Experiences with Pro-Poor Expenditure Policies
  • 1, International Monetary Fund

Contributor Notes

Author’s E-Mail Address:;

Against the background of Mexico's persistently high degree of inequality, this paper analyzes the country's experience with pro-poor policies over the last decade. A number of important government initiatives, implemented since the mid-1990s, have aimed at improving distributional equity through pro-poor expenditure programs, while at the same time seeking to increase the efficiency of public spending. This paper reviews these initiatives and outlines some additional policy options.


Against the background of Mexico's persistently high degree of inequality, this paper analyzes the country's experience with pro-poor policies over the last decade. A number of important government initiatives, implemented since the mid-1990s, have aimed at improving distributional equity through pro-poor expenditure programs, while at the same time seeking to increase the efficiency of public spending. This paper reviews these initiatives and outlines some additional policy options.

I. Introduction

Latin America remains the most unequal region of the world; Mexico is no exception. Economic inequality is as high today as it was two decades ago and poverty has been pronounced relative to regional income levels. However, changes in inequality appear to have been significantly influenced by cycles in economic growth and per-capita income (Iglesias, 1998).

In general, a serious shortcoming of the adjustment programs undertaken over the last decades in Latin America is that they have failed to improve the income distribution (Camdessus, 1997). Income inequality affects the sustainability of economic adjustment and the level and quality of economic growth. It is not always clear what policymakers can do to reduce income inequality. Traditional economic theory, for instance, along the lines of Kuznets’ (1955) well-known inverted U-curve hypothesis, studied distributional aspects of economic policies largely through their impact on economic growth. A key message of this analysis was that governments should foster rapid economic growth to get beyond the stage of economic development where income disparities widen.

It is commonly accepted now that qualitative aspects of economic growth are at least as important as economic growth itself. In general, countries that have been most successful in attacking poverty and reducing income inequality are not necessarily those with the highest growth rates but those that have promoted the efficient use of labor and invested in developing the human capital of the poor (World Bank, 1990). In this context, the distributional effects of public expenditure and its composition have received considerable attention. In contrast to traditional economic analysis, this more recent research has shown that there is not necessarily a trade-off between redistributive and efficiency goals in public expenditure policies: by improving the quality and efficiency of public expenditure, income inequality can be significantly reduced without adversely effecting economic growth. Hence, good expenditure policies can help mitigate or avert a typical Kuznets process, so that, in the process of economic development, the income distribution does not need to get worse before it can get better. Furthermore they can do so without adversely impacting future economic growth.2

As a result, research attention has shifted from the distributional implications of economic growth to the growth implications of a given income distribution, including whether and to what extent a high degree of inequality limits a country’s future economic growth potential and performance.3 It is generally accepted now that a high degree of inequality can have detrimental effects on a country’s future economic development. For example, inadequate nutrition, health, and education might easily become binding constraints to the work efforts of the poor, and improvements in these areas would contribute to raising labor productivity and enhancing the economic growth potential (Tanzi and Chu, 1992). These results are of particular importance for Latin American economies which, in general, are characterized by a high or even a growing degree of income inequality and large population segments that live in poverty.

Against this general background, this paper focuses on one specific Latin American economy: Mexico. The paper is structured as follows. First, it provides an overview of trends in Mexico’s income distribution and their broad determinants throughout the 1990s. Second, it analyzes the impact of various pro-poor and other social expenditure programs on Mexico’s income distribution, including the government’s recent efforts to improve equity. The paper concludes by offering some tentative thoughts on the links between pro-poor policies (and more generally expenditure policies) and distributional outcomes, and suggesting policy options for improving these outcomes.

