Oil Price Volatility is Effective in Predicting Food Price Volatility. Or is it?

• Author: Chatziantoniou, Ioannis

1. INTRODUCTION

   1.1 Context

   Spikes in the prices of internationally traded agricultural commodities and oil in 2008 and 2011 and the associated food inflation--not seen since the 1970s--have raised concerns about the impact of food price volatility on both consumers and producers across the world (FAO-OECD, 2011). While recent research has emphasised a strengthening of links between oil and agricultural commodities both in levels and volatilities (see, for instance, Algieri and Leccadito, 2017; Zhang and Broadstock, 2020) this has been based on in-sample evidence. In contrast, the extent to which oil price volatility impacts on forecasts of agricultural price volatility has not been established. It is this gap that we aim to fill in this paper by investigating the incremental real out-of-sample predictive content of oil price volatility on the volatility forecasts of several important agricultural commodities for the global economy.

   Understanding food price volatility is important for at least three reasons. First, volatile crop prices increase hedging costs of agricultural firms, which might, in turn, deteriorate their financial position (Wu et al., 2011). Second, in recent years, commodity markets (including energy but also agricultural commodities) have attracted the interest of financial investors, and have become more financialised as a result (Irwin and Saunders, 2011). Wu et al. (2011) and Gardebroek and Hernandez (2013) document the recent upsurge in investment portfolios that comprise agricultural commodities. This financialisation of agricultural commodity markets has led, so the argument goes, to increased volatility in crop prices, retarding investment and raising costs to famers via higher option and insurance premiums. Ordu et al. (2017) further highlight the potential impact of the financial-isation of agricultural commodities on household consumption and well-being, emphasizing the link between the price of food on one hand, and food security and malnutrition on the other. As the correlation between financial markets, energy and commodity markets becomes stronger, so does the impact from their interaction (see, inter alia, Vivian and Wohar, 2012; Silvernnoinen and Thorp, 2013; Sadorsky, 2014). Third, food price volatility is a key concern for policy makers, particularly those whose focus is food security in poorer economies (FAO-OECD, 2011). For example, in the wake of recent agricultural price volatility, G20 leaders requested the development of mechanisms to mitigate the risks that are associated with food price volatility to protect both consumers and producers, particularly those in low-income countries. As a consequence, the Agricultural Market Information System (AMIS) was established in 2011 to provide reliable market information and a forum to coordinate rapid responses to future crises (AMIS, 2020). Even in relatively prosperous economies, such as the UK, food price volatility is a pressing public policy concern owing to the importance of the agricultural industry to the rural economy and the cost of food for low income households (House of Lords, 2016).

   The intensity and extent of agricultural price volatility also has implications for the formulation and implementation of policies designed to mitigate its effect, which by their very nature, lead to significant and widespread effects in both the short and long run (Byrne et al., 2013). While almost all agricultural policy has a price stabilisation objective, explicit market intervention to address commodity price volatility, at both the national and international scale, has a long and largely chequered history (Gilbert 2011) ranging from the buffer stock schemes characteristic of now-defunct International Commodity Agreements, to the export restrictions that typified the response of many governments to the 2007-2008 price spike, which merely served to amplify the volatility on intentional markets (FAO-OECD, 2011). Most recent policy aimed at ameliorating the effects of volatility does so either indirectly (e.g. decoupled payments of the Common Agricultural Policy) or involves more market-based risk management instruments, such as the crop insurance programme in US and the promotion of futures markets among farmers (see inter alia House of Lords 2016).

   While the causes of agricultural price volatility are manifold, key among them is the linkage between the price of oil and agricultural commodities (Tyner, 2010, Tadesse et al., 2016) which Baffes (2013) argues has become stronger since 2005.

