Scott Sumner  

Bretton Woods as a "guardrails" approach to monetary policy

Adam Smith's pins... Earth 2.0?...

I now favor a monetary policy rule that I have dubbed the "guardrails" approach, although a more accurate metaphor might refer to the beeper you hear if you are about to hit a car in the front or rear when parallel parking. Under this approach, the Fed would offer to sell unlimited NGDP futures contracts at a price featuring 5% growth, and also offer to buy unlimited NGDP futures contracts at a price featuring 3% NGDP growth. Someone expecting more than 5% NGDP growth would buy these contracts from the Fed, and profit if growth did indeed exceed 5%. A bearish investor would sell 3% NGDP futures contracts to the Fed, anticipating sub-3% growth.

Because this is an unfamiliar concept, I'd like to compare it to the Bretton Woods regime. Under that system, central banks promised to keep the foreign exchange value of their currencies within a band of plus or minus 1% around the official par value.

To make things simple, let's suppose the Canadian dollar had been pegged to the US dollar at one for one. Then the Bank of Canada (or their Treasury) would promise to sell unlimited US dollars at a price of $1.01 Canadian, or buy unlimited US dollars at a price of $0.99 Canadian. As long as the actual exchange rate was within the band, traders would have no incentive to buy and sell US dollars with the Canadian government. But the plus or minus 1% exchange rate band would provide "guardrails" or limits on the amount of exchange rate volatility that the Canadian government would tolerate.

Now let's compare the two approaches, Bretton Woods and NGDP futures guardrails:

1. Under both systems, the central bank would be completely free to conduct discretionary monetary policy as they saw appropriate, as long as they adhered to their promise to buy and sell their currencies at the specified guardrails. Despite this flexibility, these are clearly rules-based approaches, which put important limits in discretionary policy. (Just as the US government promise to buy dollars at $20.67/oz. allowed some flexibility, but also put clear limits on discretionary monetary policy during the 1920s.)

2. Under both regimes the system can be "calibrated" to become either more discretionary or more rules based by widening or narrowing the width of the guardrails.

3. The Bretton Woods regime can be regarded as a wager that the Canadian economy would perform best if the exchange rate were kept roughly stable vis-a-vis the US dollar. My guardrails proposal is a wager than the US economy would perform best if NGDP futures prices remained close to a 4% growth target. That wager has two components; the assumption that 4% NGDP growth expectations are desirable, and the assumption that the market price of NGDP growth contracts is close to the market expectation of NGDP growth.

4. Neither system requires the central bank to do anything, unless asked to by the public. Thus the Fed would not have to set up a NGDP futures market, and it would not matter at all if no such market existed. The Fed would not have to watch for fluctuations in NGDP futures prices. They could simply sit back and wait for traders to approach them with offers to buy and sell NGDP futures contracts at the specified price.

5. Both regimes expose the central bank to investment risk, but only if they allow the future value of their target to move outside the guardrails. Thus if the Canadian dollar were to fall to below 99 cents, the Canadian government would suffer losses on the Canadian dollars they had purchased to prop up the exchange rate. If the future level of NGDP fell below 3% growth, then the Fed would lose money on its purchases of 3% NGDP contracts.

6. Neither regime is susceptible to the zero bound problem. The Swiss recently pegged their currency to the euro for a period of over three years, despite being hard up against the zero bound (actually negative rates.) They could have chosen to do so for much longer. At the time, speculators were also buying the Danish krone, and Tyler Cowen suggested that Denmark would provide a good test of the claim that the Swiss were somehow "forced" to revalue. Of course the Danes did not revalue, despite also being at the zero bound, and the Swiss could have also avoided revaluation if they had chosen to. Indeed the Swiss probably erred in revaluing their currency upward, which will make the zero bound problem in Switzerland even more deeply entrenched.

