Wednesday, October 24, 2012

Probably the single stimulus-related thing that will convince me that either you don't know what you're talking about or you have ulterior, potentially ideological motives if you use the Romer-Bernstein forecast as a reasonable counter-factual.

There are some people with whom I feel safe assuming they don't know what they're talking about.

There are others I know know what they're talking about, which makes me worry.

The most positive spin that I can put on those cases is:

1. They know it's not a reasonable counterfactual
2. They know a reasonable counterfactual is hard to get
3. They're pretty sure a reasonable counterfactual would show a negative impact of the stimulus
4. So fibbing about the Romer-Bernstein forecast is a convenient way of getting  your conclusion across in a way that's easy to understand, even though your analysis is complete horse shit
5. Getting your conclusion across in an environment where there is almost no chance of additional fiscal stimulus is for some reason more important than intellectual honesty

If #1 - #4 were driving you, you might be forgiven if #5 weren't the case. Let's say another stimulus was a serious possibility so it was worth being dishonest to prevent it because you sincerely believed it wasn't good for the country.

But #5 is true so these sorts of people are running out of pseudo-reasonable covers.


  1. A good counterfactual is hard to find.


  2. Why isn't it a reasonable counterfactual? Is it that you're saying there is no such thing, or that we have them, and Romer-Bernstein isn't it?

    1. No, unfortunately we probably don't have one, and Romer-Bernstein isn't it.

      "Don't have one" doesn't mean there aren't reasonable identification strategies. Right now for this crisis all we really have is state-by-state comparisons, which I'm a little skeptical of. But good work can and has been done, I just don't think it's done by comparing actual unemployment to this particular January 2009 forecast.


All anonymous comments will be deleted. Consistent pseudonyms are fine.