Budget Analysis for Expenditure Rationalization

Budget Analysis for Expenditure Rationalization Public Finance Analysis and Management CourseWorld Bank, April 23-27, 2007

Marijn VerhoevenExpenditure Policy DivisionFiscal Affairs Department, IMF

Overview• Why expenditure rationalization?• The analytical tool box for

expenditure rationalization:– A word about data– Measuring efficiency

• We found the problems—now what? (very briefly!)

Why expenditure rationalization?• To achieve macroeconomic

stability and fiscal sustainability• To create fiscal space• To increase allocative efficiency by

cutting back or reforming government activities

• To enhance X-efficiency by achieving the same outputs with less inputs

Figure 5: Fiscal SpaceIncrease of

Grant Aid in % GDP

New Borrowing in

% GDP

Improved Revenue

Effort in % GDP

Improved Expend.

Efficiencyin % GDP

0

1

2

3

4

5

Why expenditure rationalization? (cont’d)

A word about data• Expenditure analysis is typically

data driven• But data are problematic• There are several competing

sources of spending data, each with their strengths and weaknesses

• Let us look at the example of data on wage spending

Facts and figures:measuring wage spending• The wage bill is measured:

– As a share of GDP and total spending to compare across countries

– As a share of domestic revenue to assess sustainability

– At the sectoral level, compare to nonwage spending to assess efficiency

• Source is IMF Government Finance Statistics or national data

Country Group Sample SizeCentral government wages and

salaries in percent of GDP

Central government wages and salaries in percent of central

government expenditure

Africa 11 8.4 28.3Francophone Africa 6 6.3 27.7Non-Francophone Africa 5 10.9 29.0

Asia 10 5.3 20.0South Asia 3 4.6 15.1

Europe and Central Asia 21 3.9 12.6Central and Eastern Europe 12 5.1 14.4

Latin America & the Caribbean 16 5.6 25.0Caribbean countries 3 8.6 31.1

Middle East and North Africa 6 9.1 30.4

European Union 15 5.4 13.3

Low-Income Countries 19 5.7 22.6Middle-Income Countries 42 6.0 22.1High-Income Countries 30 5.9 15.6

Memorandum items:PRGF-Supported Programs 1/ 28 4.8 19.9OECD General Government 2/ 21 11.4

Sources: Government Financial Statistics database (IMF); International Financial Statistics database (IMF); World Economic Outlook database (IMF); and IMF Staff.1/ Data refers to 2001 outturn.

Table 1. Central Government Wages and Salaries, 1990-2001

2/ OECD, 2000, "Summary of the Public Sector Pay and Employment Data Analysis and Future Direction for HRM Data Collection." Data refers to 1997 total compensation cost, which includes the wage bill and employer contributions to social insurance.

Lies and statistics: mismeasuring wage spending?

• Not all compensation may be captured in wages and salaries: transfers (pension benefits and subventions for education), other goods and services (in-kind benefits and contractual workers), and capital spending (donor-financed projects) may hide substantial wage spending. In Nicaragua, out of actual wage spending of 8.6 percent in 2005, less than 4 percent is recorded as wages and salaries.

• When government is decentralized, central government wage spending is biased downward. But reliable data for general government are rare for low-income countries.

Efficiency: the issue

Source: World Bank (2004) World Development Report 2004.Spending refers to total annual public spending per child of primary school age, in 1995 US dollars.

How should we think about the efficiency of public spending?

