Scott Sumner  

NGDP futures targeting means never having to say you're sorry

Ghost World and Zombieland... Does Market Power Affect Optim...

There are many arguments for rules over discretion. Here's one that I haven't seen before.

Policymakers are only human, and don't like to admit to making mistakes. Especially when those mistakes might have harmed millions of people. In contrast, markets don't have egos. Markets don't mind admitting mistakes. Indeed hardly a week goes by without some sort of "correction" in asset prices. By 'correction' I mean retracing a previous asset price movement that was caused by expectations that, in retrospect, later seemed unwise.

One can find many examples of central banks letting their egos get in the way of good policy. In 1936 and early 1937, the Fed raised reserve requirements sharply. Later in the year it was obvious that they had made a mistake. But in the Fed minutes it is apparent that FOMC members were reluctant to do anything that would be an implicit admission that the previous policy might have contributed to the steep recession of late 1937.

Here's another example. In April and July 2011, the ECB raised interest rates under the leadership of Jean-Claude Trichet. They did so because they were worried about headline inflation. In retrospect, we know that that fear was mistaken, and indeed eurozone inflation over the past 7 years has been far too low. But in early 2011 there had been a brief inflation spike due to international factors like rising commodity prices, and also VAT increases associated with austerity.

Now let's consider how soon it took for it to become obvious that the ECB had erred. We do know that the eurozone almost immediately went into a double dip recession. But how soon was that apparent to the markets? Unfortunately we lack a eurozone NGDP futures market for the eurozone, but the German stock market is suggestive:
Screen Shot 2015-08-05 at 4.18.40 PM.png
Unfortunately the scale is cut off, but the two relevant horizontal lines are for 7500 and 5000, so soon after the second rate increase on July 13, the German stock market plunged from the low 7000s to the low 5000s, a quite sizable drop (comparable to the recent Chinese stock crash.) It seems reasonable to assume that this is about when the markets perceived the double dip eurozone recession, which we now know was beginning at roughly this time. The German market had rallied over the previous year on hopes for recovery, and then "corrected" when the double dip recession occurred.

How long did it take for the ECB to perceive its error? Maybe not too long---they could see what was happening in the markets. But a sudden reversal of ECB policy toward loosening would have been an admission of error, an error that did great damage. And human nature being what it is, that's not easy to do.

How long did it take the ECB to admit error? One indication is that they began cutting rates on November 9, 2011, about 4 months after the previous increase. And why November 9th, why not September or October? I can't be certain, but consider the following. On November 1, 2011 Trichet resigned, and was replaced by Mario Draghi. Draghi could shift course without any embarrassment, because he had no stake in the previous decision. Coincidence? Perhaps, but if not then this leads to a dilemma:

1. The "credibility" argument suggests central bankers should have very long terms, so that they can carry out their promises.

2. The "correction" problem suggests they should have very short terms, so that someone new can come in and correct any previous errors.

Or maybe we should turn policymaking over to an NGDP futures market, where egos don't get in the way. Markets can correct without any embarrassment, they are shameless.

PS. If you are too young to understand the reference in the post title, consider yourself lucky.

Comments and Sharing

CATEGORIES: Monetary Policy

COMMENTS (8 to date)
pyroseed13 writes:


Tyler Cowen recently posted on his blog this paper, suggesting that the 1936-1937 recession could not be blamed on raising the reserve requirements:;jsessionid=6BA3E5276104C4C1C1F36C91E363DED8.f02t04

I would be interested in hearing your thoughts on it. Thanks.

ThomasH writes:

I understand the hypothetical benefits of NGDP targeting (a clever way of getting around the inflation rate ceiling target) and the the theoretical argument that NGDP targeting would be better that even unconstrained inflation targeting. But how DO we "turn policy-making over to an NGDP futures market?" Does the Fed just make a (by hypothesis) credible announcement and never have to buy or sell anything but NGDP futures contacts again?

[BTW the ECB story seems to work only for an inflation rate ceiling policy. If the price level using some index exceeds the target trend one month and the monetary authority buys ST assets to raise rates and then the price level drops back below target the next month and it sells ST assets to lowers rates, it has not changed policy and has nothing to be apologetic about. It would just be following the market in the same way it would if it sold NGDP futures one month and bought them next.]

Matt writes:

I like this point, but in this particular instance it requires an crucial role for temporary decisions about interest rates - one that hasn't been fully articulated or clarified.

Suppose that the ECB was 4 months behind in cutting the short rate in response to bad news. So what? Investors could anticipate the cut, and it was quickly baked into the yield curve: if you look at the 5-year German yield in 2011, it fell from a high of 2.80% in early April to a record low of around 1.00% by early September, before any official rate cuts were put into effect. The market delivered the effective rate cut before the ECB got around to it. (That is, unless one thinks that the current value of the short rate itself, independent of its future path or its connection with long rates, is the crucial variable. But does anyone really think this?)