II. Trends in Income Distribution and Their Broad Determinants

A. Overview

Mexico’s income distribution continues to be characterized by a high degree of inequality. Mexico’s income inequality is significantly more pronounced than the Latin American average, which is the region with the highest degree of inequality in the world (Figure 1). In 1992, Mexico’s Gini coefficient4 was 0.57, according to data from Deininger and Squire (1996). In comparison, the Gini coefficient during the 1990s in OECD and high-income countries averaged 0.34 and in Latin America 0.49 (Figure 2). Also, whereas some countries and regions have experienced recent reductions in the degree of income inequality, Mexico’s income inequality has worsened.5 Although reliance on a single data source may easily give a misleading picture, all available inequality indicators for Mexico, including the various Gini coefficients reported in Table 1, show a consistent picture of a considerable degree of inequality.6 In 1994, for example, the Gini coefficient based on a broad measure of household income was 0.54 (Table 2). Similarly, the total income of Mexico’s top 20 percent of income earners was 16.4 times that of the bottom 20 percent, which compares to an average of 6.3 in OECD and other high-income countries during the 1960s-1990s.

Figure 1.
Figure 1.

Inequality in Latin America 1/

Citation: IMF Working Papers 2002, 012; 10.5089/9781451842913.001.A001

Source: Based on data from Deininger and Squire (1996), and, for Brazil in the 1990s, Clements (1997).1/ Inequality as measured by the Gini coefficient (multiplied by 100).
Figure 2.
Figure 2.

Mexico’s Inequality in Global Comparison 1/

Citation: IMF Working Papers 2002, 012; 10.5089/9781451842913.001.A001

Source: Based on data from Deininger and Squire (1996).1/ Inequality as measured by the Gini coefficient (multiplied by 100). Data reported refer to unweigthed average Gini coefficients of economies in each region. The sample includes 108 economies. Changes within regions may be caused by the fact that not all economies have observations for all decades.2/ Excludes Mexico.3/ Includes Mexico.
Table 1.

Mexico: Gini Coefficients According to Different Data Sources, 1950 2000 1/

article image
Sources: As indicated.

Gini coefficient multiplied by 100.

Table 2.

Mexico: Distribution of Household Income, 1984-2000

article image
Source: Lustig and Székely (1997a) based on adjusted income and authors’ estimates based on current income.

Gini coefficient multiplied by 100.

Share of top 10 percent divided by share of bottom 40 percent.

Share of top 10 percent divided by share of bottom 10 percent.

Share of top 20 percent divided by share of bottom 20 percent.

As in other Latin American economies, Mexico’s income inequality can largely be attributed to inequality in factor endowments and human capital formation, where the former includes factors such as land, natural resources, and physical capital, and the latter access to education and health care. Evidence suggests that low average levels and a high degree of inequality in physical and human capital formation explain a major part of Latin America’s relatively high degree of income inequality as well as the region’s “excess inequality” compared to the rest of the world (Londoño and Székely, 1997, and Inter-American Development Bank (IDB), 1997 and 1999).

Income inequality in Mexico has an important urban/rural and regional dimension: average incomes in urban areas and in the richest states remain much higher than those of the poorest states and rural areas. Separate Gini coefficients for urban and rural areas show that income inequality within urban and rural areas was less pronounced than for the country as a whole, reflecting the relatively higher degree of homogeneity. For example, calculations for 1994 by the Economic Commission for Latin America and the Caribbean (ECLAC) (1997a, 1997b) show that Mexico had urban and rural Gini coefficients of 0.41 and 0.33, respectively (Table 1), which is substantially less unequal than the Gini coefficients for the entire country. Still, calculations by Pánuco-Laguette and Székely (1996) show that inequality within urban and rural areas still account for most of the inequality in Mexico. In 1992, the “within inequality” amounted to 84 percent of total inequality and the “between inequality” to 16 percent. However, the same data also show that the importance of between inequality (that is, the urban/rural gap) significantly increased during 1984-92.

Households in urban areas are, on average, much better off than households in rural areas. This urban/rural income gap is reflected in the poverty data shown in Table 3. Accordingly, in 1994, 81 percent of Mexico’s extremely poor lived in rural areas, although the rural population share amounted only to 42 percent of the total population. Similarly, in 1994, 79 percent of Mexico’s extremely poor lived in four regions: the Center, the Center-West, the South, and the Southeast;7 in contrast, only 17 percent of Mexico’s poor lived in the northern part of Mexico (which comprises the regions North, Northeast and Northwest). Poverty rates were most pronounced in the Southeastern region that, in 1994, contained 9 percent of Mexico’s population but 19 percent of the country’s poor.

