An important new paper on NGDP targeting
Many people have sent me an excellent new paper by Wolfgang LECHTHALER, Claire A. REICHER, and Mewael F. TESFASELASSIE, Kiel Institute for the World Economy. The front page notes that:
This document was requested by the European Parliament’s Committee on Economic and Monetary Affairs.
However it clearly represents the authors’ views, and is not any sort of official statement by the EU.
As far as issues related to implementation are concerned, a strict inflation target can be simpler in certain ways to implement than either a flexible inflation target or a NGDP target, because revisions to the data on inflation are small, while revisions to the data on NGDP or real GDP are larger. Moreover there is considerable uncertainty about potential output growth. These are problems discussed in great detail by Orphanides and Williams (2002), Rudebusch (2002), and Goodhart et al. (2013). However, a counter-argument suggests that a NGDP target, even if in levels, would make it easier to avoid issues related to the measurement of the output gap. Additional arguments in favor of NGDP targeting involve the idea that it is easier to sell more stable nominal incomes to the public during bad times, and that a NGDP level target per se would increase the degree to which monetary policymakers are held accountable, by providing a measurable outcome.
To summarize, the theoretical evidence suggests that an explicit NGDP target, especially in levels, could possibly help the central bank to promote long-run price stability while allowing for a short-run response to output. However, this evidence is still relatively uncertain, and in the meantime, we find it useful to clarify the debate about what should and should not be expected to be achieved with a NGDP target.
I’m not qualified to discuss the data revision question, although Mark Sadowski challenged the conventional wisdom. I would make two points:
1. Even if measured inflation is not sharply revised, it remains a very inaccurate measure of the sort of price changes that have macroeconomic significance. For instance, a large share of the core CPI is based on rent and rental equivalents for housing. That data isn’t even a true “price”, and has no business influencing monetary policy. Last time I looked housing was 39% of the core CPI. This problem is briefly discussed later in their paper.
2. Revisions to NGDP are not large enough to be of macroeconomic significance, with one exception—changes in methodology. A recent example is the addition of R&D spending to investment, which caused a jump in NGDP. But there’s pretty general agreement that monetary policymakers would allow “base drift” in those cases; they’d raise the target by the amount of the upward bump from the new definition of NGDP. Also note that the Fed should be targeting expected NGDP 12 or 24 months out in the future, which makes near term data revisions much less important.
The theoretical case for NGDP level targeting as a forward guidance tool has been made, among others, by Woodford (2012). Similarly, in a recently published study, Coibion et al. (2012) have found strong theoretical support for price level targeting. Importantly, they take the zero lower bound into account, and they find that under inflation targeting, recessions that are deep enough so that the ZLB becomes binding are rare but costly. They Is nominal GDP targeting a suitable tool for the ECB’s monetary policy? They go on to show that price level targeting would result in less-deep recessions and stronger recoveries than would inflation targeting. Furthermore, price level targeting would imply that the ZLB would become binding less often. Therefore, switching from inflation targeting to price level targeting can lead to a substantial improvement in overall welfare, even if there is no foolproof way for such a target to always avoid hitting the zero lower bound. However, we are not yet aware of a study which compares price level targeting with NGDP level targeting, in light of the other theoretical considerations that we consider to be important. Therefore, we still consider the choice of a level target, were one to be adopted, to be an open question.
Someone should do that study!
In fact, this inability to use monetary policy to fine-tune prices or GDP motivates the debate about Taylor rules. Under a Taylor rule, the ECB would increase interest rates whenever inflation or output is above target. It turns out that something like a Taylor rule could also be used to implement NGDP targeting, at least when the zero lower bound does not bind. As Andolfatto (2013) shows, this would entail adding an additional term to represent the past deviation of the price level from its long-run path. While the specific implementation of this idea would require more thought, this idea would require relatively few changes from current operating procedures, to the extent that current policy resembles a Taylor rule but with equal weight on inflation and on output.
A more ambitious idea would be to set up a futures market in a price index or NGDP, and then for the central bank to either buy and sell these futures, or otherwise adjust monetary policy, in order to use these futures prices (rather than interest rates) as an operating instrument. To the extent that these futures prices represent accurate forecasts, then this approach should minimize fluctuations in the underlying target. Furthermore, this idea would encourage central banks to act proactively to avoid future target misses, rather than act reactively to past target misses. This idea is known as “market monetarism”, in the words of Christensen (2011) and Sumner (2011). While this approach is innovative, the likely consequences of this policy approach are not yet completely clear, and this approach would require the euro area to set up a new array of futures markets. In fact, for these futures markets to make it possible to target NGDP, financial markets would have to be efficient, in the sense of providing accurate forecasts. To the extent that financial markets are not efficient (because of bubbles, market frictions, or policy itself), then targeting futures prices would not completely solve the problems inherent in implementing a NGDP target. Nonetheless, if futures markets were to be set up, they would likely provide some information about the beliefs of market participants, and this information would be useful in implementing the target.
