Report No: 125060-XK . Social Protection and Health Expenditure Note _____________________________________________________________________________________ Kosovo May 2018 Europe and Central Asia Region . . . 1 Standard Disclaimer: . This volume is a product of the staff of the International Bank for Reconstruction and Development/ The World Bank. The findings, interpretations, and conclusions expressed in this paper do not necessarily reflect the views of the Executive Directors of The World Bank or the governments they represent. The World Bank does not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. . Copyright Statement: . The material in this publication is copyrighted. Copying and/or transmitting portions or all of this work without permission may be a violation of applicable law. The International Bank for Reconstruction and Development/ The World Bank encourages dissemination of its work and will normally grant permission to reproduce portions of the work promptly. For permission to photocopy or reprint any part of this work, please send a request with complete information to the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA, telephone 978-750-8400, fax 978- 750-4470, http://www.copyright.com/. All other queries on rights and licenses, including subsidiary rights, should be addressed to the Office of the Publisher, The World Bank, 1818 H Street NW, Washington, DC 20433, USA, fax 202-522-2422, e-mail pubrights@worldbank.org. . Acknowledgments: This paper was prepared by Anita Schwarz, Stefanie Brodmann, Francesca Lamanna, Dorothee Chen, Agim Demukaj, Olasupo Olusi, Gonzalo Javier Reyes Hartley, and Asli Senkal under the guidance of Gallina Vincelette, Manager for Macro-Fiscal Management in Eastern Europe and Central Asia region. The authors are grateful for all the support received by colleagues at the World Bank and International Monetary Fund, for data collection, analysis, and drafting. Special thanks to Rhedon Begolli, Alexandru Cojocaru, Csaba Feher, Boryana Gotcheva, Mattia Makovec, Alessia Thiebaud, and Frieda Vandeninden. The team also benefited from comments and suggestions from the Practice Manager for Social Protection and Jobs in Eastern Europe and Central Asia, Andrew Mason, the Practice Manager for Health, Nutrition and Population in Eastern Europe and Central Asia, Enis Barış, and Ekaterina Vostroknutova, Lead Economist GMTE3. The authors gratefully acknowledge comments received from the Kosovo Ministry of Finance, the Ministry of Labor and Social Welfare, and the Ministry of Health. 2 CONTENTS KEY MESSAGES ............................................................................................................................................................... 4 OVERVIEW OF SOCIAL PROTECTION MECHANISMS AND THEIR FISCAL FOOTPRINT ........................................................ 7 MAIN SOCIAL PROTECTION PROGRAMS ....................................................................................................................... 11 Pensions.................................................................................................................................................................. 11 War Related Benefits ........................................................................................................................................... 14 Social Assistance Scheme.................................................................................................................................... 16 Labor Market Programs ...................................................................................................................................... 19 HEALTH SPENDING AND THE INTRODUCTION OF HEALTH INSURANCE ......................................................................... 22 Health Spending and Financial Protection ...................................................................................................... 22 Potential Impact of the Introduction of Health Insurance ........................................................................... 25 CONCLUSIONS .............................................................................................................................................................. 27 REFERENCES ................................................................................................................................................................. 30 TABLES Table 1: The fiscal impact of the war veterans benefits and other pension increases onlast-resort assistance programs for the poor .......................................................................................................................................................... 8 Table 2: Proportion of the population in each group that receives transfers (2014) ..................................................... 10 Table 3: Numbers of beneficiaries, average benefit levels and fiscal costs associated with war-related benefits in 2016 ..................................................................................................................................................................................... 15 Table 4: Progressivity of OOP payments ............................................................................................................................ 23 Table 5: OOP payments as a share of total consumption ................................................................................................. 23 Table 6: Incidence and intensity of catastrophic OOP health payments, 2015 ............................................................... 24 Table 7: Poverty impact of out-of-pocket health expenditures, 2015 ............................................................................. 24 FIGURES Figure 1: Total social protection spending as a percentage of GDP from 2009 to 2016 .................................................. 8 Figure 2: Contributions to growth in social spending from 2013 to 2016 ......................................................................... 8 Figure 3: Coverage of social assistance benefits, by quintile, 2015 ................................................................................. 17 Figure 4:Targeting accuracy of social assistance benefits, bottom 40 and top 60 percent, 2015 ................................. 17 Figure 5: Percentage point reduction in poverty head count after social assistance transfers, several years.............. 17 Figure 6: Spending on ALMPs and employment services as percentage of GDP, several years .................................... 19 Figure 7: OOP Payments, 2015 ........................................................................................................................................... 22 Figure 8: Average OOP health expenditure per capita, by consumption quintile, 2015 ................................................ 23 Figure 9: Pen’s Parade of Per Capita Expenditures Gross and Net of OOP Payments, 2014-2015 ................................ 25 3 KEY MESSAGES Kosovo is one of the poorest countries in Europe, and a growing proportion of the population is at risk for social exclusion and long-term poverty. According to the Kosovo Agency of Statistics, 17.6 percent of the population was living below the national poverty line of EUR 1.82 a day in 2015, while 5.2 percent were living below the national extreme poverty line of EUR 1.30. Moreover, opportunities for social mobility were scarce among the economically disadvantaged, the less educated, and those facing discrimination from the general public (World Bank, 2016). Overall spending on social protection has increased significantly over recent years, largely because of increases in pensions and war related benefits resulting in increased fiscal pressure. Spending rose from 3.6 percent of GDP in 2009 to 6.1 percent in 2016, primarily due to increases in pensions for those who contributed to the former Yugoslav pay-as-you-go system (ex-contributory pensions) and war-related transfers. Ex-contributory and social pensions1 comprised over 90 percent of all social protection spending in 2016. In September 2015, a new war veterans pension was introduced (0.8 percent of GDP in cash benefits alone in 2016), resulting in increased fiscal pressure. Furthermore, the actual expenditure almost doubled in comparison to what was allocated in the budget in 2016 and reached 1.1 percent of GDP in 2017. In January 2017, the parliament amended the law on war veterans to cap this spending to 0.7 percent of GDP recognizing the increased fiscal pressure, however, the amendments have not yet been implemented. Introducing other unfunded categorical benefits could risk fiscal sustainability and crowd out essential expenditures such as education, health, and public investment. It is also important to note that the costs of such schemes are not limited to fiscal impact and can further lead to adverse incentives on labor market participation, increased informality, higher inequality and increase social pressure for further benefits. Pensions are a mechanism to substitute income due to a permanent loss of income generating capacity, such as old age or disability and should not be used as a compensatory measure for other reasons. For this reason, in the early 2000s and faced with the impossibility of using contributory pension records to pay pensions, a universal basic pension was introduced in Kosovo. However, subsequent introduction of special benefits for different motives have eroded the original intended design and contributed to expanding fiscal costs and crowds out much needed spending on other spending categories such as health and education spending which has the potential to increase growth and lift the living standard of Kosovar’s. The recently introduced pension schemes are not necessarily aligned with the criteria for evaluation of the design of pension systems used by the World Bank. Pension systems should be able to achieve adequacy (adequate enough to prevent old age poverty), affordability (the ability to remain within the financing capacity of individuals and the society), sustainability (financially sound and can be maintained over a foreseeable horizon to be functional in the long term), equitability (provides income redistribution from the lifetime rich to lifetime poor consistent with societal preferences), predictability (relying on a unique comprehensive legislation governing the system instead of ad-hoc subsequent legislations), and robustness (negative shocks to government finances or an unexpected high level of claims should not put the financing of the pension scheme at risk). To contain the expansion of pension spending going forward, the Government must refrain from introducing new benefits, disallow increasing benefit levels for all but the basic pensions, and pro-rate existing benefits 1 This includes merit-based transfers such as war-related benefits and occupational benefits. 