Sources
What the filter is doing
Both columns come from BLS via FRED. They differ only in whether seasonal variation has been removed.
X-13ARIMA-SEATS decomposes a series into seasonal, trend, and irregular components, then publishes the series with the seasonal component subtracted. For CPI, the seasonal component captures predictable annual swings — energy demand, holiday food prices, apparel cycles, school-year-driven services demand. The headlines you read ("CPI rose 0.3% in March") almost always cite the seasonally adjusted column.
- The raw column is what consumers actually paid that month.
- The filtered column is what the same month would look like if every month had the same seasonal mix.
- The Δ column is the size of the seasonal adjustment for that observation. For CPI it is typically a fraction of an index point, but the cumulative effect on year-over-year inflation calls is meaningful.
Observations
Loading…
Raw JSON: cpi.json