Economics of bank supervision

Much debate in recent years has focused on bank supervision and regulation as a result of the significant changes in policies aimed at the banking sector post-financial crisis. The 2010 Dodd Frank Act introduced a number of provisions tightening regulations for the largest and most complex banking organizations. While harder to measure, bank supervision has also undergone a significant expansion. Federal Reserve supervisory staff has risen about 50% since before the crisis, to about 4,500 employees in 2015, and now accounts for about 20% of all Fed employees. How can we evaluate such a large expansion?

In this paper, we combine a novel dataset on hours spent by Federal Reserve staff supervising banks with a model to study the economics of bank supervision — the allocation of supervisory resources across banks. At the individual bank level, we find that hours increase with bank size and risk. Interestingly, hours increase less than proportionally with size, suggesting the presence of technological scale economies in supervision. In response to increases in bank risk, however, hours increase substantially. After accounting for these factors, the data also provide evidence of disproportionate resource allocation to the very largest banks, especially in the post-2008 period.

Using the observed allocation, we then estimate structural parameters of the model and compute the model-implied shadow value of supervisory resources. We find resource scarcity to vary considerably across time and Federal Reserve districts, with the shadow value in 2014 still considerably higher than in the pre-crisis years of 2004–2006, even with the large post-crisis increase in staff. Finally, we use the model to construct instruments for supervisory attention and estimate the sensitivity of bank risk to supervisory efforts, which we find to be statistically significant and economically large.