著者

杉橋 やよい

出版者

法政大学経済学部学会

雑誌

経済志林 = 経済志林 (ISSN:00229741)

巻号頁・発行日

vol.76, no.4, pp.53-79, 2009-03-09

---

The purpose of this article is to examine both a virtue and inherent limitations in the Blinder-Oaxaca Decomposition Technique, the most popular econometric method for analysing gender wage differentials. The technique is used to decompose the gender wage gap into two parts: the gender differences in endowment and the differential remuneration of that endowment. The second part is assumed to reflect 'discrimination'. The method has a virtue in quantifying the differences in characteristics and 'discrimination' within the gender pay gap. However, there are some critical limitations that are inherent in the method. (1) The extent of the gender differences and the extent of 'discrimination' depend on variables included in the wage equations.(2) The technique ignores discrimination within or outside the labour market, or women's reaction to discrimination in the workplace, thereby decomposing the gender pay gap into two parts. In other words, the Blinder-Oaxaca decomposition method implicitly assumes that wage gaps arising from differences in endowment are separate from those arising from labour market discrimination. The inclusion of variables that reflect labour market discrimination, such as occupational segregation, increases the estimated effect of 'gender difference in endowment' and decreases the estimated effect of 'discrimination'.(3) The explanatory variables are not independent, which causes some degree of multicollinearity.(4) Breaking down the gender pay gap by individual variables is misleading since the size of the constant and the contribution of the dummy variables in the unexplained portion of the model are influenced by the choice of the base categories for the dummy variables.