POLICY RESEARCH WORKING PAPER 2741 Female Wage Inequality In three Latin American countries that introduced in Latin Am erican Labor structural reforms, quantile M arkets regression results show, female workers with less human capital saw wage Luz A. Saavedra gains relative to female workers with more human capital. The World Bank Latin America and the Caribbean Region Gender Sector Unit December 2001 POLICY RESEARCH WORKING PAPER 2741 Summary findings Saavedra uses quantile regression to document the The decline in female wage inequality can be evolution of the earnings structure of salaried and self- explained in part by changes in the premium to employed female workers in urban areas in three Latin education. Results indicate that the relative premium to American countries-Argentina, Brazil, and Costa Rica- education fell in Argentina and Brazil-that is, the after structural reforms were introduced. The analysis adjusted wage differential between more educated and covers pre- and post-reform years: in Argentina, 1988 less educated women decreased between the sampled and 1997, and in Brazil and Costa Rica, 1989 and 1995. years in these countries. In contrast, wage differentials Four primary results emerge from the analysis: arising from education increased in Costa Rica. * After other characteristics are controlled for, wage - Women earning less than their characteristics would premiums to human capital, labor experience, and other predict seemed to fare well with the economic opening: characteristics vary along the conditional distribution. domestic workers, nonwhite workers, and the least This indicates that a homoscedastic model is not suitable educated in the lower quantiles saw their wage premiums for analyzing wage differentials among working women increase relative to those of the control groups. in these countries. These results are consistent with the predictions of the * Wage inequality among women fell in the self- Heckscher-Ohlin theory of trade liberalization: those employment sector in all three countries. In the salaried with less human capital saw wage gains relative to those sector results were mixed, with wage inequality declining with more human capital. in Argentina but increasing slightly in Costa Rica. This paper-a p)roduct of the Gender Sector Unit, Latin America and the Caribbean Region-is part of a larger effort in the region to understand the role of gender in developing country labor markets. Copies of the paper are available free from the World Bank, 1818 H Street NW, Washington, DC 20433. Please contact Selphia Nyairo, room 18-110, telephone 202- 473-4635, fax 202-522-0054, email address snyairo@worldbank.org. Policy Research Working Papers are also posted on the Web at http://econ.worldbank.org. The author may be contacted at saavedra@coba.usf.edu. December 2001. (49 pages) The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas ahout development issues. An objective ofthe series is toget the findings out quickly, even ifthe presentations are less than fully polished. The papers carry the names of the authors and should he cited accordingly. The findings, interpretations, and conclusions expressed in this paper ar, entirely those of the authors. They do not necessarily represent the view of the World Bank, its Executive Directors, or the countrien r they represent. Produced by the Policy Research Dissemination Center Female Wage Inequality in Latin American Labor Markets by Luz A. Saavedra Department of Economics University of South Florida Tampa, FL.  1. Introduction The late 1980s marked a new era in the economic policy of most Latin American countries. Their goal was to give the market a predominant role in the economy by implementing structural reforms that eliminated government intervention in the production and functioning of markets and reduced trade barriers. As a consequence, both domestic and international market competition increased in the region, leading to concern over its distributional effects. This paper will examine women's wage patterns before and after structural reforms were implemented in three Latin American countries, to identify how women fared in the post-reform period as compared to the pre-reform period. Women are the unit of analysis for two reasons. First, although their labor force participation rates have been monotonically increasing over the past thirty years, very little work on women's wages in the region has been done. Second, the question of whether or not structural reforms affect women across labor markets in a similar fashion has not yet been addressed. We do not pretend to establish a causality between the structural reforms and changing wage inequality, but rather to map the direction of change throughout the 1990s to establish stylized facts to move the analysis forward. One important contribution of this work is to recognize heterogeneity within Latin American labor markets. Studies of the manufacturing sector are informative, but the informal sector constitutes 20-70% of the labor market, and the service and commerce sectors are growing in importance. Thus, we also consider the earnings of self-employed entrepreneurs and wage employees in the pre- and post-reform periods to identify those sectors that seem to be the sources of changes in income inequality. An additional contribution of this paper is that we use a different econometric method to examine the question of inequality. Rather than the typical means analysis, we use a quantile econometric approach to explore heterogeneity in female earnings and assess the changes that occurred in wage inequality after the structural reforms were implemented in each country. The benefit of this methodology is that we are able to characterize heterogeneity along the conditional distribution and thereby assess wage differentials between otherwise similar women that have lower or higher wages than their (observable) characteristics predict. This is in contrast to conditional mean analysis (OLS) that concentrates only on giving information about the central tendency of the distribution. The analysis in each country is performed for a pre- and a post-reform year: 1988 and 1997 in Argentina and 1989 and 1995 in Brazil and Costa Rica.' Again, we do not concentrate on linking our findings to trade liberalization. Because additional structural reforms and stabilization programs were implemented in these countries during the analyzed periods, which most likely introduced changes in industrial labor relations, and ' In identifying trends we tried to control for differences in the business cycle by choosing a pre- and post-reform year that were similar in terms of GDP growth. 2 consequently, in the earnings structure within different labor markets, a rather different analysis than the one presented in this paper would be needed to isolate the effects of each reform on the wage distribution. As noted above, the goal is to document wage inequality within labor markets before and after the reforms took place. The paper is organized as follows. In section 2 we briefly outline the economic reforms introduced in each country and the macroeconomic performance during the selected years. In section 3 we discuss issues related to the data and summarize the evolution of human capital and other female worker characteristics between the sampled years. Section 4 outlines the general labor market conditions in each country and presents the changes in (unconditional) wage inequality within each labor market. In section 5 we discuss some; methodological issues and present the evolution of wage inequality after controlling for differences in worker characteristics (human capital and experience) and other characteristics (regional, occupational, industry, etc). In section 6 we conclude. 2. Economic Reforms and Macroeconomic Performance2 The countries examined in this paper were selected as extreme cases during the period of liberalization. Costa Rica, a small and stable economy, had very subtle reforms during the late 1980s and early 1990s. Brazil and Argentina, on the other hand, are both big, rapidly changing economies that had massive restructuring during the period. Brazil, however, has a more flexible labor market that is adjusting to the changes in a very different manner than the more rigid Argentine economy.3 This section outlines the structural reforms and the macroeconomic situation of each country during the selected sample periods. 2.1 Argentina The main structural reforms introduced in Argentina were trade and capital liberalization, deregulation of the financial markets and privatization of publicly-owned enterprises. In addition, Argentina introduced a new monetary exchange system that fixed the exchange rate between the national currency and the U.S. Dollar at a one-to-one rate. One of the goals of this convertibility plan was to avoid the monetization of the fiscal deficit and to stabilize prices. The economy soon started showing the effects of the reforms, especially from the convertibility plan. From 1991 to 1994, the Argentine economy witnessed a significant drop in inflation that reached annual levels close to 5.5% and sustained economic activity that put annual rates of GDP growth at 11.7% (35% during the period). Employment growth was also vigorous, especially during the first years after the implementation of the reforms. Annual employment grew 5.6% but urban unemployment had also grown considerably, from 6.0% in 1991 to 12.2% in 1994. Other changes include significant increases in capital flows and imports and real appreciation of the currency. The appreciation of the currency implied that during this time the economic activity depended exclusively on the internal demand which was also being positively stimulated by capital flows. However, this trend was reversed by the Mexican crisis in 1995. Capital flows decreased considerably and Argentina entered a recession (GDP 2 These outlines are based on Baer (1995), Edwards (1995) and the World Bank (1996). Initially, we had intended to include Colombia, a stable economy with restructuring and the Dominican Republic (to include a Caribbean economy), but due to data issues, those countries were omitted over the process of the study. 3 growth during this year was -4.5%). But the crisis did not last long and by the end of 1997 it had recovered its vigorous growth (8%). Nevertheless, urban unemployment continued to grow and by the end of the year female and male unemployment had reached 17.4% and 13.4%, respectively. 2.2 Brazil Two principle economic reforms were introduced in the years 1990-1994: trade liberalization and privatization of publicly-owned enterprises. With liberalization reform there was a significant reduction of the average import tariff, which fell from 32.2% in 1990 to 12.6% in 1995. In addition, a relatively large number of state-owned firms (twenty-two) were privatized. This period was also characterized by high inflation that lead to the implementation of three main macroeconomic stabilization programs, known as Collor (1990), Collor 11 (1992) and Real (1994). The adjustment plans were designed to reduce aggregate demand and inflation, though inflation was persistent in some years during this period: 1,200% between 1989-1990 and 5000% between 1993-94. However, prices fell significantly in 1995 (inflation was 26%) and continued to fall in 1996 (11%). Deep recessions (1990-1992), partial recoveries (1993-1994) and vigorous growth (1995) characterize the 1989-1995 period. During these years annual average GDP growth and overall unemployment were 1.3% and 5%, respectively. 2.3 Costa Rica Trade liberalization, privatization and financial deregulation have been on the economic reform agenda in Costa Rica since 1986. However, reform implementation has been rather slow and incipient. For instance, the liberalization program consisted of a gradual reduction of tariffs complemented with a package of fiscal incentives to exports, and the creation of free industrial zones. During the 1986-1990 period tariffs on consumption and capital goods fell. Some sectors, like the textile industry, were subject to a slower pace on tariff reduction. By the end of 1993 this process had deepened as tariffs reached 14%. On the other hand, privatization of publicly-owned enterprises has been less important in Costa Rica. Even though some public institutions and publicly- owned firms were privatized, the emphasis of the public sector reform was concentrated more on the enhancement of efficiency rather than on privatization itself. Finally, the bulk of structural reforms also included the introduction of a new banking law in 1988. The main elements of this law were deregulation of the entire financial system, especially the liberalization of commercially-owned state banks. During this period (1986-1995) economic growth was vigorous (4% per year), non-traditional exports increased significantly (27%) and unemployment remained low (5% per year). However, inflation and the fiscal and commercial deficits were significant, especially during 1992-1993. 3. Data The data sets used in the analysis are from the household surveys for the urban areas of Argentina, Brazil and Costa Rica. For Argentina, the data are from the Permanent Household Survey (Encuesta Permanente de Hogares, EPH) for the years 1988 and 1997 (October waves). For Brazil, they are from the Pesquisa Nacional por Amostra de Domicilios (PNAD) for the years 1989 and 1995, and for Costa Rica they are from the Encuesta de Hogares de Propositos Multiples (EHPM) for 1989 and 1995. 4 The analysis considers working women, both salaried and self-employed, aged 15- 70.4 In addition, the samples from each country are restricted to those working women who report positive wages and number of hours of work, as defined in each respective survey.s The wage variable used in the analysis is the hourly wage.6 Furthermore, observations with missing data on any of the variables of interest were dropped. Table 1 (a-c) contains summary statistics for the variables used in the conditional analysis for each country.7 The human capital variables include schooling, experience, and tenure (Argentina) in the current job. The set of worker characteristics includes age, race, position in the household and marital status. The occupation variables include dummies for professional, skilled and unskilled women.8 The regional variables include three dummies for Brazil (the northeast, south and southeast regions), and five dummies for Costa Rica (San Jose metropolitan area, rest of the central region, Chorotega and Brunca, Central Pacific, Atlantic and North Huertar). No regional dummies are included in the specifications for Argentina since our sample corresponds to Greater Buenos Aires only. Industry characteristics include dummies for manufacturing, retail, personal services, financial services, professional services, public services, social services, and a residual dummy called "other" that includes non-identified industries. Finally, a dummy for informal salaried and for domestic service are included to determine differences within each labor market with respect to what is often identified as vulnerable female employment.9 3.1 Characteristics ofworking women. Over the analyzed periods, female workers increased their educational attainment across all countries. The proportion of female workers with more than twelve years of education increased significantly and there was a decline in the proportion of women with less than six years of education. However, the increase was not uniform across labor markets. In all three countries the proportion of women with a college education increased more in the salaried sector than in the self-employment sector.'0 A second pattern in all three countries is that women working in the self-employment sector were on average older than were women working in the salaried sector. The proportion of women older than 45 increased in both labor markets during the sample periods while the proportion of younger women decreased, perhaps due to the increase in educational attainment of the female working population during the sampled years. We 4The self-employed sector includes domestic service. Female family workers and employers are excluded from the analysis. s All the surveys contain information on the number of hours worked the week previous to the survey, and the monthly earnings from the job. 6 Except in Brazil, wages are deflated using the corresponding consumer index. In Argentina wages are measured at 1988 Pesos and in Costa Rica at 1989 Colones. In Brazil wages are measured in US dollars of 1995. 