Factors influencing the likelihood of overeducation: a bivariate probit with sample selection framework

Abstract

Contrary to expectations, the likelihood of overeducation is shown to be inversely related to unemployment rates when not control for selectivity. Furthermore, incidence data show that overeducation is more common among men than women and among Whites than Blacks. At issue is selectivity: employment must be selected for overeducation to occur. When using bivariate probit with sample selection models, the likelihood of overeducation is found to be positively related to local unemployment rates, higher for women than men, higher for mothers of young children than other women, and lower for fathers than other males. Race, not speaking English very well, and having a disability are found to have a greater impact on the likelihood of overeducation than incidence data suggest.

Keywords:

• educational economics
• overeducation
• undereducation
• unemployment
• job matching
• bivariate probit

Notes

1. Simple probit coefficient estimates of the models presented in Table 4 that do not control for selectivity are available upon request (not shown for brevity). Coefficient estimates are generally similar to those found in the multinomial logits (Table 2).

2. Büchel and van Ham (2003) find German women to more likely to be overeducated than their male counterparts when they have a partner and/or multiple children. Caution is advised in comparing their results with the results in this paper. Büchel and van Ham examine a different topic, their gender coefficient is not statistically significant at the 0.10 level, and gender is interacted with three other variables.