regressions in both The Bell Curve and Fisher et al. (1996)’s work, noting that the estimated
coefficients of the cross-sectional analysis are likely to be biased due to the existence of
unobservable individual characteristics. Then, we turn to Cawley et al. (1996, 1999), which
utilizes a random effects panel regression model to estimate the ceteris paribus effect of
measured intelligence on the wage rate.1 Random effects panel regression eliminates the omitted
variable bias resulting from endogenous unobservable characteristics. However, our results
indicate that the strict exogeneity assumption is violated for the panel data, so that the estimated
coefficients in Cawley et al. are not consistent. Finally, we introduce a random effects panel-
estimation technique developed by Hausman and Taylor (1981) as the appropriate regression
model to estimate the effect of intelligence on the wage rate.
II Data Set
The data used in this paper come from the National Longitudinal Survey of Youth 1979
(NLSY79), which is administered by the U.S. Department of Labor since 1979 and designed to
represent the entire population of American youth. The NLSY79 consists of a randomly chosen
sample of 6,111 U.S. civilian youths, a supplemental sample of 5,295 minority and economically
disadvantaged civilian youths, and a sample of 1,280 youths on active duty in the military. All
youths were between thirteen and twenty-three years of age in 1978. These data include equal
numbers of males and females. Roughly 16 percent of respondents are Hispanics2 and 25 percent
are black, indicating an over-sampling of racial/ethnic minorities. Sample weights are provided
1 We focus on Cawley et al. (1999) rather than their 1996 or 2001 works in terms of specification-matching, though all three
papers utilize the same data source and present related findings.
2 We recognize that “Hispanic” is an ethnicity instead of race and that the concepts of “black” and “Hispanic”, or “white” and
“Hispanic” are not mutually exclusive. For example, people can be categorized as “white Hispanic” or “black Hispanic”. The
strict definitions of “white”, “black” and “Hispanic” have very subtle and complicated social implications. However, for the
purpose of this paper, black, white, and Hispanic are treated as distinct racial/ethnic subgroups.