CONTENTS

Decomposing the Extensive and Intensive Effects of Advertising on Fluid Milk and Cheese Demand

Articles and Recent Reports by staff at NICPRE

Director’s Corner

Next Meeting

NEC-63 Spring 2001 March 22-23, 2001 Washington, DC Policy and Distribution Issues in Agricultural Commodity Promotion Programs

Decomposing the Extensive and Intensive Effects
of Advertising on Fluid Milk and Cheese Demand

by Todd M. Schmit, Chanjin Chung, Diansheng Dong, Harry M. Kaiser (Cornell Univ.)
and
Brian Gould (University of Wisconsin)

Since 1984, U.S. dairy producers have contributed $0.15 per hundredweight of milk sold to increase the demand for dairy products through generic advertising, promotion, and product research. In 1995, fluid milk processors joined the advertising efforts by enacting processor assessments of $0.20 per hundredweight on fluid milk sales through the MilkPEP program. Combined, these checkoff programs collect more than $300 million annually, with a relatively large share committed to the generic advertising programs. Over the last four years, generic advertising expenditures have averaged over $150 million annually for the fluid milk program and over $43 million annually for the cheese program.

In this application, we utilize household-level purchase data to estimate at-home demand functions for fluid milk and cheese products. We adopt a Heckman-style two-step sample selection model, where the first stage is represented by the dichotomous choice of whether to purchase, and the second stage determines the level of consumption given the decision to consume. The use of household-level data and the estimation procedure employed allow us to decompose the total effect of generic advertising into its extensive (probability of purchase) and intensive (level of consumption) components. This is particularly important when evaluating advertising programs to determine to whom the message has been successfully delivered and the type of consumer response.

Fluid milk and cheese purchase data for at-home consumption and annual household demographic data were obtained from the ACNielsen Homescan Panel sample of U.S. households from January 1996 through December 1999.

Households use hand-held scanners to record food purchase information including date of purchase, UPC code, total expenditure, and quantities purchased. Sample households also submit annual demographic information. Approximately 20,000 households were included in the estimation of the fluid milk and cheese product demand equations. National generic fluid milk and cheese advertising expenditures were merged with the household data, aggregated across media type.

Fluid milk was disaggregated into four sub-categories: whole, reduced fat (2%), light (0.5%-1%), and skim milk. Reduced fat milk was the most common milk product, followed by skim and then whole and light milk. Cheese was disaggregated into American, Mozzarella, Processed, and Other cheese categories. The Other cheese category contains numerous varieties, including Ricotta, Muenster, Farmers, brick, and cream cheese. Processed cheese had the highest average consumption, followed by Other and American cheese, while Mozzarella was a distant fourth.

Price, household income, household size, and generic advertising are the basic explanatory variables (or, “demand determinants”) included in the demand models. To account for price variation due solely to volume and store discounts, we include variables that reflect the proportion of purchases in various package sizes and store types. Differences in household composition are accounted for with variables reflecting age of the household head, whether mom works outside of the home, and the proportion of family members by age classification.

The demand models also control for race, education, geographic location, and seasonality.

Since our focus is on the decomposition of the extensive and intensive effects on total demand (particularly for price and advertising), attention will be directed to the computation of the corresponding elasticities. The probability (i.e. stage 1) elasticity represents the percentage change in probability of purchase for a 1% change in the selected variable (i.e. the extensive effect), while the conditional (i.e. stage 2) elasticity represents the percentage change in the quantity demanded, given purchase, for a 1% change in the selected variable (i.e. the intensive effect). The total, or unconditional, elasticity is the sum of the extensive and intensive effects and represents the overall impact on demand.

