Abstract: Global banks rely heavily on US money markets for short-term dollar funding. Yet, post-financial crisis regulations — designed to shield domestic investors from default risk — have constrained banks’ access to this crucial funding source. This paper uses novel quantities data to show that foreign exchange (FX) swaps emerge as alternative (“synthetic”) funding instruments when US money market funds reduce wholesale funding to banks. The resulting shift in banks’ demand for FX swaps leads to substantial deviations from covered interest parity (CIP) – the breakdown of a fundamental no-arbitrage pricing condition. Using an instrumental variables strategy that exploits idiosyncratic variation in the availability of wholesale dollars to banks, I show that (i) banks’ swap demand causes CIP deviations to worsen, and (ii) non-bank investors’ inelastic demand triggers spillover effects on their FX hedging costs. I use my empirical estimates to calibrate a model in which global banks optimally choose FX swaps to offset shortfalls in wholesale funding, generating CIP deviations in equilibrium. My model provides two quantitative insights. First, CIP deviations could be halved by increasing banks’ access to wholesale dollars, but with a 40% increase in default risk for money market funds. Second, a sharp drop in wholesale funding can disrupt global dollar credit as the marginal cost of synthetic dollars quickly outpaces the marginal revenue on bank assets. My findings suggest that regulations aimed at reducing domestic investors' default risk have contributed to the growth of synthetic dollar market, creating externalities in the form of CIP deviations and frictions in bank lending.
The Market for Sharing Interest Rate Risks: Quantities and Asset Prices
(with Jian Li, Ioana Neamtu, Ishita Sen)
Abstract: We study interest rate risk sharing across the financial system using novel data on cross-sector interest rate swap positions. We show that pension funds and insurers (PF&I) are natural counterparties to banks and corporations: PF&I buy duration, whereas banks and corporations sell duration. However, demand is highly segmented across maturities, resulting in significant imbalances at various maturity points. We calibrate a preferred-habitat investors model with risk-averse arbitrageurs to study how demand imbalances interact with supply side constraints to impact swap spreads. Our framework helps quantify the spillover effects of demand shifts, which informs policy discussions on financial institutions’ hedging requirements.
Unemployment Insurance Fraud in the Debit Card Market
(with Jetson Leder-Luis, Jialan Wang, Yunrong Zhou. NBER working paper # 32527)
Abstract: We study fraud in the unemployment insurance (UI) system using a dataset of 35 million debit card transactions. We apply machine learning techniques to cluster cards corresponding to varying levels of suspicious or potentially fraudulent activity. We then conduct a difference-in-differences analysis based on the staggered adoption of state-level identity verification systems between 2020 and 2021 to assess the effectiveness of screening for reducing fraud. Our findings suggest that identity verification reduced payouts to suspicious cards by 27%, while non-suspicious cards were largely unaffected by these technologies. Our results indicate that identity screening may be an effective mechanism for mitigating fraud in the UI system and for benefits programs more broadly.
Uninformed yet Consequential: Liquidity Shocks in FX Markets
(with Petra Sinagl)
Abstract: We study how retail liquidity shocks impact prices and volumes in the foreign exchange (FX) spot market. We model risk-averse dealers' accumulation of inventory under asymmetric information and incomplete offset across retail clients. Our model predicts that retail liquidity shocks result in inventory imbalances that are transmitted to the inter-dealer segment, increasing price volatility and trading volumes. Using month-end settlement breaks to instrument for uninformed order flow, we empirically validate these predictions: a one-standard-deviation rise in retail net volume increases volatility by 12-22% and inter-dealer volume by 10%, indicating that liquidity-driven demand interacts with intermediary constraints to determine asset prices.
Innovation Specificity
(with Jon Garfinkel, Amrita Nain)
Abstract: We study the composition of firms' innovation portfolios by machine-reading 90 million patent claims. Process-oriented patents fundamentally differ from other patents in terms of both motive and specificity: they are cost-savings-oriented, and they are rooted in firm-specific knowledge. On the former, process-oriented patents are more likely when the firm recently experienced higher costs relative to sales. To support the latter we offer several results. Process patents are more likely to cite past patents of the innovating firm, and they are undertaken by inventors who have more within-firm patenting experience. They also exploit known technologies rather than explore new ones. Finally, inventors with a greater fraction of their patents dedicated to process, are less likely to change firms. Process patents are also valued differentially in a way that reflects their specificity. Using the market for corporate control as a setting to assess the external value of innovation, we show that firms with a higher share of process patents in their innovation portfolios are significantly less likely to be acquired. Consistent with the specificity explanation, this effect reverses when there is a strong textual overlap between process patent descriptions and the acquirer’s product descriptions, indicating greater redeployability of innovation. When such overlap exists, acquisition announcement returns are also higher, and post-merger synergies—reflected in lower costs and higher operating margins—are more likely to materialize. Our study introduces a novel measure of innovation specificity and demonstrates its construct validity as well as its role in the market for corporate control.