Technical note Options for Child Benefits in Kosovo Synthesis of international experiences and considerations for Kosovo in response to Parliament Resolution on Child Allowances, No. 06-R-0141 April 10, 2019 1This technical note was prepared by Stefanie Brodmann (Sr. Economist, World Bank and Task Team Leader), Boryana Gotcheva (Consultant, World Bank), Florentin Kerschbaumer (Jr. Professional Officer, World Bank), and Erëblina Elezaj (Consultant, World Bank). The team acknowledges financial support from the Rapid Social Response Multi-Donor Trust Fund. Standard disclaimer: This technical note 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 presented in this work. 1 Main messages Kosovo is the second poorest country in Europe, after Moldova, with one in five Kosovars living in poverty. Improved economic conditions in the country have yet to translate into significant poverty reduction. The poor are overrepresented in households headed by women, in households in which the head has not completed secondary education, in households with more children, in larger households (with peak in households of 5 members), and among people whose main source of income is social assistance, specifically Category I Social Assistance Scheme beneficiaries. Child benefits are one of the most common components of the family benefit systems in the European countries, and worldwide. They are part of a broader set of measures aimed at reducing poverty and vulnerability of families with children. Practically all EU countries and five out of six Western Balkan countries (including Kosovo through a child supplement paid to social assistance recipient families with children) have monthly child benefits, either means-tested or universal, or a combination of both. Worldwide, there is a trend toward targeting child benefits to poor families or to both poor and middle- income families and, at the same time, curtailing coverage of wealthier families. In parallel, child benefits in both developed and developing countries are increasingly conditional on school enrollment, regular school attendance and/or compliance with other conditions on behalf of the benefit recipient families. The Social Assistance Scheme is Kosovo’s overarching program for protection against poverty of various vulnerable groups. Beneficiary families of the Social Assistance Scheme receive a child supplement for each child. On November 7, 2018, the Parliament of the Republic of Kosovo adopted a resolution calling on the Government of Kosovo to draft legislation for a state budget financed child benefit, “considering criteria such as family income, employment status of parents, and children’s age”. This note provides an overview of international experiences with child benefits and uses data from the Kosovo household budget survey to assess the impact of various options for protecting children against poverty. To avoid duplicating the existing child supplement, the targeting mechanism and implementation rules for any new child benefit should, at the very least, be coordinated with the targeting design and implementation arrangements of the Social Assistance Scheme. The main findings and recommendations of this note are: • Reforming the Social Assistance Scheme is the most viable and economic option for tackling child poverty in Kosovo, as opposed to a stand-alone child benefit scheme or a child supplement to the Social Assistance Scheme in its current form. • A reform of the Social Assistance Scheme (without an increase in the budget) would increase the number of families with children that are covered by the Scheme by more than 25 percent. Under a budget neutral reform that entails an elimination of the categorical filters, introduction of a new combined means and proxy means test (new targeted scheme), and a revision of the equivalence scales, the poverty gap would be reduced by 12 percent compared with the current Social Assistance Scheme. This implies that a reform of the Social Assistance Scheme would not only significantly reduce child poverty but also reduce the adverse incentives and poverty traps that many existing benefit-receiving families currently face. Moreover, it would also increase the effectiveness of Kosovo’s social assistance system as more money would be devoted to poor and vulnerable families in the country. • Increasing the current child supplement by EUR 10 per child for families covered by the Social Assistance Scheme, without reforming the Social Assistance Scheme, would also contribute to poverty reduction for families: the poverty gap would be reduced by 6 percent but at a significantly higher cost than a budget-neutral reform of the Social Assistance scheme. Therefore, this option 2 would be significantly less cost-effective and reach a significantly lower number of children (25 percent less than under a budget neutral reform of the Social Assistance Scheme). • Reforming the Social Assistance Scheme and increasing the budget would further reduce poverty by increasing coverage of the poor. Increasing the budget and allocating a share as a child supplement to poor children in SAS families and as a child benefit to children in non-SAS families, would significantly reduce the poverty gap. The estimated poverty gap reduction would be 26 percent in the case of a budget increase of EUR 17 million, of which EUR 7 million are allocated among children in SAS families, and 25 percent if the budget is the same and EUR 7 million of it are allocated among children in families which do not receive SAS. • The Government of Kosovo is considering a comprehensive reform of the social assistance system, including a common targeting mechanism for numerous means-tested programs. This would improve targeting accuracy and reduce the high administrative burden associated with using several beneficiary identification methods in parallel. A new targeting methodology could apply different eligibility thresholds/cutoff scores for different poverty targeted programs; thus, less rigorous eligibility rules could guarantee a higher coverage of the child benefit beyond the Social Assistance Scheme recipient families. As an example, reforming the Social Assistance Scheme, increasing its budget by EUR 10 million, and allocating an additional EUR 7 million to poor children who are not covered by the Social Assistance Scheme, would result in an estimated increase in coverage of the poor to 65 percent, assuming a child benefit of EUR 10 per child per month, reaching over 150, 000 (227,000) children; or – assuming a child benefit of EUR 5, would result in an estimated increase in coverage of the poor to 80, reaching over 227,000 children. • Simulations confirm that a universal child benefit would not be the best policy choice for Kosovo. A universal child benefit would have a small progressive beneficiary incidence: less than half of the child benefit recipients would belong to the poorest 40 percent of the welfare distribution and only a quarter to the poorest 20 percent. A universal child benefit would be very costly. The estimated program budget of a universal child benefit if it amounted to EUR 10 per month and was given to all children between 0 and 16 years of age, would be as high as around EUR 60 million per annum for 2019-2021, more than twice the average annual spending on the Social Assistance Scheme for 2008-2017, and close to the generous amount spent on war veteran pensions. 