The Blackwell Companion to Sociology (47 page)

(1994), Ferguson (1995), and Neal and Johnson (1995) all reproduce this

same result: skill differentials (or, for O'Neill and for Maxwell, ``schooling

quality'' differentials) obviate any important role for discrimination in labor

markets as an explanation for earnings gaps.

Thè`returns to skills'' explanation for widening earnings gaps has received

prominent attention. According to this line of argument, black males possess

significantly fewer skills on average than white males. Part of the reason for the focus on black and white males is that these groups show the largest earnings

disparities. Black±white disparities among females are not as pronounced (Dar-

ity, 1980).

Racial Economic Inequality in the USA

191

Support for the argument is not unequivocal. Using the Panel Study of Income

Dynamics (PSID) data, Card and Lemieux (1996) estimate the returns to unob-

served and observed skills for males with earnings from 1979 to 1985. Their

hypothesis is that if the cause of the widening racial wage gap between blacks

and whites in recent years is due to the increase in returns to skills, then one would expect to find that the black±white wage gap is proportional to the

returns to skills across various years for a common cohort of workers. Therefore, they conclude that the racial wage gap cannot be due to productivity differences because in their sample they estimate little change in the wage gap even in an era when returns to skills were increasing from 5 to 10 percent.

The fact that Card and Lemieux do not find an increase in the wage gap for a

common cohort of workers suggests that the observed increase in the wage gap

among different cohorts of workers must be attributable to the differing experi-

ences of younger black workers. This issue is addressed explicitly by Holzer

(1994), who examines a variety of explanations for the deterioration of earnings among young blacks. He concludes that productivity-related or skills explanations are insufficient to account for the rising racial wage gaps and points at least in part to discriminatory processes (via networks) that must be at play.

The reversal in the narrowing trends in racial earnings gaps has brought forth

new efforts to explain the phenomenon. Explanations include shifting industry

and regional employment, a decline in the real minimum wage, deunionization,

a growing supply of black educated workers relative to white workers, and

increased criminal activities among school drop-outs (Bound and Freeman,

1992). Another explanation, the spatial mismatch hypothesis, flows from the

ghetto dispersal versus ghetto development debate of the 1960s (Kain, 1968;

Harrison, 1972, 1974).

The question for the 1960s was whether residential segregation or employ-

ment discrimination caused the low incomes of blacks. The question for the

1990s is whether concentration in inner cities and/or industries where there are job losses or supply-side factors, such as high reservation wages and/or the

inducements of crime, is the cause. The spatial mismatch debate has many of

the features of the black underclass debate, where the issues of contention are

structural/demand-side factors versus behavioral/supply-side factors (Fainstein, 1987; Darity et al., 1994).

The key piece of evidence underlying the spatial mismatch hypothesis is the

apparent loss of inner-city, low-wage jobs as employment expands in the suburbs

(Kasarda, 1985; Wilson, 1987). David Ellwood (1986, p. 181) tested the hypoth-

esis and concluded: ``The problem is not space. It's race'' (Ellwood, 1986, p. 181).

Ihlanfeldt and Sjoquist (1990) do find a positive effect of proximity to a job on black and white youth employment rates. They find that 33±54 percent of the

racial gap in youth employment rates can be explained by spatial separation

from jobs. They conclude: ``poor job access is a significant contributor to the

joblessness of black youth'' (Ihlanfeldt and Sjoquist, 1990, p. 268).

Still, a comprehensive review of the recent evidence provides mixed support

for the spatial mismatch hypothesis (Moss and Tilly, 1991). In contrast,

the evidence against the skills mismatch hypothesis is more compelling. This

192

W. A. Darity, Jr and S. L. Myers, Jr

hypothesis states that technological change has been biased against low-skilled

workers, suggesting a decline in the demand for young blacks, who are dispro-

portionately low-skilled. However, the greatest shifts in demand away from low-

skilled workers occurred during the period when racial earnings gaps were

narrowing. In more recent years, while racial earnings gaps continue to widen,

the pace of skill restructuring has slowed markedly (Howell, 1994).

An incidental finding of many of the human capital-based studies of racial

earnings inequality is that factors such as neighborhood or census tract char-

acteristics, percentage black, or degree of segregation enter as statistically significant variables in earnings regressions. The justification for inclusion of

measures of segregation or concentration of minorities appeals to the social

isolation conjecture of William J. Wilson (1993) and/or the residential segrega-

tion thesis of Douglas Massey (1990). Because the effects of concentration of

blacks on employment or earnings tend to be negative, the ex-post theory about

the signs obtained relates to the various alleged disutilities associated with living in segregated or predominantly black neighborhoods.

Directions for Future Research

Research

Clearly, there are numerous unresolved issues in the labor econometrics liter-

ature concerning the reasons for the increase in racial inequality in recent years.

Five stand out prominently, on which we comment below.

Skills Mismatch and General Inequality

Is the skills mismatch hypothesis valid as an explanation for changes in general inequality? If not, then it cannot motivate the analysis of black±white male

earnings disparity in the 1980s. In particular, if a shift toward the demand for skills did take place, given the claim of these researchers that there was convergence before the 1980s, the shift must be timed precisely with the 1980s.

Times Series and Test Scores

There is an anomaly in time-series results in contrast with cross-section results.

Time-series data indicate that black standardized test scores rose relative to white scores in the 1980s, but earnings did not do so. The regression results that appear to vanquish the effects of discrimination all are based on cross-sectional data.