I’m glad they mentioned the usefulness of setting up these markets, even when they are not used as a policy instrument.
Market inefficiency is real, but very unlikely to be large enough to be of macroeconomic significance. And if I’m wrong at least there’s the silver lining that I’ll get rich trading the futures when the market price is clearly wrong.
Another issue is related to central bank communication. For instance, Sumner (2011) posits the following scenario. During a period of low inflation, an inflation target calls for higher inflation. However, higher inflation might be difficult to communicate to the public, because the public thinks of higher inflation something bad (i.e. a higher cost of living). In contrast, a NGDP target would call for increase in nominal income, and that might sound more acceptable to the broader public. This is because the public thinks of higher income as something good. The opposite would be true when inflation is high. During a period of high inflation an inflation target calls for lower inflation (which sounds good to the public). In contrast, a NGDP target would call for lower nominal income (which sounds bad to the public). In any case, policymakers who wish to implement an inflation target or a NGDP target would have to think about how they communicate these targets to the public.
Here I add that NGDP communicates more clearly all the time, both when easing and tightening. The public doesn’t understand the distinction between supply and demand side inflation. But the authors are correct that the NGDP language would be more popular when the central bank is trying to stimulate. Of course due to the zero bound problem it is precisely those times when clearer communication is most needed. When central banks want to tighten they face no zero bound problem, and hence communication is less important.
Read the whole paper, it’s an outstanding survey of the topic, and I’m glad to see that people in Europe are paying attention to this issue. The current ECB policy regime is clearly not effective in meeting the macroeconomic policy goals of the EU, low and stable inflation plus economic stability.
Tags:
30. September 2015 at 09:22
Please answer the question: if the central bank prints money, and NGDP does not rise (since velocity changes, people hoard even more money), then what does the Fed do? Continue to print money, or call it a day and stop?
30. September 2015 at 09:32
@myself – just to make the question more generic, let’s not argue over ‘why’ NGDP does not change if money is printed. Let’s either assume it does not, then what? Or, clearly state for the record that you, Scott Sumner, say it is impossible for a central bank to target NGDP via a monetary framework and for NGDP not to change. I’m saving your answers in my file, so please answer for the record.
30. September 2015 at 09:33
“…we are not yet aware of a study which compares price level targeting with NGDP level targeting…”
https://thefaintofheart.wordpress.com/2015/04/02/raise-the-banner-oops-i-mean-the-inflation-target/
30. September 2015 at 09:35
Its interesting that CPI seemed to keep shooting away in 2008 as GDP started to flatten.
Also the price-level (CPI) seems to continue almost on trend if you take out that bump, whereas the GDP line has taken a real hit in level terms before climbing back to approximate trend.
https://research.stlouisfed.org/fred2/graph/fredgraph.png?g=1ZhB
30. September 2015 at 09:35
Before the link above, meant to say:”Not a study” but a blog post with interesting contrasts”.
30. September 2015 at 09:39
Zoomed:
https://research.stlouisfed.org/fred2/graph/?g=1ZhJ
And if people want to play with the graph.
What would happen to prices if the fed had been on the pedal from mid 2008, maybe we’d get some inflation and no recession? large inflation? the reactivity, big swing up and down then correction in CPI is surprising.
30. September 2015 at 09:58
@c8to
The chart is very different if instead of Headline CPI you use Core PCE. With both in log scale. Check it out
30. September 2015 at 10:02
@c8to “Also the price-level (CPI) seems to continue almost on trend if you take out that bump, whereas the GDP line has taken a real hit in level terms before climbing back to approximate trend.”
That´s one of the reasons I argue in the link I put above against PLT. It´s not reliable! Even if, more correctly, you use the Core version.