3 Context for the introduction of a child benefit in Kosovo The Republic of Kosovo is considering the introduction of a child benefit; the details of the scheme are yet to be determined. The Parliament of the Republic of Kosovo adopted on November 7, 2018, a resolution calling on the Government of Kosovo to draft legislation for a state budget financed child benefit and to begin implementation in 2019. The resolution does not provide details about the design of the future child benefit; it specifies only the basic principle that the child benefit should be determined “considering criteria such as family income, employment status of parents, and children’s age”2. Historically, a monthly child benefit was not envisaged as a stand-alone benefit in the architecture of Kosovo’s social assistance which was set in the beginning of the 2000s3. Such a benefit was subsequently built into the design of the Social Assistance Scheme (SAS) with introduction of a supplement of EUR 5 per child per month for each child in a SAS recipient family. The SAS is Kosovo’s overarching program for protection against poverty of various vulnerable groups, including families with children. From this perspective a new child benefit should take into consideration the design of the SAS to avoid duplicating the already existing child supplement, and its targeting and implementation mechanisms should be coordinated and aligned with the design and implementation arrangements of the SAS scheme. Having two or more children increases the respective household’s or family’s risk for poverty. According to Kosovo’s most recent poverty profile4, the poverty rate among families with one or two children is similar to the average poverty headcount rate of 17.6 percent. The poverty rate among families with three and more children, however, is 22.2 percent and thus significantly higher than the average poverty headcount rate. High poverty rates are also observed among households headed by women (23.7 percent) and larger households (the peak being at five members with 20.6 percent)5. These are presumably family/household configurations with more children. The poverty rate for students and pupils is also higher than the average. The SAS has many inbuilt features that severely limit its poverty-reduction impact for families with children and pose adverse labor market incentives for working-age family members. The current SAS legislation has inbuilt categorical restrictions and near-exclusionary filters6 which restrict access to SAS to families with at least one child under the age of five (Category II). There is no evidence to suggest that the cost of childrearing decreases after that age. Excluding families with older children only constrains the poverty- reduction impact of the scheme. Furthermore, SAS benefit conditionality for Category II requires all but one household member to be classified as dependent, and the single able to work member to be 2 Parliament Resolution on Child Allowances, No. 06-R-014. 3 The key structural components of social assistance in Kosovo were introduced with a UNMIK (United Nations Interim Administration Mission in Kosovo) resolution of 2000, followed by Law No. 2003/15 on the Social Assistance Scheme in Kosovo, Law No. 2003/23 on Disability Pensions in Kosovo and Law No. 02/L-17 of 2005 on Social and Family Services. The key structural components of social assistance include: (i) last resort income support program – the Social Assistance Scheme (SAS); (ii) material support for foster families; (iii) assistance for exceptional needs; (iv) energy/electricity subsidy; (v) material support for families with children (of ages 0-18 years) with permanent disability; (vi) social and family services. 4 Kosovo Agency of Statistics and the World Bank. Consumption Poverty in the Republic of Kosovo 2012-2015, April, 2017. In 2015, the national poverty line was €1.82 per adult equivalent per day and the extreme poverty line €1.30 per adult equivalent per day. 5 Ibid. 6 According to Law No. 2003/15 on the Social Assistance Scheme in Kosovo, eligible for SAS are two categories: Category I – all family members are dependent (persons over 18 years of age with permanent and severe disabilities rendering them unable to work, persons 65 years of age and older, full-time caregivers of person(s)or of children under the age of five, persons up to 14 years of age, persons between the age of 15 and 18 (inclusive) who are in full-time education, and single parents with at least one child under the age of 15) and none is employed; and Category II - families can have one family member who is able to work but registered as unemployed with the Employment Office. All other family members must be dependent. In addition, the family must be parenting at least one child under the age of 5 or providing permanent care for an orphan under the age of 15. 4 unemployed7. This poses adverse labor market incentives for families with children as they often have more than one able to work adult family member. Hence, they can only receive SAS benefits if the second and next adult family members acquire the status of dependent, that is, fulfill Category I eligibility criteria. To the extent that this discourages their labor market participation, it creates a substantive poverty trap. If the child benefit is incorporated in the SAS and the current targeting mechanism is preserved, the exclusion will be replicated and reproduced, and the impact of the child benefit on poverty reduction would be undermined. Reforms to the SAS would reduce its adverse incentives and enhance its poverty-reducing impact. As suggested in a recent World Bank report8, a reform of the SAS would involve replacing the current SAS design9 with a new design characterized by two main features: • A two-stage poverty test with sequential application of a means test and a proxy-means test. At stage one, the means test is an assessment of income from economic activity, social protection schemes of the Ministry of Labor and Social Welfare (MLSW), other government grants and subsidies, and other verifiable sources. At stage two, a new proxy-means test (PMT) is applied to applicants/households that fall below the threshold of the means test. The new PMT is an assessment of proxy indicators of household wealth and includes observable indicators of housing conditions, ownership of assets, and demographic characteristics of households. The indicators and their weights/coefficients are determined based on econometric analysis of household survey data, and households are assigned scores according to which they either qualify or do not qualify for the SAS. • New ‘optimal’ benefit scales based on simulations to maximize the impact of the SAS scheme on poverty reduction, given a certain budget. They replace the current implicit equivalence scales which assume disproportionate economies of scale (i.e., the benefit increases only marginally with household size). 7 Law No. 2003/15 on the Social Assistance Scheme and Law No. 04/L-096 on Amending and Supplementing the Law No. 2003/15 on the Social Assistance Scheme. 8 World Bank. 2019. Kosovo Social Assistance Scheme Study, March 2019. 