Table 3.

Mexico: Poverty Profile, 1984-1994

article image
Source: Lustig and Székely (1997a).

Income inequality also has important socio-economic characteristics. Almost the entire increase in poverty rates during 1989-94, a period when Mexico continued to implement structural reforms such as privatization and trade liberalization, reflected sharp increases in poverty in households with three socioeconomic characteristics of the household head: no education beyond primary school, employment in the agricultural sector, and residing in the South or Southeastern part of the country (Table 3). Although in the more prosperous (urban) regions poverty rates declined during 1989-94, they increased in rural and marginalized8 areas, particularly in those areas with a high share of indigenous population groups. Importantly, poverty rates increased, particularly for those population groups with little formal education.

Lagging rural development and a strong pro-urban expenditure bias have frequently been indicated as being at the root of Mexican poverty and inequality;9 in addition, a fairly unstable macroeconomic environment during much of the last three decades may have had important adverse distributional consequences. Mexican poverty profiles consistently show that the extremely poor, aside from being located mostly in the rural areas and having the lowest level of educational attainment (Table 3), derive most of their earnings from self-employment and wage labor—most in agriculture and related activities. The returns to unskilled labor and land (the main asset owned by the poor) depend critically on two factors: government policies, broadly defined to include pricing and resource allocation decisions, particularly through expenditure policies; and the institutional and macroeconomic environment in which people make their decisions. Historically, the government’s policies in agriculture—which focused on providing subsidies to fertilizers, agricultural credit, and electricity, providing crop insurance, and maintaining price support schemes—probably provided large rents to higher-income producers without producing significant increases in agricultural output, higher returns to land, or higher wages for unskilled rural labor. These policies doubly discriminated against the rural poor: agriculture and rural areas received an inequitably small share of the total resources for social and infrastructure investment which were mainly geared toward urban areas; and resources channeled to the rural areas were mostly untargeted, benefiting better-off producers (Levy, 1992).

Policies that affect risk and uncertainty are key elements in the decisions people make. The poor are less able to bear risk and face uncertainty, and changes in risks and uncertainty may affect decisions to migrate, on- and off-farm labor supply, crop choice, etc. Although some uncertainties are exogenous (like the weather), others are induced by macroeconomic policies. For example, the poor, particularly the moderately poor, may hold financial assets, particularly cash balances, between the sale of a crop and the purchase of goods or inputs; inflation erodes the value of these assets, thereby limiting the ability of the poor to accumulate assets over the medium term (Levy, 1992).

B. Recent Trends

Following several decades in which the country moved toward a more even distribution of income, Mexico’s income disparities have generally widened since the 1980s. Much of the increase in income disparities happened in the 1980s. During the early 1990s, there was neither a further increase nor a significant reduction in Mexico’s income disparities, similar to what has been observed for much of the region (IDB, 1997). Since the mid 1980s, inequality has risen again.10 As the Gini coefficient shows, income inequality in Mexico increased during 1950-75 from already fairly high initial levels, and then declined during 1975-84 (Table 1).11 Although levels of inequality differ in the various studies, they all show a significant increase in income inequality from 1984 until 1994,12 a small reduction in 1996, and a further increase during 1996-2000. Income inequality in 2000 was the highest since the mid 1980s. In general, the Gini coefficient in both consumption and income increased by around 5 percentage points since 1984 (Figure 3). As indicated by the Lorenz Curve, the distribution of both income and consumption in 2000 was clearly more unequal than in 1984 (Figure 4).13 In 2000, the wealthiest 10 percent of all Mexicans received nearly 39 percent of the total income in the country, and the poorest 40 percent around 12 percent of total income (Table 4).

Figure 3.
Figure 3.

Mexico: Gini Coefficient, 1984-2000

Citation: IMF Working Papers 2002, 012; 10.5089/9781451842913.001.A001

Source: Authors’ estimates based on household surveys, several years. Consumption and income measures are unadjusted.
Figure 4.
Figure 4.

