Human Resources Development and Operations Policy HRO Working Papers The Costs of Discrimination in Latin America Harry Anthony Patrinos November 1994 HROWP 45 Papers in this series are rnt formal publications of the World Bank. Tlhey present preliminary and unpolished results of analysis duat are circulated to encourage discussion and commnent; citation and the use of such a paper should take account of its provisional character. The ftndings. interpretations, and conclusions expressed in this paper arc entirely those of the author(s) and should not be attributed in any nunner tO the World Bank, to its affiliated organizations, or to members of its Board of Executive Directors or the countries they represent. The Costs of Discrimination in Latin America by Harry Anthony Patrinos The author thanks Shelton Davis, Jorge Uquillas, George Psacharopoulos and Zafiris Tzannatos for their useful comments. Abstract Indigenous, ethnic, racial and linguistic minorities worldwide are in an inferior economic and social position vis-a-vis the "mainstream" population. The ethnic concentration of poverty and inequality is increasingly being recognized in the development literature. In this paper, studies from six Latin American countries that estimate the costs to an individual of being an economic minority are reviewed. The studies decompose the overall earnings gap into two components: one is the portion attributable to differences in the endowments of incorre- generating characteristics ("explained" differences); the other portion is attributable to differences in the returns that majority and minority workers receive for the same endowment of income-generating characteristics ("unexplained"). This latter component is often taken as reflecting the "upper bound" of wage discrimination. In economic terrns, discrimination refers to differences in economic outcomes between groups that cannot be accounted for by the skills and productive characteristics of these groups. The "upper bound" of discrimination gives an indication of the "cost" of being a minority. In two studies for Bolivia, one using a 1966 rural survey and the other a 1989 urban survey, decomposition of the differential between indigenous and nonindigenous earnings leads to the conclusion that most, if not all, of the overall differential is due to productivity and not discrimination. In other words, equalizing productive characteristics between ethnic groups will drastically reduce, if not eliminate, the entire earnings differential. However, in Guatemala, Mexico and Peru, only one-half of the earnings differential can be attributed to differences in productivity-enhancing characteristics. In other words, if indigenous workers were endowed with the same amounts of productive characteristics as are nonindigenous workers, the difference in earnings between them would narrow by about 50 percent. In Paraguay, decomposition of the overall earnings differential between monolingual Spanish speakers and Guarani speakers shows that most (80 percent) of the overall differential is explainred by human capital diffe ences. Thus, equality in educational attainment will go a long way towards equalizing labor market outcomes. In Brazil, however, there iL a significant cost to 'being non-white" irrespective of being mulwiro or black. Moreover, discrimination against non-whites appears to be increasing over time, despitc reductions in the schooling gap between whites and non-whites. Nevertheless, in most cases much of the earnings disadvantage of minority group workers is due to lower human capital endowments. While the monetary benefits of schooling are lower for minority group populations, an increase in schooling levels would lead to a significant increase in earnings. The socioeconomic condition of minority groups can be improved because policy- influenced variables such as education and occupation are largely responsible for earnings differences. This provides considerable hope for the future. The question that remains, however, is how to improve the productive capabilities of the indigenous population. Given the fact that indigerous and minority group workers have fewer years of schooling vis-i-vis nonindigenous or majority group workers, it is necessary to make an effort to increase their human capital. Contents Introduction .......................... I Poverty and Inequality ...... ................... . I Decomposing Earnings Differentials.2. . ..................... 2 Individual Costs of Discrinnation .......................... 10 D iscussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13 References . . . . . . . . . . . .. .. .. . .. . .. . . . . . . . . ............ .... . 15 Introduction Indigenous, ethnic, racial and linguistic minorities worldwide are in an inferior economic and social position vis-a-vis the 'mainstream" population. The ethnic concentration of poverty and inequality is increasingly being recognized in the development literature (see, for example, Birdsall and Sabot, 1991; Klitgaard, 1992). Empirical analyses of ethnic earnings differentials concentrated in the past on Black-white differentials in the United States. While some have examined the experiences of other ethnic groups, very little investigation has been made of the differential economic experiences of indigenous people and other minority groups within a society. This particular literature is growing, however. Recent research documents extreme conditions of poverty among the indigenous population in some Latin American countries (Psacharopoulos and Patrinos, 1994; IFAD, 1992). Only a few studies examine these issues in Latin America. Exceptions include Kelley's (1988) analysis of the 'cost of being Indian" in rural Bolivia, three studies on racial discrimination in Brazil (Webster and Dwyer, 1988; Silva, 1992; 1985), the four-country study of indigenous people in Latin America (Psacharopoulos and Patrinos, 1994), and an examination of language-earnings differentials in Paraguay (Patrinos, Velez and Psacharopou.os, 1993). The objective of this paper is to review estimates of the "costs" to an individual of being an economic