Wednesday, September 5, 2012

Horwitz and DeLong on policy in our depression

Steve Horwitz has an interesting discussion at the Adam Smith Institute on policy in this depression and specifically why "Krugman is wrong". If you are not familiar with Austrian macroeconomics it's an excellent summary. But of course, he also critiques the Keynesian response, and these portions are less strong in my view. I have two nit-picky concerns that nevertheless get to more fundamental issues - one that I'll address and one that I'll let Brad DeLong's recent post address.

Nit #1: "Stimulus failed"

Steve spends a lot of time talking about the failure of the stimulus package - not just the failure to get us out of the depression (many of us said at the time it wasn't up to that task), but the failure to do any good at all. I'm most troubled that he used the Romer-Bernstein graph to make the point. There are two reasons to show it:

1. To argue that forecasters in December 2008 did not get it right, or
2. To defend forecasters in December 2008 as presenting a viable counterfactual.

It's hard to say exactly where Steve falls because he doesn't frame it like I just did, but in the talk he seems to support the first option, and yet he also presents it as being bad news for stimulus. This seems like an extremely misleading thing to do. If you accept the first option, the only answer an economist really has any business providing on what it says for the stimulus is "we can't know exactly without a counter-factual, but we can rely on past evidence - but this graph that I am showing you understates the positive impact of the stimulus because of the poor forecast".

I could be misreading him and he could be arguing just for #2, but I don't think he's doing that. I think #1 is far more defensible myself. The problem is that Steve seems to want to have his cake and eat it too: he wants the premise from #1 but the conclusions from #2. It's not just him of course. You see this a lot. I don't understand it. Steve understands the problems around inference from non-experimental data, so why does he promote the graph as demonstrating the failure of the stimulus?

This has bigger consequences, and this is why this nit is important. In a world where "are you better off than you were four years ago" is perceived as a sensible question for evaluating a candidate, people who understand counterfactuals and inference from non-experimental data (like Steve Horwitz) ought to be clarifying this sort of thing to the public, not confusing them on what this graph means.

Nit #2: Krugman and people like him don't understand that interest rates coordinate intertemporal choice

I was surprised the first time I heard him say this, but then he revisited it several times. Does anyone honestly believe Krugman and people who agree with Krugman don't understand that the interest rate is a price signal?

A lot of people think this is what the argument is over - the market process. I disagree strongly. If you want a good example of a Keynesian understanding the interest rate as a price signal and raising other quite different concerns, I suggest you read Brad DeLong's Platonic dialogue today. The Hayek/Mises stuff is subtext until Sokrates jumps in farther down, but if you're at all familiar with the Austrian argument you should see it much earlier. This is a critical point in the exchange, from the perspective of the discussion with the Austrians (yes I know this critical "point" is like half the dialogue):

"Hypatia: A second worry is that because of tight fiscal-loose money the real rate of interest is not doing its job of signaling the true optimal first-best shadow price of intertemporal choices. It is making the price of investment too low.

Glaukon: Thus a tight fiscal-loose money economy would seem to be one that invests too much and that weights the composition of its investment too heavily toward long duration assets.

Hypatia: Yes. The price at which savers lend and at which investors borrow and choose investment projects is false. False prices produce transfer rectangles--in this case, a transfer from savers to investors--as well as excess purchases of the investment goods whose market price is less than their social resource cost.


Glaukon: But can you elucidate to me the harm from purchasing investment goods whose private purchase is profitable just because the interest rate is below the first-best optimal intertemporal tradeoff shadow price?

Hypatia: I don't think I understand: price is below shadow cost; purchases are too high; Harberger triangle; deadweight loss.

Glaukon: Let me see if I can say where I am having my problem: I understand pollution externalities--we undertake too much polluting activity because we do not factor the harm of pollution into the market price and so there is a deadweight loss Harberger triangle composed of the harm rectangle imposed on society as a whole offset by the triangle that is the benefit to private polluters minus the private cost.

Sokrates: That is completely correct, Glaukon.

Glaukon: But in the case of over-easy monetary policy I do not see this. I the benefit triangle to investors--they get to make investments that are privately profitable. I see the transfer rectangle. But where is the harm, exactly?

Tiresias: I believe that the harm must somehow arise out of the fact that consumers are non-Ricardian.

Klio: Non-Ricardian?

Tiresias: Yes, if consumers are Ricardian, then if taxes are too high they simply borrow to pay their extra taxes. Consumption and investment stay what they would be at first-best fiscal policy. There is no aggregate demand shortfall for easy money to correct."

