• Selected
• Entire List

Abstract:
This paper documents the existence of the direct and indirect (via word-of-mouth) effects of service quality on new customer acquisition, usage and retention using behavioral data from the launch of a new video on demand type service. For this technology, service quality - the quality of the signal determining the number of movies available for viewing - is exogenously determined and objectively measured. This information, coupled with location and neighborhood information for each subscriber allows us to measure both the direct and indirect effects of service quality. Our identification strategy for these effects arises from both the main effect of neighbors who have previously adopted and the interaction between the number of neighbors and their realized service quality, while controlling for other geographic and demographic covariates.

We find a direct effect of service quality on rental usage and termination behavior. In addition, we find that word of mouth affects about one-fifth of the subscribers with respect to their activation behavior. However, this indirect effect of service quality acts as a double-edged sword as it is asymmetric. We find that the effect of negative word of mouth is twice as high as the effect of positive word of mouth for consumers influenced by word of mouth. We find these effects after controlling for unobserved heterogeneity, marketing activity, distance to retail and demographics.

Consumers affected by word of mouth tend to be heavy users once they adopt the service. The implication from these results is that service quality is important for new customer acquisition in the sense that the heavy users tend to be acquired by word of mouth rather than advertising. In terms of profitability, we find that a 10% increase in service quality leads to a 7% increase in customer lifetime value. Finally, for our sample of adopters, we find that word of mouth effects also lead to acceleration in adoption behavior.

Services, New Products, Service Quality, Word of Mouth, Contagion, Social Interaction, Social Networks, High Technology, Hazard Models

Abstract:
We develop a demand model for technology products that captures the effect of changes in the portfolio of models offered by a brand as well as the influence of the dynamics in its intrinsic preference on that brand's performance. In order to account for the potential correlation in the preferences of models offered by a particular brand, we use a nested logit model with the brand (e.g., Sony) at the upper level and its various models (e.g., Mavica, FD, DSC, etc.) at the lower level of the nest. Relative model preferences are captured via their attributes and prices. We allow for heterogeneity across consumers in their preferences for these attributes and in their price sensitivities in addition to heterogeneity in consumers' intrinsic brand preferences. Together with the nested logit assumption, this allows for a flexible substitution pattern across models at the aggregate level. The attractiveness of a brand's product line changes over time with entry and exit of new models and with changes in attribute and price levels. To allow for time-varying intrinsic brand preferences, we use a state-space model based on the Kalman filter, which captures the influence of marketing actions such as brand-level advertising on the dynamics of intrinsic brand preferences. Hence, the proposed model accounts for the effects of brand preferences, model attributes and marketing mix variables on consumer choice. First, we carry out a simulation study to ensure that our estimation procedure is able to recover the true parameters generating the data. Then, we estimate our model parameters on data for the U.S. digital camera market. Overall, we find that the effect of dynamics in the intrinsic brand preference is greater than the corresponding effect of the dynamics in the brand's product line attractiveness. Assuming plausible profit margins, we evaluate the effect of increasing the advertising expenditures for the largest and the smallest brands in this category and find that these brands can increase their profitability by increasing their advertising expenditures. We also analyze the impact of modifying a camera model's attributes on its profits. Such an analysis could potentially be used to evaluate if product development efforts would be profitable.

Econometric Models, Hi-Tech Marketing, Advertising, Product Line Attractiveness, Product Development, Nested Logit Models, Kalman Filter

Abstract:
The recent marketing literature reflects a growing interest in structural models, stemming from: 1) the desire to test a variety of behavioral theories with market data, and 2) recent developments that facilitate estimation of and inference for these models. Whether one should always go through the effort of developing such tightly parameterized models with the associated computational burden of estimating them, and whether it pays off to make strict behavioral assumptions in terms of better decisions, remain open questions. To shed some light on these issues, we provide examples of structural approaches to consumer choice and demand as well as examples where the goal is to study the nature of competition in the marketplace. From that review spawns our discussion of issues in the development and application of structural models, including their estimation, testing and validation, their applicability in the practice of marketing, and their usefulness for normative, as well as descriptive purposes.

