There is something sketchy in the underlying report [0]. In figure 2, they draw what I understand to be a curve of the probability (according to their model) of a recession happening within 12 months. The strange part is that they identify a "critical threshold" (at probability 0.24), and seem to imply that when the prediction goes above that value, a recession happens. That's not how probabilities work. Are they meaning something else, or are they just lost here?
The probability of a recession rises immediately prior to a recession. When the recession occurs, the probability drops immediately back down below the threshold; as the probability of a back-to-back recession is very low and the conditions that caused the recession immediately change.
So the critical threshold is saying "at any point beyond this line, as conditions remain the same, bad things may happen at a very accelerated rate"
I think that's why the lines jump from .24 or .6 to 1.0(recession) and then back below .24.
So as far as I can tell, they're not saying a recession is guaranteed if the probability increases beyond the critical threshold, but are instead saying the probability of a recession increases more quickly up into the point of an actual recession than our ability to reliably predict and update the probabilities that would predict said recession.
[0] https://www.frbsf.org/economic-research/publications/economi...