Belarus 68791 Social Assistance Policy Note: Improving Targeting Accuracy of Social Assistance Programs in Belarus1 The World Bank, May 2011 Executive Summary Belarus has a large and extensive social protection system (SP) covering a significant share of the population. The Belarusian SP system is segmented in 2 main groups of programs. The total national SP budget equivalent to 15 percent of GDP2 divided unevenly between them. The first group, social insurance (SI) programs, accounts for 90 percent of the SP budget. The second group, social assistance (SA) programs––also known as social safety nets (SSNs)–– accounts for the remaining 10 percent (or 1.5 percent of GDP, which is just below the ECA regional average). In response to the recent global food, fuel, and finance (FFF) crisis, the Government of Belarus (GoB) decided to reform its existing SSN programs to ensure more targeted and cost- effective spending on various safety nets. The GoB considered implementing/improving a targeting mechanism to identify potential beneficiaries by replacing the country’s existing categorical approach––a typically simple method of identifying prospective beneficiaries based on a particular characteristic.3 By reforming its SSN programs in this way, the GoB expected to attain a larger reduction in poverty, without increasing the SSN budget. The targeting accuracy of the SA programs is moderate. Approximately 40 percent of the overall spending on SSN benefits goes to the poorest 20 percent of population, placing Belarus in the middle of the ECA Region’s countries in terms of targeting accuracy. This modest performance is related to the poor distribution of privileges,4 which results in a higher leakage of benefits toward nonpoor households (HH). The country also runs a specifically designed pro-poor program called the Public Targeted Social Assistance (GASP). Initially, the program relied on the categorical approach, but that practice was abolished in April 2007. GASP now allocates monthly benefits to families whose per capita income falls below the poverty line established by the country. In 2001 the program supplemented these family incomes up to 60 percent of the poverty line. Now GASP is a de facto guaranteed minimum income (GMI) program that guarantees recipients income at 100 percent of minimum subsistence. Following the launch of the 1-stop application process, the number of program beneficiaries more than quadrupled in 1 year: from approximately 60,000 in 2007 to nearly 287,000 in 2008. The Household Budget Survey (HBS) data analysis suggests that the program has the potential to reduce poverty by increasing program coverage. With an annual budget of 0.03 percent of GDP (2009), GASP is rather small and could be expanded. One way to finance this expansion would be to 1 The paper was produced by a team led by Katerina Petrina (ECSH3), including Phillippe Leite (HNDSP) and Olha Nychay, consultant, ECSH3. 2 In 2009 the country’s GDP was BYR 136.8 trillion. 3 The categorical method has the advantage of being administratively simple, but its universal appeal generates inefficiency. 4 Privileges are entitlements to certain public services rendered free of charge or at a discounted value. Privileges usually are awarded to specific categories of population (appendix 3). 2 rationalize privileges. Currently, despite recent cuts and reductions, privileges still cost the state some 0.47 percent of GDP (2009) and are very poorly targeted. Belarus has adopted a single methodology for calculating income to target GASP. This methodology also is used when testing an applicant's income/means for some of the child benefits. The targeting formula––the income test + assets test chosen by the GoB––is designed to improve targeting accuracy. However, better accuracy can be achieved if the formula is used with comprehensive and reliable information on either HH income or HH consumption. In this regard, verification techniques such as recertification and home visits are key toward ensuring the quality of the method, that is, to ensure that informed decisions are made and that the aid goes to those who need it most. To accomplish these two goals, the income or consumption levels declared by applicants for any SSN program must be confirmed by data from independent sources such as tax records. Nevertheless, verification of the declared HH income is not always straightforward because it usually entails verifying both an easy-to-verify component and a hard-to-verify component of such income. Wages, social protection benefits, and alimony, all of which can be verified against data from independent sources such as paychecks and tax records, constitute the easy- to-verify component of an applicant’s income. Entrepreneurial income, agricultural income, and in-kind consumption that often cannot be verified through these types of independent sources compose the hard-to-verify component of the income of applicant. As a result, due to the incidence of under-reported, hard-to-verify components such as in-kind and agricultural incomes, the means-test method is not entirely bias free. To reduce the leakage of benefits to the nonpoor while expanding GASP, this note assesses the usefulness of applying a Hybrid-Means-Test method (HMT), a variation of the means- testing method that combines means testing and proxy-means testing. All outcomes in this note have been estimated on the basis of the 2008 Belarusian Household Budget Survey (2008 HBS). The HMT model improves estimates of “means� by generating a predicted value for hard-to-verify incomes, which are then added to the observed (reported) values of easy-to-verify incomes. In this way, the HMT model can improve predictions of per capita HH income. Combining the use of this statistical model with the work by social inspectors could improve the detection of overpayments from the social benefit system whenever the HMT formulae are used. Under a random inspection model of a means-tested program, for each 100 cases selected for investigation, one would find approximately 26 erroneous payments. Under the HMT/client profiling, this number would nearly triple to 71 cases. Thus, the use of intelligence-led approaches such as client profiling could produce significant administrative savings in the safety nets program. The note is divided in 6 sections. In section 1, we present an overview of the current SSN programs in Belarus, their design features, number of beneficiaries, and eligibility criteria to draw the overall picture of the types of programs delivered in Belarus and the magnitude of their public spending. Section 2 reviews the targeting accuracy of existent SP programs in Belarus. Section 3 analyzes whether HMT can be an option for targeting in Belarus. Section 4 presents the HMT formulae. In section 5 we describe how HMT also can be used for client profiling of beneficiaries. In section 6, we conclude by discussing the results of some simulations about the targeting accuracy of the HMT method. 3 1. Overview of Social Assistance Programs in Belarus Belarus spends a sizable share of its GDP on social assistance (SA) programs, including through income-tested programs. These programs cover the benefits and allowances related to childbirth and care but also to disability and social care. SA programs also extend privileges (and subsidies) to specific categories of citizens, including war veterans and the disabled, for various services ranging from public transportation to housing and utilities. The Public Targeted Social Assistance (GASP) distributes benefits to low-income families and families affected by natural and economic shocks. The various types of social assistance, their size and coverage, and eligibility rules are analyzed in this note. In 20085 Belarus spent BYR 3.3 bn (US$1.17bn), or 2.56 percent of GDP, on 14 active cash transfer programs and privileges (tables 1, 2).6 The largest programs in terms of spending were the child-related allowances (1.13 percent of GDP) and allowances for temporary disability and medical treatment, including allowances for medical treatment of disabled children (0.68 percent of GDP) (table 2). The share of income-tested programs designed specifically to support low-income families has remained rather small: 0.04 percent of GDP in 2008 and 0.03 percent of GDP in 2009. From 2005 to 2008, two major trends in social assistance spending became apparent (table 1): 1. Overall spending decreased, from an estimated 2.81 percent of GDP in 2005 to 2.56 percent in 2008. As the Belarusian economy grew at relatively high rates (averaging 9 percent per annum since 2003), the gains were redistributed through higher wages and pensions, but SA financing was rather counter-cyclical. 2. GoB shifted the program mix away from income-tested programs toward categorically targeted ones in 2005–2007. The share of SA income-tested programs’ spending on 14 programs including privileges has dropped, from 9.4 percent of the total in 2005 to only 6.3 percent in 2007. The expansion of categorical programs was enabled and facilitated by the strong economic growth. Then the trend reversed, with the share of income-tested programs growing to reach 6.7 percent in 2008 and 7.3 percent in 2009. 5 As incomplete data is available for 2009 and 2010, the 2008 figures are highlighted in this section. In later sections the 2008 HBS data is analyzed. 6 Apart from social benefits and allowances extended under 14 programs, 32 various privileges are awarded to certain categories of citizens. These privileges will be discussed farther on. The data included in the analysis are estimates from the Belarusian HBSs. 4 Table 1. Size and Composition of the Social Assistance Sector, 2005–2010 Spending (mil rubels) Number of beneficiaries Income Jan-Sep Jan-Sep 2005 2006 2007 2008 2009 2005 2006 2007 2008 2009 tested? 2010 2010 All social assistance 1,828,249 2,095,225 2,516,197 3,299,222 3,191,110 2,210,340 Benefits under the Public Targeted Social 4,489 4,680 5,069 48,936 40,166 56,000* 56,966 57,845 59,201 286,771 205,906 166,000* Assistance (GASP), of which - monthly social benefit Yes 4,403 4,506 4,906 47,735 38,234 43,000* 55,313 55,642 57,229 277,123 194,808 122,000* - one-time social benefit Yes 86 173 163 1,201 1,931 5,000* 1,653 2,203 1,972 9,648 11,098 28,000* - devices for social rehabilitation of disabled No - - - - - 8,000* - - - - - 16,000* (introduced in Jan 2010) Childbirth grant No 25,033 31,389 73,043 129,508 154,520 122,292 87,223 95,604 104,737 105,739 108,896 79,393 Childcare benefit, for children up to 3 years old No 234,786 310,864 360,209 553,070 672,938 708,158 247,405 257,263 272,105 288,982 304,379 306,991 Childcare benefit, for children above 3 years old Mixed 148,800 138,400 126,300 126,100 112,437 88,116 319,837 254,893 209,138 171,264 133,844 123,579 Benefit paid to pregnant women who have registered with public medical care before their 12 No 10,395 13,292 16,044 19,152 23,016 18,271 73,033 79,405 88,295 91,991 95,709 70,126 weeks** Maternity benefit No 86,917 118,508 151,971 209,396 249,352 222,454 n/a n/a n/a n/a n/a n/a Allowance for goods of first necessity for children in No 235,160 280,213 349,801 418,689 n/a 718 807 843 949 952 n/a 879 case of multiple births (two and more)*** Benefit for taking care of disab led children under Mixed 11,793 13,899 14,791 18,601 22,454 18,477 125,543 128,002 124,214 129,720 137,105 104,135 18 Allowance for HIV-infected children under 18*** No 47 66 92 127 168.3 149 637 847 1,074 1,309 1,511 151 Allowance for temporary disability or medical treatment and rehabilitation, including for disabled No 438,437 551,519 676,596 872,182 1,102,694 873,432 n/a n/a n/a n/a n/a n/a children under 18, of which - for taking care of sick children under 14 and in case of mother's or other guardian's illness, of No 52,183 n/a n/a 114,188 155,453 120,292 n/a n/a n/a n/a n/a n/a children under 3 and disabled children under 18 Benefit for taking care of disab led category 1 or Mixed 188,349 210,381 224,244 239,292 331,707 308,907 seniors ab ove 80 7,400 9,675 11,517 26,288 57,569 57,491 Allowance for burial No 56,421 70,451 80,497 102,727 119,150 100,782 121,169 120,018 115,204 116,356 117,936 90,268 Privileges and subsidies**** 568,571 552,269 650,267 774,446 636,646 n/a 2,417,400 2,384,600 2,376,700 1,547,700 1,490,300 n/a GDP, nominal (mil rubels) 65,067,100 79,267,000 97,165,300 128,828,800 136,789,800 116,600,000 Share of social assistance programs in GDP (%) 2.81 2.64 2.59 2.56 2.33 1.90 Share of income-tested programs in total SA 9.4 8.0 6.3 6.7 7.3 7.4 spending (%) Source: Ministry of Labor and Social Protection; World Bank staff calculations. Notes: Under 'Number of beneficiaries', figures in italics are number of payments. * Forecast. ** For this benefit, unofficial data is provided. *** For this type of benefit, figures are estimates by Labor Research Institute. **** For privileges and subsidies, the HBS data is used. Number of beneficiaries is number of households who reported receiving privileges and subsidies. Income test: 'Yes' means that for eligibility the average monthly household income per capita is assessed. 'Mixed' means a combination of categorical approach and either some check of beneficiary's income against threshold for eligibility or no check but a requirement that no other income be earned nor pension or other transfers received. 5 Table 2. Level and Distribution of Social Assistance Spending, 2008 (mil rubels) (% of GDP) Social benefits to low-income families under GASP 48,936 0.04 Child-related allowances* 1,456,042 1.13 Disability allowances** 872,182 0.68 Privileges 774,446 0.60 All social assistance programs 3,299,222 2.56 GDP 128,828,800 100.00 Notes: * Does not include early registration benefit for pregnant women. ** Allowance for temporary disability or medical treatment, including for disabled children under 18 years old, only. 1.1. Social Benefits and Allowances In this section, we elaborate the three subsets of social benefits and allowances delivered in Belarus: childbirth and childcare related, disability-related, and pro-poor targeted benefits. Most social benefits are regulated by Law No. 3563-XII, On the foundations of public social insurance of January 31, 1995; and/or by Law No. 1898-XII, On the public social benefits to families with children of October 30, 1992. These benefits are financed by the Social Protection Fund of the Republic of Belarus. The Public Targeted Social Assistance (GASP) benefits are financed from the national budget and administered in accordance with Decree of the President No. 458, On the public targeted social assistance of September 14, 2009, and relevant government decrees.7 The seven benefits related to childbirth and childcare are: 1. Early registration benefit: One-time benefit paid to pregnant women who have registered with a public medical care institution before their 12th week of pregnancy 2. Maternity benefit: One-time benefit paid for 126 days to expectant women starting from the 30th week of pregnancy (or for 70 days to individuals who have adopted a child up to 3 months old or have been appointed custodial parents of the child) 3. Childbirth grant: Paid for each child born (or child adopted up to six months old) 4. Multiple birth allowance: Paid for purchase of goods of first necessity for (two or more) children born in multiple births (paid from the national budget) 5. Monthly childcare benefit: Paid for taking care of a child up to 3 years old 6. Monthly childcare benefit: Paid for children from 3 to 16 (in some cases 18) years old who are not at school and not working 7. Allowance for HIV-infected children under 18. 7 For many years, the country was running a separate program of subsidies for housing and utility services, which after 2009 was made part of the GASP. Under the program, families (or individuals) whose housing and utility bills exceeded 20% of cumulative HH income in cities, and 15% in rural areas, would receive a discount in the amount of the difference to be compensated to service providers by the state. Hence, the size of a housing and utility subsidy was calculated as the difference between the sum of a housing and utility bill (distributed among HH members, allowing for 20 sq m of lodging per capita) and the amount of 20% or 15% of HH income per capita. In 2010 this type of SA was abolished and is no longer extended to Belarusians. 