Information and Decision Systems for Production Planning

Bowman [Bowman, E. H. 1963. Consistency and optimality in Managerial decision making. Management Sci. 9 (2) 310–322.] and others have proposed that regression models derived from a manager's past behavior can serve as a basis for future managerial decisions. This paper is concerned with a comparison of this approach to more traditional modes of decision making (intuition and mathematical optimization models) in an aggregate production scheduling context under various forecast error and forecast horizon environments. The regression models developed were also used to test Bowman's “Management Coefficients” theory at the axiomatic level, as well as to examine intuitive decision processes. To do this required verification that the regression models accurately specified an individual's decision process. A verification procedure, based on the concept of predictive testing, was used conjointly with statistical goodness of fit and cross validation to validate the regression models. The experimental results (1) supported Bowman's hypotheses, and (2) indicated the effects of forecast error and horizon length on the performance of the three decision making modes and on the formation of intuitive scheduling strategies. The theoretical and practical implications of the results are discussed.

• 저널Management Science · 22(3) · 359–370
• 토픽Scheduling and Optimization Algorithms · Industrial and Manufacturing Engineering
• DOI10.1287/mnsc.22.3.359
• 저자Herbert Moskowitz, Jeffrey G. Miller