Price Discrimination with Private and Imperfect Information

Abstract

In this paper, I investigate the competitive and welfare effects of the improvements in information accuracy in markets where firms can price discriminate after observing a private and noisy signal about a consumer's brand preference. I show that when firms believe that consumers have a brand preference for them, then they charge more to these consumers, and this price has an inverse U-shaped relationship with the signal's accuracy. In contrast, the price charged after a disloyal signal has been observed falls as the signal's accuracy rises. While industry profit and overall welfare fall monotonically when price discrimination is based on increasingly more accurate information, the reverse happens to consumer surplus.

Imperfect Behavior‐Based Price Discrimination

“…This affects the first‐period demand sensitivity, thus shaping the impact of information accuracy over the equilibrium profits. In particular, the difference between our results and those in Esteves () stems from the fact that Esteves () adopts a one‐period model, whereas we adopt a two‐period model. Therefore, we have a demand‐sensitivity effect that is not present in Esteves model, and drives the difference in the results between the two models…”

“…Therefore, we allow for the possibility that the information that the firms receive about consumers’ past purchases is incomplete, that is, it does not cover all consumers. Therefore, lower imperfectness in our context means a wider scope of the information gathered, and not a lower probability of erroneous signals (as in Chen et al., , and Esteves, ) or a higher precision of the consumers’ segmentation (as in Liu and Serfes, , and Colombo, ). Furthermore, although the above‐mentioned papers assume a one‐period game, we consider a two‐stage game where in the first period the firms determine the composition of their turf and the consumers make their initial purchase.…”

“…However, in the real world, even if the firms engage in practices that aim to gather information about consumers’ past decisions, some information may end up not to be gathered. As argued by Esteves (), this may happen because marketing techniques are not perfect (e.g., some consumers may not be reached by surveys or questionnaires). Moreover, a manufacturer can sell its product both directly and by means of retailers.…”

“…In particular, Chen et al. () and Esteves () consider the case of noisy signals about the true preferences of the consumers. In these models, each firm receives a private and noisy signal about a consumer's brand preference: the less noisy is the signal, the more the model tends to perfect BBPD.…”

“…Indeed, in Chen et al. () and Esteves () all consumers are classified (maybe erroneously) as loyal to one firm or to the other, whereas in Liu and Serfes () and Colombo () all consumers are classified (correctly) into a number of subsegments within the Hoteling line and the imperfectness stems from the dimension of each subsegment. In contrast, in our model, the firm may receive no information about some consumers.…”