6 All of the above, except for the childcare benefit for children above three years old, use the categorical approach so do not require income testing for eligibility. The size of the other childcare benefit, for children up to 3 years old, is bigger if a parent is working half-time and smaller if she is working full-time or her child of 1.5 years is attending pre-school or both. Both childcare benefits may be paid with a supplement to single mothers, single parents, or single adoptive parents. Since 2008, the number of filters for single parents living in the same HH with other persons but not married to them has been increased to screen out fraudulent applications. The childcare benefit for children above 3 is refused if an able-bodied single mother or father is not working or had received a 26-week unemployment benefit in the last 12 months (see appendix 1). Social benefits related to disability are: 1. Benefit for taking care of disabled children under 18 years old 2. Allowance for medical treatment and rehabilitation of disabled children under 18 3. Benefit for taking care of sick children under 14 (and in case of mother's or other effective guardian's illness, of children under 3 and disabled children under 18) 4. Allowance for temporary disability or medical treatment and rehabilitation 5. Benefit for taking care of disabled category 1 or seniors above 80 (paid from the national budget) 6. Allowance for social rehabilitation devices for children under 18 not recognized as disabled and disabled category 3.8 To be eligible for the benefits for taking care of disabled children under 18 years old9 and of disabled category 1 or seniors above 80, an individual must not earn other income or receive other social transfers. Benefits 3 and 4, as well as the above-mentioned maternity benefit, are related to disability insurance events. In most cases, to be eligible for these benefits, applicants must be policyholders in the Social Protection Fund. Another stand-alone financial support of this kind is a burial allowance. In these cases, the benefits also are anchored to the wage level of an insured individual before the event, not to the minimum subsistence level that typically is used in SA programs (appendix 1). GASP benefits under the Public Targeted Social Assistance are (1) monthly social benefit and (2) one-time social benefit. The monthly benefit is addressed specifically to the families whose cumulative income from all sources (and after receiving all social transfers) is below the poverty line. This benefit may be awarded for 1–6 months but for no longer than 6 months in the course of 1 year. The beneficiary family is requested to take the necessary steps to improve its standing. A one-time benefit may be paid once in a year if a beneficiary family finds itself in a hard life situation, that is, circumstances that cannot be easily be coped with by the beneficiary alone 8 This allowance is legislatively related to the GASP program and financed from the national budget. While 2 out of 3 GASP benefits are designed as targeted assistance to the poor and are income tested, this type of aid is not (although de facto it may be channelled to vulnerable groups). Therefore, we choose to discuss it outside GASP. 9 In the case of families raising disabled children or HIV-infected children under 18, if a single parent in the family is disabled category 1 or 2 and not working or otherwise employed; or if a parent is in term army service, all child benefits (included those related to the child's disability) are paid, regardless of the size of HH per capita income. 7 Examples include losing the provider due to full disability; turning 80; becoming unable to care for oneself due to illness; and experiencing a natural disaster or fire. In April 2007, the categorical approach to benefit eligibility was replaced by the income test. In 2006 the practice of one-window application had been introduced for all GASP benefits. Combined with the increase in the amount of benefit itself, this innovation resulted in a spectacular growth of beneficiaries––more than quadrupling from approximately 60,000 in 2007 to nearly 287,000 in 2008 (table 1). This growth also is a result of an inflow of new types of recipients.10 Initially, families with children dominated the list. In 2009 the majority of GASP beneficiaries were still either 1-parent families raising under-aged children (31.8 percent) or families with many under-aged children (28.6 percent), followed by other categories of families with children (21.5 percent). However, since 2008, despite the noted growth in absolute terms, the share of families with children as GASP benefit recipients has been falling. The percent decline has occurred as the program has experienced a growing number of applications from representatives of other types of families and single individuals, including pensioners and disabled. Both GASP benefits are paid in cash and/or in kind. With the latter, social workers purchase food, goods for children, and medicine for beneficiary HHs. The social workers also pay for housing and utility services if the beneficiary has been found to misuse the aid money and/or fail to take good care of children.11 The eligibility and benefit formulae for income-tested programs in Belarus are very complex (table 3). An analysis of the eligibility rules suggests three categories of programs: 1. A guaranteed minimum income (GMI) program––monthly benefit under GASP––that complements the income of low-income HHs up to a minimum subsistence budget (MSB).12 In the first quarter of 2010, the MSB was BYR 255,200 (US$88) per month (table 4). 2. A childcare program, which offered an income-tested categorical benefit that topped family income up to at least 90 percent of the MSB for each child above 3 years old. 3. Two programs that offered benefits set at 65 percent of the MSB to individuals taking care of disabled children and of disabled category 1 and seniors above 80 years old and were not receiving any other income, including transfers. 10 For the composition of GASP recipients, see tables 1.7.1 and 1.7.2 in appendix 1. 11 The application process is further detailed in appendix 2. 12 MSB was introduced in 1999 as the cost of a minimum set of goods and services needed to support vital functions. MSB is a poverty line, differentiated by age and adjusted quarterly with inflation. The government also uses another indicator of living standard: the Minimum Consumer Budget (MCB). Some programs anchored benefit to the MCB until mid-2002. 8 Table 3. Key Design Features of Income-Tested Programs Program AU* Eligibility Basic formula Payment Amount** Other adjustments Income-tested programs GMI program (GASP) Fa Y < MSB B = MSB - Y per each Monthly 65,400 Awarded for up to six months; family member may not exceed six months in one year; may be in kind One-time social benefit to low- Fa Y < 150%*MSB B = up to 500%*MSB One-time 174,000 Awarded once over 12-month income families in hard period; may be in kind circumstances (GASP) Programs using mixed approach Child care benefit, for children Ind Categorical Monthly 72,700 B + 40%*B if single mother from 3 to 16 (18) years old 1) Y < 60%*MSB 1) B = 30%*MSB No benefit if able bodied single 2) Y < 80%*MSB 2) B = 15%*MSB mother or able-bodied father not working or on unemployment benefit Benefit for taking care of Ind Categorical B = 65%*MSB Monthly 161,535 B + 40% of respective child care disabled children under 18 No other income benefit (up to 3/from 3 to 16 (18)) Benefit for taking care of Ind Categorical B = 65%*MSB Monthly 161,535 disabled category 1 or seniors No other income B = 100%*MSB if two and above 80 more individuals in care Notes: * Legend for assistance unit (AU): Fa is family; Ind is individual beneficiary. ** Average benefit in BYR in 2009. Data for the first three programs is official statistics; remaining values are estimates by Labor Research Institute based on average annual MSB. MSB is national subsistence minimum budget, or poverty line; the larger of the two most recent quarterly MSBs. Y is average monthly household income per capita over the last 12 months. In case of child care benefit for children above 3 years old, household income per capita Y is calculated in prices of September last year. B is benefit. To determine the size of the benefit or allowance to be paid under the programs that use income/means testing, cumulative HH income per capita is calculated in accordance with the procedure set forth by the regulation adopted jointly by the Ministry of Labor and Social Protection and the Ministry of Finance of the Republic of Belarus. The regulation requires that the applicant submit all documents or information necessary to certify family composition; employment status of all family members; pensions and benefits assigned and paid; incomes received (or lack of income) by all family members over the 12 months preceding the application; mandatory insurance payments made to the Social Protection Fund in the past 6 months; for persons who pay them independently, income data used to calculate these insurance payments; and paid or received alimony. Essentially, GASP and other means-tested programs use the same methodology for calculation of cumulative income and per capita income and require the same sets of documents to support an application for benefits. 9 Table 4. Minimum Subsistence Level, 2005–10 (000 BYR) Year 2005 2006 2007 Date 1-Feb 1-May 1-Aug 1-Nov 1-Feb 1-May 1-Aug 1-Nov 1-Feb 1-May 1-Aug 1-Nov Average 135.2 139.2 146.2 150.8 158.1 162.5 169.6 165.8 170.5 179.1 185.4 185.7 Working age (F 55 / M 60) 147.8 151.7 157.4 165.1 174.4 179.5 186.5 183.7 188.8 197.7 203.8 206.2 Above working age 118.5 121.6 124.2 133.2 140.1 143.9 148.8 147.4 151.0 158.6 162.1 165.4 Students 141.7 145.4 150.5 158.5 166.6 171.0 177.9 175.4 180.0 189.0 195.0 197.6 Children, 0-3 years old 117.3 121.7 125.8 128.2 131.8 137.5 141.8 141.2 144.9 151.8 158.1 157.9 Children, 3-16 years old 146.9 151.9 162.4 176.2 182.2 188.7 198.0 193.1 198.4 207.6 220.4 219.1 Minimum wage, eoy 129.9 157.6 181.4 Minimum pension, eoy 134.1 165.0 189.8 Year 2008 2009 2010 Date 1-Feb 1-May 1-Aug 1-Nov 1-Feb 1-May 1-Aug 1-Nov 1-Feb 1-May 1-Aug 1-Nov Average 200.1 209.7 224.7 223.7 234.4 243.6 249.4 250.1 255.2 266.2 274.5 283.1 Working age (F 55 / M 60) 221.4 231.2 246.2 245.3 256.6 261.3 266.1 266.3 271.4 284.9 293.9 302.5 Above working age 176.7 184.9 196.1 195.8 204.4 219.1 222.9 223.3 227.4 238.4 245.3 252.2 Students 210.9 220.5 234.8 233.9 244.5 262.1 266.6 267.0 272.1 285.5 294.7 303.0 Children, 0-3 years old 168.9 179.6 190.7 191.5 200.1 215.2 222.5 223.1 227.4 237.4 245.7 251.8 Children, 3-16 years old 234.8 248.8 266.3 265.1 278.1 297.1 306.3 305.9 312.1 326 336.9 345.4 Minimum wage, eoy 214.4 229.7 282.2 Minimum pension, eoy 226.5 250.3 319.9 Source: Ministry of Labor and Social Protection. 10 The benefit formulae for the remaining programs can be found in table 5. Although requiring no income test, the child benefit for children up to 3 years old is paid taking into account the hours worked by the parent/guardian (1) at 100 percent of the MSB if working half time; and (2) at 50 percent of the MSB if working most of the month. The allowance for HIV-infected children is paid at 45 percent of the MSB per each HIV-infected child on top of other benefits related to childcare. The childbirth grant and allowance for children in multiple births are paid once for each child in the amount times the MSB. The early registration benefit for pregnant women is paid once in the amount of the MSB. The maternity benefit, benefit for taking care of sick children under 14 (and, in case of the mother's or other effective guardian's illness, of children under 3 and disabled children under 18), and allowance for temporary disability or medical treatment and rehabilitation are anchored to the level of wage of the policyholder before the event that cause the loss of ability to work and earn income. 1.2. Social privileges The system of social privileges in Belarus has seen major changes and modifications since the introduction of Law No. 239-3, On public social privileges, rights and guarantees for certain categories of citizens in June 2007. This law abolished a long list of privileges extended to some categories at the time by amending approximately 30 various laws and regulations. The following categories have been redefined, thus narrowing the number of eligible individuals:  Service personnel eligible for conscription, commanding officers of interior affairs, and national security authorities who took part in action; and service personnel with at least 25 years of service only if injured  Of family members of service personnel, partisans, and members of resistance deceased in war or military action, only their parents and children  Of invalids due to general sickness or professional illness, only disabled categories 1 and 2 and children with special needs under 18  Individuals accompanying disabled categories 1 and 2 and children with special needs under 18  Children residing in the radiation-polluted areas as a result of the Chernobyl disaster; individuals with radiation sickness as a result of other accidents. The following categories of citizens have lost all rights to privileges:  Disabled category 3  Individuals accompanying World War II disabled category 3 and war invalids who fought in other states  Citizens who were unfairly persecuted during the repressions of 1920–80s and later rehabilitated  Certain categories of service personnel  Employees sent to Afghanistan in the period from December 1979–December 1989 who completed their term of contract or left earlier for good reason  Family members other than parents and children of deceased in war or military action 11 Table 5. Key Design Features of Categorical Programs Program AU* Eligibility Basic formula Payment Amount** Other adjustments Child care benefit, for children Ind Categorical 100%*MSB if mother Monthly 193,800 If single mother: from 0 to 3 years old works half time 1) B + 75%*B before 1.5 years; 50%*MSB if 1) mother 2) B + 40%*B after 1.5 years works more than half a month; 2) at 1.5 years child starts at pre-school Allowance for HIV-infected Ind Categorical B = 45%*MSB Monthly 98,984 Paid regardless of other child- children under 18 related benefits received Child birth grant Ind Categorical 5*MSB for first child One-time 1,591,820 7*MSB for second (third...) child Allowance for goods of first Ind Categorical 2*MSB per child One-time 497,000 necessity for children in case of multiple births Pre-12th pregnancy week Ind Categorical MSB One-time 244,400 registration Maternity benefit Ind Categorical 100% of average wage One-time n/a Paid as of the 30th week of If insured plus Not more than 3*AMW pregnancy for 126 days; certain to adopting parent - for 70 days categories Min flat benefit is 50%*MSB Benefit for taking care of sick Ind Categorical 100% of average wage Term n/a Children under 14 staying at home children under 14 and in case If insured - 14 days covered of mother's or other If treated in hospital: under 5 - for guardian's illness, of children the entire period; under 3 and disabled children aged 5-14, disabled child - for the under 18 period when additional care is required Allowance for medical Ind Categorical 100% of average wage Term n/a treatment and rehabilitation of disabled children under 18 Allowance for temporary Ind Categorical 80% of average wage for Term n/a Granted for no longer than 120 disability or medical treatment If insured first 6 days, then - 100% consecutive days or 150 days if and rehabilitation Not more than 3*AMW repeated incidents in one year Allowance for social Ind Categorical - One-time n/a Cash or in kind; size of benefit is rehabilitation devices for determined on a case by case disabled children under 18 basis and disabled category 3 Allowance for burial Ind Categorical AMW two months before One-time n/a If insured Notes: * Legend for assistance unit (AU): Fa is family; Ind is individual beneficiary. ** Average benefit in BYR in 2009 (allowance for HIV-infected children under 18 is 2008). Data is official statistics and estimates by Labor Research Institute based on average MSB. MSB is national subsistence minimum budget, or poverty line; the larger of the two most recent quarterly MSBs. AMW is national average monthly wage. B is benefit. Some privileges were partially renewed during 2008–10. Pensioners annually retained a 50 percent discount on suburb commuting from May to October. Students may receive cash benefit for transportation to and from their places of study. Certain service personnel and prosecutors while on duty may receive domestic transportation cards for all means of transport except taxi. Disabled WWII participants and family members of deceased who took part in war have been allowed to have their dwellings privatized for free.13 Individuals accompanying disabled category 1 and children with special needs under 18 may use railway, waterway, and road suburb transportation for free. 13 “Privatize� means to obtain the right of private property on their apartments, which had been owned by the state. 