Believe those who are seeking the truth. Doubt those who find it. Andre Gide

Thursday, September 22, 2011

Interest rates and slumps: competing views

As Paul Krugman notes here, the nominal interest rates on U.S. Treasuries are at historic lows. He seems to take this as vindication for his view that more government "stimulus" was/is needed. (I take "stimulus" here to mean deficit-financed government purchases of goods and services.) Well, let's think about it.

First of all, I think that Krugman (along with many others) deserve credit for recognizing that money-bond swaps (Fed policy) are largely irrelevant in very depressed environments. He (again many others too) also deserve credit for understanding the special "safe haven" role that U.S. Treasury debt plays in today's world economy. But does understanding all this necessarily lead to the conclusion that what the economy needs is more (it never seems to be enough) government "stimulus?"

Some of our economic theories suggest that the answer is yes, while some suggest no. What Krugman is suggesting is that the latter group of theories should be discarded because their predictions on nominal interest rates have been completely wrong. He is getting a little ahead of himself here.

Unfortunately for Krugman, there are theories out there that generate predictions broadly consistent with the data but which do not lead to the same policy conclusion.  One such theory is the "new monetarist" model I published for the Bank of Japan (during my visit there in 2002): Monetary Implications of the Hayashi-Prescott Hypothesis for Japan. (See also here.)

The basic story is this. First, it is conceivable that "real" factors are contributing to a "growth slowdown." Here, one is free to pick your favorite bogeyman. Maybe it's becoming more difficult to expand the technological frontier (see, for example, Tyler Cowen). Maybe it's the fear that people (via their political representatives) will become more interested in appropriating wealth, rather than creating it (see, for example, The Grabbing Hand).  Whatever the case may be, the upshot is that people--investors, in particular--are rationally pessimistic (over the future after-tax return to their investment activities today).

In the model I used in my BoJ paper (an overlapping generations model), rational pessimism generates a "flight to quality"--people begin to substitute government money/debt for private assets. The effect is deflationary (driven by the increase in real money demand). The economy looks like it is suffering from "deficient demand" (it is not). There is downward pressure on the real interest rate (as the demand for investment contracts). These are not crazy predictions.

In my model economy, the central bank has control over the real interest rate, and cuts in the interest rate stimulate investment and (future) GDP. When the nominal interest rate hits zero, the central bank can no longer influence the real interest rate via money-bond swaps (i.e., there is a "liquidity trap"). Real activity may be stimulated, however, by increasing the inflation target (the operation must be undertaken by the fiscal authority in my model; the monetary authority is powerless in a liquidity trap). It might also be possible for increases in G to expand GDP (as is the case in many neoclassical models). All of this is true. And yet, it does not follow that any of these "stimulus" programs are necessarily desirable (among other things, it depends on what social welfare function one adopts).

Now, I'm not absolutely sure about the empirical relevance of my little model. This is because I can think of another theory that generate predictions that are observationally equivalent to my model, and yet delivers very different policy prescriptions.

The model is the one evidently used by Krugman, DeLong, and others (Nick Rowe?). In a nutshell, recessions are caused by an increase in money (treasury) demand. That's just like in my model. But there is a big difference. In their view (as far as I can tell), the pessimism that drives up the demand for money is attributable to "irrational" fear. And if the private sector is afraid of spending, maybe the government sector should step in and take its place.

But perhaps this is not entirely fair. There is, in fact, a literature that explains how expectations can become self-fulfilling prophesies. I could, for example, modify my model to incorporate a form of increasing returns to scale in the economy's production technology. This could generate what economists call "multiple equilibria." Each equilibrium is determined by expectations. If  people are optimistic, good things occur. If people are pessimistic, bad things occur. The pessimistic expectations are "rational" at the individual level, but not at the social level. There is potentially a role for policy here.

Krugman, DeLong and Rowe do not frame things in quite this way, so I'm not sure if this is what they are talking about. But Roger Farmer has been working in this area for a long time and I think his ideas are finally starting to gain some traction; see The Fear Factor.

Anyway, my basic point here is, as always, that we need to be more circumspect in our claims about what we know for sure. Beware of economists that make claims like this


  1. Hi David:

    1. Typo here: "Unfortunately for Krugman, there are theories out there that generate predictions broadly consistent with the data but which do [NOT?] lead to the same policy conclusion."

    2. Yep, count me in, roughly speaking, with those other two guys!

    3. Neat model. But I don't believe it for a moment!

    To my mind, the most important stylised fact about recessions is this: in a recession, it is harder than normal to sell stuff, and easier than normal to buy stuff. It takes a seller an extra long time searching to find an extra buyer; and it takes a buyer less time than normal to find a seller.

    By "stuff" I include newly-produced goods, and labour, and a lot of other stuff as well. Especially stuff the price of which *looks* sticky. Especially stuff that is normally illiquid (hard to buy and sell). It's just that that illiquidity behaves asymmetrically for buyers and sellers over the business cycle.

    The only good I exclude from "stuff" is the medium of exchange. Money becomes easier to sell, and harder to buy, in a recession. It goes the other way from all the other stuff.

    Do I have hard StatsCan data for that assertion? Nope. But massive anecdotal data.

    And I can't even begin to talk about that stylised fact except in a monetary exchange economy, where we buy and sell all other stuff for money.

    Ideally, that stylised fact should be incorporated in a search model. But a crude limiting case of a search model is Q=min{Qs,Qd}. It's impossible to find an extra buyer when there's excess supply, and impossible to find an extra seller when there's excess demand.

  2. Nick,

    Thanks for catching that (important!) typo.

    Yes, I think I understand (and appreciate) your position quite well. In fact, I just suggested today to a graduate student that he might do well to try to formalize and quantify your view (which, if I understand correctly, is subtly, but importantly, different from DeLong's).

