Wesley R. Hartmann

Retail Competition and the Dynamics of Demand for Tied Goods

We present a demand system for tied goods incorporating dynamics arising from the tied nature of the products and the stockpiling induced by storability and durability. We accommodate competition across tied good systems and competing downstream retail formats by endogenizing the retail format at which consumers choose to stockpile inventory. This facilitates measurement of long-run retail substitution effects and yields estimates of complementarities within, and substitution across, competing systems of tied goods. We present an empirical application to an archetypal tied goods category: razors and blades. We discuss the implications of measured effects for manufacturer pricing when selling the tied products through an oligopolistic downstream retail channel and assess the extent to which retail substitution reduces channel conflict.

Intertemporal Effects of Consumption and Their Implications for Demand Elasticity Estimates

Consumption of a good typically diminishes the marginal utility of consuming more, but for how long? This paper adapts a model of consumption capital to allow consumption to have a lasting effect that diminishes the marginal utility of future consumption. Estimates of the model find that it takes the 25th, median and 75th percentile of consumers 19, 32 and 43 days for their marginal utilities to return to pre-consumption levels, and they are forward-looking with respect to these effects. This generates intertemporal substitution of consumption that leads to an overestimate of the own-price elasticity of demand of ten percent when it is estimated using temporary price changes. In addition to these implications consumption effects share with those of durable and storable goods, consumption effects also raise concerns for capacity constrained industries because the timing of consumption affects capacity utilization. In the empirical application in this paper, price variation in one time period generates substantial changes in capacity utilization in that period, but minimal changes in other periods because the intertemporal substitution is spread over many time periods.

Internet Versus Television Advertising: A Brand-Building Comparison

Many advertisers are reluctant to shift a large proportion of their advertising budgets to the Internet because they still view television advertising as the main vehicle for building a brand. Using a unique and rich data set comprising 20 campaigns across a variety of industries, this study demonstrates that Internet ads perform on par with television ads on the brand-building metrics that advertisers use and trust. The authors extend traditional brand–message recall measurements to facilitate comparisons between Internet formats and television by supplementing brand–message surveys conducted during the campaign with a set of precampaign surveys to control for preexisting brand knowledge. They find that accounting for differences in preexisting brand knowledge is paramount in obtaining valid comparisons across advertising formats because people who are exposed to Internet display ads have significantly lower levels of preexisting brand knowledge than television viewers. Without considering the differences in these “initial conditions,” television advertising seems to be more effective than advertising on the Internet, but when the preexisting differences among media formats are taken into account, the brand recall lift measures for Internet ads are statistically indistinguishable from comparable television lift measures.

Empirical Analysis of Metering Price Discrimination: Evidence from Concession Sales at Movie Theaters

Prices for goods such as blades for razors, ink for printers, and concessions at movies are often set well above cost. Theory has shown that this could yield a profitable price discrimination strategy often termed “metering.” The idea is that a customers intensity of demand for aftermarket goods (e.g., the concessions) provides a meter of how much the customer is willing to pay for the primary good (e.g., admission). If this correlation in tastes for the two goods is positive, a high price on the aftermarket good allows firms to extract a greater total price (admissions plus concessions) from higher-type customers. This paper develops a simple aggregate model of discrete-continuous demand to motivate how this correlation can be tested using simple regression techniques and readily available firm data. Model simulations illustrate that the regressions can be used to predict whether aftermarket prices should be above, below, or equal to their marginal cost. We then apply the approach to box office and concession data from a chain of Spanish theaters and find that high-priced concessions do extract more surplus from customers with a greater willingness to pay for the admission ticket.

Retail Competition and the Dynamics of Consumer Demand for Tied Goods

We empirically investigate the demand for tied goods sold through competing retail channels. Tied good pricing strategies commonly involve a low price on the initial purchase (i.e. the primary good) to drive adoption, and a substantial markup on aftermarket goods to capture value. However, if the goods are sold through downstream channels, retail market power and a misalignment of incentives could distort the relative prices of primary and aftermarket goods. To evaluate whether retail competition is strong enough to prevent such distortions, we explore the commonly noted example of razors and blades, which are sold through drug, grocery, mass merchandising, and club stores. We specify a forward-looking demand model that incorporates dynamics arising from the tied good nature of the products and the stockpiling and durability aspects of razors and blades. Furthermore, we allow intertemporal substitution in the purchase of both razors and blades to occur across channels as well as time. This modeling feature enables a novel approach to measuring retail competition in single category demand analyses. Our estimates indicate that there is substantial cross-channel substitution in razors, but some retail market power in blades. However, the channel with the most market power in blades, club stores, specializes in high volume customers that would adopt a razor even if blade prices are higher. This suggests that the manufacturer can achieve its desired level of razor adoption without vertical restraints, though blade sales may be slightly reduced by double marginalization.

Information vs Automation and Implications for Dynamic Pricing

Essential resources like electricity and water can experience rapidly changing demand or supply while the other side of the market is unchanged. Short-run price variation could efficiently allocate resources at these critical times, but only if consumers exhibit short-run demand elasticity. The question for firms in these markets has always been how to enable this response. Randomized control trials are increasingly used to test dynamic pricing and technologies that can assist in response by providing information and/or automated response. But, the trials typically do not randomize short-run prices. This paper illustrates how demand from a randomly assigned control group can be used to test the effectiveness of different technologies in increasing short-term price elasticity. To do so, we use a non-parametric control function approach that eliminates the bias inherent in estimating short-term price response using only household random assignment. We find that only automation technology leads to the short-term price elasticity needed to justify real-time pricing.