Intra-firm coordination under asymmetric information: Modeling, analysis, and applications

This dissertation, motivated by the experiences of a US-based semiconductor manufacturer, presents an integrated model of incentive problems arising in forecasting and capacity allocation in a multi-plant, multi-product context. The firm studied in this dissertation allocates capacities to the competing product groups based on the demand forecasts from the product managers and the capacity forecast of the manufacturing managers. The process is characterized by incentive problems. When the capacity is scarce, a product manager inflates his forecast to gain a greater allocation of capacity. A manufacturing manager, on the other hand, deflates his forecast to cover for the uncertainties manufacturing. The aim of this thesis is to develop an incentive mechanism that induces honest forecasting by all managers.;The first part of this dissertation is concerned with single statistics reporting, where each manager forecasts only the mean of his distribution. We propose a game theoretic framework to model the strategic behavior of the managers and design a mechanism (a bonus scheme and an allocation rule to allocate available capacity to the products) that elicits truthful information from all managers. We show that the structure of the incentive scheme is rather simple with easily calculable parameters. We also show that a large class of allocation rules are manipulable. A bonus is often required for elicitation of truthful information.;The second part of the dissertation generalizes the results of Part I by assuming that the firm can ask the managers to report multiple statistics about their respective distributions (for example, mean and standard deviation). We seek to design a mechanism that elicits truthful forecasting. We have characterized the structure of the truthful bonus functions and allocation rules under multiple statistics reporting. We have compared and contrasted our multiple statistics results with those of single statistics.;The third part of the dissertation discusses the implementation aspects of our model. We have shown that it is possible to implement our schemes within the existing structure of the firm.