July 2, 1999
« More Economic Letters
Supply Shocks and the Conduct of Monetary Policy
The U.S. economy has performed remarkably well over the last few years. Real output has grown at a pace that is noticeably above average, while inflation has declined somewhat. These–and related–developments have led to an increase in uncertainty about the economy’s long-run growth rate.
In this Letter, I argue that increased uncertainty about the long-run growth rate of the economy means that one should not rely as much on some of the traditional ways of gauging the extent of inflationary pressure in the economy. Specifically, I argue that it may be unwise to rely too heavily on measures such as the “sustainable” rate of output growth or (especially) the level of “potential” output during a period in which the economy appears to have been hit by a positive supply shock of unknown magnitude and duration. A better strategy may be to pay more attention to a measure such as nominal GDP growth.
Generally speaking, the rapid growth of the last few years has surprised economic forecasters; for most of this period, they have been projecting that the economy will slow down. Thus, in December 1997, the consensus Blue Chip forecast called for real GDP to grow at a 2.2% rate over the four quarters of 1998, while actual growth came in at 4.2%. And the 3.8% growth over the previous year was almost twice the 2% growth predicted by the consensus in December 1996. The behavior of inflation has been a surprise as well. Not only did inflation come in lower than expected, but the inflation rate actually declined. For instance, the core CPI (which excludes food and energy prices) rose at a 3.0% rate in 1995 but at a 2.2% rate over the first four months of this year.
This pattern of surprises suggests that the economy is experiencing a positive supply shock. The likely source of such a shock is the technology sector, especially information processing and communications technology. The explosive growth in the use of computers over this decade hardly needs to be remarked upon; computers seem to have transformed everything from trading stocks to designing airplanes. Also commonplace is the observation that this remarkable growth has been accompanied by steadily falling computer prices.
It is not unreasonable to expect that the enormous investment that firms have made in information processing technology over this period would have a beneficial impact on the economy’s productivity. Yet until recently, such evidence has been hard to find. For example, until a few years ago, data since the early 1970s looked consistent with a long-term (or trend) productivity growth rate of around 1%. More recently, the growth rate of productivity has picked up. For instance, output per hour in the nonfarm sector has grown by an average of 2.3% per year over the 1996-1998 period.
While there is now more evidence to suggest that the average growth rate of productivity has gone up, it is still hard to tell by how much, or for how long. Much (if not all) of the recent pickup in productivity growth could be temporary; that is, it could be the economy’s usual cyclical response to high demand. Or most of it could be the result of increased use of computers. And even in this case, the resulting boost to productivity growth could be short-lived, or it could be with us for a long time.
Uncertainty about the average (or trend) growth rate of productivity translates into uncertainty about the trend growth rate of output. Uncertainty about trend output, in turn, poses a problem for those who would rely on such measures to determine the appropriate stance of policy. Suppose, for instance, that the trend rate of productivity goes up, perhaps because of the increasing use of computers. The increase in the trend productivity growth rate is unlikely to be noticed at first, however, the higher output growth will be obvious to all. Policymakers who are accustomed to output growth rates from the regime before productivity accelerated could well interpret the faster growth as evidence of excess demand and move to a tighter policy as a result.
Reliance on measures such as the “output gap” would be even more problematic, since this measure requires knowledge of the level of potential or trend output in the economy. (Taylor 1993 recommends that policymakers base the stance of monetary policy on a measure of the output gap and a target rate of inflation.) An underestimate of the productivity trend would lead to a measured output gap that would keep growing over time, even if the true gap were zero. Since the output gap is meant to measure the cyclical or temporary component of output, policymakers would interpret the growth in the measured gap as evidence of an overheating economy and would be likely to keep tightening policy in a mistaken attempt to reduce the gap to zero.
How much of a mistake could one actually make? While there is no way to provide a definitive answer, some recent work by Orphanides (1999) provides one measure of the potential for such a mistake. Orphanides looks at the conduct of policy over a 30-year period ending in 1993. Of particular interest to us is his analysis of policy during the 1960s and the 1970s, a period during which the rate of productivity growth slowed dramatically. This slowdown seems to have persisted till the end of Orphanides’s sample period; but when it began, it was hard to judge how persistent it would be. Orphanides finds that it took a long time for economists to determine that the slowdown would be long-lived, and that they persisted with old estimates of the trend well into the 1970s. Thus, real-time estimates of the output gap showed an economy that appeared to be substantially below trend in the early and, especially, the mid-1970s. He goes on to show that if policymakers had based policy on a measure of the output gap derived from then-available measures of the trend level of output, they would have generated a funds rate path close to the actual path during the 1970s (even if they also had been responding to the rising inflation rate). In other words, they would have generated a monetary policy that was similar to the actual policy over this period. Thus, a mistake about the trend growth rate could be large enough to take us from the low inflation rate of the 1960s to the high inflation rate of the 1970s.
