Evaluating a Calibration Weighting Scheme for Elementary Indexes for Commodities and Services in the U.S. Consumer Price Index

Sylvia G. Leaver, Robert A. Cage, and Darin T. Solk


The estimate of price change for an elementary cell of the U.S. CPI is a weighted average of quote-level price changes, where the quote-level weight is a function of inverse selection probabilities at each stage of selection. This paper evaluates, by means of retrospective estimation and sample simulation, an alternative weighting scheme employing calibration. The theory behind this approach is that the sum of the weights of each "useable" quote in each elementary cell should yield an estimate of the expenditure of the elementary cell. The CPI produces two distinct estimates of elementary item-area expenditure: (i) that derived from the sum of commodities and services quote weights, and (ii) that derived from Consumer Expenditure Survey (CEX) biennial data. The new formula forces quote weights in any given month to sum to biennial expenditure estimates from the CEX.