30. September 2015 at 10:12
@mn
Wow – that is smooth, gradient flattens tiny bit in 2008. Would core PCE be a candidate “price level” target – the fed would be doing not much?
the other question is how early do we know these numbers (whats the lag with non-significant revisions)
I guess thats what you want the futures markets for – is there a proxy for that, can we see in 2008 that some financial, ideally futures market fell off a cliff – other than housing i suppose.
lets not kid ourselves that we even have an inflation target (or one that fed for practical or political reasons can hit)
https://research.stlouisfed.org/fred2/graph/?g=1Zjv
30. September 2015 at 10:27
Also impressed how much this paper got right, hopefully the first of many. Officially disappointed with any OECD CB that doesn’t have de facto NGDPLT by 2017.
On the question of market inefficiency, remember this is a problem for all investnent! That’s a much, much bigger problem. But if we learned anything in the 20th century, it’s that free markets are like free elections — the worst possible system except for all the others.
30. September 2015 at 10:35
@talldave – exactly – the efficiency is a furfy. who cares about a precise but useless target, compared to a possibly flawed but useful target.
@mn – good post. I like the EC “kill the economy” B graph – i’d probably move that higher.
The israel part – guess that looks something like oz – but clarity on the inflation graph for israel: it could just possibly mean they didn’t let inflation collapse as happened in the US. or are you saying that look how smooth their NGDP is, and inflation is swinging all over the place, so forget inflation.
30. September 2015 at 10:38
Or just use nominal wage LT as a surrogate for NGDPLT.
Less revisions and higher frequency, but still quite a bit of macroeconomic nominal stability.
30. September 2015 at 12:25
Whereas I agree that the ECB’s policy regime is flawed and that an NGDP target would be an improvement, I don’t think inflation targeting is so *intrinsically* flawed that they couldn’t achieve decent macroeconomic performance. But they’re not coming close to their inflation target, indeed haven’t for several years, and I think it’s fair to say the economic situation would be better if they had. So I’m not sure that you can chalk the situation up to a bad policy regime so much as to incompetence. Is there good reason to think the ECB would be better at hitting an NGDP level target than they are at hitting an inflation target?
30. September 2015 at 12:28
>> It turns out that something like a Taylor rule could also be used to implement NGDP targeting, at least when the zero lower bound does not bind. As Andolfatto (2013) shows, this would entail adding an additional term to represent the past deviation of the price level from its long-run path.
It’s worth noting that if any central bank explores this “Taylor Rule Plus” methodology, they should really study some control theory.
“Adding an additional term to represent the past deviation” is precisely adding an “integral term” in the sense of a PI Controller. In the sense that the output gap term is supposed to correlate with the forward-looking change in the inflation rate, this “Taylor rule plus” is a full-on “PID controller” (same link).
30. September 2015 at 12:39
Ray, Keep printing.
C8to, “Interesting” as you say, and revealing. In retrospect, which indicator was more prescient?
LK, Yes, that would work too.
dbeach, Good point. I think level targeting of prices would have helped the eurozone a lot. Especially the GDP deflator net of VAT.
Majromax, I have not studied control theory, but for various reasons I don’t trust the Taylor Rule, even if applied to NGDP.
30. September 2015 at 13:33
I had to read this twice to make sure I was reading correctly: “In fact, for these futures markets to make it possible to target NGDP, financial markets would have to be efficient, in the sense of providing accurate forecasts. To the extent that financial markets are not efficient (because of bubbles, market frictions, or policy itself), then targeting futures prices would not completely solve the problems inherent in implementing a NGDP target”.
Markets are as efficient as governments and regulatores let them be, and even if “not very efficient”, they are much better to forecast anything than a dozen people in a room …
30. September 2015 at 14:25
@Ray
If the central bank prints money, and NGDP does not rise (since velocity changes, people hoard even more money), then what does the Fed do? Continue to print money, or call it a day and stop?
Continue to print money, of course. They might want to change the asset they are buying, say from ST government securities to LT or to private sector securities or foreign exchange or NGDP futures, or decrease interest paid on excess reserves. How much how long? Until NGDP gets on target. You as a rational investor, would you be willing to bet that NGDP would not rise?
30. September 2015 at 16:39
Why are rents not a true price?
30. September 2015 at 16:39
All good.
But one worrisome question: what if central banks adopt NGDP LT, but at suffocation levels, as they are wont?
Call me a heretic, but I would rather live with a central bank that honestly targeted a 2.5% to 3.5% inflation band than central bank that targeted a 3% annual increase NGDPLT.
Remember my fellow market monetarists, even after we hit a home-run we still have many innings left to play, and I rather suspect, we will have to go deep into our bullpen.
