Journal of Political Economy
We modify the Diamond and Dybvig (1983) model so that, besides offering liquidity services to depositors, banks also raise equity funding, make loans that are risky, and can invest in safe, liquid assets. The bank and its borrowers are subject to limited liability. When profitable, banks monitor borrowers to ensure that they repay loans. Depositors may choose to run based on conjectures about the resources that are available for people withdrawing early and beliefs about banks' monitoring. We use a new type of global game to solve for the run decision. We find that banks opt for a more deposit-intensive capital structure than a social planner would choose. The privately chosen asset portfolio can be more or less lending-intensive, while the scale of intermediation can also be higher or lower depending on a planner's preferences between liquidity provision and credit extension. To correct these three distortions, a package of three regulations is warranted.
Revise & Resubmit
We introduce heterogeneity in terms of workers and entrepreneurs in an otherwise standard Fisherian model to study Sudden Stop dynamics and optimal policy. We show that the distinction between workers and entrepreneurs introduces a distributive externality that is absent from the representative-agent setup. While in tranquil times redistribution is driven by the relative marginal utilities of consumption, the planner additionally favors entrepreneurs during Sudden Stops to mitigate Fisherian deflation. We show how heterogeneity adds to the understanding of how ex ante and ex post policies can best be designed to alleviate the negative effects of Sudden Stops.
"Leverage and Stablecoin Pegs", with Gary Gorton, Chase Ross, Sharon Ross, and Elizabeth Klee
Money is debt that circulates with no questions asked. Stablecoins are a new form of private money which circulate with many questions asked. We show how stablecoins can maintain a constant price even though they face run risk and pay no interest. Stablecoin holders are indirectly compensated for stablecoin run risk because they can lend the coins to levered traders. When speculative demand is strong, levered traders are willing to pay a premium to borrow stablecoins. Therefore, the stablecoin can support a $1 peg even with higher levels of run risk.
We study the interplay between indebtedness and asset price bubbles. A bubble relaxes borrowing constraints and increases borrowing capacity. Yet a deflating bubble amplifies downturns when constraints start binding. We show analytically and quantitatively that optimal macroprudential policy should respond to bubbles in a non-monotonic way, which depends on the underlying level of indebtedness. If the level of debt is moderate, policy should accommodate the bubble to reduce the incidence of a binding borrowing constraint. If debt is elevated, policy should lean against the bubble more aggressively to mitigate the externalities when constraints bind.
We study how competition between banks and non-banks affects lending standards. Banks have private information about some borrowers and are subject to capital requirements to mitigate risk-taking incentives from deposit insurance. Non-banks are uninformed and market forces determine their capital structure. We show that lending standards monotonically increase in bank capital requirements. Intuitively, higher capital requirements raise banks' skin in the game and screening out bad projects assures positive expected lending returns. Non-banks enter the market when capital requirements are sufficiently high, but do not cause a deterioration in lending standards. Optimal capital requirements trade-off inefficient lending to bad projects under loose standards with inefficient collateral liquidation under tight standards.
Banks add value by monitoring borrowers. High funding costs make banks reluctant to lend. A central bank can ease funding by purchasing loans, but cannot distinguish which loans require more or less monitoring, exposing it to adverse selection. Multi-tier pricing arises as the optimal design setting differential purchases terms given loan characteristics. The relative welfare gain compared to uniform pricing depends on three sufficient statistics: the share of loans requiring monitoring, the risk-retention ratio, and the liquidity premium. For plausible values in the data, the gain ranges from 0.02% to 0.31% of the size of central bank's loan purchases.
WORK IN PROGRESS
"Crypto-payments and runs", with Elizabeth Klee
Crypto payments can cause runs on stablecoins, even if the underlying reserve assets are perfectly safe and liquid. We illustrate this dynamic using a global games model. The frictions that can cause the run derive from congestion on the blockchain and the cost of verifying that there cannot be a circulation or double-spending of unbacked stablecoins. We show that potential policy solutions could be similar to those used in traditional payment systems, including final settlement in central bank digital currency.
"Optimal Mortgage Regulation and Income Inequality", with Nina Biljanovska