The Fed’s targeting the wrong forecast
David Beckworth has a nice interview with Ryan Avent, which touches on a number of monetary issues. In the final part of the interview they both suggest that the Fed may be treating their 2% (PCE) inflation target as a sort of ceiling, rather than the symmetrical target the Fed claims to be aiming at.
I share their frustration, but I suspect the problem lies elsewhere. I believe the Fed would actually prefer that PCE inflation average 2%. Instead, I suspect the Fed is consistently missing its target in recent years because they are relying on a flawed (Keynesian) model, which bases its inflation forecasts on concepts such as the Phillips Curve. They would have been much more successful if they had instead relied on market forecasts. (Of course NGDP level targeting is far superior to inflation targeting.)
Lars Svensson has argued that the Fed should target its own internal forecast of inflation. Two facts lead me to believe that this is exactly what they are doing. First, the Fed sees the world in a very “conventional” way. It’s a big institution full of mainstream economists. Second, mainstream economists have been forecasting roughly 2% PCE inflation over the past 11 years, since the US entered the Great Recession. Here are the 11 most recent forecasts of PCE inflation over the next two years, in each case representing the 4th quarter forecast from the Philadelphia survey of private sector economists:
(The data shows the date of the forecast, then the next year inflation rate, then the two-year forward inflation rate.)
2018:Q4 2.1% 2.1%
2017:Q4 1.8% 2.0%
2016:Q4 1.9% 2.0%
2015:Q4 1.8% 1.9%
2014:Q4 1.8% 1.9%
2013:Q4 1.9% 1.9%
2012:Q4 2.0% 2.2%
2011:Q4 1.7% 2.0%
2010:Q4 1.4% 1.8%
2009:Q4 1.3% 1.8%
2008Q4: 1.8% 2.2%
The Fed believes that its policy affects inflation with a long lag. So it seems reasonable to assume they set policy with the goal of getting inflation on target in the second year. In that case, they’ve almost perfectly targeted the Philly Fed consensus, where expected 2-year forward inflation averages 1.983% over the 11-year period. That’s not 2.0%, but it’s extremely close.
Actual PCE inflation has been much lower (averaging 1.5% from 2008:Q4 to 2018:Q4), but that’s not because the Fed was secretly targeting lower inflation; it’s because they used bad forecasts.
If the Fed had instead targeted something like TIPS spreads, minus 0.25% to account for the fact that CPI inflation runs about 0.25% above PCE inflation, then they would have done a far better job of hitting their inflation target. The TIPS spread is not perfect, with sudden oil price changes biasing it slightly, and a small risk spread, but it’s better than relying on professional forecasters.
PS. Pat Horan and I have a new piece on MMT.
Tags:
12. March 2019 at 14:12
Quick calculation: the average 5-year CPI breakeven since 2010 was 1.75%.
CPI minus 0.25 = 1.5%
The market was on to something.
12. March 2019 at 16:19
Thanks for that calculation LK.
12. March 2019 at 16:55
Agreed, but the mere fact that there is an inflation target I’d bad in context, as central bankers will always interpret a target as a ceiling.
The Reserve Bank of Australia has sidestepped recessions for several decades by having an inflation band target, that is 2% to 3%. They also appear to endure a few bouts of 4% inflation without long-term negative results— that is they did not run up into higher inflation rates.
I agree that a NGDPLT would work better. But if we are resigned to inflation targeting then inflation band and one that is a little higher than the Fed’s target might actually do the trick.
The People’s Bank of China has also had a good run and they tend to set targets at “about”, that is about 3% as presently.
Looking at results for the ECB and the Fed, I would say there’s a great deal of foolishness in a hard inflation target especially one that is too low. A peevish fixation on inflation is not a monetary policy.
How on earth a hard inflation target became accepted is beyond me.
12. March 2019 at 18:39
If the Fed prefers that PCE inflation averaged 2.0%, why don’t they just do level targeting? Level targeting doesn’t have to come with NGDP targeting; it can merely be viewed as a way to achieve the 2.0% inflation that is supposedly desired. Isn’t level targeting essentially guaranteed to be more accurate at achieving 2.0% inflation (when measured over a sufficiently long timespan, such as 5 years)?
13. March 2019 at 02:46
Why would central bankers always interpret a target as a ceiling? Why not as a floor?
In practice they do seem to treat a those targets as ceilings in the Fed and ECB perhaps. But those seem contingent cases, I don’t see why that should necessarily be the case?
13. March 2019 at 05:44
Yup. Very good post.
13. March 2019 at 10:28
Anon85, I’m not sure, but I suspect that level targeting of prices would “expose” the underlying flaw in inflation targeting, which is that inflation is the wrong variable. Under level targeting, you are forced to hit the target over time, even if there is a huge oil price shock and hitting the target is a bad idea.
Matthias, After another decade we’ll have a better answer to your question. Inflation wasn’t consistently lower than 2% during 1992-2007.
14. March 2019 at 02:02
For the same reason you will never find a central banker who will forecast the institution they work for missing their mandate (even when they privately worry about efficacy of the toolkit or the ZLB/liquidity trap), you’re unlikely to find an economist working for a Fed regulated institution implicitly criticising their employer’s regulator. Thus sincerely held private economist forecasts of a deflationary Japan-like scenario or 1980s stagflation probably get moderated to PCE 10 years at 1.7% or 2.3%.
Interestingly you can see how the market “inflation compensation” and became un-anchored to the downside and rose in day-to-day volatility in 2015-2016 just as the dispersion of the 10yr ahead CPI forecasts in Philly Fed was falling.
