Axiomatic and Economic Approaches to Elementary Price Indexes

In a 1993 paper, Marshall Reinsdorf finds that the CPI components for food and gas were biased upward by about 2% and 1% per year respectively during the 1980s. He attributes this result to outlet substitution bias. The more recent paper by Reinsdorf and Moulton [1994] presents an alternative explanation for Reinsdorf's earlier results: when the BLS moved to probability sampling of prices in 1978, the micro price quotations were aggregated together using an index number formula that generates an upward bias. This paper further explores the central theoretical issue raised by the Reinsdorf-Moulton paper: the choice of an index number formula to aggregate prices at the finest level of disaggregation. This issue is examined from both axiomatic and economic perspectives. This paper also reviews the empirical literature on alternative elementary price indexes, and the recent literature on sources of bias in consumer price indexes. The findings of this paper in conjunction with the empirical work of Reinsdorf and Moulton yield a number of recommendations for Statistical Agencies which are outlined in the final section.