minority. That is, some workers are paid rnore than others with the same endowment of productive economic characteristics by virtue of some non-economic personal characteristic (such as ethnicity, race, sex, class, caste, language). Studies in the United States have attempted to estimate this cost for females, Blacks and indigenous people. In general, for the same amount of productive characteristics, an economic minority receives lower returns to each productivity-enhancing characteristic such as education, training aiid experience. The residual is then seen as a cost to the individual. In this paper, the available studies are summarized and an attempt is made to build on the literature. Poverty and Inequality Poverty among Latin America's indigenous population is pervasive and severe (see Table 1). In Bolivia, while more than half of the total population is poor, over two-thirds of the bilingual indigenous population and almost three-quarters of the monolingual indigenous population is poor. The majority (66 percent) of the population of Guatemala is poor. The indigenous population, however, is disproportiornately poor; 87 percent of all indigenous households are below the poverty line. In Mexico, individuals in more indigenous municipios are in poorer socioeconomic condition than are individuals in less indigenous muu2icipios. A positive correlation exists between municipio indigenous concentration and the incidence of poverty; Municipios of increasing indigenous concentration experience higher percentages of poverty and extreme poverty. Most of the indigenous population of Peru is poor, at 79 percent. In fact, indigenous people are one and a half times as likely to be poor than are nonindigenous people. Conseqlently, indigenous people account for 11 percent of the sample population, yet they comprise 19 percent of poor Peruvians. In Brazil, the black and mulatto populations are overrepresented in the bottom 10 and 20 percent of the per capita income distribution 2 (Psacharopoulos et al., 1993). Paraguay, the most bilingual nation in Latin America, has a large Guarani-speaking population. While most Guarani-speakers also speak Spanish, the relatively small group of monolingual Guarani speakers have a considerably higher rate of poverty. While 11 percent of Spanish-only speakers are poor, 24 and 37 percent of bilingual (Spanish-Guarani speakers) and monolingual Guarani speakers are poor (World Bank, 1993). Table 1: Poverty in Latin America (percent of population below povertv line) All Indigenous Nonindigenous Urban Bolivia 52.6 64.3 48.1 Guatemala 65.6 86.6 53.9 Mexico 22.6 80.6 17.9 Peru 53.0 79.0 49.7 Paraguay- 20.5 36.8 10.8 Source: Psacharopoulos and Patrinos, 1994; World Bank, 1993. Note: * Refers to Guararii and non-Guarani speakers. Decomposimg Earnings Differentials For earnings differentials, the use of multivariate regression analysis allows for the simulation of alternative outcomes and the decomposition of gross differentials. The decomposition method, the technique used for an,alyzing earnings differentials, was popularized in the economics literature by Oaxaca (1973) and Blinder (1973). It was used earlier in sociology (Siegel, 1965; Duncan, 1968), and before that in demography (Kitagawa, 1955). Although in the economics literature it was first used to analyze the determinants of male/female earnings differentials, the decomposition technique has been used since to analyze ethnic earnings differentials, public/private sector earnings differentials, earnings differentials by socioeconomic background, to test the screening hypothesis, and to test the effectiveness of a job training program. Most analyses have focused on developed countries, although some studies for developing nations exist (Psacharopoulos and Tzannatos, 1992; Birdsall and Sabot, 1991). 3 The standard procedure for analyzing the determinants of earnings differentials between two groups is to fit the following two equations, or eamings functions, for employed members of the economically dominant group and employed members of the marginal group: LnY,,=b,,X,,+u (I) L_nY=b,X1 .u, (2) where subscripts n and i represent nonindigenous (majority) and indigenous (minority) workers; Y symbolizes labor market earnings; X represents measured productivity-determining characteristics of the workers, such as education, experience and other control variables. The regression coefficient, b, reflects the returns that the market yields to a unit change in characteristics such as education and experience. The error term, u, reflects measurement error, as well as the effect of factors unmeasured or unobserved by the researcher. It is known that the regression lines pass through the mean values of the variables so that, (3) Ln Y. ,X (4) where hats (^) denote estimated values and bars (-) represent mean values. If indigenous (minority) workers received the same returns as do nonindigenous (majority) workers for their endowments of wage-determining characteristics, then their average earnings would be: (5) which are the average eamnings of indigenous (minority) workers that would prevail in the absence of wage discrirnination. Subtracting Equation 5 from 3 gives the difference between average nonindigenous (majority) earnings and the average hypothetical indigenous (minority) 4 earnings that would prevail if indigenous (minority) workers were paid according to the pay structure faced by nonindigenous (majority) workers. This difference reflects their unequal endowments of income-generating characteristics, so that: ilzzY"-ilnY -iX^-6"Xi=^^(XX - ) -(6) LnY _~n -Ln6 ~X~-&s =n(X, 4X) Subtracting Equation 4 from 5 yields the difference between the hypothetical nondiscriminatory earnings of indigenous (minority) workers and their actual earnings. This difference reflects the different returns to the same income-generatin, characteristics: aYi -itwY*=^,,Xi-,Xd =X (6 -~~ (7) Adding Equations 6 and 7 yields: LAY,-Li,=b (X -X) +X4(- (-68) Thus, the overall earnings gap can be decomposed into two components: one is the portion attributable to differences in the endowments of income generating