Structural models, dynamic models, endogeneity

Abstract:
One of the major advances of the digital economy is the facilitation of building and managing individual customer relationships - a process usually referred to as "customer relationship management" or CRM. For a typical web site selling frequently-purchased consumer items, the most important stage of CRM is customer retention. This is because the long-term viability of a website is based on its ability to retain a significant customer base. In this study, we focus on a hitherto unexplored question - does banner advertising have a role to play in the customer retention phase of CRM. Using a rich behavioral database consisting of individual customer purchases at a web site along with individual advertising exposure, we measure the impact of banner advertising on customer retention (via purchase acceleration).

We formulate a model of individual purchase timing behavior as a function of advertising exposure. We model the probability of a current customer making a purchase in any given week (since last purchase) via a survival model. The duration dependence in the customers' purchase behavior is captured through a flexible, piecewise exponential hazard function. The advertising covariates enter via a proportional hazards specification. These covariates, richer than have typically been used in past research, consist of strictly advertising variables such as weight and quality as well as advertising/individual browsing variables represented by where and how many pages on which customers are exposed to advertising. Our model also controls for unobserved individual differences by specifying a distribution over the individual customer advertising response parameters. We do this by formulating our model in a hierarchical Bayesian framework. This also allows us to provide some insights into where the returns from targeted banner advertising are the highest and the extent to which the returns are higher compared to no targeting.

Our results show that the number of exposures, number of websites and number of pages on which a customer is exposed to advertising all have a positive effect on customer retention. Interestingly, increasing the number of unique creatives to which a customer is exposed lowers the customer retention probability. We also find evidence of considerable heterogeneity across consumers in response to various aspects of banner advertising. The extent of heterogeneity shows that the returns from targeting individual customers are likely to be the highest for the weight of advertising (the number of advertisements that they were exposed to in a given week) followed by the number of sites that they are exposed to advertising on. To demonstrate the value of the obtained individual response parameters, we carry out a simple experiment in which we compare sales response with and without targeting. We show that, relative to no targeting, targeting results in significant increases in the effectiveness of banner advertising on customer retention and hence, on profitability. Finally, in terms of the broader area of research on the effects of (any type of) advertising, we provide somewhat unique evidence that advertising does affect the purchase behavior of current, in contrast to new, customers.

Advertising response, banner advertising, e-commerce, internet retailing, targeting, micromarketing, survival models, hierarchical Bayesian models, Markov Chain Monte Carlo methods

Abstract:
Sales response models are widely used as the basis for optimizing the marketing mix or for allocation of the sales force. Response models condition on the observed marketing mix variables and focus on the specification of the distribution of observed sales given marketing mix activities. These models fail to recognize that the levels of the marketing mix variables are often chosen with at least partial knowledge of the response parameters in the conditional model. This means that, contrary to standard assumptions, the marginal distribution of the marketing mix variables is not independent of response parameters. We expand on the standard conditional model to include a model for the determination of the marketing mix variables. We apply this modeling approach to the problem of gauging the effectiveness of sales calls (details) to induce greater prescribing of drugs by individual physicians. We do not assume, a priori, that details are set optimally but, instead, infer the extent to which sales force managers have knowledge of responsiveness and use this knowledge to set the level of sales force contact. We find that physicians are not detailed optimally; high volume physicians are detailed to a greater extent than low volume physicians without regard to responsiveness to detailing. In fact, it appears that unresponsive but high volume physicians are detailed the most.

Response Models, Salesforce Effectiveness, Micromarketing, Pharmaceutical Industry, Hierarchical Bayes Models, Metropolis-Hastings Algorithm, Gibbs Sampler, Markov Chain Monte Carlo Methods

Abstract:
Marketing communication plays a major role in influencing consumer purchases in new product categories. An important question about this communication is whether it plays an informative or a persuasive role over the life cycle of the new product category. We expect that consumers are not well informed about product quality in the early stages of the product life cycle but they become better informed over time. The informative role of marketing communication is likely to have a much larger effect with uninformed consumers than with consumers who are better informed. Therefore, we conjecture that marketing communication plays a predominantly informative role initially and a predominantly persuasive role later.

