Research
Working papers
Signaling within the Firm [JMP] – PDF
Within-firm promotions are a large component of wage growth, yet we do not know much about how firms decide how to allocate them. Does uncertainty about worker ability play an important role in this process? Using Portuguese administrative data, I investigate the role of overtime work on careers in retail and hospitality. In particular, I ask if overtime hours are used as a signaling device inside the firm. First, I develop a model of the principal-agent interaction within the firm that investigates overtime as a mechanism for selection: workers are heterogeneous in their cost of working overtime, and the firm wants to promote those for whom such cost is lower. As overtime pay decreases, the opportunity cost of working overtime increases: overtime hours become more informative for the employer and this improves selection of promoted workers. I show the model predictions match the data. First, long working hours are correlated with future promotions. Next, I exploit a 2012 reform that reduced the overtime pay premium, creating a quasi-exogenous shift in the signaling value of overtime. The reform lead to a decrease in the number of overtime hours, particularly among workers who were about to be promoted. Importantly, workers who are promoted after the reform are better selected than workers who were promoted before the reform, in terms of total number of promotions, wage growth, and retention. These results confirm the model's predictions and show promotions respond, at least in part, to the firm's selection motive.
What Works for Working Couples? Work Arrangements, Maternal Labor Supply, and the Division of Home Production (with Martina Uccioli) – PDF
We document how a change to work arrangements reduces the child penalty in labor supply for women, and that the consequent more equal distribution of household income does not translate into a more equal division of home production between mothers and fathers. The Australian 2009 Fair Work Act explicitly entitled parents of young children to request a (reasonable) change in work arrangements. Leveraging variation in the timing of the law, timing of childbirth, and the bite of the law across different occupations and industries, we establish three main results. First, the Fair Work Act was used by new mothers to reduce their weekly working hours without renouncing their permanent contract, hence maintaining a regular schedule. Second, with this work arrangement, working mothers' child penalty declined from a 47 percent drop in hours worked to a 38 percent drop. Third, while this implies a significant shift towards equality in the female- and male-shares of household income, we do not observe any changes in the female (disproportionate) share of home production.
Work in progress
Full-time Mothers, Part-time Workers (with Martina Uccioli and Valeria Zurla) - Selected project for VisitINPS Fellowship on gender inequalities
We study indivisibility of labor as key determinant of the choice of mothers to return to work after giving birth. In Italy, new mothers have to take five months of mandatory
leave. In addition, parents are allowed up to 10 more months of leave. A 2015 law (i) gives parents the possibility of taking the voluntary leave on an hourly rather than a daily basis, and (ii) allows parents to turn a full-time contract into a part-time contract for any remaining months of voluntary leave. By comparing new parents before and after the law, we can study whether these provisions change leave length. We can then assess the effect of leave length and part-time work on the child penalty. This could go in either direction, depending on whether the compliers are mothers that in absence of the law would have worked full time or not worked. In order to disentangle the net effect into the two different treatment margins, we rely on an instrumental variable approach: the fraction of co-workers who chose different arrangements after childbirth in the previous years can be used as an instrument for individual choice, separately for the three possible choices (not working, working full-time, working part-time), and hence for the difference in utility cost of any two options.