Monday, December 31, 2012

This is a good point from Bob Murphy

The role of prediction in science is not as straightforward as I think a lot of people realize. I've criticized people on here before for assuming economists have a crystal ball and judging the veracity of their science on that basis.

Anyway, with that as context this is what Bob Murphy writes in his response to Brad DeLong:
"Finally my substantive point: I have never seen any of these guys explain why price inflation (and interest rates) are the decisive criteria for whose model and hence policy recommendations are right. Consider, for example, the infamous Christina Romer unemployment graph, showing what would happen with and without the Obama stimulus package. As far as I know, DeLong didn’t ask Romer to announce to the world that she had been wrong about everything and to spend years at the feet of Joe Salerno. No, Daniel Kuehn for example thought that anyone who wanted to use the Romer forecast as a test of the efficacy of her model was putting out “complete horsesh*t.”

So I’m not saying the following–like I said, I screwed up in a way that is relevant for economists talking to the general public. But since the following is exactly analogous to how Keynesians deal with the unfortunate Romer situation, I’m curious why they think Austrians who warned of large price inflation can’t say the following:

"Hey, it’s true, we threw out some predictions of how much prices would rise, and we were off. But our basic model wasn’t wrong, it was just the underlying forecast of the baseline. Bernanke really did create a bunch of price inflation, it’s just that in the absence of Fed action, the drop would have been bigger than we expected, so on net we didn’t see as large of an increase in absolute terms. Indeed, Krugman et al. agree with the economic model involved: they all congratulate Bernanke for having staved off massive price deflation. So what’s the argument here? We’re arguing about the counterfactual of price movements in the absence of Fed monetary inflation."

Seriously, how is the above any different from how Keynesians defended Christina Romer, to the point of saying anybody who thought her prediction should be used against her was intellectually dishonest?"
One important difference with the Romer prediction that bugged me was that it was made in late 2008 and released in early 2009 when we were still getting a sense of what the scope and scale of the problem actually was. And with Romer (as Bob alludes to here), there was also the question of which projection was wrong - the baseline or the actual trendline. [If I am losing readers please speak up in the comments - I'm assuming this is all common knowledge].

Now Bob's bet was a year after that. We knew a lot more than we knew at the end of 2008 about exactly what was going on. Now, if Bob had made the bet in 2008 and wanted to make a case that with more information his theory would have predicted years of depressed prices despite excessive money creation (either because deflation would have been severe or because he would not expect price inflation to be so closely proportional to money inflation under these at-the-time-unknown conditions), he could make a case for that. But if that is the case he wants to make for a bad made in late 2009, I'm not sure exactly what theory he's going to articulate.

Maybe something about secondary deflation?

But why wouldn't this all be laid out by late 2009?

Bob's point here is important. We have to be aware of these things when we're only dealing with counterfactuals. What frustrates me so much about the Romer thing is economists who should know better are implicitly basing their criticism on the argument that the baseline forecast was fine even thought they're too cowardly to explicitly tell you that because they know how ridiculous that is. That's what borders intellectually dishonest. If you need to say "the baseline was a great forecast" to make your criticism of stimulus but you are unwilling to come out and say "the baseline was a great forecast", something is very wrong with your argument.

For me at least, a case needs to be made. There's a good case that in late 2008 we really didn't have a complete sense of what was going on and that a baseline forecast was highly speculative. You can criticize Romer for that forecast but don't dismiss the policy effect on the basis of actual unemployment unless you are willing to embrace the baseline forecast.

Were we in the same position in late 2009? That seems more doubtful to me.

It seems to me we have a couple options here:
1. Bob is right about the impact of monetary policy and the counterfactual here is much severer deflation than we actually saw. Bob's baseline - a year into the crisis - was thus too rosy, which lead to the error of his prediction.
2. Bob is wrong about the impact of monetary policy and his baseline assumption was fine. Without Fed intervention prices would have been subdued but not crashing - he was just wrong that with Fed intervention they would be soaring.
In the case of #1 his theory of how the macroeconomy works seems problematic or his theory of how the macroeconomy works was fine and in late 2009 he just didn't have the data to make the forecast accurately. The latter seems somewhat less probably to me than the former because by late 2009 we seemed to have a better sense of things.
In the case of #2 his theory of how monetary policy works seems problematic and we are assuming his theory of the macroeconomy is fine.
Is there a #3 I'm missing?
I just think it's worth getting a grip on what Bob and his fellow travelers think is the best way to grapple with this - because let's be clear, he was not the only one worried about inflation back in 2009.
Is it just a bad forecast? Is that what Bob wants to argue? And is that because we didn't know the lay of the land in 2009 or is it because the theory behind the forecast was problematic?
These are the questions that Brad's post the other day made me curious about.
Brad may not actually be curious about the answers. Bob may not feel like giving the answers. But that's what I'm curious about.


  1. If you felt Romer's predictions would have led to much less stimulus than was needed, I can see the problem with hers, but I can;'t believe anyone thinks that. OTOH if you felt following Murphy's would have led to a much worse situation, I can see a big problem with that. Not only the magnitude is wrong, but the sign of the recommended advice. If that is not a reason for reassessment, I can't imagine what is. It says we are better off listening and inserting negatives into whatever he says.

    1. Whether we accept Bob's understanding of inflation and whether we accept his policy advice seem to be two related, but different things.

    2. Lord wrote:

      Not only the magnitude is wrong, but the sign of the recommended advice.

      Right, if we assume a priori that big government deficits boost employment during a slump, then we don't need to take Romer's bad forecast to mean that big government deficits might reduce unemployment.

      It's amazing to me how circular you guys are, and don't even realize it. You keep running around pointing fingers at your ideological, head-in-the-sand, anti-empirical opponents.

    3. If you have an argument for why you're in the same boat as Romer you're going to need to make it soon or I'm going to start assuming there isn't one.

    4. It's simple.

      1. Deflation reduces people's income.

      2. A reduction in income without a corresponding reduction in prices and reduction in nominal debt levels results in a lower level of disposable income.

      3. A lower level of disposable income results in less spending and lower aggregate demand.

      4. Less spending and lower aggregate demand results in deflation.

      If you're an Austrian economist and you believe in the validity of praxeology, then there is absolutely no reason why you should reject Keynesian economics.

  2. If you assume Romer got the sign wrong, you would also have to discard her base prediction in the first place, so it wouldn't be she got the sign wrong but was useless and irrelevant. Is that what we should assume of yours?

  3. I think there is a #3 you are missing which is that Bob underestimated the feds ability to target a specific inflation rate. It increased the money supply enough to stave off deflation, then stopped at a low positive rate of inflation.

    Bob's prediction appears to be based on a view that the fed either would have allowed , or would not have been able to prevent, 10% inflation from happening. I'm sure Market Monetarists would disagree with this view.


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