Report No. 31011-AL ALBANIA SOCIAL SAFETY NET REVIEW December 29, 2004 Human Development Sector Unit Europe and Central Asia Region FOR OFFICIAL USE ONLY Document of the World Bank _________________________ ALBANIA SOCIAL SAFETY NET REVIEW CURRENCY EQUIVALENTS (Exchange Rate as of December 29, 2004) Currency Unit = Lek Lek 1 = US$ 0.01094 US$ 1 = Leke 91.399 FISCAL YEAR January 1 ­ December 31 ABBREVIATIONS AND ACRONYMS BFP Bolsa Familia Program CCT Conditional Cash Transfers CIT Credit Income Tax DFID Department for International Development ECA Europe and Central Asia EU European Union GAASS General Administration of Assistance and Social Services GDP Gross Domestic Product GMI Guaranteed Minimum Income GTD General Tax Directorate IDA International Development Association LSMS Living Standards Measurement Survey MPC Marginal Propensity to Consume MoF Ministry of Finance MoLSA Ministry of Labor and Social Affairs MRT Marginal Rate of Tax NES National Employment Service NGO Non-governmental Organization NIT Negative Income Tax PHC Population and Housing Census PRSC Poverty Reduction Strategy Credit SIF Social Insurance Fund SII Social Insurance Institute SSDP Social Services Development Project SSN Social Safety Net SSS State Social Services TANF Temporary Assistance to Needy Families TFP Total Factor Productivity Vice President: Shigeo Katsu Country Director: Orsalia Kalantzopoulos Sector Director: Charles C. Griffin Sector Manager Hermann von Gersdorff Task Team Leader: Verdon Staines ALBANIA SOCIAL SAFETY NET REVIEW Table of Contents ACKNOWLEDGEMENTS..................................................................................................................i SUMMARY.......................................................................................................................................ii CHAPTER I: ALBANIA'S CURRENT SOCIAL SAFETY NET: ITS LOCAL AND REGIONAL CONTEXT, STRUCTURE, AND FUNCTIONING..............................................................................1 A. Evolving Local Context ..........................................................................................................1 B. Current Social Safety Net and Social Protection Programs:........................................................9 Their Structure and Functioning ....................................................................................................9 Social Insurance System....................................................................................................... 10 Social Safety Net................................................................................................................... 15 C. Regional Context: Social Safety Net Experience from Eastern Europe ....................................29 and Other Comparator Countries .................................................................................................29 CHAPTER II: ASSESSING THE IMPACT OF SOCIAL TRANSFERS ON POVERTY AND INCOME DISTRIBUTION AMONG HOUSEHOLDS......................................................................................35 A. Summary .............................................................................................................................35 B. Introduction .........................................................................................................................35 C. Coverage of Public and Private Transfers ­ Which Transfers are Targeted Better Towards Poor Households?.........................................................................................................................38 D. Distribution of Public and Private Transfers Among the Public - Which Transfers Disburse Larger Sums to Poor Households? .........................................................................................44 E. Impact of Public and Private Transfers on Poverty Reduction ­ Which Transfers are More Successful at Reducing Poverty? ...........................................................................................46 F. Impact of Social Transfers on Poverty Reduction Under an Alternative Assumption ­ Which Transfers are More Successful at Reducing Poverty, Assuming Private Remittances are Treated as Income Rather Than Transfers?.........................................................................................54 G. Efficiency Analysis - Which Transfers Were More Efficient in Terms of Their Cost Per Unit Reduction of the Poverty Gap? ..............................................................................................61 H. Overall Effectiveness of Social Transfers in Reducing Poverty - Which Transfers are More Effective Than Others?.........................................................................................................62 CHAPTER III: DIRECTIONS FOR THE FUTURE..........................................................................65 BIBLIOGRAPHY ............................................................................................................................87 List of Boxes Box S.1: Estimating How Social Safety Net Programs........................................................................iv Affect Household Incomes Relative To Poverty Lines..........................................................................iv Box S.2: Large Overall Impact of Social Safety Net.............................................................................iv on Pre-Transfer Poverty......................................................................................................................v Box 2.1: Methodology .....................................................................................................................36 Box 2.2: Estimating Total Transfers From LSMS Data.......................................................................41 Box 2.3: Determining the Marginal Impact of Social Transfers on Poverty and....................................49 Extreme Poverty After Receipt of Social Insurance Transfers..............................................................49 Box 2.4: Impact of Transfers Through Ndihma Ekonomikeand Rural Pensions, on Welfare of Rural Households Having at Least One Family Member of Pensionable Age.................................................54 List of Figures Figure 1.1: Index of GDP at Factor Cost in Constant Prices, 1987-2001 (1987 = 100) ............................1 Figure 1.2: Distribution of GDP at Factor Cost, by Sector, 1987-2001...................................................2 Figure 1.3: General Government Expenditures, Revenues and Deficit as Percentage of GDP, 1990-20022 Figure 1.4: Worker's Remittances to Albania as a Percentage of GDP, 1995-2003 .................................6 Figure 1.5: Government Current Spending and Selected Social Components, as Shares of GDP, 1993- 2002 ..................................................................................................................................................9 Figure 1.6: Total Real Social Insurance Expenditures, Benefit Costs, and Contributions,......................12 in Billions of 2003 Leke, 1994-2003 ..................................................................................................12 Figure 1.7: Social Insurance Dependency Ratios, by Pension Program, 1994-2003 ..............................12 Figure 1.8: Replacement Rates (Average Pension: Average Wage), by Pension Program, 1993-2004 ....12 Figure 1.9: Cumulative Growth Rate of Real Value of Urban Pension Since 1994................................13 Figure 1.11: Estimated Real Annual Expenditure on Ndihma Ekonomike in Billions of 2003 Leke........20 Figure 1.12: Number of Households Receiving Ndihma Ekonomike Benefit, 1996-2003 (in thousands).21 Figure 1.13: Average Monthly Ndihma Ekonomike Benefit in Real 2003 Leke, 1998-2003...................22 Figure 1.14: Monthly Unemployment Benefit Amount in Nominal Leke and Real 2003 Leke, 1993-2001 ........................................................................................................................................................22 Figure 1.15: Distribution of Households Receiving Ndihma Ekonomike Benefits, by Household Size, 1994-2001........................................................................................................................................23 Figure 1.16: Proportions of Households in Poverty and Receiving Ndihma Ekonomike,........................24 by Prefecture, 2002...........................................................................................................................24 Figure 1.17: Real Annual Program Expenditure on Disability Benefits and Services, 1996-2003...........25 Figure 1.18: Recipients of Disability Cash and Service Benefits, 1996-2003........................................26 Figure 1.19: Real Annual Program Expenditure Per Beneficiary for Disability Benefits and Services, 1996-2003........................................................................................................................................26 Figure 2.1: Distribution of Income by Source, All Households............................................................37 Figure 2.2: Distribution of Income by Source, By Type of Household .................................................38 Figure 2.3: Impact of Total Public and Private Income Transfers on....................................................47 Indicators of Poverty, Treating Remittances from Abroad as Private Transfers.....................................47 Figure 2.4: Impact of Total Public and Private Income Transfers on Indicators of Extreme...................47 Poverty, Treating Remittances from Abroad as Private Transfers.........................................................47 Figure 2.5: Impact of Income Transfers on Poverty Indicators, ...........................................................48 By Source of Transfer, Treating Remittances from Abroad as Private Transfers....................................48 Figure 2.6: Impact of Income Transfers on Extreme Poverty Indicators,..............................................49 by Source of Transfer, Treating Remittances from Abroad as Private Transfers....................................49 Figure 2.7: Poverty Indicators Before and After Total Transfers,.........................................................59 Treating Remittances from Abroad as Earnings of Recipient Household ..............................................59 Figure 2.8: Extreme Poverty Indicators Before and After Total Transfers,...........................................60 Treating Remittances from Abroad as Earnings of Recipient Household ..............................................60 Figure 2.9: Impact of Public and Private Transfers on Poverty Indicators,............................................60 Treating Remittances from Abroad as Income ....................................................................................60 Figure 2.10: Impact of Public and Private Transfers on Extreme Poverty Indicators, Treating Remittances from Abroad as Income.....................................................................................................................61 Figure 2.11: Targeting­Efficiency Analysis of Social Transfers..........................................................62 Figure 2.12: Effectiveness of Public and Private Transfers in Reducing Poverty ..................................63 Figure 2.13. Effectiveness of Public and Private Transfers in Reducing Extreme Poverty.....................64 List of Tables Table S.1: Distribution of Beneficiaries and Transfers, by Type of Transfer.......................................viii Table 1.1: Social Insurance Contributions Rates for Employees, 2004.................................................11 Table 1.2: People with Disabilities, 2004, by Category.......................................................................27 Table 2.1: Distribution of Average Monthly Household Income, Amount in Leke................................38 Table 2.2: Coverage: Percentage of Individuals Within Each Category................................................39 Receiving Public or Private Transfers................................................................................................39 Table 2.3: Distribution of Program Beneficiaries ...............................................................................43 Table 2.4: Distribution of Public and Private Transfers: Percentage of Transfers.................................45 Received by Each Category (Absolute Target Incidence).....................................................................45 Table 2.5: Percentage Point Reduction in Poverty and Extreme Poverty..............................................50 by Location of Household, Treating Remittances from Abroad as Private Transfers..............................50 Table 2.6: Percentage Point Reduction in Poverty and Extreme Poverty by Region,.............................52 Treating Remittances from Abroad as Private Transfers......................................................................52 Table 2.7: Percentage-Point Reduction in Poverty and Extreme Poverty, by Household Composition,...53 Treating Remittances from Abroad as Private Transfers......................................................................53 Table 2.8: Distribution of Beneficiaries and Transfers, by Type of Transfer,........................................57 Under Original and Alternative Counterfactual Consumption ..............................................................57 Appendices Appendix 1: What Constitutes A Social Safety Net? ..........................................................................76 Appendix 2: Impact on Poverty Headcount of Assumed Marginal Propensity to Consume from Transfers ........................................................................................................................................................78 Appendix 3: Impact of Income transfers...........................................................................................82 on Post-transfer Rates of poverty.......................................................................................................82 Appendix 4: Poverty Profile ..............................................................................................................84 ACKNOWLEDGEMENTS This task was managed by Verdon Staines (ECSHD). The World Bank team included: Philip Goldman, Monika Huppi, Lorena Kostallari, Arvo Kuddo, Nicole La Borde, Carmen Laurente, Cem Mete, Akshay Sethi, Florian Tomini, Alan Thompson, William Tracy, and Yang-Ro Yoon The team is grateful for the support of Orsalia Kalantzopoulos, Country Director (ECCU4); and Hermann von Gersdorff, Social Protection Sector Manager (ECSHD). The team benefited from comments of peer reviewers Jean-Jacques Dethier and Sandor Sipos and, in addition, from the comments of Arup Banerji, Jeanine Braithwaite, Hideki Mori, Edmundo Murrugarra, Michal Rutkowski, and Arshad Sayed. Special thanks to Carmen Laurente, who edited and finalized the document, and to Cem Mete, Akshay Sethi, and William Tracy for excellent research assistance. i SUMMARY CONTEXT (i) In 1989, Albania's rigid political and socioeconomic structure shattered beyond repair. Turbulence soon invaded every domain of life. As the state imploded, so did the state-run economy. Collectivized farming gave way to privatized agriculture, often on inefficiently small landholdings. Industrial and mining complexes ground to a halt. Together, these former employers shed thousands of jobs. Underemployment, informal employment, self-employment, and unemployment ballooned, deflating family incomes. GDP plummeted by 40 percent. Skills atrophied. First internal migration, and then external migration, redistributed the population geographically in complex patterns. Migration also changed the population's composition and gender balance, began to split individual families, and reshaped cities and villages. As the old safety net of an assigned job, a low but guaranteed wage, and subsidized consumer prices became unsustainable, economic insecurity cast a pall over the future. Calls mounted for the fledgling democratic government to take ameliorating action. Compensation payments to laid- off workers and early retirement options for older workers gave way to a new income-tested economic aid program (ndihma ekonomike) and to the enactment in 1993 of a western European- style social insurance system to subsume the previous pension system. Albania was on its way to a new social safety net strategy, although public benefits were small in a resource-scarce environment. Fortunately, a decade of largely uninterrupted robust economic growth was beginning to resume from a shrunken base. Yet economic growth did not deliver the formal employment growth that all were hoping would disperse the long lines of jobseekers. Rather, GDP growth was resourced largely from reductions in underemployment within the formal sector and from informal and self-employment. THE REVIEW (ii) Scope. This review explores ongoing consequences of this difficult transformation and of policy initiatives to mitigate or ameliorate its effects. The review builds on Albania: Poverty Assessment (World Bank, 2003), confirms or extends earlier findings about functioning of the social safety net, and proposes directions and priorities for its future development. Although Albania has been much studied, the review addresses important information gaps. It: Ø Establishes, from the representative 2002 Living Standards Measurement Survey (LSMS), which kinds of families receive public income transfers and private income flows in what amounts and from what sources; Ø Estimates the impact that these public and private income flows have in mitigating pre- transfer and post-transfer poverty among households. It does this by type of family and source of income flow, for all Albanian households and for various subsets of them; Ø Analyzes the relative importance of pensions and social assistance in alleviating poverty among rural households with aged members; Ø Documents and links data on expenditures under social programs with data on program beneficiaries for recent years; Ø Identifies current tradeoffs among the multiple objectives of social safety net policies, including those relating to coverage, adequacy, work incentives, equity, and effective government spending; and Ø Suggests options for future development of social safety net policies, including ways to strengthen overall policy linkages and programmatic coordination among social programs. ii (iii) Elements of Social Safety Net. Albania's public social safety net (SSN) consists of the following core elements: Ø Three cash social assistance programs for which the Ministry of Labor and Social Affairs (MoLSA) has policy responsibility, namely, § the income-tested cash social assistance benefit for families (ndihma ekonomike), § the monthly cash allowance for people who develop disabilities while children or students, and § price compensation paid to pensioners and their families, which was introduced when most general consumer subsidies were eliminated; Ø Social insurance programs, administered by the Social Insurance Institute; and Ø Social care services for orphaned, disabled, and elderly people, which are provided or regulated by State Social Services. Supplementary assistance comes other public programs and as private assistance from non- government organizations (NGOs). In addition, many Albanian households receive private financial flows from family members who are not living in the same household. Most commonly, family members working abroad send these funds, which are known as "remittances." In other cases, they come from family members elsewhere in Albania, such as transfers from rural areas to a family member living in an urban area. (iv) Two Views of Remittances. Remittances from abroad can be viewed from two perspectives, which have different implications for measuring the level of pre-transfer poverty and the distributional impact of transfers. One perspective sees such remittances as private transfers of income to Albanian households from outside the country that can be analyzed symmetrically with other public and private income transfers. The other perspective regards these remittances as earnings of the recipient household, not transfers, because they result from a migration-based, household employment strategy that responds rationally to depressed domestic demand for labor. Each perspective yields insights. The former perspective acknowledges that not all plans for work abroad result in actual employment, savings from earnings, and remittances that arrive safely. Hence, emigrants' households are particularly vulnerable; and remittances crucially affect their living standards. Compared with the former view, the latter view implies more income and less poverty before transfers, and a smaller role for transfers and the safety net. Analyses are undertaken from both perspectives. FINDINGS (v) Interdependent Problems. One inescapable conclusion is the causal interdependence of many of Albania's major problems. Social safety nets cannot substitute, on an ongoing basis, for strong macroeconomic performance. High rates of migration, unemployment, and labor force informality illustrate the socioeconomic stress and dislocation that derive from fundamental disequilibria in Albania's economy and labor market. Simply put, Albania has too few jobs, especially "good jobs" in the formal sector; and the geographic distributions of people, jobs, job- seekers, and infrastructure diverge. This dearth of productive job opportunities in the formal sector has induced large internal and external migration flows. These flows have had multiple and complex effects on family structure, incomes, types and location of employment, levels of unemployment and underemployment, security of housing tenure, and access to services iii including education, health, transport, water, sanitation, and power. Informal employment has been another major response to deficient demand for formal labor. To correct the disequilibria, a broad country agenda (macroeconomic performance, governance, trade, infrastructure, financial intermediation, transport, agricultural efficiency, etc.) must accompany a multi-part sectoral one. Box S.1: Estimating How Social Safety Net Programs Affect Household Incomes Relative To Poverty Lines Defining Poverty: "Poverty line" is defined as the overall expenditure level required to achieve an austere standard of per-capita consumption. "Extreme poverty line" is defined as the overall expenditure required to purchase only the food components for the same austere standard of per-capita consumption. Poverty and Extreme Poverty are each measured in three different ways: Poverty headcount ­ the proportion of individuals that are poor. Poverty gap ­ the average proportion by which poor households' incomes fall below the poverty line. (This measure reflects the depth of poverty.) Poverty severity ­ the average across poor households of the square of their poverty gap. (This measure is more sensitive to severe poverty.) Assessing How Public Transfers or Private Income Flows Affect Poverty Measures: A pre-transfer consumption level is estimated for each family, assuming all public transfers and/or private flows were removed and consumption fell by a fraction of the amount (usually 100 percent, but with results for alternative assumptions presented). Income distribution and poverty measures are recomputed on a "pre-transfer" basis. These pre-transfer measures and observed, post-transfer measures are compared. Differences are attributed to the safety net transfers or private income flows. For some analyses, remittances from abroad are treated as earnings of the recipient family, rather than transfer income sent from a different household abroad. (vi) Importance of Public Income Transfers and Private Income Flows. Public income transfers and private income flows are remarkably important within the structure of Albanian households' incomes. In 2002, averaged across all households, public income transfers reported in the 2002 Living Standards Measurement Survey (LSMS) represented 21.3 percent of household income. For the population as a whole, pensions represented the largest source of public income transfers (16.7 percent of household income), with cash social assistance through the income-tested ndihma ekonomike program (3.1 percent) contributing substantially less. Private income flows comprised 14.4 percent of household income. Dominating these were remittances from Albanians working abroad to relatives in Albania (12.8 percent of household income). Smaller income flows came from families and private NGOs within Albania. When compared with the shares of household income derived from nonagricultural employment (31.5 percent) or from agriculture (25.0 percent), the shares from public income transfers and private income flows are clearly substantial. iv Box S.2: Large Overall Impact of Social Safety Net on Pre-Transfer Poverty The combined impact of all public transfers and private income flows on the headcount measure of poverty for 2002 is striking, assuming here that these income flows are used fully to augment consumption and that remittances are viewed as transfers. National Estimates Ø Total transfers raised 18 of every 100 individuals out of poverty, lowering the estimated pre-transfer poverty rate of 43.9 percent to an observed post-transfer poverty rate of 25.4 percent. The number raised out of poverty represented 42 percent of those estimated to have been in pre-transfer poverty. Ø Total transfers raised 14 of every 100 individuals out of extreme poverty, lowering the estimated pre-transfer rate of extreme poverty of 19 percent to an observed post-transfer rate of extreme poverty of 4.7 percent. The number raised out of extreme poverty represented 75 percent of those estimated to have been in extreme pre-transfer poverty. Ø Using the income-gap measure of poverty, total transfers reduced the pre-transfer poverty gap from 20.1 to a post-transfer gap of 5.7. It reduced the pre-transfer extreme-poverty gap from 13.4 to a post-transfer gap of 0.8. Transfers therefore closed most of the poverty gap and virtually the entire extreme poverty gap. Sub-National Estimates Ø Public transfers have a greater impact on poverty headcount in urban areas (14.7 percent reduction) than in rural areas (9 percent), while private transfers have a greater impact in rural areas (7.3 percent) than in urban areas (5.9 percent). Ø Examining results for various regions separately shows that households in the central region benefit most from social transfers with respect to poverty alleviation. This holds true both for public transfers and private transfers and for reductions in both poverty and extreme poverty. Ø Households with children benefit less from public or private transfers than the average Albanian household, whereas households whose members are all of pensionable age benefit more than the average household. v (vii) Insights from Distinguishing Pre-Transfer and Post-Transfer Poverty. Analyzing how these income flows affect both pre-transfer and post-transfer poverty clarifies how households cope with the reality or risk of low income. Albania: Poverty Assessment discussed how income transfers from public social safety net programs affected households' poverty status after transfers. This review probes further to investigate what poverty levels would have been in the absence of public social safety net transfers, and how much impact each social safety net program had in reducing the poverty that would otherwise have been observed (Box S.1). Looking explicitly at pre-transfer poverty, as well as post-transfer poverty, reveals a heightened overall impact of transfers on poverty. Among households receiving urban pensions, for example, 54 percent were in pre-transfer poverty, but only 16 percent were in post-transfer poverty. This review extends that approach. It looks at implied pre-transfer poverty rates in the absence of public safety net transfers and/or private income flows. It then examines how much of the implied pre-transfer poverty was eliminated by the social safety net programs and private income flows, considered singly and in combination(Box S.2). Analyses in terms of pre-transfer poverty are a distinctive new contribution of this review. Further results are presented below. (viii) Impact of Public and Private Programs. Individual components of the safety net varied in their impact on poverty rates and living standards. Public social safety net programs had a greater impact on the poverty headcount (26 percent reduction) than that of private transfers (15 percent reduction). However, private transfers (particularly remittances from abroad, when viewed as transfers), had a relatively greater impact on the distribution of income than on poverty headcount itself. These transfers helped reduce the poverty gap by 45 percent and the severity by 78 percent ­ much larger impacts than those of public transfers. (ix) Impact of Individual Programs: Pensions. Public pensions had the broadest coverage, reaching 65 percent of extremely poor households and 59 percent of poor households. Given this and their average size, pensions made the most noteworthy contribution of all public transfers to the reduction of poverty (headcount reduction of 22 percent) and they also reduced the poverty gap by 29 percent. Although urban pensions were received primarily by urban households, they also assisted many rural households. Reflecting the basis on which they are awarded, however, pensions were less well targeted than remittances. Only 56 percent of pensioners receiving them were poor before transfers and only 27 percent were extremely poor. (x) Impact of Individual Programs: Remittances from Abroad. The most distinctive feature of Albanian transfer patterns, relative to those of other transition countries, is the key role of private remittances from family members abroad, mainly in the form of cash. On average, one Lek of household income out of every eight came from this source. However, among the 22 percent of households receiving such transfers, their average size was 13,600 Leke per month. That was equivalent to 47 percent of average monthly household income for the country. Ø If viewed as transfers, rather than as earnings of the recipient family, these remittances have been focused closely on low-income households. Of their recipients, 80 percent were poor before transfers and 66 percent were extremely poor. Remittances from abroad also achieved significant coverage, reaching 44 percent of extremely poor households and 29 percent of poor households. Because their large size eliminated much of the poverty gap as well, remittances from abroad were the safety net's most effective component overall. vi § Moreover, of all recipients of private transfers from abroad, 43 percent were extremely poor before transfers, another 22 percent were poor but not extremely poor, and the remaining 35 percent were non-poor. Those three groups received, respectively, 66 percent, 13 percent, and 20 percent of the total amount transferred as private transfers from abroad. So these transfers went disproportionately and in the largest amounts to extremely poor households. Ø A key feature of international remittances is that they usually link spouses or close relatives who have decided to take advantage of a family member's greater earning power abroad and who, in some cases, are functioning as if they have households split between two locations. Nevertheless, the Albanian-based partners are still vulnerable to low consumption and severe poverty if their emigrant members are unable to find and retain employment that generates significant, regular earnings; or if the earnings are spent abroad, remitted irregularly or in unpredictable amounts, or misappropriated while being repatriated to Albania; or if the members abroad form new primary relationships and create an alternative permanent life in their new country of residence. Ø If remittances from abroad are treated as earnings of the recipient household resulting from a migration-based household employment strategy, rather than as a transfer from outside the household, then analyses should include these earnings as pre-transfer income and should not deduct them as remittances when estimating pre-transfer consumption levels. For households with such remittances (and hence for the overall population), that results in higher estimates than before of pre-transfer consumption, lower estimates of all pre-transfer poverty indicators, and smaller estimates of the overall impact of transfers in reducing all indicators of pre-transfer poverty. § When remittances from abroad were treated this way, the estimated rate of pre-transfer poverty was 37.8 percent instead of 43.9 percent. Similarly, the estimated rate of pre- transfer extreme poverty was 12.4 percent instead of 19.0 percent. As post-transfer poverty rates were unaffected by the choice of perspective on remittances from abroad, the measured absolute impact of transfers in reducing rates of pre-transfer poverty and extreme poverty was reduced correspondingly. § However, under this approach, as well, most recipients of remittances from abroad were no longer in pre-transfer poverty. So only 34 percent of recipients of such remittances were poor before transfers; the remittances reached only 17 percent of pre-transfer poor households; and 19 percent of these remittances went to beneficiaries in pre-transfer poverty. (xi) Impact of Individual Programs: Ndihma ekonomike. The proportion of recipients who were in pre-transfer poverty was higher for the ndihma ekonomike program than for any other public transfer program. Otherwise, however, the ndihma ekonomike program was notable for assisting only 19 percent of poor families, impacting quite modestly the number of households in poverty (an overall reduction of only 2.3 percent), being no more likely to reach extremely poor households than poor ones, and channeling 36 percent of program funds to non-poor households ­ more than the 28 percent received by the much smaller group of extremely poor