP per capita would change if the employment rate were to rise to 70 percent by 2025 18 , rather than staying at its current level of around 55 percent of the working age population (Figure 1-14, left panel). Such an increase in employment, due to increased labor force participation, would be large enough to compensate for the adverse impact of the decline in the working-age population. But by itself, the higher employment rate would not raise Croatia's per capita GDP to three-quarters of the average EU25 per capita GDP by 2025 (Figure 1-14, right panel, scenario A). To meet this income target, not only would employment need to rise to 70 percent, but output per worker would also need to rise-by about 0.67 percent per year (Figure 1-14, right panel, scenario B). Figure 1·14: Convergence, labor market participation, and productivity 75 . 18,000 16,000 70 ' 65 14.000~ = 12,000. / 60 10,000 A. Employment rate rises 8,000 to Lisbon target a-'-"-''--_ _- t__6:=:..1~.~4_-+_---,-47-,::,.,-,,8_-; large shares of workers with average 1---""==~c:.==-'='=->.:.::..L_ _-+_-=13::.:.=O_-+_......:c..16=.:..:4_--I 3 514 earnings clustered around 1,800 L...!;;.==~==~=='--_...l.----"~'-!----I._--==':'';;:''''---J 2978 _ 2,000 kunas per month, and in Source: World Bank estimates based on 2002-04 LFS. general earn considerably less per month than workers living in other regions (Table 2-3 and Figure 2-11). Table 2-3: Selected labor market indicators, by region Analytical Active Participation Employment Unemployment Monthly rate (%) rate (%) rate (%) earning Region popUlation (%) (kunas) Central 63.9 72.3 60.8 11.5 2.806 Eastern 63.0 61.1 47.9 19.9 2.826 Zagreb 66.1 63.6 55.9 11.8 3.735 Adriatic North 64.4 66.1 58.5 9.9 3.498 Adriatic South 62.3 62.2 48.9 20.2 3.524 Croatia Overall - 64.0 65.2 54.5 14.5 3,276 Source: World Bank estimates based on pooled 2002-04 LFS. 26 Figure 2-11: Distribution of monthly earnings, by region o 2000 4000 6000 8000 10000 Monthlye.nlng - - Central Region - - - - - Eastern Region ·....·..·.·....·. Zagreb Region Adriatic North Region - - - Adriatic South Region Source: World Bank estimates based on pooled 2002-04 LFS. 2.26 To ascertain the underlying causes of regional differences in living conditions requires a better understanding of why average earnings vary so significantly across workers and regions. Results from a regression model based on data from the labor force surveys show that both the likelihood of getting employed and a worker's average earnings increase with age (which in tum probably acts as a proxy for the worker's level of experience), though at a steadily declining rate until this effect peaks at around 40 years of age (the impact of age on earnings continues to rise for some more years).19 Other things equal, women are less likely to find employment than are men in all regions of the country, and those women who do manage to find a job earn less on average than their male counterparts. Finally, our analysis suggests that married individuals and heads of households in Croatia are more likely than others to search for and obtain employment. 2.27 In Croatia, as in many other countries, labor markets offer significant returns to education. The multivariate analysis confirms the pattern noted above, that individuals with better educational attainment are much more likely to be employed and to have significantly higher earnings than those with little or no education (Table 2-4). 2·4: Employment, unemployment, and monthly earnings, by educational attainment Education Level Employment Unemployment Monthly Earning Rate (%) Rate (%) (kunas) No school or uncom leted basic education 30.8 12.5 1 Basic education 37.0 16.1 2 61.6 16.0 3 31.7 17.5 73.5 8.7 82.8 7.5 5252 Source; World Bank estimates based on pooled 2002 04 LFS. 19 For more details on the regression results, see Background Paper No.5. Note that since not everyone in the labor force works for wages, the estimates control for selection bias in the wage regression. 27 2.28 Workers' earnings are strongly correlated (0.8) with levels of schooling. Given differences across regions in labor market structure and performance (for example, proportions working or unemployed, earnings levels), the analysis allowed for the possibility that the impact of education on employment and earnings varies across regions. In this regard, our findings suggest that the impact of education on an individual's likelihood of finding a job is highest in the Eastern region, followed by the Adriatic South. As noted above, both these regions have among the lowest employment and highest unemployment rates in Croatia (Table 2-3). It is not surprising that in regions with limitedjob availability, workers with more education are more likely to find jobs. But in Zagreb and the Central region, labor markets appear to offer relatively higher wage premiums to educated workers, ret1ecting higher employment and hence relatively more competitive demand for labor than in other regions. 2.29 Closer examination of workers' human capital endowments, proxied by their years of schooling or their highest educational attainment-reveals considerable variation in average attainment levels across regions. For example, workers in Zagreb have at least one year more of education, on average, than workers in the Eastern and Central regions (Figure 2-12). Similarly, about two-fifths of workers in the Eastern and Central region have either "no schooling", "uncompleted basic education", or "basic education", and only 5 percent of them have completed "university education or higher", compared to 12 percent of the workers in Zagreb. Figure 2-12: People in the eastern and central regions have lower education attainment Adriatic South Region Adriatic North Region Zagreb Region Eastern Region Central Region 9 9.5 10 10.5 11 11.5 12 Source: World Bank estimates based on pooled 2002-04 LFS. 2.30 When examining regional inequality, an important issue meriting attention is the extent to which variations in living conditions are due to differences in the characteristics of workers living in these regions vs. other regional characteristics. For instance, if most people in a given region have less education than those living elsewhere, this in tum might largely explain why the region is poorer than the rest of the country. To ascertain how important this consideration may be in Croatia, we used the Oaxaca-Blinder Decomposition (Box 2-2) to study the relative importance of regional and individual characteristics in helping to explain regional wage differentials. 20 20 For more details on the analysis, see Background Paper No.5. 