characteristics (X,,-9) evaluated with the nonindigenous (majority) worker pay structure (b,.); the other portion is attributable to differences in the returns (b,,-b,) that nonindigenous (majority) and indigenous (minority) workers receive for the same endowment of income-generating characteristics ). This latter component is often taken as reflecting wage discrimination. In economic terms, discrimination refers to differences in economic outcomes between groups that cannot be accounted for by the skills and productive charactzristics of these groups (T.P. Schultz, 1991). This method, although illuminating since it allows one to deterrnine the extent of discrimination in the labor market, does not allow one to determine the origins of discrimination. Direct discrimination in the labor market can affect earnings, occupational attainment and training access; it can also be indirect, through discrimination in the acquisition of skills (schooling), prior to entering the labor market (Chiswick, 1987). The use of earnings functions to estimate discrimination means that there will always be a problem of omitted variables. This type of data problem means that the "unexplained" component is not only a measure of discrimination but also of our ignorance (Filer, 1983). It is because of omitted and unobserved factors that the "unexplained' component is seen as an Mupper bound' estimate of wage discrimination in the labor market. Included among the omitted 5 variables that are expected to account for some of the 'unexplained" component are: the quality of labor; attachment to the labor force; lack of specific trailing; interrupted work careers and tastes and personality. There is also evidence to suggest that much of the discrimination ag^anst the m'n:.ority group is due to occupational segregation, that is, the 'crowding' of the minority group into certain occupations where rates of pay and chances for promotion are low. This, of course, suggests that prior discrimination has taken place, such as lack of access to jobs, training and schooling. The results of a number of studies has shown that the greater the number of variat' s used to control for differences in productivity-related factor-., the smaller the productivity- adjusted earnings gap ("unexplained' component) relative to the unadjusted gap. Even when ar. extensive list of control variables is used, however, most studies find some residual gap th? they attribute to discrimination. Wh,tien the gap is close to zero, this usually results from the inclusion of control variables whose values themselves mry reflect prior discrimination (Gunderson, 1989' While many countries in Latin America have sizeable indigenous and ethr.cally and racially diverse populations, few include questions to identify the ethnolinguistic characteristics of individuals in their household or labor force surveys. Definitions of indigeno'i: reople differ from country to country due to the use of different survey instruments. Given ivailable data, three different variables are used to identify minority group membership: language spoken, self- perception and geographic concentration (see Table 2). Language defies. the ethnic minority population in Bolivia, Peru and Paraguay. In Bolivia and Paraguay, it is possible to distinguish between monolingual and bilingual (Spanish and minority language) individuals, while in Peru only monolingual indigenous or Spanish speakers can be isolated. ithe Guatemalan study uses the self-perception method of defuiing the indigeious population. The geographic location or concentration of the indigenous population is generally used when the indigenous population is concentrated in specific territories, and in combination with questions dealing with self- perception or language identity. This method is used in order to include Mexico, a country with a large number of indigenous people. 6 Table 2: Defining Ethnic, Racial and Linguistic Group Membership in Latin America Individual's ethnicity/race/ Perceni of Total Country language determined by: Source Population Urban Survey question: iQu, Encuesta Integrada de Indigenous: 29 Bolivia idioma(s) habla Hogares (EIH) conducted habitualmente? by the Insiuto Nacional de Estadisaic,' 9R9 Rural Survey questions on origin Head of Household Indigenous: 62 Bolivia Survey conducted by V. Rubin, L. Comitas and W.J. McEwen for the F.search Institute for the Study of Man, 1966 Guatemala Survey question: ZE's Encuesta Nacional Socio- Indigenous: 36 indfgena? Demografica (ENSD) conducted by the Instiruo Nacional de Escadisaica, 1989 Peru Survey questions: iCudI es Living Standards IWdigenous: 11 la lengua materna? and Measurement Study, iQue idioma habla? conducted by Insiruto Cuanto for the World Bank, 1991 Mexico Geographical concentration Encuesta Nacional de Indigenous: 8 of indigenous people at the Ingreso - Gasto de los municipio level based on the Hogares (ENIGH) 1990 Census. The condlucted by the Instinao percentage of indigenous de Estaistica Geografia e language speakers in each Informatica, 1989 mnunicipio was matched with the municipios included in the 1989 survey. Paraguay Survey questions on Encuesta de Hogares Guaranf-Spar:s~- 53 language (Mano de Obra), 1990 Guaranf only: 4 Brazil Color: white, Black, ye'low, Brazilian Census, 1960; 1976: Blacks: mulatto National Household MuiatOs: 3Y Survey (PNAD), !976, 1988: Blacks: 7 1988 MuLaos: 36 7 Several hypotheses regarding the role of ethnicity in society exist. The significant and positive relationship between education and earnings is well documented in the literature. The usual explanation put forward, consistent with the human capital approach, is that schooling contributes to individual productivity which, in turn, lead- to higher individual earnings. The eamnings advantage of the more educated relative to the less educated is subject to the laws of supply and demand; as the numbers of the more educated increase relative to the less educated, their earnings advantage declines and the minimum qualifications for given jobs rises in line with increased relative supplies (T.W. Schultz, 1961; Mincer, 1974; Becker, 