Scott Sumner  

When is nominal GDP targeting optimal?

PRINT
Henderson on Thaler and Sunste... Poverty and "enterprise facili...

Commenter Felipe sent me to an interesting Jeffrey Frankel post on NGDP targeting:

There are good reasons to think that NGDP targeting is better suited to emerging and developing economies than to industrialized countries. These economies are more frequently subject to adverse terms-of-trade shocks, such as increases in world oil prices or declines in prices for their commodity exports. Their economies also tend to suffer larger supply shocks from natural disasters, other weather events, social unrest, and unexpected productivity changes.

The advantage of a nominal GDP target is that adverse shocks of these sorts are reflected equally in output and inflation, rather than imposing the entire burden in the form of a loss in output. This provides the sort of response that one would want anyway, while still retaining the advantages of a rule (communicating the central bank's plans in such a way that it can live with what it has promised to do).


There's clearly some truth to this, but on balance I believe that the case for NGDP targeting in emerging economies is actually weaker than in developed economies. Let's take an extreme example of a terms of trade shock for a commodity exporter. Suppose the price of oil doubles, and the quantity of oil exports is unchanged. Under NGDP targeting a country like Saudi Arabia would have to sharply contract the non-oil sector of the economy. At first glance that might not seem so bad, one sector is booming while the other contracts. But note that oil production is far less labor intensive than other sectors such as services. Indeed even if we relaxed the constant output assumption for oil, and allowed Saudi oil output to increase, it is quite likely that NGDP targeting would lead to sharply lower total employment if Saudi Arabia was hit by a positive oil price shock.

For emerging markets with large and volatile commodity export sectors, I have suggested targeting total domestic nominal labor compensation, which is the largest component of NGDP. This will smooth out employment fluctuations and also keep inflation low in the long run. Indeed I think a good argument can also be made for targeting nominal compensation in developed countries like the US, but the advantages over the much for familiar and easy to explain NGDP targeting are less clear.

Patrick Sullivan directed me to a post that is mildly critical of NGDP level targeting, by Stephen Cecchetti and Kermit Schoenholtz:

The large gap highlights an important aspect of nominal GDP targeting: when the trend rate of real economic growth rises or falls, the implicit inflation objective also changes (in the opposite direction). The policy question is whether that is desirable. Would the Fed reduce systematic (undiversifiable) risk in the economy by stabilizing nominal GDP if doing so raises uncertainty about future inflation and the price level? Doing so conceivably protects debtors who anticipated a certain level of future income, but it also complicates decisions for households and firms who must distinguish relative price changes from inflation surprises in order to make efficient choices.

All of this leads us to conclude that returning to the price path implied by the pre-crisis trend is a realistic possibility. Returning to the earlier nominal GDP path is not. That said, the inflation overshoot that our rough calculations suggest is moderate, so the benefits are likely to be limited. But the costs could include a loss of credibility in the inflation-targeting framework. Would that really be worth it?


Let's start with the idea of returning to the pre-2007 trend line. Given the amount of wage and debt contracts that have been negotiated since 2007, I've argued that it would now be destabilizing to go all the way back to the pre-2007 trend line. Of course if NGDP targeting, level targeting, had been in effect in 2008 the recession would probably have been much milder. But that raises another problem, implicit in the Cecchetti and Schoenholtz quote---trend growth has slowed since 2007, and hence a 5% path of NGDP which would have led to roughly 3% real growth and 2% inflation over the previous 100 years, might well lead to 1% to 2% RGDP growth and 3% to 4% inflation over the next few decades. Would that be undesirable?

Put aside the question of whether 5% is too high. A number of economists have argued that a stable path of NGDP is more likely to produce economic stability and debtor/lender fairness than a stable price level path. George Selgin's "Less than Zero" is perhaps the best study of this subject. Because most debts are nominal, borrowers and lenders are more concerned about nominal income shocks than price level shocks, a point acknowledged by Cecchetti and Schoenholtz. However labor markets are also better stabilized with NGDP targeting than with price level targeting.

Consider the following example. From mid-1963 to mid-1968, inflation rose from a 1% to 2% range up to a 4% to 5% range. The same thing occurred from mid-2003 to mid-2008. But surely the public should have treated these two cases very differently. The first example was a powerful demand shock, which caused NGDP growth to accelerate. In the second case it was mostly an adverse supply shock, and by mid-2008 NGDP growth was slowing sharply even as inflation peaked at nearly 5%. Not surprisingly, workers responded to the first shock by increasing their pay demands sharply (with a lag), whereas in the second case wage growth remained fairly modest. In the long run wages follow NGDP more closely than the price level.

