Charles Goodhart might just be the best all-around monetary economist in the world, having made impressive contributions to both monetary theory and the history of monetary theory, to monetary history, and the history of monetary institutions, especially of central banking, and to the theory and, in his capacity as chief economist of the Bank of England, practice of monetary policy. So whenever Goodhart offers his views on monetary policy, it is a good idea to pay close attention to what he says. But if there is anything to be learned from the history of economics (and I daresay the history of any scientific discipline), it is that nobody ever gets it right all the time. It’s nice to have a reputation, but sadly reputation provides no protection from error.
In response to the recent buzz about targeting nominal GDP, Goodhart, emeritus professor at the London School of Economics and an adviser to Morgan Stanley along with two Morgan Stanley economists, Jonathan Ashworth and Melanie Baker, just published a critique of a recent speech by Mark Carney, Governor-elect of the Bank of England, in which Carney seemed to endorse targeting the level of nominal GDP (hereinafter NGDPLT). (See also Marcus Nunes’s excellent post about Goodhart et al.) Goodhart et al. have two basic complaints about NGDPLT. The first one is that our choice of an initial target level (i.e., do we think that current NGDP is now at its target or away from its target and if so by how much) and of the prescribed growth in the target level over time would itself create destabilizing uncertainty in the process of changing to an NGDPLT monetary regime. The key distinction between a level target and a growth-rate target is that the former requires a subsequent compensatory adjustment for any deviation from the target while the latter requires no such adjustment for a deviation from the target. Because deviations will occur under any targeting regime, Goodhart et al. worry that the compensatory adjustments required by NGDPLT could trigger destabilizing gyrations in NGDP growth, especially if expectations, as they think likely, became unanchored.
This concern seems easily enough handled if the monetary authority is given say a 1-1.5% band around its actual target within which to operate. Inevitable variations around the target would not automatically require an immediate rapid compensatory adjustment. As long as the monetary authority remained tolerably close to its target, it would not be compelled to make a sharp policy adjustment. A good driver does not always drive down the middle of his side of the road, the driver uses all the space available to avoid having to make an abrupt changes in the direction in which the car is headed. The same principle would govern the decisions of a skillful monetary authority.
Another concern of Goodhart et al. is that the choice of the target growth rate of NGDP depends on how much real growth,we think the economy is capable of. If real growth of 3% a year is possible, then the corresponding NGDP level target depends on how much inflation policy makers believe necessary to achieve that real GDP growth rate. If the “correct” rate of inflation is about 2%, then the targeted level of NGDP should grow at 5% a year. But Goodhart et al. are worried that achievable growth may be declining. If so, NGDPLT at 5% a year will imply more than 2% annual inflation.
Effectively, any overestimation of the sustainable real rate of growth, and such overestimation is all too likely, could force an MPC [monetary policy committee], subject to a level nominal GDP target, to soon have to aim for a significantly higher rate of inflation. Is that really what is now wanted? Bring back the stagflation of the 1970s; all is forgiven?
With all due respect, I find this concern greatly overblown. Even if the expectation of 3% real growth is wildly optimistic, say 2% too high, a 5% NGDP growth path would imply only 4% inflation. That might be too high a rate for Goodhart’s taste, or mine for that matter, but it would be a far cry from the 1970s, when inflation was often in the double-digits. Paul Volcker achieved legendary status in the annals of central banking by bringing the US rate of inflation down to 3.5 to 4%, so one needs to maintain some sense of proportion in these discussions.
Finally, Goodhart et al. invoke the Phillips Curve.
[A]n NGDP target would appear to run counter to the previously accepted tenets of monetary theory. Perhaps the main claim of monetary economics, as persistently argued by Friedman, and the main reason for having an independent Central Bank, is that over the medium and longer term monetary forces influence only monetary variables. Other real (e.g. supply-side) factors determine growth; the long-run Phillips curve is vertical. Do those advocating a nominal GDP target now deny that? Do they really believe that faster inflation now will generate a faster, sustainable, medium- and longer-term growth rate?
While it is certainly undeniable that Friedman showed, as, in truth, many others had before him, that, for an economy in approximate full-employment equilibrium, increased inflation cannot permanently reduce unemployment, it is far from obvious (to indulge in bit of British understatement) that we are now in a state of full-employment equilibrium. If the economy is not now in full-employment equilibrium, the idea that monetary-neutrality propositions about money influencing only monetary, but not real, variables in the medium and longer term are of no relevance to policy. Those advocating a nominal GDP target need not deny that the long-run Phillips Curve is vertical, though, as I have argued previously (here, here, and here) the proposition that the long-run Phillips Curve is vertical is very far from being the natural law that Goodhart and many others seem to regard it as. And if Goodhart et al. believe that we in fact are in a state of full-employment equilibrium, then they ought to say so forthrightly, and they ought to make an argument to justify that far from obvious characterization of the current state of affairs.