28 2.31 We find that 28 - 35 percent of the overall average wage differential between the Zagreb and Adriatic regions on the one hand, compared to the Central region on the other hand (the Eastern region is very similar to the latter), can be attributed to individual characteristics of workers, such as higher educational attainment (Table 2-5). Table 2-5: Oaxaca-Blinder Decomposition: Effect of individual vs. other characteristics on earnings Mean Monthly Individual Other Analytical Region Earning (kunas) Characteristics Characteristics Central 2,806 ... ... Eastern 2,826 64% 36% Zagreb 3,735 35% 65% Adriatic North 3,498 28% 72% Adriatic South 3524 31% 69% Note: The table shows the relative importance of individual characteristics such as age, gender, or education vs. "other characteristics"-that is, all other factors influencing wages including (but not limited to) region-specific factors not controlled for in the analysis. All regional comparisons reported are in relation to the Central region. Source: World Bank estimates based on pooled 2002 -04 LFS data. 2.32 But 65 to 72 percent ofthe observed wage differentials across regions are explained only by "other regional factors." This is another way of saying we have no explanation for these differences across regions. Lastly, we examine the hypothesis that the sector of employment-public versus private firms-affects workers' earnings and hence the income differential across regions. Data from the 2002-04 rounds of the LFS show that the average monthly earnings of workers employed in "state firms, organizations. and institutions" are about 30 percent higher than those of workers employed elsewhere. Using a Oaxaca-Blinder decomposition to analyze these differences, we find that adjusting for differences in worker characteristics across the public and private sectors helps explain about 36-59 percent of the observed wage differential among regions. Since the share of employment in public sector firms tends to also be higher in the Zagreb and Adriatic regions, this may be partly responsible for the observed inequalities in earnings across regions. Box 2-2: Understanding Differences across Groups through the Oaxaca-Blinder Decomposition In seminal papers on regression-based decomposition into subgroups, Oaxaca (1973) and Blinder (1973) analyzed gender-based wage differentials. They estimated wages for men and women, decomposing the mean wage difference into a component ascribed to differences in the characteristics determining earnings and a component ascribed to differences in the returns to such characteristics. The first component is typically called the "explained" difference, because it arises from differences in human capital. The second component is called the "unexplained" difference. because it could arise from unequal treatment of the two groups in the market place or other factors, such as differences in the quality of human or physical capital, differences in the effort level, or other socio-cultural factors. 29 3. THE TwIN CHALLENGES OF EXTERNAL AND INTERNAL INCOME CONVERGENCE Regional differences in living conditions, suggested by differences in GDP per capita, income, consumption, poverty, earnings, and education levels, confirm that a regionally focused national development strategy makes sensefor Croatia despite the country's relatively small size. In planning such a strategy, the level of aggregation is an important choice. Counties are probably too small a unit for this purpose, partly because their relative rankings vary with the indicator of living standards that is chosen, and partly because the comparative advantage of regions often spans different counties. Looking ahead, the task offaster external income convergence with the EU will be challenging, and will require both faster job creation as well as flexibility in the allocation ofjobs and workers in the economy. These will also help with more rapid improvement in living conditions in lagging regions. To these ends, the chapter highlights four sets ofinter-related policy challenges and priorities: (1) sustaining high rates of growth to permit faster external income convergence, (2) building upon local comparative advantage in the regional development strategy to help reduce internal disparities, (3) promoting greater labor mobility, including measures aimed at building human capital to allow workers access to better opportunities, and (4) improving the adequacy and effectiveness of social safety nels within a responsible fiscal framework. 3.1 Lagging regions have received much attention and concern from public policymakers across the world, but especially in Europe, where substantial European Commission (EC) funds have been used to try to reduce regional inequalities. Lagging regions often have many similar features. For instance, they tend to be characterized by slow economic growth, high unemployment and poverty, low efficiency and productivity, little external trade, and generally low fiscal capacity for local redistribution. In addition, they often share several physical characteristics, including being landlocked, having outdated industries, and often lacking (or having exhausted) natural resources. But while policymakers agree on the need for public intervention to tackle regional inequalities in living conditions, much less agreement prevails on how best to do this, even within relatively similar countries. 3.2 For Croatia, regional development is an important element of the EU accession agenda, and extensive EC grants may eventually be available to help the country's lagging regions catch up, to support economic and social conversion in areas facing structural difficulties, and to modernize systems of training to promote employment. EC funds aim at promoting income convergence not just across regions within a country but also across countries within the European Union. The National Strategy for Regional Development, whose principal goal is to enable all parts of the country "to contribute to sustainable national development and competitiveness, and to reduce social and economic disparities across the country", envisages three main programs to achieve these objectives: (1) a county and wider region development program; (2) a disadvantaged areas development program; and (3) a cross-border and interregional cooperation program. 3.3 A key requirement for better social and economic development planning and monitoring will be to develop a statistical database and a system for gathering the necessary reliable statistics. While the Central Bureau of Statistics is setting in place a framework for collecting more disaggregated statistics (for example at the NUTS II and III levels)21, reliable information is not currently collected on socioeconomic characteristics at either the county or the regional level. Estimating 30 poverty incidence at the regional level in Croatia is challenging, because the total sample size of the Household Budget Survey within each county is too small to yield estimates with the desired level of precision. To overcome the problem of small sample size, data from the 2002, 2003, and 2004 HBS rounds was pooled to derive poverty estimates at the county level. The resulting tables and maps provide useful insights into the variations in development indicators across different parts of the country. They show, for example, that while the Eastern and Central regions are both poor in monetary terms and human capital endowments, unemployment rates are quite high in the former but much lower in the latter. Similarly, while the two Adriatic analytic regions have similar levels of earnings per worker and human capital, unemployment rates are considerably higher in the south than the north. 3.4 The map of poverty incidence (Figure 3-l) clearly illustrates the relatively high concentration of the poor in the Central and Eastern parts of Croatia. Maps based on county-level estimates of the poverty gap and squared poverty gap also show a similar geographic distribution, and illustrate clearly that poverty in Zagreb as well as in the coastal counties is generally quite limited and shallow (Appendix). One of the benefits of the county-level data is that they can be aggregated up into whichever larger groupings are of interest to policymakers. Figure 3-1: Poverty incidence at the county level I Poye~~ :Cldence 6.1 ·13 13.1 ·20 20.1 - 27 27.1·33.8 Source: World Bank estimates based on pooled 2002-04 HBS. 21 Regional statistics in Europe follow the Nomenclature des Unites Territoriales Statistiques (NUTS) classification which is based mainly on the total popUlation of geographic units. Croatia corresponds to the NUTS I level, counties to the NUTS III level, and municipalities and cities fulfill the criteria for the establishment of NUTS IV regions. The issue of definition of NUTS II regions has not been resolved as yet. 31 3.1 POOR PEOI'LE OR POOR REGIONS? Key correlates ofpoverty-employment status, age of household head, and household size-help explain some but not all of the observed differences in poverty levels across regions. Given the relatively large unexplained regional d(lJerences in living standards, a regionally differentiated strategy probably makes good sense for Croatia. 