Lasa-4.univ.pitt.edu

R A C E , G E N D E R , A N D W O R K I N S Ã O P A U L O , B R A Z I L , 1 9 6 0 – 2 0 0 0 Received 4-19-2005; Revise and Resubmit 6-28-2005; Revised received 8-9-2005; Final Acceptance 11-29-2005 Abstract: This study relies on Brazilian census data from 1960–2000 to analyzelong-term trends in racial and gender wage disparities in the urban labor marketof São Paulo, one of Latin America’s most dynamic economies. Afro-Braziliansand women have made remarkable progress over the past four decades in securinghard-won legal rights and in gaining access to the highest levels of schooling,entrance into higher paying occupations, and narrowing the intraethnic genderwage gap. Despite such progress, Afro-Brazilians and women are paid less thansimilarly qualified white men, and wage discrimination is increasing. Placing theinterplay of race and gender at the center of this analysis shows how the work-place barriers people confront on the basis of skin color and sex play a fundamen-tal role in shaping social and economic inequality in contemporary Brazil. In the 1970s, feminist scholarship in Latin America challenged the male bias and gender blindness inherent in early studies of the region’sdevelopment. Influenced by Ester Boserup’s path-breaking work onwomen in Africa and drawing on forceful critiques of modernizationtheory, feminist research on women in Latin America highlighted thegendered consequences of uneven economic growth and the centralityof women’s productive and reproductive roles in the development pro-cess (Safiotti 1969; Nash and Safa 1976; Nash and Fernandez-Kelly 1983).1Landmark studies at the time provided vital evidence of the work thatwomen do and contributed decisive insights to theoretical debates thatcontinue today about the role of women in local, national, and interna-tional economies (Bose and Acosta-Belén 1995; Blumberg et al., 1995;Visvanathan 1997). The study of gender and development remains adynamic and interdisciplinary field of inquiry which draws from aca-demically grounded theories and empirical analyses of popular 1. See Visvanathan (1997), Tinker (1997), and Jaquette (1982) for a review of the ori- gins of the women and development field.
Latin American Research Review, Vol. 41, No. 3, October 2006 2006 by the University of Texas Press, P.O. Box 7819, Austin, TX 78713-7819 movements. Imbued with an activist spirit, the study of women anddevelopment has reevaluated dominant paradigms and redefined so-cial research in ways that continue to invite students and scholars intoLatin American studies.
As women and development studies evolve, new critiques have emerged. Third world women, and women of color, have challengedthe universality of feminism by underscoring the significance of race,class, and nation (Sen and Grown 1987; Mohanty et al. 1991). Akin to themale bias found in traditional studies of development is the relativeabsence of race and ethnicity from the literature on women and work inLatin America.2 Indeed, race continues to be missing from the most re-cent analyses of women’s work in the region (Hite and Viterna 2005).
When racial or ethnic differences find explicit recognition, it is oftennoted in passing, accompanied by weak admonitions about the need forgreater attention to the divisions among women.
The concept of “intersectionality” is at the heart of current postmodern and feminist perspectives. Minority women scholars and activists in par-ticular have endeavored to develop new theoretical insights that stress theinterconnected experiences of race and gender (as well as class). The inter-sectional approach has underscored the conclusion that the social posi-tion of women of color is distinct from that of either white women or menof color (Hooks 1984; Collins 1990). Most analyses in this vein rely prima-rily on historical, ethnographic, or textual treatments to illustrate differ-ences between groups. Quantitative approaches to the intersection of raceand gender are rare (but see Browne 1999). Studies of gender and race-based labor market inequalities in Latin America are rarer still.3 Race and gender assume particular relevance in the case of Brazil.
Home to Latin America’s largest African-descent population, Brazil isone of the few countries in the region where both race and gender areaccorded a prominent place in political life and official discourse. In re-cent decades, black activists and feminists have made notable strides inincreasing public awareness of race and gender inequalities and havesuccessfully struggled to secure legal rights. Despite such progress,marked race and gender disparities continue to characterize Brazil’s la-bor market. Blacks predominate in the lowest ranks of the labor force,and the wage gap between women and men ranks among the highest inLatin America and the Caribbean (Winter 1994). In 2000, women’s 2. This is also true in the United States, where research has firmly established that workers’ sex and race affect virtually every aspect of their labor market experiences.
Yet, the majority of U.S. labor studies continue to neglect the simultaneous relationshipbetween gender and race (Reskin and Charles 1999).
3. Exceptions include: Bruschini and Lombardi 2002; Guimarães 2002; Lovell 1994, RACE, GENDER, AND WORK IN SÁO PAULO 65
salaries lagged 36 percent behind men’s, with Afro-Brazilian womenbeing the most disadvantaged of all groups (IBGE 2003). Evidence ofpersistent income disparities are especially troubling given that, on av-erage, both Afro-Brazilian and white women have completed more yearsof schooling than their male counterparts and are increasingly employedin traditional male occupations (Lovell 2000).
Using Brazilian census data from 1960–2000, I analyze long-term trends in wage disparities in the urban labor market of São Paulo, Brazil’s mosteconomically developed state. Although past studies provide ample evi-dence that Afro-Brazilians (Silva 1985, 1988) and women suffer wage dis-crimination (Ometto et al. 1999; Oliveira and Hachado 2000; Scorzafaveand Menezes-Filho 2001; Soares and Izaki 2002), few examine the simulta-neous race and gender dimensions of workplace inequality. Placing theinterplay of race and gender at the center of this analysis shows howchanges in the structure of the urban labor market differentially influencethe status of subgroups of the Brazilian population. The workplace barri-ers people confront on the basis of skin color and sex play a fundamentalrole in shaping social and economic inequality in contemporary Brazil.
RACE AND GENDER IN BRAZIL
Brazil’s lack of overt racial tensions and long history of widespread miscegenation, resulting in an elaborate system of multiracial classifica-tion, make Brazil unique in comparative studies of race. These distinc-tive characteristics contributed to the widely held—but nowdiscredited—view that Brazil, unlike the United States, is a “racial de-mocracy,” free of race-based violence, segregation and discrimination.
So entrenched was the myth of racial democracy that for nearly 100 yearsafter the abolition of slavery there was virtually no public discussion ofracism. That changed with the centennial of the abolition of slavery in1988, when a new wave of black militant groups and activist scholarsrefuted the ideology of racial equality and pointed to the existence ofwidespread racial inequality.
When Brazil’s transition to democracy began in the late 1970s, the government introduced policies that would ultimately return the coun-try to civilian rule; numerous black organizations emerged includingthe Moviemento Negro Unificado, a national-level political movement.