The possibility of such mistakes in an environment of increased uncertainty about the growth rate of either actual or potential real output suggests that one should pay less attention to policy rules or strategies that rely upon such measures, and pay more attention to rules or strategies that are not affected by such problems. It seems more useful, in particular, to pay more attention to nominal GDP growth.
The recommendation to pay attention to nominal GDP growth given the likelihood of a supply shock echoes similar recommendations in the past. Hall (1983) argues for a nominal GDP target on the grounds that it can be hard to distinguish cyclical from structural changes. According to McCallum (1988, pp. 174-175), the “..most fundamental..” reason for paying attention to nominal GDP is that “…the macroeconomics profession has not produced a reliable quantitative (or even qualitative) model of Phillips-curve or aggregate-supply behavior. In other words, there is very little basis for any predictions concerning the way in which quarterly or annual changes in nominal GNP will be divided between real output growth and inflation.”
The advantage of such a strategy in the current environment is that one does not end up making a big bet on a particular rate of real output or productivity growth. A nominal GDP growth target of 5%, for example, would be enough to prevent deflation for productivity growth rates as high as 4% (assuming that the labor force continues to grow at roughly 1%). And if the productivity growth surge turned out to be temporary, and productivity growth headed back to 1%, we would have no more than 3% inflation. Thus, a strategy of responding to nominal GDP growth appears likely to be robust to (a reasonable amount of) uncertainty about long-run real output trends.
It is instructive to see how a policy that responds to nominal GDP growth would compare to one that puts most or all of the weight on real GDP in an environment where there has been an unperceived increase in productivity growth. As discussed above, a policymaker following the latter strategy would be likely to read the resulting increase in output growth as suggesting a need for tighter policy. But this response ignores the behavior of inflation, which is likely to fall as a result of the productivity increase. Looking at nominal GDP automatically internalizes both effects, and keeps policy from tightening too much.
Higher productivity growth could lead to an upward shift in demand as well. For instance, firms might like to increase investment substantially in order to attain a higher desired capital stock quickly. Or households might wish to increase consumption, in response to a perceived increase in permanent (or long-run) income, even before the expected increase in the economy’s productive capacity actually takes place. Policymakers could easily respond to the resulting increase in aggregate demand by tightening policy.
This way of responding to nominal GDP growth is similar to the way policymakers responded to money growth before velocity shifts became a problem. The most well-known example is during the period 1979-1982, when fast money growth, for instance, was interpreted as evidence that the funds rate was being held too low. So the suggestion of monitoring nominal GDP growth is not that radical; it is one way of responding to the loss of indicator variables (i.e., the monetary aggregates) caused by the recent instability in velocity.
The existence of a supply shock makes it hard to judge inflationary risk by looking at real output growth, since such shocks tend to change the output-inflation mix in the economy. One response that is robust to the resulting uncertainty is to pay more attention to the growth in nominal GDP (or spending). As supply shocks change output and inflation growth in opposite directions, they will obviously have a smaller impact on nominal GDP.
Keeping an eye on spending would allow policymakers to keep inflation within reasonable bounds, even if they were unsure about the economy’s potential growth rate. Using nominal GDP in this way–as an indicator– is similar to the way that monetary aggregates were employed in the past, before velocity shifts made them hard to interpret.
Hall, Robert E. 1983. “Macroeconomic Policy under Structural Change.” In Industrial Change and Public Policy. Federal Reserve Bank of Kansas City.
McCallum, Bennett T. 1988. “Robustness Properties of a Rule for Monetary Policy.” Carnegie-Rochester Conference Series on Public Policy 29, pp. 173-204.
Orphanides, Athanasios. 1999. “The Quest for Prosperity without Inflation.” Mimeo. Board of Governors of the Federal Reserve System.
Taylor, John B. 1993. “Discretion versus Policy Rules in Practice.” Carnegie-Rochester Conference Series on Public Policy 39, pp. 195-214.
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