Structural modeling of consumer behaviors and business strategies

Abstract

The big data trend motivates both academe and industry to focus on data-driven research work and data-intensive businesses, respectively. Structural modeling approaches are increasingly applied in research on marketing, econometrics, and information systems. Considering the key advantage of structural modeling for counterfactual analyses, this research applies structural models to analyze two business models in entertainment marketing. One model is for online entertainment shopping, whereas the other is for the consumption of cable TV digital entertainment products. Both applications produce big data of “big volume” (large number of registered online users or subscribed household TV viewers), “big variety” (structured and unstructured data), and “big velocity” (real-time online participation or cable TV return path records), allowing scholars to address research topics based on big data. Entertainment shopping supported by pay-to-bid auction mechanism has emerged as an innovative business model in recent years. Consumers expect both entertainment value and monetary return from their participation in entertainment shopping because its selling mechanism combines the features of gambling and auction. This research proposes a dynamic structural model to analyze the online shopping behavior of consumers. To the best of my knowledge, the model in this research is the first to capture the learning process of consumers from two perspectives based on the Bayesian updating framework for online entertainment shopping. The first perspective is that consumers update their beliefs about the entertainment value through their own, repeated participation experiences. The other context is that these consumers form an expectation about the anticipated monetary payoff by acquiring information signals from historical auction ending prices disclosed by the website auctioneer. The model is estimated via simulated maximum likelihood using a large data set from an online entertainment shopping website. Results show that consumers significantly overestimate the entertainment value, but underestimate the level of competition at the beginning of their participation. This finding clarifies the decreasing participation rate of consumers over time. In general, consumers exhibit risk-seeking preferences, which result in the following polarized effects: (1) many consumers quit the website early even before they learn its true entertainment value; (2) a few consumers who learn the true entertainment benefit of a website can be addicted to the games and may become increasingly committed to them. Based on the estimated parameters of the model, counterfactual analyses are performed to identify how the participation behaviors of consumers would be changed when applied. By conducting counterfactual policy simulations, this research discusses the design implications of such entertainment shopping websites and recommends strategies for generating a sustainable business model. With advances in technology, the demand for digital entertainment increases. The TV industry plays an important role in producing and diffusing digital entertainment programs. Cable TV return path data, implemented with the present-day set-top boxes, present a new opportunity for analyzing the viewing behavior of households, from which their viewing preferences can be recovered. The business model requires households to pay a fixed subscription fee monthly to purchase TV programs. This research develops a model of household viewing preference that supports the assessment of the valuation and satiation of a household for different categories of digital content within the constraints of the programs to which it subscribes. This research models household television viewing preference for content genres, rather than channels. It also incorporates household subscriptions and demographics into the unified model for identifying the effects on viewing preferences toward content genres. Instead of weekly consumer recall-based surveys employed by the previous researches, a dataset of more than 1 million observations on households is used. This dataset comes from a digital entertainment firm that offers basic and premium services. The estimation is performed with a Bayesian hierarchical model, which employs the Gibbs sampler and Metropolis- Hastings algorithm. The results show that the households have relatively homogeneous preferences for entertainment content, but they show heterogeneous preferences for content in specific packages to which they subscribe. Moreover, this study finds that incorporating household subscription information and demographics into the unified model helps analyze how these factors can affect the TV viewing preferences of households. By selecting high definition TV and upsizing their bundle contents, the households can have increased base viewing valuations and accelerated viewing satiations. The premium movies- and sports-related add-on package subscriptions yield varying effects on enhancing household preferences toward their most preferred content. In sum, the above findings provide useful insights for understanding household viewing preferences and are intended to realize certain content strategy adjustments toward the promotion and improvement of customer satisfaction. This research presents both theoretical and practical values by providing alternative solutions via structural model approaches for understanding consumer behaviors toward their online entertainment shopping and consumption of digital entertainment products. On the one hand, the theoretical implication of this study is that it extends and enriches the related literature, particularly on entertainment shopping, penny auction, consumer learning, variety-seeking, and TV viewing behaviors. On the other hand, this research yields practical values by helping the corresponding industries understand their customers better and provide them with business insights and strategies for their marketing objectives.

Date of Award	2 Oct 2015
Original language	English
Awarding Institution	City University of Hong Kong
Supervisor	Kwok Fai Geoffrey TSO (Supervisor)