A Decade of Falling Inequality in Mexico

Discussion Paper
A Decade of Falling Inequality in Mexico: Market Forces or State Action?

January 2010
United Nations Development Programme
POVERT Y REDUC TION
A Decade of Falling Inequality in Mexico: Market Forces or State Action?

Gerardo Esquivel, Nora Lustig and John Scott
January 2010
Copyright January 2010 United Nations Development Programme Bureau for Development Policy Poverty Group 304 East 45th Street New York, NY, 11375 U.S.A. E-mail: poverty.reduction@undp.org Website: www.undp.org/poverty
Abstract Mexicos Gini coefficient fell from 0.543 in 1996 to 0.498 in 2006. Standard decomposition analysis suggests that improvement in incomes at the bottom of the distribution vis--vis the upper deciles is linked to higher relative wages of low-skilled workers, a rise in the share of remittances and an expansion of government monetary transfers targeted to the poor. The fall in the skill premium appears to be associated with a reduction in the share of unskilled workers in Mexicos labour force. This change in the composition of the labour force coincided with, and was probably caused by, a significant expansion of government spending on basic education. The equalizing role of transfers, mainly driven by the conditional cash transfer programme Progresa/Oportunidades, rose over time. Benefits incidence analysis, in addition, demonstrates that government redistributive spending on education, health and nutrition became more progressive over the last decade. In 2006, taken together, taxes and transfers (including in-kind transfers) reduced the Gini by 12.6 percent. However, 58 percent of government redistributive spending was still regressive (of which 11 percent increased income inequality). In order for inequality to continue to fall in the future, Mexico needs to phase out regressive transfers and improve the quality of education, particularly for the poor. Gerardo Esquivel is a professor of economics at El Colegio de Mexico. He can be reached at gesquivel@colmex.mx. Nora Lustig is Stone Professor of Latin American Economics at Tulane University and nonresident fellow at the Center for Global Development and the Inter-American Dialogue; she can be reached at nlustig@tulane.edu. John Scott is a professor of economics at Centro de Investigacion y Docencia Economicas (CIDE) and can be reached at john.scott@cide.edu.
Acknowledgements This paper was prepared for the joint PG/BDP and RBLAC project Markets, the State and the Dynamics of Inequality in Latin America coordinated by Luis Felipe Lopez-Calva (Director of the Poverty Cluster of the Regional Bureau of Latin America and the Caribbean) and Nora Lustig (Samuel Z. Stone Professor of Latin American Economics, Tulane University and non-resident Fellow of the Center for Global Development and the Inter-American Dialogue). Project managers in PG/BDP included, at various times, Selim Jahan (Director, Poverty Practice), Rathin Roy (Acting Cluster Leader, Inclusive Development) and Shantanu Mukherjee (Policy Advisor, Microeconomics).
The coordinators and project managers are greatly indebted to Fedora Carbajal for her excellent research assistance; Mariellen Jewers, Michael Lisman, Shivani Nayyar and Anita Palathingal for their very valuable editorial recommendations; and Marina Blinova, Elia Carrasco, Patrice Chiwota, Queenee Choudhury, Jacqueline Estevez, Maria Fernanda Lopez-Portillo and Alexandra Solano for their very helpful support in the administration of this project.
Disclaimer The views expressed in this publication are those of the authors and do not necessarily represent those of the institutions to which they are affiliated or the United Nations, including UNDP, or their Member States.
1.
Introduction
Mexico is among the most unequal countries in the world. However, it is making progress in becoming more equal from 1996 to 2006, Mexicos Gini coefficient fell from 0.543 to 0.498 or by 0.8 percent per year, and from 2000 to 2006 it fell by 1 percent per year. 1 The decline in inequality coincided with important changes in Mexicos economic and social policy. In 1994, the North American Free Trade Agreement (NAFTA) with the United States and Canada went into effect, thereby establishing the largest free trade area in the world, and the most asymmetrical in terms of countries relative GDP. 2 Mexico also implemented two important government transfers programs: Procampo in 1994 and Progresa (later called Oportunidades) in 1997. Procampo is an income support programme for agricultural producers designed to help them face the transition costs resulting from the opening of agriculture under NAFTA. The second programme, Progresa/Oportunidades, is a targeted, conditional cash transfer programme. It is considered Mexicos most important anti-poverty programme. In terms of growth, barely a year after NAFTA came into effect, Mexico faced a severe macroeconomic crisis. 3 The peso crisis that began in December 1994 led to a sharp decline in economic activity during 1995 when per capita GDP fell to the tune of 8 percent. The economy recovered rather quickly from the crisis. Between 1996 and 2000, Mexicos per capita GDP grew at a rate of 4 percent per year. However, between 2000 and 2006, growth slowed down significantly; per capita GDP grew at only 1 percent per year. This low-growth period is precisely when income inequality started to decline more rapidly. Using standard non-parametric decomposition methods, this article will analyze the proximate determinants behind the reduction in income inequality between the mid-1990s and 2006. In particular, it will look at the roles played by the reduction in both labour income inequality and non-labour income inequality. It will also analyze the impact of changes in demographics, such as the ratio of adults and of working adults per household. The paper will examine to what extent the reduction in labour income inequality was due to a decline in the wage skill premium. It will explore the influence of changes in the labour force composition, in terms of education and experience, on the decline in the wage skill premium, as well as examine the relationship between the labour force composition and changes in public spending on education. Then, the paper will analyze the contribution of changes in government transfers, with particular emphasis on Progresa/Oportunidades, on the reduction in non-labour income inequality. Finally, using standard incidence analysis, the paper will look at the distributive impact of government redistributive spending and taxes.
1 The Gini reported in this paragraph is calculated using total income (it includes monetary income and imputed value of owners occupied housing). The decomposition of income inequality by the source presented below uses current monetary income (excludes capital gains and non-monetary income). The change in the Gini coefficient between 19962006, 19962000 and 20002006 was found to be statistically significant at the 95 percent level. The confidence intervals were constructed applying the bootstrap method with 150 replications. There is Lorenz dominance for the comparisons of 2006 and 1996, 2006 and 2000 and 2000 and 1996. 2 See Tornell and Esquivel (1998) for more details on these issues. 3 For an analysis of the peso crisis, see Lustig (1998). 1
Poverty Reduction Discussion Paper
The paper is organized into seven sections. Section 1 presents the evolution of Mexicos post-NAFTA distribution of income, using different measures of inequality and definitions of income at the national as well as rural and urban levels. Section 2 analyzes the proximate determinants of inequality using standard non-parametric decomposition methods. Section 3 explores the causes of the reduction in labour income inequality. Section 4 discusses the changes in government spending on education and its impact on making access to education more equitable. Section 5 analyzes the impact of Progresa/Oportunidades on income inequality in Mexico. Section 6 examines the incidence of government redistributive spending and the net impact of transfers and taxes on inequality. Finally, section 7 presents the main conclusions.
2.
Income inequality after NAFTA: 19942006
We begin our analysis by specifying what measure of inequality and what definition of income will be used in this paper, since different measures and definitions could not only lead to different estimates of inequality but also to slightly different conclusions about its determinants. 4 In this paper, we use the Gini coefficient as our preferred measure of inequality. 5 This measure satisfies all the desirable properties of an inequality indicator, and is decomposable by proximate determinants as well as income sources. 6 The paper uses total income per capita unless specified otherwise. 7 All of our estimates use information from the National Survey of Household Income and Expenditures (ENIGH, for its Spanish acronym). 8 There are comparable surveys available for the years 1994, 1996, 1998, 2000, 2002, 2004, 2005 and 2006. 9 The surveys capture income net of taxes and contributions to social security; income includes transfers from Procampo and Progresa/Oportunidades.
Our results are for the most part robust to different income concepts as well as inequality measures. Corbacho and Schwartz (2002) have a survey of Gini coefficient estimates in Mexico for different periods and different income definitions. CONAPO (2007) and Corts (2008) show recent estimates of inequality using monetary income. 5 Other measures of inequality such as the Theil index show similar trends as those described in the text. They are available from the authors upon request. 6 These principles are: adherence to the Pigou-Dalton transfer principle, symmetry, independence of scale, homogeneity and decomposability. The Gini, however, is not additively decomposable as the Theil index. 7 Income includes labour income and non-labour income. The former includes all the income that is reported as labour income in ENIGH, including labour income from the self-employed. Non-labour income includes incomes from own businesses, assets (including capital gains), pensions (public and private) and public transfers (Oportunidades and Procampo), and private transfers (e.g., remittances), as well as non-monetary income (imputed rent on owner occupied housing and consumption of own production, common in poor rural areas). Official poverty measures in Mexico use net current income: that is, capital gains and gifts and in-kind transfers to other households are subtracted from current total income. Current monetary income, the concept used in the decomposition of inequality by the National Survey of Household Income and Expenditures , does not include non-monetary income and consumption of own production (common in poor rural areas) and excludes capital gains. 8 In Spanish, Encuesta Nacional de Ingresos y Gastos de los Hogares (ENIGH). 9 Surveys for 1984 and 1989 are not as comparable but they are still used for lack of a better alternative. 2
Figure 1, Gini coefficients for alternative income definitions, 19842006
Source: Esquivel (2008). Note: Current income excludes income from sales of durables and capital gains.
Using alternative definitions of income, Figure 1 shows the evolution of the Gini coefficient in Mexico for 1984-2006. The figure clearly shows the existence of an inverted-U that peaks in 1994 and steadily declines thereafter. 10 Table 1 shows the distribution of current monetary income by deciles, the Gini coefficient and other inequality measures from 1984 to 2006. After rising by several percentage points between the mid-1980s and mid-1990s, the Gini coefficient for monetary household per capita income declined from 0.539 to 0.506 between 1996 and 2006. 11 The pace of decline accelerated between 2000 and 2006 when the Gini fell at 1 percent a year. In general, the other measures of inequality shown in table 1 follow the same trend as the Gini coefficient. But, since the Gini is more sensitive to what happens to the middle of the distribution, its pattern is slightly different from inequality measures based on top/bottom ratios. 12 One difference seen in table 1, for example, is that all the other measures tend to peak around 1998, whereas the Gini peaks in 1994. The differences between the Gini and the other inequality measures are due to important changes in the tails of Mexicos income distribution: in 1994 and 1998 both the top and the bottom deciles suffered losses in their income shares, but although the top deciles loss was greater in percentage points, the bottom decile lost more in relative terms.
10 The rapid increase in inequality that took place between 1984 and 1994 has been studied by, among others, Bouilln, Legovini and Lustig (2003) and Legovini, Bouilln and Lustig (2005). 11 The difference between the Gini coefficients shown here and those mentioned in the introduction is due to the fact that the latter are Ginis for total income whereas the former are for monetary income. 12 Sen and Foster (1973). 3
TABLE 1, Distribution of monetary household per capita income, 19842006 (percent)

