Selected Issues


Selected Issues

Improving Social Benefit Effectiveness1

While Kosovo spends a relatively high amount of resources on social welfare, the performance of benefits lags along nearly every metric. The shortcomings include low coverage, weak targeting of the poor, insufficient adequacy, and low poverty and inequality impacts. Benefits using categorical targeting – 93 percent of the total—contribute to the sub-optimal outcomes. Kosovo needs to reform existing programs and consider introducing new schemes focused on the bottom of the income distribution to meet its objectives of improving human capital and wellbeing. With the potential for worsening inequality in the future and among the highest poverty and unemployment rates in Europe, the reforms are made more urgent.

A. Overview

1. This paper analyses the distributional and labor supply effects of social benefits in Kosovo. The objectives of the analysis are to examine trends in social welfare spending over time, the distributional effect of such spending on poverty, inequality and employment, and the efficiency.

2. Despite Government commitments and moderate spending levels, Kosovo has among the highest poverty and unemployment rates in Europe. The National Development Strategy prioritizes economic growth by maximizing employment and welfare, and Kosovo spends above average amounts, with 5.9 percent of GDP on social benefits, and 2.9 percent on social assistance.2 For social assistance, this compares to 2.4 percent of GDP in the Western Balkans and 2.2 percent of GDP in Emerging Europe.3 Approximately one-sixth of the population, however, lives below the poverty line, and one-third of the working age population is unemployed, the highest rate in Europe.

3. Social benefit scheme outcomes underperform neighboring countries on most metrics, resulting in weak distributional outcomes for the amount spent. Low coverage and adequacy result in among the lowest inequality and poverty impacts of social benefits, despite Kosovo being one of the top spenders. Only 7 percent of benefits are targeted to the poor, while the remaining 93 percent are based on categorical targeting such as age or involvement in the war. Further, the design of existing policies with many schemes requiring beneficiaries to be unemployed lower incentives for employment, thereby constraining welfare gains and aggregate economic growth.

4. The analysis sub-categorizes social schemes into age, health and family pensions, war-related benefits, and poverty-targeted. Approximately 60 percent of benefits are for the first grouping, including those related to retirement, disability, and the family pension. Just over 30 percent of benefits are for war-related schemes. Less than 10 percent of benefits - one scheme - seeks to reduce poverty.

5. Based on this and other assessments, wide-ranging reforms are needed to social welfare in Kosovo. The reforms should be guided by an overarching strategy that sets the desired priorities, and well as a legal framework that ensures the durability of reforms while maintaining some flexibility where essential.

B. Socio-Economic Background

6. Despite some progress, Kosovo ranks among the bottom for socio-economic indicators, including pervasive poverty and unemployment. Median income in Kosovo is around two-thirds that of Emerging Europe and seven-tenths that of the Western Balkans averages. Kosovo has worse absolute poverty relative to the Western Balkans for all measures, including the poverty headcount, gap, and severity. This means that poverty is more pervasive, the poor are further below the poverty line, and there is more extreme poverty. Inequality is comparatively low and below both regional benchmarks. The unemployment rate is double the Emerging Europe average.4


Poverty and Unemployment Measures, Kosovo and Comparators

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff estimates based on World Bank PovcalNet, WDI and 2015 LFS.

7. While measures are improving, a large share of Kosovars are poor or vulnerable to poverty. The poverty headcount rate using the national poverty line fell nearly five percentage points from 2012 to 2015 to 17.6 percent, and so more than one in six Kosovars remain poor. Further there is high vulnerability to being poor with the poverty headcount varying each year by at least 3.5 percentage points over this time period, highlighting the annual cyclicality.5 Additionally, many Kosovars live just above the poverty line and are very vulnerable to poverty, as an increase in the poverty line of only 10 percent would make one in four people poor.6 Poverty in rural areas is slightly higher at 18.9 percent of the population in 2015 compared to 15.5 percent in urban areas, and 65.7 percent of the poor reside in rural areas. There is considerable regional variation with more than 30 percent of the population poor in Peje and Ferizaj, while less than 10 percent are poor in Gjakove and Prizen. The highest shares of the poor are in Pristina and Ferizaj with shares of 23.1 and 20.5 percent respectively.