3.5 Chapter 2 above emphasized that not only the region where a person lives, but also individual level attributes-notably age, education, and employment status-are important predictors of whether a particular person is poor. In that chapter we used multivariate analysis and data on earnings from pooled Labor Force Surveys to begin to explore the question: are the poor people who live in a particular region poor because they have unfavorable endowments of education and other such individual-level attributes linked to higher incidence of poverty, or is their poverty status primarily due to region-specific attributes of their place of residence-for example, inadequate infrastructure, poor access to basic services, and too few jobs? 3.6 The county-level estimates of poverty incidence and its correlates allow us to take this exploration further. On the face of it, the data at the county level seem to support both individual and region-specific causes for poverty. For instance, counties whose residents have a high level of educational attainment also tend to be the counties with high average incomes. This is illustrated in Figure 3-2 (bottom panel), which shows the variations in educational indicators and average incomes at the county leveL The correlation between GDP per capita and proportion of population with primary education is found to be -0.59 22 , and the correlation between GDP per capita and proportion of working population is 0.49, which would suggest that the variation in living conditions across regions is due more to individual than to regional characteristics. At the same time, however, county-level infrastructure variables also appear to be strongly correlated with county-level estimates of per capita GDP-for example, the correlation between road density and GDP per capita is 0.63. Figure 3·2: Average per capita incomes at the county level generaUy seem to be correlated with the educational attainment of the population Average per-caplta Incoma Share 01 0°-89 >16 p:~~auon 1 81.84 1 95 ·107 108 -120 .,21-133 6.1 ·10 10.1 -14 14.1-18 18.1·22 Average per·capita income (Croatia=l00, 2002·04 HBS) Post-Secondary Completion Rate (percent, 2002·04 LFS) Source: World Bank estimates based on pooled 2002-04 HBS and LFS. 22 See Background Paper No.4. 32 3.7 Using RBS data for 2004, we Figure 3-3: Simple vs. partial correlations explore further the relative impact of individual- and region-specific Simple correlation: factors on the variation in living conditions across regions. Figure 3-3 - .. '--_R_u_ra_I_.....I~------------~ Poverty Simple correlation . .1 I outlines the analytical framework Partial correlations: used. For example, suppose we wish to study whether people in rural areas Rural r- Partial correlation are poorer because of the rural area, '---Ta--..... or because people with less education Simple correlation -----====' (controlUng for education) Poverty I (which is associated with poverty) ;""';--::..l"'------'-, Partial correlation live predominantly in rural areas. The I Education j...-- (controlling for rural) simple correlation between living in rural areas and poverty can be decomposed into two effects: (1) a direct effect, as shown by the partial correlation between rural residence and poverty; and (2) an indirect effect, which runs from rural residence via education to poverty. If the partial correlation between rural residence and poverty is much smaller than the corresponding simple correlation, this would imply that much of the relationship between living in rural areas and poverty is explained by the lower educational attainment of rural residents. At the limit, if the partial correlation between region and poverty is zero, this would imply that the region of residence has no direct impact on poverty, and that the entire observed correlation between the two is due to the indirect impact of education and its correlation with region of residence. 3.8 Using a three-step process, the simple concept illustrated above can be extended to a more complex multivariate framework to investigate the interrelationships between different sets of variables.23 In the first step, we use regression analysis to estimate the relationship between consumption per adult-equivalent and five key correlates of poverty: (1) education; (2) employment status; (3) region of residence; (4) age of household head; and (5) household size. In the second step, this estimated regression model is then used to predict consumption, holding constant across the population, in tum, each of these five correlates of poverty. In the third and final step. these predicted consumption estimates are used to estimate the relative poverty risk under several different scenarios. 3.9 In the discussion below we take the relationship between the education of the household head and consumption as an example to illustrate the approach used. The main findings are shown in Table 3-1. Column 0 presents the actual relative poverty risk for individuals grouped according to the level of education of their household heads. It shows. for example, that individuals in households whose head has completed only primary education face an 87 percent higher risk of poverty than an average Croatian. To examine the pathways by which educational attainment is associated with poverty, we calculate the relative poverty risks by level of education, after controlling for the partial effects of each of the five key correlates of poverty noted above. The resulting simulated poverty risks are presented in columns I - 5. As shown in column I, the differences in relative poverty risk by education become substantially smaller after taking into account the partial effects 23 See Background Paper No.3 for more details on the methodology followed. 33 of education-thus, for example, the relative poverty risk for the "Unfinished Primary" category drops from 2.85 to \.80, while that for "General Secondary" rises from 0.24 to 0.49. Clearly, the educational level of the household head is a very important correlate of poverty. Table 3·1: Relative poverty risk by educational attainment of the household head I Educational I ACTUAL . Relative SIMULATED Relative Poverty Risk after controlliDl!: for the partial effect of: I Attainment of Poverty (J) (2) (3) (4) I (5) Household Head Risk Employment Household Education Region Age (0) status size Unfinished Primary 2.85 1.80 2.83 2.77 2.53 2.98 (0,24) (0.25) (0.23) (0,26) (0.24) (0.25) Primary 1.87 1.57 I 1.96 1.76 1.87 1.90 (0,20) (0.21) (0.19) (0,20) (0./9) (0.20) Vocational secondar~ 0.62 0.90 0.56 0.69 0.73 0.56 (0.10) (0.13) (0.09) (0,10) (0.11 ) (0.09) General Secondary 0.24 0.49 0.23 0.27 0.30 0.22 (0.07) (0. II) (0.06) (0,08) (0.08) (0,07) Post SecondillY 0.02 0.45 0.06 0.06 0.02 0.02 (0.01) (0, II) (0.04) (0.04) (0,01 ) (0,01 ) Overall Poverty I ll.l 9.1 11.6 10.0 11.0 10.8 R:.tl' (%) J (0,86) (0.8) (O.S) (0.80) i (0.9) (0.83 ) Source: Background Paper No.3. Standard errors are in parentheses. 