Deciles I II III IV V VI VII VIII IX X Top 10 / Bottom 10 Top 20 / Bottom 20 Top 10 / Bottom 40 Gini Coefficient 1984 1.2 2.4 3.3 4.3 5.6 7.1 9.0 11.8 17.0 38.3 31.9 15.4 3.4 48.9 1989 1.0 2.2 3.2 4.2 5.2 6.5 8.3 10.9 15.9 42.6 41.7 18.0 4.0 53.4 1992 0.9 2.0 3.0 3.8 4.9 6.1 7.9 10.8 16.1 44.6 51.3 21.3 4.6 55.1 1994 0.9 2.0 2.8 3.7 4.8 6.1 7.8 10.5 16.2 45.3 48.7 21.3 4.8 56.4 1996 1.1 2.2 3.1 4.0 5.1 6.4 8.2 10.8 16.1 43.1 40.2 18.2 4.2 53.9 Year 1998 0.8 1.9 2.9 3.8 4.9 6.2 7.9 10.7 15.9 45.0 54.8 22.4 4.8 55.2 2000 1.0 2.1 3.0 3.9 5.0 6.3 7.9 10.5 16.0 44.4 44.9 19.9 4.5 54.1 2002 2004 1.2 1.1 2.3 2.4 3.3 3.4 4.2 4.2 5.3 5.3 6.5 6.6 8.3 8.3 11.1 10.9 16.3 16.0 41.7 41.8 36.3 16.9 3.8 51.3 36.7 16.2 3.8 50.8 2005 1.1 2.3 3.2 4.1 5.2 6.5 8.2 10.6 15.4 43.7 40.8 17.8 4.1 51.8 2006 1.2 2.5 3.3 4.2 5.3 6.7 8.0 10.8 16.0 42.2 34.3 15.7 3.8 50.6
Source: Esquivel (2008). Note: The concept of current monetary income does not include the imputed value of owners occupied housing and income from sales of durables and capital gains. The income concept used in the decomposition by proximate factors uses total household income (monetary and non-monetary) household per capita income. Since it is different from current monetary income, the Gini coefficients need not coincide.
The evolution of Mexicos distribution of income can also be analyzed using the Growth Incidence Curves (GICs) suggested by Ravallion and Chen (2003). These curves show the percentage change in per capita income along the entire income distribution between two points in time. Figure 2 shows the GIC for 19962006 as well as for 19962000 and 20002006. 13 Looking first at the negative slope in the first graph shows why Mexicos income inequality diminished during 19962006: the income of the lower deciles grew faster than the income of the upper deciles. For example, the bottom percentiles income grew at more than four times that of the top percentile. Esquivel (2008) also presents GICs for urban and rural areas for 19942006. 14 They feature the different patterns followed by the urban and rural income distributions during this period. In the urban areas, income growth was pretty flat across the entire distribution except for the top three deciles, which experienced smaller, and in some cases even negative, income growth rates. In rural areas, the GIC had a negative slope, indicating that the bottom half of the rural income distribution had higher income growth rates than the top segment of the distribution. Finally, comparing the growth rates between rural and urban areas showed that the average income growth in rural areas was greater than in urban areas a pattern that, given the relatively large ruralurban gap, reduces inequality.
13 The GICs are constructed using total per capita income, which includes monetary and non-monetary and capital gains. 14 Rural areas are defined as townships with less than 15,000 inhabitants. 4
Figure 2, Growth Incidence Curves, national, urban and rural, 1996-2006 (percent)
Growth Incidence Curve, 19962006
Source: Authors elaboration based on ENIGHs 1996, 2000 and 2006. Note: GICs are based on total household per capita income; rural refers to households living in townships of population size less than 15,000.
Breaking out GICs for 19962000 and 20002006 also provides important details about Mexicos overall decline in inequality from 19962006. The GICs in 19962000 and 2000-2006 both have negative slopes. However, in 20002006 the negative slope was more pronounced, showing a larger increase of bottom incomes. The GICs in 19962000 and 20002006 share the following common aspects: (1) the poorest two deciles of the income distribution experienced an above-average increase in their monetary income, and (2) the income of the top decile grew at below-average rates. The increase for the lowest two deciles seems to be associated with the income growth in the rural sector. The lackluster growth of the top deciles income is associated with the dynamics of the urban sector of Mexicos economy. In fact, the pattern in the urban areas in 19962000 and 20002006 is almost identical: a very flat GIC throughout most of the income distribution, with the income of the top two deciles growing at lower rates than that of the rest of the distribution (Esquivel, 2008). Changes in rural areas account for the difference between the two GICs. In 20002006, the rural GIC curve clearly showed a pro-poor pattern, with the bottom half of the distribution enjoying substantial income growth rates (over 5 percent per year); whereas the upper half of the distribution had much smaller income gains. These results suggest that during 20002006 there must have been some factors that benefited the bottom part of the rural income distribution, as well as other factors that hurt in relative terms the upper part of the urban income distribution. The next section investigates which factors may explain this behaviour.
3.
The proximate determinants of the decline in income inequality: Labour and non-labour income and demographic factors
The purpose of this section is to identify the proximate determinants of Mexicos decline in inequality between 1996 and 2006, and quantify each proximate determinants contribution to the total decline. The proximate determinants considered in our analysis are: (1) ratio of adults to total number of members in the household, (2) proportion of adults working to total number of adults in the household, (3) labour earnings per working adult, and (4) household non-labour income (which includes government transfers and remittances) per adult. It is worth emphasizing that our analysis in this section is limited to these four proximate determinants of inequality. However, each proximate determinant, in turn, is the result of behavioural and external processes, which are not modelled here. For example, the first proximate determinant captures the impact of changes in fertility and life expectancy. The second is influenced by decisions to participate in the labour force and the demand for labour. The third and fourth labour earnings per working adult and household non-labour income are determined by many factors: for example, market forces and state action affecting the demand for labour by characteristics (education, experience, gender, sector, formal/informal); decisions by individuals (and their parents) to invest in education and other forms of capital, to participate in the labour market and how (wage earner vs. self-employed, full-time vs. part-time), to migrate and send remittances; and government transfers. A formal analysis of these factors goes beyond the scope of this paper. However, we will explore some plausible explanations for the observed trends.
Each proximate determinants contribution to the total decline in inequality was quantified by applying the method proposed by Barros et al. (2006). In essence, the method consists of decomposing the change in an inequality measure into the contributions from changes in the distribution of the proximate determinants, taken one at a time, plus the contributions from changes in the interaction (correlation) of proximate determinants with each other. The contributions are estimated through a series of sequential counterfactual simulations which assume that the distribution of the proximate determinant of interest remains the same as in the base year. 15 The method is based on the following sequence of identities: y=ar r=o+t and t=uw Hence, y = a (o + u w) (4) (3) (1) (2)
Identity (1) expresses household per capita income, with y as a product of the proportion of adults in the household, a, and household income per adult, r. Identity (2) expresses household income per adult, r, as the sum of household non-labour income per adult, o, and household labour income per adult, t. For Identity (3), household labour income per adult, t, is expressed as the product of the proportion of working adults, u, and the labour income per working adult in the household, w. Identity (4) relates per capita household income, y, to its four proximate determinants: (1) the proportion of adults in the household, a, (2) household nonlabour income per adult, o, (3) proportion of working adults, u, and (4) labour income per working adult in the household, w. Visually, these identities are presented in figure 3. For a detailed description of the methodology see Barros et al. (2006). Using this method, Alejo et al. (2009) estimated that the contribution of changes in the four proximate determinants mentioned above accounted for 1.40 percentage points of the decline in the Gini coefficient from 1996 to 2000, and 3.07 percentage points of the decline in the Gini coefficient from 2000 to 2006. 16 The results based on Alejo et al. are summarized in table 2. As one can see, the changes in all of the four proximate determinants were equalizing and the changes in all the interactions between the proximate determinants combined were unequalizing in both 19962000 and 20002006. 17 The reduction in the inequality of labour income (labour earnings per worker from wages and from self-employment) was the most important contribution to the reduction in inequality in both 19962000 and 20002006.
15 Note that although you can apply this method using any inequality indicator, the results will vary depending on the indicator. Also, the results will be sensitive to which year is chosen as the base year and the sequence selected to construct the counterfactual simulations. 16 This decomposition was based on inequality measures calculated using total (monetary plus non-monetary) income while the next decomposition exercise was based on inequality measures estimated using monetary income only. The results, however, should not be very sensitive to the use of different concepts because both monetary and non-monetary income followed the same pattern of change. 17 The individual interaction terms between pair of variables were all unequalizing too. See Alejo et al. (2009). 7
The reduction in labour income inequality (leaving out the interaction terms) accounted for 87.1 percent of the decline in inequality in 19962000 and for 65.5 percent of the decline in 20002006. Given its relative importance, we analyze below which factors explain the reduction in inequality in the distribution of labour income per worker, focusing on the skilled/unskilled wage gap and the latters relationship to trade liberalization and the educational upgrading of the labour force. 18
Figure 3, Household per capita income and its determinants
Source: Barros et al. (2009a).
According to table 2, the equalizing contribution of the proportion of adults in the household (which measures the dependency ratio) rose from 7.7 percent in 19962000 to 10.3 percent in 20002006. Also, the equalizing contribution of the proportion of working adults in total adults (which measures both the supply-side decisions to participate in the labour market and the demand-side conditions of finding employment) rose by several percentage points from 19962000 to 20002006. Between 1996 and 2000 and between 2000 and 2006, the average household size fell from 5.68 to 5.16 and from 5.16 to 4.97, respectively, and the proportion of working adults in the household rose from 58 percent to 59 percent and from 59 to 62 percent, respectively. 19 These trends may be reflecting two important changes in demographic patterns: (1) the reduction in fertility rates overall, with more pronounced declines among the poorer sectors of the population; (2) the increase in female labour force participation, particularly among the poorer sectors. Between 1996 and 2006, the average
18 Labour income includes all the income that individuals reported as labour income in the ENIGHs, including all the wages and salaries as well as the income reported by self-employed individuals. 19 See Esquivel (2008). 8
number of children per household in the lowest quintile fell from 2.3 to 1.7, whereas in the top quintile, it fell from 1.5 to 1.3. 20 The participation of adult women in the labour force during this period rose from 45.3 to 57 percent. 21
Table 2, Contribution of the proximate determinants to changes in income inequality, 19962000 and 20002006
Marginal Years 19962000 contribution of the Contribution to In percentage the change (*) proximate factor Proportion of -0.19 7.7 adults Non labor income -0.01 0.4 per adult Proportion of -0.12 4.9 working adults Labor income per -2.18 87.1 working adult SUBTOTAL -2.50 100.0 Interactions (all of 1.10 them) TOTAL (change in -1.40 Gini coefficient) Source: Authors calculations based on Alejo et al. (2009) Notes: 1. The change in Gini coefficient is in percentage points. 2. * refers to the contribution of the factor to the change in the Gini coefficient measured in percentage points. 3. The change in the Gini coefficient between 1996 and 2000 was found to be statistically significant at the 95 percent level. The confidence intervals were constructed applying the bootstrap method with 150 replications. 4. A negative (positive) sign means that a marginal increase in the source is equalizing (unequalizing). Years 20002006 Contribution to In percentage the change (*) -0.50 -0.73 -0.44 -3.19 178.7 -78.7 100.0 -4.87 1.79 -3.07 10.3 15.1 9.1 65.5 100.0 158.3 -58.3 100.0
The equalizing contribution of changes in the distribution of non-labour income to the decline in overall inequality rose sharply from 19962000 to 20002006 (table 2). In 19962000, non-labour income contributed a meager 0.4 percent to the reduction in inequality. In contrast, in 2000-2006 it accounted for 15.1 percent of the total decline in inequality, making non-labour income the second most important contributor to the decline in inequality in this period, of the four proximate determinants considered in this decomposition. Non-labour income is a very heterogeneous concept. It includes incomes stemming from the ownership of capital (such as profits, interests and rents), which tend to be concentrated at the top of the income distribution, but it also includes private transfers (such as remittances), which tend to be more concentrated in the middle and lower-middle ranges of the distribution. Finally, non-labour income also includes government transfers (such as pensions), which are concentrated in the middle and upper-middle ranges of the income distribution, as well as targeted government
20 Number of children under 12 per household classified by parental income quintile. Source: SEDLAC. Available at www.depeco.econo.unlp.edu.ar/sedlac/. 21 Share of individuals between 25 and 64 years old in the labour force. Source: SEDLAC. Available at www.depeco.econo.unlp.edu.ar/sedlac/. 9
transfers (such as the conditional cash transfer programme Progresa/Oportunidades), which are concentrated in the bottom of the distribution. 22 In what follows, we conduct a decomposition of the Gini coefficient to investigate the contribution of different income sources to the observed inequality of monetary income in Mexico in 1994, 2000 and 2006. 23 Lerman and Yitzhaki (1985) showed that the Gini coefficient for total income inequality (G) with K income sources can be expressed as:
where S k is the share of source k in total income, G k is the Gini coefficient of the income source k, and R k is the Gini correlation between the income source k and total income. 24 This decomposition of the Gini coefficient shows that the contribution of income source k to inequality depends on the interaction of three elements: (1) the relative importance of the particular income source in total income (S k), (2) the level of inequality of that income source (G k), and (3) the correlation between the distribution of that income source and that of total income (R k). Therefore, an income source (k) that represents a relative large share of total income (high S k), could have a large effect on inequality as long as it is unequally distributed (i.e. if it has a relatively high G k). However, if G k is low, it will dwarf the contribution of that income source. On the other hand, if an income source is very unequally distributed (high G k), but it is not highly correlated with total income (meaning it has a low R k, such as in the case of well-targeted transfer programs), then the contribution of such a source could in fact become negative, that is, equalizing. Stark, Taylor and Yitzhaki (1986) showed that with this type of decomposition one can estimate the effect of a small percentage change () in a given income source on total inequality (holding all other income sources constant) through the following expression:
or, alternatively
This expression means that the percentage change in inequality resulting from a marginal percentage change in income source k is equal to the relative contribution of component k to overall inequality minus the initial share in total income of income source k.
22 For a description of this programme, see section 6. 23 The base year used in this decomposition is 1994, and the income concept is current monetary income. In the previous decomposition, the base year used is 1996 and the income concept is total income. The difference in the income concept used in the two decompositions does not affect the main conclusions obtained from the results. 24 In comparison with the previous method, Lerman and Yitzhakis only allows you to see by how much inequality would change if the share of a particular income source increases but its distribution remains unchanged. It is, therefore, a static decomposition and applies to very small changes. In contrast, the previous method is designed to analyze the impact of a change in the distribution of a particular income source. It is therefore a dynamic decomposition. 10
We proceed to decompose the Gini coefficients for monetary income following the approach just described for the years 1994, 2000 and 2006. 25 The results of our decomposition exercise are shown in table 3 and the marginal effects are summarized in figure 4. At the national level, there are three inequality-increasing and three equalizing sources of income. The inequality-increasing sources of income include pensions, income from own businesses (profits) and income from property (rents). 26 The equalizing sources of labour income (since 2000) include remittances and transfers. 27 Their marginal negative (equalizing) effects on the Gini coefficient have increased over the years considered here. In the case of urban and rural areas, the signs of the marginal effects of the different income sources are similar to those at the national level but there are important differences (table 3). For example, labour income is a very important inequality-reducing force in urban areas, but not in rural areas. In fact, labour income in rural areas is inequality-increasing in 2006. Transfers are a very important equalizing income source in rural areas but less so in urban areas.
Table 3, Decomposition of Gini for monetary income and marginal effects by income source
Sk National Level Labor income Own businesses Property rents Pensions Transfers Remittances Urban Level Labor income Own businesses Property rents Pensions Transfers Remittances Rural Level Labor income Own businesses Property rents Pensions Transfers Remittances 0.62 0.25 0.02 0.05 0.05 0.01 0.66 0.22 0.03 0.05 0.04 0.01 0.51 0.34 0.02 0.04 0.07 0.03 Gk 0.70 0.87 0.99 0.96 0.95 0.98 0.63 0.89 0.99 0.95 0.96 0.99 0.70 0.81 0.99 0.98 0.93 0.97 Rk Share Marginal Effect 0.01 0.00 0.01 0.00 -0.01 -0.01 -0.02 0.02 0.01 0.00 -0.01 0.00 -0.01 0.01 0.01 0.01 -0.02 0.00 Sk 0.58 0.25 0.02 0.06 0.06 0.02 0.62 0.23 0.02 0.07 0.04 0.01 0.50 0.30 0.01 0.04 0.10 0.05 Gk 0.67 0.88 0.99 0.96 0.93 0.97 0.62 0.89 0.99 0.95 0.95 0.99 0.69 0.84 0.99 0.97 0.89 0.96 Rk Share Marginal Effect -0.02 0.02 0.01 0.01 -0.02 -0.01 -0.04 0.04 0.01 0.01 -0.01 0.00 -0.02 0.04 0.00 0.01 -0.03 -0.01 Sk 0.60 0.22 0.03 0.07 0.06 0.02 0.62 0.21 0.03 0.08 0.05 0.01 0.53 0.24 0.02 0.06 0.10 0.06 Gk 0.66 0.88 0.99 0.95 0.89 0.97 0.61 0.88 0.98 0.94 0.93 0.98 0.72 0.86 0.99 0.98 0.78 0.95 Rk Share Marginal Effect -0.01 0.02 0.01 0.02 -0.03 -0.01 -0.04 0.03 0.02 0.01 -0.02 -0.01 0.02 0.02 0.01 0.02 -0.06 -0.01
1994 0.83 0.63 0.66 0.25 0.80 0.03 0.64 0.05 0.47 0.04 0.25 0.01 1994 0.81 0.64 0.66 0.24 0.76 0.04 0.52 0.04 0.48 0.03 0.29 0.00 1994 0.73 0.50 0.67 0.35 0.74 0.02 0.71 0.05 0.42 0.05 0.46 0.03
2000 0.79 0.57 0.68 0.28 0.80 0.03 0.66 0.07 0.38 0.04 0.39 0.02 2000 0.77 0.57 0.67 0.27 0.80 0.04 0.58 0.08 0.38 0.03 0.49 0.01 2000 0.72 0.48 0.70 0.34 0.71 0.02 0.69 0.06 0.42 0.07 0.48 0.04
2006 0.80 0.58 0.67 0.24 0.82 0.04 0.70 0.09 0.32 0.03 0.31 0.01 2006 0.77 0.57 0.65 0.24 0.80 0.05 0.63 0.09 0.37 0.03 0.32 0.01 2006 0.79 0.55 0.69 0.25 0.78 0.02 0.79 0.08 0.31 0.04 0.52 0.05
Source: Esquivel (2008). Note: A negative (positive) sign means that a marginal increase in the source is equalizing (unequalizing). For more details see text and Lerman and Yitzhaki (1985).
25 In the decomposition exercise Esquivel (2009) made use of the descogini Stata programme written by Lpez-Feldman (2006). 26 Pensions include both private and public pensions and the pensions that are part of government welfare transfers. Pensions are gross, that is, contributions to social security are not subtracted. 27 Transfers include public and private (including gifts and donations) except for remittances and the pensions that are part of government welfare transfers (the latter are included under pensions). 11
Figure 4, Decomposition of the Gini coefficient by source: National, urban and rural
National Mexico: Marginal Effect on Gini Coefficient by Income Source
Urban Mexico: Marginal Effect on Gini Coefficient by Income Source
Rural Mexico: Marginal Effect on Gini Coefficient by Income Source
Source: Esquivel (2008).
12
In 1994, remittances did not seem to have a large negative marginal (equalizing) effect on inequality in either sector, although they were relevant at the national level. This apparently paradoxical result is explained by the fact that while the Gini correlation between remittances and rural monetary income is close to 50 percent, the Gini correlation between remittances and monetary income at the national level is much lower (see table 3). In that sense, remittances used to have an effect at the national level because they were heavily concentrated on the bottom half of the national income distribution. 28 Therefore, remittances worked as an equalizing source of income by reducing the rural-urban income gap and not through the sector specific income distribution. This, however, changed in 2000 and more decisively in 2006, when remittances were also equalizing for rural and for urban areas taken separately. Concentrating on the components of non-labour income, in table 3 we can observe that income derived from capital ownership (owning businesses and property rents) increased overall income inequality. Further, the inequality-increasing contribution of income from own businesses grew from 1994 to 2006. Remittances are equalizing. The equalizing contribution of remittances grew in both urban and rural areas separately, but not at the national level (which has been discussed above). Pensions (which include both private and public pensions) are inequality-increasing; their contribution to overall inequality grew over time, and at all levels, national, urban and rural. Closer scrutiny reveals that the increase in pensions contribution to overall inequality is due to an increase in pensions share of total income and, above all, the Gini correlation of pensions with total income. The (positive) Gini correlation of pensions rose from 0.64 in 1994 to 0.66 in 2000 to 0.70 in 2006. It is not clear what caused this. In the case of transfers, in contrast, their contribution is equalizing and rises over time both at the national level as well as for urban and rural areas (table 3). 29 In fact, by 2006, transfers became, over time, the income source with the largest equalizing effect of all the sources considered in this exercise: that is, a marginal increase in transfers would reduce inequality by more than a marginal increase in labour income or remittances. This was for three reasons: their share in total income rose, their own inequality fell and their Gini correlation with total monetary income fell. These changes were particularly pronounced in the rural areas where the share of transfers in total income rose from 7 percent in 1994 to 10 percent in 2000 and 2006; the Gini coefficient of transfers fell from 0.93 in 1994 to 0.89 in 2000 and 0.78 in 2006; and, the Gini correlation between transfers and total monetary income fell from 0.42 in 1994 and 2000 to 0.31 in 2006. 30 The share of transfers in total income rose because there was a significant expansion in coverage of public transfers. The percentage of households that received some source of non-labour income rose from 1994 to 2006: 23.8 percent of households received a public or a private monetary transfer in 1994; in 1996, it was 29 percent; in 2000, it was 34 percent, and in 2006, 45.5 percent of all households reported receiving part of their monetary income through a private or public transfer. 31 Although Procampo had been expanded since its creation in
