Reducing Fed discretion (dedicated to John Taylor)

John Taylor has written a number of articles that are critical of the Fed’s current discretionary regime.  It’s difficult to know what the Fed is trying to achieve, and how they intend to get there.  This creates uncertainty in the marketplace, and greater macroeconomic instability stability.  Naturally Taylor focuses on the need for something like the Taylor Rule.  In contrast, I’ve advocated NGDPLT, with the same goal—reducing discretion and making the Fed more accountable.

In this post I plan to talk about a policy that I don’t favor, in the sense that it’s not my first choice.  On the other hand it’s a policy I do favor, in the sense that I’d vastly prefer it to current Fed policy.  I’m hoping that if Taylor Rule supporters read this post they’ll have a better idea as to where I’m coming from.

My proposed rule is CPCELT, that’s “core PCE level targeting.”  Why this target?

I see two big sources of discretion and ambiguity in the current target, which is PCE inflation targeting:

1.  One problem is supply shocks.  It’s widely known that pragmatic central banks may want to allow some price level variation when there is a supply shock.  For instance, if oil prices suddenly change, then keeping overall inflation on target would require moving “all other prices” in the opposite direction.  But those other prices are often stickier than oil prices, and trying to move them suddenly can create business cycles. Thus central banks often allow some inflation variation when there are supply shocks, especially to food and energy, and in some cases they point to core inflation as being more meaningful, at least in the short run.

And yet central banks also insist that they are targeting headline inflation in the long run.  This is supposed to reassure the public, because the public cares about headline inflation.  There are two problems with catering to public opinion in this case.  First you are pandering to ignorance.  Even if the Fed prevents supply shocks from increasing the price level in the long run, they will not be able to prevent the associated loss of real income, which is why the public cares about inflation in the first place.  Economists know that the public misunderstands the effect of inflation on living standards, but we pander to that ignorance when we insist we’ll stabilize headline inflation in the long run.  The second problem is that the public thinks of inflation in terms of the CPI, but the Fed targets the PCE.  So we aren’t really targeting what the public thinks of as “inflation,” even with a headline PCE inflation target.  PCE inflation runs about 0.35% below CPI inflation.

So I propose the Fed target what they actually care about, core PCE inflation.  That increases accountability, as the Fed can no longer point to temporary oil and food price changes as excuses.

2.  The second problem is growth rate targeting.  Under growth rate targeting we have no idea where the price level will be in 10 or 20 years, making long term planning more difficult.  Suppose prices fall, as in 2009.  Is the Fed trying to make up some of that fall, or shoot for 2% inflation from that point forward?  It’s hard to say.  Perhaps errors are a sort of random walk, and the price level gradually drifts further and further from that trend line. Even worse, a central bank that pays lip service to inflation targeting, but doesn’t seriously try to get there (say the Pre-Abe BOJ) can let the price level gradually drift lower for 20 years, each year claiming they just missed by 1% or so.  At the other extreme, the dovish Bank of England can consistently run slightly above target.

Level targeting is a way of forcing central banks to do what they claim they are trying to do.  Under level targeting you always have a precise point estimate of where PCE core price level is supposed to be in 2 or 3 or 5 years.  It’s a way of holding central banks accountable.  You still allow some variation in the very short run, assuming there is the dual mandate.  But they can no longer act like magicians, directing your attention somewhere else when they persistently fail to hit their goals.  “Look at oil.”  “Look at unemployment.”  “Don’t worry, we’ll get there eventually, despite the TIPS market saying we won’t.”  Those excuses won’t work; they’ll eventually be exposed if they persistently miss.

If there is to be a dual target (which is something I don’t like, and why I favor NGDPLT instead) then central banks must also be held accountable on the unemployment front. In recent years, inflation has run below target when unemployment is high, and vice versa. The Fed should be instructed to do the opposite, and its effectiveness should be judged on that basis.  They should have to explain to Congress why inflation is below target when unemployment is high, as that’s even worse that a rigid single mandate for inflation.

