Scott Sumner  

Does the Fed need guardrails? Should they want guardrails?

Google vs the EU... Warm Fuzzies May Have Scientif...

Peter Ireland and Michael Belongia have a couple of very good posts on monetary policy, at the "E21" website. Here they discuss the recent Jackson Hole meetings:

By quick count, the foregoing discussion suggests that, as a group, Fed officials wish to direct monetary policy actions simultaneously towards (a) smoothing the volatility of equity markets, (b) preventing bubbles from arising, (c) managing the dollar's foreign exchange value so that U.S. exports do not decline, (d) slowing economic growth to prevent the economy from overheating, and (e) bringing inflation back towards the two percent target. And this list does not even include other Fed statements about achieving a specific value for the unemployment rate before making any decision to tighten the stance of monetary policy.

Even if one accepts, for the sake of argument, that the Fed can reliably influence each of these variables individually, the idea that it can achieve its goals for all of them simultaneously violates one of the fundamental principles any policymaker must confront. More than sixty years ago, Jan Tinbergen, a Dutch economist who shared the first Nobel Prize in Economics, derived this result: The number of goals a policymaker can pursue can be no greater than the number of instruments the policymaker can control. Traditionally, the Fed has been seen as a policy institution that has one instrument - the quantity of reserves it supplies to the banking system. More recently, the Fed may have acquired a second instrument when it received, in 2008, legislative authority to pay interest on those reserves.

Tinbergen's constraint therefore limits the Fed to the pursuit, at most, of two independent objectives. To see the conflict between this constraint and statements made by assorted Fed officials, consider the following alternatives. If the Fed wishes to support U.S. exports by taking actions that reduce the dollar's value, this implies a monetary easing that will increase output in the short run but lead to more inflation in the long run. Monetary ease might help reverse the stock market's recent declines - or simply re-inflate bubbles in the eyes of those who see them. Conversely, if the Fed continues to focus on keeping inflation low, this requires a monetary tightening that will be expected, other things the same, to slow output growth, increase unemployment, and raise the dollar's value with deleterious effects on US exports.

The Tinbergen constraint has led many economists outside the Fed to advocate that the Fed set a path for nominal GDP as its policy objective. Although this is a single variable, the balanced weights it places on output versus prices permit a central bank that targets nominal GDP to achieve modest countercyclical objectives in the short run while ensuring that inflation remains low and stable over longer horizons. But regardless of whether or not they choose this particular alternative, Federal Reserve officials need to face facts: They cannot possibly achieve all of the goals that, in their public statements, they have set for themselves.

I think this is right, and indeed I'd go a bit further. I don't regard reserve control and interest on reserves as being all that different (although technically they are correct.) One can think of IOR as a policy instrument that impacts the demand for reserves (and hence the demand for the monetary base.) A lower IOR leads to less base demand, and hence "excess cash balances." This sets in motion exactly the same "hot potato effect" as more supply of base money, which also leads to excess cash balances. Yes, the two policies have slightly different implications for the banking system, but at the macro level I don't think they provide nearly as diverse a set of tools as say fiscal and monetary policy. Perhaps Belongia and Ireland would agree, as they end up advocating a single target---NGDP---just to play it safe.

In an earlier post Belongia and Ireland cast a skeptical eye on forward guidance, as currently practiced by the Fed:

If forward guidance is, in fact, an important influence on spending and production, the information in these two tables appears to reveal little that would be useful for consumers or producers. Although many observers interpret the stance of monetary policy using some variant of the Taylor rule, most members of the FOMC express options of the appropriate level for the funds rate that, given their own forecasts for output and inflation, deviate considerably from the values generated by that rule. Moreover, public statements by FOMC members differ on two important points: Whether the funds rate target should be increased or left unchanged and, if that target is to be raised, when the increase should occur.
Although I continue to believe that the forward guidance offered in late 2012 was helpful, Belongia and Ireland are probably right that the overall policy isn't doing what it is supposed to do.

One alternative is level targeting, which gives business people and investors a much clearer idea of the future path of monetary policy. But the Fed opposes level targeting, as they (rightly) believe it would sharply curtail their discretion. Here I'd like to suggest a compromise, a sort of "guardrails" approach to level targeting.

Suppose we are back in 2007 and early 2008, when the Fed saw an unstable economy, but was equally worried about recession and higher inflation. Their central forecast is for continued 5% NGDP growth as far as the eye can see, but they want the discretion to adjust to things like a change in trend RGDP growth (which of course seems to have slowed after 2007.) Locking into 5% trend NGDP growth is too risky in their view, as it could lead to above target inflation. On the other hand the Fed would certainly like to prevent the sort of steep drop in NGDP, and high unemployment, which actually occurred in 2008-09.