We develop a structural model of demand that allows for the differential impact of informative and persuasive roles of marketing communication over time. In addition, we control for the possibility that product experience also contributes to learning in such categories. We estimate our model on market-level data for the prescription antihistamines category. In our model, physician learning about new antihistamines occurs through marketing communication (detailing and meetings/seminars), and experience (past prescriptions). Specifically, our model is a random coefficients discrete choice model that allows for category expansion. The model also incorporates a Bayesian learning process through which physicians learn about the efficacy of this new class of drugs. We estimate our model using a GMM-based methodology.

We find that, on average, physicians are most sensitive to detailing relative to other promotional activities. However, more interestingly, we find evidence for both informative and persuasive effects of detailing on physicians' prescription behavior. In addition, we find that detailing plays a primarily informative role in the introductory phase (typically 6-14 months post introduction) but the persuasive role dominates later on. The finding that persuasive effects are significant may explain why firms continue to detail long after a drug is introduced. In terms of resource allocation for detailing over time, our results suggest that firms should follow a pattern of heavier detailing at the introduction phase followed by lower levels.

Marketing Communication, Detailing, Learning Models, Discrete Choice Models, Structural Models, Generalized Method of Moments, Pharmaceutical Industry, Antihistamines

Abstract:
Targeting - setting marketing policy differentially for different customers or segments - is an important marketing practice. Previous approaches to quantifying the benefits from targeting have typically calibrated a response model and used the variation in response parameter estimates to compare the firm's profits under targeting schemes at different levels of aggregation. Implicit in this approach is the assumption that the data do not reflect any strategic behavior that the firm may be engaged in vis-à-vis the marketing variables being used for targeting. In this paper, we develop a method to quantify the benefits of targeting when the data reflect firm strategic behavior, i.e., when firms are i) already engaged in some form of targeting; and ii) taking into account actions of competing firms. In particular, we are interested in quantifying the improvement in profits to a firm from targeting its activities at the individual customer level as compared to the allocation of marketing resources at a more aggregate level (e.g., segment or market level).

We focus on detailing - the most important marketing instrument in the pharmaceutical industry. The pharmaceutical firm's key decision is the allocation of detailing visits across individual physicians. As firms already use the information on how detailing affects individual physician behavior in setting their detailing allocation, our proposed approach is appropriate in this context. For our analysis, we develop, at the individual physician level, a model of prescriptions as a function of detailing; and a model of detailing under the assumption that firms simultaneously maximize profits from a physician. We estimate our model on a novel physician panel dataset from the Proton Pump Inhibitor category. Estimation of the model parameters is carried out jointly using full-information Bayesian methods to obtain efficient estimates of the parameters of both models at the individual physician level. Our results suggest that accounting for firm strategic behavior improves profitability by 23% relative to segment level targeting. In addition, ignoring firm strategic behavior underestimates the benefit of individual level targeting significantly. We provide reasons for this finding. We also carry out several robustness checks to test the validity of modeling assumptions.

Targeted Marketing, Response Models, Firm Strategic Behavior, Pharmaceutical Industry, Detailing, MCMC Methods