28 See, for example, Esquivel and Huerta-Pineda (2007). 29 Transfers here include government transfers and private transfers excluding pensions and remittances. Private transfers (excluding remittances) are relatively small on average. 30 The Gini coefficient for transfers is very high because it is calculated for the entire population, including those who do not receive any transfers. 31 Esquivel (2008). 13
1994, the lions share of the expansion in households receiving non-labour income was due to implementation of the conditional cash transfer programme Progresa/Oportunidades in 1997. 32 Progresa/Oportunidades did not exist in 1996, so the percentage of households receiving transfers from this source was equal to zero in 1996; in 2006, however, Progresa/Oportunidades reached 14.8 percent of households in Mexico. 33 Alejo et al. (2009) estimate the contribution of the combined marginal effect of the changes in public transfers that occurred in terms of coverage, average benefit and distribution among recipients of all public transfers (pensions, Progresa/Oportunidades, Procampo, etc.). Given that these authors included pensions in public transfers, their results are not strictly comparable with the previous decomposition. 34 Nevertheless, some of the findings are insightful. While the combined marginal effect of what they call public transfers increased inequality for 19962000, they were equalizing for 20002006. In the latter period, the equalizing effect of the increase in coverage (percentage of households who receive public transfers) and in the magnitude of the average benefit more than compensated for the unequalizing effect caused by a rise in inequality in the distribution of public transfers among recipients. 35 Also, during 2000-2006, the equalizing marginal contribution of the changes in public transfers were large enough to compensate for the unequalizing effect stemming from changes in the interaction term that measures the association (correlation) between public transfers and total income. In contrast, during 19962000, the unequalizing effect of the interaction term dominated. In sum, starting in the late 1990s, the size of monetary government transfers became more generous, transfers became more equally distributed among recipients, and the recipients of transfers increasingly belonged to the relatively poorer segments of the population. This must undoubtedly reflect the implementation of Progresa/Oportunidades, which will be analyzed below. However, the fact that the Gini correlation between transfers and total monetary income is positive (table 3) reflects that government transfers are not as progressive as one would like them to be. If they were, the correlation would have been negative indicating that transfers are inversely allocated with respect to income. Nonetheless, an encouraging sign is that the Gini correlation between transfers and total monetary income fell quite significantly between 1994 and 2006. Given that the fall in labour income inequality (especially in urban areas) and the increase in pro-poor transfers in rural areas appear to be the two most important forces in the decline in inequality, sections 4 and 6 will analyze them more closely.
32 Based on what we know about the distributive effects of the Procampo (relatively regressive) and Progresa/ Oportunidades (very progressive) discussed below, it is quite likely that they can actually account for a large deal of the dynamics of income inequality in rural areas. For general equilibrium effects, see Taylor, Dyer and YunezNaude (2005). 33 For more details about Progresa/Oportunidades see, for example, Levy (2006). 34 In this decomposition, pensions were treated as if the full amount corresponds to a public transfer. Strictly speaking, this is not the case, since pensions include private pensions, and also because part of public pensions is personal savings (contributions by employees) and not transfers from the government. Since the public transfer of pensions tends to be regressive (see later in text), the equalizing effect of government transfers in 20022006 was probably quite strong to make the total (including pensions) to be equalizing too. 35 Alejo et al. (2009), table 16. 14
4.
Labour income inequality and the skilled/unskilled wage gap
The results of the decomposition exercises of the previous section suggest that one of the most important equalizing forces between 1996 and 2006 has been the evolution of labour income inequality. Note that labour income is basically the result of multiplying hours worked by hourly wages (here defined as including remunerations to the self-employed). If we assume that hours worked did not change much from 1996 to 2006, the change in labour income inequality must have been caused by changes in hourly wage inequality. 36 We shall focus on one of the dimensions of wage inequality: the skilled/unskilled wage gap. As shown in Bouilln et al. (2004), changes in the returns to education (the returns to years of schooling became more convex) accounted for a significant share of the rise in household per capita income inequality between 1984 and 1994. During the 1994/96-2004 period the opposite appears to have occurred. Figure 5 shows the evolution of the ratio of non-production workers wages to production workers wages 37 from 1984 to 2007. 38 The pattern of wage inequality is remarkably similar to the evolution of inequality in the various definitions of income that were shown in figure 1. Similar to the pattern of income inequality in figure 1, figure 5 shows an increase in wage inequality between 1984 and the mid-1990s followed by a steady decline since then. The rapid increase in wage inequality that occurred in Mexico between 1984 and the mid1990s has been the subject of a fairly large body of research. 39 The increase in the skilled/ unskilled gap coincided with the unilateral trade liberalization of the Mexican economy that started in the mid-1980s. In that sense, the evolution of Mexicos wage inequality was somehow unexpected; Mexico is a relatively unskilled-labour abundant country (at least from the perspective of its main trade partner, the United States), and standard theories of trade would have predicted exactly the opposite pattern (i.e., a reduction in the skilled/unskilled wage ratio). 40 The explanations that have been proposed to explain this apparent paradox can be divided roughly into two groups: the first groups explanations emphasize factors affecting the bottom part of the income distribution (that is, the segment mostly comprised by less skilled
36 Actually, between 1996 and 2006, weekly hours in all jobs fell very slightly from 45.6 to 45.1, and the decline was concentrated in low education (poorer) workers which would be an unequalizing change. This means that the equalizing changes in the distribution of hourly earnings must have been large enough to compensate for the unequalizing effect of the changes in the distribution of hours worked. Data on weekly hours and hourly wages can be found in www.depeco.econo.unlp.edu.ar/sedlac/. 37 This ratio is also usually identified as the skilled/unskilled wage ratio, where non-production workers are used as a proxy for skilled labour and production workers are a proxy for unskilled labour. This is, of course, an over simplification, since there are production workers that are highly skilled and non-production workers that are relatively unskilled. 38 The data for this graph came from the Industrial Survey in Mexico, which has monthly and annual data on total wages paid and total hours worked in the industry by both production and non-production workers. This figure is an updated version of similar figures published in Esquivel and Rodrguez-Lpez (2003) and Chiquiar (2008). 39 See, for example, Esquivel and Rodrguez Lpez (2003), Airola and Juhn (2005), Robertson (2007), Acosta and Montes-Rojas (2008), Chiquiar (2008), Verhoogen (2008), and the references cited therein. 40 For a review of the literature for Mexico and Latin America more broadly, see de Hoyos and Lustig (2009). 15
and less experienced workers); the second groups explanations emphasize factors affecting the upper part of the distribution. In the first group, for example, we have theories emphasizing the reduction in real minimum wages (Fairris, Popli and Zepeda, 2008), as well as theories suggesting that the mid-1980s reduction in tariffs disproportionally affected industries intensive in lowskilled workers (Hanson and Harrison, 1999). In the second group, for example, some theories have emphasized the role of an increase in the demand for skilled workers associated with one or more of the following: the presence of foreign investment (Feenstra and Hanson, 1997), a skill-biased technological change (Cragg and Eppelbaum, 1996 and Esquivel and RodrguezLpez, 2003), or a process of quality-upgrading due to an increase in exports (Verhoogen, 2008). Other explanations have suggested that education inequality could have also played a role (Lpez-Acevedo, 2006) or that these trends could be indicating only short-run effects (Canonero and Werner, 2002). Of course, many of the proposed explanations from both groups for the pre-NAFTA increase of wage inequality in Mexico are not mutually exclusive, and could in fact be (at least partially) correct.
Figure 5, Skilled / unskilled industrial wages, 19842007
In comparison to the 1984 and the mid-1990s inequality increase, the post-1996 reduction in wage inequality in Mexico has been much less studied. Among existing research, Robertson (2007) suggests that Mexicos manufacturing workers are now complements to, rather than substitutes for, U.S. workers. He also posits that there has been an important expansion of assembly line activities in Mexico (the so-called maquiladoras), which has
16
increased the demand for less skilled workers. 41 Campos (2008) emphasizes the supplyside explanations based on changes in the composition of the labour force. Esquivel (2008) investigates the role of demand-side and supply-side factors by looking at the patterns in male workers mean-log wages in Mexico for selected years and for different combinations of education and years of experience. Workers are classified according to the level of education achieved (less than lower secondary, lower secondary, upper secondary, and college education) and to the number of years of work experience (less or more than 20 years of experience). Between 1989 and 1994, most of the changes in the wage distribution occurred in the upper tail of the distribution (workers with high wages and high levels of education and experience). 42 The increase in wage inequality in those years was not caused by a (relative) decline in the wages of the low-skilled or low-experienced workers; rather it was the result of a rise in the wages of the high-skilled or highly experienced workers. This basically makes explanations based on changes in the lower tail of the wage distribution such as those based on a falling real minimum wage or a bias against unskilled-labour intensive industries caused by trade liberalization not convincing. In contrast, between 1996 and 2006, the reduction in wage inequality was caused by the changes in the wage distribution that took place in the lower tail of the income distribution; workers with lower levels of education and/or less years of experience had the largest increases in their average wages. This suggests that any convincing story of the post-NAFTA reduction in wage inequality has to explain the relative increase in the wages of the low-skilled/less experienced workers (as opposed to the reduction of the wages of the high-skilled/more experienced workers). 43 The discussion of the literature above confirms the impression that there is no single explanation for the evolution of wage inequality in Mexico since 1984. Indeed, the fact that the 19841994 increase in wage inequality is associated with increases in the upper tail of the income distribution, whereas the post-NAFTA reduction in wage inequality is mostly associated with increases in the bottom tail, suggest that there at least two leading forces at play. During 19841994, the only explanations that seem to be compatible with the observed trend in inequality are those suggesting the presence of skill-biased technological change, either exogenously (Cragg and Eppelbaum, 1996 and Esquivel and Rodrguez-Lpez, 2003) or endogenously via the presence of multinational firms (Feenstra and Hanson, 1997) and/or by the quality upgrading of exporting firms (Verhoogen, 2008).
41 Robertson (2007) noticed that the pattern of wage inequality in Mexico is puzzling because no single theory could explain the evolution of wage inequality before and after NAFTA. There are, however, some tentative theoretical explanations for the pattern. For example, Atolia (2007) has suggested that, under certain circumstances, even if the standard prediction from a Hecksher-Ohlin-Samuelson model works as predicted in the long run, there may be some short-run (or transitory) effects of trade liberalization that may lead to a different outcome from those of the long run. The difference between short-run and long-run effects on inequality result from two factors: first, an asymmetry in the contraction and expansion of some sectors, and second, because of the capital-skill complementarity in production. 42 The data was collected and organized by Campos (2008). 43 It should be noted that this occurred within a context in which average real and legislated minimum wages were practically flat since the mid-1990s. 17
For the post-NAFTA period, there are at least three possible explanations. Two, which are not mutually exclusive (and have already been mentioned), are: an increase in the relative supply of skilled workers (Campos, 2008) and an increase in the demand for unskilled labour resulting from an expansion in assembly line activities (maquiladoras) in Mexicos manufacturing sector (Robertson, 2007). Any of these two by itself could be sufficient to explain the reduction in the skilled wage premium that is observed in the data. However, a third explanation of the observed pattern could be consistent with a lagged effect of the standard Hecksher-Ohlin effect in an unskilled-labour abundant country such as Mexico (Chiquiar, 2008). The predicted pattern of a lower skill premium may have manifested itself with a lag either because the impact of trade liberalization on wages could have taken a few years, as suggested by Canonero and Werner (2002), or, alternatively, because an underlying effect had not showed up in the data before due to the presence of a stronger force, such as a skill-biased technological change as argued by Esquivel and Rodrguez-Lpez (2003).
Figure 6, Workforce composition by education and experience levels, 19892006
To test these alternative hypotheses one would need a much more detailed and rigorous analysis, which goes beyond the scope of this paper. However, based on the patterns of wage inequality reviewed here, we might be able to identify which hypothesis is more plausible. Figure 6 shows the composition of Mexicos workforce between 1989 and 2006, according to the levels of education and experience. The observed pattern reflects the interaction of both supply and demand factors. In general, the figure shows that from 1989 to 2006 there was both a reduction in the share of the least skilled (those with less than lower secondary education) and less experienced (those with less than 20 years of experience) workers, and an increase in the share of the most skilled (those with college education) and more experienced (those
18
with more than 20 years of experience) workers. The most dramatic changes, however, took place in the share of those workers with less than lower secondary education (but more than 20 years of experience). In fact, this group, which accounted for almost 55 percent of workforce in 1989, only represented about one third of the workforce by 2006. The share of workers with less than lower secondary education but more than 20 years of experience declined by about 20 percentage points in a 17-year span. This reduction was compensated by an increase in the shares of all the other groups of workers. These trends, which had already been present between 1989 and 1994, accelerated in the post-NAFTA period. Therefore, these results suggest that most of the relative increase in the wages of low-skilled/low-experience workers is associated with the change in the composition of the workforce in Mexico. In particular, the increase is associated with a reduction in the relative number of unskilled workers. This result is not incompatible with the hypothesis of an increase in the demand for unskilled workers as suggested by Robertson (2007). But Robertsons story, by itself, cannot explain the simultaneous increase in the relative wages and the reduction in the share of these workers in Mexicos labour force.
Figure 7 , Change in share of total workers by education and experience vs. change in log wage, by gender, 19962006
Figure 7 shows some results that reinforce our view. The x-axis shows the change in the share of the eight different groups of workers according to their levels of education and experience between 1996 and 2006. These three categories that include the less educated/ less experienced workers have seen a decline in their participation in Mexicos workforce. The y-axis indicates the average change in the log wage of male and female workers that belong to each one of these groups. As expected, the groups whose shares have diminished in the past decade are those that have had the largest increase in their wages. Notice that the increase in the wages of these workers is close to 20 percent, and in some cases even close to 30 percent,
19
for this 10-year period. On the contrary, those categories of workers that have increased their share in Mexicos workforce (the more educated/more experienced workers) tend to have had either stagnant or decreasing wages since 1996. Figure 7 supports the hypothesis that the change in Mexicos workforce composition appears to be one of the leading forces explaining the reduction in wage and labour income inequality in Mexico in the post-NAFTA period.
Table 4, Schooling inequality among adults, 19842006
Population Quintile 1 2 3 4 5 Mean Concentration Coefficient Gap decile 10-1 Source: Scott (2009a). Note: Adults are defined as people between 25 and 65 years old. Average Schooling (in years) 1984 1994 2006 2.2 2.8 4.8 3.0 3.9 6.4 3.9 5.1 7.3 5.6 6.7 8.9 7.9 9.9 12.1 4.9 6.1 8.3 0.345 6.5 0.347 8.8 0.276 9.4 Change (in percent) 19841994 19942006 27 73 30 65 32 43 19 34 25 22 27 36 1 36 -21 6
The reduction in the relative supply of workers with low levels of skills (education) might be the result of the significant increase in average years of schooling for the bottom two quintiles, which reduced educational attainment inequality considerably between 1994 and 2006 (table 4). Public spending on education changed substantially in the 1990s, which, combined with the effects of the conditional cash transfer programme Progresa/Oportunidades (discussed in section 6), significantly increased access to secondary education for the poor. The next section will analyze the changes in public spending on education and their impact on equality of access to education in more detail.
5.
The rising progressivity in government spending on education
Public spending on education in the 1970s and 1980s was heavily biased towards higher education. 44 In the 1970s, the share of educational spending allocated to upper secondary and tertiary education grew from 20 percent to 42 percent while the share of spending on basic education declined by an equivalent amount, despite an expansion in enrollment in public basic education from 9.7 to 16.5 million students. The impact on spending per student in basic education was aggravated in the 19831988 adjustment period, when basic education absorbed a disproportionate share of budgetary cuts. This bias was reversed after 1988, with an
44 For details on methodology and sources of information for this section, see Scott (2009a). 20
increasing reallocation of educational spending towards basic education. 45 Between 1992 and 2002, spending per student in tertiary education expanded in real terms by only 7.5 percent. In contrast, spending per student on primary education increased by 63 percent. The relative ratio of spending per student in tertiary versus primary education thus declined from a historical maximum of 12 in 19831988, to less than 6 in 19942000 (by comparison, the average ratio for high-income OECD countries is close to 2). 46 In addition to the budgetary allocation between educational levels, progressivity in educational spending was constrained by the limited use of post-primary public education services by the poor, even when these were fully subsidized. This is explained by supply (limited availability of secondary schools in rural areas) as well as demand constraints (high opportunity cost of attending school for basic education in poor rural households). Both of these factors were addressed in the 1990s through the expansion of basic education facilities and, most notably, the conditional cash transfer programme Progresa (called Oportunidades since 2000) launched in 1997. Progresa/Oportunidades tied monetary transfers to keeping children of poor households in school and to receiving basic health services. 47 Numerous impact evaluation studies have shown the programmes positive and statistically significant impact on poor childrens education and health. Progresa/Oportunidades and its impact are discussed in detail in section 6 below. 48 The effect of these reforms may be observed in figure 8, which presents the distribution of benefits from different levels of public education received by population deciles (ranked by per capita household income). Figure 8 also shows the distribution of total education spending, comparing the distributions for 1992 and 2006. Combining the budgetary and participation effects that is, combining the fact that coverage became more targeted to the poor and that more of the budget was allocated to these categories and programs the distribution of total public spending on education has changed qualitatively over the decade, from (mildly) regressive to progressive in absolute terms, 49 with the poorest decile obtaining a share of educational spending (12 percent) twice as large as that of the richest decile (6 percent). 50
45 Aspe and Beristin (1984) found that public spending in education was quite inequitable before the changes that started in 1988 (p. 323). 46 See OECD (2008). 47 For a description of Progresa/Oportunidades see, for example, Levy (2006). 48 See, for example, Parker (2005) and Schultz (2000) and references cited in section 6. 49 Progressivity in absolute terms means that the poor receive a disproportional share of transfers, that is, the x percent poorest population receives more than x percent of transfers, while progressivity in relative terms means that the transfers received by the poor are higher as a share of their pre-transfer income than those received by the rich. We follow a common, if somewhat confusing, practice below in using the term progressive/regressive without qualification, to mean progressive in absolute terms in the case of spending, and in relative terms in the case of taxes. 50 In order to estimate the effect of transfers in kind such as public spending on education, Scott (2009b) applies a benefit incidence analysis based on the use of public services reported in ENIGH, valued at cost of provision. This imputed distribution of transfers received, valued in monetary terms, is then used to obtain an estimate of the (monetary and in kind) post-transfer Gini coefficient, and thus by comparing it to the pre-transfer Gini of the total distributional impact of all transfers. These imputations augment the concept of non-monetary income reported in ENIGH, and differ from the non-monetary concepts already included in the latter (notably imputed owners occupied housing rent) because of the method of valuation used to obtain the relevant 21