12 Currently, 32 privileges are extended to some 27 categories of citizens (appendix 3). Some of these privileges are rather notional. They include privileged job retention, continued health care at the medical institutions of last employment, and a subset of first-served rights, However, a subset of privileges for purchase of various goods and services as well as a land tax exemption or higher tax-free income for families with children can be quantified in terms of public expenditures or revenues not received. The data on financing privileges are not collected, nor are such statistics available at the national level. An alternative way to look at privileges is to analyze the data of the Household Budget Surveys (HBS) (appendix 4). In 2009 according to the HBS, an estimated 40 percent of HHs reported privileges and subsidies, averaging 35,600 BYR per HH per month (compared to 41,700 BYR in 2008). During 2001–08, approximately 67 percent of HHs reported receiving privileges. Thus, the share of HHs receiving privileges decreased, reflecting the changes in the law and efforts to rationalize this type of program. Most frequently reported were privileges for public transportation, purchase of medicine and medical services, and housing and utility services. Despite the recent efforts to streamline the program’s rationale and budget, in 2009 privileges were estimated to cost the state approximately 0.47 percent of GDP (down from 0.87 percent of GDP in 2005).14 As the analysis below will show, privileges remain very poorly targeted. Therefore, continued rationalization of the program may help free resources to be spent more efficiently, including on SA programs that ensure support to the most in need. 2. Benefit Coverage and Incidence Analysis of Social Assistance Programs According to 2008 HBS, social protection (SP) programs––pensions, unemployment benefits, childcare benefits, and other public benefits and privileges––reach 73 percent of the Belarus population. Pensions alone cover an estimated 45.3 percent of the population. In contrast, benefits and allowances under social assistance (SA) programs, including privileges and subsidies reach 56.4 percent of the population.15 (Privileges alone reach 43.3 percent of the population.) Although they have reasonable coverage, the SSN programs are not well enough targeted to reach the poorest deciles of income distribution (table 6). To estimate such result, we ranked the sampled HHs from the 2008 HBS by their per capita HH income net of benefits, that is, before the receipt of benefits, thus approximating the pre-transfer income of a HH. We then divided the population into 10 equal groups, or deciles. Decile 1 (D1) represents the poorest 10 percent of the country's population; and decile 10 (D10) the richest 10 percent. 14 HBS-based estimates. 15 The HBS collects data on the following SA programs: 2 childcare benefits (for taking care of children up to 3 years old and of children under 16 (18) years old); early registration benefits for pregnant women; maternity benefits; childbirth grants; benefits for taking care of disabled and elderly; burial allowances; allowances for the families of servicemen killed on duty; benefits paid by local authorities; and privileges. 13 Table 6. Benefit Coverage of Social Protection Programs in Belarus Deciles of per capita income net of each SP transfer Total D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 All social protection 72.8 99.8 97.9 89.4 84.4 73.5 68.2 62.2 60.6 51.7 40.4 All social insurance 45.0 99.1 77.4 57.6 43.3 35.7 33.9 29.5 27.9 23.2 22.4 Pensions 45.0 99.1 77.4 57.6 43.3 35.7 33.9 29.5 27.9 23.2 22.4 All labor market programs 1.0 2.4 2.4 1.6 0.4 0.9 0.2 0.7 0.5 0.0 0.4 Unemployment benefit 1.0 2.4 2.4 1.6 0.4 0.9 0.2 0.7 0.5 0.0 0.4 All social assistance 56.5 81.9 68.5 61.9 56.1 54.5 55.9 54.8 46.8 46.1 38.4 Social assistance 28.0 65.5 40.3 29.9 25.2 23.1 21.9 20.8 21.3 16.0 16.2 Privileges 43.6 61.3 55.5 46.2 41.9 42.3 46.1 40.6 34.4 36.7 31.1 Notes: 1. Program coverage is the portion of population in each group that receives the transfer. 2. Specifically, coverage is: (No. of individuals in the group who live in a household where at least one member receives the transfer)/(no. of individuals in the group). 3. Program coverage is calculated setting as expansion factor the household expansion factor multiplied by the household size. SA programs demonstrate good coverage in lower deciles of the pre-transfer income distribution: 81.9 percent and 68.57 percent of the poorest HHs in D1 and D2, respectively, reported receiving some type of SSN benefits. However, coverage remains relatively high even in the richest decile, D10, of which 38.4 percent of people are receiving some SA benefit. This result is driven by the high coverage demonstrated by privileges. Unlike privileges, other SA programs that offer universal, categorical, or means-tested benefits have much better progressive coverage: 65.5 percent of HHs covered in D1 compared to 16 percent of HHs in the richest 20 percent of population, that is, D9 and D10. Another interesting indicator for the analysis of SSN programs is the incidence of benefits (“targeting accuracy�), or distribution of benefits across deciles in terms of the share of program budget spent in each decile. In table 7, estimates suggest that approximately 27.2 percent of benefits go to the poorest 10 percent and approximately 40 percent to the poorest 20 percent (D1 + D2). Such benefit incidence is not reasonable because it places the targeting accuracy of social benefits in Belarus in the middle of the group of transition economies (figure 1). This is due to poor targeting of privileges, where leakages to nonpoor HHs tend to be most pronounced as is evidenced by the incidence of privileges across the deciles of income distribution (table 7). 14 Figure 1. Targeting Accuracy: Percent of Total Social Assistance Benefits Received by the Poorest Quintile (%) (All Non-Contributory Transfers) (%) Source: the World Bank ECA Social Protection Database 15 Table 7. Benefit Incidence of Social Protection Programs in Belarus Deciles of per capita income net of each SP transfer Total D1 D2 D3 D4 D5 D6 D7 D8 D9 D10 All social protection 100.0 31.6 18.7 11.3 7.7 5.9 6.3 5.0 4.9 4.2 4.4 All social insurance 100.0 36.5 18.8 9.8 6.9 5.3 5.5 4.8 4.3 3.7 4.4 Pensions 100.0 36.5 18.8 9.8 6.9 5.3 5.5 4.8 4.3 3.7 4.4 All labor market programs 100.0 23.1 21.0 10.2 14.0 11.2 3.8 11.8 3.8 0.0 1.1 Unemployment benefit 100.0 23.1 21.0 10.2 14.0 11.2 3.8 11.8 3.8 0.0 1.1 All social assistance 100.0 27.2 13.3 9.5 8.3 7.8 8.1 7.4 6.6 6.3 5.6 Social Assistance 100.0 32.1 14.3 10.6 7.6 7.7 7.2 5.2 5.9 4.2 5.1 Privileges 100.0 15.2 12.2 10.5 7.0 8.9 10.5 9.5 9.3 10.1 6.8 Notes: Benefits' incidence is the transfer amount received by the group as a percent of total transfers received by the population Specifically, benefits' incidence is: (Sum of all transfers received by all individuals in the group)/(Sum of all transfers received by all individuals in the population). Aggregated transfer amounts are estimated using household size-weighed expansion factors. 16 The impact of SP programs on poverty also differs for each particular program (table 8). In the absence of pensions, setting the poverty line at a value of US$8816 would increase the poverty headcount from 4.9 percent to 22.9 percent. Doing so also would elevate the severity of poverty (which measures how far HHs are below the poverty line) from 0.9 percent to 11.5 percent. These results emphasize the weight of pensions for the Belarusians. In the absence of all SA programs, the poverty headcount would increase from 4.9 percent to 7.3 percent. As mentioned, privileges have lower impact on poverty rates than, for instance, child allowances were seen to have. Table 8. Social Protection Programs: Impact on Poverty Poverty Indicator Inequality FGT0 FGT1 FGT2 Gini GE(0) GE(1) GE(2) Indicator 0,049 0,009 0,003 0,245 0,099 0,100 0,113 Indicator without listed transfer All social protection 0,260 0,137 0,098 0,365 0,565 0,231 0,232 All social insurance 0,229 0,115 0,079 0,346 0,384 0,207 0,210 Pensions 0,229 0,115 0,079 0,346 0,381 0,207 0,210 All labor market programs 0,049 0,009 0,003 0,245 0,099 0,100 0,113 Unemployment benefit 0,049 0,009 0,003 0,245 0,099 0,100 0,113 All social assistance 0,073 0,022 0,010 0,259 0,125 0,113 0,125 Social Assistance 0,069 0,019 0,008 0,256 0,114 0,110 0,121 Privileges 0,054 0,011 0,003 0,248 0,102 0,102 0,115 Notes: The simulated impact is the change in a poverty or inequality indicator due to transfer, assuming that in the absence of that transfer, household welfare will diminish by its full value. These results also are represented in a cost-benefit format in figure 2, in which we calculate how much US$ reduction in the poverty gap for the US$88 poverty line we could obtain for each dollar spent on benefits under various programs. (We do not account for administrative costs here.) Note that SA (privileges excluded) and pensions are the two most cost-effective programs, with almost US$0.23 reduction in the poverty gap for each US$1 spent in the program. Such an impact seems small because the poverty line is set too low. If the poverty line were set at US$130, it would qualify 20 percent of the population at the lower end of income distribution as poor. In that case, the cost-benefit ratio would be US$0.44, US$0.40, US$0.44, and US$0.26, respectively, for pensions, unemployment benefits, SA, and privileges. 16 Exchange rate of BYR 2,987.50 and income deflated to March 2010. 17 Figure 2. Cost-Benefit Ratio of Social Protection Programs in Belarus, 2008 (%) Cost -Benefit ratio Pensions Social Assistance Unemployment benefit Privileges 0.00 0.05 0.10 0.15 0.20 0.25 Simulated Total amount poverty gap Actual poverty Difference spent in the Cost-Benefit Share of without gap (annual (dPG) program - X (dPG0/X) GDP (%) transfer (annual US$) (annual US$) US$) Privileges 112,641,507 93,948,430 18,693,077 235,759,170 0.08 0.51 Unemployment benefit 94,242,106 93,948,430 293,676 3,232,491 0.09 0.01 Social Assistance 191,741,683 93,948,430 97,793,253 480,434,289 0.20 1.05 Pensions 1,151,248,789 93,948,430 1,057,300,359 4,581,668,536 0.23 10.01 Notes: Cost-benefit is the poverty gap reduction in US$ for each unity ($1) spent in the social program. Amounts in US$ and Annual GDP BYR 136,789.8 bn a year, exchange rate BYR 2,987.50 Source: Household Budget Survey 2008. 2.1. Simulating reforms of the system of privileges Note that privileges have very little impact on poverty despite the fact that the cost of the program is estimated at 0.51 percent of GDP (figure 2). The benefit incidence analysis of privileges from table 7 also shows that more than half of the budget would go to the top 5 deciles of the distribution. Therefore, if a means-test targeting approach for privileges were chosen, some savings would be possible. To illustrate, we simulate a means-tested privilege program for the poorest 20 percent, poorest 30 percent, or poorest 50 percent of the population as follows: a. We set the actual income without privileges as the indicator of wellbeing in Belarus. b. Based on the per capita distribution of the simulated wellbeing without privileges, we selected 3 groups: (1) poorest 20 percent, (2) poorest 30 percent, and (3) poorest 50 percent. c. Then, we created 3 scenarios: 18 1) Means-tested privilege program for the poorest 20 percent: If HH is classified in group (1), it will receive the same observed amount of privileges that it had received if eligible in the past. 2) Means-tested privilege program for the poorest 30 percent: If HH is classified in group (2), it will receive the same observed amount of privileges that it had received if eligible in the past. 3) Means-tested privilege program for the poorest 50 percent: If HH is classified in group (3), it will receive as much in privileges as it had if eligible in the past. For each scenario, we estimated the cost and benefit ratio as well as savings in program budget, without accounting for administrative costs. The main findings are summarized in table 9. The first scenario would simulate a reduction in the budget allocated for privileges from 0.5 percent of GDP to 0.14 percent of GDP. This reduction would imply an increase in the cost-benefit ratio discussed above from US$0.08 to US$0.28. In scenario (2), the program cost would reach 0.20 percent of GDP, and in scenario (3), it would reach 0.28 percent of GDP. Thus, with a simple reform of privileges by applying a means-test targeting method, Belarus could save 0.23 percent–0.37 percent of GDP (US$105 million–US$169 million). This savings later could be spent on another program such as the GMI. Table 9. Simulating a Means-Tested Privilege Program Total amount spent in Cost- Share of the program - X Benefit Savings (annual US$) GDP (%) (annual US$) (dPG0/X) Privileges 235,759,170 0.08 0.51 - Second decile 66,039,749 0.28 0.14 169,719,421 Third decile 90,206,645 0.21 0.20 145,552,525 Fifth decile 129,604,208 0.14 0.28 106,154,962 Notes: Amounts in US$ and Annual GDP BYR 136,789.8 bn a year, exchange rate BYR 2,987.50 2.2. Simulating a GMI program Belarus spends only a small fraction of its social assistance budget on the GASP program. In 2008 GASP represented 0.04 percent of GDP and in 2009, 0.03 percent of GDP. In 2008 the program covered approximately 3 percent of all population, or, according to some HBS-based estimates, approximately 50 percent of low-income population. (In 2009 this figure dropped even more: to 2.1 percent of all population, or an estimated 30 percent of low-income population). The coverage is too low, and the government would like to consider expanding it to a larger share of population. Unfortunately, the 2008 HBS did not collect information on benefits paid under GASP, which is a de facto guaranteed minimum income (GMI) program. To make an assessment of the impact of a GMI program, we performed some simulations. They assumed that the HH per capita income available in the HBS is a good representation of a “true� income at the moment of joining the “simulated� program. 19 We selected all HHs with per capita incomes below the MSB threshold of US$88 as GMI beneficiaries (which simulates perfect implementation of the GMI). Of these, we estimated that 480,000 people (4.9 percent) would benefit from this type of program. This number is more than double the actual number of recipients of monthly social benefits under GASP in the official statistics––194,808 in 2009. Guaranteeing minimum income for all population under the MSB threshold, which would embody the “perfect� GMI in which all eligible HHs would receive a monthly benefit, would generate the annual cost of 0.21 percent of GDP. If the MSB is set at US$106, the GMI program would cover 10 percent of population according to the HBS data and the annual cost of benefits would increase to reach 0.55 percent of GDP. To ensure the coverage of 20 percent of population at the lower range of income distribution, the MSB would be set at US$130 and the program cost would be 1.4 percent of GDP. Therefore, an expansion of the GMI first by reaching all HHs under the MSB threshold would require reallocation of certain funds. For example, the 0.3 percent of GDP to finance shifting the MSB up from US$88 to US$106 to cover 10 percent of population with lower incomes by the GMI program could be obtained by reallocating some of the current budget for privileges, which depends on a means-test targeting approach being implemented. In summary, given the nature of pensions and unemployment benefits, Belarus’s coverage and incidence results are acceptable. With regard to privileges, the main findings suggest that a large share of benefits could be better spent to support the poorest HHs if the program were to incorporate some sort of targeting mechanism such as a means test or proxy-means test. 3. Improving Targeting Accuracy of Social Assistance Programs The Government of Belarus already has begun reforming the current approaches to the provision of social assistance to ensure pro-poorness and to improve the targeting accuracy of various cash transfers. As mentioned, the GoB gradually is phasing out untargeted privileges and is piloting a targeted SA program that uses income- and means-testing for targeting. This is a plausible strategy, and the means-test method is considered the best targeting method when complete and robust information about HH income is available. However, in the income of most HHs, there happens to be a hard-to-verify component that may compromise the quality of this targeting method. Applicants may choose to misreport or under-report their income, even more so if it is hard to verify, once they realize which declared income determines their eligibility for the program. Grosh and others (2008) strongly recommend the use of this method when full income required for eligibility is easily verifiable. Therefore, targeting the accuracy of a means-tested program relies on the weight of the easy- to-verify component of HH income. When a hard-to-verify component has some significant but still small17 weight in HH income, one possible targeting method that helps minimize under-reporting errors in hard-to-verify income is the hybrid means test (HMT). This method is robust in that it will predict hard-to-verify income based on a statistical model to be added to declared easy-to-verify income used to determine eligibility. In other words, this method enables the program administrators to assess the welfare level of a particular applicant based on a per capita income indicator that is the sum of verifiable income (from wages and social protection transfers) and the estimated hard-to-verify income. This model is being used in some transition economies, the notable examples being Bulgaria, Kyrgyzstan, and Romania. 