The Fed versus the ECB: Learning Models and Monetary Policy

By Mark Thoma

I want to continue the conversation started here and here on differences between the U.S. Federal Reserve (Fed) and the European Central Bank's (ECB) approaches to monetary policy.  The difference I want to talk about today is more speculative than the others, but it’s not one I’ve seen mentioned much elsewhere so perhaps it will be a useful issue to raise in any case.  I should warn you that the post is somewhat long and may presume some, but hopefully not too much, knowledge of terminology.

Suppose we adopt the view, as I think both central banks have, that neither private sector agents nor central bank officials have perfect information on either the structure of the economy or on the particular parameters essential to understanding how the economy operates.  Thus, acording to this view, there are two sets of agents that must learn about the economy.  Private sector agents must learn about the structure of the economy and about the policy rule, and they must understand how the two interact.  Central bank officials face a similar problem except it is assumed they know the policy rule that is in place.  Both the Fed and the ECB recognize that learning must take place, and both understand that learning is a problem faced by both central bank officials and private sector agents.  For example, here are comments recognizing each type of learning from Jean-Claude Trichet, President of the European Central Bank (ECB) made recently at the Fed Symposium in Jackson Hole:

Because the economy is never at rest, agents have to catch up with the continuous change in their environment by an ongoing process of learning. When shocks are moderate, or the underlying evolution of the economic structure proceeds at a slow pace, imperfect information and learning do not excessively complicate our interactions with the private sector. But there are times in which stormy perturbations and accelerated structural change make uncertainty more acute. These are times in which a perpetual process of learning on the part of economic agents can have implications for the overall stability of the economic system – to some extent, independently of the monetary policy regime that is in place. If agents do not possess rational expectations, but have to re-estimate continuously the coefficients of an unknown model of the economy, using rolling windows of new observations, it can well happen that a shock of sufficiently serious magnitude can unsettle expectations, even under credible monetary institutions. ... Long-term expectations ... may over-react to the shock. They may drift endogenously reflecting the impact that the unprecedented disturbance exerts on agents’ own re-assessment of the key structural features of the economy. … The difficulty lies in devising a prudent way to factor such situations into policy.

This statement clearly recognizes the role of private sector agent learning in the formulation of monetary policy, particularly in the face of large disturbances.  The ECB also recognizes a second type of uncertainty, model uncertainty, as expressed in the following statement:

Clearly, output-gap measurement issues are not a fact of the past. Central banks also need to cope with this problem at present. I take this as a warning that the central bank should not rely on any simple indicator of economic slack in taking its policy decisions. … We both do not want to rely exclusively on a single particular model of the economy, as sophisticated as it may be. I have myself said several times that the Governing Council of the ECB has no intention of being the “prisoner” of a single system of equations...

Trichet implies the Fed shares this view and Alan Greenspan, in his opening and closing remarks at the Symposium, confirms that the Fed also views model uncertainty as important.  Here’s a quote from the opening remarks:

Despite extensive efforts to capture and quantify what we perceive as the key macroeconomic relationships, our knowledge about many critical linkages is far from complete and, in all likelihood, will remain so. Every model, no matter how detailed or how well conceived, designed, and implemented, is a vastly simplified representation of the world ... Given our inevitably incomplete knowledge about key structural aspects of an ever-changing economy and the sometimes asymmetric costs or benefits of particular outcomes, the paradigm on which we have settled has come to involve … risk management. ... The risk-management approach has gained greater traction as a consequence of the step-up in globalization and the technological changes of the 1990s, which found us adjusting to events without the comfort of relevant history to guide us. ... In effect, we strive to construct a spectrum of forecasts from which, at least conceptually, specific policy action is determined through the tradeoffs implied by a loss-function…

The U.S. central bank is definitely aware of models of both types of learning.  For example, the presence of people such as Athanasios Orphanides at the Federal Reserve Board and John Williams at the San Francisco Fed make it hard to argue persuasively that these ideas are not present in policy arenas.  They are.