PS. About the guardrails vs. beeper metaphors. My proposal can be seen as like a car beeper in the sense that the "driver" (i.e. the Fed) is free to ignore the warning if he thinks the computer (i.e. market) is not accurate.

Screen Shot 2017-04-24 at 11.37.09 AM.png
On the other hand it can be seen as a guardrail in the sense that it commits the Fed to keep the market price of NGDP futures contracts within the 3% to 5% growth range.

Screen Shot 2017-04-24 at 11.37.56 AM.png

COMMENTS (12 to date)
Thaomas writes:

How do we figure out what the optimal guardrail percentages are? Presumably it is related to an estimate of what real GDP growth can be with the optimal price level target and an estimate of that optimal target (which in turn depends, I suppose, on just how "sticky" prices and wages are), but what's the model to estimate these parameters?

And as a practical matter, doesn't this imply that the inflation rate might occasionally exceed the politically acceptable ceiling? Ballard in the interview with Beckworth seemed to say that the Fed could not politically allow (or at least to admit to allowing) more than 2% inflation? So does the proposal sort of assume away (rather than overcoming) the real-world constraints on Fed policies?

Alternatively, some of Fischer's statements imply that the real constraint on Fed policy is allowing ST interest rates to fluctuate "too much" (because banker don't like big fluctuations?) But anytime the Fed were actually having to buy/sell in the NGDP futures market, interest rates would not be controlled by the Fed. Again, have we not just assumed away the political constraint?

Kevin Erdmann writes:

The guardrail analogy reminds me of the highway outside of Ouray, Colorado, which rides the edge of a cliff with no guardrails.

I suspect it's a relatively safe stretch of road, in terms of crash statistics. But, you will voluntarily keep your speed under 30 mph. Even if the lack of a guardrail creates a Peltzman effect, ironically increasing safety, the addition of a guardrail would greatly increase travel speed.

Scott Sumner writes:

Thaomas, You asked:

"Presumably it is related to an estimate of what real GDP growth can be with the optimal price level target and an estimate of that optimal target (which in turn depends, I suppose, on just how "sticky" prices and wages are), but what's the model to estimate these parameters?"

No, inflation doesn't matter, it's NGDP growth that matters. I recommend George Selgin's book "Less than Zero" for a detailed explanation. But you could argue that what really matters is NGDP growth per person, or per working age adult.

There are downsides to both very low and very high NGDP growth--I think 4% is a perfectly fine target, but a slightly different figure is also OK.

As far as interest rates, they might well be more stable under NGDP targeting than under the current regime. I see no evidence that the Fed is unwilling to allow large changes in interest rates, if needed to stabilize the economy.

Kevin, Good analogy.

Kenneth Duda writes:

Hi Scott, I'd be curious on your take about what would have happened in the 2007-2010 time period if the Fed policy was unchanged except for this feature, i.e., assuming financial crisis, Bernanke in charge, concerns about inflation leading to an IOR policy, etc.

Below Potential writes:

Is the purpose of this proposal to avoid having to set up a NGDP Futures Market in advance of the Fed switching to NGDP Futures Targeting?

I.e., by eliminating the need for setting up a NGDP Futures Market in advance, you hope to speed up the adoption of NDGP Futures Targeting by the Fed and in exchange for this you are prepared to accept a certain degree of discretion (sacrifice a bit of the ideal of a totally rule-based monetary policy)?

This is a pragmatic approach ... but maybe you don't even have to sacrifice anything:
so far proponents of NGDP Futures Targeting have taken it as given that you have to "create" a market before NGDP Futures Trading can happen.

But what if the Fed gave, say, 100 NGDP Futures featuring 4% NGDP growth (50 of them long positions, the other 50 short positions) to every citizen (theoretically they could send the futures contracts in paper-form via mail).