4 above basic pension levels contingent on funding availability. Favorable demographics and newly defined contribution pensions in Kosovo reduce the chances of an unaffordable pension bill, unlike in many countries in Eastern Europe. Although Kosovo’s current pension system is functioning well, recent changes to the pension system deviate from the original objectives for reducing poverty and threaten to make overall costs unsustainable. Policies that increase pensions significantly above the rate of inflation could backfire if current favorable demographics change for the worse. The expansion of programs targeted at specific population groups have crowded out spending of the social assistance targeted to the poor, which is both low and declining. In 2016, spending on the last-resort social assistance scheme amounted to 0.48 percent of GDP, down from 0.69 percent in 2009. Spending on social assistance for the poor was thus much lower than spending on war-related benefits (1.6 percent). In 2014, 23.8 percent of the poorest quintile and 10.2 percent of the second poorest quintile received transfers through the social assistance scheme.2 This decline in spending and the associated reduction in coverage may reflect the prioritization of social protection for groups more politically vocal than the poor. Consequently, transfers through the social assistance scheme resulted in only a 1.7 percentage point reduction in poverty, which is less than the average of 3.9 percent achieved in the region, which in turn is only half the reduction observed in the EU. Increasing programs’ coverage and benefit levels and prioritizing the most disadvantaged are key for increasing the effectiveness and equity of social assistance. A systematic reform of social assistance based on a social policy vision and strategy for poverty reduction is needed to better protect the poor and vulnerable, especially in light of ongoing reforms in sectors such as health and energy. As a first step, elements of the social assistance scheme can be strengthened. As one example, a revision of the strict eligibility criteria and its associated methodology for targeting beneficiaries is of great importance. Specifically, the criterion that excludes poor families if they have no children under the age of 5 or orphans under the age of 15 should be considered for elimination. The exclusion of families in which more than one individual is able to work should also be reevaluated. One strategy to address this issue is to apply the means-testing to all applicants rather than first sorting into Category 1 and Category 2 beneficiaries.3 Such changes to the targeting formula could serve to reduce work disincentives. Measures to incentivize the employment of work-able social assistance beneficiaries (e.g., in-work benefits) could also be explored. Finally, the tradeoff between increasing benefits to lift households above the extreme poverty line (at a minimum) and increasing coverage must be analyzed and costed. Public investments in employment promotion and active labor market programs are limited and insufficient to meet Kosovo’s labor market challenges. Recent economic growth has not spurred robust job creation and a large share of the population is either inactive or unemployed. The high levels of unemployment among women (about 31.8 percent in 2016) and youth (over 52.4 percent in 2016), and low labor force participation (38.7 percent in 2016), especially among women (18.6 percent in 2016), are significant social concerns. Additionally, about 10.7 percent of the working age population was discouraged workers4 in 2014. Despite these challenges, the government of Kosovo spent less than 0.1 percent of GDP on active 2 These figures refer to both direct and indirect beneficiaries. In terms of direct beneficiaries only, 15.2% of the first quintile and 5.6% of the second quintile received transfers from the social assistance scheme. 3 Details of the Category 1 and Category 2 beneficiaries are discussed on page 13. 4 Discouraged workers are workers who stopped looking for employment and became labor inactive due to lack of job opportunities, and who would otherwise participate in the labor market. 5 labor market programs, compared with 0.6 percent on average in the OECD. Kosovo’s public employment services are currently undergoing reform, but seeing and assessing results will take time. Given that low levels of health spending, including on drugs, is a leading cause of poverty, the social health insurance reform should not only work to improve health outcomes but also to improve financial risk protection, especially for the poor. Some progress has been achieved in the development of the outpatient drug benefit scheme, but key decisions on the parameters of the scheme have yet to be made. Immediate improvements to the Health Insurance Fund (HIF) institutional capacity and development of its information system will also be crucial to the implementation of the health insurance scheme. There is consensus that exemptions from premium contributions should rely on a combination of means-testing and proxy-means testing, but the eligibility system is yet to be operationalized. 6 OVERVIEW OF SOCIAL PROTECTION MECHANISMS AND THEIR FISCAL FOOTPRINT 1. The social protection system in Kosovo comprises income transfers and services. Income transfers that fall under the social protection budget include: (i) Pensions, which include a universal tax-financed basic pension or old-age pension payable to all those over the age of 65, with a supplement for those who can prove that they contributed for at least 15 years to the old Yugoslav pension system; also, other pensions for Kosovo Security Force and Kosovo Protection Corps retirees, Trepca miners, etc.; (ii) War-related benefits, which include various benefits for those who fought; civilians and soldiers who were disabled during the 1998-99 military conflict; and protection for surviving relatives of fallen soldiers and civilians, as well as for families of those missing; and a new scheme for victims of sexual violence during the war; (iii) A social assistance scheme, which is a tax-financed last-resort income support program that provides assistance to a limited set of beneficiaries; (iv) Other transfers, including a non-contributory disability pension for persons aged 18-64, and new pension schemes for blind, tetraplegic and paraplegic persons; benefits for children with disabilities, and payments for children in foster care. In addition to these, the social protection budget also finances services, including: (i) Social care services, organized and provided by decentralized municipal centers for social work, and focused primarily on child protection services5, as well as institutional social care organized and financed by the MLSW alone or co-administered with municipalities; and (ii) Employment services for those who are registered as unemployed and willing to work. Finally, though not part of the social protection budget, there are energy subsidies for the recipients of the social assistance scheme, which are designed to provide implicit subsidies from wealthier to poorer consumers through energy tariff differentiation. 2. Policy choices, rather than demographics or the economic cycle, have driven social protection spending. Public spending on social protection, the sum of pension spending and social assistance schemes, has increased from 3.6 percent of GDP in 2009 to 6.1 percent in 2016 (Figure 1). As a share of total public spending, overall social protection spending as a percentage of overall public spending increased from 12.7 percent in 2009 to 21.7 percent in 2016 (Table 1). The change was driven by the introduction of the war veteran pensions6 in 2015, other elections-related increases in pensions for those who contributed toward the former Yugoslav pay-as-you-go system, and the introduction of one-off payments for former political prisoners (Figure 2). Some war-related benefits were paid regardless of the beneficiary's age or ability to work, though some were determined by level of disability. Increases in spending for war-related benefits have been responsible for 45 percent of the total increase in the social protection bill (of which the newly introduced veteran benefits added 33 percent alone) between 2013 and 2016. The basic pension contributed an additional 21 percent of the increase and 26 percent came from former Yugoslav 5 Social care services also include services for the elderly and for persons with disabilities that are provided in institutional, home-based or community environments. 6 After a process of verification of war veterans which ended in 2014, and complaints were finalized by mid-2015, new benefits for war veterans were first introduced in September 2015. The number of pension beneficiaries of war veterans far exceeded expectations (reaching about 29,300) and contributed greatly to the overall cost of this merit based scheme. 7 contributor-based pensions due to the reform of the scheme and the introduction of four sub categories (based on the level of education and years of contribution) (Figure 2). However, spending on social assistance scheme as share of GDP fell from 0.7 percent of GDP in 2009 to 0.48 percent of GDP by 2016. Table 1: The fiscal impact of the war veterans’ benefits and other pension increases on last-resort assistance programs for the poor (in percent of GDP, unless otherwise indicated) 2013 2014 2015 2016 War-related benefits 0.57 0.74 0.93 1.60 War veterans’ scheme - - 0.09 0.80 Other war -related benefits 0.57 0.65 0.65 0.62 Pensions for political prisoners (one off) - 0.09 0.18 0.17 Pensions (not related to war) 3.01 3.74 4.06 4. 