Abstract:
We present a framework to measure empirically the size of indirect network effects in high-technology markets with competing incompatible technology standards. These indirect network effects arise due to inter-dependence in demand for hardware and compatible software. By modeling the joint determination of hardware sales and software availability in the market, we are able to describe the nature of demand inter-dependence and to measure the size of the indirect network effects. We apply the model to price and sales data from the industry for Personal Digital Assistants (PDAs) along with the availability of software titles compatible with each PDA hardware standard. Our empirical results indicate significant indirect network effects. By July 2002, the network effect explains roughly 22% of the log-odds ratio of the sales of all Palm O/S compatible PDAs to Microsoft O/S compatible PDAs, where the remaining 78% reflects price and model features. We also use our model estimates to study the growth of the installed bases of Palm and Microsoft PDA hardware, with and without the availability of compatible third party software. We find that lack of third party software negatively impacts the evolution of the installed hardware bases of both formats. These results suggest PDA hardware firms would benefit from investing resources in increasing the provision of software for their products. We then compare the benefits of investments in software with investments in the quality of hardware technology. This exercise helps disentangle the potential for incremental hardware sales due to hardware quality improvement from that of positive feedback due to market software provision.

High-technology products, indirect network effects, positive feedback, endogeneity

Abstract:
We present a framework to measure empirically the size of indirect network effects in high-technology markets with competing incompatible technology standards. These indirect network effects arise due to inter-dependence in demand for hardware and compatible software. By modeling the joint determination of hardware sales and software availability in the market, we are able to describe the nature of demand inter-dependence and to measure the size of the indirect network effects. We apply the model to price and sales data from the industry for Personal Digital Assistants (PDAs) along with the availability of software titles compatible with each PDA hardware standard. Our empirical results indicate significant indirect network effects. By July 2002, the network effect explains roughly 22% of the log-odds ratio of the sales of all Palm O/S compatible PDA-s to Microsoft O/S compatible PDA-s, where the remaining 78% reflects price and model features. We also use our model estimates to study the growth of the installed bases of Palm and Microsoft PDA hardware, with and without the availability of compatible third party software. We find that lack of third party software negatively impacts the evolution of the installed hardware bases of both formats. These results suggest PDA hardware firms would benefit from investing resources in increasing the provision of software for their products. We then compare the benefits of investments in software with investments in the quality of hardware technology. This exercise helps disentangle the potential for incremental hardware sales due to hardware quality improvement from that of positive feedback due to market software provision.

high-technology products, indirect network effects, positive feedback, endogeneity, econometrics, economic policy, empirical industrial organization, market research, marketing strategy, product management

Abstract:
The extant literature using household scanner data to estimate consumer choice models has identified two key sources of bias in estimated mean responses to marketing variables. Omitted heterogeneity may bias mean responses towards zero. At the same time, omitted time-varying characteristics of alternatives that influence consumer choices may also bias mean responses towards zero if these characteristics are correlated with observed factors such as price - the endogeneity bias. Both these issues have been well recognized, and methods have been proposed to address them using household scanner panel data.

However, when estimating a choice model with these data at the SKU or the UPC level, one may not observe choices for each item in each of the time periods under consideration. Without such information, one cannot control for item and time period specific unmeasured characteristics, as there is no information on alternatives during those periods in which they are not purchased by any of the panelists. In general, when a product category has many alternatives, each with fairly small shares, the household sample may not contain sufficient choices for each alternative, negatively impacting the ability to control for endogeneity with household data. In contrast, as aggregate store-level data are the true aggregation of purchases by all households visiting the store, they contain the time-period specific item level information required to account for endogeneity as long as each item has some sales in each time period. Given the relative merits of household data to estimate the distribution of heterogeneity and store-level data to address the endogeneity problem, we propose an integrated estimation procedure that uses the information in both sources. Our approach provides consistent estimates of the mean responses to marketing variables and the heterogeneity distribution and also controls for potential endogeneity due to correlation between unmeasured item-level characteristics and prices.

Household scanner data, store-level scanner data, price endogeneity, heterogeneity

Abstract:
This paper develops a framework to measure "tipping" - the increase in a firm's market share dominance caused by indirect network effects. Our measure compares the expected concentration in a market to the hypothetical expected concentration that would arise in the absence of indirect network effects. In practice, this measure requires a model that can predict the counter-factual market concentration under different parameter values capturing the strength of indirect network effects. We build such a model for the case of dynamic standards competition in a market characterized by the classic hardware/software paradigm. To demonstrate its applicability, we calibrate it using demand estimates and other data from the 32/64-bit generation of video game consoles, a canonical example of standards competition with indirect network effects. In our example, we find that indirect network effects can lead to a strong, economically significant increase in market concentration. We also find important roles for beliefs on both the demand side, as consumer's tend to pick the product they expect to win the standards war, and on the supply side, as firms engage in penetration pricing to invest in growing their networks.

