Improving the Median CPI: Maximal Disaggregation Isn't Necessarily Optimal

For decades, the Federal Reserve Bank of Cleveland (FRBC) has produced the Median Consumer Price Index (CPI). It has proven useful in various contexts, such as forecasting and understanding post-COVID inflation dynamics. Historically, revisions/improvements to the FRBC methodology have involved increasing the level of disaggregation in the CPI components. Thus, it may be reasonable to assume that further disaggregation improves the properties of the median CPI. We theoretically demonstrate: not necessarily. We then empirically explore the impact of further disaggregation by examining fifteen candidate baskets of CPI items that vary by the level of disaggregation. In line with prior literature, we find that greater disaggregation in the shelter indexes improves the ability of the Median CPI to track the medium-term trend in CPI inflation and its predictive power over future CPI movements. In contrast, increasing disaggregation in the remaining components leads to a deterioration in performance. Our preferred Median CPI measure suggests lower trend inflation pre-pandemic, a faster acceleration in trend inflation in 2021, and a faster deceleration in trend inflation after 2022.