Discriminating factors of women's employment

Abstract

Italy exhibits a dramatic level of territorial heterogeneity in terms of socio-economic dynamics and in the economic position of women. We employ this territorial variance to assess the impact of selected policies and institutions on men's and women's employment using microeconomic data. Such an analysis provides results partly different from what was expected on the basis of cross-country aggregate evidence on industrialized countries. Aggregate growth and tertiarization of the economy are surprisingly found beneficial only to men's employment, while culture and discrimination are relevant for women's. Social Assistance is found highly significant too, with the provision of services being more beneficial to women's employment than monetary transfers.

Acknowledgements

We wish to thank Anna Conte, Marcella Corsi, John D. Hey, Federico Lucidi, Furio C. Rosati, Annalisa Rosselli for fruitful discussions and for their help. All remaining errors are responsibility of the authors only.

Notes

¹For extensive evidences of socioeconomic differences within Italian regions, see Putnam (1993) and Tabellini (2006).

²Researches showed that countries with more liberal attitudes toward gender roles, higher work orientation of women and higher acceptance towards young children's working mothers, show higher women's employment rates (compare Reimers, 1985; Vella, 1994; Pencavel, 1998; Antecol, 2000; Fernández et al., 2004; Fernández, 2007).

³The role of social policies in facilitating the conciliation of work and family lives was analysed, for example, by Anderson and Levine (1999), Conelly (1991, 1992), Hofferth and Wissoker (1992), Kimmel (1995) and Powell (1998). Apps and Rees (2004) note that countries supporting motherhood by means of childcare facilities rather than monetary benefits exhibit both higher rates of women's labour supply and higher fertility rates.

⁴This last index is different from the Human Development Index (HDI) in two respects: on one side aggregation is obtained by harmonic instead of arithmetic mean, in order to prevent Regions with high unbalances in the component variables from obtaining extreme values of the index, and, in turn, to rank higher those Regions exhibiting a more equilibrated development of all the selected variables (Casadio and Palazzi, 2004). On the other side, the variables chosen for the construction of the Modified Human Development Index (MDI) are slightly different from those of the standard HDI in order to better catch relevant differences among Italy's Regions, and to avoid the inclusion of some variables that in our framework have instead been adopted as micro determinants in the probability of employment.

Table 1. Selection and aggregation of context variables

⁵Nardo et al. (2005), Nicoletti et al. (1999), Kline (1994).

⁶We used the GLAMM module for STATA 9 (compare Rabe-Hecketh and Skrondal, 2005).

⁷We also estimated a fixed-effect probit model including the same first-level variables: we find that under the multilevel model, the coefficients of individual and household variables increase in absolute magnitude, and some in statistical significance. In particular, the impact of family-related variables strengthens for both women and men, enhancing the estimated impact of traditional gender roles and sexual division of labour on the probability of being employed. Further results are available from the authors upon request.

⁸In other words, the RI model can be considered as a special case of the RC, where the indirect effects are constrained to zero. In our estimates, interaction effects – i.e. the impact of contextual conditions on the coefficients of the variables denoting individual characteristics – are significant, especially for women (detailed results are available upon request). Focusing on women's sample, RCP emerges as the macro index that most significantly affects the impact of being married and of co-living with an old-aged person. Interestingly, the extent of private provision of services differently impacts on women's and men's employment: it significantly affects the probability of employment for married women, while it impacts only on men co-living with an old-aged person.