Job Market Paper
"What Caused Chicago Bank Failures in the Great Depression? A Look at the 1920s" (revise and resubmit, Journal of Economic History) A number of researchers have shown that bank failures in the US Great Depression can be linked to fundamental weaknesses already apparent before the start of the depression, around June 1929. This view apparently contrasts with the one pioneered by Friedman and Schwartz, in which all banks faced very large, non-discriminating deposit withdrawals, which caused thousands of unwarranted failures across the country. Looking at the long-term behaviour of balance sheet items from 1923 to 1933 provides new insights into the causes of bank failures. All Chicago banks indeed suffered tremendous deposit withdrawals during 1930-1933 in what seems to have been an indiscriminate run. However, using ordered logistic regression by cohort, I find that banks which failed the earliest in the 1930s had invested more in inherently less liquid assets (in particular, mortgages) in the 1920s. The main cause of banks' failure was therefore a combination of deposit losses on the liability side and of illiquid mortgages on the asset side. Banks heavily engaged in mortgages did not have enough liquid assets to face the withdrawals and failed. This paper thus reasserts banks' responsibility in their investment choices in terms of liquidity risk management.
Research in Progress
"Debt Dilution in 1920s America: Lighting the Fuse of a Mortgage Crisis" (EHES Working Paper, March 2014) An explanation of the Great Depression based on mortgage debt via the banking channel has been downplayed due to the conservatism of mortgage contracts at the time. For instance, loan-to-value ratios often did not exceed 50 per cent. As I show in a previous paper, the main problem with mortgages at the time was their intrinsic lack of liquidity. Nevertheless, using newly-discovered archival documents and a newly-compiled dataset from 1934, this paper uncovers a darker side of 1920s US mortgage lending: the so-called "second mortgage system." As borrowers often could not make a 50 percent down payment, a majority of them took on second mortgages at usurious rates. As theory predicts, debt dilution, even in the presence of seniority rules, can be highly detrimental to both junior and senior lenders. The probability of default on first mortgages was likely to increase, and commercial banks were more likely to foreclose. Through foreclosure they would still be able to retrieve 50 percent of the property value, but often after a protracted foreclosure process -- a great impediment to bank survival in case of a liquidity crisis. This paper is thus a timely reminder that second mortgages, or "piggyback loans" as they are called today, can be hazardous to lenders and borrowers alike. It provides further empirical evidence that debt dilution can be detrimental to credit.Out of the Shadows: Commercial Bank Mortgage Securitization and the Great Depression in the Chicago Area In a previous paper I showed that unsecuritized mortgage loans were a significant predictor of failure among Chicago banks in the Great Depression. Would mortgage securitization have solved the problem? Here I show that it would not in the way undertaken by banks in the Chicago area. First, using new archival material I show that most banks in the Chicago area did engage in some form of mortgage securitization in the 1920s. Second, I provide evidence that the two conditions for efficient mortgage securitization based on Snowden (1995) - the existence of a regulated securities exchange and full legal guarantee on the part of the issuer - were not met, thereby reducing monitoring incentives and putting banks' solvency under threat. Finally, I emphasize an aggravating factor in the special case of commercial, deposit-taking banks: the fact that many investors in the mortgage bonds were also depositors in the bank, thereby increasing the likelihood of a bank run. Those inefficiencies are not dissimilar to those affecting US mortgage securitization in the 2000s.