19 Ex-contributory pensions (former YU pay as you go) (over 65) 0.93 1.11 1.18 1.45 Basic pensions (over 65) 1.64 1.95 2.09 1.97 Disability pension 0.28 0.30 0.32 0.31 Other pensions 0.17 0.29 0.29 0.29 Social assistance scheme 0.52 0.48 0.44 0.48 Total social protection spending 4.14 4.86 5.24 6.09 Total social protection spending (% of overall public spending) 14.4 17.9 19.2 21.7 Source: MLSW and World Bank Staff calculations. Note: Other war-related benefits category includes ten categories related to war law excluding the war veterans’ benefits scheme and is presented in Table 3. Other pensions category includes Kosovo Protection Corps (KPC) early retirement, Trepca early retirement benefits, Kosovo Security Force (KSF) pensions, and pensions for the blind. Figure 1: Total social protection spending as a percentage of GDP from 2009 to 2016 Social Protection Spending (% of GDP) 7 6 5 4 3 2 1 - 2009 2010 2011 2012 2013 2014 2015 2016 Ex-contributory pensions Basic pensions Disability pension Pensions for political prisoners Other pensions Other war-related benefits War veterans scheme Social assistance Source: Ministry of Labor and Social Welfare data, World Bank staff calculations. 8 Figure 2: Contributions to growth in social spending from 2013 to 2016 100% 90% 80% Other pensions 70% Social assistance 60% Political prisoners 50% Disability pension Basic pensions 40% Ex-contributory pensions 30% War Veterans scheme 20% Other war-related benefits 10% 0% 2013-2016 Source: Ministry of Labor and Social Welfare data, World Bank staff calculations. 3. Coverage of population by the social protection programs remains low. In 2015, 32.5 percent of the population (direct and indirect beneficiaries) was covered by some form of social protection in Kosovo (Table 2). This aggregate figure masks significant differences in coverage rates by programs: high coverage with categorical merit-based benefits, expanding coverage with pensions, and low coverage with social benefits and services. Coverage of the social assistance scheme was low, both in terms of direct beneficiaries (5.4 percent) and direct and indirect beneficiaries (9.6 percent) and among the poor (15.1 percent of direct and 23.8 percent of direct and indirect beneficiaries). Table 2 also shows that remittances emerge as an alternative to the public safety net in Kosovo, with double the coverage compared with social assistance and about a third of the overall social protection coverage. 4. Pensions represent a large share of spending and have increased especially over recent years. The pension system was initially designed to be simple and fiscally sustainable, but benefits, coverage, and the overall complexity of the system are increasing and could jeopardize sustainability in the long run. In 2016, the amount of the basic pension stood at EUR 75 per month (up from EUR 60 in 2013) and that for contributory at a minimum of EUR 158 (up from EUR 112 per month in 2013).7 Total payments reach a 1.4 percent of GDP for the basic pension for a total of 119,000 beneficiaries (up from 116,512 in 2013) and 2 percent of GDP for the contributory pension for approximately 41,300 beneficiaries. It is important to stress that the lowest contributory benefit of EUR 158 per month is approximately 44 percent of the average wage of Kosovo, which is slightly above ILO standard for an average contributory pension of 40 percent of average wage after 30 years of service (note that in Kosovo, it is provided starting from 15 years of contributions). In addition to elderly pensions, almost 20,000 individuals (equivalent to about 6 percent of the employed or 1.1 percent of the total population) receive pensions for permanent disability of EUR 75 per month at a cost of 0.3 percent of GDP annually. For certain categories the disability pensions are higher, 7 This supplement was introduced in January 2018 for those who could prove an over 15-year contribution to the former Yugoslav pay-as-you-go system. 9 e.g. the pension for the blind is EUR 125 (and an additional EUR 125 for care by other person), and the Material Support for Children for Permanent Disability is EUR 100 per month. Table 2: Proportion of the population in each group that receives transfers (2014) Panel A: Direct beneficiaries ONLY Poverty Status Total Poor Non Poor All social protection 32.5 43.2 29.8 Pensions 27.1 28.1 26.9 All labor market programs n.a. n.a. n.a. All social assistance 5.4 15.1 3.0 All remittances (from abroad and from family members within 10.9 7.7 11.7 Kosovo) Panel B: Direct and indirect beneficiaries Poverty Status Total Poor Non Poor All social protection 51.6 66.2 47.9 Pensions 46.0 50.5 44.9 All labor market programs n.a. n.a. n.a. All social assistance 9.6 23.8 6.1 All remittances (from abroad and from family members within 16.9 13.7 17.8 Kosovo) Source: World Bank ADePT based on Kosovo National Household Budget Survey (HBS) 2015. Note: Poor category corresponds to the lowest quintile. 5. Coverage of and spending on active labor market programs (ALMPs) remain insufficient to meet the needs of the unemployed. Labor market programs in Kosovo are small-scale and mainly donor-driven. The number of beneficiaries of ALMPs increased by 50 percent from 2012 to 2016, but still remains low with respect to the high rate of unemployment among vulnerable groups like women and youth. The overall spending on ALMPs is the lowest in Eastern Europe and Central Asia, comprising less than 0.10 percent of GDP. On average, OECD countries spend 0.56 percent of GDP on administration of public employment services (PES) and ALMPs. 6. Looking forward, these programs need to be reevaluated to address the risks of the poor and the unemployed and to increase incentives for job search and formal employment. The current targeting formula for beneficiaries of the social assistance scheme contains a number of disincentives as well as trade-offs between job search, formal employment, and saving on one hand, and maintaining social benefits on the other. As explained in more detail in section C, the social assistance scheme distinguishes between two categories of eligible households: Category 1 is composed of dependent household members only and 10 Category 2 requires one household member to be registered as unemployed, but all others must be dependent. Overall, there are strong exclusion criteria (e.g., child age limit, number of unemployed able- bodied household members, but also temporary access to public works or one-time payments/remittances) and the means test is implemented in a way that can place eligibility at the discretion of the social worker. Activation policies should focus on promoting (and rewarding) positive behavioral changes towards job search, including with in-work benefits. MAIN SOCIAL PROTECTION PROGRAMS Pensions 7. The simplicity and sustainability of Kosovo’s pension system are increasingly at risk. The pension system was designed to be straightforward, fiscally sustainable, and reduce economic insecurity during old age for all residents of Kosovo. Over the years, however, various political demands have resulted in large increases in the cost of the system every electoral cycle, and in additional benefits for various special groups, undermining not only the simplicity but also the fairness of the original design. 8. The original design provided a pension sufficient to bring all poor elderly 65 years and older to the poverty line. Under the former Yugoslav system, employers sent pension contribution records to the Pension Fund. These had to match records documented in work history booklets that were issued to each worker. During the independence conflict, the official Pension Fund records were removed, leaving partial copies only; some individuals lost their work history booklets; and individuals who worked in other ex- Yugoslav countries had trouble verifying their work history. The decision was thus made to simplify the system and ensure that no elderly individual falls below the poverty line. 9. Earning-related pensions with mandatory contributions were introduced to supplement basic pensions. As of 2002, the basic pension is supplemented for future retirees with a defined contributions scheme. It is financed with mandatory contributions of 5 percent of wages each from employers and employees. The funds are held in a savings account and made available at retirement to supplement the universal pension. This system, administered by the Kosovo Pension Savings Trust, has worked well and currently pays beneficiaries EUR200 per month until the account runs out of money. 10. The number of additional pensions was limited. There were only a limited number of additional pensions paid from the State budget, one to unemployed Trepça miners (currently about 3,300 beneficiaries at a total cost of 0.07 percent of GDP annually) and the other to those downsized from the Kosovo Protection Corps (currently almost 1,000 beneficiaries at a total cost of 0.04 percent of GDP annually), as it was converted to the Kosovo Security Force (KSF). Individuals retiring from the KSF are also now receiving pensions (currently just under 300 beneficiaries at a total cost of 0.01 percent of GDP). 11. Just before independence, Government introduced an additional pension. In 2008, the Government introduced an additional pension, designed to recognize those who had contributed the minimum period of 15 years to the former Yugoslav pension fund. The basic pension was held constant since 2014, when it was increased to EUR 75 per month (over 35 percent above the poverty line). But this new “ex- contributory” pension increased repeatedly until 2016, and stands at almost 300 percent of the poverty line. As a result, by the end of 2016, approximately 119,000 individuals received only the basic pension of EUR 75 per month at a total fiscal cost of 2 percent of GDP annually; approximately 41,300 received the ex- 11 contributory pension of at least EUR 158 per month at a cost of 1.45 percent of GDP annually; and almost 20,000 individuals received disability pensions of EUR 75 per month at a cost of 0.31 percent of GDP annually. Note that the EUR 158 per month is approximately 93 percent of the minimum wage and 44 percent of the average net wage, more than meeting the ILO standard of an average contributory pension of 40 percent of average wage after 30 years of service. Kosovar workers are receiving this benefit if they worked for a minimum of 15 years during ex-Yugoslavia regardless of their current employment and retirement status, and they are eligible to receive this pension after they reach the retirement age of 65. 