households. One qualification to these results is that participation in the ndihma ekonomike program appeared to be substantially underreported within the LSMS survey. Ø For Albania's main means- and income-tested anti-poverty program, ndihma ekonomike clearly scores poorly on targeting efficiency, coverage effectiveness, and impact on the vii incidence of poverty. Moreover, the program's coverage, benefits, and budget are simply too small to make a significant difference to the main indicators of poverty. Currently, that budget cannot finance even the full benefits to which eligible people are entitled. Ø This highlights the need: to review the budget; to maintain the new priority on improving the program's targeting; to monitor and evaluate closely the future impact of recent changes in program design and administration; and to consider whether additional initiatives are required to raise program benefit levels and expenditures (both of which have fallen in real terms), to increase program coverage, or to complement ndihma ekonomike with additional assistance for extremely poor households that is channeled through alternative program designs. Table S.1: Distribution of Beneficiaries and Transfers, by Type of Transfer Percentage of Percentage of Percentage of transfer amount Social Transfers poor who are beneficiaries who beneficiaries that goes to poor are poor households Total Transfers 83.9 67.2 55.5 Total Public 73.8 56.9 56.1 Total Private 35.8 78.8 63.5 Ndihma Ekonomike 18.6 63.6 65.2 Urban 28.4 55.1 54.4 Rural 27.6 54.8 58.0 Other 12.2 62.3 62.0 Pensions Transfers Total Pensions 58.7 56.0 55.7 Public Unemployment Benefits 1.7 52.6 59.5 Maternity Benefits 1.2 58.4 56.8 Social Care 0.5 68.7 59.7 Other Public 3.8 64.7 57.1 Within Albania 5.1 69.3 54.7 From Abroad 28.7 79.6 64.8 Remittances Transfers All Remittances 33.7 78.8 63.0 Private Other Pvt. 2.1 77.0 73.1 Note: Poverty Status is estimated in the absence of all social transfers assuming a 100% marginal propensity to consume out of transfer income. (xii) Targeting of Transfers. The transfer system is better targeted to poor families than conventional wisdom had thought but significant failings in targeting remain, as the discussion above has highlighted. In terms of their coverage, targeting, average size, and contribution to reducing the poverty gap, private remittances from abroad (if viewed as transfers) are clearly the most effective among all transfers. They are also the most effective in their coverage of the viii extremely poor, second only to private transfers from NGOs in their target incidence, and have by far the greatest impact in reducing the extreme poverty gap. Even though only 43 percent of extremely poor households receive these transfers, remittances from abroad seem to be the most efficient transfer in helping extremely poor households to close the poverty gap. Although eligibility for pensions depends mainly on age rather than income, pensions deliver substantial assistance to poor households but nevertheless reduce the efficiency of public transfers overall in targeting households in pre-transfer poverty. More importantly, one-third of income-tested ndihma ekonomike payments in 2003 went to non-poor households and the program reached only one-fifth of all poor households. (xiii) Population Coverage Through Transfers. Of all households in pre-transfer poverty, 84 percent received transfers and 16 percent did not. Of the total amount distributed in transfers, households in pre-transfer poverty received 67 percent and non-poor households received 33 percent. Of all households receiving transfers, 55 percent were poor before receiving transfers and 45 percent were not. (Table 1 provides disaggregated results.) (xiv) Safety-Net Function of Pensions in Rural Areas. Policy makers have been considering how far and fast to proceed in trying gradually to equalize contribution and benefit rates between the rural and urban pension programs for self-employed workers. One concern is that pensions might play a larger safety-net function in rural areas than ndihma ekonomike. The associated risk is that, if rural contributors cease contributing as contribution rates for rural pensions rise sharply, the safety-net function that pensions play in rural areas will be undercut significantly in future years when current rural contributors retire without a full contributory history of coverage. An analysis directed at this issue found that rural and urban pensions had more impact than ndihma ekonomike on poverty among those rural households with at least one family member of pensionable age. Sixty eight percent of individuals living in such households reported receiving rural pensions, 30 percent reported receiving urban pensions, and 8 percent reported receiving ndihma ekonomike. For these rural households, urban and rural pensions, considered jointly, reduced the poverty headcount by 16 percentage points, whereas ndihma ekonomike reduced the headcount by only 0.3 percentage points. Similarly, pensions reduced the extreme poverty headcount by 12 percentage points, whereas ndihma ekonomike reduced it by only 0.4 percentage points. Moreover, although only one-quarter of these rural households reported receiving remittances from abroad, the average size of the remittances they received was 15,385 Leke per month. As a result, remittances from abroad reduced the poverty headcount by a further 7.4 percentage points and the extreme poverty headcount by 9.1 percentage points. These findings imply a strongly cautionary note as policy makers consider how far and fast to proceed in equalizing the contribution rates of the rural and urban pension programs for self-employed workers. (xv) How Does the Social Safety Net Work? For individual Albanian families seeking to maintain acceptable living standards and lifestyles and for policy makers seeking to manage or improve the social safety net, a basic logic is evident in how the social safety net interacts with people's lives. Ø Family members' current labor incomes and pension entitlements provide the foundation for their current living standards and, initially, can be taken as given. Ø Families, however, can respond to this initial situation by exercising one or more of three related choice variables. ix § One choice is to seek additional or different employment that is more lucrative. § A second choice is for the family as a whole, or a member of the family, to migrate internally or abroad. That creates the possibility of remittances to the remaining family members if one family member migrates and succeeds in finding new employment. § A third choice is for the family to alter its composition in other ways by joining with relatives or another family (to take advantage of household economies of scale) or by requiring a family member to become economically independent (to reduce the economic demands placed on the remaining family members). § Clearly, decisions about employment, migration, and other family composition changes will often be interdependent. Ø Once decisions about these choice variables have been made, families can take advantage of other available social safety net programs (including ndihma ekonomike) to fill income gaps and augment their resources. Ø Policy makers must identify the pattern of family decisions they observe and respond to these through changes in the social safety net programs that are publicly managed, in the policies and infrastructure that affect how well migration or other household composition changes meet family and social needs, or through parallel income- and employment-enhancing changes in other policy areas. Those responses could include: § Active or intensified monitoring and evaluation of changes already being introduced in public programs within the social safety net; § New parametric changes in the structure of these programs that broaden conditions of eligibility, introduce graduated income-testing provisions, improve program administration and outreach, or reverse recent declines in real program spending and in the real value of benefit ceilings by modifying benefit levels or rules that govern these (like the link to relatively static unemployment benefit levels); § Introduction or piloting of new safety net program structures or significant new linkages between existing programs, such as making cash income transfers conditional upon school attendance by school-aged children, or exploring the potential of proxy means testing or of stronger linkages between benefit eligibility and indicators of potential consumption; or § Changes in other relevant program areas, such as upgraded financial sector infrastructure for simpler and more reliable international transmission of remittances and other funds, international agreements that make migration and overseas job search less difficult, and measures that could enhance options for income generation ­ for example, improved transportation and marketing arrangements for agricultural products that could boost agricultural incomes, or improved systems for land transfer and agricultural credit that could facilitate consolidation of agricultural land-holdings into larger and economically more efficient parcels. PRIORITY POLICY RECOMMENDATIONS (xvi) Scope for Improvement. The post-transition social safety net, as a whole, directs substantial assistance to families that would be poor or extremely poor in the absence of transfers. Nevertheless, there is scope to improve significantly the structure, administration, and x performance of Albania's social safety net and social protection programs. Fiscal, social, and governance pressures will make such changes increasingly urgent. Chapter 3 identifies key recommendations for social insurance and social safety nets. (It also presents their rationale and discusses key issues and a more detailed agenda for future action.) (xvii) Social Insurance. For social insurance, the priority is to create a sound guidance system for future development of policies and a congruent capacity-building initiative to reorganize the policy development process and to reengineer and strengthen critical implementation and administrative systems. Ø Although Albania's most recent parametric pension reforms were based on competent modeling and simulation by consultants, that capacity is no longer regularly available. Given known uncertainties, it is imperative that Albania develop and retain the capacity (including trained staff or a long-term consulting contract) to undertake skilled long-term modeling and simulation of the social insurance system. This capacity should be directed urgently (i) to reassess predicted outcomes under current policy parameters; (ii) to adopt a conservative indexation rule for pensions already awarded, in combination with any further parametric changes needed to ensure long-run fiscal balance that permits a gradual reduction in total payroll contribution rates; (iii) to plan coordinated changes in pensions policy and health care financing reform; and (iv) to achieve a transparent new configuration of social insurance contribution rates and other financing sources for the various branches of social insurance that equates the revenues and expenditures of each branch. Pending the completion of these tasks, there should be no change ­ up or down ­ in the total payroll tax contribution rate or in the allocation of this revenue stream between pension and health functions. Ø To supplement this capacity-building initiative for policy development, there should be a parallel institutional-strengthening package for the Social Insurance Institute (SII) and the Ministry of Labor and Social Affairs (MoLSA) that (i) redefines policy-making processes and responsibilities and implements the necessary changes and capacity building activities; (ii) reviews and reengineers operating procedures within SII and related responsibilities, as the basis for defining and operationalizing new processes of information management, financial management, and business transactions that strengthen existing systems and that assure fiscal and informational integrity and operational efficiency through systems that incorporate independent cross-checks. (xviii) Social Assistance: Monitoring and Evaluation. For social safety nets, Albania lacks good ways to tell what impact its social assistance programs are having and whether numerous policy and administrative changes that it is making are helping. To supplement analysis of occasional household surveys, it should establish, and make operationally effective, a monitoring and evaluation system that measures, assesses, and, where appropriate, audits: Ø the impact of the changes in national and local government responsibilities for cash social assistance and social service programs, including the ndihma ekonomike program; Ø the piloting and later introduction of work-availability requirements within this program; and Ø the accuracy and effectiveness with which individual local governments implement and administer national program rules, including those about program eligibility and allocation of funding for cash social assistance and social service programs. xi (xix) Social Assistance: Program Design Elements. Moreover, the ndihma ekonomike program is designed to assist the poorest families who lack other resources and opportunities. Of the main income transfers within Albania's social safety net, the ndihma ekonomike program is the least effective at reaching poor individuals and raising them above the poverty line. It is only modestly better than other public programs at restricting funding to those who are poor and who qualify. Declining real budgets, linkages to unemployment benefit rates that have been declining in real value, inconsistent program administration, and unclear work incentives are among the factors apparently contributing. Accordingly, MoLSA should review: (i) maximum benefit levels and real growth rates of benefits and program funding levels within cash social assistance and other social safety net programs, including linkages between parameters of these and other programs such as unemployment benefit; (ii) the appropriateness of indexing or routinely adjusting these benefit levels; and (iii) the relationship between benefit levels and small amounts of private income within these programs. To improve targeting of cash social assistance for low- income households, MoLSA should also consider alternative or complementary measures (e.g., conditional cash transfers linked to school attendance, or proxy means-testing with community assessment) as well as possibilities for piloting these. xii xiii CHAPTER I: ALBANIA'S CURRENT SOCIAL SAFETY NET: ITS LOCAL AND REGIONAL CONTEXT, STRUCTURE, AND FUNCTIONING A. Evolving Local Context 1. Overview. Albania has undergone radical political, economic and social transformation since 1989. Most fundamentally, the totalitarian political structure which directed a command economy and imposed tight social control has been replaced. Its downfall, however, unleashed pressures and processes that are still powerful interconnected forces driving continuing change. Processes of political decision-making now look very different but continue to evolve. That is equally true of the former industrial and occupational structures, the population's geographic distribution, the gender balance within Albanian households, and the organization of commercial and human services. As in other countries of Europe and Central Asia (ECA), the prerequisites for higher living standards and increased economic opportunity also brought with them new social and economic risks for individuals and families. Economic contraction during 1990-1992 slashed GDP by 40 percent (Figure 1.1). Parallel changes included an end to collective farming, privatization of agricultural land, widespread closure of industrial enterprises, shifts in the sectoral composition of output, and sharp increases in underemployment and open unemployment (Figure 1.2). The emerging shortage of employment ­ especially of good jobs in the formal sector ­ became Albania's fundamental socioeconomic problem for the next decade. Figure 1.1: Index of GDP at Factor Cost in Constant Prices, 1987-2001 (1987 = 100) 140 120 115.9 100 108.3 108.9 100.098.6 97.9 101.0 94.2 80 93.9 86.0 87.3 79.0 60 71.0 72.2 65.9 40 20 0 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Source: World Bank economic database. 1 Figure 1.2: Distribution of GDP at Factor Cost, by Sector, 1987-2001 100% Cost 21 22 23 16 19 22 23 25 22 23 19 19 20 21 23 80% Factor 24 23 22 22 24 24 25 26 27 28 at 60% 47 38 46 46 45 GDP 40% of 54 54 54 56 53 57 56 54 52 49 20% 33 31 32 37 42 Percent 0% 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 Agriculture Industry Year Services, etc Source: World Bank economic database. Not only did the shortage of formal employment lead to unemployment, major expansion of the informal economy, a consequent drop in the government's ability to collect various tax revenues (Figure 1.3), and perhaps an enlarged sector of criminal economic activity. The shortage also prompted major flows of migration, both domestically and abroad, by people looking for work. Figure 1.3: General Government Expenditures, Revenues and Deficit as Percentage of GDP, 1990- 2002 60 50 40 30 GDP of 20 10 Percent 3 0 0 0 -2 -2 -2 -1 0 -10 1990 1991 1992-111993 1994 1995 1996-61997-91998-51999 2000 2001 2002 -20 -22 -30 Year Total General Gov't Expenditures Total General Gov't Revenues General Government Deficit Source: World Bank economic database. 2 Much of this migration involved husbands or sons leaving their families to go to Albania's cities or to go abroad. So migration, in turn, increasingly altered household composition, as many households whose men went abroad became female-headed or combined with relatives' households, including through multigenerational households. (These patterns of compositional change within households merit further examination during the next poverty assessment.) The impact of economic change on employment, migration, and household composition in turn precipitated a need for new approaches to the provision of social safety net programs. 2. Employment. Prior to 1989, the political leaders in power managed a centrally planned economy and assigned people to jobs in specified geographic locations. The economy followed a Chinese model and traded with a small number of partner countries but was otherwise segregated from the rest of the world economy. Albania's economy was based heavily on industrial production in various regional centers, mining, and agricultural production on collective farms. Technology in industry and agriculture at that time is perceived by officials today as having been competitive relative to Albania's trading partners and competitors in those years. Living standards were low, as managed consumer prices helped to offset low wages for mandatory work. The onset of transition saw the termination of collective farming and the subsequent privatization of land, although into parcels that are now seen as uneconomically small. It also saw the collapse of heavy industrial production in the regional industrial complexes. In Elbasan alone, for example, about 12,000 jobs disappeared in the industrial complex that once produced iron, steel, fabricated heavy engineering products, chemicals, and cement. The smog-shrouded complex is now a wasteland, where a single foreign concessionaire employs about 100 people. Foul smoke from underground escapes even from the surrounding hills. Yet, today, Elbasan's population is larger than in 1989, although almost 20,000 people are registered as unemployed. Similarly, in Shkoder, the once-thriving plant that produced copper wire for export is idle and chained shut in the middle of a depressed and decaying area. The composition of Albania's output changed sharply, as industrial production slumped and the recently privatized agricultural sector absorbed residual, newly released labor (Figure 1.2). Following the economic contraction in 1990-1992, GDP did not pass its estimated pre-transition level until 1999, despite strong measured economic growth from 1993, which was interrupted by a further economic and social crisis in 1997 when the collapse of widely patronized pyramid investment schemes triggered a period of anarchy. During that period, pension payment (from a reserve fund) was reportedly the only government function that continued to operate. 3 Smoke haze over Elbasan's industrial zone in January 2004 Shkoder's mothballed copper wire factory, guarded by an ancient steamroller Buildings used for housing across the street from the copper wire factory 4 3. Economic Growth and Its Sources. Measured rates of economic growth since 1992 have generally been high and recent analysis attributes nearly all of this growth to more efficient use of labor. The recent Albania country economic memorandum concluded that. "economic growth in Albania during transition has been driven primarily by high rates of total factor productivity growth (i.e., by the improved allocation of resources). During the initial years of transition, the disorganization or chaos resulting from the removal of central controls and coordination produced negative total factor productivity growth rates as output fell and large parts of the capital stock were idled. Subsequently, as the economy achieved macroeconomic stability and introduced structural reforms, the reallocation of resources to more productive activities allowed the economy to generate rapid growth with low rates of investment, so that total factor productivity growth rates increased. Growth accounting reveals that during the period 1993-2003, total factor productivity growth contributed 6.14 percent of the annual average annual real GDP growth (6.27 percent)."1 4. Implications for Demand for Labor. The growth decomposition story, however, can also be interpreted in a less benign way. This is that the shift in output composition away from the industrial sector reflected a massive write-off of heavy industrial capital in cities like Elbasan and Shkoder. This ­ and probably other employment changes linked to the ending of the collective farm system ­ created or perpetuated widespread, sustained underemployment and unemployment in the country which underlie the current disequilibrium in the labor market. A gradual decrease in this underemployment has been the main source of growth in GDP, with little role for employment increases as a source of economic growth. Two key questions flow from this analysis. First, how much underemployment remains to be eliminated and, by implication, how soon might more extensive use of labor replace more intensive use of labor as the source of growth? Second, what could be sources of an increase in the number of "good" jobs within the formal sector and what are the prerequisites for generating those jobs? 5. Migration. Within this context of unemployment and underemployment, migration has served as a major response mechanism ­ one of the few over which individuals have had personal control ­ as people sought employment opportunities elsewhere. The recent poverty assessment for Albania reported: Four types of migration can be discerned since 1990. First, a comparison of the 1989 and 2001 census shows large-scale migration from rural to urban areas and from villages and small cities to bigger cities, particularly to Tirana. Second, short term international migration (for periods of days, weeks, or months) is very common, almost exclusively to Greece, particularly from those regions bordering Greece. Third, there is the most common type of long-term international migration, to Greece and Italy as well as other countries of the European Union. And finally, legal long-term international migration to the USA and Canada has started in recent times.2 6. Emigration and Remittances. Emigration has taken one-third of Albanian males aged 20­35 to both neighboring and distant countries, from which many send home significant remittances (Figure 1.4). Albania has effectively exported much of its best labor to countries 1World Bank. Albania: Sustaining Growth and Improving Competitiveness ­ A World Bank Country Economic Memorandum, World Bank: Washington D.C., 2004, p.i 2World Bank. Albania: Poverty Assessment. Washington D.C.:2003, p.122. This analyzed data from the 2002 Living Standards Measurement Survey (LSMS) and the 2001 Population and Housing Census (PHC). 5 with much higher per capita incomes where the income returns to Albanian workers are probably much greater than if they had remained in Albania. As the Country Economic Memorandum notes, Albanian growth has been fueled by remittances and an apparently large, though difficult to quantify, injection of resources from illicit trafficking in humans and drugs. Official estimates indicate that remittances are the largest source of foreign exchange, greater than the combined value of exports and foreign direct investment, currently constituting 14 percent of GDP. The amount of remittances from Albanians living abroad and recorded by the banking system is estimated to be over US$ 200 million per year. This stimulus has generated a domestic boom in non-tradable activities--principally construction and services. Since 1990, approximately one- fifth of the total population of the country has left and is living abroad, and there have been large- scale movements of population from rural to urban areas. Information from Greece and Italy, the two principal receiving countries, puts the number of Albanians who are legal residents in these countries at approximately 600,000 in 2000-2001, or one-sixth of the Albanian population.3 However, short-term and shuttle migration flows are sensitive to distortions and economic incentives in the labor markets of both Albania and neighboring countries. Consequently, the impact of remittances from Albanian emigrants (many of whom have low skills) might decline over time (with assimilation) or fluctuate pro-cyclically for low-skilled workers if labor demand falls in neighboring countries (e.g., as construction drops after the Greek Olympic Games). Figure 1.4: Worker's Remittances to Albania as a Percentage of GDP, 1995-2003 25.0 19.7 20.0 18.6 18.4 GDP of 15.0 12.3 12.8 11.9 12.3 12.8 10.0 9.5 Percentage 5.0 0.0 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year Source: World Bank economic database. 7. Internal Migration. Internal migration also displays multiple patterns that are fundamentally reshaping the distribution of Albania's population. 3World Bank, 2004. op. cit., p.i 6 First, Albania is undergoing a fundamental demographic transition, from rural to urban, and from the North Eastern Mountains to the districts of the Coast and Tirana. Second, Tirana is by far the principal destination of internal migrants, a process that appears to have accelerated in the second half of the 1990s. Third, there is evidence of a local rural-urban migration within the North Eastern region. Whether these urban centers in the north are only interludes to more distant and lucrative types of migration for these households remains unclear, although the large numbers of migrants from these provincial urban centers in the North to Tirana seems to suggest that many, indeed, move on. Given the internal migration to the urban areas of the Coast, as well as local rural-urban migration, creating economic opportunities in urban areas beyond Tirana could help in re-establishing a more sustainable rural-urban flow, as Tirana alone is struggling to accommodate and service the increasing number of migrants flocking to its poorest neighbourhoods.4 Almost half of all internal migrants go to Tirana, which has received about 160,000 people and in which two peri-urban areas ­ the poorest mini-municipalities in Tirana ­ have doubled in population since 1990. In these peri-urban communities around Tirana and other main central cities, squatter families live tenuously on land that they do not own and that lacks basic services. As the peri-urban areas lie outside the Municipality of Tirana, however, the city government is not responsible for providing services to their communities. Migration from surrounding rural areas to the formerly industrialized regional towns has left these towns with larger, but on average less skilled, populations than in their industrial heyday, despite the departure of most of the former industrial workers for Albania's larger central cities or for countries abroad. In part, this net outflow of residents has been offset by immigration from the north and west. However, it has contributed to an unbalanced and under-skilled domestic labor force, with many of the most skilled and potentially most entrepreneurial workers going abroad and others gravitating to the informal or illegal sectors of the economy. 8. Other Migration Patterns. A recent INSTAT analysis of 2001 Census data reported that, between 1989 and 1991: Of 253,700 internal migrants, 146,000 (or 58 percent) migrated to urban areas. Migrants tended to be young, with 46 percent aged 30 or under. Although 86 percent were aged 15- 64, only 60 percent worked. About 710,000 people ­ 390,000 of them men ­ emigrated during the same period. The cohort aged 20 to 40 is thought to have provided the most emigrants. Because internal migrants go disproportionately to the main departure points for emigration, one hypothesis views emigration as a continuation of prior internal movement. The mass exodus modified the age structure of the population considerably. As young adults ­ mostly men ­ emigrated, the decline in Albania's birth rate strengthened and a process of population ageing appeared to have begun. The internal movements also intensified the urbanization of the country, although local labor markets did not effectively integrate the new arrivals.5 Three other elements round out the migration picture. Albania also receives immigrants, mainly from north and east. It took in up to 450,000 refugees during the 1999 Kosovo crisis, and others 4World Bank. Albania: Poverty Assessment. op. cit. , pp.127-8. 5Albanian Institute of Statistics (INSTAT). "Migration in Albania." 2001 Population and Housing Census Research Publication Number 18. INSTAT: Tirana, Albania. 2004 7 during the Macedonia conflict in 2001. In addition, Albania is a transit country for illegal immigration from South East and East Asia and it is an important link, as well, in the trafficking of Eastern European and Albanian women for prostitution.6 9. Changing Social Safety Net Needs. These and other changes in Albanian society created a very different environment for families in economic difficulty or without strong informal social support systems. Moving towards a market-oriented economy requires acceptance that long- term increases in the flexibility, dynamism, and efficiency of resource use and the resulting increase in average living standards will nevertheless be achieved at a cost. That cost includes greater economic uncertainty, fluctuations in employment levels, market-induced restructuring of firms and retraining of workers, changes over time in relative wage rates, and a loss of apparent government control over important economic outcomes. This prospect of greater uncertainty during the transition to a market-oriented economy and thereafter has motivated the urgent development of social safety nets within transition states across eastern Europe, as well as the introduction of active and passive labor market programs (including unemployment benefits) and targeted cash benefits of last resort. But it has also discouraged enthusiasm for reform among actual and potential losers under the transition process. Prior to transition, the implicit safety net strategy had been to place a meager floor under living standards by first requiring people to work for low wages in assigned jobs at specified locations within the centrally managed economy, and then setting consumer prices sufficiently low to make an austere lifestyle possible. Once the transition to a market-oriented and democratic system began, however, employment was no longer guaranteed and prices and wages were subject to market forces. Unemployment and underemployment have reduced economic security directly. Growth in the informal economy has made employment more tenuous, and wage rates and earnings less predictable. Internal migration separates people geographically from their traditional support networks and moves people to areas where schools, health care, community services, and other amenities may be overstretched or absent. External migration takes capable adults from their families and substitutes the possibility of intermittent receipt of remittances. Not surprisingly, therefore, like other countries in the region, Albania recognized and began to address the need for changes in its social safety net systems. 