DYNAMIC MODELS OF SINGLE-LEVEL INVENTORY MANAGEMENT

Our research analyzes stochastic dynamic models applied to single-echelon inventory management, comparing them with classical models such as Wilsons and deterministic approaches. The study assumes that demand follows a log-normal distribution, an assumption supported by available data and validated through statistical goodness-of-fit tests. Several external factors influencing demand were integrated, including product price, substitute product price, and past demand observations. Empirical studies and numerical simulations indicate that stochastic dynamic models can reduce inventory management costs by 10-30% on average compared to traditional policies based on periodic replenishment or fixed thresholds, particularly in uncertain environments. Moreover, incorporating constraints close to real operating conditions allowed the identification of more effective management policies, especially in sectors such as pharmaceuticals and agri-food. The analysis shows that the selected external variables explain more than 90% of demand variations. Product and substitute prices influence demand at the 10% significance level, while past demand exerts a significant effect on future demand at the 1% level, thereby improving estimation accuracy.