7 The figures on the tables are non-weighted by the survey inflation factors. The weighted sample proportions were very similar to the non-weighted proportions. 8 Unfortunately, our samples did not include data on family background and characteristics, which have been shown to be important in determining wage differentials in developing countries (see for example Lam and Schoeni (1993) and the references there in). 9 See the Appendix for a definition of all variables. "o Because there are some differences in the education systems, it is not appropriate to make exact comparisons of the education variables across countries. 5 also observe that the incidence of headship increased in all three countries between the sample years. There were not significant changes in the average value of potential experience, measured as (age - the number of years of schooling - six). In Brazil and Costa Rica for the sample years, self-employed women had more work experience than salaried women. In Brazil and Argentina the average potential years of experience decreased for salaried and self-employed women, while in Costa Rica, it did not change for salaried women, but it increased for self-employed women (from 23 to 25 years). In all three countries the service sector, and in particular, the public sector had the biggest share of the population of working women. However, in our sampled periods, the proportion of women working in the public sector decreased significantly in Costa Rica while it slightly increased in Argentina and Brazil. Even though privatization has been rather gradual in Costa Rica, the implementation of stabilization programs that reduced public expenditures and cut public employment during 1989-1995 explains the downward trend in this sector. In Argentina, and especially in Brazil, most public workers enjoy tenure rights, which constitute a constraint to the reduction of public employment. In contrast, the proportion of women working in the manufacturing sector declined in all three countries within each labor market. For instance, in Argentina, where the manufacturing sector had an important share of the working population in 1988, the proportion of women working in this sector, whether as wage employees or as self- employed, declined by almost half by 1997. In Brazil and Costa Rica there was also an important relative decline of women in this sector." However, in the post-reform years, the contribution of this sector to female employment was still relatively important in these two countries. Finally, "vulnerable" female employment in the unprotected salaried sector, proxied by the proportion of women without government-mandated benefits working in small firms, slightly increased in all three countries. The biggest increase occurred in Brazil, where the proportion of informal salaried increased from 16% in 1989 to 17.2% in 1995.12 Furthermore, there was a significant increase in the proportion of women working as domestic servants in this country. In contrast, this proportion decreased in Argentina and Costa Rica. 4. General Labor Market Indicators. 4.1 Female participation, employment by sector and unemployment Women's labor force participation over the past decade has increased in all three countries, as well as in the region as a whole, but it remains low in comparison to the US and other developed countries. In Brazil and Costa Rica participation rates increased from 43.9% to 51.0%, and from 59.3 to 61.1% between 1989 and 1995, respectively, while in Argentina the rate increased from 40.2% to 44.9% between 1988 and 1997. These figures indicate that the structural reforms undertaken by these countries did not reverse the trend of increasing female labor participation observed in the years previous " At the time being we are not able to determine if this fall in manufacturing employment is due to a self- selection process or to a fall in demand due to increases in competitive pressures in the manufacturing sector. Further research is needed to understand this trend. 12 See the Appendix for the definition of informal salaried women. 6 to the reforms. In fact, Pessino (1998) reports that female participation peaked in Argentina after the convertibility plan was put in place in response to an increase in male unemployment.13 The participation of females in each sector of the labor market also differs across countries. Self-employment is clearly less important for Argentine and Costa Rican women than for Brazilians. For the pre- (post-) reforms years, the percentage of women working as self-employed was approximately 23.5 (20.4) in Argentina, 23.0 (20.8) in Costa Rica, while it was 36.2 (41.7) in Brazil, revealing an upward trend in the proportion of women working in the self-employment sector in Brazil, while the opposite occurred in Argentina. and Costa Rica.14 Furthermore, the ratio of women working in the salaried sector to women working in the self-employment sector rose in both Argentina and Costa Rica, but decreased in Brazil. Finally, female unemployment increased in all three countries but at different rates. Argentina experienced the biggest increase of approximately 1.2% per year, Brazil ranked second with an annual increase of 0.95%, but in Costa Rica, female unemployment did not change significantly. These figures, together with the overall female participation rates, suggest that the increase in Argentine unemployment is due to both an increase in labor force participation of women and job loss. On the other hand, in Brazil, even though unemployment grew between the sampled years, it did not surpass the growth observed in female participation.'5 In conclusion, female labor force participation rates continued to grow throughout the structural reform period. In Brazil, this is associated with an expansion of the self- employment sector, while in Argentina and Costa Rica it is associated with an increase in the informal salaried sector. Even though we cannot establish a direct connection between female informal employment, self-employment and the structure reforms, the figures indicate that after the implementation of the reforms the employment incidence in these sectors increased more in the two countries that implemented major changes. In addition, we observe a high incidence of female unemployment during these periods. In both Argentina and Brazil, unemployment rates rose, but in Costa Rica, a country with sluggish economic restructuring, female unemployment rates did not change. Unemployment may have increased due to the observed increased in wages (see next section) and to a decreased in demand for workers due to competitive pressures. A within country comparison reveals that after the economic reforms, female participation in these countries is still very low as compared to male labor force participation. For the same periods male participation was between 80.6% - 81.3%, 84.6% - 83.7% and 81.6% - 82.0% in Argentina, Brazil and Costa Rica, respectively (Arias, 2000). 1 A similar trend observed in all three countries is the upward increase in the proportion of women within the other sector. This group includes family workers and non-specified categories. In all three countries, this proportion increased by about two percent during the sampled years. Is Comparing the evolution of female unemployment with male unemployment, women had higher unemployment rates than men during both the pre and post-reforms years (except in Costa Rica, where unemployment rates were similar). In the Post-reform years the differences are bigger in Argentina and Brazil. 7 4.2 Earnings, Wages, and Inequality16 This section highlights the growth patterns of wages (remuneration provided by employers to their employees), earnings (business profits, without excluding the returns to capital or the opportunity cost of inputs), and inequality measures within and between sectors. The sector of participation is divided along many lines. First, we consider wage workers and the self-employed where the former includes informal, formal, and public sector workers and the latter is both self-employed and domestic servants.17 A second slice to the data breaks the wage sector into formal wage workers (including public servants) and informal wage workers and treats self-employed and domestic servants separately. We use several exercises to map the trends, including means tables, inequality measures, and wage (earnings) densities. Tables 3, 4, 5, and 6 report average real hourly wages for the sampled years sectors. In addition, the table shows the evolution of seven quantiles of the hourly wage (unconditional) distribution in each sector. The 5th and 10th quantiles identify "low wage earners," the 25th, 50th and 75th quantiles identify the "moderate" and the 90th and 95th quantiles identify "high wage earners." Figures 1 (a-c) show plots of the estimated wage densities for formal and informal salaried and for self- employed.'8 These plots describe changes of the wage distribution along all quantiles during the sampled periods. Figures 2(a-f) also present these plots for salaried and self- employed women according to their educational attainment and age. We will discuss comparisons between quantiles (holding year and sector fixed), between years (holding quantile and sector fixed), and between sectors (holding quantile and year fixed). 4.2.1. Wage Earners Mean hourly wages increased in all three countries across the period, but a breakdown by quantile shows that a means analysis is not particularly informational since the rates of increase were not uniform across the distribution (Figures 1 (a-c) & Table 3). In Argentina and Brazil, wages of women below the median increased significantly more than wages of women above the median.19 Moderate earners also obtained greater wage increases than high earners did. This pattern of wage growth implied a reduction in wage inequality that is in line with the predictions of a Heckscher-Ohlin model that predicts that in developing countries, where unskilled labor is abundant and skilled labor is scarce, trade affects relative prices in favor of the abundant factor, i.e. it increases the demand for (and thus the price of) the abundant factors while it reduces the demand for the scarce factors. Consequently, trade should reduce wage inequality. In Costa Rica the story is a bit different. Women at the bottom (5th quantile) and at the top (90th-95th quantiles) benefited the most. These patterns in wage growth can be 16 We present several measure of inequality, however, it is important to keep in mind that different measures of income inequality can give different and competing results in terms of the magnitude of the changes in inequality. See Karoly (1992) for a comparison of different measures of wage inequality in the US. 1 Family workers and firm owners are omitted due to sample size considerations and because their labor force participation and returns patterns are distinct from individuals who supply their own labor in return for remuneration. 's Densities are estimated using kernels. The bandwidth was chosen using 1.144*var(logwage)*(n)^(-1/5), Siverman (1986) 'Bear in mind that for Argentina the analyzed periods is 1988-1997, while for Brazil and Costa Rica it is 1989-1995. 8 further illustrated by the change in the inequality measures presented in Table 5. The ratio between the wage at the 0.90th and 0.1P0 quantiles decreased in Argentina (from 5.78 to 4.78) and Brazil (from 9.96 to 9.36) and slightly increased in Costa Rica (from 4.53 to 4.74). A rather different picture emerges for Argentina when we break down the sample between informal salaried and formal salaried. For formal salaried, the figures show a reduction in wage dispersion that is bigger than the one observed in the sector as a whole. In addition, because wage growth consistently decreased as the quantiles increased in this sub-sector (see table 4), wage inequality decreased among all women within this group. In contrast, in Brazil and Costa Rica, there were not remarkable differences within the formal salaried sector between the pre- and post-reform years. On the other hand, wage inequality in the informal sector increased in Argentina but decreased in Brazil. In Argentina, the bigger increase in wages occurred at the higher quantiles. In fact, except for wages at the 5th quantile, wages increased monotonically with the quantiles. Consequently, wage dispersion increased significantly within this sub- sector. In Brazil, though, wage inequality decreased significantly within this subsector. In fact, the reduction in wage inequality observed in the salaried sector as a whole can be attributed to the significant reduction in the informal sector. In contrast, wage inequality did not change in Costa Rica within the informal salaried sector. The 90th/10th wage ratio decreased from 4.2 in 1989 to 4.0 in 1995. Comparing earnings growth of salaried formal with salaried informal, the data from Brazil and Costa Rica indicate that informal wages increased more than formal wages did. Figure 3a-c shows that the rightward shifts of the (estimated) densities are substantially bigger for informal salaried than for formal salaried. This trend was uniform along the whole distribution (except in Brazil for the 25th and 75th, where formal salaried did better, see Table 4). In Argentina, the trend was different. Only informal salaried at the 75th-90th quantile did better. The opposite is true at the other quantiles. In conclusion, our data show that after the structural reforms the observed patterns in wage dispersion varied across countries and across labor markets within each country. While in Argentina wage inequality decreased in the formal sector, it increased significantly in the informal sector. In Brazil, wage inequality decreased in the informal sector, but did not change in the formal sector. On the other hand, there were not significant changes in Costa Rica across labor markets.20 In addition, while wage differentials between formal and informal decreased in Brazil and Costa Rica, they tended to increase in Argentina. The implication of this is that there is not a clear pattern of changes in wage dispersion across countries and across labor markets after the implementation of the structural reforms. 20 A clear pattern in this country is that wages grew sluggishly along the whole wage distribution causing little change in wage dispersion between the sampled years. 