Live FRED data · refreshed daily
The U.S. economy, at a glance.
Inflation, jobs, growth, rates, and the recession signals economists watch, from the St. Louis Fed’s FRED database, recomputed every morning.
Headline indicators
Latest reading, change, and recent trend. Tap a card to jump to its chart.
Recession & policy signals
Derived from the raw series. Method notes are at the bottom of the page.
History
Shaded bands mark NBER-dated U.S. recessions.
Yield-curve inversions since 1976
Spells of 30+ days with the 10-year yield below the 2-year, and the NBER recession that followed, if any.
Method & data notes
Sahm rule recession indicator
Three-month average of the unemployment rate minus its lowest three-month average over the previous 12 months. Since 1970, a reading of 0.50 percentage points or more has occurred early in every U.S. recession (Claudia Sahm, 2019). Computed here from UNRATE in real time on the current vintage.
Yield curve (10-year minus 2-year)
Daily DGS10 minus DGS2 on matching dates. A negative spread (an inverted curve) has preceded each recession since the late 1970s, typically by 6 to 24 months. Inversion spells shorter than 30 calendar days are ignored in the table to filter noise.
Inflation, growth, and jobs
Inflation is the 12-month percent change in CPI (CPIAUCSL) and core CPI excluding food and energy (CPILFESL); the card also shows core's 3-month annualized pace, which turns faster. Real GDP growth is the quarter-over-quarter change in GDPC1 at a compound annual rate, the headline BEA convention. Job gains are the monthly change in nonfarm payrolls (PAYEMS), with a 3-month average to smooth revisions. The real policy rate is the effective fed funds rate minus core CPI inflation.
Calendar alignment and missing releases
Every lag is matched on the calendar, not by row position, so a skipped release can never cause a comparison against the wrong month. When a single base month is missing, it is interpolated log-linearly from its neighbours for growth-rate math only; the missing month itself stays blank.
Pipeline
A Python job (pandas, requests) pulls 11 FRED series every morning in GitHub Actions, derives the metrics above, and writes two compact JSON files: a small summary that paints this page first, and the full histories behind the charts. Pure transforms are unit-tested with pytest. Older daily Treasury data is thinned to weekly to keep downloads small; the last two years stay daily.