Black militants successfully campaigned for constitutional amendmentsto end racial discrimination and began an unprecedented public dia-logue on the role of skin color in structuring contemporary opportuni-ties and rewards in society.4 For all of the nuance and complexity of race 4. Although rarely enforced, Federal antidiscrimination laws were passed in Brazil as and the social meaning of skin color, there is now a wealth of empiricalevidence that racial inequality, prejudice, and discrimination are fea-tures of everyday life in Brazil (Hasenbalg and Silva 1992; Andrews 1992;Reichmann 1999; Lovell and Wood 1998; Telles 2004). Under the recentpresidency of Fernando Henrique Cardoso, Brazil began to acknowl-edge and combat these inequalities and established Latin America’s onlynational affirmative action policies aimed at increasing Afro-descendants’access to education and employment (Telles 2004; Htun 2004). Indeed,no other country in Latin America has seen the same degree of blackpolitical mobilization as has taken place in Brazil (Andrews 2004).5 Issues of gender inequality also emerged as important rallying cries in the opposition politics of Brazil’s “new social movements” (Tabak1994; Alvarez 1990). Women’s efforts focused, in general, on the demandfor better conditions in daily life and a more egalitarian society. Many ofwomen’s petitions were also transformed into constitutional rights. Forinstance, provisions on equality between women and men, a chapter onthe family, maternity and paternity leaves, new rights for domestic andrural workers, and antidiscrimination legislation were all included inthe rewritten 1988 constitution. To protect women from domestic abuse,Brazil created the world’s first women’s police stations (delegacias). In1996, the country approved a women’s quota law requiring that politi-cal parties must reserve a minimum of 30 percent of candidate slots forwomen (Htun 2002). Councils that represent women’s concerns currentlyexist in all state and local governments. From small groups of women inpopular social movements to national and international networks, Bra-zilian feminists have pioneered some of the world’s most advanced leg-islation and innovative mechanisms to promote gender equity.
Yet, despite common concerns and objectives, black and white women often engage in separate political struggles. Afro-Brazilian feminists ini-tially shunned predominantly white women’s groups because suchgroups seldom addressed the specificity of black women’s situations.
Likewise, some Afro-Brazilian women felt the male-dominated blackmovement has not adequately addressed their needs or concerns. Thecombined effects of racism in the feminist movement and sexism in theemerging black movement led Afro-Brazilian women to found autono-mous groups (Alvarez 1994; Hanchard 1994). To address their specific 5. Race-based collective action in Brazil has never reached the scale of the U.S. civil rights movement or the South African liberation movement. Research suggest that ob-stacles to popular black mobility in Brazil include a weak “racial consciousness,” (Mitchell1985; Hanchard 1994), religious divisions (Burdick 1998), class divisions between theactivists and those whom they seek to mobilize, the still powerful ideology of racialdemocracy and the popular alienation from activist discourse about “blackness”(Andrews 2004).
RACE, GENDER, AND WORK IN SÁO PAULO 67
issues, Afro-Brazilian women have organized national conferences, semi-nars, and created black women’s organizations and institutes. GeledésInstituto da Mulher Negra and Fala Preta are two of the best know blackwomen’s nongovernmental organizations in São Paulo that confront is-sues of reproductive rights, labor market discrimination, and health con-cerns. Afro-Brazilian women also participate in the national women’smovement (Safa 2003). As a consequence, since the early 1990s there hasbeen a growing awareness of and attention to race and gender concernswithin both feminist and black discourses.6 RACIAL IDENTITY AND CENSUS DATA
As in many countries, racial terminology in Brazil is multifaceted and reflects changing cultural and political ideas and struggles. Today inBrazil, the terms negro and afro-brasileiro are frequently used to signifythose of African descent. The black movement has promoted the use ofnegro as an affirmation of African origin and identity and as a rejectionof the assimilationist and whitening ideal of mestizaje (Maggie 1989).
Putting aside any pretense of measuring “race” in the genetic sense, theBrazilian Census Bureau has traditionally asked respondents to choosetheir own identity among four color categories: branco (white), pardo (lit-erally translated as grey but interpreted as brown), preto (black) andamarelo (yellow).
There is considerable controversy regarding the census’ method of data collection. Anthropological research on racial self-identificationamply documents the numerous racial distinctions Brazilians apply topeople who vary with respect to hair texture, skin tone, and variousfacial markers (Harris 1964). The Census Bureau conducted its own sur-vey in 1976 and found that an open-ended question on racial self-classi-fication elicited nearly two hundred different responses (Silva 1988). Thewide variety of terms Brazilians use to identify race questions the valid-ity of the simplified census scheme. Another concern is the interplaybetween a person’s social class and color and the associated mutabilityof racial identity. Very dark-skinned persons who are also poor are likelyto be thought of—and classify themselves—as black, but high-statuspersons of the same skin tone are more likely to be thought of—and toclassify themselves—as closer to the white end of the color continuum.
The fact that social class plays a role in racial classification suggests thepossibility that upwardly mobile individuals may reclassify themselvesfrom darker into lighter racial categories after rising in socioeconomicstatus.
6. Two Brazilian journals devoted to examining gender and racial dynamics are Estudos Feministas and Estudos Afro-Asiáticos.
To estimate the mobility of people from one color category to another, De Carvalho, Wood, and Andrade (2004) studied the consistency and sta-bility of the Brazilian census’ self-classification of racial categories from1950 to 1980. They found evidence of considerable reclassification out ofthe preto and into the pardo category. They did not, however, find evi-dence of reclassification from preto or pardo into white. Their study con-cludes that the boundary between pretos and pardos is ambiguous andunstable, while the boundary between white and nonwhite (the combina-tion of pretos and pardos) is relatively unambiguous and remarkably stable.
Their findings provide compelling reasons to collapse pardos and pretosinto a single Afro-Brazilian category when census data are used to ana-lyze changes in racial disparities over time. Given the multidimensionalbasis of racial identity, it is not surprising that the strategy of convertingthe already simplified census typology to a white/Afro-Brazilian di-chotomy has been criticized for imposing an inappropriate bipolar racialscheme on Brazil (Harris et al. 1993). Others contend that, despite the pro-fusion of racial terms, the great socioeconomic divide is nonetheless be-tween white and Afro-Brazilian (Andrews 1992; Lovell and Wood 1998).
When faced with the questionable reliability and accuracy of statistical data on race in Brazil, why use census data? As this and previous researchhas shown, race matters in determining access to education, jobs, andwages. Only with data on racial disparities can we begin to identify thedeep inequalities that divide Brazilian society. The individual-level data Iuse in this study from the 1960, 1980, 1991, and 2000 public use samples ofthe Brazilian Censuses of Population make it possible to estimate howrace and gender together affect the schooling, jobs, and wages of Afro-Brazilian and white women and men. My sample is limited to wage earn-ing employees aged 18–64 in São Paulo’s urban labor market.7 Followingan approach used by England, Christopher, and Reid (1999), my focus ison whether, and how much, race-based gaps in labor market predictorsand outcomes vary by sex and whether sex-based gaps vary by race.
LABOR MARKET INDICATORS IN SÃO PAULO: SCHOOLING, OCCUPATIONAL
SORTING, AND WAGES
Racial and gender labor market inequality have typically been ana- lyzed in terms of differences in individual-level qualifications (Mincerand Polachek 1974). Human capital explanations of the gender gap focus 7. Because my sample omits workers who are self-employed or unemployed, my con- clusions do not explain racial or gender inequality in access to jobs. My results insteadpertain only to the factors that affect wages among current employees.
RACE, GENDER, AND WORK IN SÁO PAULO 69
on how differences in men’s and women’s plans for work will lead themto different levels of investments in their own skills and education (i.e.,human capital). Women are assumed to plan noncontiguous employmentso that they can accommodate childbearing and homemaking. As such,human capital theorists argue that women will invest less in acquiringschooling and skills than men and choose occupations that can accom-modate their entrance into and exit from the labor force. Thus the genderwage gap can be explained if women have lower levels of education thando men and are concentrated in lower paying occupations.