9 4.2.2. Self-employed Mean hourly earnings, i.e. profits taken home by the self-employed and domestic servants, increased in all countries across the periods by even more than wages did. Hourly earnings of women in the self-employment sector followed the same pattern observed in Brazil and Costa Rica's salaried sector, but there was an opposite pattern in Argentina. Brazil's low earning women in the self-employment sector obtained bigger increases than high earning did, and Costa Rica's low and high earners enjoyed the biggest gains (see Table 3). Argentine self-employed women who were high earners experienced higher earnings growth than did the low earners. Except for women at the 95th quantile, who obtained the smallest wage increase in this sector (12.7%), high earning women enjoyed increases that were nearly twice as much as the increase obtained by low earning women. Consequently, self-employed women experienced reductions in wage inequality in Brazil and Costa Rica, while the opposite was observed in Argentina (see Table 6). However, after excluding domestic servants, wage inequality decreased within this sector in Argentina. The same patterns were observed in Brazil and Costa Rica. Comparisons of the hourly earnings across labor markets at different points of the earnings distributions reveal that in all three countries, salaried workers earn more than self-employed workers (except average wage and high wage women in Argentina), but the rate of increase of wages along all quantiles was higher for self-employed than for salaried, as illustrated by the larger rightward shifts of the earnings density (except in Argentina at the lower quantiles. See Figure la-c). Brazil presents the biggest wage differential across these two sectors. Furthermore, the earnings differential across sectors is bigger for low wage than for high wage. For example, in Brazil in 1989, salaried women at the 5th and 10th quantiles earned roughly 3.4 times more than the self-employed did at the same quantiles, while salaried women at the 90-95th quantiles earned between 2.0 and 1.8 times more than their self-employed counterparts. In sum, wage inequality within the self-employment sector decreased in Brazil and Costa Rica, and increased slightly in Argentina. After excluding domestic servants from the sample of self-employed women, the patterns changed in Argentina and Costa Rica. In Argentina inequality decreased and in Costa Rica it did not change. In addition, before and after the structural reforms, women working in the salaried sector earned more than women working in the self-employment sector. However, increases in real earnings during the analyzed periods were significantly greater for self-employed. This reduced the relative wage differentials across these labor markets.21 Overall, except for the increase in wage inequality among the informal salaried in Argentina, the post-reform years show that wage dispersion did not change or decrease 21 One qualification is important at this point. Note that the described earnings differentials across sectors are "unconditional differences." In other words, they are not obtained from a conditional wage analysis, and therefore, these differentials are not informative to conclude that employment in one sector is better than in the other. For a discussion about the controversy of estimating wage differentials among labor markets see Maloney (1998). 10 within any labor market. In addition, wage differentials between formal and informal salaried and between salaried and self-employed also tended to decrease. 4.2.4. Demographic comparisons When splitting the samples by educational attainment and age some interesting patterns are revealed (see figures 2(a-f)). In all three countries and in both labor markets, working women with less than six years of education appear to have gained more than working women with higher levels of human capital. Furthermore, there is not a clear trend in the evolution of the distribution of hourly earnings by age (a proxy for experience). While in Argentina and Costa Rica, the young (less than 25) and older (more than 45) workers did relatively better than middle-aged women (26-44), in Brazil, middle-aged and older women did relatively better than younger women. In fact, the pictures indicate that within these specific groups some working women experienced wage deterioration. For example, within the group of self-employed women with college education, some low earning women experienced real wage deterioration across the three countries. This is also the case in Costa Rica for some low to moderate wage women in the salaried sector. The previous analysis indicates that there were substantial reductions in wage inequality among salaried women in Argentina and Brazil, while there were no significant changes in Costa Rica. However, while wage inequality increased among informal salaried workers in Argentina, it decreased significantly in Brazil. In addition, wage inequality decreased among the self-employed in all three countries (in Argentina after excluding domestic servants). Furthermore, in all three countries wage differentials between the formal salaried and the informal and self-employed decreased. Except for the significant increase in wage dispersion within the informal salaried sector in Argentina, these patterns are consistent with the view that liberalization reforms tend to decrease wage inequality. The conditional analysis presented in the next sections illustrates wage differentials after controlling for observable characteristics. 5. Wage equations 5. 1 Econometric Approach This section presents estimates on wage determinants and how they contribute to the changes in wage inequality between the pre- and post-reform sampled years. We control for standard variables that measure worker human capital, experience (tenure in Argentina), occupation, region and industry. The dependent variable is the logarithm of the hourly wage. Instead of estimating conditional means by using OLS, we use quantile regression techniques (Koenker and Basset (1978)) to estimate returns to characteristics of different types of women along the conditional wage distribution. When using this econometric method, we can estimate the coefficient of any conditional ith quantile (known as a regression quantile). A good characterization of the conditional distribution is obtained by estimating a set of "representative" quantiles, i.e., the 0.10th, 0.25th, 0.50th 0.75th, and 0.90th quantiles. Intuitively, these regression quantile estimates convey I ( information on wage differentials arising from non-observable characteristics among women who are otherwise observationally equivalent. In other words, by using quantile regression we can determine if women that rank in different positions in the conditional distribution (i.e., women that have higher and lower wages than predicted by observable characteristics) have different premiums to education, experience, or to other relevant observable variables. In addition, by using quantile regression, we can compare the returns to these variables for different quantiles across the period, to identify those characteristics that contributed to the changes in wage differentials after the implementation of economic reforms.22 In the specification of the model we include several binary variables for different levels of education, tenure and age.23 This specification is convenient to determine differentials in wage premiums within and between education, experience and age groups. The base group (i.e. the omitted dummy variables) is composed of women with college, older than 45, with more than 10 years of tenure, unmarried, working in the manufacturing sector and professional. Except for Argentina, where we observe tenure on the current job, we will treat experience as a non-observable characteristic.24 Quantile regression estimates and their corresponding p-values are presented in tables 8(a-b) for Argentina, 9(a-b) for Brazil and 10(a-b) for Costa Rica. Tables with the letter a contain estimates for the salaried sector, and tables with the letter b contain the estimates for the self-employment sector. 5.2 Heterogeneity Quantile regression estimates in Tables 8, 9 and 10 indicate that a conditional mean model was not suitable to analyze wage differentials among working women in all three countries. We find heterogeneity along the conditional wage distribution in both years. The estimated wage differences between different groups of working women differed along the quantiles. An illustration of this heterogeneity in Argentina is given in Tables 8a and 8b. Note that the adjusted premium to educational attainment increased monotonically over the quantiles. We can see this by looking at the quantile regression estimates of the dummies element and high in table 8a. Recall that each dummy measures the log wage differential among women within the group represented by the dummy and those in the omitted group. For instance, in 1988, the estimates for element indicate that working females with college education ranking at the bottom of the 22 Recently, robust models of sample selection are becoming an active research topic in the context of quantile regression estimation. However, at the time of writing this paper there is not an estimator available that will allow us to control not only for selectivity into the labor force, but also into each labor market. A multivariate participation model in the context of quantile regression estimation is still an open research question. This imposes a constraint in our quantile regression estimates. Mainly, that we can only make inferences about the female population working within each labor market. It is important to note that the use of parametric sample selection methods (mainly, Heckman (1974)) to measure earnings differentials across formal and informal labor markets in Latin American has not been satisfactory due to the sensitivity of the approach to statistical assumptions. For a discussion of this issue see Maloney (1988, 1998) and the references therein. 23 See the Appendix for a description of the dummy variables. 24 We estimate a second specification for each country in which we use the standard measure of potential experience, which is defined as age, minus the number of years of schooling minus six. In addition, Following Buchinsky (1998), we include the interaction between this variable with number of children in the family. The idea is that child bearing can be seen as the main alternative use for women's time. Results were similar and therefore they will not be discussed, however, the estimates are available upon request. 12 (conditional) wage distribution earned 74% more than low wage workers with elementary education. This indicates that the premium for a college education over an elementary school education is 74% for those types of women who tend to be under-performers compared to their peers who have otherwise similar characteristics. On the other hand, the premium for college is 111% for individuals who tend to be high earners, relative to their demographic group. This evidence of heterogeneity is also present in the evolution of wage differentials arising from demographic and regional industry characteristics as well. 5.3 Salaried Women 5.3.1 Within Year Estimates Most of the control variables are significant in all three countries. As expected, education and wage differentials among different education and age groups are significantly different from zero at different quantiles.25 Similarly, wage differentials among women with different educational attainment varied along the conditional distribution. For instance, after controlling for other factors, in Costa Rica in 1989 the relative premium of college education over elementary education ranged from 33.9% at the O 10t quantile to 86.1% at the 0.90th quantile. The premium over high school ranged from 17.9% to 39.6%. This indicates the higher the level of education the lower the variation across quantiles in a given year. A similar pattern is observed in Argentina and Brazil (see Tables 9b and 1Ob). This implies that non-observable characteristics increase wage inequality within groups with relatively low education (along quantiles) more than within higher education groups. Wage differences among women in different age groups decreased with age. In addition, wage differences among women within the same age group increased over the quantiles. For instance, in Brazil in 1989 women 45 years and older at the 0.90th quantile earned 92.1% more than women less than 25 years old at the same quantile. The same group earned 47.8% more than high wage aged 25-35 and 14.7% more than high wage aged 35-45. The same patterns were observed in Argentina and Costa Rica (see Table 8a- 10a). Wage differentials between mid-age women (25-35) and otherwise observationally women older than 45 are also significant, especially at the higher quantiles (except in Argentina in 1988). We estimate significant wage differentials arising from specific industry and from regional variation.2 After controlling for other factors, women working in financial services enjoy a wage premium over women in the manufacturing sector. This premium is similar across quantiles in Argentina, but increases over the quantiles in Brazil and Costa Rica. Furthermore, Argentine and Costa Rican women in the public sector also enjoy a positive premium over women in the manufacturing sector, though in Argentina is only at the low quantiles (0.10-0.50). The opposite is observed in Brazil, though the premium decreases over the quantiles. Furthermore, while in Brazil and Costa Rica there are 25 A third specification in which we assume that the premium to education does not change with educational attainment, age or experience shows that in 1989 the estimated marginal return to education ranged between 10% and 14% in Brazil and between 4.0% and 6.6% in Costa Rica for women at the low (0.1 0'h) and high (0.90th) quantiles, respectively. 26 See the Appendix for the definitions of these dummies. 