My paraphrase of Brad's response to the Austrians is that yes, ultra-expansionary monetary policy, if we were to adopt it, would move us below the natural rate, and that will throw investment and consumption off balance and produce elongated capital structures like you're worried about. But it's critical to recognize that that is a cost associated with government spending that is below trend in a non-Ricardian world and of credit market problems. Loosening monetary policy is the best alternative to the fiscal and credit market problem. Another way of putting it is the old saying that "it takes a lot of Harberger triangles to fill an Okun's gap".

Yesterday he presented a somewhat different argument, which I am more amenable to: that we are actually above the natural rate right now, which is why the Wicksellian perspective calls for monetary expansion. There is no false signal being sent. In fact, if I were to put my Hayekian hat on I would be worrying that firms are getting the wrong signals to invest too little and to have production processes that are too short because we are above the natural rate.

How to square Brad's post yesterday with his post today? I'm assuming the monetary policies he's discussing today are genuinely "ultra-loose" in the sense that White meant it - below the natural rate - in order to compensate for the failure of fiscal policy makers.

So there are two responses to the Austrians:

1. You're right but those are costs that we incur because we're in second-best territory. If your argument is that first-best is better than second-best we agree. But these Harberger triangles still don't outweight the Okun gap.

2. You are right about the response of entrepreneurs to the interest rate but wrong that we are below the natural rate right now: all signs suggest we are above it.

Both are appropriate, but I think #1 is unnecessarily deferential to the Austrian assumptions about the natural rate.

Anyway, one thing is clear: Brad DeLong understands that interest rates are signals that coordinate intertemporal decisions.


  1. 1. Re: "Steve understands the problems around inference from non-experimental data, so why does he promote the graph as demonstrating the failure of the stimulus?"

    Could it be because Steve Horwitz is not in the education business but the propaganda business?

    2. Re: "'Krugman and people like him don't understand that interest rates coordinate intertemporal choice.' I was surprised the first time I heard [Steve Horwitz] say this, but then he revisited it several times. Does anyone honestly believe Krugman and people who agree with Krugman don't understand that the interest rate is a price signal?"

    See my answer to (1).

    There comes a point when you have to divide people into those who are trying to understand what is going on and those who are trying to pull the wool over their (and your) readers' eyes.

    I see no evidence that Horwitz has not long ago crossed that boundary and left honest discourse behind--and here today we have two more examples...


    Brad DeLong

  2. I would say that the natural rate in Wicksell's sense is the rate that matches aggregate demand to potential output, and that that natural rate will only be equal to the first-best rate of intertemporal transformation is fiscal policy gets it right (or if households are Ricardian). If not, than there is a distinction between them. Thus right now the first-best rate of intertemporal transformation would be a positive interest rate, but--because fiscal policy and banking policy are too tight--the Wicksellian natural rate is less than zero...


    Brad DeLong

  3. ""we can't know exactly without a counter-factual, but we can rely on past evidence - but this graph that I am showing you understates the positive impact of the stimulus because of the poor forecast"."

    Of course, if posits a negative multiplier, which many Austrians do, the forecast without the stimulus could have been just right, and the stimulus simply had the exact opposite effect that RB thought it would have.

    (*I* don't think a negative multiplier is likely, but it would make Steve's position consistent.)

    1. I should review that section again, but it seemed pretty clear he agreed it was reasonable that the forecast wasn't right. If that's the case, then even if you expect a negative stimulus the naive reading of the graph will still give you a lower multiplier than is true, right? It will still look more negative than it actually is.

      Ya, obviously what a good counterfactual will be is a question we can (and do) argue about until the cows come home. What concerns me is statements about the graph that make the general public feel like they can just take the graph and do policy evaluation with it. That is wrong and nobody should be giving that impression regardless of what you think the right multiplier was.

      And I think that point is value-free.

  4. Daniel,

    ==> I can't watch Horwitz's talk; I have too much stuff going on. Can you point me to the discussion about the forecast?

    ==> Regarding interest rates: Brad DeLong is *way* more knowledgeable about this stuff than Paul Krugman, at least from my reading. So if Horwitz said, "Krugman doesn't know about intertemporal structure of production, he just thinks interest rate regulates total spending" (or something), you're not refuting this by pointing to DeLong posts.

