This is a question I have wanted to see an answer to for a long time. What is the minimum efficient scale for financial institutions? This is an important question with respect to breaking up large banks into smaller entities. Some have argued, based on very little compelling evidence as far as I can tell, that breaking up big banks would be costly because large banks are able to exploit economies of scale. Others disagree, but again evidence for either point of view is unclear. I don't mean there is no evidence at all, the existing research is described in the introduction to this paper, but the results do not point strongly in any particular direction. Hopefully, more work on the topic will shift the weight of the evidence in one direction or another:
The Evolution of Scale Economies in U.S. Banking, by David C. Wheelock and Paul W. Wilson, August 2015: Abstract Continued consolidation of the U.S. banking industry and general increase in the size of banks has prompted some policymakers to consider policies to discourage banks from getting larger, including explicit caps on bank size. However, limits on the size of banks could entail economic costs if they prevent banks from achieving economies of scale. The extent of scale economies in banking remains unclear. This paper presents new estimates of returns to scale for U.S. commercial banks based on nonparametric, local-linear estimation of bank cost, revenue and pro t functions. We present estimates for both 2006 and 2012 to compare the extent of scale economies in banking some four years after the financial crisis and two years after enactment of the Dodd-Frank Act with scale economies prior to the crisis. We find that most banks faced increasing returns to scale in cost in both years, though results for the very largest banks in 2012 are somewhat sensitive to specification. Further, most banks faced decreasing returns in revenue in both years, though nearly all banks could still increase revenue and pro t by becoming larger.
[As I've written many, many times, I do not think that breaking up big banks will do a lot to reduce our susceptibility to bank crises. After all, we had a financial crisis about every 20 years in the 1800s, and this continued through the Great Depression, and at that time banks were relatively small. Thus, it seems that crises have more to do with the diversity of activity and connectedness than bank size. I favor breaking up the biggest banks to reduce their political power, which I believe is excessive, and to reduce their economic power. If the above results had shown that the minimum efficient scale was much smaller than the typical large, systemically important bank, breaking them up would be an easy call. But that's not what the results imply. Thus, in this case, there is a tradeoff between the benefit or reducing political and economic power versus losing economies of scale (not sure how steep the cost function is at the existing size -- if it's relatively flat the loss of scale economies could be small). The other alternative is to treat them along the lines of a public utility. We allow them to be large to exploit scale economies, then regulate pricing and other behavior. However, this is where the political power of the large banks matters, and it's not clear that a policy of "large but with regulatory oversight" is the best option to pursue.]