13 significant adjusted wage differentials between women working in manufacturing and women working in service-related activities (personal and social services) they are not significant at most quantiles in Argentina. Finally, after adjusting for other characteristics of the female working population, quantile regression estimates indicate that there was a positive and significant wage premium for the formal salaried over the informal salaried in all three countries. However, the premium is more important at the low quantiles, i.e. it decreases or it is not significant at the higher quantiles. For instance in Argentina and Costa Rica this wage premium does not exist at the high quantiles. This indicates that after controlling for other characteristics, wage differentials between formal and informal salaried are important among those women that earn less than what their observed characteristics predict. For those women that earn much more than their characteristics predict there are not significant (adjusted) wage differentials. Wage differentials also arise,from women living in different regions in each country. Working women living in urban areas in the south of Brazil enjoyed (at the median) 35% higher wages than otherwise observationally similar women living in urban areas in the northeast of the country. The premium decreased along the quantiles. Similarly, in Costa Rica the median wages of women living in the most developed area of San Jose area were significantly higher than wages for women in the other (poor) urban areas of the country. It is important to note that regional differences in Costa Rica were significant only at a few quantiles (mainly right tail quantiles). Significant wage differentials arise from variations in ethnic background in Brazil (race was not available for Argentina and Costa Rica). After controlling for other factors, black and mulatto women earn significantly lower wages than white women. On the other hand, there are no significant wage differentials between whites and Asians. This is in contrast to the positive wage premium of Brazilian males with Asian roots who enjoyed a high wage premium over whites (Arias, 1999). Furthermore, we estimate a significant and positive wage premium to marriage that increases over the quantiles in Brazil (1995) and Costa Rica, and decreases over the quantiles in Argentina. However, in Argentina the quantile regression coefficient is only significant at the left tail quantiles. Finally, there is a significant wage return to tenure in the current job in Argentina, which did not vary significantly over the quantiles. However, the premium varies among women with different tenure levels, especially between new hires and old hires. For instance, we find significant wage differentials between (otherwise) similar women with more than 10 years and women with less than 5 years of tenure, while there are no significant differences between those with more than 10 years and those with 5 - 10 years of tenure (see Table 8b). These estimates indicate that wage inequality is important not only among different groups of salaried women but among women within specific groups. The following section shows the evolution of these wage differentials in the post-reform years. 5.3.2 Between Year Differences. 14 Education Comparing across years, wage differentials between (otherwise similar) women with different levels of education in Brazil and Argentina decreased between the pre- and post- reform years. The premium for high school relative to elementary and college relative to elementary or high school fell for nearly all quantiles. For instance, the adjusted wage differentials between women with elementary education and high school decreased at the median quantile from 60% to 39% in Argentina, and from 59.2% to 54.3% in Brazil. In contrast, in Argentina the adjusted wage premium between women with some college education and women with a high school degree increased at all quantiles. In contrast to the results found for Argentina and Brazil, in Costa Rica human capital contributed to increased wage differentials among (otherwise similar) women with different levels of education between the sampled years. The relative premium between women with elementary education and women with high school and college education increased at all quantiles between 1989 and 1995.27 For instance, in 1989, women with an adjusted median wage with college education (the omitted dummy) earned 33.4% more than the corresponding middle wage women with elementary education (element) and 51.6% more by 1995. Finally, adjusted wage differentials between women with technical education (univ1) and women with college increased at the right tail quantiles. Note that while in 1989 the adjusted wage differentials between these two groups were not significant at any quantile, they were significant at the higher quantiles in 1995. Therefore, those with particularly specialized technical skills are losing out with respect to college graduates in the new market. Furthermore, wage inequality within each education group also decreased for salaried women in Argentina and Costa Rica and for the self-employed in Argentina and Brazil. We can see this by comparing the variation of the quantile regression coefficients along the conditional distribution in each year. Therefore, education contributed to decrease wage differentials not only among women in different education groups (after controlling for other characteristics) but also among women within these groups. Experience As noted before, we treat labor experience as non-observable and include age as a proxy for experience in general. After controlling for other factors, the wage premium for older women over younger women increased at most quantiles between the sampled years in all three countries.28 In addition, (adjusted) wage inequality between women within the same age group increased in Argentina for the youngest (less than 25), did not change in Brazil and decreased in Costa Rica. 27 Only wage differentials between women with secondary and elementary education decreased at nearly all quantiles in this country (except at the 0.75t). 28 Except in Brazil and Costa Rica where there were reductions in wage differentials between the oldest (women older than 45) and the youngest (women less than 25 and between 25-35) at the upper quantiles. 15 Experience on the job (tenure) in Argentina contributed to increased wage inequality among the newer and more experienced workers, but only at the low quantiles. Wage differentials arising from tenure between new entrants and women with more than 5 years of tenure increased at the low quantiles but decreased at the high quantiles. On the other hand, wage differentials among more experienced women were not significant in either year. Skill Level Our estimates indicate that there was a significant wage premium to specific skills on the job in all three countries. The premium increased over the quantiles (except in Brazil) and across years. For instance, in Argentina after controlling for other characteristics, the wage differential between women working in professional jobs and women working in low- or high-skilled jobs increased between the pre- and post-reform years over the estimated quantiles. In particular, wage differentials increased between professionals and skilled workers, and between professionals and unskilled workers. In addition, wage differentials between skilled and unskilled workers increased along all quantiles (see Tables 8a-b). A similar pattern is estimated in Brazil and Costa Rica (except at the 10- 25t quantiles between professionals and white-collar workers). Industrial Sector The adjusted wage premium that women working in financial services enjoyed over women working in the manufacturing sector in the pre-reform years decreased at nearly all quantiles in the post-reform years in Argentina and Brazil (except at the 0.10th quantile). In Costa Rica this premium vanished for women at the upper quantiles and became negative for women at the lower quantiles. Furthermore, while the adjusted wage premium of women working in the public sector over women in the manufacturing sector decreased in Argentina, it increased at all quantiles in Costa Rica. On the other hand, the (adjusted) wage advantage that Brazilian women working in the manufacturing sector had over women in the public sector decreased significantly in the post-reform year, especially at the low quantiles. Informality After controlling for other observable characteristics, quantile regression estimates of the coefficient of the dummy for informal salaried workers are consistent with the reduction in wage differentials between women in the formal and informal sectors documented in Table 4. For instance, in 1989 in Brazil and Costa Rica, formal salaried women with an adjusted median wage earned 34.4% and 33.8% more than the corresponding middle wage informal salaried women. In 1995, the estimated adjusted wage differential between these two groups decreased to 29.9% and 9.2%, respectively. Similar changes were observed at the other quantiles, but they are not monotonic. For example, in Argentina, the inequality at the low quantiles in 1988 decreased at the 0.25- 050' quantiles, but increased slightly at the 0.1 0th quantile. The increase in wage inequality among informal salaried women in Argentina and the decrease in Brazil documented in previous sections is also illustrated by the increase in variation of the quantile regression estimates along the conditional distribution (see Tables 8a and 9a). 16 Demographic Characteristics The positive premium to marriage over single women estimated in Argentina in 1988 decreased at the low quantiles in 1997. As in 1988, quantile regression estimates were not significant at the right quantiles in 1997. This is in contrast with the findings for the male population, for which a premium to marriage appeared to be consistently significant during these years (Arias (2000)). In Costa Rica the positive premium increased at the higher quantiles and did not change at the lower quantiles. Marital status was not available for Brazil in 1989, but the quantile regression coefficient is significant at all quantiles in 1997. The adjusted wage premium that working women in the southern urban areas of Brazil had over working women living in the northeastern urban areas decreased between 1989 and 1995 at the low quantiles, but increased at the high quantiles. In other words, after controlling for other characteristics, regional wage differentials decreased for low earning women but increased for high earning women. For instance, in 1989, low wage women (at the 10th quantile) working in the south earned 62% more than (otherwise similar) women earned in the north. This wage premium differential decreased to 42% in 1995. In contrast, high earning women in the south earned 13% more than (otherwise similar) women earned in the north in 1989, while they earned 18% more in 1995. Similar changes in wage differences are found between women in the southeast and northeastern regions (see Table 9a). On the other hand, there is no clear pattern of change of the regional wage differences estimated in Costa Rica. They decreased at some quantiles (025, 0.50 and 0.75th), but increased at others. Finally, in Brazil the wage differentials arising from different ethnic backgrounds consistently decreased at all quantiles between 1989 and 1995 (see Tables 9a-b). 5.3.3 Summary In summary, results from the conditional analysis indicate that part of the reduction in wage inequality in the salaried sector in Argentina and Brazil can be explained by the reduction in the relative wage premium between women with low and high levels of human capital. This reduction in wage differentials due to education must have compensated, in part, for the increase in inequality due to higher wage premiums over new entrants, younger women and unskilled workers. For instance, in Argentina the overall inequality due to observable characteristics decreased during the sampled years. The wage ratio between the low and high earners decreased from 3.84 in 1988 to 3.75 in 1997.29 Furthermore, wage inequality decreased among women within the base group. Estimated intercepts in Table 8a, which represent the logarithm of hourly wage for this group across quantiles, indicate that the wage ratio between the 10th and 90th quantiles decreased from 2.19 to 1.76 in 1997. Similar results are obtained for salaried women in Brazil. In contrast, human capital contributed to increased wage inequality in Costa Rica. After controlling for other factors, the relative wage premium of more educated women over less educated women increased at all quantiles between the pre- and post-reform years. 29 This ratio is computed using the predicted wage at each quantile, where the dummies are evaluated at the means, and the continuous variables are evaluated at the medians. 17 Finally, the decrease in wage differentials between formal salaried and informal salaried workers documented in section 4.2 (except in Argentina at the 10th quantile) is also illustrated by the reduction in the estimates of the informal dummy variable for all quantiles. Thus the non-observable characteristics contributed to decreased wage differentials between women in these two sectors. 5.4 Wage Equations for Self-employed Women While for Argentina and Costa Rica the quantile regression estimates of many of the conditioning variables were not significant in any of the specifications of the earnings equations for this sector, most of them were significant for Brazil.30 In the discussion of the results that were statistically significant in each country, comparisons between quantiles within years and across years are presented in the same subsection. 5.4.1 Within and between Year Estimates Education The returns to education and wage differentials between different education groups were significant in all three countries at most quantiles.31 As in the salaried sector, wage differentials arising from variation in educational attainment increased along the quantiles in Argentina. On the contrary, they decreased along the quantiles in Brazil and were not significant at all quantiles in Costa Rica.32 For instance, in Argentina in 1988 the adjusted wage premium of college over elementary education increased from 125% for women at the median to 384% for women at the top quantile (0.90th), while in Brazil in 1989, the premium decreased from 298.4% to 206.2%. Similar patterns are observed for the estimated (adjusted) wage premium of college over high school and incomplete college (Argentina). Furthermore, wage differentials within each education group (given by the variation of the quantile regression estimates of a given education dummy variable) were bigger for self-employed than for salaried women. This implies that in these two countries (conditional) wage differentials within education groups in the self-employment sector are more severe than in the salaried sector. Adjusted wage differentials between the most educated (women with college) and the least educated (women with elementary education) decreased at all quantiles (except at the 10th quantile) between 1988 and 1997 in Argentina (see Table 8b). In addition, the adjusted wage differential between (otherwise similar) women with college and secondary education decreased at the top quantiles and increased at the bottom quantiles. Estimated (adjusted) wage differentials arising from educational attainment in Brazil showed the same trend observed in the salaried sector: they decreased between the most 3o Most of the estimates at the extreme quantiles (I 'h and 951h) could not be estimated with precision (p-values are close to one). This is due to the fact that there were not enough data points at these quantiles. Because some of these variables were significant at the median quantile we also estimated wage equation by OLS. Results are discussed along with the significant quantile regression estimates. OLS estimates are available upon request. 3 The third specification of the model shows that self-employed Brazilian and Costa Rican received a marginal return to their education that ranged between 8.9-9.0% and 7.0-5.8% along the 0. 10" and 0.90th quantiles, respectively. Note that these estimates are smaller that those obtained for salaried women (see footnote 23). 32 Similarly, OLS estimates indicate that there was not a (statistically) significant adjusted wage premium to college education over elementary and high school education. In other words, after adjusting for other factors differentials in educational attainment do not explain wage differentials between self-employed women in Costa Rica. This result obtains along the whole conditional distribution. 