Estimating wage functions and wage discrimination using data from the 1995 Swiss labour force survey: a double‐selectivity approach

Abstract

The data from the Swiss Labour Force Survey (SAKE) have been widely used to estimate wage functions, which in turn have been applied for the determination of wage discrimination between genders. One serious problem with the SAKE data is that about 17 per cent of employed individuals did not report wages. Those studies which use the SAKE data to estimate wage functions simply ignore these non‐respondents. Such an approach could lead to a serious selectivity bias if the response decision is not purely random. In this study this issue is analysed in a double‐selectivity framework, in which both this response decision and the usual market‐participation decision are modelled. Although the response decision can be partially explained by certain socio‐economic variables, a large degree of randomness/unexplained variation exists. The authors therefore conclude that, in the absence of a better model, the standard approach at estimating wage functions (i.e. only correcting for the selectivity bias arising from women’s participation decision) is the most appropriate one.