Bounded Rationality and Consumer Choice: An Evaluation of Consumer Choice of Mobile Bundles in China

Abstract

This paper is an examination of the rationality of consumers’ choice of alternative mobile bundling plans comprising voice, short message service and internet data offered by a major wireless provider in China. Using a large dataset of consumer usage, billing information and demographics, we demonstrate that the vast majority of consumers chose non-optimal bundles, most choosing more expensive bundles that the one warranted by their actual level of usage. We also find that the probability of non-optimal selection increased with the complexity of bundling plans, and decreased with the length of time the user has been in a subscription relationship with a service provider, suggesting a ‘learning effect’ by which users optimized their choice over repeated subscription cycles. As a means of attracting more consumers and increasing average revenues per user, product bundling has become a widespread business strategy in every type of industry. Though numerous prior studies have examined consumer valuation, consumer surplus, and the relative distribution of social welfare in bundling transactions from a theoretical perspective, there have been relatively few empirical analyses with large datasets, since access to data is often restricted on account of proprietary information and consumer privacy. The few available papers utilize very small data sets, or lab experiments to assess the rationality of consumer choice. This paper uses a large dataset of operational information from China Telecom, one of China’s largest mobile telecommunication providers, obtained through a research agreement between the company and one author’s university. Based on the review of theoretical and empirical literature, we propose three hypotheses: the risk aversion hypothesis, the complexity hypothesis and the learning effect hypothesis. The risk aversion hypothesis states that users will prefer to subscribe to a more expensive bundle to avoid possible overages in usage, and the consequent unpleasant surprises. The complexity hypothesis states that the probability of non-optimal selection will increase with the complexity of bundles. The learning effects hypothesis states that users will eventually ‘learn’ to better optimize their bundle selection over repeated subscription cycles, with the result that the probability of non-optimal selection will decrease with the duration of a consumers continuous subscription relationship with a service provider. We use a value model for assessing the rationality of consumer bundling choice, which compares the pricing for a consumer’s amount of usage in voice, SMS and Internet surfing under the currently chosen bundle, and available alternative bundles. We assess the degree of risk aversion inherent in consumer choice by comparing the optimal bundling plan to their chosen bundle for each individual customer, and estimate the influence of factors such as gender, age, user level, etc. The findings suggest significant differences between customers in terms of their choice in telecommunication bundling. Most users (85.9%) did not select their optimal bundle given their usage; in general, they paid more than they should. The complexity hypothesis too was supported with simpler bundles having lower rates of non-optimal selection. However, there was no significant learning effect, possibly because the high levels of churn resulted in few users having long continuous tenures with the service provider. Demographics too did not help predict non-optimal selection, with no significant differences by gender or age on decision making. In addition to the obvious benefits to consumers, prior research has suggested that service providers too can benefit from enabling consumers to choose appropriate bundles and service levels by increasing retention rates and reduced churn. Our recommendations in this paper will therefore be of interest to consumers as well as to mobile telecommunications firms.