Descriptive analyses undertaken on a sample of pensioners born in post-colonial Tunisia and having worked in the Paris region from the 1960s to the end of the 1980s highlight large wage variations according to individuals' gender and nationality. Although controversial, regression models prove particularly useful for decomposing the disparities observed within our sample. Within this family of models, multilevel regression models lead to results somewhat different from those obtained using conventional models: in particular, they show that the effects of pure gender- and nationality-related discrimination tend to disappear as we control more precisely the effects of endogeneity and unobserved heterogeneity. Whatever the chosen model, regression analysis conversely underlines the decisive influence of the occupational space on wage differentials - sectoral differences playing a negligible role in this regard compared to the effect of the market segment.

[No Keywords available]

[No Keywords available]