« Ontario and its Physicians: Richer Than They Think No Longer | Main | The federal deficit is shrinking quickly »


Feed You can follow this conversation by subscribing to the comment feed for this post.

If you promise to go to Toronto, most of us will cut you some slack on ways and means. We will evaluate your promise according to how you fulfilled similar promises in the past - did you go to Montreal when you promised to do that? etc.

However, if you promise to go to the moon, that is naturally going to provoke some curiosity about the details. Your record of going to Toronto is not very impressive in that context.

So once again you are indulging in a rather self-flattering question-begging, since the debate turns not on expectations versus actions, but rather on whether getting the expectations you want is more like going to Toronto or going to the moon. People like you and Scott Sumner and Ben Bernanke the professor think the former. But it seems that Ben Bernanke the Fed governor thinks the latter.

The exponential assert purchases would present quite the game of chicken.

Does it matter that it is hard to measure NGDP accurately?

What does the CB use to measure expectations--NGDP futures? How well would such a futures market work, given that NGDP is hard to measure accurately?

The CB can only really target one expiration for NGDP futures, like 2 years out, correct? Shorter expirations may then deviate from the LPT...

Phill: I can imagine that if Bernanke declared that either he will be on moon next year or he will drown the world in dollars, this "promise" could convince many people willing to make his trip reality. Of course, only if his threat is credible.

Or imagine some CEO in technology company who just saw Kennedy's speech about getting to the Moon. Do you think that they would tell themselves "Nah, he did not tell what concrete steps is he willing to take to deliver on this promise. I doubt that there is even technology available for this. Just ignore him he is crazy." Or they would tell themselves - "Great, give me my chief technology officer on the phone. Immediately start research and investment in areas that are needed for this moon trip, we need to beat our competitors in this once in a lifetime opportunity"

What if Central Banks aren't all-powerful? I was listening to an online chat by Paul Krugman who claims to have invented the phrase "Confidence Fairy", it was a satirical take on the concept.

What if Central Banks get caught in bad promises, like inflation targeting when we need higher inflation? It happens.

Or what if our entire economic response to this depression, and I use that word purposefully, has been wrong? Like the fact that the bulk of the US and problem has been due to public-sector austerity, in particular teacher layoffs?

Paul Krugman had other good points that Ben Bernanke the professor isn't Ben Bernanke the Fed Chairman, and the current Chairman has been "assimilated" by the Fed Board. Therefore we should "Let Ben be Ben" or "Free the Beard".

Phil: Are you 100% certain that the US economy will be at the ZLB liquidity trap forever, or 100% confident that the Fed's words mean absolutely nothing about what it will try to do? If not, you have given me all the leverage I need. I can promise to increase the stock of base money one hundredfold for ever. Or hold nominal interest rates at 0% forever. Or keep on buying up everything that moves forever. And I will raise your expectation of the future price level path by some positive amount. Which means you will start spending earlier, and expect the economy to escape the ZLB liquidity trap earlier. Rinse, repeat.

(Thanks for being not too harsh BTW; I guessed you probably wouldn't like this post, sniff.)

Andrew F: "Does it matter that it is hard to measure NGDP accurately?"

It might, if we use NGDP futures markets to implement the policy. We have the same problem in principle with the CPI and inflation targeting. We know the CPI is not a perfect measure of the price level, both for theoretical reasons and because the sample can never be fully representative of the population. The only difference is that the CPI is a one-off survey, so the data never gets revised.

"The CB can only really target one expiration for NGDP futures, like 2 years out, correct? Shorter expirations may then deviate from the LPT..."

The question of the targeting horizon is one that has been largely ignored by NGDP people, AFAIK (I might have missed it). The BoC is reasonably explicit about its inflation target horizon, but even so cuts itself some slack to vary that horizon a little depending on circumstances. Being a pragmatist, I would suggest the Fed start out with a rather vague targeting horizon. Because of the level-path nature, the length of the horizon matters less than it does for inflation targeting. It will try to get onto the NGDP level path as soon as possible, but is unsure about when it will succeed.

Determinant: central banks aren't all powerful, but there is no limit on the quantity of money they can create. If Zimbabwe can do it, then...damn! I just lost the argument!

"What if Central Banks get caught in bad promises, like inflation targeting when we need higher inflation? It happens."

It might happen. But we are not 100% certain it will happen before the economy escapes the ZLB.

Being wrong is always a hazard of offering policy advice.

Re: measuring NGDP, would measuring net nominal wages be easier? If so, this counts as a point in favour of targeting net nominal wages, despite some of the specific political/communication problems involved in doing so.

On economics, why target wages? They have hardly had anything to do with inflation over the past 30 years which has been driven far more by primary product price increases such as the heavy correlation with oil.

