Sunday, September 6, 2015

Bad Ideas?

Paul Krugman concludes that "hiking rates now is still a really bad idea." So, his opinion is clear. What's not so clear is his argument, which is this:
When the Fed funds rate was 5 percent, there was room to cut if a rate hike turned out to be premature — that is, the risks of moving too soon and moving too late were more or less symmetrical. Now they aren’t: if the Fed moves too late, it can always raise rates more, but if it moves too soon, it can push us into a trap that’s hard to escape.
So, suppose we're in the pre-financial crisis era, and the fed funds rate is 5%. As a thought experiment, suppose the FOMC decided at its regular meeting to hike the fed funds rate target to 5.25%. Then, at its next meeting it decided that the previous hike was a mistake, and undid it, reducing the fed funds rate target to 5%. I think Krugman is telling us that, in those circumstances, ex post we would prefer the policy that stayed at 5% to the one that went up a quarter point and then back down. I think he's also telling us that, once we discover the mistake, the best policy would be to reduce the fed funds rate below 5%. That's the basis for the asymmetry argument he's making - there's no problem if you're at 5%, but when you're at zero (essentially), you can't correct the mistake. So, fundamentally, this argument revolves around the assumption that there is an economically significant difference between going up to 5.25% this meeting, then down to 5% at the next meeting, vs. having stayed at 5%.

If that's the crux of it, Krugman needs to do a better job of making the case. In terms of modern macroeconomic theory, we don't think in terms of "too early" and "too late." Policy is state-dependent, i.e. data-dependent.
The policymaker takes an action based on what he or she sees, and what that indicates about where the economy is going. The question is: What is Krugman's desired policy rule, and where would that lead us? What exactly is the nature of the "hard to escape" trap that might befall us? As is, Krugman's not giving us much to go on.

Addendum: Here's another thought. Krugman seems to like the "normal" world of 5% fed funds rate better than the zero-lower-bound world - because, as he says, the normal world allows you more latitude to correct "mistakes." So why wouldn't he use that as an argument for liftoff?

42 comments:

  1. Cullen Roche - There is no Defensible Argument for Raising Rates at Present http://www.pragcap.com/there-is-no-defensible-argument-for-raising-rates-at-present/

    ReplyDelete
    Replies
    1. Here's Cullen Roche's argument:

      "First, the natural overnight rate is 0% because a banking system with excess reserves will bid the overnight rate down to 0% naturally."

      That's incorrect. A banking system with excess reserves will bid the overnight rate down to the interest rate on reserves - if there are no frictions. There are some technicalities in the U.S., which is what the ON RRP facility deals with.

      "Second, raising rates 25 bps does not provide ammo for later on. If cutting rates by 500 bps over the last few years didn’t spark a recovery then why would cutting from 25 bps?"

      So, it looks like he's saying it doesn't matter one way or the other. So he should have no objection to liftoff.

      "Third, the Fed is contributing to global financial instability by watching the dollar climb ever higher so that argument doesn’t hold much water."

      I thought changes in domestic interest rates don't matter much for the U.S. economy. But apparently this is a big deal for the rest of the world. Seems inconsistent.

      Delete
    2. Hi Stephen,

      Long time (lurking) reader. Thanks for everything you do.

      Yes, the banking system will bid down the int rate to whatever floor the Fed sets (if it chooses to). Otherwise, excess reserves will send the rate to 0%. My point is that the Fed has to LIFT the overnight rate from 0%. People who say the overnight rate has been "manipulated" lower are misunderstanding the dynamics here.

      It's true that I don't think the liftoff will make a huge difference either way, but I am not positive about that. I see more risks than benefits though which is why I would hold off.

      - Cullen

      Delete
  2. Perhaps Krugman is concerned about reputational damage to the Fed which might hinder future policy implementation? Markets might get the idea that the Fed doesn't know what it's doing. It could do expectational damage - raising expectations that inflation is coming more quickly than it might otherwise - effectively firing the inflation starting gun too early? All this might be cause for unneccessary instability?

    Henry.

    ReplyDelete
    Replies
    1. So the argument would be that you stay at zero because you're terrified of making an error?

      Delete
    2. I don't know. I'm just saying what I am guessing that Krugman might have had at back of mind in making his post.

      Are the concerns I raised above valid?

      Henry.

      Delete
  3. " Policy is state-dependent, i.e. data-dependent. "

    What about managing expectations?

    Henry.

