Krugman trying to revive the credentials of economists.

This was certainly true of the most recent poll, which asked whether the American Recovery and Reinvestment Act — the Obama “stimulus” — reduced unemployment. All but one of those who responded said that it did, a vote of 36 to 1. ...

I actually went into many of the reports. The summary conclusion was exactly as I expected, no economist could calibrate any model to fit the conditions of the test. The only effort I saw to combat that effect was research to break up the stimulus along state lines, and find some small series of data, prior to the crash that might be valid at the peak of unemployment.

This is the Romer and Romer method, find some stationary period, calibrate some coefficient, then skip over the intervening events, and apply the coefficients.  The authors of the report did this, after eliminating much of the stimulus, and focusing on a short period.   Results which were not a resounding success, and likely suffer all the same problems Romer and Romer have. Namely eliminating data that does not conform to priors. That 36 economists are incompetent in statistics is not astounding.

 Lets look at the big states.  Did the orange line make the unemployment rate turn, for each state? Well they all seem to be rising together with the same slope. Did they turn back in any way that can be correlated with the orange line.  No, they are  more decorrelated going down than going up. Except that California was the first one to rise. Notice that New York had the double dip?  It had a smaller double dip in the 2001 recession.

Notice the the slopes in which all the big states recovered and unemployment dropped.  Anything correlated between them?  They all recovered at different rates.  Their asymmetry is worse than the 81 recession, taking longer to recover, but maybe a little better than the 92 recession.

Absolutely nothing of consequence show up that is related to the stimulus. The only thing we got from economists was suppositions, and invented priors, none of which was useful. Any supposition that might work for one of these states, likely would not work for the others.  The double dip in New York would have screwed any results anyway, for that state. Texas Unemployment seemed the same all the way thru the peak.
 
Wouldn't the stimulus obviously help?  No, there are all sorts of reasons the stimulus would not help, too many to list. There is nothing in the data, no reason to create some fiction.