Working Paper: CEPR ID: DP13004
Authors: Alessandro Bonatti; Gonzalo Cisternas
Abstract: A long-lived consumer interacts with a sequence of firms in a stationary Gaussian setting. Each firm relies on the consumer's current score--an aggregate measure of past quantity signals discounted exponentially--to learn about her preferences and to set prices. In the unique stationary linear Markov equilibrium, the consumer reduces her demand to drive average prices below the no-information benchmark. The firms' learning is maximized by persistent scores, i.e., scores that overweigh past information relative to Bayes' rule when observing disaggregated data. Hidden scores--those only observed by firms--reduce demand sensitivity, increase expected prices, and reduce expected quantities.
Keywords: Price Discrimination; Information Design; Consumer Scores; Signaling; Ratchet Effect; Persistence; Transparency
JEL Codes: C73; D82; D83
Edges that are evidenced by causal inference methods are in orange, and the rest are in light blue.
| Cause | Effect |
|---|---|
| consumer scores (D16) | firms' pricing strategies (L11) |
| firms' pricing strategies (L11) | consumer demand (D12) |
| score persistence (C29) | lower average prices (P22) |
| lower average prices (P22) | consumer demand (D12) |
| consumer manipulation of purchasing behavior (D18) | diminished firm learning from scores (L25) |
| hidden scores (C70) | reduced demand sensitivity (R22) |
| reduced demand sensitivity (R22) | higher prices (D49) |
| hidden scores (C70) | lower quantities purchased (L42) |
| score persistence (C29) | informativeness of scores (C52) |