Purchasing Scarce Products Under Dynamic Pricing: An Experimental Investigation

Abstract

W hereas theoretical studies on dynamic pricing typically assume that consumers are either fully strategic or fully myopic, systematic empirical investigations into how consumers behave under dynamic pricing contexts are relatively rare. Focusing on scarce products, we constructed and experimentally tested a two-stage model in which a firm sells a seasonal good under exogenous inventory constraints to a market of strategic buyers. In our experiment, subjects assigned the role of buyers made purchase decisions in response to prices set by an automated seller. We find that equilibrium predictions assuming fully strategic buyers largely accounted for aggregate behavior in the experiment, and the ex post optimal decisions for subjects were overwhelmingly consistent with equilibrium prescriptions. Moreover, subjects tended to become individually more strategic as the session progressed. However, there were also nuanced systematic patterns of deviations from equilibrium that had profit and pricing implications for the seller. First, a nonnegligible minority of subjects exhibited completely myopic buying behavior even with practice. Second, when the product was relatively more scarce, myopic buying had a stronger impact on demand at higher prices; the upshot is that the seller's season-profit-maximizing price could be considerably higher than what would be optimal with fully strategic buyers.

Dynamic Pricing Decisions and Seller-Buyer Interactions under Capacity Constraints

“…Firstly, it connects with experimental works that focus on seller's pricing decisions in dynamic pricing contexts, such as [16] on dynamic pricing of durable goods with no inventory constraints, and [9] on dynamic pricing decisions when sellers face a simulated market of buyers with heterogeneous levels of strategic sophistication. Secondly, it also connects with several experimental ( [13,17,18] and empirical ( [19,20]) studies that focus on buyers' strategic sophistication in dynamic pricing contexts.…”

“…Our model is a major extension of an earlier model [13]-the earlier model only analyzes buyers' reactions to the price offers, whereas our present model incorporates the seller as a strategic price-setting player. In our model, the inventory is disclosed to all the buyers before they place their orders (see, e.g., [5,14,15]).…”

“…Our work is most closely related to [13] which, in contrast to the present study, examines strategic considerations on the buyers' side only. [13] leaves open the question whether the observed buyer behavior, and the systematic differences in buyer behavior with respect to capacity constraints could be attributed to the fact that the seller was automated.…”

“…[13] leaves open the question whether the observed buyer behavior, and the systematic differences in buyer behavior with respect to capacity constraints could be attributed to the fact that the seller was automated. This problem is not unique to their study; experiments in other decision-making areas (for a recent example, see [21]) already have shown how human-to-human interactions could make a crucial difference relative to human-to-bot interactions.…”

“…This problem is not unique to their study; experiments in other decision-making areas (for a recent example, see [21]) already have shown how human-to-human interactions could make a crucial difference relative to human-to-bot interactions. In the present context, because the dynamic pricing game in the experiments in [13] was iterated multiple times with the same set of buyers, there is the distinct possibility that their buyers might have learned basic features of the program used to determine the seller's pricing mechanism and responded accordingly. It also appears in many recent dynamic pricing experiments that focus on only one side of the seller-buyer market, assigning the other side to bots, which are not programmed to respond to the changes over time in the behavior of the genuine players.…”