    ReplyDelete
  4. "In terms of modern macroeconomic theory, we don't think in terms of "too early" and "too late." "

    Why not? Thinking otherwise suggests you know what you are doing with a great deal of certainty. Can't see that would be the case in any policy making situation.

    Henry.

    ReplyDelete
    Replies
    1. Steve is saying that policy is state dependent, not time dependent. You don't lift off too early, you lift off in some state that perhaps is too high or low relative to some optimum.

      Delete
  5. Isn't he saying liftoff to early = liquidity trap while liftoff too late doesn't entail any such risk?

    Of course he's assuming everyone agrees with the idea of the liquidity trap.


    All these economists saying liftoff too late carries no risk makes me nervous.

    ReplyDelete
    Replies
    1. It's clear he thinks the zero lower bound is important. It's important, I think, to distinguish between liquidity traps and lower bounds on the nominal interest rate. As long as there is a significant stock of excess reserves outstanding, there is a liquidity trap - in the sense that, given the interest rate on reserves, a swap by the central bank of reserves for Treasury bills does nothing. But changing the interest rate on reserves matters, though the interest rate on reserves can't go below zero. Krugman appears to think it's important that the interest rate on reserves can't go below zero, in terms of evaluating the costs and benefits of liftoff.

      Delete
  6. Krugman's implicit idea here is that the consequences of mistakes by the Fed have some kind of unspecified endogenous persistence, especially on the downside: if you raise rates too high when they're near zero, you might push the economy into recession, and then that recession might continue even if you lower rates back to zero again.

    Is Krugman right about this? I don't know. Certainly there is no direct support for it in the New Keynesian model, as long as the "natural rate" is assumed to be exogenous: the model is entirely forward-looking, and bygones are entirely bygones. Some additions to the model even push in the opposite direction - if you add capital to the model, for instance, then by pushing rates too high at time 't' a policymaker will leave a lower capital stock at time 't+1', leading to a higher natural rate at 't+1'.

    With time-to-build or a relative investment friction it might go the other way, but it's not clear whether this would be quantitatively significant: if forward-looking bond investors knew that the mistaken interest rate policy would be reversed quickly, then long-term rates would barely be affected in the first place. (And this is a more general problem with all views of the form "X central bank made a horrible mistake by briefly having short rates slightly too high": unless you believe that the Fed's current short rate alone - regardless of longer-term rates determined in the bond market - has a very large impact on the economy, it's not clear that the Fed is even capable of doing much through these short-lived mistakes. One possibility is to say that the market is stupid and incorrectly thinks that the Fed won't reverse its policy once warning signs appear - which is fair enough, but probably wouldn't be endorsed by many market monetarists. Another possibility is that by raising rates, the Fed is revealing something damaging about its implicit policy rule, and that this revelation will continue to have consequences even once the initial mistake is reversed. This is also possible, but without further clarification it's maddeningly vague.)

    The difficult reality is that we don't have a good theory of macroeconomic inertia. We don't have many workable mechanisms that produce inertia endogenously, rather than as a reflection of inertial fundamentals - and we have especially little backing for the notion that recessions specifically have inertia of their own. The kinds of models that one might think would produce inertia, like labor search models, actually converge back to the steady state far too quickly to produce inertia under standard parameterizations.

    ReplyDelete
  7. [Continued from the previous comment.]

    What to do in this situation? Well, I am still somewhat sympathetic to Krugman - because I believe that rational policymakers cannot just wave away all ideas that currently lack a clear model to back them. (Accordingly, I believe that model fundamentalists - although they do science a great service by insisting that we codify our ideas into models - are irrational and even somewhat deranged at policy. There have been many times in economic history when relying on the best model available would have produced disaster, and we're lucky that policymakers were a little more openminded than that.)

    And it is easy enough to suggest what forces might ultimately support Krugman's view. If prematurely higher interest rates prompt a downturn and in the process destroy public confidence, lower rates afterward might not be enough to fix the damage. Now, this is terrifyingly difficult to model, because we have no idea what determines possibly non-RE "public confidence", and even granting that we also need some kind of aggregate demand externality for individually suboptimal waves of optimism and pessimism to inflict broader social damage. But many parts of the strong ring true, and prudent policymakers can't await the economics profession's modeling epiphany 20 years from now. If the costs of waiting aren't that high, then Krugman could easily be right.

    (Let me add, to further muddle the discussion, that Krugman's greatest omission is his continued focus on the ZLB itself, even when Switzerland and other countries have proven that it is possible to go below zero. He is extremely ready to promote conventional liberal policies as a second-best remedy due to the ZLB, but is puzzlingly hesitant and dismissive when it comes to the possibility of breaking the ZLB entirely - raising practical questions of implementation, for instance, and studiously avoiding any followup with potential answers. Obviously, this is a case where politics has infected his judgment.)

