A postponement model for demand management

Ananth V. Iyer, Vinayak Deshpande, Zhengping Wu

Research output: Contribution to journalArticlepeer-review

37 Scopus citations

Abstract

In this paper, we analyze demand postponement as a strategy to handle potential demand surges. Under demand postponement, a fraction of the demands from the "regular" period are postponed and satisfied during a "postponement" period. This permits capacity to be procured to satisfy the postponed demands. A reimbursement per unit is paid to customers whose demands are postponed. The basic idea is that by preempting stockouts through demand postponement, we can reduce overall stockout costs. We formulate and solve a two-stage capacity planning problem under demand postponement. We propose a power range class of distributions to capture the nature of demand surges. We establish the scalability and conjugate properties of the power range distributions under demand postponement, which leads to a tractable analysis of the problem. We analytically solve the problem of determining the optimal regular and postponement period capacities, and the demand splitting rule to minimize the supplier's expected cost. We show that (a) the value of postponement may be significant depending on cost and demand parameters, (b) a postponement strategy may lead to reduced investment in initial capacity, and (c) it may be optimal to do no demand postponement over a range of demands even after observing a higher demand signal. We then relax several model assumptions and provide results for these extensions. We conclude with managerial insights.

Original languageEnglish (US)
Pages (from-to)983-1002
Number of pages20
JournalManagement Science
Volume49
Issue number8
DOIs
StatePublished - Aug 2003
Externally publishedYes

Keywords

  • Capacity planning
  • Demand postponement
  • Value of information

ASJC Scopus subject areas

  • Strategy and Management
  • Management Science and Operations Research

Fingerprint Dive into the research topics of 'A postponement model for demand management'. Together they form a unique fingerprint.

Cite this