How Many and What Instruments Are Needed to Estimate Returns to Education? The Case of Taiwan

Abstract

This paper applies the 2SLS instrumental variable method to estimate the rate of return to education using data from the 1990, 2000, and 2010 Taiwan’s Manpower Utilization Survey. Tests are conducted to determine the most effective valid instrument from all combinations of the four IVs. The estimated rate of return to education is relatively higher by the IV method than by the OLS method. However, adding more IVs do not imply a better estimation and institutional, geographic, and cultural variables are more plausible candidate for valid instruments. Moreover, downward bias by OLS estimation is found to be greater for females than for males