And then simply let the free market do its magic: people pessimistic about NGDP growth will want to swap their long contracts for short contracts, people expecting NGDP growth to be higher than 4% will want to do the opposite. You would have an NGDP Futures Market instantly. It would start out rather primitive (people swapping futures via mail), but soon financial service providers would realize an opportunity to make money by making NGDP Futures Trading easier for people. After a couple of months one could expect NGDP Futures to be traded at major derivative exchanges across the US.

Long story short: if only market agents had NGDP contracts, a NGDP Futures market would automatically emerge through the process of spontaneous order.

P.S.: Instead of handing out Futures contracts to each US citizen, it would, of course, be easier and more efficient to simply give a bunch of them to certain financial institutions (e.g. to all market agents with access to Open-Market Operations). The principle is the same. They would soon figure out a way to trade these contracts efficiently.

Scott Sumner writes:

Ken, I believe there would have been a period of stagflation, with very low real growth and above 2% inflation, for several years. The policy would have been widely viewed as a failure, because people would not have understood how much worse the alternative would be. Of course we actually experienced the alternative, so we all know how bad 2008-09 turned out to be, but if there had been an optimal 4% NGDP growth policy then people would not have understood just how much worse the alternative would have been.

Below Potential, It is a sort of pragmatic compromise, which also addresses several of the criticisms of NGDP futures targeting that I often encounter.

I would still favor the Fed setting up a NGDP futures market, and subsidizing trading, but now primarily for research purposes.

Thaomas writes:

Yes, I should have said real GDP/capita. If there are downsides to very high NGDP growth, I guess that is because of high inflation, so it seems that inflation does matter; the Fed does have a dual mandate. If so, what's the optimal long term rate of increase in the price level? Or more generally, what are the criteria for setting the NGDP/capita target [taking population growth as exogenous] if not building it up from optimal/maximum GDP/capita dot/GDP/capita * PL dot/PL? Or do we just assume that NGDP growth/capita in the "Great Moderation was optimal?

Gordon writes:

Scott, I know you're playing catch up as you've only recently returned from vacation. But I did want to give you a heads up on one of David Beckworth's recent podcasts. He spoke with James Bullard. Bullard said that he is open to the idea of NGDP targeting which is good. But he also said that currently the Fed targets headline inflation. So it looks like they didn't learn anything at all from 2008.

Below Potential writes:


In my opinion you had already developed the perfect monetary regime (apart from Free Banking, maybe).

At the moment I don't see which "criticisms of NGDP futures targeting" are supposed to be addressed with this approach.

Do you mind writing another post explaining which criticisms of NGDP futures targeting are addressed and how they are addressed with this new approach ?

Maybe that will convince me and/or others.

Thaomas writes:


Ken Duda asks a good question but I think we need to go deeper. What should the Fed have done to maintain a 4% NGDP growth in 2008 and 2009 and 2010 etc?

And if the Fed had had the political independence sufficient to have bought enough LT bonds or foreign exchange or taxed reserves enough to in fact keep NDGP expectations growing at 4% I can't see why we would have had low real GDP growth + inflation at all. [OK it might have taken a few months for people to realize that the Fed was not going to let the economy fall into recession.] What real shock reduced Real GDP growth potential?

Lorenzo from Oz writes:

Your proposal makes central banks much more accountable for their performance. Obviously a good thing public policy wise, and a bad thing institutional self-interest wise.

But you target is agreement among economists, so a selling point.

Scott Sumner writes:

Thaomas, Again, inflation doesn't matter, it's NGDP growth that matters. High rates of NGDP growth lead to excessive taxation of capital income. Low rates lead to labor market inefficiency. I don't know where the happy medium is, but I suspect it's around 4% per year.

Regarding your second question, "level targeting" is the single best thing they could have done in 2008, to prevent the severe recession.

Thanks Gordon, that's on my to-do list.

Below Potential. There were criticism that NGDP futures markets would lack liquidity, or that people would try to manipulate those markets, or that the Fed needed to retain some discretion for unusual situations. My new proposal addresses all three of those objections.

Lorenzo, Good point.


Return to top