Economic Forecasts and Monetary Policy

William Poole*
President, Federal Reserve Bank of St. Louis

Arkansas Business and Economic Society and
The Central Arkansas Chapter of the Risk Management Association
Little Rock
Feb. 15, 2001

*I appreciate comments provided by my colleagues at the Federal Reserve Bank of St. Louis. I take full responsibility for errors. The views expressed are mine and do not necessarily reflect official positions of the Federal Reserve System.

Of all the things that people ask me after speeches, at meetings, at parties and during casual conversation over lunch, no topic arises more often than that of my outlook for the economy. Strangely enough, this is a topic on which I can shed little light for anyone who is modestly well informed. My purpose today is to discuss this disconnect-why it is that people believe that I have a very special economic forecasting crystal ball and why I know that I do not.

The first question is pretty easy. People seem to assume with little thought that I must have some forecasting expertise, or otherwise I wouldn't be in my current position. I'll challenge that assumption by explaining why I know I do not have such expertise. However, toward the end of my remarks I'll explain how I apply my professional expertise to the output of expert economic forecasters.

I'll explore my topic by first discussing the accuracy of economic forecasts. Perhaps you are already chuckling at that phrase-"the accuracy of economic forecasts." If so, I'm off to a good start. Then I'll dig into why forecasts are not very accurate. I think that what I have to say in this regard will ring true to you. If forecasts are not very accurate, how can the Fed conduct monetary policy? In discussing this question, I'll emphasize that what the Fed can do is to set a stable long-run path for policy, yielding low and stable inflation on the average. Finally, I will speak to the current outlook, concentrating on conveying the nature of the mainstream, consensus forecast. I share that forecast, because I am a consumer rather than a producer of forecasts.

Before proceeding, I want to emphasize that the views I express here are mine and do not necessarily reflect official positions of the Federal Reserve System. I thank my colleagues at the Federal Reserve Bank of St. Louis for their comments, but I retain full responsibility for errors.

Forecast Accuracy

In discussing the accuracy of economic forecasts, I'm going to rely heavily on the Blue Chip compendium of forecasts. As many of you probably know, Blue Chip is a monthly newsletter reporting forecasts from a panel of economic forecasters. The newsletter is dated the tenth of every month, and I'll be referring to the February 10, 2001 issue.

The Blue Chip consensus forecast for the U.S. economy is that in 2001 total real gross domestic product will grow by 2.1 percent over its total for last year. It is important to note that this forecast refers to the total GDP for the year, and not the growth from the fourth quarter of 2000 to the fourth quarter of 2001, which is another common way of measuring real growth for the year. Blue Chip defines the "consensus" as about the middle, or the median, of the 51 forecasters surveyed.

I think that forecast is reasonable, but suppose I did not? Suppose, hypothetically, I were to tell you that I believe the U.S. economy will be very much weaker than the consensus forecast. Suppose, also, that I were to tell you that as a consequence of weak GDP growth corporate profits are going to be much lower than the consensus forecast this year and next year as well. Moreover, I'd tell you-this story is all hypothetical, remember, to prove a point-that with the miserable outlook for corporate profits I believe that the stock market is grossly overvalued at current levels.

I'm sure I'd get your attention by spinning out this dreary forecast. How would you react? Would you, for example, pull out your cell phone to tell your broker to sell all your common stock? Would you use the cash you raised as collateral for large additional short sales of common stock? I think not. I hope not.

Why not? The answer is pretty simple. You know that economists' forecasts are not to be taken that seriously. Whatever might be my forecast, you know that there are other economists with respectable credentials who are forecasting continuing solid growth after a slow first half of this year. Just as you would not sell short on the basis of my forecast, you would not mortgage your house to the hilt to buy stock on margin if I were instead to offer a very upbeat forecast. My economic forecast-every economic forecast-must be treated with extreme caution.

In the February Blue Chip, 51 respondents offered forecasts for 2001 and 47 for 2002. Although the consensus forecast was 2.1 percent real GDP growth in 2001 over 2000, the average of the top ten was 2.8 percent growth and the average of the bottom ten was 1.3 percent. The most optimistic of the 51 forecasts for real GDP growth in 2001 was 3.6 percent and the most pessimistic was 0.9 percent. For 2002, the consensus forecast was 3.5 percent real growth; the average of the top ten was 4.1 and for the bottom ten 2.8 percent. The most optimistic real growth forecast for 2002 was 5.3 percent and the most pessimistic was 2.2 percent.

The range of forecasts is obviously quite substantial. The names in the Blue Chip list are mostly names that you would recognize. They include major banks and other companies that maintain economic forecasting units. In addition, there are specialized commercial forecasting firms in the list. All of these forecasters have an intense interest in getting the forecast right. They devote substantial time and resources to their forecasts.