3.10 The main reason why the poverty risk differs across the various educational groups shown in Table 3-1 is because part of the association between education and poverty takes effect via other variables such as employment status, region, age, and household size. For example, since poorer regions tend, on average, to have residents with relatively little education, the simulated poverty risks taking into account the correlation between region of residence and education (column 3) are lower than those simulated on the basis of the simple correlations alone (column 0). Similarly, the younger population in Croatia has, on average, higher educational attainment than the elderly, so taking this correlation into account (column 4) narrows the differences in poverty risks across educational groups. 3.11 Similar analysis was carried out to help isolate the direct impacts of the other key correlates of poverty-employment status, region of residence, age of household head, and household size on the risk of poverty. The findings confirm that people in households with elderly (aged 65 years and older), unemployed, or inactive heads have a significantly higher risk of poverty than the overall population. 24 The risk of poverty is also strongly affected by a person's region of residence. As shown in Table 3-2, while the regional variations in simulated poverty risks partly depend on the education, employment status, age, and household composition of the regional populations, these correlates of poverty explain only part of the observed regional variation in poverty. The differences in relative poverty risk attributable to region-specific factors (column 3) are considerably greater than those attributable to the correlation between region of residence and other variables considered in our analysis. 24 See Background Papcr No.3 for more details. 34 3.12 In other words, the other factors identified as key correlates of poverty--employment status, age of household head, and household size-help explain some but not all of the observed differences in poverty levels across regions. The results suggest that other region-specific factors, such as variations in the quality of regional infrastructure, play an important role in differences in regional living standards. Table 3-2: Relative poverty risk by region of residence ! I (0) I SIMULATED ACTUAL . Relative Poverty Risk after controlling for the partial effect of: L Region I Relative (I) (2) (3) (4) (5) Poverty Risk I Education I Employment Region Age ! Household I status I size Central 1.65 1.71 1.65 1.29 i 1.68 1.65 i i (0.20) (0.22) (0.20) (0.20) (0.20) (0.20) I Eastern 1.69 1.61 1.67 1.29 1.71 1.67 (0.23) (0.24) (0.23) (0.18) (0.23) (0.23) Zagreb 0.30 0.22 0.35 0.61 0.25 0.33 (0.07) (OJ)6) (0.07) (0.10) (0.06) (0.07) Adriatic North 0.41 0.46 0.34 0.65 0.32 0.37 I (0.14) (0.15) (0.11) (0.19) (0.12) (0.12) f-Adriatic South 0.79 0.85 0.79 1.07 0.85 0.78 (0.16) (0.17) (0.15) (0.19) (0.16) (0.15) Overall Poverty ~e(%) 11.1 9.1 11.6 10.0 11.0 10.8 (0.86) (0.8) (0.9) (0.80) (0.9) (0.83) Source: Background Paper No.3. Standard errors are in parentheses. Box 3-1: Regional Inequalities in Croatia: Legacy of the Past Factors such as the impact of the war and the specificity of past development legacies have had implications for regional inequality in Croatia. Many of the counties comprising the Central and Eastern Croatia regions were war-affected, with severe impacts on certain municipalities reflected through county-level indicators. Although displacement, destruction of infrastructure and housing, and loss of livelihoods have been addressed through various programs and initiatives, the impact persists in terms of lagging development and is reflected in higher unemployment and lower incomes and household expenditures. This legacy is reinforced by the presence in these regions of declining industries in metal processing, chemical engineering, food processing and textiles. Many of the state-owned enterprises in these sectors have ceased to function or operate only with reduced workforces. Lacking concomitant job opportunities and linkages to opportunities, responses to job loss by workers have included: return to small-scale agriculture, family-based coping strategies, job search "discouragement" and participation in the informal economy. The impact of industry decline has had knock-on effects in certain areas as well. 3.2 How SIGNIF1CANT ARE REGIO!'uth Eastern Croatia Overall 0.4 0.2 0.0 Zagreb Central Adriatic North Adriatic South Eastern Croatia Overall Social Assistance Benefits Unemployment Benefits .·· though there seems to be scope for further improvement Note: A comparison of the relative heights of the two bars for each region thus illustrates graphically the extent to which benefits are geographically well-targeted in that region-for instance. if the two sets of bars are equal in each region, this would indicate perfect geographic targeting of benefits. Source: Beneficiaries: Background Paper No.4. A similar picture emerges on comparing survey-based estimates of unemployment with the regional distribution of unemployment beneficiaries (Figure 3-7, right panel). At the same time, however, the graphs show that the Central region is relatively under-served in terms of social welfare benefits, and the Eastern and Adriatic South regions are relatively underserved in terms of unemployment benefits. 3.3 SUSTAINING HIGH GROWTH THROUGH FASTER JOB CREATIO:>l To sustain high rates ofgrowth to permit faster external and internal income convergence. Croatia needs to improve its labor utilization, which in turn will require adoption of a fairly wide menu of policy measures: addressing the remaining discouraging features of the investment climate, including by facilitating the registration ofproperty and by improving the efficiency of the legal system, helping real wages to adjust to productivity and local market conditions (particularly wages in public enterprises. where the restnicturing agenda remains incomplete); and subsidizing employment of young workers or those in economically depressed regions-in preference to their having no jobs at all. The challenge ofjob-creation would also require the government to address its fiscal problem, which in tum has been due to problems with an unreformed health sector, rising social spending. and continued state support to loss-making state-owned enterprises. 3.22 Stimulating the creation of more jobs-and reducing the proportion of jobless households-is a key policy challenge for the government. And only when the jobs created are "good jobs," paying high wages to reward high worker productivity, do higher employment rates necessarily help reduce the risk of poverty. Various factors are known to influence the pace and quality of job creation (Box 3-2). While the specific mix of policies needed to create more and better jobs clearly depends on a country's particular economic and labor market situation, a recent World Bank regional cross-country study on employment opportunities concludes that, in general, improving labor utilization requires a two-track strategy: (1) increasing the pace of job creation and (2) flexibility in the reallocation of jobs and workers. 3D Croatia's overall investment climate and business environment are perceived as relatively favorable and improving. 3 ! In 2003, the government introduced a responsible wage policy, which reversed the trend of rising unit labor 40 costs, and reduced non-wage labor costs. In addition, a new Labor Code was introduced, which is better aligned with EU guidelines and practices in EU member countries. Box 3-2: Polices, Institutions, and Job Creation: Some Lessons from Cross-country Experience The factors below play an important role in influencing the decision of firms to create more, and more productive, jobs, and the decision of workers to stay in the labor market and seek more rewarding jobs: I. Macroeconomic policy setting: Macroeconomic instability and financial crises can discourage firms from undertaking new investment and creating new jobs, as can excessively high interest rates. 