As July rolls in, it’s interesting to note that the heightened likelihood of a recession in 12 months comes not from interest rate spreads, nor from financial conditions, but from oil prices.
figure 1: Probability regression of 10-2 year spread (blue), 10-3 month spread (green), and 10-2 year spread plus Chicago Fed Financial Conditions Index (tan) leading to a recession in the 12-month range Probability, estimated from June 1976 to July 2022. The NBER uses shades of grey to define the peak and trough dates of the recession. Red horizontal line at 33%. Source: Author calculations based on data in Figure 2.
While the probability of a recession in 2007-23 was higher in the 10-year to March period (due to the fall in this spread), it was still lower than the 28.8% in the 10-year to 2-year period. Notifications including a financial condition index reduce the probability.
figure 2: 10-year-2-year Treasury spread (blue), 10-year-3-month Treasury spread (green), both expressed as a percentage, and the Chicago Fed National Financial Conditions Index (tan). Three-month treasury bonds are secondary market yields, FRED series TB3MS. The NBER uses shades of grey to define the peak and trough dates of the recession. Source: Treasury, Federal Reserve Bank of Chicago, all from FRED, NBER, and author’s calculations.
So why are many forecasters saying the odds of a recession are high.Under certain circumstances foreign yield curve, inflation or unemployment allocation. An important determinant is oil prices. If the relative price of oil (WTI vs core CPI) is added to the 10- to 2-year specification, the red line below is added to the forecast.
image 3: 10-2 year spread (blue), 10-3 month spread (green), 10-2 year spread plus Chicago Fed Financial Conditions Index (tan), 10-2 year spread plus relative oil prices (bold red) probability regression recession probability), 12-month time horizon, estimated 1976M06-2022M07. July WTI oil prices and July core CPI as of 7/23 are nowcasts from the Cleveland Fed. The NBER uses shades of grey to define the peak and trough dates of the recession. Red horizontal line at 33%. Source: Author calculations based on data in Figure 2.
Using a threshold of 33%, the norm will miss the 1990, 2001, and 2020 recessions, but catch others. Of course, a broader search yields a norm that covers other recessions – the 3-month change in nominal oil prices covers 1990, 2001, but misses the 2007 and 2020 recessions (and is close to forecasting 2021) M03 recession).
The bottom line is that term spreads and financial conditions are currently not the reasons most people think a recession is highly likely. Oil prices are the key to pushing probability estimates beyond the threshold.
(by contrast, note that I’m getting over 40% chance of M06 recession by 2020back in 2019, using only 10-year to 3-month term spreads).