    ReplyDelete
    Replies
    1. Well, that was all very thoughtful. I think the bottom line is that there is a lot we don't know, but would like to.

      "The difficult reality is that we don't have a good theory of macroeconomic inertia."

      Yes, exactly. In the models, inertia is either built into the shocks themselves (as for example in the stripped down RBC model) or built into the propagation mechanism via ad-hoc partial adjustment, and typically both. But a recession appears to have a huge amount of inertia - policymakers can mitigate the effects, but seem incapable of getting rid of the downturn altogether.

      I like a story that can be backed up by a serious model - that keeps the policymaker honest, and consistent. But we know that good modelers can be bad policymakers, and that no model is ready to be taken seriously in a quantitative sense for doing monetary policy. Someone with long central banking experience can be very useful. These people have seen a lot, and understand through experience the pitfalls of particular policy moves.

      In this instance, thought, we haven't got a lot to go on in terms of past experience. I like this:

      "... it's not clear that the Fed is even capable of doing much through these short-lived mistakes."

      If you look at the history of central banking in the U.S. over the last 46 years (Volcker and beyond), there have been some very large policy moves that have not plunged us into disaster. Early in Volcker's term the fed funds rate goes down 8 percentage points, up 10, down 5, then up 4 before declining continuously. In the last tightening cycle, the fed funds rate went up 4.25% from 2004-2006, with a quarter point increase each FOMC meeting. Currently, the Fed appears to be contemplating a quarter point increase, and Janet Yellen is on record saying that any further increases will be data-dependent and measured. That doesn't seem like a policy prone to disaster.

      Delete
  8. I'm see three different "ideas" in various places.

    One says the monetary base is the medium of account (MOA). That is all that matters.

    Another says the monetary base is the medium of account (MOA). However, demand deposits of the commercial banks are medium of exchange (MOE). It also says the commercial banks work to keep the demand deposits near the same value as the monetary base like a foreign currency where there is a fixed one way peg, but not both ways. I believe this means the commercial banks will not produce too many demand deposits causing price inflation because there would be a "bank run" as demand deposits get "cashed in" for currency. The central bank will not allow the "bank run" because it would violate the price inflation target.

    A third "idea" says that currency and demand deposits of the commercial banks are both MOA and MOE. It is like a foreign currency where the fixed peg is two ways. If there is a "bank run", the central bank will maintain the fixed pegged rate by increasing currency using its lender of last resort / elastic currency function even if price inflation goes above target as long as the commercial bank(s) is/are solvent.

    Which of these is correct?

    ReplyDelete
    Replies
    1. Well, I don't recognize anything that I'm very familiar with in any of those three stories. What do you think? What is it that you want to do with these ideas?

      Delete
    2. I think it is the last one.

      There are some people who say a recession is when the demand for "money" goes up. There are some people who say a recession is about prices of goods/services being too low. In both cases, they say all the central bank needs to do is buy a bond and sell currency or central bank reserves to supply more "money" and/or raise the price level of goods/services to fix the recession.

      Delete
    3. In terms of the macroeconomic models people like to think about, we would say a recession is caused by some aggregate shock to the economy. In some theories, the economy need not respond to the shock in an optimal fashion on its own, and then there may be a role for monetary policy in correcting an inefficiency. So the case for staying at zero would be that this is more efficient than the alternative, not just for today, but thinking about the entire future.

      Delete
    4. Let's assume there was some shock to the economy. It requires more "money". This is believed to be temporary. It may or may not be temporary.

      How does the more "money" happen?

      Delete
    5. I'm not sure what you're getting at.

      Delete
    6. There is some aggregate shock to the economy. Price inflation comes in below target. Market interest rates fall. The central bank goes along and lowers the fed funds rate also. More private debt is created thru the commercial banking system so demand deposits go up. With the third "idea", that means "money" goes up, hopefully increasing quantities and prices. Price inflation goes back to target or above.

      Delete
    7. Try it this way. There is some shock to the economy and inflation goes down. Short-term nominal interest rates don't go down unless the central bank reduces its policy target interest rate. So, suppose we're in the pre-financial crisis world. Then, the central bank reduces the fed funds target. What happens next? Well, the conventional idea is that, in the short run, inflation will go up. But, we know that, in all of the monetary models we have, that if the central bank reduces nominal interest rates permanently, that this will actually reduce inflation. So the long-run effect works the other way. Typically, at best we're relying on a short-run effect to push inflation back to target, if the fed funds rate is going down in an attempt to increase inflation.