9 The current SAS design involves categorical identification with application of exclusionary filters (for defining Category I and Category II SAS beneficiaries) and a questionnaire combining income and proxy-means indicators which can be exclusionary and are not based on a rigorous econometric model that links it to poverty. It also involves equivalence scales which are not favorable for larger households as the benefit levels increase only marginally with household size. 5 Two basic models of child benefits worldwide: universal and targeted to the poor There are two prevailing models of child benefits or allowances10 worldwide: universal and targeted to the poor. A universal child benefit is granted to every child irrespective of the income, assets and employment status of the family in which this child is raised. Its proponents emphasize advantages such as promoting horizontal equity and reducing income inequality which could arise from the costs of parenting; also, administrative simplicity and absence of moral hazard. Opponents point to its high cost and low effectiveness in poverty reduction. A targeted child benefit is granted based on specific poverty-related criteria, usually a means test, or other type of poverty test. Its proponents emphasize its capacity to target those most in need and to effectively reduce poverty. The strengths and weaknesses of these models are summarized in Table 1. Table 1: Main characteristics (strengths and weaknesses) of universal and poverty targeted child benefits Universal child benefit Poverty targeted child benefit Ensures basic income security for all children Provides income security for a targeted but narrower irrespective of differences in need group of children in need Costly – requires significant resources to support all Cost effective – results in a larger reduction in poverty children with a meaningful benefit; alternatively – with less spending by focusing the resources to the small benefits for all would have a meager impact on poorest, underprivileged and needy poverty reduction Horizontal equity - all children are treated in the same Aims at vertical equity – only children who have less manner and the state recognizes the contribution (are poor) receive child benefit made by all parents to the task of rearing children Absence of moral hazard – the receipt of benefits High marginal tax rates associated with income-tested regardless of parents’ employment or income does not benefits could reduce incentives to enter and stay in contribute to unemployment or poverty trap; receiving employment; also, there may exist an incentive at the benefits on top of in-work income does not create a margin not to earn more or save if this meant loss of disincentive to work the child benefit Administrative simplicity and efficiency Administratively complex beneficiary identification and increasing complexity as family structures become more fluid Full, or close to full, take-up due to easy verification of Low take-up due, in the case of administrative barriers, entitlement to access, stigma, lack of information, etc. Source: Compiled after: Child Benefits in Central and Eastern Europe. A comparative review / Jonathan Bradshaw, Kenichi Hirose; International Labor Organization, ILO DWT and Country Office for Central and Eastern Europe. Budapest: ILO, 2016 and Child benefits in the European Union, CPAG Poverty Magazine 139. 10 The terms ‘monthly child benefit’ and ‘monthly child allowance’ are used as synonyms interchangeably. 6 Possible child benefit design scenarios for Kosovo Poverty targeted child benefit The Government of Kosovo is considering the introduction of a poverty targeted child benefit. The entry point for selecting design scenario(s) for a new child benefit in Kosovo is the provision in the Parliament’s resolution stating that the child benefit should be determined by, inter alia, family income (i.e., be means tested). In general, family benefits, including monthly child benefits, are ‘loaded’ with multiple objectives such as reducing poverty among families with children, compensating parents for some of the costs of child rearing, and even birth promotion. While there is no robust empirical evidence that the monthly child benefits impact the decision to have more children, they are increasingly considered a viable instrument for reducing poverty among children and boosting investment in human capital11. The long-term social and economic costs of child poverty are estimated as high12. To the extent that child benefits contribute to reducing these costs, they could have a large financial return. As evident from Kosovo’s poverty profile, in the case of Kosovo, there is an argument in favor of greater financial protection for poor children. Greater financial protection for poor children in Kosovo could be achieved in several ways: Option 1: Reforming the Social Assistance Scheme with variations of reform options with respect to financing (budget neutral option and option with increased budget). Option 1a: Reforming the Social Assistance Scheme without increasing its budget (budget neutral) The key elements of the proposed reform involve a new poverty test, elimination of the categorical ‘filters’ (Category I and Category II), and a new equivalence scale which is ‘optimal’ in that it has the strongest effect on poverty given the available budget. 13 This new equivalence scale is more favorable for larger families; it assigns a weight of 1 to the first adult in the family, a weight of 0.25 for each additional adult, and a weight of 0.2 for each child. With this new optimal equivalence scale, the 5 EUR child supplement will be removed. Reforming and modernizing the SAS to eliminate discrimination in program design, promote inclusion, expand coverage, and reduce poverty of social assistance recipients would bring many benefits through poverty reduction. The reform would benefit specifically those who are excluded from receiving social assistance under its current design (i.e., those who do not meet the strict eligibility criteria), namely: poor families with children in which all children are older than 5 years; and families who have more than one able-bodied, but not working, adult member. Through a reform of the SAS, poor families with children would be more adequately covered with social assistance. Option 1b: Reforming the Social Assistance Scheme and increasing budget Ideally, a reform of the SAS as outlined above would go hand in hand with rationalization of funds within the social transfers, and reallocation from categorical benefits towards programs targeted to the poor. Any increase in the budget for the reformed SAS would increase the number of children receiving SAS due to the new poverty targeting mechanism and increase the transfer amount to families of children due to the optimal equivalence scale. 11 The number of countries which provide poverty tested child benefits and also link their provision to conditions like school enrollment and attendance, vaccinations, and regular medical checks of young children with primary health providers. 12 For example, in studies undertaken in the UK, these costs are projected to be over GBP 35 billion, or 3 percent of GDP by 2020. See: Child Benefits in Central and Eastern Europe. A comparative review / Jonathan Bradshaw, Kenichi Hirose; International Labor Organization, ILO DWT and Country Office for Central and Eastern Europe. Budapest: ILO, 2016. 13 World Bank. 2019. Kosovo Social Assistance Scheme Study, March 2019. 