Earnings regressions for married and cohabiting Swedish males in 1993 indicate that controlling for family background reduces the measured return to education by about 9 percent, net of measurement error bias. The Swedish evidence is generally consistent with the hypothesis that family...