"A Static Capital Buffer Is Hard to Beat" with Matthew Canzoneri, Behzad Diba, and Arsenii Mishin, March 2026 — conditionally accepted at American Economic Journal: Macroeconomics. This paper was previously circulated with the title "Optimal Dynamic Capital Requirements and Implementable Capital Buffer Rules."
There is a tenuous predictive relationship between the ratio of nonfinancial credit to GDP and GDP two years ahead. Setting bank capital requirements following a simple rule based on the credit-to-GDP ratio is not a good idea in our model even if the model encapsulates a predictive relationship in line with the observed data.
A Static Capital Requirement is Hard to Beat
with Matt Canzoneri, Behzad Diba and Arsenii Mishin — latest version March 2023 — conditionally accepted at American Economic Journal: Macroeconomics
Rules that respond to cyclical conditions fail to prevent excessive risk taking, whereas a static capital buffer performs nearly as well as the Ramsey rule.
TFP dispersion across firms rises sharply during recessions. The 90th–10th percentile spread within industries widens in each of the four recessions in the sample, consistent with the model's prediction that credit rationing intensifies when aggregate conditions deteriorate.
Cyclical Fluctuations, Financial Frictions, and Productivity Differences across Firms
with Jinill Kim and Arsenii Mishin — June 9, 2026
We bridge two literatures that have largely developed in parallel: the endogenous growth literature, which uses tractable representative-agent models to study the evolution of aggregate TFP, and the misallocation literature, which studies why financial frictions prevent capital from flowing to its most productive uses across heterogeneous firms. We build a tractable model that imports key features of the misallocation literature into a representative-agent framework. Credit rationing arising from adverse selection and moral hazard prevents capital from flowing efficiently to the most productive firms. As credit rationing becomes more or less intense, aggregate TFP acquires an endogenous component, accounting for about 30 percent of the variance of TFP at business cycle frequencies. We show that our tractable model can match key features of the observed distribution of productivity across firms and its co-movement with output growth and credit conditions in the data.