12. Before the 2014 election, the Government added another more generous provision to the existing ex-contributory pension scheme. This provides for differentiation among the ex-contributory pensioners by level of educational attainment, with those having completed only elementary school receiving EUR 158 per month, those with high school education receiving EUR 172 per month, those with higher education receiving EUR 186 per month and those with tertiary and post-tertiary education receiving EUR 230 per month. Almost half (47.8 percent) of the contributory pensioners receive the lowest amount, with 32.1 percent receiving the second category and only 10.5 percent and 9.6 percent receiving the two highest categories, respectively. The additional cost for implementation of changes to pension law was estimated to about Euro 18.5 million or 0.3 percent of GDP. The implementation of the law was delayed until April 2016, which increased fiscal pressure going forward. 13. Ex-contributory benefits were also extended to education workers who continue to work in the parallel education system after they were put out of work by the Milosevic regime in the 1990’s. These workers had not completed 15 years of contributions before they were ousted from work. Around 2,000 additional workers are benefiting from this extension, which has been implemented since April 2016. Other workers thrown out of work during this period are also lining up to access the contributory pensions, with health workers next in the queue. 14. A new draft law on the status of Education worker’s is currently under discussion can lead to a large and uncertain fiscal cost going forward, and can create additional demands from other groups. This additional benefit, if passed, would be given in addition to the other benefits that were introduced such as the ex- contributory pensions and can lead to uncertain fiscal costs, as well as, labor market disincentives, or work in the informal sector and would crowd out other essential social spending, such as health and education. 15. Benefits for those considered disabled through work injury under the old Yugoslav law and family benefits for contributory pensioners were added in the 2014 legislation, but not implemented until March 2017. About 112 disabled who received benefits under the Yugoslav regime are now receiving EUR 75 per month, and 2,150 family pensioners are receiving EUR 75 per month with 20 percent added per child. These family benefits are only available for widows and widowers between the ages of 55 and 65, when the basic pension becomes available, and for children under the age of 18 or up to the age of 26 if they are studying. A compensation for the blind was also added beginning in January of 2014, which provides EUR 125 per month per beneficiary with an additional EUR 125 for a caretaker, benefiting more than 3,600 individuals. A separate additional benefit for police is being considered, and other occupations are also requesting additional benefits. 16. Kosovo’s pension system has morphed from a simple design intended to keep the elderly out of poverty into an arrangement in which many groups demand and receive special benefits. Pensions are a mechanism to substitute income due to a permanent loss of income generating capacity, such as old age or disability and should not be used as a compensatory measure for other reasons. For this reason, in the early 2000s and faced with the impossibility of using contributory pension records to pay pensions, a 12 universal basic pension was introduced in Kosovo. However, subsequent introduction of special benefits for different motives have eroded the original intended design and contributed to expanding fiscal costs and crowds out much needed spending on other spending categories such as health and education spending which has the potential to increase growth and lift the living standard of Kosovar’s. As a result, pension spending, excluding what is spent on war veterans’ pensions, now exceeds 4 percent of GDP. Although this is low compared to other countries in the region, the growth in spending is not sustainable. 17. The recently introduced pension schemes are not well aligned with the criteria for evaluation of the design of pension systems used by the World Bank. Pension systems should achieve adequacy (adequate enough to prevent old age poverty), affordability (the ability to remain within the financing capacity of individuals and the society), sustainability (financially sound and can be maintained over a foreseeable horizon to be functional in the long term), equitability (provides income redistribution from the lifetime rich to lifetime poor consistent with societal preferences), predictability (relying on a unique comprehensive legislation governing the system instead of ad-hoc subsequent legislations), and robustness (negative shocks to government finances or an unexpected high level of claims should not put the financing of the pension scheme at risk). 18. Cost-reducing measures improved eligibility verification. In response to the recent costly legislation, the Government undertook a number of cost-saving measures, including (i) better elimination of dead pensioners; (ii) better enforcement of the residency requirement; and (iii) better enforcement of restrictions against collecting multiple pensions. Many people in Kosovo die at home resulting in no death certificate being issued. Following a series of unsuccessful arrangements, the Ministry of Labor and Social Welfare reached an agreement whereby the Islamic societies notify the Civil Registration Authority (CRA) and the CRA notifies the Ministry of Labor and Social Welfare of a death. As a result of this exchange which began in October 2015, the number of cases terminated in 2016 was 9,489, more than 50 percent higher than the 6,012 cases terminated in 2015. A second measure enforces the Kosovo residency requirement to be eligible for the basic pension. By administrative order, individuals are required to show proof of their residence in Kosovo. Previously, a large number of individuals resided in Turkey, Albania, and other countries, but returned to Kosovo every 6 months to collect the basic pension. As a result of stronger enforcement of the residency requirement, 8,235 basic pensions have been terminated since the beginning of 2016. The third cost-reducing measure involves identifying individuals illegally collecting multiple benefits from the state. An additional 2,700 cases were removed from the roster because of double benefits. Recommendations: 19. Substantial efforts have been made to contain the burgeoning fiscal costs of the pension system, but further steps are necessary to ensure its sustainability. The Government is advised to focus on two courses of action: (i) limiting the costs of the additional benefits that have already been granted; and (ii) preventing new benefits from being granted. Limiting the costs of the formerly granted additional benefits will require advocating that all elderly residents in Kosovo are entitled to the basic pension and estimating its cost. For all additional pensions, the Government should consider imposing a limit of around 1 percent of GDP, with all currently granted benefits prorated to fit within that limit, and adequately enforce it. Additionally, the benefit entitlement should be fixed at today’s nominal terms to be adjusted downward to fit within the 1 percent limit. It will also be important for residents to understand that the provision of additional benefits to additional groups results in fewer benefits for all. 13 War-Related Benefits 20. Kosovo offers a wide and still growing plethora of war-related benefits which at end-2016 stood at more than 1.6 percent of GDP. These include pensions for families of individuals who died fighting, pensions for families of civilians who died, pensions for families of fighters who are missing since the war, pensions for families of civilians who are missing since the war, pensions for fighters who became invalid in the course of fighting, pensions for civilians who became invalid during fighting, pensions for families of invalid fighters once they have died, pensions for families of civilian invalids once they have passed away, caregiver benefits for the invalid fighters, and caregiver benefits for the invalid civilians. On top of these, the Government passed a law providing pensions to healthy fighters in recognition of their service to the nation. The implementation of this law began in 2016. 21. Benefit levels, as shown in Table 3, are substantial and can provide strong incentive for individuals to remain outside of formal labor markets. War-related disability benefits are provided for lower rates of disability and at higher amounts than the civilian disability benefits, contributing to inequity in disability protection. Significant shares of war-related benefits are received by family members (of war martyrs, invalids, missing soldiers), including by persons of working age and able to work. Table 3 provides details on the number of individuals and/or households receiving benefits, the monthly value of those benefits, and the cost to the budget of the cash benefit portion only. It is useful to note that the benefit levels are relatively high, with families of dead and missing fighters receiving benefits more than 25 percent above the average wage, measured as Euro 457 for pension versus Euro 365 average wage, leaving families with little incentive to participate in the labor market. Military invalids receive well above minimum wage in cash benefits. War-related civilian deaths and injuries are awarded less generously, but still receive 200- 400 percent of the poverty line. These benefits may reduce beneficiaries’ motivation to join the formal labor market, and encourage participation in informal labor markets. Given the difficulty of securing employment in Kosovo, individuals may find it easier to procure income by petitioning the Government for benefits. 22. The fiscal costs shown in Table 3 exclude all the noncash benefits that war-related beneficiaries receive, greatly underestimating the total impact of war-related benefits on the budget. Noncash benefits include free primary, secondary, and tertiary health care, including health services abroad if not available in Kosovo; privileged access to employment in the civil service both at the national and local levels; advantages in enrollment in educational institutions for the individual and family members; scholarships for education; housing in student dorms at no cost; free textbooks; exemption from transportation costs, both urban and interurban; free electricity if in need; exemptions from certain judicial and administrative taxes, customs and business taxes; professional retraining; and housing provided in collective housing buildings if needed. The costs of these noncash benefits are spread throughout various ministry and agency budgets and difficult to assess, but are likely to be significant. 