7 Option 2: Increasing the child supplement for SAS recipient families without reforming the Social Assistance Scheme Without reforming the SAS, there is little room for increasing support for children. It would only be possible to: (i) increase the nominal amount of the child supplement, or (ii) differentiate it based on the child’s age or birth order. Any increase in the budget for the child supplement without a reform of the SAS scheme would preserve the scheme’s limitations - its limited coverage and significant exclusionary errors would apply equally to the child supplement. The absence of a rule-based approach to determining the amount of the child supplement, as opposed to an equivalence scale, would undermine equity (both within the SAS and child supplement scheme, and across the groups of poor who receive SAS and poor who do not receive SAS). The possible differentiation of the child supplement (e.g., by age or birth order) would not increase the coverage but would, however, unnecessarily increase administrative costs. Option 3: New child benefit which is coordinated with the Social Assistance Scheme in terms of beneficiary identification methodology This is not a stand-alone option per se, rather a next step in the development of Option 1, in which the child benefit can be provided both to poor and possibly low-middle income families that do not qualify for SAS. Within a national social assistance system, the child benefit which is provided based on a means test or another form of poverty test, could be separate and independent from the poverty test used by other schemes (primarily the SAS scheme in the case of Kosovo). Alternatively, the child benefit’s targeting method could be aligned with and coordinated in one way or another with the targeting methodology for the SAS. Both approaches have pros and cons, but there are strong arguments in favor of setting up an overarching and harmonized poverty test (that involves family incomes and assets, and other family characteristics) with possibilities for variations in certain elements of the targeting formula (such as eligibility thresholds, proxy means test scores, demographic and/or categorical criteria). The implementation of such an approach is becoming increasingly feasible with enhancements of the social assistance management information systems and, in particular, the SAS management information system (MIS), and even more so in the case of a future social registry for identification of the poor and vulnerable individuals and families. This approach is viable and can be applied in Kosovo after reforming the SAS. The same scoring formula used for SAS beneficiary identification, but with less rigorous access thresholds/score, could be used to identify children eligible for the child benefit who live in poor but non-SAS recipient families, or in low-middle income families. 8 Simulation results To assess the relative importance of the various recommendations for reform, this section uses microsimulations to evaluate the impact of modifications to the existing SAS.14 Option 1: Reforming the Social Assistance Scheme Reforming the SAS would be the most efficient solution to address child poverty in Kosovo. When simulating a poverty targeted child benefit, the current SAS with its current targeting mechanism and benefit level (EUR 60 for the first adult, EUR 22.5 for the second, EUR 7.5 for each additional household member, plus EUR 5 for each child) is used as a baseline. Then the child benefit performance (in terms of reduction of poverty headcount and poverty gap) is simulated with the new targeting criteria as proposed for the reform of the SAS and an ‘optimal’ equivalence scale15 that is chosen by maximizing the new scheme’s impact on poverty reduction. Table 2 summarizes the simulated impact of all reform options which are discussed further in the note: the changes in SAS performance (in terms of targeting accuracy, adequacy, and coverage); the changes in the number of SAS direct and indirect beneficiaries; and the number of children in and outside SAS families who would benefit from each of the reform options. Table 2: Changes in poverty impact, performance indicators and beneficiary numbers under different SAS and child benefit design and spending scenarios Number Change in Change in Change in Change in Number Number of of Head- Poverty Targeting Benefit Coverage Number of Scheme Head-count Poverty of children in children count Gap Accuracy Adequacy of poor beneficiaries (pp) Gap (pp) Families SAS not in (percent) (percent) SAS Current Scheme 1.776 1.197 0 0 64% 39% 26% 106,416 28,276 55,285 0 Current Scheme with EUR 2.06 1.385 2% 6% 64% 47% 26% 106,416 28,276 55,285 0 15 instead of 5 per child Reformed SAS: optimal 3.642 1.965 13% 12% 76% 53% 40% 150,648 22,289 69,730 0 benefit level Reformed SAS: optimal benefit level plus EUR 7 4.215 2.173 16% 18% 71% 53% 44% 162,582 26,330 80,172 0 million Reformed SAS: optimal benefit level plus EUR 10 4.348 2.216 17% 20% 68% 53% 46% 174,649 28,547 85,969 0 million Reformed SAS: optimal benefit level plus EUR 10 4.759 2.407 20% 26% 68% 61% 46% 174,649 28,547 85,969 0 million, plus EUR 7 million for children Reformed SAS: optimal benefit level plus EUR 10 million, plus EUR 7 million 4.912 2.39 21% 25% 49% N/A 65% 334,893 56,454 85,969 65,211 for children not in SAS (10 EUR/child) Reformed SAS: optimal benefit level plus EUR 10 million, plus EUR 7 million 4.884 2.383 20% 25% 40% N/A 80% 506,561 86,843 85,969 141,108 for children not in SAS (5 EUR/child) Source: World Bank estimations using HBS 2016. 14 A detailed description of the data and methodology can be found in: World Bank. 2019. Kosovo Social Assistance Scheme Study, March 2019. 15 The optimal scale is the following: = 60 ∗ [1 + max( − 1,0) ∗ 0.25 + ℎ ∗ 0.2] 9 Option 1a: Reforming the Social Assistance Scheme without increasing its budget (budget neutral) The number of children in SAS recipient families will increase by over a quarter only by reforming the SAS (with the new beneficiary identification method and revised equivalence scale), even without any budget increase. This reform option is budget neutral (highlighted in Table 2 in orange) and any changes in SAS performance would come solely from changes in its design. Such changes include: • The number of children in SAS recipient families would increase by 26.1 percent (from 55,285 to 69,730 children), or from around 9 percent to around 11.3 percent of Kosovo’s child population under the age of 1916. • Poverty would be reduced significantly – the poverty headcount rate by 12 percent and the poverty gap by 13 percent compared to the baseline scenario (current SAS scheme including a child supplement of EUR 5 per month per child up to the age of 18). • The targeting accuracy (share of SAS budget accrued to the poorest population quintile) would increase to 76 percent (from the current 64 percent). • The SAS benefit adequacy would also increase notably – from 39 percent to 53 percent of the consumption of the poorest quintile. • The coverage of the poorest quintile would increase from 26 percent to 40 percent. • The number of direct and indirect beneficiaries would increase from 106,416 to 150,648. At the same time, this effect would lead to a decrease in the number of SAS beneficiary families (SAS direct beneficiaries) from 28,276 to 22,289. Figure 1 presents the expected changes in numbers of beneficiaries under different reform options for the SAS and child benefits. Figure 1. Changes in number of SAS recipient families and beneficiaries under different reform options Reformed SAS: optimal benefit level plus Euro 10 million, plus 86,843 141,108 85,969 7 million for children not in SAS (5 Euro/child) 506,561 Reformed SAS: optimal benefit level plus Euro 10 million, plus 56,454 65,211 85,969 7 million for children not in SAS (10 Euro/child) 334,893 Reformed SAS: optimal benefit level plus Euro 10 million, plus 0 28,547 85,969 7 million for children 174,649 28,547 Reformed SAS: optimal benefit level plus Euro 10 million 0 85,969 174,649 26,630 Reformed SAS: optimal benefit level plus Euro 7 million 0 80,172 162,582 22,289 Reformed SAS: optimal benefit level 0 69,730 150,648 28,275 Current SAS Scheme 0 55,285 106,416 0 100,000 200,000 300,000 400,000 500,000 600,000 Number of families Number of chilldren not in SAS Number of children in SAS Number of beneficiaries Source: World Bank estimations using HBS 2016. 