美国经济衰退风险：衰退概率更新

Recession probability update:Sputtering We revisit our recession "tipping points" model followingrecent payroll revisions. Various specifications suggest thatthe US is likely in a stall state, and perhaps for a year or more.With this heightened susceptibility, the model places odds ofa recession within eight quarters at about 50%. Jonathan Millar+1 212 526 4876jonathan.millar@barclays.comBCI, US Marc Giannoni+1 212 526 9373marc.giannoni@barclays.comBCI, US •The latest update of our "tipping points" model suggests that the underlying pace of USgrowth has decelerated to a pace that makes it vulnerable to a recession.Our findings arerobust to various specifications, including those that draw signals from private job growthand those that work from the unemployment rate. This is especially true when we factor inearly estimates of the upcoming payroll survey benchmark revision, which nearly alignsresults. Pooja Sriram+1 212 526 0713pooja.sriram@barclays.comBCI, US Colin Johanson+1 212 526 8536colin.johanson@barclays.comBCI, US •These results would support Chair Powell's argument from Jackson Hole that risk of asharp deterioration in the labor market now outweighs those of high inflation. •Our results draw from a regime-switching model that assesses the probability that the USeconomy is in each of four states (rapid expansion, expansion, stall speed, and recession) andthe probability of transitioning between them. The key feature is the "stall" state, whichidentifies when the economy is susceptible to entering a recession, but it is not a foregoneconclusion. •We apply this approach using two main indicators: one that takes signal from the ratio ofnonfarm payroll employment to the labor force, and one that looks at the measuredunemployment rate (à lathe Sahm Rule). Both give signals about how employment growth isevolving relative to supply. While neither specification suggests that the US is currently inrecession, both indicators imply an elevated probability that the economy is in a stall state.When we factor in our early estimate of the upcoming benchmark revision, probabilitiesrange from 47-90%. •Although our model assumes a stall as necessary for a recession, it alone is notsufficient:theeconomy can transition to a recession, remain in a stall, or return to expansion. Estimatesimply that the probability of experiencing a recession is 33-39% by Q2 2026, and 50-56% byQ2 2027. •Our resultsoffersome support for expectations that the FOMC will pivot to rate cuts inSeptember, on the grounds that dual-mandate risks are becoming more acute for thelabor market than for inflation.Although we do not think a September cut is a foregoneconclusion given potential surprises in the upcoming dataflow, these results support ourcurrent baseline that the FOMC is most likely to deliver 25bp cuts in September andDecember. Note: Unsmoothed probabilities of being in stall state. Shaded regions denote NBER recessions. "Very preliminarybenchmark" estimates use our estimate that the upcoming benchmark revision will mark down monthly private nonfarmjob gains by 75k/m from Apr 24-Mar 25.Source: Barclays Research Recession signals: Job revisions versus slack Since the July employment report, markets have aggressively priced in a resumption of theFOMC easing cycle at the upcoming September meeting. In many ways, this seems reasonable,with incoming estimates and revisions having slowed the pace of headline job gains from athree-month moving average (3mma) of 147k/m at the time of June's report to just 35k/m inJuly. This was jarring in comparison to what markets have become accustomed to, with jobgains having routinely clocked in at about 200k/m through 2019, and seemingly having returnedto a similar trajectory in 2023-24 following pandemic-related swings. Market pricing alsoappears to be factoring in upcoming downward revisions, including this September'spreliminary announcement of the 2025 benchmark revision to the payroll survey jobsestimates. According to available estimates, this preliminary estimate will be unusually large,possibly marking down the aggregate number of jobs in March 2024 by about 950k, whichwould roughly halve the monthly pace of job gains from April 2024-March 2025, which currentlystands at nearly 150k/m. Job revisions are not a reliable signal of recession risks Could these revisions provide information about recession risks that might pull the Fed in fromthe sidelines? Markets seem to think so, reflecting a general perception that sharp downwardrevisions to NFP estimates — especially from benchmark revisions — tend to coincide witheconomic turning points. To be sure, this was true in theaftermathof the Great Recession, whenlarge revisions marked down sample-based estimates of job gains in 2008 and 2009 by a total of about 1.3mn.1That being said, we think perception is inconsistent with experience, as there isno reliable correspondence between large benchmark revisions and recessions since the BLSbegan