Sunday, November 17, 2013

Larry Summers at the IMF

Miles Kimball seems very excited about Larry Summers's talk at the annual IMF research conference. For me, this just reveals (if it wasn't already clear) that we dodged a bullet when Summers decided he did not want to be a candidate for Chair of the Fed.

There is really nothing new in Summers's talk. It seems to be either the standard Keynesian narrative on the financial crisis and post-crisis sluggishness in the U.S. economy, or borrowed from Paul Krugman. Here's the basic story. According to Summers, the pre-crisis period was not one of "excessive aggregate demand," but an episode in which asset price bubbles masked underlying secular stagnation. Post-crisis, we're in a state where the "real interest rate consistent with full employment" is very low. In this state, according to Summers,
…We may well need, in the years ahead, to think about how we manage an economy in which the zero nominal interest rate is a chronic and systemic inhibitor of economic activity, holding our economies back, below their potential.

Summers is very good on his feet. If you don't listen too carefully, everything fits together nicely, and he doesn't contradict himself. He knows his audience, and gets a chuckle when he jokes about Minnesota/Chicago (whatever he thinks that means). But what he's actually saying is shallow, and unsupported by serious economic research. Who wants a guy running the Fed who thinks like that?

The standard New Keynesian narrative at the core of Summers's talk is outlined in this post. New Keynesians (NKs) argue that we can model the financial crisis as a preference shock - the rate of time preference fell. The optimal monetary policy response to such a shock in a NK model is for the nominal interest rate to fall to zero (if the shock is large enough). The "natural rate of interest" falls, but the actual real rate falls short of that because the zero lower bound binds, according to the story. The problem for this story in replicating recent U.S. data is that New Keynesian (NK) models revert to trend. Ultimately, as prices and wages adjust, real GDP growth will be determined by exogenous TFP (total factor productivity) growth. As I argued here, it seems implausible to argue that there is enough price and wage rigidity to give us the level of NK inefficiency that some want to argue exists, more than 5 years after the Lehman collapse.

Maybe Summers is thinking the same thing, as what he wants to add to the story is some kind of Alvin Hansen "secular stagnation" mechanism. This seems puzzling, as Hansen predicted in the late 1930s that the economy would stagnate forever, without sustained government spending to support it. Given that the U.S. economy returned to its pre-recession growth trend after the big reduction in government spending following World War II, that idea seems not to have panned out. But Summers wants to somehow connect secular stagnation to the zero lower bound. This is even more puzzling, as it runs into the same problem as in the standard Keynesian narrative. In NK theory, the key problem is that wage and price stickiness misaligns relative prices, including intertemporal prices (i.e. the real interest rate), and if the zero lower bound binds, then monetary policy can't correct the problem. Summers seems to want us to think that we can have permanent relative price distortions - that if monetary policy is hemmed in by the zero lower bound we'll have inefficiency forever. Apparently Summers does not think prices will ever adjust, or he's unaware of some of the simple tax policy solutions that are available, even if we buy into the NK framework.

So, secular stagnation does not seem to be consistent with NK models, as we know them. Maybe Summers has something new in mind. If so, I haven't seen it, and he certainly did not make us aware of any such new work in his talk. That's why this is a bad policy talk, representing only the glib ramblings of a man who hasn't addressed the problem at hand in a serious way.

12 comments:

  1. Interesting. Do the terms "rate of time preference" and the "natural rate of interest" mean the same thing in your post?

    ReplyDelete
  2. No, but if the rate of time preference falls, then the natural rate falls. In a NK model, if wages and prices are flexible, then we get the equilibrium allocation for the underlying neoclassical growth model, and that equilibrium allocation is efficient. The natural rate of interest is the interest rate in that equilibrium. Thus, the natural rate is determined by whatever the fundamentals are in the model - preference shocks, TFP shocks, etc. One property that the neoclassical growth model has is that, in the deterministic steady state, the real interest rate is equal to the rate of time preference. Thus, in the absence of stochastic shocks, the real rate - the natural rate - converges to the rate of time preference.

    ReplyDelete
  3. " . . . we can model the financial crisis as a preference shock - the rate of time preference fell."

    Is there a way to measure how much time preference falls? Or more generally, the distribution of these preference shocks? Surely, such preference shocks must have other observable implications. Otherwise we must regard the distribution of preference shocks as a free parameter.

    ReplyDelete
    Replies
    1. I'm not advocating this approach. This is what NK people are saying. I don't think it makes any sense to think of the financial crisis in this way, but one approach would be to take a basic NK model, and fit it to the data in a way that allows you to back out the shocks.

      Delete
    2. Any thoughts on what sort of data might allow you to back out the distribution over shocks?

      Delete
    3. Just to clarify, my impression is that macroeconomists (and even, say, people who do applied IO) who use preference shocks aren't very careful in justifying exactly how they picked a particular distribution of shocks. It is typically hard to back out the shocks from the model that they are applied in. You have to look at some other kinds of tradeoffs to figure out the shocks. And I'm not sure what macroeconomists who use preference shocks actually do.

      Delete
    4. Again, I think that, if you're trying to model the effect of the financial crisis, and you have to represent it by a preference shock, then you have the wrong model. However, if you buy into this approach, there are ways to measure the shocks. For example, you can write down the model in such a way that it's identified. Then, you construct time series for the shocks that will reproduce the data exactly.

      Delete
    5. I don't disagree that this seems like a lousy way to model the financial crisis. But I was wondering if the proponents of NK do try and take their models seriously. One step towards that would be to try and say something about the distribution of shocks.

      My birds eye view is that the way NK models are written down, there is no way to identify the preference shocks. Using time series to reproduce the data also seems not quite right -- I suspect there is a large parameter space of distributions that will reproduce the same data, that is, unless you have fantastic data, and data outside of the financial crisis.

      What I'm suggesting is that NK people should look at some other (testable) implications of the model that don't have anything to do with the financial crisis, and see if these other implications are consistent with the preference parameters they choose. I suspect this isn't being done. If it is, could you mention a reference? If it isn't, then preference shocks are just a free parameter.

      Delete
    6. Giorgio Primiceri has a paper where they back out something that looks like a preference shock. But it is still insane to think of the Great Recession as a collective bout of patience.

      Delete
  4. That the Wicksellian natural rate of interest is negative right now is pretty obvious and while wage and price rigidity enhance the problem wage and price flexibility will not necessarily solved it.

    Once neoclassical economists can offer a decent explanation for persistent involuntary unemployment I am all ears. Until then I stick with Wicksell and Keynes.

    ReplyDelete
    Replies
    1. Until Keynes can offer a decent explanation for anything that doesn't rely on hand-waving away inconvenient facts I stick with Diamond-Mortenson-Pissarides.

      Delete
  5. "That the Wicksellian natural rate of interest is negative right now is pretty obvious"

    What is the evidence for this?

    ReplyDelete