My compromise would be for the Fed to set "guardrails" at a band around 5%, say 4% and 6%, or 3% and 7%. These band lines might extend out 3 to 5 years, at which time the Fed would re-evaluate the trend, based on new information about trend RGDP growth in the US. The idea is that the lower bound (let's say 3%) would be a floor on NGDP growth, and the Fed would commit to, at a minimum, returning to that trend line if growth fell below 3%. And ditto for a overshoot of 7%. That doesn't mean they commit to return exactly to the trend lines, rather they would commit to do at least that much stabilization. Obviously they'd be free to do even more, including going back to the 5% trend line.

At one level this compromise might seem pointless. If the Fed doesn't want to have its hands tied, why would the guardrails approach be any better than a single NGDPLT trend line of 4% or 5%? The answer is that while the Fed doesn't want its hands tied, it also genuinely doesn't like wild swings in NGDP growth. Recall that these swings make its job much harder, and put it in the spotlight as it adopts emergency policies like QE to deal with the undershoot. (Or conversely the Fed would need very high and unpopular interest rates to deal with an overshoot of 7%.) It would prefer to avoid these extremes, which guardrails help them to do.

My claim is that the Fed itself sees, or should see, a tradeoff. Yes, it wants discretion, but it also wants success. There is some band so wide that the Fed would view movements outside that band as unacceptably large. I claim that 2008-09 was one of those unacceptable movements. But because they didn't already have a guardrail regime in place, they had trouble communicating a policy that would get us back into the acceptable range. That communication would have had to use their inflation targeting language (Paul Krugman's 4% inflation, for example), or perhaps would have required an amount of QE that was politically unacceptable. With the 3% and 7% guardrails they could promise to "do whatever it takes" without seeming to violate previous commitments.

I do have a hidden agenda here. I believe that over time they'd become more comfortable with this policy approach, and the guardrails would gradually narrow. And as NGDP growth, not inflation, became better understood as "the real thing", the Fed would become more and more comfortable with keeping its NGDP target stable, even as trend RGDP growth (and hence inflation) fluctuated. Or perhaps they'd only adjust the NGDP target for labor force changes, which would move us closer to George Selgin's productivity norm--a policy approach that's probably superior to simple NGDP targeting.

PS. With the recent slowdown in trend RGDP growth, I believe that today the Fed would choose 4% as the center of the band, which might be set at 2.5% and 5.5%

Comments and Sharing


COMMENTS (14 to date)
Michael Byrnes writes:

How would the level targeting come into play here - what trend line is the Fed committed to returning to in the event? Over a few years, 3% and 7% annualized NGDP growth would diverge quite a bit.

Benoit Essiambre writes:

Like Michael, I find the details of your proposal unclear. I don't understand how levels fit into this, I assume it can't be based on diverging level bounds.

Would it mean targeting inflation except when current GDP relative to a point in the past is out of bounds? Or would it mean targeting an NGDP path but with a wide margin relative to near future or a near past levels?

I remember reading a Bank of Canada paper where the reason they gave for not adopting PLT was the possibility of having to adopt too tight a stance if prices ever overshot the path.

But it seems to me it would have been easy to avoid this problem by using a level floor, or simply adding language to the effect that they keep discretion to modify the path under exceptional circumstances to avoid large price or unemployment gyrations.

It does seem like some kind of gradual transition to better or higher targets will be needed. Changing suddenly may be too destabilizing and probably too scary for central bankers.

Scott Sumner writes:

Michael and Benoit, Yes, I should have been more clear. I envisioned an NGDP target, not inflation targeting. But obviously this idea could be applied to inflation as well.

As far as the gradual divergence of the two trend lines, I foresee resetting the target path periodically to keep the trend lines reasonably close together.

Here's an example. A the beginning of the year you set a 3% to 7% range. Now assume that over 12 months NGDP has grown by only 2%, so it's 1% below the bottom of the range. For the next year you set another 3% to 7% range, but starting from a point one percent above current NGDP. Thus in practice NGDP would have to grow by 4% to 8% in the second year. That's the point of level targeting---to allow for "catch up" or "fallback."

Michael Byrnes writes:

Ah, now that makes sense. And even 3% (or 2.5%) or whatever will prevent a recession or, failing that, demand a robust monetary response to one.

Peter Gerdes writes:

I think one should ask whether it would be good for the fed to publicly endorse that their overriding goal is ensuring NGDP remains between the bounds. Sure, I suspect at a theoretical level it would be beneficial for the markets to know this. However, I worry that publicly commiting to such a policy risks both reducing the fed's ability to change course when markets don't seem to be reacting as expected or to reevaluate the wisdom of ngdp targeting itself.

