Working Paper:

  • Does Labor Market Concentration Decrease Wages? Evidence from a Retail Pharmacy Merger (Job Market Paper)

"Increasing labor market concentration is a topic of recent policy concern. However, there is lack of direct evidence showing how labor market power affects workers. In this paper, I study the effects of a large merger in the retail pharmacy sector in Brazil on workers’ wages. I use a matched employer-employee dataset from 2007 to 2018 that allows me to follow workers and establishments over time. The effect of market power is estimated using a difference in difference (DD) approach that compares the labor market effects of the merger in regions where firms overlapped and concentration increased relative to those in which it didn’t. To account for composition effects, I implement a DD estimator that includes establishment and worker fixed effects. I find that increasing market power lowers wages, but less than previously thought. There are two reasons for that. First, failing to account for composition effects biases the estimates of the effects of concentration. Second, the negative labor market effects of a merger are offset by competitors’ responses. The results show a great deal of heterogeneity across occupations and worker tenure, highlighting the benefits of studying a particular industry."

Work in Progress:

  • Effectiveness and Consequences of Occupation and Industry-specific Minimum Wages (with Joaquin Serrano)

Overview: We study the effects of minimum wages on employment levels and on the wage distribution. Most empirical studies in the minimum wage literature use county-level variation and ignore that different firms will respond to minimum wages in different ways. In theory, the employment response of each firm will depend on three conditions: (i) how binding is the minimum wage; (ii) how competitive are the labor markets that the firm participates; and (iii) how substitutable are different types of labor within each firm. In this paper, we evaluate the effects of occupation and industry-specific minimum wages in Brazil on employment and wages. Using an employer-employee matched dataset (RAIS) we estimate a reduced form equation that incorporates a measure of bindingness and a measure of labor market concentration. We then use the variation in minimum wages across occupations to verify the role of substitution in determining the minimum wage effects. Preliminary results show the importance of controlling for the bindingness of the minimum wages when determining its effects on wages and employment. Furthermore, zero and positive employment effects of the minimum wage are associated with some measures of labor market concentration.

  • Defining Labor Markets from Worker Flows (with Ryan Boone)

Overview: We use worker flows to define better approximations to labor markets. Labor market definitions are important to predict the effects of mergers, labor market policies (like minimum wages), migration, trade shocks, and other events on labor market outcomes. Researchers have typically relied on ad-hoc approximations such as geographic boundaries, occupational codes, or industry codes at differing levels of granularity. For any given job, it is not clear which definition is most appropriate. For example, the relevant market for airline pilots might be a detailed occupation-industry cell at the national level. On the other hand, the relevant market for pharmacy cashiers might consist of many occupations and industries but only within a narrow geographic area. The paper follows three steps. First, we use modularity maximization approaches from the network literature to identify labor markets. We use worker flows across occupations, industries, and geographies to define N markets. The defined markets maximize the modularity measure, i.e., the measure of the density of links within markets. Second, we describe and evaluate the resulting networks. For example, we show that labor markets often cross one-digit occupation codes. We are currently working on the third step, in which we validate the labor market approximation by applying the definitions to empirical frameworks.

  • Has the College Premium really Flattened? The Entry of New Schools (with Ariadna Jou)

Overview: The ratio between the average wages of individuals with a college degree and the average wages of individuals with a high school degree has flattened in the last two decades in many Latin American countries (Fernandez and Messina, 2018). Economists commonly use this statistic as a proxy for the college premium, i.e., the counterfactual wage that a worker with a high school degree would earn if she were to graduate in a bachelor’s program. However, this flattening happened at the same time as new schools entered the education market. This paper shows that the college premium in Brazil has not flattened, instead it is increasing. We do this by matching novel data with the names of around one million college graduates from 50 schools and 20 different cohorts with the RAIS, a Brazilian employer-employee matched dataset. First, we show that students from incumbent schools perform better in the end-of-degree standardized exams, which suggests that these are better schools. Second, we show that new cohorts of students graduating from incumbent schools receive higher wages than previous cohorts and same-age workers from other schools. Combining these two facts, we infer that a composition change in the labor force is driving the flattening of the college premium proxy. There are more workers with a college degree, but with lower quality degrees, which reflects into lower average wages. The findings in this paper are relevant for any study that uses college premium proxies in countries, or periods of time, with increasing access to lower quality colleges. More precisely, we are concerned that general equilibrium effects of access to college have been overestimated by not accounting for these changes in quality.


  • Modernização da Agricultura em Moçambique: determinantes da renda agrícola, with Carlos E. Guanziroli. 2015. Revista de Economia e Sociologia Rural, V53, S1. DOI:10.1590/1234-56781806-94790053s01009