Every US recession in the past 50 years was preceded by a negative yield spread; there were no recessions without one.

True: Treasury yield-curve data (e.g. the 10-year minus 3-month or 10-year minus 2-year spread) together with NBER recession dates show that every US recession from about 1973 to January 2023 was preceded by an inverted (negative) spread, with no recession lacking one. Mostly true: the pattern holds, with a minor qualification such as the 2020 recession being preceded by an inversion but caused by COVID, or a spread that only briefly touched zero. Mixed: the result depends heavily on which spread measure is used. Mostly false: one or more recessions in the period had no prior inversion on standard measures, though most did. False: several recessions had no prior inversion.

Set before the research began.

1. E1 The Inverted Yield Curve and Next US Recession Independent Institute · 10 July 2023

   “Negative spreads preceded all recessions, regardless of whether yields generally were high or low, and regardless of whether inversion resulted from short-term (bill) yields rising above long-term (bond) yields”

2. E2 The Inverted Yield Curve and Next US Recession Independent Institute · 10 July 2023

   “The yield curve spread that most accurately forecasts recessions is that between the 10-year Treasury bond yield and the 3-month Treasury bill rate.”

3. E3 US Business Cycle Expansions and Contractions NBER · 14 March 2023

   “Recessions — contractions in economic activity — start in the month after a peak in the business cycle, and end in the month of the trough.”

E1, an independent analysis from July 2023, states that negative yield spreads preceded all recessions, whether yields were high or low overall. E2 shows that the same analysis treats the 10-year minus 3-month Treasury spread, one of the measures the criteria name, as the best recession forecaster. E3 gives NBER's definition of recession timing but no list of dates. The quoted text does not say which years E1 covers, and no Treasury data or NBER recession dates were provided to check the pattern directly. That limits the verdict to mostly true rather than true.