Journal Issue
Share
Article

Italy: Selected Issues

Author(s):
International Monetary Fund. European Dept.
Published Date:
September 2014
Share
  • ShareShare
Show Summary Details

Improving Public Spending Allocation and Performance in Italy: an Efficiency Analysis1

Budget allocation in Italy will need to increasingly rely on an efficiency analysis to find savings and improve performance. Achieving sizable expenditure savings will require addressing the large social spending, in particular, current pensions. In the education and non-pension social protection sector, the room for savings may be limited, but there exists scope for improving outcomes. In the health sector, the analysis shows room for savings, particularly in the areas of goods and services by reducing cross-regional variation in spending efficiency. Improving allocation of regional transfers to sub national governments using standard costs based on more efficient benchmarks could provide an important avenue for savings.

A. Background

1. Italy faces a challenging situation to reduce its high public debt. Public debt, at over 134 percent of GDP, is the second highest in the euro area. Refinancing needs of over 24 percent of GDP leave Italy vulnerable to financial market swings. Sizable efforts to curtail spending and raise taxes have been undertaken over the past decade. Nevertheless, the debt overhang and the interest burden means that Italy needs to keep running a sizable primary surplus in order to lower debt. The tax burden is also very high stifling the economic recovery. Furthermore, low trend growth and disinflation are leading to unfavorable debt dynamics.

2. Italy’s medium-term fiscal strategy relies heavily on spending reforms to lower the tax burden while also undertaking fiscal consolidation to strengthen fiscal sustainability. In the April 2014 DEF, the authorities aim to reduce the fiscal deficit by 2¾ percent of GDP by 2017, in keeping with its fiscal rule of zero structural balance. Spending levels will be lowered to 3¼ percent of GDP while revenues will also decrease by ½ percent of GDP. Under the reform plans, the tax wedge will be reduced further by 0.7 percent of GDP annually from 2014 to 2017.

3. Fiscal reforms will require an in-depth spending analysis to identify savings while preserving public spending quality. The government has undertaken a spending review and identified savings of EUR 34 billion over 2014–16 to lower the tax burden and support fiscal consolidation. Going forward, the priority will be to take better account of the quality concerns in the resource allocation framework. Accordingly, this paper seeks to examine the efficiency of spending on a cross-country and cross-regional basis to identify potential areas of savings and further improvements.

B. Comparison with Euro Area Spending and Efficiency Indicators

Spending Structure in Italy

4. Primary spending level in Italy is average by Euro area standards, but is heavily biased towards pension spending. Overall spending is slightly above the euro area average reflecting the large debt servicing burden, the highest in the euro area. A breakdown by economic classification shows that public spending is concentrated in social benefits spending with a large share in public pensions (about a third of primary spending). This reflects both a high share of elderly population as well as a generous pension system with a high replacement rate (OECD, 2013). As a result, Italy has the highest elderly bias in social spending in the euro area after Greece (Vanhuyess, 2013). Indeed, Italy currently spends about 7 times more on each elderly (above age 65) compared to a non-elderly. In terms of functional classification, spending in Italy also exceeds euro area average spending in general services, defense, public order and environmental spending, although the excess relatively small. In contrast, the main areas where Italy under spends are in education and economic affairs.

Primary GG Expenditure, 2013

(Percent of GDP)

Source: Eurostat.

Age-Related Spending, 2013

Source: OECD: and staff calculations.

Interest Expenditure, 2013

(Percent of GDP)

Source: Eurostat.

Elderly Bias in Social Spending, 2007-08

Source: Vanhuyess (2013).

5. Further decomposition across classifications (COFOG) shows pockets of overspending relative to euro area average. In particular, intermediate consumption spending (mainly comprising goods and services spending) is more than 25 percent higher in environment and health sectors compared to the euro area, but is much lower in education. Similarly, wage spending is at least 15 percent higher in defense, health, and public order and safety. Capital spending also exceeds euro area averages in defense, general services, environment, and health.

Italy and Euro Area: General Government Spending, 2000–12(Percent of GDP)
Functional Classification
Total

Expenditure
General

Services
DefensePublic

Order and

Safety
Economic

Affairs
Environmental

Protection
HousingHealthRec/CultureEducationSocial

Protection
Economic Classification
Total ExpenditureItaly50.69.11.41.93.40.90.77.30.74.221.0
Euro area49.86.81.31.84.30.80.87.41.15.020.5
Intermediate ConsumptionItaly5.71.00.30.30.30.60.21.90.30.40.3
Euro area5.50.90.50.30.70.40.20.90.30.70.5
Compensation of EmployeesItaly10.61.40.91.50.40.10.12.40.23.20.3
Euro area10.51.50.71.30.60.20.21.60.33.30.8
SubsidiesItaly1.00.00.00.00.90.00.00.00.00.10.0
Euro area1.30.00.00.00.90.00.00.10.00.10.1
Other current transfersItaly1.60.90.00.00.10.00.00.10.20.10.1
Euro area2.21.00.00.00.20.00.00.10.10.30.4
Other social benefitsItaly22.60.00.00.00.00.00.02.50.00.020.1
Euro area23.40.00.00.00.10.00.04.50.00.218.6
Capital TransfersItaly1.20.00.00.00.90.00.10.00.00.00.0
Euro area1.50.20.00.01.00.00.10.00.00.10.0
Gross Capital FormationItaly1.90.30.10.10.70.20.20.20.10.10.0
Euro area2.10.20.00.10.70.20.20.20.20.30.1
Color code:upto 10 percent more< 10 percent

Evolution of Public Spending

6. Since 2000, overall spending in Italy, as a share of GDP, has increased broadly in line with the euro area. This reflects not only the higher spending growth, which increased mostly during pre-crisis period, but also the decline in nominal GDP since the crisis. In real terms, primary spending growth has been the lowest in the region, declining by 5 percent over 2009–13. The largest spending increases were in social protection and health followed by economic affairs. These increases have been partly offset by reduced spending on education and general services.

GG Spending, 2000-13

(Percent of GDP)

Source: Eurostat. and staff calculations

Real Primary Spending, 2000-13

(Percent change)

7. The rapid increase in current spending has crowded out capital spending. Most of the current spending increase was in social benefits. Intermediate consumption and compensation of employees also increased, mainly in general services, defense and health sectors. The wage bill increased modestly despite significant reductions in public employment, which at 14 percent of the labor force, is below OECD average. Public wages have risen significantly over the past decade driven by collective wage bargaining process and automatic grade progression (Report on Expenditure of Central Government Administration, 2012). The post-crisis wage freeze has helped to curtail this growth. Current transfers also increased, mainly in the general services sector. In contrast, capital spending has declined at a slightly higher pace than in the euro area.

