Pricing Information Cascades

Опубликовано 17 августа 2016, 20:43

We consider the problem of optimal pricing of a common-value product in the presence of social learning effects. A new product reaches the market and agents obtain private signals that partially inform them about the value of this product. Agents decide sequentially whether to purchase this product. Before making their own decisions, they also observe the purchasing decisions of agents who acted previously and learn from those actions in a Bayesian rational fashion. We address the problem of how a firm should price the product when taking social learning into account. Our first result shows that firms do best asymptotically if the firms select prices that lead the customers to learn the true value of the product. We show how the firm can induce learning at a low cost by inducing a vanishing fraction of the agents to act according to their private signals. We also show a lower bound on the agents' regret of $T^{2/3}$ for a society of size $T$. We finally show a pricing policy that achieves this lower bound.