17 When a hard-to-verify component has a large weight in HH income, a proxy-means test is recommended. 20 The main question is whether a hybrid means test is appropriate for Belarus. To find the answer, we ran some tests to measure the size of the hard-to-verify income in the full income. First, given the fact that consumption data are derived from a very detailed questionnaire, we compared the actual HH per capita consumption and declared HH per capita income in search of inconsistencies in the declaration of income. Figure 3 shows that incomes are a little higher than consumption levels and that, across the income distribution, poorer HHs either do not save or save less than richer HHs. The shape of the curves indicates good and reliable data as consumption and income lines match well over the x-axis. Moreover, according to the data, consumption and income are highly correlated at 0.7080. Figure 3. Household Per Capita Income and Consumption over Income Distribution Quintiles As to the weight of easy-to and hard-to-verify components on HH income in Belarus, the data suggests that 85 percent of HH income falls into the easy-to-verify component (table 10). As expected, there is some regional variation for this income component. In Minsk and other cities, the share of easy-to-verify income is above the average, and wages dominate other income sources in this component. In rural areas, pensions have larger weights in the easy-to- verify component than in any urban area. The former represent more than 35 percent of this component compared to 12 percent–22 percent in urban areas. Moreover, it is notable that rural areas also depend heavily on agricultural and in-kind income. More than 50 percent of hard-to-verify income in rural areas comes from these two sources. 21 Table 10. Distribution of HH Income Components by Geographic Region Total Minsk city Large city Small city Rural Verifiable Income 84.7 88.7 87.2 86.1 79.0 Wages 58.7 69.4 63.3 62.4 45.9 Pensions 20.6 15.4 18.9 17.7 27.2 SSN 3.3 2.5 3.0 3.6 3.8 Alimony 0.8 0.6 0.9 0.9 0.6 Privileges 1.3 0.8 1.2 1.4 1.4 Non Verifiable Income 15.3 11.3 12.8 13.9 21.0 Entrepreneurial Income 2.5 2.3 3.5 2.7 1.4 Agriculture Income 1.4 0.0 0.1 0.4 4.2 In-Kind consumption 6.2 2.6 3.4 5.5 11.3 Other 5.2 6.4 5.7 5.3 4.0 Total 100.0 100.0 100.0 100.0 100.0 In addition to table 10, we estimated the shares of verifiable and nonverifiable income for each ventile of HH income distribution, or for each of 20 groups that represents 5 percent of population, ranked from the poorest 5 percent to the richest 5 percent. Figure 4 indicates that the share of verifiable income remains quite stable over the distribution of income, although poor HHs tend to have slightly higher levels of hard-to-verify income than wealthier HHs. Figure 4. Easy-to-Verify and Hard-to-Verify Income across Income Distribution Ventiles (%) A more in-depth analysis of the composition of both easy-to-verify and hard-to-verify income is presented in figures 5 and 6. Notably, in figure 5, wages represent approximately 50 percent of the easy-to-verify income of the poorest 5 percent (ventile 1), whereas pensions and safety nets each represent approximately 20 percent of this income. Among the richest HHs, wages plus pensions comprise almost 95 percent of the total easy-to-verify income. For the composition of hard-to-verify income (figure 6), in-kind and agricultural incomes are more significant among the poorest while the wealthier HHs depend more on entrepreneurial activities for income. 22 Figure 5. Easy-to-Verify Income Composition across Income Distribution Ventiles (%) Figure 6. Hard-to-Verify Income Composition across Income Distribution Ventiles (%) To conclude, some of the preconditions for a successful implementation of means-tested programs hold in Belarus. In rural areas, the higher share of hard-to-verify income makes means-tested targeting more problematic, and then precautions are needed to reduce targeting errors. Hence, a combination of the easy-to-verify component with a statistical model to estimate the hard-to-verify income along the line of the hybrid-means-test model may work in identifying the poorest Belarusians. 4. A Model to estimate the Hard-to-Verify Component of Income The HBS provides a rich set of variables that can be associated with the hard-to-verify component of HH income. However, the 2008 HBS data available to us have limitations since the HBS does not represent raw data, and few variables are available for this exercise. We have information about the possession of durable goods, employment status of HH members, land ownership, livestock, and other data about HH members. Therefore, we estimated the best statistical model to explain the hard-to-verify income conditional on the available, and verifiable, variables. This model must be seen as an attempt to develop a more complex and robust estimate for the hard-to-verify component of HH income. 23 Considering that some HHs may not have hard-to-verify income, we ran a two-step model to estimate the hard-to-verify income. The first model, a probabilistic discrete model that informs whether a HH has high or low probability of posting hard-to-verify income, has 16 variables including age of the head, education of the head, and employment status of the head. Next, using the predicted probability as a control, we estimated a model on the logarithm of hard-to-verify incomes of 35 variables, including controls for geographic area. The set of variables and respective weights are presented in table 11. A positive weight means that the characteristic is associated with higher income, which translates into a higher predicted income. Conversely, negative coefficients, as for HHs with children, are associated with lower per capita income. 24 Table 11. Hard-to-Verify Income Estimation: Two-Step Model Model 1. Probit of the presence of hard-to-verify income Coef s.e. p-value Age of the head -0.025 0.002 0.000 Education of the head: Secondary specialized education -0.040 0.071 0.574 Education of the head: General basic education 0.273 0.145 0.060 Education of the head: General primary education 0.071 0.200 0.723 Status of the head: Employer -0.943 0.307 0.002 No. of other members with higher education 0.360 0.080 0.000 No. of other members with secondary education 0.058 0.044 0.188 No. of other members with basic education 0.222 0.128 0.082 Density of HH members per room 0.009 0.049 0.846 Do you have central heating at home? Yes -0.298 0.190 0.117 Do you have bath or shower at home? Yes -0.403 0.164 0.014 Do you have hot water supply at home? Yes 0.456 0.135 0.001 Does your family have any land in use? No -1.112 0.093 0.000 Area of residence: Big city 0.153 0.069 0.027 Area of residence: Small city 0.275 0.080 0.001 Area of residence: Rural 0.937 0.178 0.000 Constant 3.850 0.258 0.000 Pseudo R-square 0.206 Model 2. OLS model on the log of hard-to-verify income Coef s.e. p-value No. of children 0–5 years old -0.028 0.023 0.225 No. of children 6–12 years old -0.295 0.020 0.000 No. of children 13–18 years old -0.130 0.020 0.000 No. of adults 19–59 years old -0.228 0.014 0.000 No. of elderly, above 60 years old -0.404 0.022 0.000 Education of the head: Higher education 0.216 0.052 0.000 Education of the head: Secondary specialized education 0.199 0.044 0.000 Status of the head: Employer 2.630 0.102 0.000 Status of the head: Unemployed 0.522 0.076 0.000 No. of other members unemployed 0.230 0.054 0.000 No. of other members employed as employees -0.210 0.031 0.000 No. of other members who are employers 2.112 0.143 0.000 No. of other members who are self-employed 0.890 0.061 0.000 No. of other members who have higher education 0.044 0.041 0.283 No. of other members who have secondary education 0.067 0.029 0.019 Do you have central heating at home? Yes 0.044 0.039 0.259 Do you have gas at home? Yes 0.129 0.046 0.005 Do you have land-line phone at home? Yes 0.057 0.032 0.079 Do you have hot water supply at home? Yes -0.151 0.054 0.005 Do you have sewage at home? Yes 0.043 0.043 0.308 Do you have bath or shower at home? Yes 0.170 0.056 0.003 Do you have garage at home? Yes 0.131 0.025 0.000 Does your family have any land in use? No 0.183 0.042 0.000 Does your family keep any poultry, cattle, or bees? Yes 0.410 0.030 0.000 25 Area of residence: Small city -0.011 0.033 0.746 Area of residence: Rural 0.307 0.039 0.000 Area of residence: Vitebsk oblast -0.124 0.040 0.002 Area of residence: Gomel oblast 0.072 0.035 0.040 Area of residence: Grodno oblast 0.063 0.038 0.099 Area of residence: Minsk city 0.217 0.048 0.000 Area of residence: Minsk oblast -0.169 0.038 0.000 Area of residence: Mogilev oblast 0.079 0.043 0.067 Log expenditure with fuel for heating dwellings 0.016 0.003 0.000 Log expenditure with fuel for health care 0.055 0.012 0.000 Hazard function from model 1 -3.696 0.301 0.000 Constant 2.339 0.128 0.000 R-square 0.188 Source: Authors. The HMT model generates a prediction of hard-to-verify income that must then be added to the declared easy-to-verify income, generating a HMT score that predicts full HH income per capita. Figures below indicate that the model looks adequate and capable to predict per capita income quite well. The diagram in figure 7a plots the observed and HMT income and suggests high correlation, 0.90, of the 2 variables. On the y-axis, the actual HH per capita income is plotted; and on the x-axis, the HMT per capita income. A good model should generate predictions that match the true values of income represented by the diagonal line in figure 7a. The fact that the points are concentrated around the diagonal is a good demonstration of the predictive power of the model, despite the fact that predicted values are lower than the actual values of per capita income. Figure 7b confirms that the density function of the HMT (dashed line) reproduces fairly well the density function of the true per capita income. Figure 7a. Scatter Plot of Actual and HMT Income Figure 7b. Density Function of Actual, HMT and Standard PMT Scores 1.5 8 7 1 True per capita income 6 f(x) 5 .5 4 3 0 0 2 4 6 8 2 l1 Observed HMT PMT 2 3 4 5 6 7 8 HMT per capita income Source: Authors. Note: PMT is proxy-means-test. The cross-tabulation of the HMT HH income and the actual income is presented in table 12. Comparison of the actual deciles partition versus the HMT income deciles confirms that the 26 method is a good fit (a perfect prediction implies that all cases are found on the diagonal). For the HMT, 90 percent of the cases are predicted either in the correct decile or in the next (lower or upper) decile (sum of gray area). Table 12. Accuracy of Client Profiling Model HMT Prediction 1 2 3 4 5 6 7 8 9 10 Observed 1 6.8 2.5 0.1 0.0 0.1 0.0 0.0 0.0 0.0 0.0 2 1.2 5.2 3.1 0.4 0.0 0.0 0.0 0.0 0.0 0.0 3 0.6 1.3 4.6 3.3 0.2 0.0 0.0 0.0 0.0 0.0 4 0.4 0.6 1.4 3.9 3.2 0.5 0.0 0.0 0.0 0.0 5 0.3 0.2 0.3 1.3 3.9 3.7 0.3 0.0 0.0 0.0 6 0.2 0.1 0.2 0.6 1.6 3.4 3.9 0.2 0.0 0.0 7 0.2 0.1 0.1 0.3 0.6 1.3 4.2 3.4 0.0 0.0 8 0.1 0.1 0.1 0.2 0.3 0.7 1.0 5.3 2.4 0.0 9 0.2 0.0 0.1 0.1 0.2 0.3 0.4 0.9 6.5 1.5 10 0.1 0.0 0.1 0.0 0.0 0.1 0.2 0.3 1.1 8.5 Total 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 10.0 5. A GMI Program Simulation: Applying New Targeting System Due to the hard-to-verify component of income, the HBS observed income may not be a good representation of the actual HH income. Thus, to complement the GMI simulation above, we added a targeting instrument to estimate targeting errors. A large component of income comes from verifiable sources, but the hard-to-verify component can be underestimated, which would increase targeting errors. Therefore, we will simulate the implementation of a GMI program alongside three different targeting approaches: 1. We set only verifiable income as a variable for targeting (verifiable means testing, or VMT). In this case, if a HH had a verifiable per capita income below the MSB, it would receive the difference between the MSB and the verifiable income as a GMI benefit. 2. We used a standard proxy-means-test (PMT) estimator as the proxy for the true income. In this case, if a HH had its predicted per capita income (PMT) below the MSB, it would receive the difference between the MSB and the predicted income as a GMI benefit. 3. For each HH, we predicted a hard-to-verify income to be added to the verifiable income to generate a HMT predictor to be used as a proxy for the true income (hybrid means-testing, or HMT). As before, if a HH had a predicted per capita income (HMT) below the MSB, it would receive the difference between the MSB and the predicted income as a GMI benefit. After selecting potential beneficiaries and assigning amounts of benefit under the simulated GMI if selected, we would do a sort of ex-post analysis in which the benefit amount would be added to the true income to estimate the targeting errors, targeting accuracy, and impact on poverty of the simulated program. Figure 8 shows that, with verifiable income as a means-test indicator for the MSB threshold, the program would reach 1.3 million people, of whom 480,000 actual poor (blue bar) would be selected to participate, whereas almost 900,000 nonpoor (65 percent of inclusion error, or red bar) also would be selected, generating a significant budget increase. If the PMT were used for selection of beneficiaries with the MSB 27 as threshold, the program would reach only 136,000 people, but 47 percent of those would be wrongly included (red bar). Therefore, PMT would have a large inclusion error as well as a large exclusion error (80 percent). The reason is that only 98,000 poor people would be in the program, meaning that almost 80 percent of actual poor would be excluded from the program. If the HMT were used to select beneficiaries, the program would reach approximately 660,000 beneficiaries, of whom 45 percent would not be poor and only 25 percent of the poor would be excluded from the program. The annual cost of HMT-based monthly benefits would be 0.1 percent of GDP higher than that of a perfectly simulated GMI program. Figure 8. Number of Beneficiaries and Targeting Errors by Targeting Method Source: Authors. In summary, targeting that is based on verifiable income would lead to a large inclusion error but no exclusion error. Both PMT and HMT yield similar inclusion errors, but exclusion errors under PMT are three times higher than under HMT. The fact that HMT has a better targeting performance in terms of inclusion/exclusion errors than do the other two methods is due to the fact that less than 20 percent of HH income in Belarus is associated with hard-to-verify income. That said, the use of this sort of model in Belarus for targeting looks promising. Moreover, if the GoB keeps collecting both easy-to- verify and hard-to-verify income as part of application process, this model could be very helpful in protecting Belarus's benefit system, as shown in the next section. 