Dynamic Oligopoly, network effects, antitrust, concentration, durable goods, penetration pricing

Abstract:
The growth of online word-of-mouth via user ratings, blogs, etc. has prompted an emerging area of research into the effects of such factors on offline product performance. Measuring the actual effects of such factors on offline sales remains a challenge due to the presence of unobserved factors that can be correlated with both the online word-of-mouth online WOM measure and the offline sales measure. The key objective of this study is to measure the effects of online word-of-mouth, specifically online user ratings, on offline sales for new products that are sequentially released across geographic markets while accounting for the possible endogeneity of these ratings. We achieve this by exploiting the sequential rollout of products across geographic markets, in conjunction with local market-level information, and by constructing plausible instruments with these data for the user ratings (that are typically available only at the national level).

Our empirical application is set in the context of the movie industry, where we investigate the impact of user ratings (valence, volume and variance) from sites such as Yahoo! Movies on movie box-office performance. We use a unique data set that has box-office information at the local market level for all movies released between November 2003 and February 2005. We find that mean user rating (valence) has a significant and positive impact on box-office earnings. Moreover, other online WOM measures (volume and variance) have no significant impact on box-office performance. We check the robustness of our results to (i) the choice of instruments; and (ii) the website from which we obtain user ratings. In addition, we compare our estimates obtained using local market-level data with those obtained using aggregate national-level data that are commonly used in the extant online WOM literature. The key finding is that mean rating is not significant in the aggregate level models; a finding that replicates the results in previous studies but with our data. These findings provide new managerial insights while also shedding light on the potential limitations of using aggregate national data to measure online WOM effects in this market.

Our proposed measurement approach and identification strategy can be readily applied to other industry settings as well, where the sequential release of a new product across markets is common and the issue of the endogeneity of online WOM is a potential problem to be addressed.

Online Word of Mouth, Sequential New Product Release, Endogeneity, Instrumental Variables, Generalized Method of Moments, Motion Pictures

Abstract:
We propose an alternative approach to obtaining SKU-level preferences and response sensitivities. An attribute-level model in which the unit of analysis is the market share for an alternative created by aggregation e.g., Colgate toothpaste) is distinguished from a truly disaggregate SKU-level model and develop an analytical relationship between parameters obtained from these two models is established. We show that the researcher can recover SKU-level parameters via calculation from estimated attribute-level parameters, circumventing the need for direct estimation of the more complex true SKU-level model.

Our market share model is calibrated using 98 weeks of data for 10 brands and 168 SKUs in the toothpaste category. We show that instead of estimating 168 preference parameters (when we have an outside alternative in addition to the 168 inside ones), one need only estimate 10 brand preference parameters from which the 168 parameters can be computed as long as share and marketing mix data are available at the SKU level. Marketing mix response parameters can be recovered in a similar fashion. Estimation on holdout samples demonstrates superior predictive performance compared with other available methods. Implications for the derivation of SKU-level elasticities and forecasts of market share and response sensitivity for new items introduced to the category are discussed.