10. What Constitutes a Social Safety Net for a Market Economy? Within market-oriented economies, a social safety net is a network of programs and support systems that helps people who are poor or vulnerable or falling into socioeconomic difficulties and who would otherwise have what society judges to be unacceptably low access to resources and services. The programs and support systems can be publicly or privately sponsored. Typically, an SSN has components reflecting up to three different emphases. An SSN can include elements of: (i) a net to catch people when crises overwhelm their usual support systems and their ability to meet basic needs falls to critically low levels; (ii) an austere ongoing income floor for families with inadequate resources; and (iii) risk-reducing social programs designed to reduce the likelihood that vulnerable families will experience crises that have economically or socially debilitating effects. While incorporating some or all of these elements, social safety nets: show considerable diversity in their components; can include assistance from private individuals, private organizations and government agencies; and draw heavily on a range on complementary programs and public policies. Appendix 1 explores and illustrates these dimensions of social safety nets. 6World Bank. Albania: Poverty Assessment. op. cit. , pp.123-4 8 B. Current Social Safety Net and Social Protection Programs: Their Structure and Functioning 11. Current Programs and Related Expenditures. The social safety net in Albania today consists of three major groups of public programs, plus private transfers between and within families. The public programs consist of a comprehensive social insurance system, social safety net programs, and labor market policies. These are discussed in detail below. Social insurance programs can play an important role within social safety nets, delivering substantial assistance to low-income households through programs that are not explicitly income- or asset-tested. This potential is accentuated by the scale of expenditures on social insurance, which dominate those on social assistance in most countries including Albania. Although comprehensive, consistent, multi-year time series showing spending for social purposes are not easy to construct from Albanian data, Figure 1.5 shows spending on key components of Albania's social safety net from 1994 through 2002. Several distinctive patterns are evident. One is the high level of social insurance expenditures ­ around one-quarter of GDP ­ when the new social insurance law was introduced in 1993 at the depths of the transition recession and when policy-makers were trying to assist the large share of the workforce whose traditional jobs had suddenly and permanently disappeared. This share of social insurance spending declined dramatically within a decade to more typical levels for transition countries of around 5 percent of GDP. Another pattern is the much greater scale of spending on pensions and social insurance benefits than on social assistance programs of cash assistance to families with limited incomes and assets or to people who have become disabled while young. A further pattern, reflecting expenditure management goals, is a decline of five percentage points in the GDP share of Albanian government sector spending for recurrent purposes. Figure 1.5: Government Current Spending and Selected Social Components, as Shares of GDP, 1993-2002 35 Total 30 government spending 25 Social insurance GDP expenditures of 20 Pensions and benefits 15 Percent Other social 10 Insurance Ndihma 5 ekonomike 0 Disability 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 cash benefit Year Source: Data from INSTAT, the Social Insurance Institute, the Ministry of Labor and Social Affairs, and the World Bank. 9 Social Insurance System 12. Scope of Albanian Social Insurance. Elements of pension and social insurance arrangements were in place for some Albanians by 1928, and more comprehensive arrangements with steadily growing coverage of public sector employees developed from 1947. However, Albania's current system of social insurance was established through a new Law on Social Insurance in 1993, which created a mandatory, publicly managed, contributory system with universal coverage. All economically active individuals, including those self-employed in agriculture (farmers), are required to contribute. The Law provides for comprehensive protection against income loss due to old age, disability, loss of a primary earner, death, unemployment, general sickness, occupational injury and work related illnesses, childbirth and maternity. There are approximately 20 different types of benefits provided to the insured. Different groups of economically active people are provided different types of benefits. One pension program covers employees. Another covers the self-employed, but in practice, this encompasses distinct schemes for urban and rural self-employed workers. The 1993 Law tightened eligibility criteria for retirement, gradually raising the retirement age to 60 for men and 55 for women. In 2002, as agreed during discussions for the first IDA-supported Poverty Reduction Strategy Credit, the Government initiated a further increase of 5 years in male and female retirement ages (to 65 and 60 respectively), beginning from July 2002, which are being phased in at the rate of 6 months per year. Linked to this were several additional decisions. One was to permit early retirement from ages 62 for men and 57 for women, if they have completed 40 and 35 years of contributions respectively, with a permanent reduction in their retirement benefit of 0.7 per cent for each month by which they retire before the standard retirement age. Others were to raise the ceiling on contributory income, lower payroll contribution rates by four percentage points, and strengthen mechanisms for enforcing collections compliance. 13. Pension Benefits. Pensions are of a defined-benefit type. The two-tier pension formula for employees includes a basic pension and an earnings-related part. The latter is equal to one percent of the pension calculation base for each year of insurance. The base is calculated as an average of gross earnings on which social insurance contributions have been paid for the individual for three consecutive years out of the last 10 prior to retirement. The base is adjusted annually for the changes in the average contribution paid by all contributors. So pensions for employees depend on years of contributory employment, the income level on which contributions were based, and workers' average contributions in each year. 14. Rural Pension Issues. Although the law treats farmers as self-employed and does not provide a separate benefit formula for them, in practice, the government has financed most of their "contributions" and has paid them a flat-rate old-age pension, independently of their working history or contribution base, if they have sufficient years of contributory employment. However, the thrust of recent policies has been to try to establish a more contribution-linked pension system for rural residents and, over time, to raise pension contribution and benefit rates for self-employed rural workers to levels commensurate with those for self-employed urban workers. This approach runs the risk that rural participation rates could fall and undercut the SSN role that rural pensions appear to play in rural areas, a risk of which SII and the government are aware and for which relevant evidence is presented in Chapter 2. 10 15. Scale and Financing. Social insurance is financed by: (i) payroll contributions, borne both by employers and employees (see Table 1.1); (ii) transfers from the central government budget for some specific benefits;7 (iii) deficit financing for the remaining deficit, with most of the cost of rural pensions effectively being covered by the budget, and (iv) special arrangements in relation to unemployment and maternity benefits (discussed below). The total contribution rate for the employed, including health insurance, is 41.9 percent. In 2003, social insurance provided income to 552.4 thousand beneficiaries, or 17.8 percent of the Albanian population. Expenditures under compulsory social insurance represented 5.4 percent of GDP, compared with 5.7 percent in 1994. The Social Insurance Institute (SII) reports that the deficit of the social insurance system had fallen to 0.25 percent of GDP in 2003, compared with 3.3 percent of GDP in 1993. This needs to be considered, however, in conjunction with the arrangements noted below, whereby the state budget now finances unemployment and maternity benefits, for which the Social Insurance Institute continues to collect social insurance contributions. The social insurance system faces a tension common to all transition countries of the region in wishing to maintain robust and sustainable financing while providing pension and benefit levels that support a reasonable standard of living. Thus, social insurance expenditures exceed revenues consistently and the gap must be bridged by structural or temporary transfers from the budget or other sources (Figure 1.6). At the same time, the ratio of beneficiaries to contributors is disturbingly high (Figure 1.7) ­ a key factor that prompted the 2002 decision to raise retirement ages by a further five years. Nevertheless, replacement rates, represented by the ratio of the average pension to the average wage, are lower than was once the case, although again they are at levels ­ at least for the urban pension scheme ­ that seem typical for social insurance systems in many transition countries (Figure 1.8). Moreover, the average pension is 19 percent higher in price-adjusted terms than it was in 1993 (Figure 1.9). Table 1.1: Social Insurance Contributions Rates for Employees, 2004 Insurance Branch Employer's Employee's Total contribution contribution (%) (%) Pensions 21.0 8.9 29.9 Maternity benefit 1.63 0.67 2.3 Sickness benefit 0.57 0.23 0.8 Subtotal 23.2 9.8 33.0 Unemployment 5.0 - 5.0 Work accidents & illnesses 0.5 - 0 .5 Health insurance 1.7 1.7 3.4 TOTAL 30.4 11.5 41.9 Note: Rates are expressed as a percentage of the gross wage subject to payroll tax 7Social insurance contributions for soldiers and people receiving unemp loyment compensation; certain allowances for war veterans, merit pensions, and work invalids allowance. 11 Figure 1.6: Total Real Social Insurance Expenditures, Benefit Costs, and Contributions, in Billions of 2003 Leke, 1994-2003 40 Real total 35 Lekes expenditures 30 25 2003 20 Real total benefit of 15 costs 10 Billions 5 Real total 0 contributions 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year Source: Data from Social Insurance Institute. Figure 1.7: Social Insurance Dependency Ratios, by Pension Program, 1994-2003 7 6 5 Contributor Total 4 per Urban 3 Rural 2 1 Beneficiaries 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year Source: Data from Social Insurance Institute. Figure 1.8: Replacement Rates (Average Pension: Average Wage), by Pension Program, 1993-2004 70 60 58 60 52 49 48 48 50 45 46 44 41 40 27 30 22 21 18 20 Percent 17 20 17 16 14 16 in 10 Ratio 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 Year Urban Rural Source: Data from Social Insurance Institute. 12 Figure 1.9: Cumulative Growth Rate of Real Value of Urban Pension Since 1994 40 34.4 30 27.5 Real 24.2 Growth 20 15.4 16.5 10 7.0 8.8 2.9 Cumulative 0 Percentage -2.5 1995 1996 1997 1998 1999 2000 2001 2002 2003 -10 Year Source: Social Insurance Institute 16. Administration. The social insurance system, including collection of contributions, is administered by SSI through its central, district and local offices. The SII is established by Law as an "independent, state public institution", managed by a Board composed on a tripartite basis. District branches of the SII collect social insurance contributions. They also maintain records for individual contributors and pensioners, primarily using paper systems. The social insurance is organized in five branches: pensions, sickness, maternity, work accidents and injuries, and unemployment insurance. Unemployment benefits were originally to be financed from the 6 percent payroll tax collected by SII. Instead, the benefits are now financed by the central budget and administered by the national employment offices. SII collects the unemployment insurance tax and uses the proceeds to fund other social insurance benefits, mostly urban pensions, by recording loans from the unemployment fund to the pension fund. A similar arrangement applies to maternity benefit. SII maintains a reserve fund sufficient for three months expenditures. During the pyramids crisis in 1997, the existence of this reserve fund enabled SII to continue paying pensions ­ often reported as the only government activity that continued to function during this period. 17. Collections Responsibility. In October 2002, the Government issued a decree initiating a phased transfer of responsibility for collecting social insurance contributions from SII to the General Tax Directorate of the Ministry of Finance. A schedule for the phased transfer has been agreed and is being implemented progressively. However, a decision has been made that the transfer of responsibility will not be complete and will leave SII with responsibility for collecting contributions from the very large number of self-employed farmers working in the agricultural sector. 18. Recent Pension Developments. By 2001, policy-makers had already implemented social insurance on a western European model, with distinct pension schemes for employees, urban self-employed workers, and rural self-employed workers (farmers). They had created an organization (SII) to manage this, a payroll tax base to support it, and transparent supplementary 13 budget financing through explicit, intentional cross-subsidies. Retirement ages were rising to 55 for women and 60 for men; the contributory period required for full pension eligibility had been lengthened; SII outlays were a falling share of GDP; urban replacement rates for age pensions were manageable at around 40 percent; and more self-employed farmers had been induced to contribute. 19. Pension Pressures. Nevertheless, the pension system in 2001 remained vulnerable to multiple pressures. First, the pension system's deficit was large ­ about one-fifth of the overall budgetary deficit. Pensioners outnumbered contributors ­ an extraordinary situation for a pay- as-you-go system in a demographically young country and a legacy of early retirement provisions induced by the initial unemployment shock. Most self-employed rural workers still did not contribute to the system. Nor did most urban workers employed outside government and state-run enterprises. New pensioners retiring at ages 55/60 had long life expectancies. The informal sector was large and had the potential to grow. Uncertainty about future migration and fertility patterns added to concern about the contributor base. The pension system's deficit, although declining, was about one percent of GDP. This deficit was largely financed by contributions designed originally to fund unemployment insurance benefits. Only 1 out of every 7 or 8 Leke of SIF revenues came from private-sector contributions. Finally, any sustained slow- down in previously high rates of growth would significantly increase financial pressures on the system. 20. Key Pension Issues. Accordingly, pension policymakers developed a strategy of further reform to address five key issues. The first was this deficit. Second, rural pensions were largely financed by budget subsidies and contributed to the deficit. Third, social insurance contribution rates were so high that both voluntary participation and formal employment had been discouraged. Fourth, the purchasing power of pensions remained a source of concern. Fifth, the share of the retired population covered by the pension system was projected to drop substantially in the longer term (30 years) because a large part of the current workforce did not contribute. That would create political pressure to expand coverage or assistance for retirees who did not contribute, putting further pressure on the system's finances. Thus, building on past and ongoing technical analyses, a second round of pension reforms was developed as a package, within the basic constraint of an "unfunded" or "pay-as-you-go" pension system: Number of contributors x Average contribution = Number of pensioners x Average pension. 21. Initiatives Since 2001. Albania's social protection programs, and their financing, have been significantly strengthened since 2001. An IDA Poverty Reduction Strategy Credit (PRSC) supported this process. Over a ten-year period, policy makers are now phasing in a second 5- year increase in the retirement ages for men and women, taking these to 65 and 60 years respectively. The increases are accompanied by agreed benefit reductions for certain people retiring early after long contributory histories. Policy makers have also taken measures: Ø to reduce disincentives to participation in the formal sector by lowering payroll contribution rates by four percentage points; Ø to raise the ceiling on contributory incomes; Ø to improve the efficiency of collections (including through the phased and partial transfer of responsibility for collection of social insurance contributions to tax authorities within the Ministry of Finance); 14 Ø to initiate both triennial reports by the Social Insurance Institute (SII), and periodic independent expert auditing, to international accounting standards, of SII's funds; Ø to eliminate price-compensation elements of pensions in excess of EU guidelines; Ø to improve performance in collecting pension contributions owed to SII; Ø to narrow the gap in pension levels for self-employed rural and urban workers; Ø to increase rural pension levels in real terms; Ø to pilot the automation and periodic processing of pension payments; Ø to prepare a 3-year work and investment program for improving the ongoing performance and the administrative efficiency and effectiveness of the SII; and Ø to plan indexation of pensions based on price inflation. Social Safety Net 22. Components. The social safety net in Albania includes three cash social assistance programs and a program of social care services for orphaned, disabled, and elderly people. The social assistance programs are: a targeted poverty benefit in cash (called ndihma ekonomike or economic aid); a regular monthly allowance to those with disabilities that are congenital or acquired while children or students; and price compensation paid to pensioners and their families. 23. Ndihma Ekonomike and Its Basic Provisions. This is the main income-tested cash social assistance program in Albania. The Government has decided to implement changes, described below, in its structure. The program has provided a cash benefit for categorically eligible families with little or no earned income. The program is structured through conditional grants from the national government that local governments are to administer using nationally specified eligibility and benefit rules. Larger local government areas are established as municipalities. Smaller local government areas are administered as communes. Local governments receive and process applications, including through a mandatory visit to each applicant family to check its eligibility. The final decision on whether to grant assistance is then made by municipal and communal councils, which also set the benefit amount. The amount of assistance awarded to a family is to be based on a national Government decision dating from November 1996. This set the benefit ceiling for the recipient family at 2.5 times the unemployment benefit. The base benefit (awarded to each eligible family) cannot be higher than 95 percent of the individual unemployment benefit. Subject to the overall ceiling, aged or disabled family members attract an additional amount of 95 percent of the individual unemployment benefit; people below working age attract 25 percent of the base amount; and people of working age attract 20 percent. The program identifies categorically eligible families that satisfy various selection screens designed to exclude families on the basis of family composition (e.g., no economically dependent family members, and no employable adult family member living away from the household) or of ownership of assets that imply wealth or income-generating capacity. For eligible families, the program specifies a maximum allowable benefit for families with no other income. For families with limited private income, a partial benefit may be payable. The partial benefit is reduced from the maximum level to offset all private income (including imputed income) that the family generates through home production (e.g., from land or garden plots) or that it receives from outside employment. In practice, payment of benefits is also limited by the 15 program budget available. When the program budget is inadequate to meet the total cost of the benefits for which families are eligible, local governments often classify families deemed eligible into subgroups that will receive different proportions (e.g., 100 percent, 75 percent, or 50 percent) of the benefit amounts awarded. 24. Prior Empirical Analysis. A 1998 assessment, based on 1996 data, found that 36 percent of all program expenditures went to families whose per capita incomes placed them in the lowest decile, and three-quarters went to families in the bottom four deciles. However, half of all families in the lowest decile received no assistance; the allocation of central government funds to local governments was less well targeted than the local governments' allocation of funds to families; and local governments that received relatively higher levels of funding used this primarily to extend coverage to less needy families rather than to raise benefits to the poorest families.8 Subsequent anecdotal accounts suggested that targeting by local governments has deteriorated since 1996. 25. Recent Social Safety Net Concerns. By 2001, policymakers had identified six concerns about SSN programs. One was that some local governments targeted ineffectively the ndihma ekonomike funds for income-tested cash social assistance that the national government provides to them through a conditional grant. Another was whether the distribution of these funds conformed with nationally legislated eligibility criteria in this core, but decentralized, anti- poverty program. A third concern was that the benefit structure created an incentive for recipients to conceal resources and income, making accurate administration difficult, as well as creating a disincentive for them to increase their income. A further worry was that the benefits were inadequate for families with no other resources. In addition, caseload growth rates suggested that the administration of the determination of eligibility for disability assistance required tighter procedures. Finally, the system of social services was underdeveloped and, partly for that reason, poorly coordinated with cash assistance. 26. Targeting Concerns. The first two of these concerns relate to poor targeting or inappropriate allocation of funds for the ndihma ekonomike program. One indirect source of relevant evidence is to see whether the average monthly benefit amount per person under the program varies systematically with the incidence of poverty across geographic districts, when these districts are ordered from most poor to least poor and are then grouped into five quintiles. As Figure 1.10 shows, there is a systematic relationship of the expected kind, with average benefits being higher in poorer districts. This finding does not preclude instances of inappropriate allocation of program funds, but it does suggest that the problem is not pervasive and systematic. 27. Related Developments. Beginning in 2001, the Ministry of Labor and Social Affairs (MoLSA) systematically reviewed: (a) the geographic distribution of social assistance spending, relative to that for poverty; (b) the continuing eligibility of all ndihma ekonomike beneficiaries; and (c) administrative processes for these programs. MoLSA also began to examine the potential feasibility and benefits of conditioning income-tested cash assistance on the availability of household heads for work at specified times. Progress was quite limited, however, in 8Harold Alderman. Social Assistance In Albania: Decentralization and Targeted Transfers. Living Standards Measurement Study Working Paper Number 134. World Bank: Washington D.C. July 1998. 16 implementing the IDA-supported Social Services Delivery Project, which aimed to develop a policy framework and sustainable financing system for restructuring the provision of social services and to expand their availability within this framework. MoLSA recently translated this groundwork into five initiatives, undertaken in conjunction with the Ministry of Local Government and Decentralization. Ø A new framework Law on Social Assistance and Services redefines program structures and the roles of implementing bodies at national and local levels. Ø The ndihma ekonomike program of income-tested cash social assistance is piloting a new provision making receipt of benefits conditional on availability for local work. It also offers wages that are higher than the current benefit rates for participants who accept work through the revised program. Ø The program of cash assistance for those with disabilities before adulthood will introduce a new national level of medical review for compliance of local authorities' assessments with program guidelines on claimant's disability levels. Ø Funding of social services will occur through an extension of current grant arrangements to local governments, which will then be responsible for contracting local service providers and for financing service provision. Ø Energy subsidies have been introduced for defined categories of eligible low-income households. MoLSA will prepare a nonduplicative list of eligible households that purchase electricity from KESH, the national supplier. The households will be eligible for a subsidy on electricity consumption below a threshold level. KESH will bill eligible households at the discounted rate and the Ministry of Finance will reimburse KESH for the aggregate cost of the discounts. The scheme is designed to comply with a budgetary ceiling on subsidies. Figure 1.10. Average Ndihme Ekonomike Spending per Capita per Month, by Quintiles of Districts for Poverty Incidence 300 250 (Lekes) 200 NE 150 Average 100 50 spending/capita/month 0 Richest Fourth Middle Second Poorest Quintiles of districts Source: World Bank staff calculations based on data from Ministry of Labor and Social Affairs, INSTAT, and LSMS 28. Current Changes in Program Structure. The Government is legislating changes in the structure of this cash social assistance program and, concurrently, is transferring to local 17 governments additional responsibility for program administration as part of its decentralization reforms. To implement the planned change in program structure, the Government has begun to pilot a requirement, for a subset of social assistance recipients, that they must be available to work for the local community on a number of specific days each month designated by program administrators. Otherwise, they would lose their eligibility for benefits. One reason for this change is to identify recipients, or their family members, who are already working but who are concealing their employment from program administrators. Households concealing employment would become ineligible for assistance. A second intention is to use a supplementary allocation of State budget funding to finance a wage rate for this community work that would leave social assistance recipients who undertake the community work with more income and assistance in total than they currently receive through benefits alone. Currently, budgeted social assistance funds are insufficient to finance the full benefit entitlement of all people found to be eligible and, as a result, some or all of those eligible have their actual benefit amount reduced to (say) 50 percent or 75 percent of the calculated amount for which they are eligible. Under the proposed changes, program funds released if some people lose eligibility would be reallocated to increase the social assistance benefits for those people who are deemed eligible but who are currently being paid at reduced amounts. 29. Assessment of Initial Impact on Program Coverage and Benefits. The relevant Ministries are planning information systems they could put in place to assess the impact of the changes in program structure and to monitor the ongoing performance of the program in each local government area. Policymakers have planned for some time to assess the impact of adopting provisions requiring availability for work by monitoring two variables: the number of recipients of cash social assistance benefits (which should fall); and the average benefit payment per recipient who remains eligible (which should rise). These indicators will measure key targeting outcomes. 30. Monitoring Other Issues. Even if the proposed work-availability requirements have these intended effects, however, they might also have other effects that are of policy interest because they are viewed as desirable or undesirable. On the positive side, for example, they might: Ø Lead to ongoing employment, and higher combined earnings and benefits, for some families currently receiving only cash transfers; Ø Lead to a more positive self-image for household members who have become newly employed; Ø Create or rehabilitate community assets through their community-sponsored employment; or Ø Provide services to other people in the community. On the negative side, for example, they might: Ø End eligibility for, and in this way, disadvantage families with young children for whom alternative child care arrangements are unavailable, or families where members are sick or disabled and thus either unable to work or in need of adult care, or families lacking transport to proposed work sites; 18 Ø Be applied disproportionately by local administrators to families from groups that are minorities or are subject to social exclusion, possibly as a way to encourage them to move elsewhere; Ø Be used to substitute work by program participants (which is financed from Republican budget funds) for work previously undertaken by local government officials (which is financed from local government budget funds); or Ø Lead to costs for program administration and for materials required as inputs for work projects that become large relative to participants' earnings. Ultimately, that might result in eligibility rules for the social assistance program being tightened so that available administrative budgets for the community work program will not be exceeded. 31. Relevance of National Criteria. Attention to these possibilities will be of greater importance to the extent that there are not clear, nationally specified criteria (a) for which subset of recipients will be subject to the new requirements and (b) for what kinds of work, working conditions, and hours of work could be required by local government authorities. If such criteria either are absent or are not followed consistently across the country, the introduction of the new social assistance provisions could lead to selective patterns of implementation, or to differences between local government areas in the approach adopted, that could be viewed as unfair or discriminatory. Accordingly, additional monitoring indicators could be used to enable these or other significant patterns in program outcomes to be detected and addressed. In addition, if national criteria are not specified for who is required to be available for work and for the kinds of work, working conditions, and hours of work considered acceptable, then each local government might be required to specify what criteria it is adopting on these matters and to advise the national Ministry of Labor and Social Affairs (MoLSA) about the criteria adopted. In that case, MoLSA could also analyze variability across local government areas in the criteria used and should assess the implications of this variability. 32. Administrative Aspects. Key characteristics of the ndihma ekonomike program are that it is totally financed by the state budget; management is shared between central government (MoLSA) and local government (municipalities and communes); eligibility is not subject to a time limit, provided families remain eligible for support; and the target of the program is the family rather than the individual. To administer the new program in 1993, MoLSA created an Ndihma Ekonomike department. In 1995, to improve administration, MoLSA merged this department and the department for people with disabilities into the General Administration of Assistance and Social Services (GAASS), which operated as a General Directorate reporting to MoLSA. In 2002, GAASS was named "State Social Services" (SSS) and became an independent self-managing body, which is an executive institution for MoLSA and which is administered and directed from the Administrative Council and the General Director of SSS. It has a central directorate and regional units in all 22 centers of the regions (prefectures). Its responsibilities are: to implement government policies in the field of social assistance and social care services; to administer, allocate and monitor the use of state budget funds for ndihma ekonomike, the disability benefits, and social care services; to administer the residential institutions of social care; to formulate service standards and proposals for new type of services; and to coordinate the other institutions that deliver public and non-public social care. The regional units of SSS monitor the implementation of the ndihma ekonomike program, the delivery of social care services from the local government units, and the quality of the services 19 offered by public and non-public institutions. They also develop partnership relations with civil society representatives. Management responsibility for the ndihma ekonomike program is divided between central government (MoLSA and SSS) and local governments units. The latter use official records to evaluate qualitative and quantitative indicators of poverty (population, employment and unemployment, property ownership, and adequacy of agricultural land ownership) and, on this basis, to assess regional needs for future funds. Based on those assessment reports, SSS programs and allocates the block-funds for each local unit. 33. Assessment of Pilot Program. The government formulated a decree of November 27, 2003, on piloting the ndihma ekonomike NE bloc grant transferred to local governments. The pilot was implemented in 10 communes during May-September, 2004. The main areas of work involved road infrastructure, environment, and delivery of social services (such as support to elderly people without support, orphans, etc). Project preparation is the responsibility of the local government, while implementation can be done by NGOs, private entities, state institutions, municipalities, and so on. Experience in the pilots revealed mostly part-time involvement of beneficiaries in community works. Overall, the results have been viewed as positive, by creating new employment possibilities, improving transparency in use of allocated funds, strengthening monitoring systems, increasing the income level of targeted beneficiaries, and also influencing or improving infrastructure and social services. Figure 1.11: Estimated Real Annual Expenditure on Ndihma Ekonomike in Billions of 2003 Leke 8.00 7.38 6.74 7.00 6.24 5.49 Lekes 6.00 5.24 4.77 5.00 2003 4.00 of 3.00 2.00 Billions1.00 0.00 1998 1999 2000 2001 2002 2003 Year Source: Ministry of Labor and Social Affairs. 