Here is a recession probability plot from probabilistic regression of 10yr-3mo and 10yr-2yr term spreads, The latest thoughts from Cam Harvey.
figure 1: 10-year to 3-month treasury bond spread (blue, left scale) and 10-year to 2-year spread (tan, left scale), both in %; and 10yr-3mo recession probability (dark blue , right axis), 1970-2023M05 (light blue, right axis), and 10yr-2yr (brown, right axis) during 1986-2023M05. Dates of peak-to-trough recessions as defined by the NBER are shaded in gray. Regression assumes no recession through May 2023. Source: Treasury Department calculations based on FRED, NBER, and authors.
Note that omitting the 2020 pandemic-related recession and subsequent time periods does not change the underlying result; then the probability for May 2024 (using the 1986+ sample) rises from 0.86 to 0.89. The results are also not sensitive to including time trends. The estimated probability does rise significantly when the Chicago Fed’s National Financial Conditions Index is included. Figure 2 shows the details of the above projections along with the FCI expansion series.
figure 2: 10-year to 3-month treasury bond spread (blue, left scale) and 10-year to 2-year spread (tan, left scale), both in %; and 10yr-3mo recession probability (dark blue , right axis), Financial Conditions Index (pink, right axis), 10yr-3mo term spreads from 1970-2023M05 (light blue, right axis) and 10-2 years from 1986-2023M05 (brown, right scale) . Dates of peak-to-trough recessions as defined by the NBER are shaded in gray. Regression assumes no recession through May 2023. Sources: Treasury Department via FRED, Chicago Fed via FRED, NBER and authors’ calculations.
I haven’t checked to see if the foreign term spread a la is included Ahmed Chin (2022)but I believe it raises the implied probability, as in this march post.
This means that if the historical correlation persists, then a recession (of some kind) is likely. (So, it sure makes sense Bill McBride/CR Under recession watch. )