I am not in the forecasting business-I am a consumer of economic forecasts. What should I make of the substantial range of forecasts produced by serious forecasters who make a living at this enterprise? To me, the only reasonable conclusion is that there is substantial room for significant differences of professional opinion on this matter. Different professionals have different views on the outlook, and that fact must reflect gaps in economists' knowledge. The fact is that economics is not so well developed as a discipline that these forecasters are all compelled by confirmed knowledge to report forecasts clustering within a few tenths of a percent of each other. Instead, what economists can reasonably say they "know" permits wide differences of opinion on forecasts. Forecasters themselves will say that the differing forecasts of many of their rivals are "reasonable."

Let me now look at another dimension of this range of forecasts. The February Blue Chip contains a table of consensus forecasts for 2001 made during the course of 2000. The January 2000 consensus for 2001 was real growth of 3.0 percent. The consensus forecast rose over the course of 2000 reaching a high of 3.5 percent in September. In October the consensus was also 3.5 percent real growth. In November, the consensus forecast began to fall, but it slipped only slightly, to 3.4 percent. Thus, not only do various forecasters differ about the outlook at any moment of time, but they change their mind as time goes on. Just to emphasize this point, note again that the consensus for 2001 last November was 3.4 percent but now, only three months later, the consensus for this year has fallen to 2.1 percent real growth.

What should we conclude from the fact that forecasts are often revised substantially over the span of a few months, as illustrated by the recent significant downward adjustment in the forecast for this year? For one thing, certainly, one reason why you ought not to sell your portfolio, or mortgage your house to buy additional stock, on the basis of economic forecasts is that these forecasts differ a lot from one forecaster to another, and change significantly with the passage of time, even over a couple of months. For another thing, let's recognize that forecasts are not now and never have been highly accurate.

There is a substantial professional literature on the accuracy of economic forecasts. To discuss this literature in any detail would require that I get into a mind-numbing exposition of exactly what is being forecast-for example, the preliminary or finally revised GDP estimate-and other such matters. But let me give the flavor of this research. Every forecast ought to have a standard error attached to it. When discussing annual average data, which I have been doing so far today, the standard error for a real GDP forecast is in the neighborhood of one percentage point. That is, when we say that the forecast for 2001 is 2.1 percent real growth, what we mean is that there is about a two-thirds probability that real growth will be 2.1 percent plus or minus 1.0 percent. That means, of course, that there is a one-third probability that the actual outcome will be either higher or lower than the range from 1.1 to 3.1 percent real growth.

We also need to attach an error band to the inflation forecast. For 2001, the Blue Chip consensus for the consumer price index is that its annual average will be 2.6 percent above 2000, with a range from high to low among the panel of forecasters of 1.8 to 3.5 percent. Based on my reading of the professional literature on this matter, I think that the reasonable standard error to attach to the year-ahead inflation forecast is probably a bit smaller than for the year-ahead real GDP forecast, but perhaps not all that much smaller.

Let me summarize this discussion briefly before continuing. Economic forecasts are subject to an inherent range of error. In the United States, my assessment of the evidence is that every real GDP forecast should have an error band around it of about plus or minus one percentage point for a forecast made at the beginning of the year. For inflation, the error band should be only a little smaller. Even with these large error bands, outcomes outside the forecast range will be observed fairly often. Failure to understand the limits of economic forecasting can only produce problems for anyone acting on the basis of the forecasts.

Why Forecasts Go Astray

Where do forecast errors come from? Why do forecasts go astray? Understanding the answers to these questions helps us to better appreciate what economic forecasts mean.

Let me begin with an analogy. Suppose you were to ask a professor of chemistry at your local university what will happen in the chemistry lab next September 27, in the afternoon. The professor might look a bit puzzled, and then go on to explain that if the experiment concerns the effect of pressure on the boiling point of water then the answer is very well known. But, I insist, I want a simple answer: What is going to happen? Don't confuse me, professor, with more questions.

My question to the chemistry professor, in fact, is either foolish or meaningless. Chemistry can provide good predictions when the experiment is known. Without knowledge of what experiment is to be run, it's impossible to know what the outcome is going to be.

Economic forecasters face the same problem. The discipline of economics provides substantial insights about what will happen if some specified event takes place. But, not knowing what events might take place-what economic experiments might be run-we have a very weak basis for making a forecast. In current circumstances, we are being asked to offer a forecast even though we do not know what the nature of possible tax legislation will be, what events might occur abroad, what the resolution of the California electricity situation will be, and so forth. Clearly, we face thousands of possible events that could derail even a very sensible forecast. So, one reason forecasts go astray is that things happen that the forecaster had not foreseen, and in most cases could not foresee.

