Financial stress and realized volatility: The case of agricultural commodities

Abstract

Given recent debates about the financialization of commodity markets, we analyze the predictive power of financial stress for the realized volatility of agricultural commodity price returns. We estimate realized volatility from high-frequency intra-day data, where the sample period ranges from 2009 to 2020. We study the in-sample and out-of-sample predictability of realized volatility using variants of the popular heterogeneous autoregressive (HAR) model for realized volatility. We analyze the predictive value of financial stress by region of origin and by financial source, and we also control for various realized moments (leverage, realized skewness, realized kurtosis, realized jumps, realized upside tail risk, and realized downside tail risk). We find for several commodities evidence of in-sample predictive value of financial stress for realized volatility, consistent with the financialialization hypothesis. This in-sample evidence, however, does not necessarily extend to an out-of-sample forecasting environment.