34. Implications of Patterns in Program Functioning. Important patterns in program functioning raise other issues that policy-makers might wish to consider. In particular, real program outlays for the ndihma ekonomike program (that is, outlays adjusted for price changes over time) are declining (Figure 1.11). Similarly, the number of households assisted and the average real benefit provided to participating households are also declining (Figures 1.12-1.13). Thus, the program has been shrinking as a source of assistance to poor households. To some degree, the decline in the number of people assisted reflects a policy objective to target 20 assistance more accurately only to those people who are strictly eligible. The program changes outlined in paragraph 23 are also designed to assist with that objective, as well as with the objective of increasing the level of assistance for households in the program. Nevertheless, shrinking real benefit amounts per recipient household suggest that there could be other structural aspects of the program that warrant reexamination. For example, program benefits are constrained by the level of unemployment benefits, with a view to maintaining appropriate work incentives. However, the real value of unemployment benefit has also been declining over time (Figure 1.14). This suggests that policymakers could raise assistance provided through the ndihma ekonomike program either by raising ­ or at least maintaining ­ the real level of unemployment benefits. Alternatively, they could eliminate or modify the link between the levels of unemployment benefit and the maximum level of ndihma ekonomike. In this context, it is relevant that the maximum household level of ndihma ekonomike is 2.5 times the maximum unemployment benefit for an individual. Figure 1.12: Number of Households Receiving Ndihma Ekonomik e Benefit, 1996-2003 (in thousands) 210 149 150 150 150 141 141 135 140 123 thousands) (in 70 Households 0 1996 1997 1998 1999 2000 2001 2002 2003 Year Source: Ministry of Labor and Social Affairs. 35. However, program data show that most families receiving ndihma ekonomike have from three to five family members (Figure 1.15). Thus, the link between the levels of unemployment benefit and the maximum level of ndihma ekonomike could be modified ­ for example, by raising the maximum household level of ndihma ekonomike to, say, four times the maximum unemployment benefit for an individual without creating either equity or incentive concerns relating to unemployment benefit. The relevance of these and related options is given added weight by analysis in Chapter II. This indicates that the ndihma ekonomike program reaches only one family in five that would be poor in the absence of transfers. Moreover, the program has a comparatively small impact on the scale and depth of poverty. These finding arise notwithstanding that two-thirds of the program's recipients would be poor in the absence of all public and private transfers. 21 Figure 1.13: Average Monthly Ndihma Ekonomike Benefit in Real 2003 Leke, 1998-2003 4000 3464 3500 3176 Month 2974 3000 2754 2714 2682 per 2500 2000 Lekes 1500 1000 2003 500 Real 0 1998 1999 2000 2001 2002 2003 Year Source: Ministry of Labor and Social Affairs. Figure 1.14: Monthly Unemployment Benefit Amount in Nominal Leke and Real 2003 Leke, 1993- 2001 5000 4501 4696 Lekes 4500 3974 4155 4000 3643 2003 3415 3500 3208 3070 29433100 Unemployment Real 3000 2500 2500 2500 2500 2500 Benefit Nominal or Month 2500 2200 Monthly Amount 1920 per Person per 2000 1500 1237 Lekes 1000 Unemployment 500 Benefit in Real 0 Nominal 2003 Lekes 1993 1994 1995 1996 1997 1998 1999 2000 2001 Year Source: INSTAT. 36. Work Incentives. Work incentives are a crucial issue in social assistance policy. They depend on the proportion of any additional earnings that a worker can keep without an offsetting reduction in the benefit (or, equivalently, on the rate at which benefits are withdrawn when private earnings increase). They also depend on the worker's overall resources and consequent ability to "afford" to engage instead in non-work alternatives including home production and leisure. Benefit adequacy, work incentives, and program affordability are inherently competing objectives of cash transfer programs that generate a fundamental and unavoidable tradeoff. So 22 policymakers must strike an optimal balance among them. High levels of assistance to the poorest groups, strong incentives for them to work, and low overall expenditures on social assistance conflict because they involve three policy variables--the maximum benefit, the effective marginal tax rate, and the ceiling income level for benefit eligibility--but only two of these can be set independently. Unfortunately, raising the basic benefit amount for someone without other income will also raise total program spending, unless higher spending is offset by faster withdrawal of benefits as private income increases. But that would worsen work incentives! More generally, policy changes to achieve any of the three goals more completely would worsen attainment of at least one of the other two. Figure 1.15: Distribution of Households Receiving Ndihma Ekonomike Benefits, by Household Size, 1994-2001 of over 6 100% members 90% 80% 6 members 70% Distribution 60% 5 members 50% Households 40% 4 members 30% Cumulative 20% 3 members 10% 0% 2 members 1,994 1,995 1,996 1,997 1,998 1,999 2,000 2,001 1 member Year Source: INSTAT. 37. Lack of Clarity on Work Incentives in Ndihma Ekonomike Program. Rules and practice for determining social assistance benefit levels for families with small amounts of income should be clear. It is currently unclear how the ndihma ekonomike benefit is affected if an otherwise- eligible family has a small amount of actual or imputed income that is less than the relevant ndihma ekonomike benefit. In these cases, is the family excluded from eligibility altogether, is the benefit reduced by the full amount of the income earned, or is the benefit reduced by only a fraction of the income received (so as to create an incentive for recipients to seek additional income)? Program administrators appear to intend a Lek-for-Lek offset. However, it is possible that each approach might happen in different areas. If a family in these circumstances is deemed ineligible for any ndihma ekonomike benefit, then that family would also appear to be unable to benefit under the work-testing provisions that assist families qualifying for ndihma ekonomike to generate earned income that is greater than the ndihma ekonomike alone would have been. There are important equity and incentives issues raised by this issue. 38. Relevance of Work Incentives in Ndihma Ekonomike Program. Do families receiving ndihma ekonomike face appropriate incentives to work? Balancing the desire to provide adequate income support with a concern to maintain adequate work incentives is a challenge for 23 all social assistance programs. Providing public benefits reduces the direct pressure on families to pursue all available sources of private income or assistance. Moreover, denying benefits if any earnings are received, or reducing benefits to fully offset increases in earnings, can also reduce correspondingly the incentive to work. Albania has defined a partial benefit ceiling under the ndihma ekonomike program for low-income families that have land or other potentially productive resources. This is equivalent to imputing to such families income from those resources. Families migrating to peri-urban areas near the cities are affected by this provision. As noted above, a further possible response to this inherent dilemma would be to reduce benefits by only a proportion of any increase in earnings. Such a reduction would impose an implicit tax rate on earnings up to the ceiling level of income for receipt of any benefits. (Lithuania used this approach in the early 1990s; and the US foodstamps program also does so.) It is currently unclear whether income is offset fully against benefits or whether any income results in exclusion from the program. Both might happen in different areas. Other countries have responded by imposing work requirements on some or all recipients as a condition for receiving benefits, as Albania is planning to do. (Highly visible US welfare reforms in 1996 combined this approach with job training, subsidies for child and dependent care, retention of health insurance assistance benefits, and a ceiling on duration of benefits.). Figure 1.16: Proportions of Households in Poverty and Receiving Ndihma Ekonomike , by Prefecture, 2002 45.0 40.0 35.0 30.0 Household Households Poverty Rate 25.0 of (LSMS 20.0 2002) 15.0 10.0 Families Percentage 5.0 Receiving 0.0 Ndihma Ekonomike Tirana Berat Fier Korce Diber Vlore Elbasan Kukes Lezhe Shkoder Durres Gjirokaster Prefecture Source: Ministry of Labor and Social Affairs. 39. Targeting Accuracy Within the Ndihma Ekonomike Program. For several years, there have been anecdotal accounts suggesting that some local government areas have not followed the nationally prescribed rules when allocating ndihma ekonomike program funds. Analyses reported in Chapter II imply that about one-third of the funds distributed through the program are received by families that would not have been in poverty in the absence of transfers. Does this reflect cases where administrators tried to apply the program rules but did so inaccurately while 24 undertaking a challenging task? Conversely, does it, instead, represent a subset of local governments that systematically used allocation criteria differing from the national program rules? Evaluating the ndihma ekonomike program requires knowing what proportion of poor families are ineligible and, more particularly, whether those poor families that do not receive benefits are indeed ineligible. If so, does the exclusion of many poor families result legitimately from the national program criteria, or illegitimately from differences between national legislative criteria and local governments' de facto administrative criteria? A related factual question is whether families in peri-urban areas get ndihma ekonomike benefits or not. This is unclear, as reports on the status of provision of cash social assistance there from residents and program administrators are not consistent. Alternatively, is the problem underlying non-receipt of ndihma ekonomike by many poor families simply that the program's scale is inadequate? In that case, would raising the program's budget substantially increase its coverage of poor households ­ especially those that are extremely poor? Simple comparisons of the distribution by prefecture of poverty and of the proportion of households receiving ndihma ekonomike suggest that there is a moderately close correlation, with a few prefectures not following the pattern (Figure 1.16). Figure 1.17: Real Annual Program Expenditure on Disability Benefits and Services, 1996-2003 Lekes 5.000 3.844 4.000 3.498 2003 3.000 2.594 2.644 2.075 2.000 1.729 Constant 0.971 1.088 of 1.000 0.000 Billions 1996 1997 1998 1999 2000 2001 2002 2003 Year Source: Ministry of Labor and Social Affairs. 40. Regular Monthly Allowance for People Disabled Since Childhood. The regular monthly allowance in cash for people disabled since childhood is granted to a person certified as disabled prior to 21 years of age (24, if attending university). It is a flat rate benefit, amounting to 70 percent of the minimum wage. The amount of allowance is doubled if the disabled person attends secondary school, and tripled if he or she attends university. Since 1995, the benefit has been administered in broadly the same way as the ndihma ekonomike program: at the local community level and with the central and regional GAASS offices performing general monitoring and supervision functions. Under a recent Government decision to be submitted for legislative enactment, this program will introduce a new level of medical review at the national level for compliance of local authorities' assessments with program guidelines on claimant's disability levels. 25 41. Program Growth. The number of beneficiaries under this program has grown by a factor of 2.5 within the seven-year period from 1996 to 2003, and real (price-adjusted) program expenditures quadrupled over the same period (Figures 1.17 and 1.18). Although real expenditure per beneficiary has also grown over this period, its level reached a plateau from 1998 through 2001, before resuming its increase in 2002 and 2003 (Figure 1.19). It should be recognized, however, that good independent data are not available on the number of people who might be eligible. Figure 1.18: Recipients of Disability Cash and Service Benefits, 1996-2003 60000 50000 40000 Recipients 30000 of 20000 10000 Number 0 1996 1997 1998 1999 2000 2001 2002 2003 Year Source: Ministry of Labor and Social Affairs. Figure 1.19: Real Annual Program Expenditure Per Beneficiary for Disability Benefits and Services, 1996-2003 80.0 71.3 74.2 2003 70.0 58.4 60.0 56.1 50.9 53.6 55.2 Real 50.0 47.3 of 40.0 Lekes 30.0 20.0 10.0 Thousands 0.0 1996 1997 1998 1999 2000 2001 2002 2003 Year Source: Ministry of Labor and Social Affairs 26 42. Situation Concerning People with Disabilities. Statistical coverage of Albanian people with disabilities is incomplete. Available data comprise the number of people with disabilities who are evaluated and certified by state disability assessment commissions. The total number of people defined as disabled is derived from the state budget funds allocated for them (Table 1.2). These funds cover state allowances for people with disabilities, and several support services provided to them in the community and at residential centers. Gender-disaggregated data are needed but not yet available. Statistics on the employment rate of people with disabilities are also problematic. Their employment rate is very low and currently only 2,275 out of 9,533 who are partially able to work found employment people. In most cases, this work is of a secondary and unqualified nature. Growth in disability allowances was concentrated in urban areas, suggesting that rural areas might have had less access to social security programs. Informed observers generally agree that the number of people in Albania with disabilities is significantly underestimated. Table 1.2: People with Disabilities, 2004, by Category Category Total Urban Rural Caretakers 1. People with Congenital or subsequent disabilities under 21 44,794 19,669 25,125 7,288 a. Mental, physical , senses 35,547 15,166 20,381 3,593 b. Blind 8,621 4,197 4,424 3,142 c. Paraplegic or Quadraplegic 626 306 320 553 2. Disabled 30,250 26,274 3,976 1,500 a. Work-related 29,676 25,904 3,772 1,500 b. War-related 574 370 204 Total 75,044 45,934 29,101 8,788 43. Price Compensation to Pensioners and Their Families. Compensation to pensioners and their dependents for bread, electricity, and food price increases was introduced in conjunction with the elimination of price controls and subsidies, to cushion the impact and help maintain the consensus in favor of economic reform.9 Since 1994, the price compensation program has either been discontinued or consolidated into wages (for government sector employees) for all eligible categories, except for pensioners. In 2001, the Government reviewed the continued relevance of price compensation for pensioners. As a result, it limited the category of pensioners eligible for 9Food price compensation was introduced in conjunction with the August 1992 liberalization of most consumer goods prices, including all food other than bread; bread price compensation was introduced in July 1993, when bread prices were raised to cost-covering levels; electricity compensation was introduced in April 1994, with a sharp increase in electricity price. In 2000, the monthly amount of electricity compensation was Lek 350 per dwelling, while bread compensation was Lek 150 per pensioner per month. The amount of compensation given to pensioners' dependents ­ that is, each family member supported by the pensioner, irrespective of age ­ was Lek 414 per month. According to the Social Insurance Institute there were 0.6 dependants per pensioner. Rural pensioners were entitled only to electricity compensation. 27 these compensation payments to those with dependents who are unemployed, children up to age 18, or students up to age 25 who are not in employment--categories chosen because of their consistency with the European Convention on Social Security. This measure, effective from January 1, 2002, was expected to reduce pension outlays in 2002 by an estimated 250 million Leke. Price compensation is financed from the central budget. It is administered by the Social Insurance Institute, which identifies beneficiaries (pensioners and their dependents) and pays the benefit along with the pension payment. For most of the pensioners, the compensation is part of their regular pension payment. 44. Social Care Services. Albania has a system of residential social care services for orphaned, disabled, and aged people in need of assistance. It is administered by the General Administration of Assistance on Social Services (GAASS). Albania's Social Services Development Project, discussed below, is supported by an IDA credit and is seeking to strengthen, broaden, and restructure the current system of services. A recent new framework law on social assistance and services, together with a related Ministerial policy paper on social assistance and decentralization, constitute a significant policy initiative by Albanian authorities to achieve a more systematic social safety net framework for economic assistance and social services. The policies clarify the responsibilities of each level of government within the system of social assistance and services and could also promote more diversified, balanced, and supportive social services. The policy intends social services to be given priority over assistance in cash. It specifies that, although local governments will have a complementary role in funding services, responsibility for funding the social services will lie with the central government, which will provide for the sustainability of the services. Under full funding of social care and services at the level of assessed needs, the division of roles would appear to be that the national government defines the overall policies; specifies the eligibility conditions for cash assistance; identifies the scale of funding needed to meet local governments' determination of funding requirements within their areas for cash social assistance, disability benefits, and social services; provides that funding, subject to budget constraints; and assesses the performance of local governments. Local governments would then play the main role in delivering each of these benefits, assembling administrative and monitoring data needed for an assessment of program performance, identifying new service providers, and beginning the same cycle for the next year. 45. Social Services Development Project. Social services represent a potentially valuable resource that can assist families with members who are frail or aged, or who have disabilities or a variety of special needs. The Government is engaged in an IDA-supported Social Services Development Project that aims to improve the living standards of poor and vulnerable population groups by: (i) increasing their access to well targeted and effective social care services; (ii) supporting development, monitoring and evaluation of a more effective social policy; and (iii) improving capacity for planning, managing and delivering social care services with increased involvement of local governments, communities and civil society. DFID has been providing related technical assistance. A new framework law for social assistance and social care services was approved in June 2004. This will complement a new social care services strategy and development of a regulatory framework and of social care services standards. The State Social Services (SSS) agency would have responsibility for quality assurance and for monitoring the quality and relevance of the services provided. The regional SSS offices of four regions (Durres, Shkoder, Tirana and Vlore) are involved in a pilot program to institutionalize regional 28 community care planning committees. Although progress in delivering new services has been slow, funding for new services had been included within the medium term expenditure framework. SSS also has responsibility for running a series of state-sponsored residential care facilities for children and adults requiring care because they are orphans or have disabilities or are unable to care for themselves. 46. Active Labor Market Programs. The National Employment Service (NES) has responsibility for: (i) a national network of labor market offices (currently limited to urban areas); (ii) the National Labor Inspectorate, which seeks primarily to enforce work safety standards and to require employers of informal labor to contribute to the Social Insurance Fund; (iii) passive labor market programs (administration of unemployment insurance benefits, as well as income support for defined categories of people made unemployed by mass privatization of state-owned enterprises); (iv) active labor market programs including (a) the Employment Promotion Program, which offers either a year of temporary employment or hands-on training to unemployed job-seekers, as well as support for creation of relevant institutions, and (b) the Public Works Program until its funding was terminated in 2000; and (v) professional training. These programs were developed in part through the Albanian Labor Market Development Project, which received World Bank support. One legacy of this project is a valuable capacity to assemble statistical data for the formal portion of the labor market. However, current experience indicates that, when job vacancies occur, they are nearly all filled directly by interactions between employers and job-seekers without use of the NES as an intermediary and that the NES is usually not an active participant in this process. Nevertheless, thousands of people register with the NES as unemployed and looking for work. They do so partly because being registered in this way is a prerequisite for one's family to be eligible for ndihma ekonomike benefits if it meets all other eligibility conditions. Thus, the NES plays a major role in the administration of the ndihma ekonomike program, generating associated costs and benefits. An NES strategy was approved through a tripartite agreement among the Government and employer and employee representatives. The strategy aimed to reform the NES at all levels and it proposed creation of an Employment Fund within the framework of the Social Insurance Institute. So far, the Government has not accepted proposals for an Employment Fund. Its rationale for rejecting this proposal has been that specific proposals for employment-related initiatives should have to compete with all other proposals for new government spending within the standard framework of annual budget processes. Such proposals, therefore, should not have access to a separate, privileged, and less contested funding source. C. Regional Context: Social Safety Net Experience from Eastern Europe and Other Comparator Countries10 47. The Challenge. The collapse of centrally planned economies led to falls in production and income of up to 60 percent in the new transition countries of the ECA region. As government revenue promptly plummeted, the old social protection strategy ­ guaranteed public employment, heavily subsidized consumer prices, and generous social benefits for pensioners and socially dependent groups ­ became unsustainable. New programs were clearly needed to 10 This section is based on V. Staines and D. Isamiddinova, "Social Assistance in Transition Countries of Europe and Central Asia." Spectrum. The World Bank: Washington D.C., Summer 2004, pp. 21-25. 29 assist burgeoning groups of poor and vulnerable people, although an adequate resource base for these programs was not apparent. Moreover, popular expectations for living standards had fallen more slowly than production. Economically sustainable reform options then seemed politically unattractive when compared with what was promised, but not deliverable, under past policies. One key element of the policy response was the introduction of new social assistance programs. 48. Trends and Issues. A wide variety of basic models for providing social assistance has been used, singly or in combination, within ECA countries. In general, their relative importance has changed over time and also varies across countries and between ECA's European and Eurasian sub-regions. As a social protection strategy, the combination of guaranteed public employment, widespread price subsidies, and generous earnings-related pensions gave way to alternatives more quickly in the European sub-region than the Eurasian sub-region. Children's benefits were a popular alternative strategy early on. They gradually lost prominence, however, to approaches that assist individuals or families based on their personal or household characteristics ­ particularly, measures, indicators or judgments about their levels of income, assets, or consumption. Subsequent experience has highlighted the difficulty of measuring or estimating family income accurately in transition countries with limited administrative capacity, especially when the informal sector is very large. Families have incentives to misreport their circumstances; and administrative systems for verifying this information easily are limited. Another major challenge, it is now clear, is to find an acceptable balance between including too many families and including too few. 49. Children's Benefits. European transition countries often substituted universal payments for children as an initial step but found, by the mid-1990s, that their sharply constrained budgets could not finance adequate payments of this kind for all families (Social Protection Team, 2000). More narrowly targeted approaches were then adopted by many countries, which experienced significant challenges in implementing them successfully. For example, in 1993, Turkmenistan introduced an income-tested family benefit to accompany scaling back of its universal children's benefit for nonworking mothers of children aged under seven. However, the income test ended eligibility abruptly at a threshold level of per capita income. That created inequities between families just above and below the threshold. It also created incentives for families to reduce or hold down their incomes, or at least to report that they had done so. Such incentives imply "poverty traps," where people's attempts to improve their economic position by earning more actually make them worse off.11 50. Income-tested, Asset-tested, or Means-tested Payments. Some countries, beginning with Estonia in 1990, adopted a "guaranteed minimum income" (GMI) approach. This paid families the difference between their actual incomes and a low family income level that allowed only an austere living standard. However, this approach also created, below that income threshold, an effective marginal rate of tax (MRT) of 100 percent that served as a disincentive to work. Lithuania, an innovative reformer, introduced instead a means-tested social benefit structured as a negative income tax (NIT) in 1993. Initially, this paid half of the difference between a state- 11These problems do not arise because benefits are made to depend on an income test. Rather, they arise because the size of the benefit is reduced too quickly as the private income of the benefit recipient ris es. The solution is to restructure the income test so that benefits vary more slowly than incomes and so that increases in income lead to benefit reductions that are smaller than the income changes that caused them. 30 supported family income level and families actual earnings. It thereby reduced the MRT to 50 percent but, in consequence, also raised the income ceiling for benefits to twice the state- supported family income level. The broader eligibility was affordable only if the supported income level was lower than that affordable under a GMI design. Over time, to make possible a larger state-supported income level, Lithuania raised the MRT to 90 percent, where it remains. (Lithuania also addressed fairness and cost issues by imputing income to household plots of land.) In 1995, Kyrgyz Republic became the first Eurasian country to adopt such a benefit. Although legislated to look like an austere GMI program for the poorest households, the Unified Monthly Benefit program is administered by making payments only in respect of children and dependents. This has made the benefit function like an NIT program with an MRT that varies with the proportion of adult family members. Albania also has an income-tested benefit for eligible poor families. Although three-quarters of Albanian families receiving this benefit are poor, three-quarters of poor Albanian families do not receive it. The Kyrgyz experience also reflects a similar pattern of benefits that go predominantly to poor households but that nevertheless miss most poor families. In both countries, however, some poor families not receiving income-tested benefits are recipients of pensions or other forms of assistance. 51. Means-Testing Using Proxy Indicators. An alternative approach, which some countries have adopted, predicts which families are likely to be poor using proxy indicators that supplement or replace measured income. In the simplest form of this approach, statistical analysis identifies household characteristics and assets that correlate strongly with consumption levels reported in household surveys. Those indicators are used to select who will be assisted by the program and to what extent. The estimated underlying statistical relationships capture averages rather than special circumstances; and information on the proxy indicators can itself be vulnerable to intentional misreporting. Armenia's family poverty benefit is more complex and evolved from the prior Paros system used to allocate humanitarian aid first after the 1988 earthquake and then after the war with Azerbaijan. It replaced 26 fragmented, categorically targeted uncoordinated benefits. The scheme constructs a family need score by multiplying together factors representing (a) family information on each member's social risk category and ability to work, (b) the household's place of residence and housing status, and (c) the household's income, expressed as a proportion of its estimated minimum income needs. (The first factor is the most important.) A family is eligible if (i) the family need score exceeds a threshold level, (ii) it has been assessed as eligible for social benefits by a process involving community and government representatives, and (iii) it does not own a car or business. If eligible, the family qualifies for a base benefit amount and a flat-rate supplement per child under 18. A recent analysis concluded that about 80 percent of the benefits paid went to families that would have been poor if those transfers had been suddenly withdrawn and that three-quarters of recipient poor families would have been extremely poor in their absence (Posarac, 2002). However, once again, 80 percent of families poor prior to transfers, and over 70 percent of extremely poor families, received no assistance. Real transfers under this program have fallen by almost one-third since its introduction in 1999. 31 52. Benefits Based on Community Judgments within Guidelines. Another unique program within the ECA region is based on the Mahalla system in Uzbekistan (see Box 1.1). The Mahalla tradition involves a group of respected elders who help to solve social problems and conflicts within the community. It has been adapted to provide a framework for allocating funds for social assistance benefits based on community judgments within guidelines. After the state abolished many social allowances in 1994, it delegated to the mahallas responsibility for the social assistance function of allocating cash assistance for low-income households with children. The mahallas' role was augmented to include, from 1997, targeting children's benefits and, from 1999, targeting maternity benefits to non-working mothers of children under two. As well, Mahallas can identify elderly people living alone and in need of care and recommend in-kind benefits ­ mainly foodstuffs. Each benefit is means-tested and a law provides guidance on how this should be done and on authorized benefit levels. (The main criterion is monthly per capita average income for the last 12 months.) Yet the mahallas also have considerable discretion; and all benefits for individual families are publicly discussed and approved in general Mahalla meetings open to the community. 53. Selected Price Discounts. Price discounts for a broad range of wide range of consumer goods and services, also known as "privileges," were widespread within the region in the early 1990s. Eligibility was diverse, reflecting meritorious conduct or social need or other political judgments. Most ECA countries have substantially reduced the range of goods and services covered, the categories of people eligible, and the size of the subsidies. Nevertheless, many such privileges remain within the region, in Ukraine as well as in the Eurasian transition states. Some countries have narrowed their scope and targeted the subsidies partly by limiting them to poorer social groups or even to poor families within such groups. Targeting energy price subsidies has proved a particularly challenging area in countries that are raising energy prices substantially as part of reforms to achieve financial viability within restructured energy sectors. This is particularly so when these countries have very cold winters or mountainous regions. 