On the other hand, research in this area assumes that equally quali- fied workers will be paid at the same rate regardless of their sex or skincolor because they are similarly productive and of equal value to anemployer. This is a supply-side approach which explains wage differ-ences between women and men—and blacks and whites—as a productof individual-level skills or choices. Important for my analysis is thenotion that workers can increase their earnings by investing more intheir human capital. For example, Afro-Brazilians and women can ob-tain more schooling to better compete with white men. The appeal ofthis perspective is that we can measure differences in worker character-istics by sex and race and also measure the impact of these differenceson wages. Indeed, educational differences have been found to explainsizable shares of race and gender-based differences in pay (Corcoranand Duncan 1979; Kilbourne, England and Beron 1994; Farkas et al. 1997).
Table 1 presents the distribution of completed years of schooling for urban male and female workers by race. From 1960 to 2000, schoolcompletion rates increased for all workers in São Paulo. Gender differ-entials show that working women of both races have consistentlyachieved higher levels of education than men. The percentage of Afro-Brazilian women who had completed nine or more years of schoolingwas 2 percent in 1960 and rose to 37 percent in 2000. Similar increasesfor Afro-Brazilian men were from 1 to 29 percent. Likewise, the percent-age of white women who had completed nine or more years of school-ing increased from 18 to 61 percent over these four decades while a similarincrease for white men ranged from 1 to 50 percent. Despite women’sconsistent overall educational advantages, racial inequality in school-ing persisted for both sexes. By 2000, only 6 percent of employed Afro-Brazilian women and 4 percent of employed Afro-Brazilian men hadcompleted 12 or more years of schooling, compared to 23 percent and 18percent respectively for employed white women and men.
To summarize the racial differences in schooling, table 1 shows indi- ces of dissimilarity by race for each decade.8 This index represents the 8. The racial index of dissimilarity, or “segregation index,” developed by Duncan and Duncan (1955) is calculated as: 1/2 ∑ | w - b |, where w is the proportion of whites Table 1 Years of Completed Schooling for Employed Women and Men, Urban São Paulo, White Brazilian White Brazilian White Brazilian White Brazilian Sources: Public Use Samples from the 1960, 1980, 1991 and 2000 demographic censuses.
percent of Afro-Brazilians or whites who would have to change catego-ries in order for racial schooling distributions to be the same. Over time,the racial index of dissimilarity remains nearly unchanged for Afro-Brazilian women (25 percent in 1960 and 23 percent in 2000) while itincreases substantially for Afro-Brazilian men (from 6 percent in 1960 to20 percent in 2000). This suggests that the racial gap in education amongwomen has changed little over forty years while such differences haveincreased among men. Nevertheless, men are more similar in their dis-tribution among the schooling categories than are women.
Inequality in schooling between women and men is measured by the gender index of dissimilarity. This index represents the percent of womenor men within each racial group who would have to change categoriesin order for gender schooling distributions to be the same. In contrast to who are in category i (e.g., schooling or occupational category), and b is the proportion of Afro-Brazilians in that same category i. The gender index of dissimilarity is calculated inthe same way using the proportion of men and women in category i.
RACE, GENDER, AND WORK IN SÁO PAULO 71
Table 2 Occupational Distribution for Employed Women and Men, Urban São Paulo, White Brazilian White Brazilian White Brazilian White Brazilian Sources: Public Use Samples from the 1960, 1980, 1991 and 2000 demographic censuses.
Note: The 6 broad occupational categories used here follow a classification scheme usedby the Census Bureau. Specific census codes are available by request from the author.
racial inequality, gender inequality in schooling is much lower. For ex-ample, the gender index of dissimilarity for white women was 17 per-cent in 1960 and 11 percent in 2000. The same index for Afro-Brazilianwomen was 2 percent in 1960 and 8 percent in 2000. Thus the gender gapin schooling is narrowing somewhat for whites and widening for Afro-Brazilians. Unlike the racial index of dissimilarity, however, whereAfro-Brazilians were underrepresented in the highest schooling categories, the gender index of dissimilarity reflects both Afro-Brazilian and whitewomen’s higher levels of schooling relative to men.
These descriptive results begin to suggest the ways in which race and gender can simultaneously affect labor market outcomes. Mostimportant for an intersectional analysis is the finding that there is nota generic account of either race or sex-based differences in education.
Afro-Brazilian women workers had higher levels of completed school-ing than Afro-Brazilian men, yet racial inequality was higher amongwomen than men, largely due to Afro-Brazilian women’s under-representation in the highest levels of the education system. Racialobstacles are thus likely to be more acute than gender barriers for Afro-Brazilian women’s access to education. White women, in contrast, arethe most advantaged of all groups. Thus white women benefit fromtheir race and seem to confront no obstacle to schooling on the basis oftheir sex.
Occupational Segregation by Race and Sex Rates of completed schooling are important indicators determining access to better paying jobs. We know from numerous studies (Englandet al. 1996; Kilbourne et al. 1994) that the process of confining women tosome jobs (usually lower paying) and men to others (usually higher pay-ing) is one of the most important mechanisms for maintaining genderinequality in wages. There is evidence that occupational segregationbased on race plays a similar role in generating disparities in pay(Lieberson and Waters 1990). Some studies suggest that occupationalsegregation is a double disadvantage for minority women because theytend to be sorted into jobs that are segregated by both race and sex (Reskin1999). Given that Afro-Brazilian workers in São Paulo have lower levelsof education than whites and women workers have higher levels ofschooling than men, we can expect two possible outcomes consistentwith the human capital theory: occupations in São Paulo are likely to bestratified by race and, in the absence of gender bias, women should haveequal or greater access than men to more prestigious jobs. Table 2 pre-sents the occupational distribution of urban São Paulo workers for theyears 1960–2000. The racial and gender dissimilarity indices for eachyear of the census measure the proportion of either women/men or Afro-Brazilians/whites that would have to change occupations for the groupsto be distributed identically across occupations.
In 1960, Afro-Brazilian and white women’s occupations were distinct.
Afro-Brazilian women were almost exclusively employed (94.2 percent)in three blue collar occupations: manufacturing, service, and domesticwork. Notably, 66 percent of Afro-Brazilian women worked as domesticservants compared to 24 percent of white women. Four decades ago, Afro- RACE, GENDER, AND WORK IN SÁO PAULO 73
Brazilian women were virtually absent in the higher paying administra-tive, professional, and clerical occupations that together represented 37percent of white women’s employment. Women’s racial index of dissimi-larity for 1960 tells us that 42 percent of Afro-Brazilian or white womenwould have had to change occupations to eliminate segregation by race.
Over the next two to three decades, as São Paulo developed a highly diversified industrial economy, Afro-Brazilian women began to makesignificant inroads into more skilled occupations. For instance, the pro-portion of Afro-Brazilian women employed in the better paying “pink“-and white-collar occupations (clerical, professional, and administrativejobs) increased by 28 percent between 1960 and 1991. White women ex-perienced a similar 27 percent gain in the proportion employed in thehigher-status occupations. Correspondingly, the racial index of dissimi-larity among women declined from 35 percent in 1980 to 30 percent in1991, indicating an overall narrowing of the racial gap among womenin occupational distribution. Despite these gains, the percentage of Afro-Brazilian women employed as domestic servants remained at about 38percent in both 1980 and 1991.
