Chicago Fed President on Inflation Forecasting
CHICAGO — The June 15 edition of The Wall Street Journal editorial page featured an opinion piece by Chicago Fed President Michael Moskow on inflation forecasting.
Following is the text:
The Inflation Game
Over the past three years, the Federal Reserve's accommodative monetary policy has supported both of our primary goals of maximum sustainable growth and price stability. Not only did it provide useful stimulus to real economic growth, it helped prevent inflation from falling to undesirably low levels. However, as the economy has improved, concerns about the sustainability of growth and falling inflation have been replaced by concerns about the possibility of rising inflation.
One of the perks of my job is that I get to talk to many different people — from international investment bankers to local community groups. The insights and concerns that people share are an important complement to the statistical analyses that Fed economists produce. In recent months, people have been worried about increases in the prices of many goods and services; gasoline, metals, housing, and health care prices are the most frequently cited.
Of course, such reports aren't just anecdotes. This year, broad measures of consumer inflation have risen from the extremely low rates we saw late last year. One of the key challenges for Fed policymakers is judging to what extent such developments are symptomatic of a more general increase in inflationary pressures and to what extent they are transitory fluctuations in what are often volatile monthly data.
Many analysts like to point to a single economic indicator that they believe to be a good barometer of underlying inflationary pressures. Common favorites include gold and commodity prices. Although there's a venerable history of tracking such indicators, I've always found it difficult to know when their movements reflect general inflationary pressures or when they're being driven by more idiosyncratic forces. Often it seems to be the latter. Though rising gold prices are often claimed to foreshadow higher inflation, since 1984 movements in gold prices and general inflation have been virtually uncorrelated. Thus, it's not surprising that formal statistical analyses show that single indicators like gold have little ability to predict inflation. Indeed, given its frequent ups and downs, one could say — with apologies to Paul Samuelson — that the price of gold has predicted nine out of the last five bouts of inflation.
Other analysts see results like this, throw up their hands and say we might as well give up on forecasting changes in inflation and simply predict that next year's inflation rate will be the same as this year's. In fact, core inflation in the U.S. generally does not change very quickly and — as some of my colleagues at the Minneapolis Fed reported a few years ago — forecasting next year's inflation with this year's rate did remarkably well between 1985 and the late '90s. Not surprisingly, however, this method did not do well during less tranquil inflationary periods. And, obviously, such forecasts offer no guidance on when inflationary pressures are building or waning.
There is a useful observation to make at this point. Predicting inflation as accurately as possible is only one component of our forecasting exercise. Another is learning from our mistakes. Understanding the reasons that forecasts are off the mark often yields important insights into the workings of the economy. This requires a view or theory of the inflation process — something that is missing from forecasts simply based on the price of gold or last year's inflation rate.
An important case in point was the recent acceleration in productivity. Inflation forecast errors provided early indication of the staying power of the extraordinary step-up in productivity growth that occurred in the second half of the '90s. Many forecasters thought that the robust economic growth during the period would generate higher inflation. But inflation kept coming in lower than expected. Over time, we began to understand that the economy was not experiencing the resource constraints and associated cost pressures that we had expected because the underlying trend in productive capacity had increased. Although many economists had conjectured that higher productivity growth was here to stay, the inflation data provided useful confirmation on its persistence.
Another way of describing the lessons from the inflation forecast errors of the late '90s is that we learned that actual output was not as high relative to potential output as we'd thought. Many statistical studies show that indicators of resource utilization can help predict inflation. The output gap — the difference between potential and actual output — is simply one framework used to gauge resource slack. Of course, the gap is difficult to measure, and factors other than the level of resource slack affect inflation. Indeed, inflation is the outcome of complicated interactions among the structure of the economy, inflationary expectations, monetary policy, and unforeseen events. This means that forecasting and explaining inflation requires a combination of economic theory, statistical analyses, and judgment. Accordingly, monetary policymakers look at a wide variety of economic data and models in their search for general patterns of consistency that signal incipient inflationary pressures.
One way to find and quantify these patterns is to use statistical techniques. At the Chicago Fed, we have been publishing our Chicago Fed National Activity Index (CFNAI) since March 2001. The index is a statistical summary measure of 85 economic indicators that provides useful predictive information about turning points in inflation. Still, we wouldn't advocate exclusively using the CFNAI to forecast inflation; it's just one tool in the box.
I've discussed several alternative approaches to forecasting, and it is fair to ask how well these methods have done over time. For example, rising gold prices in 1979 may have provided a useful signal of inflationary pressures then, but a track record of continued success is crucial. How good has the Fed been at predicting inflation? Recent studies by David and Christina Romer of the University of California and Christopher Sims of Princeton have found that the Federal Reserve Board staff's inflation forecasts have performed better than those of professional forecasters over the last 20 years. Perhaps this is because Fed forecasts are informed by a wide range of economic indicators, a variety of analyses, a healthy dose of anecdotal information, and judgment.
Over the past 25 years, inflation has come down from double-digit rates to a pace consistent with effective price stability. This is an important achievement. Of course, we must continue to be vigilant in monitoring developments that pose a risk to this accomplishment. Currently, the federal funds rate is very low, and this degree of policy accommodation cannot be maintained indefinitely. Sifting through the data and analyzing them from a variety of perspectives will continue to be an important input into our assessment of the appropriate stance of monetary policy.