Two weeks ago, the IMF organised a major research conference, in honour of Stanley Fischer, on lessons from the crisis. Here is my take. I shall focus on what I see as the lessons for monetary policy, but before I do this, let me mention two other important conclusions.
One, having your macro house in order pays off when there is an (external) crisis. In contrast to previous episodes, wise fiscal policy before this crisis gave emerging market countries the room to pursue countercyclical fiscal policies during the crisis, and this made a substantial difference. Second, after a financial crisis, it is essential to rapidly clean up and recapitalise the banks. This did not happen in Japan in the 1990s, and was costly. But it did happen in the US in this crisis, and it helped the recovery.
Now let me now turn to monetary policy, and touch on three issues: the implications of the liquidity trap, the provision of liquidity, and the management of capital flows. On the liquidity trap: we have discovered, unfortunately at great cost, that the zero lower bound can indeed be binding, and be binding for a long time—five years at this point. We have also discovered that, even then, there is still some room for monetary policy. The bulk of the evidence is that unconventional policy can systematically affect the term premia, and thus bend the yield curve through portfolio effects. But it remains a fact that compared to conventional policy, the effects of unconventional monetary policy are very limited and uncertain.
There is therefore much to be said for avoiding the trap in the first place in the future, and this raises again the question of the inflation rate. There is wide agreement that in most advanced countries, it would be good if inflation was higher today. Presumably, if it had been higher pre-crisis, it would be higher today. To be more concrete, if inflation had been 2 percentage points higher before the crisis, the best guess is that it would be 2 percentage points higher today, the real rate would be 2 percentage points lower,