Pre-COVID trend regression (2010–2019 baseline)

CPI All Items vs Pre-COVID Trend

Log CPI for all items plotted against the trend line fit through 2010–2019, showing the extent of the price-level gap that opened during the pandemic.

CPI Goods vs Pre-COVID Trend

Goods prices surged well above trend through 2021–2022 as supply chains seized and durable goods demand was pulled forward.

CPI Services vs Pre-COVID Trend

Services inflation was slower to lift off than goods but has proven more persistent, reflecting the labor-intensity of the sector.

Retail Sales vs Pre-COVID Trend

Nominal retail sales relative to the pre-pandemic trend — the fiscal stimulus effect is visible as a sharp break from the 2010–2019 trajectory.

Average Hourly Earnings vs Pre-COVID Trend

Nominal hourly earnings against the 2010–2019 trend — wages have largely reclaimed their pre-pandemic path, though the timing differed across sectors.

Long-term baseline (2000–2019)

CPI Goods — Long-Term Baseline (2000–2019)

A longer trend fit through the full 2000–2019 period captures the secular disinflationary drift in goods prices before the pandemic break.

CPI Services — Long-Term Baseline (2000–2019)

Services inflation has run at a more stable positive trend over two decades — the post-pandemic overshoot is visible against this longer baseline.

Filters and cycles

Hodrick-Prescott Filter Trend

The HP filter extracts a smooth trend from log CPI, separating the long-run price level from cyclical fluctuations using a smoothing parameter of 129,600.

HP Trend vs Raw CPI — YoY % Change

Comparing the year-over-year growth rate of the HP-filtered trend against the raw CPI — divergences reveal periods where inflation is running above or below its structural pace.

Christiano-Fitzgerald Cycles (1975–Present)

The CF bandpass filter isolates the business-cycle frequency component of CPI, goods, and services — cycles of 18 to 96 months — stripping out both the long-run trend and high-frequency noise.