Modeling strategic investment decisions under sequential technological change

Modeling strategic investment decisions under sequential technological change

0.00 Avg rating0 Votes
Article ID: iaor199646
Country: United States
Volume: 41
Issue: 2
Start Page Number: 282
End Page Number: 297
Publication Date: Feb 1995
Journal: Management Science
Authors:
Keywords: allocation: resources, equipment, forecasting: applications, programming: dynamic
Abstract:

Strategic decisions to invest in new equipment are critical not only because of the large initial capital costs incurred but even more importantly because they affect future unit production costs, revenues, and the ability of the firm to perform operations that were not possible earlier. Thus these decisions determine the very competitiveness of the firm. Further, decisions regarding the choice of technology are very expensive to correct if incorrect decisions are identified. These decisions have become increasingly urgent and complex because the state of the art in technology is changing rapidly. However, many models available for evaluating these decisions are either too complex and inefficient or too restrictive in the number, types and the way appearance of future technologies is modeled. This research attempts to relax some of these restrictions using forecast horizon procedures to model capital investment decisions where any number of technologies may appear in the future with purchase costs and revenues that may vary over time. The appearance of these future technologies are considered uncertain with probabilities that may also vary with time. However, we assume that the order in which they appear is sequential, much like the different generations of microchips for microcomputers. We develop a new approach using nonunique terminal rewards to solve a dynamic programming model of the problem by introducing ‘converse’ difference functions and present an algorithm that is both simple and efficient. Despite the increase in state space from the use of ‘converse’ functions, we show that the computational burden of the algorithm does not increase. Numerical examples are given to illustrate our algorithm. Sensitivity of the optimal decision to changes in probabilities, costs, and revenues are also discussed.

Reviews

Required fields are marked *. Your email address will not be published.