W Peden: I don't know. My guess is that wages are even harder to measure, given the fuzziness between wages and benefits, and wages and non-labour income. But that's just a guess.

Determinant: let's just suppose that what you said is true. Suppose there is a very low correlation between nominal wage inflation and nominal price inflation. Why would that be an argument against targeting nominal wages? If you wanted to keep inflation stable, then sure it would be. But if what you wanted was to keep inflation stable, you would target inflation anyway.

Nick Rowe,

I notice that the US government publishes monthly wage and salary disbursements. Anyway, even given the fuzzinesses you mention, I'm sure a large representative sample of wages could be drawn from unproblematic cases on a regular basis.


Nominal wages provide a nominal anchor causally related to aggregate demand and the business cycle-


If what you want is to prevent excessive swings in aggregate demand, then nominal wages are just as good as NGDP and (if data is more easily obtained) superior as a target variable.

And I don't want monetary policy to target inflation, except insofar as it relates to aggregate demand: if prices are rising purely because oil is dear, then I want those prices to rise. Higher aggregate prices under supply-shock situations encourage innovations and investments; they should be allowed to play this role.

I agree about supply shocks, but Nick insists such shocks and contractions don't exist in macro. He and I agree to disagree.

If you're going to target wages, then how to you propose to run counter-cyclical policy when wages are depressed but corporate profits are high? The obvious answer seems to be inflation and unionization to rectify power imbalances but somehow I think that's politically unpalatable to many.


The same thing I'd do in a Long "Depression" scenario where wages are outperforming profits: nothing. It's not a monetary policy matter, just like the NGDP/RGDP split is not a monetary policy matter.

Note that the same situation can occur under strict inflation targeting when we accept the possibility of supply shocks: if we have steady NGDP but weak real GDP growth or a contraction in RGDP, then we have a situation where NGDP and RGDP have to be squeezed in order to keep inflation on target. You can go down the route of price controls under a scenario, just like you can go down the route of unionisation in the nominal wage case. My recommendation would be to do neither and tolerate inflation in both cases.

The more I think about it, the more a nominal wage target seems to have benefits that NGDP targeting does not e.g. if credible it provides a stable starting basis for wage contracts, especially long-term wage agreements. There's a big existing literature on the matter (which I haven't read at all).

Good post/good song.

All WCI posts should come with a song.

Well I think Phil Koop made the main point I would make. Of course we can influence people's actions by making promises that influence their expectations. But some of our promises will have little influence, if they are not grounded in prior beliefs on the part of our hearer about the kinds of outcomes we capable of achieving through means other than promising. My promise to meet my friend in Boston will succeed in getting him to go to Boston, but only because he believes that I have the mechanical means of getting to Boston.

Also, the set of people who are influenced by a person's promises is a subset of the people who are (a) listening to that person and (b) understand what he says. For 90% of people who hear "NGDP blah blah" on the radio on their way hope from work there will be no psychological impact at all. And only a fraction of the remainder will believe the central banker can do what he says.

Dan: change the metaphor: "The party will be in Boston. BYOB." If enough people hear the message, they all go to Boston, and there is a party in Boston. I don't make the party happen. They make the party happen, just by showing up. Hmmm. I like that metaphor. And we can always afford a party. The valid version of Say's Law.

Suppose the Fed chose *one year* as its “targeting horizon”: it professed to aim for NGDP one year from now to be (say) 5% above what it is now, and it promised to continue to aim at such a target indefinitely. If this promise were credible, the market would expect that NGDP would in fact be 5% higher in a year, and 10.25% higher in two years, and 15.76% higher in three years, etc. In fact, if it were easier to measure NGDP, and it could be done continuously, the Fed may as well target NGDP *tomorrow* or, indeed, *one second from now*, at a rate that, compounded, would produce growth per annum of 5% (or whatever).

Of course, the Fed should be *targeting the forecast*: the forecasted *level* of NGDP. Sumner wants to create a futures market for NGDP, which would make the Fed’s job trivially easy. The *horizon* for this futures market--three months, one year, two years, whatever—would be pretty nearly irrelevant—again, assuming Fed credibility.

There are some issues that come to mind about targeting NGDP

1) Revisions - It's hard to have a futures market without a fixed sales price. Nobody goes back a year later and revises the delivered price of soybeans or pork bellies. If you look at futures as a feedback mechanism, control theory would say that delays in the feedback loop are generally a bad thing. It costs phase margin. You pay performance penalties to compensate for this. Furthermore, a feedback system where the loop gain and delay have a large noise component is very ugly. Better and more timely statistics might fix this problem.