Figure 8, Distribution of benefits of public spending on education by income decile, 1992 and 2006 (percent)
Source: Scott (2009b). Note: Population deciles ordered by pretransfer household income per capita.
monetary values: cost of provision (in kind public services) versus self-reported valuation (imputed rent). It would of course be possible to apply a standard decomposition analysis on the (post-transfer) Gini coefficient obtained from this augmented concept of total income, thus allowing in principle a direct and full comparison of the effects on inequality and its evolution of private vs. public income sources. Given the noted measurement issues, however, such a direct comparison would have to be interpreted with care. This would also imply a substantial revision of the (total income) Gini coefficients reported for countries. The present analysis reports the estimated effect of these transfers on the Gini coefficient in purely accounting terms, following common practice in benefit incidence analysis. 22
Spending on all levels has become more progressive (or, in some cases, less regressive), but the most important change is observed in the case of lower secondary education. This is explained by at least three factors: (1) most importantly, the dynamics of educational expansion: as full coverage of primary education of the relevant age group was achieved by the early 1990s, even among the poor, these cohorts were at least formally qualified to access the next level; (2) the conditional scholarships of Progresa/Oportunidades, with increasing payments to lower secondary students and upper secondary education (since 2001); and, less encouragingly, (3) public education at the basic level (and higher secondary education from the 7th decile) being progressive in part because higher-income groups opt out of public schools. This is because private schools are perceived to be of better quality (as is confirmed by standardized evaluation surveys). In other words, public spending on basic education is progressive in part because it is self-targeted through low quality. An immediate corollary is that efforts to improve the quality of public education would, if successful, necessarily be so at the cost of equity. Access to tertiary education, on the other hand, is still highly regressive, improving only slightly since 1992. The participation of the poorest quintile is insignificant, and among the lowest in the region (Scott 2002b). As in the case of secondary education, this is slowly improving and should be expected to increase in the future simply as a consequence of advancing coverage in the earlier educational levels. But there are two further constraints explaining the failure to reach the poor that will require decisive policy reforms to make this potential demand effective. First, the high opportunity cost of tertiary education will require a reform in university financing. One reform will be targeting public subsidies to the poor through scholarships or educational credits, rather than simply offering free tuition to middle- and upper-income groups. But not everything is a problem of money. The poor are not able to attend public universities because of the low quality of their pre-university education, given that they have to compete for scarce university places with students who graduated from private high schools. So, if we want to improve equity of access at the tertiary level, the second area of attention needs to be improving the quality of upper secondary education opportunities for the poor. It is also important to note that health might have also contributed to a change in the relative returns to skills. Between the mid-1990s and the mid-2000s, health spending and spending on food subsidies and nutrition became considerably progressive (figure 9). Government policy focused both on expanding the supply of health services and the demand for health services on the part of the poor, the latter through the programme Progresa/ Oportunidades (discussed below). 51 More equitable access to health services might have contributed to improving the productivity of low-skilled workers: for example, improved access to health services may have translated into fewer days of work missed due to illness. Better access to health services might have also improved cognitive development of children in poor households, thereby improving their educational achievement and productivity. 52
51 Scott (2009b). 52 See, for example, Lustig (2006, 2007a). 23
6.
Pro-poor government spending: The distributional impact of Progresa/Oportunidades
As discussed above, in addition to changes in the distribution of labour income, transfers in particular, government transfers became an important equalizing force from 2000 to 2006, compared to the earlier years. Transfers equalizing contribution has been increasing over time both at the national level as well as for the urban and rural areas (table 3). 53 According to the decomposition results presented in table 3, transfers became the income source with the largest equalizing effect of all the sources considered in this exercise: that is, a marginal increase in transfers would reduce inequality by more than a marginal increase in labour income or remittances. Transfers became more equalizing over time because their share in total income rose (mostly because of a significant increase in coverage) and their own inequality and Gini correlation with total monetary income fell. These changes were particularly pronounced in the rural areas. The sharp rise in the role and equalizing impact of government transfers was a consequence of a significant policy shift in 1997, when the government launched the conditional cash transfer programme Progresa (whose name was changed to Oportunidades in 2000). 54 Progresa/Oportunidades is an innovative federal government programme that targets rural and urban households in Mexico that fall within the extreme poverty category. 55 The programme transfers cash to beneficiary households provided they take their children for periodic visits to health centres and send them to school. The logic behind the cash transfer is to compensate for credit market failures and/or parenting failures. 56 Progresa/Oportunidades has three components: education, nutrition and health. The education component grants cash transfers based on school attendance and high school completion, and for school supplies. The nutrition component offers cash or in-kind transfers (nutritional supplements, vaccinations, preventative treatments, etc.), based on regular visits to a health clinic. The average monthly transfer is about $35, and estimated total transfers are equivalent to, on average, 25 percent of eligible rural households average monthly income. 57 By the end of 2005, Progresa/Oportundiades granted benefits to 5 million families (about 24 percent of the Mexican population). The budget for Progresa/Oportunidades in 2005 equaled
53 Transfers here include government transfers and private transfers excluding pensions and remittances. Thus, it is not equivalent to government transfers only. 54 For a detailed analysis of Programa de Educacin, Salud y Alimentacin (Progresa) see, for example, Levy (2006). 55 This section is based on Lustig (2007). 56 Making the transfers conditional to a certain behaviour has the purpose of ensuring that additional resources are really used by beneficiary households to improve childrens levels of health, nutrition and education, thereby breaking intergenerational poverty traps. The conditionality logic resides in the preoccupation that parents, because of ignorance or indifference, are not good agents for the principal (their children), and (without conditionality) may devote the transfers to expenditures that do not have an impact in improving their childrens human capital. 57 This figure of $35 can increase in families with school-age children. In 2005, Progresa/Oportunidades granted money and species equivalent to direct monetary monthly assistance of $44.30 per family. Source: Levy (2006). 24
0.36 percent of GDP (compared with 0.02 percent in 1997), and it commanded 2.29 percent of the programmable public expenditure budget. 58 Numerous studies have estimated that, among other dimensions, the programme has had significant impacts on health and education. In offering beneficiaries scholarships to attend post-primary school, Progresa/Oportunidades addresses a principal restriction to access to post-primary education among the rural poor. Comparing enrollment before the programme (1996-97) and after the programme (2002-03), an average increase of 24 percent was found in rural areas. 59 Of note was enrollment in secondary education in rural areas, which rose by 11 percent for girls and 7.5 percent for boys after two years of launching the programme. 60 By providing demand-side subsidies and empowering public service users, rather than providers, it complemented traditional supply-side spending in health and education with demand-side subsidies. In fact, demand for health services among beneficiaries was 67 percent higher than demand in communities not participating in the programme. 61 Infant mortality was found to fall at a rate 11 percent higher among beneficiaries compared to non-beneficiaries. 62 Another study estimated a reduction of 11 percent in maternal mortality and a 2 percent reduction in infant mortality in rural communities compared to those not in the programme. 63 Likewise, many studies show that Progresa/Oportunidades has had a positive impact on household consumption, savings and investments, which helped reduce poverty and inequality in Mexico. For 2004, the poverty incidence of programme participants (percentage of population associated with the programme that is below the poverty line) fell by 9.7 percent in rural areas and 2.6 percent in urban areas. In the same year, there was a reduction in the poverty gap (18.7 percent) and a reduction in the severity of poverty (28.7 percent) in rural areas, 64 as well as reductions in the urban poverty gap (4.9 percent) and urban poverty severity (1.7 percent). 65 In terms of its impact on the distribution of income, table 5 shows that the direct effect of Progresa/Oportunidades transfers reduced the Gini coefficient from 0.502 to 0.494, which is the equivalent of close to one fifth of the decline experienced in the Gini coefficient between 1996 and 2006 (table 8). 66
58 Although the Social Development Secretariat is in charge of the programme, most of the health budget, mainly associated to the provision of health services, is part of the conventional definition used in health public expenditure. 59 Parker (2005). 60 Schultz (2000). 61 Bautista et al., (2004). 62 Barham, (2005). 63 Hernndez et al., (2003). 64 Poverty gap measures the mean distance of the poor with respect to the programmes poverty line, and reflects a worsening or improvement in the conditions of the poor. On the other hand, povertys severity measures the average of the squared distances of the poor with respect to the poverty line and reflects inequality among the poor population. 65 Cortes and Banegas 2006. 66 The impact on the Gini coefficient takes account only the direct effect. The effects on inequality of changes in behavior or through higher human capital among the poor are not contemplated in this calculation. 25
Politically, Progresa/Oportunidades set new standards for social policy in Mexico. 67 It is the first social programme in Mexico to apply transparent targeting mechanisms; it effectively identifies the poorest rural localities and households, using, at the latter level, proxy-means tests based on a full census of socioeconomic characteristics and economic assets within these localities. Progresa/Oportunidades is also the first social programme in Mexico to be subject to rigorous impact evaluations which were planned from the beginning of the programme. 68 The programme is also notable in having survived not only a change of administration (no other major anti-poverty initiative over the past two decades has done this), but also in having survived the first change in 70 years of the political party in power. In fact, rather than discard the programme, the new partys administration changed its name from Progresa to Oportundiades, and starting in 2001 the new government increased coverage from 2.3 to 4.2 million households (mainly in rural areas), and added semi-urban and urban localities to the already established rural ones.
Table 5, Direct distributive impact of targeted monetary transfers: Total and Progresa/Oportunidades, 2006 (percent)
Deciles 1 2 3 4 5 6 7 8 9 10 Gini Change in Gini All Monetary Targeted 1.8 2.7 3.6 4.4 5.3 6.7 8.2 10.7 15.7 41.0 49.2 -2.1 Progresa / Oportunidades 1.7 2.7 3.5 4.4 5.3 6.7 8.1 10.7 15.7 41.1 49.4 -1.7 Primary or Market Income 1.5 2.5 3.4 4.3 5.3 6.7 8.2 10.8 15.8 41.4 50.2
Source: Authors elaboration based on ENIGH (2006). Note: 1. Distribution of total primary current income per capita before and after targeted monetary transfers. For more details, see Scott (2005). 2. The programmes in this table refer to monetary transfers only, corresponding to the first item under D in table 6.
Progresa/Oportunidades also marked a significant shift in the distribution of food subsidies in Mexico. As shown in figure 9, the reallocation of food subsidies through Progresa/ Oportunidades transformed the broadly neutral distribution of government spending in food subsidies in Mexico into a highly progressive one: with the share benefiting the poorest decile increasing from 8 to 33 percent between 1994 and 2000. Several factors may explain
67 This section is based on Scott 2006. 68 See, for example, www.ifpri.prg for the principal evaluation results. Also, see Parker (2005) and Schultz (2000). 26
this shift. First, the discovery of a redistribution technology: the piloting of Progresa in 1997 showed that distributing food subsidies (and cash transfers) to poor families in remote rural localities was operationally feasible. Second, the empowerment of rural voters as a consequence of Mexicos democratization process gave them potentially more voice in the allocation of government resources. Third, unlike health and education, food subsidies are not under the purview of Mexicos powerful public sector unions, which have successfully blocked pending reforms in the case of education and health services. Finally, in contrast to education and health service provision, the reallocation of food subsidies is not constrained by labour shortages (lack of qualified doctors and teachers in poor rural areas) or physical infrastructure. 69
Figure 9, Distribution of public spending on food subsidies, Procampo and health services for the uninsured by income decile, 1990s2000s (percent)
Source: Scott (2003). Note: 1. Food subsidies: population deciles ordered by pretransfer household income per capita. 2. Users of health services: population deciles ordered by pretransfer household income per capita. The use of health services is measured in millions of users as a percentage of the corresponding decile.
The political economy of Progresa/Oportunidades survival through various administrations is also a result of a perfect combination of factors that merit note. In contrast to the intensive government media campaigns accompanying most flagship anti-poverty programs in Mexico, the government refrained from such a campaign in the case of Progresa, which facilitated its political survival beyond the administration and the PRI regime. 70 Second, the rapid expansion of the programme (with 2.5 million direct beneficiary households by the end of the 2000) ensured support for it from a large constituency. Third, just as importantly, the decision to make the programme transparent and invest in ambitious and highly credible external impact evaluations also contributed significantly to the political survival of Progresa/
69 For a discussion of the political economy of this reform see Scott (2009b). 70 The PRI (Partido Revolucionario Institucional, Revolutionary Institutional Party) regime refers to the (almost) one-party regime that characterized Mexico for 70 years. 27
Oportunidades. However, it was not completely immune to damaging political economy issues. In 2001, when it was rebranded as Oportunidades, the programme was expanded to urban areas and upper secondary education, inevitably reducing its targeting efficiency from 71 percent of its resources benefiting the poorest quintile in 2000, to 55 percent in 2006. In sum, Progresa/Oportunidades is an example of redistributive efficiency. With as little as 0.36 percent of GDP and 4 percent of redistributive spending (table 7), the programme accounts for 18 percent of the change in the post-transfers Gini, and 81 percent of the change in the Gini after including programmes targeted to the poor (table 8). 71 If one compares the redistributive impact of this programme with the redistributive impact of all monetary transfers and subsidies, the impacts are equal. This implies that the progressive incidence of the remaining transfers targeted to the poor and other spending categories is wiped out by the regressive effect of all other untargeted monetary transfers and subsidies. Thus, unfortunately, the redistributive efficiency of Progresa/Oportunidades is an isolated case among redistributive instruments currently operating in Mexico. Even though government spending became undoubtedly more pro-poor during the last decade, a lot still needs to be changed to make redistributive spending a powerful instrument to reduce inequality in Mexico. This will be the topic of the next section. In what follows, we will analyze the incidence of all redistributive spending and identify which spending categories must be changed if fiscal resources are to be used more progressively.
7.
The distributional impact of state action: Government spending and taxes
Since the early to mid-1990s, social spending, and particularly spending on programs targeted to the poor, has expanded considerably (figures 10 and 11). As discussed above, and shown in figures 8 and 9, spending on education, health and nutrition became more progressive. In addition, with the introduction of Progresa/Oportunidades, the Mexican government found ways to redistribute income through transfers in an efficient and cost-effective way. Despite this progress, 58 percent of government redistributive spending in 2006 was regressive (of which 11 percent increased income inequality). In advanced countries, government transfers are able to reduce primary (i.e., market) income inequality by 30 to 50 percent. In contrast, in Mexico, the post-transfers Gini coefficient in 2006 was only 9.3 percent and a meager 1.7 percent if we exclude transfers in-kind lower than the primary income pre-transfers Gini.
71 The figure of 18 percent is obtained by dividing the percentage change in the Gini due to Progresa/ Oportunidades in table 5 (-1.7 percent) by the percentage change in the Gini due to transfers shown in table 8 (-9.3 percent). 28
Figure 10, Evolution of social spending, 19252006 (percent)
Source: Scott (2009b). Note: Programmable spending excludes debt servicing.
Figure 11, Evolution of extreme poverty, anti-poverty spending, GDP and Gini (average annual change in percent)
Note: 1. Poverty is measured using the official extreme (pobreza alimentaria or food poverty) poverty line. 2. Poverty programmes for 20002006 are those listed under D in table 6. For 19891994, the list was different; in particular, it included Pronasol and excluded Progresa/Oportunidades (launched in 1997).
Estimating the redistributive impact of government spending faces considerable methodological challenges. Most income and expenditure surveys report government monetary transfers as a component of household income. The contribution of these transfers to overall income inequality can then be estimated through the application of standard decomposition techniques on inequality measures by income source. This is the path followed above. However,
29
while in mature welfare states, monetary transfers represent between a third and a half of total social spending, and account for reductions in inequality on the order of 20 to 50 percent in developing countries monetary transfers are a small part of total government transfers, and in the case of Latin America account for reductions in inequality of around 2 percent. 72 The largest share of redistributive spending in Latin America occurs through government transfers in-kind, which are not included, in general, in the income concept measured in household surveys. In Mexico, the two cash transfers programmes reported in the ENIGH survey (Oportunidades and Procampo) represent a mere 5 percent of the portion of public spending devoted to redistributive objectives (table 6). 73 The remaining 95 percent of government transfers is not covered by the Gini decomposition analysis presented above. This means that standard analysis of inequality dynamics excludes a significant portion of a true measure of income, which, if measured correctly, would have to include all government transfers. To illustrate this, in 2006, average household income increased by 4.1 percent when all monetary transfers and subsidies were included, and increased by 12.8 percent when in-kind transfers (education and health) were added to the latter. Table 6 presents the 2006 totals of government transfers in cash and in kind as well as in the form of subsidies (including value added tax [VAT] exemptions or fiscal spending) for the main redistributive categories and programmes. 74 The programmes are described in table 7. The categories and programmes covered here include all public education and health spending (at the federal and state levels), as well as all federal public spending on pensions (the subsidized component), energy and agricultural subsidies, and the main programmes targeted to the poor. Altogether, these items comprise a total of 25 programmes or specific spending categories which, in 2006, amounted to close to $100 billion, or 10 percent of Mexicos GDP, 12.8 percent of primary (before taxes and transfers, also called pre-fiscal) household income, 60 percent of total public spending (the remainder includes spending on administration, defence, etc.) and 97 percent of total social spending. Using ENIGH 2006, we analyze their distributive impact using standard benefit incidence analysis. 75
72 These reductions in mature welfare states are measured in purely accounting terms: i.e., as the difference between the pre-transfer Gini and post-transfer Gini without taking into account any behavioural responses. See Ervik (1998), Smeeding and Ross (1999). 73 We put redistributive in quotation marks because not all these transfers are progressive. 74 In-kind transfers, which include mainly education and health services, may be highly valued by some (though not all) beneficiaries. They are non-tradable, highly labour intensive, depend on locally available infrastructure for access, and are thus highly variable in quality. Thus, one relevant implication for the present analysis is that the gap between the public cost of the transfer, and the monetary benefits to recipients, is likely to be larger for in-kind transfers than for monetary transfers or direct subsidies. Another is that the low quality of services may act as an implicit, but effective, self-selection targeting mechanism. 75 Public spending data on all programs reported in table 6 are obtained from the Public Accounts of the Federation for the relevant years. For health spending at the state level, we use the National and State Health Accounts, published by the Health Ministry. Education spending at the state level is estimated from federal per student spending rates and the total of students in schools financed by states as reported by the Education Ministry. Spending on education and health at the state levels includes spending on education and health financed through earmarked federal transfers, revenues from federal revenue sharing programs and fiscal revenues collected by the states themselves. 30
Table 6, Government redistributive spending by category and programme, 2006

