Do Supermarkets Have Market Power?

Back in 2013, a former graduate school colleague of mine at K-State, Veronica Pozo (now at Utah State) presented a paper on price transmission along the beef supply chain (retail, wholesale, and farm prices). The authors used data from the BLS and retail scanner data to look for asymmetries in the price adjustment process.

On the issue of BLS and scanner data, the authors note some problems with BLS data:

Evidence suggests that the BLS retail price data may be biased. Hausman (2003) showed that the methodology BLS uses to calculate the food Consumer Price Index (CPI) may overestimate the price of food. The omission of random-weight food items (BLS collects only price data, but does not collect quantity data) and supercenter purchases (reflecting shifts in shopping patterns to lower-priced stores) may cause a significant upward bias on price estimates. In addition, BLS data do not account for large volumes sold at discounted prices during retail specials (Rojas et al., 2008; Lensing and Purcell, 2006). Therefore, this issue raises the question of whether findings from previous studies that have used BLS retail price data are reliable.

The authors examine price transmission along the supply chain because it can indicate several problems with the functioning of markets. The authors write:

Several potential explanations have been offered for asymmetric price responses in vertically linked markets, including market power and concentration at processing and retail levels (e.g., Bailey and Brorsen, 1989; Azzam, 1999; Peltzman, 2000; Xia, 2009); adjustment and menu costs (e.g., Bailey and Brorsen, 1989; Levy, 1997); inventory adjustment practices (e.g., Blinder, 1982); government intervention (e.g., Kinnucan and Forker, 1987; Mohanty et al., 1995); consumption inertia (Xia and Li, 2010); and the empirical methodology employed in testing for asymmetry (Miller and Hayenga, 2001). However, a necessary condition for assessing market failure is that the data used to test for asymmetry are adequate (Bailey and Brorsen, 1989; von Cramon-Taubadel, 1998). 

They conclude that there is no asymmetry in price transmission from the farm or wholesale level to the retail level. If prices fall at the farm level, the speed of adjustment at the wholesale level is no faster or slower than at the retail level. This has important implications, as the authors conclude:

In this study, we examine price transmissions among farm, wholesale and retail U.S. beef markets using two types of retail level price data, one collected by the Bureau of Labor Statistics (BLS) and the other one collected at the point of sale using electronic scanners. In particular, we compare BLS and scanner price adjustments to changes in upstream prices (i.e., farm and wholesale prices). Although these two retail price series differ in the way they are constructed (e.g., data collecting methods, volume sales and discounted price considerations), we find no evidence of asymmetry in the response of retail prices to shocks in upstream prices. Our results have important implications for the U.S. beef market efficiency. Particularly, since retailer adjustments to farm and wholesale price changes are symmetric, the U.S. beef market is not as inefficient as found in previous studies (e.g., Goodwin and Holt, 1999). Finally, our analysis suggests that the U.S. beef market has become more efficient in recent years. That is, information is transmitted more efficiently along vertically coordinated beef markets.