Every quarter since 2015, FOMC participants have published a median forecast for where unemployment and core inflation will land — one, two, three years out. Here's how those forecasts stacked up against what actually happened, vintage by vintage.
Unemployment rate
Each line is one SEP's median projection path, made at a single meeting. The red line is what actually happened (Q4 average, UNRATE).
SEP vintage, Sep 2015 → Sep 2026 (older → newer)
Actual (realized)
Hover a line to trace one vintage · hover the chart to compare all forecasts for a year
Core PCE inflation
Same read, for core PCE (Q4/Q4 % change, PCEPILFE) — the Fed's preferred inflation gauge.
SEP vintage, Sep 2015 → Sep 2026 (older → newer)
Actual (realized)
Hover a line to trace one vintage · hover the chart to compare all forecasts for a year
Accuracy by forecast horizon
Bias = actual − median forecast, averaged across all vintages at that horizon. Positive means the Fed underestimated; negative means it overestimated. MAE is the average size of the miss, regardless of direction.