28 Table 13. Targeting Accuracy of Simulated GMI Program Share of Share of No. of MSB threshold Targeting method beneficiaries GDP beneficiaries (%) (%) US$88 "Actual" 488,073 4.9 0.21 MT: Verifiable income 1,386,587 14.0 0.90 Poorest 5% PMT 191,131 1.9 0.07 HMT: Verifiable income + prediction for nonverifiable 664,836 6.7 0.31 income Besides this simulation, we also simulated consequences for the GMI program should the MSB be raised to US$106, the “poverty line,� for the poorest 10 percent; and then if raised to US$130, the “poverty line� for the poorest 20 percent. After shifting up the MSB, the target population will increase to include the poorest 10 percent or poorest 20 percent, whereas the GMI benefit still will be paid to top their incomes up the MSB. Main findings here support the earlier conclusions about the HMT method having better targeting outcomes than the other two targeting methods. The main implications of the MSB increase are the higher number of beneficiaries and larger cost of the program. Table 14. Number of Beneficiaries and Cost of Increasing MSB Share of Share of No. of MSB threshold Targeting method beneficiaries GDP beneficiaries ( %) (%) US$88 "Actual" 488,073 4.9 0.21 MT: Verifiable income 1,386,587 14.0 0.90 Poorest 5% PMT 191,131 1.9 0.07 HMT: Verifiable income + prediction for nonverifiable 664,836 6.7 0.31 income US$106 "Actual" 1,006,168 10.2 0.55 MT: Verifiable income 2,254,309 22.8 1.75 Poorest 10% PMT 452,944 4.6 0.21 HMT: Verifiable income + prediction for nonverifiable 1,233,583 12.5 0.74 income US$130 "Actual" 2,023,388 20.5 1.41 MT: Verifiable income 3,472,699 35.1 3.38 Poorest 20% PMT 1,166,508 11.8 0.63 HMT: Verifiable income + prediction for nonverifiable 2,211,936 22.4 1.72 income 29 6. Client Profiling for Oversight and Controls The use of statistical models as HMT also can be effective to protect the benefit system in Belarus against the fraud and errors common to targeted benefit systems. Countries usually invest greatly in the design of an efficient benefit system, but social workers and social inspectors deal daily with such a large number of cases that this overload eventually affects the quality of service delivery and the recertification process. Social inspectors play an important role in detecting fraudulent or erroneous cases, mainly through recertifying cases via home visit checks and data matching. Even with a very comprehensive audit system in place, detection rates can be low if the inspection regime is too general and ill-defined instead of employing a more discriminating approach. For example, across Ukraine’s 27 oblasts, the error detection rate correlated negatively with the percentage of cases checked, meaning that the social inspectors obviously were swamped with work for recertification. If instead they had discarded permanently some of their caseloads and focused on particular cases that deserved attention as fraud- or error-prone, the inspectors likely would have yielded better results on each inspection. Client profiling is exactly the approach that uses statistical models to find cases that are more at risk for fraud, errors, and omissions; and helps generate a narrower list of cases to be reviewed by social inspectors. There are other ways to generate such targeted lists, but we concentrate here on a method that exploits the difference between reported and predicted per capita income. The basic idea behind this client profiling model is to direct inspections to the applications that post large differences between reported and estimated income. Larger differences imply a higher probability that the applicant has misreported or under-reported her/his HH income per capita. We demonstrate how client profiling works for a GMI program designed to cover the poorest population in the country. Our assumption was that some applicants have submitted untrue income data during their applications. The false reporting resulted in leakages of benefit money to the nonpoor. We randomly generated this situation for the GMI program based on the 2008 HBS. It is important to note that social inspectors recertified all HHs receiving this GMI benefit to determine their true incomes. In other words, for each HH randomly classified as eligible for the benefit, we set the observed HH income in the 2008 HBS as the “true� income confirmed by a social inspector. The outcome of this recertification is presented in the last column of table 15: 74 percent of all HHs receiving benefits under the program belonged to the lowest quintile, Q1, or the poorest 20 percent of population. Conversely, 1 of 4 HHs selected for the program would not be eligible. Keep in mind that for this recertification, social inspectors had to visit all the HHs participating in the program. 30 Table 15. Beneficiaries of the Simulated GMI Program Ranked by Their Actual and Predicted Per Capita Income GMI Program: HMT formulae Q1-Bar Q2-Q5 Bar Total Q1 63 10 74 Q2-Q5 1 25 26 Total 64 36 100 Deterrence 1 71 26 This extensive recertification process can be optimized by using statistical models. For example, once again we used an income-based means test as the targeting mechanism. In this case, a HMT formula was applied to the caseload to estimate the predicted HH income of each applicant. We then ranked all cases according to the HMT predicted income of the respective applicant and highlighted only the cases with predicted income above a certain threshold. These would come to the attention of social inspectors. As in table 15, we grouped the simulated GMI beneficiaries into a 2 x 2 table. Horizontally, we listed cases according to the true HH income: Q1 for the poorest 20 percent and Q2-Q5 for the rest of population. Vertically, we ranked cases by their HMT-generated income, also for the poorest quintile and the remaining quintiles. The simulation indicates that the majority of GMI beneficiaries (63 percent) would be correctly classified as poor in the HMT procedure, which compares the prediction with the “true� income. The HMT model suggests that 36 percent of the applicants could be considered suspicious cases because they are not among the poorest 20 percent according to the HMT. By going through only these 36 percent of cases to recertify the HH, a social inspector could: a. Reduce by 67 percent the number of cases for inspection b. Increase the probability of detecting over-payments from 26 percent (26/100) to 71 percent (25/36) of cases. In other words, 2 of 3 inspections could deter erroneous payments. Therefore, combining statistical models with social inspector work greatly raises the chances for the SSN system to detect overpayments. Under a random inspection model of a means- tested GMI program, the odds of finding a case of fraud would be approximately 1 in 4. The use of the HMT formulae for client profiling would change these odds to 2 in 3. In other words, for each 100 cases for investigation, the random model would find approximately 26 erroneous payments, whereas the HMT model would signal 71 such cases. Thus, the use of intelligence-led approaches such as client profiling could lead to significant administrative savings in a SSN program. In conclusion, the reform of privilege programs would generate savings that could be allocated to any other more efficient social assistance programs such as the GMI (as discussed in simulations above) or child-related benefits (tables 6 and 7). A more in-depth analysis requires access to the raw data on specific SSN programs or a new survey that would list all benefits disaggregated and collect information on each individual program. 31 Appendix1. Additional Data on Some Social Assistance Programs Table A1.1 Childbirth Grant 2001 2002 2003 2004 2005 2006 2007 2008 2009 No. of beneficiaries 85,698 83,259 84,091 85,368 87,223 95,604 104,737 105,739 106,796 Ave. benefit 97,914 178,707 265,236 331,811 406,177 467,355 850,578 1,405,915 1,591,820 Total spending (mil BYR) 8,391 14,879 22,304 28,326 35,428 44,681 89,087 148,660 170,000 Notes:  Before mid-2002, benefits were calculated based on the MCB; since mid-2002, the MSB has been used.  From mid-2002 to 2006, a single benefit was twice the MSB for the 1st, 2nd, 3rd … child; from 2006 to 2008: 2xMSB for 1st and 2nd child and 3xMSB for 3rd … child; since 2008: 5xMSB for 1st child and 7xMSB for next child. Table A1.2 Childcare Benefit, for Children, Aged 0-3 2001 2002 2003 2004 2005 2006 2007 2008 2009 No. of beneficiaries 253,693 249,959 247,125 245,085 247,405 257,263 272,105 288,982 296,010 Ave. benefit (with supplement) 22,000 45,300 64,300 78,500 91,700 106,200 116,200 169,200 193,800 Total spending (mil BYR) 56,949 117,063 165,134 200,098 234,786 310,864 360,209 553,070 680,000 Notes:  Before mid-2002 benefits were calculated based on the MCB; since mid-2002, the MSB has been used.  From mid-2002 to 2007 a single benefit was 65 percent of the MSB; since 2008: 80 percent of the MSB; since 2010: 100 percent of the MSB.  Until 2007 working women, students and women unable to work for good reason were paid higher benefit; since 2007 it has become equal for all beneficiaries.  Benefits are differentiated by the character of employment after birth and in case the child is attending pre- school. Table A1.3 Childcare Benefit, for Children, Aged 3-16 (18) 2001 2002 2003 2004 2005 2006 2007 2008 2009 No. of beneficiaries 649,225 451,130 419,570 408,137 319,837 254,893 209,138 171,264 137,138 Ave. benefit (with supplement) 9,400 20,600 29,700 36,200 42,300 49,000 53,700 63,500 72,700 Total spending (bn BYR) 69,4 103,5 134,8 163,8 148,8 138,4 126,3 126,1 109,0 Note: Before mid-2002, benefits were calculated based on the MCB; since mid-2002, the MSB has been used. 32 Table A1.4 Benefit for Taking Care of Disabled Children under 18 2001 2002 2003 2004 2005 2006 2007 2008 2009 No. of monthly payments of benefit 98,590 102,784 110,591 121,173 125,543 128,002 124,214 129,720 Average benefit (no supplements)* 51,375 68,414 81,723 94,682 108,013 119,769 142,977 161,535 No. of disabled children 29,538 30,253 29,282 29,297 28,895 28,403 27,662 26,632 25,867 *Estimates by the Labor Research Institute (Belarus). Note: Before mid-2002, benefits were calculated based on the MCB; since mid-2002, the MSB has been used. Table A1.5 Allowance for HIV-Infected Children under 18 2001 2002 2003 2004 2005 2006 2007 2008 2009 No. of monthly payments of allowance 212 261 303 365 637 847 1,074 1,309 Average benefit* 35,567 47,364 56,578 65,549 74,778 82,917 98,984 *Estimates by the Labor Research Institute (Belarus). Note: Before mid-2002, benefits were calculated based on the MCB; since mid-2002, the MSB has been used. Table A1.6 Benefit for Taking Care of Disabled Category 1 or Seniors above 80 2001 2002 2003 2004 2005 2006 2007 2008 2009 No. of monthly payments of benefit 116,102 119,371 132,202 157,915 188,349 210,381 224,244 239,292 Average benefit* 12,515 19,760 26,313 31,432 36,416 41,544 46,065 142,977 161,535 * Estimates by the Labor Research Institute (Belarus). Notes:  Before mid-2002 benefits were calculated based on the MCB; since mid-2002, the MSB has been used.  Until 2007 a single benefit was 25 percent of the MSB and 50 percent of the MSB if taking care of two and more individuals; since 2007: 65 percent of the MSB and 100 percent of the MSB. 33 Table A1.7 GASP Benefits 2001 2002 2003 2004 2005 2006 2007 2008 2009 No. of beneficiaries* 174,663 101,130 42,575 45,399 56,966 57,845 59,201 286,771 205,906 Average monthly benefit 7,800 12,600 16,800 21,400 32,500 27,000 28,600 57,400 65,400 Average one-time benefit - - - 62,300 52,300 81,300 82,800 124,500 174,000 Total spending (mil BYR)** 4,091 3,824 2,155 2,914 4,489 4,680 5,069 48,936 40,166 Share of population covered, percent 1.8 1.0 0.4 0.5 0.6 0.6 0.6 3.0 2.1 Share of low- income population covered, percent 5.4 2.9 1.3 2.1 4.4 4.7 8.4 50.1 29.1 Notes: *Both monthly and one-time benefits. ** Including in kind benefits.  Poverty line for monthly benefit was 60 percent of MSB in 2001 and is now equal to the MSB. Table A1.7.1 Composition of GASP Beneficiaries* 2001 2002 2003 2004 2005 2006 2007 2008 2009 No. of beneficiaries 174,700 101,100 42,600 45,400 57,000 57,800 59,200 286,800 205,900 Families with many children under age 94,400 46,500 7,500 7,800 14,000 16,100 16,000 84,000 58,800 Single-parent families with under-aged children 70,000 46,800 31,100 32,800 34,000 32,600 32,500 90,800 65,400 Families with disabled children 1,200 83 156 449 328 237 5,900 4,100 Other categories of families with children 4,800 2,500 3,100 7,000 7,400 8,900 68,000 44,200 Other categories of families 9,210 1,000 500 600 700 800 1,000 19,900 16,800 Individuals 1,100 800 900 1,000 800 600 600 18,200 16,600 Single disabled category 1 and 2 109 156 158 334 299 134 114 2,600 2,700 Single disabled category 3 receiving social pensions 72 33 53 130 99 103 75 170 136 Single pensioners 725 389 548 538 397 364 394 10,100 8,100 Other 213 172 178 5,300 5,600 Note: *Both monthly and one-time benefits. 34 Table A1.7.2 Composition of GASP Beneficiaries (% of All Beneficiaries)* 2001 2002 2003 2004 2005 2006 2007 2008 2009 Families with many children under age 54.0 46.0 17.7 17.2 24.6 27.9 27.1 29.3 28.6 Single-parent families with under- aged children 40.1 46.3 73.0 72.2 59.7 56.4 54.9 31.7 31.8 Families with disabled children 0.0 1.2 0.2 0.3 0.8 0.6 0.4 2.1 2.0 Other categories of families with children 0.0 4.8 5.8 6.8 12.2 12.7 15.1 23.7 21.5 Other categories of families 5.3 1.0 1.2 1.2 1.3 1.3 1.6 6.9 8.2 Individuals 0.6 0.7 2.1 2.2 1.4 1.0 1.0 6.3 8.1 Single disabled category 1 and 2 0.1 0.2 0.4 0.7 0.5 0.2 0.2 0.9 1.3 Single disabled category 3 receiving social pensions 0.0 0.0 0.1 0.3 0.2 0.2 0.1 0.1 0.1 Single pensioners 0.4 0.4 1.3 1.2 0.7 0.6 0.7 3.5 3.9 Other 0.1 0.2 0.3 0.0 0.0 0.0 0.0 1.8 2.7 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 Note: *Both monthly and one-time benefits. 35 Appendix 2. Administration of GASP Benefits 1. The Labor, Employment, and Social Protection Office accepts and registers application form and supporting documents; it notifies the applicant of his/her responsibility for failure to provide complete and true information and authentic documents influencing the decision and size of benefit. 2. Standing commission18 reviews the application and makes decision GASP process: from on granting or refusing the benefit in five days. application to decision 3. Labor, employment and social protection office notifies the applicant of the decision in three days. 4. Labor, employment and social protection office has the right to check information provided by the applicant and informs the applicant about the coming check. In this case the decision about the benefit may be made in 25 days upon receiving the application. Quantity of staff tasked with assigning N/a and paying benefits Heads of municipal authorities, of the agency administering the program are liable for untargeted use of local budget money channeled to finance GASP benefits according to the law. Responsibility of staff tasked with assigning Officers of municipal authorities and of the agency administering the and paying benefits program can be relieved from their positions and given administrative sanctions for red tape, violating the rules and terms for application review, including requesting supplementary documents. In case of misreporting or under-reporting incomes or assets by family Responsibility of members, providing false information on family composition, place of applicants submitting residence, need in social rehabilitation devices and other information information for influencing the benefit and its size, applicant is deprived of the right to apply benefit application for assistance for 12 months from the date of refusal. 18 Commission is called upon by the district (municipal) executive body (local administration) and consists of district (city) councillors, specialists of the department (office) of labor, employment and social protection of the executive body, the department of social protection at the local administration, specialists of other departments of the district (municipal) executive body (local administration), local center for social service, other representatives of local government and civil society. Labor, employment and social protection office is the technical arm of the commission. 36 1. Application for monthly social benefit under GASP To the Department (office) of labor, employment and social protection at the executive committee (local administration) [Applicant's last name, first name] Address of residence: [City, village or town, street, number] [Identity card, passport] [Series, number, issued by, date of issue] APPLICATION I would like to apply for public targeted social assistance for myself (my family) in the form of monthly social benefit. I submit the following information. CHAPTER I GENERAL 1. Last name __________________________________________________________________ First name ____________________________________________________________________ Patronymic ____________________________________________________________________ [Applicant] 2. Place of registration: City, town, village ______________________________________________________________ Street ________________________________________________________________________ No. _____________________________ Apartment _______________________ 3. Home telephone _________________ 4. Organization tasked with servicing applicant's dwelling _______________________________________ 5. Number of co-inhabitants family members as of application date ________________ persons. CHAPTER II FAMILY COMPOSITION Last name, first name, Place of work, name of No. Family relationship Date of birth patronymic educational institution Number of family members* _________ Coefficient to account for co-inhabitancy effect* _____ * To be filled out by administrator at labor, employment and social protection office. 