The make-up of a regression coefficient: An application to gender

In this paper we illustrate the potential use of an old/new methodology which combines the use of concentration curves in order to investigate the components that make up a regression coefficient. The illustration is based on examining gender differences in the effect of age on labor market participation in Italy. Women participation rate is substantially lower than men, but their age profile is similar. The most striking difference is in terms of hours of work: while Italian men increase their work effort until the age of 35, Italian women reduce it until the age of 39. These results do not differ substantially when we split the working population into employed and self-employed. Earnings increase with age for both men and women, however the local regression coefficient is negative for Italian women in the age of 38–42. This evidence is accentuated when we focus on the employees.