Estimating Consumer Response to Package Downsizing:an Application to Chicago Ice Cream Market
We estimate a random utility model of demand to measure consumer response to packagedownsizing in consumer goods markets. To our knowledge this paper provides the firstestimates of consumer response to package size. We perform the analysis using NielsenHomescan data on bulk ice cream purchases of a panel of households in Chicago, between1998 and 2007. The estimation framework involves modeling household heterogeneity, ad-dressing price endogeneity and dealing with unbalanced choice alternatives. We adapt aBayesian approach for estimation. The main finding is that consumers are less responsiveto package size than to package price; the demand elasticity with respect to package sizeis approximately one-fourth the magnitude of the demand elasticity with respect to price.This result implies that manufacturers can use downsizing as a hidden price increase in or-der to pass through increases in production costs, i.e. cost of raw materials, and maintain,or increase, their profits.
Keywords: consumer behavior, package downsizing, demand analysis, discrete choicemodel, hierarchical Bayesian analysisJEL: D12, M31, C11, C33, C35
Package downsizing, the practice of reducing the volume of product per package such that
the new size replaces the old one, is commonly observed in U.S. consumer goods markets,
and has been especially common in food manufacturing. For example, Breyers downsized
its half-gallon ice cream container from 64 oz. to 56 oz. in 2000, and again from 56 oz.
to 48 oz. in 2007. Similarly, Dannon reduced the size of its yogurt from 8 to 6 oz. in 2003,
and Hellmans reduced its mayonnaise from 32 to 30 oz. in 2006. Similar examples can be
found for numerous other brand names in various food and non-food product categories.
Package downsizing may be consistent with profit-maximizing firms in competitive
markets responding to higher prices of primary inputs by shifting the input mix towards
marketing inputs. Alternatively, it may serve as a strategy to increase prices implicitly; if
a given percentage reduction in product volume is not accompanied by an equivalent
percentage decrease in package price the resulting unit price of the new product is higher.
Notably, consumer product markets are often characterized by imperfectly competitive
market structures that would allow for such pricing strategies.
The objective of this study is to investigate manufacturing firms incentives for package
downsizing. If firms downsizing strategy aims to implicitly increase the unit price, then it
should be the case that firms expect higher returns to package downsizing than to raising
the package price directly. The relative returns to these alternative strategies depend on
consumer response to package size and package price. While estimates of price elasticities
of demand are common for a range of consumer goods, this paper presents the first
estimates of consumer response to package size. We measure these responses in a
simultaneous demand and supply analysis. Specifically, we seek to estimate and compare
demand elasticities with respect to package size and price in order to shed light on
manufacturing firms incentives for downsizing.
Empirical evidence on the causes and effects of package downsizing is sparse. Notable
work exist in an indirectly related literature that analyzes consumer perceptions of
quantity and price indicators of packaged goods. For example, early experimental studies
by Granger and Billson (1972) and Russo (1977) studied consumer preference for different
package sizes when unit price information is made explicit. They find that when unit
price is made explicit consumers tend to switch to the larger sizes that provide quantity
discounts. This result suggests that consumers base their decisions on relative differences
in both package price and package size. In contrast, Lennard et al. (2001) argue that
consumers do not process unit price information perfectly. They use accompanied
shopping interviews and in-store questionnaires to analyze consumers purchase decisions,
and conclude that consumers are unaware of volume indicators, such as weight of content
or number of servings, and often rely on the physical size of the package. Binkley and
Bejnarowicz (2003) find a similar result in their analysis of consumer price awareness.
Using data on actual purchases Binkley and Bejnarowicz (2003) explore why consumers
buy surcharged products. They conclude that time-constrained consumers substitute a
general knowledge of prices, such as the belief that larger packages are cheaper, for careful
price comparisons. Similarly, in his experimental study Wansink (1996) shows that
consumers perceive larger packages as less expensive than small packages on a per-unit
basis, and, therefore, that large packages (holding volume constant) accelerate
consumption which can potentially increase overall sales.
Although these studies provide insights to consumer perceptions of small and large
packages and their implications to marketing, they do not explain manufacturers
incentives to downsizing. In this article we fill this void by performing a demand analysis,
which involves estimation and comparison of elasticities of demand with respect to
package size and price. To date, no study provides these elasticities with a focus on
downsizing. Therefore, we advance the literature by addressing two important questions:
What is the consumer response to downsizing? And how does it compare to the consumer
response to price changes?
We perform a simultaneous demand and supply analysis that allows to derive testable
hypotheses on consumer sensitivity to price and package size. Specifically, we characterize
the demand system with a random coefficient logit equation, where household purchase is
a function of both price and package size. We account for the household heterogeneity
and price endogeneity in demand estimation. We model the household heterogeneity as a
function of household demographics and address the price endogeneity with a reduced
form supply equation. We take a Bayesian approach to estimate the model and focus on
the U.S. bulk ice cream market, which is a typical oligopolistic market with differentiated
products. We perform the analysis using Nielsen Homescan data on bulk ice cream
purchases of a panel of households in Chicago, between 1998 and 2007.
The U.S. Bulk Ice Cream Industry
The U.S. bulk ice cream industry is characterized by concentration and branding, as is
typical of oligopolistic, differentiated-product markets. We document this using Nielsen
Homescan data from 2002-2007. The top two brands in the market hold approximately
32% of the national market; the top six brands account for half the market. Other brands
have low national market shares, but some, especially the store brands, hold significant
shares in some cities. Each of the leading brands offer its ice cream in myriad flavors and
in a variety of package sizes. We focus on the half gallon1 category for two reasons. First,
the half gallon category accounts for approximately 80% of the total volume of the bulk
ice cream market. Second, in the Nielsen Homescan dataset between 1998 and 2007 we
observe that across all size categories downsizing is only prominent in half gallon products.
For example, the leading brand with the largest national market share downsized its half
gallon products twice within the period of study, initially reducing package size from 64
oz. to 56 oz. in 2000, then again reducing package size from 56 oz. to 48 oz. in 2007. The
second largest brand reduced the half gallon package size at approximately the same time
in the same way. Other ice cream manufacturers have also applied downsizing, while a
few brands have not applied downsizing and still offer a 64-oz. package.
In table 1 we report the annual average package size in the half gallon category for each
major brand.2 At this level of aggregation the transition from 64 oz. to 56 oz. is not
discrete. The gradual replacement of old package sizes could be a result of inventory lags,
as brands replace an old package with the new one in any particular store or market.
Also, the smaller package sizes are not introduced at the same time in all cities, so that
national average size changes gradually. As discussed above, the timing of package size
changes differs across brands. Some brands complete the shift to the smaller package as
early as 2004, while others are slower to move to the smaller package, and still others
maintain the larger package throughout the sample.
Insert Table 1 about here
In table 2 we compare average annual package price, price per oz., and package size of the
brands with the three largest national market shares. In the case of national brand 1 a
large drop in average package size in 2002 is accompanied by a decrease in price per
package, however price per oz. increased by 0.2 cents, or approximately 4%. In the case
of national brand 2 a large drop in average package size in 2003 is accompanied by an
increase in the average price per oz. of 0.5 cents, or approximately 10%. Levels and
changes of store brand 1s package size and pricing appear to differ from the the trends of