As we all consider how to avoid future crises, there is a lesson in Mr Greenspan's legacy: it may be unwise to let Fed chairmen serve for very long periods. A more reasonable term limit would be 10 years (inclusive of prior service), in recognition that the chairman needs a lot of on-the-job learning unless he or she has already been a senior Fed official.

There is a tendency in American politics to impose an institutional solution on an individual problem. Towards the end of Mr Greenspan's term, he screwed up; therefore, limit Fed chairmen's terms. But that tendency does not mean term limits are a bad thing. They are common in democracies for government positions of significant power. Because of the tradition of deference by other governors and the open market committee, which sets US interest rates, the Fed chairman is an immensely powerful official with de facto control over trillions of dollars in spending power.

Term limits involve a trade-off: on the one hand, we might lose the occasional exceptional executive; also, term limits might encourage an executive to act strategically near the end of his term to improve chances for a good subsequent job. On the other hand, without term limits, longstanding popularity and prestige may entrench a leader, masking human frailties that gradually erode his or her performance, such as age, fatigue, staleness, complacency, overconfidence, greed or hunger for power - Mr Greenspan may have exhibited some of these pathologies in the latter part of his tenure. History is full of examples of leaders whose long service eventually became counterproductive, such as J. Edgar Hoover at the FBI. (There are also exceptions, such as Robert Morgenthau, New York County district attorney for 34 years.)

Another problem with long-tenured leaders is that they may get stuck in a strategic rut, keeping strategies that have worked in the past, even though conditions have changed. This is a common challenge: chief executives of successful companies fail to change course when needed, intelligence agencies miss new trends, investors stick with a certain style too long. For Mr Greenspan, cutting rates to avert crisis was an effective strategy during the recession of 1991 and the global currency crisis of 1998. Lowering rates after the internet bust seemed logical, but keeping them so low for so long was a mistake. Rotating leadership positions is a way to bring in a fresh set of eyes, facilitating change.

The Fed faces unique risks from long-standing leadership, because the market not only reacts to policy decisions but anticipates its next steps. Mr Greenspan's successes led to widespread confidence in the Fed's power to forestall an economic collapse, consequent on a bubble, through interest rate cuts - the Fed's commitment to protect asset prices in this way came to be called the "Greenspan put". If too many people in the markets predict the same outcome (in this instance by relying on Mr Greenspan) and then are proved wrong, the day of reckoning will be painful, as we have just experienced. More frequent rotation of the chairman's position would help prevent excessive confidence in a single strategy or personality. Market participants would be less certain that they could predict the actions of a new chairman. A dose of uncertainty might encourage companies to hold more capital and take less risk.

Finally, a shorter term limit might diffuse some of the Fed's political unpopularity, lessening the risk that Congress will step up political pressure on the Fed, which could lead to unhealthy pro-cyclical policies. This is an argument against the alternative to a 10-year term - of limiting the chairman to two five-year terms. The possibility of renewal would impinge on the Fed's independence; Senator Harry Reid implied last year, when Mr Bernanke's nomination for a second term was pending, that Mr Bernanke had assured him the Fed would not risk increasing unemployment by raising interest rates, though exactly what Mr Bernanke said is not known.