Re-ordering policies for inventory systems with a fluctuating economic environment – Using economic descriptors to model the demand process

Abstract

Customer demand is highly dependent on external environmental factors that are mainly economic in nature. We build on the financial literature to identify two key economic factors that impact the demand process, which are the gross domestic product (GDP) and the economic policy uncertainty (EPU). We accordingly categorize the economic environment into states that jointly capture different realizations of the GDP and EPU. We resort to data from the technology, automotive and oil industries to further validate the state categorizations by illustrating the relationship/dependency between the economic states and the demand process. We utilize a regularization technique to capture the fluctuation in the economic environment by a continuous time Markov Chain. We observe that the demand rates are dependent on the economic environment and accounting for this dependency improves the performance of the inventory system, especially in the technology industry. We numerically investigate the impact of the fluctuations on inventory control systems with continuous (s, S) inventory control policies. Algorithmic approaches, based on matrix representations of the system, are presented to compute the inventory performance measures. Our numerical study shows that savings are highest for the technology industry where the demand process is highly dependent on fluctuations in the economic environment.

Keywords:

• Stochastic demand
• economic policy uncertainty
• gross domestic product
• correlated demand
• MMPP
• inventory systems
• ordering policies

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1 Available for download from: https://www.policyuncertainty.com

2 Our sample consists of 396 months. However, our total number of observations for the transition matrix is 395 observations since the transition from December 1984 to January 1985 is missing due to the exclusion of December 1984 from our sample.