Having said all that, I do have some sympathy with the following point made by Goodhart et al.
Given our uncertainty about sustainable growth, an NGDP target also has the obvious disadvantage that future certainty about inflation becomes much less than under an inflation (or price level) target. In order to estimate medium- and longer-term inflation rates, one has first to take some view about the likely sustainable trends in future real output. The latter is very difficult to do at the best of times, and the present is not the best of times. So shifting from an inflation to a nominal GDP growth target is likely to have the effect of raising uncertainty about future inflation and weakening the anchoring effect on expectations of the inflation target.
That is one reason why in my book Free Banking and Monetary Reform, I advocated Earl Thompson’s proposal for a labor standard aimed at stabilizing average wages (or, more precisely, average expected wages). But if you stabilize wages, and productivity is falling, then prices must rise. That’s just a matter of arithmetic. But there is no reason why the macroeconomically optimal rate of inflation should be invariant with respect to the rate of technological progress.
HT: Bill Woolsey
Uneasy Money 
David,
Is there even any sense in NGDP targeting as an activity of a central bank? As I see it Fed could target whatever it wants, but it doesn’t give it new instruments for actually reaching this target. I am sort of confused by this hype for the NGDP targeting because I don’t understand how it is relevant for the usual open-market operations of the central bank.
LikeLike
There are circumstances where inflation could go above the NGDLT rate. This is when RGDP falls. This could happen for 2 reasons.
1. A supply shock reduces output capabilities. Inflation would be the level of the NGDP target + the fall in RGDP. For a mild shock this may not matter too much, but a larger and longer lasting shock could indeed cause inflation worries to make recovery harder (for example businesses will find economic calculation difficult)
2. When RGDP has dropped because poor expectations about the future have caused investment levels to be fall. In this case , in as much as NGDPT will keep the supply of money in line with demand then its effects will be positive. But Inflation targeting would have the same effect here. Where NGDPLT differs from IT is what happens when hitting the IT fails to keep RGDP on trend. IT will accept a drop in RGDP at the point while NGDPT will push ahead with additional nominal growth until the nominal spending target is hit. This (it is hoped) will induce a combination of real growth and additional inflation. How do we know with certainty that at this point the additional uncertainty about inflation will not weaken investment further in comparison with the situation with inflation targeting?
In short The point where : NDGPT and IT diverge is the point where inflation becomes higher under NDGPT. Proponents of NGDPLT believe that in these circumstances a higher degree of certainty about future nominal spending plus higher expectations about inflation will have a net benefit. I probably agree with this but still have concerns that we know less than we would like about the reasons why IT sometimes fails to keep RGDP on target and therefore we cannot know with certainty that allowing higher inflation will indeed help and not hinder RGDP stabilization.
LikeLike
Roman P., Central banks need at least one target and at least one instrument. Without a target, there is no point in having an instrument. Without an instrument, having a target is futile. I don’t think that central banks lack instruments in their tool kits, which is not to say that there might not be better ones, but I do think that their main problem has been not having the right targets and understanding what makes a good target. Open market sales tend to reduce NGDP and open market purchases tend to increase NGDP, so I don’t exactly understand what you are missing.
Ron, A supply shock that causes the rate of inflation to rise should not cause a tightening of monetary policy, which I think is what you are suggesting. There is no reason for monetary policy to tighten just because there has been a supply shock. If you keep prices rising at the same rate in the face of a supply shock, you have to force wages down. That’s a prescription for rising unemployment on top of the rising prices, AKA stagflation.
LikeLike
“A supply shock that causes the rate of inflation to rise should not cause a tightening of monetary policy,”
I agree with that statement. My point was rather that one cannot safely say that NGDPT “caps” the inflation level. There are situations where this isn’t true (for example when RGDP falls due to external factors that affect supply capacity).
I would also be interested in your views on my point 2) where RGDP falls due to lack of business optimism. I think this is different from the supply shock scenario but would also lead to IT and NGDPT having different outcomes.
LikeLike
“Given our uncertainty about sustainable growth, an NGDP target also has the obvious disadvantage that future certainty about inflation becomes much less than under an inflation (or price level) target.”
