Brad DeLong's Notes of Jan de Vries's Talk on the Purpose of Economic History
Why Should Economists Study Economic History? After-Action 210a Note

Are Modern Recessions Different on the Labor Side?

Paul Krugman thinks they are. Recovery from the old pre-1990 the-Federal-Reserve-has-hit-the-economy-on-the-head-with-high-interest-rates recessions was quick once the Fed had concluded that inflation was under control and had eased up. But he thinks (and I more than half agree) that these days recessions are credit-channel-overinvestment-bubble-bursting events, and that on the labor side at least recovery from them is a much more lengthly and difficult process.

He may well be right.

Here is Paul Krugman:

Deep? Maybe. Long? Probably. - Paul Krugman - Op-Ed Columnist - New York Times Blog: I still keep reading articles asserting that the last two recessions were brief and shallow. Formally, that’s true. But both were followed by prolonged “jobless recoveries” that felt like continuing recessions. Below is the employment-population ratio since 1989, with shading showing the official recessions. In both cases the employment slump went on for a long time after the recession was supposedly over.

There’s every reason to think that the same thing will happen this time. There’s a huge overhang of excess housing inventory; it will probably take several years before housing prices fall to realistic levels; and it’s not at all clear what will fill the gap left by weak housing and consumer spending.

There’s still the question of how deep the slump will be. I can see the case for arguing that it will be nasty. The 1990-91 recession was brought on by a credit crunch, the 2001 recession by overinvestment; this time we’ve got both. I guess we’ll see. In any case, whatever happens will probably last quite a while.

Here is the graph that Paul attaches:

Deep? Maybe. Long? Probably. - Paul Krugman - Op-Ed Columnist - New York Times Blog

The extremely slow employment rebounds of the "Jobless Recovery" of 1991-1993 and of the "Bush Boom" of 2001-2005 are only two data points, but they are two data points.