To summarize, with CPCELT central banks will no longer be able to offer the following excuses:

1.  We missed because of temporary factors like food and energy prices.

2.  We keep missing by 1/2% on the low or high side, year after year, but in the future we promise to do better.

But there are other advantages as well.  When prices fall, as in 2009, inflation expectations automatically rise (they actually fell under the current regime) and this reduces real interest rates, boosting the economy. During the housing boom the market would have understood that core inflation was overshooting, and that awareness would have led to more bearish expectations—reducing speculation.  In addition, some macro models suggest that you want to target the stickiest price, which might be wages.  Core inflation is more closely correlated with wage inflation than is headline inflation.

Everyone knows that a NGDP target would have called for a more expansionary monetary policy in recent years.  But the same is true for core PCE.  Core PCE inflation averaged 2.03% between December 1990 and December 2007, then the inflation rate fell to 1.4% between December 2007 and December 2014.  Under the dual mandate you’d prefer slightly lower inflation during the booming 1990-2007 period, and slightly above 2% inflation during the depressed 2007-14 period.  Even under a single inflation mandate this sort of variation is unacceptable, as it creates macroeconomic instability. Under a dual mandate it’s even more unacceptable.

To conclude, my critique of the Fed goes far beyond the fact that they are targeting inflation and I’d prefer NGDP.  Even under inflation/price level targeting there could be vast improvements, which would reduce discretion, increase accountability, and make the economy more stable.

Of course there’d still have to be a conversation about the instrument rules needed to hit this target.  But I firmly believe this sort of regime would vastly reduce the discretion/accountability/ambiguity problems that John Taylor worries about, even before we moved on to the policy instrument issue.

PS.  This post is a response to commenter SG.



13 Responses to “Reducing Fed discretion (dedicated to John Taylor)”

  1. Gravatar of Michael Byrnes Michael Byrnes
    14. February 2015 at 09:14

    Level targeting dramatically lowers the consequences of Fed error, correct? Because of the commitment to offset past misses. Maybe the Fed is opposed to it because it makes their policy more mechanical. (As you have often pointed out, under an NGDPLT regime there are no Hawks and Doves.)

  2. Gravatar of Morgan Warstler Morgan Warstler
    14. February 2015 at 10:03

    Scott if I’m not mistaken, you’d take almost anything that ends in LT, over one that doesn’t end in LT, right?

    If that’s the case, isn’t the optimal ask the one that is easiest to steer back to LT (high or low) the fastest?

    And even if I’m wrong, which thing is easiest to steer back to LT? Which is most steerable the fastest?

  3. Gravatar of Steve Steve
    14. February 2015 at 12:04

    Someone needs to tell Mother Nature to engage in Snowfall Level Targeting.

    The consequences of allowing each blizzard to become a bygone followed by the resumption of random storm development is a real hardship, and is preventing the formation of rational transportation expectations.

  4. Gravatar of Kenneth Duda Kenneth Duda
    14. February 2015 at 13:18

    I cannot press 1+ enough times. What will it take to get the CB to adopt some form, any form, of nominal level targeting?

  5. Gravatar of Lorenzo from Oz Lorenzo from Oz
    14. February 2015 at 13:27

    “This creates uncertainty in the marketplace, and greater macroeconomic stability.” Is there an ‘in’ missing at the start of the last word?

  6. Gravatar of Zack Zack
    14. February 2015 at 13:37

    Nice post. I’m interested to hear you’re response to Morgan Warstler’s question too.

  7. Gravatar of Major.Freedom Major.Freedom
    14. February 2015 at 15:05

    “The second problem is growth rate targeting. Under growth rate targeting we have no idea where the price level will be in 10 or 20 years, making long term planning more difficult.”

    But no business integrates price levels into their production processes. At most, price levels are treated as one factor among many in the setting of interest rates. But in terms of where to produce, what to produce, who to produce it, how to produce it, and yes, even what price to set for their specific goods, price levels are all but ignored. Businessmen care about the demand for their specific products, of which aggregate price levels have little to nothing to say.

    If a perfect predictor told any individual businessman what the CPI will be next year, and the year hence, that price level does not tell thaw businessman what the price of HIS products will be. A rise in the price level of say 5% in two years, cannot provide the individual businessman with any information on what HIS prices will be. The prices of HIS products may rise by more, the same, less, and even much much less.