Discrete Choice, Market Share Models, Marketing Mix, Product Attributes, SKU

Abstract:
We provide a descriptive study of a household's decisions on where to shop, when to shop and how much to spend for grocery products in the same grocery chain's online and offline stores. We observe households that shop interchangeably in the online and offline stores and make a majority of their purchases in the chain. The data cover all product categories purchased by households at the chain (packaged goods, perishables, etc.). For each household we identify the top 30 categories that account for a majority of that household's expenditures in the chain. These categories vary across households. We relate a household's shopping decisions to the household's needs in each of its top 30 categories, price promotions in those categories, their perishability, and household and store specific characteristics. We specify a multinomial probit model for store visit and channel choice decision, a competing-risks continuous-time proportional hazard model for trip timing, and a regression model for basket expenditures. We allow all drivers of the shopping decisions to have channel-specific effects and take a hierarchical Bayes approach to estimate the parameters of the proposed models. We find channel-specific effects of inventory levels and price promotions on all three decisions. Depletion of inventories of categories that account for a larger proportion of household expenditures drives a household to the online store while depletion of inventories of lower expenditure categories drives a household to the offline stores. Price promotions influence households to visit the chain, increase the probability of households' visiting the chain, and reduce offline trip expenditure, but do not affect online expenditure much. All these indicate the more planned nature of online shopping trips, which are influenced more by households' internal needs and less by the retailer's marketing activities. There is evidence of sorting of trips whereby households make stock-up trips to the online store and fill-in trips to the offline stores.

Channel choice, e-commerce, Grocery shopping, Trip timing, Multinomial probit model, Competing-risks proportional hazard model, Hierarchical Bayes model

Abstract:
Normative models typically suggest that prices rise in periods of high demand and cost. Yet in many markets, prices fall when demand or costs rise. This inconsistency occurs because the normative models assume that competitive intensity does not change with demand and cost conditions over time. We therefore introduce the notion of time varying competition by suggesting that it is important to not account for the direct effect of demand and cost on prices (e.g., higher demand means higher prices), but also the indirect effect of demand and cost changes on competition (e.g., higher demand could cause more competition and hence lower prices).

We develop a general, unified framework to empirically model the direct and indirect effects of demand and cost shifts on pricing in differentiated product markets. Our approach allows us to measure the indirect effect of multiple demand and cost drivers on competitive intensity and test predictions from alternative theories of repeated games. The empirical application is to the U.S. photographic film industry where there are two main players, Kodak and Fuji.

We find that the indirect effects are highly significant and comparable in magnitude to the direct effects. Competitive intensity is greater in periods of high demand and lower cost and is moderated by whether demand or costs are expected to grow or decline. Interestingly, we find asymmetries in the competitive responses of Kodak and Fuji. While Kodak is sensitive to demand factors, Fuji is sensitive to costs. Our results suggest that market characteristics such as observability of competitor prices can be an important determinant of how competitive intensity is affected by demand and cost conditions.

Pricing Research, Competition, Competitive Strategy, Game Theory, Estimation and Other Statistical Techniques

Abstract:
In the context of introducing new products around the world, it is important to understand the relative attractiveness of various countries in terms of maximum penetration potential and diffusion speed. In this paper, we examine these market characteristics for a new category of prescription drugs in both developing and developed countries. Using data from fifteen countries, and a logistic specification in the Hierarchical Bayesian framework, we report the differences in diffusion speed and maximum penetration potential between developing and developed nations. Our methodology accounts for the limited number of data observations as well as serial correlation and endogeneity problems that arise in the analysis. The principal findings include: (i) Compared to developed countries, developing nations tend to have lower diffusion speeds and maximum penetration levels; (ii) Laggard developed countries have higher speeds. However, laggard developing countries do not have higher diffusion speeds; (iii) Per capita expenditures on healthcare have a positive effect on diffusion speed (particularly for developed countries), while higher prices tend to decrease diffusion speed. The paper concludes by identifying useful avenues for additional research.

diffusion, cross-country analysis, pharmaceuticals, Hierarchical Bayes, econometrics, economic policy, empirical industrial organization, market research, marketing strategy, product management

Abstract:
In this research, we provide a new method to estimate discrete choice models with unobserved heterogeneity that can be used with either cross-sectional or panel data. The method imposes nonparametric assumptions on the systematic subutility functions and on the distributions of the unobservable random vectors and the heterogeneity parameter. The estimators are computationally feasible and strongly consistent. We provide an empirical application of the estimator to a model of store format choice. The key insights from the empirical application are: 1) consumer response to cost and distance contains interactions and non-linear effects which implies that a model without these effects tends to bias the estimated elasticities and heterogeneity distribution and 2) the increase in likelihood for adding non-linearities is similar to the increase in likelihood for adding heterogeneity, and this increase persists as heterogeneity is included in the model.