Perhaps this is true in theory, but in practice, the experience seems quite different considering that the inflation Chicken-littles have had lots of time in the spotlight in recent years. As an average Jane, I care more about how much money I will have coming in than about whether inflation will be 2% or 3% or 4% for that matter. The difference between 2% and 3% is really just pennies on the dollar while losing a job costs a lot more even in the long run.
In addition, Goodhart’s argument around production sounds good, but it leaves out an important point that Marcus Nunes made a while back, and I hope I understood him correctly, that to stabilize the price level (if we could all agree on an appropriate measurement of it) NGDP growth would necessarily have to constantly be some degree below trend.It does not seem to me that a monetary focus on inflation/price level in that regard is consistent at least with the Humphrey-Hawkins mandates, or with the kind of world most of us would like to live in with plentiful opportunity.
LikeLike
Yeah, who says inflation-targeting works? As in Europe, Japan or the USA?
BTW, six of the last eight monthly readings for CPI have been down. Check it out. The index has been down in six of the last eight readings. Deflation.
And this after a four-year stretch of very low to zero inflation.
And Goodhart is blubbering about inflation.
More and more, I understand Japan….
LikeLike
David,
I am not sure that NGDP is a suitable target for the central banks and that it is well-matched to their usual instruments. All they could do is… set capital and reserve controls, an overnight discount rate, buy and sell bonds or other financial assets (and that, to my knowledge, more or less exhausts the list). Arguably, nothing of that affects the wider economy much, not to the effect of central banks having direct control over such parameters as NGDP, unemployment, inflation, etc…
In that light, whether B. Bernanke announces to the crowds that Fed now targets ten percent inflation or that they will go for 5% increase in NGDP, he will get the same dubious results. He might as well target RGDP, industrialization or peace in the world, for all the good it does.
There is, as well, another objection to the actual implementation of NGDP targeting. Because it takes time to calculate the GDP of the economy, the central banks won’t be able to get the timely and precise information on the state of the economies and the results of their actions. You can’t control any system like that; and that’s the reason why the firms invest heavily in the systems of managerial accounting even if they already have the financial accounting implemented! One can’t manage the firm based only on the regular financial statements, and I imagine that central banks need real-time controls as well. AFAIK, they are aware of that and on day-to-day scales target something over which they have direct and clear control – overnight interest rates.
Am I wrong?
LikeLike
David, I tend to belive that there is some flawed in NGDPT. Goodhard answers some of my doubts.
As an monetary objetive I only can see a infinite sequence of ex post rectifications. I Don’t understand this sacralization of a variable that is in most of its components a stuff of judgemental criteria (I know because I worked a Lot with National Accout, mainly US’). GDP and NGDP are only wishful thinking. A necessary conventionality, not a reallity. And its periodical revisions are not a good card against uncertitude.
What if after two years NGDP growth of two years ago was not actually 5% but 7%, whereas inflation has grown and long term interest rates have doubled? What if agents see that monetary Policy can be revisited each two years in the opposite sense?
NGDPT is suspect for much more uncertitude than a simpler objetive that, in any case, could be suspended if necessary. The mistake of FED (until the last FOMC) and ECB has been not neglect NGDP but not have changed they objective at time.
LikeLike
Ron, You said:
“My point was rather that one cannot safely say that NGDPT “caps” the inflation level. There are situations where this isn’t true (for example when RGDP falls due to external factors that affect supply capacity).”
Ok, but aren’t you implying that the rate of inflation should be capped in such a case? If we are talking about the rate of output inflation, the answer it seems to me is clearly no. If we are talking about the rate of wage inflation, I believe that I showed in a post that I did a year to a year and a half ago, that a constant rate of NGDP growth will actually cause wages to fall slightly (or fall below the implicit trend of wage increase), which I think is too contractionary.
I agree that RGDP can fall because of business pessimism, and if it does, I think that the optimal rate of inflation rises. Also something I think I have discussed in a post (I am guessing) in the summer or fall of 2011.
dajeeps, Well said. There is a widespread misconception (it occurs to me) that the only uncertainty that matters is uncertainty about the rate of inflation. That makes no sense to me. There are all kinds of uncertainties that we have to try to cope with and navigate, and inflation uncertainty is far from the most important of those. Now that’s OK, unless it turns out that minimizing inflation uncertainty actually increases other types of uncertainty. What I have been arguing for a long time is that, when an economy is stuck in a low-expectations trap, low inflation reinforces the low-expectations trap, thereby worsening other types of uncertainty.
Benjamin, I was really surprised to see the piece by Goodhart. Maybe there’s some kind of central bank virus that infects people who go to work for a central bank. Switching topics, kudos to you and Marcus Nunes on your new book, which I hope to write about some time soon, although I am running way behind on a number of projects.