      Delete
    8. "So, suppose we're in the pre-financial crisis world. Then, the central bank reduces the fed funds target. What happens next? Well, the conventional idea is that, in the short run, inflation will go up."

      Why does price inflation go up here when the fed funds target is lowered?

      I'm going to say it is because more private debt is created (third "idea").

      Delete
    9. Before the financial crisis, if the New York Fed received a directive to hit a lower fed funds rate target, they would attempt to do so through open market purchases. So initially reserves balances would go up, and Treasury securities held in the private sector would go down (typically this would have been a repo market intervention, but this gets the idea). Then we have to work through what happens in asset markets, and the initial open market purchase doesn't tell you what the path of open market operations is required to, say, support a lower fed funds rate forever. There is an old story about money multipliers (I think this is your "private debt" argument) which I don't entirely buy into.

      Delete
    10. "Before the financial crisis, if the New York Fed received a directive to hit a lower fed funds rate target, they would attempt to do so through open market purchases. So initially reserves balances would go up, and Treasury securities held in the private sector would go down (typically this would have been a repo market intervention, but this gets the idea)."

      Let's assume interest on central bank reserves is 0%. The fed funds rate is 4%, and the central bank wants to go to 2% on the fed funds rate. Central bank buys a bond and sells central bank reserves. The fed funds rate heads toward interest on central bank reserves. The central bank then reverses the process by buying central bank reserves and selling a bond. However, the fed funds rate remains lower at 2%.

      Also, I believe there can be an "announcement affect". The central bank announces the target, and it goes there.

      In both of those cases, the monetary base (currency and central bank reserves) stays the same, while the fed funds rate went down.

      In my scenario, the fed funds rate and market rates both fell with no change in the monetary base.

      “There is an old story about money multipliers (I think this is your "private debt" argument) which I don't entirely buy into.”

      I am pretty sure my private debt argument is not about money multipliers. Now with both market and fed funds interest rates lower, is that an attempt to increase private debt and increase demand deposits from the ***banking system*** so that “money” increases (“third idea”) and is spent so that quantities and prices increase, hopefully lifting price inflation back to target?

      Delete
  9. Krugman's point is that the correction requires more than a reset but an offset or overcorrection. His assumption that this is problematic at the ZLB is that to (conventional?) policy becomes ineffective and such overcorrection is not possible.

    In other words, tight money today leads to lower rates tomorrow. The ECB is a fine example of this.

    ReplyDelete
    Replies
    1. "...the correction requires more than a reset but an offset or overcorrection."

      But why?

      "The ECB is a fine example of this."

      Assuming what they did was right. How do we know that?

      Delete
  10. "But why?"

    Expectations and signaling: http://economistsview.typepad.com/timduy/2015/08/does-25bp-make-a-difference.html

    "How do we know that?" Fair enough, but what should the ECB have done?

    ReplyDelete
    Replies
    1. Duy says: "First, it will help clarify the Fed's reaction function. Second, if the experience of Japan and others who have tried to hike rates in the current global macroeconomic environment is any example, the Fed will only get one shot at pulling the economy off the zero bound. They better get it right."

      1. If people thought they were clear on the Fed's "reaction function" prior to the financial crisis, i.e. the Taylor rule, that rule would tell you that liftoff is long overdue.
      2. The experience of "Japan and others" I think can be summarized by: lliftoff from zero is a very hard thing to do, as a lot of people are in the grip of ideas like those in Duy's post.

      "...what should the ECB have done?"

      http://newmonetarism.blogspot.com/2014/09/theories-of-inflation-and-european.html

      Delete
    2. What the market's knew, or didn't, 7 years ago doesn't seem relevant. Market uncertainty today leads to slower growth today. If clarity can alleviate the uncertainty, we should expect as much from policy makers.

      That's a good post, but is it really plausible in the short run? As in your other recent post "Intuitive Neo-Fisherism", non-neutrality scenarios show the liquidity effect dominates in the short run. If so, is that acceptable and how long until the Fisher effect dominates?

      A couple other questions:
      The Fed raises the target rate tomorrow by 1% - what happens to inflation?
      The Fed never raises the target rate - what happens to inflation?