2. The cost of doing business: Private investment is often discouraged in the region's countries by a set of factors that increase the cost of new investment, the risks associated with it, and barriers to competition faced by firms. 3. Wage flexibility: While wages have generally beeome more flexible during the transition in the region, there are still many instances in which wage floors and government intervention prevent wages from adjusting. 4. Employment protection legislation: Labor reallocation is influenced by regulations on hiring and tiring. Although several countries have reformed these regulations to better conform to market requirements, the reforms have often focused only on creating flexibility at the margin (for example, through temporary contracts). 5. Social benefits: While benetits have in general played an important role in smoothing the cost of the transition, overly generous benefits, especially during the early phases of transition, have weakened job search incentives. Source: Enhancing Job Opportunities: Eastern Europe and the Former Soviet Union, November 2005. World Bank, Washington DC. 3.23 In many other respects, however, Croatia still faces important challenges. Government spending accounts for about half of GDP-a large share even by European standards. Fiscal problems with an unreformed health sector, rising social spending, and continued state support to loss-making state-owned enterprises have led to persistently high government deficits. Most public enterprises depend on extensive support from the state, for example through subsidies, transfers, or guarantees, as well as periodic bailouts. Subsidies to the enterprise sector constituted 3.4 percent of GDP in 2003 and accounted for well over half the government deficit of 6.2 percent during that fiscal year. Not only is the development impact of this high level of spending an open question, but this eontinued support may well be responsible for impeding the internal reallocation of workers in the Croatian economy from old, less productive firms towards newer more productive firms. 3.24 A fairly wide menu of policy measures could help foster job creation in Croatia: addressing the remaining discouraging features of the investment climate (including by facilitating the registration of property and by improving the efficiency of the legal system); helping real wages 30 World Bank: Enhancing Job Opportunities: Eastern Europe and the Former Soviet Union, November 2005. World Bank, Washington DC. EBRD-lBRD BEEPS study for 2002 and 2005. 41 to adjust to productivity and local market conditions (particularly wages in public enterprises, where the restructuring agenda remains incomplete); and subsidizing wages for young workers, or workers in economically depressed regions-in preference to their having no jobs at all. Strengthening social assistance benefits could potentially be an effective means to protect the incomes of workers displaced from restructured enterprises; one way to afford this is through cutting uproductive social spending and channeling a larger share into targeted benefits. Box 3·3: Doing Business: Objective Measures of Business ReguJations and Their Enforcemeut Since 2004, the World Bank has been preparing an annual ease of doing business index that summarizes in an index the relative performance of different countries calculated on the basis of the country percentile rankings in various areas of business activity. For example, the 2007 Doing Business covers 10 topics, including starting and closing a business, licenses, taxes, trade, property registration, contract enforcement, access to credit, protection of investors, and t1exibility of labor regulations. Croatia ranked as one of the top ten reformers in 2005/06 based on this index, significantly improving its overall ranking in three of the ten categories covered (see table below). For instance, the costs and procedures involved in importing and exporting goods declined appreciably between 2005106, with time taken for export and imports from final contractual agreement to delivery of goods are now over 10 and 50 percent less than the regional averages. Ease of... 2006 rank 2005 rank Change in rank I Doing Business 124 134 +/0 I Starting a Business 100 112 +12 I Dealing with Licenses 170 17l +1 Employing Workers 130 131 +1 Registering Property J09 J09 0 Getting Credit 117 117 0 Protecting Investors 156 156 0 Paying Taxes 58 72 +14 Trading Across Borders 92 ]40 +48 Enforcing Contracts 28 28 0 Closing a Business 80 76 -4 Notwithstanding these recent improvements, Croatia still has much ground still to cover to further improve its overall ranking with regard to ease of doing business. For instance, the costs of dealing with licenses, as proxied by the procedures and time taken to build a warehouse, are considerably higher than the regional average, and almost twice the average for OECD countries. Similarly, rigidity of employment index in Croatia is still considerably higher than the regional and OECD averages. Source: See www.doingbusiness.org for more details. 3.25 Finally, while Croatia has made good progress in 2005/06 with improving regulations governing various aspects of everyday business activities (see box), there remains considerable room for further improvement to narrow the gap in ease of doing business in the country compared to other more business-friendly economies in the region (e.g. Denmark, Norway, Ireland, Baltic states). 42 3.4 BUILDING ON LOCAL COMPARATIVE ADVANTAGES 3.26 In the context of regional development analysis and planning, an important question is how best to define regional disparities and at what level of territorial aggregation. An important finding from our analysis is that the ranking of different territories in Croatia can vary considerably depending on the development indicator used for this purpose. This is particularly true at the county level, where it points to the need for caution in relying on anyone indicator for ranking purposes. For instance, the lists of highest ranked counties based on many development indicators all include the counties of Zagreb city, Primorje-Gorski kotar, and Istria. But the lists of the five poorest counties contain different counties depending on whether the ranking criterion is per capita 2003 GDP or the survey-based estimates of per capita incomes, expenditures, or monthly earnings (only 3 of 15 possible matches, see Table 3-4). Not one of the five counties with the highest poverty rates as calculated using the HBS data features in the list of five poorest counties as ranked by per capita 2003 GDP. Table 3·4: Alternative lists offive nmlrp,n counties in Croatia Ranking GDP per capita Per capita Per capita Ave. monthly Headcount (l=poorest) (2003) income consumption earnings Poverty Rate (% ) Virovitica Karlovac Karlovac Podravina Virovitica 2 Podravina VaraZdin Sisak-Moslavina Bjelovar 3 Sisak-Moslavina Bjelovar-Bilogora Bilogora Koprivnica 4 Krizevci 5 Osijek-Baranja Source: World Bank estimates based on 2002-04 HBS and LFS. GDP per capita from background paper No.4. 3.27 While counties in Croatia correspond to the EC NUTS III level of disaggregation, other European countries typically assess regional inequalities at a higher level of aggregation (the NUTS II level). Indeed, many observers have argued that in the case of smaller EU members such as the Baltic states, Czech Republic, Hungary, Ireland, Slovakia, or Slovenia, the pertinent unit should be the entire countryY A similar argument might be made in the case of Croatia, given the country's relatively small overall size and population. This report argues in favor of the middle ground. Given the relatively large unexplained regional differences in Croatia's living standards, a regionally differentiated strategy probably makes good sense. Diverse factors such as the impact of the recent war, varied geographic terrain, and special problems faced by those regions with a higher concentration of traditional industries currently facing problems have resulted in considerable variation in living standards across different parts of the country. However, while a focus on regional development is indeed justified, there is some need for caution to avoid focusing on too small a geographic unit of analysis for this purpose. Michele and Boldrin in B. Funck and L. Pizzati European imegration, Regional Policy, and Growth. 