By 2000, we begin to see some reversals of the above trends as women of both racial groups exit clerical positions and enter lower-paying ser-vice jobs. For instance, the proportion of white and Afro-Brazilian womenemployed in clerical jobs declined by 12 and 8 percent respectively from1991 to 2000. During this decade, the proportion of white women em-ployed in the service sector grew from 7 to 22 percent with a similarincrease from 12 to 28 percent for Afro-Brazilian women. Declines ineducational attainment clearly do not explain this exodus from clericaljobs, as women’s formal education increased during these years.
Women’s movement back into the traditional female service occupa-tions may instead be indicative of broader shifts in São Paulo’s indus-trial economy.9 Despite the steady decline in the racial index of dissimilarity among women over these forty years, one in three Afro-Brazilian women con-tinued to work as a domestic servant in 2000. At the opposite end of thejob hierarchy we find that in 2000 white women were twice as likely asAfro-Brazilian women to be employed in the highest paying adminis-trative and professional occupations. Consistent with the human capi-tal perspective and other studies of occupational segregation, these 9. Major shifts in the industrial structure of the U.S. economy since the 1970s had led to deindustrialization, increased demand for skills, and a decline in the availability oflow-skill jobs in the central cities. Many scholars suggest these structural shifts are anexplanation of the persistent race and ethnic inequality in the U.S. labor market. To myknowledge, no similar studies exist which examine the effects of recent structural changesin the economy on racial inequality in Brazil. This is an avenue for future investigation.
findings show that occupations are clearly segregated by race amongwomen. What is unknown is whether this stratification is a consequenceof Afro-Brazilian women’s lower levels of education or whether, aftercontrolling for educational differences, Afro-Brazilian women face race-based exclusion from the higher paying occupations.10 Nevertheless, itis clear that Afro-Brazilian women confront triple forces that push theirwages downward: segregation into lower-paying female-dominated po-sitions, segregation into occupations structured by human capital, andsegregation in occupations structured by race.
Among men the index of racial dissimilarity shows that differences in occupational sorting have always been narrower for men than forwomen. In 1960 the index of dissimilarity between Afro-Brazilian andwhite men was 19 percent, less than half of the index for women (42.2percent). Over time, however, the direction of the racial dissimilarityindex for men moved in the opposite direction than that for women.
While racial inequality in job sorting steadily declined for women, ra-cial inequality rates for men increased from 1960 (19 percent) to 1980 (24percent) and then declined to their lowest level in 2000 (17 percent).
Despite the variability in racial inequality among men, men’s occupa-tional structure changed significantly over time with a decline in manu-facturing employment and an increase in service sector and professionalemployment. These changes, however, disproportionately benefitedwhite men, who exited manufacturing at a much more rapid pace andentered administrative and professional occupations in much greaternumbers than Afro-Brazilian men. Nevertheless, throughout every de-cade, racial differences in the distribution of jobs remained much smalleramong men than among women.
The gender index of dissimilarity in table 2 measures the extent to which intraethnic women and men hold different occupations. The gender seg-regation index is substantially higher than the racial index, with highsranging from 65 percent in 1960 for Afro-Brazilian women to 40 percent in1991 for white women. Over time, sex segregation in occupations hasvaried by race. Sex segregation among whites increased from 1960 to 1991(from 31 to 40 percent) followed by a decline in 2000 (35 percent). In con-trast, gender inequality between Afro-Brazilians, although higher than thatfor whites, has steadily declined from 65 percent in 1960 to 43 percent in2000. Sex segregation for both Afro-Brazilian and white women is causedmainly by men’s overrepresentation in manufacturing and administra-tive jobs. In 2000, white and Afro-Brazilian men were four to five times as 10. To better understand how and why race and gender simultaneously affect occupa- tional segregation, we need to move beyond these descriptive techniques and insteaduse models such as logistic regression or multinomial logit analysis that allow us toseparate the effects of race, sex, and human capital on occupational distributions.
RACE, GENDER, AND WORK IN SÁO PAULO 75
likely to work in manufacturing and roughly twice as likely to hold ad-ministrative positions as intraethnic women. Contrary to the predictionsof human capital theory, women’s overall higher levels of education donot guarantee their access to these high-status jobs.
Although a more nuanced investigation of the causes of occupational segregation is beyond the scope of this paper, these findings show thatoccupational segregation in Brazil occurs along three interconnected axes:race, sex, and levels of human capital endowments. Overall, occupa-tions are more segregated by sex than by race. This is counterintuitive tothe human capital model which would predict that because of the sig-nificant lower levels of education among Afro-Brazilians, jobs shouldbe more segregated by race. Findings instead suggest that jobs in Brazilare stereotyped more on the basis of sex than skin color.
Brazil’s gender wage gap is among the highest in Latin America and the Caribbean (Winter 1994). In 2000, women workers throughout Bra-zil earned only 64 percent of men’s wages (IBGE 2003). Gender and racewage ratios for all urban São Paulo workers are show in table 3. Genderwage ratios are calculated by dividing women’s mean monthly wagesby men’s mean monthly wages. Race wage ratios are calculated by di-viding Afro-Brazilian mean monthly wages by white mean monthlywages within gender groups.
The gender wage gap has narrowed among urban workers in São Paulo over the past four decades, although it has varied by race. Thelargest change was for Afro-Brazilian women as their share of Afro-Bra-zilian men’s earnings rose from 48 percent in 1960 to 71 percent in 2000.
The decline in the gender wage gap was also significant for white women,although the pattern of change varied over time. In 1960, white womenearned 63 percent of white men’s wages, with their earnings falling to54 percent in 1980 and then increasing for the remaining decades. By2000, white women working in urban São Paulo earned 68 percent ofmen’s wages (4 percentage points higher than the national average). Forboth Afro-Brazilian and white women, the greatest gains were concen-trated in the post-1991 period. Yet, given that working women havehigher levels of education, their persistent lower salaries vis-à-vis mensuggests that women’s returns to wage related characteristics are lowerthan that of their male counterparts.
The racial wage gap, on the other hand, has remained remarkably consistent over these forty years. Among women, the racial wage gappersisted at roughly 60 percent from 1980 to 2000. During this same time,Afro-Brazilian men earned roughly 59 percent of white men’s wages.
The consistency of the racial wage gap for both women and men is explained, in part, by the persistent educational and occupational dis-parities between Afro-Brazilians and whites in the labor force in SãoPaulo. As previously shown, Afro-Brazilian women and men completedfewer years of schooling than whites and were disproportionately con-centrated in the lowest paying occupations. In unreported results, I foundthat both the gender and racial wage gap persisted even when educa-tion and occupation are taken into account, and indeed increased withmore years of schooling and higher status occupations.11 Overall, these findings suggest that over the past forty years, the gen- der wage gap has narrowed among workers in São Paulo (more so forAfro-Brazilians than whites), yet the racial wage gap has remained fairlyconstant. The convergence in the wages of intraethnic women and menis most likely due to women’s greater investments in education andmovement into higher status occupations, both of which were trans-lated into better average wages. What remains to be seen is what pro-portion of the gender and racial wage gap is due to unequal humancapital versus unequal pay.
