Consumer Reactions to Unobserved Changes in Price Schedules

Economic theory presumes that individuals respond to true marginal prices when deciding on the amount of goods and services they buy and many other economic decisions. However, learning about these marginal prices is often costly in terms of search time, cognitive effort or monetary outlays. This is likely to be true of price changes in subscription plans. Consumers may therefore opt to be satisfied with only an approximate knowledge of the relevant marginal prices. This paper presents an experiment that studies repeated consumer purchase and price information updating and acquisition decisions when parameters of the price schedule are serially correlated but unknown. Subjects have an option to acquire the pricing information at a cost, or otherwise just update their beliefs based on the observation of the total cost of purchase. We find the following: (1) conditional on information acquisition decisions, the model of Bayesian updating provides a good approximation for revealed mean beliefs about the per-unit price held by subjects who appear to understand the experiment and/or report their expected cost of purchase accurately; it is not a good approximation for other subjects; (2) the demand for information decreases with the cost of information, as expected; (3) controlling for Bayesian beliefs and cost of information, the demand for information does not vary with the length of the remaining time horizon in which the information can be used, contrary to the theoretical prediction; (4) large positive surprises in the cost of purchase in the most recent period increase information demand, whereas negative surprises decrease it, relative to the no-surprise baseline, which is contrary to the theoretical prediction