Who pays the taxes?
Our analysis dismisses the – commonly held – belief that middle-sector families are the ones supporting the heaviest total tax burden (Figure 4.11). Of course, this is relatively large, and there is considerable variation in the total amount of tax paid by particular families within it. But the bulk of the overall tax take (51% in Chile and 53% in Mexico) is generated in the highest deciles, with affluent families being net taxpayers in both countries. This overall behaviour may not be reflected across indirect taxes, health-care contributions and personal income tax. We have analysed the incidence of each of these – though the results should be treated with caution given incompleteness in the data.
Figure 4.11. Tax incidence by household income decile (weighted average, percentage of mean disposal income, 2006)
The indirect taxes are principally VAT and excise duties, the former having the greater take. Such consumption taxes have the greatest impact on the income of middle-sector households, accounting for 13.8% and 9.8% of the mean per capita income for Chilean and Mexican families respectively – personal income tax being mainly paid by the affluent (see also Box 4.2). When measured relative to decile disposable income, indirect taxes exhibit a different pattern in Chile from that in Mexico. While in Chile the top-two and bottom-two deciles pay a lower share of their income than the rest, in Mexico the share of income taken is essentially similar across income groups.
Mexico exempts many goods regarded as essential, such as food or medicine, from VAT in an effort to make the tax less regressive. In practice this proves to be a poorly targeted (implicit) subsidy and the absolute benefits from these exemptions increase with household income.
Social-security contributions for health care present different patterns in the two countries. While they are neutral in Mexico (accounting for about 1% of income in each decile), in Chile they are regressive – something explained by the fact that in Chile households higher up the income scale tend to opt for private insurance.
The top two deciles pay the bulk of the take from income tax. This reflects both the skewing of the income distribution in the region and the fact that more than 60% of income earners have sufficient exemptions to mean they pay nothing.
21 Their burden is still low nonetheless: 3.3% in Chile and 10.8% in Mexico as a proportion of the mean income in their decile. For middle-sector families, the net effect is even lower, and – given the effect of tax credits on salary – low-income groups, in Mexico at least, have effective negative contributions.
Box 4.2. Who pays personal income tax in Latin America? Not the working middle sectors
Compared with OECD countries revenues from personal income tax in Latin America are very low. Only a small proportion of the population is a net payer of this tax – and almost nobody within the middle sectors. This is the result of the region’s highly concentrated income profile, a tendency to under-report income, and tax codes full of credits and exemptions.
This small tax take is a problem for the region. Of course, it limits the public sector’s potential for redistributive policies. It also has a less obvious impact in removing a useful stabiliser from the economy. Daude et al. (2010) estimate that the automatic stabilisers inherent in Latin America’s tax systems are around half the size of their OECD equivalents. To these can be added, from a political economy perspective, the additional legitimacy that a stronger personal income tax would bring to the fiscal systems of the region.
So who does pay this tax? To find out we have modelled its incidence in seven countries of the region, according to the following methodology. First, a distribution of potential tax payers is computed using the latest available national household surveys. These have data from 2005 in Uruguay, 2006 in Argentina, Chile, Costa Rica, Mexico, Peru, and 2008 in Colombia. The “adjusted first-earner income” distribution is then calculated by taking into account household composition, using the OECD methodology for estimating structural balances (Girouard and André, 2005). The analysis is restricted to labour income (whether from employment or self-employment), and the sample is limited to households with at least some income of this type. All households with income above 6 times the national median are grouped together – on average these households earn from 8.6 times the median in Uruguay to 12.1 times in Colombia. Figure 4.12 shows the resulting distribution of households.
Figure 4.12. Distribution of households by income bracket (relative to national median labour income)
Given the high levels of informality and income inequality in the region, the conventional OECD analysis (calibrated within OECD countries for those earning from 0.5 to 3 times the median income) is extended to households earning from 0.05 times the median income (so from almost the first peso, sol or real of labour income) to more than 6 times the median income – De Mello and Moccero (2006) follow a similar procedure in their analysis for Brazil.
The effective tax burden is then computed for some 120 representative household types, assuming they differ only in their income level. Figures for Chile and Uruguay were provided by the respective finance ministries, while rates for Mexico were calculated using the OECD Taxing Wages simulator, developed by the OECD Centre for Tax Policy and Administration. For the remaining countries, calculations were based on the legislation in force during fiscal year 2006, a relatively neutral year in cyclical terms. For Uruguay survey figures were updated with the observed CPI up to 2009 to permit the incorporation of the new personal income tax framework introduced from 2008. In those cases where fiscal legislation allows individual and household declaration, the option more beneficial to the tax payer was chosen. (Tax declarations are at the individual level in Chile, Colombia, Peru and Uruguay, and by household in Argentina, Costa Rica and Mexico.) Allowances for both spouse and children were included in Argentina and Mexico.
Figure 4.13 shows the computed average effective rate by income level for each country. It is apparent that personal income tax in all countries of the sample is formally progressive, with average tax rates increasing with income. However, labour-income earners only become net payers of personal income tax at levels well above the national median wage – ranging from 1.7 times the reported household median labour income in Chile and Costa Rica, to 5.5 times in Colombia. The only outlier is Mexico, owing to the interaction of limited exempted income and tax credits. Here net tax becomes payable at about 0.85 times median income.
Figure 4.13. Average personal income tax rates by income (relative to national median labour income, percentage)
These very high effective thresholds combine with the concentration of households in the lower part of the income distribution to mean that only a very small proportion of households pay net income tax (Figure 4.14). The largest tax base is 60% of households in Mexico, and this dwindles to less than 10% in Colombia and Peru.
Focusing on the working middle sectors, Mexico gets net taxes from about half of this group (those earning from 50% to 150% of the median national household labour income). But south of here no working household from the middle sectors pays any net personal income tax – on average at least.