    ==> I didn't go read the whole thing, but from the excerpt you've given here, it sounds like DeLong is saying, "Yes, I understand you Austrians claim there is an intertemporal coordination problem, but I just don't see what the problem could be." So even on your own terms, how is this supposed to reassure Horwitz? He claims Krugman (and all Keynesians?) don't get the intertemporal coordination problem. It sounds like you should say, "Right, we don't, because there is no such thing." Not, "We do too get it!"

    1. All good questions.

      1. Discussion of RB starts at 13:30. At about 14:30 he seems willing to accept the forecast was bad, but doesn't strongly commit to it. There is another strange way that Steve reacts to this too. Like a lot of people looking at this graph, he seems to assume that forecasting models and models producing multiplier estimates are the same thing. This clearly isn't true. Keynesian macroeconomists and macroeconomic forecasters are two very different groups of people working with very different tools and models.

      2. Go a little before the start of that discussion (13:15 or so), and then also throughout the discussion, and you'll see he refers to both Krugman and DeLong, and he does throughout. He starts discussing interest rates as a signal at around 20:00. I can't find the detailed interest rate discussion right now, but I don't think it makes a difference if he said K & D or just K. Come on - do you really think Krugman doesn't understand this? He's talked about it a lot on the blog, including recent posts on DeLong's discussion of Wicksell. This is sort of first year micro and macro stuff right? I don't think there's an economics undergraduate out there that fails to understand this, much less a Nobel laureate. I really don't understand this criticism.

      3. I might need to reread, but I thought DeLong was saying that the arguments make sense but the resulting inefficiency is insignificant relative to the other problems: it takes a lot of Harberger (or in this case, Hayekian) triangles to fill an Okun gap. Now what they think of Austrian business cycle theory is VERY different from what they think of interest as an intertemporal coordinator. I don't see anywhere where he's downplayed that point but perhaps you could clarify. I don't see why ABCT and interest as a coordinator are one and the same. You can rely on the latter and not think the former is significant: that's modern mainstream macroeconomics!!

    2. Daniel wrote:

      This is sort of first year micro and macro stuff right? I don't think there's an economics undergraduate out there that fails to understand this, much less a Nobel laureate. I really don't understand this criticism.

      No, absolutely not. I can't even believe you said this. (I'm not offended, just astounded.) I am willing to entertain the idea that Krugman knows about it, since he does talk about Wicksell (via DeLong, and I'm not sure Krugman's IS-LM interpretation of Wicksell is right), but give me a break. No, first year micro and macro students absolutely do NOT have any freaking clue of what Austrians mean when they warn that screwing with interest rates will mess with the intertemporal structure of production. I think maybe 5% tops of professional economists could even reproduce to my satisfaction what we mean by such a worry, and not all of them would agree it's a problem.

    3. Right, but the interest rate as an intertemporal coordination signal and the interest rate as a determinant of the capital structure are two very different things!!!

      I agree the capital structure point is not a widely appreciated concept.

      The idea that the interest rate is a price signal that coordinates intertemporally is.

      I think your 5% figure is high if you're talking about the capital structure (not because the concept is beyond them, but just because they don't know about it). I think the percent that understand that the interest rate coordinates intertemporal decisions is pretty much 100%

    4. Daniel, look: The first (and last?) time Krugman publicly responded to me, he actually said that empirical research showing that the Fed can influence the timing of the business cycle, is a strike against Austrian business cycle theory. (!!!!!) <=== I can use a lot of punctuation marks just like you!

      So don't tell me principles students know all this stuff, and understand how a false interest rate can lead to distortions. The only thing principles students are taught is that if the central bank has "too easy" monetary policy (which, at that level, is associated with keeping interest rates "too low") then you get too much inflation. That's it, end of story. 95%+ of principles classes wouldn't say one word about, "And this of course sets up an unsustainable intertemporal structure of production, leading to a necessary recession in a few years because of underlying real distortions."

    5. You and I seem to be talking about extremely different things.

      I was talking about whether the interest rate is a price that coordinates intertemporal decisions.

      You seem to be talking about a particular model of the capital structure and the question of whether a certain monetary policy is good or bad.

    6. I agree - most don't know about the more specific argument that you are talking about.

    7. OK maybe we're talking about different things then; I haven't listened to Horwitz yet.

      But even your weaker claim doesn't work for intro classes, unless by "intertemporal decisions" you mean "how much to save?"

      Krugman seems to think the only real problem is if the market interest rate doesn't yield full employment. This isn't what Horwitz means by "intertemporal allocation" or "intertemporal coordination."


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