A Critique of the Use of BLS Data for Market Structure‐Price Analysis

Abstract

Listen

Translate article

Food prices, and the factors affecting them, are subject to continuing examination by academicians and policy makers. As in most areas of economics, analysis is dependent on the available data. The Bureau of Labor Statistics (BLS) provides in Estimated Retail Food Prices by City the only publicly available and ongoing data set on retail food prices. These data are widely used (and misused) in economic analysis. A recent study (Grinnell, Crawford, Feaster) that used the BLS data received considerable attention at hearings on prices and profits in food retailing conducted in the spring of 1977 by the Joint Economic Committee. Ignoring disclaimers by the study's authors, industry representatives cited the study as evidence that market concentration and retail food prices are unrelated and once again aroused interest in the validity of using BLS price data for cross-sectional market structuremarket price analysis.' It should be noted at the outset that BLS price data were never intended to answer the questions of industrial organization economists. BLS price data for food consumed at home attempt to reflect the average food costs of consumers in different metropolitan areas and the nation. Differences among city average prices may be due to differences in competitive environments, the products consumed, or the size and type of stores. All are possible causes. Determining the extent to which price differences are due to the different factors, a primary concern of industrial organization economists, is impossible with present BLS data. In examining the performance of markets, both operational efficiency (i.e., cost per unit of output) and allocative efficiency (i.e., the equivalence of prices and costs) are important and rather traditional dimensions of performance considered. Available data indicate that operational efficiency in supermarkets is much greater than in convenience and specialty stores. For entire markets, the great r the importance of supermarkets, the lower the expected cost of the retailing function. Analysis of allocative efficiency in retail food markets is particularly concerned about questions such as (a) Does Safeway, A&P, or other supermarket chains charge lower (higher) prices in their st res in competitively structured markets than in n ncompetitively structured markets? (b) Are average supermarket prices in competitively structured markets lower (higher) than in noncompetitively structured markets? Data to examine these questions are generally not available. This note considers the relevance of BLS price data for such analyses. That is, if store size and type is held approximately constant by examining only supermarkets, how are food prices affected by market concentration, barriers to entry, product (or enterprise) differentiation, and differences in factor costs? When used as an indicator of allocative efficiency across markets, BLS data have two major shortcomings. This paper discusses the nature and i plications of these shortcomings, but is not written as a criticism of BLS, because this agency has e phasized that its price data are collected for comparison of prices over time, not across markets (Rothwell). In developing retail food price data, BLS follows two procedures which would be expected to lead to an inverse relationship between market concentrati n and BLS market average prices. Briefly these are: (a) The brand of individual products selected for pricing is based upon volume and, hence, is allowed to vary from store to store and from market to market. The lack of comparability that results is particularly serious when private label products are selected in some cases and national brand in others. Lower priced private label products tend to be included more frequently in BLS calculations in concentrated than in unconcentrated markets. (b) BLS market average prices reflect the prices of all types of food stores in a market, weighted approximately in accord with their market share. However, if allocative efficiency is the performance dimension of interest, it is inappropriate to combine the prices of different types of stores into a market average. Because supermarkets represented 67% of all food store sales and 72% of all grocery store sales in 1972 and tend to set the competitive tone in most Frederick E. Geithman and Bruce W. Marion are research specialist and professor, respectively, Department of Agricultural Economics, University of Wisconsin. The helpful comments of Willard F. Mueller and Russell C. Parker on earlier versions of this paper are gratefully acknowledged by the authors. An additional BLS data set, the Urban Intermediate Food Budget, was introduced at the recent Joint Economic Committee hearings by food chain spokesmen as evidence that prices in local markets are not correlated with market concentration. The method of calculation has similar brand and outlet biases as BLS food price data and is plagued by additional limitations as well (Sherwood). All in all, the food budget data are as objectionable for use in market structure-price analysis as the currently published BLS food prices.