Mitigating Systemic Shocks: AI-Augmented Forecasting in Quick Response Fashion Supply Chains

To cope with challenges raised by large-scale systemic shocks, which significantly amplify demand uncertainties, many fashion brands have adopted artificial intelligence (AI) for demand forecasting, but not all. Compared to traditional forecasting (TF), AI-augmented forecasting (AIF) stands out due to efficiency, accuracy, and adaptability. Capturing these features, we develop an analytical model based on Bayesian conjugate pair theory in a quick response (QR) fashion supply chain. This supply chain consists of a retailer, who determines the ordering quantity based on demand forecasts, and a manufacturer, who produces the required quantities in response. The retailer’s incentive for AIF adoption is identified by evaluating its weakening effect on the basic forecast error (BFE), which reflects the notable reduction in systemic shocks on the demand side. We also investigate the impacts of AIF-based forecasting from stakeholders’ perspectives. Challenging the common belief that AIF is welfare-enhancing, our analysis reveals a counter-intuitive insight: Even when AIF is costless, its adoption may reduce social welfare when the expected demand is sufficiently low. This outcome reflects the heterogeneous incentives of different supply chain members, including the retailer, manufacturer, and consumers. When AIF generates a positive social value but private incentives are misaligned, government sponsorship schemes can play a critical role in facilitating AIF adoption and resolving incentive conflicts among supply chain members. We further show how key factors in supply chain management, such as operational flexibility and contractual constraints, posed impacts on the value of AIF in QR fashion industries, and show that most of the results remain robust in the extensions. Finally, we discuss the implications of these findings and explain how they can help different fashion brands improve their operations in the presence of AIF.