Extended Abstract

Agricultural commodities are vital in providing food and raw materials, making their price stability crucial. Post-2008, the agricultural market has experienced heightened price volatility due to global crises, like COVID-19 and the Russo-Ukrainian war, which disrupted supply chains and heightened geopolitical risks. This vulnerability has sparked interest in what causes price volatility in the agricultural commodity market.

The study analyzes the realized volatility of the global agricultural commodity index, represented by the iPath Bloomberg Agriculture Subindex (JJA), across calm and turbulent periods (COVID-19 and the Russo-Ukrainian war) from 28.09.2009 to 07.06.2023. Symmetric and asymmetric Heterogeneous Autoregressive (HAR) models, both univariate and multivariate, are developed, incorporating decompositions of realized volatility into continuous, jump, positive, and negative components as predictors. The models are extended with GARCH processes in the error terms to capture volatility clustering. The HAR framework assesses the impact of own and cross-market volatilities (crude oil futures, US dollar index, and stock market fear index) and jump components on agricultural commodity volatility at daily, weekly, and monthly horizons.

Symmetric HAR models show that own volatilities significantly increase agricultural volatility, with persistent effects at longer horizons, highlighting the importance of macroeconomic factors, such as supply chain dynamics and weather. Asymmetric HAR models reveal an inverse leverage effect at daily and weekly horizons, transitioning to a leverage effect at the monthly horizon. During the war, leverage effects are observed at daily and weekly levels, while monthly volatility effects become symmetric. Leverage effects at shorter horizons reveal heightened sensitivity to geopolitical risks, while symmetric responses at the long-term suggest that extended uncertainty creates more balanced market reactions.

Daily crude oil, exchange rate, and fear index volatilities significantly elevate agricultural commodity volatility. Weekly fear index volatility, however, has a stabilizing effect in the medium term. During the pandemic, weekly exchange rate volatility becomes significantly positive, while monthly volatility is negative, reflecting short-term reactivity and longer-term adjustments. During the war, weekly crude oil volatility significantly increases agricultural volatility, while monthly crude oil volatility exhibits a stabilizing negative effect. This dual role shows the complicated interdependencies between energy and agricultural markets under various geopolitical conditions. Concurrently, the monthly fear index volatility becomes significantly positive, highlighting the role of prolonged uncertainty.

We also evaluate the out-of-sample forecasting performance of our models, comparing them with benchmark counterparts and demonstrating that the proposed models achieve superior out-of-sample forecasts. Therefore, the inclusion of cross-market volatilities and crisis periods improves predictive accuracy. Additionally, models that incorporate volatility asymmetries and decompose cross-market volatilities into continuous and jump components provide the most accurate projections, emphasizing the importance of distinguishing between regular fluctuations and abrupt shocks when developing risk assessment models.

Our findings provide valuable insights into the dynamics of agricultural commodity volatility across varying time horizons and during calm and turbulent periods. The findings emphasize the importance of risk managers to design hedging strategies that address both short-term shocks and long-term volatility patterns.