Italy and Euro Area: Change in General Government Spending, 2000-12

(Percent of GDP)

8. Spending has increased at all levels of government. At the central government level, the largest increase was in other current transfers in social protection and general services sectors which may reflect the high internal transfers under decentralization. In contrast, capital transfers and subsidies declined as a share of GDP. At the social security level, pension benefits increased by more than 2 percent of GDP compared to the euro area. For local governments, increases in current spending more than offset the decline in capital spending. This increase was mostly concentrated in the health sector, offsetting the decline at the central government level.

9. The above trends mask a sizeable decline in spending over the past few years. As in the rest of the region, spending has declined as a share of GDP in 2009–12. This decline was across the board, except in the areas of general services (reflecting interest payments) and social protection where the increases were much larger than the regional average, partially offsetting the cuts in other areas especially in capital transfers.

Italy and Euro Area: Change in General Government Spending, 2009-12

(Percent of GDP)

Figure 1.Italy and Euro Area: Increase in Spending by Level of Government, 2000–12

Source: Eurostat; and staff calculations.

10. Looking ahead, Italy fares relatively well compared to the euro area in terms of future spending pressures. With pension and health care reforms already underway, the pace of increase in Italy’s age-related spending is projected to be one of the lowest in the euro area over the coming decades (OECD, 2013). Pension spending is expected to decline over 2010–60, most of it before 2025. While pension benefits are expected to fall significantly, this mainly affects future pensioners while current pensioners are largely unaffected by the 2012 pension reforms.

Change in Age-related Spending, 2010-60

11. To summarize, an analysis of the spending level and growth shows that additional savings is likely difficult without addressing current pensions. Pensions and health have been the main sources of spending pressures since 2000 and have crowded out other areas including education and capital formation. Over the past few years, health care reforms have been implemented and pension reforms have addressed long-run sustainability concerns, but leaving existing pensions largely unaffected. Significant savings is likely to be difficult without addressing the large spending on current pensions.

12. Beyond an analysis of the level and growth in spending, identification of possible savings or reallocations requires an assessment of efficiency and effectiveness. Such an analysis would assess the effectiveness of public spending in achieving specific policy goals. Furthermore, public spending reductions can be achieved without sacrificing outputs by cutting inefficient and wasteful spending. Below, we examine some broad indicators of efficiency of public spending in key social spending areas to gauge potential efficiency gains and possible savings.

Benchmarking Using Efficiency Indicators

Methodological issues

13. We analyze spending efficiency in some key social sectors such as health, education, and social protection using a Data Envelopment Analysis (DEA) methodology. Together, these account for 64 percent of total expenditure. The DEA uses cross-country efficiency scores in a sample of EU/euro area countries to analyze the relative performance in different sectors. Furthermore, the DEA method is also used to analyze spending efficiency at the regional level in Italy.

14. DEA efficiency scores provide a theoretical benchmark of potential savings that can be achieved without sacrificing outcomes. The methodology uses linear programming techniques to construct an efficient frontier of countries that have the maximum output for a given level of input and production technology. This frontier “envelopes” a set of countries based on their input/output combinations. The distance to each of these countries’ input relative to the distance to the frontier provides the efficiency score. The countries on the frontier have an efficiency score of 1. The efficiency scores can be measured in two ways: the input efficiency score provides the horizontal distance to the frontier and the output efficiency score measures the vertical distance to the frontier. So, for example, an input inefficiency score of 0.7 would imply that one could achieve the same output with 30 percent less input.

15. Efficiency analysis at the macro level should be viewed as indicative and requires further in-depth analysis. While this methodology has the advantage of not requiring assumptions on a specific functional form, there are important caveats. The DEA score is a relative concept that is subject to sample bias and to outliers. Importantly, outcomes do not solely reflect public spending. The distance to the frontier may reflect not only inefficiencies, but also idiosyncratic conditions. Since exogenous non-discretionary factors can also impact outcomes, a second stage regression analysis is undertaken to analyze what factors explain the efficiency scores. Where significant, a corrected efficiency score is also calculated. Given these caveats, the efficiency scores below should be considered as indicative rather than an absolute level of savings.

Efficiency of Public Spending in Italy

16. Studies of Italy’s public spending pointed efficiency scores that were relatively low with significant variations across regions. For example, a composite indicator of public sector performance looking at administrative, education, health and public infrastructure outcomes as well as indicators of economic distribution, stability and performance showed Italy with a relatively low efficiency in the OECD (Afonso, et. al., 2005). Specifically, public administration efficiency ranked very low. These scores have likely improved over time given the changes in public administration. Nevertheless, more recent studies of public sector efficiency at the regional level show a consistently lower efficiency in the southern regions (Giordano and Tomassino, 2011). Below, we estimate the efficiency scores for health, education and social protection sectors using the latest available data and discuss policy implications.

Input and Output Efficiency Measurements

Health

17. Reforms have helped to reverse the rapid rise in health spending over the past decade. Health care accounted for much of the spending increase at the regional government level. Since 2009, the Health. Pact has been implemented with the goal of cost containment through supply-side reforms such as limiting hospital beds, admission and length of stay and procurement reforms. Consequently, health care spending has declined over the past few years and is currently slightly below the EU average.

Euro Area: Correlation Between Public Sector Efficiency and Economic Growth

18. Efficiency indicators of public health spending suggest scope for improvement. Outcome indicators such as life expectancy are high in Italy, but adjusted for quality (i.e. adjusted for both mortality and morbidity), health adjusted life expectancy (HALE) indicators and mortality rates appear weaker. Efficiency indicators using the latter indicators (HALE) suggest the same outcomes could be achieved with 25 percent less public health care spending (Text table and Table A1.1). To control for exogenous non-budget factors, these efficiency scores were regressed on per capita income (PPS), old-age dependency ratio and private spending on health as a share of GDP. In this fairly homogenous sample, however, none of these factors were significant in explaining the variation in efficiency scores.