8. Inequality is low and has declined in recent years. The Gini coefficient was 0.26 in 2012 and fell to 0.23 in 2015. Inequality is higher in urban area by 2.4 Gini points. There is substantial regional variation in inequality, lowest in Prizren and highest in Gjilan. By quintile, more than a quarter of the richest quintile reside in urban areas. The rural population is relatively equally distributed across all but the top quintile, with only 16.4 percent of the rural population in this group.

9. An unemployment rate that is double the EM Europe average lowers economic welfare and growth. With 30.6 percent of the labor force unemployed in 2017, Kosovo severely lags EM Europe and the Western Balkans with rates of 16.2 and 21.1 percent respectively. Inactivity rates are also high at 61.8 percent of the population age 15-64 and close to 80 percent for females. This compares to an average of 40 percent for the Western Balkans. With nearly two-thirds of the working age population not engaged in employment, the lack of labor income increases the risk of poverty, limits revenue mobilization though PIT, VAT, and CIT, and reduces the growth potential of the economy. Among household members that have a recipient of social benefits, only 29.9 percent are employed, while 70.1 percent are not employed, suggesting social payments discourage work, and issue that will be examined in more detail later. Further there are currently no passive unemployment or active employment benefit programs.


Gini Coefficient by Region

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff estimates based on HBS 2015.

C. Fiscal Context

10. The composition of social spending has significantly changed, pension and social assistance increasing in relative size. At the aggregate level, social spending has been an expanding component of expenditure in recent years, with its share of primary current spending increasing by 1.7 times in the period 2010-2017. However, the overall trend masks heterogeneous developments of the two main components. While pensions and social assistance expenditure as a share of GDP has nearly doubled, other transfers and subsidies have been declining.7


Primary Current Expenditure Composition

(shares of primary current expenditure)

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff calculations on MoF data.

Government Expenditure – Social Spending

(Millions of Euros; 2010-2017)

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff calculations on MoF data.

11. When examining social assistance spending relative to peers, spending is higher than both the EM Europe and the Western Balkans average. Kosovo spends 2.9 percent of GDP on social assistance, above the 2.4 percent and 2.2 percent averages for the Western Balkans and EM Europe respectively. Of the 10 countries in the EM Europe sample, Kosovo social assistance spending ranks the fourth highest.


Government Expenditure – Social Assistance

(Percent of GDP; 2017 or most recent year)

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff calculations based on MoF. IMF and World Bank data.Notes: 1/ No data for Macedonia.

12. Spending on social benefits has more than doubled since 2010. The rate of increase for social spending has been rapid, coinciding with new schemes, expanded number of beneficiaries, and higher benefit amounts. Social benefit spending, excluding administrative costs, increased by 124 percent between 2010 and 2017.


Social Benefits

(Millions of euros)

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff calculations based on data from authorities.

13. The composition of spending is heavily skewed towards categorical benefits. Only 7 percent of benefits are allocated based on economic welfare, which explains much of the weak distributional performance. Almost two-thirds of benefits are for age, health, and family benefits, while close to 30 percent are for war-related benefits.


Social Benefits Amounts by Categories

(in percent)

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF Staff calculations based on data from authorities.

D. Performance of Social Benefits


Social Assistance Spending and Gini Reduction

(percent of GDP and in percent)

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff estimates based on IMF, World Bank and HBS 2015.

14. While expenditure on social assistance is high by regional standards, the distributional effect is weak. Among the 10 countries in EM Europe, Kosovo spends the fourth highest on social assistance as a share of GDP. The estimated impact on inequality, however, ranks seventh lowest. The distributional efficiency of social assistance spending in Kosovo is therefore quite weak relative to its peers. Further social benefit receipt appears to have a negative effect on employment, likely due to some benefits requiring recipients to be unemployed.

Employment Status and Benefit Receipt, Kosovo

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff estimates based on HBS 2015.