Mexico: Lorenz Distribution in Income and Consumption, 1984 and 2000

Citation: IMF Working Papers 2002, 012; 10.5089/9781451842913.001.A001

Source: Authors’ calculations based on the 1994 and 2000 household surveys (INEGI, 1995 and 2001a).
Table 4.

Mexico: Distribution of Household Income by Income Decile, 1950-2000

article image
Sources: Székely (1996), Lustig and Székely (1997) and authors’ estimates.

Data from Lustig and Székely (1997) are corrected for estimated capital income received.

The marked increase in income inequality in the 1980s reflected largely an increase in the income gap between the very rich and the rest. This pattern is similar to what has been observed elsewhere in Latin America, where the richest 10 percent of all households have generally been able to hold on or increase their income share, while the poorest 40 percent just managed to maintain theirs or suffered a decline (ECLAC, 1997b). There are few differences between the lower tail and middle of the income distribution in Mexico and in other countries, although Mexico has larger differences between the richest 10 percent of the population and the rest: when excluding the top 10 percent of the income distribution, Mexico only ranks 13th rather than 5th in terms of inequality in Latin America. Moreover, inequality among the first 90 percent of the population is even lower than inequality among the same group in the United States (Székely, 1999).

It has been argued that government expenditure played an important role in increasing income inequality during the period of economic adjustment that followed the Mexican debt crisis of 1982, particularly through transfers that largely benefited better-off population groups (Székely, 1994; and Pánuco-Laguette and Székely, 1996). This is also supported by evidence presented by Carral Cuevas and Chávez Presa (1982), who show that subsidized agricultural credit has, traditionally, been more unequally distributed than income or land holdings: rich households received an even larger share of total credit subsidies than they received of total income or had in total land holdings. In general, the main losers during the adjustment period were industrial workers and the rural poor who bore the brunt of economic adjustment and saw their income share diminish. The positive effects of economic growth in the late 1980s were largely absorbed by the very rich who experienced a further increase in their income share as they captured much of the increase in enterprise profits. Other population segments failed to recover their positions.

In the past, adverse economic shocks, like the recession during 1984-89, have affected the income distribution in high-density urban areas more than in rural areas. As a result, during 1984-92, income inequality in urban areas became more pronounced relative to rural areas, notwithstanding the fact that rural poverty remained pervasive. Given the relatively higher income level in urban areas, adverse shocks may create much larger income disparities there than in rural areas, where the income level is lower to begin with. Hence, during 1984-92, income inequality between urban and rural areas rose as inequality within urban areas became considerably more pronounced in comparison to inequality in rural areas (Pánuco-Laguette and Székely, 1996).14

The December 1994 crisis seems to have reduced income disparities but increased poverty in its immediate aftermath. The share of income of the top quintile fell in the aftermath of the crisis, resulting in reduced income disparities in 1996. Estimates suggest that extreme poverty increased from 16 percent of the population in 1994 to 19 percent in 1995, and dropped to 18 percent in 1997; similarly, moderate poverty increased from 32 percent of the population in 1994 to 36 percent in 1995, and dropped to 34 percent in 1997 (Lustig and Székely, 1997b).

The increase in poverty following the 1994 crisis can be accounted for by several factors. Real minimum wages dropped significantly in 1995: in the fourth quarter of 1995, the real minimum wage had dropped to 81 percent of its average 1994 level and then remained fairly stable at around 80 percent (Table 5). Also, during 1995-96, the percentage share of salaried workers who received an income below the minimum wage rose lightly. Many sectors experienced dramatic drops in real wages. There was a steep drop in incomes in the agricultural sector, which probably led to a significant increase in the incidence of poverty in rural areas (Lustig and Székely, 1997a). In the manufacturing and construction sectors, real earnings per worker declined even after the crisis until 1997, when they were about 78 percent and less than 70 percent of their average 1994 level, respectively (Table 5). The 1994 crisis also had a big impact on the unemployment rate, which increased from 3.6 percent in the fourth quarter of 1994 to 7.4 percent in the third quarter of 1995 (Table 5). By the end of 1995, the unemployment rate had started to decline, reaching a low 2.2 percent in 2000. In urban areas, manufacturing and construction sector workers probably bore much of the adverse employment effects of the financial crisis. While employment in the manufacturing sector recovered considerably after 1996, employment in the construction sector remained low compared to pre-crisis levels.