Price effects were negative and highly significant for nearly all fluid milk and cheese products (Table 1). Only light milk had a positive conditional price elasticity, but this was not significantly different from zero. Unconditional price elasticities for the aggregate products revealed cheese demand to be approximately twice as sensitive to price changes as fluid milk, largely the result of a substantially higher purchase probability effect. This seems logical given the differences in these products; e.g. perishability, purchase frequency, and role in the household diet. Larger unconditional price elasticities on the individual products were predominantly due to the high purchase probability effects. This is likely reflective of product switching, which would not affect the aggregate model estimates. For example, in response to a price reduction for light milk, a household may switch their purchase from reduced fat to light milk. This portrays little price response for the total milk category, but relatively large price responses for the individual products.

Long-run generic advertising elasticities were positive and significant for all fluid milk products. The generic advertising message appears to have a predominantly intensive effect on fluid milk demand. For total fluid milk, the unconditional long-run advertising elasticity was 0.081, 88% of which can be attributed to increasing the conditional demand for milk from current consumers, while only 12% can be attributed to increasing the probability of households to purchase. This is consistent across all categories, with the largest relative response from the generic campaign shown for reduced fat (0.081) and skim milk (0.082), followed closely by whole (0.074) and light milk (0.072). Whether the results would be similar from a more differentiated campaign is an empirical question and cannot be gleaned from these results; however, the relative similarity in response across all products seems consistent with the generic advertising message and the relatively homogeneous nature of the fluid products.

While positive and significant in the total cheese category, the total generic cheese advertising elasticities are lower than that estimated for fluid milk. The total cheese advertising elasticity was estimated to be 0.024; however, the entire amount of this was realized from the extensive margin. That is, cheese advertising appears to be effective at increasing the probability of purchase, but has no significant effect on increasing the conditional demand of current consumers. This is consistent across all cheese products in which no conditional demand elasticities were significantly different from zero. The largest contributors to the total cheese result were from Other (0.069) and American cheese (0.063), while the Processed (0.021) and Mozzarella (0.021) advertising elasticities were lower. It is interesting that while both programs are generic and do not target any specific products, the individual product advertising elasticities are more variable for the relatively differentiated cheese products, but are much more similar for the relatively homogeneous milk products.

The use of more micro-level data allows the researcher to capture additional household heterogeneity effects not available in more aggregate analyses. Furthermore, one can decompose the total effect on demand into that which induces purchase and that which changes the demand levels of current consumers.

Both advertising programs have significantly increased the demand for their respective products; however from the wide disparity in extensive versus intensive advertising response, it is clear that the fluid milk and cheese generic advertising campaigns are inducing different types of response. Fluid milk advertising seems most effective at increasing the demand of current consumers; i.e. when consumers buy, they're buying more. The effect on the probability of purchase is also positive but is less pronounced. Conversely, the effect that cheese advertising has had on total household demand is clearly from increasing the probability of purchase. Therefore, even though the level of purchase is unchanged at each purchase occasion, households are purchasing cheese more frequently and, in turn, increasing the total at-home demand for cheese.

Information such as this provides valuable information to dairy product marketers in determining where past efforts have been most effective and in developing future advertising strategies with respect to their target audience.

AUTHOR'S ACKNOWLEDGEMENTS

This article is part of a larger research effort on the effectiveness of generic dairy advertising on household-level demand. Additional model results are available upon request. This research was sponsored by the Agricultural Marketing Service, U.S. Department of Agriculture and was funded by Dairy Management, Inc. and MilkPEP. We wish to thank John Mengel and Madlyn Daley for coordinating this research. We also acknowledge ACNielsen in providing the household data used in this study. All statements in this article are the conclusions of the authors; ACNielson does not support or confirm any conclusions made by the authors which the authors claim to be based on AC Nielsen information.

[ return to text] [ top ]

Prob.

Probability elasticity; i.e. the percentage change in probability of purchase for a 1% change in price or advertising

Condtl.

Conditional elasticity; i.e. the percentage change in quantity demanded, given purchase, for a 1% change in price or advertising

Uncondtl.

Uncondtional elasticity; i.e. the percentage change in total quantity demanded for a 1% change in price or advertising