Mark Thoma directs us to David Warsh on Gorton and Holmstrom’s view of the role of banking. I’ve written about this view in several places. My own view of banking is very different and here is a quick summary of my key points.
The source of Gorton and Holmstrom’s errors: Taking U.S. banking history as a model
In my view Gorton and Holmstrom err by basing their view of what banking is on the pre-Fed U.S banking system. Nobody argues that the U.S. represented a “state-of-the-art” banking system in the late 19th century. In fact, in the late 19th century the U.S. banking system was still recovering from the reputational consequences of the combination of state and bank defaults in the 1840s that had led many Europeans to conclude that American institutions facilitated fraud. By the end of the 19th century, however, the U.S. did have access to European markets and there is evidence that the U.S. banking system relied heavily on the much more advanced European banking system for liquidity (e.g. the flow of European capital during seasonal fluctuations). Indeed, the crisis of 1907, during which the none-too-respected U.S. banking system was at least partially cut off from the London money market, was so severe, it led to the decision to emulate European banking by establishing the Federal Reserve.
What Gorton and Holmstrom get right: the fundamental difference between money market and capital market liabilities, or as Warsh puts it: “Two fundamentally different financial systems [are] at work in the world”
In particular, it is essential for the debt that circulates on the money market to be price stable or “safe.” This distinguishes money markets are from capital markets, where price discovery is essential. Holmstrom writes:
Among economists, the mistake is to apply to money markets the lessons and logic of stock markets. … Stock markets are … aimed at sharing and allocating aggregate risk … [and this] requires a market that is good at price discovery. … [By contrast,] The purpose of money markets is to provide liquidity for individuals and firms. The cheapest way to do so is by … obviat[ing] the need for price discovery.
What Gorton and Holmstrom get wrong:
1. The historical mechanisms by which the banking system created “safe” money market assets.
Holmstrom writes: “Opacity is a natural feature of money markets and can in some instances enhance liquidity.” This is the basic thesis of Gorton and Holmstrom’s work.
A study of the early 20th century London money market indicates, however that the best way to create safe money market assets is to (i) offset the implications of “opacity” by aligning incentives: any bank originating or selling a money market asset is liable for its full value, and (ii) establish a central bank that (a) has the capacity to expand liquidity and thereby prevent a crisis of confidence from causing a shift to a “bad” equilibrium, and (b) controls the assets that are traded on the money market by (1) establishing a policy of providing central bank liquidity only against assets guaranteed by at least two banks, and (2) withdrawing support from assets guaranteed by low-quality originators. (ii)(b) plays a crucial role in making the money market safe: no bank can discount its own paper at the central bank, so it has to hold the paper of other banks; at the same time, no bank wants to hold paper that the central bank will reject. Thus, the London money market was designed to ensure that the banks police each other — and there is no American-style problem of competition causing the origination practices of banks to deteriorate.
The Gorton-Holmstrom approach is based on the historical U.S. banking system and sometimes assumes that deterioration of origination quality is inevitable — it is this deterioration that is “fixed” by financial crises, which have the effect of publicizing information and thereby resetting the financial system. In short, by showing us how a banking system can function in the presence of both opacity and misaligned incentives, Gorton and Holmstrom show us how a low-quality banking system, like that in the late 19th century U.S. which could only create opaque (not safe) assets, can be better than no banking system.
Surely, however, what we want to understand is how to have a high-quality banking system. The kind of system represented by the London market is ruled out by assumption in the Gorton-Holmstrom framework which focuses on collateralized rather than unsecured debt. An alternative model for high-quality banking may be given by the 1930s reforms in the U.S. which improved the origination practices of U.S. banks and — temporarily at least — stopped the continuous lurching of the U.S. banking system from one crisis to another that is implied by opaque (rather than safe) money market assets.
2. Gorton and Holmstrom err by focusing on collateral rather than on overlapping guarantees.
Holmstrom writes: “Trading in debt that is sufficiently over-collateralised is a cheap way to avoid
adverse selection.” His error, however is to use both language and a model that emphasize collateral in the literal sense. The best form of “over-collateralization” for a $10,000 privately-issued bill is to add to the borrower’s liability the personal guarantee of Jamie Dimon — or even better both Jamie Dimon and Warren Buffett. This is the principle on which the London money market was built (and because both extended liability for bank shares and management ownership of shares was the norm until the 1950s in Britain, personal liability played a non-negligible role in the way the banking system worked). This is rather obviously an excellent mechanism for ensuring that money market debt is “safe.”
The fact that it may seem outlandish in 21st century America to require that a bank manager have some of his/her personal wealth at stake whenever a money market asset is originated, is really just evidence of the degree to which origination practices have deteriorated in the U.S.
Note also that there is no reason to believe that the high-quality money market I am describing will result in restricted credit. Nothing prevents banks from making the same loans they do now; the only issue is whether the loans are suitable for trade on the money market. Given that our current money market is very heavily reliant on government (including agency) assets and that these would continue to be suitable money market assets, there is little reason to believe that the high-quality money market I am describing will offer less liquidity that our current money market. On the other hand, it will offer less liquidity than, say, the 2006 money market — but I would argue that this characteristic is a plus, not a minus.
3. Holmstrom errs by focusing on debt vs. equity, rather than money markets vs. capital markets
Holmstrom claims that: “Equity is information-sensitive while debt is not.” He clearly was not holding GM bonds in the first decade of the current century. A more sensible statement (which is also consistent with the general theme of his essay) is that capital market assets including both equity and long-term debt are information sensitive, whereas it is desirable for money market assets not to be informationally sensitive.
In short, I argue that in a well-structured banking system money market assets are informationally insensitive because they are safe. For institutionally-challenged countries, a second-best banking system may well be that presented by Gorton and Holmstrom, where money markets assets are “safe” — at least temporarily — because they are informationally insensitive.
In my view, however, we should establish that a first-best banking system is unattainable, before settling on the second-best solution proposed by Gorton and Holmstrom.