This is a recent talk she gave on estimating multipliers, and if you have ever caught yourself saying something like this hypothetical from her talk:
"If you press people for why they think this they will probably say something like, “It’s not rocket science—all you need are two good eyes to look around you. We spent all of this money and the economy is still terrible. It obviously didn’t work.”
Then you should really read the part that she writes next:
"Well, the theme of my talk this evening is that it is not that easy. Estimating the effects of fiscal policy may not be rocket science, but it is incredibly hard. The reason that it is hard is that fiscal actions are often taken in response to other things happening in the economy. Separating the impact of those other factors from the impact of the tax changes or spending decisions is very difficult. It requires many of the sophisticated techniques in the economist’s tool kit—along with a big dose of creativity, and plenty of plain old-fashioned hard work."
As well as the rest of it.
HT Brad DeLong