Canback Dangel calculates weekly the economic recession probability for the United States with an 8 month horizon using the "term spread method."

Predicting recessions used to be notoriously difficult. However, over the past 25 years a new method has evolved that offers good predictive power (Estrella (2005) gives an easy-to-read overview). The term spread method (also called "yield curve method") uses the known information-intensity of the term spread between a long-term bond and a short-term bill. Based on the term spread method, all recessions since 1953 have been accurately forecasted with only one false recession signal (1967). No other method has performed better.

The term spread is used as the independent variable in a probit analysis, with the existence of a recession as the dependent variable. The probit estimates the probability of a recession: if the probability is above 50%, a recession is likely at a chosen time horizon (here, 8 months); if it is less than 50%, a recession is unlikely.

Often, but not necessarily, the spread between the 10-year Treasury bond (TCMNOMY10) and the 3-month Treasury bill (TBM3) is used. This is also the spread used by Canback Dangel. The recession variable is based on the NBER’s definition of a recession. Data are available from April 1953.

Other independent variables can be added. Examples include the S&P 500 return, the composite leading indicator, and consumer sentiment (Filardo 2004). Canback Dangel calculates the recession probability based on such extended models as well, but does not publish them except when they disagree materially with the term spread method (which has yet to happen). In general, the term spread has the highest explanatory and predictive power of all conceivable variables (Bernanke 1990).

The term spread method has been shown to work in other countries as well (Bernard and Gerlach 1996). Canback Dangel calculates probabilities for several countries and provides them as a fee-based service. Likewise, more detailed recession analyses for the U.S. are fee-based.

No predictive method is perfect. While the term spread method is statistically robust in most respects and has given accurate signals so far (1953-2005), it is not guaranteed to work in the future.

Estrella, A. 2005. The yield curve as a leading indicator: Frequently asked questions. Federal Reserve Bank of New York. [Includes an extensive bibliography]

Filardo, A. J. 2004. The 2001 recession: What did recession prediction methods tell us? BIS working paper no. 148. Bank for International Settlements.

Bernanke, B. 1990. On the predictive power of interest rates and interest rate spreads. NBER working paper no. 3486. National Bureau of Economic Research.

Bernard H. and S. Gerlach. 1996. Does the term structure predict recessions? The international evidence. BIS working paper no. 37. Bank for International Settlements.