On estimating earnings functions for LDCs

A technique is developed for estimating earnings functions from data in which both wage-earners and self-employed persons are included; it is argued that the return to schooling is overestimated if only wage-earners are in the sample. A procedure for simultaneously imputing labor incomes to the self-employed is also developed and carried out with 1971 data for Bangkok, Thailand. The evidence suggests that ceteris paribus the self-employed have labor earnings at least as high as their wage-earning counterparts.