    You have anecdotal data. Wonderful. So do I. Try tuning in to CNBC and listen to the business executives who consistently and repeatedly point to "policy uncertainty," "leadership faiures," "vague rules of the game, subject to capricious change," etc. etc.

    Are these executives blowing smoke to cover their asses. Maybe. But maybe...maybe not!

    Btw, what do you think of Roger Farmer's model and how it relates to your own view? (He does not rely on sticky prices because he is an Old Keynesian.)

  3. Nick,

    By the way, if one was to endogenize the search and matching in product markets, the shock I describe may very well lead to thinner markets, more difficult trading, etc.

    The observational equivalence is not so easy to overcome.

  4. David: yep, but don't those business executives also say it's hard to sell stuff, and easy to hire labour, right now? And the observation that recessions are a "general glut" goes back an awfully long way.

    I still can't quite get my head around Roger's model. I get the idea that there's ex post bilateral monopoly/monopsony in the labour market, and so a zone of indeterminacy for wages, between the demand price and the supply price of labour. But from that point, I would want to say that wages are determined by a Schelling focal point, and that history provides a strong focal point. Sticky wages, in other words. If I understand him correctly, Roger is saying that with one equation knocked out, he is free to impose another equation, and take expectations as exogenous?? Because it can't violate RE?? I would instead say there's a thick vertical AS curve, but the AD curve is still well-defined.

    Dunno. My head is still not clear on it.

  5. Oh, and sure thinner markets would likely increase search costs, but why would a reduction in quantities traded mean bigger marginal search costs for sellers and *lower* marginal search costs for buyers?

    We can't even make sense of the distinction between "buyers" and "sellers" in a non-monetary economy.

  6. You have a job, it's easy to quibble over models when the wolf isn't at the door.
    U6 is hovering around 16%, that's millions of citizens that desperately need a good paying job.
    Corporate American is/has/will not be creating millions of jobs any time soon. Only the federal government can create good paying jobs. Deficit spending so people can have money to spend and pay taxes makes sense if you care about people.
    When your life implodes and your living off food stamps and unemployment tell me how your models will save you.

    George K

  7. George K is a crackpot.

  8. Nick:

    Oh, and sure thinner markets would likely increase search costs, but why would a reduction in quantities traded mean bigger marginal search costs for sellers and *lower* marginal search costs for buyers?

    I am not even sure this is a fact, but let us assume that it is. Moreover, let us assume that the model I just described cannot replicate this fact. My claim is this: then your model cannot replicate this fact either.

    Reason? Because the shock that generates a recession in your model is an increase in the demand for money. But that's basically the shock that creates a recession in my model too! (The only difference is that the increase in money demand from my perspective is reacting endogenously to something.)

  9. @George K: Everyone has a model in their mind of the way the macroeconomy functions (whether they realize this or not). We don't expect our models to feed us directly. Our hope is that they will help us interpret the world around us in the best way possible; and that this knowledge will help us avoid any unintended consequences of well-meaning policies. (You may recall the saying that the road to hell is paved with good intentions.)

  10. Hi David,

    I don't think you need increasing returns to get multiple equilibria. If each firm's output and hiring decisions are based on forecasts of sales revenue at alternative levels of production; and if this revenue forecast is driven in part by expectations about the hiring decisions of other firms (or, more simply, whether economic conditions will improve or not), then you've got "pessimistic" and "optimistic" equilibria without regard to increasing returns.

    It's important to differentiate between perfect competition (with diminishing returns) where firms can sell as much as the wish at prevailing prices and, for lack of a better term, "pure competition," where firms are price takers, but nevertheless must forecast prices and sales revenue.

    A reluctance to part with cash in this circumstance is not irrational if you think others will do likewise. The trick is to create some assurance that, if I part with my cash, others will do likewise.

  11. Hi Greg:

    Yes, you are right--don't "need" increasing returns for the result (that's why I said "for example").

    The point remains, however, that there appears to be an observational equivalence across these competing theories. They generate very similar behavior, but have very different policy implications.

  12. Well one thing for fiscal stimulus is that it's not dependent on inflationary expectations.

  13. CL, neither would a stimulus package that released the entrepreneurial spirit from it's shackles.

  14. To Bosephus Jones:
    Sorry to disappoint but I'm an old bond trader, in fact I was one of the first people to hedge bank liquitidy prtfolios in the 80's. We did a lot of yeild curve inhancement trade that made my banking customers a lot of money and did alright myself.
    Told everyone last summer to get out of equities and build a ladder of T bills and short to longterm bonds.
    Your right I'm not an ecoonmist, I guess you would say I'm a "Fat Tony".
    One of my econ professors in the 70's told the class everytime you heard an economist talking about unemployment that person has a good job as an expert and usually doesn't understand what a trageity it is to not have a good paying job.
    He was right.

  15. GK

  16. David:

    As you said, the interest rates on bonds are already low, hence the limited effect of QE. A fiscal stimulus coordinated with a monetary expansion would most likely have the desirable effect. Fiscal stimulus has the advantage of increasing the money supply without actually pushing the interest rates.

  17. CL,

    Well, I'm not sure. I certainly do believe that it might be a very good time for the federal government to issue more debt to finance infrastructure investment. I don't base this view on any "Keynesian" principles of "deficient demand," however. Simple NPV cost-benefit analysis will do.

  18. "One of my econ professors in the 70's told the class everytime you heard an economist talking about unemployment that person has a good job as an expert and usually doesn't understand what a trageity it is to not have a good paying job.
    He was right."

    But he can probably spell "tragedy". And while he may be right about not knowning what it is like to be unemployed, he was wrong about the economist not understanding unemployment.

    Again, GK is a crackpot.