FRBSF Economic Letter

1999-18 | May 28, 1999

Output and Inflation: A 100-Year Perspective

Kevin J. Lansing and Jeffrey Thalhammer

While economists generally accept that monetary policy can influence nominal variables such as the price level and inflation, they continue to debate the relationship between monetary policy and real variables such as the unemployment rate and real GDP. During the early 1960s, many economists and policymakers believed that policy could exploit a stable trade-off between inflation and real economic activity. One version of the hypothesized trade-off, originally described by A.W. Phillips (1958) using U.K. data from 1861-1957, implied that policymakers could permanently lower unemployment by generating higher inflation. Some years later, economists Edmund Phelps (1967) and Milton Friedman (1968) argued persuasively that any such trade-off was bound to be short-lived: once people came to expect the higher inflation, monetary policy could not keep unemployment below its long-run equilibrium or “natural” level. This claim was later borne out by the experience of the 1970s, when rising U.S. inflation did not bring about the lower unemployment rates promised by the Phillips curve. On the contrary, higher inflation coincided with higher unemployment—a combination that became known as “stagflation.”

In the late 1990s, the situation is precisely the reverse—the U.S. economy exhibits low inflation combined with low unemployment. Some commentators seem to view this combination as a puzzle or breakdown in the traditional relationship between inflation and real economic activity. This Economic Letter challenges such a view by putting the recent data into a 100-year historical perspective.

Stabilization policy

Validation of the Phelps-Friedman argument by the experience of the 1970s still left open the possibility that policymakers might exploit a short-term or transitory link between inflation and real activity to smooth business cycle fluctuations. For example, the Fed might pursue an expansionary monetary policy to stimulate aggregate demand when real activity was “too low” and adopt a contractionary monetary policy to reduce aggregate demand when real activity was “too high”—all the time recognizing that no permanent trade-off between inflation and real activity exists. Indeed, one can see how this principle might be used nowadays to guide the Fed in pursuit of its legally mandated goals to promote “maximum” employment and “stable” prices (see Federal Reserve Bank of San Francisco, 1999).

Efforts to smooth business cycles using monetary policy face some difficult problems. First, history has shown that monetary policy affects the economy only through long and variable lags. Second, a policy of “leaning against the wind” requires policymakers to decide in real time (using preliminary data) whether economic activity is “too low” or “too high” relative to some benchmark. Benchmarks such as the natural rate of unemployment or the long-run trend of real GDP cannot be observed directly, however. These must be estimated from available data by combining economic theory, statistical analysis, and sound judgment.

Estimating the long-run trend in real output

One very simple way to estimate the long-run trend of real GDP per person is to fit a straight line through the data. This would be appropriate if one believes that technological change occurs at a constant rate, while transitory shocks generate fluctuations around the long-run trend. Many economists believe, however, that technological change occurs unevenly and is influenced by many factors, such as the training and education of the workforce, the amount of resources devoted to inventive activity, and tax policy, to name a few. If this were true, then the assumption of an invariant long-run trend could lead one to the mistaken conclusion that observed movements in real GDP are due to transitory shocks pushing the economy away from trend (thus calling for a monetary policy response) when in fact the long-run trend itself has shifted. To avoid such a mistake, it is desirable to employ a trend measure that is capable of accounting for shifts in the long-run trend of real GDP. One way this might be done, for example, would be to fit a piecewise straight line through successive midpoints of business cycle expansions. This approach has a serious drawback, however, because the midpoint date is not known until after the expansion has ended. Policy decisions must be made in real time without the benefit of such hindsight.

John Cochrane (1994) has suggested a way of using consumption data to help account for shifts in the long-run trend of real GDP. Cochrane’s idea is based on the permanent income hypothesis, which says that people’s consumption decisions depend primarily on their “permanent” income, i.e., income that is expected to persist into the future. According to this theory, transitory changes in income do not have much impact on consumption; people use saving and borrowing to maintain a smooth pattern of consumption when hit by transitory income shocks. If we observe a change in income that is not accompanied by a change in consumption, then we can infer that people view the income change as transitory. On the other hand, if we observe simultaneous changes in income and consumption, then we can infer that people view the income change as permanent. Individual consumers have first-hand knowledge regarding the future prospects of their employer (and their employer’s industry) and hence are in good position to judge whether changes in their own income are likely to be permanent or transitory.

A shift in the long-run trend of real GDP will affect people’s permanent income and should thus be accompanied by a change in consumption. By regressing output data on consumption data (both of which are observable in real time), these shifts can be automatically incorporated into a measure of trend output. The resulting “output gap,” i.e., the deviation of real GDP from trend, should provide an informative signal about where the U.S. economy is operating relative to its long-run potential (for additional discussion, see Cogley and Schaan 1995).