54. Other Approaches. Other approaches have been, or could be, considered. (For a recent review of international experience, see Coady, Grosh, and Hoddinott, 2002) Benefits Linked to Other Behavioral Goals. Conditional cash transfers (CCT) are cash payments to the families of poor children who regularly attend school or, in some countries, health clinics. Many countries have experimented with CCT for specific programs or are conditioning social assistance generally.12 Brazil's recent experience with this approach is described below. Turkey also has a large CCT program that combines this approach with proxy means-testing and elements of community judgment in rural some areas. Work-fare and Work-tested Benefits. Some countries have responded to concerns about work incentives by imposing work requirements on some or all recipients as a condition for receiving benefits. Bulgaria's recent initiative along those lines is described in a companion article in this volume on public works (Kuddo, 2003). Highly visible US welfare reforms in 1996 combined this approach with job training, subsidies for child and 12See Braithwaite (2003) for a discussion emphasizing the experience in Europe and central Asia. 32 dependent care, retention of health insurance assistance benefits, and a ceiling on duration of benefits.13 Mandatory Child-Support Payments. Like many other transition countries, Turkmenistan inherited from the Soviet era a system of mandatory child support following divorce. Husbands' child support obligations are set by law at 25 percent of salary for one child, 33 percent of salary for two children, and 50 percent of salary for three or more children. A child support payment of 50 Manats per month per child is payable from a Central Bank account if the husband pays no child support. This benefit is also available to unmarried women with children (for example, in second, unofficial families). 55. Conditional Cash Transfers in Brazil. Brazil, a country in which the richest one percent of the population and the poorest 50 percent each receive ten percent of total monetary income, has recently used conditional cash transfers as the model for social assistance reform. Its new Bolsa Familia Program (BFP) integrates and reforms a series of separate programs all targeting the same broad subgroup of the population through assistance programs that, along with somewhat different emphases (education, health, food consumption, or compensation for adjustment), also had different administrative structures, beneficiary selection processes, and banking-sector contracts for issuing payments. The BFP is a unified conditional cash transfer program that seeks to achieve short-term reductions in poverty through cash transfers and long- term reductions in poverty through encouraging investments in human capital. At the same time, the BFP would achieve more effective targeting, administrative efficiencies, stronger incentives for participant behavioral change, and program simplification. The program targets the extremely poor and the moderately poor. Payments are usually made to the mothers. The transfers are conditional on all relevant family members complying with key human development conditionalities: enrolment and regular school attendance for children aged 6-15 years; having up-to-date vaccinations for children aged 0-6 years; and prenatal visits for pregnant women. Program administration and implementation will build on Brazil's experience with earlier programs and the systems these used. Rigorous and extensive monitoring and evaluation form part of the program design. 56. Tax-Transfer Linkages. As transition countries develop stronger administrative systems, policy linkages between systems of taxation and social benefits could become more attractive. Family benefits or children's allowances illustrate the possible benefits from integrating or coordinating the tax and income-transfer systems. The income tax structure could potentially assist all families with children, including families with too little income to pay tax, by permitting a refundable tax credit for each child. But this mechanism requires a sophisticated system of tax administration and a well-informed, low-income population that applies successfully for the credit to be refunded. However, because the social insurance system can deliver cash assistance to a large part of the population, an administratively simpler alternative would be to pay each family a children's allowance that varies with family size in the same way as the tax credit. If the income tax structure included no other preferences for children (such as higher exempt levels of income), then that allowance should be nontaxable. Paying the allowance would be equivalent to providing a refundable income tax credit within the income tax 13United States Congress. Personal Responsibility and Work Opportunity Reconciliation Act of 1996, which replaced the old welfare system with a new program, Temporary Assistance to Needy Families (TANF). 33 structure. (In all other cases within an integrated tax-transfer system, however, social security benefits should be taxable, with benefit rates set accordingly.) Alternatively, if policymakers wished to provide assistance to families in a more targeted and selective way, they might prefer choose instead to do so through means-tested or categorical social assistance benefit systems. The latter approach would reduce the number of families assisted but, for a given level of spending on benefits, would enable larger benefits to be paid to each eligible family. Administrative costs, however, would be higher. 34 CHAPTER II: ASSESSING THE IMPACT OF SOCIAL TRANSFERS ON POVERTY AND INCOME DISTRIBUTION AMONG HOUSEHOLDS A. Summary 57. Distribution of Transfers. Public and private transfers account for almost 35 percent of the average Albanian household's income. Together, their contribution to total income increases, both as a percentage and in absolute terms, as the household grows poorer. Although public transfers, whether in the form of social assistance or social insurance programs, are more likely to reach poor households than rich ones, there is much scope for better coverage of poor and extremely poor households. Only one-fifth of poor households report receiving assistance through ndihma ekonomike, the Government's main income-tested cash social assistance program, although there appears to be substantial underreporting of participation in the program. Moreover, it appears that more than one-third of the funds `leak' to non-poor households. However, the true number might be considerably higher if underreporting was concentrated among ineligible households, or considerably lower if it was concentrated among eligible households. Even though, as expected, pensions have broader coverage of poor households, their target incidence is lower than that of ndihma ekonomike and their program beneficiaries are less likely to be poor. However, the average size of pensions is much higher than the average benefit received through ndihma ekonomike. Pensions, therefore, play an important role in poverty alleviation and in income redistribution. 58. Remittances. Remittances from abroad are highly effective and efficient transfers that play a vital role in poverty alleviation. They are much better targeted at poor and extremely poor households than even public transfers. Because their average size is also relatively large, they help substantially to reduce both the poverty headcount and the poverty gap. Remittances from abroad can be viewed as either transfers from outside the Albanian household or income from a migration-based household employment strategy. 59. Variability in Patterns. The effectiveness of social transfers in reducing poverty varies to some degree with differences in household composition and region. Generally, however, similar patterns are observed when assessing the impact that various social transfers have on income distribution in Albania. B. Introduction 60. Functions of Transfers. Social transfers that are controlled by governments can be viewed largely as measures to redistribute income: from current to past workers, in the case of pensions; and from taxpayers to poor families, in the case of social assistance programs. Private transfers similarly redistribute income between families, or between individuals residing in different regions within the country or abroad. Together, public and private transfers play a 35 significant role as safety nets that help alleviate poverty and improve living standards among their recipients. 61. Questions Addressed. This chapter analyzes the impact of transfer income on the economic well-being of poor and extremely poor households. It looks first at the percentage of the total income of all Albanian households that is received as transfer income from each source. In six ensuing sections, the analysis then addresses questions about (i) which transfers are targeted better towards poor households; (ii) which transfers disburse larger sums to poor households; (iii) which transfers are more successful at reducing poverty; (iv) what implications follow for the impact of transfers on poverty if remittances from abroad are interpreted as income of the recipient household rather than as a private transfer from outside; (v) which transfers are more efficient in terms of their cost-to-benefit ratio; and (vi) which transfers, overall, are more effective than others. Box 2.1 describes the basic methodology used. Box 2.1: Methodology To assess the impact a particular social transfer has on a household's economic welfare, (if consumption is used as the measure of welfare), one must first estimate the consumption level of the household assuming that the transfers did not exist. Accordingly, this analysis constructs a `counterfactual consumption' level that is consistent with the absence of both public and private transfers. It does so by making an assumption about what proportion of the transfers were used by households to finance additional consumption. The change in consumption relative to the change in income (in this case, transfer income) is defined as the marginal propensity to consume out of transfer income. It can range between 0% and 100%, depending on each household's behavior. That is to say, if the marginal propensity to consume were 100%, the household would consume all of its transfer income, while at 0% it would not change its consumption level at all. It is difficult, however, to predict the behavior of households in the absence of these transfers. Many might reduce their consumption by the full amount of the transfer income withdrawn. If a household is faced with the loss of transfer income, however, there are several reasons why it might reduce its consumption by less than the reduction in its income. It might finance part of its continuing consumption by increasing its indebtedness. Some household members might choose to work more or to begin working, thereby offsetting part of the lost transfers through increased wages or earnings. Faced with a sustained and foreseeable loss in income from transfers, some households might withdraw a child from school, or expand home production activities so as to maintain consumptions levels closer to their prior levels before the loss of income. Other households might join together or change their composition so as to exploit economies of scale from a larger household, although doing so could entail a loss of independence that the household members see as detracting from their overall well-being. Thus, it is clear that households would not always change their consumption levels by the full amount of the change in their incomes. It is hard to obtain reliable empirical estimates of households' average marginal propensity to consume out of transfer income. Accordingly, the analysis reported here uses alternative calculations to estimate the counterfactual consumption level. Counterfactual consumption levels are defined for four different assumptions about the marginal propensity to consume out of transfer income ­ 25%, 50%, 75% and 100%. The impact of transfers is then estimated under each assumption. For simplicity, while maintaining consistency in the analysis, results for most aggregated analyses are reported in this chapter using only the assumption of a 100% marginal propensity to consume. Appendix 2 provides tables showing the impact of social transfers on poverty and economic welfare under alternative assumptions about the marginal propensity to consume from transfer income. These tables reveal the extent to which a different counterfactual would have yielded different results. 36 62. Public transfers accounted for approximately 21 percent of the total income of all households, while private transfers accounted for another 14 percent (Figure 2.1). Within public transfers, urban pensions were the largest contributors to total income (10 percent of total income), while remittances from abroad were the major private source of transfer income (13 percent of total income). Figure 2.1: Distribution of Income by Source, All Households other public transfer other pension 1% 2% other income rural pension 1% 4% Urban pension 10% Non Agricultural Ndihma ekonomike Employment 3% 32% Remittances from abroad 13% Remittances from Albania 1% Agriculture Non agriculture 26% 7% Source: World Bank staff tabulations from Albania 2002 LSMS. 63. On average, public transfers contributed much more to the income of pre-transfer poor households (30 percent) and to that of pre-transfer extremely poor households (34 percent) than they did to the average income of all households (21 percent). This is evident from examining the contribution of the major sources of income to the total incomes of pre- transfer poor and extremely poor households, relative to their corresponding contributions to the incomes of non-poor and average Albanian households (Figure 2.2). (Unless otherwise specified, references below to "poor" and "extremely poor" households imply pre-transfer poverty.) Private transfers also accounted for a greater share of average income for poor households (20 percent) and extremely poor households (30 percent) than they did for non-poor households (11 percent). 64. The absolute sums of money received as public and private transfers are higher for extremely poor households than for poor households, which in turn have higher absolute transfers than the average household (Table 2.1). 37 Table 2.1: Distribution of Average Monthly Household Income, Amount in Leke Overall Pre-transfer Pre-transfer Poor Extremely poor Public Transfers 3,777 5,353 6,280 Private Transfers 3,343 6,556 11,877 Source: World Bank staff tabulations from Albania 2002 LSMS. Figure 2.2: Distribution of Income by Source, By Type of Household 100% 8 9 6 5 17 80% 25 24 26 Non-agriculture 14 & other income 60% 21 income Agriculture 31 30 39 total 40% 20 of Employment 14 20% 11 Share 30 34 Private transfers 21 16 0% Overall Non-poor Poor Extremely Public transfers poor Type of household Source: World Bank staff tabulations from Albania 2002 LSMS. 65. The importance of transfer income, both as a percentage of total income and as an absolute amount, increases for households that are poorer (Table 2.1, Figure 2.2). Transfer income contributes a large share of the average household's total income and makes even greater contributions to the incomes of poor and extremely poor households. On average, total public and private transfers contribute fifty percent of poor households' incomes and two-thirds of extremely poor households' incomes, but only about one-quarter of non-poor households' incomes. Given that, the impact of social safety net transfers on the income distribution warrants in-depth analysis. In part, this can help identify which households are more likely or less likely to improve their status through public or private transfers. C. Coverage of Public and Private Transfers ­ Which Transfers are Targeted Better Towards Poor Households? 66. Fifty-eight percent of the population receive public transfers and about one- quarter of the population receives private transfers (Table 2.2). Whereas 80 percent of the extremely poor households and 74 percent of poor households receive public transfers, just 55 percent of 38 the extremely poor households and 36 percent of poor households receive private transfers. The fact that two-thirds of the population receive a public or private transfer is itself a sufficient reason to study the impact that transfers have on the income distribution of households. Table 2.2: Coverage: Percentage of Individuals Within Each Category Receiving Public or Private Transfers Extreme Non Household Category Poorest Second Middle Fourth Richest Average Poor Poor Poor Urban Rural Total Transfers 93.7 75.8 67.1 53.3 42.1 66.4 94.4 83.9 52.7 65.4 67.2 Total Public 79.9 69.1 60.0 45.3 34.2 57.7 79.9 73.8 45.1 57.1 58.1 Total Private 53.5 21.6 17.8 15.8 14.9 24.7 55.1 35.8 16.1 24.8 24.7 Ndihma Ekonomike 19.2 19.3 11.4 10 2.8 12.5 19.0 18.6 7.8 11.7 13.1 Urban 32.5 25.1 22.0 18.2 16.6 22.9 32.3 28.4 18.6 35.1 14.2 Rural 31.4 24.5 21.7 14.4 12.6 20.9 30.5 27.6 15.7 6.7 31.1 Pensions Other 15.6 9.4 8.8 5.5 3.9 8.6 15.7 12.2 5.9 7.8 9.2 Total Transfers Pensions 65.5 53.0 48.1 34.7 29.9 46.2 65.4 58.7 36.5 45.4 46.8 Unemployment Public Benefits 1.7 1.3 2.2 0.2 0.8 1.2 1.7 1.7 0.9 2.0 0.7 Maternity Benefits 1.2 1.1 1.0 0.6 0.6 0.9 1.3 1.2 0.7 1.0 0.8 Social Care 0.8 0.4 0.3 0.2 0.3 0.4 0.8 0.5 0.3 0.3 0.5 Other Public 3.9 3.8 3.1 1.9 2.0 2.9 4.1 3.8 2.3 3.6 2.5 Within Albania 7.7 3.0 2.1 4.0 3.5 4.1 8.0 5.1 3.3 5.1 3.3 From Abroad Transfers Remittances 42.4 17.5 15.2 11.0 10.9 19.4 43.5 28.7 12.2 18.6 20.0 All Private Remittances 50.1 20.5 17.3 15.1 14.4 23.5 51.5 33.7 15.5 23.7 23.3 Other Private. 3.4 1.2 0.4 0.7 0.5 1.2 3.6 2.1 0.6 1.1 1.3 Note: Consumption quintiles and poverty status are estimated in the absence of all social transfers, assuming a 100 percent marginal propensity to consume out of transfer income. 67. Ndihma ekonomike, the main income-tested social assistance public transfer, was reportedly received by only 19 percent of all poor individuals and the same percentage of all extremely poor individuals (Table 2.2). Although not all poor households meet the program's categorical eligibility criteria, the program's low apparent level of coverage and its failure to assist the poorest households disproportionately are both clear signs that a review of the program's coverage and administration is warranted. Moreover, almost 8 percent of non-poor households receive ndihma ekonomike, which further supports the need for better targeting of the program. The total money reportedly being received through ndihma ekonomike by the non-poor is estimated to be about 70 million Leke14 (Box 2.2). As discussed there, however, the level of participation in the ndihma ekonomike program that is reported in the LSMS survey is only about half the level shown by administrative data; and it is unclear to what 14 The amount of money transferred is estimated from the survey and does not include administrative and other overhead expenses made by the Government. 39 extent under-reporting has affected the distribution of recipient families across poverty categories that is estimated from the survey data. 68. For people belonging to a poor household, the most commonly received transfers are remittances from abroad, followed closely by urban and rural pensions. Remittances from abroad reach 29 percent of poor households, while each pension category is received by 28 percent. Taken together, pensions overall are received more commonly than remittances. 69. Individuals living in rural locations and those in urban locations are equally likely to receive some form of public transfer. The same is true for private transfers. One striking pattern is that 14 percent of individuals living in rural locations report receiving urban pensions, while about 7 percent of those living in urban locations report receiving a rural pension. This can arise from occupational changes during an employment career, migration between rural and urban locations, or changes over time in families' composition, including through household mergers across generations when family members retire. 70. Public transfers, whether in the form of social assistance or social insurance programs, are more likely to reach poor households than non-poor ones. Despite this, there is much scope for better coverage of poor and extremely poor households. The poorer quintiles of the population are proportionally more likely to receive public transfers than the richer quintiles. On average, 80 percent of households among the poorest fifth receive public transfers, compared with 34 percent of those among the richest fifth). Ndihma ekonomike is reportedly received by approximately 20 percent of the poorest fifth of the population, but by less than 3 percent of the richest fifth. Overall, 12.5 percent of households reported receiving ndihma ekonomike. 71. Pensions also play an important role, because 46 percent of people live in households receiving some form of pension. Poor households are more likely to receive both urban and rural pensions (about 28 percent each) than non-poor households (19 percent for urban pensions, and 16 percent for rural pensions). 72. Among public transfers, pensions are more successful in covering extremely poor households than even ndihma ekonomike. Almost two of every three extremely poor households report receiving either an urban or rural pension. This compares with one in five of such households that report receiving ndihma ekonomike. Other public transfers such as unemployment benefits and maternity benefits were being received by relatively small sections of society, approximating 1 percent of the population. 40 Box 2.2: Estimating Total Transfers From LSMS Data Albania's LSMS provides an opportunity to estimate the total sums received by households through public and private transfers. The total amounts that households report receiving are: ndihma ekonomike, 2.30 billion Leke; urban pensions, 18.94 billion Leke; rural pensions, 5.56 billion Leke; other pensions, 3.68 billion Leke; unemployment benefits, 513 million Leke; remittances from within Albania, 2.15 billion Leke; and remittances from abroad, 26.79 billion Leke. For all public transfers received other than ndihma ekonomike, the estimated amounts are close to the amounts reported through official data sources such as MoLSA and INSTAT. For ndihma ekonomike, however, the estimated amount of benefits received that is based on reports from LSMS households (2.30 billion Leke) is well below the official expenditures reported by MoLSA (4.37 billion Leke). Similarly, the implied estimate for the number of households that received ndihma ekonomike, which is 80,272, is about 40 percent lower than the official number of 135,378 households. This 47 percent difference between survey and administrative data on implied expenditures can only be partly explained by the fact that neither administrative nor overhead expenditures are captured by the LSMS and thus the survey-based expenditure estimates for ndihma ekonomike are underestimated. A second reason could be that, although the sample is designed to be representative nationally, it might not be representative of households receiving ndihma ekonomike. Consequently, findings of this report with respect to the impact of ndihma ekonomike on poverty should be interpreted with considerable care and caution. If the official sources of data are accurate, findings reported in this study underestimate the level of coverage of the ndihma ekonomike program. Moreover, the report's analyses of the distributional incidence of program benefits could be biased in either direction, depending on whether households failing to report receipt of this transfer were predominantly poorer, eligible households or households that failed to meet program eligibility requirements for income or household composition. Unfortunately, it is not possible to ascertain whether or not the `missing' households (those which purportedly received ndihma ekonomike transfers but did not report them in the survey), displayed the same distribution of characteristics as those that did report receiving ndihma ekonomike benefits. If those `missing' households followed the same distribution as households reporting receipt of ndihma ekonomike benefits, then estimates reported in this study for coverage, target incidence, and beneficiary incidence of ndihma ekonomike would not change. The marginal impact of ndihma ekonomike on poverty, however, would roughly double from the very small 2.3 percent to a small 4.6 percent. In the extreme case where all `missing' households were indeed poor households, then the estimates of coverage, target incidence and beneficiary incidence would increase, as would the marginal impact of ndihma ekonomike on poverty. Conversely, if all `missing' households were non-poor, then the estimates of coverage, target incidence and beneficiary incidence would decrease, even though the marginal impact of ndihma ekonomike on poverty would increase. Nevertheless, although these issues should be pursued through future research, the LSMS data are the best data currently available with which to explore related policy questions. The Albania Poverty Assessment also used the LSMS data to estimate that private remittances received from persons living abroad for more than a year totaled US$203 million. To this , it added another US$156 million as an estimate of the earnings brought back by household members who migrated in the course of the year. Thus, total remittances were estimated at US$359 million. 41 73. Poor households were more than twice as likely to receive remittances from abroad (29 percent) as non-poor households (12 percent). About 43.5 percent of extremely poor individuals receive remittances from abroad. These private transfers, which clearly are very well targeted, can be important income sources that help alleviate poverty, especially for those who are very poor. The important role of remittances is not surprising, given the scale and patterns of migration among Albanians. Many Albanian families appear to use migration to countries with higher average wages and stronger demand for labor as a strategy through which a husband or adult child can increase total family income from employment. Earnings abroad can then be returned as remittances. It is likely that migration designed to generate remittances from abroad represents a coping mechanism through which poor households, and those with key adults unemployed, can seek to improve their circumstances and maximize their income from employment. So the departure abroad in this way of potential income earners helps explain two things: why households whose members remain behind are more likely to be poor; and why poor households in Albania are more likely than non-poor ones to receive remittances as transfers. On the whole, remittances from abroad were being received by about a fifth of the population, while remittances from within Albania were being received by about 4 percent of the population. As in the case of remittances from abroad, the poorest fifth were more likely to receive remittances from within Albania (about 8 percent) than the richest fifth (about 3.5 percent). Possibly, remittances within Albania, which predominantly go to urban areas, represent a subsidy from rural households for job search by family members who are seeking employment in Tirana or other cities. 74. An alternative way to analyze `target incidence' ­ the incidence of individuals' receiving each benefit ­ is by calculating the percentage of all people receiving the benefit who are found within each quintile or other population subgroup (Table 2.3). Naturally, a social transfer that is received disproportionately by households in the poorer sections of society implies better targeting towards poor households. In particular, if more than 20 percent of the people receiving a transfer are members of the poorest quintile, then the transfer's target incidence is better than random. 75. Although, relative to the richer quintiles, the poorer quintiles contained a disproportionately high number of beneficiaries receiving transfers, there was still considerable `leakage' of transfers to non-poor sections of society. While individuals in pre- transfer poor households represented 44 percent of the population, they comprised 55 percent of beneficiaries receiving total transfers, 56 percent of beneficiaries receiving public transfers and 63 percent of beneficiaries receiving private transfers. Similarly, households that were extremely poor before transfers accounted for about 20 percent of the population, whereas they represented 27 percent of the beneficiaries getting any transfer, 26 percent of the beneficiaries getting public transfers, and 42.5 percent of the people getting private transfers. 76. More than one in every three reported beneficiaries of ndihma ekonomike is non- poor. Under-reporting within the survey of those receiving the benefit could affect this finding, however, in either direction. Nevertheless, this is a major leakage in the system and leaves significant scope for improvement. Although the dataset is too small to answer the question, it would be valuable to know whether there is a uniform rate of leakage across the country, or whether the leakage is concentrated in a subset of local government areas. 42 Table 2.3: Distribution of Program Beneficiaries Non- Quintiles Total Poorest Second Middle Fourth Richest Extreme Poor Poor Poor Urban Rural Total Transfers 100.0 28.2 22.9 20.2 16.0 12.7 27.0 55.5 44.5 41.1 58.9 Total Public 100.0 27.7 24.0 20.8 15.7 11.8 26.3 56.1 43.9 41.3 58.7 Total Private 100.0 43.3 17.5 14.4 12.8 12.0 42.4 63.5 36.5 41.8 58.2 Ndihme Ekonomike 100.0 30.7 30.8 18.1 16.0 4.4 28.8 65.2 34.8 39.0 61.0 Urban 100.0 28.4 21.9 19.2 15.9 14.5 26.9 54.4 45.6 63.9 36.1 Rural 100.0 30.0 23.5 20.7 13.8 12.0 27.7 58.0 42.0 13.4 86.6 Pensions Other 100.0 36.2 21.8 20.2 12.8 9.0 34.4 62.0 38.0 37.7 62.3 ansfers Total Tr Pensions 100.0 28.3 23.0 20.8 15.0 12.9 26.9 55.7 44.3 41.0 59.0 Unemployment Benefits 100.0 27.8 20.5 35.2 3.9 12.7 26.5 59.5 40.5 68.5 31.5 Public Maternity Benefits 100.0 26.9 24.8 22.5 13.0 12.8 26.9 56.8 43.2 47.3 52.7 Social Care 100.0 40.7 19.0 15.9 11.2 13.2 40.7 59.7 40.3 30.0 70.0 Other Public 100.0 26.2 26.2 21.0 12.8 13.9 26.2 57.1 42.9 50.6 49.4 Within Albania 100.0 37.9 14.7 10.4 19.9 17.1 37.5 54.7 45.3 52.5 47.5 Remittances From Abroad Transfers 100.0 43.7 18.0 15.7 11.4 11.2 42.6 64.8 35.2 39.9 60.1 All Remittances Private 100.0 42.7 17.5 14.8 12.8 12.2 41.7 63.0 37.0 42.0 58.0 Other Pvt. 100.0 54.5 18.5 6.7 11.9 8.4 54.5 73.1 26.9 37.0 63.0 Share of Households in Population 100.0 20.0 20.0 20.0 20.0 20.0 19.0 43.9 56.1 41.7 58.3 Note: Consumption quintiles and poverty status are estimated in the absence of all social transfers assuming a 100 per cent marginal propensity to consume out of transfer income. 77. Urban and rural pensions were well targeted at poor households, with 54 percent of the beneficiaries of the former and 58 percent of the beneficiaries of the latter belonging to poor households. Among beneficiaries of urban pensions, 36 percent belong to rural households. Of those receiving unemployment benefits, 68.5 percent belong to urban households and 31.5 percent belong to rural households. The Poverty Assessment for Albania (World Bank, 2003) shows that the unemployment rate was estimated to be 11 percent in urban areas as against 2.2 percent in rural areas. In this regard, unemployment benefits seem to follow the same pattern. 78. In terms of reaching the extremely poor, private transfers were much better targeted than public transfers. More than 6 out of every 10 individuals who received remittances from abroad were poor and more than 4 out of every 10 such individuals were extremely poor. Remittances from within Albania were received relatively more evenly by poor and non-poor households although, even in this case, 55 percent of those receiving remittances from within Albania were poor. 43 79. Individuals in rural households are significantly more likely to receive remittances from abroad than those living in urban households (t ratio: 2.5). In rural areas, 46.7 percent of the population are in pre-transfer poverty, compared with 40 percent of the urban population. Because pre-transfer poverty is higher in rural than urban areas, it is possible that individuals in rural households are more likely to receive remittances from abroad because they are more likely to be poor. However, much deeper analysis is needed to identify what factors lead members of rural households to be more likely than those in urban households to receive remittances from abroad. D. Distribution of Public and Private Transfers Among the Public - Which Transfers Disburse Larger Sums to Poor Households? 80. What differences exist in the volume of funds transferred as public and private transfers to different sections of society? How do these volumes vary across different quintiles of the population and between urban and rural locations? The absolute target incidence is calculated for each transfer by measuring the average benefit received by each subpopulation. 81. A related concept used here classifies each public and private transfer into one of three categories on the basis of how the transfer's incidence compares with the target group's share of consumption and its share in the population. Specifically, transfers are classified as: § Regressive if the poorer subpopulation receives a smaller share of the total transfer than its share in total consumption; § Mildly Progressive: if the poorer subpopulation receives a larger share of the total transfer than its share in total consumption, but less than its share in the total population; and § Highly Progressive: if the poorer subpopulation receives a larger share of the total transfer than both its share in total consumption, and its share in the total population. 82. Total public and private transfers, considered jointly, were highly progressive. The poorest fifth of the population received 48 percent of all transfers, whereas the richest fifth received only 11.1 percent of the transfers (Table 2.4). In terms of poverty status, poor households received 67 percent of the total transfers, while extremely poor households received 47 percent of the transfers. Fifty-seven percent of the public transfers went to households in pre- transfer poverty (compared with their share in the population of just under 44 percent), while the non-poor households received the rest (43 percent). Also, the share of public transfers received by extremely poor households was almost 31 percent, whereas their share in total population was only 19 percent. So public transfers were also highly progressive. 