The impact of different sources of uncertainty on the Newsvendor model

This thesis explores the sensitivity and robustness of the Newsvendor, a fundamental model in inventory management. Despite its importance, a comprehensive analysis addressing the model's robustness to variations in distribution and economic parameters is still missing: existing literature focuses on specific scenarios, overlooking a general view of model sensitivity. Through an experimental methodology and a simulative approach, this study first verifies simulation reliability with known distributions. Then, it tests the theoretical model in counterintuitive scenarios, confirming that the cumulative probability of the demand distribution, linked to the optimal quantity, is function only of the economic parameters and it is independent from the distribution shape. This thesis also demonstrates that the optimal expected profit depends only on the portion of the distribution up to the purchased quantity. Finally, sensitivity analyses reveal that mean and standard deviation are the most critical factors influencing profitability, with higher means leading to increased expected profits and higher standard deviations resulting in reduced profits due to increased variability risk. Skewness and kurtosis effects are instead notably less significant. In conclusion, profitability is favourably impacted when the optimal quantity is slightly higher the mode of the distribution and in a highly concentrated area of the demand curve.