The Price-change Statistics We’ve Weighted For 
Price stickiness is a key feature of New Keynesian macroeconomic models, with a higher degree of price stickiness implying larger monetary non-neutrality—that is, a greater pass-through of nominal shocks to real output. Previous research measuring price-change distributions using US consumer price index (CPI) microdata has kept the consumption basket fixed, considering only time variation in product-level price-change statistics. However, several studies document that shifts in expenditure weights—due to demographic changes, changes in product networks, and changes in industrial composition—have likely affected the level of price stickiness. This paper evaluates price stickiness from 1978 to 2024 while allowing for variation in both price-change statistics and expenditure weights at a granular product level, using the same expenditure weights employed by the Bureau of Labor Statistics in computing the CPI.
Key Findings
- When the measurement incorporates time variation in expenditure weights from 1978 to 2024, the average frequency of price changes remains very similar to its level 50 years earlier, contrasting with the fixed-weights approach, which yields a higher price-change frequency for 2024 compared with 1978.
- Decomposition analysis reveals that this difference arises because, while the “within” effect (time variation in individual product frequencies) raises the frequency distribution as product-level prices become more flexible over time, the “between” effect (time variation in expenditure weights) largely counteracts this increase as the composition of the consumption basket shifts toward products with relatively low price-change frequencies.
- Services excluding travel is the most important sector driving the between effect throughout the sample period. Short-term fluctuations in the between effect are driven by specific sectors: the decline in relative expenditures on gas and used cars in the late 1980s, increased gas expenditures in the early 2000s, and reduced relative expenditures on gas and lodging after the Great Recession, with additional effects from expenditure shifts induced by the COVID-19 pandemic.
- The between effect also reduces the absolute size of price changes but is counterbalanced by a positive cross effect due to co-movement between changes in the absolute sizes of price changes and product weights, resulting in a relatively muted net impact of weights on the size distribution.
Implications
The study demonstrates that changes in expenditure weights are economically meaningful when measuring monetary non-neutrality. Fixing expenditure weights overstates the decline in monetary non-neutrality after 1978 by approximately 20 to 25 percentage points across multiple measurement approaches, including sufficient statistics for the cumulative output response to monetary policy shocks and Phillips-curve slopes from calibrated multisector models. These results indicate that monetary policy transmission remains more potent than implied by fixed-weight analyses, as the economy’s shift toward stickier-price products has offset the tendency of individual products to adjust prices more frequently. For policymakers and researchers, the findings underscore the necessity of incorporating time-varying expenditure weights to accurately assess monetary policy effectiveness, with the authors’ granular data set of monthly price-change statistics and expenditure weights from 1978 to 2024 providing a valuable resource for future analysis.
Abstract
The real effects of monetary policy depend on price stickiness. Existing studies that measure aggregate stickiness using US consumer price index microdata hold the consumption basket fixed. This yields a lower level of stickiness in 2024 compared with 1978. We show instead that stickiness is unchanged. Although individual products now change prices more frequently, the effect is largely offset by shifts in consumer spending, notably toward services with stickier prices. These consumption-basket shifts reduce the estimated decline in monetary non-neutrality by 25 percentage points, suggesting that monetary policy remains far more effective than methods used in existing studies imply.