In 2003, econometrician Clive Granger, along with econometrician

Robert Engle, received the Nobel Prize in economics. Granger’s award was “for methods of analyzing economic time series with common trends (cointegration).”

Trained in statistics, Granger specializes in the behavior of time-series data (i.e., data that are recorded in calendar sequence, annually or at shorter or longer intervals). Early in his career, the best-developed statistics assumed that time series were stationary—that is, that they tended to vary randomly around a common long-run mean (or average) value or around a nonrandom trend. Many economic time series, however, appear to be nonstationary—to follow processes related to the random walk. The term “random walk” is suggested by the metaphor of a drunken man stumbling in the street—just as likely to go one way as another. A time series is a random walk when the next period’s value is as likely to be higher as it is to be lower, so that the best forecast of the next period’s value is just whatever today’s value happens to be.

For lack of better techniques, economists often applied statistics designed for stationary data to nonstationary data. In 1974, Granger and coauthor Paul Newbold, building on the 1920s work of the English statistician G. Udny Yule, showed that pairs of nonstationary time series could frequently display highly significant correlations when there was no causal connection between them. For example, the U.S. federal debt and the number of deaths due to AIDS between 1981 and 2000 are highly correlated but are clearly not causally connected. Such “nonsense correlations” called into question the meaningfulness of many econometric studies.

Short-run changes in time series are frequently stationary, even when the time series themselves are nonstationary in the long run. So one strategy in the face of nonstationary data was to study only short-run changes. But Granger (working with Engle) realized that such a strategy threw away valuable information. Not all long-run associations between nonstationary time series are nonsense. Suppose that the randomly walking drunk has a faithful (and sober) friend who follows him down the street from a safe distance to make sure he does not injure himself. Because he is following the drunk, the friend, viewed in isolation, also appears to follow a random walk, yet his path is not aimless; it is largely predictable, conditional on knowing where the drunk is. Granger and Engle coined the term “cointegration” to describe the genuine relationship between two nonstationary time series. Time series are “cointegrated” when the difference between them is itself stationary—the friend never gets too far away from the drunk, but, on average, stays a constant distance back.

Many economic time series are nonstationary. For example, over long periods, federal revenues and spending appear to be nonstationary, but they also appear to be cointegrated, in the sense that when they are far out of line, they tend to be drawn back into close proximity. Granger developed econometric methods for testing whether the relationships among these time series were genuine cointegrating relationships or nonsense, and for correctly estimating the genuine relationships.

In addition to his work on cointegration, Granger is famous for his earlier development of the concept of Granger causality, an idea with roots in the work of the mathematician Norbert Wiener. The current value of a time series is often predictable from its own past values. For example, GDP this quarter is imperfectly predicted from information about GDP over the past few years. A second time series is said to “Granger-cause” another if its past values improve the prediction one would get just from the past values of the first time series. Granger causality is related to cointegration. Granger and Engle demonstrated that when two variables are cointegrated, then at least one of them must Granger-cause the other. The first important application of Granger-causality to economics appears in a 1972 article by Christopher Sims in which he showed that money Granger-causes nominal GNP, apparently bolstering the monetarist idea that fluctuations in money are the major cause of business cycles (see monetarism and milton friedman). In the debate that followed, the limits of Granger-causality were clarified: the concept concerns predictability and not control, so that a finding that money Granger-causes GNP does not imply that the Federal Reserve has an effective instrument to steer the economy. While Granger himself had referred simply to “causality,” the adjective “Granger” is now always attached to his idea to distinguish it from causality based on control.

Granger was born in Wales. He attended the University of Nottingham, where he earned a B.A. in mathematics and economics in 1955 and a Ph.D. in statistics in 1959. He was on the faculty of the University of Nottingham until 1973, with occasional visiting positions at other universities. In 1973, he became a professor at the University of California at San Diego. He retired in 2003, just two months before winning the Nobel Prize. He was knighted in 2005. In reminiscing about his childhood, Sir Clive wrote, “A teacher told my mother that ‘I would never become successful,’ which illustrates the difficulty of long-run forecasting on inadequate data.”

## About the Author

Kevin D. Hoover is professor in the departments of economics and philosophy at Duke University. He is past president of the History of Economics Society, past chairman of the International Network for Economic Method, and editor of the *Journal of Economic Methodology.*

## Selected Works

1969. “Investigating Causal Relations by Econometric Models and Cross-Spectral Methods.” *Econometrica* 37: 424–438.

1974 (with Paul Newbold). “Spurious Regressions in Econometrics.” *Journal of Econometrics* 2: 111–120.

1981. “Some Properties of Time Series Data and Their Use in Econometric Model Specification.” *Journal of Econometrics* 16: 121–130.

1987 (with Robert Engle). “Co-integration and Error-Correction: Representation, Estimation and Testing.” *Econometrica* 55: 251–276.

## Footnotes