Can the Term Spread Predict Output Growth and Recessions? A Survey of the Literature
Wheelock, David C., Wohar, Mark E., Review - Federal Reserve Bank of St. Louis
This article surveys recent research on the usefulness of the term spread (i.e., the difference between the yields on long-term and short-term Treasury securities) for predicting changes in economic activity. Most studies use linear regression techniques to forecast changes in output or dichotomous choice models to forecast recessions. Others use time-varying parameter models, such as Markov-switching models and smooth transition models, to account for structural changes or other nonlinearities. Many studies find that the term spread predicts output growth and recessions up to one year in advance, but several also find its usefulness varies across countries and over time. In particular, many studies find that the ability of the term spread to forecast output growth has diminished in recent years, although it remains a reliable predictor of recessions.
(JEL C53, E37, E43)
Federal Reserve Bank of St. Louis Review, September/October 2009, 91(5, Part 1), pp. 419-40.
Information about a country's future economic activity is important to consumers, investors, and policymakers. Since Kessel (1965) first discussed how the term structure of interest rates varies with the business cycle, many studies have examined whether the term structure is useful for predicting various measures of economic activity. The term spread (the difference between the yields on long-term and short-term Treasury securities) has been found useful for forecasting such variables as output growth, inflation, industrial production, consumption, and recessions, and the ability of the spread to predict economic activity has become something of a "stylized fact" among macroeconomists.
This article surveys recent research investigating the ability of the term spread to forecast output growth and recessions.1 The article briefly discusses theoretical explanations for why the spread might predict future economic activity and then surveys empirical studies that investigate how well the spread predicts output growth and recessions. The survey describes the data and methods used in various studies to investigate the predictive power of the term spread, as well as key findings. In general, the literature has not reached a consensus about how well the term spread predicts output growth. Although many studies do find that the spread predicts output growth at one-year horizons, studies also find considerable variation across countries and over time. In particular, many studies find that the abil- ity of the spread to forecast output growth has declined since the mid-1980s. The empirical lit- erature provides more consistent evidence that the term spread is useful for predicting recessions. Furthermore, the relationship appears robust to the inclusion of other variables and nonlinearities in the forecasting model.
A LOOK AT THE DATA
Yields on long-term securities typically exceed those on otherwise comparable short-term securities, reflecting the preference of most investors to hold instruments with shorter maturities. Hence, the yield curve, which is a plot of the yields on otherwise comparable securities of different maturities, is typically upward sloping. Analysts have long noted, however, that most recessions are preceded by a sharp decline in the slope of the yield curve and frequently by an inversion of the yield curve (i.e., by short-term yields rising above those on long-term securities).
Figure 1 shows the difference between the yields on 10-year and 3-month U.S. Treasury securities for 1953-2008. The shaded regions indicate recession periods as defined by the National Bureau of Economic Research.2 As Figure 1 shows, every U.S. recession since 1953 was preceded by a large decline in the yield on 10-year Treasury securities relative to the yield on 3 -month Treasury securities, and several recessions were preceded by an inversion of the yield curve. Moreover, the only occasion when the 3 -month Treasury security yield exceeded the (constant-maturity) 10-year Treasury yield without a subsequent recession was in December 1966. …