UC Berkeley

This dissertation addresses empirically, theoretically and experimentally issues of business taxation. The first chapter studies firms’ margins of response to a historically large payroll tax cut that affects a subset of Brazilian firms. Difference-in-differences estimates based on plausibly exogenous legal variation indicate that the payroll tax reduction causes an increase in employment, wages, and profits, while capital decreases. Responses are substantially more pronounced among small firms, and workers’ earnings gains are concentrated at the top of the distribution. Reduced-form estimates reveal that consumers pay 65% of payroll taxes, firm owners 23%, and workers 12%. These results establish not only that payroll tax cuts primarily benefit consumers, but also exacerbate within-firm earnings inequality. This evidence cannot be reconciled within a competitive framework. I estimate a model that allows for product and labor market power to explain these findings.

The second chapter, in joint work with Thiago Scot and Pedro Zuniga, we study corporate responses to a minimum income tax, using the universe of corporate tax filings in Honduras. The policy design allows us to separately estimate cost misreporting under profit taxation and the elasticity of reported revenue. Large corporations overreport true costs when taxed on profits. Taxing revenue leads to a substantial decrease in reported revenues: we estimate an elasticity in the range 0.35-1. The elasticity of revenue is at tenuated when third-party information on the revenue of firms is available, suggesting misreporting plays an important role. Our results inform trade-offs when broadening tax bases to curb evasion.

The third chapter of this study utilizes an experiment to examine the consequences of violating the stable unit treatment value assumption (SUTVA), a fundamental concern in the empirical evaluation of business tax policy. One significant issue is that changes to the tax code often generate general equilibrium effects, potentially biasing empirical estimates of tax responses. To study this risk, I collaborated with David Holtz, Ruben Lobel, Inessa Liskovich, and Sinan Aral to design an experiment specifically aimed at quantifying the biases introduced by general equilibrium spillovers. Cluster randomization, i.e., the practice of randomizing treatment assignment at the level of “clusters” of similar individuals, is an established experiment design technique for countering interference bias in social networks, but it is unclear ex ante if it will be effective in marketplace settings. In this paper, we use a meta-experiment or “experiment over experiments” conducted on Airbnb toboth provide empirical evidence of interference bias in online market settings and assess the viability of cluster randomization as a tool for reducing interference bias in market-place TATE estimates. Results from our meta-experiment indicate that at least 19.76% of the TATE estimate produced by an individual-randomized evaluation of the platform fee increase we study is attributable to interference bias and eliminated through the use of cluster randomization. We also find suggestive, non-statistically significant evidence that interference bias in seller-side experiments is more severe in demand-constrained markets, and that the efficacy of cluster randomization at reducing interference bias increases with cluster quality.