Concept
Household income before taxes and transfers Taxes (including social security contributions) Social Spending (official classification) Total Transfers Analized, of which Transfers in kind (lines A+B) Other Transfers* (lines C+D) Memo item: Total untargeted to the poor Total targeted to the poor A. Education Preschool Primary Lower Secondary Upper Secondary Tertiary B. Health Services for Insured Services for Uninsured (SSA), of which Seguro Popular IMSS-Oportunidades C. Subsidies* Pensions** , of which IMSS ISSSTE Consumption Subsidies, of which Residential Areas Electricity Subsidy Gasoline and other fuels subsidy Memo item: Consumption Subsidies ***, of which VAT exemptions on food and prescription drugs Agricultural Subsidies, of which Procampo D. Programs Targeted to the Poor, of which Targeted monetary (mainly), of which Oportunidades Vivienda (Tu Casa) DIF/ Feeding Programs Liconsa Programa de Empleo Temporal (PET) Opciones Productivas Other Scholarships (excl. Oportunidades) Piso Firme Hbitat Microrregiones
Millions Current Mexican Pesos

7,521,608 1,038,283 988,369 964,567 654,675 309,892 916,696 47,871 402,385 44,583 135,352 86,817 52,932 82,701 252,290 159,986 92,304 11,700 5,716 262,021 85,230 50,004 35,226 107,153 64,935 42,218 270,102 162,949 69,638 15,025 47,871 45,179 33,526 4,234 2,806 1,300 1,090 401 1,571 251 1,992 700
Total Transfers (%)
Household inc. before taxes and transf. (%)