43 Figure 3-8: Robust regional rankings based on various development Indicators 140 140 Income per capita (2002104) Consumption per capRa (2002104) 120 120 100 100 80 80 I I 60 60 40 40 20 20 0 Eastern Central Acriatic Mialic Zagreb Central Eastern AQiatic AQiatic Zagreb South North South North 2S 160 Poverty Rate (2002104) GOP per capita (2003) 140 20 120 15 100 80 ·· III 10 60 ' 0 Central Eastern I Mimic South Zagreb Adriatic North 40 ' 20 Eastern Adriatic South Central Aaialic North Zagreb 4,000 20 Ave. Monthly earnings (2002104) Post-Secondary Completion (~ 2002104) 3,800 18 3,600 18 3,400 14 3,200 12 3,000 10 I 2,800 I I II 8 2,600 2,400 4 2,200 2,000 0, Central Eastern Adriatic Adriatic Zagreb Eastern C«lraI Atiialic Aliialic ZIJr;pl) North South Seith North Note: Vertical axis of income per capita, consumption per capita, and GDP per capita are expressed as percentage of national average. Poverty rates and post-secondary rates are expressed as percentages, Average monthly earnings are in Kunas per month. Source: World Bank estimates based on 2002-04 HBS and LFS data, GDP per capita estimates are from background paperNo. 4. 3.28 In contrast to the considerable variability in county-level rankings, we find that the ranking by level of prevailing living standards differs much less across the analytic regions used in this report, no matter which development indicator or data source is used. In general, average incomes and expenditures in Central and Eastern Croatia are below the national average, those in 44 the Adriatic South are at par with, and those in the Adriatic North and Zagreb regions are higher than the overall national average (Figure 3-8). 3.29 Initiatives with the greatest chance of success tend to work with the forces of economic geography, through building on local comparative advantages and drawing on local knowledge and initiativeY They are typically based on a diagnosis of a region's comparative advantages and the constraints on its development. The national regional development strategy focuses on identifying key growth sectors and actions to be taken by various stakeholders, including local and regional governments, private firms, and nongovernmental organizations. For instance, the OECD's analysis of the Teruel region in Spain cites the region's unspoiled countryside and agricultural traditions to advocate an economic strategy based on the niche marketing of local products-ham, olives, peaches-and farm-based tourism. 34 Box 3-4: Regional Operational Programs in Croatia The EU ROP methodology and approach are being used in Croatia to access EU CARDS (Community Assistance for Reconstruction, Development, and Stabilization) program funding for the development of counties. The main aim of this exercise is to assess and address the development needs of each county, and to help allocate resources, both local and international, for this purpose. The main ROP principles agreed between the EC and counties in Croatia include: (1) ownership, i.e. that each county has the responsibility for preparing and implementing the ROP, (2) commitment, i.e. that each county is committed to the ROP process and principles, including making available the needed counterpart funding, (3) transparency and openness in the preparation and implementation of these programs, and (4) partnership and stakeholder consultation, i.e. that all relevant stakeholders at the national and county level should be consulted at each key stage of program preparation and implementation. Note: More details pertaining to the allocation of EU funds based on the ROP principles can be found in various EU regional policy documents and publications. 3.30 In Croatia, regional operational programs (ROPs) are being developed at the county level mostly over the period 2004-10 (some regions started earlier). As part of this exercise, various local projects are planned through a highly consultative and collaborative process (Box 3-4), and compete for funding. In addition, steps are taken throughout the ROP process to help ensure that the strategic framework developed is well aligned with the county's comparative and competitive advantages. Hence in this respect, the ROP exercise is fully consistent with the strategic principles outlined above. Looking ahead, however, as the Croatian government moves towards preparing higher-level regional development projects under the county and a wider-region development program, it will be important to accord high priority to initiatives that build upon commonalities across contiguous counties (that is, in terms of areas of comparative advantage identified), so as to avoid bogging down the overall regional development effort in an undue proliferation of mini projects?5 33 Dillinger, W. Concept paper on "Regional Development: What Works and What Doesn't?" 34 On the whole, agriculture tends not to be the focus of regional development initiatives, given the sector's generally low employment-generation potential, though as illustrated above, there clearly are exceptions to this rule. 33 In fact, Boldrin and Canova (2003) have argued that in the case of the small new member states of the European Union, the NUTS-I level may well be the most pertinent for regional policy. However, as argued in this report, a more disaggregated approach may well make sense for Croatia, which appears to have significant regional heterogeneity and bottlenecks to growth. 45 3.5 LABOR MOBILITY AND GREATER LABOR MARKET FLEXIBILITY 3.31 As seen in Chapter 2, as much as one-third ofthe wage differentials between leading and lagging regions in Croatia can be attributed to differences in worker characteristics, including education. These findings clearly point both to the importance of raising human capital in lagging regions as well as of facilitating greater labor mobility as a means of reducing inter-regional earning disparities. Classical migration theory suggests that wage differentials across regions will spur workers to move from low- to high-wage regions, thus reducing regional inequalities in living conditions. However, there are several reasons why in practice this may not happen. Since workers in lagging regions tend to have less education than those in more prosperous regions, and since most of the labor demand in the latter regions is likely for higher-skilled positions, unemployed or low-wage workers in poorer regions may not gain anything by migrating to the richer regions; staying at home may well be a rational decision on their part. 3.32 Public policy can also help to reduce barriers to the interregional migration of workers by reducing market imperfections. Poor information about job opportunities in other regions or a low perceived likelihood of finding employment may result in low interregional mobility, even among the relatively highly skilled. Difficulties associated with finding housing in other regions as well as other housing market imperfections and distortions may also lead to similar problems. Poor transport links, particularly with nearby regional growth centers such as medium-sized cities, may result less-than-optimal commuting within regions. Finally, overly generous social assistance programs with inadequate attention to work-disincentive issues may likewise reduce the mobility of labor across regions. Policies aimed at correcting these market imperfections can have an important role to play in facilitating freer flows of workers across regions, which in tum can be an extremely powerful channel for equalizing interregional differences in living standards. 