« Enough is Not Enough | Main | Paul Krugman: The Big Meltdown »

Thursday, March 01, 2007

Explicit Inflation Targeting as a Commitment Device

Given the large amount of discussion recently about whether inflation targeting by central banks hurts employment, I want to talk about why a government might want to announce an explicit inflation target. This is not the only reason why a central bank might want to commit to an inflation target, but it is one of the reasons.

[Responses by James Galbraith and Dean Baker follow the main post.]

One problem that explicit inflation targets can help to overcome is know as time inconsistency. A time inconsistent policy is one that, though it is optimal at the time the plan is made, once the time comes to actually execute the plan the individual wants to make another choice. Wikipedia gives one example:

Students, the night before an exam, often wish that the exam could be put off for one more day. If asked on that night, such students might agree to commit to paying, say, $10 on the day of the exam for it to be held the next day. Months before the exam is held, however, students generally do not care much about having the exam put off for one day. And, in fact, if the students were made the same offer at the beginning of the term, that is, they could have the exam put off for one day by committing during registration to pay $10 on the day of the exam, they probably would reject that offer. It is the exact same choice, just made a different points in time, and because the outcome would change depending on the point in time, the students would exhibit time inconsistency.

For another example, consider the marriage market and the decision of whether to get married at time t. How do you know that later, at time t+i, someone better won't come along making you regret your original choice? And if that's a possibility on both sides, why get married? The problem is one of commitment to a plan - that you can trust that once the decision to get married has been made, your potential spouse won't reoptimize at a later date if a better alternative comes into the picture. Those pieces of paper you sign, ceremonies, rings, wedding gifts, etc. are partly about enforcing this commitment. [To push the analogy a little further, if the agreement is broken later it's possible for everyone to be better off, but it's also possible the reoptimization will only satisfy a narrow set of interests (and hence the reason it is pursued). When the interests of all affected parties are considered, the outcome can be inferior to the outcome under commitment. Another consideration in comparing the outcomes under discretion and commitment is that without commitment mechanisms, there would be less willingness to get married in the first place leading to a reduction in potentially beneficial matches.]

Central banks face the same problem, the absence of commitment mechanisms can cause the economy can end up in a sub-optimal outcome. If the central bank promises to deliver low inflation today, and private agents believe them, then at some point in the future, with the belief in a low inflation policy still in place, it will be optimal for the central bank to deviate from its announced plan, hit the economy with surprise inflation and increase output and employment. So long as the value of the increase in output exceeds the cost of the inflation surprise, the bank will have an incentive to inflate.

Because private sector agents understand this incentive, they will not expect low inflation, they will expect the Fed to pursue high inflation as described above. With high inflation expected, it's optimal for the central bank to deliver high inflation with the result that the economy experiences higher inflation but, because the inflation is fully anticipated by the public, there is no output gain to show for it.

In a model with price and wage rigidities, the economy will be worse off with higher inflation than lower inflation (because the sticky prices cause resource misallocations, and this is worse when other prices are changing faster), so the high inflation outcome is not the optimal policy.

So how might we get the Fed to pursue the optimal policy and the public to believe the Fed won't deviate from its announced intentions? There is a rich literature on this topic (e.g. see Walsh chapter 8), but I will only touch on one aspect.

Suppose you want to quit smoking or go on a diet. Why do you announce your intentions to other people (at least some of you anyway) instead of keeping your plans to yourself? It's a commitment device. If you tell people you are quitting smoking, and then get caught smoking later (when you've decided another plan is optimal), it's embarrassing. Announcing your plans is an attempt to stop yourself from changing your mind later. Telling your friends you are going on a diet and making a really big deal out of it only to be seen scarfing candy bars tomorrow undermines your credibility.

The Fed can use the same device. If it doesn't tell us what its target inflation rate is, then it's easy to reset the target later and give into impulses of the moment. As an analogy, suppose you tell people you are going on a diet, but don't announce your target weight. Say it's 160 initially but only you know that. But as time goes on, sticking to the plan gets harder and harder, and when you get to 175 you can announce, joyfully, that you've hit your target weight without suffering any credibility loss since nobody knows what your target is. It's not the optimum weight, but it's good enough at the time.

Greenspan was good at this before Congress. Ex-post, no matter what the Fed did, Greenspan would explain why it was optimal for the Fed to have proceeded as it did. But without knowing the Fed's plans a priori, there was no way to judge whether the Fed was, in fact, hitting its target rate of inflation, or any of its other goals.