nonparametric, discrete choice, heterogeneity, random effects, store choice, panel data

Abstract:
We present a framework for modeling consumer adoption of multiple categories of technology products that may be related as complements (or substitutes). The context of technology products as well as the relationship between categories poses some unique challenges. First, the declining prices (and the corresponding increase in quality levels) over time imply that consumers anticipate these changes and make a trade-off between adopting the product early on and consuming the product for a longer time versus adopting later at lower prices. Second, the durable nature of technology products implies that even if two categories are related as complements, consumers may stagger their purchases over several periods; unlike in the case of packaged goods, one cannot infer complementary relationships between these categories based on joint purchases. Third, the adoption decision for some categories (such as printers) may be contingent upon adoption of another related category (such as a personal computer). We illustrate how our proposed modeling framework is flexible enough to address these issues in the context of two related categories and discuss how it can be extended to multiple categories. We apply our modeling framework to a unique dataset that contains information on consumer adoption of three related categories of technology products - personal computers, digital cameras, and printers. The results reveal a strong complementary relationship between these categories. As a result, the probability that a consumer would adopt a given category increases significantly if she has already adopted one or more of the related categories. Policy simulations based on a temporary price decrease in any one category provide interesting insights into how consumers would modify their adoption behavior over time as well as across categories as a consequence of the price decrease.

Technology products, product adoption, multiple categories, complementarity, forward-looking consumers

Abstract:
An important issue facing managers of firms marketing several prescription drugs in different categories to the same primary care physicians is to understand the responsiveness of these physicians to detailing across categories. In this paper, we show that, for the categories we consider, (i) physicians’ prescription decisions are correlated across categories; and (ii) if this correlation is ignored, then one can arrive at incorrect inferences regarding physicians’ detailing sensitivities in the various categories. Since detailing sensitivity is an important basis for physician segmentation, inferences regarding this aspect of a firm’s marketing can be impacted. We propose a multicategory model of physician prescription while accounting for the endogeneity of firms’ detailing efforts. Within a category, the proposed Bayesian Multivariate Poisson – Mixture of Multivariate Lognormals prescription model allows for physicians’ intrinsic brand preferences and detailing sensitivities to be correlated across brands. Across categories, the model accounts for correlations in: (i) physician’s responsiveness to detailing; (ii) brand preferences and detailing sensitivities for brands from the same firm; (iii) correlations due to unobserved (by the researcher) factors. The detailing model combines the “heuristic” approach of allowing for detailing levels to vary depending upon physicians’ demand parameters; the “limited information” approach that allows for dependence of current period detailing levels on previous decisions and outcomes; and the recently proposed Bayesian approach to the instrumental variables problem. From a substantive perspective, we are able to quantify the impact of a multicategory analysis when segmenting physicians based on their responsiveness to detailing. The results have implications for possible reallocation of detailing across product categories.

Abstract:
Discrete choice models of aggregate demand, such as the random coefficients logit, can handle large differentiated products categories parsimoniously while still providing flexible substitution patterns. However, the discrete choice assumption may not be appropriate for many categories in which we expect consumers may purchase more than one unit of the selected item. We derive the aggregate demand system corresponding to a discrete/continuous household-level model of demand. We also propose a Method-of-Simulated-Moments procedure that provides consistent estimates of the structural parameters when only aggregate data are available. The procedure also enables the researcher to control both for the potential endogeneity of marketing variables as well as potential heterogeneity in consumer tastes. Using our aggregate estimates, we can measure the decomposition of price elasticities into incidence, brand choice and purchase quantity components. We also propose several empirical tests to assess the validity of the discrete/continuous demand system versus the logit model. In several simulation experiments, we demonstrate the robustness of this model across datasets in which quantity choices may or may not be important. Our empirical calibration to store-level data in the refrigerated orange juice category indicates a considerable improvement in fit of the observed aggregate sales using the discrete/continuous model.