18 and least educated self-employed between 1989 and 1995. Furthermore, adjusted wage differentials also decreased between women with college and secondary education, and between women with secondary and primary education (see Table 9b). Furthermore, the difference between the quantile regression estimates for a given education group (dummy) are smaller in the post-reform period in Argentina and Brazil. This implies that wage differentials within each educational group decreased between the pre- and post- reform periods for self-employed women in these two countries. Estimates for Costa Rica are not conclusive since quantile regression estimates were not significant. These results are similar to those obtained for salaried women: adjusted wage differentials between and within different education groups decreased in the post-reform periods. Skills In Argentina, after controlling for other characteristics, there was a significant wage premium for specific skills on the job, which increased along the quantiles in both the pre- and post-reform years. As was the case for Argentine salaried women, the relative premium increased at all quantiles between these years. Similarly, the professional self- employed in Brazil earned more than the non-professional self-employed did. However, in contrast to the patterns estimated for salaried women, the relative premium for Brazilian women decreased at all quantiles between the pre- and post-reform years. In contrast to the findings for the salaried sector, self-employed women working in the manufacturing sector in Brazil earned less than (otherwise similar) self-employed working in retail, service and financial sectors. However, these adjusted wage differentials decreased at most quantiles (except at the 0.90th quantile for women working in social services) between 1989 and 1995. On the other hand, there were significant sectorial wage differentials in Argentina only at the median and at some top quantiles between women working in the manufacturing sector and those in retail, finance and personal services. These wage differentials decreased for women in retail but increased for women working in services (see Table 8b).3 Ethnicity After adjusting for other characteristics, we also find significant wage differentials between Brazilian self-employed women with different ethnic backgrounds. Like salaried women, black and mulatto self-employed women earned less than white self-employed women. Even though there was no clear pattern in the variation along the quantiles in each year, it is clear that the relative premium differential decreased at all quantiles between white and mulatto and at the higher quantiles between white and black between 1989 and 1995. On the other hand, and as was the case for salaried women, after controlling for other characteristics, there were no significant wage differentials between white and Asian self-employed. Maids 3 OLS estimates of these sectorial dummies were significant in 1988 but insignificant (except personal services) in 1997. This indicates that (adjusted) wage inequality arising from sectoral variation decreased at the mean. 19 A similar result found in Argentina and Brazil is that domestic servants that ranked at the bottom of the conditional earnings distribution had a wage premium in comparison with (otherwise similar) self-employed women. In contrast, high earning women in the self-employment sector (mainly women ranking at the 90th quantile) earned more than high earning domestic servants did (after controlling for other characteristics). For instance, in Argentina, in 1988, domestic servants at the 0.25'h quantile earned 120% more than (otherwise similar) self-employed non-domestic servants. In contrast, domestic servants at the 0.90h quantile earned 172% less than (otherwise similar) self- employed in 1988. By 1997, the wage premium differential at the low quantile increased to 238% but decreased to 30% at the high quantile. This suggests that for the post-reform year (after controlling for other factors) domestic servants at all quantiles did better than all other self-employed did. However, in Brazil the adjusted wage differential decreased along all quantiles between 1989 and 1995 (see Tables 8b and 9b). In contrast, domestic servants in Costa Rica earned less than all other self-employed did along all quantiles in both the pre- and post-reform years. In addition, the adjusted wage differential increased at all quantiles between these years (except at the median where the premium for not being a domestic servant decreased from 91.2% in 1988 to 77.8% in 1995). Region Finally, adjusted regional wage differentials were also significant for Brazilian and Costa Rican self-employed women. As was the case for salaried, self-employed women living in urban areas in southern Brazil and in the urban parts of developed San Jose enjoyed a positive (adjusted) wage premium over self-employed living in other urban areas in each country. The premium decreased in Brazil for women at the lower quantiles but increased for women at the higher quantiles between the pre- and post-reform years. On the other hand, the premium decreased at all quantiles in Costa Rica.34 In sum, quantile regression estimates indicate that after controlling for other factors, education contributed to decrease wage inequality between self-employed women in all three countries. Furthermore, wage differentials arising from ethnicity, especially between white and mulatto and between white and black Brazilian women decreased along most of the quantiles between the pre- and post-reform period. Moreover, maids, who are considered poor working women, did relatively better than other self-employed women in the post-reform periods (except in Costa Rica). On the other hand, skills on the job contributed to increased wage inequality in Argentina, but decreased wage inequality in Brazil. 6. Conclusion Quantile regression estimates indicate that a conditional mean model was not suitable for analyzing wage differentials among working women in these countries. After controlling for other characteristics, wage premiums to human capital, labor experience and other characteristics varied along the conditional distribution. In other words, the 34 Note that in this country, regional wage differentials were significant in 1989 but insignificant in 1995. 20 estimated changes in wage inequality between the pre- and post-reform periods due to specific factors differed across (otherwise) observationally working women. The unconditional and conditional analysis indicate that poor and less educated working women did relatively well in the post-reforms periods, especially in Argentina and Brazil. These general results are consistent with the Hecksher-Ohlin model, which indicates that in countries with abundant unskilled labor, trade liberalization should reduce wage differentials. Female informal sector participation continued to grow throughout the analyzed period, coinciding with a decrease in wage differentials with the formal sector, however, the decrease was not uniform along the estimated quantiles. For instance, while in Argentina and Costa Rica the reduction in wage differential between formal salaried and informal salaried was more important between women at the high quantiles, in Brazil it was more important between women at the low quantiles (after adjusting for other factors). In any case, the data indicate that during the post-reform period wage growth in the informal and self-employment sector was relatively better than in the formal salaried sector. These results may suggest that policymakers should not be concerned with the increase in these sectors. During the post-reform period, jobs in these sectors offered women better wage performance. However, wage disparities within the informal sector increased in Argentina. This suggests that policy measures in this country could be oriented to decrease wage disparities within this sector rather than to decrease employment within it. In all three countries there was a decrease in the proportion of working women with low education (proportion of women with less than six years of education decreased in all three countries). In addition, the proportion of working women with college and more than college education increased between the sampled years. The better relative performance of low skill workers is also consistent with the decrease in the relative supply of less skilled working women (the same trends have been found in other countries (see for instance Gottschalk and Joyce (1998)). Wage inequality decreased in the formal salaried and the self-employment sectors in Argentina, decreased in the self-employment sector in Brazil and Costa Rica, and slightly increased in the salaried sector in Costa Rica between the sampled years. This evolution in wage inequality is explained in part by changes in the premium to education. Quantile regression estimates indicate that there was a reduction in the relative premium to education in Argentina and in Brazil. In contrast, wage differentials arising from education increased for salaried women in Costa Rica. These results show that working women in the two countries with major reforms experienced important reductions in wage differentials. Those with less human capital (less educated) enjoyed wage increases over those with more human capital (more educated). This suggests that a reinforcement of the economic reforms could mean additional wage improvements in wage inequality. On the other hand, there was an increase in wage differentials arising from different skills on the job in the salaried sector. After controlling for other characteristics, the wage differential between women working in high-skilled jobs and women working in low-skilled jobs increased at all quantiles between the pre- and post-reform years in all 21 three countries. The increase in differentials was greater at the higher quantiles. The same patterns were observed in the self-employment sector in Argentina, while the opposite was observed in Brazil. This suggests that increased competition is rewarding women with better skills on the job. Job training programs targeted to women in low skilled jobs may reduce wage differentials arising from variation of skills on the job. 22 7. Tables TABLE la: Sample Means (Proportions) of All Explanatory Variables ARGENTINA Salaried Workers Self-Employed 1988 1997 1988 1997 married 0.411 0.468 0.570 0.524 element- (incomplete and complete) 0.367 0.274 0.580 0.480 high1 =(complete high school) 0.211 0.202 0.109 0.122 high2 =(incomplete high school). 0.168 0.157 0.130 0.182 univ1 =(complete College) 0.107 0.163 0.047 0.054 univ2 =(incomplete College) 0.147 0.204 0.135 0.162 age 35.34 36.89 41.60 43.55 agel =(years<=25) 0.295 0.259 0.104 0.071 age2 =( <25 years >=35) 0.238 0.235 0.262 0.216 age3 (3545) 0.232 0.278 0.363 0.432 tenure 6.090 5.545 7.876 7.074 tenure^2 93.8 82.6 143.3 136.0 tenurel =(years<1) 0.306 0.334 0.246 0.267 tenure2 =(10) 0.186 0.173 0.267 0.193 head 0.179 0.203 0.236 0.260 manuf 0.215 0.128 0.150 0.084 retail 0.130 0.138 0.223 0.274 finan 0.091 0.109 0.060 0.095 servper 0.188 0.193 0.132 0.149 pubserv 0.201 0.223 - - servsoc 0.149 0.172 0.085 0.088 other 0.026 0.037 0.010 0.014 domserv - - 0.339 0.297 prof 0.097 0.084 0.122 0.105 skill 0.412 0.536 0.332 0.348 unskill 0.492 0.379 0.547 0.547 h35 0.302 0.342 0.516 0.503 salinf 0.248 0.256 - - lahwage 1.873 2.250 1.792 2.172 23 TABLE lb: Sample Means (Proportions) of All Explanatory Variables. BRAZIL Salaried Workers Self-Employed 1989 1995 1989 1995 married - 0.342 - 0.364 eduyears 8.538 10.049 4.286 5.950 elementl =(years<4) 0.127 0.061 0.424 0.241 element2= (4=35) 0.326 0.316 0.269 0.274 age3 ( 3545) 0.118 0.133 0.228 0.219 potexpl 17.06 16.59 25.02 23.48 potexp2 435.6 416.7 828.5 739.3 children 0.769 1.340 1.053 2.165 potchil 13.19 34.58 22.73 67.83 potchi2 304.7 1069.0 638.4 2456.0 workage - 16.30 - 14.04 workexp - 16.34 - 21.39 workexp^2 - 404.1 - 645.9 head 0.154 0.164 0.175 0.185 manuf 0.209 0.171 0.040 0.032 retail 0.142 0.148 0.117 0.150 serv 0.102 0.109 0.341 0.233 finan 0.051 0.036 0.000 0.001 pfserv 0.049 0.054 0.021 0.023 pubserv 0.273 0.310 - - servsoc 0.133 0.145 0.027 0.025 other 0.041 0.027 0.005 0.004 domserv - - 0.448 0.531 prof 0.234 0.260 0.042 0.039 h35 0.267 0.257 0.384 0.402 salinf 0.160 0.172 - - northeast 0.288 0.293 0.352 0.362 south 0.105 0.105 0.091 0.094 southeast 0.606 0.602 0.557 0.544 lahwage 0.124 0.451 -0.692 -0.084 24 TABLE Ic: Sample Means (Proportions) of All Explanatory Variables. Salaried Workers Self-Employed 1988 1997 1988 1997 married 0.350 0.347 0.348 0.347 eduyears 10.12 10.471 6.525 7.100 element= (years<=6). 0.267 0.246 0.618 0.568 highl = (7=35) 0.360 0.337 0.279 0.268 age3 (3545) 0.101 0.115 0.210 0.270 potexpl 15.56 15.85 23.27 24.86 potexp2 358.0 366.4 732.1 812.4 children 1.224 1.010 1.458 1.172 potchi 1 18.38 14.92 29.25 25.45 potchi2 374.2 301.4 774.4 724.6 head 0.189 0.228 0.261 0.273 manuf 0.232 0.202 0.266 0.176 retail 0.215 0.253 0.192 0.222 finan 0.047 0.054 0.007 0.015 servsoc 0.472 0.438 0.527 0.581 other 0.034 0.053 0.009 0.006 domserv - - 0.424 0.392 pubsec 0.444 0.369 - - prof 0.314 0.327 0.054 0.052 whitecollar 0.229 0.243 - - bluecollar 0.452 0.429 - primasect 0.005 0.002 - - h35 0.118 0.137 0.533 0.542 salinf 0.091 0.100 - - metro 0.391 0.455 0.348 0.440 rest 0.264 0.217 0.188 0.162 chorotega 0.075 0.068 0.136 0.127 pacific 0.108 0.100 0.136 0.089 brunca 0.050 0.060 0.083 0.041 hatlant 0.068 0.064 0.080 0.106 hnorth 0.044 0.035 0.029 0.035 lahwage 4.505 4.568 3.923 4.171 25 TABLE 2: Female Participation, Employment and Unemployment* Argentina Brazil Costa Rica 1988 1997 1989 1995 1989 1995 Employed women (%) 37.4 37.1 42.3 46.3 56.7 57.7 Salaried 72.6 74.4 59.3 51.8 71.0 71.0 Self-employed** 23.5 20.4 36.2 41.7 23.0 20.8 Other** 3.9 5.2 4.5 6.5 6.1 8.2 Unemployment Rate 6.5 17.4 3.8 9.3 4.4 5.7 Overall Participation Rate 40.0 44.9 43.9 51.0 59.3 61.1 Out of labor Market (%) 60.0 55.1 56.1 49.0 40.7 38.9 *Employed women is a percentage of women aged 15-70. Unemployment rate corresponds to unemployed women as a percentage of all women participating in the labor markets (ie. employed plus unemployed). **Self-employed includes domestic Service, and "Other" includes family workers, and non-specified sector. Source: EPH, Argentina; PNAD, Brazil and EHPM, Costa Rica. Figures are author's estimations. 