|
Article Excerpt
In this paper we describe the pass-through behavior of a major U.S. supermarket chain for 78 products across 11 categories. Our data set includes retail prices and wholesale prices for stores in 15 retail price zones for a one-year period. For the empirical model, we use a reduced-form approach that focuses directly on equilibrium prices as a function of exogenous supply- and demand-shifting variables. The reduced-form approach enables us to identify the theoretical pass-through rate without specific assumptions about the form of consumer demand or the conduct of a category-pricing manager. Thus, our measurements of pass-through are not constrained by specific structure on the underlying economic model. The empirical pricing model includes costs of all competing products in the category on the right-hand side (not only the cost of the focal brand) and yields estimates of both own-brand and cross-brand pass-through rates.
Our results provide a rich picture of the retailer's pass-through behavior. We find that pass-through varies substantially across products and across categories. Own-brand pass-through rates are, on average, more than 60% for 9 of 11 categories, a finding that is at odds with the claims of manufacturers about retailers in general. Importantly, we find substantial evidence of cross-brand pass-through effects, indicating that retail prices of competing products are adjusted in response to a change in the wholesale price of any given product in the category. We find that cross-brand pass-through rates are both positive and negative. We explore determinants of own-brand and cross-brand pass-through rates and find strong evidence in multiple categories of asymmetric retailer response to trade promotions on large versus small brands. For example, brands with larger market shares, and brands that contribute more to retailer profits in the category, receive higher pass-through. We also find that trade promotions on large brands are less likely than small brands to generate positive cross-brand pass-through, i.e., induce the retailer to reduce the retail price of competing smaller products. On the other hand, small share brands are disadvantaged along three dimensions. Trade promotions on small brands receive low own-brand pass-through and generate positive cross-brand pass-through for larger competing brands. Moreover, small share brands do not receive positive cross pass-through from trade promotions on these larger competitors. We also find that store brands are similarly disadvantaged with respect to national brands.
Key words: pricing; promotion; retailing; channels of distribution; econometric models
History: This paper was received July 20, 2001, and was with the authors 9 months for 4 revisions; processed by Dick Wittink.
1. Introduction
In the packaged goods industry, over 60% of manufacturers' marketing budgets is now channeled through the retailer in the form of trade promotion spending (Cannondale Associates 2001). This amounts to 16% of the revenues of these manufacturers. The annual Cannondale survey shows that year after year, the single biggest concern of manufacturers is the inefficiency of this trade promotion spending. Manufacturers believe that "nonpass-through" of trade-promotion money to consumers is a major contributor to the inefficiency. They contend that only about half of their trade spending is passed through to the consumer, while retailers claim that percentage is substantially higher. In the 2001 Cannondale survey, for example, food manufacturers claimed that 52% of trade funds were passed through to consumers, 21% covered retailers' promotion costs, and 27% were applied to the bottom line of the retailer. By contrast, food retailers said they passed through 62%, used 24% to cover promotion costs, and 14% went to the bottom line. Based on the estimated trade-promotion expenditure of $75 billion, the gap between manufacturers' and retailers' reported amounts of pass-through exceeds $7.5 billion.
Retailers mediate the marketplace impact of trade promotions, which are a key competitive instrument for manufacturers. Hence, it is essential for manufacturers to understand retail pass-through behavior. Our goal in this paper is to describe the pattern of pass-through of a major Chicago supermarket chain. For our analysis we use a scanner data set for 11 product categories that includes weekly retail shelf prices and wholesale prices. Data on the latter are rarely available for academic studies. For each of the stores in the chain, we also have a set of characteristics that describes the consumers and the local competition in the store's trading area. In each product category, we estimate own-brand and cross-brand pass-through elasticities for each of the brands. Pass-through is defined as the rate at which changes in the wholesale price of a product are passed through by the retailer to the shelf prices. To measure pass-through, we use the reduced form of a retail-pricing model as the basis of our econometric specification. This reduced-form approach permits us to estimate pass-through without constraints on the range of pass-through that are implicit in a structural model. The econometric model controls for the potential role of multiproduct pricing decisions by a category manager. Thus, we consider the impact of changes in a given brand's wholesale price on the shelf prices of both, the same product, and all other products in the category.
This paper has two primary objectives--one descriptive and the other exploratory. The descriptive component documents the magnitudes of own-brand and cross-brand pass-through elasticities and rates in several large supermarket categories. As we discuss below, there is little published evidence on retail pass-through, especially on cross-pass-through rates. We find that many of our results would not be predicted by extant theoretical models. This suggests there is a need and opportunity for marketing scientists to develop more general theoretical models of retail pass-through.
Given the broad range of pass-through rates that we discover across products and categories, we carry out an exploratory second-stage analysis. Here we examine the determinants of own-brand and cross-brand pass-through elasticities across products and across retail stores. The set of covariates we examine includes market share, share of category profits, store brands versus national brands, and demographic and competitive characteristics of the stores' trading areas. Our results reveal patterns of covariation that may be used to assess the plausibility of theoretical models of retail pass-through.
1.1. Comparison with Previous Empirical Studies of Retail Pass-Through
The importance of trade promotions for manufacturers and retailers has motivated a number of academic studies in marketing. However, in contrast with the substantial empirical evidence on consumer response to retail promotions, there are only a few empirical studies that study retail pass-through, which is a measure of retailer response to manufacturer trade promotions. Specifically, we find three studies (1) that provide estimates of retail pass-through: Chevalier and Curhan 1976, Walters 1989, and Armstrong 1991. These papers provide valuable insights into the magnitude and range of retail pass-through rates. However, there are important differences between our approach and the methodology employed by past studies.
First, unlike the econometric approach used in this paper, past studies are typically based on accounting measures of pass-through. The approach used in these studies is to identify trade promotion events and compute the ratio of retail price change to wholesale price change during this event. This creates potential problems. For example, price changes that occur outside the predefined window of the event are ignored when in fact, due to inventorying behavior, the impact of cost changes may spill over outside the trade promotion event. Further, the reported pass-through is confounded by alternative drivers of retail price changes, such as seasonality. Our use of an econometric model and pooled cross-section, time-series data alleviates these concerns.
Another critical difference is that we model the retail pass-through decision in the context of category management by the retailer, while previous empirical studies of retail pass-through consider each product independent of other products in the category. In effect, therefore, past work makes the (unstated) assumption that cross-brand pass-through rates are zero. (2) Our model of retail pass-through allows the wholesale prices of one brand to influence the retail prices of all products in the category. This aspect of our model has two important implications. First, estimates of own-brand pass-through rates are biased if competitive costs are not controlled, because trade promotions on competing products are likely to be correlated clue to strategic manufacturer interactions. We provide evidence of this bias in the paper. Second, there are significant cross-brand pass-through effects that indicate the retailer's pass-through is not only on the trade-promoted brand but is also on competing brands in the category. Our analysis, therefore, provides a more comprehensive and accurate picture of retail pass-through.
1.2. Preview of Findings
We conduct our analysis on 78 products in 11 product categories across 15 price zones. We find that although category average own-brand pass-through rates (3) range from as low as 22% for toothpaste, to as high as 558% for beer, on average the estimated pass-through rates for this chain are much higher than the percentage claimed by manufacturers in the Cannondale (2001) study for all retailers. There are substantial differences in pass-through estimates across retail pricing zones and categories, and between products within categories. We find that the vast majority of own-brand pass-through estimates are positive. As many as 14% of the own-brand pass-through rates are significantly greater than one, implying that in these cases, on average the retailer offers a larger discount to the consumer than the retailer receives from the manufacturer. These findings challenge the "empirical generalization" that most products display pass-through much smaller than one (Blattberg et al. 1995), although our results are consistent with empirical findings of Armstrong (1991) and Walters (1989). A notable finding from our analysis is that as many as two-thirds of the estimated cross-brand pass-through rates are statistically significant. This implies that the retailer responds to a trade promotion for one brand by changing retail prices of multiple products in the category. Interestingly, the cross-pass-through effects of a given brand's wholesale price change are positive for some competing products in the category, and negative for others.
We explore the determinants of pass-through via a pooled analysis across categories of estimated own-brand and cross-brand pass-through rates. In multiple categories we find evidence that the retailer's pass-through response on large versus small brands is asymmetric. Larger brands, as measured by share of category volume or by share of category profits, receive higher pass-through. Moreover, these brands are unlikely to generate positive cross-brand pass-through, i.e., induce the retailer to reduce prices of competing smaller brands. Conversely, manufacturers of small brands suffer three...
|
