Seller preferences for risk seeking and limited information in an evolutionary price demand game

The paper introduces evolutionary dynamics into a two-agent price demand game, in which sellers observe past period transactions before announcing a price, and buyers either accept or reject the announced price. Under the assumption of homogeneous clients, and large enough number of past-period observations, the process almost surely converges to a stable or continuously-reoccurring convention. Increased risk seeking in the class of sellers results in a long-term convention price at least as close to the buyer valuation as when the class of sellers is more risk averse. However, increasing risk seeking among sellers is often not feasible. I show that introducing imperfect information into the game by limiting the number of past period observations can also result in long-term convention prices closer to the buyer valuation, thereby increasing ex-ante expected utility among sellers. Thus sellers may have preferences for limited rather than perfect information of past transactions. However, buyers are never made better o ¤ by limiting seller memory. Preliminary and incomplete. I appreciate the comments of Ani Guerdjikova, Finn Christensen, and mem-bers of Cornells grad student workshop in economics. All mistakes remain my own. Contact information