Date of Award
Doctor of Philosophy (PhD)
Chair and Committee
Chakravarthi Narasimhan, Song Yao, Arun Gopalakrishnan, Xiang Hui,
The general topic of my dissertation is customer relationship management. Specifically, I use quasi-experimental causal inference methods, randomized field experiments, and machine learning methods to study and measure consumer response to co-branded credit cards and email communications promoting subscriptions.
In Chapter 1, “The Impact of Co-branded Credit Card Adoption on Customer Loyalty”, we estimate the treatment effects of adopting a co-branded credit card on spending and loyalty behaviors using a comprehensive longitudinal dataset from a North American airline. Our data set contained detailed records of both airline credit card adopters and non-adopters, including their travel and loyalty program activities over a four-year horizon. We deal with the self-selection of card adopters by (1) using rolling-based matching procedure, (2) conducting difference-in-differences estimation on the matched sample with a two-way fixed effects specification, and (3) dividing treatment effects into three phases of time and argue that the endogenous timing of card adoption will most likely manifest in the short-term effect and is least likely to affect long-term effect. We find statistically significant and economically meaningful effects of card adoption on a multitude of behaviors. Specifically, flight spend was lifted by 42% when considering spend more than 12 months after adoption, demonstrating the persistence of the effect. These flight spend increases were largely driven by more flights purchased rather than higher prices paid per flight, which is indicative of increasing share-of-wallet among adopters. Card adopters also increased award flight redemption to a greater extent than redeeming loyalty program points with airline partners. Finally, card adopters who experienced the highest increase in flight spend, tended to live near hub airports of the airline firm or were already existing members of the loyalty program.
In Chapter 2, “An Experimental Investigation of Price vs. Non-Price Messaging in Subscription Programs”, we study how firms can attract and retain customers for subscription services. Subscriptions of digital and physical goods are becoming increasingly popular, and firms often compete heavily on price for customer acquisition. However, the challenge associated with advertised price discounts is substantial, as the featured price discounts highlight price savings and this might create an adverse selection problem with some customers signing up for just the price discount and then churning soon after. We worked with a major retailer that sells pet products, and we launched a four-week field experiment where we randomized price and non-price messaging in email advertising of subscription. We find that the non-price messages perform as well as the price messages in terms of sign-up rates and outperform price messages for reorder rates. This pattern also holds for number of orders, revenue and profit margin. We find that the inferior performance of the price message is primarily due to price attracting lower quality customers. Our findings suggest that one of the most dominant strategies of selling subscriptions is very suboptimal. Firms would be better off with the messages that make non-price motivations more prominent. Further, firms could also use previous purchase history to better target customers who could be a good match for the subscription services. Our results suggest that the price message should be sent to customers who are less familiar with the online channel, customers who are new to subscription, customers who have more regular purchase history, and customers who are less deal-prone. The rest should be sent non-price messages. Customers with no prior engagement with the firm should be targeted with the risk message by default. Finally, those who are most deal-prone and most familiar with online channels should not be sent any messages at all.
Zhao, Nan, "Essays on Customer Relationship Management" (2023). Olin Business School Electronic Theses and Dissertations. 35.