It probably lends support to your point that the survey of professional forecasters is the wrong thing to target.
14. March 2019 at 14:03
This is somewhat off-topic, but the Fed issued a notice of proposed rulemaking to pay lower IOER to “Pass-through Investment Entities.” Clearly PTIE’s only target TNB and the ANPR was issued on the same day as the New York Fed’s motion for dismissal.
I’m wary of sounding shrill, but it seems like blatant protection of existing TBTF banks and regulators. The monetary policy arguments are tenuous. The money multiplier has decreased substantially due to IOER but the Fed has still increased NGDP.
But even if the monetary policy arguments are true, then they should go back to 0% reserves for *all* banks. Instead, the proposal is to pay higher-than-T-bill rates to banks except for TNB, but pay TNB a lower rate (likely 0%).
Edit: The link to the motion to dismiss and the ANPR swallowed the comment. I’m sorry if the comment is just being moderated. John Cochrane, who I disagree with a lot, is right on with his outrage about the Fed’s action re TNB.
14. March 2019 at 19:40
A good post, but it is also a bit frustrating because of your earlier post on 29 November 2018 titled “What about the TIPS spread?”.
15. March 2019 at 11:19
OG, You said:
“an economist working for a Fed regulated institution”
These economists making private sector forecasts were not working for a Fed regulated institution.
Matthew, I am just as outraged as John Cochrane.
HL, Explain.
15. March 2019 at 22:18
RE Pat Moran/Sumner MMT piece in Mercatus
Well, nice round-up, but I sense some issues are getting blurred.
1. For example, some could be “for” MMT, but look askance at social welfare spending. Some might think MMT would work, but by instituting deficit-inducing tax cuts whenever unemployment rose about 4%. In fact. I would prefer a much smaller federal budget both social and military. It is the deficit spending that matters, not what the money is spent on (in fact, Japan sidestepped the Great Depression, but federal outlays for war and occupation rose dramatically, not social welfare).
James Galbraith says as much in a Project Syndicate column:
“MMT is not, as its opponents seem to think, primarily a set of policy ideas.”
So, as in 1930s Japan, one could be a MMT’er who wanted to be super-patriotic, wave flags, militarize, euthanize weaklings, and occupy China and Pacific Asia. Macro-economically, the plan worked. Japan’s economy grew through the Great Depression, while other Western economies moldered. Non-macroeconomically, Japan’s imperialism was a horrific endeavor.
2. There still seems to be some blurriness between Ben Bernanke’s calling for money-financed fiscal deficits in Japan in 2003, and modern-day MMT. Is there a difference? Is a money-financed fiscal deficit (as advocated at times by Bernanke, George Selgin and Adair Turner) the same thing as MMT? What is the difference?
The MMT crowd, including Galbraith (see link), keep talking about the federal government as financed by taxes and borrowing (albeit, possibly the borrowing is offset by QE, another fuzzy point).
If QE+federal deficits equals MMT, and MMT is a horrible catastrophe in waiting, then when the US went to $4 trillion of QE after 2008, with mildly positive results—what does that say about MMT?
We saw no inflation, let alone hyperinflation. Europe and Japan are sinking back into deflation, despite lots of QE.
https://www.project-syndicate.org/commentary/modern-monetary-theory-opponents-misunderstanding-by-james-k–galbraith-2019-03
17. March 2019 at 18:53
Prof. Sumner,
Have any MMT proponents responded to the Mercatus piece you co-authored? I’ve seen many pieces that critique other MMT skeptics like Dean Baker, Ken Rogoff, Doug Henwood et al. It seems to me that you laid out what you perceive to be the major weaknesses of MMT in a clear and concise manner, such that MMT advocates should be both eager and able to refute your arguments.
18. March 2019 at 07:56
The Fed, or a regional Fed, needs to set up an official core PCE inflation prediction market. Then these forecasters will have to explain why their forecasts are more bullish than the market, and why they aren’t going long inflation.
You have PredictIt gaining credibility in the political prediction arena, with journalists routinely citing it, this makes me think “we” might be able to get this idea into the head of a regional Fed president. I’m imagining five years worth of monthly contracts set up, with an additional monthly contract added each month to replace the expiring contract.
Is this realistic? I’m thinking it might be better to focus any new Hypermind contracts on PCE, rather than NGDP, as that’s sadly how people think. Incidentally, Hypermind has not responded to my inquiries to fund a new NGDP market, nor has PredictIt.
20. March 2019 at 10:16
I don’t understand the Fed’s forecasting. They have an estimate for NAIRU, non-accelerating inflation rate of unemployment, which AFAICT they make up and sometimes raise or lower when events contradict it. They currently think that we will “remain well below NAIRU until 2021,” yet without inflation accelerating past their 2.0% target, only barely reaching it. There doesn’t seem to be much meaning to their NAIRU concept.
20. March 2019 at 15:40
Ben, You said:
“If QE+federal deficits equals MMT”
And if it doesn’t?
TheNumeraire, I’m certain they’d just say we misunderstood their theory.
John, I believe they think the PC is pretty flat right now.
24. April 2019 at 09:38
[…] I believe Occam’s razor works in the opposite direction, in favor of my hypothesis. In this earlier post I pointed out that over the past decade the private consensus forecast has also consistently […]
24. April 2019 at 10:01
[…] fact, I believe Occam’s razor works in the opposite direction, in favor of my hypothesis. In this earlier post I pointed out that over the past decade the private consensus forecast has also consistently […]