Volume 15, Number 1, June 2020
| An Empirical Study of Commodity Inventory Effect on Risk Management |
Abstract
Previous studies focus on the commodity inventory effect in modelling conditional return and volatility, and also its importance in optimal hedging. In this study we examine the importance of inventory effect in computing VaR forecasts for an equally-weighted commodity portfolio. To this end, accuracy and comparative predictive analyses are both used to evaluate the performance of portfolio VaR forecasts by univariate and multivariate models. The accuracy analysis includes the unconditional and conditional coverage tests, and the comparative predictive ability involves two economic loss functions.
The empirical investigation is conducted with eight commodity futures contracts and assessed with whole period and two sub-periods. The results indicate that, on the accuracy perspective, multivariate VaR model outperforms univariate model, and the inventory effect plays an important role in computing portfolio VaR forecasts with multivariate models. In terms of comparative predictive ability, the inventory effect is substantial in calculating market risk capital requirements for portfolio VaR forecasts. However, when assessing with the asymmetric linear loss function, the inventory effect loses its remarkable importance.
Keywords: Commodity Inventory Effect, Portfolio VaR Forecasts, Accuracy and Comparative Predictive Analyses, Market Risk Capital Requirements
JEL Classification: G11, G17