Seems to me the more sensible thing to do would be for the fed to continue to publicly endorse a long list of incompatible policy goals while trying out ngdp targeting informally. This lets them back out of the choice if they later reconsider and they may never want to publicly commit to one overriding policy goal even if it's just keeping ngdp between two bounds.

Politically a public guiding rule makes it much harder for the fed to defend against attacks like: the fed is favoring the interests of (insert disliked/privileged group) over (insert sympathetic group) not to mention revoking that principle.

Scott Sumner writes:

Michael, Yes, that's right.

Peter, In the end, I believe the least controversial path for the Fed is success. The Fed was not very controversial during the "Great Moderation" (1984-2007) and became very controversial when NGDP plunged.

I would add that some of your concerns also apply to inflation targeting, the current policy.

As far as not being certain NGDP targeting is the right option, I think that's a good point. That's why you approach it gradually, starting with a fairly wide band to allow some discretion.

ThomasH writes:

Interesting that the 5 "objectives" listed did not include what supposedly is their objective, to keep the price level trend on a constant track. If you have an inflation target it implies you take undershooting as seriously as overshooting. The financial press has been very negligent in reporting this failure. How are people supposed to make long run plans if the future price level is what ever results from inflation wandering between 0% and 2%?

You really need to address as well the political obstacles to your proposals. There is a powerful group that wants tight money, Some of it is just that, most, I think, has been has been wanting tight money to prevent a decent recovery during the administration of a Democrat.

Rajat writes:

The width of the guardrails seems to still leave open a lot of uncertainty. Ex ante, the market wouldn't know whether it would be facing a 5-year path of NGDP rising at 4% pa or 6% pa. Maybe that's no worse than now... But under a simple rate of change target, the market might at least expect that if NGDP were 4% in one year, the central bank would take action to lift NGDP in the following year, if it was doing its job properly of course.

Speaking of which, I have a more general question on level targeting. How does one make it credible? All of the Fed, the ECB and the BoJ have been undershooting for 3 years, the market is predicting they will keep undershooting for the next 5-10 years, and yet... nothing happens. Under LT, they would be facing rising accumulated price level shortfalls. But apart from looking a little more embarrassing, why would they behave differently? What would be the sanction?

Rajat writes:

Thinking further, I guess you might say that if NGDP were 4% in the first year, the central bank would still have a 5% central target in the following year and so would have an obligation to loosen policy even though NGDP didn't undershoot the lower guardrail in the first year. But I wonder if the very existence of the guardrails would make the central bank inclined to see the target as 4-6% pa rather than 5% pa.

bill writes:

Perhaps the guardrails could work like this? In lieu of say a 4% NGDPLT, year 1 would have bumpers at 2% and 6%. And then the two bumpers would each grow at 4% after that. So (ignoring compounding), year 2 target would be 6% and 10%, end of year 3 would be 10% and 14% cumulative, etc.

Jose Romeu Robazzi writes:

If this was not such a serious matter, I would think that the prevalent strong belief on central bankers discretion was funny. Despite all recent failures, the academic community still believe that discretion is needed. I think that we should give other paradigm a shot. Your bands idea, Prof. Sumner, is good only if it makes a strict nominal targeting process feasible, otherwise it is a compromise that can serve only as an excuse to dismis the model later ...

B Cole writes:

Well, perhaps the Fed does have multiple objectives, such as exchange rates or financial stability.

But at Jackson Hole there was only one topic: inflation.

All six Jackson Hole panel discussions were on inflation. In a word count of Stanley Fischer's presentation, it was something like 70 for "inflation" and one for "aggregate demand."

Exchange rates? The dollar has been soaring and no action from the Fed. Financial stability? Where was the Fed when the stock market collapsed in 2008?

I contend the Fed has one object: to keep inflation about at 1% , or less if possible

Scott Sumner writes:

Thomas, I don't think it's political with Yellen, Fischer and the rest, I think it's simply the wrong regime.

Rajat, Actually, failure would be far more embarrassing. Suppose that because of persistent undershoots the BOJ was aiming for 8% inflation. And suppose it came in at 0%. That would be a huge embarrassment. Actually, I don't think they'd ever get in that situation under level targeting. If they always missed by 1% low, that would mean the rate of change was exactly on target after the first year.

Obviously I'd prefer guardrails as narrow as possible, at least narrow enough to prevent another 2008-09.

Bill, That's a very good idea, I may use that.

Tobyw writes:

[Comment removed pending confirmation of email address. Email the to request restoring this comment. A valid email address is required to post comments on EconLog and EconTalk.--Econlib Ed.]

Comments for this entry have been closed
Return to top