Projects
📃Research Statement
Projects
📃Research Statement
Job Market Paper
Taxing the Untaxable: Cryptocurrencies, Tax Evasion, and Optimal Policy
Presented at KDSA 2025 (Best Presentation Winner Award)
This paper develops a search-theoretic general equilibrium model of money to analyze optimal fiscal and monetary policy when tax evasion is linked to payment methods. In a tax evadable market, a sales tax is automatically collected on monitored electronic money but can be evaded using unmonitored cash. Buyers endogenously choose payment methods, determining the effective tax base. A benevolent Ramsey planner optimizes the sales tax, cash audit intensity, and fiat money growth rate, balancing revenue needs against evasion incentives and distortions. Cryptocurrency splits into centralized exchanges (taxable, with tax pass-through shaped by regulatory clarity; CEX) and decentralized exchanges (where the government does not have the technology to tax; DEX). The policy regime combines regulatory stance and rule clarity, which together affect price risk and the share of transactions routed through CEX. The analysis shows: (i) clear rules raise pass-through and revenue without large evasion; (ii) restrictive cryptocurrency regulations can backfire by shifting flow to DEX and amplifying risk; (iii) revenue collected on CEX is hump-shaped in as regulations tighten; and (iv) credibility matters, since policy surprises that shift the perceived regime reduce future tax capacity. The framework delivers implementable comparative statics and policy mixes tailored to country characteristics.
Privacy Concerns in Digital Currency Adoption
Presented at Summer Workshop on Money, Banking, Payments, and Finance (Federal Board-2024), Midwest Macro Meetings (2024), ESA (2024), SEA (2024), KDSA (2023) (Best Presentation Winner Award)
The digital transformation of money has introduced new dynamics in how currencies are adopted and used, especially concerning privacy. This research investigates the interplay between privacy concerns, transaction histories, and price discrimination in the adoption of new digital currencies. The framework is based on a multiple currency monetary search model where agents’ decisions determine the acceptance and the value of the currencies. Buyers are heterogeneous in their valuations over the goods and sellers have the ability to engage in price discrimination based on transaction histories. Theoretical findings indicate that higher privacy preferences can increase cash usage, however buyers may strategically switch payment methods to avoid future price discrimination by concealing their types. The theory is coupled with laboratory experiments which vary factors such as the value of privacy, the magnitude of the valuation shocks and transaction costs.
On Cross-Border Payments and the Industrial Organization of Correspondent Banking
with Garth Baughman, Cathy Zhang, Ashley Zhao
Presented at the Bank of Canada (2025)
Despite advances in domestic payments arrangements, cross-border payments remain costly and slow. This paper builds a model of cross-border payments where market power in correspondent banking relationships reduces efficiency. We consider two policy experiments: the introduction by one country of an internationally held CBDC, and development of an internationally interoperable settlement system. Both policies can at least attenuate the inefficiencies in cross-border payments, but neither is automatically a complete solution nor are they without difficulties. For an international CBDC, we identify a political economy barrier: banks of the country potentially introducing the CBDC disproportionately suffer losses while benefits tend to accrue to foreign depositors, so a central bank concerned with its domestic banks' profitability may be unlikely to take such an action. Interoperability does not face the same barrier, as benefits accrue more symmetrically, but we argue that technical and other issues inherent to interoperability may be difficult to overcome.
Liquidity, Networks, and Financial Stability: A Model of Monetary Policy and CBDC Flows
📜Draft SOON
I study whether a potential retail CBDCs information flows about which banks customers are converting deposits into CBDC, can help spot and contain bank runs in a system with many banks. This differs from FedNow/Fedwire, which record transfers between banks rather than households pulling funds from a particular bank. In a simple network model, I show that better-connected banks require less support, and that bank-specific outflow thresholds can trigger earlier action and limit knock-on effects across the system which cannot be captured with a single-bank setting. I also examine a big-bank scenario where people move to a dominant private bank instead of CBDC and explain why CBDC data alone can then miss stress and how to adjust for it. With simulations I show these mechanisms and map policy choices to the risk of runs and the size of support needed during stress.
Macroeconomic Policy When Inequality is Taken Seriously
with James Bullard
Our focus is on economies with non-state-contingent nominal contracts (NSCNC) and heterogeneous agents. We define a nominal-risk wedge that measures the gap between exante real returns and expected growth, showing this wedge is the key sufficient statistic for welfare losses. Under optimal price-level targeting, the wedge is zero and the first-best allocation obtains. Realistic Taylor rules generate a non-zero wedge with closed-form variance expressions under AR(1) processes. We derive formulas for optimal Taylor rule coefficients that depend on the covariance between inflation and growth shocks. Importantly, uninsurable idiosyncratic risk amplifies the welfare cost of the wedge but does not change the policy target, bridging representative agent and heterogeneous agent frameworks.
Work in Progress
Experimental Evidence on Currency Competition in Integrated Economies
with Marcos Cardozo, Yaroslav Rosokha, Cathy Zhang
When Do Managers Delegate to AI? Task Type, Stakes, and Accountability in Human–AI Assignment
We examine how accountability shapes the delegation of tasks to AI. In many industries, penalties for AI errors are more severe than for equivalent human mistakes. We model how firms readily delegate low-stakes tasks to AI but avoid it for medium-stakes tasks where the liability costs outweigh the performance benefits. For the highest-stakes tasks, however, "human-in-the-loop" systems, where a human verifies an AI’s proposal, become optimal. This framework helps explain differing AI adoption rates across sectors and points to specific policy levers.
Racing to the Edge: AI Development, Alignment, and Catastrophic Risk
This project models the collective action problem in AI safety. I argue the key issue is the distinction between two risk types: firm-specific disasters (like lawsuits), which firms are incentivized to prevent, and systemic catastrophes (like existential risks), which create a large negative externality. We show that firms do manage private risks, however, for systemic risks since they can only internalize a fraction of their safety investment’s social benefit they under-invest, which worsens as competition intensifies. When safety and capability are complements but firms treat them as substitutes, this dynamic pushes firms toward a "catastrophic frontier" where AI progress outpaces alignment. The framework provides a basis to evaluate competing policy interventions such as minimum safety standards, liability rules, and research subsidies.