The FRED® Blog

Modeling recession forecasts

New insights from the Research Division

The FRED graph above shows a data series that has been featured in recent FRED Blog posts: ”Dating a recession,” “Are we in a recession (yet)?” and “Assessing recession probabilities.”

Each data point represents the probability of the US economy being in a recession during the preceding month. In other words, these are backward-looking probabilities. These probabilities can range between 0 (complete confidence the economy is expanding) and 100 (complete confidence the economy is contracting).

As of October 2023, the latest observation available at the time of this writing, the data signaled a 2.2% probability the US economy was in recession during September 2023.

But what about forward-looking probabilities? Organizations of professional forecasters such as Consensus Economics synthesize available economic data to estimate the likelihood of an economic downturn occurring in the near future. Recent research from Christopher Neely at the St. Louis Fed investigates what variables that organization appears to use to predict the probability of recession occurring in the next 12 months.

Neely finds that although 10 economic variables are useful when forecasting recessions, they do not explain the Consensus Economics estimated probability of recession very well. Moreover, Treasury yield spreads are not among those best predictors, though they have a well-established record in predicting recessions. You can learn more about forecasting from yield spreads.

For more about this and other research, visit the website of the Research Division of the Federal Reserve Bank of St Louis, which offers an array of economic analysis and expertise provided by our staff.

How this graph wase created: Search FRED for and select “Smoothed U.S. Recession Probabilities.”

Suggested by Diego Mendez-Carbajo.