EXPLAINING RACE AND GENDER DIFFERENCES IN PAY
In the burgeoning literature on labor market inequality in the United States, analysts have presented numerous hypothesis concerning per-sistent racial and gender differentials in earnings among similarly quali-fied workers. Important to my analysis is the research on discriminationthat shows that women and blacks earn less than white males, after sta-tistical adjustments are made for years of schooling, occupation, workexperience and other explanatory characteristics (Corcoran and Duncan1979; Farley 1984; England 1992).12 This observation is critical to the studyof labor market inequality because returns to workers’ characteristics 11. Tables that disaggregate wages by race, sex, occupation, and education are avail- 12. Discrimination may be the outcome of two additional and complementary pro- cesses: tastes and statistical discrimination. Many contemporary economists begin withBecker’s (1957) neoclassical taste model which posits that employers, workers or cus-tomers may have a taste for discrimination which results in a preference in favor of oragainst hiring, working with, or buying from women. Thus discriminators are willingto pay more to hire members of the preferred group. In contrast, models of statisticaldiscrimination assume that hiring decisions are made on the basis of group averagesrather than individual attributes. Lester Thurow says that statistical discrimination “oc-curs whenever an individual is judged on the basis of the average characteristics of thegroup or groups to which he or she belongs rather than upon his or her own character-istics” (1975, 172). Hence, assumptions that men (or whites) have higher average levelsof education or lower job turnover than women (or blacks), results in mistaken predic-tions about individual women or minorities and privileges males and whites in the la-bor market. Neither “tastes” nor judgments are easily tested with census data.
RACE, GENDER, AND WORK IN SÁO PAULO 77
Table 3 Sex and Race Wage Ratios, Workers Aged 18–64, Urban São Paulo, 1960–2000 Sources: Public Use Samples from the 1960, 1980, 1991 and 2000 demographiccensuses.
Note: Wage ratios are based on mean monthly wages for all workers between the agesof 18 and 64 except for students, members of the armed forces, unpaid workers, andworkers who received self-employment income. Sex wage ratios are calculated by di-viding women’s mean monthly wages by men’s mean monthly wages within racialgroups. Wage ratios by race are calculated by dividing Afro-Brazilian mean monthlywages by white mean monthly wages within gender groups.
are controlled by employers not employees. Thus, unequal returns toworkers’ endowments allow us to quantify the amount of the wage gapdue to either race- or sex-based discrimination.
Beginning with studies from the late 1970s, racial and gender wage discrimination in Brazil is well researched. Silva (1978) and Hasenbalg(1979) studied labor market inequality between Afro-Brazilian and whitemen and found that Afro-Brazilian men were concentrated in the lowesteconomic strata and those who attempted to climb the social ladder con-tinued to experience discrimination. Silva and Hasenbalg’s landmarkstudies clearly demonstrated that racial differences could not be reducedto class factors as researchers in the 1940s and 1950s contended, but thatdiscrimination was also at work.13 In 1980, discrimination accounted for24 percent of the wage gap between white and Afro-Brazilian men (Lovell1994). Studies of the gender wage gap in Brazil also find that a substan-tial part of the sex differential in pay is attributed to discrimination. AWorld Bank sponsored research program on Brazilian women’s payfound that, in 1989, discrimination accounted for 81–89 percent of theearnings differential between wives and husbands (Tiefenthaler 1992).
In a recent study of São Paulo (Lovell 2000), I investigated wage differ-entials between white women and men and found that by 1991, the pro-portion of the gender wage gap attributable to discrimination hadincreased to 99 percent. Despite major structural transformations inBrazil’s socioeconomic base, these studies show that discrimination inBrazilian workplaces is increasing.
13. See the work of Azevedo 1953; Pierson 1942; Wagley 1969; and Harris 1964.
In this study, I extend current research on discrimination in Brazil in two ways. First, I take advantage of four decades of data to determinewhether the social transformations taking place in Brazil during this pe-riod narrowed or widened racial and gender pay disparities. Second, incontrast to the emphasis in the literature that separates race and gender, Ifocus on how race and gender combine to ameliorate or intensify labormarket disadvantage. To test for discrimination by sex and race, I esti-mated separate wage regression equations for 18- to 64-year–old employedAfro-Brazilian and white women and men for the years 1960–2000. Thedependent variable is the log of monthly wages. The independent vari-ables included in the regressions are standard wage predictors: experi-ence, experience squared, years of completed schooling, occupation, maritalstatus, and hours worked.14 In light of my previous findings on schoolingdifferences and occupational segregation, it is important to note that botheducational and occupational differences are accounted for in this model.15I use the results from the estimated wage regressions as input into an econo-metric model following Jones and Kelly (1984). 16 This technique separatesthe wage gap into 3 parts: differences in individual-level characteristics(the “explained” or what is referred to as the “composition” portion ofthe wage gap); differences due to unequal pay (the “unexplained” or 14. Results of the regression are available from the author on request.
15. Other researchers have noted that inclusion of occupation in wage regression analy- sis may lead to results that underestimate the extent of wage discrimination. As Oaxaca(1973) argued, if jobs are assigned on the basis of sex and race, then including occupa-tion in a wage equation will in effect control for a portion of labor market discrimina-tion. We have every reason to believe that the determination of jobs is mediated by labormarket discrimination in Brazil, thus my estimates of wage discrimination are most likelyunderstated.
16. I use a modified version of the standard Oaxaca method proposed by Jones and (Yh – Yl) = [(αh – αl ) + ΣX l (β h – β l )] + Σβ l (Xh – Xl) + Σ(β h – β l) (Xh – Xl).
Part (a), “discrimination” is the amount due to the difference between the intercept of the white’s (male’s) and Afro-Brazilian’s (female’s), plus the difference in coefficients.
The substantive interpretation is how much of the income gap results from group mem-bership and how much results from differential returns on human resources. In otherwords, how much of the wage differential is a result of being Afro-Brazilian and receiv-ing less pay compared to an equally qualified white worker. This is a conservative mea-sure of discrimination because it includes only intercepts and slopes unlike Blinder (1973)that adds part (c) to this term, thus inflating the proportion of the wage gap attributedto discrimination. Part (b), “composition,” represents the amount of the wage gap thatis due to differences in human resources, such as different levels of education or jobexperience. It estimates the amount by which Afro-Brazilian’s average income is de-pressed because of human capital deficits (e.g., having lower levels of schooling). Part(c), “interaction,” represents the combination of both differential returns and differencesin human resources.
RACE, GENDER, AND WORK IN SÁO PAULO 79
discrimination portion of the wage gap); and the combined effect of com-position and discrimination (the interaction portion of the wage gap).