Efficiency of Health Spending

19. A regional analysis of health care spending shows cost inefficiencies, particularly in the south. Health care spending is significantly higher in the southern regions, although outcomes are not particularly worse and outputs are about average (Text table and Table A2.1). This partly reflects the fact that regional health care spending is funded by central government transfers that are determined primarily on a per capita basis. Indeed, efficiency scores based on health spending per capita suggest lower regional variation. However, efficiency indicators using per capita health spending as a share of PPS based per capita GDP suggest relatively large regional variation. Bringing efficiency scores to national average levels in the low performing regions could provide savings of almost 1 percent of GDP. The magnitude of savings is likely lower now since the analysis is conducted using 2011 data and health care spending has declined over the past few years. Second stage regressions results suggest that these efficiency scores are strongly correlated with socio-economic conditions such as per capita income and population at risk of poverty. Population density is also positively correlated likely reflecting ease of providing services. At the same time, fiscal variables such as transfer dependency and size of borrowing are negatively correlated (Tables A2.5-A2.7) suggesting possible room for savings through better allocation of fiscal transfers.

DEA Efficiency Scores for Health Sector
InputOutput
Public Health

Exp/GDP

2011, or latest
HALE Female

2011
HALE Male

2011
Mortality rate,

cancer

2010
Mortality rate,

heart disease

2010
Efficiency (input-based)Rank
Italy6.962.763.4383.3907.40.7614/16
Average6.764.164.5390.8893.10.91
Sources: Eurostat; OECD and staff calculations.
Sources: Eurostat; OECD and staff calculations.
DEA Efficiency Scores for Health Sector, by Regions
OutputInputEfficiency
Life

expectancy

at birth,

2012
Available

beds (per

1000

inhabitants),

2011
Physicians

(per 1000

inhabitants),

2011
Public health

spending as a

share of

regional GDP,

2011
Input-

based
Output-

based
Nord-Ovest82.63.63.96.60.941.00
Nord-Est83.13.83.66.00.951.00
Centro82.83.34.46.91.001.00
Sud82.23.24.010.20.530.99
Isole81.93.24.810.50.800.99
Mean82.63.54.07.90.821.0
Sources: Eurostat; ISTAT; and staff calculations.
Sources: Eurostat; ISTAT; and staff calculations.

20. With recent reforms, these scores have likely improved. Important reforms to base transfer of health care funds on standard costs have been approved. The 2012 health care reform also promoted the prescription of generic drugs, cut the hospital bed ratio, and reduced public financing by €2.1 billion by 2015. At the regional level, measures have also improved primary care; updated health care tariffs; restructured governance of hospitals; revised the list of reimbursable pharmaceuticals; and introduced health technology assessment (HIT, 2013). The Health Pact at the regional level based on standard costs alongside restructuring plans is also being strictly monitored.

21. Policy measures could focus on further supply side reforms. Francese and Romanelli (2013) find that improvements in public administration, use of generic drugs and supply-side reforms could help improve efficiency and gain savings. Regional health technology assessments can be stepped up and more market mechanisms for health care services such as in contracting between regional health care units and providers can be used (HIT, 2013). Additional gains particularly in goods and services spending are also under consideration. An evaluation and follow-up on the recent measures should be undertaken.

Education

22. Increased education spending while improving outcomes is needed to enhance productivity, employability and growth. Education spending has declined over the past decade and is currently the second lowest in the euro area. At the same time, education outcome indicators are weak, particularly for tertiary education attainment, completion rate of upper secondary education, employability rate and PISA scores. Tertiary graduation rates are particularly low. DEA efficiency scores using these indicators and education spending input variables, show Italy is close to the efficiency frontier (Text table and Table A1.3). This implies that room for savings is likely not large, while there is possible scope to increase spending to move along the efficiency frontier and improve outcomes.

Efficiency of Tertiary Education

DEA Efficiency Scores for Education Sector
InputOutput
Public Education

Expenditure/GDP

2011
Total education

spending per

pupil in PPS

2010
Public expenditure on

tertiary education per

pupil (in percent of

GDP per capita)

2010
Tertiary education

attainment (percent of

population, 30-35 yrs)

2011
Upper sec edu

(percent of

Population, 25-64 yrs)

2011
pupil

teacher

ratio

2011
Employability

Rate

2009
PISA

Reading

2012
Public

Spending

Efficiency

score 1/
Total Per

Pupil

Efficiency

score 2/
Tertiary

Education

Spending per

Pupil Efficiency

score 3/
Italy4.26097.325.820.356.012.059.2489.80.991.001.00
Mean5.97685.838.239.070.712.076.8498.60.890.940.92
Sources: OECD; Eurostat; WDI; and staff calculations.

Input: Public Education spending to GDP; Output: All five indicators

Input: Total Education Spending per Pupil, PPS; Output: PISA score, pupil teacher ratio and employability rate

Input: Total Spending on Tertiary Education per Pupil, PPS; Output: PISA score, tertiary education attainment and employability rate

Sources: OECD; Eurostat; WDI; and staff calculations.

Input: Public Education spending to GDP; Output: All five indicators

Input: Total Education Spending per Pupil, PPS; Output: PISA score, pupil teacher ratio and employability rate

Input: Total Spending on Tertiary Education per Pupil, PPS; Output: PISA score, tertiary education attainment and employability rate

DEA Efficiency Scores for Education Sector, by Regions
OutputInputEfficiency
Invalsi

Score
Tertiary

education

attainment,

25-64 yrs

(percent)
Enrollment

in sec and

post-sec

education,

15-24 yrs

(percent)
Employment

rate,

20-64 yrs

(percent)
Public spending

on education per

pupil as a share

of per capita

GDP
Input-

based
Output-

based
Nord-Ovest1.017.347.067.621.40.800.98
Nord-Est1.116.348.370.519.50.970.99
Centro1.018.947.264.919.60.940.99
Sud1.015.145.848.632.00.450.92
Isole1.013.644.847.333.50.410.89
Mean1.016.446.860.424.80.730.96
Sources: Eurostat; ISTAT; and staff calculations.
Sources: Eurostat; ISTAT; and staff calculations.

23. Significant regional variation in education efficiency is seen with the south being worse. Sibiano and Agasisti (2013) also find similar regional variations. While the northern regions are close to the frontier, the southern regions could achieve the same outcomes with less than half the resources currently being spent. Spending per student as a share of per capita income is high whilst outcome indicators such as test scores, enrollment rates and completion rates are weak (Table A2.2). The low efficiency scores are negatively correlated with socioeconomic conditions such as per capita GDP (PPS), and dependency on central government transfers (Table A2.4).

