Deposit insurance, bank capital structures and the demand for liquidity

This paper provides an economic explanation for the extraordinary and historically unprecedented accumulation of liquid assets by the banking system in the aftermath of the Great Depression. At the end of the decade (1938 to 1939) the banking system held over 35 percent of their assets in non-interest bearing cash. Why are these holdings so high and why didn't we observe the same phenomenon in Canada? My theory is that, contrary to what happened in Canada, U.S. banks came out of the Depression severely undercapitalized and they did not immediately replenish the capital account because, at the time, it would have been extremely expensive to do so. In order to calm depositors' fears, bank managers increased the share of liquid assets in their portfolios to reduce their risk exposure on the asset side. In order to shed some light into this observation I construct a banking model that generates some empirically testable implications. The model generates a negative correlation between the equity-to-assets ratio and the liquidity ratio. Using aggregate time-series data (1929-1940), I find that this prediction is borne out in the data. The banking system held a higher share of liquid assets during periods of more severe undercapitalization. With Call Report data (1985Q1-1989Q4) on over 10,000 U.S. banks, I test the cross-sectional implications of the model. I perform a pooled time-series, cross-section, fixed-effects regression across different asset size classes and find evidence of the negative correlation suggested by the model. In fact the banks with higher shares of liquidity were the ones with more precarious capital positions. I also do a theoretical experiment on the optimal design of deposit insurance policies. I derive six propositions characterizing bank risk-taking and portfolio composition under three different deposit insurance arrangements: (i) Fixed-rate (FDIC type), (ii) Actuarially fair, and, Bank-neutral. Banks would never hold actuarially fair deposit insurance voluntarily and there is a moral hazard problem with fixed-rate deposit insurance. Bank-neutral deposit insurance eliminates the moral hazard incentive and allows implementation without coercion. It has an imbedded subsidy given by the FDIC but with good countercyclical properties.