44 Table 2.4: Distribution of Public and Private Transfers: Percentage of Transfers Received by Each Category (Absolute Target Incidence) Extr- Non- Total Poorest Second Middle Fourth Richest Poor eme poor Urban Rural Quintiles Poor Total Transfers 100 47.6 16.5 13.6 11.1 11.1 67.2 47.0 32.8 46.3 53.7 Total Public 100 31.6 21.4 18.8 14.5 13.7 56.9 30.6 43.1 56.8 43.2 Total Private 100 65.7 10.9 7.7 7.4 8.3 78.8 65.4 21.2 34.5 65.5 Ndihme Ekonomike 100 30.7 28.9 18.5 15.7 6.3 63.6 28.4 36.4 54.6 45.4 Urban 100 31.7 19.3 19.0 15.4 14.6 55.1 30.8 44.9 72.0 28.0 Rural 100 28.6 22.3 18.9 14.0 16.1 54.8 27.3 45.2 13.1 86.9 Pensions Other 100 38.8 20.0 18.8 13.2 9.1 62.3 38.0 37.7 47.5 52.5 Total Transfers 100 32.0 20.0 19.0 14.8 14.2 56.0 31.0 44.0 57.2 42.8 Pensions Public Unemployment Benefits 100 20.4 25.0 31.4 3.6 19.5 52.6 19.2 47.4 82.2 17.8 Maternity Benefits 100 31.4 23.4 22.6 7.3 15.4 58.4 31.4 41.6 65.7 34.3 Social Care 100 54.8 13.8 10.4 11.2 9.8 68.7 54.8 31.3 39.0 61.0 Other Public 100 26.7 36.2 11.7 11.0 14.4 64.7 26.7 35.3 41.7 58.3 Within Albania 100 51.6 13.9 7.1 15.9 11.6 69.3 50.8 30.7 66.3 33.7 From Remit- 100 66.7 10.9 7.9 6.7 7.8 79.6 66.4 20.4 31.9 68.1 Abroad Transfers tances All 100 65.5 11.1 7.9 7.4 8.1 78.8 65.2 21.2 34.5 65.5 Private Remittances Other Pvt. 100 74.7 2.3 0.6 7.5 14.9 77.0 74.7 23.0 32.5 67.5 100 20.0 20.0 20.0 20.0 20.0 43.9 19.0 56.1 41.7 58.3 Share of households in category Share of households in total 100 1.0 10.5 15.7 23.0 49.8 14.0 0.7 86.0 49.5 50.5 consumption Note: Consumption Quintiles and Poverty Status are estimated in the absence of all social transfers assuming a 100% marginal propensity to consume out of transfer income. 45 83. Of reported transfers through ndihma ekonomike, only 28 percent reached extremely poor households, although 64 percent of transfers reached poor households. However, 36 percent were received by non-poor households. So ndihma ekonomike is highly progressive as a transfer, although the `leakage' of funds to the non-poor is also large. Pensions were also highly progressive, with about 56 percent of pension transfers going to poor households. Most other public transfers, such as unemployment benefits and maternity benefits, are also highly progressive. In addition, 82 percent of spending on unemployment benefits goes to urban areas, which perhaps reflects higher unemployment rates in urban areas than in rural areas. 84. For both poor and extremely poor households, remittances from abroad were the most progressive of all public and private transfers that delivered more than one percent of household income overall. Eighty percent of remittances from abroad went to poor households and an astonishingly high 66 percent of such transfers went to extremely poor households, with non-poor households receiving only 20 percent. Private transfers by NGOs and similar institutions in the country were even better targeted at poor and extremely poor households, but they formed only a small percentage of total private transfers. 85. Two-thirds of remittances from abroad were received by individuals in rural households. This was in direct contrast to local transfers, for which two-thirds went to recipients in urban households. Individuals in rural households were likely to receive larger sums as transfers from abroad than those in urban households; and individuals in urban households were likely to receive larger transfers from within Albania than those living in rural households.15 E. Impact of Public and Private Transfers on Poverty Reduction ­ Which Transfers are More Successful at Reducing Poverty? 86. This section assesses the impact of public and private transfers in reducing poverty, measured by the three indicators - poverty headcount, gap, and severity. Even though not all public and private transfers have poverty reduction as their primary or implicit objective, their size and coverage are such that they frequently raise consumption levels and lower poverty rates. 87. Income from transfers helped raise about 18 out of every 100 individuals out of poverty. For Albania as a whole, the incidence of poverty, or poverty headcount, after receipt of transfer income, was estimated at 25.4 percent. The poverty gap index, which measures the average income shortfall of poor households in relation to the poverty line, was 5.7 percent. The poverty gap squared, a measure of severity of poverty, was 1.9. As described above, estimates were recalculated for these three indicators by stripping all households of all transfer incomes, based on the assumption that 100 percent of the transfer income was used for consumption purposes. The revised distribution of consumption aggregates for 2002 implied that 43.9 percent 15Both results were statistically significant, with t-ratios of 3.4 and 2, respectively. 46 of all individuals in Albania were below the (pre-transfer) poverty line, with the poverty gap index increasing to 20.1, and poverty gap squared increasing to 25. Similarly, the post-transfer extreme poverty headcount stood at 4.7 percent but, on a pre-transfer basis, its level was much higher at 19 percent. This implies that public and private transfers were able to reduce extreme poverty by almost 75 percent (Figures 2.3 and 2.4). Figure 2.3: Impact of Total Public and Private Income Transfers on Indicators of Poverty, Treating Remittances from Abroad as Private Transfers 50.0% 43.9% 40.0% 30.0% 25.4% 25.0% 20.1% 20.0% 10.0% 5.7% 1.9% 0.0% Headcount Poverty Gap Poverty Severity post-transfer pre-transfer Source: World Bank staff tabulations from Albania 2002 LSMS . Figure 2.4: Impact of Total Public and Private Income Transfers on Indicators of Extreme Poverty, Treating Remittances from Abroad as Private Transfers 50.0% 40.0% 38.3% 30.0% 19.0% 20.0% 13.4% 10.0% 4.7% 0.8% 0.2% 0.0% Headcount Poverty Gap Poverty Severity post-transfer pre-transfer Source: World Bank staff tabulations from Albania 2002 LSMS. 47 88. Public transfers, in total, had a large impact on poverty reduction, mainly through social insurance transfers such as pensions. However, ndihma ekonomike had only a small impact on reducing the poverty headcount (Figures 2.5 and 2.6). In the absence of all income transfers, the estimated poverty headcount was 43.9 percent, as noted earlier. If all public transfers are added back, the poverty headcount drops to 32.5 percent, implying a 26 percent reduction in poverty incidence. Of all public transfers, pensions reduced poverty the most (poverty reduction of 22 percent). Pensions also reduced the poverty gap by 29 percent, although they were overshadowed by private transfers in their impact on reducing poverty severity. Recall that poverty is measured for the population as a whole and not for a specific sub-population that receives the transfer income. Although it is likely that consumption levels would increase for families receiving ndihma ekonomike benefits, the impact is very small in the context of Albania as a whole (2 percent reduction in poverty incidence). 89. Private transfers also had a substantial impact on the poverty headcount, but they had an even greater impact on the distribution of income (Figure 2.5). Private transfers reduced the poverty headcount by 15 percent, less than the 26 percent impact of public transfers. However, private transfers reduced the poverty gap by 45 percent and the severity of poverty by 78 percent ­ the latter, a far larger impact than for public transfers. Figure 2.5: Impact of Income Transfers on Poverty Indicators, By Source of Transfer, Treating Remittances from Abroad as Private Transfers 100.0 80.0 78 73 poverty only public in 60.0 only pvt 45 decrease 40 40.0 35 Ndhima only 29 26 28 25 22 Percentage 20.0 15 14 Pensions only 2 4 2 Remittances 0.0 from abroad Headcount Poverty Gap Poverty Severity only Source: World Bank staff tabulations from Albania 2002 LSMS. 90. The impacts that income transfers from each individual source had on indicators of extreme poverty correspond broadly to their impacts on indicators of poverty (Figure 2.6). Once again, public transfers brought about the highest percentage reduction in poverty headcount (44 percent). However, they were not as effective in redistributing income (poverty gap reduction of 37 percent) as private transfers (poverty gap reduction of 74 percent). Reported 48 ndhihma ekonomike transfers were only one-seventh as effective in reducing poverty as remittances from abroad. Figure 2.6: Impact of Income Transfers on Extreme Poverty Indicators, by Source of Transfer, Treating Remittances from Abroad as Private Transfers 100.0 92 88 80.0 74 only public 67 poverty in 60.0 only pvt 4441 decrease 40.0 36 35 37 Ndhima only 33 22 20 Pensions only 20.0 percentage 6 3 Remittances 1 from abroad 0.0 only Headcount Poverty Gap Poverty Severity Source: World Bank staff tabulations from Albania 2002 LSMS. Box 2.3: Determining the Marginal Impact of Social Transfers on Poverty and Extreme Poverty After Receipt of Social Insurance Transfers A parallel analysis was done to measure the marginal impact of transfers through ndihma ekonomike and remittances from abroad, assuming that the population was receiving no other social transfers except urban or rural pensions (again using a 100 marginal propensity to consume out of transfer income). Based on this, a new counterfactual consumption aggregate was calculated, subsequent to which transfers from ndihma ekonomike and remittances from abroad were added one at a time. The results showed that, at the counterfactual consumption level, poverty headcount was 34.4 percent, which decreased by 1 percentage point if only transfers from ndihma ekonomike were added back, and decreased by 6.3 percentage points if only remittances from abroad were added back. Similarly ndihma ekonomike reduced the poverty gap by only 0.7 percentage points, while remittances from abroad reduced the poverty gap by almost 7 percentage points. This shows that the marginal impact of remittances from abroad is much greater than that of ndihma ekonomike, even when the population receives social insurance transfers (pensions). 91. So far, analysis of the impact of social transfers has been reported at a national level, which can hide variations by type of household or even region. Disaggregated analyses were also undertaken to see whether the impact on poverty of transfer income from each source varied by (a) u rban or rural location, (b) region, or (c) household composition. Although not all results are discussed in depth, some interesting results are shared in this section. As before, the results all presuppose that the marginal propensity to consume out of transfer income is 100 percent. 49 By urban-rural location 92. Public transfers have a greater impact in urban areas (14.7 percent reduction in poverty) than in rural areas (9 percent), while private transfers have a greater impact in rural areas (7.3 percent) than in urban areas (5.9 percent; Table 2.5). Ndihma ekonomike also benefited urban households more than rural households, with a poverty headcount reduction of 1.2 percent in urban areas as against 0.6 percent in rural areas. Not surprisingly, urban pensions had a much larger impact in urban areas (10.3 percent reduction in poverty headcount), but they nevertheless impacted poverty in rural areas, as well (3.3 percent reduction). Similarly, rural pensions were more effective in rural areas than in urban areas. Most other income transfers had an equal impact in reducing poverty in urban and rural areas alike. The impact of transfers from specific sources in reducing extreme poverty was similar to their impact in reducing poverty. Table 2.5: Percentage Point Reduction in Poverty and Extreme Poverty by Location of Household, Treating Remittances from Abroad as Private Transfers Absolute %-point reduction in povertyAbsolute %-point reduction in extreme poverty Overall Urban Rural Overall Urban Rural Total Public 11.4 14.7 9.0 8.3 9.4 7.5 Total Private 6.7 5.9 7.3 7.8 6.8 8.5 Ndihma Ekonomike 0.9 1.2 0.6 1.1 1.1 1.0 Urban 6.0 10.3 2.9 4.0 6.4 2.2 Rural 2.2 0.6 3.3 1.8 0.4 2.7 Pensions Other 1.4 1.4 1.4 1.4 1.2 1.5 Transfers Total Pensions 9.5 11.9 7.7 6.8 7.6 6.2 Unemployment Benefits 0.3 0.3 0.4 0.2 0.1 0.2 Public Maternity Benefits 0.2 0.2 0.1 0.1 0.1 0.1 Social Care 0.1 0.0 0.1 0.1 0.1 0.0 Other Public 0.5 0.8 0.3 0.3 0.4 0.3 Within Albania 0.6 0.8 0.4 0.8 1.1 0.6 From Abroad 6.1 5.1 6.8 6.7 5.6 7.4 Remittances Transfers All Remittances 6.6 5.9 7.2 7.5 6.8 8.0 Private Other Pvt. 0.1 0.1 0.1 0.3 0.0 0.5 Note: Poverty headcount is estimated in the absence of all social transfers assuming a 100% marginal propensity to consume out of transfer income. By region 93. Public and private income transfers had the greatest overall impact in alleviating poverty for households in the central region (Table 2.6). This held true for both public and private transfers and for reductions in both poverty and extreme poverty. Public transfers accounted for a reduction of 12.6 percentage points in poverty incidence in the central region, 50 compared with a reduction of only 8.7 percentage points in the mountain region. Similarly, private transfers reduced poverty by 8.3 percentage points in the central region, as against 3.2 percentage points in the mountain region. Urban pensions had a greater impact on poverty and extreme poverty in Tirana, which is primarily urban, than in other regions. Following the same trend as public transfers, private remittances from abroad also had a greater impact in the central region than in other regions. However, ndihma ekonomike had a greater impact in the mountain region than other regions, in reducing both poverty and extreme poverty. This could reflect a pattern whereby many former residents of the mountain region have moved to other regions, applied for ndihma ekonomike, and been awarded only a partial benefit because they still own land in the mountain region. By household composition 94. Restricting the analysis to specific household types reveals considerable variation in the impact that public and private transfers have on poverty headcount for those household types. The three specific household types chosen for disaggregated analysis were (i) those having adults of pensionable age, (ii) those having children (up to age 15), and (iii) those having only members of pensionable age. Results for these groups are compared against the poverty reduction impact of social transfers on the average household. 95. Households that have elderly dependants are significantly more likely to benefit from public income transfers than the average household (Table 2.7). Pensions as a whole helped reduce poverty by 20 percentage points for households with members of pensionable age, although they reduced poverty by only 9.5 percentage points for the population as a whole. Evidently, urban pensions played a more important role than rural pensions in reducing the poverty headcount for households with elderly dependants. Private remittances from abroad, on the other hand, did not have a significantly greater impact on the poverty headcount of households having elderly members than they did on the average household. 96. Households having children are, on average, less likely to benefit from public or private transfers than the average households (Table 2.7). While ndihma ekonomike plays a similar role for the economic welfare of such households as it does for the average household, both pensions and remittances from abroad have a smaller impact on economic welfare for households with children than for the average household. 97. Households composed only of members of pensionable age benefited much more from social transfers than the average household, in terms of poverty reduction (Table 2.7). Again, urban pensions seemed to have had a much greater impact on poverty reduction for households having only members of pensionable age than they did for the average household. Rural pensions also reduced poverty more among households having only members of pensionable age than for the average household. One hypothesis suggested by these findings is that only those people of pensionable age who have significant transfer income from pensions or remittances are able to live independently of their younger relatives. 51 Table 2.6: Percentage Point Reduction in Poverty and Extreme Poverty by Region, Treating Remittances from Abroad as Private Transfers Absolute % point reduction in poverty Absolute % point reduction in extreme poverty Coastal Central Mountain Tirana Coastal Central Mountain Tirana Total Public 10.6 12.6 8.7 11.7 5.4 9.8 9.2 8.9 Total Private 6.3 8.3 3.2 5.0 7.1 9.8 4.3 5.3 Ndihma Ekonomike 0.4 0.9 1.9 0.7 0.7 1.2 2.2 0.4 Urban 5.1 6.8 3.2 8.3 2.8 4.1 3.9 6.6 Rural 2.2 2.5 2.0 0.8 1.2 2.3 1.8 0.9 Pensions Other 1.9 1.0 1.7 1.2 1.1 1.8 1.3 0.4 Transfers Total Pensions 9.1 10.4 6.6 9.9 4.6 8.0 7.0 7.8 Unemployment Benefits 0.1 0.5 0.1 0.6 0.0 0.2 0.5 0.2 Public Maternity Benefits 0.2 0.2 0.1 0.0 0.2 0.1 0.1 0.0 Social Care 0.2 0.0 0.1 0.0 0.0 0.1 0.0 0.0 Other Public 0.8 0.4 0.4 0.3 0.0 0.4 0.2 1.0 Within Albania 0.5 0.7 0.2 0.5 0.5 1.2 0.1 1.0 From Abroad 5.8 7.5 2.9 4.4 6.5 8.1 3.8 4.3 Remittances Transfers All Remittances 6.3 8.2 3.2 4.9 7.1 9.2 3.9 5.3 Private Other Pvt. 0.0 0.2 0.0 0.1 0.0 0.6 0.3 0.0 Note: Poverty headcount is estimated in the absence of all social transfers assuming a 100% marginal propensity to consume out of transfer income. 52 Table 2.7: Percentage-Point Reduction in Poverty and Extreme Poverty, by Household Composition, Treating Remittances from Abroad as Private Transfers Absolute % point reduction in poverty Absolute % point reduction in extreme poverty Overall Have old Have child All old Overall Have old Have child All old Total Public 11.4 22.0 9.2 41.1 8.3 14.6 7.4 24.7 Total Private 6.7 7.2 5.9 12.1 7.8 9.5 7.3 14.7 Ndihma Ekonomike 0.9 0.4 0.9 0.0 1.1 0.3 1.3 0.0 Urban 6.0 13.5 4.1 30.3 4.0 8.5 2.9 17.9 Rural 2.2 4.8 1.8 8.2 1.8 4.1 1.5 5.5 Pensions Other 1.4 2.2 1.2 1.7 1.4 1.7 1.2 1.6 Transfers Total Pensions 9.5 20.4 7.0 40.7 6.8 13.8 5.6 24.3 Unemployment Benefits 0.3 0.2 0.4 0.0 0.2 0.0 0.2 0.0 Public Maternity Benefits 0.2 0.2 0.1 0.0 0.1 0.1 0.1 0.0 Social Care 0.1 0.1 0.1 0.0 0.1 0.1 0.0 0.0 Other Public 0.5 0.8 0.5 2.3 0.3 0.6 0.3 0.7 Within Albania 0.6 0.2 0.5 1.9 0.8 0.6 0.7 2.4 From Abroad 6.1 7.0 5.3 9.9 6.7 8.4 6.2 12.0 Remittances Transfers All Remittances 6.6 7.1 5.8 11.8 7.5 9.0 6.9 14.5 Private Other Pvt. 0.1 0.0 0.1 0.3 0.3 0.6 0.4 0.3 Note: Poverty headcount is estimated in the absence of all social transfers assuming a 100% marginal propensity to consume out of transfer income. 53 Box 2.4: Impact of Transfers Through Ndihma Ekonomike and Rural Pensions, on Welfare of Rural Households Having at Least One Family Member of Pensionable Age A similar analysis was undertaken in which the sample was restricted to only those rural households that had household members of pensionable age. Sixty-eight percent of individuals living in such households reported receiving rural pensions, while 30 percent reported receiving urban pensions, and 8 percent reported receiving ndihma ekonomike. Households that received urban pensions, on average, received 7,984 Leke; households that received rural pensions, on average, received 3,296 Leke, and those households receiving ndihma ekonomike received, on average, only 1,984 Leke. Ndihma ekonomike was neither very successful in reducing the poverty headcount, nor in reducing the extreme poverty headcount for such households. Urban and rural pensions combined reduced the poverty headcount by 16 percentage points, whereas ndihma ekonomike only reduced the headcount by 0.3 percentage points. Similarly, pensions reduced the extreme poverty headcount by 12 percentage points, whereas ndihma ekonomike reduced it by only 0.4 percentage points. Although a smaller percentage of such households reported receiving remittances from abroad (23 percent), the average size of these remittances, for those households receiving them, was 15,385 Leke per month. Remittances from abroad were able to reduce the poverty headcount by 7.4 percentage points and the extreme poverty headcount by 9.1 percentage points. F. Impact of Social Transfers on Poverty Reduction Under an Alternative Assumption ­ Which Transfers are More Successful at Reducing Poverty, Assuming Private Remittances are Treated as Income Rather Than Transfers? 98. Analyses reported in this chapter demonstrate unequivocally that remittances from abroad play a critical role in the economic well-being of one-fifth of all Albanian households and almost half of the poorest households. As the chapter summary noted, remittances from abroad can be viewed from two perspectives. · One perspective sees each such remittance as a private transfer of resources from outside the country to the residents of a particular household dwelling unit within Albania. As such, the remittance can be analyzed in the same way as any other income transfers that might flow to the household from private or public sources outside the household. The resources can make a substantial difference to the family's living standards, but uncertainties remain about whether the remittance will be repeated and, if so, with what regularity, on what scale, for what duration, and with what level of reliability concerning successful transmission of the funds. · The other perspective sees remittances not as a safety net but as an employment strategy. Within this perspective, each remittance from abroad should be viewed as an internal transfer within a single family that, in the face of persistently inadequate demand for labor nationwide, has adopted a migration-based household employment strategy. That strategy enables the family to maximize the income earned from its members' 54 human capital. The strategy necessitates, however, that family members function as a dual-residence household, with one or more family members working abroad and remitting home part of their earnings. For each remittance, within this perspective, there is just a single family that pools its income and resources but that requires two residences, the way some other families might require two cars. 99. Both perspectives have legitimacy and yield insights. But they imply different ways of structuring related analysis, because the second perspective views the remittance as earnings of the Albanian-based recipient household unit, rather than as a transfer from an independent private household abroad. 100. The distinction between the two perspectives is important because contributory pension systems, remittances, and social assistance imply different "coping strategies" on the part of poor people. Pensions represent, in principle, a strategy of insurance, savings, and deferred consumption ­ albeit one that, in practice, is heavily intertwined with taxation and mediated largely by the state. Remittances represent a decision on the part of the family to seek work outside the country that complements domestic earnings and income, because labor demand is insufficient within the country. Social assistance entails a much higher degree of reliance on the state, with transfers redistributing resources from taxpayers to poor or disadvantaged households. 101. For policy makers, it is important to understand how public and private transfers affect consumption levels at the margin if remittances from abroad are treated as part of recipient households' earned income. What impact does that change in perspective have on implied values of the three main indicators of pre-transfer poverty? How much impact would ndihme ekonomike then have in reducing pre-transfer poverty, if the level of pre-transfer poverty is significantly lower because remittances from abroad are now seen as substantially boosting household earnings? In that environment, how important would pensions continue to be in alleviating pre-transfer poverty? Given the prevalence of pensions, what would be the incremental impact on poverty ofndihme ekonomike transfers, once pension payments have been made? Would ndihme ekonomike transfers then be targeted more efficiently, or more inefficiently, towards poor households? This section addresses these questions, and other similar ones. In interpreting data to answer these questions, however, it is prudent to recall that the family and local community are always able to adjust their own transfers to poor households to take account of the structure of any public social assistance and social insurance programs. Although government policymakers might see themselves as determining the final adjustments to the income distribution through their decisions, they do not have the final word. 102. Until this point, Chapter II has estimated the impact of public and private income transfers on the poverty status and consumption levels of people using an estimated pre-transfer level of consumption. This pre-transfer level of "counterfactual" consumption has been calculated by stripping all public and private transfers (including remittances from abroad) from the observed consumption of households, assuming a specific marginal propensity to consume out of transfer income. That approach makes explicit and transparent the sources of all the resources that the family has available to support its consumption and yields insights into the marginal contributions of each transfer income source to household consumption. It is the 55 appropriate way to explore household consumption dynamics if one adopts the first perspective towards remittances. 103. This section, however, studies the impact of public and private transfers on households' poverty status and consumption levels under the second perspective on remittances from abroad. To do this, it defines an alternative counterfactual level of consumption that treats private remittances from abroad as labor income, rather than as a transfer. The analysis retains the assumption that households used a specified proportion of transfer income to augment consumption and continues, for illustrative purposes, to use 100 percent as the specific value for that proportion. However, because remittances are now treated as part of household income, the counterfactual consumption of each household is estimated by subtracting from the observed household's consumption the sum total of all public transfers, private transfers received from other households in Albania, and aid received from NGOs and other private individuals in Albania. Using this new counterfactual, this section considers the coverage, distribution, and target incidence of transfer income with respect to `poor' and `non- poor' households. Table 2.8 summarizes the three indicators of pro-poor targeting of public and private transfers under the alternative counterfactual and the original counterfactual. 104. How does coverage of those in pre-transfer poverty change when, under the new counterfactual, remittances from abroad are treated as earnings of the recipient family? For public transfers, coverage remained at levels similar to those under the original counterfactual. The percentage of pre-transfer poor who received ndihma ekonomike under the new counterfactual was about 2 percentage points more, at 20.7 percent, than under the original counterfactual. Coverage of pensions and other public benefits, however, remained almost the same as before. In relative terms, of all the transfers, pensions ­ both urban and rural ­ are seen to have had the best coverage of poor households. 105. Private remittances from abroad, however, reached only an estimated 17 percent of pre-transfer poor households, as against the earlier estimate of 29 percent. This sharp decline in coverage for private remittances from abroad arises because, under the new counterfactual with its redefinition of pre-transfer income, fewer people were categorized as being in pre-transfer poverty once private remittances from abroad were included as part of the earnings of all recipient households. Of the remaining households that were still below this redefined pre-transfer poverty line, the proportion receiving private remittances from abroad was much smaller than under the original counterfactual, because most who received such remittances were reclassified out of poverty because of the income the remittances were considered to represent. 106. Targeting performance sometimes varied under the new counterfactual. The most notable fall was for private remittances from abroad. Under the new counterfactual, only one of each three recipients of remittances from abroad was deemed poor. This reinforces the observation above that remittances from abroad, once treated as earnings, were themselves responsible for raising many households above the poverty line. With the resulting new distribution of earned income, only a small share of the remaining poor households received remittances from abroad. 56 Table 2.8: Distribution of Beneficiaries and Transfers, by Type of Transfer, Under Original and Alternative Counterfactual Consumption Percentage of poor receiving this Percentage of beneficiaries who are poor Percentage of transfer amount that goes to transfer poor Alternative Counter- Original Original Counter- Original Counter- factual Counter-factual Alternative Counter- factual Alternative Counter- factual Consumption Consumption factual Consumption Consumption factual Consumption Consumption Total Transfers 81.4 83.9 46.3 55.5 37.8 67.2 Total Public 76.0 73.8 49.8 56.1 50.0 56.9 Total Private 25.5 35.8 39.0 63.5 24.0 78.8 Ndihma Ekonomike 20.7 18.6 62.4 65.2 59.8 63.6 Urban 28.8 28.4 47.7 54.4 47.6 55.1 Rural 27.9 27.6 50.4 58.0 45.8 54.8 Other 13.3 12.2 58.2 62.0 57.9 62.3 Pensions Transfers Total Pensions 59.4 58.7 48.6 55.7 48.6 56.0 Unemployment Benefits 1.9 1.7 57.5 59.5 50.3 52.6 Public Maternity Benefits 1.3 1.2 52.9 56.8 53.8 58.4 Social Care 0.6 0.5 59.7 59.7 68.7 68.7 Other Public 3.9 3.8 50.1 57.1 59.3 64.7 Within Albania 5.9 5.1 54.7 54.7 69.3 69.3 From Abroad 17.2 28.7 33.5 64.8 18.9 79.6 ces Remittan All Remittances 23.1 33.7 37.2 63.0 22.7 78.8 Private Transfers Other Private 2.4 2.1 73.1 73.1 77.0 77.0 57 107. The targeting efficiency of virtually all public transfer programs deteriorated somewhat. Presumably, this happened because, when remittances from abroad were redefined as earnings of the recipient household and, in consequence, many recipient families were reclassified as non-poor, some of the families reclassified from "poor" (before transfers) to "non- poor" had also been receiving public transfers. For example, the targeting incidence of pensions dropped by about 7 percentage points under the new counterfactual; and only about half the beneficiaries of urban and rural pensions were found to be poor. However, three-fifths of poor people receive pensions, highlighting how categorical targeting can also be an effective form of targeting when administrative systems are relatively weak. 108. Ndihma ekonomike, with 62 percent of its beneficiaries being poor (down from 65 percent), continued to be the best targeted income transfer except for private transfers from NGOs. 109. Defining remittances from abroad as earnings also changed patterns for which transfers delivered most assistance to people in pre-transfer poverty. Once again, remittances from abroad ceased to be prominent in this role, for the same reasons discussed above. Instead, less than one-fifth of funding from these remittances was delivered to individuals in pre-transfer poverty, compared with four-fifths percent under the original counterfactual. Private transfers from NGOs, private remittances from within Albania, and social care took over as the transfers that delivered the highest proportion of their funding to households in pre-transfer poverty. For transfers and private income flows from all sources, the percentage going to poor individuals dropped from 67 percent to 38 percent. 110. Concerning ndihme ekonomike, however, it is instructive to try to understand why redefining the counterfactual resulted not only in higher estimated coverage of pre-transfer poor people but also in declines in the share of beneficiaries who were poor and in the share of benefits directed to those who were poor. One possible explanation might be that a relatively small proportion of recipient families had members working abroad, even though such families were thereby ineligible for benefits. Given that assumption, when redefining the counterfactual shifted a large group of families with migrant members out of pre-transfer poverty, it excluded from poverty in the process a sizeable group of people who were categorically ineligible for ndihme ekonomike benefits, despite having low income from sources within Albania. Thus, of people who were poor, the proportion covered by ndihme ekonomike rose. However, the people who were ineligible for ndihme ekonomike, but were receiving it nevertheless, and who had been removed from the group designated as poor then, by definition, increased the number of recipients who were non-poor and the volume of transfers received by the non-poor. The bad news for the program, under this possible explanation, is that some ineligible families with migrant members were receiving benefits. The good news is that that number might have been relatively small. The credence attributed to this possible explanation, however, should take into account the extensive underreporting within the LSMS survey of participation in the ndihme ekonomike program. 