1975). Institutional hypotheses emphasize the centrality of the functioning of labor markets (Doeringer and Piore, 1972). In this tradition, labor market segmentation theories view the labor market as being divided between the primary (high productivity, high wage) and the second&ry (low productivity, low wage) sectors. Individuals in the secondary labor market are locked into that sector and barriers exist to their moving into the high wage, high productivity, primary labor market (Carnoy, 1980). The hypothesis here is that minority group members receive lower earnings and have a higher incidence of poverty because they are locked into the secondary sector of the economy. Theories of internal colonialism postulate that the conditions of colonialisrii can exist within a nation-state when one group dominates a previously independent nation within its borders (Welch, 1988). In such a case, a dual economy, with a dual wage and labor market, is in place. Also present are the conditions of "unfree' labor, a dual occupational structure and dual wage scales, with the more rewarding occupations reserved for the minoritv -opulation. Indigenous and ethnic minority populations often play the role of a reserve labor force. Poverty, a lower standard of living, lower expectations and a lack of knowledge of labor laws are just some of the reasons why the indigenous or minority group labor force may agree to sell its labor cheaply. Also, in many cases, indigenous worke:s wish to return to their families and homes and may be willing to tolerate Jiscrimination and low wages in order to facilitate their return (sirnilar to "target" workers theory). The "screening" hypothesis states that, in general, employers pay higher salaries to the more educated because they use schooling level as a proxy for other characteristics that "signal' which individuals could be more productive. Thus, it is not the content of their education that makes individuals more productive, but rathier that years of schooling demonstrate to employers which potential employees are more produictive since the more able will attain higher levels of schooling (Arrow, 1973). The hypothesis here is that indigenous people receive !ower ?.,nings because they are screened on the basis of their schooling, which reflects their ethnolinguistic characteristics. Other explanations of outcornes are concerned with the productivity of schooling. That is, for the same level of schooling and the same level of ability, different outcomes can result due to the application of "skills iui the labor market. Individual skills may be developed both in and out of school. Group variations in rates of return to schooling arise from differences in the ability to convert the schooling process into earnings (Chiswick, 1988). This may be a 8 consequence of parental investments in the home-produced components of child q- .lity, although one can think of many other reasons. It would appear that members of more successful ethnic groups had parents with higher levels of schooling, fewer siblings to compete with for parental time and other family resources, and had mothers who were less likely to work when young children were in the household (Chiswick, 1988). A positive relationship between educational attainments across generations reflects the intergenerational transmission of human wealth. In the case of indigenous people, if their parents had low levels of schooling and other forms of human capital, then this will be reflected in their level of human capital attainment. Lower stocks of human capital will be converted into lower relative earnings and a higher incidence of poverty. A similar hypothesis would be that difftrences are due to class background rather than discrimination against minority groups. This hypoLhesis has been verifi5ed for the case of Bolivia (Kelley, 1988), and often has been put forward for the case of Brazil, but has not been verified (Webster and Dwyer, 1988; Silva, 1985). 'Mh.w is, the great differences between ethnic groups could be due to the natural working of economic forces, rather than discrimination. According to this hypothesis, an individual's socioeconomic background in terms of family income and father's and mother's education and occupation are more important factors in determining present socioeconomic conditions than is ethnicity. Differential outcomes, of course, may be due to outright discrimination. Discrimnination against ethnic groups may work to deleteriously affect an individual's access to schooling, the quality of schooling that individual receives and labor market performance. This leads to lower schooling levels, lower returns to schooling, lower earnings and ultimately higher levels of poverty. Becker's (1971) seminal work on discrimination attempts to explain segregation in the workplace. Becker postulates that earnings differentials between ethnic groups would be due to individual "tastes" for discrimination against other labor market participants. Competitive forces in the economy would lead to a gradual elimination of wage discrimination over time. Related to this, assimilation theory, or the industrialization hypothesis, is the classical sociological theory of ethnic relations. It suggests that divisions based upon race and ethnicity will whither away in the long run in modern societies. This outcome is supposed to reflect modern industrial organization, where social mobility is based upon achieved rather than ascribed status (Hirschman, 1983). Also known as acculturation theories, they predict that inequality based on "traditional" criteria are being replaced by rational or legal criteria, and that 'particularistic" criteria are being replaced by 'universalistic' criteria such as education and ability. The implication is that the significance of race and ethnicity will decline as sciety devc"ps. It is argued that indigenous people who are "traditional' will place less importance on the labor force and will use it only to achieve a specific, short term end, such as obtaining cash to finance a lengthy period out of the labor market. Such individuals have been labelled