16The number of children 0-19 is 617,558, according to the population projections of the Kosovo Agency of Statistics (medium variant) for 2017. 10 Option 1b: Reforming the Social Assistance Scheme and increasing its budget The allocation of additional budget to the reformed SAS would be cost-effective in terms of poverty reduction; the SAS performance in terms of targeting will remain good, and – most importantly - would result in a significant increase in the number of children who would benefit from the reformed SAS. The simulated results of reforming the SAS together with increasing its budget (two sub-options: (i) budget increase by EUR 7 million and (ii) budget increase by EUR 10 million, respectively; Table 2, highlighted in blue) indicate that: • The number of children in SAS beneficiary families would increase to 80,172 (by 45 percent, to 13 percent of the child population under the age of 19) in case of budget increase of EUR 7 million, and to 85,969 (by 55.5 percent, to 13.9 percent of the child population under 19 years of age in Kosovo) if the budget is increased by EUR 10 million. • The SAS coverage would increase – respectively to 44 percent and 46 percent of the poorest quintile, compared to 40 percent under the budget neutral option. Coverage of the reformed SAS under the financing scenarios discussed above, as well as under the budget neutral scenario, and compared to the performance of the current SAS are presented in Figure 2 (right). • The SAS benefit adequacy would remain high, similarly to the budget neutral reform option (at 53 percent of the consumption of the poorest quintile). • The SAS benefit incidence (share of SAS budget going to the poorest quintile) would decline to 68 percent and 71 percent compared to 76 percent in the budget neutral option (Figure 2, left). There is a tradeoff between the targeting accuracy and coverage. Increase in coverage increases also the probability of inclusion errors. In this case, despite the decline, targeting accuracy remains high. • The poverty impact of the reformed and more generously financed SAS would increase to 16-17 percent of poverty headcount reduction and 18-20 percent of poverty gap reduction compared to the baseline (current SAS scenario). The schedules of reduction in the poverty headcount and poverty gap under different SAS design and spending scenarios are compared in Figure 4. Figure 2. Comparison between the targeting accuracy (left) and coverage (right) of the reformed SAS (without budget increase and with two options of increased financing) compared to the current SAS TARGETING ACCURACY COVERAGE Percentage beneficary HH that are non-poor Percentage of poor households that receive SAS Percentage of poor households that do not receive SAS Percentage beneficary HH that are poor 120% 80% 74% 70% 100% 60% 56% 54% 24% 29% 32% 60% 80% 36% 46% 50% 44% 40% 60% 40% 30% 26% 40% 76% 71% 68% 64% 20% 20% 10% 0% 0% Current SAS Reformed SAS: Reformed SAS: Reformed SAS: Current SAS Scheme Reformed SAS: Reformed SAS: Reformed SAS: Scheme optimal benefit optimal benefit optimal benefit optimal benefit optimal benefit optimal benefit level level plus Euro 7 level plus Euro level level plus Euro 7 level plus Euro 10 million 10 million million million Source: World Bank estimations with ADePT using HBS 2016. 11 Option 2: Increasing the child supplement for SAS recipient families without reforming the SAS Increasing the current child supplement by EUR 10 per child for families covered by the SAS, without reforming the SAS, would also contribute to poverty reduction for families: the poverty gap would be reduced by 6 percent and the poverty headcount by 2 percent, but at a significantly higher cost than a budget-neutral reform of the SAS. Therefore, this option would be significantly less cost-effective and reach a significantly lower number of children (26 percent less than under a budget neutral reform of the SAS). Other parameters of the SAS would remain unchanged, only SAS adequacy would increase from the current 39 percent to 47 percent of the post-transfer value of consumption in the poorest quintile. Option 3: Reforming the SAS and increasing the SAS budget with variation based on how the budget increase is allocated Reforming the SAS and increasing the budget would further reduce poverty by increasing the coverage of the poor. Increasing the budget and allocating a share as a child supplement would significantly reduce the poverty headcount rate and the poverty gap. These simulations assess the impact of additional financing for the reformed SAS at the amount of EUR 17 million and consider three options for allocating this financing for all beneficiaries, children in SAS beneficiary families and also children outside SAS beneficiary families. The simulations look at three sub- options for allocation of additional financing to the reformed SAS (Table 2, highlighted in green), namely: • Option 3a: reformed SAS with optimal benefit level plus EUR 10 million, plus EUR 7 million allocated only for children in SAS recipient families; • Option 3b: reformed SAS with optimal benefit level plus EUR 10 million, plus EUR 7 million for children not in SAS (10 EUR/child); and • Option 3c: reformed SAS with optimal benefit level plus EUR 10 million, plus EUR 7 million for children not in SAS (5 EUR/child). The changes in SAS beneficiaries under Options 3a-3c are presented in Figure 1. The most significant increase in the numbers of beneficiaries would be achieved if the eligibility for the child benefit is ‘’opened’’ for children living outside SAS recipient families (Options 3b and 3c), and especially in case of a lower level benefit (EUR 5, Option 3c) per child per month. This approach would allow extending the child benefit to over 227 thousand children (36.8 percent of all children in Kosovo 0-19 years old), including 85,969 children in SAS beneficiary families and 141,108 children who are not in SAS families. Option 3b would allow covering of 85,969 children in SAS beneficiary families and 65,211 children who are not in SAS families, or 150,180 children in total (24.3 percent of all children in Kosovo 0-19 years old). Option 3a, on the other hand, would not increase the number of eligible children; eligibility will remain restricted only to children in SAS recipient families. It would however have a significant impact on increasing benefit adequacy – to over 60 percent of the post-transfer consumption of its recipients. The coverage of the poor (share of recipients in the bottom consumption quintile) would be also significantly higher when the child benefit is extended to children beyond SAS recipient families, as high as 80 percent when the benefit level is set at EUR 5 per child per month (Option 3c), and 65 percent when the benefit level is EUR 10 per child per month (Option 3b) (Figure 3, right). Conversely, the targeting