An important caveat is that the decomposition analysis attributes that portion of the wage gap not explained by gender or racial differences inindividual-level attributes (such as unequal levels of education) to “dis-crimination.” Discrimination is therefore the result of unmeasured dif-ferences. It is possible that there are other unidentified factors notincluded in the model that also contribute to this wage gap (e.g., qualityof schooling). Thus, the measure of discrimination obtained using theeconometric model is an estimated, not an actual, measure of discrimi-nation and should be taken as broadly representative of the extent ofwage discrimination. Despite the shortcomings of this measure, the dis-crimination component offers a robust measure of the relative wagesworkers are paid on the basis of their sex and race.
The results in table 4 decompose the wage gap between workers us- ing white men as the comparison group. The analyses conducted in eachof the four decades yield a remarkably consistent finding. After adjust-ing for education, occupation, hours worked, and other standard pre-dictors of wages, the main explanation for the racial and gender gap inwages is discrimination. In every decade since 1960, similarly qualifiedAfro-Brazilians and women are paid less than white men in urban SãoPaulo workplaces.
Differences in individual level qualifications (composition) explain only a small part of the wage gap between the groups. The portion ofthe wage gap due to compositional disadvantages has remained fairlysteady for Afro-Brazilian women and men over time. Since 1980, com-positional differences explain about 28 percent of the gap for Afro-Bra-zilian men and about 12–13 percent of the gap for Afro-Brazilian women.
As seen in the previous findings on education and occupation, despiteAfro-Brazilians absolute gains in the labor market, the relative gap be-tween blacks and whites has remained remarkably consistent over time.
The stability of the racial wage gap due to compositional differencesreflects this relative inequality. White women, in contrast, have outpacedall other groups in education and gained greater access to higher pay-ing occupations. As a result, the percentage of the wage gap explainedfor white women in terms of a male compositional advantage has beennegative since 1980.
For four decades, between 37 and 42 percent of the wage gap be- tween Afro-Brazilian and white men is explained by unequal pay. Amongwomen, wage discrimination has increased. In 1960, discrimination ac-counts for 53 percent of the wage gap between Afro-Brazilian womenand white men and 80 percent of white women’s differential pay. By2000 the proportion of the wage differential attributable to discrimina-tion increased to 63 percent for Afro-Brazilian women and 115 percent Table 4 Decomposition of Average Wage Differentials for Urban Workers by Race and Sex, São Paulo, 1960–2000 (Base Group = White Men) 1960 1980 1991 2000 1960 1980 1991 2000 1960 1980 1991 2000 32.19 28.84 27.74 27.65 23.17 12.03 11.68 13.01 42.22 37.32 39.85 41.37 52.76 45.09 51.61 63.02 79.67 90.71 99.80 114.77 25.58 33.84 32.41 31.22 24.07 42.88 36.71 23.97 12.35 12.73 6.70 -6.58 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% 100% Sources: Public Use Samples from the 1960, 1980, 1991 and 2000 demographic censuses.
for white women.17 These estimates show that there is a persistent andperhaps even increasing cost to being black and female in Brazil. De-spite absolute gains in education and jobs, Afro-Brazilians and womendo not receive equal rewards to their human capital in the labor market.
CONCLUSION
Labor market research remains one of the most insightful approaches to understanding social inequality in contemporary Latin America. Justas the pioneering studies of women and work challenged our percep-tions about the development process, studies of the ways in which raceand sex intersect challenge our understanding of the collective roles theseattributes play in stratifying societies. Women’s and black’s unequal treat-ment in the labor market influences many processes, including the dis-tribution of jobs, earnings, social status, and economic security. Theseemployment outcomes are central to creating and reducing social andeconomic inequality.
Afro-Brazilians and women have made remarkable progress over the past four decades in securing hard-won legal rights and in gaining ac-cess to the highest levels of schooling, entrance into higher paying occu-pations, and narrowing the intraethnic gender wage gap. Yet, womenand Afro-Brazilians are still paid less than similarly qualified white men.
17. The percentage of the wage gap attributed to discrimination for white women exceeds 100 percent in 2000 because both the compositional and interaction terms to-taled a negative 14.7 percent indicating that white women had higher levels of humancapital relative to white men.
RACE, GENDER, AND WORK IN SÁO PAULO 81
Discrimination by race and gender is not restricted to the poor. Evenamong those with the highest level of education, wage inequality exists.
Wage discrimination has remained fairly steady for Afro-Brazilian menand actually increased for Afro-Brazilian and white women working inurban São Paulo, one of Latin America’s most dynamic economies. Thispattern of absolute improvement but relative inequality has long char-acterized Brazil’s development.
Forty years of economic growth in São Paulo has neither erased the relative gap between whites and Afro-Brazilians nor has it favoredwomen’s equal pay. Racial and gender divides are not transitory as pre-dicted by early development theories. There is every reason to believethat the development process in Brazil may actually increase racial andgender inequality especially among those at the top part of the socialstructure. As Darity and Mason (1998) have suggested, discriminationis even more likely to occur when minority workers’ access to the higherpaying jobs cannot be denied on the basis of their observed productiveattributes. Studying the intersection of race and gender in the work-place shows how women and blacks are simultaneously integrated intothe economy but excluded through processes of discrimination.
Despite women’s and black’s shared disadvantages, the relative dif- ferentials between groups vary. Generic discussion of gender oftenconflates white and minority women as sharing a status relative to men.
This study finds that the predictors and outcomes of wage disadvan-tages facing Afro-Brazilian and white women are different. Likewise,generic discussion of race often conflates minority women and men assharing a status relative to whites. I find that the barriers to racial equal-ity are not the same for women and men. The two explanations of earn-ings inequality that I examined, changes in workers’ human capitalinvestments and changes in the returns to human capital, offer theoreti-cally competing explanations for the persistence of racial and genderearnings inequality among workers and a window onto the race/gen-der nexus.
Consistent with the human capital perspective, I found substantial although declining differences in workers’ wage-related characteristicsover the past four decades Unequal levels of education and occupationalsegregation are important in explaining racial wage gaps in urban SãoPaulo, but such racial differences explain a higher portion of inequalityin pay for men than for women. Consistent with the discriminatory paystructure explanation, I found statistically significant differences in thewages paid to equally qualified black and white workers. Yet, racial dis-crimination by employers varies by sex. Thus, a generic account of race-based differences does not accurately describe the situation of eitherAfro-Brazilian women or men, in part as a result of the higher level ofdiscrimination that Afro-Brazilian women confront.
As for the gender gap, human capital differences were irrelevant in explaining sex-based differences in pay for white women and have hada declining effect on Afro-Brazilian women’s wages since 1960. Employ-ers penalize women of both races, independent of their human capitalcharacteristics, by paying lower rates of return for women’s work-re-lated characteristics relative to white men. Surprisingly, white womenwere the most discriminated against of all groups. Since 1980 discrimi-nation has accounted for the entire wage gap between white womenand men. Sex discrimination is therefore unequal in magnitude withineach racial group. Thus, a generic account of gender differences in paycannot accurately explain the situation of either Afro-Brazilian or whitewomen. These findings make clear that analyses of labor market inequal-ity and polices designed to address such inequality must consider howgender and race jointly determine labor market outcomes.