• What is the mix of public programs that best meets government objectives?– Where to invest the marginal dollar across

sectors• For example, can education goals be reached by

investing the marginal dollar in other sectors?– Where to invest the marginal dollar within

sectors• Primary versus secondary education• Primary health care versus secondary health care

How should we think about the efficiency of public spending? (cont’d)• Given allocative decisions, is output

maximized with given inputs?– Common problems:

• Inappropriate student/teacher ratios• Shortage of medicine or nurses relative to

doctors• Shortage of textbooks• Waste, leakage of funds• Labor and utility costs crowding out

maintenance and capital spending

Assessing efficiencyMany roads lead to Rome:• “Basic” benchmarking• PETS• Randomized evaluations• Absenteeism studies• Sectoral efficiency analysis

Choice depends on data availability,objectives, and priors

Assessing efficiency always begins with...Review the basics of public spending• Functional classification

– Primary, secondary, tertiary education• Inputs, programs, types of intervention

– Education: Teachers, textbooks– Health: Spraying, information and education

campaign, etc.• Economic classification

– Wage versus non-wage– Recurrent versus capital (investment) spending

• Central and local government budgets, other Ministries

• Planned versus actual, nominal versus real• Share of private, NGO, and donor spending

“Basic” benchmarking• Selected useful descriptive statistics

– Budget data– Unit costs– Ratios of teachers, students, non-teaching staff– Distribution of teachers among levels of

qualifications; percentage meeting basic government standards

– Actual maintenance budget versus engineering estimates for routine maintenance

– Enrollment rates, repetition rates, dropout rates– Absenteeism, informal payments, etc.

“Basic” benchmarking (cont’d)• Comparisons

– Sub-national units, clinics, schools– Private versus public schools– Private versus public health facilities– Comparator countries

• Cross-country information on real resources and output– UNESCO education indicators– Program for International Student Assessment

(PISA) – WHO Indicators of Health System Attainment– Trends in International Mathematics and Science

Study (TIMSS) – Progress in International Reading Literacy Study

(PIRLS) and the International Survey of Adults (ISA)

Randomized evaluationsRandomized evaluations of educational reformprograms:

• Random selection of schools for the reform– Colombian voucher program, Angrist and others

(2002)

• Randomized phase-in of programs– Argentina: Decentralization took place across all

provinces, but at different periods and intensities, Galliani and Schargrodsky (2002)

Public expenditure tracking surveysTrace the flow of resources through thebureaucracy from the central governmentdown to the service facility:• Comparing originally allocated funds with funds

that actually arrive at the facility• Amount of time required for fund to arrive• Reinikka and Svensson (2001): Uganda in the

1990s, significant leakage existed

Sectoral efficiency analysis: basic concepts• The measurement of efficiency

generally requires the following: – (i) information on inputs and associated

costs;– (ii) an estimation of output or benefit; and – (iii) a comparison of (i) and (ii)

• Key question:– Could the same level of output be achieved

with less input?– Equivalently, could more output be

generated with the same level of input?

Sectoral efficiency analysis: basic concepts (cont’d)

Health Expenditure

Public health expenditure Private health expenditure

Real Health Resources (examples)

Hospital beds Physicians/health workers Immunizations Doctors’ consultations In-patient admissions Lengths of stay Bed occupancy rate

Health Outcomes

Health adjusted life expectancy

Standardized death rate Infant mortality rate Child mortality rate Maternal mortality rate Incidence of tuberculosis

Cost effectiveness

System efficiency

Overall efficiency

Sectoral efficiency analysis: Best-practice frontier

0

0.5

1

1.5

2

2.5

3

0 0.2 0.4 0.6 0.8 1 1.2 1.4Input item

Prod

uct i

tem

AD

FC

B

E

Sectoral efficiency analysis: measuring efficiency• Basic idea: measuring distance from the best-practice

frontier• Regression analysis

– Corrected ordinary least squares (COLS)• Evans et al (2000), WHO (2000): Efficiency of national health systems

– Alternative: Greene (2004): Stochastic frontier analysis

• Nonparametric analysis:– Free disposal hull analysis (FDH)

• Gupta and Verhoeven (2004) (Chapter 11): Efficiency of health and education spending in 85 countries, 1984-95

– Data envelopment analysis (DEA)• Herrera and Pang (2005): Efficiency of health and education spending

in 140 countries, 1996-2002• Affonso and St. Aubyn (2004): Efficiency of health and education

spending in OECD countries

Sectoral efficiency analysis: problems• Lack of insight in nature of relationship

between inputs and outputs:– How to measure inputs and outputs?– Lags?– Impact of environmental factors?– Externalities across sectors?