15. The combination of comparatively low adequacy and scarce coverage of lower income individuals result in social protection having a limited impact on poverty and inequality. The performance of social benefits in Kosovo lags on nearly all metrics when comparing against EM Europe. The adequacy of benefits ranks 10th lowest of the 24 countries, while Kosovo has the fourth lowest coverage of the poorest quintile at 58.4 percent, compared to a regional average of 75.3 percent. Kosovars are therefore receiving relatively low benefit amounts. As a result, the poverty effect from social protection transfers Kosovo is 7 percentage points lower than the regional average reduction of 52 percent, placing it in the bottom 30 percent of the countries. The distributional impact is also relatively weak, reducing the Gini coefficient by 23.1 percent, 4 percentage points below the EM Europe average.8 Note that the performance is likely worse in 2017 due to the further expansion of non-poverty targeted transfers, namely the war veteran pension. Receipt of benefits appears to have perverse work effects.

Social Assistance Performance Measures, Kosovo and EM Europe

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source: IMF staff estimates based on HBS 2015 and World Bank.

16. There are a number of issues and reforms that could markedly increase the impact of social benefits. While each benefit scheme attempts to improve social outcomes, the results are mixed. Age and disability benefits respectively suffer from weak eligibility checks and inequitable designs. A large share of benefits are allocated to war-related benefits, despite the weak correlation with war participation and well-being. The one scheme focused on poverty -social assistance – suffers from weak targeting and low benefit amounts. Further, there are currently no large labor market program despite the very high unemployment. The table below, based on international best practice, highlights some of Kosovo’s social benefit shortcomings and possible reforms options.

Table 1.

Social Benefit Issues and Reform Options

article image
article image
Source: IMF staff

17. Remittances have a large distributional effect, though these are not public transfers, and represent a high-risk form of social protection. In 2015, remittances had a higher distributional effect, as measured by the Gini reduction, than social assistance, though lower effect than social benefits since the larger effect of social insurance improve social protection overall. Private transfers reduced inequality by 3.2 Gini points, compared to 1.3 points and 5.6 points for social assistance and insurance respectively.

Figure 1.
Figure 1.

Kosovo: Social Benefits and Remittances

Citation: IMF Staff Country Reports 2018, 031; 10.5089/9781484340509.002.A002

Source:IMF calculations based on 2015 HBS

E. Conclusion and Policy Recommendations

18. Social benefit spending in Kosovo has increased substantially since 2010 though the performance for the expenditure level remains weak by regional standards. This is due to low coverage, weak targeting and insufficient adequacy of many benefits. Further the composition of spending has shifted towards war-related rather than a poverty, inequality and employment enhancing focus. Social benefit reform is macro-critical to ameliorate future increases in inequality, while addressing current widespread unemployment and low income of individuals, which should increase aggregate growth. A series of policy recommendations should be considered.

All schemes:

  • Set overarching goals and parameters for social benefits, including increasing coverage and benefit levels for bottom of distribution.

  • Introduce an enforceable residency eligibility requirement, such as requiring individuals to pick-up benefits in person every month.

  • Remove disincentives to work by removing requirement of being unemployed to receive a benefit, the exception being for employment benefits.

  • Subject all benefits to personal income tax.

Age, disability and family benefits:

  • Apply uniform disability criteria across schemes and equalize benefit amounts downwards towards civilian scheme.

  • Revise occupational scheme from general government financing to defined contribution and base benefit amounts on career average earnings.

  • Prevent any reforms that would allow for early retirement, unless actuarily and budget neutral.

Poverty focused benefits:

  • Index benefits to inflation to avoid the erosion of value over time.

  • Reform the current social assistance scheme to improve the poverty-reduction goal by increasing the benefit amount and improving targeting criteria - to increase the share of beneficiaries and benefits allocated to the bottom of the distribution - by removing categorical requirements for age, unemployment, and that recipients are a family.

War-related benefits:

  • Enforce the 0.7 percent of GDP budget cap on the veteran’s pensions, and credibly complete reclassification and verification of beneficiaries.

  • Restrict benefits to the current generation.

  • Consider one-off payments in lieu of permanent benefits.

See the May 2016 IMF TA report for other useful recommendations on how to improve social assistance - “Kosovo : Technical Assistance Report-Enhancing Social Protection Cash Benefits”,

Appendix I. Assessing Social Benefit Performance—Methodological Issues

1. This appendix provides details on how household survey data are used in the report for the assessing social benefit performance. The general methods of survey analysis are first discussed, followed by an explanation of their application.