Why do we want to impose time-consistency on the Fed through a commitment to an announced target? There is a loss of discretion in doing so but it is precisely that loss of discretion (the inability to choose to smoke or eat junk food) that leads to a superior outcome in this model (commitment isn't always preferred to discretion, it depends on model specifics). The benefit is that the optimal policy is maintained. The cost is, as Greenspan always emphasized, loss of flexibility and discretion.

Another reason for pre-commitment to an inflation target, and Barney Frank has been providing us with a good example of this, is that political pressure can cause central banks to deviate from their optimal plan (oh, go ahead, have a piece of cake). With pre-commitment, such temptations are easier to resist.

Before going on, I should note that I want what Barney Frank and everyone else wants, high, stable employment and high-paying jobs. The debate is how best to get there, and I have disagreements on this with many who have been writing on this topic recently.

Okay, now let's turn to this commentary by James Galbraith. I should note that my comments will refer to the entire set of recent commentary along these lines, the comments are not directed at James in particular:

Bernanke throws in the towel?, by James K Galbraith, Comment is Free: In a remarkable article in the Wall Street Journal on Monday, writer Greg Ip reports that the Federal Reserve has shifted the focus of its concerns with inflation. The shift is from actual inflation - the movement of prices as measured by statisticians - to "expected inflation", which is akin to a forecast estimated by economists.

The logic is that actual inflation is quite volatile. An increase this month could be followed by a decline the next. But expected inflation is quite stable: a chart showing the past decade has it barely budging from three percent. So the shift helps keep the Fed from chasing rabbits up and down their rabbit holes by reducing the pressure to react to the news. The Fed would only react, so the new story goes, if rising prices started to affect expectations of future inflation.

That is not the only, and perhaps not even the most important argument for targeting expected inflation. Monetary policy has considerable lags before its effects are fully felt - as long as a year and a half after policy is implemented. What matters is inflation that is expected, say, six to eighteen months ahead, not the current inflation rate, it's too late to do anything about the current inflation rate (after a policy change, there's a delay of one or two quarters before inflation moves at all). The inflation rate to use in the policy rule is also an area of theoretical debate, and there are questions about how the equilibrium properties of these models vary when you use lagged, current, or expected inflation in the monetary policy rule. The point is that the reason for targeting expected inflation is not just because it's more stable than actual inflation, it's because of the delay in the effects of policy and because of theoretical considerations (in any case, the Fed uses the trimmed mean PCE which solves some of the instability problem).

James goes on:

But there's a catch. How do we know if any particular change in prices is being incorporated in inflation expectations? By the Fed's own methods -so far as we understand them - we don't know that at all. Over any short period, there is no independent basis for ever thinking that changes in inflation expectations have occurred. If actual inflation goes up, in other words, the rational forecast is simply that it will come down again.

It's rational to think inflation will come down again only under inflation targeting. If the Fed varies its rule as many are pushing it to do and quits worrying about the inflation rate as much, why would expectations remain anchored? The argument in the commentary is that inflation expectations do not react to current inflation, but that is becasue of the Fed's commitment to a low inflation target.  Without that commitment, expectations would behave differently.

How does this relate to the Fed's monetary policy rule?

Most economists think the Federal Reserve has a reaction function with two influences: the deviation of inflation (or, now, expected inflation) from its target, and the deviation of unemployment from its target. In an spat last week, House Financial Services Committee Chairman Barney Frank took Bernanke to task for placing too little weight on unemployment. Bernanke held his ground: The Fed, he said, would continue to focus mainly on inflation.

But now, in effect, Bernanke has conceded to Frank. For if the practical impact of inflation news on monetary policy falls to zero, the Fed will react only to deviations of unemployment from its target. And so long as unemployment is higher than the full employment rate, the Fed should be fostering credit expansion - usually with low interest rates. If the Fed doesn't think we're above full employment now, Frank should be asking why. After all, we had much fuller employment just seven years ago, without inflation.

The news was that deviations of unemployment from target have a smaller impact of future inflation in the past. But the news did not say that an increase in the money growth rate no longer causes inflation or that inflation is no longer costly to the economy. I just don't see why the practical impact of inflation news on Fed policy will now be zero as claimed because of this new information (somewhat new anyway, e.g. the Stock and Watson paper on the change in the relationship between inflation and unemployment is a couple of years old now and was given at a Fed conference on this topic, so the Fed has been aware of this and incorporating the results into policy for some time). The Fed will keep reacting to inflation news to keep expectations anchored.