100 13.8 13.1 12.8 8.7 4.1 12.1 0.6 5.3
100 68 32 95 5 42 5 14 9 5 9 26 16 10 1 1 27 9 5 4 11 7 4
3.4
3.5 1.1
1.4
3.6 7 2 5 4 3 0.4 0.3 0.1 0.1 0.0 0.2 0.0 0.2 0.1 0.9 0.6
Source: Scott (2009b). See description of programmes in table 7. Note: 1. * Excludes VAT (value added tax) exemptions (also known as fiscal spending). It also includes subsidies and monetary transfers to agriculture such as Procampo and Ingreso Objetivo. 2. ** Government subsidy, that is, pensions net of contributions from workers and employers. 3. *** Includes VAT exemptions (also known as fiscal spending).
31
Table 7, Description of main categories and programmes

Programme Health and Pensions (Social Security)
IMSS ISSSTE Health Services SSA Seguro Popular IMSS-Oportunidades
Description
The principal social security institution providing health services (and other benefits) to private sector workers, created in 1943. Its pension system was reformed in 1997 to a fully funded, defined contributions scheme administered by private entities. The principal social security system covering federal and state government employees. Health services for the uninsured provided by the state and federal government Health Ministries. Health programme created in 2004 to provide basic health insurance to the (formally) uninsured, with a projected coverage of 12 million households by 2010. System of rural health clinics for the uninsured targeted at poorer localities, administered by IMSS but funded by the Health Ministry, created in 1970's as IMSSCoplamar. Special tax on gasoline (IEPS) became a negative tax, or a subsidy, since 2006, as price adjustments lagged the increase of international oil prices. Fiscal spending principally associated with VAT exemptions on food and prescription drugs. Agricultural subsidy created in 1994 to compensate producers of basic crops for the opening up of agricultural markets under the North American Free Trade Agreement. Provides a direct transfer per hectare. Administered by the Agricultural Ministry. Agricultural subsidy based on a target price. Administered by the Agricultural Ministry. Conditional cash transfer programme created in 1997, currently covering 5 million households, providing direct monetary transfers conditional on school attendance and use of health services. Originally targeted to poor rural communities, and for basic education in 2001. It was gradually extended to urban localities and for uppersecondary education. Administered by the Social Development Ministry. Housing credit programme targeted to the uninsured. Administered by the Social Development Ministry. School breakfast and kitchen programme. Targeted milk subsidy programme, providing rations of milk at half price in urban localities. Administered by the Social Development Ministry. Workfare programme created in 1995, providing a maximum of 88 days of work for low wages (originally 90 percent of the minimum wage, at present 99 percent). Support for productive projects: organization, technical assistance, credit. Administered by the Social Development Ministry. Universal rural non-contributive basic pension for people 70 years or older created in 2007, with a $ 500 pesos (37 US dollars) benefit per month. This category refers to the scholarship income reported by households in the ENIGH income and expenditure survey, excluding Oportunidades scholarships. Provides a cement finance to purchase material inputs to built cement floors for houses with dirt floors in poor rural localities. Urban development, infrastructure and participation programme. Social Development Ministry. Rural development programme providing infrastructure and productive inputs to promote local economic development. Administered by the Social Development Ministry.
Consumption Subsidies
Gasoline and other fuels subsidies VAT exemptions on food and prescription drugs
Agricultural Subsidies
Procampo Ingreso Objetivo
Programs Targeted to the Poor
Progresa/Oportunidades
Vivienda (Tu Casa) DIF feeding programmes Liconsa Programa de Empleo Temporal (PET) Opciones Productivas Rural Old Age Pension Other Scholarships (excl. Oportunidades) Piso Firme Hbitat Microrregiones
Source: Scott (2009a).
32
It is important to note that social spending as defined in official government budgetary classifications and the concept of redistributive spending used in this incidence analysis are not the same. The fact that, in 2006, they are so close in actual amounts is a coincidence. In addition to spending on education, health and programmes targeted to the poor, redistributive spending includes spending on agricultural and consumption subsidies (items excluded from social spending). In the case of pensions for social security beneficiaries, redistributive spending just includes the subsidy component (i.e., it excludes the contributions made by workers and employers) whereas social spending includes all spending on social security pensions regardless of how they are financed. Finally, redistributive spending excludes some of the spending items included in social spending because they cannot be mapped to specific households. This includes social spending on public goods such as rural roads, for example. If we disaggregate the total redistributive spending presented in table 6 by category, inkind transfers of education and health represent 42 and 26 percent of the total, respectively, and together they represent 68 percent of the total. The distribution of the remaining 32 percent is as follows 76: 11 percent on general subsidies on electricity for residential areas and gasoline and other fuels, 9 percent on pensions (the subsidized component), 77 7 percent on agricultural subsidies, and 5 percent on monetary transfers targeted to the poor. 78 In order to estimate the incidence of redistributive spending, first the distribution of total spending on each programme or category by population deciles ordered by total current income per capita before taxes and transfers (primary or pre-fiscal income) is recorded; then, the degree of progressivity is measured using concentration coefficients (CC). 79 The CCs are a Gini
76 The benefit incidence analysis does not include VAT exemptions on the spending side. They are included in the analysis of tax incidence. 77 Although pensions are reported in the ENIGH survey, the totals cannot be treated as government transfers. This is because ENIGH data does not allow disaggregation between public pensions from private ones, or identifying the subsidy (vs. saving) component in the public pensions. Pension income must therefore be decomposed as a separate income component and, as shown in the decomposition of inequality by income source above, it is unequalizing. The incidence analysis presented includes only the tax-financed subsidies to the public pension systems, and uses information on social security affiliation reported in the ENIGH (for active workers only) to impute them. 78 Item D in table 6. Although some of the transfers (Vivienda, DIF desayunos, Liconsa, Programa de Empleo Temporal, Opciones Productivas, Piso Firme, Hbitat, Microrregiones) are not reported in the ENIGH published database, they were included in a special module of social programs commissioned with ENIGH (2002, 2004, 2006) by Sedesol. Thus we were able to include them in the incidence analysis presented here. 79 For details, see Scott (2009a, 2005). The main data source for the analysis is the 2006 ENIGH household income and expenditure survey. In addition to being the most detailed source for household income available, the ENIGH survey reports the main public monetary transfers (Progresa/Oportunidades and Procampo), the use of public education and health services by household members, whether household members contribute to the social security systems, spending on electricity and gasoline and other fuels, and on consumption goods that are subject to VAT (or exempted from it). As is common in household surveys, total household income in ENIGH tends to be underreported by a large margin when compared to the closest equivalent concept of total household income in Mexicos National Accounts (the ratio of total household income in National Accounts to total household income in the survey was equal to 1.87 in 2006). To estimate the incidence and redistributive effect of public transfers it is necessary to ensure comparability between public transfers obtained from the National Accounts and private income reported in ENIGH. To do this, the data in the National Accounts is adjusted 33
coefficient of the distribution of transfers (or taxes, depending on the case) with the population ranked according to total pre-transfers current income per capita. 80 A negative CC means that the category or programme is progressive in absolute terms: that is, the poor receive more than the rich in per capita terms. A positive CC, which is lower than the Gini of the primary (pre-fiscal) income, means that the category or programme is regressive in absolute terms: that is, the poor receive more than the rich in proportion to their respective incomes but less in per capita terms (this is sometimes called progressive in relative terms). A positive CC that is higher than the Gini of the primary (pre-fiscal) income means that the category or programme is regressive in relative (and thus, absolute) terms: that is, the post-programme distribution is more unequal than the distribution of primary income. By definition, if a programme is regressive in relative terms, it must be regressive in absolute terms but the converse is not true. The CCs of redistributive spending by programme or spending category are shown in figure 12. The CCs range from the most progressive programme in absolute terms, Oportunidades (-0.53) to the most regressive (in relative terms) programme Ingreso Objetivo (0.81). 81 Most of the programmes targeted to the poor (item D in table 6), the pro-poor health insurance programme Seguro Popular, health services provided by the Ministry of Health for the uninsured, and basic education (pre-school, primary and lower secondary) are progressive (pro-poor) in absolute terms. The programmes that make the post-transfer distribution worse than the pre-fiscal distribution are primarily the cash transfer programs to agricultural producers (for example, Ingreso Objetivo) and subsidized pensions to employees of some state-owned companies. In between, there is a large number of programmes and categories which are regressive in absolute terms (but progressive in relative terms): for example, energy and other generalized consumption subsidies (gasoline, LP gas, residential electricity, VAT exemptions), social security benefits, and tertiary education.
proportionately by the ratio described above to ensure consistency with household income estimated from ENIGH surveys and the totals in the National Accounts. With the exception of Oportunidades, which is reported in the ENIGH survey, the information for the targeted programs included in the analysis is obtained from a special module on social programmes commissioned by the Social Development Ministry as part of the 2006 ENIGH. The distribution of agricultural public expenditures, and in particular Procampo and Ingreso Objetivo, are obtained from the administrative beneficiary data base, and reported as producer deciles ordered by the size of land holdings. The inclusion of the latter results with the ENIGH-based estimates is justified on the assumption that the size of land holdings is positively correlated with income. The only agricultural subsidy reported in ENIGH is Procampo, but the survey is not designed to report the distribution of this programme accurately. A large fraction of Procampos benefits are concentrated on a small group of producers at the top end of the land and income distribution. The ENIGH survey is particularly limited in capturing income at the top end of the distribution, for well-known reasons of small samples and problems of underreporting (see footnote 4, above). Therefore, it significantly underestimates the concentration of Procampo transfers. 80 CCs are also sometimes called quasi-Ginis. They are calculated based on the transfers received by population decile. CCs are also calculated for taxes. 81 CCs can range from -1 to 1. The closer they are to -1 (1) means that the fiscal resources are allocated in a propoor (pro-rich) way. 34
Figure 12, Concentration coefficients for redistributive public expenditure, 2006
Source: US: Piketty and Saez, 2003; France: Piketty, 2001.
CCs, by definition, do not capture nuances in the actual distribution of benefits. A more complete picture can be obtained by looking at the benefit incidence by deciles or quintiles. Scott (2009b) finds that 16 programs transfer an amount at least equal to their population share to the bottom quintile that is, they are progressive in absolute terms (fig. 12). At the other end are 12 programs that are regressive in absolute terms and 5 that are regressive in relative terms (i.e., their share is lower for the bottom quintile in the post-programme distribution compared with the bottom quintiles share of primary income). 82 Programmes that are regressive in relative terms do not reach the poor and, in effect, worsen income inequality. Only 42 percent of all the redistributive spending categories (monetary transfers, subsidies and in-kind transfers) in 2006 were progressive in absolute terms (i.e., the poor received more than the rich in per capita terms). 83 Of the remaining 58 percent of redistributive spending, 47 percent was regressive in absolute terms and 11 percent was allocated to programs that were regressive in relative terms that is, the distribution of income worsened as a result of these programmes.
82 Procampo, strictly speaking, is probably not regressive in relative terms as discussed here, which would bring the total of regressive in relative terms to three instead of four. 83 Subsidies includes general subsidies to gasoline and other fuels, subsidies and monetary transfers to agricultural producers and the subsidized part of social security pensions. 35
The overall impact of total redistributive spending can be seen in table 8. Given that regressive programs use a larger share of fiscal resources, the former effectively cancel the propoor impact of the progressive ones, producing a slightly regressive (in absolute terms) global distribution of public spending. In-kind transfers are generally neutral (in absolute terms), monetary transfers targeted to the poor are progressive in absolute terms and the rest of the transfers are either regressive in absolute or relative terms (fig. 13). However, because the bulk of government redistributive spending (89 percent) is allocated to programmes that are either progressive in absolute terms (42 percent) or to programmes that, even though regressive in absolute terms, make the post-transfer distribution less unequal than the pre-transfer one (47 percent), government redistributive spending reduced the pre-fiscal Gini coefficient by 9.3 percent in 2006. 84 If we exclude the impact of transfers in kind, the reduction in the Gini coefficient becomes much smaller: a mere 1.7 percent.
Figure 13, Distribution of government transfers by decile, 2006
Sources: Alvaredo, 2009, Atkinson, 2007, Atkinson and Leigh 2007a,b, Nolan, 2007, Piketty and Saez, 2003, Alvaredo and Pisano, 2010, Alvaredo and Saez, 2009, Dell, 2007, Dell, Piketty and Saez, 2007, Moriguchi and Saez, 2008, Piketty, 2001, Roine and Walderstrm, 2008, Saez and Veall, 2005, Salverda and Atkinson, 2007.
Let us now turn to the redistributive impact of taxes (including social security contributions). Table 8 presents the estimates of tax incidence for 2006 calculated by the Finance Ministry (SHCP, 2008) used to obtain net benefits. 85 The incidence of taxes is progressive but only mildly so: taxes paid by the poorest decile are equal to 6.7 percent of their pre-fiscal income while the figure for the top decile is 13.9 percent. Taxes reduce the pre-fiscal Gini by 2.8 percent. The combined effect of all government redistributive spending and taxes reduces the Gini from
84 Since this estimate includes the valuation of social services based on government spending (transfers in kind in education and health), the redistributive impact must be viewed as an upper bound that would have to be adjusted downwards to the extent that the value of services received is less than the budgetary cost of provision. 85 The tax incidence analysis uses the same database (ENIGH 2006) and applies the same methodology as the spending incidence analysis reported here, so the tax and transfer incidence estimates are comparable. 36
0.5024 to 0.4387, or by 12.6 percent. If we exclude the impact of in-kind transfers and consider the effect of monetary transfers and subsidies only plus the effect of taxes, the Gini falls from 0.5024 to 0.4794, or by 4.6 percent. In sum, although government redistributive spending has become more progressive over the last decade, and although both spending and taxes reduce the pre-fiscal inequality, there still is a lot of room for improvement in making state action more progressive on the fiscal front.
Table 8, Incidence and redistributive impact of transfers and taxes, 2006 (percent)
Distribution (as a proportion of categorys total) Household income Post monet. transf. and subsid. & taxes (7) Post all transf. (col 5) & taxes (8) Transfers (1) Post- monet. transf., subsid., and tranfers in kind (5) Pre-transf. & taxes (3) Post-monet. transf. and subsid. (4) Post-tax (6) Transfers (9) Taxes (2) Tax (10) -6.7 -5.9 -6.9 -7.6 -7.5 -8.7 -8.5 -10.2 -11.5 -13.9 -11.1 Incidence (as a proportion of pre-fiscal household income) Net (after taxes and transf.) (11) 68 37.8 25.6 19 14.4 9.8 7.9 3.4 -1.6 -9 37
Deciles
1 2 3 4 5 6 7 8 9 10 Total CC/G * Percentage Change in G
8.2 0.9 8.4 1.3 8.4 2.1 8.7 3 8.7 3.5 9.3 5.2 10.1 6.2 11.1 9.9 11.9 16.4 15.1 51.6 100 100 0.1047 0.6132
1.5 2.5 3.4 4.3 5.3 6.7 8.2 10.8 15.8 41.4 100 0.5024
1.6 2.7 3.5 4.4 5.4 6.7 8.2 10.8 15.8 40.8 100 0.4937 -1.7
2.3 3.2 4 4.9 5.7 7 8.4 10.8 15.4 38.3 100 0.4558 -9.3
1.5 2.7 3.6 4.5 5.5 6.8 8.4 10.9 15.8 40.1 100 0.4885 -2.8
1.7 2.8 3.7 4.6 5.6 6.9 8.5 10.9 15.7 39.5 100 0.4794 -4.6
2.4 3.4 4.2 5.1 5.9 7.2 8.7 11 15.3 36.9 100 0.4387 -12.6
74.7 43.7 32.5 26.6 21.9 18.5 16.4 13.6 10 4.8 13.3
Source: Based on Scott (2009a). Note: 1. Transfers include all the categories described in table 6. In the case of pensions, it includes the subsidized component only. 2. Taxes include all taxes and contributions to social security. 3. Transfers in kind correspond to categories A and B in table 6. 4. Monetary transfers and subsidies correspond to categories C and D in table 6. 5. * not in percent. 6. CC=concentration coefficient; G=Gini coefficient. For a definition of CC see text
7. Conclusions
After more than a decade of rising income inequality between the early 1980s and early 1990s, Mexicos inequality finally began to decline in the post-NAFTA years. Between 1996 and 2006, the Gini coefficient for total income fell from 0.543 to 0.498, or at 0.8 per cent per year; the decline was higher, 1 percent a year, in the more recent period between 2000 and 2006. A more detailed analysis of income patterns reveals that during the past decade the bottom quintile experienced an above-average increase in its income while the top decile grew at belowaverage rates. The increase for the bottom quintile seems to be associated with income growth in rural areas. The lackluster growth of the top deciles income is associated with the dynamics of the urban sector of the economy. Therefore, from the mid-1990s to 2006, there must be factors that benefited the low end of rural households and hurt in relative terms the upper deciles in urban areas. The results of the decomposition exercises discussed in this paper suggest that the improvement in incomes at the bottom of the distribution is linked to higher relative wages of low-skilled workers, a rise in the share of remittances in rural areas and an expansion of government monetary transfers targeted to the poor. The reduction in the skill premium appears to explain the loss in the relative position of the upper deciles, particularly in urban areas. The reduction in the premium occurred because wages (or labour earnings, more precisely) had increased more for low-wage workers than for the rest. A preliminary exploration of the causes indicates that this flattening of the wage distribution coincided with a significant change in the composition of the labour force in terms of education. Between 1989 and 2006, the share of workers with less than lower secondary education fell from 55 percent to around 33 percent. This educational upgrading of the labour force is associated with important changes in government spending on education. First, since the end of the 1980s, overall education spending increased, with the largest increases in spending going to basic education. Between 1992 and 2002, spending per student in tertiary education expanded in real terms by 7.5 percent while it rose by 63 percent for primary education. The relative ratio of spending per student in tertiary versus primary education thus declined from a historical maximum of 12 in 19831988, to less than 6 in 19942000 (by comparison, the average ratio for high-income OECD countries is close to 2). In addition to the budgetary allocation between educational levels, progressivity in educational spending was improved by addressing two important constraints to access: limited availability of secondary schools in rural areas as well as education-demand constraints (high opportunity cost of attending school for lower secondary and secondary aged children in poor rural households). In the 1990s, the government expanded basic education facilities, particularly in rural areas. In addition, the government implemented the conditional cash transfer Progresa (launched in 1997), which made the cash transfers to poor households conditional on school enrollment and attendance, with emphasis on the lower secondary level. The combined effect of the changes in budgetary allocations and coverage at the post-primary level changed the distribution of total (federal and state) public spending on education from mildly regressive (in 1992) to progressive in absolute terms (in 2006). By 2006, the poorest decile received 12 percent of total education spending while the highest decile received 6 percent.
38
The persuasiveness of the labour-supply side of the story is strengthened by the fact that for workers whose share in the labour force declined (rose), wages tended to increase (decline) or increase more rapidly (slowly). Of course, an increase in the demand for lowskilled workers could have reinforced the supply-side changes in labour force composition. As mentioned in the paper, there are indications that the greater integration of production with the United States made Mexicos manufacturing workers complements rather than substitutes for U.S. workers. The greater integration also contributed to expanding assembly-line activities in Mexico, which increased the demand for low-skilled workers. In addition, it is possible that remittances and government transfers also increased demand for low-skilled workers, as recipients used the resources for home improvements or demanded more goods and services in local communities something that could have had a local multiplier effect. 86 Demographic changes in the proportion of adults per household, and working adults among households during 19962006 were also equalizing; they also raised the importance of demographic factors in accounting for the observed decline in inequality. One change in particular is worth noting: the relatively larger decline in average household size among the poor. Between 1996 and 2006, the average number of children per household in the lowest quintile fell from 2.3 to 1.7; whereas in the top quintile, it fell from 1.5 to 1.3. This led to an increase in the number of income earners per dependent and thus, by itself, would have led to an increase in household per capita income ceteris paribus. The decomposition of inequality by income source clearly showed that, at the national level, there were three inequality-increasing income sources: pensions, income from own businesses and income from property, and three equalizing income sources: labour income (since 2000), remittances and transfers. Labour income became a very important equalizing force in urban areas (the specific dynamics were summarized above), but not in the rural sector, where it was even inequality-increasing in 2006. In contrast, transfers became a very important inequality-reducing income source in rural areas and, to a lesser extent, in urban areas. The equalizing contribution of transfers increased over time (both at the national level as well as for urban and, especially, rural households). By 2006, transfers became the income source with the largest equalizing effect of all the income sources considered in the paper; of note, a marginal increase in transfers would reduce inequality by more than a marginal increase in labour income or remittances. The share of transfers in total income rose because there was a significant expansion in coverage of public transfers. 87 The sharp rise in the role and equalizing impact of public transfers was a consequence of a significant policy shift in 1997, when the government launched the
86 The actual picture may not be as positive. Income inequality in Mexico has been falling during a period of low growth. If low growth is a consequence of the fact that not many jobs are generated in the high productivity sectors and this is a consequence of the high costs faced by firms in the formal sector, then low growth and the fall in wage inequality might both be a result of the same undesirable process. Levy (2008) argues that this is in fact what has happened in Mexico, and that the main culprit is, in some aspects, Mexicos social policy which has created disincentives for employment generation in the formal sector. Pages et al., (2009) find evidence that job creation in Mexico and other Latin American countries occurred disproportionately in low-productivity and low-wage sectors. They call this growthless jobs. 87 Although the concept of transfers used in the decomposition analysis includes both private and public transfers, given that remittances are analyzed separately, they are probably mainly public transfers. 39
