On the standard errors of Oaxaca-type decompositions for inter-industry gender wage differentials

Horrace and Oaxaca (2001) treat the regressors in gender wage gap by industry measures as non-stochastic when computing the corresponding standard errors. However, the non-stochastic regressors assumption is thought to be inappropriate in modern econometrics. In this paper, we derive the correct standard errors for the measures proposed by Horrace and Oaxaca (2001). We then empirically apply the derived correct standard errors in regard to the March 1998 Current Population Survey data adopted in Horrace and Oaxaca (2001), as well as the Manpower Utilization Survey in the Taiwan area conducted by the Census Bureau over the years from 1978 to 2003. The empirical results suggest that the researchers would be better to use the correct standard errors derived in this paper, accompanied by the White correction, to arrive at a more accurate statistical inference. I would like to thank an anonymous referee of this journal and Kevin Hallock (Associate Editor) for helpful comments and suggestions. I also would like to thank William Horrace and Hwei-Lin Chuang for kindly providing the data. Ta-Sheng Chou and Jiu-Yuan Chen offered helpful research assistance. Financial support from the National Science Foundation in the form of grant NSC-95-2415-H-007-008 is gratefully acknowledged. Of course, all errors are mine. Citation: Lin, Eric S., (2007) "On the standard errors of Oaxaca-type decompositions for inter-industry gender wage differentials." Economics Bulletin, Vol. 10, No. 6 pp. 1-11 Submitted: January 12, 2007. Accepted: May 1, 2007. URL: http://economicsbulletin.vanderbilt.edu/2007/volume10/EB-07J00001A.pdf