Galbraith goes on to reiterate Barney Frank's call for the Fed to worry more about unemployment and less about inflation:

Frank's victory is good news: it represents a real triumph - perhaps too late - for my recommendation that Bernanke "find a creative excuse to do nothing at all". In a world in which globalization conquered inflation more than twenty years ago, perhaps it's slowly dawning that the inflation-mongering we still get from the Fed is just so much hot air. We would be much better off if they focused on producing jobs rather than keeping a bogey in a bottle to scare children with. (The example is from the European Central Bank, but the point remains.)

Again, inflation is low because of the Fed's policy rule. If anyone doubts that the rule the central bank follows matters, see Zimbabwe. Or see this criticism of Japan's monetary policy from Anil Kashyap. I'll emphasize again that the Fed's focus is on producing jobs. Remember how people reacted when the Fed refused to stop the tech bubble or the housing bubble? In these instances, the Fed was criticized for not stabilizing asset markets by some of the same people who are now complaining that they are stabilizing asset markets.

There is the view that pursuing inflation stability is giving into the wealthy over workers, that it is as an abandonment of high employment goal. I disagree. The goal of stable overall economy and employment is a primary focus of Fed concern.


The Fed's shift - if it's for real - hasn't been fully explained to the people who write testimony for the chairman. ... Perhaps he should have dumped the inflation scare-talk months ago. Perhaps he should have focused then on the actual dangers facing the economy. These include a housing collapse leading to a manufacturing recession, and now a downward revision in GDP growth. Those dangers don't include surging inflation.

We shall see, soon enough, if worse is yet to come.

On the housing collapse, didn't the housing bubble come, at least in part, from low interest rate policy attempting to spur employment and the economy just as those who are concerned over employment would want (there was and is a serious debate over whether the Fed took interest rates too low for too long)? Is the argument that the Fed was too concerned about employment and, because of that, allowed a housing bubble to occur? It can be argued that the Fed allowed a bubble to occur to protect employment, and that it is now doing its best to slowly rebalance the economy (move resources out of housing and into other sectors)  through slow, deliberate, well-publicized and cautious tightening. I think that's the right thing to do. The air needed to be let out of the housing bubble as soon as we could do so (without causing a big crash) to avoid potentially bigger problems later. So far, so good, but there are certainly signs of problems ahead to caution is required.

The Fed cannot cure all ills in the economy, I think we sometimes expect far too much. When unemployment goes up, or inflation kicks up, that is not necessarily a sign the Fed has made big policy mistakes. Economies are volatile and despite the best efforts of policymakers, recessions and booms will occur. The question is whether the Fed can do even better.

As I look back through time, I see an economy that became considerably more stable, on the order of a 50% reduction in volatility, and inflation became considerably more anchored with the adoption of an interest rate (inflation targeting) rule in the early 80s. It may be that the Great Moderation since the 1980s is due to other factors and the change in monetary policy had little to do with it. But I cannot, in good conscience, recommend taking the chance of returning to the kinds of policy we had in the 1960s and 1970s when inflation did not play the same role in Fed policy that it plays today. "It's just a little inflation, and think about the unemployed" is hard to resist in the short-run, but a recipe for a time-inconsistent disaster over the longer haul.

If I thought the Fed would do better by relaxing its weight on the inflation term in its reaction function, I would push hard for such a change. You would get tired of hearing about it. But there's good reason on both the theoretical and empirical fronts to believe that doing so would constitute a return to pre-1980s policy and cause a deterioration in the economy's performance.

Update: In comments, James Galbraith responds:

Mark has correctly intuited that my comment was inspired by, and partly directed at, his post of a few days back, on Barney Frank and the reaction function.

Mark's response rests, though, on a strongly maintained hypothesis: that the Fed does, in fact, control inflation through its control of... well, something. Monetarists used to say "money," but no one says that with a straight face anymore. So we have an amorphous doctrine according to which the Fed controls inflation by controlling inflation expectations, which they control by (exactly how?) reacting to inflation. Mark says: "Inflation is low because of the Fed's policy rule."

There are a number of difficulties with this.

The first is that we have no serious theory on the formation of inflation expectations. By a serious theory, I mean something that could actually predict, in advance, a change in those expectations, or that could measure them if they occurred, independent of some long-running past movement in actual inflation. Mark imagines that the Fed's reaction to actual inflation controls inflation expectations, but there is no way to distinguish that view from an alternative, which says that inflation expectations are simply highly inertial.

