Recession fears are rising:
source: Google Trends, accessed April 30, 2022.
What do we know, using a metric with some predictive content? Using an ordinary probability probability model estimated during 1986-2022M04, 12-month time horizon:
figure 1: Specify the recession probability for the month, using the 10-year to 3-month spread (blue) and the 10-year to 2-year spread (brown). Recession dates as defined by NBER are shaded from peak to trough in gray. Source: Treasury via FRED, NBER, author’s calculations.
These are the spreads.
figure 2: 10yr-3mo Treasury spread (blue), 10yr-2yr spread (brown), both expressed as a percentage. Recession dates as defined by NBER are shaded from peak to trough in gray. Source: Treasury via FRED, NBER.
Using a 50% threshold seems natural, but using 10 years to 3 months would miss the 1990 recession. 30% between 10 and 2 years seems like a reasonable threshold, although one would like to have a formal assessment of false positives and false positives (ie AUROC). Using a threshold of 30% means that there will be no recession. The 20% threshold could be defended, although then one would have to say there were false positives in 1999 and that in 2020 a recession would have happened even without the coronavirus.
The horizon I’m looking for is 12 months. Deutsche Bank’s forecast is late 2023. Here is the 10-year to 2-year 18-month forecast (light blue) and the 12-month forecast:
image 3: Recession probability for the specified month, using the 10-2 year spread over the next 12 months (brown) and the next 18 months (sky blue). Recession dates as defined by NBER are shaded from peak to trough in gray. Source: Treasury via FRED, NBER, author’s calculations.
This spread predicts a 21% chance of a recession in October 2023.
