Thus, this model explains why groups of workers with identical expected productivity could earn different wages, and therefore provides yet another reason why labor market information may generate wage differences between similar workers. This is perhaps most pertinent for sex differences in wages, because women do not, on average, receive lower performance ratings than men. While simple statistical discrimination therefore cannot explain women’s lower wages, such extensions of the statistical discrimination model, coupled with worse information about women, can generate group discrimination against women.

To this point, a single reliability ratio for information regarding new hires is estimated. The question of differential information, however, hinges on whether the reliability ratio differs across groups. This question can be addressed by estimating the regression
for each subgroup, obtaining both OLS and IV estimates.

The reliability of information on new workers can be estimated as the ratio of the OLS estimate of a (aoLS)t0 the IV estimate (aIV), since
To carry out a statistical test for differences in this estimated ratio across demographic groups, an estimate of the variance of this ratio is required. A first-order Taylor-series expansion yields an approximation for the variance legitimate payday loans

The covariance term in equation (8) is straightforward to estimate, since both aiV and aOLS are linear estimators. In particular,

where e0LS and eIV are the OLS and IV residuals. In fact, equation (6) ends up being estimated for more than one demographic group, in which case P in equation (9) is simply the matrix including the performance rating as well as the dummy variables for demographic subgroups. The estimates from each demographic group (or set of groups) can be treated as coming from independent samples, making it easy to test for differences in the estimated ratios aoLS/aIV across groups.

There is no way to determine how much differences in labor market information would shift average wages for a group. Nonetheless, evidence of differences would suggest that better information about particular groups of workers could raise their average wages.