​​

• Modeling the Distribution of Tax Evasion: Implications for the Tax Gap (Job Market Paper) with Andrea Modena. Selected for presentation at 40th European Economic Association (EEA), Bordeaux, France (Aug. 25, 2025) & 31st International Conference on Computing in Economics and Finance (CEF), Santiago, Chile (July 7, 2025). 

Understanding the drivers of tax evasion is critical for designing effective policies to reduce the persistent gap between taxes owed and collected. This paper develops a dynamic heterogeneous-agent model in which taxpayers choose between simple and sophisticated evasion under income shocks and enforcement risk. Sophisticated evasion becomes attractive only for high-income individuals who can absorb its fixed cost and exploit legal ambiguities in the tax system, creating a productivity-dependent threshold that shapes non-compliance across the distribution. Calibrated to U.S. data, the model reproduces the aggregate tax gap and its concentration at the top. Simulations show that traditional deterrence tools, such as audits and fines, reduce simple evasion but have little influence on sophisticated evasion, whose incentives are largely insulated from enforcement intensity. As a result, the remaining tax gap becomes increasingly composed of sophisticated, harder-to-detect behavior among high-income taxpayers. Greater income dispersion expands the set of individuals for whom sophisticated avoidance is accessible, which flattens the fiscal-capacity (Laffer) curve and lowers the revenue-maximizing tax rate under imperfect detection. These findings highlight that effective policy must target the mechanisms that allow avoidance to appear legitimate, such as legal ambiguity or insufficient fine-enforcement credibility, rather than relying solely on traditional deterrence instruments..

​​

• The Offshore Rotation: Which Domestic Risks Shift Deposits from Safe Banks to Tax Havens (Preliminary Title)

Foreign bank deposits are a standard vehicle for liquidity, safety, and precautionary saving, yet residents under domestic stress often hold them in tax havens rather than only in transparent foreign banking centers. This paper studies this compositional reallocation, or rotation: which domestic risks shift residents' foreign deposits toward tax havens rather than ordinary foreign banks? Using quarterly bilateral non-bank deposits from the BIS Locational Banking Statistics for 168 source countries over 1995-2020, I compare destinations within the same source country and quarter. Market-priced indicators of domestic risk, including sovereign-yield volatility and the real sovereign yield, rotate deposits toward havens relative to safe transparent foreign banking centers. When priced risk rises, havens gain relative to the destinations that a pure safety, liquidity, or precautionary motive would favor. Slow-moving balance-sheet stocks, such as government debt over GDP and corporate and household loans, do not generate the same rotation and instead resemble a standard safe-bank response. The haven differential builds over time and is strongest for continuous measures of priced risk. A stylized portfolio model clarifies the test: foreign-safety demand should favor safe transparent banks, whereas jurisdictional-insulation demand predicts haven gains relative to those same centers. The evidence is consistent with haven demand reflecting jurisdictional insulation, not only foreign safety, liquidity, or yield.

​

• Tone Asymmetries in IMF Board Discussions: An Machine Learning Approach with J. Araujo, M. Rousset, and S. Chauhan.

Using supervised and unsupervised machine-learning techniques, we analyze publicly available IMF Executive Board transcripts (2005-19) to gauge perceptions of lack of evenhandedness in policy and country discussions.

​

• Political Bureaucrats? Evidence from Chile on Political Partisanship in Central Banking with Y. Carriere.

Do political views influence the deliberations of independent technocratic institutions? Using newly released transcripts from the Central Bank of Chile’s monetary policy meetings (2000–2012), we analyze over two million words exchanged between board members and staff economists. Chile’s central bank provides a unique setting: despite strong formal independence, its board appointments rotate among the main political parties, allowing us to infer members’ political leanings. Using natural language processing and topic modeling methods inspired by Hansen, McMahon, and Prat (2018), we associate speech with individual board members and study how discussion priorities vary by political affiliation. While all members emphasize price stability, those aligned with left-leaning parties refer more often to domestic demand and labor conditions. Applying Gentzkow and Shapiro’s (2010) techniques, we find that polarization in speech rises during periods of policy trade-offs, such as the 2006–08 commodity price surge, and declines during synchronized downturns like the Global Financial Crisis. Overall, political diversity shapes the tone and focus of deliberations but does not translate into systematic differences in voting behavior.

​

• Climate Change and the Macroeconomic Efficiency of Public Spending with D. Riera-Crichton.

This paper looks at the heterogeneous efficiency gains of public spending in infrastructure in countries with varying degrees of vulnerability and readiness to adapt to climate change.  Using a local projections model, our empirical model finds a multiplier effect of up to 2.5 dollars in infrastructure investment over a four-year period in countries with lower scores in both vulnerability and readiness to climate change adaption. We find heterogeneous results in countries with different levels of both readiness and vulnerability, finding lower multiplier effect results in countries with lower vulnerability scores. Overall, given other possible multiplier effects from other investments, we find that infrastructure investment is a very cost-efficient way to ensure long output gains in countries that are ranked as very vulnerable to the effects of climate change.

​

​