23. The newest benefit, the veterans’ pension continues to place significant fiscal pressure on the budget as the amendments to the war veterans law to cap this spending to 0.7 percent of GDP is still not implemented to this day. The veterans’ pension is provided for life for those who fought during the conflict, no matter for how long. Currently non-public employees verified as war veterans receive half the monthly benefit of EUR 170, and those employed by the Government receive no benefit while employed. The Government originally estimated 13,000-14,000 beneficiaries. As of March 2018, however, more than 37,000 individuals have received this benefit, and there are still significant number of applications still pending. The amendments to the war veterans law that was passed in January 2017 categorized 14 beneficiaries into three groups: (i) those who had been part of the KLA since 1991 would receive EUR 250; (ii) those who joined the fighting in 1997 would receive the current EUR 170; and (iii) those who joined during the general mobilization in March 1999 would receive EUR 120. However, this classification of war veterans into three groups has not been implemented to this day. Consequently, the Veteran’s benefit scheme ran out of money in August 2017, and a reallocation of about 23 million was necessary through the mid-Year-Budget Review process in 2017to meet the shortage. The 2018 budget allocated more than 0.7 percent of GDP suggesting that the reform will not take place. The overall war-related benefits currently account for about 1.8 percent of GDP in 2017, up from about 0.7 percent in 2014. 24. Table 3: Numbers of Beneficiaries, Average Benefit Levels and Fiscal Costs Associated with War- Related Benefits in 2016 Number of Average Benefit Level Cost as a Percentage Beneficiaries (EUR per month) of GDP, per year (individual and/or households) Families of war martyrs 1,964 457 0.180% Families of missing fighters 20 453 0.002% KLA war invalids 3,423 295 0.203% Caregivers for war invalids 74 225 0.003% Families of dead war invalids 153 264 0.008% Families of civilians killed during 4,573 174 0.161% conflict Families of missing civilians 363 177 0.013% Civilian invalids 1,975 131 0.052% Caregivers for civilian invalids 33 50 0.0003% Families of dead civilian invalids 103 101 0.002% New war-related benefit 28,207 170 0.8 % (Healthy war veterans pension), only came after 2015 Political prisoners (one off) 0.17 % TOTAL* 40,888 1.60% Source: MLSW data and World Bank staff calculations. Note: The total cost as % of GDP per year in 2016 includes the Benefits for political prisoners although that category is not covered by war law but implemented through a separate law. The benefits are one-off and are based on days spent in prison. Recommendations: 25. The Government has limited options in the short run aside from prorating benefits to recipients across all sub-categories to fit within the agreed upon budget constraint. Claims have been made by individuals younger than 16 at the time of the conflict, but time and political will are necessary to eliminate such fraudulent cases. It will also be important for residents to understand that increases in the number of claims are associated with lower benefits for all. The government should implement the amendments to the war veterans law to correct the fraudulent cases and maintain the fiscal pressures arising from this scheme. The burden that these payments put on the budget also lowers the growth potential in the economy, by limiting other developmental expenditures that could promote growth and poverty reduction. 15 Social Assistance Scheme 26. Kosovo operates a single “last-resort” targeted social assistance scheme (Asistenca Sociale).8 This social assistance scheme is the only program in Kosovo targeted to the poor, combining categorical eligibility criteria with a means test. As a first step, households are classified into one of two categories. Households in Category 1 consist of members that are disabled, dependent or care-givers. Households in Category 2 must have one household member registered as unemployed. The household must also have a minimum of either one child under the age of five or one orphan younger than 15 years. All other household members must be dependent. As a second step, households classified as Category 1 or Category 2 are subject to a proxy means test. Poverty is widespread among disabled persons and pupils/students,9 which aligns with the priorities of the social assistance scheme. However, gaps in coverage of the poor exist because the program operates under strict eligibility criteria, including binary exclusion filters. For example, households in Category 2 are no longer eligible once the youngest child turns five, which leads to exclusion of families with school-age children. The use of such exclusion filters is unclear because the cost of raising a child older than five years of age is not expected to decline in the short run. Children are more likely to live in poverty and families with multiple children have a higher likelihood of living in poverty and extreme poverty. 27. Benefit eligibility is tied to employment, specifically, only one able-bodied family member can be registered as unemployed for the family to qualify as a Category 2 social assistance recipient. This requirement indicates that employment (particularly in the formal sector) and the receipt of benefits cannot be combined, regardless of the level of a family’s assessed income. This could potentially create disincentives to work or incentives to work informally in order not to lose benefits. In addition, it discourages the second able-bodied family member to search for formal employment, and instead declare him-/herself as a full-time caretaker of other dependent family members.10 28. Spending on last-resort social assistance in Kosovo is low, both by regional standards and in comparison, to other benefit categories; it has also been declining both in terms of volume and coverage. In 2016, spending on the last-resort social assistance scheme amounted to 0.48 percent of GDP (or 1.71 percent of total public expenditure). Spending on social assistance for the poor is thus much lower than spending on war-related benefits (1.6 percent of GDP). With time, the last-resort social assistance for the poor is increasingly crowded out by the rise in spending on pensions and war-related benefits. Between 2009 and 2016, spending on the last-resort social assistance scheme fell by 12.3 percent in real terms, from 19 percent of total social protection spending to 8.13 percent, or from 0.69 to 0.48 percent of GDP. In December 2016, 26,302 households benefited from the last-resort social assistance scheme, down from 35,781 at end-2010. Over the same period, the number of beneficiary households receiving war-related benefits increased from just under 13,000 to 40,887. This decline in spending and coverage of the social assistance scheme may reflect policy choices and priorities that focused social protection on more politically vocal groups – such as war veterans and select groups of pensioners – rather than the poor. 8 In this section, “social assistance” refers to the social assistance scheme (Assistencia Sociale) only. Other welfare programs, like the war-related benefits or benefits for “children in custody and parental care” are either discussed in a separate section of this report or not included due to small program size. 9 In 2015, disabled persons made up a third of the poor and students/pupils another third (Kosovo Agency for Statistics and World Bank, 2017). 10 Csaba Feher, La-Bhus Fah Jirasavetakul, Alain Jousten, February 2016, Kosovo Enhancing Social Protection Cash Benefit, International Monetary Fund, Fiscal Affairs Department. IMF Country Report No. 16/123. 16 Household-level data from the household budget survey (HBS) 2015 confirm that social assistance benefits are relatively well-targeted. In 2015, 23.8 percent of the poorest quintile and 10.2 of the second poorest quintile were covered by social assistance benefits (Figure 3). Over 77 percent of the budget allocations on social assistance benefits went to the bottom 40 percent of the population (Figure 4). However, ‘leakage’ to the non-poor is not insignificant and implies flaws in the design or/and implementation of the methods with which the poor are targeted. 29. Even with recent reforms, benefits of the last-resort social assistance are too modest to bring beneficiaries above the poverty line. The decline in the number of beneficiaries was also due to the tightened eligibility criteria, which reduced the number of beneficiaries by 26.2 percent between 2009 and 2016. Combined with largely unchanged program expenditures, this amounted to an increase in the average benefit amount from around EUR 75 per month in 2013 to EUR 91 in 2016. Benefits are inefficient in protecting the poor; these transfers resulted in only a 1.7 percentage point reduction in poverty (Figure 5), which is less than the average of 3.9 percent achieved in the region, which in turn is only half the reduction observed in the EU. For a family with four members, which is the average size of beneficiary families (IMF, 2017), this translates to EUR 0.76 per family member per day in 2016. The daily benefit amount corresponds to 37 percent of the national poverty line (EUR 1.82 per adult equivalent per day), 58 percent of the extreme poverty line (EUR 1.30) and 13 percent of minimum wage (EUR 170 per month). Moreover, benefits decrease comparatively as family size increases, due to the use of equivalence scales which are well below those commonly used in OECD countries and emerging economies. Figure 3: Coverage of social assistance benefits, by Figure 4:Targeting accuracy of social assistance quintile, 2015 benefits, bottom 40 and top 60 percent, 2015 Source: World Bank ADePT based on Kosovo HBS 2015. Note: Figures refer to both direct and indirect beneficiaries. Coverage of direct beneficiaries ONLY is much lower. 