    Do you know why almost all macroeconomists are useless to private business firms that produce goods and services? It is their largely useless models that are tailored to politicians who are asked to care about aggregate chimeras that only macro economists are willing to pretend are important.

  8. Gravatar of benjamin cole benjamin cole
    14. February 2015 at 16:43

    Excellent blogging. But make sure the price level target is high enough to not pinch off real growth. Central banker culture is divorced from the real economy. Plosser’s price level target in 10 years is 90, on current index of 100.

  9. Gravatar of SG SG
    14. February 2015 at 17:45


    Great post, and I honestly don’t know what some of the hardcore Taylor rule proponents would make of this. Maybe you can get John Cochrane to respond (does he even support the Taylor rule?).

    I’m curious, though, at your refusal to demonize Bernanke. Is it because you don’t think it would work, or because it would be wrong? In other words, is this an example of your general utilitarianism or an exception?

    It seems to me that certain other prominent public intellectuals have gotten a lot of traction by using tribalist tactics to drum up support for their positions, and I’m wondering what makes you reluctant to do so.

  10. Gravatar of ssumner ssumner
    14. February 2015 at 20:10

    Michael, Good point.

    Morgan, I certainly don’t want to level target gold prices. I’d have to think about whether it’s easier to level target NGDP or prices. I’m not sure.

    Steve, Amen.

    Thanks Ken.

    Lorenzo, Thanks, I fixed it.

    SG, It would be wrong. If I thought Bernanke was a demon, or even a bad guy, I’d say so. Now if you asked me about Trichet . . .

    From a pragmatic perspective, the only possible hope for reform is to get both liberals and conservatives to buy in. Fortunately we are having some success so far.

  11. Gravatar of Johannes Fritz Johannes Fritz
    15. February 2015 at 06:50

    I follow Kenneth Duda on the +1.
    Would it be necessary to establish a PLT with some sort of time window to close gaps and/or a “target-the-forecast” implementation?

    I got to this conclusion because I found it surprisingly hard to go after the ECB with a PLT this morning. Applying your insight on Euro area inflation, it’s not clear one can argue with the ECB on its own turf until mid-2013.

    That is, the observed price path in the Euro area does not deviate much on the downside from a fictional PLT established at various points in history. This is true for a PLT on HICP with an inflation of 1.9% and a core-HICP with 1.6%.

    The point is that the ECB was pretty far above the path in summer 2008. So I guess the argument would be that they should have never gotten away that far on the upside. But even with a PLT established in 2011, the ECB only goes off path in early 2013. Sure, still 2 years and counting. But given they were above all this time, the argument is not as clear as I thought it was last night.

    The pics for HICP and core-HICP here:
    Is there any other way to leave pictures?

    (I tried various decision points for the intro of the PLT: 1999 [ECB operational], 2002 [Euro circulates], 2003 [ECB “formalized” the “close to 2%”], 2008-Dec [first cut], 2011-Apr [first hike].)

  12. Gravatar of ssumner ssumner
    16. February 2015 at 05:57

    Johannes, You said:

    “I got to this conclusion because I found it surprisingly hard to go after the ECB with a PLT this morning. Applying your insight on Euro area inflation, it’s not clear one can argue with the ECB on its own turf until mid-2013.”

    This is a common mistake. You can’t figure out the impact of a 2% PLT by looking at data reflecting the trends in the economy when there was no PLT in effect. If a PLT had been in effect during that period everyone else would have been different, wages, expectations, interest rates, etc.

    Nonetheless, I do agree that a PLT would have been far from optimal, mostly due to higher prices from oil, VATs etc. NGDPLT is much better.

    In the US, under a PLT inflation would have been lower during the housing boom, and higher during the period since.

  13. Gravatar of Are we not monetarists? A brief overview of nominal GDP targeting | Economic Studies Consortium @ UCF Are we not monetarists? A brief overview of nominal GDP targeting | Economic Studies Consortium @ UCF
    19. April 2015 at 14:12

    […] One of the most prominent advocates of the modern monetarist position, Scott Sumner argues that discretionary policy “creates uncertainty in the marketplace” and identifies nominal GDP […]

Leave a Reply