Paul Krugman recently commented that the extent of the stimulus provided by increases in government purchases has been significantly over-estimated:

"That is, for all the talk of a failed stimulus, if you look at government spending as a whole you see hardly any stimulus at all. And with federal spending now trailing off, while big state and local cutbacks continue, we’re going into reverse."

Just how accurate is this? In the 4th quarter of 2007 (that is, at the peak of the business cycle as defined by the NBER Business Cycle Dating Committee), combined Federal, state, and local government purchases of goods and services totaled 18.4% of GDP. In the second quarter of 2010, that total was 19.4%--of a smaller GDP...In fact, government pruchases as a percent of GDP peaked in the second quarter of 2009, at 19.9%, and have declined since then.

In fact, government purchases grew at about a 2.1% annual rate between 1960 and 2007. At that rate, government purchases in the second quarter of 2010 would have been about $2,602 billion (annual rate). Actually, government purchases were less than that--$2,567 billion (annual rate). [One can argue whether having government purchases grow at that rate is appropriate or not--government pruchases grew more slowly than did GDP (which grew at a 3.1% annual rate), but that's a different question.]

So Krugman is unquestionably correct. Counting state and local government purchases (which have actually declined by 2.1% since the beginning of the recesion) as well as Federal government purchases (which have grown significantly, but the increase only slightly offsets the S&L decline--combined, state and local governments purchase considerably more than does the Federal government), the amount of fiscal stimulus was quite small.