Time-varying Capital Requirements: Rules vs. Discretion (again)

Among its numerous innovations, Basel III proposes that national authorities use countercyclical capital buffers to temper booms in credit growth and asset prices. The idea is that when financial conditions are overly accommodative, one way to tighten them and prevent a bust is to require banks to hold additional capital. During downturns, the banks would then be allowed to release this buffer, easing financial conditions even without lowering interest rates. That’s the idea, anyway.

Will it work? Is such a system either practical or desirable? Our view is that regardless of how theoretically attractive, making such time-varying capital regulation discretionary is unlikely to work in practice. Rules would serve us all much better.

There are three arguments that tip the balance against the discretionary approach:  high information requirements, long transmission lags and significant political resistance.

On information requirements, figuring out when the regulator should act is going to be very difficult. What constitutes a credit boom or property price boom that is sufficient to justify raising capital requirements on banks? 

Even if the regulator knew when, she would still need to know how much. Calibration will be more complex than for a fixed capital ratio because an increase in capital requirements cannot be implemented instantly. You would have to give banks some time to raise capital – whether by issuing equity or reducing dividends and stock buybacks. If you give them a year or more, which seems reasonable, the effective tightening of financial conditions wouldn’t peak until then. And the impact on the ultimate objective – asset prices and credit – probably would occur with an even longer transmission lag. If you had been concerned about financial conditions when you first considered the need to raise requirements, you will probably have added a couple of years of compounding the trouble even if you had acted expeditiously.

Next comes politics, a problem in two parts. The first is that you are targeting debt-intensive industries. While this seems similar to the way interest rate policy operates, it differs in one important respect: raising interest rates benefits savers, helping to make it somewhat more palatable to at least one fairly broad constituency. In contrast, virtually no one benefits from higher countercyclical capital requirements, at least not in the short run. It is like giving everyone a painful vaccination –  eventually it may be good for you and for others, but not when you get it.

The second part of the political problem is that policymakers are effectively telling a bunch of people that they are getting too rich. These are people whose house prices are rising, and who think that they have met the market test. They are shareholders of profitable banks, whose balance sheets are expanding. More generally, they are investors in risky assets whose prices are stretched beyond what fundamentals warrant in the long run. Policymakers are saying to these groups that it would be good to acknowledge now that they aren’t really as rich as they think. Perhaps true, but who wants to listen? And, how can policymakers really be sure?

These technical and political problems create doubts about the time consistency of a discretionary framework. Can policymakers really credibly commit to raising capital requirements when the time comes? If banks doubt the resolve of future policymakers, they have an incentive to take on greater risks today, making the system more (not less) vulnerable.

How can we overcome these regulatory challenges? One answer is to make the financial system sufficiently robust so that it can weather occasional booms and busts. This suggests a combination at all times (good and bad) of high capital requirements and effective stress tests implemented in a rules-based framework.

Capital requirements would always include what in Basel III is called a conservation buffer. In effect, there are two levels of capital that matter, a high level above which a bank’s operations are unrestricted, and a lower one below which the bank would get shut down. In between those two, a bank’s dividend, share buyback and compensation policies would be restricted so that it would retain earnings until it returned to the higher level. You might think of this as an automatic reduction of capital requirements in bad times that avoids the implementation lags of discretionary, time-varying requirements.

Bank stress tests would still be needed to reduce regulatory arbitrage. They also would allow regulators to assess the vulnerability of those financial sectors where asset prices may be out of whack. From this perspective, stress tests include a hidden element of discretion, but the communication and decision-making process (not to mention the politics) would be far more routine. Indeed, because banks know that stress tests are coming, the impact should be smaller if and when the stress focuses on sensitive matters like property lending.

Economists don’t get surprised when rules beat discretion, but many policymakers still do.