Roman P., My view is that by conducting open market operations the central bank can determine the amount of currency held by the public which means that it can control the nominal price level. That’s all a central bank has to do to determine NGDP. It’s true that as nominal interest rates fall, the central bank has to work harder and harder to affect the price level because the demand to hold currency becomes increasingly elastic as the nominal interest rate tends to zero, but I don’t believe that there is ever an infinite demand for currency so that an increase in the stock of currency will always have some tendency to raise the price level. I agree that targeting NGDP must allow for a significant margin of error. Scott Sumner has addressed this issue by proposing that the target be expected NGDP which could be targeted via a futures contract with the central bank conducting open market operations by buying or selling the contract. This is a variation on the wage index target (labor standard) that I proposed in my book Free Banking and Monetary Reform.
Luis, I agree that national income accounts are not easily measured. But I think that there is more variability in estimating price levels and real GDP than there is measuring NGDP. But you are right that there are ongoing revisions even in NGDP which means that there would be a problem in defining a futures market in NGDP. Maybe Scott, who has given it a lot more thought than I have, has addressed the mechanics of implementing such a market in light of endless revisions in the data, I don’t have anything more to add on that .
LikeLike
Would I be more favorable ro NGDPT in the case of better measurement, without endless revisions ? Well, in this case I can not see any objection, I think.
LikeLike
“Another concern of Goodhart et al. is that the choice of the target growth rate of NGDP depends on how much real growth,we think the economy is capable of.”
“But Goodhart et al. are worried that achievable growth may be declining. If so, NGDPLT at 5% a year will imply more than 2% annual inflation.”
Please explain to me why there is a limit to the amount of real growth an economy is capable of? For instance why can’t you have 5% nominal growth and 10% real growth?.
Also,
“Open market sales tend to reduce NGDP and open market purchases tend to increase NGDP, so I don’t exactly understand what you are missing.”
Really, that’s funny because nominal GDP and nominal interest rates tend to be positively correlated:
http://research.stlouisfed.org/fred2/graph/fredgraph.pdf?&chart_type=line&graph_id=&category_id=&recession_bars=On&width=630&height=378&bgcolor=%23b3cde7&graph_bgcolor=%23ffffff&txtcolor=%23000000&ts=8&preserve_ratio=true&fo=ve&id=GDP,DGS10&transformation=pc1,lin&scale=Left,Left&range=Custom,Max&cosd=1960-01-01,1962-01-02&coed=2012-10-01,2013-01-29&line_color=%230000ff,%23ff0000&link_values=,&mark_type=NONE,NONE&mw=4,4&line_style=Solid,Solid&lw=1,1&vintage_date=2013-01-30,2013-01-30&revision_date=2013-01-30,2013-01-30&mma=0,0&nd=,&ost=,&oet=,&fml=a,a&fq=Quarterly,Daily&fam=avg,avg&fgst=lin,lin
LikeLike
Luis, OK, thanks for the clarification.
Frank, The limit on real growth is the limitation of real resources and technology. To get more output, you need more input or a better technology. That limit on real growth, in principle, is independent of the what is happening to nominal GDP growth; in practice they are not independent.
Open market sales reduce the amount of currency or high-powered money because the public turns over currency in exchange for the Treasuries. That tends to reduce nominal GDP, which tends to reduce nominal interest rates. Does that help?
LikeLike
David,
“Open market sales reduce the amount of currency or high-powered money because the public turns over currency in exchange for the Treasuries. That tends to reduce nominal GDP, which tends to reduce nominal interest rates. Does that help?
No it does not help. Currently the U. S. federal reserve is holding over $3 trillion of US Treasury debt and other interest bearing securities. Are you arguing that if the Federal Reserve sold all of those securities tomorrow to the public:
1. Nominal GDP would fall
2. Nominal interest rates would fall
Your statement:
“Frank, The limit on real growth is the limitation of real resources and technology. To get more output, you need more input or a better technology.”
To get more output you need more input or a technology different from what is currently available. To get people to buy that output you need salesmen to convince the public that the technology is better.
LikeLike
Frank, Yes, I am asserting that if the Fed sold a significant portion of its holding of Treasuries, it would cause NGDP to fall. The immediate effect on interest rates is not so clear. If the policy, caused a liquidity shortage, interest rates might rise, but over time interest rates would fall if the policy were maintained,
I agree that marketing resources are also necessary to sell products after they have been produced, and I don’t think that is inconsistent with the basic point I was trying to make.
LikeLike