Initially Assuming The Final Conclusion

Louis Woodhill nails it on the evaluation of the Congressional Budget Office on the effect of Obama's stimulus on output and unemployment:

"The striking thing about the CBO analysis is that it is "reality proof". Rather than presenting evidence that "stimulus" works, the CBO employed economic models that assume that "stimulus" works. As a result, no matter what happened to 2Q2010 GDP and employment, the CBO's calculations would always show that "stimulus" worked exactly as intended....

To estimate the impact of "stimulus" upon 2Q2010 GDP and employment, the CBO applied "output multipliers" to the reported expenditures in eight different categories. Their calculations did not include any actual economic data. Accordingly, when the BEA reduced its estimate of second quarter GDP downward by 33% on August 27, there was no need for the CBO to revisit their calculations."

(Actually it was GDP growth that was revised down by a third, but that was probably what Woodhill meant, and it doesn't affect the underlying point)

Thus, the report wasn't really a study of how the economy performed after the "stimulus" was implemented. All it showed was that if you use the assumption that Keynesian stimulus work while analyzing a specific case of Keynesian stimulus, the conclusion will be that this particular case of Keynesian stimulus worked.

While the report might nevertheless make interesting reading for those who believe in the models and their alleged justification in the form of previous similar studies, it is not really an argument against those of us who have argued against such models when they were presented before the "stimulus" was implemented.