Fiscal Incidence Analysis. B. Essama-Nssah World Bank Poverty Reduction Group Washinton D.C. June 03, 2008

Transcription

1 Fiscal Incidence Analysis B. Essama-Nssah World Bank Poverty Reduction Group Washinton D.C. June 03, 2008

2 Introduction Key questions Who benefits from public spending? Who bears the burden of taxation? How desirable is the distribution of net benefits from the operation of a tax-benefit system? Evaluation framework Identification of winners and losers Estimation of gains and losses Shifting and incidence assumptions or formal modeling to account for individual behavior and social interaction. Progressivity Criterion for social desirability Progressive policy favors the poor relative to the non-poor Burden distributed according to ability to bear. Benefits distributed according to need Indicators Relative burden/benefit plot against income Concentration/Lorenz curves

3 The Burden of Taxation Basic tax incidence analysis Informational basis: Tax laws and statutes, data on tax collected by categories (e.g. direct and indirect taxes) from tax administration, and household survey data. Estimation of burden distribution Depending on type, taxes affect either the sources (earnings) or the uses (expenditure) side of the household account Importance of shifting: one is less likely to bear the tax burden when one faces better alternatives to what is taxed (substitutes). In given market degree of shifting depends on elasticity of demand and supply.

4 The Burden of Taxation Estimation of burden distribution, continued Assign tax payments to socioeconomic groups on the basis of their sources and uses patterns and maintained incidence assumptions Typical incidence assumptions (1) income tax paid by income recipients; (2) payroll and social security tax borne by workers; (3) corporate income tax by shareholders or all capital owners; (4) consumption tax by consumers. Progressivity If ratio of burden to income is increasing with income: tax is progressive. If decreasing: tax regressive. Else: proportional. (Alternative test: if concentration curve of tax payments lies entirely below the Lorenz curve for pretax income, tax is progressive).

5 The Burden of Taxation Case of Chile in 1996 shows that tax burden disproportionately borne by poor Second decile pays highest percentage of its income in taxes, 16 percent compared to 12 percent paid by the wealthiest. Table 1. The Distribution of the Tax Burden in Chile (1996) Decile Pre-Tax After-Tax Whole Tax Income VAT Other Income Income System Tax Taxes Source: Engel, Galetovic and Raddatz (1999); Note: the entries are in percentage.

6 The Burden of Taxation The Presumption of Progressivity Individual income and wealth taxes are the most capable of redistributing income Institutional weaknesses limit progressivity of income tax in developing countries: informal sector and weak tax administration. Hence importance of indirect taxes on goods and non-factor services (such as sales, excise and value added taxes) Taxes on consumption tend to be regressive Poor people generally spend a higher proportion of their income than the non-poor. To counter this, authorities could reduce tax rate or provide exemptions for a handful of basic foodstuffs. Taxes on luxury goods are believed to be progressive since these goods are mostly consumed by high income households.

7 Incidence of Public Expenditure Valuation issues Difficult to measure benefits accruing to individuals or household from publicly provided goods and services Short cut: unit cost of provision. Basic approach Estimate unit cost from public expenditure accounts Estimate rate of use from household survey and impute benefits on the basis of these two estimates Report results by policy-relevant socioeconomic groups (quantiles of the distribution of some welfare indicator, regions, gender or ethnic groups) Assess progressivity Similar to the case of taxation. Extent of targeting: compare concentration curve for benefits to 45 degree line, the further the curve lies above the 45 degree line the better the targeting.

8 Incidence of Public Expenditure Example: Incidence of public spending on health in Ghana in 1992 Progressive because for the poorest quintile the subsidy represents 3.5 percent of total spending compared to 1.8 percent for the richest quintile However poorly targeted because the poorest receive only 12 percent of the overall public health spending compared to 33 percent for the richest. Table 2. Incidence of Public Spending on Health in Ghana (1992) Quintile Per Capita Share of Total Subsidy as Share of Total Subsidy Subsidy (%) Household Expenditure (Cedis) Ghana Source: Demery (2003)

9 Incidence of Public Expenditure Further considerations Simpler approach: note only whether or not service is used by household (no need for unit cost); base analysis on the distribution of dummy variable indicating use. Important to account for needs in order to better assess equity. Accounting for interaction between demand and supply factor is important for policy making (regional variability in unit cost or variation by level of service); demand is influenced income, quality and cost (direct and opportunity).

10 Incidence of Public Expenditure Ignoring behavior responses to a spending program can lead to bias in the estimates of the impact of the program Behavioral approaches relevant to the construction of a counterfactual and in the context of marginal incidence analysis which seeks to determine who benefits from the expansion of an expenditure program or who loses from a cut in public spending (standard approach yields average incidence). Marginal incidence can be assessed on the basis of changes in average incidence induced by program expansion over time. Need repeated cross-sectional data (at least) to compute changes in incidence over time across quantiles. Al-Samarrai and Zaman(2007) compare the incidence of public expenditure on education in Malawi between 1990/91 and 1997/98. They found that the abolition of fees decided by the government in 1994 led to pro-poor changes in incidence. Microsimulation and general equilibrium modeling are common ways of account for individual behavior and social interaction.

11 Background Paper