Measuring Public Expenditure Efficiency. Yong Yoon Faculty of Economics Chulalongkorn University

Transcription

1 Measuring Public Expenditure Efficiency Yong Yoon Faculty of Economics Chulalongkorn University

2 Introduction Governments of developing countries typically spend resources equivalent to between 15 and 35 percent of GDP. We will quickly look at health and education expenditure (since they typically absorb the largest share of most countries budgets) How do we measure expenditure efficiency?

3 Public Expenditure on Health and Education (2010)

4

5

6

7 Efficiency Farell (1957): technical vs allocative efficiency We look at (technical) efficiency : defined as the distance from the observed input-output combinations to an estimated efficient frontier. Empirical and theoretical measures of efficiency are based on ratios of observed output levels to the maximum that could have been obtained given the inputs utilized. The maximum/efficient frontier is the benchmark for gauging the relative efficiency of the observations.

8

9 Empirical Methods There are multiple techniques to estimate the efficient frontier to examine the efficiency of public spending. Data Envelopment Analysis (DEA) DEA assumes that linear combinations of the observed input-output bundles are feasible. Free Disposable Hull (FDH) The FDH method imposes the least amount of restrictions on the data, as it only assumes free-disposability of resources Stochastic Frontier Analysis (SFA)

10 Data Envelopment Analysis (DEA) CRS: Constant return to scale VRS: Variable returns to scale Input-oriented technical efficiency is TE=YV/TC Scale efficiency is YN/YV

11 Free Disposable Hull (FDH) Input efficiency is X A /X B Output efficiency is Y B /Y A

12 Stochastic Frontier Analysis (SFA) Limitations of the non-parametric methods (e.g. EDA, FDH) derive mostly from the sensitivity of the results to sampling variability, to the quality of the data and to the presence of outliers. Aigner, et al. (1977) ln(y it ) = ln f(inputs it, time, controls it ) + (v it - u it ) Then technical inefficacies are explained by determinants z it e.g. u it = f(z it ) One-step approach by Wang & Ho (2010)

13 Input indicators The main input variable is the public spending per capita on education, health, infrastructure, etc. As in many cross-country comparisons we assume that the quality of inputs is similar among countries (factor homogeneity). This is a shortcoming as crosscountry differences in factor quality could be significant (Herrera and Pang, 2005). Moreover, input is used more intensively in richer countries (i.e. public spending is highly correlated with education, health, etc. -- Balassa-Samuelson effect; Wagner s hypothesis). We could split the sample by groups of countries (Gupta and Verhoeven, 2001). We could use per capita GDP as a second input in the model as a control variable (Jarasuriya and Wodon, 2002). A third option consists in using as an input the orthogonal component (or residual of linear regression) of public expenditure to GDP.

14 Output indicators (Herrera and Abdoulaye Ouedraogo, 2018) 7 health output indicators life expectancy at birth, immunization (DPT13 and measles), and the disability-adjusted life expectancy (DALE), maternal survival rate, infant survival rates and Tuberculosis Free Population. 14 education output indicators primary school enrollment (gross and net), secondary school enrollment (gross and net), gross tertiary school enrollment, literacy of youth, average years of school, second level complete, PISA learning scores for math, reading and science and the WEF Indices for Quality of Match and Science Education, Quality of Primary Education, and Quality of the education system.

15 Source: Herrera & Ouedraogo (2018, p.20)

16 Source: Herrera & Ouedraogo (2018, p.16)

17 Source: Herrera & Ouedraogo (2018, p.23)

18 Multiple input-output In health, for example, there are multiple possible combinations of inputs (public expenditure, private expenditure, and literacy of adults) and outputs (life expectancy at birth, immunization DPT, immunization measles, and Disability Adjusted life expectancy (DALE)). Source: Herrera & Ouedraogo (2018, p.27 & 30)

19 from Lavado & Domingo (2015)

20 Explaining inefficiencies In the literature, larger education and health spending tends to reduces public sector efficiency in each of the sectors. Why? Endogeneity? Determinants of public spending efficiency: Economic determinants, e.g. per capita income (+/-), commodity exporter (-), inflation (-), openness (+), income inequality (-), share of pubic sector in service provision, external aid, etc Institutional determinants, e.g. accountability (+), corruption (-), democracy (+), social infrastructure (+), years of schooling (+) Demographic & geographic determinants, e.g. age distribution, population density (+), linguistic fractionalization (-), geographical size (-), climate (+/-) Tobit model Estimated efficiency scores = f(determinants)

21 References Subal Kumbhakar and C Lovell (2000) Stochastic Frontier Analysis, Cambridge: Cambridge University Press. Rouselle Lavado and Gabriel Domingo (2015) Public Service Spending: Efficiency and Distributional Impact Lessons from Asia, ADB Economics, Working Paper Series, No IMF Staff Report (2015) Making Public Investment More Efficient. Santiago Herrera and Abdoulaye Ouedraogo (2018) "Efficiency of Public Spending in Education, Health, and Infrastructure: An International Benchmarking Exercise", World Bank - Macroeconomics, Trade and Investment Global Practice.