Uncertainty spillovers and portfolio resilience in agricultural commodity markets: a quantile VAR and cross-quantilogram approach

Yesser Drira, S. Boutouria, Mouna Boujelbène

Abstract

This study investigates the impact of macroeconomic uncertainty and green finance on the return dynamics of ten key agricultural commodities and dairy products – including cereals, vegetable oils, soft commodities and livestock – over the period from January 2013 to January 2023. It examines how geopolitical risk, financial stress, Twitter-based Economic Policy Uncertainty (EPU) and green bonds influence commodity markets across different states of uncertainty. The empirical strategy combines a quantile vector autoregression (QVAR) framework and the cross-quantilogram (CQ) method to capture nonlinear, asymmetric and quantile-dependent effects. The QVAR model analyzes how shocks to uncertainty and green finance propagate across commodities at different points of the conditional distribution, while the CQ approach identifies directional and lead–lag dependencies. In addition, a quantile-based dynamic portfolio optimization framework is implemented using a baseline 12-month rolling window, complemented by 24- and 36-month estimation windows to assess the robustness of portfolio performance across varying macro-financial regimes. The results show that financial stress and green bond shocks exert significant long-term effects on several key agricultural commodities, particularly cotton, canola and sunflower oil. In the short term, positive geopolitical shocks increase sunflower oil returns, while negative shocks are associated with declines in sugar returns. The CQ analysis reveals a pronounced negative dependence between geopolitical risk and a broad set of commodities (canola, coffee, maize, cotton, dairy and meat) during bullish market phases, a relationship that reverses under less favorable conditions. The portfolio analysis further shows that the mean–CVaR strategy delivers superior risk-adjusted performance under elevated uncertainty, whereas minimum variance and risk parity are more suitable in stable environments. This study contributes to the literature by integrating quantile-based connectedness models with a dynamic portfolio optimization framework grounded in the empirical distribution of uncertainty. Unlike traditional approaches based on arbitrary binary classifications (e.g. calm versus crisis), the proposed framework captures heterogeneous market conditions in a continuous manner. The findings highlight the importance of accounting for asymmetric and distributional dynamics in commodity risk management and provide actionable insights for investors, fund managers and policymakers concerned with sustainable finance and global food security.

Source: semanticscholar

Read the AI summary, key takeaways and discussion on WOBR Quant Research.