UBS Economics-Global Economic Perspectives _US recession probability stab...-117585922

ab2 September 2025Global ResearchGlobal Economic PerspectivesUS recession probability: stable but elevatedUS hard data in July overall weak, but stable: recprob at 93%With the BEA's monthly personal income (and personal consumption expenditures) release, we have all the information in hand for a full hard data picture for July, as seen through the lens of our leading indicator factor model (see here for details). Recall that the factor extracts a signal net of surveys, PMIs and financial data...just the "hard" data. Accordingly, the data flow is such that the first 'hard' item for August is the employment situation report released this coming Friday. In July, the hard factor moved sideways, at roughly the same level since May and clearly in negative territory (Figures 1 and 2). Cumulating the factor leads to a hard data cycle that continues to contract (Figure 3). The recovery in the factor at the end of last year held the promise of a new dawn after more than two years of contraction, likely the sign of a successful soft-landing. However, the tip back into negative territory since February prompted the turning point algorithm to revise away the fledgling expansion. In turn, the leg down in the cycle this year now translates into a recession probability of 93% (Figure 4). That has been stable since May, when the incoming data was weakening, and sits at historically worrying levels given the signal's track record of identifying NBER turning points. That said, Figure 5 shows that most contributions to the factor were negative, but in a case of 'mile wide inch deep' malaise. No leading hard series is yet showing signs of rapid unravelling (for which we would expect negative contributions closer to 1 standard deviation right ahead of a recession). Soggy, soft, weak, yes, but not collapsing.US yield curve, partially inverted front end, but also stable: recprob at 23%We also track recession signals from the Treasury yield curve, specifically the share of the curve out to 10 years that displays a negative slope (see the reference above). Figure 6 shows how the signal historically leads NBER recessions. The average monthly read for July suggests 23% of the curve is inverted, stable since May like the hard data signal, but up roughly 20 pp up since January's yield curve reading. The more timely daily data suggests the signal remained level throughout August (up to last week, Figure 7).Credit stress builds up slowly but surely: recprob at 41%The US credit team's recession model relies on aggregate corporate credit metrics: leverage, interest coverage, and non-performing loans as well as the Senior Loan Officers Survey. The signal has moved up from 23% at the start of the year to 41% through Q2, driven mainly be declining interest cover as net interest expense rises and domestic non-financial profit growth remains in the low single digits. Putting all the pieced together: aggregate recession probability at 52%Figure 8 shows how the dynamics of the credit