Unobserved Household and Community Heterogeneity and the Labor Market Impact of Schooling: A Case Study for Indonesia

Abstract

Studies have claimed that schooling investments have high rates of return in developing countries. Researchers however fail to control for all of the unobserved household and community effects which may affect estimates of returns. Using data from the 1986 Indonesian Labor Force Survey wage functions are estimated for Indonesia with and without fixed effects to see if controls for outside factors which might be correlated with years of schooling and wages make a difference. The impact of schooling on wage rates and hours supplied to the paid labor market is assessed with separate consideration given to males and females. Results suggest that estimates of the impact of schooling on wage rates and hours supplied to the paid labor market which do not control for household fixed effects are quite misleading. The directions and magnitudes of biases caused by failing to take control measures differ according to sex between wage rates and hours supplied and among schooling levels. Specifically ordinary least squares estimates of the impact of schooling are biased to overestimate returns to school returns to lower schooling levels relative returns of schooling for males relative to females and the positive impact of schooling on hours supplied to the paid labor force. Advocating schooling investments in other societies based upon the standard estimates may therefore create unrealistic expectations.