accuracy would remain highest when the additional budget of EUR 17 million is allocated to the reformed SAS plus EUR 10 million, and plus EUR 7 million for children only in SAS recipient families (Option 3a) (Figure 3). This option would achieve high targeting accuracy (68 percent of the poorest quintile) but this would be the effect of not expanding coverage – about 40 percent of the poor will be covered. The other two options will compromise on targeting accuracy (under Option 3b, 50 percent 12 of the SAS and child benefit budget will accrue to the poorest 20 percent of the population; under Option 3c, only around 40 percent of the budget will be allocated to the poor) while coverage of the poor would expand significantly. The impact of SAS reform options on poverty, including Options 3a-3c, are presented in Figure 4. In all three cases of budget increase by EUR 17 million, the reduction in poverty is impressive: the poverty headcount rate would be reduced by 20-21 percent, the poverty gap by 25-26 percent. Figure 3. Targeting accuracy (left) and coverage (right) of the reformed SAS under different financing scenarios TARGETING ACCURACY COVERAGE Percentage beneficary HH that are non-poor Percentgae of poor households that receive SAS Percentage beneficary HH that are poor Percentage of poor households that do not receive SAS 120% 90% 80% 80% 100% 70% 60% 65% 24% 32% 54% 80% 60% 50% 59% 46% 60% 50% 40% 36% 40% 40% 76% 30% 20% 68% 20% 20% 50% 41% 10% 0% 0% Reformed SAS: Reformed SAS: Reformed SAS: Reformed SAS: Reformed SAS: Reformed SAS: Reformed SAS: Reformed SAS: optimal benefit optimal benefit optimal benefit optimal benefit optimal benefit optimal benefit optimal benefit optimal benefit level level plus Euro 10 level plus Euro 10 level plus Euro 10 level level plus Euro 10 level plus Euro 10 level plus Euro 10 million, plus 7 million, plus 7 million, plus 7 million, plus 7 million, plus 7 million, plus 7 million for children million for children million for children million for children million for children million for children in SAS families not in SAS (10 not in SAS (5 in SAS families not in SAS (10 not in SAS (5 Euro/child) Euro/child) Euro/child) Euro/child) Source: World Bank estimations with ADePT using HBS 2016. Figure 4: Reduction in the poverty headcount and poverty gap for families with children under different SAS design and spending scenarios 3.0 Plus 10 million Plus 17 million 2.5 Reduction in poverty gap (percentage points) Same Budget 2.0 Plus 17 million Current scheme Plus 17 million 1.5 Plus 7 million 1.0 0.5 0.0 0.0 1.0 2.0 3.0 4.0 5.0 6.0 Current SAS Scheme Reduction in headcount (percentage points) Reformed SAS: optimal benefit level Reformed SAS: optimal benefit level plus Euro 7 million Reformed SAS: optimal benefit level plus Euro 10 million Reformed SAS: optimal benefit level plus Euro 10 million, plus 7 million for children in SAS families Reformed SAS: optimal benefit level plus Euro 10 million, plus 7 million for children not in SAS (10 Euro/child) Reformed SAS: optimal benefit level plus Euro 10 million, plus 7 million for children not in SAS (5 Euro/child) Source: World Bank estimations using HBS 2016. 13 Comparisons with a simulated Universal Child Benefit A universal child benefit does not emerge as the best policy choice for Kosovo. Simulations suggest that: • A universal child benefit will have small progressive benefit incidence: less than half of the budget would be allocated to the bottom 40 percent of the welfare distribution. Conversely, over half of the budget would be ‘wasted’ on non-poor and affluent families (Figure 5); • Similarly, a universal child benefit would have small progressive beneficiary incidence: less than half of the child benefit recipients would belong to the poorest 40 percent of the welfare distribution and only a quarter to the poorest 20 percent; • A universal child benefit would be costly. The estimated program budget of a universal child benefit if it amounted to EUR 10 per month and was given to all children 0-16 years of age, would be as high as around EUR 60 million per annum for 2019-2021 - more than 2 times the average annual spending on the Social Assistance Scheme for 2008-2017, and close to the generous amount spent on war veteran pensions. Figure 5: Simulated beneficiary incidence (left) and benefit incidence (right) of a universal child benefit, by consumption quintiles (%) 30% 30% 25% 25% 25% 23% 23% 22% 21% 19% 19% 19% 20% 20% 15% 15% 15% 15% 10% 10% 5% 5% 0% 0% Q1 Q2 Q3 Q4 Q1 Q1 Q2 Q3 Q4 Q1 Source: World Bank calculations using HBS 2016. 14 Annex 1. Main characteristics of the child benefit models in Europe Monthly child benefits are one of the most common elements of the family benefit systems in the European countries, and worldwide. They are a part of a broader set of measures that aim at reducing poverty and vulnerability of families with children17. All EU and EFTA member states deliver monthly child benefits in cash, except for Iceland where support for families with children exists only in the form of tax refunds. The Western Balkan countries also have monthly child benefits, except for Albania. In Kosovo, as already mentioned, the child benefit exists as a child supplement within the SAS scheme. The main characteristics of child benefits in the EU, EFTA and Western Balkan countries are summarized in Table 2. Roughly half (13 out of 28) EU Member States have universal child benefits, and in 3 of these cases the universal child benefit includes certain elements of means-testing18. 12 EU Member States have only means-tested child benefits, and 3 countries have two child benefits, one of them universal and the other means-tested. The child benefits are means-tested in the four Western Balkan countries where they exist as stand-alone programs (Bosnia and Herzegovina, Montenegro, North Macedonia, and Serbia). Usually the means test for the child benefit is less rigorous than the means test for the last-resort income support benefit, which is usually narrowly targeted while the child benefit is designed to reach not only low but also (part of) middle-income families. Worldwide, there is a trend toward targeting child benefits to poor families or to poor and middle-income families and, at the same time, curtailing coverage of wealthier families. In parallel, child benefits in both developed and developing countries are increasingly conditional on school enrollment, regular school attendance and/or compliance with other conditions on behalf of the benefit recipient families. The child benefits are predominantly tax-financed schemes19. Eligible usually are children who are citizens, or permanent or habitual residents of the respective country, living in the country and being raised in a family or household which are responsible for raising and maintaining them. The most common age limit is 18 with possibilities to prolong the receipt of the child benefit if the child remains in education or training. There could be variations in the benefit amounts depending on the number of children in the family, their age and/or birth order. Higher benefit amounts and higher age limits for child benefit receipt exist when the child is an orphan or with special needs. Child benefits for children with special needs are not subject to means-test. Some of the legal regulations explicitly point at the child benefit should be paid to the mother, or to the family member who takes care of the child (usually the mother)20. 