How does the relationship between earnings and schooling change with the introduction of comprehensive economic reform? This paper sheds light on this question using a unique data set and procedure to reduce sample selection bias. Our evidence is from consistently coded, non-retrospective data...