Discrete/continuous demand, logit demand system, aggregate data, price endogeneity, primary and secondary demand, econometrics, economic policy, empirical industrial organization, market research, marketing strategy, product management

Abstract:
Before the deregulation of Digital Subscriber Line (DSL) services by the FCC in 2005, phone companies were required to share their DSL bandwidth with independent DSL providers. However, despite the large number of independent providers that entered the market, phone companies accounted for 95.3% of all DSL subscribers in 2005. A common explanation for this is based on supply-side factors such as the costs faced by these providers to lease the telephone lines from the phone companies with whom they then had to compete with, as well as price discounts offered by the local phone companies. In this paper, we look for a demand-side explanation for this market outcome. In particular, we study the demand for home broadband Internet services to understand the factor(s) that may have contributed to the competitive advantages of phone companies in the broadband market and use household panel data for the years 2003-2005 prior to the deregulation for our empirical analysis. Given the large shares of the local phone companies, investigating choices of households within the broadband category alone might lead us to the conclusion that households have a much stronger preference for DSL provision from local phone companies - a conclusion that would have been at odds with the accolades received by independent providers for their service. Since local phone and cable companies offer bundled price discounts and the convenience of a single bill across their services, and since these services could be intrinsic substitutes or complements, we investigate the demand for broadband services in conjunction with related services such as cable television and local telephone. We find evidence of strong complementarities between consumption of broadband (cable modem and DSL) and of those related categories. The main source of such complementarities, in our data, is the convenience benefit to consumers from having a single provider for multiple services. Based on counterfactual experiments, our results also indicate that the share of phone companies on the broadband market would have been 48% smaller without complementarities stemming from such a single-provider effect, while in comparison it would have been 19% smaller if there was no price discount offered on the bundle of DSL local telephone.

Abstract:
The recent withdrawal of Cox-2 Inhibitors has generated debate on the role of information in drug diffusion: can the market learn the efficacy of new drugs, or does it depend solely on manufacturer advertising and FDA updates? In this study, we use a novel data set to study the diffusion of three Cox-2 Inhibitors - Celebrex, Vioxx and Bextra - before the Vioxx withdrawal. Our study has two unique features: first, we observe each patient's reported satisfaction after consuming a drug. This patient level data set, together with market level data on FDA updates, media coverage, academic articles, and pharmaceutical advertising, allows us to model individual prescription decisions. Second, we distinguish across-patient learning of a drug's general efficacy from the within-patient learning of the match between a drug and a patient. Our results suggest that prescription choice is sensitive to many sources of information. At the beginning of 2001 and upon Bextra entry in January 2002, doctors held a strong prior belief about the efficacy of Celebrex, Vioxx, and Bextra. As a result, the learning from patient satisfaction is gradual and more concentrated on drug-patient match than on across-patient spillovers. News articles are weakly beneficial for Cox-2 drug sales, but academic articles appear to be detrimental. The impact of FDA updates is close to zero once we control for academic articles, which suggests that FDA updates follow academic articles and therefore deliver little new information to doctors. We find that drug advertising also influences the choice of a patient's medication. A number of counterfactual experiments are carried out to quantify the influence of information on market shares.

Abstract:
Researchers have recently developed models for determining which market conduct best describes observed data. We apply these techniques from the "new empirical industrial organization" literature to the competitive product line pricing decision, where a firm strategically prices its brands when determining the profit-maximizing conduct in the market. Demand, cost, and market structure are estimated endogenously. Empirical results from analyzing price competition in the laundry detergent market between Procter and Gamble selling Tide and EraPlus, and Lever Brothers offering Wisk and Surf, indicate that each firm positions its strong brand as a Stackelberg leader, with the rival's minor brand being the follower.