3.33 Similar analysis of the earnings differentials between public and private sector workers shows that the average monthly earnings of workers in state firms are about 30 percent higher than earnings of those employed by the private sector. Furthermore, differences in employee characteristics-such as age, years of education, and gender-at best explain only about one-third of the earning differential between these two sectors, while the remainder is unexplained. This in tum points to the problem of incomplete restructuring of public sector enterprises in Croatia, which continue to pay their workers salaries higher than those received by workers with similar qualifications and characteristics employed in other sectors, which may hinder faster internal reallocation of workers between various sectors. Policies aimed at correcting these market imperfections can have an important role to play in facilitating freer flow of workers across regions and sectors of the economy. When implementing such policies, it would be important to ensure the presence of effective social safety nets to protect incomes of workers displaced in the wake of enterprise restructuring. 46 3.6 IMPROVING THE ADEQUACY AND TARGETING OF SOCIAL SAF'ETY NETS 334 A key feature of the government's program to reduce poverty and social inclusion is to improve the effectiveness of the social protection system, by (1) rationalizing spending and improving targeting, and (2) strengthening system administration, As pointed out in Chapter 2, poverty in Croatia is quite shallow and the incomes of the poor are not far below the poverty line, This means that the resources involved in eliminating poverty are reasonably affordable: a perfectly targeted transfer of approximately 1.5 billion kunas (about 0,7 percent of GDP) would have been sufficient to eliminate poverty in the country in 2004, Further, even relatively modest increases in average incomes in poorer regions would lead to a considerable narrowing in regional poverty differentials. For example, if average per capita expenditures in rural Central Croatia were to increase by the equivalent of only 10 percent of the national average, the prevailing poverty rate would be halved (i.e. drop from 22 to 11 percent). If the increase were 20 percent, this would eliminate entirely the differential in poverty rates with the urban Zagreb region. 3.35 Social spending in Croatia is high by international standards, but only a small share of it goes towards means-tested programs. For instance, recent data show that while total social spending has increased to 31 percent of GDP (17 percent on social transfers, 5 percent on education, and 9 percent on health care), only a very small fraction (0.7 percent of GDP) is used for the poverty related social assistance program. The social assistance program is the best targeted to people in need, but continues to have low coverage rates, even among the poor. For its beneficiaries it is an important source of income. 6.36 Another important dimension of targeting is the extent to which, within each region, benefits are channeled to the people who most need assistance. To assess this, we use data from the Household Budget Survey series. As well as comparing the targeting performance of programs, comparing data for 2002 with those for 2004 allows us to assess how targeting has changed over time. The surveys include data on the recipients of various types of social transfers, which have been grouped into the following broad analytic categories: (1) child allowances, (2) other family allowances, (3) social assistance, and (4) unemployment benefits. 337 We compare the performance of these various types of transfers in reaching the poorest one-fifth of the population (henceforth referred to as the "target group") using three related criteria: (1) targeting (that is, the share of total program spending accruing to the target group); (2) coverage (that is, the share of this group receiving benefits), and (3) adequacy (that is, the share of their total consumption provided by this transfer). Population quintiles are derived on the basis of per capita consumption (minus transfer receipts). 3.38 Our analysis indicates that social assistance (support allowance) is the best-targeted program in Croatia. More than two-thirds of the total spending under this program accrues to the poorest one-fifth of the Croatian population (Figure 3-9). Unemployment benefits are found to be the next -best targeted program, with about 40 percent of total benefits reaching the target group of interest; moreover, the targeting of unemployment benefits appears to have improved significantly between 2002 and 2004, probably as a result of the employment legislation changes that occurred in 2003. By contrast, neither child nor other family allowances appear to be particularly well targeted: the poorest fifth of the population receive only 34 and 29 percent of the spending on these allowances, respectively--only slightly more than the 20 percent this group would receive if the program were completely untargeted. 47 Figure 3-9: Social assistance is the best targeted program in Croatia 80 70 60 50 44 40 j40 34 30 29 30 20 10 o Child Allowance Other Family Social Unemployment Any of these Allowances Assistance benefits benefits . · 2002 .2004 Share of Spending Accruing to Poorest One-Fifth of the Population Source: World Bank estimates based on 2002 and 2004 HBS. 3.39 Overall, about 35 percent of the population in Croatia is covered by transfer programs (Figure 3-10). Child allowances are much the largest of the various programs considered here, and reach about 26 percent of the country's population. The population coverage rates of other family allowances (10 percent), the social assistance program (5 percent), and unemployment benefits (6 percent) are considerably lower. The social assistance program does not reach the great majority of poor people: it covers only about 13 percent of the poorest one-fifth. Figure 3-10: All programs have relatively low coverage rates, even among the poor 70 Program CowFage Rates (2004) 60 60 50 40· 35 J 30 · Overall population. Poorest one-fifth .. .ill 20 12 13 10 10 o Cllild Allowance Other Family Social Unemployment Any of these Allowances Assistance benefits benefits Source: World Bank estimates based on 2004 HBS. 48 3.40 Total transfers under social assistance are small in relation to other programs, at less than 2 percent of total social spending by the government, but are an important income source for the poor, supplying 16 percent of its beneficiaries' consumption per-adult equivalent (Figure 3-11). Figure 3-11: Importance of transfers for beneficiaries of the social assistance program 90 18 _ Consumption per Equivalent Adult -+- Social Assistance 80 16 70 14 ~ E <>'l i" 60 12 sa, § l 50 10 '" ~ '0 40 ~ 8 i· -a ~ .8 B 30 6 ~. F 20 4 ~ ,.. 