as "target" workers: they work only as long as necessary to obtain a fixed sum of wages (Sandefur and Scott, 1983). The reasons put forward for this behavior include a desire to work at one's own pace, and the importance of kinship and community in Amerindian society. Traditional Amerindians see themselves as members of communities first, and are driven more for the good 9 of the community utan for individual achievement. This characteristic is expected to have a negative effect on labor force attachment and, ultimately, wages. Level of education, however, is expected to lead to a decline in traditional activities, and this is confirmed for the Canadian north (Stabler, 1990). Many Aymara who now live in urban environments maintain ties with the rural communities to their mutual advantage (Hardman, 1981). Indigenous people who reside in the cities normally maintain their rural ties and landholdings (Saavedra, 1981). The Aymara value education highly, which meshes with their traditional values of individualism, hard work and communal and private advancement (Hardman, 1981). Open competition and forceful self-expression, however, are missing from Aymara culture (Saavedra, 1981). The theory of "target' workers was developed from analyses of the work behuvior of peasants and has been applied to the case of Amerindians. It is argued that peasant agr.cultural workers can either work more hours, more intensely, or both. They seek output adequate to meet their basic needs. Since this work involves drudgery, their effort is not pushed beyond the point when increases in output are outweighed by the irksomeness of the extra work. A rough equilibrium is struck between the degree of satisfaction of family needs and the degree of the drudgery of labor (Chayanov, 1966). Peasants may have a certain target level of income, after which they begin consuming leisure. Thus, efforts to increase income might result only in an increase in the amount of leisure consumed. Traditional community values have persisted among Amerindians. Prior to European contact, these included entrepreneurial activity, which was crushed by the Europeans. When this entrepreneurial spirit became active again, innovation was effected on behalf of tribal enterprises, rather than for individual economic advancement (Hagen, 1968). Among indigenous peoples in Latin America, kinship systems, comuneros, and the institution of compadrazgo are very important for economic security and well-being (IFAD, 1992; Collins, 1983). Most theories, however, predict that discrimination will eventually decrease in society in the long run for the following reasons: it is inefficient from the perspective of profit maximizing employers; differences due to ethnicity will decline over time as ethnic groups become assimilated; and ethnic groups will achieve equality in terms of productive characteristics such as education, training and experience. Free markets and access to 4.ality education should lead to less discrimination in society over time. But segmented labor markets, as a result of such factors as ethnic and linguistic differences, can restrain the equalizing forces of competition. Research based in modern industrial societies points to the eradication of discrimination over time. Little research effort has gone into examining these issues in less developed societies, where, theoretically, discrimination is most likely due to the nature of the market and the great linguistic and ethnic differences not dissipated by schooling. The fact that indigenous populations have remained distinct after centuries of assimilation policies, violence, rural-urban migration, and increasing levels of schooling reflects the insufficiency of most theories. 10 Individual Costs of Discriniination Summaries of studies of six Latin American countries for which the overall earnings differential between ethnic groups has been estimated are presented in Table 3. Using a 1966 survey of about 1,000 male household heads in rural Bolivia, Kelley (1988) decomposes the differential between indigenous and nonindigenous individuals in terms of education, occupation and income. With background data on father's and own education and occupation, Kelley (1988) concludes that all (between 95 and 100 percent) of the overall differential is due to 'class" components (family background, education and occupation), rather than ethnic differences. In other words, equalizing human capital and family backgrounds of individuals would result in virtual elimination of ethnic socioeconomic inequalities. Kelley (1988) suggests that the main reason for the change from ethnic to class inequality in Bolivia is due to the 1952 Revolution, the most important event at the time of the survey. The Revolution resulted in a considerable increase in the power and opportunities available to Bolivia's indigenous population. A more recent analysis for urban Bolivia (1989) shows that the equalization of human capital and other productive characteristics would result in the near elimination of earnings differentials between indigenous and nonindigenous workers in Bolivia. The statistical decomposition of the ethnic earnings differential between indigenous and nonindigenous workers demonstrates that the portion of the overall earnings differential due to disparities in the productive characteristics of indigenous and nonindigenous working males is 72 percent in Bolivia. In other words, based on observed characteristics, the earnings differential between indigenous and nonindigenous working males would narrow by 72 percent if each group were endowed with the same productive characteristics. The remaining 28 percent difference ir earnings is "unexplained," and r-.flects both measurement error and unaccounted factors such as disparities in ability, quality of e6tucation, labor force participation, culture, and labor market discrimination. 'herefore, discrim nation could account for only as much as 28 percent of the overall earnings differential betw.-en indigenous and nonindigenous workers in the urban Bolivian labor market. 