30. Although not part of the social protection budget, poor Kosovars benefit from explicit and implicit energy subsidies. The objective of the subsidies is to protect poor groups from rising energy prices or reduce “energy poverty” (i.e., spending more than 10 percent of household expenditure on energy) and ensure sufficient access to energy for poorer households. Beneficiaries of the last-resort social assistance scheme, in addition to beneficiaries of some war-related benefits (e.g., martyrs’ families, war veterans, and civilian victims) received explicit annual energy subsidies in the range of EUR 143 per household per year in 2015. These subsidies are distributed at a cost of around EUR 4.5 million per year or 0.1 percent of GDP. Future energy cost increases are foreseen, with negative consequences for poor households and those just above the poverty line. Even today, a fifth of the poorest quintile reports no spending on energy (compared with 6.3 percent in the top quintile). 17 Figure 5: Percentage point reduction in poverty headcount after social assistance transfers, several years Source: World Bank ASPIRE (The Atlas of Social Protection Indicators of Resilience and Equity) and SPEED (Social Protection Expenditure and Evaluation Database). Note: Numbers next to country names refer to the year of the household survey used in the analysis. Recommendations 31. A key priority is increasing the effectiveness of Kosovo’s last-resort social assistance program in protecting the poorest. Spending and coverage are low by international comparison and have been decreasing over recent years. An important step is introducing a refined targeting methodology based on a combination of means and proxy-means testing, which would allow for the accounting of certain informal incomes, as well as the capturing of remittances. The removal of certain strong exclusion criteria (specifically when the youngest child turns five) would improve the coverage of the poor without compromising targeting accuracy. The amount of the benefit could be reconsidered to at least lift households over the extreme poverty line of EUR 1.30 per person per day; the applied equivalence scales could be reconsidered in accordance with the OECD and other international practices to reinforce the protection of families with more than two members. The current scheme offers a number of disincentives for taking up work (e.g., only one household member can be registered as unemployed, others must be dependents). Changes in the targeting formula can alleviate some of these disincentives, and other measures, such as in-work benefits, could be considered. 32. Future energy and health care reform will raise costs and premiums for the currently uncovered poor - due to rigid eligibility criteria - which will adversely impact those vulnerable households. Existing coverage, generosity and planned spending on social assistance are not sufficient to address the needs of the poor. Going forward, social protection systems should ensure their capacity to protect the poor and vulnerable and mitigate the impact of modernization in the sectors (e.g., energy) and changes in policies (e.g., health). For that to occur, the Ministry of Labor and Social Welfare must increase its capacity to target poor families and to scale up the coverage and generosity of programs as need arises. Caution will have to be continuously exercised to avoid promoting disincentives for taking up work. 18 Labor Market Programs 33. Activation and intermediation services have a key role in facilitating transitions to and increasing productivity in the Kosovo labor market. These services can help reduce the negative effects of labor market exclusion as well as structural challenges (e.g., failure of the education system to adequately address the needs of the labor market). Limited capacity and funding for public employment services results in low coverage of ALMPs, unclear results, and highly fragmented (largely donor funded) projects. 34. Despite the high rate of unemployment, the level of public funds dedicated to the implementation of intermediation and various active labor market programs (ALMPs) is low. Unemployment in Kosovo increased from 27.4 percent in 2016 to 30.4 percent in 2017. Kosovo also stands out as the country in the region with the highest rate of youth unemployment (over 50 percent), the largest share of youth among the long-term unemployed (over 20 percent), and the highest share of youth neither in employment, education nor training (NEETs) (almost 30 percent) (World Bank, 2018). In 2016, spending on ALMPs (e.g., on-the-job training, wage subsidies) was less than 0.05 percent of GDP, and overall spending, which includes employment services (PES) administration and training in vocational training centers (VCTs) was still less than 0.1 percent of GDP. Even when comparing total spending (PES administration and ALMPs) across regions, Kosovo spends less than the regional average of 0.32 percent of GDP (Figure 6). The difference in spending between Kosovo and more advanced economies is greater still. For example, in 2013, Poland spent 0.49 percent of GDP on ALMPs and another 0.08 on PES and administration. The OECD average spending on ALMPs is 0.42 percent and an additional 0.14 percent on PES and administration (OECD, 2015). Figure 6: Spending on ALMPs and employment services as percentage of GDP, several years Source: World Bank ASPIRE and SPEED. Note: Numbers next to country names refer to the year of the household survey used in the analysis. 35. Labor market programs in Kosovo are small-scale, donor-driven, and consist of active measures only. ALMPs include vocational training, on-the-job-training, internships, employment incentive programs (wage subsidies), public works, and entrepreneurship grants. Coverage of ALMPs in Kosovo increased from 6,412 beneficiaries in 2012 to 9,558 beneficiaries in 2016, but still covers only a fraction of the registered unemployed (8.5 percent). In 2016, the largest share of beneficiaries by far was in vocational training programs (6,826), followed by on-the-job training (951), public works (819), wage subsidies (474), internships (434), and self-employment support (54 beneficiaries). From 2012 to 2015, the share of 19 beneficiaries in training programs more than doubled, whereas the share of beneficiaries in public works was three times the size in 2012 compared with 2016. 36. PES in Kosovo is currently undergoing a thorough reform process, including the establishment of a separate employment agency (Employment Agency of the Republic of Kosovo (EARK)). Going forward, the PES will be responsible for implementing intermediation services and ALMPs. A number of new processes are being established, such as a new service delivery model that distinguishes between jobseekers and those registered as unemployed; profiling of the registered unemployed into three risk categories; creation of an individual employment plan for those identified at high risk of longer-term unemployment; and a performance measurement plan for branch offices and a broader monitoring system for ALMPs. 37. There is limited evidence on the gross and net impacts of existing ALMPs on employment outcomes in Kosovo. In addition to implementing and monitoring the ongoing reforms and services, a careful review of the impact of existing programs is necessary in the design of new programs that effectively address the needs of its beneficiaries. Of equal importance is the development of a system that will allow for the regular monitoring and evaluation of ALMPs. Recommendations 38. The development of more effective intermediation services and ALMPs could be instrumental in reducing unemployment and improving the employability of the labor force, particularly for women, youth, and the long-term unemployed who lack informal networks. In 2015, a total of 11,506 vacancies and 112,179 unemployed (48,960 females and 63,219 males) were registered with PES. The public employment offices enabled 6,174 job placements in 2015, and have the potential for more placements provided that resources are made available and accountability criteria are established. Given limited coverage of employment services and ALMPs, the Government should explore increased partnerships with non-public providers, including NGOs and private intermediation providers. For instance, this could include the outsourcing of selected training activities or other ALMPs, especially for specific target groups such as youth, women, or jobseekers with disabilities, and/or in certain geographic areas. The Government has already started exploring the legal framework for such public-private partnerships. In addition, operationalizing these partnerships would require developing adequate regulation and procedures for selecting, contracting, and monitoring providers. 39. It would be important to improve both intermediation and activation measures through improved design and evaluation of services and programs. The design of an ALMP menu should build not only on Kosovo’s country-specific experience but also on the larger international experience. Evidence suggests that some ALMPs have larger success rates for specific targeted populations (e.g., youth), particularly interventions that emphasize human capital formation and provide a mix of training and job experience. It is recommended that the design of effective ALMPs must: (i) ensure frequent and robust needs assessment analysis of the private sector, (ii) involve employers’ and workers’ organizations in ALMPs to improve job prospects, (iii) increase the role of ALMPs especially for youth and women facing social exclusion, (iv) include a package of services that address both demand and supply of the labor market which will require working directly with the private sector on the delivery of programs, and (v) ensure that interventions are evidence-based; the process of expanding and scaling down programs must respond to the close monitoring and careful evaluation of the different programs. 20 40. Going forward, it will be important to ensure alignment of social assistance, war-related benefits and labor market programs. The generosity provided from war-related benefits in particular and the strict eligibility criteria for the last resort social assistance program both discourage employment in the formal labor market and encourage participation in the informal sector. 