37 CHAPTER III INFORMATION ON RECEIVED INCOMES* From ____________ till _______________ Type of income Size of income, in BYR * To be filled out by applicant in case of independent submission and by administrator at labor, employment and social protection office in case of requesting data from public and other institutions. CHAPTER IV OWNED ASSETS Information on real estate: Yes No Type of asset Dwellings (apartments) Garages Other buildings (cottages, summer houses, etc.) Land plots Information on transport: Model of automobile or other transport Production date Purchase date Notes CHAPTER V ADDITIONAL INFORMATION ON MATERIAL STANDING Yes No Does the family (citizen) have debts on payments for housing and utility services? Did the family (citizen) or family members receive the following incomes during the twelve months preceding the application: income derived from rent of dwellings, lease of land and/or annuity; proceeds from sale of agricultural produce (except for the sale of milk); proceeds from sale of fruits, vegetables, etc. of own production (seed, flowers, honey, poultry, bees, etc.); proceeds from sale of other produce (from hunting, foresting, fishing, of feed, herbs, berries, mushrooms, etc.); dividends on shares and other financial income; money inherited or gifted; social assistance in cash paid by public organizations, charity, religious organizations? Did a family member acquire a discounted voucher for medical treatment and rehabilitation (for free or partially free)? Do the family members own more than one dwelling in Belarus? Do the family members rent out or sub-rent their dwellings? 38 Do the family members study at their own cost? Did a family member travel as a tourist outside the country in the twelve months preceding the application? Does the family use the land plot for farming? Do the family members work half-time by their own request? I attach documents on _____ pages. I hereby confirm being notified about responsibility for failure to provide true information and authentic documents. I agree to the check of my submitted information and visit at home. ____________ ________________________ [Date] [Signature] Documents accepted on ____. _______________________________________________ [Signature of the specialist who accepted the documents] Registration No. ___________________ 39 2. Application for one-time social benefit under GASP To the Department (office) of labor, employment and social protection at the executive committee (local administration) [Applicant's last name, first name] Address of residence: [City, village or town, street, number] [Identity card, passport] [Series, number, issued by, date of issue] APPLICATION I would like to apply for public targeted social assistance for myself (my family) in the form of one-time social benefit due to hard life situation. _____________________________________________________________ [Describe the circumstances] I submit the following information. CHAPTER I GENERAL 1. Last name __________________________________________________________________ First name ____________________________________________________________________ Patronymic ____________________________________________________________________ [Applicant] 2. Place of registration: City, town, village ______________________________________________________________ Street ________________________________________________________________________ No. _____________________________ Apartment _______________________ 3. Home telephone _________________ 4. Organization tasked with servicing applicant's dwelling _______________________________________ 5. Number of co-inhabitants family members as of application date ________________ persons. CHAPTER II FAMILY COMPOSITION Last name, first name, Place of work, name of No. Family relationship Date of birth patronymic educational institution CHAPTER III 40 INFORMATION ON RECEIVED INCOMES* From ____________ till _______________ Type of income Size of income, in BYR * To be filled out by applicant in case of independent submission and by administrator at labor, employment and social protection office in case of requesting data from public and other institutions. I attach documents on _____ pages. I hereby confirm being notified about responsibility for failure to provide true information and authentic documents. I agree to the check of my submitted information and visit at home. ____________ ________________________ [Date] [Signature] Documents accepted on ____. _______________________________________________ [Signature of the specialist who accepted the documents] Registration No. ___________________ 41 3. Decision on GASP benefits _______________________________ [Name of authority] DECISION ________________________________________________________________ [On granting/on refusing to grant] public targeted social assistance in the form of monthly benefit and/or one-time social benefit ____________ 2010 Protocol No. ____ The commission on public targeted social assistance of ______ persons has reviewed the application for monthly and/or on-time social benefit by ____________________________________________________________________. [Applicant's last name, first name, patronymic] Calculation of average income per capita, amount of monthly and/or one-time benefit: Poverty line, in rubels Average HH income per capita, in rubels Coefficient to account for co-inhabitancy effect Coefficient to account for state of health Average HH income per capita corrected for coefficient to account for co- inhabitancy effect or for state of health Amount of monthly benefit per capita per month, in rubels Amount of monthly benefit for the entire family per month, in rubels Amount of monthly benefit for the entire family for _____________ months, in rubels Amount of one-time benefit for the entire family (for individual), in rubels Decision has been made to ________________________________________________________________________ Head of commission _______________________ _________________________ [Signature] [Name] Calculation made by _______________________ _________________________ [Signature] [Name] _____________ 2010 42 Appendix 3. Privileges and Eligible Categories in Belarus 9392 35417 51998 43900 45035 1326 23600 4600 26196 1373000 1308645 18 345131 1984300 1950000 65300 52300 32878 751096 4132 n/a n/a n/a n/a n/a n/a n/a Number of eligible persons / families per category Former prisoners of Nazi emergency situations who service), including relieved accompanying investigations, Parents, spouses who did War veterans who fought in Individuals who took part in Servicemen, judges and prosecutors with at least 20 Heroes of Belarus, of the of Motherland, of Glory and internal Active-duty servicemen and and and financial investigations and disabled Category 1 and 2 and children with special radiation radiation of colonel (colonel of police, camps, prisons, ghettos, Soviet Union, of Socialist from service for age, state of health, due to staff cuts, other Children with special needs conscription, not remarry, children of Radiation sick as a result of consequences with at least 5 years of of of of Invalids of childhood injured Labor, full holders of Orders Servicemen up to the rank authorities of interior affairs, authorities of interior affairs, Categories of citizens students of military colleges Families with many children Disabled category 1 and 2 in WWII or other hostilities family officers officers of Chornobyl accident Families with children emergency situations deceased in WWII or WWII participants police, Children under 7 of in years of service needs under 18 Labor veterans WWII veterans Retired by age of of Labor Glory polluted areas polluted areas commanding commanding for were injured War invalids Servicemen Servicemen other states Employees Inhabitants eliminating Individuals Chornobyl deceased accidents Members under 18 Orphans financial finance service eligible etc. Privileges 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Free-of-charge privatization of dwelling, ≤20 sq.m 1 Free-of-charge on-the-job professional training or enhancing qualification at public educational 2 institutions, with salary at the last position level paid throughout the training period Free-of-charge catering at school 3 Higher level of tax-free personal income 4 Longer maternity leave 5 First-served in medical and public social care institutions, post offices, canteens, housing and utility managing units, automotive technical service 6 and repair, cultural and sports organizations, whle purchasing transport tickets, priveleged service at shops, dry cleaner's, etc. Off-the-list offer of social housing to individuals 7 whose living conditions require improvement Specific scholarships for students of higher education institutions, of secondary special and 8 professional schools Use of vacation time in summer or other season as 9 requested Health care at the medical institutions of last 10 employment, military or civil service after retiring Offer of alternative lodging when requested to 11 move out of current social lodging Exemption from land tax 12 First-served while entering garage and parking 13 cooperatives Off-the-list supply of feed for cattle in household 14 First-served while acquiring land plots for construction of one-apartment houses to individuals 15 whose living conditions require improvement First-served while looking for a job after moving to 16 new area Preferential loans, subsidies for construction (reconstruction), purchase of apartment, cottages, 17 etc. cuts labor law Notes: rehabilitation plot buildings Privileges Increased pensions For public transportation For purchase of medicine coming from another area For communication services For housing and utility services For medical treatment and rehabilitation or training courses, priority while entering Increased allowance for temporary diability For purchase of vehicles (devices) for social with temporary housing offered to individuals Breaking labor contract due to moving without giving employer prior notice typically required by Increased benefit for taking care of child under 3 Renovation of apartment paid for by local budget Off-the-contest enrollment in professional schools Privileged job retention ceteris paribus during staff Supply of wood for construction of house or on-the- 50% discount on catering at pre-school institutions 32 31 30 29 secondary special and high education institutions, 28 27 26 25 24 22 21 20 19 18 Categories of citizens Number of eligible persons / families per category 23 * 1 War invalids 9392 * Individuals in respective categories have to be not working. * 2 WWII participants 35417 * 3 WWII veterans 51998 Parents, spouses who did 4 not remarry, children of deceased 43900 Former prisoners of Nazi 5 camps, prisons, ghettos, etc. 45035 * 6 Invalids of childhood injured in WWII or other hostilities 1326 7 War veterans who fought in other states 23600 * Radiation sick as a result of 8 Chornobyl or other accidents 4600 9 Children with special needs under 18 26196 Individuals who took part in 10 eliminating consequences of Chornobyl accident n/a Servicemen, judges and 11 prosecutors with at least 20 years of service n/a 12 Families with children 1373000 43 Inhabitants of radiation 13 polluted areas 1308645 Employees in radiation 14 polluted areas n/a * Heroes of Belarus, of the Soviet Union, of Socialist 15 Labor, full holders of Orders of Motherland, of Glory and of Labor Glory 18 16 Disabled category 1 and 2 345131 17 Labor veterans 1984300 18 Retired by age 1950000 19 Families with many children 65300 Servicemen up to the rank of colonel (colonel of police, finance police, internal service), including relieved 20 from service for age, state of health, due to staff cuts, with at least 5 years of service n/a Active-duty servicemen and 21 eligible for conscription, students of military colleges 52300 Servicemen and commanding officers of 22 authorities of interior affairs, financial investigations, emergency situations n/a * Servicemen and commanding officers of authorities of interior affairs, 23 financial investigations and emergency situations who were injured n/a 24 Orphans 32878 25 Children under 7 751096 * Members of family of 26 deceased in WWII 4132 Individuals accompanying disabled Category 1 and 2 27 and children with special needs under 18 n/a Continued from previous page 44 Appendix 4. Information on Privileges and Beneficiaries of Privileges Based on HBS Data Table A4.1. Number of HHs who reported privileges and subsidies (000 families) Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 All privileges and 2,466.2 2,358.3 2,325.9 2,405.0 2,417.4 2,384.6 2,376.7 1,547.7 1,490.3 subsidies, of which for purchase of meals 334.7 316.0 307.1 328.6 357.6 364.0 352.8 386.0 397.7 for public transportation 1,677.5 1,630.0 1,608.2 1,695.4 1,746.9 1,704.9 1,752.8 746.0 698.7 for housing and utility 550.2 485.5 429.9 448.1 365.0 371.4 334.2 215.3 178.4 services for medical treatment and 346.3 366.1 426.0 459.3 447.0 338.0 367.6 330.3 349.3 rehabilitation for purchase of medicine 754.1 670.5 548.9 582.6 614.6 657.4 713.0 348.9 345.6 for services of 296.2 211.9 214.9 253.9 204.9 178.3 144.8 152.2 100.3 kindergartens other privileges and 73.1 53.9 57.6 115.8 115.5 111.4 122.5 85.4 66.9 subsidies Table A4.2. Share of HHs who reported privileges and subsidies Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 All privileges and subsidies, of which 64.1 61.2 60.6 64.4 64.9 64.2 64.0 41.7 40.1 for purchase of meals 8.7 8.2 8.0 8.8 9.6 9.8 9.5 10.4 10.7 for public transportation 43.6 42.3 41.9 45.4 46.9 45.9 47.2 20.1 18.8 for housing and utility services 14.3 12.6 11.2 12.0 9.8 10.0 9.0 5.8 4.8 for medical treatment and rehabilitation 9.0 9.5 11.1 12.3 12.0 9.1 9.9 8.9 9.4 for purchase of medicine 19.6 17.4 14.3 15.6 16.5 17.7 19.2 9.4 9.3 45 for services of kindergartens 7.7 5.5 5.6 6.8 5.5 4.8 3.9 4.1 2.7 other privileges and subsidies 1.9 1.4 1.5 3.1 3.1 3.0 3.3 2.3 1.8 Table A4.3. Total monthly value of reported privileges and subsidies (number of beneficiary HHs multiplied by average value of privileges and subsidies) (mil BYR) Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 All privileges and 18,496.4 25,941.8 32,795.9 43,048.7 47,380.9 46,022.4 54,188.9 64,537.2 53,053.8 subsidies, of which for purchase of meals 4,117.1 4,739.8 5,680.5 7,558.4 8,832.2 8,881.6 9,172.6 4,014.3 14,832.5 for public 1,845.2 3,423.1 5,307.1 7,290.3 9,782.8 10,229.2 13,146.1 14,994.4 5,100.4 transportation for housing and utility 1,320.4 2,573.4 4,341.8 4,974.2 4,891.4 5,608.6 5,882.3 1,248.5 2,854.2 services for medical treatment 7,964.1 10,726.2 13,505.5 17,179.0 16,985.1 12,708.8 16,139.7 2,939.8 20,122.1 and rehabilitation for purchase of 2,865.5 3,687.8 3,073.6 4,660.5 5,715.7 7,100.3 8,413.5 3,279.4 9,401.0 medicine for services of 325.9 699.4 752.3 965.0 1,065.3 998.4 984.8 623.9 521.8 kindergartens other privileges and 131.6 167.2 178.5 416.8 369.5 568.3 477.9 196.3 327.8 subsidies Table A4.4. Value of privileges and subsidies per HH per month (000 BYR) Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 All privileges and subsidies, of which 7.5 11.0 14.1 17.9 19.6 19.3 22.8 41.7 35.6 for purchase of meals 12.3 15.0 18.5 23.0 24.7 24.4 26.0 10.4 37.3 for public transportation 1.1 2.1 3.3 4.3 5.6 6.0 7.5 20.1 7.3 for housing and utility services 2.4 5.3 10.1 11.1 13.4 15.1 17.6 5.8 16.0 46 for medical treatment and rehabilitation 23.0 29.3 31.7 37.4 38.0 37.6 43.9 8.9 57.6 for purchase of medicine 3.8 5.5 5.6 8.0 9.3 10.8 11.8 9.4 27.2 for services of kindergartens 1.1 3.3 3.5 3.8 5.2 5.6 6.8 4.1 5.2 other privileges and subsidies 1.8 3.1 3.1 3.6 3.2 5.1 3.9 2.3 4.9 Table A4.5. Value of privileges and subsidies per capita per month (000 BYR) Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 All privileges and subsidies, of which 8.5 10.2 12.6 14.3 for purchase of meals 3.6 4.3 5.3 6.6 7.3 7.2 7.5 10.5 10.9 for public transportation 0.6 1.0 1.5 2.0 2.7 3.0 3.7 3.5 3.8 for housing and utility services 1.5 3.6 6.7 7.1 8.7 10.1 12.5 10.1 10.9 for medical treatment and rehabilitation 8.8 10.0 11.0 13.2 14.1 13.1 15.2 18.3 20.6 for purchase of medicine 2.1 3.2 3.1 4.3 5.1 5.7 7.0 11.9 14.7 for services of kindergartens 0.3 0.9 0.9 0.9 1.3 1.4 1.7 1.2 1.1 other privileges and subsidies 1.0 1.5 1.5 2.1 1.7 3.0 2.3 2.7 2.3 Table A4.6. Change of Value of Privileges and Subsidies per HH Year 2001 2002 2003 2004 2005 2006 2007 2008 2009 All privileges and subsidies, in 2001 prices 7.5 7.7 11.0 15.2 17.8 18.0 21.0 36.3 31.5 (000 BYR) Y-o-y change, percent 102.9 142.4 138.0 117.2 101.5 116.6 172.7 86.7 Change since 2001, percent 102.9 146.4 202.1 236.9 240.5 280.4 484.3 420.1 47 Appendix 5. Making GMI More Efficient In Fighting Poverty: Good Incentives and Social Inclusion. Lessons from Europe Although modest by the European standards,19 the share of its GDP that Belarus spends on social protection (SP) is sizable (15 percent). These programs have been shown to reach almost 75 percent of the country’s population. Most of the SP budget is paid in pensions to approximately 45 percent of Belarusians, and pensions have been shown to play an important role in reducing poverty. Social assistance (SA) programs are financed at 2.3 percent of GDP and reach 56.4 percent of the population. SA programs tend to be poorly targeted toward those truly in need and exhibit notable leakages to the higher deciles of income distribution due to extensive privileges inherited from the Soviet era. Policymakers around the world face the challenge of adopting the right mix of safety net policies and programs that will help reduce poverty and vulnerability in their countries. Programs that guarantee minimum income usually offer income support to those without sufficient means to meet the necessary costs of living. Designed explicitly to fight poverty, these programs have become a very popular and efficient instrument to achieve this end. However, in Belarus, despite the recent attempts to modernize the country's social safety nets (SSNs), categorical aid prevails, and a very small share of programs is means tested. As this study demonstrates, income support through GASP (still miniscule at 0.03 percent of GDP) could be expanded to amplify the impact that a good SSN system is expected to have on poverty and vulnerability. In the European Union, almost every country has some type of minimum income scheme that provides income support, but these programs vary markedly in structure and coverage. As a result of a recent EU attempt to classify these programs,20 three distinct groups emerge. The first group comprises programs that utilize minimum income as the only (or most important) universal measure that is open to all of those who lack sufficient resources and that is not limited to specific target groups. Using GMI as a universal measure is typical of Austria, Luxemburg, Malta, Poland, Romania, and the Slovak Republic. By and large, Finland, France, Germany, Ireland, and the United Kingdom belong to the second group: the use of minimum income as a measure of last resort for all of those who already have exhausted all other possible claims for targeted measures. Both categorical assistance schemes and a general minimum income support are provided in these countries. (The Belarusian SSN shows a similar pattern.) Finally, Greece, Hungary, and Italy have only categorical schemes and no last-resort measures. Spain has no national GMI scheme. Belgium, the Czech Republic, Netherlands, and Sweden have systems that contain a rather broad measure as described under the first group above. In contrast, Bulgaria, Cyprus, Denmark, Estonia, Latvia, Lithuania, Portugal, and Slovenia tend to employ minimum income as a last-resort measure (second group). 