24. Performance based budgetary reforms in the education sector could improve outcomes. Italy stands out for its lack of outcome based policy in education, Gounard (2007). Italy’s challenge is particularly acute in higher education (OECD, 2011). Fixing funding rules, evaluation systems and institutional autonomy for staffing policy can help reduce cross regional variation in inefficiencies and strengthen effectiveness of spending allocations, which in turn can enhance economic productivity and competitiveness (St. Aubyn, et. Al 2009). Furthermore, adult literacy, numeracy and technology skills are also relatively weak in Italy (OECD, 2013). Consequently, beyond formal education, more emphasis is needed on continuing education to improve technological skills.

Social Protection and Active Labor Market Policies

Despite low spending, efficiency of social protection spending and labor market policies can be enhanced to improve outcomes. The efficiency indicator using social protection spending (excluding pensions) as input, and multiple outcome indicators such as income inequality and population at risk of poverty indicators shows that the same outcomes could be achieved with 30 percent less spending. Italy ranks relatively well in the euro area, likely reflecting the below average spending on non-pension social expenditures and the higher reliance on intra-family transfers for social assistance (Text figure and Table A1.2). Outcome indicators are however weak resulting in relatively low ranking in terms of output based efficiency indicator.

Composition of Social Protection Spending

(Percent of total expenditure)

Source: Eurostat.

DEA Efficiency Scores for Social Protection Spending(excluding Pensions)
InputOutput
Social

protection

spending,

2011
Population

at risk of

poverty afte

transfers

2011
Gini

coefficient

2011
Income

quintile share

2011
Efficiency,

input based 1/
RankEfficiency,

output

based 1/
Rank
Italy4.419.631.90.20.706/190.8917/19
Mean6.815.028.60.20.560.94
Sources: Eurostat; and staff calculations.

Input: Social protection spending, excl pensions; outputs: poverty risk (inverted), Gini coefficient (inverted) and share of income of lowest quintile to highest quintile.

Sources: Eurostat; and staff calculations.

Input: Social protection spending, excl pensions; outputs: poverty risk (inverted), Gini coefficient (inverted) and share of income of lowest quintile to highest quintile.

25. The regional variation in efficiency reflects the highly decentralized and fragmented social assistance system. Municipalities play an important role in financing and administration of the social assistance system with different institutions responsible for transfers to different vulnerable groups. Southern regions generally have weaker outcomes in terms of poverty risk and the share of long term unemployed. The average efficiency score across regions suggest that the same outcomes could be achieved with almost 25 percent less resources. Alternatively, the same resources could be used to improve outcomes by 12 percent. The regional efficiency indicators are strongly correlated with GDP per capita (PPS) (Table A2.4). Given the relatively low level of spending in this area, the scope for savings could be more limited. Indeed, it is estimated that bringing the low performing regions to national average levels would yield only around 0.14 percent of GDP in savings.

DEA Efficiency Scores for Social Protection Spending, by Regions
OutputInputEfficiency
At-risk of

poverty rate

(inverted),

2012
Share of Long-

term unemployed

(inverted), 2013
Social protection

spending as a

share of regional

GDP, 2011
Input-

based
Output-

based
Nord-Ovest88.30.21.30.91.0
Nord-Est88.40.41.30.81.0
Centro85.60.20.90.80.9
Sud70.50.11.00.70.8
Isole68.70.11.50.50.7
Mean80.80.21.10.80.9
Sources: Eurostat; ISTAT; and staff calculations.
Sources: Eurostat; ISTAT; and staff calculations.

26. The weaker outcomes could reflect the limited impact of social transfers on the poor. Studies have shown that the poverty reduction impact of taxes and transfers in Italy is weak (Caminada and Goudswaard, 2010, OECD, 2014). A disproportionately low share of social transfers accrues to the low-income working-age population. Tighter means testing of benefit eligibility could help improve outcomes. For example, implementing administrative efforts to limit fraud, such as the legal requirement to cross check beneficiaries with the tax register, can also help. This pattern of benefit incidence, however, may also reflect the lack of progressivity in pension transfers and tax expenditures.

Cash Social Transfers Received by Low and High Income Groups

(Percentage of Average Transfers, 2010)

Sources: OECD Income Distribution Database.

27. Given the high share of long-term unemployed, strengthening the coverage of active labor market policies (ALMP), can provide an important pathway for exiting unemployment. Efficiency indicators on ALMP using as input indicator, public spending on labor market policy (that include training, labor market services, employment incentives and direct job creation) and as output indicators, share of participants in the program among jobseekers and the inverse of share of long term unemployed of the total unemployed, show that Italy is average by euro area standards. This is mainly due to the relatively high share of long-term unemployed in Italy. However, when adjusted for GDP per capita, which is significant with the efficiency scores, it lags behind most euro area members (Table A1.4).

DEA Efficiency Scores for Active Labor Market Policies
InputOutputInput basedOutput based
Public Spending

on Labor Market

Policy Measures

(percent of GDP)

2003-11
Participation in

Labor Market

Policy Measures

(percent)

2003-11
Share of LT

Unemployed

2012
Efficiency,

corr. for PC

GDP PPS 1/
RankEfficiency,

corr. for PC

GDP PPS 1/
Rank
Italy0.4331.10.530.4022/280.5125/28
Mean0.4326.70.420.560.73
Sources: Eurostat; and staff calculations.

Input: Spending on LMP Measures; output: Participation in LMP measures, inverse share of LTU

Sources: Eurostat; and staff calculations.

Input: Spending on LMP Measures; output: Participation in LMP measures, inverse share of LTU

28. Reforms to improve central coordination for collecting performance information and evaluation should be implemented to better inform resource allocation. Indicators on job placement and employability of participants over the longer run for participants in the ALMP were not available for Italy. Reforms to improve central coordination which can help in the resource allocation process were approved in 2012, but implementation is still lagging. The majority of unemployed are ineligible for unemployment benefits and services for job search, resulting in a weak link between work activation and benefits. As reforms to expand unemployment benefits get underway, conditioning continued unemployment benefit receipts to participation in job search assistance and training could help disincentivize passive unemployment. Combining activation policies with public works may also be self-financing as it reduces unemployment benefits. These reforms are key priorities under the Renzi government’s reform agenda.