111. As the marginal impact of each public or private transfer, given the new counterfactual, is of greater interest than the overall impact of these transfers, the new counterfactual consumption is used to calculate the baseline poverty levels and then to measure by how much each of the 58 social transfers would have reduced the poverty rate if it were added back to the household's consumption. Table 2.8, as well as Figures 2.7 and 2.8, summarize the three indicators of pro- poor targeting of total public and private transfers under both the alternative and original counterfactuals. Similarly, the marginal impacts of each social transfer on the poverty gap and on severity of poverty have also been estimated (Figures 2.9 and 2.10). The marginal impact of remittances from abroad is not estimated here, because it is already included in the counterfactual consumption of the households receiving them. 112. The marginal impact of Ndihma Ekonomike on poverty headcount and poverty gap remained small, at about 2 percent and 5 percent respectively, even under the alternative higher counterfactual consumption. This reflected the low coverage and small transfer amount of the social assistance transfer program. Urban pensions reduced poverty by 15.7 percent while rural pensions reduced poverty by 6 percent and, altogether, pensions eliminated over one-quarter of the prevailing poverty. This is higher than the marginal impact of pensions when calculated at the original counterfactual consumption level, where all pensions together reduced only a little more than one-fifth of poverty. It is worth noting here that this comparison involves percentage reductions in poverty and not absolute reductions in poverty. Although the percentage-point reduction in poverty from all pensions combined, under the original counterfactual, was 9.5 percentage points ­ almost equal to the 9.7 percentage point reduction in poverty under the new counterfactual ­ the reduction in poverty, as a percentage of the baseline poverty level, was higher under the new counterfactual than under the original counterfactual. Figure 2.7: Poverty Indicators Before and After Total Transfers , Treating Remittances from Abroad as Earnings of Recipient Household 40.0 37.8 35.0 30.0 25.4 25.0 20.0 15.0 12.1 10.0 5.7 6.7 5.0 1.9 0.0 headcount gap severity pre-transfer poverty post-transfer 59 Figure 2.8: Extreme Poverty Indicators Before and After Total Transfers, Treating Remittances from Abroad as Earnings of Recipient Household 14.0 12.4 12.0 10.0 8.0 6.0 4.7 4.4 4.5 4.0 2.0 0.8 0.2 0.0 headcount gap severity pre-transfer extreme poverty post-transfer extreme poverty Figure 2.9: Impact of Public and Private Transfers on Poverty Indicators, Treating Remittances from Abroad as Income 60.0 Ndihme 55.0 52 Urban pensions 50.0 45.0 poverty Rural pensions 39 in 40.0 37 Other pensions 35.0 30.0 26 Pensions total decrease 25.0 24 Other public 20.0 16 15.0 Pvt transfers (from 9 9 11 11 Albania) Percentage 10.0 8 64 6 7 5 6 5.0 2 Other pvt transfers 1 2 0 2 2 2 (NGOs) 0.0 Headcount Gap Severity 113. A similar pattern is observed with the impact of social transfer on extreme poverty. As expected, the marginal impact of ndihma ekonomike is higher on extreme poverty than it is on poverty. Once again, urban pensions are the most efficient social transfer in reducing extreme poverty headcount, gap and severity than all other public transfers. 60 Figure 2.10: Impact of Public and Private Transfers on Extreme Poverty Indicators, Treating Remittances from Abroad as Income 80.0 72 70.0 Ndihme poverty 59 60 60.0 Urban pensions extreme 50.0 46 Rural pensions in 40 40.0 Other pensions 30.0 26 Pensions total decrease 20 20.0 Other public 14 1210 12 14 9 11 11 Pvt transfers (from 10.0 8 7 7 6 7 Albania) 2 3 3 2 percentage Other pvt transfers 0.0 (NGOs) Headcount Gap Severity G. Efficiency Analysis - Which Transfers Were More Efficient in Terms of Their Cost Per Unit Reduction of the Poverty Gap? 114. Ndihma ekonomike was little more than half as efficient as remittances from abroad in reducing the extreme poverty gap, but it was more efficient than pensions (Figure 2.11, which shows a targeting­efficiency ratio for selected specific transfers). Private remittances from abroad, if viewed as transfers, appear to have been the most efficient, irrespective of the poverty line chosen. While private remittances were able to achieve a one-Lek reduction in the extreme poverty gap by transferring 2.3 Leke, ndihma ekonomike had to transfer 4.1 Leke to achieve the same result. As observed, the difference between the targeting­efficiency ratios of private remittances from abroad and ndihma ekonomike transfers was much narrower for poverty reduction than for reduction of extreme poverty. Unsurprisingly, pensions stood out as inefficient social transfers in reducing the poverty gap, in comparison with other transfers. Pensions cost almost 4 Leke to reduce the poverty gap by 1 Lek, and almost 5.4 Leke to reduce the extreme poverty gap by one Lek. Again, it is important to reiterate that poverty reduction is not the principal aim of some of these transfers, and their efficiency should be judged in that light. 115. To measure the efficiency of social transfers, a targeting­efficiency ratio is calculated as the ratio of the total volume of benefits transferred to the contribution of that transfer in reducing the poverty gap. Due to a lack of data on administrative costs, the cost of transfers is estimated from the self-reported transfer income that each household receives, which is summed for the entire population. In reality, the cost should also include administrative costs and overhead expenses and, in this respect, the results should be interpreted with caution. It is possible that the 61 administrative costs of some transfers may outweigh the marginal benefits from the transfer. Also, social transfers have multiple benefits and should not be judged solely on the basis of their targeting­efficiency ratio. Figure 2.11: Targeting­Efficiency Analysis of Social Transfers Remittances from Cost to Benefit abroad only Ratio for reducing Pensions only Extreme Poverty Gap Ndhima only Cost to Benefit only pvt Ratio for reducing Poverty Gap only public 0 1 2 3 4 5 6 Cost per percentage point reduction of poverty gap (Leks per 1 Lek reduction in poverty gap) H. Overall Effectiveness of Social Transfers in Reducing Poverty - Which Transfers are More Effective Than Others? 116. Overall effectiveness is measured by looking comprehensively at three indicators ­ coverage, absolute target incidence and their contribution to reduction of the poverty gap. All three indicators are plotted on a single graph as follows: the coverage, or percentage of poor households receiving the social transfer, is plotted on the x-axis; the absolute target incidence, or the percentage of the total transfer amount going to poor households, is plotted on the y-axis; while the reduction in total poverty gap (or extreme poverty gap), is depicted by the size of `bubble' in the graph. A `perfectly-targeted program' would therefore be located on the upper- right hand side of these graphs, with a large bubble. Figure 2.12 shows the effectiveness of all social transfers relative to one another. 117. Although ndihma ekonomike disburses a greater percentage of its transfers to poor households, it covers a smaller proportion of the poor households and has a smaller impact on reducing the poverty gap than do urban or rural pensions (Figure 2.12). Urban pensions have a similar coverage to private remittances from abroad (viewed as transfers); however, they are not as effective as private remittances from abroad either in target incidence or in reducing the poverty gap. Rural pensions are similar to urban pensions in their distribution but less than half as effective in terms of reducing the poverty gap. 62 Figure 2.12: Effectiveness of Public and Private Transfers in Reducing Poverty Key Label Remittances from Effectiveness of Social Transfers in reducing Poverty RO abroad 100.0% UP Urban Pensions transfer RP Rural Pensions 80.0% RO P total RA of S NE Ndihme Ekonomike poor O NE OP to 60.0% M UP U RP OP Other Pensions going percentage- Remittances from 40.0% RA within Albania amount O Other Public Transfers Incidence 20.0% M Maternity benefits Target 0.0% 0.0% 20.0% 40.0% 60.0% 80.0% 100.0% U Unemployment benefits Coverage - percentage of poor receiving benefit P Other private transfers S Social Care benefits Note: Size of bubble depicts the relative contribution of social transfer in reducing the poverty gap. 118. It is clearly evident that in terms of coverage, target incidence and contribution to reduction of poverty gap, private remittances from abroad (if viewed as transfers) are the most effective among all transfers. There is no transfer close to the top right corner of the chart, showing that there is a dearth of transfers that can not only target poor or extremely poor households effectively but can also transfer their funds to poor households without leakages. 119. Remittances from abroad are most effective in their coverage of the extremely poor and second only to private transfers from NGOs in their target incidence. They also have by far the greatest impact in reducing the extreme poverty gap (Figure 2.13). Remittances from abroad seem to be the most efficient transfer in helping extremely poor households close the poverty gap, even though only 43 percent of the extremely poor receive such transfers. 120. Ndihma ekonomike is only able to reduce the extreme poverty gap by half a percentage point and is less effective than urban or rural pensions in terms of coverage of the extremely poor. Even in the case of target incidence, ndihma ekonomike is only marginally better than urban and rural pensions at disbursing the transfers to extremely poor households. 63 Figure 2.13. Effectiveness of Public and Private Transfers in Reducing Extreme Poverty Key Label RO Remittances from abroad Effectiveness of Social Transfers in reducing extreme poverty UP Urban Pensions 100.0% RP Rural Pensions total poor of 80.0% NE Ndihme Ekonomike P extreme RO to 60.0% OP Other Pensions S percentage- RA Remittances from going RA 40.0% OP within Albania M Other Public NE RPUP O Incidence amount O 20.0% Transfers U Target transfer M Maternity benefits 0.0% Unemployment 0.0% 20.0% 40.0% 60.0% 80.0% 100.0% U benefits Coverage - percentage of extreme poor receiving benefit P Other private transfers S Social Care benefits Note: Size of bubble depicts the relative contribution of social transfer in reducing the extreme poverty gap. 64 CHAPTER III: DIRECTIONS FOR THE FUTURE 121. Focus. What should policy-makers do? This chapter identifies the big picture for SSN performance ­ its strengths, weaknesses, and problems requiring solution. Probing deeper for each main program area, it asks what constitutes the priority agenda for action. That agenda is based on careful appraisal of key policy design and implementation issues, program performance, and available options. The appraisal process has involved not only the analysis of LSMS data reported in this study. As a complement, it has also entailed four years of cooperative work with Albanian officials from 2000 through 2004, while preparing a series of three Poverty Reduction Strategy Credits, to identify the relevant agenda for improving policy-making and implementation for both social insurance and social assistance. As most public funding to reduce poverty comes through pensions, continuing to achieve an impact on poverty will require not only further improvements in social assistance policies and programs, but also continuation and strengthening of an effective, sustainable pension system. 122. Big Picture. For 15 years, Albania has been experiencing a torrent of turbulent economic and social change. This was unleashed by the rupturing of political barriers that had sought to contain human energy, material resources, and creative imagination within a rigid and anachronistic societal structure that bred mistrust of government and neighbors and that denied an array of human possibilities within the changing world around it. The economic and social change unleashed has entailed a highly disruptive transition that has forced tens of thousands of families to make changes in their occupations, their locations, their social support networks, their level of economic self-sufficiency, and their relationship to government institutions and their new communities. New economic possibilities have brought with them major new economic risks, especially insufficient jobs and the possibility of inadequate incomes. 123. New Social Safety Net Functions. Within this changed and challenging world for families, a new social safety net has emerged through both private intra-family initiatives and complementary public ones. That safety net has multiple components. Of these, the component focused on public income transfers and private income flows has performed moderately well, providing substantial help to a fortunate minority of households and modest help to many others, but none at all to some households in economic hardship. Viewed as a whole, this safety net delivered income transfers from at least one source to five out of six poor households; those beneficiary households received two-thirds of the total funds distributed; and 56 percent of those receiving transfers were poor. One reason why 44 percent of beneficiaries are non-poor is that some important income transfers ­ particularly pensions ­ are not targeted specifically to poor households, although their categorical targeting (to aged or disabled people and survivors in the case of pensions) results in quite high coverage among poor households. 124. Dual Nature of Remittances. Both data analysis and policy-making, however, require recognizing the dual nature of private remittances from abroad. 65 ØFrom one perspective, where social safety nets are viewed broadly as the set of arrangements through which a society meets the consumption needs of its poorest and most vulnerable members, private remittances from abroad can be seen as playing the role of a critically important form of private income transfer. This income source can be dramatically helpful but is also risky. The number of households that expect remittances but that do not receive them is not known. Households whose economically active members live abroad clearly remain vulnerable to low consumption and severe poverty if their emigrant members are unable to find and retain employment that generates significant, regular earnings; or if the earnings are spent abroad, remitted irregularly or in unpredictable amounts, or misappropriated while being repatriated to Albania; or if the members abroad form new primary relationships and create an alternative permanent life in their new country of residence. ØFrom an alternative perspective, in a country like Albania with too few good jobs available, remittances from abroad can be seen as the consequence of households' adopting an economically rational employment strategy that maximizes returns to human capital by sending one or more family members to work abroad where the probability of employment and the expected wage rate are both higher. However, an employment strategy that requires long-term separations from spouses, children, or parents is far more radical than one that involves working in a local store. Except among people with advanced, marketable skills or a pronounced willingness to accept short-term risks or costs for long-term financial gains, an employment strategy of this kind might easily be viewed with anxiety as one that generates major non-economic costs, as well as economic risks, for both the emigrant(s) and those remaining behind. ØAccordingly, this study has framed its analysis using alternative methodologies that capture the different views. § The view of remittances from abroad as private transfers is reflected in analyses that treat remittances symmetrically with all other transfers. These analyses estimate pre- transfer consumption by deducting the aggregate value of all transfers from the observed level of consumption to infer pre-transfer consumption levels (for the convenient illustrative case where all transfers are assumed to be allocated to raising consumption). Because remittances from abroad are typically large, this tends to make estimated pre-transfer consumption low for households receiving such remittances, to raise estimates of the pre-transfer poverty headcount, and to create a large pre-transfer poverty gap. Conversely, these remittances also appear effective at reducing the poverty headcount and the poverty gap on a post-transfer basis, while also substantially reducing poverty severity and inequality. § The view of remittances from abroad as an employment strategy leads to their inclusion as earnings of the Albanian household unit, rather than as transfers of income from abroad. Analyses adopting this perspective do not deduct such remittances when estimating pre-transfer consumption levels. For households with such remittances, this results in higher estimates than before of pre-transfer consumption, lower estimates of all pre-transfer poverty indicators, and smaller estimates of the overall impact of transfers in reducing all indicators of pre-transfer poverty. 66 125. Impacts of Main Income Transfer Programs. Individual components of the safety net varied in their impact on poverty rates and living standards. Ø Private remittances from family members abroad, if viewed as transfers, focused closely on low-income households. § When remittances from abroad are treated as transfers, then 42 percent of their recipients were poor before transfers and 63 percent were extremely poor. These remittances also achieved significant coverage, reaching 29 percent of poor households and 44 percent of extremely poor households. Eighty percent of remittances from abroad went to beneficiaries who were poor. Because their large size eliminated much of the poverty gap as well, they were the safety net's most effective component overall. However, as noted above, the data do not reveal whether they were received regularly by their recipient households, or how many households with relatives abroad received nothing. § When remittances from abroad are treated as earnings of the Albanian family unit, which reduces estimates of pre-transfer poverty, then only 34 percent of their recipients were poor before transfers, the remittances reached only 17 percent of pre- transfer poor households, and 19 percent of these remittances went to beneficiaries in pre-transfer poverty. Ø Public pensions had even broader coverage than private remittances from abroad, reaching 65 percent of extremely poor households and 59 percent of poor households. Reflecting the basis on which they are awarded, however, pensions were less well targeted than remittances. Only 56 percent of pensioners receiving them were poor before transfers and only 27 percent were extremely poor. When remittances from abroad are viewed as income of the recipient household, thereby lowering the overall estimate of pre-transfer poverty, the proportion of pension recipients who are poor falls by several percentage points. Ø Ndihma ekonomike, as a program, was notable for assisting only 19 percent of poor families, impacting quite modestly the number of households in poverty, being no more likely to reach extremely poor households than poor ones, and channeling 36 percent of program funds to non-poor households. Clearly, for Albania's main means- and income- tested antipoverty program, it scores poorly on targeting efficiency, coverage effectiveness, and impact on the incidence of poverty. Consequently, the impact of recent structural and administrative changes in this program requires careful monitoring and evaluation to see whether they are improving targeting and increasing assistance to beneficiary households. Trends in maximum inflation-adjusted benefit levels, eligibility rules defining the target group, and overall inflation-adjusted budget allocations for the program also warrant reexamination. So do possible alternative program models for helping the target group for the program. 126. Large Overall Impact of SSN on Poverty. Based on Chapter 2's methodology and using the headcount measure of poverty, the combined impact of all public and private transfers on poverty was striking. Ø Consider first the case where private remittances from abroad are viewed as transfers, rather than as income of the recipient family. Total transfers raised 18 of every 100 individuals out of poverty, lowering the estimated pre-transfer headcount poverty rate of 43.9 percent to an observed post- 67 transfer poverty rate of 25.4 percent. The number raised out of poverty represented 42 percent of those estimated to have been in pre-transfer poverty. Total transfers raised 14 of every 100 individuals out of extreme poverty, lowering the estimated pre-transfer headcount rate of extreme poverty of 19 percent to an observed post-transfer rate of 4.7 percent. The number raised out of extreme poverty represented 75 percent of those estimated to have been in extreme pre-transfer poverty. Using the poverty-gap measure, total transfers reduced the pre-transfer poverty gap from 20.1 percent of the minimum aggregate income needed to eliminate poverty to a post-transfer gap of 5.7 percent; and it reduced the pre-transfer extreme-poverty gap from 13.4 percent to a post-transfer gap of 0.8 percent. Transfers therefore closed most of the poverty gap and virtually the entire extreme poverty gap. Ø Consider next the case where private remittances from abroad are viewed as income of the recipient family in Albania, rather than as a transfer from members of a separate household abroad. Rates of pre-transfer poverty are lower, when viewed from this perspective, although rates of post-transfer poverty are the same. Total transfers raised 12 of every 100 individuals out of poverty, lowering the estimated pre-transfer headcount poverty rate of 37.8 percent to an observed post- transfer poverty rate of 25.4 percent. The number raised out of poverty represented 33 percent of those estimated to have been in pre-transfer poverty. Total transfers raised 8 of every 100 individuals out of extreme poverty, lowering the estimated pre-transfer headcount rate of extreme poverty of 12.4 percent to an observed post-transfer rate of 4.7 percent. That raised out of extreme poverty 63 percent of those estimated to have been in it before transfers. Total transfers reduced the pre-transfer poverty gap from 12.1 percent of the minimum aggregate income needed to eliminate poverty to a post-transfer gap of 5.7 percent; and it reduced the pre-transfer extreme-poverty gap from 4.4 percent to a post-transfer gap of 0.8 percent. Transfers therefore closed most of the poverty gap and even more of the extreme poverty gap. 127. Post-Transfer Poverty Remains Substantial. Even so, post-transfer poverty rates continue to generate concern, as more than one-quarter of all households are poor and 4.7 percent remain in extreme poverty after transfers. The question that arises is how the Government's overall strategy or its safety net programs could be made more effective. 128. Headline Recommendations. Key recommendations for social insurance and social safety nets are summarized here, with the detailed case for these set out below. (a) Social Insurance Ø It is dangerous to drive a car that lacks steering or brakes. The same is true for social insurance systems. Although Albania's most recent parametric pension reforms were based on competent modeling and simulation by consultants, that capacity is no longer regularly available. Given known uncertainties, it is imperative that Albania develop and retain the capacity (including trained staff or a long-term consulting contract) to undertake skilled long-term modeling and simulation of the social insurance system. This capacity should be directed urgently (i) to reassess predicted 68 outcomes under current policy parameters; (ii) to adopt a conservative indexation rule for pensions already awarded, in combination with any further parametric changes needed to ensure long-run fiscal balance that permits a gradual reduction in total payroll contribution rates; (iii) to plan coordinated changes in pensions policy and health care financing reform; and (iv) to achieve a transparent new configuration of social insurance contribution rates and other financing sources for the various branches of social insurance that equates the revenues and expenditures of each branch. Pending the completion of these tasks, there should be no change ­ up or down ­ in the total payroll tax contribution rate or in the allocation of this revenue stream between pension and health functions. Ø To supplement this capacity-building initiative for policy development, there should be a parallel institutional-strengthening package for SII and MoLSA that (i) redefines policy-making processes and responsibilities and implements the necessary changes and capacity building activities; (ii) reviews and reengineers operating procedures within SII and related responsibilities, as the basis for defining and operationalizing new processes of information management, financial management, and business transactions that strengthen existing systems and that assure fiscal and informational integrity and operational efficiency through systems that incorporate independent cross-checks. (b) Social Safety Net Ø Albania lacks good ways to tell what impact its social assistance programs are having and whether numerous policy and administrative changes that it is making are helping. To supplement analysis of occasional household surveys, it should establish, and make operationally effective, a monitoring and evaluation system that measures, assesses, and, where appropriate, audits: § the impact of the changes in national and local government responsibilities for cash social assistance and social service programs, including the ndihma ekonomike program; § the piloting and later introduction of work-availability requirements within this program; and § the accuracy and effectiveness with which individual local governments implement and administer national program rules, including those about program eligibility and allocation of funding for cash social assistance and social service programs. Ø The ndihma ekonomike program is designed to assist the poorest families who lack other resources. Of the main income transfers within Albania's social safety net, the ndihma ekonomike program is the least effective at reaching poor individuals, raising them above the poverty line, and channeling funding only to those who qualify and are poor. Declining real budgets, linkages to unemployment benefit rates that have been declining in real value, inconsistent program administration, and unclear work incentives are among the factors apparently contributing. Accordingly, the Ministry of Labor and Social Affairs (MoLSA) should review: (i) maximum benefit levels and 69 real growth rates of benefits and program funding levels within cash social assistance and other social safety net programs, including linkages between parameters of these and other programs such as unemployment benefit; (ii) the appropriateness of indexing or routinely adjusting these benefit levels; and (iii) the relationship between benefit levels and small amounts of private income within these programs. To improve targeting of cash social assistance for low-income households, MoLSA should also consider alternative or complementary measures (e.g., conditional cash transfers linked to school attendance, or proxy means-testing with community assessment) as well as possibilities for piloting these. 129. Key Design Issues in Pensions. Albania's economic and demographic circumstances signal that several key parameters of the pension system are sources of concern. Ø In any demographically young country, a ratio of pensioners to contributors (dependency ratio) that is close to unity is cause for serious concern about fiscal sustainability. Past modeling has suggested that, with disciplined policy choices, Albania could circumvent this challenge. As many assumptions for modeling are themselves highly uncertain, however, the need is urgent to create an internal capacity to update modeling results regularly. Recognition among all policymakers of the discipline required in framing pensions policy is also vital. One illustrative policy issue is the need to limit increases in pensions, once awarded, to no more than indexation based on consumer prices. Ø Among the uncertainties are future patterns of emigration, return migration, fertility, life expectancy, and coverage of the workforce as registered contributors to social insurance (and later as its beneficiaries). These imply a need both for relevant research and for policy prudence. Ø In such circumstances in a relatively low-income country, policymakers would usually consider adopting a flat-rate pension system rather than one with earnings- linked benefits. The benefit structure in Albania's pension system for employees and urban self-employed workers, however, has a maximum benefit provision set at three times the minimum benefit. Consequently, the pension system functions in many respects like a flat-rate system with a more compressed distribution of pension benefits than of either earnings or contributions. Ø Albania works hard to register workers in the informal sector within its social insurance, but it has had only limited success. This raises concerns not only about the adequacy of current system revenues, but also about the long-term level of pension coverage among future retirees, given that work history prior to the transition was counted as contributory service for the current system. A related concern is that a relatively small share of contributors is from the private formal sector. Ultimately, through taxation to finance general government expenditures, this sector must bear indirectly much of the burden of government contributions on behalf of public sector employees. Ø The pension scheme for self-employed rural workers currently has low contributions, pays low benefits, and is not fully transparent. Much of what is deemed to be workers' contributions is ultimately financed from the national budget sector, so even the small benefits are not fully financed by contributors. The government would like to raise both benefit and contribution levels over time so that, to the extent feasible, 70 they come to match those in the pension scheme for urban self employed workers. It has initiated this process. However, the strategy runs the risk that rural self-employed workers will drop out of the rural pension scheme completely and the further risk that the rural pension system plays a de facto social assistance function in rural areas comparable to that of ndihma ekonomike in the urban areas. Thus, this policy could ultimately undercut the current social assistance function of the rural pension program. Evidence reported in Chapter II suggests that rural pensions, and also urban pensions, do indeed function like a social assistance program in rural areas. 130. Key Institutional Issues in Pensions. Although the Social Insurance Institute has managed the successful development of a social insurance system based broadly on a western European model, there remains a series of important issues and concerns relating to improving future implementation of the system. Ø Long-term modeling and simulation work for the most recent set of parametric changes in pensions was undertaken by the Institute for Social Research, based in Milan, on a consultant basis. The government now needs to develop an internal and bureaucratically sustainable capability to undertake pension modeling and policy simulation, either directly or through ongoing and consistent consultant support from a stable group of advisers. Ø Linked to this issue is the intention of the Ministry of Labor and Social Affairs to strengthen and restructure its policy development processes for all aspects of social policy within its jurisdiction. It envisages that pensions policy development would be an important part of this policy-making unit. A key issue is where the pension modeling capability will be located and what the respective roles of MoLSA and SII (and their staffs) will be in relation to this function. A further issue is how the sophisticated skills required will be further developed through training and then retained over time as staff change. Ø Although it manages financial flows of the order of 5 percent of GDP, SII has not had the resources to modernize its business processes, and thereby increase the procedural and fiduciary safeguards within its operations, to the extent that would be desirable. Most regional office transactions are processed in paper mode, with summary information being transmitted electronically between regional offices and the national office. Various initiatives are underway or planned to modernize office procedures, but additional resources would be required to refine and extend these plans and to implement them quickly. Ø SII and the General Tax Directorate (GTD) of Ministry of Finance (MoF) are jointly undertaking the partial transfer of responsibility for collecting payroll-based social insurance contributions from SII to GTD. Clarifying the long-term collection processes that will be used by SII for collection in rural areas, as well as other aspects of the remaining transfer of responsibilities, is an important task. 131. Action Agenda for Pensions Policy and Institutional Strengthening. Desirable actions include the following. Ø Development of a sustainable internal capability to update and maintain one or more computer models of the Albanian social insurance system, to use these tools to project 71 financially the operations of the current pension system over 30-50 years, and to model and simulate alternative policy options; Ø Development of a clear and sustainable division of responsibilities between MoLSA and SII concerning responsibilities and processes for policy development in both social insurance and social safety net aspects of social security; Ø Systematic institutional strengthening of SII focusing on HQ/regional office interrelationships, re-engineering (and selective automation) of most administrative processes, and solidifying the financial management and control of funding flows equivalent to 5 percent of GDP; Ø Completion and evaluation of the pilot for automated processing of benefit payments in the Elbasan region; Ø Completion of the planned phased and partial transfer of collection responsibilities to the General Tax Directorate of MoF and strengthening of SII's capacity to undertake its residual collections role in the rural sector and informal economy; and Ø Annual updates of long-term pension projections and triennial projections and simulations of major policy options or issues, with future publication within SII's planned triennial report series. 