Essays on capacity management under uncertainty and information asymmetry.

Abstract

This dissertation studies firms' optimal operational decisions on capacity and production under uncertainty and information asymmetry about demand and supply. The first essay considers the optimal capacity investment decisions of a, firm facing uncertain demand and fixed costs associated with purchasing and salvaging capacity. I introduce a new concept, called (<italic>K₁, K₂</italic>)-concavity, and show that the profit-to-go function satisfies this property. It leads to a structural characterization of the optimal policy in which the initial capacity levels are segmented into multiple regions where purchasing/salvaging decisions can be simplified. It is also shown that this property is a generalization of concavity, <italic> K</italic>-concavity and sym-<italic>K</italic>-concavity, and how special cases of this problem reduce to well-known results. The second essay studies firms' production decisions under a competitive environment where firm's capacity is private information. I analyze how information asymmetry about competitor's capacity influences the set of equilibria in a two-period capacitated Cournot game. I show that, when price functions are stationary or decreasing across two periods, updating the beliefs in the second period does not affect either firm's production decision and each player pursues its short-run optimal (myopic) policy. However, when the price functions are increasing, a firm with high capacity has incentive to differentiate itself from the low type and it may sacrifice part of its short-run profits to reveal its type to its opponent. The last essay considers a manufacturer facing two options: (1) to procure components via auctions and integrate them in-house; and (2) to auction the integration to one of the component suppliers who obtains the other component by running an other auction. For option (2), it is shown that the second-stage bid is independent of the publicly announced winning price in the first stage; and the first-stage bid can be decomposed into the expected costs associated with integration and individual components. The condition under which option (2) dominates option (1) is explicitly characterized when the suppliers' cost distributions are uniform and the integration costs are linear. Particularly, when integration costs are constant, both options yield the same expected costs, but the cost is stochastically less variable in option (2).