   There are two main channels by which oil price volatility can exert impact on the volatility of food prices. On the one hand, there is a direct impact; in many parts of the world commercial agriculture is oil-intensive reflecting a reliance on fertilizers, irrigation and mechanisation in the pursuit of higher yields. For example, Lott (2011) claims that 10 calories of oil are required to produce 1 calorie of food in the US. While few agricultural systems are as oil-dependent as they are in the US, commercial farming would not exist without it, with the result that oil price volatility is transmitted into the markets of all internationally traded agricultural commodities whose production depends on it (see, Gardebroek and Hernandez, 2013; Ordu et al., 2017, among others). On the other hand, there is an indirect channel arising from the demand for widely traded agricultural commodities (principally maize, soybeans, wheat and sugar) as biofuels (see, Harri et al., 2009; Zhang et al., 2010; Du et al., 2011). Biofuel demand, particularly in times of high oil prices, has been significant. FAO-OECD (2011) report that during 2007-09, biofuels accounted for some 20% of the global consumption of sugar cane and 9% of vegetable oil. Furthermore, the adoption of biofuel mandates (mandatory obligations to blend fixed proportions of biofuels with fossil fuels) in both the US and European Union are widely thought to have hardened the inelasticity of demand that gives rise to agricultural price volatility in the first place. In countries where biofuel mandates were most aggressive, effects were stronger, Hertel and Beckman (2011) reporting that in 2010 about 40% of maize production in the US ended-up in biofuels. Other developments, such as the use of index investments (in which oil is an important component) in agricultural commodity market trading (Tang and Xiong 2012) and the fact that oil and agricultural commodities are typically traded in the US dollar (Harri et al. (2009), give credence to the view that the prices of oil and agricultural commodities and the variability of those prices are inextricably linked in international markets.

   1.2 Commodities, volatilities and the futures market

   Overall, while the linkage between the volatilities of oil and agricultural prices is widely discussed in the literature, evidence is still rather scarce. This view is highlighted by Serra (2011) and Gardebroek and Hernandez (2013) who are able to document only a handful of studies investigating the transmission of volatility between energy and commodity prices. Despite the scarce empirical literature in the field, relevant studies occupy the full spectrum of potential linkages between oil and agricultural commodities, in the sense that the research relates to both spot and future prices and/or their respective volatilities. By focussing on the volatility of futures prices, the present study is positioned within an even scarcer strand of extant relevant work. Prominent in this literature is Algalith (2010) who provides evidence that higher levels of oil price volatility results in higher food prices. More particularly, Algalith (2010) argues that hedging oil quantities in futures contracts would likely reduce oil price uncertainty and as a consequence, reduce food commodity prices. In addition, Du et al. (2011) utilise crude oil, corn and wheat future contracts in order to investigate the impact of crude oil volatility on agricultural commodities, providing evidence of a positive link. More recently, Trujillo-Barrera et al. (2012) employ futures prices in order to examine spillovers from crude oil to a number of commodity markets in the US economy and show that there are important spillover effects running from oil price volatility to corn price volatility, which intensify during economic turbulent periods (specifically, the Global Financial Crisis of 2007-09).

   1.3 Forecasting commodity price volatility in the futures market

   Following from above, relevant studies that investigate the volatility of commodity prices in futures markets include Giot (2003) who was among the first to forecast agricultural commodity volatility--showing that implied volatilities of future contracts provide predictive gains to conditional volatility forecasts. Some fifteen years later, Tian et al. (2017a) forecasted the volatility of five agricultural commodity futures, namely Soybean, Soybean oil, White Sugar, Gluten Wheat and Cotton, using a two regime-switching Markov model. Their findings suggested that the dynamics of regime switching are capable of providing superior predictive accuracy relative to an AR(1) model or even a Markov-Switching AR(1) model.

   In turn, Tian et al. (2017b) estimate Heterogeneous AutoRegressive (HAR) models, with both static and time-varying parameters, considering the daily realized volatility, the range estimator, jump components (i.e. discontinuities in the underlying process) and other realized volatility (RV) measures of agricultural commodity volatility, such as the realized threshold multi-power variation and the realized threshold bi-power variation, as potential predictors. Their study concentrates on China and comprises six of the most rapidly expanding agricultural commodity futures markets; namely, soybean, cotton, gluten wheat, corn, early Indica rice and palm. Their findings show that while jump components play a role, it is the HAR models...