Intergroup disparity: economic theory and social science evidence.

• Author: Darity, William A. Jr.
• Position: Transcript

1. Introduction

   Consider a hypothetical community where two groups are distinguished by an ascriptive characteristic or a set of ascriptive characteristics that are understood as separating markers by the members of each of the groups. Indeed, the distinguishing characteristics may be recognized by others external to the two groups or external to the community. Furthermore, these separating markers are associated with the social categories of race or ethnicity.

   Credible surveys taken by researchers reveal that, on average, the members of the two groups experience widely distinct economic outcomes. Members of group A are found to have significantly lower earnings, a higher incidence of unemployment, less prestigious occupations, lower levels of total income, and substantially lower levels of wealth than members of group B.

   Conventional economics offers two major explanations for the observed disparity between groups A and B. First, average productivity-linked endowments (or a human capital gap) correlate strongly with membership in each of these ascriptively bounded groups, producing the differences in outcome. The origins of these background differences have been attributed to a variety of causes. The four most prominent reasons offered for the wedge in human capital acquisition have been (i) an unequal allocation of social resources across the two groups, (ii) the intergenerational transmission of past disadvantage or advantage across the groups, (iii) somewhat invidiously, cultural differences between the groups, and (iv) most invidiously, genetic differences between the groups that generate not only the ascriptive distinctions but also differences in intelligence and motivation.

   Second, at any moment, members of each group with given productivity-linked characteristics can be confronted with differential treatment of those characteristics in markets, especially labor markets. In short, conventional theory suggests that the observed gap in outcomes could be due to discrimination against group A and nepotism on behalf of group B.

   The two major explanations for group A's inferior positions - a human capital deficiency and discrimination - need not be mutually exclusive. The typical member of group A may have less of the stuff that makes for economic success than the typical member of group B, but there could be some members of group A competing for certain economic rewards with a similar or superior human capital profile as the contestants from group B seeking similar economic rewards who still are excluded from the positions. Nevertheless, conventional economic theory privileges the human capital explanation over the discrimination explanation.

   Orthodoxy denies the persistence of discrimination as a market-based phenomenon. Indeed, I submit that neoclassical economics cannot be manipulated to produce a convincing story of why groups or individuals who differ ascriptively but who share similar productivity-linked attributes experience differential treatment in markets over time. Even the best neoclassical story, the statistical discrimination hypothesis, is not satisfactory.

   The statistical discrimination story says profit-motivated employers (or, more generally, selectors) in an environment of imperfect information believe they can distinguish visually between candidates from groups A and B. Furthermore, selectors believe candidates from groups A and B are drawn from different frequency distributions for ability to perform. They assume, for example, that the mean of group B's ability distribution is markedly higher than the mean of A's ability distribution, and there is little difference, if any, in the variance. Therefore, they display a preference for members of group B.

   Suppose selectors are wrong; the two distributions actually are identical. Then that fact should be learned over time. Profit-motivated selectors will become indifferent between A and B members. On the other hand, suppose selectors are correct. Then we do not have a theory of discrimination after all. At base, the subsequent inequalities in intergroup outcomes are due to average differences in ability. This has led Barbara Bergmann to observe trenchantly that the statistical theory of discrimination really is "human capital theory in drag."

   To compound matters, the. informational assumption that leads selectors to rely on knowledge, whether accurate or not, about group affiliation in making individual hiring decisions seems tenuous. It suggests that selectors are incapable of making a sound approximation about a candidate's future potential to perform without relying on the additional signal of group affiliation. This is implausible given the vast resources corporations devote to hiring decisions and the design of screening mechanisms. Over time, an appropriate set of questions or tests should emerge that will facilitate selection, regardless of group affiliation. Therefore, when group affiliation is taken into account, it is not likely to be done to enhance the accuracy of prediction of an individual's performance.

   Orthodox microeconomics invariably reverts to the position that persistent intergroup differences in economic outcomes must be due to induced or inherent comparative human capital deficiencies for members of group A. In the long run in neoclassical or marginalist theory, market processes only can produce differential outcomes between groups if the groups are fundamentally different in attributes salient to performance. Employers preferring profits to prejudice are expected to prevail over those preferring prejudice to profits; according to conventional theory, the latter presumably will be driven out of business. In making a critical conceptual distinction between extra-market discrimination (e.g., unequal group access to quality schooling), and in-market discrimination, conventional theory clearly gives precedence to the former.

2. Race Coefficients and the Blinder-Oaxaca Decomposition

   However, when we examine the empirical record in social science research, we find strong evidence of intergroup discriminatory differentials that appear to be generated in the market across countries, time, and region. Contrary to the forecast of conventional theory, discriminatory differentials do not decline consistently over time in market-based economies. In short, competitive market activity has not succeeded in rooting out in-market discrimination. Indeed, I contend that the empirical, international record of discrimination indicates the need for reconstruction of the way in which economists view and understand intergroup disparity.

   What is the evidence of ongoing discrimination? Indirect statistical tests detect in-market discrimination in two ways. The first type of indirect statistical test is the race coefficients approach. The coefficients are generated from log earnings, log wages, or other labor market outcomes as the left-hand-side variable in regression equations with race or ethnicity dummies on the right-hand side. Other variables included on the right-hand side serve as controls for human capital differences between members of the groups. A statistically significant and negative sign on the race or ethnicity coefficients is then taken as evidence of discrimination in the labor market against members of the corresponding group. This approach necessarily constrains estimated coefficients on the remaining included variables to be the same for both groups.

   The second type of indirect statistical test is the Blinder-Oaxaca approach (Oaxaca 1973; Blinder 1974), an approach that exploits the potential for differences in estimated coefficients between the groups. The starting point for this approach is estimation of separate labor market outcome equations for each group. The Blinder-Oaxaca approach allows the analyst to isolate the extent to which outcomes vary between groups because of differences in mean human capital characteristics and differences in treatment of given characteristics in labor markets. The latter effect - interpreted as evidence of discrimination - is captured by differences in the estimated constant term and coefficients on the variables included in the regressions between groups A and B. Thus, the Blinder-Oaxaca approach facilitates decomposition of the gap in group outcomes between differences in average group characteristics and differences in returns to average group characteristics - a juxtaposition between the roles of an intergroup human capital gap and an intergroup pattern of discriminatory treatment.

   Gottschalk's (1997) study of the recent history of inequality in the U.S. economy uses the first, or the race coefficients, approach. In addition to including race as a right-hand-side variable in a log earnings equation using Current Population Survey (CPS) data, Gottschalk also controls for schooling, experience, age, gender, and regional location. He finds a pure negative effect on log earnings associated with being black; his race coefficient estimates were (-0.115) for 1980 and (-0.119) for 1990.

   In conjunction with David Guilkey and William Winfrey (Darity, Guilkey, and Winfrey 1996), I have used a variant of the Blinder-Oaxaca technique using decennial census data. In a study using self-reports on the ancestry question asked in both the 1980 and 1990 census, in conjunction with responses to the race question, we partitioned the U.S. population into 50 ethnic/racial groups. We also examined males and females separately so that all of our comparisons were intragender. We used all males and all females as the reference groups for the variant of the decomposition we performed. The variables included in our log-wage regressions were education (years of schooling, college completion), age, experience, metropolitan versus nonmetropolitan location, English fluency, disability status, born in the U.S. or not, assimilated or not, occupational category, and regional location.

   We found no systematic...