19 In 2004 the average spending on SP in the EU (excluding administrative costs) represented 26.2% of GDP. SP expenditure by countries ranged from 12%–20% in the Baltic States, Czech Republic, Hungary, Ireland, Malta, Poland, and Slovakia to approximately, or even above, 30% in Denmark, France, Germany, and Sweden. 20 “Role of Minimum Income for Social Inclusion in the European Union.� Policy Department Economy and Science, DG Internal Policies, European Parliament, December 2007. IP/A/EMPL/ST/2007-01 PE 401.013. 48 The poverty-reducing effect of social transfers is most evident in Austria, the Czech Republic, France, Hungary, the Netherlands, and Nordic countries, where all social transfers reduce poverty by half or more. Sweden is the country that is most effective in reducing poverty. This goal is achieved almost entirely by social transfers; pensions have little relevance. In the Swedish SA system, means- tested benefits complement the quasi-universal social insurance (SI) and extensive social services. All but the old age assistance measure are contributory. GMI is designed for those who lack sufficient means to meet the necessary costs of living. The duration of GMI payments is unlimited. Some active measures to achieve gainful employment are provided, but GMI has no specific social inclusion programs. A good GMI program is not only one that is well targeted, or pro-poor. Obviously, a good program also would support the poor and at-risk-of-poverty cost efficiently and sustainably. Above all, the program would give these groups the right incentives and help them seek social reintegration. Indeed, many factors affect poverty. However, “...while all these factors are important, getting people back into the labour market––and the education and training that can help them find jobs––is key, because it is the one dynamic factor capable of changing individuals’ conditions and enabling them to become self-sufficient on a lasting basis.�21 It is critical to get individuals back into jobs. For this reason, the EU’s social policy sets ambitious goals for reducing poverty and increasing social inclusion, as well as policy measures that promote the provision of a guaranteed minimum income (to all residents, not just nationals) with accompanying labor activation measures. For unemployed and inactive (other than retired) individuals, who tend to have a higher risk of falling into poverty, access to employment is a key aspect of social inclusion. Here, the challenge for the social service is to combine an economic and a social dimension in one policy that will guarantee an adequate standard of living but also will encourage job seeking. Alongside access to employment, setting the incentives right is important because remaining as benefit recipients can bias HHs’ behavior. In this regard, limiting benefit duration and linking its renewal to job seeking effort is highly plausible. Another option may be to add a requirement for public service, work, or enrollment in some sort of training. Conditional cash transfer (CCT) programs help build the human capital of HHs. Thus, a good GMI program is a social scheme with a strong integration component. This component provides personalized support to beneficiaries to enhance their earning opportunities and also helps them activate themselves toward social inclusion, while keeping labor disincentives that some beneficiaries might develop under control. The key question that remains is whether a policy such as minimum income is capable of creating a long-lasting exit from poverty, social integration, and permanent employment. This question makes one aspect of SA programs in the EU––social activation––particularly interesting. Most EU countries combine an economic allowance with programs aimed at helping 21 OECD, “Combating Poverty and Exclusion through Work,� Policy Brief, Paris, March 2005. 49 beneficiaries to find a job or to improve their labor skills such through training and internships (Bulgaria, Czech Republic, Denmark, France, Germany, Ireland, Latvia, Luxembourg, Malta, Netherlands, Portugal, Slovak Republic, Slovenia, Spain, Sweden, and UK). A few countries do not complement SA with built-in social and employment services, or economic incentives, although in Austria people capable of work are expected to be willing to perform reasonable work; and in Lithuania and Poland, to be registered at the labor office and available for training and work. In all of the countries that associate the economic benefit with a strong emphasis on labor reintegration, the support is aimed at fostering the resources and competencies that the most disadvantaged people have and creating opportunities for them to apply their strengths to benefit their lives and their society. Beneficiaries are not left on passive income support. Moreover, in principle and in most cases, beneficiaries are directly involved to account for personal objectives and preferences in the elaboration of the activation plan. Minimum income is an integral part of any workfare strategy designed to facilitate the activation of public resources. Guaranteeing the minimum income goes beyond helping to improve the circumstances of the poorest. By calling all stakeholders including trade unions and private firms to the table, GMI also enhances local social capital. The workfare process benefits from the supply side’s knowledge of local labor-market problems and wins them as responsible partners in developing new workfare policies. Denmark has the highest expenditure level in Europe for housing and for reducing social exclusion, and the most effective transfers (excluding pensions) in reducing poverty. In addition, the country’s SSN system has been effective in enhancing the flexibility and skills of the labor force. At the same time, the SSN succeeded in creating enough sheltered jobs for those who cannot compete in the normal labor market. Approximately half of those who have been exposed to the activation component of the program claim that their qualifications, self-confidence, and job chances have notably improved. As a result, the number of unemployed among SA recipients decreased. France is another country with a high level of expenditure and effectiveness of its policies implemented to reduce poverty. In 1998 France developed a comprehensive policy to combat exclusion, similar to the Irish and British antipoverty programs. The core of the French SA program remains the guaranteed minimum income (RMI). By way of encouraging RMI recipients to get back to work, the program includes exemptions from SI contributions and cash payments to employers for taking on the long-term unemployed. The German means-tested benefits supplement SI benefits, which are based primarily on previous earnings and contributions rather than on individual needs. In Germany, a refusal to accept a job may result in a reduction of at least 25 percent in the basic rate payable. The country has a high total expenditure, well distributed across the target groups, and a medium effectiveness in reducing poverty, with social transfers and pensions being of the same relevance. Italy does not have a national GMI program. Income support delivered in certain regions has helped families to recover from certain potentially dangerous lifestyles. Families have recovered 50 from arrearage (avoiding eviction) and debts (also with the public administration); and the number of minors leaving school early has been reduced. Just as in any other SA program, a good GMI program must be adequate, equitable, cost- effective, incentive compatible, and sustainable.22 Based on an assessment of SA programs (GMI among them) carried out in the EU countries,23 we discuss specific solutions to make programs operate along these lines. The analysis demonstrates that the weaknesses and risks of GMI schemes usually are linked more to irregularities of implementation than to the characteristics of the measures themselves. This finding stresses the need to systematically review GMI implementation. Adequacy The three dimensions to consider are coverage, level of benefit, and duration. The profile of poverty and vulnerability will provide an indication of the level of need. In real life, considering possible disincentive effects or fiscal constraints may limit the program's volume and reduce payments. To guarantee an adequate minimum income for all to live in dignity, social protection systems must have sufficient coverage and levels of payment. This is one of the main problems in most European countries: the coverage is insufficient to guarantee either decorous levels of payment or support to all who are in need and eligible for it, or both. Insufficient coverage that may account for an ineffective alleviation of poverty often is due to the fact that a threshold set by the schemes is too low, thus making some parts of population not eligible. As was shown in this study, a low threshold also is the case in Belarus, where given the poverty line of US$88, approximately 4.9 percent of population would be eligible for GASP. Raising the poverty line to cover 10 percent of population would increase program expenditure from 0.21 percent of GDP to 0.55 percent of GDP. In addition, means-tested benefits often are not fully taken up by those eligible. In Belarus, while approximately 480,000 would qualify for GASP at the poverty line of US$88, only 194,808 were actually receiving it in 2009.24 While coverage and benefit size depend on the program design, non-take-up is determined both by individual perceptions or behavior and by program design/administration process. In fact, non-take-up may be inherent in the program design whenever governments accept stigma/transaction costs as a way to reduce program enrollment. Administrative toughness fulfills the purpose of screening applicants to exclude those with higher permanent income (such as the self-employed) and to target those with the most urgent (and long-term) needs for assistance. In UK, for example, different researchers have identified two main types of barriers in claiming Income Support, particularly among pensioners: (a) a “stigma� dimension associated with 22 M. Grosh, C. del Ninno, E. Tesliuc, and A. Ouerghi. For Protection and Promotion: The Design and Implementation of Effective Safety Nets, World Bank, 2008. 23 “Role of Minimum Income for Social Inclusion in the European Union.� Policy Department Economy and Science, DG Internal Policies, European Parliament, December 2007. IP/A/EMPL/ST/2007-01 PE 401.013. 24 The difference may not be fully due to the issues with take-up. In practice, the GASP benefit is awarded to those whose income still falls short of the MSB after they have received all other transfers. The estimate based on the 2008 HBS cited here could have left out this subtlety. 51 claiming income-related benefits and (b) a “process� dimension consisting of objections to (or negative perceptions of) various aspects of the claim process, for instance, stemming from bad past experiences with the social security system. These findings emphasize the importance of regular evaluation of means-tested benefit schemes to introduce changes to the programs that will positively influence the program’s public image and perception. As evidenced in Finland, changes in behavior could drive the observed downward trend in take- up rates during the post-recession period, highlighting the increasing stigma of relying on SA during economic upturns. The Finnish SA scheme, Toimeentulotuki, is a relatively generous program that provides a financial safety net for those with no or very limited incomes from other sources. In a different study of the claiming patterns, a set of stable determinants of claiming behavior was identified.25 Measuring take-up is a difficult task. The authors point out that existing evidence does not make full use of the information recorded by social service administering benefits. Consequently, they combined national administrative data and eligibility simulations based on the tax-benefit calculator used by the Finnish authorities to derive more accurate measures of non-take-up. Rates of non-take-up were found to be both substantial and robust: 40 percent–50 percent of those eligible and of working age did not claim. According to this study, stigma effects associated with claiming welfare benefit may be felt more acutely when unemployment declines during periods of economic recovery. Alternatively, the increase in non-take-up could be due simply to a composition effect. A decrease over the recession period in the number of long-term unemployed as a group characterized by a higher propensity to claim, drives down the rate of non-take-up. However, the change in the composition of the eligible population over the studied period was found to be marginal in Finland. While the share of long-term unemployed in the population saw a marked decline, this did not carry over to the subpopulation of people entitled to SA (the core poverty group). Hence, in Finland the decreasing number of SA recipients was not a direct consequence of lower unemployment but more likely a result of a change in take-up patterns during the economic recovery. Households claiming behavior may not be the only factor affecting take-up. Many OECD countries recently have moved toward a "rights and responsibilities" approach. It emphasizes the responsibility of the recipient to comply with certain behavioral requirements, job-seeking among them, to retain the benefit. While sanctions often are partial, it is likely that an increasing number of low-income individuals fail to receive benefits not because they do not claim them but because they are denied benefits as a result of behavior that cannot easily be observed. The practice of limited duration of GASP benefits is plausible if the purpose is to curb benefit dependency and make beneficiaries focus on job seeking. In seven countries of the EU, the duration also is limited and varies between 3 months and 24 months. In most countries, however, GMI is paid as long as the beneficiary meets the eligibility criteria. In view of the program's adequacy in addressing the depth of poverty and the risks of re-entering poverty, increasing the 25 O. Bargain, H. Immervoll, and H. Viitamäki, No Claim, No Pain: Measuring the Non-Take-up of Social Assistance Using Register Data, September 2010. 