Forecasts also go astray because economists have not been able to pin down accurately how the economy will respond to particular events. Of course, the likely effects of some events are better understood than others. That is, even after some events occur we are often uncertain about their likely effects. In current circumstances, for example, we do not have solid predictions of the likely effects of the decline in the stock market after last March, or of the mild depreciation of the dollar against the Euro in recent months, or of the apparent downturn in the Japanese economy, or of numerous other things that have happened or seem likely to have happened. Knowledge just isn't all that complete. We can poke fun at economists, or laugh at their plight. In fact, most of us economists do both. Still, at the end of the day, we have to take account somehow of the fact that knowledge is incomplete. After all, we are taking about a deadly serious business because the consequences of policy mistakes, as we all know, can be very unfortunate.

Another reason forecasts go astray is that behavior depends importantly on expectations about the future. The decisions people make about changing jobs, about saving and spending, and that firms make about hiring workers and investing capital depend critically on their expectations about the future, and in some cases the fairly distant future. Economists understand something about how expectations are formed, but there are huge gaps in what we know. I suspect that for the indefinite future economists will be faced with uncertainties about sudden shifts of public opinion, or non-shifts when changes in opinion might seem logical.

I've discussed the sources of forecast errors not because I believe that we have any realistic chance of wiping them away but rather because they are so obviously relevant when we spend a few minutes thinking about them. Understanding why forecasts go astray is an important part of understanding what value forecasts have.

Implications for Monetary Policy

What are the implications of limited forecast accuracy for monetary policy? The place to start is with the observation that professional forecasters do in fact forecast more accurately than simple alternatives one might consider. For example, forecasters on average beat naïve forecasts such as that growth will be constant or the same as last year. Forecasts are a valuable input to economic policy, provided consumers of forecasts like me understand their limitations.

Where I think my professional expertise matters is in reading and assessing the work of economic forecasters. I know why I think some forecasts-almost always the ones at the extremes-are "off the wall." To understand why forecasts have the characteristics they do requires extensive knowledge of business cycle dynamics and long-run economic growth processes. This understanding is an important input to making monetary policy decisions.

As you may know, twice each year each member of the Federal Open Market Committee (FOMC), the Fed's main monetary policy body, submits an economic forecast in preparation for the Federal Reserve's Monetary Policy Report to the Congress and the Chairman's congressional testimony on that Report. Chairman Greenspan submitted such a report and testified just two days ago. Although the forecasts of individual FOMC members are not made public, the Monetary Policy Report has a table showing the range and central tendency of those forecasts. At the St. Louis Fed, we prepare our forecast by studying what professional forecasters, both inside and outside the Fed, are saying and examining whether we might have some reason to differ from the consensus forecast. By comparing the central tendency of the FOMC forecasts with the Blue Chip consensus, for example, it is evident that the FOMC forecasts are in the mainstream of professional forecasters in general. No one should be surprised at this outcome; the professional forecasting fraternity bases its work on an accumulated body of forecasting techniques broadly shared among economists who specialize in this area.

Given what I've said so far, though, I hope you agree that I'd be crazy to make my policy position depend on whether a forecast is a few tenths of a percent higher or lower. Nor should I pay great attention to arguments among forecasters about a few tenths of a percent real growth or inflation. Given the record of forecast errors, a forecast of real growth of 2 percent is basically indistinguishable from one of 2 ý percent, or 1 ý percent. Quite frankly, it is ludicrous to present a forecast to the tenth of percentage point when the discussion is intended for a non-professional audience. Doing so provides a completely misleading picture of what a forecast might mean. I've used tenths of percentage point today in discussing the Blue Chip forecasts because that is how Blue Chip reports the forecasts. But whenever I discuss the outlook, I talk in terms of rough ranges, or forecasts rounded off to the nearest half percentage point. Thus, today I would not speak of 2.1 percent real growth but of about 2 percent real growth. And I would hope to get across the point that what that forecast really means is growth most likely in the 1 to 3 percent range. I am not saying that it makes no difference whether growth is 1 percent or 3 percent, but only that at this time we must be prepared to deal with this range of uncertainty.

Even if near-term forecasts were not subject to considerable uncertainty it is important to recognize that monetary policy adjustments cannot have precise and quick effects on the economy. There has been lots of talk about whether the Fed can engineer a "soft landing" for the economy. In early December I experienced a true soft landing; after touring the Boeing F-18 assembly plant in St. Louis, I had a short session on the F-18 simulator. The instructor helped me to land an F-18 on an aircraft carrier deck. That maneuver is routine in the Navy, but I can assure you that the Federal Reserve cannot land the economy on a carrier deck in the economic ocean we face. Instead, the Fed's job is to keep the economy on a sound long-run track; we cannot avoid most short-run economic fluctuations but we can help to prevent those inevitable jolts from becoming cumulative and pushing the economy far off its long-run growth track.

