| Even though steel is a relatively homogeneous good, considerable price dispersions are observed in the sales of a steel service center (SSC). Cost related factors, which by and large are time specific, cannot explain all the variations. Buyer specific factors and variables associated with bargaining and price discrimination, such as quantity of demand and freight, significantly contribute to the price dispersions.; The SSC in fact has a well-established mechanism for price discrimination—each buyer has to bargain with a sales agent on a transaction-by-transaction basis. We model such bargaining game as an alternating-offer game whose subgame perfect equilibrium converges to the generalized Nash bargaining solution. We find that large quantity buyers pay a lower price because they have a lower reservation value as well as a stronger bargaining power. Also, buyers paying more freight, buyers from the wholesale trade industry, and buyers receiving a better price quote from an alternative supplier get a more favorable deal because their reservation value is lower. Finally, buyers purchasing a product that constitutes a large portion of the total order have a stronger bargaining power and pay less.; Under the assumption that the SSC is rational, pricing through bargaining should be the profit maximizing pricing strategy. We simulate what the sales quantity and revenue would be if the SSC were to impose a fixed markup. In all cases considered, we cannot reject the hypothesis that the SSC is rational in adopting the optimal pricing strategy.; In addition to the static cooperative bargaining model, we also present a dynamic strategic price discrimination model. On each business day, the SSC first makes the order decision of choosing the size of additional inventory to order. If a buyer then arrives for a price quote, the SSC will make a take-it-or-leave-it offer at a price based on its expectation of the buyer's reservation price, which positively correlates with the SSC's spot purchase cost and negatively correlates with the buyer's fixed quantity of demand. In a calibrated example of the model, we are able to simulate several stylized facts observed in the actual data. |
