A Model-based Approach to Variance Estimation for Fixed Weights and Chained Price Indices

Abstract

There is presented a model-based approach to variance estimation for both short-term fixed weights and long-term chained price indices. The individual price observations are treated as random variables whose probability distribution in part depends on some unknown index parameters that are postulated as the underlying theoretical inflation rates. The variance estimate summarizes the amount of variation in the potentially noisy price signals, providing an intuitive measure of uncertainty surrounding the estimated average price development.