2. Estimates of social assistance performance are based on data from the most recently publicly available HBS household survey. Surveys such as the HBS contain useful information on the demographic characteristics of individuals and households, as well as details on their income and expenditure. Statistical techniques allow for the survey to be used to calculate estimates of at the national level, as well as by population subgroups (e.g. region, urban/rural, and decile).1

3. The unit of analysis and topics covered make the survey well-suited for analysis. Approximately 12,000 individuals are included in the survey, which is updated on a quarterly basis and released each year. The unit of observation is the household and individuals in each household.2

4. For the analysis expenditure is used as a proxy for income. Expenditure is widely viewed as a more accurate measure than income of economic well-being, particularly in developing countries, for reasons including the presence of large informal economies.3 As survey information is collected at the household level, household expenditure is divided by the number of per adult household members to obtain the welfare indicator (expenditure per adult equivalents). Quintiles and deciles are derived from per capita household expenditure, where the bottom decile (D1) or quintile (Q1) represent the poorest 10 and 20 percent of the population respectively, and the best off are D10 and Q5. For poverty analysis, the poverty line is set at 17.6 percent.

5. Sampling techniques allow for the estimation of statistics representative at the national level. A two-stage random sampling design was used by the Statistical Office for data collection, which first included random selection of Primary Sampling Units (PSUs) and then randomly selecting households within the PSUs.

6. As any data source, household surveys have some limitations. There is a risk of systematic errors that decrease the representativeness and precision of results. There may be low response rates for certain questions, that reduces the precision of estimates. There may be intentional mis-statements by respondents, such as the level of income, more common at the top of the distribution.

Social Assistance Assessment

7. A range of performance indicators are used to assess the performance of social assistance. Coverage is defined as the share households that receive a benefit. Adequacy (or Generosity) is the benefits as a share of expenditure. For example, if expenditure is 1,000 Euros and a transfer is 100 Euros, then the generosity would be 10 percent (100/1,000). Benefit incidence describes how benefits are distributed across deciles and quintiles. Beneficiary incidence shows how benefit recipients are distributed across deciles and quintiles. The estimated poverty and inequality impact is calculated as the percentage change in the measures (poverty headcount, poverty gap, severity, and Gini coefficient) if one or all social transfers are removed. Finally, the benefit-cost ratio (BCR) is an efficiency measure, which indicates the share of benefits that reduce the poverty gap. A value of 0.0 would indicate that none of the benefits reduce the poverty gap, while a value of 1.0 means that all benefits reduce poverty.

Appendix II. Improving Targeting Performance

1. The targeting of social assistance transfers improves distributional outcomes by more efficiently allocating resources to those most in need. Unlike universal transfers, targeted social assistance programs seek to concentrate resources on certain population sub-groups, often low-income households. International evidence indicates that top performing programs allocate up to 60-80 percent of benefits to the bottom income quintile. The size of benefits also tends to increase by up to four times compared to if benefits were randomly allocated. As country budgets often have scarce resources, more targeted transfers can result in considerable fiscal savings while achieving more effective distributional outcomes. A list of resources and references on the issue of targeting is provided at the end of this Appendix to support the authorities internal debate on this issue.

2. To improve accuracy, more than one targeting method should be used. There are multiple types of targeting, though some with better accuracy, and programs that layer more than one targeting approach tend to be most effective. Methods include categorical, geographical, self-selection, community-based, proxy-means testing (PMT), and means testing (see Appendix Table 4).1 A program may for example combine geographic targeting, if certain regions are relatively impoverished, and then use PMT in these areas, and community-based targeting for final verification.

3. PMT has been found to be effective in developing countries, where income is difficult to verify. Relying on income, particularly if self-reported, leads to poor targeting with both exclusion and inclusion errors.2 Income often is not an accurate measure of welfare when verification is difficult.3 Verification difficulties arise from large agricultural sectors where accurate income reporting can be challenging, substantial informal non-agricultural employment, high reliance on remittances, and low administrative capacity to check income. Recognizing the inaccuracy of using income in these contexts, many programs use PMT, which utilizes proxies for economic welfare. These proxies tend to be easy to verify, such as housing quality, family composition, location, and education, which can be used to more accurately identify low-income households.

