Answers are grounded in this paper and cite their sources.
This study examines the application of extreme value theory (EVT) to the newsvendor problem under censored demand conditions. It investigates whether classical economic order quantity (EOQ) logic can approximate stochastic inventory decisions and identifies when such deterministic reasoning remains reliable. The objective is to minimize expected total cost (holding and stockout costs) by utilizing EVT to more accurately estimate the survival function of the demand tail, thereby refining the optimal order-up-to level (Y* ) in the presence of censored data. Traditional inventory models often fail to capture extreme demand fluctuations, particularly during stockouts. EVT, including the generalized Pareto distribution (GPD) and Fisher–Tippett–Gnedenko convergence, enhances tail-end demand estimation. Results show KM is generally more robust in balancing costs, while a split algorithm combining GPD for backorders and a parametric distribution (e.g. Poisson) for holding costs may outperform KM. EVT’s predictive potential for optimal reorder quantities with specific asymptotic distributions (e.g. Weibull, Gumbel and Fréchet) offers stability for distinct inventory scenarios. The study offers decision criteria for method selection, positioning EOQ-style logic as a useful approximation under moderate service levels and stable demand and EVT models as more appropriate in high-reliability settings where tail risk and extreme demand realizations dominate.
No takes yet. Share an insight, caveat, or question.
Westland et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: