What Brings a Consumer Back for More? Evidence from Quantifiable Gain and Loss Experiences in Penny Auctions

Evidence increasingly points to the importance of reference-dependence in predicting consumer behavior. We utilize detailed data from penny auctions, which first appeared as an internet phenomenon in the late 2000's, to uncover how consumers' prior experiences predict their willingness to try a relatively new experience again. The penny auction setting allows us to concretely quantify consumers' positive and negative experiences, which avoids the measurement challenges typical in other consumer experience settings. We find the following empirical patterns with respect to nominal gains and losses, which bear intuitive resemblance to reference-dependent behavior: First, consumers' tendency to return is at least weakly increasing in the outcome from their prior experience; Second, consumers' return rate on marginal loss experiences drops more steeply than it increases for marginal gain experiences; Third, consumers have an aversion to loss at any level, making them discontinuously less likely to return after any loss compared to after a gain. Developing a theoretical model of consumers' decisions to return to a service after prior experiences, we derive a reasonable set of sufficient conditions on utility functions and reference point formation which can account for the intuitive empirical findings. Finally, we examine the bracketing tendencies of consumers in calculating their gains and losses after repeated experiences. Consumers tend to bracket narrowly, with the above patterns strengthening the more recently the experience occurred. This is consistent with the intuition that a recent negative experience risks driving consumers away permanently in spite of previous positive experiences