Paul Krugman’s wrap-up of the Minsky Conference on financial panics ends with a mention of Gary Gorton’s cartel hypothesis that banking got too competitive:
Deposit insurance was certainly part of the answer. But how did we mostly avoid, until the savings and loan scandals, the moral hazard such insurance can create? Regulation, especially limits on leverage, surely helped. But Gorton and other have suggested that an important part of the story was the simple fact that banking wasn’t very competitive: with limits on the interest banks could pay, coupled with barriers to entry, banks had a large franchise value — and this made bankers reluctant to take risks that could kill the cash cow that kept laying golden eggs, or something like that.
You can see Gorton develop this idea in a variety of venues, but in a lot of ways I think the clearest explanation is in his 1994 paper “Bank Regulation When ‘Banks’ and ‘Banking’ Are Not the Same” which despite botching the use-mention distinction lays out the relevant details in a way that should be accessible to the layperson. The fact that the paper was written 16 years ago also makes it valuable, in my view. Too much of the conversation has tended to focus on the specific details of the Panic of 2008 (subprime loans, etc.) which are unlikely to precisely recur rather than on the structural features of the situation which made the Panic of 2008 possible.
Gorton notes that in part because of deregulation, and in part just because of changes in the marketplace banks started to face more competition. Competition from foreign banks, competition from non-bank constitutions, competition from untraditional instruments, etc. This disrupted an earlier paradigm in which banking was not just regulated, but through regulation substantially protected from new entrants coming in and competing with them. New competition reduces the value of bank charters, and this creates problems:
Limiting entry into banking creates capital in the form of a valuable charter. Entry into banking by non-banks, i.e. firms which from the point of view of the government are not ‘banks’, or by new debt markets, causes charter values to fall. The difficulties for bank regulation with such a situation are enormous. When charter values are declining due to entry, there is a difference between the privately optimal and socially optimal levels of risk in the banking system. Without charter value, banks engage in new activities which are risky and, possibly, difficult to assess. These activities may trigger panic because they are intertwined with lending and deposit-taking. To the extent that non-banks engage in banking, they may also be subject to runs. The first scenario would occur, for example, if a bank were subject to a run because of defaults on its derivatives book. The second scenario would occur, for example, if a money-market mutual fund were subject to large withdrawals because of publicized declines in the value of money-market instruments. In both cases, the problem is that activities which are not related to lending and deposit-taking become informationally intertwined with banking in the eyes of the public. It is not only that banks may engage innew activities which are risky, they may also engage in old activities which are riskier than previously, activities such as commercial real-estate construction and development loans.
In a sense, the results of this are the same as they would be in any other business. More competition in Sector X is a boon for consumers of whatever it is Sector X is selling (in this case, lower interest rates) but it also means that firms in Sector X are now more prone to failing. The problem, however, is that regulators are loathe to let banks go bankrupt, lest there be economy-wide panic. The main response to this has been to directly subsidize banks via bailouts in order to compensate for the loss of stability caused by de-cartelization:
There is a problem with increasing subsidies to banks when entry is not limited. When banking is less profitable and riskier than before, the only way that the loss in value can be made up when entry is not limited is by directly transferring resources to banks, that is, by subsidizing their attempts to remain as large and as profitable as they were prior to entry by competitors. The subsidies are counter-productive because they do not result in a valuable asset which is lost if the bank becomes insolvent, like the charter. Rather than creating an incentive to limit risk, the subsidies are increased when the bank becomes insolvent, creating an incentive to engage in risk-taking. The defence of these policies is that they are needed to protect the provision of banking services. For example, it is argued that the loss of information produced in the course of developing bank relationships would be sizeable if the particular bank were allowed to fail. (The Slovin et al., 1993, study is consistent with this.) Sprague (1986), the former chairman and director of the US Federal Deposit Insurance Corporation, is fairly explicit about this rationale.
While it is true that there may be sizeable costs to bank failure, the policy is fundamentally confused. In the current situation of reduced charter values, such policies can only increase the likelihood of the very events, risk-taking and failure, that they are supposed reduce. These policies subsidize bank risk-taking in an environment where banks can be expected to increase risk.
Essentially, this is the story of the policy response to the Panic of 2008. Financial institutions in a competitive marketplace engaged in risky activity and some of them went belly-up. In response we tried a mix of panic-unleashing bankruptcies (Lehman Brothers) and long-term counterproductive bailouts.
At any rate, there’s a lot that could be said about this but the post is long already and I’d like to encourage people to read the original paper for themselves. Suffice it to say that I think this is persuasive in part because, as I said, it seems to pretty presciently forecast our current problems over a decade in advance and to do so in the kind of way that matters most — with attention to the abstract structure of the situation rather than the particular details of the housing boom. Note also that Gorton’s diagnosis seems to jibe with the observation that Canada’s banking system is unusually solid.