In this example Mincer includes work experience as a covariate:
People with less schooling but much more work experience might earn more than people with more schooling but no work experience.
Omitted variable bias, OVB, leads us to underestimate the returns to schooling.Â
Lets' determine Why?
The two regressors -- schooling and experience can be correlated with each other!
1. Schooling and work experience are negatively correlated with one another
2. Schooling and experience are positively correlated with earnings.
People with more schooling have less experience (negatively correlated)
When we increase schooling, earnings don't increase as much as they would if we were holding experience constant as a covariate.
Example Ability
How do we know when we've got the right set of covariates so that we've got an unbiased estimate?
We can't know!
We would need an experiment that randomly sends some students to more years of school than others.
Then every possible omitted variable would be uncorrelated with years of schooling, eliminating OVB.