Reading it. Each ISM series is averaged over 3 months, then weighted 25% manufacturing and 75% services, roughly their shares of the economy. Contraction: weighted headline below 50. Inflationary boom: headline above 50 while employment is below 50 and prices paid sit at or above the 80th percentile of their own history. Otherwise Expansion. New orders rising or falling is the momentum read. Source: ISM.

Reading it. Claims are averaged by month; the level match compares today's initial and continuing claims, scaled by payrolls, with past months to imply a U3 range. The cross-check pairs claims with the S&P 500's YoY direction. The 2020 spike is clipped.

Reading it. Bars below zero push unemployment down; bars above push it up. Employment is demand; population and participation are supply. January population-control breaks are removed.

Reading it. Spending leads jobs. Real retail sales flat to negative for several months has preceded recessions; spending well above real aggregate payrolls is being financed by savings or wealth. A saving rate turning up from its low has come a few months before recessions (exceptions 1973–74 and COVID).

Reading it. Headline minus core shows how much of inflation is energy and food. Core 3-month annualized above the 12-month rate means accelerating. Real wages are nonsupervisory wage growth minus CPI.

Reading it. Gas up 40%+ YoY has been stagflationary; oil up 80%+ YoY recession-grade. Affordability is the sturdier test: spring 2026 cleared the YoY trigger but not affordability, and no recession followed. Monthly averages throughout.

Reading it. Recessionary when all three are 10%+ below their 3-year high and 10%+ down YoY. Single-family permits turning positive YoY has come within five months of every recession's end in 50+ years; units under construction improving is the last shoe.

Reading it. Housing is usually the first sector to weaken before a recession and the first to recover after one. Each line is a sector's contribution to GDP growth over the last four quarters, measured against that sector's own previous 10 years, so no reading uses hindsight. Below −1 is a drag. A drag matters less when consumers or government are still spending above normal: that cushion is why the 2019 and 2022 drags didn't become recessions. The chart is cut off at ±6 so 2006–08 and the 2020 shutdown don't flatten everything else. Net exports is left out on purpose: it rises when imports fall, which happens in downturns. Tested with no hindsight, the drag came before 6 of the last 9 recessions (0 to 7 quarters ahead) and also fired in 1995, 2018 and 2021 without one, so it's one input, not an alarm on its own. Source: BEA via FRED; method after Leamer (2007).

Reading it. When bond investors demand more than default risk justifies, they're pulling back on risk, and that has tended to come before weaker growth. Calm is the lowest quarter of readings since 1973, Elevated the top 25% (excluding the most extreme 5%, which is Stressed). Housing drag and stressed credit together is the stronger warning. The recession probability is the Federal Reserve Board's own model, shown as published; the Fed revises recent months as company balance sheets come in. Source: Gilchrist and Zakrajšek, Federal Reserve Board.