17 The set of measures includes, apart from cash payments, also tax reliefs, maternity, paternity and parental or child-raising leaves, school start allowances, childcare services, health and education services, housing, utility, sanitation and other public services related to children. 18 Austria provides a monthly supplement to the child benefit for the third and subsequent children in families which are identified as poor based on their income. Also, in Austria, if a child eligible for child benefit has reached 19 years of age and has own income above certain threshold, this income is subtracted from the benefit. Belgium pays a ‘’social supplement’’ to the child benefit for low-income categories - parents who are pensioners, recipients of unemployment benefit or benefit as self-employed, also to single parents. Lithuania has a child benefit supplement for poor families. 19 Italy is the only EU Member State where the child benefit is financed by employers’ and workers’ contributions. In Austria and France, the financing is combined – social insurance contributions and taxes. 20 When paid to the mother, the child benefit is considered a secure and independent source of income at times of family disruption and gives a degree of independence to women where resources are not shared. Evidence suggests that mothers spend the money directly on their children. 15 Annex 2: Main characteristics of child benefits in the EU, EFTA and Western Balkan countries Variation Variation Variation Basic principles Means test with with birth with child’s Other provisions income order and age number of children Western Balkan countries Albania No specific child benefit scheme Bosnia and 1.Federation BiH: 1. Means- 1. No 1. Flat rates 1. No 1. Child benefit rates Herzegovina financed from tested 2. No 2. Variation 2. No vary by cantons cantonal budgets, 2. Means by number of 2.For the 2nd thru 4th for employed and test, with children child. Paid to 1st child if unemployed exempted with developmental 2.Republika Srpska: categories delay, orphan or comes financed by the from family receiving republican Child LRIS Fund, for all citizens North Tax-financed, for Means No Flat rate for Yes, two The benefit amount is Macedonia economically active: tested each age tiers – for determined as employed, farmers, bracket up to 15 percentage of craftsmen, army years old minimum wage. Ceiling officers, also for and for 15- on the benefit amount pensioners, war 18 years old regardless of number veterans and of children or their age recipients of LRIS Montenegro Non-contributory, Yes, No, but Flat rate per No Extended till the child state budget indirectly - rates vary child, up to 3 reaches 18, or above if financed. Eligible beneficiaries by type of children in a still in secondary are children in of means categorical family education. No means families receiving tested LRIS benefits test for disabled LRIS; care and which children. Conditions for assistance determine receipt of LRIS, incl. allowance, personal eligibility activation, apply to the disability allowance, child benefit as well also orphans Serbia Non-contributory, Means No Flat rate paid No Administered by the state budget tested; to up to 4 municipalities, not the financed separate children in a CSWs. Regional and from LRIS family local governments can means test provide top-ups EU and EFTA Member States Austria 1.Universal scheme No No Yes, increase Yes, 3 age Paid mainly to the financed by Exception Income- according to brackets mother. Combined employers’ for child’s tested is ‘’sibling with tax allowances for contributions and own income the scales’’ families with taxes after the supplement children/tax relief for 2.Supplement for age of 19 childcare the 3rd and subsequent child(ren) in low income families Belgium 1.Compulsory social 1. No 1. No 1. Yes 1. No Covers working insurance scheme 2. Yes, 2. Yes 2. Yes 2. Yes persons, self-employed financed by a income- and civil servants. Paid federal grant to the person raising 16 2.Social supplement tested (and the child, usually the for parents who are categorical) mother. Regional pensioners, receive variations. Higher unemployment benefit for orphans. benefit, benefit for Tax rebates self-employed, or single parents with low income Bulgaria Tax-financed Yes, income Yes, three Yes, increase No Regular school system - monthly test except income till 4th child attendance, allowance for for orphans, groups / vaccinations and health raising a child until disabled levels prophylactic controls. graduation from children, in Combined with tax high school, but not foster care allowances after the age of 20 Croatia Tax-financed Yes, income Yes, three Flat rate per No Parents of children scheme covering all test income income absent from Croatia for residents groups / bracket more than 3 months levels are not eligible. Conditional to school enrollment. Income tax deductions Cyprus Tax-financed Yes, total Yes, four Variations by No Families with 1 or 2 scheme gross income birth order children are paid income and groups / and income annually; families with household levels brackets 3 or more children are assets paid monthly. Additional entitlement to childcare Czech 1.Tax financed Yes, income Yes, two Flat rate Yes The child should be in Republic scheme for test rates / varying with full-time education at qualifying residents income age secondary school or 2.Increased amount levels university. Tax credit in case of low for childcare in the income from work form of tax or receipt of sickness benefits, unemployment benefits, or care allowance Denmark Tax financed Yes Yes, (1) is 1. Flat rate 1. Yes, 4 age For (1), the parent with scheme covering all reduced for 2. Flat rate brackets parental authority must residents families 2. No pay taxes in Denmark. 1.Child and Youth with higher Paid mainly to the Benefit income mother – monthly (1) 2.Child Allowance or quarterly (2) Estonia Tax financed No No Same rate for No Tax credits. The child schemes 1st and 2nd care allowance 1.Universal child child; same becomes payable after benefit higher rate the end of the 2.Child care for 3rd and 4th maternity benefit, the allowance – for all child. parental benefit and children up to 3, Allowance for the adoptive benefit for and from 3 to 8 in large families the same child certain cases Finland Tax financed No No Increase with No All children (10 months universal scheme birth order, – 6 years) are eligible addition for for municipal childcare 17 single Cash benefits for home parents care or private daycare. Temporary tax credits for low income families in 2015-17 fiscal years France Universal scheme Yes Since July Variation Yes, age Paid for at least 2 financed by 2015, the with number supplement dependent children. contributions from allowances of children Tax reliefs. Supplement employers, self- are for Child Care. Early employed and from reduced for childhood benefit - a portion of the families partial payment of care generalized social with a high costs for children contribution income younger than 6. Depends on income, age and number of children and type of chosen child care Germany Tax-financed No No Variation No Tax allowances for scheme for general with birth families with children / taxpayers who are order tax breaks for childcare resident in are available Germany or are liable to income tax Greece Tax-financed Yes Yes, three Flat rate for No Legal and permanent scheme income the 1st and residence in Greece for groups / 2nd