So did the stimulus work?

Dylan Matthews claims that the stimulus worked and he points to some selective studies.  The first is by  James Feyrer and Bruce Sacerdote at Dartmouth.  Here is basically what I wrote up in my book entitled "Debacle" published by Wiley:

One expects that taking resources from the rest of the country and giving it to a particular state will also transfer jobs. The more the favored states get, the more jobs one expects that they will gain. But despite what Feyrer and Sacerdote as well as Krugman claim, that is not the same thing as saying that increasing federal government spending increases the number of jobs at the national level. Indeed, the federal government can end up, on net, destroying jobs at the same time that it moves jobs from those states that don’t get much money to those that get a lot.

The Feyrer and Sacerdote estimates do not consider the distinction between the benefits that individual states might get from receiving more money and that, in the aggregate, the Stimulus can reduce employment. They just look at whether the states getting more Stimulus money end up with relatively more employment. The negative intercept coefficients that they have for their state-level regressions in their Table 3 (columns 1 to 3) are certainly consistent with the Stimulus destroying jobs in the aggregate. Indeed, the intercept is so negative that only Alaska has a net increase in employment as a percent of the population.

Feyrer and Sacerdote's state level data is driven by two states Alaska and North Dakota, and even then primarily driven by North Dakota.  Alaska has almost twice the per capita stimulus dollars as North Dakota and is a real outlier.  No one can serious argue that North Dakota's success is due to the stimulus and not its oil boom.  Eliminating both states finds essentially a zero relationship between stimulus dollars going to a state and Feyrer and Sacerdote measure of percentage of the population employed.

With the later stimulus data released by recovery.gov, I have looked at the results and North Dakota now explains all the obtained relationship across states.

Steve Bronars has some additional notes here.

Some other notes:

I wrote a whole book on this issue, but I can point to some other pieces.
Would the economy have been in worse shape if the various stimulus packages had never seen the light of day?
Canada's cold proof: Job growth without 'stimulus'
Stimulus Spending Is Making Things Worse Not Better