Tyler Cowen and Eric Falkenstein raised doubts about the efficacy of leverage regulation of banks. Here is how I think about it.

1. We assume that banks have a government backstop that is either explicit or implicit. the object of the game for the government is to minimize the moral hazard this creates, and in some sense the object of the game for the banks is to to take maximum advantage of the moral hazard.

2. Leverage regulation is “stupid” regulation. You require, say, a ratio of capital to assets of 15 percent. The way that banks respond is to raise the riskiness of assets. So even though the bank holds a lot of capital, it takes a huge amount of risk, and wins the game.

3. Risk-based capital regulation is “clever” regulation. In my view, the best form of clever regulation is stress testing. You apply tests, such as “what happens if interest rates jump 200 basis points?” or “what happens if house prices fall 20 percent?” The problem with stress testing is that the results are model-dependent. As the financial crisis hit, the Basel rules were in the process of pivoting toward stress testing, with banks doing their own modeling (as Tyler would say, Yikes!). The problem with clever regulation is that whatever stress test you use, the bank can structure its balance sheet to just pass the stress test even though it holds practically no capital. In a sense, that is what the whole mortgage securitization structure was all about, except that the regulatory regime relied on AAA ratings to serve the function of stress tests.

4. I think that a combination of both forms of regulation would be much better than either one alone. Clever regulation can spot the worst abuses of stupid regulation. Stupid regulation can serve as a backstop as banks learn to how to game clever regulation. So the question of which type of regulation should be employed should not be answered “either-or,” it should be answered “both.”