26 TABLE 3: Distribution of Real Hourly Wages ARGENTINA (Peso, $1988=1.00) Salaried Labor Market Self-Employed 1988 1997 Change* 1988 1997 Change* Mean 8.4 11.5 4.0% 9.8 13.8 4.5% 5th 2.1 3.1 5.7% 1.3 1.8 4.2% 10th 2.8 4.2 5.8% 1.9 2.8 5.1% 25th 4.1 6.3 5.9% 3.9 4.6 2.1% 50th 6.0 9.6 6.7% 5.8 9.6 7.3% 75th 10.1 14.4 4.8% 9.1 15.9 8.4% 90th 16.0 20.1 2.9% 18.8 28.8 6.0% 95th 21.8 26.6 2.5% 35.2 39.7 1.4% BRAZIL (U.S. Dollars, $1995=1.00) Salaried Labor Market Self-Employed 1989 1995 Change* 1989 1995 Change* Mean 1.95 2.46 3.4% 1.07 1.59 5.5% 5th 0.29 0.51 7.2% 0.09 0.24 10.8% 10th 0.44 0.57 3.9% 0.13 0.33 10.1% 25th 0.54 0.82 5.6% 0.26 0.49 7.8% 50th 0.99 1.36 4.6% 0.47 0.82 7.2% 75th 2.11 2.78 4.0% 0.91 1.63 7.4% 90th 4.35 5.34 3.1% 2.17 3.06 4.8% 95th 6.52 7.83 2.8% 3.63 4.90 4.3% COSTA RICA (Colones, $1989=1.00) Salaried Labor Market Self-Employed 1989 1995 Change* 1989 1995 Change* Mean 112.0 119.9 1.2% 83.6 102.4 3.8% 5th 33.7 37.3 1.8% 11.7 17.5 8.3% 10th 45.5 48.3 1.0% 15.6 22.0 7.0% 25th 57.8 61.7 1.1% 25.9 35.3 6.0% 50th 85.5 88.8 0.7% 50.5 60.6 3.3% 75th 138.4 150.5 1.5% 93.3 110.2 3.0% 90th 206.4 227.0 1.7% 157.5 201.6 4.7% 95th 265.1 302.8 2.4% 210.5 301.9 7.2% *Annual average 27 TABLE 4: Distribution of Real Hourly Wages by Sector ARGENTINA (Peso, $1988=1.00) Salaried Labor Market Self-Employed Formal Informal Excluding Dom. Service Domestic Service 1988 1997 Change* 1988 1997 Change* 1988 1997 Change* 1988 1997 Change* Mean 9.6 12.1 2.8% 4.8 9.7 11.4% 12.2 13.7 1.4% 5.1 14.0 19.4% 5th 2.6 4.1 6.6% 1.3 2.3 8.1% 1.2 1.6 3.3% 2.1 4.1 10.2% 10th 3.5 5.1 5.3% 2.0 2.6 3.6% 1.7 2.4 4.3% 2.9 4.6 6.3% 25th 4.7 6.9 4.9% 2.9 4.5 6.1% 3.6 3.7 0.2% 4.1 7.2 8.4% 50th 7.3 10.1 4.3% 4.1 7.7 10.0% 7.0 8.5 2.3% 4.9 10.8 13.4% 75th 11.4 14.4 2.9% 5.8 13.9 15.5% 12.7 18.0 4.7% 6.1 14.4 15.3% 90th 18.2 21.5 2.0% 7.4 18.0 16.1% 26.6 32.2 2.3% 7.3 18.5 17.2% 95th 24.2 27.0 1.3% 9.2 23.1 16.8% 43.6 42.5 -0.3% 7.9 22.7 20.7% Standard. Dev. 8.52 8.67 0.15 4.15 7.33 3.19 17.28 14.55 -2.73 1.82 22.15 20.33 BRAZIL (U.S. Dollars, $1995=1.00) Salaried Labor Market Self-Employed Formal Informal Excluding Dom. Service Domestic Service 1989 1995 Change* 1989 1995 Change* 1989 1995 Change* 1989 1995 Change* Mean 1.97 2.66 5.8% 1.16 1.51 5.1% 1.58 2.39 8.6% 0.44 0.88 16.5% 5th 0.43 0.54 4.2% 0.14 0.31 18.5% 0.09 0.26 32.2% 0.08 0.23 31.1% 10th 0.48 0.61 4.5% 0.22 0.41 14.6% 0.15 0.41 27.8% 0.11 0.30 28.1% 25th 0.58 0.92 9.6% 0.37 0.56 8.1% 0.33 0.68 17.7% 0.20 0.41 16.9% 50th 1.03 1.53 8.0% 0.54 0.86 9.6% 0.70 1.22 12.7% 0.36 0.61 11.5% 75th 2.03 3.06 8.5% 1.04 1.47 6.8% 1.45 2.45 11.5% 0.54 1.02 14.7% 90th 4.35 5.71 5.2% 2.13 2.85 5.7% 3.26 4.90 8.4% 0.85 1.84 19.1% 95th 6.52 8.16 4.2% 3.48 4.77 6.2% 5.80 8.16 6.8% 1.09 2.45 20.9% Standard. Dev. 2.95 3.81 0.86 3.52 2.85 -3.2% 5.23 4.31 -2.9% 0.40 0.89 0.49 COSTA RICA (Colones, $1989=1.00) Salaried Labor Market Self-Employed Formal Informal Excluding Dom. Service Domestic Service 1989 1995 Change* 1989 1995 Change* 1989 1995 Change* 1989 1995 Change* Mean 117.4 124.9 1.1% 58.0 74.7 4.8% 109.5 137.3 4.2% 48.5 48.5 0.0% 5th 42.4 44.8 0.9% 20.2 24.1 3.2% 14.1 22.0 9.3% 10.1 15.7 9.3% 10th 49.0 51.2 0.7% 24.3 32.8 5.8% 19.4 29.4 8.5% 12.1 17.6 7.6% 25th 61.1 64.6 1.0% 33.3 39.1 2.9% 39.0 48.9 4.2% 19.6 25.5 4.9% 50th 91.9 93.2 0.2% 49.3 58.8 3.2% 67.4 88.2 5.2% 32.4 39.8 3.8% 75th 144.2 155.9 1.4% 63.1 87.4 6.4% 111.1 162.6 7.7% 58.3 60.6 0.6% 90th 213.7 241.7 2.2% 101.3 132.3 5.1% 194.4 290.9 8.3% 93.4 88.2 -0.9% 95th 269.2 313.3 2.7% 137.4 159.7 2.7% 340.4 407.3 3.3% 105.7 108.3 0.4% Standard. Dev. 91.2 92.3 1.12 40.7 75.1 34.40 162.4 186.3 23.96 51.9 35.0 -16.90 'Annual Average. TABLE 5: Hourly Wage Differentials ARGENTINA (Peso, $1988=1.00) Salaried Labor Market Self-Employed 1988 1997 Difference 1988 1997 Difference 5Oth/1Oth 2.16 2.29 0.12 3.00 3.40 0.40 50th/25th 1.45 1.52 0.07 1.50 2.08 0.58 75th/25th 2.45 2.29 -0.17 2.34 3.45 1.11 90th/50th 2.67 2.09 -0.58 3.23 3.00 -0.23 90th/75 1.58 1.39 -0.19 2.07 1.81 -0.26 90th/Oth 5.78 4.78 -0.99 9.69 10.20 0.51 95th/5th 10.50 8.48 -2.02 27.29 22.29 -4.99 Variance of Log Wage 0.55 0.44 -0.12 0.88 0.98 0.10 Coefficient of 0.94 0.73 -0.21 1.48 1.24 -0.24 Variation BRAZIL (U.S. Dollars, $1995=1.00) Salaried Labor Market Self-Employed 1989 1995 Difference 1989 1995 Difference 50th/10th 2.26 2.38 0.12 3.63 2.50 -1.13 50th/25th 1.82 1.67 -0.15 1.78 1.67 -0.12 75th/25th 3.88 3.41 -0.47 3.47 3.33 -0.14 90th/50th 6.60 5.76 -0.84 7.80 6.00 -1.80 90th/75 2.06 1.92 -0.14 2.40 1.88 -0.52 90th/Oth 9.96 9.36 -0.60 16.95 9.38 -7.57 95th/5th 22.50 15.36 -7.14 42.17 20.00 -22.17 Variance of Log Wage 0.97 0.78 -0.19 1.25 0.88 -0.37 Coefficient of 1.60 1.50 -0.10 3.69 1.96 -1.73 Variation COSTA RICA (Colones, $1989=1.00) Salaried Labor Market Self-Employed 1989 1995 Difference 1989 1995 Difference 5Oth/Oth 1.88 1.85 -0.02 3.25 2.75 -0.50 50/25th 1.48 1.44 -0.04 1.95 1.72 -0.23 75th/25th 2.39 2.44 0.04 3.60 3.13 -0.47 90th/50th 2.42 2.55 0.14 3.12 3.33 0.21 90th/75 1.49 1.51 0.02 1.69 1.83 0.14 90th/10th 4.53 4.74 0.20 10.12 9.14 -0.98 95th/5th 7.87 8.11 0.24 18.04 17.25 -0.79 Variance of Log Wage 0.40 0.42 0.02 0.89 0.80 -0.09 Coefficient of 0.80 0.77 -0.03 1.57 1.49 -0.08 Variation TABLE 6: Hourly Wage Differentials by Sector ARGENTINA (Peso, $1988=1.00) Salaried Labor Market Self-Employed Formal Informal Excluding Dom. Service Domestic Service 1988 1997 Difference 1988 1997 Difference 1988 1997 Difference 1988 1997 Difference 50th/10th 2.1 2.0 -0.12 2.1 2.9 0.88 4.1 3.5 -0.53 1.7 2.4 0.69 50th/25th 1.5 1.5 -0.06 1.4 1.7 0.31 2.1 1.5 -0.56 1.4 1.6 0.17 75th/25th 2.4 2.1 -0.30 2.0 3.1 1.09 3.5 4.9 1.38 1.5 2.0 0.52 90th/50th 2.5 2.1 -0.37 1.8 2.3 0.52 3.8 3.8 0.00 1.5 1.7 0.23 90th/75 1.6 1.5 -0.10 1.3 1.3 0.02 2.1 1.8 -0.32 1.2 1.3 0.09 90th-10th 5.2 4.2 -1.05 3.7 6.9 3.14 15.4 13.4 -2.02 2.5 4.1 1.57 Variance ofLog Wage 0.51 0.37 -0.14 0.37 0.55 0.19 1.16 1.19 0.03 0.24 0.47 0.23 Coefficient of 0.9 0.7 -0.17 0.9 0.8 -0.11 1.42 1.06 -0.35 0.36 1.59 1.23 Variation BRAZIL (U.S. Dollars, $1995=1.00) Salaried Labor Market Self-Employed Formal Informal Excluding Dom. Service Domestic Service 1989 1995 Difference 1989 1995 Difference 1989 1995 Difference 1989 1995 Difference 50th/10th 2.1 2.5 0.35 2.5 2.1 -0.40 4.6 3.0 -1.55 3.2 2.0 -1.20 50th/25th 1.8 1.7 -0.11 1.5 1.5 0.09 2.2 1.7 -0.49 1.8 1.4 -0.46 75th/25th 3.5 3.3 -0.15 2.8 2.6 -0.14 4.4 3.6 -0.79 2.7 2.5 -0.17 90th/50th 4.2 3.7 -0.47 3.9 3.3 -0.59 4.7 4.0 -0.69 2.4 3.0 0.64 90th/75 2.1 1.9 -0.28 2.0 1.9 -0.10 2.3 2.0 -0.25 1.6 1.8 0.22 90th-10th 9.0 9.3 0.30 9.8 7.0 -2.79 21.3 12.0 -9.34 7.6 6.1 -1.54 Variance of Log Wage 0.90 0.74 -0.16 0.90 0.67 -0.23 1.49 1.05 -0.44 0.63 0.50 -0.13 Coefficient of 1.48 1.43 -0.04 2.67 1.88 -0.79 3.32 1.80 -1.52 0.92 1.01 0.10 Variation COSTA RICA (Colones, $1989=1.00) Salaried Labor Market Self-Employed Formal Informal Excluding Dom. Service Domestic Service 1989 1995 Difference 1989 1995 Difference 1989 1995 Difference 1989 1995 Difference 50th/10th 1.87 1.82 -0.05 2.03 1.79 -0.24 3.46 3.00 -0.46 2.68 2.26 -0.42 50th/25th 1.50 1.44 -0.06 1.48 1.50 0.02 2.01 1.66 -0.34 1.62 1.44 -0.18 75th/25th 2.36 2.41 0.05 1.89 2.23 0.34 2.85 3.32 0.48 2.97 2.38 -0.59 90th/50th 2.33 2.59 0.27 2.06 2.25 0.19 2.89 3.30 0.41 2.88 2.21 -0.67 90th/75 1.48 1.55 0.07 1.60 1.51 -0.09 1.75 1.79 0.04 1.60 1.46 -0.14 90th-10th 4.36 4.72 0.36 4.17 4.03 -0.14 10.00 9.90 -0.10 7.71 5.00 -2.71 Variance of Log Wage 0.37 0.40 0.03 0.34 0.37 0.03 0.90 0.79 -0.11 0.62 0.41 -0.22 Coefficient of 0.78 0.74 -0.04 0.70 1.00 0.30 1.48 1.36 -0.13 1.07 0.72 -0.35 Variation TABLE 7: Average Hourly Earnings Ratios by Education and Age Argentina Brazil Costa Rica Salaried Self-Employed Salaried Self-Employed Salaried Self-Employed 1988 1997 1988 1997 1989 1995 1989 1995 1989 1995 1989 1995 College/Element. School Overall 3.03 1.82 4.44 2.38 5.38 4.77 9.07 7.00 2.68 2.75 1.54 3.50 Age less than 25 1.91 1.50 2.13 0.64 4.05 3.69 16.95 6.89 2.09 2.15 1.70 4.05 Age between 25-35 3.00 1.86 2.50 2.15 4.92 4.55 7.78 6.13 2.99 2.70 2.56 2.56 Age between 35-45 3.86 2.03 7.21 2.65 5.66 4.97 8.90 6.50 3.04 3.10 1.28 2.00 Age older than 45 3.07 1.91 4.64 2.84 6.90 5.79 5.63 7.38 2.17 2.72 0.20 2.68 College/Secondary School Overall 1.44 1.45 2.63 2.44 2.56 2.66 2.47 3.49 1.85 2.04 1.02 2.40 Age less than 25 1.17 1.66 1.35 1.17 2.27 2.32 4.93 3.38 1.53 1.80 0.74 3.80 Age between 25-35 1.35 1.44 2.41 2.62 2.28 2.47 1.91 3.19 1.72 1.95 1.31 1.52 Age between 35-45 1.81 1.25 3.42 1.64 1.98 2.25 2.62 3.34 1.69 2.16 0.86 1.31 Age older than 45 1.15 1.35 2.00 3.93 2.24 2.33 1.67 2.93 1.61 1.37 0.15 2.13 Secondary School/Element. School Overall 2.10 1.25 1.69 0.97 2.11 1.79 3.67 2.00 1.45 1.35 1.50 1.46 Age less than 25 1.64 0.91 1.58 0.55 1.78 1.59 3.44 2.04 1.36 1.19 2.31 1.07 Age between 25-35 2.22 1.28 1.04 0.82 2.16 1.84 4.07 1.92 1.74 1.39 1.96 1.68 Age between 35-45 2.13 1.62 2.11 1.61 2.85 2.21 3.39 1.95 1.80 1.43 1.49 1.52 Age older than 45 2.67 1.42 2.32 0.72 3.07 2.48 3.37 2.52 1.35 1.99 1.37 1.25 TABLE 8a: Quantile Regression Estimates, Salaried Women, Argentina, 1988 & 1997 1988 1997 0.1 0.25 0.5 0.75 0.90 0.10 0.25 0.50 0.75 0.90 a 1.479 1.743 2.218 2.499 3.142 1.542 2.001 2.383 2.793 3.113 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 elemen -0.501 -0.519 -0.633 -0.640 -0.852 -0.263 -0.338 -0.495 -0.466 -0.457 0.000 0.000 0.000 0.000 0.000 0.016 0.000 0.000 0.000 0.000 high1 -0.175 -0.120 -0.180 -0.181 -0.182 -0.141 -0.089 -0.186 -0.172 -0.188 0.092 0.030 0.002 0.004 0.075 0.085 0.296 0.003 0.013 0.036 high2 -0.357 -0.349 -0.494 -0.507 -0.609 -0.142 -0.223 -0.368 -0.421 -0.442 0.003 0.000 0.000 0.000 0.000 0.112 0.013 0.000 0.000 0.000 univ2 -0.129 -0.009 -0.085 -0.121 -0.236 0.158 0.083 -0.057 -0.006 0.045 0.304 0.908 0.235 0.051 0.059 0.042 0.266 0.385 0.930 0.697 tenure 0.026 0.025 0.021 0.031 0.029 0.022 0.028 0.027 0.022 0.028 0.034 0.003 0.000 0.000 0.001 0.097 0.000 0.000 0.005 0.014 tenures -0.001 -0.001 0.000 0.000 0.000 0.000 -0.001 -0.001 0.000 -0.001 0.171 0.099 0.272 0.011 0.128 0.546 0.085 0.000 0.220 0.173 EDAD -0.002 0.001 0.003 0.005 0.005 0.003 0.003 0.006 0.004 0.008 0.676 0.414 0.005 0.010 0.052 0.232 0.184 0.000 0.033 0.000 married 0.320 0.262 0.169 0.095 0.069 0.024 0.094 0.042 0.100 0.037 0.000 0.000 0.000 0.007 0.209 0.697 0.044 0.208 0.008 0.351 salinf -0.390 -0.241 -0.190 -0.080 -0.078 -0.398 -0.241 -0.097 0.011 0.179 0.000 0.000 0.000 0.295 0.480 0.000 0.001 0.016 0.893 0.206 retail 0.329 0.202 0.077 0.008 0.042 0.178 -0.023 -0.130 -0.016 -0.172 0.000 0.002 0.148 0.904 0.658 0.348 0.755 0.000 0.860 0.009 finan 0.385 0.406 0.358 0.346 0.385 0.520 0.318 0.187 0.248 0.091 0.000 0.000 0.000 0.000 0.000 0.002 0.000 0.000 0.001 0.354 servper 0.117 0.100 0.035 -0.078 -0.048 0.301 0.349 0.404 0.651 0.355 0.480 0.185 0.461 0.395 0.639 0.059 0.002 0.000 0.000 0.105 pubser 0.175 0.210 0.102 0.027 -0.057 0.392 0.184 0.054 0.125 0.019 0.097 0.003 0.153 0.602 0.354 0.029 0.011 0.428 0.038 0.817 servsoc 0.099 0.121 0.052 0.005 -0.012 0.341 0.081 0.029 0.155 0.062 0.386 0.080 0.357 0.940 0.879 0.032 0.295 0.545 0.073 0.517 serv 0.357 0.214 0.149 0.150 0.046 0.462 0.341 0.288 0.363 0.120 0.419 0.035 0.154 0.352 0.775 0.015 0.018 0.006 0.000 0.424 skill -0.198 -0.219 -0.266 -0.234 -0.518 -0.245 -0.289 -0.266 -0.406 -0.474 0.048 0.000 0.000 0.015 0.000 0.003 0.001 0.001 0.000 0.010 unskill -0.322 -0.360 -0.384 -0.449 -0.655 -0.417 -0.434 -0.445 -0.668 -0.718 0.005 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 Table 8b: Quantile Regression Estimates, Salaried Women, Argentina, 1988 & 1997 1988 1997 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 c 1.674 2.132 2.585 3.149 3.676 2.042 2.456 2.886 3.254 3.605 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 elemen -0.555 -0.554 -0.644 -0.668 -0.747 -0.167 -0.351 -0.501 -0.480 -0.419 0.000 0.000 0.000 0.000 0.000 0.073 0.000 0.000 0.000 0.001 highl -0.161 -0.103 -0.171 -0.209 -0.162 -0.120 -0.114 -0.171 -0.164 -0.100 0.131 0.160 0.005 0.038 0.079 0.168 0.271 0.020 0.010 0.294 high2 -0.377 -0.388 -0.461 -0.529 -0.561 -0.058 -0.220 -0.364 -0.448 -0.376 0.000 0.000 0.000 0.000 0.000 0.558 0.045 0.000 0.000 0.004 univ2 -0.123 -0.018 -0.068 -0.114 -0.141 0.191 0.071 -0.037 0.013 0.098 0.277 0.841 0.306 0.193 0.128 0.009 0.460 0.594 0.796 0.329 agel 0.016 -0.132 -0.150 -0.228 -0.215 -0.056 -0.148 -0.182 -0.166 -0.321 0.899 0.023 0.005 0.000 0.022 0.602 0.014 0.002 0.000 0.000 age2 -0.001 -0.049 -0.067 -0.086 -0.094 -0.016 -0.079 -0.114 -0.091 -0.157 0.992 0.315 0.250 0.106 0.301 0.832 0.123 0.048 0.076 0.089 age3 0.091 0.033 0.007 0.004 0.027 -0.055 -0.145 -0.050 -0.033 -0.152 0.260 0.501 0.887 0.938 0.754 0.565 0.006 0.304 0.487 0.017 tenurel -0.298 -0.224 -0.245 -0.329 -0.374 -0.325 -0.340 -0.260 -0.195 -0.194 0.029 0.000 0.000 0.000 0.000 0.002 0.000 0.000 0,009 0.021 tenure2 -0.196 -0.147 -0.152 -0.217 -0.261 -0.064 -0.092 -0.145 -0.147 -0.071 0.058 0.006 0.000 0.000 0.001 0.605 0.129 0.022 0.011 0.326 tenure3 -0.043 -0.095 -0.122 -0.162 -0.151 0.009 -0.038 -0.038 -0.083 -0.055 0.651 0.137 0.059 0.026 0.105 0.931 0.517 0.456 0.358 0.544 married 0.276 0.229 0.160 0.061 0.083 0.051 0.104 0.064 0.081 0.044 0.000 0.000 0.000 0.100 0.105 0.455 0.031 0.048 0.055 0.485 salinf -0.347 -0.218 -0.180 -0.055 -0.102 -0.391 -0.182 -0.101 0.048 0.225 0.000 0.000 0.000 0.473 0.337 0.001 0.023 0.061 0.601 0.138 retail 0.328 0.160 0.088 0.008 0.116 0.094 -0.020 -0.138 -0.011 -0.167 0.000 0.022 0.137 0.905 0.291 0.580 0.807 0.004 0.886 0.093 finan 0.368 0.340 0.421 0.346 0.368 0.443 0.263 0.183 0.236 0.139 0.001 0.001 0.000 0.000 0.001 0.012 0.000 0.004 0,001 0.199 servper 0.103 0.026 0.060 -0.081 -0.034 0.282 0.304 0.405 0.650 0.334 0.517 0.741 0.329 0.391 0.716 0.068 0.008 0.000 0.000 0.140 pubser 0.173 0.166 0.131 0.028 -0.027 0.295 0.161 0.063 0.107 0.024 0.107 0.050 0.042 0.602 0.612 0.120 0.028 0.262 0.095 0.811 servsoc 0.117 0.074 0.077 0.044 0.046 0.212 0.067 0.014 0.170 0.080 0.275 0.324 0.243 0.597 0.558 0.202 0.408 0.777 0.021 0.482 serv 0.433 0.219 0.170 0.176 0.093 0.392 0.355 0.268 0.356 0.091 0.118 0.034 0.185 0.116 0.588 0.042 0.002 0.016 0.000 0.631 skill -0.169 -0.230 -0.238 -0.282 -0.488 -0.355 -0.264 -0.260 -0.416 -0.392 0.075 0.001 0.000 0.001 0.000 0.001 0.005 0.001 0.000 0.022 unskill -0.266 -0.352 -0.347 -0.454 -0.658 - -0.548 -0.444 -0.409 -0.688 -0.659 0.031 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.001 Table 8c: Quantile Regression Estimates, Self-employed Women, Argentina, 1988 & 1997 1988 1997 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 a 0.734 1.945 2.420 2.838 3.858 2.263 2.758 3.038 2.989 4.485 0.226 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 elemen -0.467 -1.055 -0.760 -0.884 -1.791 -0.993 -0.655 -0.585 -0.545 -0.801 0.443 0.001 0.004 0.004 0.000 0.001 0.011 0.008 0.039 0.042 high1 0.295 -0.285 -0.215 -0.646 -1.412 -0.651 -0.336 -0.261 -0.357 -0.246 0.534 0.351 0.426 0.010 0.002 0.002 0.239 0.238 0.098 0.442 high2 -0.316 -0.864 -0.741 -0.913 -1.780 -0.956 -0.683 -0.441 -0.420 -0.740 0.362 0.018 0.006 0.000 0.001 0.006 0.001 0.037 0.149 0.012 univ2 0.240 -0.266 -0.564 -0.600 -1.482 -0.849 -0.412 -0.313 -0.054 -0.590 0.768 0.317 0.011 0.040 0.016 0.003 0.347 0.315 0.893 0.043 tenure 0.030 0.018 0.025 0.027 0.010 0.046 0.018 0.013 -0.007 -0.013 0.281 0.373 0.006 0.007 0.634 0.162 0.312 0.339 0.720 0.402 tenures -0.001 -0.001 -0.001 -0.001 0.000 -0.002 0.000 0.000 0.001 0.000 0.499 0.472 0.000 0.085 0.660 0.578 0.626 0.594 0.190 0.105 EDAD -0.004 0.000 0.002 0.001 -0.001 -0.011 -0.007 -0.006 -0.002 0.003 0.640 0.949 0.333 0.644 0.846 0.184 0.213 0.071 0.710 0.412 married 0.510 -0.043 0.028 0.090 0.026 0.175 -0.060 0.004 0.017 -0.079 0.000 0.744 0.656 0.100 0.819 0.219 0.636 0.965 0.883 0.580 domser 0.680 0.774 0.444 -0.265 -0.946 1.310 1.126 1.114 1.213 -0.062 0.022 0.041 0.119 0.296 0.019 0.004 0.020 0.000 0.008 0.802 retail -0.133 0.275 0.470 0.091 -0.430 0.391 0.207 0.101 0.459 -0.352 0.605 0.440 0.087 0.701 0.271 0.519 0.638 0.659 0.325 0.006 finan 0.626 0.393 0.589 0.748 0.695 0.348 0.416 0.662 0.888 -0.388 0.458 0.422 0.124 0.002 0.191 0.525 0.397 0.005 0.032 0.016 servper 0.281 0.198 0.545 0.614 0.663 0.543 0.706 0.811 0.853 0.430 0.177 0.584 0.000 0.007 0.047 0.032 0.031 0.000 0.068 0.089 servsoc 0.575 0.390 0.551 0.769 0.162 0.233 0.386 0.461 0.627 -0.356 0.389 0.395 0.049 0.048 0.807 0.561 0.311 0.025 0.152 0.240 serv 0.196 0.882 0.380 0.119 0.287 0.935 0.614 0.680 0.372 -0.330 1.000 1.000 0.453 1.000 1.000 1.000 1.000 0.029 1.000 1.000 skill -0.297 -0.160 -0.552 -0.307 0.226 -0.488 -0.840 -0.686 -0.488 -0.809 0.377 0.427 0.011 0.154 0.644 0.356 0.057 0.003 0.004 0.000 unskill -0.158 -0.361 -0.728 -0.065 0.831 -0.805 -1.056 -0.984 -0.925 -0.835 0.718 0.243 0.137 0.821 0.094 0.083 0.121 0.003 0.001 0.000 Table 8d: Quantile Regression Estimates, Self-employed Women, Argentina, 1988 & 1997 1988 1997 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 a 0.930 1.938 2.678 3.146 3.816 1.637 2.625 2.678 3.135 4.808 0.234 0.000 0.000 0.000 0.000 0.008 0.000 0.000 0.000 0.000 elemen -0.600 -0.848 -0.811 -0.887 -1.577 -1.175 -0.645 -0.528 -0.648 -0.849 0.186 0.020 0.000 0.010 0.000 0.000 