The Congressional Budget Office recently estimated that the so-called stimulus generated jobs and growth. I addressed some of the profound shortcomings in CBO’s Keynesian model in a previous post, pointing out that the model is structured to produce certain results regardless of what happens in the real world.

Interestingly, the Director of the CBO, Doug Elmendorf, basically agrees with me. In a recent speech, recorded by C‑SPAN, he was asked during the question-and-answer session whether the model simply spits out pre-determined numbers. After some hemming and hawing and a follow-up question, he confessed “that’s right” when asked if the model would be unable to detect whether the stimulus failed. The relevant exchange begins around the he 39-minute mark of this recording, and Elmendorf’s confession takes place shortly after the 40-minute mark (I selflessly watched the entire thing so you wouldn’t have to suffer waiting for the key moment).

I’m not sure whether this admission is good news or bad news. It is a sign of progress, I suppose, that CBO’s Director is now on the record acknowledging that the model is useless (at least for purposes of measuring the effectiveness of more government spending). But it is perhaps an even more troubling indication of what’s wrong in Washington that nobody is concluding that the time has come to junk Keynesian analysis. This is either an updated version of The Emperor’s New Clothes or a perverse form of the joke about the drunk looking for his keys under the streetlight because there’s light, even though he lost them someplace else.