Table 5.

Mexico: Labor Market Indicators in 43 Urban Areas, 1994-2000

article image
Source: Instituto National de Estadistica, Geografia e Informátics (INEGI), and authors’ estimates.

As percentage of all persons aged 12 and over.

Persons aged 12 and over who in reference period: a) did not work; b) were available for employment, and c) unsuccessfully sought employment in the 2 months prior to reference period. As percent of economically active population; data refer to quarterly averages.

Openly unemployed (as defined in 2/) plus those working less than 15 hours/week. As percent of economically active population; data refer to quarterly averages.

Nominal minimum wage deflated by the consumer price index.

Preliminary estimates.

Government expenditure programs did not help to reduce the effect of the crisis on the poor. To a large extent, this reflects the impact of crisis-related expenditures: for example, government programs geared toward providing support to the banking system directly benefited those population groups with outstanding consumer and mortgage credits, who tend to be middle- and upper-income groups. From September 1995 to February 1997, the government’s support scheme for small debtors (ADE) transferred about 0.1 percent of GDP, and the support scheme for mortgage debtors transferred another 0.1 percent of GDP in 1996. Other government programs that restructured loans or provided support for specific interest groups, such as highway concessionaires, also benefited better-off income groups. In addition to these direct transfers, public finances also had to absorb the cost of banking system restructuring operations, which may have crowded out other potential pro-poor social expenditures.15

III. Social Expenditure and Income Distribution

A. Overview of Social Expenditure

Government policies can be important tools for shaping a country’s income distribution.

Generally, as pointed out by Tanzi (1998), “there is much room, especially in fiscal policy, for reforms that are both pro-growth and pro-poor. Often, many of the policies that benefit the lowest-income groups are also those with the highest social rate of return.”

The expenditure side of the budget offers greater scope for income redistribution than the tax side. Even a moderately progressive tax system is unlikely to be a major determinant of the post tax income distribution. Major taxes, like the VAT, are often not progressive, and, while some progression can be built into them, this usually comes at a high cost in terms of greater administrative effort needed or evasion itself, In contrast, expenditure policies can potentially have a much stronger differential impact on various income groups.16 This is especially true for social expenditure, because of its direct impact on income via transfer payments to households, and by affecting key elements that determine the income distribution over time, like human capital accumulation or the acquisition of physical assets.

During 1990-2000, social expenditures increased sharply, although more so in the first half of the 1990s. Social expenditure in Mexico includes spending on health, nutrition, sanitation, housing, social assistance, and pensions. Total social expenditure amounted to about 9.5 percent of GDP in 2000, compared to 6.1 percent of GDP in 1990, which implies an increase of 83 percent in real terms over the decade. Overall, and as shown in Figure 5, social expenditure increased in real terms and relative to GDP during 1990-94, but following the December 1994 crisis, contracted sharply in 1995-1996. During 1997-2000 social expenditures relative to GDP increased again. In 2000, 89 percent of all social expenditures were for health, education, and social security (mainly pensions), a proportion that has been roughly stable throughout the 1990s. Although government spending on social welfare and social assistance is fairly small, it was the only social expenditure item that was not reduced relative to GDP in the years immediately following the December 1994 crisis, mainly to maintain a social safety net for the most vulnerable segments of society.17 Given the large share of wages in social spending, the contraction of social expenditure in the crisis year of 1995 is in part attributable to restrictive wage policies. Notwithstanding the crisis, the government safeguarded social expenditures, and even expanded the maximum duration of health benefits for the unemployed from 3 months to 6 months and to broaden some education scholarship programs for the needy.

Figure 5.
Figure 5.

Mexico: Social Expenditure, 1990-2000

Citation: IMF Working Papers 2002, 012; 10.5089/9781451842913.001.A001

Source: Mexico, Federal Executive (2001), and authors’ estimates.

