Maintaining Adequate Protection in a Fiscally Constrained Environment Measuring the efficiency of social protection systems May 27, 2013 Brussels, Belgium Ramya Sundaram. rsundaram@worldbank.org The World Bank Europe and Central Asia Region
Social sector spending accounts for the major share of government spending in EU countries Social sector spending as % of total government spending 2010 Denmark Finland Norway Sweden Austria France Germany Luxembourg United Kingdom Estonia Italy Slovenia Lithuania Belgium Portugal Spain Netherlands Malta Poland Czech Republic Greece Bulgaria Slovakia Hungary Latvia Romania Iceland Cyprus Ireland 0 10 20 30 40 50 60 70 80 Health Education Social Protection 2
and boomed over 2004-2008 Real growth in social spending, contribution by sub-sector, during boom (2004-2008) Romania Slovakia Lithuania Poland Estonia Latvia Czech Republic Ireland Greece Cyprus Spain Norway Bulgaria Netherlands Luxembourg Slovenia Malta Finland France Austria Belgium Hungary Italy Portugal Denmark Sweden United Kingdom Germany Iceland Driven by increases in social protection (mainly pensions) -40-20 0 20 40 60 80 100 120 Health Education Social Protection 3
Social spending has to adjust not just because of the crisis, but also because of aging 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Large Increase in EU15 Old Age Dependency Rate Aging will add to pressures on pensions + long term care Doubling of elderly populations and sharp declines in workforces Increased dependency ratios (biggest increase in EU15 in Italy, Portugal and Spain) 1970 2010 2050 4
Social Protection Framework Labor market services Training Employment incentives Supported employment and rehabilitation Direct job creation Start-up incentives Unemployment benefits Early-retirement benefits Labor Market Programs Social Assistance Last-resort social assistance programs (LRSA) Family and child benefits Social pensions Heating and housing allowances Disability allowances Old age Disability Survivor Sickness Family/children Social Insurance 5
What efficiency to measure? Effectiveness of social assistance programs in protecting the poor Policy coherence, and efficiency of the mix of social safety net programs Administrative efficiency, and links across programs 6
EFFECTIVENESS OF SOCIAL ASSISTANCE PROGRAMS IN PROTECTING THE POOR 7
Typically most social assistance spending is not targeted to the poor Spending as a share of GDP Denmark Sweden Norway Cyprus Finland Luxembourg United Kingdom France Iceland Germany Austria Hungary Netherlands Ireland Belgium Italy Greece Slovenia Slovakia Malta Spain Czech Republic Lithuania Romania Bulgaria Portugal Latvia Estonia Poland Share of non-means-tested programs in total SA spending Spending on Social Assistance (2009) Non-means-tested Means-tested Share of non-means-tested programs in total SA spending 12 10 8 6 4 2 0 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 Source: ESSPROS data, Staff calculations. 8
Social assistance covers many of the poor in some countries, but the rich benefit too Social Assistance: Coverage of the poorest and richest quintiles (%, 2009) 100 Poorest Quintile 90 Richest Quintile 80 70 60 50 40 30 20 10 0 Performance indicators are generated using standardized methodology that includes the use of: 1. EU-SILC surveys 2. Ranking households into quintiles based on pre-all social assistance-income 3. AdePT software (Social Protection module) developed by DECRG team 9
Detailed program information is important to measure efficiency of spending Public spending on Social Assistance by Quintile (% of GDP) in 2010, Latvia Family state benefits Municipal GMI benefit Poorest Quintile Housing allowances Child care benefits Second Quintile Third Quintile Fourth Quintile Richest Quintile 0.00 0.05 0.10 0.15 0.20 0.25 0.30 Percentage of GDP 10 Source: Latvia EU SILC 2011, Central Statistical Bureau of Latvia, World Bank staff estimates. Expenditure data is from MoW
Consolidating social protection information makes measurement difficult EU-SILC data consolidates a lot of information on social protection programs Social assistance not elsewhere classified includes recurrent and lump-sum benefits, and makes it hard to measure the impact of recurrent benefits For Bulgaria Monthly benefit for low income households Lump-sum benefit for occasional needs arising Target benefit for heating Target benefit for free traveling For Latvia Guaranteed Minimum Income Lump-sum municipal benefit in emergency situation Benefit for politically repressed Compensation paid to persons engaged in work in the galleys And so on The targeting accuracy (amount of benefits received by the poorest quintile) for the GMI benefit in Latvia is 77 percent. The targeting accuracy for social assistance not elsewhere classified is 33 percent! 11
POLICY COHERENCE, AND EFFICIENCY OF THE MIX OF SOCIAL SAFETY NET PROGRAMS 12
Index (2003 = 100) Expenditure on targeted programs decline; expenditure on disability benefits expand 300 Social protection program expenditure, Bulgaria 250 200 150 100 50 0 2003 2004 2005 2006 2007 2008 2009 2010 Pensions Last Resort Social Assistance Housing & utility benefits Disability benefits Active labor market programs Source: World Bank ECA Social Protection Database 13
Thousands Disability assistance outpaces disability pensions (both in terms of recipients 180 Number of Beneficiaries of Disability Programs in Albania 160 140 120 100 80 60 40 20 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Disability pension Disability allowances Source: World Bank ECA Social Protection Database 14
and spending as a percent of GDP 1.6% Spending on Disability Programs in Albania, % of GDP 1.4% 1.2% 1.0% 0.8% 0.6% 0.4% 0.2% 0.0% 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 Disability pension Disability benefits Source: World Bank ECA Social Protection Database 15
ADMINISTRATIVE EFFICIENCY, AND LINKS ACROSS PROGRAMS 16
Georgia: unified method of targeting o All households can apply for benefits o IDs are assigned to households who apply o Almost half of Georgian households are in the database o The application form includes measurable characteristics of households o Households are assigned a score o Score is a consumption proxy o Score is the linear combination of these characteristics complemented by subjective assessment of the household s living conditions o Households are ranked based on the aggregate score
The Data Base of the Socially Vulnerable Families The family has five members (three children). Single family. Thepensioner s household has Receives a pension and monthly income of GEL 20 The household two social assistance Ranking score with - by 39social 550 members. The household category hasfamily monthly income with three of RankingGEL score - 47has 950a members. The 80 family disabled child, who Ranking score -has 64 four 300 receives a pension and The household social assistance members.by social category has a The household monthly income of470 Ranking score - 155 GEL 1,050 DATA BASE Ranking score - 665 960 5 4 3 18 1 2
The Data Base of the Socially Vulnerable Families DATABASE 5 Electricity subsidies score - 200 000 4 Health insurance score - 70 000 3 Monetary benefits score - 57 000 1 2
60.0 50.0 40.0 30.0 20.0 10.0 Share of beneficiaries in the number of applicants 55.3 29.0 Share of applicants in extremely poor HHs 72.9 27.1 0.0 Receiving health insurance Receiving cash benefit Not applied Applied
Moving towards a systems approach to social protection 1. Having coherent objectives for social protection programs 2. Ensuring that the program is well designed, given the stated objective 3. Evaluating the program against the stated objective 4. Providing robust protection against poverty 5. Providing the right incentives to work, when jobs are available 6. Understanding and monitoring linkages across programs 7. Aligning administrative subsystems across various programs 8. Aligning institutional arrangements that enhance coordination across agencies 9. Developing information systems that are flexible, converts data into information, and that is well designed so as to capture the needed level of detail.