21 HEALTH SPENDING AND THE INTRODUCTION OF HEALTH INSURANCE Health Spending and Financial Protection 41. Public health expenditures account for 71.1 percent of total health expenditures in Kosovo and the share of out-of-pocket payments is relatively high. Private spending, in the form of out-of-pocket (OOP) payments at the point of service, contributes an estimated 28.9 percent in 2015 of total spending in the sector.11 This proportion of OOP payments is similar to the average for Western Balkans countries, but high in comparison with the EU average. In 2014, the OOP expenditure represented on average 27.3 percent of total spending in Western Balkans countries, and 13.9 percent share in the EU.12 42. Kosovo does not protect its citizens from the financial costs associated with OOP payments on health, a guideline established by the World Health Organization (WHO). Financial protection is typically measured in terms of OOP payments against a household budget threshold. If out-of-pocket health spending exceeds this certain threshold of total household or total non-food consumption, then it is considered “catastrophic”. Health spending high enough to make a household fall below the poverty line is considered “impoverishing”. Kosovo fails to meet the WHO’s criterion for financial protection, as its share of OOP payments in total health spending exceeds the recommended upper limit of 15-20 percent. 43. The largest share of OOP payments covers the cost of pharmaceuticals and other medical goods. In 2015, individuals spent an average of €37.7 per capita on health services in Kosovo. The largest share of out-of-pocket spending went towards the purchase of drugs. As illustrated in Figure 7, about 70.1 percent of total OOP payments were for pharmaceuticals and other medical devices.13 Figure 7: OOP Payments, 2015 5.2%0.4% 20.0% Pharmaceuticals Public Health Services Private Health Services 4.4% Health Services Abroad Other 70.1% Source: World Bank Staff calculations using HBS 2015. 11 World Bank Staff calculations using HBS 2015. 12 HFA-DB (WHO). 13 World Bank Staff calculations using HBS 2015. 22 44. OOP payments of the wealthiest quintile are higher than those of the poorest quintile, and are moderately progressive, although this may reflect self-rationing by the poor. In 2015, the average OOP expenditure averaged €19.2 for those in the poorest quintile, and €74.8 in the wealthiest quintile (Figure 8). Individuals below the poverty line spent about €31 on health OOP payments, compared to €39 on average for individuals above the poverty line. The Kakwani index measures financing progressivity as the difference between the concentration index and the gross consumption Gini coefficient. In 2015, the Kakwani index was approximately 0.06, indicating that wealthier individuals spent a slightly greater share of their income on OOP payments for health than the poor. OOP payments totaled 3.5 percent of total consumption expenditure for the bottom quintile, and 4.2 percent for the top quintile (Table 5). This suggests that the poor may forgo treatment due to limited payment capabilities. Figure 8: Average OOP health expenditure per capita, by consumption quintile, 2015 80 74.8 70 60 50 40.0 Euros 40 29.8 30 24.5 19.2 20 10 0 Bottom Second Third Fourth Top Source: World Bank Staff calculations using HBS 2015. Table 4: Progressivity of OOP Payments Table 5: OOP Payments as a Share of Total Consumption Per capita consumption quintile 2015 Per capita consumption quintile 2015 Q1 (lowest) 10.2 Q1 (lowest) 3.5 Q2 13.0 Q2 3.1 Q3 15.8 Q3 3.0 Q4 21.2 Q4 3.3 Q5 (highest) 39.8 Q5 (highest) 4.2 Concentration Index 0.294* Total 3.5 Kakwani Index 0.060* Source: World Bank Staff Calculations using HBS 2015 Source: World Bank Staff Calculations using HBS 2015 and ADePT. and ADePT. Note: *Index is significant at 5%; **index is significant at 1%. 45. Catastrophic spending incidences are more common among the poor. In 2015, 8.6 percent of households exceeded the 10 percent threshold of total household consumption on health. Relative to non- food consumption, the share was 25.5 percent in 2015 (Table 6). The concentration index indicates how 23 concentrated catastrophic spending was across various wealth groups. In 2015, catastrophic payments were somewhat more concentrated among the poor (Table 6). Table 6: Incidence and intensity of catastrophic OOP health payments, 2015 Threshold share of total household consumption 5% 10% 15% 25% 30% 40% 2015 Headcount 21.7 8.6 3.6 1.2 0.8 0.4 Concentration index -0.054* -0.024* 0.059 0.249 0.241 0.615 Threshold share of non-food consumption 5% 10% 15% 25% 30% 40% 2015 Headcount 41.2 25.5 16.1 6.9 4.9 2.1 Concentration index -0.048* -0.065* -0.096* -0.113 -0.070 0.048 Source: World Bank Staff Calculations using HBS 2015 and ADePT. Notes: *Concentration Index is significant at 5%; **Concentration Index is significant at 1%. For example, 25.5 percent of the households spent more than 10 percent of their non-food expenditure on health and the lowest wealth quintile was overrepresented in these households. 46. Health expenditures are impoverishing a number of households. In 2015, 17.6 percent of the population of Kosovo was below the national poverty line. However, if health expenditures are considered essential and the poverty threshold is adjusted accordingly, the poverty headcount increases to 18.8 percent, which represents a 9 percent increase. This indicates that about 2 percent of the population was not counted as living in poverty but would have been classified as such if health payments were taken into account. The estimated poverty gap in 2015 also increased – by about 17 percent – when health payments are taken into account. Deducting health payments from non-food expenditures, the normalized mean positive gap increases by about 6.6 percent – from 21.2 percent to 22.6 percent (Table 7). Overall, this suggests that the widening poverty gap is primarily due to households falling below the poverty line rather than a deepening of poverty among the already poor.14 Table 7: Poverty Impact of Out-Of-Pocket Health Expenditures, 2015 Net of health Gross of health payments Change % change payments 2015 Poverty headcount 17.6 18.8 1.6 9.3% Poverty gap 24.1 28.2 4.1 17.0% Source: World Bank Staff Calculations using HBS and ADePT. Note: Poverty headcount is measured as the fraction of people under the poverty line. Poverty gap is the mean shortfall of the total population from the poverty line. 47. Middle-class households are the most vulnerable to impoverishing health expenditures. While health payments are largest at higher values of total consumption, it is the households in the middle of the consumption expenditure distribution that are most likely to drop below the poverty line as a result of health payments (Figure 9).15 14 Poverty gap reflects the depth of poverty as well as its incidence. The poverty gap increases both because those already counted as poor appear even poorer once health payments are netted out of household resources and because some who were not counted as poor on the basis of gross expenditures are assessed as poor after OOP payments are taken into account. 15 For each household, the vertical bar (or “paint drip”) shows the extent to which health payments reduce consumption. 24 Figure 9: Pen’s parade of per capita expenditures gross and net of OOP payments, 2014-2015 2014 2015 pre-OOP consumption pre-OOP consumption 4 post-OOP consumption 4 post-OOP consumption Consumption as multiple of PL Consumption as multiple of PL 3 3 2 2 1 1 0 0 0 .2 .4 .6 .8 1 0 .2 .4 .6 .8 1 Cumulative proportion of population, ranked from poorest to richest Cumulative proportion of population, ranked from poorest to richest Source: World Bank Staff Calculations using HBS and ADePT. Potential Impact of the Introduction of Health Insurance 48. The Health Insurance Law (HIL), which was approved in 2014, provides the legal basis for the creation of a social insurance scheme and the associated Health Insurance Fund (HIF). The scheme is expected to raise additional revenues for the health sector and increase financial protection (particularly of the most vulnerable) through risk pooling. It will be financed through a combination of payroll tax revenues and flat- rate citizen premiums. The coverage will be mandatory, with penalties for those who do not contribute, although households receiving last resort social assistance and other categories of the population will be exempt from premiums and co-payments. The initial benefit package will include outpatient drugs. 49. A new outpatient drug benefit would likely be the first sign of progress achieved under the HIL. As discussed earlier, out-of-pocket expenses can be large, and poor and middle-class households are vulnerable to impoverishing health expenditures. The largest share of OOP spending is on medications and reached 70.1 percent of OOP in 201516. In addition, there is some evidence that poor households under- spend on health, with both individual and social costs. The social health insurance scheme should initially consist in a limited outpatient drug benefit, which would be highly visible and pro-poor. The cost of this is difficult to estimate: while patients currently spend an estimated €47 million per year on OOP for pharmaceuticals17, the total amount will likely rise when the cost for households reduces following the introduction of the benefit. The benefits package should then be gradually expanded. 16 World Bank Staff Calculations using HBS. 17 World Bank Staff Calculations using HBS. 25 Recommendations 50. The design of the reform is in line with the objectives to improve health outcomes and financial risk protection, especially for the poor. Some progress has been achieved in the development of the outpatient drug benefit scheme, but key decisions on the parameters of the scheme have yet to be made, including (i) Defining the reimbursement list—which medicines will be covered; (ii) Defining eligible populations – providing the same benefits to everyone may not be affordable; (iii) Defining prices and/or reimbursement rates, budget caps or other rules to control spending; (iv) Setting rules for co-payments, with exemptions and limits for vulnerable populations; (v) Defining eligibility of pharmacies to be included in contracts with HIF; and (vi) Setting rules for the compensation of pharmacists and distributors, substitution rights, etc. Urgent improvements to the HIF institutional capacity and immediate development of its information system will also be crucial to the health insurance scheme implementation. This includes an insured population management module allowing for automatic and (semi) manual manipulations on groups of records or individual records and a transaction and payment module, which are prerequisites for the establishment of a full-fledged scheme. Finally, there is consensus among the Government that exemptions from premium contributions should rely on a combination of means-testing and proxy-means testing. But, the eligibility system is yet to be operationalized, which will require, inter alia, the definition of detailed operational procedures, capacity building of the Centers for Social Work (including training and procurement of equipment), and the development and deployment of adequate software. This is critical for ensuring that the reform benefit the poor and vulnerable population. 