Indicators of social well-being—which, to some extent, measure the effectiveness of social expenditure—have shown significant improvement over the last few decades (Table 6). However, the distribution of social indicators shows a pattern of continued significant inequality (Table 7). The population in the wealthiest quintile has on average 7 more years of education than the population in the poorest quintile, nearly 70 percentage points more social security coverage, nearly 60 percentage points more health insurance coverage and is exposed to a lower unemployment rate. The following sections provide overviews of the main social expenditure items, examine distributive effects, review recent government initiatives and suggest policy options for improving equity.

Table 6.

Mexico: Indicators of Social Weil-Being, 1940-2000

article image
Sources: Illiteracy rale in 1940-60: Lustig and Székely (1997b); in 1970-2000: World Development Indicators, World Bank (2001), Average years of formal education in 1940-94: Lustig and Székely (1997b); for 1996-2000: INEGI (2001b). Infant mortality rate in 1940-80: Lustig and Székely (1997b); for 1990-2000: World Development Indicators, World Bank (2001). Under-Age 5 mortality rate: INEGI (2001b). Coverage of vaccination: INEGI (2001b)

For population over 15 years of age.

Table 7.

Mexico: Distribution of Social Indicators by Income Quintiles, 1996

article image
Source: Inter-American Development Bank (2001).

B. Education


Standard education system output indicators generally improved over the last decade, particularly for the primary level (Tables 6 and 8). For example, in 2000, the average in years of formal education was 7.6 years, compared to 6.3 years in 1990.18 Despite the improvements in recent years, educational attainment levels remain low: for example, in 1994, 54 percent of the overall population had not completed primary school or was without formal education; in rural areas, this amounted to 74 percent of the population. Also, the distribution of educational attainment is very unequal (Table 7).

Educational attainment levels remain particularly low in communities with a high share of indigenous population groups:19 whereas in 1989, the average years of formal education amounted to 4.9 years in communities where less than 30 percent of the population belonged to the indigenous population; it was only 2 years in communities where over 70 percent of the population belonged to the indigenous population (Panagides, 1994). The states with the highest concentration of indigenous population groups (Yucatan, Quintana Roo, Oaxaca, Chiapas, and Campeche) are all in Southern Mexico, and are commonly recognized as being generally poor or having large regions with a high incidence of poverty. Some of these states also have had traditionally poor educational attainment levels: in 1970, for example, the illiteracy rate in Oaxaca and Chiapas was more than 45 percent of the population over 15 years of age, while in the wealthier state of Nuevo León and in the Federal District it was below 15 percent (Lustig and Székely, 1997a). Even in 1990, the illiteracy rate in the 8 poorest states (all in the South) averaged 22 percent, while the national average was 12.5 percent. Differences in educational attainment have remained considerable in recent years. In 1999, the illiteracy rate in Oaxaca and Chiapas was over 23 percent, compared to the national average of 10 percent. In the wealthier areas of Nuevo León and the Federal District illiteracy rates were below 3.6 percent (Mexico, Secretariat of Health, 2001).

Public education remains largely financed by the federal level of government. Up to 1992, the provision and financing of public education were largely federal responsibilities. In 1992, all operational aspects of the school system were decentralized to the states; still, the federal government remained in charge of financing the system and operating the public school system in Mexico City (the Federal District). With public education financed by the federal government, regional differences in per-student expenditure and in other indicators (such as student/teacher ratios) are relatively small compared to some other Latin American countries where financing responsibilities lie with sub-national levels of government. Data from the Secretariat of Public Education (SEP) show total education expenditure amounted to 6.1 percent of GDP in 2000 (Table 8). Of the total, 4.1 percent of GDP was spent by the federal system, while the rest corresponds to states, municipalities and the private sector. After a sharp reduction in education spending in the mid 1980s, federal spending rapidly increased during 1990-94 from 3.0 percent of GDP to 4.6 percent of GDP, reflecting the government’s emphasis on the education sector as a core element of future economic growth. Following the December 1994 crisis, federal education spending was reduced to 4 percent of GDP in 1996 and rose again slightly by the end of the decade.

Table 8.

Mexico: Education Expenditure, 1980-2000

article image
Source: Mexico, Federal Executive (1996a and 2001); Mexico, Secretariat of Public Education (1996b), and authors’ estimates.