0.001 0.046 0.006 0.057 highl 0.170 -0.228 -0.371 -0.589 -1.180 -0.828 -0.475 -0.250 -0.559 -0.411 0.721 0.512 0.202 0.030 0.002 0.005 0.096 0.314 0.009 0.261 high2 -0.382 -0.694 -0.794 -0.952 -1.577 -1.085 -0.592 -0.419 -0.540 -0.780 0.384 0.054 0.001 0.003 0.001 0.023 0.000 0.073 0.038 0.030 univ2 0.420 -0.342 -0.669 -0.635 -1.082 -0.940 -0.624 -0.265 -0.182 -0.811 0.547 0.168 0.044 0.092 0.056 0.000 0.072 0.387 0.633 0.001 agel -0.126 -0.039 -0.096 0.001 0.081 0.459 0.283 0.328 0.118 -0.190 0.688 0.857 0.492 0.993 0.596 0.252 0.220 0.248 0.421 0.790 age2 0.005 0.112 -0.013 0.065 0.251 0.368 0.044 0.002 0.000 -0.007 0.988 0.483 0.860 0.502 0.143 0.239 0.814 0.990 1.000 0.978 age3 -0.045 -0.025 -0.039 0.000 0.024 0.285 0.198 0.165 0.065 0.050 0.780 0.844 0.702 1.000 0.847 0.440 0.200 0.053 0.595 0.818 tenurel -0.354 -0.048 -0.012 -0.028 -0.069 -0.119 -0.170 -0.184 -0.109 0.027 0.2'25 0.805 0.917 0.780 0.612 0.827 0.458 0.145 0.483 0.920 tenure2 0.042 0.008 -0.077 -0.091 -0.182 0.139 0.114 -0.004 -0.019 0.152 0.807 0.960 0.481 0.260 0.265 0.760 0.643 0.966 0.879 0.620 tenure3 0.037 0.223 0.172 0.165 0.292 0.313 0.146 0.027 0.140 0.104 0.866 0.231 0.020 0.272 0.183 0.422 0.461 0.835 0.329 0.682 married 0.510 0.128 0.002 0.065 0.081 0.182 -0.087 0.012 -0.095 -0.260 0.033 0.289 0.976 0.380 0.439 0.388 0.432 0.913 0.373 0.215 domser 1.000 0.790 0.516 -0.330 -1.001 1.181 1.217 1.164 1.003 -0.230 0.001 0.042 0.010 0.239 0.001 0.108 0.000 0.000 0.055 0.267 retail 0.258 0.336 0.457 -0.012 -0.450 0.126 0.401 0.171 0.351 -0.558 0.258 0.370 0.021 0.973 0.105 0.816 0.217 0.521 0.482 0.006 finan 0.313 0.491 0.404 0.570 0.365 0.265 0.260 0.653 0.768 -0.557 0.654 0.175 0.248 0.112 0.465 0.607 0.612 0.009 0.066 0.030 servper 0.337 0.385 0.580 0.500 0.514 0.563 0.755 0.837 0.744 0.433 0.156 0.231 0.000 0.032 0.226 0.173 0.001 0.000 0.142 0.242 servsoc 0.369 0.451 0.510 0.655 0.142 0.053 0.331 0.671 0.491 -0.598 0.590 0.190 0.040 0.111 0.821 0.860 0.509 0.017 0.255 0.122 serv 0.140 0.940 0.477 -0.077 -0.434 0.765 1.087 0.626 0.626 -0.292 1.000 1.000 0.344 1.000 1.000 1.000 1.000 0.110 1.000 1.000 skill -0.361 -0.588 -0.567 -0.418 0.040 -0.225 -1.088 -0.605 -0.457 -0.808 0.311 0.065 0.018 0.097 0.916 0.608 0.007 0.004 0.001 0.003 unskill -0.588 -0.652 -0.757 -0.142 0.653 -0.448 -1.345 -0.972 -0.867 -0.878 0.194 0.085 0.044 0.687 0.236 0.396 0.021 0.000 0.000 0.010 Table 9a: Quantile Regression Estimates, Salaried Women, Brazil, 1989 & 1995 1989 1995 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 Q -2.248 -1.894 -1.644 -1.407 -1.208 -1.859 -1.682 -1.528 -1.332 -1.125 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 eduyear 0.103 0.106 0.120 0.132 0.143 0.091 0.100 0.112 0.124 0.133 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 age 0.019 0.019 0.021 0.025 0.030 0.014 0.016 0.020 0.024 0.026 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 married na na na na na 0.065 0.061 0.070 0.067 0.084 na na na na na 0.000 0.000 0.000 0.000 0.000 nocard -0.485 -0.363 -0.269 -0.197 -0.155 -0.292 -0.292 -0.235 -0.153 -0.060 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.065 south 0.546 0.399 0.314 0.239 0.138 0.433 0.389 0.323 0.278 0.192 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000 0.000 0.000 s. east 0.440 0.337 0.309 0.281 0.248 0.363 0.331 0.321 0.302 0.301 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 retail -0.219 -0.232 -0.255 -0.232 -0.124 -0.109 -0.123 -0.138 -0.160 -0.143 0.000 0.000 0.000 0.000 0.004 0.000 0.000 0.000 0.000 0.000 servsoc -0.227 -0.197 -0.189 -0.182 -0.142 -0.075 -0.124 -0.092 -0.095 -0.081 0.000 0.000 0.000 0.000 0.000 0.006 0.000 0.000 0.001 0.022 servI -0.207 -0.214 -0.210 -0.205 -0.162 -0.098 -0.113 -0.105 -0.119 -0.081 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.043 other 0.090 0.096 0.187 0.205 0.277 0.194 0.122 0.097 0.022 0.045 0.046 0.008 0.000 0.000 0.000 0.000 0.000 0.016 0.658 0.427 finan 0.360 0.490 0.583 0.580 0.724 0.324 0.492 0.568 0.550 0.559 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 pfserv -0.116 -0.076 -0.063 0.005 0.091 0.025 0.040 0.056 0.074 -0.015 0.014 0.046 0.121 0.915 0.103 0.504 0.134 0.149 0.054 0.714 pub -0.338 -0.189 -0.164 -0.190 -0.189 -0.082 -0.095 -0.099 -0.140 -0.126 0.000 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000 0.000 prof 0.169 0.192 0.144 0.136 0.128 0.206 0.258 0.232 0.188 0.190 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 black -0.090 -0.105 -0.183 -0.235 -0.290 -0.087 -0.099 -0.102 -0.128 -0.174 0.148 0.000 0.000 0.000 0.000 0.006 0.000 0.000 0.000 0.000 mixed -0.137 -0.139 -0.167 -0.182 -0.219 -0.105 -0.108 -0.122 -0.153 -0.179 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 asian 0.070 0.145 0.087 0.139 0.010 -0.042 0.076 0.196 0.153 0.019 0.440 0.249 0.431 0.024 0.957 0.670 0.493 0.044 0.042 0.910 Table 9b: Quantile Regression Estimates, Salaried Women, Brazil, 1989 & 1995 1989 1995 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 at 0.134 0.601 1.085 1.681 2.308 0.250 0.718 1.180 1.680 2.265 0.036 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 elemen -1.140 -1.200 -1.311 -1.414 -1.468 -0.956 -1.095 -1.199 -1.340 -1.462 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 high -0.767 -0.758 -0.777 -0.801 -0.822 -0.641 -0.709 -0.749 -0.791 -0.827 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 agel -0.236 -0.307 -0.357 -0.487 -0.653 -0.250 -0.344 -0.416 -0.505 -0.629 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 age2 -0.074 -0.121 -0.154 -0.250 -0.391 -0.114 -0.162 -0.191 -0.224 -0.332 0.034 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 age3 0.034 0.008 -0.028 -0.064 -0.135 -0.019 -0.051 -0.070 -0.042 -0.079 0.354 0.768 0.320 0.102 0.010 0.525 0.015 0.001 0.132 0.067 married na na na na na 0.051 0.049 0.060 0.069 0.085 na na na na na 0.006 0.001 0.000 0.000 0.001 nocard -0.521 -0.423 -0.296 -0.252 -0.215 -0.318 -0.288 -0.262 -0.182 -0.085 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.024 south 0.482 0.330 0.278 0.182 0.124 0.348 0.314 0.280 0.232 0.169 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000 0.000 0.000 s. east 0.446 0.314 0.298 0.291 0.253 0.326 0.293 0.312 0.310 0.292 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 retail -0.150 -0.125 -0.144 -0.092 -0.007 -0.051 -0.063 -0.090 -0.083 -0.096 0.000 0.000 0.000 0.000 0.846 0.013 0.002 0.000 0.000 0.015 servsoc -0.194 -0.176 -0.143 -0.086 -0.032 -0.052 -0.061 -0.057 -0.042 -0.057 0.000 0.000 0.000 0.001 0.543 0.044 0.008 0.013 0.107 0.145 serv -0.205 -0.227 -0.233 -0.219 -0.185 -0.107 -0.114 -0.145 -0.127 -0.127 0.000 0.000 0.000 0.000 0.001 0.000 0.000 0.000 0.000 0.002 other 0.012 0.095 0.244 0.328 0.352 0.171 0.147 0.130 0.106 0.018 0.763 0.020 0.000 0.000 0.000 0.000 0.003 0.000 0.010 0.762 finan 0.345 0.483 0.653 0.680 0.803 0.377 0.490 0.594 0.614 0.591 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 pfserv -0.079 -0.038 0.027 0.091 0.217 0.031 0.063 0.093 0.125 0.013 0.050 0.264 0.450 0.122 0.001 0.519 0.056 0.001 0.000 0.806 pub -0.339 -0.190 -0.103 -0.076 -0.071 -0.071 -0.053 -0.091 -0.058 -0.088 0.000 0.000 0.000 0.015 0.122 0.003 0.010 0.000 0.005 0.013 prof 0.176 0.215 0.180 0.152 0.118 0.224 0.229 0.264 0.202 0.172 0.000 0.000 0.000 0.000 0.006 0.000 0.000 0.000 0.000 0.000 black -0.102 -0.138 -0.235 -0.287 -0.317 -0.091 -0.102 -0.119 -0.167 -0.206 0.005 0.000 0.000 0.000 0.000 0.013 0.000 0.000 0.000 0.000 mixed -0.149 -0.139 -0.184 -0.223 -0.278 -0.124 -0.120 -0.130 -0.144 -0.176 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 asian 0.046 0.123 0.150 0.045 0.070 -0.071 0.158 0.186 0.162 0.161 0.712 0.250 0.198 0.597 0.583 0.509 0.142 0.089 0.036 0.407 Table 9c: Quantile Regression Estimates, Self-employed Women, Brazil, 1989 & 1995 1989 1995 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 a -3.765 -3.341 -2.429 -1.224 -0.734 -2.500 -1.810 -1.308 -0.899 -0.529 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 eduyear 0.089 0.097 0.086 0.083 0.090 0.060 0.061 0.062 0.069 0.075 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 age 0.014 0.015 0.013 0.013 0.014 0.013 0.012 0.013 0.016 0.019 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 married na na na na na 0.069 0.081 0.121 0.155 0.177 na na na na na 0.004 0.000 0.000 0.000 0.000 domser 0.629 0.720 0.411 -0.311 -0.440 0.419 0.066 -0.130 -0.342 -0.354 0.000 0000 0.001 0.000 0.000 0.000 0.406 0.004 0.000 0.000 south 0.840 0.656 0.541 0.512 0.420 0.499 0.425 0.457 0.551 0.482 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 s. east 0.607 0.507 0.497 0.416 0.384 0.433 0.404 0.395 0.496 0.451 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 retail 1.266 1.470 1.326 0.905 0.955 0.696 0.506 0.502 0.400 0.403 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 servsoc 0.287 0.764 0.824 0.703 0.463 0.499 0.271 0.459 0.464 0.581 0.078 0.000 0.001 0.003 0.016 0.013 0.042 0.008 0.005 0.000 serv1 0.887 0.982 0.748 0.186 0.185 0.436 0.292 0.222 0.085 0.078 0.000 0.000 0.000 0.010 0.126 0.000 0.000 0.000 0.289 0.302 finan 2.521 2.100 1.938 1.526 0.948 1.255 0.988 1.044 0.769 1.335 1.000 1.000 0.000 1.000 1.000 1.000 0.000 0.000 0.067 1.000 pfserv 0.756 1.124 1.286 1.083 1.162 0.449 0.465 0.631 0.523 0.587 0.073 0.000 0.000 0.000 0.000 0.089 0.000 0.000 0.000 0.000 other 1.084 1.187 1.132 1.032 1.003 0.666 0.337 0.696 0.668 0.388 0.000 0.000 0.000 0.001 0.000 0.019 0.045 0.001 0.000 0.261 prof 0.782 0.569 0.539 0.282 0.461 0.422 0.534 0.489 0.598 0.517 0.000 0.009 0.002 0.145 0.001 0.018 0.000 0.002 0.000 0.000 black -0.094 -0.079 -0.136 -0.174 -0.195 -0.031 -0.086 -0.102 -0.095 -0.080 0.079 0.055 0.000 0.000 0.000 0.393 0.000 0.000 0.009 0.120 mixed -0.179 -0.200 -0.153 -0.175 -0.160 -0.072 -0.075 -0.103 -0.056 -0.055 0.000 0.000 0.000 0.000 0.000 0.002 0.000 0.000 0.018 0.055 asian 0.360 0.315 0.515 0.580 0.338 -0.007 -0.140 0.133 0.118 0.253 0.058 0.358 0.003 0.000 0.512 0.952 0.530 0.520 0.707 0.122 Table 9d: Quantile Regression Estimates, Self-employed Women, Brazil, 1989 and 1995 1989 1995 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 -1.537 -1.152 -0.298 0.949 1.198 -0.894 -0.0005 0.578 1.091 1.699 0.000 0.000 0.051 0.000 0.000 0.000 0.996 0.000 0.000 0.000 elemen -1.497 -1.383 -1.382 -1.349 -1.119 -0.833 -1.066 -1.110 -1.014 -1.012 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 high -0.940 -0.816 -0.833 -0.796 -0.456 -0.565 -0.806 -0.812 -0.690 -0.629 0.000 0.000 0.000 0.000 0.004 0.000 0.000 0.000 0.000 0.000 agel -0.346 -0.388 -0.322 -0.278 -0.288 -0.337 -0.300 -0.325 -0.410 -0.511 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 age2 0.088 0.059 0.038 0.026 0.000 -0.055 -0.061 -0.063 -0.063 -0.128 0.117 0.129 0.205 0.463 1.000 0.142 0.020 0.004 0.023 0.002 age3 0.072 0.056 0.079 0.080 0.148 0.108 0.028 0.034 0.037 -0.009 0.234 0.186 0.029 0.047 0.011 0.005 0.274 0.166 0.151 0.816 married na na na na na 0.006 0.057 0.076 0.135 0.128 na na na na na 0.822 0.014 0.000 0.000 0.000 domser 0.755 0.819 0.438 -0.378 -0.442 0.441 0.105 -0.095 -0.336 -0.434 0.000 0.000 0.003 0.000 0.000 0.001 0.234 0.116 0.000 0.000 south 0.830 0.684 0.594 0.510 0.405 0.525 0.452 0.477 0.548 0.492 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 s. east 0.632 0.570 0.549 0.458 0.426 0.459 0.413 0.424 0.512 0.502 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 retail 1.373 1.592 1.423 0.864 0.985 0.723 0.596 0.567 0.452 0.409 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 servsoc 0.530 0.827 0.920 0.438 0.708 0.572 0.450 0.409 0.453 0.710 0.000 0.000 0.000 0.056 0.000 0.000 0.001 0.000 0.014 0.000 serv1 0.952 1.076 0.809 0.181 0.272 0.510 0.373 0.299 0.134 0.058 0.000 0.000 0.000 0.019 0.000 0.000 0.000 0.000 0.078 0.480 other 0.877 1.163 1.049 1.093 1.119 0.980 0.791 0.992 0.863 1.384 0.001 0.000 0.001 0.000 0.000 1.000 0.036 0.000 0.037 1.000 finan 2.009 1.787 2.298 1.472 0.847 0.593 0.519 0.459 0.538 0.550 1.000 1.000 0.000 1.000 1.000 0.002 0.000 0.000 0.001 0.000 pfserv 0.733 1.073 1.169 0.849 1.222 0.693 0.395 0.860 0.689 0.461 0.002 0.000 0.000 0.000 0.000 0.075 0.025 0.001 0.000 0.207 prof 0.641 0.532 0.397 0.292 0.257 0.337 0.446 0.424 0.473 0.352 0.001 0.008 0.064 0.177 0.046 0.054 0.001 0.000 0.001 0.002 black -0.093 -0.044 -0.144 -0.238 -0.298 -0.069 -0.095 -0.110 -0.089 -0.041 0.134 0.332 0.000 0.000 0.000 0.055 0.001 0.000 0.016 0.462 mixed -.0.226 -0.196 -0.182 -0.198 -0.203 -0.085 -0.105 -0.098 -0.083 -0.073 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.048 asian 0.374 0.207 0.620 0.597 0.511 -0.179 -0.272 0.087 0.151 0.460 0.107 0.536 0.000 0.001 0.102 0.137 0.243 0.692 0.606 0.043 Table 10a: Quantile Regression Estimates, Salaried Women, Costa Rica, 1989/1995 1989 1995 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 a 3.389 3.705 3.934 3.782 3.726 3.810 3.712 3.902 3.944 4.014 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 eduyear 0.040 0.036 0.035 0.059 0.066 0.042 0.053 0.054 0.070 0.072 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 age 0.004 0.005 0.009 0.013 0.015 0.006 0.009 0.009 0.011 0.014 0.084 0.003 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 married 0.131 0.115 0.070 0.058 0.120 0.158 0.110 0.133 0.154 0.192 0.011 0.000 0.055 0.138 0.023 0.000 0.001 0.000 0.000 0.000 infsal -0.450 -0.412 -0.232 -0.157 -0.191 -0.380 -0.286 -0.119 0.071 0.216 0.001 0.000 0.001 0.003 0.040 0.000 0.000 0.008 0.448 0.024 white -0.180 -0.160 -0.259 -0.216 -0.222 -0.191 -0.186 -0.266 -0.313 -0.246 0.005 0.002 0.000 0.000 0.002 0.000 0.001 0.000 0.000 0.002 blue -0.128 -0.242 -0.370 -0.291 -0.251 -0.371 -0.308 -0.367 -0.355 -0.302 0.102 0.000 0.000 0.000 0.021 0.000 0.000 0.000 0.000 0.000 prim -0.698 -0.766 -1.037 -0.774 -1.012 -0.142 -0.280 -0.510 -0.737 -0.769 1.000 1.000 0.001 1.000 1.000 1.000 1.000 1.000 1.000 1.000 retail -0.199 -0.034 -0.007 -0.028 0.064 -0.282 -0.193 -0.186 -0.209 -0.199 0.016 0.523 0.875 0.519 0.295 0.000 0.000 0.000 0.000 0.006 finan 0.102 0.176 0.274 0.327 0.379 -0.180 -0.156 -0.065 0.075 0.009 0.647 0.006 0.019 0.001 0.003 0.032 0.011 0.280 0.491 0.928 servsoc 0.113 0.159 0.157 0.236 0.359 -0.289 -0.164 -0.113 -0.154 -0.184 0.483 0.000 0.095 0.006 0.000 0.000 0.015 0.001 0.011 0.027 other 0.208 0.183 0.109 0.013 0.521 -0.201 -0.065 0.024 0.090 -0.011 0.064 0.001 0.130 0.930 0.064 0.001 0.415 0.739 0.316 0.909 pubsec 0.240 0.147 0.154 0.079 0.053 0.178 0.217 0.241 0.218 0.242 0.115 0.000 0.049 0.364 0.388 0.003 0.002 0.000 0.000 0.002 metro 0.150 0.098 0.037 0.088 0.124 -0.020 -0.001 0.019 0.023 0.055 0.002 0.006 0.286 0.013 0.011 0.672 0.972 0.577 0.637 0.278 chob -0.023 -0.071 -0.106 -0.087 -0.105 -0.158 -0.105 -0.110 -0.033 -0.054 0.844 0.243 0.040 0.044 0.124 0.070 0.075 0.030 0.674 0.418 pacific 0.062 -0.026 -0.092 -0.111 -0.051 -0.029 -0.018 -0.073 -0.096 -0.003 0 0.339 0.584 0.069 0.010 0.347 0.561 0.676 0.080 0.170 0.983 huetar 0.053 0.039 -0.055 -0.063 0.045 -0.053 -0.037 -0.071 -0.042 -0.092 0.611 0.416 0.314 0.215 0.704 0.497 0.415 0.258 0.529 0.091 Table 10b: Quantile Regression Estimates, Salaried Women, Costa Rica, 1989/ 1995 1988 1995 0.10 0.25 0.50 0.75 0.90 0.10 0.23 0.50 0.75 0.90 a 4.115 4.430 4,825 5.219 5,526 4.651 4.884 5.174 5.575 5.676 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 element -0.292 -0.238 -0.288 -0.431 -0.621 -0.328 -0.373 -0.416 -0.634 -0.551 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 highl -0.165 -0.113 -0.158 -0.265 -0.334 -0.237 -0.321 -0.358 -0.429 -0.372 0.002 0.010 0.009 0.000 0.000 0.000 0.000 0.000 0.000 0.000 univI -0.044 0.036 -0.113 -0.099 -0.296 0.008 -0.105 -0.225 -0.143 -0.270 1.000 0.775 0.396 0.405 1.000 0.880 0.114 0.034 0.053 0.045 agel -0.002 -0.112 -0.209 -0.319 -0.392 -0.127 -0.195 -0.226 -0.279 -0.268 0.9816 0.019 0.000 0.000 0.000 0.089 0.000 0.000 0.000 0.000 age2 0.055 0.001 -0.114 -0.166 -0.228 -0.115 -0.150 -0.119 -0.125 -0.022 0.638 0.987 0.050 0.005 0.035 0.134 0.000 0.003 0.042 0.796 age3 0.037 0.001 -0.054 -0.051 -0.044 0.001 -0.046 -0.028 0.003 0.059 0.737 0.987 0.204 0.437 0.709 0.992 0.263 0.425 0.960 0.423 married 0.134 0.118 0.068 0.080 0.111 0.135 0.119 0.151 0.159 0.153 0.008 0.000 0.040 0.039 0.031 0.002 0.000 0.000 0.000 0.000 infsal -0.531 -0.461 -0.291 -0.162 -0.155 -0.473 -0.311 -0.090 -0.001 0.178 0.000 0.000 0.000 0.000 0.181 0.000 0.000 0.127 0.990 0.174 white -0.192 -0.214 -0.249 -0.223 -0.255 -0.183 -0.185 -0.257 -0.280 -0.314 0.000 0.000 0.000 0.000 0.007 0.000 0.000 0.000 0.000 0.000 blue -0.192 -0.326 -0.396 -0.387 -0.353 -0.407 -0.400 -0.470 -0.445 -0.423 0.011 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 prim -0.702 -0.942 -1.116 -1.148 -1.229 -0.221 -0.377 -0.641 -0.779 -1.075 1.000 1.000 0.000 1.000 1.000 1.000 1.000 1.000 1.000 1.000 retail -0.133 -0.039 0.012 -0.016 0.022 -0.208 -0.136 -0.172 -0.204 -0.129 0.129 0.439 0.750 0.726 0.771 0.000 0.000 0.000 0.000 0.020 finan 0.128 0.174 0.245 0.252 0.395 -0.242 -0.111 -0.026 0.010 0.074 0.520 0.052 0.034 0.040 0.011 0.034 0.097 0.554 0.910 0.444 servsoc 0.087 0.151 0.209 0.270 0.302 -0.289 -0.166 -0.156 -0.173 -0.039 0.573 0.010 0.003 0.000 0.000 0.000 0.031 0.001 0.001 0.605 other 0.212 0.197 0.157 0.130 0.485 -0.159 -0.123 -0.009 -0.002 0.100 0.025 0.003 0.001 0.237 0.044 0.017 0.096 0.913 0.979 0.399 pubsec 0.294 0.154 0.169 0.090 0.051 0.259 0.250 0.285 0.201 0.209 0.072 0.000 0.003 0.223 0.423 0.000 0.000 0.000 0.000 0.009 metro 0.070 0.089 0.030 0.080 0.117 -0.015 0.026 0.044 -0.003 -0.003 0.151 0.006 0.414 0.044 0.009 0.749 0.509 0.112 0.935 0.959 chob -.01151 -0.098 -0.123 -0.104 -0.096 -0.169 -0.051 -0.088 -0.095 -0.137 0.197 0.018 0.020 0.073 0.231 0.132 0.357 0.023 0.200 0.089 pacifico 0.003 -0.057 -0.083 -0.150 -0.097 0.018 -0.041 -0.077 -0.120 -0.099 0.960 0.249 0.098 0.013 0.107 0.800 0.179 0.083 0.016 0.517 huetar -0.018 0.015 -0.021 -0.100 -0.046 0.031 -0.024 -0.100 -0.066 -0.190 0.842 0.737 0.682 0.002 0.788 0.745 0.583 0.045 0.090 0.060 Table 10c: Quantile Regression Estimates, Self-Employed Women, Costa Rica, 1989 & 1995 1989 1995 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 c 3.033 3.435 3.452 3.900 4.667 3.866 3.646 3.909 4.578 5.468 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 eduyears 0.070 0.070 0.070 0.060 0.058 0.021 0.038 0.032 0.034 0.022 0.005 0.001 0.000 0.019 0.038 0.393 0.057 0.020 0.015 0.064 age 0.000 0.004 0.011 0.013 0.007 -0.002 -0.001 0.005 0.000 -0.007 0.949 0.464 0,015 0.005 0.340 0.616 0.896 0.343 0.906 0.051 married 0.157 0.097 0.149 0.221 0.442 -0.478 0.002 0.193 0.124 0.055 0.418 0.428 0.157 0.049 0.016 0.001 0.987 0.039 0.277 0.667 domser -0.286 -0.465 -0.374 -0.615 -0.451 -0.649 -0.712 -0.596 -0.838 -0.936 0.470 0.003 0.057 0.000 0.077 0.000 0.000 0.000 0.000 0.000 prof -0.045 -0.054 0.050 -0.173 -0.374 0.238 -0.055 0.083 0.770 0.502 0.939 0.881 0.843 0.145 0.286 0.622 0.841 0.665 0.010 0.416 retail -0.227 -0.095 -0.096 -0.069 -0.175 -0.367 0.055 0.268 0.393 0.474 0.439 0.511 0.517 0.674 077 0.014 0.734 0.143 0.018 0.025 finan -0.009 -0.774 0.244 0.383 0.085 -0.454 0.665 0.376 -0.193 0.140 1.000 1.000 0.751 1.000 1.000 1.000 0.299 0.239 0.544 1.000 servsoc -0.193 -0.186 -0.057 -0.035 -0.326 -0.145 0.163 -0.014 0.056 0.093 0.667 0.229 0.788 0.730 0.200 0.421 0.151 0.906 0.673 0.539 other -1.198 -1.056 -0.865 0.414 0.202 -0.265 -0.210 -0.456 0.778 0.336 1.000 1.000 0.420 1.000 1.000 1.000 1.000 0.512 1.000 1.000 metro -0.490 -0.145 -0.017 0.088 0.158 -0.177 0.011 0.090 0.195 0.060 0.148 0.325 0.899 0.357 0.403 0.288 0.902 0.435 0.161 0.735 chob -0.594 -0.558 -0.538 -0.281 -0.488 -0.278 -0.145 -0.250 -0.157 -0.408 0.011 0.000 0.000 0.060 0.052 0.040 0.225 0.061 0.294 0.078 pacifico -0.384 -0.373 -0.336 -0.427 -0.544 -0.043 -0.079 0.157 0.234 -0.158 0.101 0.036 0.001 0.005 0.005 0.787 0.692 0.374 0.156 0.512 huetar -0.665 -0.484 -0.503 -0.273 0.078 -0.338 -0.155 -0.195 -0.173 -0.352 0.012 0.050 0.004 0.137 0.825 0.153 0.186 0.177 0.323 0.235 Table 10d: Quantile Regression Estimates, Self-Employed Women, Costa Rica, 1989.& 1995 1989 1995 0.10 0.25 0.50 0.75 0.90 0.10 0.25 0.50 0.75 0.90 U 3.904 4.280 4.765 5.062 5.680 4.756 4.385 4.732 5.006 5.543 0.000) 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 element -0.691 -0.486 -0.423 -0,078 -0.373 -0.921 -0.603 -0.483 -0.369 -0.439 1.000 0.154 0.055 0.849 1.000 0.001 0.005 0.067 0.341 0.505 highl -0.083 -0.174 -0.041 0.131 -0.373 -0.686 -0.445 -0.467 -0.282 -0.410 1.000 0.581 0.867 0.741 1.000 0.006 0.027 0.064 0.432 0.474 univI na na na na na -0.985 -1.385 -1.749 -2.012 -2.643 na na na na na 1.000 1.000 1.000 1.000 1.000 agel -0.011 -0.163 -0.307 -0.357 -0.037 -0.130 -0.060 -0.108 0.088 0.336 0.970 0.296 0.122 0.027 0.872 0.365 0.637 0.443 0.688 0.078 age2 0.249 0.179 -0.163 -0.269 0.042 0.052 0.182 0.099 0.114 0.128 0.311 0.182 0.376 0.083 0.851 0.568 0.231 0.469 0.434 0.374 age3 0.198 0.067 -0.090 -0.227 0.042 0.132 0.232 0.230 0.258 0.088 0.325 0.625 0.663 0.175 0.779 0.432 0.085 0.035 0.107 0.525 married -0.083 0.100 0.101 0.227 0.518 -0.443 -0.059 0.182 0.107 0.267 0.610 0.307 0.398 0.142 0.003 0.000 0.683 0.024 0.301 0.135 domser -0.228 -0.319 -0.648 -0.659 -0.725 -0.586 -0.697 -0.575 -0.837 -0.896 0.382 0.022 0.009 0.000 0.012 0.000 0.000 0.000 0.000 0.000 prof -0.064 -0.112 -0.178 -0.107 -0.333 -0.018 0.179 0.200 0.636 0.338 0,894 0.702 0.542 0.635 0.462 0.966 0.596 0.359 0.040 0.585 retail -0.189 -0.076 -0.103 -0.081 -0.206 -0.208 -0.064 0.282 0.107 0.337 0.284 0.678 0.437 0.606 0.553 0.233 0.689 0.081 0.545 0.040 finan -0.014 -0.533 0.354 0.619 0.367 -0.944 0.149 -0.117 -0.210 0.163 1.000 1.000 0.641 1.000 1.000 1.000 0.821 0.742 0.560 1.000 servsoc -0.215 -0.257 0.090 -0.028 -0.121 -0.105 0.152 0.011 0.059 -0.019 0.595 0.079 0.733 0.871 0.596 0.638 0.180 0.929 0.722 0.904 other -0.980 -0.825 -1.048 0.274 0.148 -0.270 -0.241 -0.431 0.690 -0.072 1.000 1.000 0.281 1.000 1.000 1.000 1.000 0.540 1.000 1.000 metro -0.254 -0.093 -0.009 0.030 0.112 -0.326 -0.034 -0.003 0.225 0.203 0.280 0.235 0.954 0.747 0.559 0.003 0.684 0.978 0.087 0.237 chob .0.525 -0.542 -0.430 -0.434 -0.631 -0.309 -0.223 -0.274 -0.092 -0.307 0.000 0.000 0.006 0.002 0.006 0.028 0.103 0.018 0.592 0.239 pacifico -0.398 -0.366 -0.363 -0.448 -0.577 -0.156 -0.185 0.007 0.253 0.039 0.013 0.009 0.006 0.008 0.020 0.310 0.353 0.970 0.241 0.876 huetar -0.796 -0.505 -0.402 -0.393 0.009 -0.316 -0.188 -0.355 -0.210 -0.156 0.000 0.001 0.060 0.034 0.982 0.164 0.152 0.010 0.257 0.492 References Arias Omar, 2000, "Marginalization of men in Argentina: Evidence from changes in wage distributions," working paper, World Bank, Washington DC. Beyer, H., P. Rojas and R. Vergara, 1999, "Trade liberalization and wage inequality," Journal of Development Economics, Vol. 59, 103-123. Buchinsky Moshe, 1994, "Changes in the U.S. wage structure 1963-1987: Application of Quantile Regression," Econometrica, Vol. 62, 2, pp. 405-458. Buchinsky Moshe, 1998, "The dynamics of changes in the female wage distribution in the USA: A quantile regression approach," Journal ofApplied Econometrics, Vol. 13, 1, pp. 1-30. Cunningham W. and W. Maloney, 1997, "Heterogeneity in small-scale LDC enterprises: The Mexican case," Unpublished working paper, World Bank, Washington DC. Cunningham W. and W. Maloney, 1998, "Heterogeneity among Mexico's Micro- Enterprises, An application of factor and cluster analysis, Policy Research Working paper, World Bank, Washington DC. Hill, Anne M., 1988, "Female labor supply in Japan: Implications of the informal sector for labor force participation and hours of work," The Journal of Human Resources, 26, 1, pp. 143-161. Kazt, L. and K. Murphy, 1992, "Changes in relative wages, 1963-1987: Supply and demand factors," Quarterly Journal of Economics 107 (1), 35-78. Killingsworth Mark R. and James J. Heckman, 1986, "Female labor supply: A survey," Handbook of Labor Economics, Vol. 1, pp. 103-204, 0. Ashenfelter and R. Layard, Eds. Karoly Lynn A., 1992, "Changes in the distribution of individual earnings in the United States: 1967-1986," The review ofEconomics and Statistic 74(1), 107-115. Koenker Roger and G. Bassett, 1978, "Regression Quantiles," Econometrica, 46, pp. 33- 50. Lam D. and R. F. Schowni, 1993, "Effects of family background on earnings and returns to schooling: Evidence from Brazil," Journal of Political Economics, Vol. 101, No. 4. Maloney William F., 1998, "Are LDC Labor Markets Dualistic?" working paper, World Bank. Montenegro, C. E., 1998, "The structure of wages in Chile 1960-1996: An application of quantile regression, Estudios de Economia, Vol. 25, 1, pp. 71-98. Siverman B., 1986, "Density estimation for statistical data analysis," Chapman Hall. Wood Adrian, 1997, "Openness and wage inequality in developing countries: The Latin American challenge to East Asian conventional wisdom," The World Bank Economic Review, Vol. 11(1), 33-57. APPENDIX Definitions of Dummy Variables Informal: In Argentina informal salaried is defined as women working in small firms (less than 6 workers) without work insurance and retirement plan coverage. In Brazil, informal salaried are women without signed federal card (excluding public servants). In Costa Rica, informal is defined as those salaried women working in small firms (less than 6 workers) that do not have social security. Education: In Argentina the dummy element groups women with elementary education, highl and high2 group women with complete and incomplete high school, respectively, and the dummies univl and univ2 group women with complete and incomplete college education, respectively. Completed college (univ1) is the omitted category. Tenure (Argentina): the dummies are less than 1 year (tenurel), between I and 5 (tenure2), between 5 and 10 (tenure3) and more than 10 years (tenure4, the excluded dummy). Industry: eight dummies capture inter-industry wage differentials. These dummies are manuf (the omitted dummy), retail, sersoc, serv1, finan, pfserv, pub and other. Manuf groups women working in the production of food, beverages, cigarettes, textiles, clothing, chemical and metallic products, and other manufacturing; retail groups women working on sales and commerce; servsoc groups women working in social services, excluding public social services; servl groups women working in non professional services; finan groups women working in the banking and real state sectors; pfserv groups women working in professional services; pub groups women working in the public sector. Finally, other is a residual dummy that groups women working in other sectors not included in the other dummies (communications, construction, utilities, etc.). Regional: In Brazil three regional dummies were included: south, southeast, and northern. The south dummy includes Parana, the southeast dummy includes Minas Gerais, Rio de Janeiro and Sao Paulo, and the northern region (the omitted dummy) includes Bahia, Ceara, Pernambuco and Maranhao. In Costa Rica we included five regional dummies: chobrun, pacific, huertar, metro and san Jose (the omitted dummy). The dummy metro groups women working in the metropolitan area; chobrun groups women working in Chorotega and Brunca; pacific groups women working in the Central Pacific region; huertar groups women in the Atlantic and Northern Huertar regions. Skills on the job: Each occupational dummy groups women according to three different levels of reported skills: professional (the omitted dummy), skilled and unskilled. The first dummy groups women that reported to be working in jobs that required professional qualifications. The dummy skilled groups women that reported to be working in jobs that required high levels of qualification. The dummy unskiled groups women that reported to be working in jobs that did not required any qualification. Policy Research Working Paper Series Contact Title Author Date for paper WPS2717 Bridging the Economic Divide within Raja Shankar November 2001 A. Santos Nations: A Scorecard on the Anwar Shah 31675 Performance of Regional Development Policies in Reducing Regional Income Disparities WPS2718 Liberalizing Basic Carsten Fink November 2001 R. Simms Telecommunications: The Asian Aaditya Mattoo 37156 Experience Randeep Rathindran WPS2719 Is There a Positive Incentive Effect Truman G. Packard November 2001 T. Packard from Privatizing Social Security? 89078 Evidence from Latin America WPS2720 International Migration and the Global Andres Solimano November 2001 A. Bonfield Economic Order: An Overview 31248 WPS2721 Implications for South Asian Countries Sanjay Kathuria November 2001 M. Kasilag of Abolishing the Multifibre Will Martin 39081 Arrangement Anjali Bhardwaj WPS2722 Japan's Official Development Masahiro Kawai November 2001 J. Mendrofa Assistance: Recent Issues and Shinji Takagi 81885 Future Directions WPS2723 Using Development-Orineted Equity Helo Meigas November 2001 S. Torres Investment as a Tool for Restructuring 39012 Transition Banking Sectors WPS2724 Tropical Bubbles: Asset prices in Santiago Herrera November 2001 R. lzquierdo Latin America, 1980-2001 Guillermo Perry 84161 WPS2725 Bank Regulation and Supervision: James R. Barth November 2001 A. Yaptenco What Works Best? Gerard Caprio Jr. 38526 Ross Levine WPS2726 Applying the Decision Rights Florence Eid November 2001 A. Santos Approach to a Case of Hospital 31675 Institutional Design WPS2727 Hospital Governance and Incentive Florence Eid November 2001 A. Santos Design: The Case of Corporatized 31675 Public Hospitals in Lebanon WPS2728 Evaluating Emergency Programs William F. Maloney December 2001 A. Pillay 88046 WPS2729 International Evidence on the Value Luc Laeven December 2001 R. Vo of Product and Geographic Diversity 33722 Policy Research Working Paper Series Contact Title Author Date for paper WPS2730 Antidumping as Safeguard Policy J. Michael Finger December 2001 R. Simms Francis Ng 37156 Sonam Wangchuk WPS2731 An Alternative Technical Education Gladys Lopez-Acevedo December 2001 M. Geller System in Mexico: A Reassessment 85155 of CONALEP WPS2732 The Unbalanced Uruguay Round J. Michael Finger December 2001 R. Simms Outcome: The New Areas in Future Julio J. Nogues 37156 WTO Negotiations WPS2733 Trade Policy Reform and Poverty Bernard Hoekman December 2001 R. Martin Alleviation Constantine Michalopoulos 39065 Maurice Schiff David Tarr WPS2734 Agricultural Markets in Benin and Marcel Fafchamps December 2001 P. Kokila Malawi: The Operation and Eleni Gabre-Madhin 33716 Performance of Traders WPS2735 Shifting Tax Burdens through Bernard Gauthier December 2001 H. Sladovich Exemptions and Evasion: An Empirical Ritva Reinikka 37698 Investigation of Uganda WPS2736 Social Policy and Macroeconomics: F. Desmond McCarthy December 2001 J. Turner The Irish Experience 81767 WPS2737 Mode of Foreign Entry, Technology Aaditya Mattoo December 2001 R. Martin Transfer, and Foreign Direct Marcelo Olarreaga 39065 Investment Policy Kamal Saggi WPS2738 Assisting the Transition from Emanuela Galasso December 2001 C. Cunanan Workfare to Work: A Randomized Martin Ravallion 32301 Experiment Agustin Salvia WPS2739 Poverty, Education, and Health in Peter Lanjouw December 2001 P. Sader Indonesia: Who Benefits from Public Menno Pradhan 33902 Spending? Fadia Saadah Haneen Sayed Robert Sparrow WPS2740 Are Men Benefiting from the New Omar Arias December 2001 S. Nyairo Economy? Male Economic 34635 Marginalization in Argentina, Brazil, and Costa Rica