conditional cash transfer programme Progresa. During 19962006, the size of public transfers increased; they became more equally distributed among recipients, and the recipients of transfers were increasingly from the relatively poorer segments of the population. In spite of its relatively small use of fiscal resources and thanks to its targeting and administrative efficiency, Progresa/Oportunidades has had an important effect on equity in the use of government resources. The programmes resources represent a mere 0.36 percent of GDP and 4 percent of government redistributive spending. 88 Nonetheless, its direct impact (through monetary transfers) has been estimated to reduce the pre-fiscal (before transfers and taxes) Gini coefficient by 0.8 percent or the equivalent of 18 percent (close to one fifth) of the decline between 1996 and 2006. Progresa/Oportunidades essentially shifted resources that were used to subsidize food in the urban areas (Mexico City in particular) to the rural poor. In just half a decade, the reallocation from food subsidies to Progresa transformed an essentially neutral distribution of food subsidies into a highly progressive distribution of spending on nutrition and health, with the share of spending benefiting the poorest decile increasing from 8 to 33 percent between 1994 and 2000. In addition, through its positive impact on low secondary school enrollment and usage of public health services among the poor, the programme made government spending on education switch from mildly regressive to progressive in absolute terms, and increased the progressivity of government spending in health for the uninsured. As shown in the paper, since the mid-1990s, social spending and spending on programmes targeted to the poor has expanded considerably and government spending on education, health and nutrition has become more progressive. In addition, with the introduction of Progresa/Oportunidades, the Mexican government found ways to redistribute income through transfers in an efficient and cost effective way. However, despite this undeniable progress, when all government redistributive spending is considered and not just the portion of redistributive transfers captured by the income concept in household surveys which is, as mentioned above, a mere 5 percent of total redistributive government spending benefits incidence analysis shows that 58 percent of government redistributive spending is still regressive in absolute terms (of which 11 percent increased income inequality). In advanced countries, government transfers are able to reduce primary or market income inequality between 30 and 50 percent. In Mexico, in contrast, the post-transfers Gini coefficient in 2006 was only 9.3 percent and a meager 1.7 percent smaller than the pre-transfers Gini if we exclude transfers in kind. Unfortunately, the redistributive efficiency of Progresa/Oportunidades is an isolated case among redistributive instruments currently operating in Mexico. Even though government spending became undoubtedly more pro-poor during the last decade, a lot still needs to be changed to make redistributive spending a powerful instrument to reduce inequality in Mexico. In order to make government spending more progressive, at least two things need to happen. Mexico needs to eliminate regressive monetary transfers, subsidies and spending categories, and improve the poors access to post-secondary education (from grades 10 and above).
88 Redistributive spending includes monetary transfers, subsidies and transfers in kind. For more details, see tables 6 and 7 and section 7. 40
The most regressive programmes those that worsen the distribution of income are primarily those associated with agricultural subsidies and monetary transfers as well as (the subsidized component of ) pensions to state-owned enterprise employees. Other programmes and spending categories are regressive but in absolute terms alone. One of them is government spending on tertiary education. Although it has been improving, it is regressive because tertiary school enrollment among the poorest quintile is insignificant (and among the lowest in Latin America). In order to change this in a fundamental way, the government must compensate the poor for their high opportunity cost of tertiary education. Perhaps the government should implement large-scale scholarship or educational credit programs targeted to the poor, in partnership with the private sector. In addition, the availability of tertiary-level facilities will have to be expanded, given that, at present, not all students who wish to pursue a tertiary-level education are admitted because of supply-side constraints. More importantly, and perhaps the most difficult to tackle, the quality of public education will have to be improved at all levels, and at the upper secondary level in particular. Nowadays, the poor in Mexico are not able to attend public universities because of the low quality of their pre-university education. Poor children who graduate from low quality high schools lose out when they compete for scarce university places with students who graduated from private high schools. However, even if the competition to attend public universities were to be eliminated, students who attended poor quality public schools would either fail or become second- or third-rate professionals. Substantially improving the quality of public education is essential to making state action more progressive. To a large extent, public spending on basic public education is progressive in absolute terms because the higher income groups opt out of public schools, given the lower quality. In other words, public spending on basic education is progressive in part because it is self-targeted through low quality. An immediate corollary is that efforts to improve the quality of public education, if successful, would necessarily be at the cost of equity in the use of public resources. However, in this case, this would be a welcome cost. What are the chances that such reforms might occur? Between the mid-1990s and mid2000s, the state became more redistributionist, but not by strengthening the power of labour and peasant organizations. Large chunks of the traditional corporatist labour unions and agrarian organizations lost power and access to economic resources during the period of structural reforms and as part of the democratization process. This meant that corporate organizations were unable to capture part of the economic rents as they used to do before the reforms. However, other institutions representing the interests of the poor and disenfranchised did not replace them. Thus, state-led redistribution took place mainly through straightforward fiscal policy. In particular, in the areas of education and health and via targeted programs such as Progresa/ Oportunidades, fiscal policy expanded and redirected resources away from the top and middle ranges of the distribution towards the poor. State-led redistribution was a top-down process led by enlightened technocrats, made possible by the weakening of the power of corporatist organizations, and it had less to do with the empowerment of the poor and disenfranchised. However, not all corporatist organizations were weakened. One of the remaining diehards is the powerful teachers union, which has succeeded in blocking most attempts to improve the quality and accountability of the public school system. In addition, the democratization
41
process itself has weakened the ability of the state to implement top-down reforms because the ruling party does not have a congressional majority, and in order to create viable coalitions the Executive has to make concessions that will inevitably block reforms designed to eliminate entitlements of some politically powerful groups (such as agricultural producers, workers in state-owned oil and electricity companies and the teachers unions). Although the political economy outlook is not encouraging, it does not mean that the government should not try to push for reforms. In doing so, however, there is one aspect which is usually overlooked in the discussion of equity and state action. Policy recommendations often emphasize the elimination or reduction of spending categories or programmes that are regressive in absolute terms (that is, the poor receive less than the rich in per capita terms). The problem with this is twofold: first, even though regressive in absolute terms, the incidence of these spending categories is progressive in relative terms, and in the absence of compensatory mechanisms, eliminating them would make the poor worse off. This is the case, for example, of the VAT exemptions on food and prescription drugs. Second, the section of the public that receives higher per capita benefits includes large chunks of the middle- and lower-middle classes, who may be right in resisting a reform that could potentially affect their or their childrens social mobility or welfare. In addition, progressive leadership may want to financially support a burgeoning middle class because of the role it plays in fostering stable and democratic institutions.
42
References
Acosta, Pablo, and Gabriel V. Montes-Rojas (2008), Trade Reform and Inequality: the Case of Mexico and Argentina in the 1990s, The World Economy 31, 6: 763780. Airola, Jim, and Chinhui Juhn (2005), Wage Inequality in Post-Reform Mexico, IZA Discussion Paper No. 1525 (March), Institute for the Study of Labour (IZA). Alejo, Javier, Marcelo Bergolo, Fedora Carbajal and Guillermo Cruces (2009), Background paper prepared for the UNDP project Markets, the State and the Dynamics of Inequality in Latin America coordinated by Luis Felipe Lpez-Calva and Nora Lustig. lvarez, Jess, Vicente Garca Moreno, and Harry Anthony Patrinos (2007), Institutional Effects as Determinants of Learning Outcomes: Exploring State Variations in Mexico, Policy Research Working Paper No. 4286, World Bank. Aspe, P., and J. Beristin (1984), Distribution of educative and health services, in The Political Economy of Income Distribution in Mexico, eds. Pedro Aspe and Paul. E. Sigmund, Homes and Meier. Atolia, Manoj (2007), Trade Liberalization and Rising Wage Inequality in Latin America: Reconciliation with HOS Theory, Journal of International Economics 71, no. 2 (April): 467-494. Barham, Tania (2005), Providing a Healthier Start to Life: The Impact of Conditional Cash Transfer on Infant Mortality, Dissertation/Working Paper, Department of Agriculture and Resource Economics. U.C. Berkeley. Barros, Ricardo Paes de (2009), Markets, the State and the Dynamics of Inequality: Brazils Case Study, Mimeo. Barros, Ricardo Paes de, Mirela de Carvalho, and Samuel Franco y Rosane Mendona (2006), Uma anlise das principias causas da queda recente na desigualdade de renda brasileira, Econmica (Braslia) UFF 8, 1: 117147. Bautista, Sergio, PJ Gertler, M. Hernndez-Avila, SM Bertozzi (2004), Impacto de Oportunidades en la Morbilidad y el Estado de Salud de la Poblacin Beneficiaria y en la Utilizacin de Servicios de Salud: Resultados de Corto Plazo en Zonas Urbanas y de Mediano Plazo en Zonas Rurales. In Evaluacin externa de impacto del programa Oportunidades 2004. Tomo II. Salud, Bernardo Hernndez and Mauricio Hernndez eds., Mxico D.F.: CIESAS/INSP. Bouilln, Csar, Arianna Legovini and Nora Lustig (2004), Can Education Explain Changes in Income Inequality in Mexico? In The Microeconomics of Income Distribution Dynamics, Franois Bourguignon, Francisco H. G. Ferreira and Nora Lustig eds., Washington DC: World Bank/Oxford University Press. Bouilln, Csar, Arianna Legovini. and Nora Lustig (2003), Rising Inequality in Mexico: Household Characteristics and Regional Effects, Journal of Development Studies 39, 4: 112133.
43
Campos, Raymundo (2008), Why did Wage Inequality Decrease in Mexico after NAFTA? Mimeo, UC-Berkeley (October). Canonero, Gustavo, and Alejandro Werner (2002), Salarios Relativos y Liberalizacin del Comercio en Mxico, El Trimestre Econmico 69, 273: 123142. Chiquiar, Daniel (2008), Globalization, Regional Wage Differentials and the Stolper-Samuelson Theorem: Evidence from Mexico, Journal of International Economics 74, 1: 7093. CIDE-ITAM (2004), Anlisis de las Finanzas Pblicas en Mxico. CONACYT-CIDE-ITAM-Foro Consultivo Cientfico y Tecnolgico. Coady, David (2000) The application of social cost-benefit analysis to the evaluation of PROGRESA, Working Paper, International Food Policy Research Institute. Corbacho, Ana, and Gerd Schwartz (2002), Mexico: Experience with Expenditure Pro-poor Policies, Working Paper WP/02/12. IMF (International Monetary Fund). Cortes, Fernando, Patricio Sols e Israel Banegas (2006). Oportunidades y pobreza en Mxico: 20022004, Ciudad Mxico: El Colegio de Mxico. Cragg, Michael, and Mario Epelbaum (1996), Why has Wage Dispersion Grown in Mexico? Is it the Incidence of Reforms or the Growing Demand for Skills? Journal of Development Economics 51: 99116. Eberhard, Juan, and Eduardo Engel (2008), Decreasing Wage Inequality in Chile, Mimeo. Esquivel, Gerardo (2008), The Dynamics of Income Inequality in Mexico since NAFTA. Background paper prepared for the UNDP project Markets, the State and the Dynamics of Inequality: How to Advance Inclusive Growth, co-ordinated by Luis Felipe Lopez- Calva and Nora Lustig. Available at undp.economiccluster-lac.org. Esquivel, Gerardo, and Alejandra Huerta-Pineda (2007), Remittances and Poverty in Mexico: A Propensity Score Matching Approach, Integration and Trade Journal, JulyDecember 2007. Esquivel, Gerardo, and Jos Antonio Rodrguez-Lpez (2003), Technology, Trade and Wage Inequality in Mexico before and after NAFTA, Journal of Development Economics 72: 543565. Fairris, David, Gurleen Popli and Eduardo Zepeda (2008), Minimum Wage and Wage Structure in Mexico, Review of Social Economy, June, 66: 181208. FAO (Food and Agriculture Organization of the United Nations), (2005), Evaluacin de la Alianza para el Campo 2004. Mexico: FAO. Feenstra, Robert, and Gordon Hanson (1997), Foreign Direct Investment and Relative Wages: Evidence from Mexicos Maquiladoras, Journal of International Economics 42: 37193. Ferreira, Francisco H. G, Phillipe G. Leite and Matthew Wai-Poi (2007), Trade Liberalization, Employment Flows and Wage Inequality in Brazil, Working Paper 4108. World Bank Policy Research.
44
Gasparini, Leonardo, and Guillermo Cruces (2008), A Distribution in Motion: the Case of Argentina. A Review of the Empirical Evidence, Document prepared for the UNDP project Markets, the State and the Dynamics of Inequality: How to Advance Inclusive Growth, coordinated by Luis Felipe Lpez-Calva and Nora Lustig, Available at www.undp.org/latinamerica/inequality. Gill, Indermit S., Truman G. Packard and Juan Yermo (2004), Keeping the Promise of Old Age Income Security in Latin America: A Regional Study of Social Security Reforms, Washington, DC: World Bank. Goldberg, Pinelopi K., and Nina Pavcnik (2007), Distributional Effects of Globalization in Developing Countries, Journal of Economic Literature, March, 45: 3982. Grosh, Margaret (1994), Administering Targeted Social Programs in Latin America: From Platitudes to Practice, Washington, DC: World Bank. Guerrero, Isabel, Luis Felipe Lpez-Calva, and Michael Walton (2006), The Inequality Trap and its Links to Low Growth in Mexico, paper presented at conference on Equity and Growth of the World Bank at the Harvard Rockefeller Center in Mexico, Mexico City. Hanson, Gordon, and Ann Harrison (1999), Trade Liberalization and Wage Inequality, Industrial and Labour Relations Review 52, 2: 271288. Hernndez, Bernardo et al. (2003), Evaluacin del Impacto de Oportunidades en la Mortalidad Materna e Infantil. In Evaluacin Externa del Impacto del Programa Oportunidades 2003, eds. Bernardo Hernndez Prado and Mauricio Hernndez vila, vol. 2, chapter 2. Cuernacava, Mxico: Instituto Nacional de Salud Pblica. de Hoyos, Rafael, Nora Lustig (2009), Apertura comercial, desigualdad y pobreza. Resea de los enfoques metodolgicos, el estado del conocimiento y la asignatura pendiente, El Trimestre Econmico (Fondo de Cultura Econmica) (AprilJune) 302: 283328. IMSS (Instituto Mexicano del Seguro Social) (2005), Informe al Ejecutivo Federal sobre la Situacin Financiera del IMSS. IMSS. Komives, Kristin, et al. (2009), Residential Electricity Subsidies in Mexico: Exploring options for reform and for enhancing the impact on the poor. Washington, DC: World Bank. Legovini, Arianna, Csar Bouilln and Nora Lustig (2005), Can Education Explain Changes in Income Inequality in Mexico, in The Microeconomics of Income Distribution Dynamics, eds. Franois Bourguignon, Francisco H. G. Ferreira and Nora Lustig, pp. 138. World Bank/Oxford University Press. Lerman, Robert, and Shlomo Yitzhaki (1985), Income Inequality Effects by Income Source: A New Approach and Application to the U.S., Review of Economics and Statistics 67, 1: 15156. Levy, Santiago (2006), Progress against poverty, Sustaining Mexicos Progresa-Oportunidades Programme, Brookings Institution Press. __________ (2008), Good Intentions, Bad Outcomes: Social Policy, Informality, and Economic Growth in Mexico, Brookings Institution Press.
45
Leyva-Parra, Gerardo (2005), El Ajuste del Ingreso de la ENIGH con la Contabilidad Nacional y la Medicin de la Pobreza en Mxico, Research Paper no. 19, Mexico: Sedesol. Lpez-Acevedo, Gladys (2006). Mexico: Two Decades of the Evolution of Education and Inequality, Working Paper No. 3919, World Bank Policy Research. Lpez-Feldman, Alejandro 2006 Decomposing Inequality and Obtaining Marginal Effects, The Stata Journal 6, 1: 106111. Lustig, Nora (1992), Mexico: The Remaking of an Economy, Reprint, Washington: The Brookings Institution, 1998. __________ (2006), Invertir en salud para el desarrollo econmico. Mxico D.F.: Comisin Mexicana de Macroeconoma y Salud (Fondo de Cultura Econmica). __________ (2007a), A Note on Markets, the State and the Dynamics of Inequality, Concept note for the UNDP project Markets, the State and the Dynamics of Inequality: How to Advance Inclusive Growth, coordinated by Luis Felipe Lpez-Calva and Nora Lustig, Available at www.undp.org/latinamerica/inequality. __________ (2007b), Health and economic development: The Mexican case, English version of Salud y desarrollo econmico. El caso de Mexico, El Trimestre Econmico 296 (OctoberDecember). Mancera, C (2008), Resultados preliminares de los estudios de calidad de los servicios educativos que atienden a la poblacin beneficiaria de Oportunidades en reas rurales, Cuarta Sesin del Comit Acadmico de Expertos en Poltica Social, Sedesol (Secretaria de Desarrollo Social). OECD (Organisation for Economic Co-operation and Development) (2006), Agricultural and Fisheries Policies in Mexico: Recent Achievements, Continuing the Reform Agenda, OECD (November). __________ (2008), Education at a Glance 2008: OECD Indicators, OECD, Available at www.oecd.org/edu/eag2008. Pnuco-Laguette, Humberto and Miguel Szkely (1996), Income Distribution and Poverty in Mexico, in The New Economic Model in Latin America and its Impact on Income Distribution and Poverty, ed. V. Bulmer-Thomas, pp.185221, London: Macmillan Press. Parker, Susan (2005), Evaluacin del Impacto de Oportunidades sobre la Inscripcin, Reprobacin y Abandono Escolar, in Evaluacin Externa del Impacto del Programa Oportunidades 2004, eds. Bernardo Hernndez Prado and Mauricio Hernndez vila, chapter 1, Mexico: Instituto Nacional de Salud Pblica. Perry, Guillermo E. et al.( 2007), Informality: exit and exclusion. World Bank Latin American and Caribbean Studies. PNUD (Programa del las Naciones Unidas para el Desarrollo) (2008), ndice de Desarrollo Humano Municipal en Mxico 20002005.,Mexico: PNUD. Presidencia de la Repblica (2003), Tercer Informe del Gobierno. __________ (2008), Segundo Informe de Gobierno.
46
Ravallion, Martin and Shaohua Chen (2003), Measuring Pro-Poor Growth, Economics Letters 78, 1: 9399. Robertson, Raymond (2007), Trade and Wages: Two Puzzles from Mexico, The World Economy 30, 9: 13781398. Robinson, James A (2010), The Political Economy of Redistributive Policies, In this series. Schultz, Paul (2000), The Impact of Progresa on School Enrollments. Final Report, Washington: International Food Policy Research Institute. Scott, John (2002a), Distribucin de la ayuda alimentaria en Mxico: La revolucin de los noventa, Working Paper 240, (Divisin de Economa). CIDE (Centro de Investigacin de Docencia Econmica). __________ (2002b), High Inequality, Low Revenue: Redistributive Efficiency of Latin American Fiscal Policy In Comparative Perspective, paper prepared for a Conference on Fiscal Aspects of Social Programs, Studies On Poverty and Social Protection, Regional Policy Dialogue, Poverty Reduction and Social Protection Network. Inter-American Development Bank, Available at idbdocs.iadb.org/wsdocs/getdocument.aspx?docnum=616098. __________ (2004), Eficiencia redistributiva de los programas contra la pobreza en Mxico, Working Paper 330, (Divisin de Economa). CIDE (Centro de Investigacin y Docencia Econmicas). __________ (2005), Social Security and Inequality in Mexico: From Polarization to Universality, Well-being and Social Policy 1: 5982. __________ (2006a), Health and Health Care Inequality in Mexico, Working Paper 4, Innovations in Health Financing, Funsalud-INSP, Technical Background paper for the Report of the Mexican Commission on Macroeconomics and Health. __________ (2006b), Poverty and Inequality in Mexico under NAFTA: Trade Liberalization and Policy Failures, Unpublished, Mexico: CIDE. __________ (2007), Agricultural policies and rural poverty, CIDE Working Paper and Background Paper for the OECD Report on Agricultural and Fisheries Policies in Mexico: Recent Achievements , Continuing the Reform Agenda. OECD (November 2006). __________ (2008a), Diagnstico y Propuesta de Reforma para PROCAMPO: Poltica de Apoyos para un Crecimiento con Equidad en el Campo Mexicano, Working Paper for ASERCA-BID project on Procampo reform. __________ (2008b), Mexico Agriculture Public Expenditure Review: Assessment of ARD Public Expenditures, Background Paper for the Mexico Agriculture Public Expenditure Review, World Bank. __________ (2008c), The Political Economy of Inequality in Mexico, Mimeo. __________ (2009a), Gasto Pblico y Desarrollo Humano en Mxico: Anlisis de Incidencia y Equidad, Working Paper for Informe de Desarrollo Humano de Mxico 2008/2009, Mxico: PNUD.
47
__________ (2009b), Redistributive Constraints under High Inequality: The Case of Mexico, document prepared for the UNDP project Markets, the State and the Dynamics of Inequality: How to Advance Inclusive Growth, coordinated by Luis Felipe Lpez-Calva and Nora Lustig, Available at www.undp.org/latinamerica/inequality. __________ (2009c), The Incidence of Agricultural Subsidies in Mexico, paper for project on Agricultural Trade Adjustment and Rural Poverty: Transparency, Accountability and Compensatory Programs in Mexico, Woodrow Wilson International Center for Scholars, Mexico Institute. SEDLAC (Socio- Economic Database for Latin American and the Caribbean) CEDLAS and World Bank (2009), Databases for Latin America and the Caribbean. Available at www.depeco.econo.unlp.edu.ar/sedlac/eng/index.php. Sen, Amartya (1995) The Political Economy of Targeting, in Public Spending and the Poor: Theory and Evidence, D. van de Walle and K. Nead eds., Baltimore: Johns Hopkins University Press. Sen, Amartya, James E. Foster (1973), On Economics and Inequality. Reprint, USA: Oxford University Press, 1997. SHCP (Secretaria de Hacienda y Crdito Pblico) (2008), Distribucin del pago de impuestos y recepcin del gasto pblico por deciles de hogares y personas. Resultados para el ao de 2006., SHCP. Stark, Oded, J. Edward Taylor and Shlomo Yitzhaki (1986), Remittances and Inequality, Economic Journal 96 (September): 72240. Stefano, Scarpetta (2009), Job Creation in Latin America and the Caribbean: Recent Trends and Policy Challenges, Palgrave Macmillan and World Bank. Szkely, Miguel (2003), Es posible un Mxico con menor pobreza y desigualdad, Research Papers 3, Mxico: Sedesol (Secretara de Desarrollo Social). __________ (2005), Pobreza y Desigualdad en Mxico entre 1950 y 2004, El Trimestre Econmico 72, no. 288 (OctoberDecember). Thomas, Vinod, Yan Wang, and Xibo Fan (2000), Measuring Education Inequality: Gini Coefficients of Education, World Bank. Tornell, Aaron, and Gerardo Esquivel (1997), The Political Economy of Mexicos Entry into NAFTA, in Regionalism versus Multilateral Trade Arrangements, Takatoshi Ito and Anne O. Krueger eds., Chicago: University of Chicago Press. van Doorslaer, Eddy, and Owen ODonnell (2008), Measurement and Explanation of Inequality in Health and Health Care in Low-Income Settings, Discussion Paper No. 2008/04, UNU-WIDER (World Institute for Development Economics Research of the United Nations University). Verhoogen, Eric (2008), Trade, Quality Upgrading, and Wage Inequality in the Mexican Manufacturing Sector, Quarterly Journal of Economics 123, 2 (May): 489530.
48
World Bank (2004), Reforma del financiamiento para pensiones y actividades de asistencia social: evaluacin del pasivo fiscal global de los sistemas pblicos de pensiones en Mxico, Mimeo. ________ (2006), Decentralization, Poverty, and Development in Mexico, in Working Paper 35692, Decentralized service delivery for the poor (vol. 1 of 2), chapter 1. Mexico: World Bank.
49
United Nations Development Programme 304 East 45th Street New York, NY, 11375 U.S.A. E-mail: poverty.reduction@undp.org Website: www.undp.org/poverty