Regional Spending Efficiency and Transfers to Sub national Governments

29. An important policy question is whether spending efficiency is affected by dependency on central government transfers. Regional efficiency scores are negatively correlated to transfer dependency and external borrowing ratios (Table A2.4). One possible explanation is that higher transfers from the central government can weaken spending discipline with negative consequences on spending efficiency. There is little incentive to reduce spending as it would imply possible risk of losing transfers. Indeed, overall health and education spending are positively correlated with transfers (while this is not the case with social protection spending).

30. This positive relation with transfers may also reflect the regional income level (Table A2.4). This would be the case if transfers are distributed primarily for equalization goals. Econometric tests using cross-sectional data show that socioeconomic conditions such as per capita income are indeed strongly correlated with efficiency indicators (Table A2.5). Furthermore, while fiscal variables, such as transfer dependency and external borrowing to GDP, are in general negatively correlated with health and education efficiency scores, they lose significance when controlling for per capita income (Table A2.5). Other demographic and institutional variables are also used as control variables. While population density appears to be associated with higher efficiency scores, other variables are not significant.

31. Regional efficiency scores show a stronger negative correlation with initial transfer dependency. While fiscal variables and spending may change annually, policy outcomes likely change only gradually over an extended period. Therefore, to assess whether efficiency scores are affected by lagged transfers rather than contemporaneous ones, efficiency scores are calculated using averages of input and output variables over two time horizons: the first over 2002–08 and the second over 2009–11. This breakdown marks the period when legislative changes to the criteria for transfers were introduced in 2009. Furthermore, large cuts in transfers were initiated as a crisis related measure. As above, regressions show a negative correlation with transfer dependency. The correlations with the initial (first period) transfer dependency are significant especially for health and education and are robust across different specifications. Regressions of the change in efficiency scores between the two periods on initial transfers also suggest that regions with lower transfer dependency have higher improvement in efficiency scores, although the significance level drops.

32. To enhance the performance impact of transfers, legislative decrees to implement the 2009 Framework Law 42 need to be approved. Several important decrees have already been approved to enhance autonomy and accountability, such as the definition of standard cost for the health care sector. Efforts are ongoing on the definition of expenditure needs and funding for municipalities; the funding and tax system for ordinary regions; and the harmonization of sub-central accounting principles. The law envisages setting of standard costs (fabbissogno standard) to meet minimum essential services, which determine the formula for an equalization fund. Indicators to determine these standard costs need to be finalized and approved by Parliament (OECD, 2012). At the same time, strict discipline is needed to ensure against ex-post bail outs.

C. Conclusions and Policy Recommendations

33. Budget allocation in Italy will need to increasingly rely on an efficiency analysis to find savings and improve performance. In a context where spending has already been tightened significantly, an in-depth analysis on the efficiency of spending is crucial to reduce spending without adversely impacting outcomes. Furthermore, it can also help inform how resources can be reallocated or outcomes improved within the existing resource envelope.

34. Achieving sizable expenditure savings will require addressing the large social spending, in particular, current pensions. Public pension and health care have been the main sources of spending pressures over the past decade. Intergenerational imbalances in Italian social spending are among the highest in the OECD. Pensions was the only area, other than interest payments, where spending increased in the post-crisis period. Although reforms have been implemented in these areas, there is scope to do more to reorient the budget in a growth-friendly manner from the current high pension benefits towards investments, education and active labor market policies for younger generations.

35. To reduce cross-regional variation in spending efficiency, greater use of performance targets in resource allocation decision-making is needed.

  • In the education sector, the room for savings may be limited, but there exists scope for improving educational outcomes. More reliance on outcome targets is needed in policy-making and reallocation of resources.

  • In the health sector, the analysis shows significant room for savings, particularly in the areas of goods and services spending where growth has been the highest. In 2011, improving efficiency in the low-performing regions to the national average could potentially save as much as 1 percent of GDP. Efficiency scores have likely improved with the reforms such as standard costs recently introduced. Nevertheless, given the regional variation in health spending efficiency, scope to improve performance in less efficient regions remains.

  • While non-pension social protection spending is relatively low, there is room to improve outcomes. Better central coordination in collecting performance information and evaluation is needed to improve effectiveness of active labor market programs.

  • Regional analysis also shows that spending efficiency is generally weaker in the South that is poorer and more reliant on outside transfers. Improving allocation of regional transfers to sub national governments using standard costs based on more efficient benchmarks could provide an important avenue for savings. Early parliamentary approval of the legislation determining standard costs will be important.

Annex 1. Selected European Countries: DEA Efficiency Scores
Table A1.1.DEA Efficiency Scores for Health Sector
Public Health

Exp/GDP

2011, or latest
HALE Female

2011
HALE Male

2011
Mortality rate,

cancer

2010
Mortality rate, heart

disease

2010
Efficiency (input-

based)
Rank
Euro area6.8
United Kingdom6.865.265.2347.8681.20.76
Austria7.660.359.8391.4512.80.71
Belgium6.963.663.4370.21091.70.75
Denmark7.259.463.6318.1
France7.863.662.7388.21739.11.00
Germany6.758.757.9387.1621.50.80
Italy6.962.763.4383.3907.40.7614
Luxembourg5.267.165.8377.91161.41.00
Netherlands5.75964330.51290.31.00
Norway7.27069.9371.1803.91.00
Sweden6.770.271.1416.7622.71.00
Switzerland64.766.3435.4856.2
Finland6.958.357.7434.6410.71.00
Greece5.966.966.2404.7950.60.97
Iceland8.167.769.1377.4
Ireland6.568.366.1349.7554.30.85
Malta6.370.770.3411.7467.31.00
Portugal7.258.660.7405.21314.11.00
Spain5.765.865.4408.81213.61.00
Cyprus6161.6507.4876.4
Sources: Eurostat; OECD and staff calculations.
Sources: Eurostat; OECD and staff calculations.
Table A1.2.DEA Efficiency Scores for Social Protection Spending(excluding Pensions)
Social

protection

spending,

2011
Population

at risk of

poverty after

transfers

2011
Gini

coefficient

2011
Income

quintile share

2011
Efficiency,

input based 1/
RankEfficiency,

output

based 1/
Rank
United Kingdom 5.8516.2330.190.520.92
Austria6.3212.626.30.260.480.96
Belgium7.1515.326.30.260.430.96
Denmark10.9313.027.80.230.280.96
France9.2914.030.80.220.330.95
Germany (until 1997.1215.8290.220.430.93
Italy4.3619.631.90.180.7060.8917
Luxembourg9.0013.627.20.250.340.95
Netherlands4.1811.025.80.260.730.98
Norway8.9910.522.90.301.001.00
Sweden9.3614.024.40.280.330.98
Finland11.3213.725.80.270.270.96
Greece5.4721.433.50.170.560.87
Iceland3.059.223.60.301.001.00
Ireland9.7915.229.80.220.310.93
Malta4.9515.427.40.240.620.95
Portugal3.3118.034.20.180.920.90
Spain5.7522.234.50.140.530.86
Cyprus3.4114.829.20.230.890.94
Sources: Eurostat; and staff calculations.