• Parametric approaches are very data intensive and require more assumptions about the relationship

• Nonparametric approaches are less robust (e.g., small sample bias) and sensitive to outliers

• The analysis is only as good as the data—and data are weak (e.g., on quality and policy objectives)

Examples of FAD sectoral efficiency analysis• FAD research:

– The efficiency of education, health, and social assistance spending in EU New Member States

– The efficiency of education and health spending in the G7

– The efficiency of government investment in Latin America

• Focus here on the efficiency of health spending in the Slovak Republic

Output-Oriented Efficiency Relative to the OECD (distribution by quartiles of the ranking of OECD bias-corrected output-oriented

efficiency scores) 1/

Percentile 1-25 26-50 51-75 76-100 Public expenditures Bulgaria Czech Republic Estonia Hungary Latvia Poland Lithuania Slovak Republic Romania Slovenia Public and private expenditures Bulgaria Estonia Lithuania Hungary Czech Republic Romania Slovenia Latvia Slovak Republic Poland

Source: IMF staff calculations 1/ Slovak Republic’s output-oriented efficiency scores for public expenditures ranked, on average, at the 54th percentile of the overall ranking of efficiency scores in the sample of OECD countries. This places Slovak Republic in the third (51-75) quartile of the OECD ranking distribution. The rankings are based on each country’s average of the individual point estimates of the bias-corrected output-oriented efficiency scores for various outcome indicators, including infant, child, and maternal mortality, the incidence of tuberculosis and HALE (see Appendix II).

Rank of Health Efficiency Scores Relative to the OECD 1/

System Efficiency 2/ Overall Efficiency 3/

Intermediate Resources to

outcomes Public expenditures

to outcomes

Public and private expenditures to

outcomes Slovak Republic 1.7 1.1 0.4 Bulgaria 2.0 0.5 0.5 Czech Republic 1.4 0.7 0.5 Estonia 1.9 1.4 0.7 Hungary 1.9 1.5 1.4 Latvia 2.2 1.0 1.5 Lithuania 2.0 1.6 1.1 Poland 1.6 1.0 0.5 Romania 2.0 1.5 0.6 Slovenia 0.7 1.1 1.0

NMS-10 average 1.7 1.1 0.8 EU-15 average 0.9 1.0 1.1

Source: IMF staff calculations 1/ Ratio of output-oriented efficiency rankings of NMS-10 and EU-15 countries and the average ranking in the sample of OECD countries. The ratio is 1 if the country is as efficient as the average for the OECD, and is higher if the country is less efficient (see Appendix II). 2/ Based on output-oriented efficiency rankings using as inputs the average ranking of various real resources (Table 3) and as output various outcome indicators, including infant, child, and maternal mortality, the incidence of tuberculosis and HALE. 3/ Reflecting the output-oriented efficiency rankings of Table 6.

SVK health efficiency: sources of inefficiencies• Low co-payments• Unproductive spending on

administration and collective care• High spending on pharmaceuticals• High doctors’ consultations, outpatient

contacts, and inpatient hospital care

Key challenge is changing the mix ofreal resources!

SVK health efficiency:recommendations• Restrain pharmaceutical spending

– introducing a national procurement system– introducing incentives for generics – improving the pharmaceutical pricing and

reimbursement policy of the Ministry of Health• Reduce the reliance on hospitals and

contain the cost of hospital care– Eliminate excess hospital beds– Impose hard budget constraint on public hospitals – Restart hospital privatization

• Reintroduce co-payments for doctors’ visits and hospital care

SVK health efficiency:recommendations (cont’d)• Enhance incentives for competition

and more cost-effective administrative arrangements– Introduce incentives for practitioners to be

cost-effective– Define a stricter basic health care package,

thereby allowing some variations in basic insurance premiums

– Increase the power of the Antitrust authority and enhance the autonomy and independence of the Health Care Supervisory Board

– Refrain from introducing new limitations on profits of private insurance companies

Thank You!