A second problem is a collective action problem: even if the Fed has a firm inflation target, and everyone believes it, it is not rational for any particular economic agent, who has the opportunity to violate the rule by demanding and receiving a higher price or a higher wage, to forego that certain gain just because the Fed may react. This story is usually confused with talk of a "representative agent," but the real economy doesn't consist of representative agents. No oil company ever forewent a rise in the gas price because they feared it would be followed by a rise in the interest rate.

A third problem is that the maintained hypothesis itself -- that the Fed's policy rule has material effect on inflation (except in the extreme case when the Fed pushes the economy into an actual recession, triggering mass unemployment) -- has no foundation in evidence. It is a pure assertion, an article of faith.

I believe, on the contrary, that the decisive turn against inflation came with the rise of the dollar in the early part of the 1980s, the crushing of industrial unionism in the American Midwest, and the opening of our economy to an almost perfectly elastic supply of imported wage goods. This is why I said -- in a phrase Mark quotes but passes over -- that "globalization" conquered inflation.

I'm not saying monetary policy had nothing to do with this: of course it did. I'm saying that it was a one-and-for-all affair, changing the underlying structure of the supply curve. Therefore the notion that the economy rests on some fragile knife-edge, always ready to slip over into rising inflation if the Fed becomes incautious, is not only wrong but pathological. In the JEP in 1997 -- TEN years ago and BEFORE the economy actually reached full employment without inflation -- I called it a "self-inflicted wound."

The belief that inflation was entirely out of the system led me to predict, along with almost no one else, that one could have full employment -- "chock-full employment" as Bill Vickrey used to call it -- without inflation. In the late 1990s, I got a test of this proposition, thanks to Alan Greenspan. It was proven correct. The Fed acted in a way that virtually the whole economics profession believed would lead to higher inflation. But it didn't.

Still, the economics profession, broadly speaking, remains unwilling to face the lesson of the late 1990s, and to abandon the legacy of Friedman and Phelps on which the present generation of macroeconomists was raised and weaned.

(Economists seem to think that because the Phillips Curve was wrong--which it was--that the Friedman-Phelps natural rate revision must have been right. This is illogical. There are other possibilities. One is that the macro-economists I most admired -- Nicky Kaldor, Robert Eisner, and Bill Vickrey -- were right in distrusting the Phillips Curve from the very beginning. Events, let me suggest, have borne them out handsomely.)

As a result of failing to face the evidence, the Fed misses every single downturn, including the one now getting underway. Alan Greenspan missed every one, but he eventually got quite good at reacting to them pretty fast -- with a truly major achievement along those lines following September 11.

Will Bernanke have the nerve, the wit, and the legs to do the same? It's an open question, but his allegiance to the anti-Keynesian shibboleths and inflation terrors of the late 1960s is distinctly worrisome. We could all help him, I think, by thinking through this problem a bit more clearly.


Dean Baker adds:

As one of those who was perhaps intended to be implicated by Mark's comment that the Greenspan/Bernanke critics wanted them to both pursue high employment and attack the bubbles, let me point out that I don't see any contradition.

I have argued that the Fed could have burst both bubbles by using the Fed's platform (along with the accompanying arithmetic) to show that both the stock bubble and housing bubble were not sustainable. I know that economists like to ridicule the idea that this could have made any difference, but that position is based solely on their prejudices, not on any evidence.

For example, it was easy to show back in the late nineties, when PEs crossed 30, that unless investors had radically different expectations of future growth than most economic forecasters, they would have been willing to accept real returns of around 3 percent in holding stock. I am willing to bet that if Greenspan had repeated made this point in his public testimonies and speeches, then the bubble would not have persisted (or better yet, if he started when PEs crossed 20).

Maybe fund managers and otehr investors would have completely ignored the Fed chair, but I doubt it. It would make an interesting law suit, if the market subsequently plunged, and mutual fund shareholders sued fund managers for negligence. Of course, the managers could have no coherent response to the hypothetical Greenspan statements, so would they say that they don't pay any attention to the Fed chair? Watch fund managers lose their life savings, their homes and all their other property.

Anyhow, none of us know what would have happened if Greenspan had engaged in systematic attacks on the bubble. I will note that Greenspan and Bernanke certainly act as though they words make a difference. Until a Fed chair tries this sort of bubble attack, we are all just speculating on the market response.

    Posted by on Thursday, March 1, 2007 at 01:22 PM in Economics, Monetary Policy | Permalink  TrackBack (0)  Comments (27)


    TrackBack URL for this entry:

    Listed below are links to weblogs that reference Explicit Inflation Targeting as a Commitment Device:


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