A general solution for the newsboy model with random order size and possibly a cutoff transaction size

Abstract

In a recent paper, Dekker et al developed a solution procedure for the newsboy problem with a cutoff transaction size (CTS), such that customers with orders larger than the cutoff value are satisfied in an alternative way, against additional cost. A compound Poisson demand with discrete order sizes is assumed, and a computational procedure developed to identify the optimal solution. In this paper, we discard all distributional assumptions, and, given that order size is random, only the first three moments of the arrival rate and the order size are specified. A general optimal solution is developed, using Shore's piece-wise linear approximations. For cases where a CTS exists, the derived optimal solution is extended to also identify the optimal CTS. The new solution approach requires minimal distributional information, and its merits relative to current approaches are discussed and numerically demonstrated. The new methodology, based on a ‘Calculus of moments’ combined with a distributional approximation, may be easily extended to other computationally intractable problems.

Keywords:

• calculus of moments
• distribution fitting
• inventory analysis
• newsboy problem
• Shore's approximations

Acknowledgements

I am indebted to an anonymous reviewer, whose insights and comments helped appreciably in improving the quality of the paper.