11 Table 3. Characteristics of Adult Males in Latin America by Ethnic Group and Country with Estimates of Discrimination Rate of Upper Bound Country and Minority/Majority Age Schooling Return to Discrimination Ethnic Group Earnings Ratio (years) (years) Schooling (percent) N Urban Bolivia (1989) Indigenous 38.5 7.4 5.7 2394 Norindig. .61 36.0 10.1 8.6 28 4070 Rural Bolivia (1966) Indigenous 1.2 675 Nonindig. .66* 4.5 12 421 Guatemala (1989) Indigenous 36.2 1.8 9.1 2459 Nonindig. .42 34.5 4.9 10.5 52 6029 Mexico (I989) Indigenous 34.5 3.8 8.7 476 Nonindig, .30 33.6 7.3 9.3 48 8343 Peru (1991) Indigenous 39.3 6.7 2.6 316 Nonindig. .43 37.6 10.0 6.2 50 1863 Paraguay (1990) Guarani .64 34.7 8.2 8.2 21 1084 Spanish 36.8 11.2 8.2 647 Brazil (1988) Black .50 39.8 5.1 10.8 51 1212 Brown .55 38.8 5.4 10.1 46 6857 White 39.9 7.9 13.3 1121 5 Source: Psacharopoulos and Patrinos, 1994; Kelley, 1988; Silva, 1992; Patrinos, Velez and Psacharopoulos, 1993. Note: 0 Refers to wealth. 12 Approximately one-half of the earnings differential can be attributed to differences in productivity-enhancing characteristics in Guatemala. The portion of the differential that is due to the productive characteristics or endowments of individuals is equivalent to 52 percent of the differential in log of wages between workers in indigenous and nonindigenous areas of Mexico. In other words, if those in indigenous areas were endowed with the same amounts of productive characteristics as those in nonindigenous areas, the difference in earnings between them would narrow by 52 percent. However, the remaining 48 percent difference in earnings is "unexplained." For Peru, the proportion of the overall nonindigenous-indigenous earnings differential that is due to the productive characteristics of individuals is equivalent to about 50 percent of the differential in log of earnings between indigenous and nonindigenous men. In other words, if indigenous workers were endowed with the same productive characteristics as nonindigenous workers, the earnings differential between them would narrow by 50 percent. The remaining difference in wages is 'unexplained." In Paraguay, decomposition of the overall earnings differential between monolingual Spanish speakers and Guarani speakers shows that much of the overall differential is explained by human capital differences. Thus, equality in educational attainment will go a long way towards equalizing labor market outcomes. For Brazil, wage differences between white and non-white males remain after controlling for education and estimated experience (Webster and Dwyer, 1988). In fact, the income gap between the two groups widens with increased schooling. Silva (1985) estimates a significant cost to 'being non-white' irrespective of being mulatto or black; non-whites are equally discriminated against in Brazil relative to whites (Table 4). Table 4: Discrimination in Brazil over Time Upper Bound of Discrimination (percent) Mulatto Black Nonwhite 1960 17.6 14.6 16.3 1976 32.9 26.3 31.2 1988 45.6 50.9 Source: Silva, 1985; 1992. The "upper bound' of discrimination presented in Tables 3 and 4 gives an indication of the "cost" of being a minority. In monetary terms, this is illustrated by the Bolivian case. The mean difference in nonindigenous-indigenous earnings is approximately 2,784 bolivianos 13 a year (Psacharopoulos and Patinos, 1994). From the decomposition results presented in Table 3, it is known that 28 percent of this differential is due to discrimination ("upper bound"); 72 percent is, therefore, due to productivity characteristics differences. In monetary terms, the mean difference that is attributed to productive characteristics differences is 1,992 bolivianos a year. The "unexplained' component, therefore, accounts for 792 bolivianos, and is interpreted as reflecting the costs to indigenous people of discrimination. Discussion Much of the earnings disadvantage of minority group workers is due to lower human capital endowments. While the monetary benefits of schooling are lower for the minority group populations, an increase in schooling levels would lead to a significant increase in eamings. The results compare with those obtained from studies of the earnings differential between indigenous and nonindigenous workers in the United States. These studies suggest that much of the earnings differential would disappear if indigenous workers had the same characteristics as nonindigenous workers. The same results, however, would not occur in the case of Canada, where the "unexplained" component of the earnings differential between indigenous and nonindigenous workers is very high. Discrimination accounts for only as much as 10 percent of the overall differential between indigenous and nonindigenous workers in the United States (Sandefur and Scott, 1983), while in Canada it accounts for 59 percent (Patrinos and Sakellariou, 1992). The Bolivian findings suggest that the socioeconomic condition of indigenous people can be improved because policy-influenced variables such as education and occupation are largely responsible for earnings differences. This provides considerable hope for the future. The question that remains, however, is how to improve the productive capabilities of the indigenous population. One obvious solution is to improve the educational level. The educational level of the Bolivian population has been increasing rapidly over the last few decades. The average schooling level of indigenous people increased continuously over time, with a sharp rise for cohorts born ir 1959 and later. The increase is particularly strong for the post-1952 Rzvolution cohorts born during 1949-53 and 1954-58. Given the fact that indigenous workers have fewer years of schoolung vis-a-vis nonindigenous workers, it is necessary to make an effort to increase their human capital. This is justified by the fact that productivity-enhancing characteristics are positively related with earned income, so that an increase in human capital will result in an increase in earnings. Access to better jobs could also be attained by improving the informnation network so that both employers and indigenous workers are better informed. In this way, employers would be better appriseJ about the qualifications of indigenous labor force participants, thus hopefully preventing discrimination, and indigenous workers would be better apprised about labor market opportnities (see Klitgaard. 