Of course wages are only one of many relative prices in the economy. But I'd argue they are the most important, largely responsible for the business cycles. If inflation is unstable then consumers will have a difficult time ascertaining relative price changes for toothpaste and cars. But is this actually a big problem? On the other hand if NGDP growth is unstable then the relative aggregate wage level will move to a suboptimal position---which really is a big deal.

Over at TheMoneyIllusion I just did a post showing that Keynes favored thinking in terms of NGDP and employment, not inflation and real growth. Back in the 1970s I also studied the models of inflation shocks, and I do understand the underlying assumptions that led most New Keynesians to favor inflation targeting. But I fear those assumptions are wrong, and I believe the events since 2008 have exposed a serious weakness in the inflation targeting approach. Some of those flaws could be fixed if we moved to price level targeting, as discussed by Cecchetti and Schoenholtz, but not all.


Comments and Sharing





COMMENTS (13 to date)
George writes:

Just an observation: If anyone could actually "target" NGDP (or any other variant of GDP) do you think the EU and the US would have looked so bad the last few years? I think not. You economists need to get off of these silly discussions that have NO basis in reality.

Just because you can approximately measure something doesn't mean that you can make it happen...

James writes:

Scott:

Exactly how much (as in numbers) of a difference would it make would the world look if the Fed adopted NGDP targeting?

Clearly this is an important issue to you, but I don't think you've done the best job explaining why it should be important to anyone else. If NGDP targeting would decrease average unemployment and RGDP volatility by a basis point each, that's still nothing to get excited about.

Even proponents of returning to gold as money can plausibly claim that their idea would reduce inflation by about 100%. That's actually exciting.

ThomasH writes:

Scott,

Withing the context of this exposition could you address the difference (if any) in how NGDP would work for a large quasi closed economy (tradable output makes up a small part of GDP) like the US and a small open economy. Perhaps that is what you meant by "developing" v "advanced" as shown by the Saudi Arabia example.

Also, (in the two classes of economies) is there a role for "automatic" fiscal stimulus meaning undertaking investments whose PV have increased because of unemployment that develops (?) following a positive TOT shock if the MA is following a NGDP target. Does the choice of target depend at all on the MA's expectation of whether fiscal policy will in fact carry out the "automatic" stimulus suggested?

dannyb2b writes:

Maybe a dumb question.

Could ngdplt unhinge inflation expectations if the the economy falls away from its target for sustained period of time and then ngdp starts to recover excessively fast to catch up?

Scott Sumner writes:

George, A few decades ago the major central banks decided to target inflation at 2%. Since then inflation has averaged roughly 2%, except in Japan, which targeted a rate of zero until recently. Now Japan has adopted a 2% inflation target, and inflation is rising there as well. Amazing coincidences!!

James, If you are one of the millions of unemployed you might feel differently about the advantages of NGDP targeting. I have some papers on the subject over at the Mercatus Center, which you can google if you are interested in learning about the many, many advantages of NGDP targeting.

And no, gold does not stabilize the price level. The price of gold was fixed from 1929-33, and dollars were fully redeemable during that period.

Thomas, Fiscal policy should be judged strictly on "classical" grounds, not on its impact on demand. Strictly cost/benefit.

The closed/open economy distinction is not important for NGDP targeting, but heavy commodity exports is.

Danny, Inflation expectations would vary a bit, but not too much.

ThomasH writes:

My question is what conclusions there are for monetary policy if the Monetary Authority thinks that fiscal policy is insufficiently "classical," that it will NOT increase investment in response to the lower opportunity costs of investment goods and at least temporarily a lower discount rate. And should fiscal policy increase expenditures beyond this "classical" formula if MA are expected not to follow NGDP (acting, for example, as if they were constrained by a zero nominal interest rate)?

James writes:

Scott:

I'm looking for the executive summary, not advice to read some papers. If any of your papers give precise and falsifiable point estimates of the impact of NGDP targeting, I can't imagine any good reason for you not to cite those estimates here. Let me quote a sharp guy on this kind of thing:

18. Have you seen Garrison’s Powerpoint slides?

Yes, several times. Using his terminology, we now face a secondary depression. BTW, please don’t ask me to read such and such a book on Austrian economics. The comment section is where you get to show me how useful ABCT really is, by making thought-provoking comments on the post. Until I finish my other projects, I won’t have much time to read anything."