The scarcity of statistical data on race and ethnicity is one explanation for the limited information on race in the literature on women and workin Latin America. In 2000 only 12 Latin American nations gathered censusdata on race and ethnicity (Buvinic 2003). Early feminist research on LatinAmerica that challenged traditional assumptions about the sexual divi-sion of labor was also concerned with the scarcity of data on women. Thesescholars and activists raised questions that directly resulted in the improvedmeasurement and collection of data on women’s work (Aguiar 1986). Justas feminists lobbied for additional and better data on women, so too areblack activists, scholars and policy makers throughout Latin America andthe Caribbean calling for additional national level data on race and ethnicity.
The United Nations and the Inter-American Development Bank convenedan international conference in 2000 to discuss the need for racial data inLatin American censuses (Andrews 2004). Improving the quality of infor-mation on race and ethnicity is a basic step governments can take to mea-sure and counter social exclusion.
Census data, however, tell us nothing about the concrete practices by which Afro-Brazilians and women are excluded from jobs and higherwages. Neither do census data give priority to social agency. By empha-sizing process (rather than outcomes) and human agency, participatoryresearch and qualitative approaches are also needed to further our under-standing of racial and gender intersections in the workplace. Still morecan be learned from additional empirical studies that focus on shifts in theindustrial structure of the Brazilian economy as a source of racial and gen-der inequality. Despite the need for additional studies, the findings fromthis research have important implications for policy makers and activistsworking to improve the lives of women and Afro-Brazilians.
Discrimination in the labor market is a shared experience that should unite feminists and black activists in their struggle for social justice.
RACE, GENDER, AND WORK IN SÁO PAULO 83
The discrimination that white middle class women confront demon-strates that inequality is not solely a product of race and class in Bra-zil. Rather, labor market discrimination cuts across gender, race, andclass boundaries. Hence, gender and racial activism need not be in-compatible. For purposes of social equity, black activists, feminists,and the state together must address wage discrimination. The Brazil-ian government’s endorsement of antidiscrimination laws in 1988 andaffirmative action policies in 2001 suggests that arguments about theconnections between race, gender, equality, and democracy are salientand persuasive.
There remains, however, a deep chasm between the federal antidis- crimination laws passed in 1988 and their effective enforcement. It isunlikely that wage discrimination will disappear without additionalgovernmental and collective intervention. Activists must rally to ensurethat employers adhere to labor discrimination laws by strengtheningcurrent enforcement mechanisms and identifying necessary judicial re-forms to encourage workers to seek reparation if their labor rights areviolated. A further priority is to examine the potential for existing lawsto actually limit women’s employment opportunities. For example,maternity protection laws and some child care provision laws may workto raise the costs of female labor relative to male labor. These factorssuggest that activists and policy makers need to review policies andlaws currently in place so that the benefits accorded to workers, are, asfar as possible, gender and race neutral. Strong, carefully conceived, andwell-implemented policies and laws can be effective in reducing racialand gender labor market discrimination. Such laws will have little ef-fect, however, if their target population has limited knowledge of them.
Women and Afro-Brazilians must be informed about labor rights. To-gether, black activists and feminist groups could assist in promotingworker education about labor laws.
Overall interventions that focus on raising educational attainment and access to better paying jobs of the entire work force, Afro-Braziliansin particular, should alleviate some racial and gender differences in la-bor market inequality. Nevertheless, this analysis shows that Brazil’sexemplary affirmative action legislation designed to increase Afro-Brazilian’s access to education and jobs will not alone eliminate racialwage differences. Women have surpassed men in educational levels, butthis has not assured either white or Afro-Brazilian women access to bet-ter wages. The best intervention for the elimination of gender and raciallabor market inequality is to end discrimination. Unequal pay on thebasis of sex and skin color has far-reaching economic and social costs,which both impede the development of Brazil and ensure continuedracial and gender stratification.
“Research Guidelines: How to Study Women’s Work in Latin America.” InWomen and Change in Latin America, edited by June Nash and Helen Safa, 22–33.
South Hadley, MA: Bergin & Garvey.
Engendering Democracy in Brazil: Women’s Movements in Transition Politics.
Princeton: Princeton University Press.
“The (Trans)formation of Feminism(s) and Gender Politics in DemocratizingBrazil.” In The Women’s Movement in Latin America: Participation and Democracy,edited by Jane S. Jaquette, 13–63. Boulder: Westview Press.
“Racial Inequality in Brazil and the United States: A Statistical Comparison.”Journal of Social History 26 (2): 229–63.
Afro-Latin America, 1800–2000. New York: Oxford University Press.
As elites de côr numa cidade brasileira: Um estudo de ascensão social. São Paulo:Companhia Editora Nacional.
The Economics of Discrimination. Chicago: University of Chicago Press.
“Wage Discrimination: Reduced Form and Structural Equations.” Journal ofHuman Resources 8 (4): 436–455.
Blumberg, Rae Lesser, Cathy A. Rakowski, Irene Tinker and Michael Monteón Engendering Wealth and Well-Being: Empowerment for Global Change. Boulder:Westview Press.
Bose, Christine E., and Edna Acosta-Belén Women in the Latin American Development Process. Philadelphia: Temple Univer-sity Press.
Latinas and African American Women at Work: Race, Gender and Economic Inequal-ity. New York: Russell Sage Foundation.
Bruschini, Cristina, and Maria Rosa Lombardi “Instruídas e trabalhadeiras: Trabalho feminino no final do século XX.” CadernosPagu 17 (18): 157–196.
“The Lost Constituency of Brazil’s Black Movements.” Latin American Perspec-tives 98 (25): 136–55.
“Social Inclusion in Latin America and the Caribbean: Experience and Lessons.”Paper presented at the seminar Good Practices in Social Inclusion: A Dialoguebetween Europe and Latin America and the Caribbean. March 21–22, Milan, Italy.
Black Feminist Thought: Knowledge, Consciousness and the Politics of Empowerment.
Boston: Unwin Hyman.
“Work Experience, Labor Force Attachment and Earnings Differences Betweenthe Races and Sexes. Journal of Human Resources 14 (1): 3–20.
“Evidence on Discrimination in Employment: Codes of Color: Codes of Gen-der.” Journal of Economic Perspectives 12 (2): 63–90.
De Carvalho, José Alberto Magno, Charles H. Wood, and Flávia Cristina Drumond Adrade “Estimating the Stability of Census-Based Racial/Ethnic Classifications: The Caseof Brazil.” Population Studies 58 (3): 331–44.
“Residential Distribution and Occupational Stratifications.” American Journal ofSociology 60 (5): 493–503.
RACE, GENDER, AND WORK IN SÁO PAULO 85
Comparable Worth: Theories and Evidence. New York: Aldine.
England, Paula, Karen Christopher, and Lori L. Reid “Gender, Race, Ethnicity and Wages.” In Latinas and African American Women atWork: Race, Gender and Economic Inequality, edited by Irene Browne, 139–82. NewYork: Russell Sage Foundation.
England, Paula, Lori Reid, and Barbara S. Kilbourne “The Effect of the Sex Composition of Jobs on Starting Wages in an Organiza-tion: Findings from the NLSY.” Demography 33 (4): 511–21.
Farkas, George, Paula England, Keven Vicknair, and Barbara S. Kilbourne “Cognitive Skill, Skill Demands of Jobs, and Earnings Among Young European-American, African America, and Mexican-American Workers.” Social Forces 73(3): 913–40.
