October 29, 1999
« More Economic Letters
Risks in the Economic Outlook
Robert T. Parry
This Economic Letter is adapted from a speech delivered by Robert T. Parry, President and Chief Executive Officer of the Federal Reserve Bank of San Francisco, at the Annual Meeting of the National Association of Business Economists on September 27, 1999, in San Francisco.
It’s always a pleasure and an honor to speak at the NABE annual meeting. One reason is that it’s a chance for me to get together with so many professional forecasters. And this is an especially good time to reflect on an interesting fact about the profession: while the last three years have been very good years for the U.S. economy, they haven’t been such great years for a lot of forecasters. Frankly, for much of that time, many forecasts predicted that the combination of fast growth, low unemployment, and low inflation was about to end any minute.
It’s easy to understand why. It was natural to assume that there was little change in the structure of the economy. So it also was natural to assume that economic performance would return to historical norms. Over the last three years or so, that left most forecasts centered on a real GDP growth rate of 2% or a little higher, which was thought to be the long-run trend. Specifically, the median one-year-ahead forecasts from the Blue Chip survey were 2.1% in January 1996, 2.1% in January 1997, and 2.2% January 1998. But, in fact, the actual rates were almost twice that! At the same time, with the economy growing rapidly and labor markets apparently tight, most forecasts overshot actual inflation in 1997 and 1998. For example, the Blue Chip survey showed CPI inflation of 2.9% and 2.3% in 1997 and 1998, respectively, versus actual figures of 1.9% and 1.5 %.
In my remarks today, I want to focus on what these forecast errors could be telling us about the current economy. And I’ll also discuss what the errors mean for the conduct of monetary policy.
So, let me turn to the first issue–what could be going on in the economy that would be consistent with fast output growth, low unemployment, and low inflation? I can point to several special factors that have helped keep inflation down during this period. For example, global financial crises weakened foreign economic activity. That led to a stronger dollar–and therefore lower import prices–as well as falling commodity prices worldwide and a drop in capacity utilization rates in U.S. manufacturing. In addition, energy prices were falling during 1997 and 1998, and so were the costs of health care, as the industry restructured itself. Finally, there’s a technical point: the CPI is down 1/2% or a bit more because of the improvements the Bureau of Labor Statistics has made in measuring prices.
But beyond these special factors, there’s a more fundamental issue–and that’s the nation’s productivity. Certainly, several developments in the last couple of years suggest that we may be in the midst of a supply shock related to more rapid, and more dispersed, technological change. And that would be consistent with the fast growth, low unemployment, and low inflation we’ve seen. One obvious indicator is that measured productivity growth has picked up–rising to just under 2-1/2% on average. This compares to an estimated trend rate of only around 1 to 1-1/2% for the 1980s and the first half of the 1990s. Another indicator is the faster growth we’ve seen in real labor compensation. This result is something we’d expect to go along with higher labor productivity. A third indicator is the strength of corporate profits. This also can help explain at least part of the extraordinary rise in stock market values.
It’s easy to see why people would point to technological advances as the source of much of this increase in productivity growth rates. For one thing, we’ve seen very rapid increases in investment in computers and other information processing equipment–since 1995, it has ranged from just under 20% a year to over 30% a year, in real terms! For another, there are plenty of examples of the difference technology can make–not only in labor-saving devices, but also in changing the way people do business. For example, I saw firsthand how technology could increase the flexibility of production processes when I visited a lumber mill in Oregon. They demonstrated how they used lasers to define the geometry of a log, and they then cut it based on the latest price information for different cuts. If there’s a shortage of two-by-fours, then prices on them rise and the mill cuts more of them and fewer of other sizes.
Such improvements in production flexibility and real-time information flows illustrate how technology can make a difference for U.S. firms. They can help eliminate bottlenecks, streamline production, and fine-tune specifications so firms can better match–and even anticipate–customers’ needs. And this all could translate into faster productivity growth for the economy.
But even though there are indicators–and anecdotes galore–suggesting that a productivity shock has been driving the recent performance of the U.S. economy, there also are serious uncertainties–many of them revolving around the productivity data themselves.
For example, just consider recent productivity numbers. Frankly, they don’t stand out as all that robust when you compare them to recent decades. There are lots of instances–especially early in past business cycle expansions–when the productivity growth rate was much higher than it is now.
Another issue is where the productivity is showing up in today’s data. Part of it is in the computer industry itself, and thus does not reflect efficiencies that are spreading throughout the economy.
Of course, it’s well-known that the data suffer from some real problems and could be missing much productivity growth. For example, the data don’t measure productivity in the services sector very well, and in some cases not at all. To resolve some of these problems, the GDP accounts are undergoing some major revisions. The new data will be released in late October, and they’ll reflect a number of changes, such as beginning to include the measurement of productivity in banking and calculating software as investment spending, rather than simply as a raw material to the production process. As a result, estimates of productivity growth will be higher–though we don’t know by how much. As welcome as these revisions are, however, they’re not a complete solution to the problem of measuring productivity.