I'm not even asking you for thought provoking. I just want you to make a public point estimate of the impact of NGDP targeting, something like "Unemployment will be 4% on average with NGDP targeting and 6% on average otherwise." This way, if some central bank embraces NGDP targeting, we can know if NGDP is performing as well as promised, sort of like when the people calling for fiscal stimulus give public estimates of fiscal multipliers.

Your comment about "if you are one of the millions of unemployed..." seems odd coming from an economist, almost like you were trying to make an emotional appeal. The economic case for any policy should be in how it affects everyone in society, not how it affects a specific class of people, right?

Michael Byrnes writes:

James wrote:

"Exactly how much (as in numbers) of a difference would it make would the world look if the Fed adopted NGDP targeting?

Clearly this is an important issue to you, but I don't think you've done the best job explaining why it should be important to anyone else. If NGDP targeting would decrease average unemployment and RGDP volatility by a basis point each, that's still nothing to get excited about."

Scott has been answering this question for the past 5-6 years, no?

What you would get from NGDP targeting is long-term inflation comparable to what was seen during the Greenspan Fed era and a moderated business cycle.

"Even proponents of returning to gold as money can plausibly claim that their idea would reduce inflation by about 100%. That's actually exciting."

Not really. Under the pre-Fed gold standard, long-term inflation was minimal, but the price level and business cycle were anything but stable in the short-term.

George writes:

Scott,

The US has indeed hovered around 2% for quite a while - point taken. However, Japan is at almost 2x that rate (high 3's). I have no doubt that they can raise inflation (you can always print more money), and when there are no transient phenomena causing trouble - oh, say like a debt crisis - you can approximate the target inflation rate. However, Japan is highly leveraged...if their interest rates rise very far they will trigger a crisis. Their ability to maintain price stability may or may not survive the ensuing economic troubles.

If NGDP targets are simply inflation targets, then there is a much easier task at hand. The RGDP part of NGPD is not as easy to control.

In the US, things appear stable, but an awful lot of money is just sitting in bank vaults (I know, the Fed is paying them to keep it). A change in bank behavior could trigger inflation - if that money gets out to the general economy.

There have been very interesting discussions by your fellow economists at GMU. My comments are most likely ridiculous over-simplifications, but I don't think that the debt levels around the world bode well for long term stability.

James writes:

Michael Byrnes:

Look at what you have done:

"Not really. [In response to the claim that gold as money would reduce inflation by about 100%] Under the pre-Fed gold standard, long-term inflation was minimal"

So inflation would "not really" be reduced by about 100% but it was "minimal" the last time there was a gold standard?

And here you've changed the subject:

"...but the price level and business cycle were anything but stable in the short-term."

If you think macroeconomic instability was greater before central banking, that's between you and the data but you can't refute a claim about a first moment by making a claim about a second moment.

"Scott has been answering this question [point estimates for the impact of NGDPLT] for the past 5-6 years, no?"

If Scott has provided numerical point estimates, (like "1.6%," and "15% less variance" and unlike "moderated" and "comparable to the Greenspan era") I'm not aware of them. If you can quote any such point estimates made by Scott, that would be great.

Michael Byrnes writes:

@ James

By "not really", I meant "not really that exciting" (in response to your comment that minimal inflation over the long-term would be exciting. Difference of opinion, n'est pas?

"If Scott has provided numerical point estimates, (like "1.6%," and "15% less variance" and unlike "moderated" and "comparable to the Greenspan era") I'm not aware of them. If you can quote any such point estimates made by Scott, that would be great."

Would it have been a good thing if the 2008 recession had happened but looked a lot more like the mild 1980 recession (2.2% rise in unemployment from trough to peak)?


James writes:

Michael Byrnes:

Less unemployment is better than more, other things equal. If there are tradeoffs, then it depends what the tradeoffs are.

If some workers are brought out of unemployment by nominal wages that they would reject if they had more accurate expectations about the future purchasing power of money, then those workers will be worse off.

It's ok if you can't answer my question about point estimates for the impact of NGDPLT but it would look a lot less suspect if you said so explicitly, rather than changing the subject again and again.

Felipe writes:

Hi Scott (same Felipe here), thanks for taking a look at the article. In the comments you say:

The closed/open economy distinction is not important for NGDP targeting, but heavy commodity exports is.

This is I think very important, but has so far been unexplored. Perhaps you could do a post about this distinction, why it matters and (most importantly) what to do about it? I think Lars Christensen would disagree with you (as far as I have managed to understand from commenting on his blog). He advocates an Export Price Norm, which links monetary policy even more strongly to the commodity export price.

Comments for this entry have been closed
Return to top