Tariffs, stablecoins, and the demand for dollars (with Anantha Divakaruni)
Interview in Central Banking. Blog coverage by Vincent Arnold and Joachim Klement.
Several studies have shown that aggregate demand for US dollars fell following the announcement of tariffs by the US government on April 2, 2025. Using data on stablecoins as a proxy for dollar trading, we find that the decline in dollar demand is smaller for investors in countries that saw larger increases in tariffs. Our interpretation is that, as foreign investors anticipate that tariffs will make it more expensive to acquire US dollars in the future, they buy dollars today. This channel is stronger for more liquid stablecoins and for countries with tighter capital controls, consistent with the idea that, when actual dollars are hard to acquire, stablecoins may be regarded as a substitute. Our findings cast light on the effects of the tariffs on global foreign exchange markets, as well as on the degree to which stablecoins are considered a close substitute for dollars.
Flash loans: lending without trust, collateral, or risk (with Farshid Abdi and Cornelius Johnson)
Empirical section coming soon!
Flash loans are a means of unsecured lending in the decentralised finance (DeFi) system. A smart contract creates conditionality so that, if the borrower does not repay, the flash loan is never advanced. Such a contract is infeasible in traditional finance. This arrangement eliminates credit risk for the lender, but also means the borrower can costlessly default. Therefore, a flash loan insures a trader against loss: if the payoff from a trade would have been negative, then the trade does not settle. We examine the effect on market liquidity of arbitrage financed by flash loans. Bid-ask spreads widen when there are more flash-financed traders, but a market can exist so long as there are some uninformed traders. However, if flash-financing is cheap enough, then spreads are so wide that informed traders drop out of the market, and private information is no longer incorporated into prices. Our predictions are testable using blockchain data.
A monetary model of cryptocurrency
Interview on New Money Review podcast (article here) and in CoinDesk. Other media coverage: Irish Independent, Bitcoin Exchange Guide, Decrypt, TronWeekly.
I show price formation for cryptocurrencies is very different to standard assets. Blockchains have limited settlement space, creating competition between users of the currency. I adapt this insight into a Lagos-Wright model featuring a settlement constraint and a financial market. Speculative activity can crowd out monetary usage. This means higher speculative demand for cryptocurrencies can reduce prices, contrary to standard economic models. Crowding-out also raises the riskiness of investing in cryptocurrency, explaining high observed price volatility.
An earlier version of this paper, which used a different model, was circulated under the title 'Blockchain structure and cryptocurrency prices'. Click here for a non-technical summary. Also published as a 'Bank Underground' blog post.
The sterling unsecured loan market during 2006-08: insights from network theory (with Anne Wetherilt and Kimmo Soramäki)
Bank of England working paper 2010, no. 398.
We model the unsecured overnight market in the United Kingdom as a network of relationships and examine how the structure has changed over the recent period of crisis.
Interest rate risk at US credit unions (with Grant Rosenberger)
Media coverage: California and Nevada credit union news
Rising interest rates have prompted concerns about losses on bank assets, especially following the failure of Silicon Valley Bank (SVB) in March 2023. In this working paper, we examine whether US credit unions could be subject to similar losses as banks and analyze how their regulatory capital would be affected. We estimate that after realizing losses from assets that have decreased in value and not yet been sold the overall net worth of the credit union industry would have fallen by 40 percent in 2023:Q1. Unrealized losses were most severe at the largest credit unions. Nonetheless, the bulk of deposits at credit unions were insured, suggesting limited risk of an SVB-style run. In addition, credit union deposit rates are relatively insensitive to market interest rates, providing credit unions with a hedge against a rising rate environment. Overall, credit unions’ balance sheet positions seemed to be more resilient to unrealized interest rate risk than banks’.
We present a novel demand-side explanation for credit crunches following financial crises, based on an adverse selection problem. When the probability of distress at a relationship lender is high, borrowers insure against the risk of not being financed by choosing smaller projects that other, less informed, lenders are willing to lend against. We show that, when the probability of distress reaches a critical level, there is a discontinuous drop in the borrower's chosen project size, and in welfare. This means that even a small shock to financial stability can have very large real consequences.