Input: Social protection spending, excl pensions; outputs: poverty risk (inverted), Gini coefficient (inverted) and share of income of lowest quintile to highest quintile.

Sources: Eurostat; and staff calculations.

Input: Social protection spending, excl pensions; outputs: poverty risk (inverted), Gini coefficient (inverted) and share of income of lowest quintile to highest quintile.

Table A1.3.DEA Efficiency Scores for Education Sector
Tertiary
Public expenditure onUpper sec eduPublicTotal PerEducation
Total educationtertiary education perTertiary education(percent ofpupilSpendingPupilSpending per
Public Educationspending perpupil (in percent ofattainment (percent ofPopulation, 25-teacherEmployabilityPISAEfficiencyEfficiencyPupil Efficiency
Expenditure/GDPpupil in PPSGDP per capita)population, 30-35 yrs)64 yrs)ratioRateReadingscore 1/score 2/score 3/
20112010201020112011201120092012
Euro area5.16900.133.233.769.3
United Kingdom6.48334.625.745.876.417.774.8499.31.001.001.00
Austria5.62639.739.123.882.510.284.0489.60.890.880.96
Belgium6.28036.535.642.671.310.579.4508.60.800.960.96
Denmark7.89604.856.941.276.911.481.3496.10.630.810.60
France6.07337.438.443.371.614.470.2505.50.800.990.72
Germany4.47737.930.786.315.780.2507.71.001.00
Italy4.26097.325.820.356.012.059.2489.80.991.001.00
Luxembourg5.2:48.277.39.782.0487.81.00
Netherlands5.88522.842.741.172.316.387.9511.21.001.001.00
Norway5.610403.942.748.881.310.187.5503.91.000.821.00
Sweden6.88311.741.846.881.611.875.9483.30.830.870.70
Switzerland6.1:41.643.885.687.0509.01.00
Finland6.4737938.546.083.713.374.8524.01.001.001.00
Greece4.1:28.964.58.667.3477.21.00
Iceland8.17226.728.744.670.710.775.9482.50.611.000.94
Ireland5.1:33.249.773.615.170.2523.21.001.00
Malta5.87645.460.521.438.19.378.0
Portugal6.3:31.426.135.09.178.6487.80.680.98
Spain4.86865.228.540.653.811.459.8487.90.970.890.90
Cyprus7.29144.638.346.275.211.481.2
Sources: OECD; Eurostat; WDI; and staff calculations.

Input: Public Education spending to GDP; Output: All five indicators

Input: Total Education Spending per Pupil, PPS; Output: PISA score, pupil teacher ratio and employability rate

Input: Total Spending on Tertiary Education per Pupil, PPS; Output: PISA score, tertiary education attainment and employability rate

Sources: OECD; Eurostat; WDI; and staff calculations.

Input: Public Education spending to GDP; Output: All five indicators

Input: Total Education Spending per Pupil, PPS; Output: PISA score, pupil teacher ratio and employability rate

Input: Total Spending on Tertiary Education per Pupil, PPS; Output: PISA score, tertiary education attainment and employability rate

Table A1.4.DEA Efficiency Scores for Active Labor Policies
Input basedOutput based
Public Spending

on Labor Market

Policy Measures

(percent of GDP)

2003-11
Participation in

Labor Market

Policy Measures

(percent)

2003-11
Share of LT

Unemployed

2012
Efficiency,

corr. for PC

GDP PPS 1/
RankEfficiency,

corr. for PC

GDP PPS 1/
Rank
Belgium1.0683.00.450.920.93
Bulgaria0.2811.50.550.510.70
Czech Republic0.1410.20.430.610.81
Denmark1.3051.50.280.170.80
Germany0.6327.50.450.150.44
Estonia0.082.20.540.770.73
Ireland0.5730.50.620.170.34
Greece0.1510.40.590.570.65
Spain0.6369.80.440.600.91
France0.7246.60.410.310.66
Italy0.4331.10.530.40220.5125
Cyprus0.1612.00.300.981.02
Latvia0.204.40.520.420.68
Lithuania0.187.30.490.430.71
Luxembourg0.3976.50.310.280.35
Hungary0.3217.40.450.480.71
Malta0.056.70.471.071.07
Netherlands0.8145.90.340.190.65
Austria0.5328.00.260.460.68
Poland0.4418.50.410.410.68
Portugal0.4929.70.480.450.62
Romania0.075.30.460.940.96
Slovenia0.2116.30.480.550.70
Slovakia0.1622.80.670.920.89
Finland0.7726.00.210.510.86
Sweden0.8328.40.190.890.90
United Kingdom0.052.20.340.990.99
Norway0.5227.10.190.580.63
Mean0.4326.70.420.560.73
Sources: Eurostat; and staff calculations.

Input: Spending on LMP Measures; output: Participation in LMP measures, inverse share of LTU

Sources: Eurostat; and staff calculations.

Input: Spending on LMP Measures; output: Participation in LMP measures, inverse share of LTU

