"The Gender Gap in Earnings Between and Within Firms: Evidence from Linked Employer-Employee Data*"

This paper investigates the role of firms in determining the gender gap in earnings on average and at different quantiles of the earnings distribution. Using a linked employer-employee dataset for Italy, we show the existence of firm-specific premia which differ across gender and explain on average 30% of the gender pay gap in the period 1995-2015. We decompose the gap in premia into sorting of women across firms and differences in bargaining power within firms by means of a standard Blinder-Oaxaca decomposition. We find that sorting explains a larger fraction of the gender pay gap than bargaining on average and at the bottom of the distribution, whereas bargaining dominates at the top. This is consistent with what we find across different occupations: sorting explains a larger fraction of the gender pay gap for blue- and white-collar workers, whereas bargaining dominates for managers. The two channels of sorting and bargaining have similar weight in explaining gender differences in pay in the early stages of workers' careers, but for older cohorts sorting dominates. On the other hand, bargaining has increased in importance over time. We replicate our decomposition exercise in four overlapping intervals and find that while sorting is the only driving force of the firm contribution to the gender pay gap in 1995-2000, bargaining explains as much of the firm contribution as sorting in 2010-2015. We explain the rise in importance of bargaining as the result of an evolution in the Italian labour market that (i) favoured an increase in female labour force participation, (ii) introduced mechanisms of decentralised wage setting. We propose sorting as the outcome of gender differences in mobility rates and provide evidence that women have a lower probability of moving towards firms with higher premia. We show that this gap in the probability of moving to higher premia firms is larger for high-skilled women, but smaller for women moving from high to low risk firms, where we defined riskiness of firms based on the standard deviation of their wages. Finally, we exploit a natural experiment on the gender composition of corporate boards and find no evidence that an increase in female presence in boards reduces differences in bargaining power between men and women.

*This paper is based on joint work with Alessandra Casarico, Bocconi University.