      Delete
    3. 1. Yes, I agree, clarity is good.
      2. Plausibility? It's plausible to me, or I wouldn't have written it. Whether you buy it or not is another question.
      3. "...non-neutrality scenarios show the liquidity effect dominates in the short run." This is always the question. If the change in policy is permanent, and everyone believes that, then the Fisher effect is there from the beginning. Then, the big question is how large the non-neutrality is, and there's no good reason to think that inflation will necessarily go down on impact when the nominal interest rate goes up permanently. And so what if it did go down? We still have to be cognizant of the dynamics in thinking about the policy. You can't focus only on the very short run.
      4. "how long until the Fisher effect dominates?" You can see the correlation in the data - nominal interest rates and inflation are positively correlated. So, even if the inflation rate declines on impact, you don't have to wait long for it to go up.
      5. "The Fed raises the target rate tomorrow by 1% - what happens to inflation?" Look at the chart in the European inflation post. That looks about right.
      6. "The Fed never raises the target rate - what happens to inflation?" More of the same. Japan post 1995?

      Delete
    4. Is there anything more than correlation analysis to support the idea?

      Delete
  11. Krugman in "Debt is good" also writes: "What can be done? Simply raising interest rates, as some financial types keep demanding (with an eye on their own bottom lines), would undermine our still-fragile recovery. What we need are policies that would permit higher rates in good times without causing a slump. And one such policy, Mr. Kocherlakota argues, would be targeting a higher level of debt".
    Luciano Priori F.

    ReplyDelete
    Replies
    1. "...still-fragile recovery..."

      The recession ended, according to the NBER, six years ago. Since then real GDP has been growing at a sustained pace, albeit at a lower rate than average performance over the post-WWII period. It's troubling that productivity growth is low, but it would seem that's something that monetary policy can't address. I discussed the state of the labor market here:

      http://newmonetarism.blogspot.com/2015/08/the-state-of-labor-market-in-us.html

      That's not completely "normal," but if you look at it closely, you could characterize it as "tight." I'm not sure where to find this "fragility." Why would a modest increase in short-term nominal interest rates push this thing off a cliff?

      Debt may be good, but that's the fiscal department, not the monetary department.

      Delete
  12. I'm surprised nobody has mentioned what seems to me the obvious interpretation of Krugman's concern: raising interest rates may cause deflation (or rather, disinflation) which cannot be offset by lowering rates below 0. This makes the ZLB constraint tighter, since with lower inflation a 0 nominal interest rate now corresponds to a higher real interest rate.

    Now admittedly formalizing this argument requires making assumptions about how people form inflation expectations. Krugman seems to think in terms of simple backward-looking adaptive expectations, where the rate of change in inflation depends on the gap between the real interest rate and the natural rate.

    One critique of Krugman's position is the lack of a deflationary spiral in any country at the ZLB over the last several years. But he seems to be concerned about a sort of "low-inflation trap", produced by similar logic to the classic deflationary spiral, but in a model augmented with a lower bound on inflation.

    ReplyDelete
    Replies
    1. "...raising interest rates may cause deflation (or rather, disinflation) which cannot be offset by lowering rates below 0."

      Two problems: (i) maybe inflation goes up when the nominal interest rate goes up. (ii) If inflation goes down when nominal interest rates go up, why doesn't it go back up when nominal interest rates go down?

      "...seems to be concerned about a sort of "low-inflation trap", produced by similar logic to the classic deflationary spiral..."

      What is that "classic deflationary spiral" anyway? I see this mentioned from time to time, but I know of no model of a deflationary spiral.

      Delete
    2. "What is that "classic deflationary spiral" anyway? I see this mentioned from time to time, but I know of no model of a deflationary spiral."

      For a self-proclaimed Fisherite you are surprisingly unfamiliar with Fisher's most important paper.

      Delete
    3. Do you mean the debt deflations idea? In early versions of Bernanke Gertler (AER 1989), they tried to formalize that without success. If that's what you're talking about, but maybe you have something else in mind. To be more specific, I should say "serious model."

      Delete
    4. I checked Fisher's 1933 Econometrica paper, just to make sure I had this right. What Fisher wrote was a verbal description of a process. It's an idea - which has the same status as any idea we haven't yet formalized. In 2015 we have much higher standards than we did in 1933. We expect the idea to be fleshed out in terms of a formal model, so that we can check for consistency, to see if the logic is correct, and to evaluate how this fits with what we already know. So, I'm afraid Fisher's debt-deflation theory doesn't make it.

      Delete
    5. Is this the right paper for Fisher's debt deflation?

      https://fraser.stlouisfed.org/docs/meltzer/fisdeb33.pdf

      Delete