Thursday, September 06, 2012

Fed announcements as "cheap talk"

As any semi-loyal KPC reader knows, I find the "just do your job" critique of the Fed baffling. In our current institutional framework, the Fed cannot credibly commit to future actions that will conflict with their period by period utility function.

That is to say (more or less accurately), the Fed cannot credibly promise to tolerate higher inflation than it prefers after the economy recovers because, when the economy recovers the Fed will still not like inflation and there is nothing to prevent them from not tolerating it. Knowing this, people will not believe the initial announcement.

So, why does the Fed make announcements about the future at all? Could they ever "work".

I believe the relevant economic theory here is the literature on "cheap talk" in games.  A readable approach can be found in Ferrell & Rabin.

For cheap talk to be effective, Ferrell & Rabin argue that it must be self-signalling meaning that the sender only wants to send the message if it is true (or if it is at least correlated with the truth). They also argue that it must be self-committing, meaning if the receiver believes the message, the sender has incentives to fulfill it.

I think that the messages people want the Fed to send are not self-committing, so that such "cheap talk" won't work.

However, there is a famous paper by Jeremy Stein on cheap talk and the Fed that argues that the Fed can make imprecise announcements which will have some effect on expectations. I haven't fully figured that paper out yet.


James Oswald said...

The task of market monetarists is to alter the preference itself, both with the Fed and with Congress. If you believe higher nominal growth would lower unemployment and expand real growth, there is no time consistency problem at all. The Fed raises NGDP now, people love it, and tomorrow they simply maintain the higher path. If you think higher nominal grow will be horrible politically, then yes, it will probably be reversed. The argument then doesn't come down to time consistency at all, but what level of inflation/NGDP growth should be preferred.

Whether the Fed could cause higher NGDP is not a concern for market monetarists. If you want to argue that the Fed could buy up all outstanding Treasury debt and abolish interest on excess reserves, and NGDP wouldn't budge one inch, then explicitly make that argument. And honestly, if that were reality, it wouldn't be so bad. Market monetarists don't focus on the interest rate channel. If you argue that nominal interest rates won't drop, that won't convince them that NGDP won't change.

Jeff said...

James nails it. The most persuasive critic of the fed's actions is coming from the market monetarists who want to change the preferences of the fed from that of 2% inflation targeting to 5% NGDP targeting.

Others, such as Cochrane have argued that if the Fed changes its policy today what is stopping it from changing tomorrow. This is much different from a time inconsistency problem and personally I don't it all that convincing. I'm sure Fed officials truly believe what they are doing what is best and the task is to convince them that NGPD targeting is better. If later, someone discovers and even better policy, then yes, the Fed should change their policy once again and I don't see any reason why this should destroy their credibility.

John Thacker said...

The Fed's promises certainly seem to have "worked" in the sense of persuading the market in the short run, on many occasions.

Whether or not the theory is that people won't believe the initial announcement, the evidence seems to be that people do believe it.

Phil Arena said...
This comment has been removed by the author.
Unknown said...

Very informative article. I've found your post via Yahoo and I?m really glad about the information you provide in your posts. Thank You for sharing this very informative article...

ewald struggl

Angus said...

JT: two things.

First the Stein paper seems to say imprecise cheap talk actually can be partly effective. The problem I'm having embracing that result is that the problem in his model is not exactly the same as the problem we are discussing.

Second, most studies that show Fed announcements "work" are looking at very short term movements in stock prices. That's a far cry from saying they have the macroeconomic consequences their proponents claim.

If QE worked the way it was supposed to according to the "create inflation expectations" people, rates should have gone up, not down.

kebko said...

Angus, during QE1 & QE2, rates went up.

James Oswald said...

It's better to look at the TIPS spread, because nominal rates can go either way. On one hand, the Fed pushes them down by buying bonds, but on the other hand the Fisher effect pushes rates up due to expected inflation. But like I said before, market monetarists don't pay attention to the interest rate channel.

Wonks Anonymous said...

Karl Smith gives a layman-friendly explanation of effective cheap talk.

I second James that the Fed's utility function cannot be assumed known and set in stone.