Steve Williamson on Taylor Rules

Steve Williamson responds to my post asking whether it was sticking to the Taylor Rule that got us into the mess [the opposite of John Taylor’s contention, that it was departing from it that was the problem!].

Two points by way of a response.

My first arises, I think, out of a misunderstanding Steve has of my ambiguous drafting.  In my post I explain the result than in the NK models religious adherence to the Taylor Rule produces two steady-states, one involving being trapped at the zero bound.  I argue that this wasn’t the reason the Fed wound up trapped at the zero bound, because those at the Fed ‘don’t use rules like this’.  By which I mean ‘don’t use rules in the sense of believing them to be religiously adhered to’ and not ‘don’t use Taylor Rules for any purpose’ which Steve takes that to mean.  Monetary policy rules are ubiquitous in conversations about monetary policy in the Fed and all central banks.  But talking about them is not enough.  SGU’s paper doesn’t have anything to say about central banks that follow a Taylor Rule unless they think it’s leading them into a liquidity trap and then deviate from it.  Or about central banks that use Taylor Rules as a ‘cross check’ [a frequently used term that in my experience meant not really using them at all].

The point being that without total commitment to the rule, this pathology of the model goes away.

Steve’s second point is a reprise of his contentions in previous blogs that the Fed could raise inflation by raising the nominal interest rate.  And, relatedly, that perhaps inflation is too low because the Fed lowered rates.

This has been debated on the blogosphere extensively.  I don’t have anything new to add to it. A summary of the case against Steve is something like this:

1.  Steve’s proposition is true in flex-price versions of the standard monetary business cycle model.  But in that model, business cycles are of no concern, and the Fed would have no business attempting to smooth them, or doing anything with interest rates.  Rather, interest rates at zero forever is Nirvana, this being the Friedman Rule which equalises returns from holding monetary and real assets.

2.  Micro evidence suggests – though there are a few who dispute it – that prices are sticky.  In the sticky price version of the model in 1. above, Steve’s contention is not true.  Raising nominal rates would raise real rates and depress demand, lower inflation, raising real rates, accentuating that recession, and so on.  Steve and his peers in the ‘new monetarist’ literature often say:  so what?  This model is full of made up stuff that can’t be justified from first principles.  I’m in the camp that worries about this and much admires what Steve and coauthors are doing to construct alternative and ‘deeper’ macro-models.  But for now, the conventional model seems to be the best we have.

3.  Empirical VAR evidence on the effects of identified monetary policy (ie interest rate) shocks suggests that raising rates would have effects similar to those described in 2.  Taken individually, there are lots of shots one can take at papers in this literature.  But taken as a whole, the evidence is pretty compelling.  If you raise interest rates, inflation falls.  Some (like Uhlig, for example) quibble about whether output falls.  Most don’t.


This entry was posted in Uncategorized. Bookmark the permalink.

7 Responses to Steve Williamson on Taylor Rules

  1. Steve Williamson says:

    Hi Tony,

    On (1), (2), and (3), I’m not denying that there are short-run non-neutralities of money. Further, the existence of short-run non-neutralities is consistent with my story. Central bank actions affect two things: inflation and real economic activity. The effects on inflation can be permanent; the effects on real economic activity (other than super-non-neutralities) are temporary. First, suppose I were a central banker who cared only about real economic activity – I’m not concerned with inflation at all. Then, presumably I’m trying to smooth real economic activity over time. I move short-term interest rates up when I want real activity to go down, and I move interest rates down when I want real activity to go up. I understand that I can’t always be doing good things; sometimes my immediate actions will be causing pain. I’m also constrained by the ZLB, and when I hit the ZLB, I can’t do anything but cause pain, unfortunately. Once we’re at the ZLB for a long time, all the positive effects – the non-neutralities – have dissipated. I have to get off zero if there is to be any hope of smoothing in the future. I’m powerless to smooth real activity in the future unless I cause some short-term pain by raising the nominal interest rate.

    Second, suppose that I’m a central banker who cares about inflation as well. Now when I think about the dynamics of the economy, I have to worry about the short and long-run effects of my policies. If I think about policy in terms of pegging a short-term nominal interest rate, I understand that, if the nominal interest rate goes down, and stays there, that the short run effect is more inflation (that’s the liquidity effect at work), but in the long run the Fisher effect takes over, and inflation actually goes down. Again, the liquidity effect ultimately dissipates. So, suppose we have been at the ZLB for a long time. The liquidity effect has dissipated, and the inflation rate is too low. How do I get the inflation rate up? I certainly can’t rely on a short-run liquidity effect, as the nominal interest rate can’t go down. There’s nowhere to go but up, and to rely on the long-run Fisher effect to take over ultimately. Of course, you may have to bear the effects of even lower inflation in the short run because of the liquidity effect.

