A STRATEGY TO PREVENT FUTURE CRISES: SAFELY SHRINK THE BANKING SECTOR Adam S. Posen (Senior Fellow Institute for International Economics) I was reading an article in the New Yorker the other day, and it was a history of how people have tried to reduce the number of traffic fatalities in the United States. Essentially, there were two basic strategies undertaken: there was the strategy of making safer cars and there was the strategy of trying to make better drivers. If your goal is to reduce fatalities, these both seem like legitimate means. And so we have things like airbags and crumple zones to make cars safer and we also have Mothers Against Drunk Driving and "wear your seatbelt" regulations to make safer drivers. But what never really occurred to anybody, this being the United States, was to reduce the number of drivers. If you have fewer drivers, you have smaller accidents and fewer fatalities. Like traffic fatalities, financial crises are bad, so the fewer, the better. When we look at financial crises and really think about the historical record (and what was discussed at this conference), financial crises are in essence banking crises. So as with traffic, you can try to make a safer banking sector by having prudential rules and credit standards, and so on. You can also try to make safer bankers and better bank supervisors by exchanging experiences at conferences like these and having the BIS' Financial Stability Institute train everybody, and instilling market discipline. I am going to suggest instead that the best strategy to prevent future crises should be reducing the number of banks. If traditional deposit-taking, long-lending banks were to play a reduced role in most economies, and therefore the amount of damage bank failures could do was limited, the extent of financial crises would be more limited. Of course, this raises questions of what economies give up when you reduce the number of banks, and how feasible or likely it is to reduce the number of banks. I will argue on the first point that you are not giving up all that much by reducing the number of banks, that other financial services can replace them easily for both borrowers and depositors. On the second point, I will suggest that the trends in corporate finance and technology are so strongly against traditional banking, that the public policy issue is how to assure sufficient safe exit of banking firms-feasibility of exit is not a problem.The first issue is: why? Why should we be shrinking the banking sector? The answer is simple: we economists now seem to have an exceptionally good understanding of financial crises. Everybody agrees1: imperfect information, regulatory forbearance, connected lending, illiquid collateral, untransparent accounting, deposit insurance not fully creditable, shake it up, add a macroeconomic shock, and presto-a banking crisis. And the consensus on this analysis is actually pretty wide. This is why regulators can call for prediction,2 and then we can talk about making use of early warning.3 Yet, with all due respect, early warning does not seem to work, right? We have known all these things for some time, and we are constantly improving our forecasting, but the problem is not so much warning as acting on the warnings. At issue is not whether there is time enough to react, but instead, if banks and regulators have too much time to react, what do they do? We all understand that the basic danger is that when an economy has undercapitalized regulated banks, they behave badly, they are subject to a moral hazard, which results in adverse selection in the financial markets. The resulting rolling over and accumulation of bad loans is an enormous drag on growth. You either suffer the cost of recapitalizing the banks quickly at large cost in terms of real output (which was repeatedly the case in Latin America), or you suffer the costs of having wasted a lot of your capital, foregoing growth, and eventually paying a larger bill (which is arguably a major part of what has happened in Japan for the last ten years). This is serious stuff. And we all knew ahead of time both that bank crises are serious and where they were occurring. So that is my second point: learning, unfortunately, does not seem to affect behavior when it comes to financial crises. Now as a think tank person, I am supposed to believe in the power of learning. My colleagues' and my only hope of having influence, is that people listen to one of us and say: "Ah ha! Good idea, I'll do that." But this does not seem to overcome the incentives underlying bank crises. Consider US-Japan foreign economic relations, specifically financial relations, in the last 20 years.4 What is very interesting is that, going through the history, the amount of transpacific and international exchange in the happy community of bank supervisors was already quite great by the late 1980s. Yet, we still had the US S&L crisis, and prior to that we had the UK land-bubble, and immediately following that we had the Swedish banking crisis, and then we have the ongoing Japanese financial crisis. This lists only some of the instances in only wealthy countries, just to remind everyone this is a universal problem. So even with the best of intentions, real public oversight, best training, best of supervision and, assumedly, mostly non-corrupt civil servants, learning about bankingcrises does not seem to take place. The US savings and loan crisis ended when there was a political demand ex post; the Japanese banking crisis we are hoping will be ending now that there is a political demand ten years ex post. And there was no shortage of advice or warning to Tokyo from Washington and elsewhere in the bank supervisory community by the early 1990s. There are very fundamental economic and political incentives for why this is the case.5 One need not be an experienced political cynic, however, to model these processes. There are reasons why banks rollover bad loans and why bureaucrats engage in regulatory forbearance, and it is very difficult to get beyond those. As so often in economic policy, one is caught between the choice of either removing discretion totally, which seems self-defeating since the whole point of the bank supervisory exercise is the judgment of imperfect information, or leaving regulators the discretion to engage in regulatory forbearance (which is sometimes euphemized as "prompt corrective action").6 So, that is the why. There does not seem to be any easy