10 2 Poorest 2 3 4 Richest Income Group Note: Left axis show the consumption per equivalent adult in thousands of kuna, while the right axis shows social assistance transfers received as a percentage of consumption per equivalent adult of the group. Source: World Bank estimates based on 2004 HBS. 3.41 A high proportion of Figure 3-12: Who benefits from social transfer programs? social assistance beneficiaries labor Market Status of Household Head (2004) report the main labor market 100 status of their household 90 head as "retired," "other," 80 or "unemployed" (Figure 70 3-12) Does Croatia's social assistance program discourage 60 people from working? Before 50 possibly recommending an 40 expansion of the social as 30 sistance program because of 20 its relatively good targeting performance and low cove to rage rate, one must first con 0 sider the adverse impact that Pensions Family Social Unemployment Overall Allowances Assistance benefits Population an expansion might have on · Employed · Unemployed C Retired C Other Croatia's already low labor Source: World Bank estimates based on 2004 HBS. market activity rates. 49 3.42 The Croatian government has introduced a number of measures during the past few years to improve the performance and effectiveness of the overall social welfare system. Following the modification of the applicable law and rationalization of expenditures, total expenditures on child allowances decreased from HRK 2.415 billion in 2001 to HRK 1.509 billion in 2004. Moreover, HBS data suggest that the government succeeded in protecting the poorest segments of the population from bearing a disproportionate share of these cuts in spending, and the share of total spending on child allowances accruing to the poorest one-fifth of the population increased during this period, from 28 percent of total expenditures in 2002 to 34 percent in 2004 (Figure 3.9). 3.43 Three years after those measures were issued, several of the recommendations of studies that were commissioned as part of this overall reform effort remain valid today (Box 3-5). Box 3-5: Reforming the Social Welfare System in Croatia In the course of preparing the Social Protection Project, the Ministry of Labor and Social Welfare commissioned a series of studies in 2003. Among these, a study by the Labor and Employment team proposed ways to restructure programs and policies so as to help welfare recipients move into productive employment. The recommendations covered four areas: (I) Policies to help social welfare beneficiaries back to work: To help reduce both the flows into unemployment as well as the stock of long-term unemployed: · Target active labor market measures (ALMM) (employment subsidies, labor market training, and measures to promote jobs for disabled workers and youth to those still on the unemployed register after six months; · Intensify "activation" for people without work for more than six months, including the introduction of compulsory job-search workshops after 12 months of unemployment; · Develop new measures to combat employers' negative perceptions of the long-term unemployed; · Focus ALMM more on improving the basic skills of the long-term unemployed; and · Introduce a major new workfare program for long-term unemployed recipients of welfare, so that they can gain work experience and be reconnected with the world of work. (2) Integration (J{ social welfare and employment programs: To achieve more client-centered service provision and better integration of the centers for social welfare (CSW) and employment (CES): · Allow CES direct access to the welfare payment system to be able to allow/stop welfare payments for registered unemployed people; · Achieve greater clarity and emphasis on the role of CES in relation to long-term unemployed people and welfare recipients in particular; and · Strengthen joint work between CES and CSW, particularly with regard to helping ensure better work opportunities for the long-term unemployed. (3) Cost-effective active labor market programs (ALMP): To remedy current weaknesses in these programs: · Administer unemployment compensation in an integrated manner with other services such as counseling, mediation; · Target ALMP on the long-term unemployed as well as those at high risk; and · Make the source of funding for ALMP more secure. (4) Tackle social exclusion through jobs: Introduce a national program of workfare. Source: Final Report: Labor and Employment Team, May 2003. Predrag Bejakovic and Ray Phillips, Croatia Social Protection Project, Ministry of Labor and Social Welfare. 50 3.44 Measures that are likely to yield a high payoff include efforts to further improve system administration (for example through integration of the centers for social welfare and the centers for employment as well as with the local government units), and improving information technology and record-keeping and to improve the orientation of the various programs (for example by using fewer passive and more active measures, better counseling, and job-placement services). Similarly, Croatia has considerable scope for strengthening the poverty impact of social spending by improving geographic targeting through increasing the program coverage rates in poorer regions. 3.45 The social assistance program in Croatia clearly does much better than other programs in reaching the poor-69 percent of total spending on this program accrues to the poorest one-fifth of the population (Figure 3.9) compared to only 34 percent for child allowances, and 29 percent for family allowances respectively. Can the government improve the overall mix of total social spending by spending relatively less on child allowances and family benefits, and more on well targeted programs like social assistance? A direct comparison of these programs is complicated by the fact that providing assistance to poor families is only part of the main objectives of child and family allowances-an important additional policy goal of these programs is to provide incentives to boost the overall population growth rate. Further study of the impact of reduced spending on child allowances between 2001 and 2004 on the overall fertility rate in Croatia thus merits serious consideration, so as to ascertain the extent to which this program provides a cost-effective means of increasing the country's birth rate. 51 52 APPENDIX List ofDerived Indicators and Data Sources Output and Incomes: 1. GDP per capita (Croatia=lOO, using 2003 Regional Accounts data) 2. Average per capita income (Croatia=100, using 2002-2004 HBS data) 3. Average per capita expenditures (Croatia=IOO, using 2002-2004 HBS data) Labor Market Indicators: 4. Employment Rates (2002-2004 LFS) 5. Unemployment Rates (2002-2004 LFS) 6. Average Earnings from Primary Job (2002-2004 LFS) Living Conditions: 7. Poverty Incidence (2002-2004 HBS) 8. Poverty Gap (2002-2004 HBS) 9. Squared Poverty Gap (2002-2004 HBS) Share of Population 16 years and older with Highest Educational Attainment: 10. Unfinished Primary (2002-2004 HBS) 11. Primary (2002-2004 HBS) 12. Vocational Secondary (2002-2004 HBS) 13. General Secondary (2002-2004 HBS) 14. Post-Secondary (2002-2004 HBS) 53 A1: Poverty Gap at the County-Level (2002-04 HBS) Poverty gap 0-1.5 1.5 - 4 4.1 - 6.5 6.6 - 9 9.1 · 11.3 A2: Poverty Incidence at the County-Level (2002-04 HBS) Squared gap 0-1 1.1 - 2 2.1 - 3 3.1 - 4 4.1 - 5.3 54 A3: GDP per capita at the County-Level (2003 regional accounts data, Croatia=l00). G D P per-capita 0-76 77 - 102 103 - 128 129 - 154 155 - 179 A4: Average per capita consumption at the county level (2002-04 HBS) Average per-capita consumption 0-76 77 - 89 90 -104 105 -119 120 - 131 c/ -' ""'"-"~"~"-~ ( 55 AS: Average monthly earnings from primary job at the county level (2002-04 LFS) Average earnings 0-2650 2651 - 2950 2951 - 3250 3251 ·3550 3551 ·3895 A6: Employment Rates at the county level (2002-04 LFS) loyment rate 0·43 44·50 51 ·57 58·64 65·70 56 A7: Unemployment Rates at the county level (2002-04 LFS) Unemployment rate 0-8 9 - 13 14 - 18 19 - 23 24 - 28 AS: Share of population with incomplete primary education (2002-04 LFS) Share of >16 population 0-6 6.1 - 12 12.1-18 18.1 - 24 24.1 - 26.8 57 58