    So with the short run effects (on real activity, and the liquidity effect on inflation) in there, the story is the same. You’re boxed in at the ZLB. Nowhere to go but up – and staying put does not help.

    • Tony Yates says:

      I see.
      Two thoughts.
      1. If your prescription were followed, then, relative to a rule that implemented optimal policy, or even just a good rule, [both hypothetical things calculated in the absence of the zero bound] this would be a series of very large tightening monetary policy shocks, I surmise of ever increasing magnitude. So there would be a succession of ever worsening short runs.
      2. If for the sake of argument you believed literally in the sticky price RBC model, what accounts for not going with the prescription for optimal policy that Eggertson and Woodford figured out, namely a sharp fall in the nominal rate, hold at zlb for a long time, then slowly raise again as the natural rate rises?

      • Steve Williamson says:

        1. The “prescription” I’m talking about has to do with what happens when you follow a bad rule, get into a bad state, and then have to change the rule. The first step in getting out has to be raising the nominal interest rate. What is done after that is not clear. You have to ask what the optimal rule should be. I’m not sure there are good answers to that question – that depends on what the non-neutralities are. It might be a nonlinear rule by which the policy rate responds less aggressively to the inflation rate as you get closer to the ZLB.
        2. Well, if you think that the financial crisis was all about an increase in patience, that might help you. I don’t think it was. In the paper I cited with Andolfatto in my post, we show that, if the real interest rate is low because of a collateral shortage, then it’s suboptimal to go to the zero lower bound. To understand this you need to be thinking about all the assets and what they do in financial markets. E/W neglects all that, which is too bad.

      • Tony Yates says:

        1. OK, I follow. EW is out the window, because that was ignored when it was appropriate, and we are now trapped, and we take i=0 as part of our initial conditions to program optimal policy. But, a conjecture. In EW what caused rates to need to fall was a temporary but large fall in the flex price equilibrium real rate. Suppose we stay in that world, but ignore the EQ prescription, following instead some not very good policy rule, which leads interest rates eventually down to zero anyway. If the natural real rate is going to rise back to SS, then I’d guess that the optimal policy computed in the same fashion as EW, but starting late, is to hold rates at the ZLB for a while, and let the rise in the eq’m real rate guide nominal rates back eventually. Of course, who knows if what drove rates down can be described in this way – as an exogenous fall in the natural real rate that will go away.
        2. I havent’ read your paper with DA yet. I will. Your comment 2 rang alarm bells though in the following sense. I thought that in an RBC type model with two assets, one safe and another risky, and increase in risk drives the riskless rate down and the risky rate up. In which case if we overlaid a monetary model on top of this, we would find that the average nominal rate consistent with a given inflation target fell. I’ll stop there, though and read your paper before commenting further.

        Thanks a lot for your comments. I understand your position a lot better now.

  2. Anon says:

    The increase-interest-rates-to-increase-inflation scenario is a long-run one, so in a NK model you might want to think of it not as a positive error/shock to the Taylor rule but rather as a one-and-for-all increase in the constant/intercept term.

  3. Tony Yates says:

    I don’t think thinking of it as a constant changes the view that raising rates would be a tightening. Since the TR prescription for rates would fall on account of current inflation being further below the new higher inflation traget.

  4. Steve Williamson says:


    There was no reply button on your last comment above, so I’m putting my reply to your reply down here.

    1. I guess a lot depends on what experiment we are thinking about, whether a change in the discount factor is a good way to capture what is going on, etc. This is complicated enough that you have to write it all down and work it out, I think.
    2. “…we would find that the average nominal rate consistent with a given inflation target fell.”
    Exactly. That’s in part how we can explain why the ZLB policy doesn’t cause deflation. Tight financial constraints raise the liquidity premium on government debt, and lower the real interest rate, so at the ZLB, the inflation rate is higher than it would otherwise be. Then, as constraints are relaxed (in the recovery, with the “production” of more safe private assets) the inflation rate should fall, if you stay at the ZLB. Thus, you could be at the ZLB and still hitting the 2% inflation target, if constraints are tight enough. But that’s not what appears to be happening now.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s