Extreme Volatility in Agricultural Commodity Markets and Implications for Food Security
Athanasios Triantafyllou, George Dotsis, and Alexandros Sarris
Unexpected price changes and large upward/downward price swings have become very frequent and very common in the volatile agricultural markets. Sudden jumps in agricultural prices denote undesirable events for both policy makers and commodity producers, and create difficult situations for countries facing food security challenges. This is because unpredictable price increases raise the cost of food imports. The infrequent nature of these price changes makes it difficult to identify, anticipate and hedge them in a proper and timely fashion. Nevertheless, the nature of such events is important for food security planning. This is because low- income food dehcit countries, which number 54 according to the latest 2015 list of the Food and Agriculture Organization (FAO) of the United Nations, may find it difficult to import at reasonable cost what they need in periods of international food commodity price spikes. The purpose of
A. Triantafyllou (*) • G. Dotsis • A. Sarris
© The Author(s) 2017
G. Mergos, M. Papanastassiou (eds.), Food Security and Sustainability, DOI 10.1007/978-3-319-40790-6_9
this chapter is to explore the nature of large basic food commodity price changes using extreme value theory tools.
Insuring for food price spikes in the context of importing food commodities by food-insecure countries entails hedging strategies which depend a lot on the assumed underlying frequency distribution of such events. An assumed normal distribution of commodity price changes (or returns if percentage changes are considered) implies infrequent occurrence of extreme price events, which appears not to be consistent with the experienced frequencies of such changes. Hence it appears reasonable to assume that the distribution of food price returns is not normal. Mandelbrot (1963) is perhaps the hrst who showed empirically that commodity returns deviate significantly from the Gaussian normal distribution. He found that the tails of the empirical distribution of cotton price changes were much thicker compared to the tails of the normal distribution, implying that prices spikes are more frequent than predicted by simple Gaussian distributions. The empirical results in the literature regarding the behavior of commodity price changes indicate that the distribution of agricultural commodity returns are fat-tailed and that the large positive and negative changes occur more frequently than they would if the returns were drawn from a normal distribution. This is important for risk management in agricultural markets in the context of food security planning.
Recent studies show that tools from Extreme Value Theory are more suitable for modeling the risk in agricultural markets. Extreme Value Theory is a branch of statistics that deals with the modeling of extreme deviations and rare events using heavy-tailed distributions. Hilliard and Reis (1999) find that the returns of agricultural commodity futures are not normally distributed while Koekebakker and Lien (2004) show that agricultural price movements significantly deviate from the normality assumption because they exhibit sudden and unexpected jumps. In a recent article, Xouridas (2015) examines the empirical distributions of the returns of 60 agricultural commodities and finds that these distributions are significantly fat-tailed (exhibit a large kurtosis value). Other studies in the literature develop risk management tools that take into consideration heavy tails in commodity returns. Such risk management approaches are particularly important for food security planning, such as hedging import expenditures by low-income food-importing countries. Sam (2010) develops a nonparametric kernel method that accommodates fat tails and asymmetry in returns to calculate potential maximum losses in agricultural markets. Odening and Hinrichs (2002) find that the traditional value-at-risk methods fail to adequately capture the tail risk in the agricultural sector because of the fat tails in the empirical distributions of agricultural products. They show in their empirical analysis that the tools from Extreme Value Theory significantly improve the tail risk forecasts when used as a complementary tool to the traditional Value-at-Risk (VaR) methods in the agricultural sector. Morgan et al. (2012) provide further empirical support to Odening and Hinrichs’ (2002) findings, by applying some techniques of Extreme Value Theory in tail quantile-based risk measures (e.g., VaR and Expected Shortfall) applied to the estimation of extreme agricultural financial risk. Martins-Filho et al. (2012) propose fully nonparametric estimators for conditional VaR and Expected Shortfalls. They show that the proposed estimators have reasonable finite properties and they capture tail risk in the returns of agricultural commodities.
The remainder of the chapter is structured as follows. In Sect. 2 we describe some simple tools from Extreme Value Theory that can be used in order to quantify tail events and in Sect. 3 we apply these tools in the context of food import risk management for three basic food commodities. Section 4 sums up the conclusion to this chapter.