30. September 2015 at 16:49
Rent is a true market price. “Owner equivalent rent” is not a true market price.
“Owner equivalent rent” is an estimate of the benefit you get from owning your own home. If you own your home outright, and I estimate that it would rent for $2,000 per month, then your “owner equivalent rent” is $2,000. That’s not a market price because you aren’t actually paying it. (And there is no real reason to be cnfident that you would be paying that if you were going to rent housing).
Capturing housing in a price index is difficult because some people are renting (and thus paying a market price), others own their homes outright (and are thus paying nothing), while others still are making mortgage payments. And in that latter group, what those payments actually are depend not only on the value of the home but on other factors like the level of long-term interest rates at the time of purchase, etc.
Owner equivalent rent is a compromise to try and get housing costs into measures of inflation, but the flaws in this approach have become very evidenct over the past decade.
30. September 2015 at 16:50
@Ray
More importantly, if NGDP does not rise initially, accelerate the rate of money printing. Japan has shown that changing the rate of money printing does make an effect on NGDP. The thing is, it may cause only a rise in the price level, and not RGDP growth (or maybe only an ephemeral growth in RGDP that goes away quickly, that is, money is superneutral in the long run, but has real effects in the short run).
I take the liberty to copy here the link to a Mark Sadowsky post at Marcus Nunes blog, that shows exactly that happening in Japan:
https://thefaintofheart.wordpress.com/2015/08/18/what-monetary-policy-can-and-cannot-do-japan-edition/#comments
1. October 2015 at 05:26
Don’t bury the lede! The most cited source in the paper is this blog.
1. October 2015 at 06:41
@ThomasH – to keep printing money, hoping that NGDP will rise, is to risk hyperinflation. Though I believe money is neutral or super-neutral, if you print too much of it, it can possibly cause hyperinflation. Model: put a flat eraser or pencil on a hardback book lying flat. Start raising one end of the book so that it goes vertical. The pencil or eraser should stay static and immobile (nothing happening) until such time that the coefficient of static friction changes, then the eraser or pencil quickly slides off the book. That hissing, sliding sound is hyperinflation.
@Jose Romeu Robazzi -thanks, but the Sadowski piece was utter nonsense. Japan’s rise in NGDP was not due to Abenomics, but simple recovery from the 2008 recession, which every country has seen, even countries practicing austerity and obviously not doing Abenomics. Once again Sadowski cannot see the forest for the trees.
1. October 2015 at 13:25
[…] of NGDP targeting or, rather, a debate that had quietened down seems to building up steam again in Europe and in the […]
1. October 2015 at 18:01
cma, Suppose you bought a car and had to pay $400 a month for 6 years. Would you call those monthly payments the price of the car? If the monthly payments didn’t change for 6 years would that make car prices sticky? Obviously not. Rent is fairly similar, it’s more like installment payments on a debt that a true market price.
For the purposes of macroeconomic stabilization, the price that matters is the price of newly constructed housing units. That’s the price you need to stabilize in order to stabilize the labor market. If you want to consider rent on existing units, it should be the rent on newly occupied apartments.
And yes that does conflict somewhat with NGDP targeting, it suggests we should target total labor compensation, not NGDP.
2. October 2015 at 04:47
@Ray
Abenomics is a series of things (at least in words) I was just thinking about the monetary part. And hey, was the tax hike in Japan an austerity move? Maybe you are right, it was just correlation and not causation… But then, maybe not.
2. October 2015 at 10:01
[…] Sumner annotates a paper on nominal GDP targeting written for the European […]
22. October 2015 at 10:54
[…] we may be looking soon at the mandate and there were hints that the European Parliament is also looking into the mandate. At least talk about NGDP Targeting and you can then “Feel The Power” in time for the […]
23. October 2015 at 22:01
[…] we may be looking soon at the mandate and there were hints that the European Parliament is also looking into the mandate. At least talk about NGDP Targeting and you can then “Feel The Power” in time for the […]
10. December 2015 at 18:20
[…] the Euro Area on NGDP Targeting. The first of three papers was leaked in September and Scott Sumner commented on its positive case for NGDP Targeting. The other two papers presented to the European Parliament […]
15. February 2016 at 12:14
[…] hope so. It is a real shame no Parliamentarians took him up on the recommendation of one of their own reports strongly endorsing NGDP […]
17. February 2016 at 23:01
[…] hope so. It is a real shame no Parliamentarians took him up on the recommendation of one of their own reports strongly endorsing NGDP […]