Competition Between Private Labels And National Brands: Empirical Evidence From Homescan Data On Fluid Milk Markets

The purpose of this study is to empirically examine the nature of the pricing relationship between private labels (or store brands ) and national brands. To accomplish this, we control for the exogenous variation in the farm price of a commodity that serves as the main agricultural input common to both private labels and national brands (any shocks to this farm price will pass through to both retail prices - private label and retail). The product of choice is fluid milk, as the farm price of milk comprises a large share of producer and retailer cost. Two examples of underlying theories concerning this relationship are that (1) the introduction of private labels into a market lowers relative national brand prices, and (2) the introduction of private labels raises the relative national brand prices. The intuition following these two theories about private label/national brand competition tells us the patterns we should look for in the farm-to-retail price transmissions - whether relative national brand prices rise or fall. Hence, the models we estimate are standard price transmission models, each appropriate under specific assumptions about the data. We obtain results that are inconsistent with either of the chosen theories. Furthermore, the results are robust across model specification. Within the results, we do observe that for small number of markets, the price transmission patterns for private labels are statistically different from those of national brands. Using Chow tests, these markets can be identified and set aside for further investigation. Finally, we estimate the models again using private label retail prices from retailers for whom private labels are a relatively larger share of sales, and private label prices from retailers for whom private labels are a relatively lower share of sales. On average, we observe no difference in price-setting by retailers who feature national brands and those who don\u27t, which is consistent with the preliminary findings