Revisiting The Bell Curve Debate Regarding the Effects of Cognitive Ability on Wages

Regression Results and Interpretation

Regressions 3A to 3E present our results from applying the Hausman-Taylor model to the
NLSY79 data set. Regression 3A limits the sample to the survey years 1979 to 1994. The large
increase in the coefficient of g from 0.042 (2A) to 0.277 (3A) is surprising; note it even exceeds
the coefficients found in the cross-sectional analysis reported in Table 1. Note that the two
regressions use the same set of data, the same regression equation, and the same independent and
dummy variables. Hence, any difference in the estimated coefficients must be the result of the
bias caused by the violation of the strict exogeneity assumption in regression 2A. Therefore, by
using the random effects model, we have potentially underestimated the marginal effect of g on
the wage rate. Moreover, in regression 3A, the demographic group interaction terms are not
statistically significant. In comparison with regression 2A, in which all of the dummy interaction
terms are significant at the 5 percent significance level, regression 3A casts some doubt on the
claim by Cawley et al. (1996) that the wage return to intelligence is different across the race and
gender demographic groups.

Regression 3A confirms one important claim of The Bell Curve’s critics: socio-economic
background variables have significant explanatory power in predicting wage differentials. For
example, the coefficients on human capital variables such as education, tenure, and labor market
experience remain statistically significant in 3A and are close to those in regression 2A in
magnitude. Compared with the large difference in the estimated coefficients of g, these variables
are largely unaffected by the use of the random effects versus the HT estimators. Herrnstein and
Murray (1994) ignore the effect of the human capital measures on the wage rate; hence, their
estimation of g’s explanatory power suffers from omitted variable bias.

17

<<   17   18   19   20   21   22   23   >>

More intriguing information