Test Scores as Proxies for Race

Take any standardized test on which blacks do worse on average than whites.

One suspicion is that any of these scores included in a regression analysis would have the effect of eliminating evidence of discriminatory residuals. But if, for example, AFQT scores are endogenous outcomes generated by processes similar

to those generating earnings, these models are misspecified (Oust-like models

Racial Economic Inequality in the USA

193

that use IQ as a proxy for `àbility''). Rodgers and Spriggs (1996) demonstrate

how powerfully results are altered by recognizing the endogeneity of AFQT

scores. Moreover, when psychological capital is also taken into account in the

process of wage determination (by psychological capital we mean motivation

and self-esteem), discriminantory residuals re-emerge even when AFQT scores

are included in wage equations (see Goldsmith et al., 1997).

Unmeasured Effort

Patrick Mason (1994) argues that a previously unobserved variable that had

been excluded from wage equations is effort. He attempts to demonstrate that

black effort is greater for attainment of given characteristics because blacks face greater obstacles. The obstacles are embodied in disadvantageous ``social capital,'' not in the narrow sense of neoclassical economists, but in the richer

sociological sense. For Mason (1994), ``social capital'' should be understood

as: ``The importance of having access to individuals embedded . . . in positions of power and authority. Consequently, equivalent scores between black and

white males may actually correspond to greater black male personal productiv-

ity-linked skills.''

Rodgers and Spriggs (1996) show that there is a systematic racial difference in

the capacity of AFQT scores to predict earnings when they estimate separate

structural equations for AFQT for blacks and whites with education, age, school

quality, and family background measures as the independent variables. AFQT is

a much weaker predictor of earnings outcomes for blacks than for whites. This is consistent with the general observation that most standardized tests are weaker

predictors of black than white performance in a variety of areas. Previous

research has been predicated on the assumption that there is no racial difference in the predictive power of AFQT scores.

Coupled with the fact that employers would not observe the AFQT scores in

the NLSY, Rodgers and Spriggs (1996) argue that AFQT is a biased predictor of

black skills, since the coefficients in the black AFQT equation tend to generate lower scores for given education, age, school quality, and family background

characteristics. They propose that a non-discriminatory weighting of African

American characteristics would involve generating hypothetical black scores on

AFQT by using the coefficients from the white equation. When the hypothetical

scores are inserted in the wage or earnings equation, racial wage gaps re-emerge that point toward indirect statistical evidence of discrimination against blacks (see also Maume et al., 1996, for a similar finding). Rodgers and Spriggs (1996) also point out that the military itself ``does not use the AFQT score as the only indicator in placing military personnel in different occupations'' because it is a far from comprehensive predictor of performance.

Wealth as an Omitted Variable

We remain curious about what results would be generated in earnings equations

that decompose racial earnings differentials if another omitted variable entered 194

W. A. Darity, Jr and S. L. Myers, Jr

into the analysis: wealth. However, oddly enough, in a case where reported

wealth from the 1978 NLSY is included in a simultaneous equation model of

wages and psychological well-being, the wealth variable is grossly insignificant (Goldsmith et al., 1997). It remains to be seen if that result holds with an older cohort with self-reports on wealth. The most powerful adverse effects of the

racial wealth disadvantage fall on patterns of self-employment, access to higher education, and access to quality health care.

Conclusions

In summary, the recent explanations for the widening gap between black

and white earnings have focused on the alleged human capital roots of low

black wages and salaries. Alternative explanations that focus on institutional

and structural deficits of labor markets and employers have largely been ignored.

While many analysts claim by deduction that the widening earnings gaps

between whites and blacks must not be caused by discrimination, racism, or

other institutionalized factors, we remain unconvinced. There is persuasive

evidence that the US anti-discrimination effort was nearly dismantled precisely

during the period when racial gaps earnings reopened in the 1980s. We are

also dissuaded because the process of eliminating explanations by focus-

ing on individual deficiencies of minorities themselves leaves much to be

desired.

More reasonable as an explanation of how and why racial earnings gaps

widened is that white males lost jobs. Put simply, the widening of the racial

earnings gap is inextricably linked to the widening of overall inequality and the loss of white middle-class employment opportunities. White males were

squeezed out of the vanishing middle-class jobs, which had been their purview,

especially well paid blue-collar jobs. They, in turn, were then crowded into a

lower tier of occupations that they would not otherwise have held. They

squeezed black males out of those jobs.

In fact, white males have begun to appropriate a set of jobs that were

previously held by blacks and are making them ``white male'' jobs. Black males

and females in the most tenuous labor market positions have been pushed out,

both by whites and by immigrants from South America and Southeast Asia.

Furthermore, those black males who had held blue-collar jobs and made the

transition into white-collar employment actually went from operative positions

to sales positions that paid less. In contrast, white males who made a similar

transition went from blue-collar operative positions to sales positions that paid them more (Cotton, 1988).

In a sense, the argument here is that exclusion (discrimination) is endogen-

ously linked to the employment needs of non-black males. As the occupational

distribution eliminates jobs traditionally held by white males, they secure less attractive jobs at the expense of black males. The force of discrimination is then seen as instrumental. Its application intensifies when the dominant group's status is threatened, either because of improved productivity-linked characteristics of Racial Economic Inequality in the USA

195

members of a potential rival group or because of a diminution in the job

opportunities of members of the dominant group.

The endogenous model of discrimination also finds support in other contexts.