Blacks and Whites: Narrowing the Gap? Cambridge, MA: Harvard University Press.
“Os desafios da eqüidade: Reestruturação e desigualdades de gênero e raça noBrazil.” Cadernos Pagu 17 (18): 237–266.
Orpheus and Power: The Movimento Negro of Rio de Janeiro and São Paulo, Brazil,1945–1988. Princeton: Princeton University Press.
Patterns of Race in the Americas. New York: Walker and Company.
Race Relations in Post-Abolition Brazil: The Smooth Preservation of Racial Inequali-ties. PhD diss., University of California, Berkeley.
Hasenbalg, Carlos, and Nelson Do Valle Silva Relações raciais no Brasil contemporâneo. Rio de Janeiro: Rio Fundo Editora.
Hite, Amy Bellone, and Jocelyn S. Viterna “Gendering Class in Latin America: How Women Effect and Experience Changein the Class Structure.” Latin American Research Review 40 (2): 50–82.
Feminist Theory: From Margin to Center. Boston: South End Press.
“Puzzles of Women’s Rights in Brazil.” Social Research 69 (3): 733–751.
“From ‘Racial Democracy’ to Affirmative Action: Changing State Policy on Racein Brazil.” Latin American Research Review 39 (1): 60–89.
Censo Demográfico 2000. Rio de Janeiro: IBGE “Women and Modernization Theory: A Decade of Feminist Criticism.” WorldPolitics 34 (2): 267–84.
“Decomposing Differences between Groups: A Cautionary Note on MeasuringDiscrimination.” Sociological Methods and Research 12 (3): 323–43.
Kilbourne, Barbara S., Paula England, and Kurt Beron “Effects of Individual and Occupational Characteristics on Earnings: Intersec-tions of Race and Gender.” Social Forces 72 (4): 1149–76.
Kilbourne, Barbara S., Paula England, Georege Farkas, Kurt Beron, and Dorothea Weir “Returns to Skills, Compensating Differentials and Gender Bias: Effects of Oc-cupational Characteristics on the Wages of Women and Men.” American Journalof Sociology 100 (3): 689–719.
From Many Strand: Ethnic and Racial Groups in Contemporary America. New York:Russell Sage Foundation.
“Race, Gender and Development in Brazil.” Latin American Research Review 29(3): 7–35.
“Gender, Race and the Struggle for Social Justice in Brazil.” Latin American Per-spectives 27 (6): 85–103.
“Skin Color, Racial Identity and Life Chances in Brazil.” Latin American Perspec-tives 25 (3): 90–109.
Catálogo da abolição. Rio de Janeiro: Federal University of Rio de Janiero.
“Family Investments in Human Capital: Earnings of Women.” Journal of HumanResources 13 (2): 118–34.
“Blacks and the Abertura Democrática.” In Race, Class and Power in Brazil, ed-ited by Pierre-Michel Fontaine, 95–119. Los Angeles: Center for Afro-AmericanStudies, University of California.
Mohanty, Chandra Talpade, Ann Russo, and Lourdes Torres, eds.
Third World Women and the Politics of Feminism. Bloomington: Indian UniversityPress.
Nash, June, and Maria Patricia Fernández-Kelly, eds.
Women, Men and the International Division of Labor. Albany: SUNY Press.
Sex and Class in Latin America. New York: Praeger Publishers.
“Male-Female Wage Differential in Urban Labour Markets.” International Eco-nomic Review 14 (3): 693–709.
Oliveira, Ana Maria H. C., and Ana Flávia Machado “Mobilidade ocupacional e rendimentos no Brasil metropolitano: 1991/96.”Pesquisa e Planejamento Econômia 30 (1): 117–156.
Ometto, Ana Maria Holland, Rodolfo Hoffmann, and Marcelo Corrêa Alves “Participação da mulher no mercado de trabalho: Discriminação em Pernambucoe São Paulo.” Revista Brasileira de Economia 53 (3): 287–322.
Negroes in Brazil: A Study of Race Contact at Bahia. Chicago: University of Chi-cago Press.
Race in Contemporary Brazil. University Park: The Pennsylvania University Press.
“Occupational Segregation by Race and Ethnicity Among Women Workers.” InLatinas and African American Women at Work: Race, Gender and Economic Inequal-ity, edited by Irene Browne, 183–204. New York: Russell Sage Foundation.
“Now You Seem ‘Em, Now You Don’t: Race, Ethnicity and Gender in LaborMarket Research.” In Brown, Latinas and African American Women at Work, 380–407. New York: Russell Sage Foundation.
“Challenging Social Exclusion: Afro-Descendent Women in Latin America andthe Caribbean.” Unpublished paper prepared for the Inter-American Develop-ment Bank, Washington, D.C.
Women in Class Society. New York: Monthly Review Press.
Scorzafave, Luiz G., and Naércio A. Menezes-Filho “Participação feminina no mercado de trabalho brasileiro: Evolução edeterminantes.” Pesquisa e Planejamento Econômico 31 (3): 441–477.
Development, Crises and Alternative Visions: Third World Women’s Perspectives. NewYork: Monthly Review Press.
“White-Nonwhite Income Differentials: Brazil.” PhD diss. University of Michi-gan.
RACE, GENDER, AND WORK IN SÁO PAULO 87
“Updating the Cost of Not Being White in Brazil.” In Race, Class and Power inBrazil, edited by Pierre-Michel Fontaine, 42–55. Los Angeles: Center for Afro-American Studies, University of California.
“Cor e o processo de realização socio-econômico.” In Estrutura social, mobilidadee raça, edited by Carlos Hasenbalg and Nelson do Valle Silva, 144–63. Rio deJaneiro: Instituto Universitário de Pesquisas do Rio de Janeiro.
“A participacão feminine no mercado de trabalho.” IPEA Working Paper No.923:Rio de Janeiro.
“Women in the Struggle for Democracy and Equal Rights in Brazil.” In Womenand Politics Worldwide, edited by Barbara J. Nelson and Naoma Chowdhury,127–141. New Haven: Yale University Press.
Race in Another America: The Significance of Skin Color in Brazil. Princeton: PrincetonUniversity Press.
“Female Labor Force Participation and Wage Determination in Brazil, 1989.” InCase Studies in Women’s Employment and Pay in Latin America, edited by GeorgePsacharopoulos and Z. Tzannatos, 89–118. Washington: World Bank.
Generating Inequality. New York: Basic Books.
“The Making of a Field: Advocates, Practitioners and Scholars.” In The Women,Gender & Development Reader, edited by Nalini Visvanathan, Lynn Dugan, LaurieNisonoff and Nan Wiegersma, 33–42. London: Zed Books.
“Introduction to Part 1.” In The Women, Gender & Development Reader, edited byNalini Visvanathan, Lynn Dugan, Laurie Nisonoff and Nan Wiegersma, 17–32.
London: Zed Books.
“ From Cast to Class in North Brazil.” In Comparative Perspectives in Race Rela-tions, edited by Melvin Tumin, 47–62. Boston: Little, Brown.
“Working Women in Latin America: Participation, Pay and Public Policy.” Wash-ington, D.C: The World Bank.

Source: http://lasa-4.univ.pitt.edu/LARR/prot/fulltext/vol41no3/lovell.pdf