Nonparametric estimation of market risk: an application to agricultural commodity futures

Purpose - While the extant literature is replete with theoretical and empirical studies of value at risk (VaR) methods, only a few papers have applied the concept of VaR to quantify market risk in the context of agricultural finance. Furthermore, papers that have done so have largely relied on parametric methods to recover estimates of the VaR. The purpose of this paper is to assess extreme market risk on investment in three actively traded agricultural commodity futures. Design/methodology/approach - A nonparametric Kernel method was implemented which accommodates fat tails and asymmetry of the portfolio return density as well as serial correlation of the data, to estimate market risk for investments in three actively traded agricultural futures contracts: corn, soybeans, and wheat. As a futures contract is a zero‐sum game, the VaR for both short and long sides of the market was computed. Findings - It was found that wheat futures are riskier than either corn or soybeans futures over both periods considered in the study (2000‐2008 and 2006‐2008) and that all three commodities have experienced a sharp increase in market risk over the 2006‐2008 period, with VaR estimates 10‐43 percent higher than the long‐run estimates. Research limitations/implications - Research is based on cross‐sectional data and does not allow for dynamic assessment of expenditure elasticities. Originality/value - This paper differs methodologically from previous applications of VaR in agricultural finance in that a nonparametric Kernel estimator was implemented which is exempt of misspecification risk, in the context of risk management of investment in agricultural futures contracts. The application is particularly relevant to grain elevator businesses which purchase grain from farmers on a forward contract basis and then turn to the futures markets to insure against falling prices.