2)NGDP isn't all that great a measure of welfare (pretty much regardless of your definition of welfare). It ignores changes in asset prices, is filled with odd accounting choices, mismeasures imports and is massively imputed... This could probably be corrected if enough people agreed what right correction was, and there was good data with which to make it.

Philonous: I think a very short targeting horizon would be dangerous. It takes time for quantities to adjust, and time for prices to adjust, and they all have different adjustment speeds. If you made the horizon too short, that very small subset of prices and quantities that could adjust would have to do a lot of adjusting to get aggregate P.Y on target. So you might get a lot of fluctuations in relative prices and quantities. One or two years would be my preference. As long as people expect P.Y to come back to the target path one or two years in the future, actual P.Y can't depart from it much.

Peter N: I think that if we use Scott's method we would have to agree on which revision to use for the futures market. As long as there is no bias in the revisions (is there?) it shouldn't matter much which version is used, as long as it is agreed on in advance.

Agreed that NGDP is a bad measure of welfare. RGDP is better, but still bad. But I don't think the point is to try to target welfare, which can't be done anyway with monetary policy, since welfare, however you define it, is a real rather than a nominal variable. What matters is to target monetary exchange, and NGDP is a better proxy for that.

“As long as people expect P.Y to come back to the target path one or two years in the future, actual P.Y can't depart from it much.” I.e., actual P.Y can’t depart much from the desired path, *even on a daily basis*. But then your worry about possible big short-term fluctuations seems misplaced.

By the way, Sumner doesn’t care intrinsically about NGDP; his target is *the expectation for NGDP one year (?) in the future*. I still doubt that it would matter if this were altered to *one DAY in the future* (taking NGDP *per day* rather than *per year*, and leaving aside practical difficulties in measurement). (Admittedly, Fed actions aren't instantanious; they'd need at least a few seconds to respond to changes in the Sumnerian futures market.)

C'mon Nick. Now the Fed is the national party master who succeeds in elevating spending purely by calling a spending party and giving people the social permission to spend?

Could I suggest that one of the sources of disconnect in a lot of these discussions is that those of you who have professional training in macroeconomics and central bank policy have acquired along the way a wildly exaggerated estimation of the social position of central bankers?

Just because all 10 people in your Money & Banking graduate seminar on Wednesday might know what Ben Bernanke said on Tuesday doesn't mean a substantial number of other people throughout the economy are paying close attention to him or are disposed to alter their spending behavior one iota in response to anything he says.

Dan: do asset prices matter for investment and consumption? Do traders in asset markets pay close attention to Ben Bernanke? Do you think there might be a positive feedback process (call it "the Keynesian multiplier" if you like)? I would say "yes" to all three.

Yes, I think some asset traders shuffle their money back and forth among different assets in response to things Ben Bernanke says, as well as in response to hundreds of other tips, news items and signals. They pump up bubbles in one market and deflate bubbles in other markets along the way. But I don't think all that trading has a large macroeconomic effect, since I think it mostly amounts to mere shifts that cancel each other out, and the asset traders are just figuring out how to acquire claims on productive businesses that already exist. I believe the economy mainly grows from the ground up, not from The Street or The City down. Most of the traders are not engaged in long-term investing or creating, and don't have a vision that extends more than 6 months out, but are just looking for places to park dollars to make the maximum amount of free money they can get. They are not growers of the economy but either parasites upon it or the agents of parasites, at worst, or at best play an essential intermediation role with savings, but are not really creating or driving anything.

More important than the money-jobbers are the actual operators of business who frequently invest out of their own savings. I believe they respond most to demand signals within their own markets, not to grandiose and unverifiable claims and expressions of hope about national aggregates. More important than all of them are 100 million householders who base their spending decisions on their actual income and financial position primarily, and their expectations of future income which are based on real, local conditions they can see all around them.

You want to grow the economy? Sprinkle more seeds on the ground. The monetarist approach is to have some shaman do a rain dance, which they hope the farmers will confuse with actual rain, and in response to which the farmers will coax miracle sprouts from the dry, seedless soil in through a sheer act of hope and will.


There is definitely some bias. On average it's not dreadful,


but it may be that it's at its worst when it is most important


"What matters is to target monetary exchange, and NGDP is a better proxy for that."

No, it's not that great for that. It's a measure of certain sorts of production real and imputed. It ignores all other spending. Added value isn't spending (and it isn't added value either in any useful sense, but that's a different problem). In particular it misses things like sales of non-new housing.

And it's got a problem with imports. Suppose you found out that US productivity growth was really %0.8. Would that matter to your prescription for the economy? You get these kinds of problems when you're trying to calculate added value. Like aggregated demand it's beautiful from a distance, but up close, not so much.

The comments to this entry are closed.

Search this site

  • Google

Blog powered by Typepad