My final points about the uncertainties surrounding productivity are yet more fundamental. One problem is: at this early stage, we can’t tell whether the surge in productivity growth is a cause of the fast output growth–or a consequence of it. And it may be a consequence because, historically, productivity growth has followed a pro-cyclical pattern. So, it’s possible that the recent strength in productivity won’t last very long–it might largely be due to the strong business cycle upswing we’ve been in. In that case, continued strong real GDP growth and tight labor markets eventually would create pressures for inflation to increase. Furthermore, even if the productivity surge is a cause of the fast output growth, we don’t know how long it will last. While the data admit the possibility of a sustained increase in productivity, I can’t say that I find them convincing by themselves. So, until we have enough data to see the rapid growth sustained for a long time, we won’t know for sure.
The uncertainty about recent productivity growth appears to be the major uncertainty in the outlook for the U.S. economy, and also for the conduct of monetary policy. For policy, this uncertainty complicates the question of where the Fed should be along the spectrum of more pre-emptive action or more cautious action.
Ideally, policy should be more toward the pre-emptive end of the spectrum because of the long lags between policy actions and their effects on the economy. For example, if policymakers wait until inflation actually begins to pick up steam, they face a problem that was all too familiar in the late 1970s and early 1980s. In order to quell inflation, interest rate increases have to be bigger, which means the output losses are bigger, and the employment losses are bigger.
But, if a central bank reacts early and correctly, it can alter inflation expectations and cut off the rise in inflation before it gets started. It looks like that’s what happened in the U.S. in 1994. At that time, we were dealing with forecasts of higher inflation that were based not only on increasingly tight labor markets but also on low short-term real interest rates. So the Fed responded by raising interest rates substantially. In that case, inflation didn’t take off, and the economy moved smoothly into the favorable conditions we’ve enjoyed in recent years.
But, suppose there are reasons to doubt the forecasts. Suppose, like now, we’re uncertain about the underlying model of the economy. When there’s a high degree of uncertainty about forecasts, it could be best for policy to tend more toward the cautious end of the spectrum. With high uncertainty about the future, a somewhat delayed action could be preferable to running the risk of tightening when it’s not warranted.
The appropriate degree of caution depends on an analysis of the risks of the two approaches in any particular circumstance, and it will vary over time. In 1994, there were several key indicators of rising inflation in the future; in addition, there was no major source of uncertainty to make us doubt the normal relationships in the economy.
In the current situation, we face uncertainty about whether the good news on productivity growth is a cause or a consequence of our current performance–and if it’s a cause, how long it will last. Since most forecasts of output and inflation have been off the mark recently, it makes sense to place somewhat less weight on them than we normally have in the past. And the same is true of our interpretation of some of our usual indicators. For example, labor markets have been tight for several years. And the so-called output gap, which compares real GDP with an estimate of its long-run trend, also has suggested rising inflation for a few years. But the possibility of a productivity shock renders these indicators less reliable. As a result, the Fed has taken a fairly cautious approach in reacting to such indications of a higher future inflation, especially since actual inflation has been so well-behaved.
But it’s important to make a distinction here. While the Fed has reacted cautiously to signs of building inflationary pressures, that doesn’t mean that the Fed has reacted cautiously to everything. For example, last fall, there was no doubt that the financial crises abroad were putting a strain on U.S. markets, and we lowered rates promptly. And conversely, as the financial crises abated and foreign demand began to strengthen, we moved to raise interest rates gradually — but still pre-emptively. So far, this strategy seems to have worked well. Inflation has remained subdued, while the expansion has not only continued for eight years–it has even strengthened!
I’m certainly pleased with the Fed’s contribution to this outstanding economic performance. But I’m also well aware that monetary policy is only part of what’s made the U.S. economy the envy of the world, especially in recent years when we have experienced a technology boom. By moving toward price stability, the Fed has helped to provide a healthy, stable environment for people and businesses to manage their economic affairs. At the same time, sound fiscal and regulatory policies certainly can claim credit for some share of the nation’s economic and financial success.
These policies provide an environment in which the drivers of America’s productivity and prosperity–hard work, innovation, and entrepreneurship–can flourish.
Robert T. Parry
More Economic Letters
Opinions expressed in FRBSF Economic Letter do not necessarily reflect the views of the management of the Federal Reserve Bank of San Francisco or of the Board of Governors of the Federal Reserve System. This publication is edited by Sam Zuckerman and Anita Todd. Permission to reprint must be obtained in writing.
Please send editorial comments and requests for reprint permission to