4. Even with accurate targeting, program implementation that minimizes direct and indirect costs is critical to ensure the intended outcomes. There is wide variation in the performance of targeting, largely due to implementation. Administrative costs can be high if a program is very complicated, and so a balance of the trade-off between accuracy and efficiency is needed.4 Efforts should be made to minimize private costs, such as opportunity costs.5 Targeting can lead to perverse incentive effects, such as individuals altering their savings or labor supply behavior to qualify for a program. There may be social costs, namely stigma attached to being in a program, and political considerations may affect the efforts to mobilize public support if a large portion of the population is excluded.6

Table 1.

Targeting Methods, Key Characteristics and Effectiveness

article image
Source: IMF staff based on the World Bank and other literature.

Appendix III. Basic Features of Main Social Benefit Schemes

article image
article image

Indicative amount

Source: IMF based on authorities.


  • Cojocaru, A. 2017. “Kosovo Jobs Diagnostic.” World Bank, Washington, DC.

  • Feher, C., La-Bhus, F., Jousten, A. 2016. “Kosovo Technical Assistance Report - Enhancing Social Proetection Cash Benefits”. International Monetary Fund. Washington D.C.

    • Search Google Scholar
    • Export Citation
  • Government Republic of Kosovo. 2016. National Development Strategy 2016-21 (NDS). Pristina.

  • Kosovo Agency of Statistics. 2016. “Results of the Kosovo 2015 Labour Force Survey”. Pristina.


  • World Bank and Kosovo Agency of Statistics. “Consumption Poverty in the Republic of Kosovo 2012-2015”. Washington D.C. and Pristina.

    • Search Google Scholar
    • Export Citation
  • World Bank and Vienna Institute for International Economic Studies (WIIW). 2017. “Western Balkans Labor Market Trends 2017”. Washington D.C.

    • Search Google Scholar
    • Export Citation

Prepared by Brooks Evans.


Values for 2016. Social benefits include all benefits, while social assistance includes the basic pension, social assistance scheme, SH benefits, and benefits for the blind. Non-contributory and contributory benefits not listed are excluded in this definition of social assistance.


World Bank ASPIRE database,


Absolute poverty measures are based on a $1.90/day PPP poverty line in each country. The poverty headcount is the share of the population below the poverty line, poverty gap the average distance of the poor from the poverty line expressed relative to the poverty line, and poverty severity the squared gap, thereby placing more weight on the poor furthest from the poverty line. Inequality is measured by the Gini coefficient, where a value of 1.0 means that one person in a country has all income, and a value of 0.0 every individual has the same income.


The poverty line in 2012 was €1.79 a day per adult equivalent and €1.82, €664 per year, in 2013-2015.


The poverty line would increase from €1.82 to €2.00 per day, or €731 per year, resulting in a poverty headcount rate of 24.3 percent.


Other transfers and subsidies includes subsidies to public enterprises.


Note that part of the weak inequality effect is Kosovo can be explained by its comparatively low Gini index, which means that more fiscal effort is needed than in a country with higher inequality.


These techniques include sampling for random selection of households and the use of survey weights.


Servants and temporary members are excluded.


Accurate income measurement in the informal sector, which often includes agriculture and services, can be a challenge due to barter trading, less accurate bookkeeping, misreporting, and other factors.


Community based targeting refers to community leaders or members identifying those most in need based on program criteria.


Exclusion errors are benefits not received by the targeted population, while inclusion errors are benefits received by the non-target group. Exclusion errors can lead to frustration among those not receiving a benefit, while excessive inclusion errors may increase questions of the efficient use of funds and fraud.


Reasons include intentional misreporting and the difficulty for many individuals, such as poorly educated subsistence farmers or street vendors, to accurately determine income. Consumption expenditure is therefore often used instead of income in developing countries.


Administrative costs are highest in initial years as the program is set-up and scaled. Efforts should be made to pool functions across programs.


In some cases, high opportunity costs may be desirable for self-selection, though should not disproportionately affect certain sub-groups, such as individuals living far from social assistance enrolment offices.


Social costs can be lowered by awareness campaigns, while political support can be bolstered by measures such as the inclusion of some middle-income households, in addition to low income households.

Republic of Kosovo: Selected Issues
Author: International Monetary Fund. European Dept.