132. Key Design Issues in Social Safety Nets. These issues include the following. Ø Evaluating the ndihma ekonomike program requires knowing what proportion of poor families are ineligible and, more particularly, whether those poor families that do not receive benefits are indeed ineligible. If so, does the exclusion of many poor families result legitimately from the national program criteria, or illegitimately from differences between national legislative criteria and local-governments' de facto administrative criteria? Alternatively, is the problem underlying non-receipt of ndihma ekonomike by many poor families simply that the program's scale is inadequate? In that case, would raising the program's budget substantially increase its coverage of poor households ­ especially those that are extremely poor? Ø It is unclear if families in peri-urban areas get ndihma ekonomike benefits or not. Reports on the status of provision of cash social assistance there from residents and program administrators are not consistent. Ø Rules and practice for determining social assistance benefit levels for families with small amounts of income should be clear. It is currently unclear how the ndihma ekonomike benefit is affected if an otherwise-eligible family has a small amount of actual or imputed income that is less than the relevant ndihma ekonomike benefit. In these cases, is the family excluded from eligibility altogether, is the benefit reduced by the full amount of the income earned, or is the benefit reduced by only a fraction of the income received (so as to create an incentive for recipients to seek additional income)? It is possible that each approach might happen in different areas. If a family in these circumstances is deemed ineligible for any ndihma ekonomike benefit, then that family would also appear to be unable to benefit under the work-testing provisions that assist families qualifying for ndihma ekonomike to generate earned income that is greater than the ndihma ekonomike alone would have been. There are important equity and incentives issues raised by this issue. Ø Do families receiving ndihma ekonomike face appropriate incentives to work? Balancing the desire to provide adequate income support with a concern to maintain 72 adequate work incentives is a challenge for all social assistance programs. Providing public benefits reduces the direct pressure on families to pursue all available sources of private income or assistance. Moreover, denying benefits if any earnings are received, or reducing benefits to fully offset increases in earnings, can also reduce correspondingly the incentive to work. Albania has defined a partial benefit ceiling under the ndihma ekonomike program for low-income families that have land or other potentially productive resources. This is equivalent to imputing to such families income from those resources. Families migrating to peri-urban areas near the cities are affected by this provision. As noted above, a further possible response to this inherent dilemma would be to reduce benefits by only a proportion of any increase in earnings. Such a reduction would impose an implicit tax rate on earnings up to the ceiling level of income for receipt of any benefits. (Lithuania used this approach in the early 1990s; and the US foodstamps program also does so.) It is currently unclear whether income is offset fully against benefits or whether having any income results in exclusion from the program. Both might happen in different areas. Other countries have responded by imposing work requirements on some or all recipients as a condition for receiving benefits, as Albania is planning to do. (Highly visible US welfare reforms in 1996 combined this approach with job training, subsidies for child and dependent care, retention of health insurance assistance benefits, and a ceiling on duration of benefits.)16 Ø Other possible changes in structure could relate to eligibility rules, the targeting mechanism (measured income or consumption vs. proxies or both in combination), compliance with monitoring systems, and the program's response to internal migration. 133. Key Institutional Issues in Social Safety Nets. There is an important agenda for institutional strengthening in the social safety net arena. Ø The Law on Social Assistance and Services envisages that social services will be provided by private or not-for-profit organizations and also by State Social Services. In principle, it is desirable to structure the responsibilities of public bodies so that no single public body is responsible both for providing services and for regulating the provision of such services at the same time. This principle is intended to maintain transparency in public administration and to avoid conflicts of interest in the roles that public bodies play. It avoids, for example, situations where a public body (a) must regulate itself, or (b) could be seen as a competitor of the service-providing organizations that it regulates, or (c) defines the service standards to which it ­ along with other service providers ­ is subject, or (d) determines the areas in which funding will be made available for new services that it might itself provide. Accordingly, greater clarity would be helpful in relation to the respective current and future roles of MoLSA and State Social Services for the provision and regulation of publicly provided social services. Ø Service standards should be developed for the delivery of cash social assistance benefits, where local governments' performance both in including eligible families within the program, and excluding ineligible ones, needs to be assessed 16United States Congress. Personal Responsibility and Work Opportunity Reconciliation Act of 1996, which replaced the old welfare system with a new program, Temporary Assistance to Needy Families (TANF). 73 systematically against nationally legislated criteria. In each case, systems to measure and record the information necessary for monitoring these standards would be required. That information would need to be collected and recorded for applicants denied authorization for assistance or services, as well as for those approved for assistance or services. Ø To assess the impact of introducing new arrangements within the ndihma ekonomike program, monitoring indicators should be systematically collected in relation to: eligibility changes and their incidence; household income changes; community work program inputs and outputs; patterns of employment of local government staff; program efficiency and effectiveness indicators; views of social assistance and community work among program participants; and geographic variation across local government areas (computed at national level). 134. Action Agenda for Social Safety Net Policy and Institutional Strengthening. Desirable actions include the following. Ø Develop improved policy development structures and processes, as noted above for pensions. Ø Clarify, through nationally specified criteria, (a) which people within families otherwise eligible to receive ndihma ekonomike benefits will be expected to work and which will not, (b) what limits will be placed on the kinds and duration of work they may be asked to do and on the minimum acceptable working conditions, and (c) how the ndihma ekonomike benefit would be affected if an otherwise-eligible family has a small amount of income that is less than the relevant ndihma ekonomike benefit. Ø Seek ways to improve the ease and efficiency of arrangements for Albanians abroad to send remittances back to their families. Ø Improve systems for evaluating the impact of planned changes in ndihma ekonomike in the nine planned pilot areas and for monitoring the continuing functioning of this and other programs as local governments assume a larger, more complex, and more discretionary role. Require early definition and regular measurement of a range of indicators that monitor the broad impact of the new policies and proposed law in each local government area. Frame policy development for social assistance jointly with policy development for labor markets in a high-migration environment. Ø Coordinate social assistance policies with developments in the financing of health care and/or health insurance as they affect poor households via health insurance contributions or health care co-payments. Ø Examine the appropriateness of linking receipt of new or existing social assistance payments to educational enrollment of children, or health services for children or pregnant women, once access to satisfactory schooling and health services is geographically convenient throughout the country. Ø Examine the potential of proxy means-testing linked to community review or assessment as a complement or alternative to targeted assistance through the ndihma ekonomike program. Ø Require the Ministry of Labor and Social Affairs (MoLSA) to set and implement service standards for assessing how well local governments, when delivering cash social assistance benefits, conform with nationally legislated criteria, including those 74 relating both to inclusion of eligible families within the program and exclusion of ineligible families. Ø Accelerate progress in development of community-based social services and development of coordinated packages of cash assistance and community services for households requiring this. Ø Monitor the planned roles of State Social Services and other public bodies so that no single public body is responsible both for providing services and for regulating the provision of such services at the same time. 75 APPENDIX 1: WHAT CONSTITUTES A SOCIAL SAFETY NET? 135. What Constitutes a Social Safety Net for a Market Economy? Within market- oriented economies, a social safety net (SSN) is a network of programs and support systems that helps people who are poor or vulnerable or falling into socioeconomic difficulties and who would otherwise have what society judges to be unacceptably low access to resources and services. The programs and support systems can be publicly or privately sponsored. Typically, an SSN has components reflecting up to three different emphases. An SSN can include elements of: (i) a net to catch people when crises overwhelm their usual support systems and their ability to meet basic needs falls to critically low levels; (ii) an austere ongoing income floor for families with inadequate resources; and (iii) risk-reducing social programs designed to reduce the likelihood that vulnerable families will experience crises that have economically or socially debilitating effects. 136. Diverse Components of SSNs. SSN programs vary in their time orientation, their approaches, the forms of assistance they offer, and the types of organizations sponsoring them. For example, to catch people falling into difficulties, nongovernmental organizations could offer a net through soup kitchens for hungry individuals, emergency shelters for abused women and children or for families evicted from housing, coal for those without winter heating, or free medical referral programs for mothers without access to health care for a sick child. Government social assistance programs, like the ndihma ekonomike program, can offer an austere ongoing income floor to families with inadequate resources and no short-term opportunities to increase their incomes, or they can link receipt of such income transfers to participation in work or training activities. Relatives can support one another through cash remittances from family members abroad or through combining households to exploit the economies of scale available to larger households. Social insurance programs can provide planned longer-term consumption support for former employees who are retired, disabled, unemployed, or having a baby through programs that smooth consumption over the life cycle by shifting income from periods of employment to periods without employment. Publicly or charitably sponsored social service programs can provide delivered meals or home help that enable a frail elderly person or someone with disabilities to remain in her home rather than move into an unfamiliar and expensive institutional setting. Related programs (including respite care) can support relatives who act as ongoing caregivers for such frail aged or disabled people so that the caregivers can sustain this demanding but valuable social role. Alternatively, social service programs might help a family in crisis develop solutions to financial, child care, or health care needs. However, creating an adequate, coordinated, and financially sustainable system of SSN programs that does not undercut incentives for self-help and family help is difficult for any country. Ultimately, it is the cumulative impact of all of these programs on families' economic circumstances that matters. 137. Sources of Assistance Within SSNs. Social safety nets include assistance from three distinct sources. One is private individuals, particularly friends and extended family of the people being assisted. Intrafamily transfers are the most basic form of safety net and typically are the most important safety net in countries with very low 76 incomes. A second source is private organizations that offer programs, goods and services, discounts or assistance to poor or vulnerable people either as their primary function or as a secondary function. These can be charitable, religious, community- based, self-help, or other not-for-profit organizations, as well as private for-profit organizations. The third source of assistance is government agencies, whether at national, state or local level. Policymakers responsible for social safety nets must recognize that providing government assistance to someone can potentially induce changes in the assistance that person receives from private individuals and organizations. Consequently, the "targeting" impact that policymakers wish to achieve through a program can potentially be offset by changes in private assistance flows that the program benefits induce. Stated differently, private assistance can potentially determine the distribution, at the margin, of aggregate assistance through safety nets. For example, extended families might rechannel their transfers towards family members who do not receive assistance from public programs. For this reason, the distributional impact of a government social safety net program need not be identical to the distribution of assistance across households. 138. Complementary Programs. The public components of a social safety net draw heavily on social security programs. However, they can also incorporate assistance from the much broader range of public policies ­ for example, those towards economic management, good governance and political accountability, taxation, employment, health insurance or health care subsidies, other price subsidies, training, social services, housing, child care, and access to credit. At the broadest level, social safety net programs can be seen as one aspect of systems of social risk management (Holzmann and Jorgensen 1999).17 This three-part conceptual framework distinguishes policies to lower the probability that people experience various risks, policies to mitigate (i.e., reduce) the impact of those risks if they are experienced, and policies to help people cope with the impact of the risk when it does occur. For example, ...labor market policies reduce the risk individuals face by fostering a flexible formal labor market that rewards labor productivity and allow labor to be allocated to its most efficient use. Similarly, active labor market policies include a variety of measures to retrain workers in accordance with labor market demands. Second, pensions and social insurance schemes--such as affordable unemployment insurance--allow individuals to mitigate risk by smoothing consumption over their lifetimes. Finally, social assistance-- including cash transfers (with built-in work incentives), in-kind benefits and services, and subsidies--are risk coping mechanisms that allow the state to reduce the impact of a negative income shock or persistent low income by providing an affordable standard of living for the poor.18 17 Holzmann, R., and S. Jorgensen. 1999a. "Social Protection as Social Risk Management: Conceptual Underpinnings for the Social protection Sector Strategy Paper". Social Protection Discussion Paper No. 9904. World Bank, Washington, D.C. 18 Social Protection Team. Balancing Protection and Opportunity: A Strategy for Social Protection in Transition Economies (World Bank, 2003), p.2 77 APPENDIX 2: IMPACT ON POVERTY HEADCOUNT OF ASSUMED MARGINAL PROPENSITY TO CONSUME FROM TRANSFERS Table A2.1 shows the baseline poverty headcount assuming all transfers are removed from the household's income using different marginal propensities to consume out of transfer income (MPC). The table then presents the marginal impact of each social transfer assuming that that particular transfer is added back to the household using the same level of MPC. For example, at 50 percent marginal propensity to consume, the baseline poverty headcount, assuming all transfers are removed, would be 34.4%. If only remittances from abroad are added back to the household's income, the poverty headcount would decline to 30.6 percent, that is, a 3.8 percentage point reduction in poverty. The same methodology is applied to all subsequent tables in this appendix. Table A2.1: Impact on Poverty All households Marginal propensity to consume out of transfer income 25 50 75 100 Total Transfers 29.7% 34.4% 39.0% 43.9% Total Public 27.4% 29.2% 30.9% 32.5% Total Private 27.6% 30.2% 33.4% 37.2% Ndihma Ekonomike 29.5% 34.0% 38.4% 43.0% Urban 28.5% 31.7% 35.0% 37.9% Rural 29.2% 33.2% 37.2% 41.7% Pensions Other 29.4% 33.7% 37.9% 42.5% Transfers Total Pensions 27.6% 29.9% 32.3% 34.4% Unemployment Benefits 29.6% 34.2% 38.8% 43.6% Public Maternity Benefits 29.6% 34.3% 38.9% 43.7% Social Care 29.7% 34.4% 39.0% 43.8% Other Public 29.6% 34.1% 38.7% 43.4% Within Albania 29.5% 34.0% 38.6% 43.3% Remittances From Abroad 27.9% 30.6% 34.0% 37.8% Private All Remittances 27.7% 30.2% 33.6% 37.2% Transfers Other Pvt. 29.5% 34.3% 38.9% 43.8% 78 Table A2.2: Impact on Extreme Poverty All households Marginal propensity to consume out of transfer income 25 50 75 100 Total Transfers 6.7% 10.8% 14.6% 19.0% Total Public 5.6% 7.5% 9.3% 10.7% Total Private 5.6% 7.4% 9.1% 11.2% Ndihma Ekonomike 6.5% 10.2% 14.0% 18.0% Urban 6.5% 9.4% 12.0% 15.0% Rural 6.4% 10.0% 13.4% 17.3% Pensions Other 6.6% 10.2% 13.7% 17.6% Transfers Total Pensions 5.9% 8.2% 10.2% 12.2% Unemployment Benefits 6.7% 10.8% 14.4% 18.8% Public Maternity Benefits 6.7% 10.8% 14.6% 18.9% Social Care 6.7% 10.8% 14.6% 19.0% Other Public 6.7% 10.8% 14.5% 18.7% Within Albania 6.7% 10.6% 14.3% 18.2% Remittances From Abroad 5.6% 7.7% 9.7% 12.4% Private All Remittances 5.6% 7.5% 9.3% 11.5% Transfers Other Pvt. 6.7% 10.7% 14.4% 18.7% Table A2.3: Impact on Poverty Urban Households only Marginal propensity to consume out of transfer income 25 50 75 100 Total Transfers 23.8% 28.6% 34.5% 40.0% Total Public 20.9% 22.2% 23.8% 25.3% Total Private 22.3% 25.2% 29.3% 34.1% Ndihma Ekonomike 23.4% 28.2% 33.6% 38.7% Urban 21.4% 23.6% 27.1% 29.6% Rural 23.8% 28.4% 34.1% 39.4% Pensions Other 23.6% 28.3% 33.4% 38.6% Transfers Total Pensions 21.1% 23.2% 25.8% 28.1% Unemployment Benefits 23.7% 28.5% 34.2% 39.6% Public Maternity Benefits 23.8% 28.5% 34.3% 39.7% Social Care 23.8% 28.6% 34.5% 40.0% Other Public 23.6% 28.2% 34.0% 39.2% Within Albania 23.6% 28.0% 33.8% 39.2% Remittances From Abroad 22.5% 25.8% 30.1% 34.9% Private Transfers All Remittances 22.3% 25.3% 29.4% 34.1% Other Pvt. 23.8% 28.6% 34.4% 39.9% 79 Table A2.4: Impact on Extreme Poverty Urban Households Only Marginal propensity to consume out of transfer income 25 50 75 100 Total Transfers 6.1% 9.7% 13.3% 18.2% Total Public 5.3% 6.3% 7.7% 8.8% Total Private 5.0% 6.7% 8.7% 11.5% Ndihma Ekonomike 6.0% 9.1% 12.7% 17.2% Urban 5.7% 7.5% 9.5% 11.8% Rural 6.0% 9.3% 13.1% 17.8% Pensions Other 6.0% 9.5% 12.6% 17.0% Transfers Total Pensions 5.4% 6.9% 8.5% 10.6% Unemployment Benefits 6.1% 9.6% 13.2% 18.1% Public Maternity Benefits 6.1% 9.6% 13.3% 18.1% Social Care 6.1% 9.6% 13.2% 18.1% Other Public 6.1% 9.7% 13.0% 17.9% Within Albania 6.0% 9.3% 12.6% 17.1% Remittances From Abroad 5.0% 7.1% 9.5% 12.6% Private Transfers All Remittances 5.0% 6.8% 8.8% 11.5% Other Pvt. 6.1% 9.6% 13.3% 18.2% Table A2.5: Impact on Poverty Rural Households Only Marginal propensity to consume out of transfer income 25 50 75 100 Total Transfers 33.9% 38.5% 42.3% 46.7% Total Public 32.0% 34.2% 36.0% 37.7% Total Private 31.4% 33.7% 36.3% 39.4% Ndihma Ekonomike 33.9% 38.2% 41.7% 46.1% Urban 33.6% 37.4% 40.7% 43.8% Rural 33.2% 36.6% 39.5% 43.4% Pensions Other 33.6% 37.6% 41.2% 45.4% Transfers Total Pensions 32.3% 34.7% 36.9% 39.0% Unemployment Benefits 33.7% 38.3% 42.1% 46.3% Public Maternity Benefits 33.8% 38.4% 42.2% 46.6% Social Care 33.9% 38.5% 42.2% 46.6% Other Public 33.9% 38.4% 42.1% 46.4% Within Albania 33.8% 38.3% 42.1% 46.3% Remittances From Abroad 31.7% 34.0% 36.8% 39.9% Private Transfers All Remittances 31.6% 33.8% 36.6% 39.5% Other Pvt. 33.7% 38.4% 42.1% 46.6% 80 Table A2.6: Impact on Extreme Poverty Rural Households Only Marginal propensity to consume out of transfer income 25 50 75 100 Total Transfers 7.2% 11.6% 15.6% 19.6% Total Public 5.8% 8.4% 10.4% 12.1% Total Private 6.0% 7.8% 9.4% 11.0% Ndihma Ekonomike 6.9% 11.0% 14.8% 18.5% Urban 7.0% 10.8% 13.9% 17.4% Rural 6.6% 10.5% 13.7% 16.8% Pensions Other 7.0% 10.7% 14.4% 18.1% Transfers Total Pensions 6.2% 9.1% 11.5% 13.3% Unemployment Benefits 7.1% 11.6% 15.4% 19.3% Public Maternity Benefits 7.2% 11.6% 15.6% 19.5% Social Care 7.2% 11.6% 15.6% 19.6% Other Public 7.2% 11.6% 15.5% 19.3% Within Albania 7.2% 11.5% 15.5% 18.9% Remittances From Abroad 6.0% 8.2% 9.8% 12.2% Private Transfers All Remittances 6.0% 8.0% 9.7% 11.5% Other Pvt. 7.2% 11.4% 15.3% 19.0% 81 APPENDIX 3: IMPACT OF INCOME TRANSFERS ON POST-TRANSFER RATES OF POVERTY This appendix presents the results of analyses of baseline rates of Albanian post-transfer poverty and extreme poverty in 2002 and the impact on these baseline rates of income transfers, considered singly or in groups, by alternative assumed marginal propensities to consume from transfers. The results are drawn from work done for Albania: Poverty Assessment (World Bank, 1993). Table A3.1. Estimated Impact of income transfers on post-transfer rates of poverty, if specified sources of transfer income had not Public Transfers Private Transfers Pensions Remittances BaselinePovertyRateandEstimated Total Total Total Ndihme Unemp Maternity Social Other Other Impact Transfers Public Private All Ekonomike Urban Rural Other Total Benefits Benefits Care PublicWithin From Remitta Pvt. Pensions Albania Abroad nces Baselinepost-transferpovertyrate(%) 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 25.4 Impactassumingconsumptionincreasedby100percentofTransfers Pre-TransferPovertyRate(%) 43.9 37.2 32.5 26.5 31.2 27.5 27.0 35.1 25.7 25.6 25.5 25.7 26.1 31.6 32.4 25.5 %PointReduction 18.5 11.8 7.1 1.1 5.8 2.1 1.6 9.8 0.3 0.2 0.1 0.3 0.7 6.3 7.0 0.2 % of implied pre-transfer poverty rate 42.2 31.7 21.9 4.0 18.6 7.6 6.1 27.7 1.0 0.7 0.5 1.2 2.7 19.8 21.5 0.6 Impactassumingconsumptionincreasedby75percentofTransfers Pre-TransferPovertyRate(%) 39.0 33.4 30.9 26.1 29.0 26.8 26.7 32.0 25.6 25.4 25.5 25.6 25.9 30.3 30.8 25.5 %PointReduction 13.6 8.0 5.5 0.7 3.6 1.4 1.3 6.6 0.2 0.0 0.1 0.2 0.5 4.9 5.4 0.1 % of implied pre-transfer poverty rate 35.0 24.0 17.9 2.8 12.4 5.1 4.8 20.7 0.9 0.2 0.5 0.7 1.9 16.2 17.5 0.6 Impactassumingconsumptionincreasedby50percentofTransfers Pre-TransferPovertyRate(%) 34.4 30.2 29.2 25.7 27.7 26.3 26.1 29.3 25.5 25.4 25.5 25.5 25.8 28.7 29.0 25.5 %PointReduction 9.0 4.8 3.8 0.3 2.3 0.9 0.7 3.9 0.1 0.0 0.1 0.1 0.4 3.3 3.7 0.1 % of implied pre-transfer poverty rate 26.1 15.8 13.0 1.3 8.2 3.4 2.8 13.4 0.5 0.2 0.3 0.5 1.4 11.4 12.6 0.6 Impactassumingconsumptionincreasedby25percentofTransfers Pre-TransferPovertyRate(%) 29.7 27.6 27.4 25.5 26.5 25.6 25.6 27.1 25.5 25.4 25.4 25.4 25.6 27.0 27.2 25.5 %PointReduction 4.3 2.2 2.0 0.1 1.1 0.2 0.3 1.7 0.1 0.0 0.0 0.0 0.2 1.6 1.8 0.1 % of implied pre-transfer poverty rate 14.5 7.9 7.2 0.6 4.2 1.0 1.0 6.3 0.5 0.0 0.0 0.2 0.7 6.1 6.7 0.6 existed, by sources of transfer and assumed share of transfer used to increase consumption, for all households in 2002 82 Table A3.2. Estimated impact of income transfers on post-transfer rates of extreme poverty, if specified sources of transfer income had not existed, by source of transfer and assumed share of transfers used to increase consumption, for all households in 2002 Public Transfers Private Transfers BaselineExtremePovertyRateand Total Total Total Pensions Remittances Ndihme Unemp Maternity Social Other Other EstimatedImpact Transfers Public Private Total Within From All Ekonomike Urban Rural Other Benefits Benefits Care Public Remitta Pvt. Pensions Albania Abroad nces Baselinepost-transferextremepoverty 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 4.7 rate(%) Impactassumingconsumptionincreasedby100percentofTransfers Pre-TransferPovertyRate(%) 19.0 11.2 10.7 5.5 7.7 5.8 5.6 10.0 4.9 4.8 4.8 4.8 5.1 10.2 10.6 4.9 %PointReduction 14.3 6.5 6.0 0.8 3.0 1.1 0.9 5.3 0.2 0.1 0.1 0.1 0.4 5.5 5.9 0.2 %ofimpliedpre-transferpovertyrate 75.3 58.0 56.2 14.3 39.0 19.5 16.0 52.8 4.4 1.8 1.7 3.0 8.7 53.9 55.7 3.5 Impactassumingconsumptionincreasedby75percentofTransfers Pre-TransferPovertyRate(%) 14.6 9.1 9.3 5.3 6.5 5.5 5.4 8.1 4.9 4.7 4.8 4.8 5.0 8.9 9.2 4.9 %PointReduction 9.9 4.4 4.6 0.6 1.8 0.8 0.7 3.4 0.2 0.0 0.1 0.1 0.3 4.2 4.5 0.2 %ofimpliedpre-transferpovertyrate 67.9 48.4 49.5 11.7 27.7 15.2 12.2 42.2 4.3 1.0 1.4 1.4 5.5 47.4 48.7 3.5 Impactassumingconsumptionincreasedby50percentofTransfers Pre-TransferPovertyRate(%) 10.8 7.4 7.5 5.3 5.5 5.3 5.1 6.6 4.8 4.7 4.7 4.8 4.9 7.2 7.4 4.9 %PointReduction 6.1 2.7 2.8 0.6 0.8 0.6 0.4 1.9 0.1 0.0 0.0 0.1 0.2 2.5 2.7 0.2 %ofimpliedpre-transferpovertyrate 56.6 36.1 37.4 10.7 13.9 11.3 7.5 28.5 2.2 1.0 0.9 1.4 4.3 34.8 36.3 3.5 Impactassumingconsumptionincreasedby25percentofTransfers Pre-TransferPovertyRate(%) 6.7 5.6 5.6 5.0 4.9 4.9 4.9 5.3 4.8 4.7 4.7 4.7 4.8 5.6 5.6 4.7 %PointReduction 2.0 0.9 0.9 0.3 0.2 0.2 0.2 0.6 0.1 0.0 0.0 0.0 0.1 0.9 0.9 0.0 %ofimpliedpre-transferpovertyrate 30.2 15.5 16.2 6.7 3.8 4.6 4.6 11.3 1.4 0.9 0.9 0.9 1.8 15.5 16.2 0.9 83 APPENDIX 4: POVERTY PROFILE Average household size in Albania is 4.3 members per household (Table A4.1). The average household size is 5.7 among poor households compared with 4 among non-poor households. The dependency ratio is consistently higher among the poor households, irrespective of location. Household heads in poor households are on average 3.6 years younger than their non-poor counterparts, and possess only 6 years of schooling, compared to 8 years among the non-poor. With respect to location, poverty in Tirana city is 17.8 percent as compared to 29.6 percent in rural areas. 19 Table A4.1: Main Characteristics of Poor Households Poor Non-poor Total Tirana Urban Rural Total Tirana Urb Rur Tot Tir Urb Rur Tot Mean H'hold size 5.3 5.4 5.8 5.7 3.6 3.7 4.2 4.0 3.8 4.0 4.6 4.3 % 60 and over 13.2 9.7 10.0 10.2 19.4 19.5 19.9 19.7 18.6 18.1 17.6 17.8 No. aged < 15 yrs 1.8 1.9 2.2 2.1 0.8 0.9 1.2 1.0 0.9 1.1 1.4 1.2 Dependency ratio 1.04 1.06 1.16 1.12 0.66 0.73 0.83 0.77 0.71 0.78 0.91 0.84 % Female- headed h'hold 10.5 12.1 8.1 9.3 18.0 13.5 11.6 13.1 17.0 13.3 10.7 12.4 Age h'hold head 49.1 49.0 47.8 48.2 53.4 51.7 51.4 51.8 52.8 51.3 50.1 51.1 Years of schooling h'hold head 6.8 6.7 6.0 6.2 10.1 8.6 6.9 7.9 9.5 8.3 6.6 7.4 Mean No. of workers in h'hold 1.0 1.0 2.3 1.9 1.1 1.1 2.0 1.6 1.1 1.1 2.1 1.6 Mean number of workers includes workers 15 and over only. Mean years of school of household head assumes that the household head completed the maximum years possible in the second highest level of schooling completed. Dependency ratio=(number of children age 0-14 + number of elders age 60 and over) / number of household members age 15-59 Primary school enrolment rates are lower among the poor, and lowest among the extremely poor (Table A4.2). The pattern is much more pronounced for enrolment rates in secondary school. Among the poor and extreme poor, only about 2 children in 10 enroll in secondary school, versus 5 in 10 among the non-poor. The incidence of poverty among adults holding a primary school diploma or less is double the poverty rate among individuals with a vocational diploma or higher. More than 80 percent of the poor did not complete secondary school or hold a vocational school diploma. 19This appendix is based on the analysis previously done in Albania: Poverty Assessment carried out by the World Bank in 2003 84 Table 4.2: Poverty Measures and Shares by Educational attainment and Location Educational attainment for Poverty measures people 21 and older % of Poverty Tirana Other Urban Rural Total population share No schooling 19.0 20.5 30.7 26.6 3.5 4.1 Primary 26.7 26.1 30.8 29.3 61.6 79.4 Secondary 10.5 14.3 19.5 15.6 15.8 9.7 Vocational 2y 14.2 14.4 17.1 15.0 1.5 0.8 Vocational 4y 10.9 8.2 16.5 12.0 10.7 5.2 University and higher 2.9 2.6 3.8 2.9 6.9 0.7 Total 15.9 18.8 28.1 23.7 100.0 100.0 The poor exhibit substantially higher unemployment rates when compared to the non-poor (Table A4.3). The unemployment rate among the poor based on the standard ILO definition ­ at 14.2 percent ­ is almost double the rate among the non-poor. Among the extreme poor the unemployment rate is virtually triple. Table 4.3: Income shares by poverty status (percent) - ers care, work Non transfers Other Dependent business business Ndihme Benefits Agricultural agricultural Transf Pensions Ekonomike Maternity, social Unemployment other Non-Poor 34.6 24.8 8.5 14.4 2.0 13.7 0.4 1.3 0.5 Poor 26.8 36.7 5.7 7.9 7.4 14.1 0.5 0.6 0.3 Total 32.6 27.8 7.8 12.7 3.4 13.8 0.4 1.1 0.4 In Table A4.4, families are classified according to the main source of income, i.e., the one that provides more than 50% of the total.20 More than one-third of the families participating in the LSMS get most of their income from wage employment (this share is much higher in urban areas, particularly in Tirana ­ 64%). In this group, the headcount index of poverty is about 21%, varying from 19% in urban areas to 24% in rural ones. As expected, poverty is most severe for households relying on social assistance ("Ndihme Ekonomike", unemployment benefits, maternity, social care and other public transfers), where the headcount index is 46%, and is higher in urban areas (48%) than in rural ones (43%). Poverty is also severe for households relying mostly on agricultural businesses (34%, all rural poverty) and among rural pensioners (32%). 20This implies that about 5% of the households are not considered in the table, as no single source reaches this threshold. 85 Among those relying on pensions, rural households are remarkably poorer, with a headcount index of poverty reaching 32%, compared to 22% in urban areas. The incidence of poverty is lower in families depending on non-agricultural business (14%) and on transfers (17%). Poverty rates among households relying on transfers are higher in urban areas (21%) than in rural areas (15%). Table A4.4: Incidence of poverty by main source of income (at least 50%) Urban Rural Total Headcount Headcount Headcount Poverty N. Obs. Poverty Index N. Obs. Poverty IndexN. Obs. Index Dependent work 1029 18.8 341 23.7 1370 20.5 Agricultural business 24 --n/a-- 696 34.0 720 33.6 Non-agricultural business 264 11.6 72 17.1 336 13.5 Transfers 134 21.2 168 14.6 302 16.7 Pensions 339 21.5 178 32.0 517 26.3 Social assistance: Ndihme Ekonomike, unemployment, maternity, social care, other 77 48.1 55 43.1 132 45.6 Other 10 --n/a-- 3 --n/a-- 13 --n/a-- 86 BIBLIOGRAPHY Albanian Institute of Statistics (INSTAT). "Migration in Albania." 2001 Population and Housing Census Research Publication Number 18. INSTAT: Tirana, Albania. 2004 Albanian Institute of Statistics (INSTAT). "People and Work in Albania." 2001 Population and Housing Census Research Publication Number 19. INSTAT: Tirana, Albania. 2004 Albanian Institute of Statistics (INSTAT). "Living Conditions and Inequality in Albania." 2001 Population and Housing Census Research Publication Number 20. INSTAT: Tirana. Albania. 2004 Albanian Institute of Statistics (INSTAT). "Gender perspectives in Albania." 2001 Population and Housing Census Research Publication Number 21. INSTAT: Tirana, Albania. 2004 Alderman, Harold. Social Assistance In Albania: Decentralization and Targeted Transfers. Living Standards Measurement Study Working Paper Number 134. The World Bank: Washington D.C. July 1998. Braithwaite, Jeanine. Russia: Targeting and the Longer-Term Poor. Report No. 19377-RU The World Bank: Washington, D.C. May 1999. Coady, David, Margaret Grosh, and John Hodinott. "The Targeting of Transfers in Developing Countries: Review of Experience and Lessons." World Bank Social Safety Net Primer Series: Processed. April, 2002 Holzmann, R., and S. Jorgensen. "Social Protection as Social Risk Management: Conceptual Underpinnings for the Social Protection Sector Strategy Paper." Social Protection Discussion Paper No. 9904. The World Bank: Washington, D.C. 1999. Smolensky, E., Siobhdn Reilly, and Eirik Evenhouse. "Should Public Assistance Be Targeted?" Paper presented to the World Bank Conference on Designing and Implementing Social Safety Nets. Williamsburg, 1994. Social Protection Team. Balancing Protection and Opportunity: A Strategy for Social Protection in Transition Economies. The World Bank: Washington, D.C, 2003. Staines, Verdon and Dilnara. Isamiddinova. "Social Assistance in Transition Countries of Europe and Central Asia." Spectrum. The World Bank: Washington D.C., Summer 2004. pp. 21-25. World Bank. Albania: Poverty Assessment. The World Bank: Washington D.C.:2003. World Bank. Albania: Sustaining Growth and Improving Competitiveness ­ A World Bank Country Economic Memorandum. The World Bank: Washington D.C., 2004. 87