He starts out with the most important point: we really don't have a lot of hard evidence for any question of fiscal policy in a recession (I would add that we perhaps have somewhat more empirical knowledge about monetary policy). The reasons he gives are good: not much data, things change over time, every recession is different, etc. etc.
But he bypasses one of the most important factors: the endogeneity of fiscal stimulus. We stimulate the economy when it is weak, which means that any simple examination of the data is going to bias impact estimates downwards. Glaeser doesn't mention this, which is perhaps understandable - you don't necessarily want to drone on about endogeneity in a forum like the New York Times. But even if he doesn't mention that, he proceeds to plot change in the unemployment rate against stimulus dollars, by state, as if there is no endogeneity problem at all! Which means he fails to mention one of the most important problems with empirical work on fiscal policy, and then proceeds to do empirical work that deliberately flaunts this concern!
There's another issue I have with this post - like lot's of people today, Glaeser completely ignores the fact that we live in a federal republic. He writes:
"There is no equivalent consensus about fighting unemployment and economic downturn. For decades, the economics profession had been moving away from Keynes, but when the recession hit, no one had much of a viable alternative to Keynesian countercyclical spending. We’ve had a $787 billion recovery act — a great burst of Keynesian activity — and unemployment remains at 9.9 percent."No observer that is both objective and well informed can claim that we've seen a "great burst of Keynesian activity". Perhaps we've seen this at the federal level, but it has largely been canceled out by spending cuts at the state and local level leaving no real "burst of Keynesian activity" to speak of. I forget what the GDP figures looked like in mid-2009. Perhaps we had a "spurt" of Keynesian activity as the bulk of the stimulus was pushed out, but it was nothing to write home about. Americans today blissfully ignore their states and counties. In normal times, that's a shame because these governments that are the closest to the people often provide the most innovative solutions to public problems. In downturns, it's a shame because the one thing states can't seem to do right is macroeconomic policy. We look to the federal government only, and ignorantly assume that there's been a "burst of Keynesian activity" when there hasn't been.