But as I observe the two central banks, I see a difference in who is presumed to be doing the learning and in how much learning there is to do.  The U.S. central bank seems far more concerned with the lack of knowledge that policymakers have about the structure of the economy, and less concerned with private sector learning, than the ECB.  If this is true, then the question is why.  Trichet offers one explanation in his comments at the Symposium:

…we have to fully accept that our economies, as well as the global economy, are experiencing a period of very rapid structural changes. I would fully subscribe to Alan’s remark that “the economic world … is best described by a structure whose parameters are continuously changing”. It is even truer in Europe. Because in addition to the impact of science and technology, in addition to the effect of globalisation, in addition to the very rapid demographic changes that characterise the industrial world, we also have to take into account the structural transformations that Europe has boldly marshalled for itself: the single market, the single currency, the historic enlargement. But the lesson is valid for all. Robustness in the present situation is key and we should not depend on a single model…

According to this view, because change is more dramatic in Europe, learning is more important and the chance that expectations will lose their anchor when there are large disturbances is much larger in Europe than in the U.S.  This would translate into a less activist role for the ECB, except in extraordinary times when extraordinary measures would be required to keep expectations from drifting from fundamentals (but see Edward’s comments to the previous post).

I found three things interesting here.  First, that there does appear to be a difference in who is presumed to be doing the learning, with the ECB appearing to place more emphasis on private sector learning than the Fed.  I am, however, fully willing to concede that this perception may be due to my own lack of knowledge and would be pleased to hear other views on this.  But I have never heard Greenspan refer to rolling regressions.  Second, assuming there is a difference the importance each bank attaches to each type of learning, and a difference in the importance of learning generally, why do such differences arise and how do they affect policy? I had presumed the difference in activism between the two banks was due to a difference in philosophy regarding activism, but the suggestion here is that it is due to fundamental differences in the degree of structural change present in the two economies.  There is more to learn about the economy in Europe.  This difference in structural change makes the private sector learning problem more important for the ECB and leads to a more conservative, less activist view for policy.  Finally, I believe the role of model uncertainty and learning and its effect on policy does not receive enough attention from analysts, though the idea of risk management does seem to be discussed more since Greenspan mentioned it at the Symposium (e.g. Bloomberg uses the term in this story). There has been a wealth of theoretical work on learning in recent years, the corresponding empirical work is catching up, and it will interesting to see the degree to which the learning literature further affects the conduct of monetary policy.

Comments

[Edward Hugh]

Mark, you are clearly giving a lot of thought and consideration to all this. One of the funny things I note is the fact that from over in Europe Greenspan looks pretty good, whereas people in the US want to believe in Trichet. I have a short piece on this and I think I'll put it up as a post. In the bloomberg article this quote from Trichet caught my eye:

``This model has served us extraordinarily well,' said Trichet, 62, in an interview. A clearly defined monetary strategy ``reduces uncertainty about the central bank's ultimate motives, provides a stronger anchor for inflation expectations and makes it easier for the market to map the expected path of the policy rate.''

The issue really boils down to whether or not it is so desireable to reduce uncertainty to such an extent, given that the central bankers themselves may be uncertain about the evolution of the variables to which they are responding. I really admired Greenspan most when he kept people guessing. This made it difficult for the markets to front run him. This was Greenspan at his best. I think during the deflation scare period the accommodative policy was thoroughly justified, and I suspect many of his critics don't understand just what a headache deflation would have been. Fortunately he over estimated the extent of the danger, but this is only known post hoc.

Since we got into the 'measured increase cycle' he has been a lot more predictable, and this is where the magic has gone. The markets now can front run him, until and unless there has to be an abrupt change, then we will really see in the way people handle this just how successful predictability has been.

[Edward Hugh]

On the other hand there is the simple pragmatic detail that as a newborn institution with a relatively short history learning may need time, and the difference in emphasis may be related to this.

Incidentally you may find this paper, given at a recent ECB conference, interesting:

ttp://www.ecb.int/events/pdf/conferences/ecbimop/Preston_Hydra1.pdf

[Pandora Braclets]

I like how the charms professor is a bearded little elf with snowy white hair & in all the other movies is ala charlie chaplin?