Trying to ascertain scientifically the existence and importance of gender-based salary discrimination runs into conceptual and technical difficulties, the implications of which are not always fully appreciated: depending on the approach used, our analysis of a large sample for Switzerland in 1997 concludes that there is significant discrimination against women; or that no discrimination can be detected; or that there is significant discrimination against...men. The approaches used are: direct general regression (for given attributes such as qualification levels, do salaries differ according to gender ?); inverse general regression (for a given salary level, do employers require different attributes, such as qualification levels, depending on gender ?); the Oaxaca-Blinder method (the effect of a given attribute - e.g. age or marital status - on salaries may differ depending on gender and one must therefore estimate and compare separate equations for women and for men); Heckman's correction (focusing on working persons only may give rise to a selection bias). Our general conclusion is agnostic : given the present state of techniques and knowledge, it is not (yet ?) possible to ascertain scientifically the existence of gender-based salary discrimination in Switzerland. This conclusion differs from that reached by pre-existing studies, which tended to underpin the existence of salary discrimination against women.
More papers in Cahiers de Recherches Economiques du Département d'Econométrie et d'Economie politique (DEEP) from Université de Lausanne, Faculté des HEC, DEEP Address: Université de Lausanne, Faculté des HEC, DEEP, Internef, CH-1015 Lausanne Contact information at EDIRC. Series data maintained by Claudine Delapierre Saudan ().