2164. Using Causal Inference to Assess the Value of Dual Sourcing
Invited abstract in session TD-7: Quantitative methods for systemic and climate risk, stream Risk Management in Commodities and Financial Markets .
Tuesday, 14:30-16:00Room: Clarendon GR.01
Authors (first author is the speaker)
| 1. | Thomas De Backker
|
| Research Center for Operations Management, KU Leuven |
Abstract
Recent disruptions from events like COVID-19, climate change and trade embargoes have driven firms to adopt strategies to implement supply chain resilience. One such measure is the adoption of multi-sourcing strategies, where components and products are procured from multiple suppliers to mitigate risks associated with failures from the primary supplier. We aim to investigate the effectiveness of dual sourcing in mitigating disruption risk and provide insight into which components benefit most. We differentiate between using a second supplier as a regular supplier (supplier diversification) or an emergency (backup) supplier, activated only if a disruption occurs. We employ causal inference methodologies to examine the causal effect of dual sourcing on stockout risk. This allows us to isolate the independent causal relationship between the presence of a second supplier and a component’s stockout vulnerability. A case study is conducted with an OEM, analyzing 597 components (22 dual-sourced), using historical stockout data(01/2022 - 01/2025). By controlling for other component characteristics and considering the potential mediating effect of inventory policies, we investigate the causal impact on the predicted number of stockout days over a predefined period. To inform recommendations for or against dual sourcing, we analyze conditional treatment effects with component characteristics to determine which types of components are more susceptible to the effects of dual sourcing.
Keywords
- Risk Analysis and Management
- Supply Chain Management
- Machine Learning
Status: accepted
Back to the list of papers