A Decade of Falling Inequality in Mexico

Hochgeladen von

Dokumentinformationen

Originalbeschreibung:

Copyright

Verfügbare Formate

Dieses Dokument teilen

Dokument teilen oder einbetten

Freigabeoptionen

Stufen Sie dieses Dokument als nützlich ein?

Sind diese Inhalte unangemessen?

Copyright:

Verfügbare Formate

A Decade of Falling Inequality in Mexico

Hochgeladen von

Copyright:

Verfügbare Formate

Discussion Paper

A Decade of Falling Inequality in Mexico: Market Forces or State Action?

POVERT Y REDUC TION

A Decade of Falling Inequality in Mexico: Market Forces or State Action?

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

Income inequality after NAFTA: 19942006

Gerardo Esquivel, Nora Lustig and John Scott

Figure 1, Gini coefficients for alternative income definitions, 19842006

Poverty Reduction Discussion Paper

TABLE 1, Distribution of monetary household per capita income, 19842006 (percent)

Gerardo Esquivel, Nora Lustig and John Scott

Growth Incidence Curve, 19962006

Growth Incidence Curve, 19962000

Growth Incidence Curve, 20002006

Poverty Reduction Discussion Paper

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

Source: Barros et al. (2009a).

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

National Mexico: Marginal Effect on Gini Coefficient by Income Source

Urban Mexico: Marginal Effect on Gini Coefficient by Income Source

Rural Mexico: Marginal Effect on Gini Coefficient by Income Source

Source: Esquivel (2008).

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

Gerardo Esquivel, Nora Lustig and John Scott

Labour income inequality and the skilled/unskilled wage gap

Poverty Reduction Discussion Paper

Source: Esquivel (2008).

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

Source: Esquivel (2008).

Gerardo Esquivel, Nora Lustig and John Scott

Source: Esquivel (2008).

Poverty Reduction Discussion Paper

The rising progressivity in government spending on education

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

Gerardo Esquivel, Nora Lustig and John Scott

51 Scott (2009b). 52 See, for example, Lustig (2006, 2007a). 23

Poverty Reduction Discussion Paper

Pro-poor government spending: The distributional impact of Progresa/Oportunidades

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

Gerardo Esquivel, Nora Lustig and John Scott

Poverty Reduction Discussion Paper

The distributional impact of state action: Government spending and taxes

Gerardo Esquivel, Nora Lustig and John Scott

Figure 10, Evolution of social spending, 19252006 (percent)

Source: Scott (2009b). Note: Programmable spending excludes debt servicing.

Poverty Reduction Discussion Paper

Gerardo Esquivel, Nora Lustig and John Scott

Table 6, Government redistributive spending by category and programme, 2006

Millions Current Mexican Pesos

Total Transfers (%)

Household inc. before taxes and transf. (%)

Poverty Reduction Discussion Paper

Table 7, Description of main categories and programmes

Programs Targeted to the Poor