child; at least 5 years before levels increased flat the year of submission rate for 3rd of application. Tax and next allowances for families child with children/ tax relief for childcare Hungary Tax financed No No Variation No Supplements for universal scheme with number disabled children, of children - single parents, poor increase till pensioners, foster 3rd child parents. Tax allowances for families with children Iceland* No cash child Yes, for the No Yes No For living legally in benefit tax refund Iceland and subject to Only tax refund taxable taxation of income income of from employment or parents is insured counted Ireland Tax financed No No Flat rate per No Children should be scheme child under 18 and in full- time education. Benefit is paid on first place to the mother or step- mother. Early Childhood Care and Education Scheme for early childhood care and education for children in pre-school Italy Earnings-related Yes, Yes, 3 Variation No Child-raising benefit for the counted is income with income vouchers for additional family, financed groups and expenses of raising 18 mainly by the household composition children (baby-sitting employers' income of family unit or other child care contributions and services) within 11 also partly by months following the workers’ compulsory maternity contributions leave and as an alternative to optional supplementary parental leave. Tax deductions for family expenses (childcare, university fees) Latvia Tax-financed No No Increase with No Supplements for universal scheme birth order families with 2 and more children Liechtenstein Compulsory, No No Flat rate for Yes, If both parents are contribution- 1st and 2nd supplement entitled to benefits, the financed social child; for each payment is made to the security scheme for increased flat child over parent who primarily residents or rate for next 10 years old cares for the child and employed in children in whose household he Liechtenstein or she lives Lithuania 1.Tax-financed 1. No 1. No 1. No 1. No Tax allowances for universal scheme 2. Yes 2. Yes 2. Yes 2. Yes families with with benefits for all children/tax relief for residents childcare 2. Means-tested benefit for poor families with children Luxemburg Universal tax- No No Flat rate per Yes, age Tax relief for families financed scheme child supplement with children and for childcare costs Malta A universal system Yes Yes Flat rate per Yes For Maltese citizens financed by general child whose children reside taxation providing in Malta. On first place, an earnings-related the benefit is paid to allowance the mother Norway Tax financed No No Flat rate per No, except Infant supplement for universal scheme child for an infant single parents with providing a flat-rate supplement child/ren under 3 benefit for all years. Cash-for-care- children benefit – universal and tax financed. Tax relief for childcare Poland Tax-financed 1. Yes 1. Yes 1. Flat rate 1. Yes, 3 age 1. From 1 January scheme which 2. No, but 2. No, per child brackets 2016, families with includes: no benefit is except in 2. Flat rate 2. No income a little above 1. Family allowance paid for the the case of per child the threshold can 2.Benefit for child 1st or only 1st or only receive allowance. Tax raising 500 Plus child if child relief for childcare income per 2. If income is low capita (€183 net per person), exceeds a 500 Plus is paid for the threshold 1st or only child 19 Portugal Compulsory Yes, income Yes, 5 Variation Yes Subsidy for nurseries universal protection and value of income or with family with assistance based system financed by movable earning composition, on household income. taxes assets levels income and Tax relief for families child age with children/tax relief for childcare costs Romania Tax-financed 1.No 1.No 1.Flat rate 1. No Family Support schemes with cash 2.Yes, 2.Yes, 3 per child 2. Yes Allowance (2) is and in-kind household income 2.Flat rate conditional to regular benefits: income levels per child per school attendance and 1.State Allowance income level no misconduct. Tax for Children deductions for families 2.Family Support with children Allowance Slovakia Tax financed No No Flat rate per Supplement Insurance -based Child universal scheme child for Care Allowance for providing flat-rate pensioners child care expenses. benefits and taking care Child Tax Credit – covering all of a child universal scheme, flat residents with rate deduction, high- dependent income families are child/ren exempted Slovenia Tax financed Yes, Yes Variation by Two rates: Reduction in payment scheme incomes and birth order – before / in of nursery fees for assets of all increase till primary families with per capita family 4th child education income below the members and in average wage. Tax secondary allowance for education dependent members Spain Tax financed non- Yes, family Yes Flat rate, Yes, only for Income from labor, contributory benefit income and increased in disabled capital, economic scheme number if case of children activities, assets and dependents disability rights of a contractual nature is counted. No income limit in case of disabled child. Benefit levels differentiated based on disability status. Tax allowances for families with children/tax relief for childcare Sweden Tax financed, No No Flat rate child Yes, regular Extended Child compulsory and allowance; and Allowance when the universal scheme differentiated extended child is over 16 but still large family child in compulsory supplements allowance education; paid directly to the child. Subsidy for childcare Switzerland Federal level: No, except No Flat rates per No variation The federal schemes Scheme for for cantonal child with age, pay higher benefits in agricultural workers level (2) except for 3 mountain regions and self-employed benefit – cantons compared to lowland farmers, financed persons not regions. In principle the by contributions in paid payment is received by and taxes employment the parent who works. Cantonal level: which is Tax deductions for Schemes for (1) families and for 20 employees and self- means childcare costs, both at employed not in tested the federal and the agriculture cantonal levels (financed by contributions) and (2) for persons not in paid employment with low income (financed by taxes) The Tax financed 1. No 1. No 1. Flat rate 1. Yes, 3 age Additional Child-related Netherlands scheme for insured 2. Yes, 2. Yes per age brackets allowance for single persons: counted is bracket 2. Yes, 3 age parents. Means-tested 1.Child benefit household 2. Variation brackets tax-financed childcare 2.Child-related income, with number allowance for parents allowance assets and of children for cost of childcare. composition Tax allowances for families with children/ tax relief for childcare United Tax financed (non- No No, but tax Yes, higher Yes Complemented by Kingdom contributory) charge in rate for the Child tax credit - tax system for all case of high eldest child financed, non- parents family contributory, income- income related system, for all parents *No data in MISSOC, other online sources are used for reference: https://www.rsk.is/english/individuals/child-benefits/; and https://dendax.com/en/iceland-child-benefit. Source: MISSOC database, extraction as of July 1, 2018 for EU ad EFTA countries, https://www.missoc.org/missoc- database/comparative-tables/results/; MISSCEO database 2018 for Albania, Bosnia and Herzegovina, North Macedonia, Montenegro and Serbia, http://www.missceo.coe.int 21