Annex 2. Italian Regions: DEA Efficiency Scores
Table A2.1.DEA Efficiency Scores for Health Sector, by Region
OutputInputEfficiency
Life

expectancy

at birth,

2012
Available beds

(per 1000

inhabitants),

2011
Physicians

(per 1000

inhabitants),

2011
Public health

spending as a

share of

regional GDP,

2011
Input-

based
Output-

based
Input-

based
Output-

based
Nord-OvestPiemonte82.43.73.76.60.830.990.941.00
Valle d’Aosta/Vallée d’Aoste82.73.83.66.91.001.00
Liguria82.33.34.67.50.911.00
Lombardia82.83.83.75.21.001.00
Nord-EstProvincia Autonoma di Bolzano/Bozen83.33.93.15.81.001.000.951.00
Provincia Autonoma di Trento83.94.03.36.41.001.00
Veneto83.03.53.45.60.971.00
Friuli-Venezia Giulia82.53.63.96.70.790.99
Emilia-Romagna82.94.14.35.51.001.00
CentroToscana82.93.14.46.51.001.001.001.00
Umbria82.93.04.47.61.001.00
Marche83.43.53.77.01.001.00
Lazio82.03.64.96.61.001.00
SudAbruzzo82.63.24.28.10.670.990.530.99
Molise82.73.84.19.80.551.00
Campania80.82.84.111.10.480.97
Puglia82.73.33.810.70.500.99
Basilicata82.63.13.510.50.510.99
Calabria82.03.04.211.20.490.99
IsoleSicilia81.42.84.710.80.600.990.800.99
Sardegna82.33.65.010.21.001.00
Mean82.63.54.07.90.820.99
Std. Dev.0.70.40.52.10.220.01
Table A2.2.DEA Efficiency Scores for Social Protection Spending, by Region
OutputInputEfficiency
At-risk of

poverty rate

(inverted),

2012
Share of Long-

term unemployed

(inverted), 2013
Social protection

spending as a share

of regional GDP,

2011
Input-

based
Output-

based
Input-

based
Output-

based
Nord-OvestPiemonte86.40.170.960.730.940.870.96
Valle d’Aosta/Vallée d’Aoste92.10.342.681.001.00
Liguria83.20.200.950.740.91
Lombardia91.50.240.701.001.00
Nord-EstProvincia Autonoma di Bolzano/Bozen88.50.831.431.001.000.780.97
Provincia Autonoma di Trento86.50.452.130.450.95
Veneto89.00.261.030.700.97
Friuli-Venezia Giulia86.80.290.840.910.95
Emilia-Romagna91.20.260.850.861.00
CentroToscana87.70.240.830.850.960.800.93
Umbria86.80.200.880.800.95
Marche86.00.180.930.760.94
Lazio81.70.140.900.780.89
SudAbruzzo78.40.151.090.640.860.730.77
Molise73.00.110.920.760.80
Campania63.60.070.920.760.70
Puglia70.90.090.950.740.77
Basilicata67.40.100.900.780.74
Calabria69.60.070.970.720.76
IsoleSicilia57.70.071.410.500.630.470.75
Sardegna79.60.101.590.440.87
Mean80.80.221.140.760.88
Std. Dev.10.00.20.50.160.1
Table A2.3.DEA Efficiency Scores for Education, by Region
OutputInputEfficiency
Invalsi

Score
Tertiary

education

attainment,

25-64 yrs

(percent)
Enrollment

in sec and

post-sec

education,

15-24 yrs

(percent)
Employment

rate, 20-64 yrs

(percent)
Public spending

on education per

pupil as a share

of per capita GDP
Input-

based
Output-

based
Input-

based
Output-

based
Nord-OvestPiemonte1.0615.947.866.518.80.780.970.800.98
Valle d’Aosta/Vallee d’Aoste1.0616.147.669.833.70.430.97
Liguria1.0119.947.264.819.51.001.00
Lombardia1.0417.245.469.313.81.001.00
Nord-EstProvincia Autonoma di Bolzano/Bozen14.845.976.626.51.001.000.970.99
Provincia Autonoma di Trento16.650.270.523.51.001.00
Veneto1.0415.048.067.815.10.980.98
Friuli-Venezia Giulia1.1117.248.667.017.60.850.99
Emilia-Romagna1.0517.848.970.615.01.001.00
CentroToscana1.0417.047.768.018.80.780.970.940.99
Umbria1.0220.045.765.223.11.001.00
Marche1.0518.149.365.320.41.001.00
Lazio1.0120.546.061.216.31.001.00
SudAbruzzo1.0317.344.958.822.00.630.920.450.92
Molise0.9816.646.151.029.70.470.93
Campania0.9614.345.143.432.10.430.90
Puglia0.9513.246.545.931.90.450.93
Basilicata0.9415.147.749.938.70.380.95
Calabria0.9313.944.242.337.70.370.88
IsoleSicilia0.9513.343.942.835.20.390.880.410.89
Sardegna0.9513.845.651.731.80.440.91
Mean1.0116.446.860.424.80.730.96
Std. Dev.0.052.21.710.78.10.270.04
Table A2.4.Cross Correlation between Regional Efficiency and Exogenous Factors
Correlation

