Economists like to use phrases like: “sure, poverty and unemployment are on the rise (or whatever) but things would have been even worse if POLICY X had not been implemented.” Conversely, political scientists use phrases like: “sure, with the recession the incumbent party was set to lose seats but they would have lost many more seats if CAMPAIGN X had not been implemented.”

You guessed it: these counter-factuals are never observed–academics and experts just estimate these counterfactuals with a little help from theories and statistical methods.  But what if the estimated counterfactuals are really off the mark? In real-world policymaking, this makes accountability an almost impossible task…

Greg Mankiw tackles this very issue here.