Fed policy has been extremely successful in recent years in maintaining confidence that the rate of inflation will remain low and stable over a period of years. This stable long-run outlook is what enables the markets to work through necessary short-run adjustments in constructive fashion. Market interest rates respond sensitively to current conditions and do much of the stabilization work. The Fed sets a stable long-run environment within which the markets can make short-run adjustments in response to current developments.

Let me now return to my opening paragraph where I noted the intense interest people show in my views about the economic outlook. People do seem to expect that I will be able to shed some light on where the economy may be going, and some light on future Federal Reserve policy. I hope I have convinced you that I do not have any gems of unconventional wisdom to offer on where the economy is going. I accept the judgment of expert forecasters, just as I do expert lawyers. I question and probe to better understand expert advice, but do not try to construct from scratch either my own economic forecast or my own legal analysis.

Nor can I provide any help in forecasting the Fed's next policy adjustment. The issue here is that economic conditions do sometimes change rapidly, and the job of the FOMC is to collate all the scraps of information up to the time of a meeting. I do not even make final my own position at an FOMC meeting until I quite literally get there. Of course, I am accumulating information all along, but I know from experience that new information and new staff evaluation of existing information can and should affect my thinking. For example, I read carefully the Fed staff economic forecast, which is available only a couple of days before the FOMC meeting.

No one should have great confidence in what action the FOMC will take four weeks, for example, ahead of a meeting. The world is just too full of surprises for that. As we come up to the day of the meeting itself, the market and the Fed will have digested all of the information available, and in recent years have most often come to a common judgment about the appropriate policy action. This is as it should be given the uncertainties of the world we face.

The Economic and Policy Outlook Today

As I already noted, the consensus outlook is about 2 percent real growth and about 2 ý percent CPI inflation in 2001, measured by the annual average for 2001 over the annual average for 2000. Looking at quarterly data, the Blue Chip consensus for the first quarter is 1 percent real growth and then about 2 percent for the second quarter. Thereafter, Blue Chip sees the economy gradually picking up speed to a growth rate of about 3 ý percent in the fourth quarter. Growth is expected to continue at about that rate into 2002.

This outlook should be regarded as very heartening. Forecasters believe that the economy's slower first half will be followed by growth that is more or less consistent with the economy's estimated long-term growth potential. The estimate of growth potential reflects the underlying demographics of the labor force, which point to growth in labor input of about one percent per year, and estimated trend productivity growth in the ballpark of 2.5 percent. The productivity growth figure must be regarded with considerable uncertainty, as productivity processes are not very well understood. What I think we can say about productivity growth is that every passing year brings additional convincing evidence that U.S. productivity growth is now measurably higher than it was in the 1970s and 1980s. The issue is how much higher and, more importantly, how monetary policy should deal with the range of uncertainty over productivity growth.

Most importantly, for me, is that through all these current economic adjustments and uncertainties the inflation rate remains well controlled. In energy markets, futures prices indicate that the best guess is that energy prices will be trending lower over the next several years. Current information continues to suggest that firms believe they have very little pricing power. Data on inflation expectations over the longer run indicate that households and firms continue to believe that inflation will remain in the range of recent years.

I'll summarize by emphasizing three points. First, the outlook I've outlined reflects the middle of the range of best professional judgment. That is the range where it is prudent for me as a policymaker to hang my hat. Second, the outlook is subject to uncertainty, and will most likely be revised in coming months as new information arrives. Third, because new information will in all likelihood change the outlook in some respects, monetary policy cannot be locked into a given path for the principal policy instrument-the target federal funds rate set by the FOMC. Everyone should understand that the policy forecasts for upcoming FOMC meetings currently built into the market may well need to be revised, either up or down. No policy actions are ever "baked into the cake" long in advance of FOMC meetings.

I do not want to finish by leaving the impression that FOMC decisions are driven by short-run data. The Fed has a commitment to low and stable inflation as its most fundamental policy goal over the long run. Short-run policy adjustments must always be consistent with that fundamental goal. This point is especially important in the current environment, where understandable concerns about the current state of the economy may tend to obscure consideration of long-run policy objectives. Our job is to filter out the short-run noise in the data, about which we can do nothing anyway, and maintain a stable policy environment focused on keeping inflation low in the long run. These are the conditions in which individuals, firms and markets can make necessary short-run adjustments in an efficient manner.

I hope I've given you a perspective on how economic forecasts fit into my conception of the Fed's policy process. I'd be pleased to take your questions.

Back to top