t-Statistic
ProbabilityEDUEFF1 1EDUEFF2HEALTHEFF1HEALTHEFF2SOCEFF1SOCEFF2GDPPCPPSPOPDENSITYPOPLNELDERLYUNEMPLTAXDEPJUDICIALPOVRISKEXTBORR
EDUEFF11
-----
-----
EDUEFF20.89 ***1.00
8.35-----
HEALTHEFF10.76 ***0.77 ***1.00
5.125.31-----
HEALTHEFF20.60 ***0.68 ***0.82 ***1.00
3.244.096.21-----
SOCEFF10.260.47 **0.200.161.00
1.182.330.870.69-----
SOCEFF20.76 ***0.85 ***0.87 ***0.77 ***0.47 *1.00
5.136.947.595.332.33-----
GDPPCPPS0.76 ***0.84 ***0.81 ***0.67 ***0.53 *0.90 ***1.00
5.026.686.043.932.749.17-----
POPDENSITY0.260.07−0.02−0.310.15−0.080.021.00
1.180.30−0.08−1.440.66−0.340.10-----
POPLN0.16−0.030.03−0.170.12−0.050.060.85 ***1.00
0.69−0.120.14−0.730.54−0.200.267.07-----
ELDERLY0.43 *0.49 *0.37 *0.41 *0.160.46 *0.27−0.09−0.221.00
2.062.461.751.960.732.231.24−0.39−0.99-----
UNEMPL−0.79 ***−0.88 ***−0.79 ***−0.70 ***−0.44 *−0.91 ***−0.93 ***0.120.13−0.45 *1.00
−5.65−8.22−5.55−4.23−2.13−9.57−10.740.510.55−2.18-----
TAXDEP−0.39 *−0.52 *−0.64 ***−0.65 ***−0.11−0.65 ***−0.58 ***0.41 *0.29−0.49 *0.62 ***1.00
−1.87−2.65−3.63−3.76−0.48−3.73−3.081.951.32−2.463.40-----
JUDICIAL0.50 *0.52 *0.48 *0.240.310.64 ***0.73 ***0.030.090.19−0.67 ***−0.42 *1.00
2.512.632.351.091.443.604.650.120.370.84−3.95−2.00-----
POVRISK−0.75 ***−0.84 ***−0.87 ***−0.78 ***−0.46 *−1.00 ***−0.89 ***0.070.03−0.48 *0.90 ***0.66 ***−0.62 ***1.00
−4.94−6.75−7.61−5.37−2.26−64.48−8.370.290.14−2.398.833.82−3.46-----
EXTBORR−0.44 *−0.60 ***−0.58 ***−0.85 ***−0.13−0.64 ***−0.53 *0.39 *0.28−0.45 *0.66 ***0.54 *−0.140.64 ***1.00
−2.14−3.29−3.09−7.16−0.59−3.63−2.721.851.25−2.203.862.81−0.603.63-----
Table A2.5.Multivariate Cross-section Regression on Regional Efficiency Scores
Censored Normal Tobit Dependent Variable:Education EfficiencySocial Protection EfficiencyHealth Efficiency
C0.74 (***)0.00.74 (***)0.130.99 (***)0.20
5.43−1.19.80.69.00.5
GDPPCPPS0.0 (**)2.4E-053.0E-05
−1.70.00.0
POPDENSITY0.0022 (**)0.0020 (**)0.0005−0.00020.0012 (**)0.0003
2.52.41.10.71.90.7
TAXDEP−3.1 (***)−1.0−0.30.7−2.1 (***)−0.7
−3.02.4−0.60.2−2.80.4
EXTBORR−3.4 (*)−1.00.8−3.7 (***)−1.2
2.6−0.70.5−2.10.4
No. of obs212121212121
S.E.0.2355010.162510.17030.14920.1480.121
Table A2.6.Multivariate Panel Regression on Regional Efficiency Scores, Levels
Dependent Variable: Levels Censored TobitEducation EfficiencySocial Protection EfficiencyHealth Efficiency
coeff.t-statcoeff.t-statcoeff.t-stat
Lagged transfer dependency−1.26−2.41 (**)−0.50−0.98−1.48−2.28 (**)
GDP per Capita PPS0.00−1.620.00−0.910.00−1.55
Population density0.000.000.000.200.001.05
Judicial−0.15−0.62−0.03−0.14−0.05−0.16
Constant1.725.46 (***)1.113.85 (***)1.834.24 (***)
S.E. of regression0.200.230.19
No. of obs191919
Table A2.7.Multivariate Panel Regression on Regional Efficiency Scores, Change
Dependent Variable: Change Pooled OLSEducation EfficiencySocial Protection EfficiencyHealth Efficiency
coeff.t-statcoeff.t-statcoeff.t-stat
Lagged transfer dependency−1.26−1.61−1.35−2.44 (**)−0.86−1.44
GDP per Capita PPS0.00−0.580.00−1.620.00−1.26
Population density0.00−0.600.001.91 (*)0.002.01 (*)
Population0.00−1.69 (*)0.00−1.99 (*)0.00−0.77
Elderly−0.02−0.60−0.05−1.82 (*)−0.01−0.41
Judicial−0.31−0.890.100.400.100.36
Constant1.131.441.452.62 (**)0.520.88
R-squared0.390.490.45
No. of obs191919
References

    Afonsoet. al.2005. “Public Sector Efficiency: An International ComparisonPublic Choice12332147.

    BenhocineN.2013Assessing Efficiency of Public Spending on Health and Education in Iceland” in Selected Issues Iceland IMF Country Report No. 13 257.

    Bossaertet. al2012The Impact of Demographic Change and its Challenges for the Workforce in the European Public SectorsEuropean Institute of Public Administration Working Paper 2012/01.

    Caminada and Goudswaard2010How Well is Social Expenditure Targeted to the Poor?” in P.Saunder and R.Sainsbury(eds)Social Security Poverty and Social Exclusion in Rich and Poorer Countries International Studies on Social Security volume 16Morstel, Belgium: Intersentia pp. 97112.

    European Commission2012The Quality of Public Spending in the EUEuropean Economy Occasional Papers 125December2012.

    FranceseM. and M.Romanelli2011Healthcare in Italy: Expenditure Determinants and Regional Differentials” (October142011). Bank of Italy Temi di Discussione (Working Paper) No. 828.

    GiordanoR. and P.Tommasino2011Public Sector Efficiency and Political CultureBanca D’Italia Working Paper 786.

    GonandF.I.Goumard and R. W.Price2007Public Spending Efficiency: Institutional Indicators in Primary and Secondary EducationOECD Economics Department Working Papers No. 543OECD Publishing.

    HaunerD.2007Benchmarking the Efficiency of Public Expenditure in the Russian FederationIMF Working Paper 07/246.

    European Observatory of Health Systems and PoliciesItaly Health System Review, 2009Health in Transition2009.

    OECD2010Reforming Fiscal Federalism and Local Government Beyond the Zero Sum GameOrganization of Economic Cooperation and DevelopmentParis.

    OECD2012Human Resource Management Country Profiles. Organization of Economic Cooperation and DevelopmentParis.

    OECD2013Health at a Glance. Organization of Economic Cooperation and DevelopmentParis.

    OECD2014Demographics Public AdministrationOrganization of Economic Cooperation and DevelopmentParis.

    OECD2013Skilled for Life? Key Findings from the Survey of Adult SkillsOrganization of Economic Cooperation and DevelopmentParis.

    OECD2013Education at a GlanceOrganization of Economic Cooperation and DevelopmentParis.

    OECD2013Pensions at a GlanceOrganization of Economic Cooperation and DevelopmentParis.

    Piperno2012Implementing Fiscal Decentralization in Italy between Crisis and Austerity, Challenges AheadPerspectives on Federalism Vol. 4 issue 32012.

    SibianoP. and T.Agasisti (2013) Efficiency and Heterogeneity of Public Spending in Education among Italian RegionsJournal of Public Affairs Volume 13Number1 pp 1222 (2013)

    St. AubynM.2007Modernizing Public Administration and Economic GrowthGPEARI Ministério das Finanças e da Administração Pública Working Paper 001.

    St. AubynM.et. al. 2007Study on the Efficiency and Effectiveness of Public Spending on Tertiary EducationEconomic Papers 390| November 2009European Commission.

    VanhuysseP.2012”Does Population Aging Drive Up Pro-Elderly Social Spending?European Social Observatory (OSE) Research Paper Series No. 7February2012.

Prepared by Anita Tuladhar (EUR).

Other Resources Citing This Publication