52 duration of payments may prove beneficial. Nevertheless, it should be done only after careful analysis of the poverty profiles of the pertinent target groups. Equity Equity analysis examines the distribution of benefits across pertinent groups, showing both who is included in the program and who is excluded. The benchmarks to be used in judging whether the patterns of benefit are acceptable are not absolute. They depend on how the target groups are defined. Should GMI be the last resort extended to the most marginalized with no other means, or should it be a measure for the unemployed who still have residual personal and professional resources? Each country makes its choice. In any case, minimum income should be part of a broader policy against poverty and social exclusion and should take into account the peculiarities of each specific target group. Specific attention should be focused on the difficulties associated with activating weak and marginalized groups, a large group of beneficiaries who are difficult to activate because they have severe social, psychological, or health problems. Incentive Compatibility In many countries, debate has concentrated on the relationship between social policies (minimum income, unemployment compensation) and work. Policies providing income for those out of work are thought to create inactivity or the “poverty/unemployment trap,� a term that became fashionable in all European countries in the second half of the 1990s. These disincentives to work may be summarized as follows: a. The size of allowance allows HHs to live in dignity without working. b. To work in “integration jobs� is far less demanding than to work in regular jobs, for almost the same level of income. c. If minimum income is equal to minimum wage, there is no need to work. d. The integration in the labor market has costs that can make it unattractive to beneficiaries, especially if the job is part time or short term. For example, in some countries, if they take these kinds of jobs, beneficiaries lose free health care or housing allowances. Solutions and measures typically introduced to diminish/neutralize disincentives to work could be direct, such as introducing financial incentives––negative income tax, unemployment insurance systems, adjusting levels of minimum wage and minimum income; or indirect–– enhancing job search efficiency, or providing better working conditions. Among specific strategies to avoid the poverty trap and to stimulate beneficiaries to accept job opportunities, a number of countries exclude earnings (in part or total earnings for a limited period) from income calculation. A fixed amount is discounted from the income in Belgium (for a maximum of 3 years), Cyprus, and in the UK (fixed discounts per week). Moreover, in Denmark, earnings from work performed in the framework of an activation measure are completely deducted, except the pay per working hour capped at 160 hours per month. In comparison, a share of earnings is not taken into account in Portugal (20 percent) and in the Slovak Republic (25 percent). 53 France, Ireland, Latvia, Malta, and UK follow the strategy of “backtowork� allowances, which envisage gradually scaling down minimum income support when a beneficiary begins to work. The British Income Support pays benefits to single parents for an additional 2 weeks and to beneficiaries receiving help with mortgage interest for 4 weeks after such beneficiaries have started to work and moved out of the income support scheme. In Latvia, when a recipient of income support has started to work, her benefit is granted for an additional 3 months at a rate that decreases from 75 percent to 50 percent to 25 percent of the initial benefit. In addition, in many countries, participation in employment and social programs is a condition of entitlement. In Denmark, for example, an unemployed person who rejected a fair offer of activation while on benefit loses the right to unemployment benefits for four weeks. A fair offer of activation rejected during the activation period leads to an immediate loss of the right to unemployment benefits. In UK, to avoid the “poverty trap,� a low level of payment has been set for all adults of working age. To ensure that people are better off working, changes have been made in eligibility rules, with part of earnings disregarded for Income Support. In particular, the tax credits are supposed to be useful not only as incentives to work but also to improve the take- up rates of entitlements and reduce stigma by switching from a benefit to an entitlement to retain more of one's earned income. In certain cases, steps taken to create incentives for new employment opportunities with the supply-side representatives of the labor market prove less successful. In France, a practice of direct or indirect subsidies to companies was recognized as inefficient as in safeguarding long- term employment because it would encourage setting the wages low. As a result, subsidies would push ever larger segments of the population to poverty thresholds located between RMI (GMI) and minimum wage. The new employment policies introduced in France since the late 1990s along with several socio-fiscal system reforms emphasize, among others, financial incentives to work to boost the financial attractiveness of low-paying jobs. In the Czech Republic, to address the disincentive caused by the equal level of minimum income and minimum wage, a lower level of minimum income was introduced in 2007 for those avoiding employment. Many countries including Denmark, France, and Germany also have made downward corrections to the levels of health care service or housing allowances or, in some cases, the compensation for the transport costs to raise the costs of nonworking and influence the labor decisions of those staying on benefit. So far as parental labor market decisions are made, financial incentives to work matter. The majority of OECD countries now have individualized tax systems, but nearly all OECD countries also have either some form of tax relief for unemployed spouses or some form of family assistance that aggregates incomes of spouses to determine levels of assistance. These arrangements are found to produce weak financial stimuli to work (more) for (potential) second earners, as the effective marginal tax rate of the second earner is close to that of the primary earner.26 The general policy approach toward parents on income support also is a crucial determinant of to what extent such families, among which are many single-parent families, rely on benefit support. 26 OECD, “Babies and Bosses: Reconciling Work and Family Life,� Paris, 2007. 54 The OECD study fairly emphasizes that to engage in paid work is in the long-term interest of all families, including single-parent families, “as this is the most effective way of reducing the risk of family poverty, enhancing child development, and generally giving children the best possible start in life.� Single-parent employment rates in the Ireland and UK (also in Australia and New Zealand) are relatively low, at 45 percent–55 percent, compared to an estimated 70 percent–80 percent in the Nordic countries. The result is poverty, which damages the future life-chances of children. Therefore, in terms of policy approach, Nordic countries treat single parents on income support the same as any other parent: parents who are no longer entitled to paid parental leave (or home- care payments) are work-tested to receive benefit. This policy choice requires active and early interventions toward labor market reintegration of (single) parents on income support. These interventions include investment in childcare, in-work benefits to make work pay, and employment supports through intensive counseling and training programs. To address this issue, since 1997, the United Kingdom has successively increased both the level of child payments and in-work benefits, and childcare supports. These interventions have positively affected the employment incidence among sole parents. Sustainability and Cost-Effectiveness Sustainability usually has three dimensions: fiscal, political, and administrative. In Italy, the measure has not been implemented due to, among other reasons, its high cost. In Germany, the reform has greatly increased the number of job seekers. The Danish Flexicurity has been commented on in the literature as an expensive model due in particular to the costs of activation of the most marginalized. In some cases, sustainability and cost-effectiveness are seriously undermined by fraud associated with the access to the measure. The UK has introduced a “monetary fraud indicator,� which audits the share of fraudulently claimed benefits. In Denmark, the system is based on the trust among all parties involved. In general, the Danish workfare strategy does not aim to repress and punish but rather to get beneficiaries involved in the planning of counseling, training, and education activities that might empower them. In Italy, fraud has represented a real risk, and many resources have been used to find solutions to the problem. A measure such as minimum income needs to be implemented on different institutional levels: in municipalities, which administer active social policy; in regional and central governments, which structure and fund the measure; and on the level of labor market and private businesses at which job reintegration finally takes place. Many problems are linked to the difficulties in cooperation among the different levels of implementation involved. Interestingly, in the French and German contexts, the decentralization of SA would be perceived differently. Decentralization raises concerns about the lack of economies of scale and less mobility, about the use of administrative resources, and about the incentives for cost-efficiency. If administrative units are too small, they will be unable to offer a full range of services; less likely to dedicate professional staff; and too vulnerable to financial risks beyond their control. 55 Almost inevitably, rural districts face bigger challenges in service delivery than do urban centers. A decentralized system generates more administrative costs than if only one level of government were involved. Similarly to other Nordic countries, in Finland in recent years, transfer of responsibilities from central to local government has taken place, possibly accompanied by more restrictive handling of SA claims. Most cited among the benefits of decentralization is the increased flexibility to reflect local preferences and needs. Moreover, a devolved structure allows for local initiative and facilitates policy competition and innovation. In Bulgaria’s search for a more efficient safety net policy, the attempts to decentralize the financing of the GMI program were less successful. From 1998 to 2003, the government experimented with different formulae for cost-sharing between the central and local authorities. In 1999 the cost-sharing arrangement called for a 50 percent national contribution and 50 percent local contribution. However, this agreement led to arrears or to partial payment of benefits because some local governments did not contribute their shares. By the end of 1999, local arrears represented 10 percent of the entitlements. In 2002 the cost-sharing rule was changed to 75 percent and 25 percent by the central government and local authorities, respectively, but the arrears persisted. In 2003 a new social assistance law centralized the financing and implementation of means-tested programs, and the arrears stopped. Social Activation A program that associates economic support with a strong activation policy may pose a number of challenges. To begin with, offers of general and individual programs and plans for activation and/or labor reintegration, particularly in the most deprived areas of the country, may be difficult to develop. Individual action plans and activation offers may be of uneven quality. There may be difficulties associated with activating weak and marginalized groups whose activation is really costly and requires a certain amount of sheltered jobs in the private and public sectors. In some cases (Italy), the discretionary power left to municipalities as to the modalities of implementation of the measure makes possible special treatment of single cases. Often, these difficulties could be adequately sorted out with increased funding and enhanced administrative capacity as well as qualification of the staff who are asked to elaborate and implement activation plans. In some cases, officers to deal with activation plans simply are not there. Adequate attention to activation may be a problem when social protection staff work in heavily deprived areas at which their concern may be focused on the most marginalized and needy groups of the population. In Italy, for example, the elaboration of inadequate activation plans never became a real and durable social inclusion of the beneficiaries. In Denmark, specific solutions have been adopted to address this issue such as projects aimed at training and bringing social workers closer to the citizens concerned and creating local authority systems. To ensure good implementation of the activation component of such programs, relevant procedures must be in place and staff well trained and sufficient. One risk of activation plans is that, in case of lack of real job opportunities or other forms of activation, many people are introduced to programs of low interest and low effectiveness in the 56 path to social inclusion. In Denmark, this problem has been tackled directly by a law that clearly states that activation offers must improve the employment possibilities of the unemployed. Activation policies integrated into minimum income schemes are likely to work in an environment with job opportunities. Otherwise, the risk is that, especially in depressed areas, minimum income schemes will be demanded to solve the problem of unemployment linked to the low job demand. The performance of the measure is strongly influenced by context. In the areas with severe unemployment, minimum income cannot be asked to solve structural imbalances of the local economy, so the measure risks inducing assistance dependency. Where active employment opportunities are limited or exhausted, social programs often come to the rescue. Usually, these include education programs (Denmark, Ireland); counseling and support to beneficiaries with specific needs related to daily life (France, Slovenia, UK); occasions of active participation in community life for pensioners (Bulgaria, Germany); socially useful activities such as gardening in public spaces, assistance in office work at local councils (Bulgaria, Italy); medical treatment and rehabilitation for beneficiaries with problems of alcohol or drug abuse (Estonia, Latvia, Slovenia); and family care and support such as looking after old people and children (Italy). Public works programs should be designed to accommodate the complexity of the country's infrastructure and the character of its labor market. Such programs usually select works at the local government level and may have a lower labor content. Alternatively, the labor hours may be used for a completely different range of public service activities. For example, beneficiaries may work in parks, libraries, schools, or hospitals; or may act as home aides for the elderly or those with disabilities. This usually means that workers are allocated across many different agencies rather than working in large groups on a few construction sites. Where the labor market is mostly formal, public works programs sometimes subsidize employment in private firms or count training or job search as labor effort. Such programs augment the beneficiary’s earning potential alongside immediate income support. All of these variations to the basic public works scheme make supervising the labor effort, finding suitable placements for workers, monitoring the program, and evaluating it on its multiple objectives much more difficult.27 Coupled with measures to facilitate integration in the labor market for those capable of working, income support not only provides relief from poverty but also helps individuals and families durably elude poverty. GMI is most effective whenever adequate health care, education, housing, and social services are available. In the countries that established an integrated approach to SA social expenditure appears more effective in reducing poverty, and unemployment (as one of the most important causes of poverty) has been consistently reduced: this is the case for Denmark and Germany, but also for France and the UK in relation to specific target groups (single-parent HHs). Evidence from the EU countries, the new EU members including Bulgaria among them, reconfirms that successful safety net systems and programs should be dynamic, with the balance and implementation of programs evolving accordingly over time. The mix and design of 27 M. Grosh, C. del Ninno, E. Tesliuc, and A. Ouerghi, For Protection and Promotion: The Design and Implementation of Effective Safety Nets, World Bank, 2008. 57 programs should respond as needs change, and the implementation of individual programs should involve a constant search for improvements.