Really, what was the harm of those few months of delay? One could say "it demonstrated that the ECB had a lower implicit target for inflation or output growth", which had more lasting harm - but that really begs the question, because then the damage came from the mindset signaled by the delay rather than the delay itself.

The "don't have to say sorry" point is more viable in other cases, where the delay itself really does make a big difference. Effective stabilization of the business cycle can require large and rapid changes in the short rate; modern central banks, which inertially adjust the rate in neat little increments, are loath to do this. Pushing the short rate from 6% to 0% instantly, rather than slowly descending to 3% over the course of 2 years, does make a big difference by any measure; that's the promise of an automatic, market-driven system like NGDP targeting. (Of course, the question here is whether there might be any hidden wisdom underlying central banks' traditional lethargy; might such rapid swings in the short rate be dangerous for some other, outside-the-model reason? I don't know.)

This would apply to more recent events too, if not for the zero lower bound (you could lower rates from 1% to -5%). But given their perceived ZLB constraint (putting aside the question of whether this is really a constraint), I have a tough time seeing the profound harm from the Fed's dithering in mid-2008 or the ECB's dithering in mid-2011. They didn't have very far to adjust rates anyway, and at most their failure to act immediately raised long-term yields by a handful of basis points.

Scott Sumner writes:

pyroseed, I just did a post on that:

Thomas, Yes, the delay would not be very costly under price level targeting.

Here's my paper on NGDP futures targeting:

Matt, During that 4 month period the Wicksellian equilibrium rate fell very very sharply. So the rate cut had much less effect than if it had come earlier. But yes, under a proper monetary regime like PLT or NGDPLT the delay would make little difference. Under the actual regime it was very costly.

ThomasH writes:

Thanks for the link. I certainly hope I am not the only participant in this who had never understood exactly what you mean by NGDP targeting.

I had not previously understood NGDP targeting as a way of using the market to prevent a monetary authority from using discretion in employing policy instruments. I think the paper makes a good case and I am not unsympathetic to advocating first best solutions to problems. But I noticed that in the introduction in which you consider various monetary policy rules to which NGDP targeting will be shown superior, the actual policy of central banks (or constraint on policy) –inflation rate ceilings – was not there.

For example, you treat the Fed decision in 2008 not to reduce ST interest rates as an error based on a mis-estimate of future inflation rates (and such a mis-estimate was certainly there), but it seems more enlightening to think about that action as a failure to follow its (supposed) price level target and to consider the reasons that led to that failure.

It seems to me that the error was not just a mis-estimate of behavior of the price level if rates had been lowered, but rather an (also possibly mistaken) estimate that the implicit inflation rate ceiling would be breached. And later the reluctance to use more vigorous QE (right up to today) seems more than just conservatism about using a new policy instrument, but again fears that it would lead to future breaches in the inflation rate ceiling. Finally talk about raising ST interest rates in the near future (when the price level is still below its supposed target) implies that more than errors in discretion is involved. The fundamental problem seems to be a sub-optimal constraint on following the supposed price level target. (Why losses from uncertainly about future price levels ever loomed so large relative to losses from lack of aggregate demand is a separate but probably not unrelated problem.)

Without doubting that NGDP targeting would be superior to price level targeting, I think a more complete analysis would have to deal with likely constraints on the policy rather than to assume them away. If I am not mistaken, the previous attempt to make monetary policy by a strict rule – increasing M2 at a constant rate -- foundered not because it produced erratic movements in the price level, (or even inflation rates) but because it led to violation of an unacknowledged constraint, “unacceptable” variability in ST interest rates.

Scott Sumner writes:

Thomas, In my view NGDP targeting would lead to more stable interest rates than current policy.

As far as M2 targeting, the Fed paid lip service to the idea, but never took it very seriously.

ThomasH writes:


I did not mean that interest rate instability, per se, was a possible hidden defect of NGDP targeting, just that it could be aborted if we do not understand the reasons/political forces behind the failure of the Fed to actually follow the supposed PL target.

Can you point me to an examination of the supposed M2 targeting experiment? I think my impression is a common one. [I'm definitely of an age to understand your reference, my Sophomore year, I think, and I was a great fan of Friedman and in my field, Harry Johnson. But by the time of the "experiment" I was not paying much attention to US macroeconomics.]

Dustin writes:

NGDPLT - or any inflexible rule-based policy - will have to overcome not only intellectual scrutiny, but also human psychology, given our bias to control our own fate.

Planet Money recently published a podcast in which the issue of discretionary human intervention was discussed in the context of technology. The episode explores elevators, Google Car, and airplanes as case studies to assess the impact on safety when human intervention is allowed. The gist: 1) human intervention is bad once technological capabilities exceed our own and 2) even when technological supremecy is established, we are reluctant to yield discretionary control.

In an academic setting, where folks tend to think very highly of themselves, the bias toward discretionary control is probably even stronger

Comments for this entry have been closed
Return to top