英语论文网

ing this particular research technique.

SECTIONS: Literature review, methodology, software, empirical methods,

Empirical procedures

Discrimination has caused a lot of social distress throughout the history of humankind. Its existence is still prevailing because humans tend to base their judgment on prejudice and stereotype. In labour economics, when membership in a particular group conveys valuable information about person’s skills or productivity, another individual with no personal preconception might still be inclined to think it is rational to statistically discriminate (Oaxaca and Dickinson, 2006). However, the income inequality is not something that cannot be eradicated to some extent. It is one of the policy and legislation objectives which nowadays economies have to follow to decrease the gap between the genders and races (Jones, 1984). It has become less challenging to detect discrimination and understand the extent of this issue as more and more micro data is readily available. A wide range of empirical techniques have also been used to analyse the gender or race pay gap and to explore the magnitude of discrimination. The aim of the empirical methodology is to control for productivity related factors as much as possible, including ones for which data is conventionally not available (Gunderson, 2006).The main methods, their implementations, as well as strengths and weaknesses, are described here as an introduction to the evidence on gender pay gap in the subsequent sections.

Regression decompositions

The traditional modelling of the gender wage gap is mainly based on the distinction between the productivity and discrimination elements, and often attempts to estimate the relative proportions of the two. This methodology was consolidated by the utilisation of the Oaxaca-Blinder decomposition procedures (Blinder, 1973; Oaxaca, 1973), and remains to be used with more sophisticated derivations (Blau and Kahn, 1997; Juhn et al., 1991, 1993; Nielsen, 2000; Oaxaca and Ransom, 1994, 1999). It has become so widely used among the econometricians because this method is employed to decompose the pay differential (or any other relevant differential) into one accountable for differences in production related characteristics and the other attributable to differences in expected returns to characteristics (Breunig and Rospabe, 2005). The first component of the prominent “two-fold” decomposition (note that a “three-fold” decomposition is also possible) is comprised of differences in productivity-related pay-determining factors such as level of education, experience, and health status as well as control factors such as occupation, industry, and firm size (Gunderson, 2006). The second term denotes the differences in wage when males and females have the same productivity attributes. For this specific reason the second summand is conventionally referred to as a measure of pay discrimination.

The major benefit of Oaxaca-Blinder decomposition is the fact that it helps in identifying the causes of the wage gap, so it can be used in gender policy applications. This technique is parametric and is based on the average wage density function which could be both an upside and a downside. It is relatively easy to implement and gives a mean market outcome. However, some would argue that the downside of a parametric approach is that it only looks at the mean values and does not