No abstract

No abstract

Merit-rating structures in automobile-insurance systems require the insured to decide whether to file a claim for an accident when he is at fault. This decision can only be analyzed in the light of future developments and future decisions, and thus must be formulated as a sequential decision process that can be solved by dynamic programming. If future developments are estimated in light of new experience, the decision problem can be analyzed by adaptive programming. The approach is illustrated by sample calculations providing a number of observations that are of interest from the insurance industry's point of view.

A model is developed for planning optimal production and employment levels in multiproduct, multistage production systems. The market requirements for each product over the planning period are given. Backorders and/or shortages are permitted. Backorders and shortages must be considered in order to determine the amount of each product's demand that should be filled, backlogged, or go unsatisfied if the production capacity is insufficient to fill all market requirements. Backorders and shortages, on the other hand, are desirable under certain dynamic market conditions.

Coordinating information and material flows are key to effective supply chain management. The complexity of interactions in and the uncertainties surrounding supply chains make such coordination difficult. However, coordination can be realized by optimizing the flows in supply chains with analytical approaches. A mixed integer programming model is presented to support the tactical decisions of ordering, producing and transporting under various conditions of information availability at the loci of decision making. The model is applied to a modified version of MIT's well known Beer Distribution Game. The performance of the modeling approach is contrasted with the results of human decision making under identical conditions and underlines the enormous potential for performance improvement analytical decision support can provide. Several methodological aspects for coping with the difficulties of solving rather large mixed integer models are presented and it is shown that they can contribute significantly in dealing with the inherent computational problems.