Figure 1 plots private real GDP per person (real GDP less government expenditures) for the period 1890-1998 together with a consumption-based trend. This approach places the U.S. economy close to its long-run trend during the late 1990s. The strong growth of real GDP over this period could thus be interpreted as reflecting an upward shift in the economy’s ability to produce goods and services—a shift that people view as permanent based on the strong growth of their consumption. Given this interpretation of the data, the current U.S. output gap is actually quite small in comparison to some other periods of fast economic growth, such as the early to mid-1960s.

Estimating the long-run trend in inflation

Figure 2 plots U.S. inflation over the period 1890-1998. As with real GDP, it is useful to construct a long-run trend of inflation so that short-term or transitory movements can be defined as deviations from the trend. Here we adopt a trailing five-year moving average as our measure of trend inflation, which approximates an exponential smoothing technique. As noted by Cogley (1998), this technique accounts for periodic shifts in what might be viewed as the Fed’s long-run inflation target and has the advantage that the trend can be computed in real time (unlike a centered moving average trend).

The trailing five-year moving average also can be interpreted as a measure of expected inflation. In particular, people are likely to use observations about past inflation to help make predictions about future inflation. According to the Phelps-Friedman argument, only the unexpected component of inflation (the deviation from trend) should influence real economic activity. Figure 2 shows that U.S. inflation is slightly below trend in the late 1990s—a situation that has been labeled a puzzle by some commentators, given the strong growth of real GDP over the same period.

Short-term movements in output and inflation

Figures 3 and 4 plot short-term movements in output and inflation. The “output gap” (private real GDP minus the consumption-based trend) consistently falls during recessions. The “inflation gap” (inflation minus the moving average trend) usually falls during recessions but is sometimes observed to rise during these periods (Figure 3). An example is the 1974 recession when OPEC production cutbacks led to a 68 percent increase in the price of crude oil. Thus, the presence of transitory supply-shocks (which by definition should be separated from shifts in the long-run trend) help to explain why the output and inflation gaps do not always move in the same direction. Notice that the two gap measures appear to be going in opposite directions at the end of our data sample in 1998 (Figure 3). It is worth noting that the price of crude oil dropped by nearly 40 percent in 1998—an event that can be viewed as a favorable transitory supply shock.

Despite periods when the two gap measures move in opposite directions, the correlation between short-term movements in output and inflation is positive for the whole sample period (correlation coefficient of 0.18), the post-WWII sample period (correlation coefficient of 0.20), and the 1917-1946 sample period, which includes two world wars and the Great Depression (correlation coefficient of 0.34). This positive correlation accounts for the “traditional” view of the short-term output-inflation relationship. It also suggests that transitory demand shocks (which cause the two gap measures to move in the same direction) are an important feature of the data.

Data for 1998 place the U.S. economy close to the center of the scatter diagram with an inflation rate that is slightly below the best-fit regression line (Figure 4). Given the statistical uncertainty in the position of the best-fit line and the possibility of a transitory supply shock from falling oil prices, the 1998 output-inflation combination should not be viewed as “puzzle” or a breakdown in the historical relationship.

Kevin Lansing

Jeffrey Thalhammer
Research Associate


Cochrane, John H. 1994. “Permanent and Transitory Components of GNP and Stock Prices.” Quarterly Journal of Economics 109(1), pp. 241-265.

Cogley, Timothy. 1998. “A Simple Adaptive Measure of Core Inflation.” Federal Reserve Bank of San Francisco Working Paper 98-06.

___, and Desiree Schaan. 1995. “Using Consumption to Track Movements in Trend GDP.” FRBSF Weekly Letter 95-28 (September 1).

Farmer, Roger E.A. 1999. Macroeconomics. South-Western College Publishing.

Federal Reserve Bank of San Francisco. 1999. “U.S. Monetary Policy: An Introduction.” FRBSF Economic Letter 99-01 (January 1).

Friedman, Milton. 1968. “The Role of Monetary Policy.” American Economic Review 58, pp. 1-17.

Phelps, Edmund S. 1967. “Phillips Curves, Expectations of Inflation and Optimal Unemployment Over Time.” Economica 34, pp. 254-281.

Phillips, A.W. 1958. “The Relation Between Unemployment and the Rate of Change of Money Wage Rates in the United Kingdom, 1861-1957.” Economica 25, pp. 283-289.

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
Research Library
Attn: Research publications, MS 1140
Federal Reserve Bank of San Francisco
P.O. Box 7702
San Francisco, CA 94120