1992). 14 In terrns of the "upper bound" estimates of discrimination that one gets from the individual-level analysis, it can be concluded that there can be significant costs to discrimination at the individual level. However, another question is whether the majority group also suffers a "cost' when there is discriimination against a minority group. When the majority group is very large compared to the minority group, then discrimination by the majority hardly lowers its income (Becker, 1993). But when the minority group is a sizable fraction, then discrimination by members of the majority injures them as well. An obvious example given by many authors is the case of South Africa. Discrimination against Blacks allowed whites to gain for only so long. Evertually, Apartheid became a cost to the economy. The reason for this is the resulting skill bottleneck. The manufacturing sector was dependent upon skilled workers, but as these jobs were reserved for people of European extraction and the education system made it impossible for the majority to compete for higher- level positions, then skill bottlenecks formed, acting as a brake on further expansion of the economy. Discrimination against Blacks ended up hurting whites, so the system of Apartheid fell apart (Lundahl, 1992; Becker, 1993). In Guatemala, the indigenous population (economic minority) makes up a very large - if not the majority - of the total population of the country. Does discrimination against indigenous people hurt the (economic) majority nonindigenous population as well? It is estimated that in 1960, only 7 percent of secondary school-aged children in Guatemala actually attended secondary school. Had the Guatemalans invested in education to increase attendance to a relatively modest 50 percent in 1960, the country's growth rate per capita from 1960 to 1985 might have increased an amazing 1.3 percent per year (Barro, 1991). Investing in the human capital of the large minority population and decreasing the level of discrimination against it will go a long way towards improving the standard of living of Guatemala's indigenous and nonindigenous population. This is clearly a priority research area. References Arrow, K. 1973. "Higher Education as a Filter." Journal of Public Economics 2:193-216. Barro, R.J. 1991. "Economic Growth in a Cross Section of Countries." Ouarterly Journal of Economics 106:407-43. Becker, G.S. 1971. The Economics of Discrminlation. Chicago: University of Chicago Press. Becker, G.S. 1975. Human Capital. New York National Bureau of Economic Research. Becker, G.S. 1993. "Nobel Lecture: The Economic Way of Looking at Behavior.' Journal of Political Economy 101:385-409. Birdsall, N. and R. Sabot, eds. 1991. Unfair Advantage: Labor Market Disrimnination in Developing Countries. Washington, D.C.: The World Bank. Blinder, A.S. 1973. "Wage discrimination: Reduced form and Structural Estimates." Jouual of Human Resources 8:436-65. Carnoy, M. 1980. 'Segmented Labour Markets: A Review of the Theoretical and Empirical Literature and its Implications for Educational Planning." In M. Carnoy et al., eds., Education. Work and Emnloyment Volume II. Paris: UNESCO, International Institute for Educational Planning. Chayanov, A.V. 1966. "On the Theory of Non-Capitalist Economic Systems." In D. Thorner, ed., The TheorEy of Peasant Economy. Homewood, IL: Richard D. Irwin, Inc. Chiswick, B.R. 1987. "Race Earnings Differentials." In G. Psacharopoulos, ed., Economics of Education: Research and Studies. Oxford: Pergamon Press. Chiswick, B.R. 1988. 'Differences in Education and Earnings across Racial and Ethnic Groups: Tastes, Discrimination, and Investrnents in Child Quality." Ouarterly Journal of Economic (August):571-97. Collins, J.L. 1983. "Fertility Determinants in a High Andes Community." Poulatiol and Development Review 9:61-75. Doeringer, P.B. and M.J. Piore. 1972. Internal Labor Markets and Manpower Analysis. Lexington, MA: D.C. Heath. Duncan, O.D. 1968. "Inheritance of Poverty or Inheritance of Race?" In D.P. Moynihan, ed., On Understanding Poverty. New York: Basic Books. 16 Filer, R.K. 1983. "Sexual Difference in Eamings: T'he Role of Individual Personalities and Tastes." Journal of Human Resources 18. Gunderson, M. 1989. "Male-Female Wage Differentials and Policy Responses." Journal of Economic Literature 27:46-72. Hagen, E.E. 1968. The Econofics of Development. Homewood, IL: Richard D. Irwin, Inc. Hardman, M.!. 1981. "Introducti,ry Essay." In M.J. Hardman, ed., The Avyara Language in Its Social and Cultural Context: A Collection of Essays on Aspects of Aymara Laneuage and Culture. Gainesville: University Presses of Florida. Hirschrnan, C. 1983. "Labor Markets and Ethnic Inequality in Peni.,sular Malaysia, 1970.' Journal of Developing Areas 18:1-20. International Fund for Agricultural Development (IFAD). 1992. The State of World Rural Poverty; An Inguirv into Its Causes and Conseguences. New York: New York University Press. Kelley, J. 1988. "Class Conflict or Ethnic Oppression? The Cost of Being Indian in Rural Bolivia." Rural Sociolo 53:399-420. Kitagawa, E.M. 1955. "Components of a Difference between Two Rates." Journal of American Statistical Association 50:1168-1194. Klitgaard, R. 1992. Adjustin, to Reality: Beyond "State versus Market" in Economic Development. San Francisco: ICS Press. Lundahl, M. 1992. Apartheid in Theory and Practice: An Economic Analysis. Boulder: Westview Press. Mincer, J. 1974. Schooling. Experience and Earnings. New York: National Bureau of Economic Research. Oaxaca, R. 1973. "Male-Female Wage Differentials in Urban Labour Markets." International Economic Review 14:693-709. Patrinos, H.A. and C.N. Sakellariou. 1992. "North American Indians in the Canadian Labour Market: A Decomposition of Wage Differentials." Economics of Education Review 11:257-66. Patrinos, H.A., E. Velez and G. Psacharopoulos. 