Optimal portfolio choices to split orders during supply disruptions: An application of sport's principle for routine sourcing

• Published in: Decision sciences 2022-12, Vol.53 (6), p.1068-1087
• Main Authors: Padhi, Sidhartha S., Mukherjee, Soumyatanu
• Format: Article
• Language: English
• Subjects: Cricket; Disruption; Duckworth–Lewis–Stern method; mean–variance decision‐theoretic model; Polls & surveys; portfolio of risky options; Quantiles; Regression models; Risk; Risk factors; risk preference structures; Risk preferences; Side effects; Sourcing; Suppliers; Supply; Supply chains; supply disruption
• ISSN: 0011-7315; 1540-5915
• DOI: 10.1111/deci.12511

Sourcing in the face of supply chain disruptions has been one of the most challenging tasks in supply chain management, particularly when such disruptions occur due to natural calamities, such as flood, fire, and earthquake, affecting both the primary and the backup suppliers. Invariably, such disruptions lead to reduced supply from the primary supplier, encouraging the supplier to place fresh orders with the backup suppliers. In order to mitigate the adverse effect of supply disruption, in this article we use the concepts underlying the well‐known Duckworth–Lewis–Stern method, used in cricket, to revise the supply target of the primary supplier and to decide a target for the backup supplier. We simulated the supply disruption scenarios in an experimental setting by conducting a two‐round questionnaire survey among 300 purchase managers. The means and variances of the participants’ estimates of probabilities of meeting the revised targets within the scheduled time for various model‐generated supply scenarios were used to find the participants’ risk preferences. In the second‐round survey, the participants, clustered in groups of 10, ranked their own risk preferences. These ranks were used to find the optimal portfolio choices. Finally, we validated the theoretical predictions for the risk options using two approaches—one, at the group level by estimating the within‐ and the between‐group risk preferences of buyers, and, two, at the aggregate level, by considering all the participants, fitting quantile regression model to the experimental results, and estimating the risk preference structures for different quantiles of the relative risk–return trade‐off distributions.