26 CONCLUSIONS 51. Increased spending on pensions and war-related transfers served to almost double social protection spending as a share of GDP between 2008 and 2016. Recent social policies, mainly driven by the electoral cycle, have increased the fiscal burden of social spending and squeezed budget resources for productive capital investments. The introduction of war veteran benefits to a larger than expected number of beneficiaries disrupted fiscal stability in 2016 and triggered legislation to cap total spending on this program at 0.7 percent of GDP. 52. At the same time, spending on programs for the poor and other vulnerable groups comprised a smaller proportion of the budget than in neighboring countries. The new and expanding war veterans’ scheme and pension increases do not directly target the poor. Moreover, they have crowded out the existing last-resort social assistance targeted and means-tested to the poor. The poverty rate in Kosovo is high, with 17.6 percent of the population living on less than €1.82/day, and Kosovo’s social assistance overall is poorly targeted, resulting in little reduction or mitigation of poverty. As discussed, active labor market programs received a smaller share of total funding than in similar countries. Also, health spending is a leading cause of falling into poverty, including for the middle class, but neither social programs nor health insurance schemes have been put in place to mitigate the situation. 53. The recent expansion of non-targeted social programs has not effectively reached poor and vulnerable populations and also led to concerns about fiscal sustainability. The fiscal burden on social spending has almost doubled as a share of GDP between 2013 and 2016 but was primarily driven by new benefits to war veterans of KLA and the introduction of four sub-categories of former ex-contributory pensions to the ex-YU pay-as-you-go system. This has triggered some action to preserve fiscal stability by imposing a 0.7 percent of GDP budget ceiling for war veteran spending, however this measure was not implemented in 2017. Additional steps must be taken to preserve fiscal sustainability and ensure social policy serves to alleviate poverty. Reform options need to be costed and impacts simulated to improve their effectiveness in protecting the poor. The following recommendations are therefore made in this note: Pensions The Government should make efforts to contain the fiscal costs of the burgeoning pension system. The Government has already made efforts to limit pension spending by moving to a defined contribution scheme. To further this work, the Government is advised to focus on two tracks: (i) limiting the costs of the additional benefits that have already been granted; and (ii) blocking the granting of new benefits. Limiting the costs of additional benefits already granted will take a double-pronged approach: a. working to ensure that all elderly residents in Kosovo are entitled to the basic pension and estimating its cost; for all additional pensions, a limit of around 1 percent of GDP can be imposed, with all currently granted benefits prorated to fit within that limit; b. to further limit costs, the benefit entitlement should be fixed at today’s nominal terms to allow for a downward adjustment that fits within the 1 percent limit. The Government must also 27 work to help residents understand how the granting of additional benefits to additional groups results in fewer benefits for all. War-related benefits The Government has limited options in the short run aside from prorating benefits to recipients across all sub-categories to fit within the agreed upon budget constraint. The government should implement the amendments to the war veterans law to correct the fraudulent cases and maintain the fiscal pressures arising from this scheme. It will be important for residents to understand that increases in the number of claims are associated with lower benefits for all. The burden that these payments put on the budget also lowers the growth potential in the economy, by limiting other developmental expenditures that could promote growth and poverty reduction. No additional benefits should be implemented, apart from the ones already in existence. Social assistance scheme The Government must work to increase the effectiveness of Kosovo’s last-resort social assistance program in protecting the poorest and most vulnerable groups, including by widening its coverage and increasing spending, preferably at the expense of existing non-targeted programs. Targeting accuracy could be improved further by introducing a refined targeting methodology based on a combination of means and proxy-means testing. Importantly, the design would allow for the accounting of certain forms of informal incomes, as well as the capturing of private transfers. The removal of strong exclusion criteria (specifically when the youngest child turns five) would improve coverage of the poor without compromising targeting accuracy. The benefit amounts could be reconsidered to at least lift households above the extreme poverty line of EUR 1.30 per person per day. The current scheme offers a number of disincentives for taking up work, but improvements in the targeting formula and in-work benefits, for example, could serve to alleviate some of these disincentives. Labor market programs Effective intermediation services and ALMPs are instrumental in reducing unemployment and improving the employability of the labor force, especially for women, youth, and the long-term unemployed who lack informal networks. As a first step, it would be important to enhance institutional structures and systems to strengthen the functioning of the public employment services, and to leverage the provision of services through non-public providers (e.g., based on performance-based contracts). Second, the Government must improve both intermediation and activation measures through evidence-based design and rigorous evaluation of services and programs. The Government needs to ensure appropriate financing for intermediation services and ALMPs. The design of an ALMP menu should not only build on Kosovo’s country-specific experience but also on the large international experience. There is evidence to suggest that some ALMPs have greater success for specific targeted populations (e.g., youth), especially interventions that emphasize human capital formation and combine training with job experience. Factors to be reviewed and considered in the design of effective AMLPs for Kosovo include: (i) ensuring frequent and robust needs assessment analysis of the private sector, (ii) involving employers’ and workers’ organizations in ALMPs to improve job prospects, 28 (iii) increasing the role of ALMPs especially for youth and women facing social exclusion, (iv) including a judicious package of services that address both demand and supply of the labor aspects which will require working directly with the private sector on the delivery of programs, (v) ensuring that interventions are evidence based; the process of expanding and scaling down programs must respond to the close monitoring and careful evaluation of the different programs. Going forward, it will be important to ensure the alignment of social assistance, war-related benefits and labor market programs to help eliminate disincentives for labor market participation, by for example adapting the benefit amount to align it with available fiscal resources as well as with other social programs and average wage or poverty line. Health reform The design of the reform is in line with the objectives to improve health outcomes and financial risk protection—particularly for the poor. Some progress has been made in the development of the outpatient drug benefit scheme, but key decisions on the parameters of the scheme have yet to be made, including: (i) Defining the reimbursement list—which medicines will be covered; (ii) Defining eligible populations – providing the same benefits to everyone may not be affordable; (iii) Defining prices and/or reimbursement rates, budget caps or other rules to control spending; (iv) Setting rules for co-payments, with exemptions and limits for vulnerable populations; (v) Defining eligibility of pharmacies to be included in contracts with HIF; and (vi) Setting rules for the compensation of pharmacists and distributors, substitution rights, etc. Urgent improvements to the HIF institutional capacity and the immediate development of its information system will also be crucial to the health insurance scheme implementation. This includes an insured population management module allowing for automatic and (semi) manual manipulations on groups of records or individual records and a transaction and payment module, which are prerequisites for the establishment of a full-fledged scheme. Finally, there is a consensus among the Government that exemptions from premium contributions should rely on a combination of means-testing and proxy-means testing, but the eligibility system is yet to be operationalized. This is critical ensure that the reform benefits the poor and vulnerable population. 29 REFERENCES ASPIRE: The Atlas of Social Protection Indicators of Resilience and Equity: http://datatopics.worldbank.org/aspire/ Csaba Feher, La-Bhus Fah Jirasavetakul, Alain Jousten, February 2016, Kosovo Enhancing Social Protection Cash Benefit, International Monetary Fund, Fiscal Affairs Department. IMF Country Report No. 16/123 Grosh, Margaret, Carlo del Ninno, Emil Tesliuc and Azedine Ouereghi. 2008. For Protection and Promotion: The design and implementation of effective safety nets. Washington, DC: World Bank Kosovo Agency for Statistics and World Bank. 2017. Consumption and Poverty in the Republic of Kosovo 2012-2015 Kosovo Agency of Statistics, Household Budget Survey databases Organisation for Economic Co-operation and Development (OECD). 2015. OECD Employment Outlook 2015 Social Protection Expenditure and Evaluation Database (SPEED) World Bank. 2016. Republic of Kosovo Systematic Country Diagnostic World Bank and Vienna Institute for International Economic Studies (wiiw). 2018. The Western Balkan Labor Market Trends 2018 30