1993. "Language, Education and Earnings in Asuncion, Paraguay." 17 Psacharopoulos, G. and H.A. Patrinos, eds. 1994. Indigenous People and Povern in Latin America: An Empirical Analysis. Washington, D.C.: The World Bank. Psacharopoulos, G. and Z. Tzannatos. 1992. Women's Employment and Pay in Latin Arnerica: Overview and M1ethodologv. Washington, D.C.: The World Bank. Psacharopoulos, G., S. Morley, A. Fiszbein, H. Lee and B. Wood. 1993. "Poverty and Income Distribution in Latin America: The Story of the 1980s." Report No. 27. Latin America and the Caribbean Technical Department, The World Bank, Washington, D.C. Saavedra, C. 1981. "Social and Cultural Context of the Aymara in Bolivia Today." In M. J. Hardman, ed., The Aymara Language in Its Social and Cultural Context: A Collection of Essays on Aspects of Avmara Languaze and Culture. Gainesville: University Presses of Florida. Sandefur, G.D. and W.J. Scott. 1983. "Minority Group Status and the Wages of 'ndian and Black Males." Social Science Research 12:44-68. Schultz, T.P. 1991. "Labor Market Discrimination: Measurement and Interpretation." In N. Birdsall and R. Satht, eds., Unfair Advantage: Labor Market Discrimination in Developing Counties. Washington, D.C.: The World Bank. Schultz, T.W. 1961. "Investment in Human Capital." American Economic Review 51:1-17. Siegel, P.M. 1965. "On the Cost of Being a Negro." Socioloeical Inquiry 35:41-57. Silva, N.V. 1985. 'Updating the Cost of Not Being White in Brazil." In P.-M. Fontaine, ed., Race. Class and Power in Brazil. Los Angeles: University of California. Silva, N.V. 1992. "Racial Discrimination in Income." Paper presented at the International Seminar, Labor Market Roots of Poverty and Inegualitv in Brazil. August 12-14, Rio de Janeiro. Stabler, J.C. 1990. "A Utility Analysis of Activity Patterns of Native Males in the Northwest Territories." Economic Development and Cultural Chanye 39:47-60. Webster, P.L. and J.W. Dwyer. 1988. "The Cost of Being Nonwhite in Brazil." Sociolov and Social Research 72:136-142. Welch, A.R. 1988. "Aboriginal Education as Internal Colonialism." Comparative Education 24:203-15. World Bank. 1993. 'Poverty and the Social Sectors in Paraguay: A Poverty Assessment." Report No. 12293-PA. Washington, D.C. Human Resources Development and Operations Policy Working Paper Series Contact for Title Author Date paper HROWP14 Labor Markets and Market- Arvil V. Adams October 1993 S. Khan Oriented Refoxms in Socialist 33651 Economies HROWP15 Reproductive Tract Infections. May T.H. Post October 1993 0. Shoffner HIV/AIDS and Women's 37023 Health HROWP1 6 Job Security and Labor Ricardo D. Paredes November 1993 S. Khan Market Adjustment in 33651 Developing Countries HROWP17 The Effects of 4aye Luis A. Riveros November 1993 S. Khan Indexation on Adjustment, 33651 Inflation and Equity HROWP1 8 Popular Participation in Philip R. Gerson December 1993 L. Malca Economic Theory and Practice 37720 HROWP19 Economic Returns from Edwin Mansfield January 1994 I. Dione Investments in Research and 31447 Training HROWP20 Participation, Markets and Deepak Lal January 1994 L. Malca Democracy 37720 HROWP21 Sale Motherhood in Patricia Daly January 1994 0. Shoffner Francophone Africa Michael Azefor 37023 Boniface Nasah HROWP22 Indigenous People and George Psacharopoulos February 1994 I. Conachy Poverty in Latin America Harry Anthony Patrinas 33669 HROWP23 Is Grameen Bank Sustainable? Shahid Khandker February 1 994 S. Davii Baqui Khalily 33752 Zahed Khan HROWP24 Concepts of Educational Marlaine E. Lockheed March 1994 M. Verbeeck Efficiency and Effectiveness Eric Hanushek 34821 HROWP25 Scientific Research for Erik W. Thulstrup March 1 994 1. Malca Development 37720 HROWP26 Issues in Education Finance Stephen P. Heyneman Apnl 1994 B. Cassorla and Management in ECA and 371 72 OECD Countfies HROWP27 Vocational Education and Julian Schweitzer Apnl 1994 A. Gonzalez Training: The Role of the 37799 Public Sector in a Market Economy HROWP28 Social Security Issues and Nguyen X. Nguyen May 1994 M. Espinosa Elements of Reform 37599 HROWP29 Health Problems and Policies Mary Eming Young May 1994 0. Shoffner for Older Women: An 37023 Emerging Issue in Developing Countries Human Resources Development and Operations Policy Working Paper Series Contact for Title Author Date paper HROWP30 Language and Education in S.M. Cummings May 1994 M. Espinosa Latin America: An Overview Stella Tamayo 37599 HROWP31 Does Participation Cost the Jesko Hentschel June 1994 D. Jenkins World Bank More? Emerging 37890 Evidence HROWP32 Research as an Input into Harold Alderman June 1994 P. Cook Nutrition Policy Formation 33902 HROWP33 The Role of the Putlic and Deepak Lai June 1994 M. Espinosa Private Sectors in Health 37599 Financing HROWP34 Social Funds: Guidelines for Soniya Carvalho July 1 994 K. Labrie Design and Implementation 31001 HROWP35 Pharmaceutical Policies: Graham Dukes July 1994 0. Shoftner Rationale and Design Denis Broun 37023 HROWP36 Poverty, Human Development Harsha Aturupane August 1994 P. Cook and Growth: An Emerging Paul Glewwe 30864 Consensus? Paul lsenrnan HROWP37 Getting the Most out of Helen Saxenian September 1994 0. Shoffner Pharmaceutical Expenditures 37023 HROWP38 Procurement of Denis Broun September 1 994 0. Shoffner Pharmaceuticals in World 37023 Bank Projects HROC%P39 Notes on Education and Harry Anthony Patrinos September 1994 1. Conachy Economic Growth: Theory 33669 and Evidence HROWP40 Integrated Early Child Mary Eming Young October 1 994 0. Shoffner Development: Challenges and 37023 Opportunities HROWP41 Labor Market Insurance and Deepak Lal October 1994 M. Espinosa Social Safety Nets 37599 HROWP42 Institutional Development in Alberto de Capitani October 1994 S. Howard rhird World Countiits: The Douglass C. North 30877 Role of the World B.nk HROWP43 Public and Private Secondary Marlaine E. Lockheed November 1994 M. lierbeeck Schools in Developing Emmanuel Jimenez 34821 Countries HROWP44 Integrated Apprcaches to T. Paul Schultz November 1994 M. Espinosa Humnan Resource 37599 Development HROWP45 The Costs of Discrimination in Harry Anthony Patrinos November 1994 1. Conachy Latin America 33669