Foreign Capital and Gender Differences in Promotions: Evidence from Large Brazilian Manufacturing Firms

Widespread consensus asserts that occupational segmentation is the chief determinant for gender differences in wages in the Brazilian labor market (see, for example, Oliveira 2001). Although women represent 39 percent of the formal workforce in Brazil, they amount to only 23 percent of the employees in the sectors that pay relatively higher wages in 2004. As well, the predominance of segmentation increases if one restricts attention to managerial positions, of which women hold only 14 percent. This suggests that limited promotion prospects for women in the Brazilian job market may present barriers to their promotion to the upper rungs of the corporate ladder.

This paper contributes to the literature by examining data from the largest firms of the Brazilian manufacturing industry. In particular, our goal is to determine whether obstacles to women’s ascension exist by tracking the amount of time it takes for a woman to get a promotion to a managerial position. The aim is to complement the literature on gender differences in promotions, whose papers mainly focus on computing the promotion rates. Time to promotion offers a different angle in that samples with a large time span could well present similar likelihoods of promotion for both genders even if the promotion durations are different. We construct the duration variable by following from January 1996 to December 2005 every individual who enters the Annual Report of Social Information (RAIS) database between January 1991 and December 1995. We denote as nonpromoted the individuals who do not obtain a promotion to a managerial position within the sample (that is, the right-censored observations), whereas we assign duration zero to the workers who already are in managerial positions in January 1996. Accordingly, we label as promoted the individuals within our sample who rise to a managerial position (that is, those with uncensored positive durations). Investigating differences in promotions is particularly difficult because observationally similar male and female workers may display equal promotion rates and durations even in the presence of gender differences (for example, promotions may differ in quality).

As far as we know, this is the first study to examine gender differences in promotions using microdata from a developing country. Apart from the obvious interest in finding whether the main stylized facts hold in a Latin American country, we were also motivated by the quality and availability of the Brazilian data. In contrast to the many studies that employ data from individual firms (see, for example, Cabral, Ferber, and Green 1981; Gerhart and Milkovich 1989; Baker, Gibbs, and Holmstrom 1994), our study relies on a homogeneous sample of recently hired workers from the Brazilian manufacturing industry in the period from 1996 to 2005. Our data set is particularly convenient. First, the data include a wide array of controls for worker and firm characteristics. Second, unlike Blau and deVaro (2007), we observe multiple workers per establishment at different occupations and hierarchical levels as well as their career paths in terms of occupation and wage. A potential limitation is that our data set does not include any direct measure of on-the-job productivity and hence we must come up with indirect controls.

Interestingly, we find that there are significantly fewer gender differences in the time to promotion within foreign-owned firms than within domestic firms in the Brazilian manufacturing industry. Our findings complement to some extent the literature on the differences between multinationals and domestic firms (Doms and Jensen 2006; Greene and others 2009) as well as the evidence that gender differences may depend on the nature of the firm.¹ We argue that such gender differences are consistent with statistical discrimination and self-selection. We reason as follows. Suppose there are relatively more women than men who prefer to dedicate more time to their family than to their careers. Berk (2001) shows that gender differences in average career concern do not ensure statistical discrimination against the group with lower average career concern if employees optimally choose which jobs to apply for. A career-minded woman who applies for a job in a firm...