A recent paper by Cerra and Saxena, two economists at the Bank for International Settlements, argues that the economic costs of recessions and more serious crises tend to be permanent. According to their analysis, post-recession recovery does not generally return an economy to its pre-recession growth trend, but shifts the trend line down a notch. That is, periods of negative growth are not usually followed by periods of above-trend growth. Poor countries may be poor not because their growth rates during healthy periods are lower, but because they have more and deeper disruptions.
I am not in position to adjudicate, except to notice (1) most economists simply assume that recovery returns an economy to trend, (2) the issue is of enormous importance, and (3) if Cerra and Saxena are right, we need to rewrite the macro textbooks (again).
UPDATE: DeLong endorses the view that short run effects don't last:
Which side am I on? I tell my undergraduates:
At a time horizon of 0-3 years, be a Keynesian: the most important things are the fluctuations in unemployment, in real demand, and in capacity utilization.
At a time horizon of 3-8 years, be a demand-side monetarist: you can assume (provisionally) that fluctuations in employment, real demand, and capacity utilization die out; the most important things are the fluctuations in the composition of real demand (investment vs. consumption vs. government vs. net exports) and in inflation- and deflation-causing nominal demand assuming (provisionally) stable growth of the economy's productive capacity.
At a time horizon of 8 years or greater, be a sane supply-sider: the most important things are the processes of investment in physical, human, and organizational capital that raise the economy's productive capacity