Inflation measurement in the presence of stockpiling and smoothing of consumption

A chained price index is said to suffer from chain drift bias if it indicates an overall price change, even though the prices and quantities in the current period have reverted back to their levels of the base period. The empirical relevance of this bias is well documented in studies that apply sub-annual chaining to scanner data. There it is shown that stockpiling can lead to downward chain drift bias. The present paper draws attention to the fact that smoothing consumption causes substantial upward chain drift. In addition, this study introduces a simple utility framework consistent with stockpiling and smoothing. Building on this framework, a "stress test" is conducted that examines whether rolling window variants of multilateral indices (GEKS, TPD, and GK) effectively curtail the chain drift problem.