Is there nonlinear causality between grain and meat prices?

The relationship between input and output products has always been of interest in an agricultural production. However, according to our knowledge, none of the previous studies have analysed nonlinear Granger causality between grain and meat prices. It has been widely noted in the literature that a linear approach to causality testing can have low power in the case of nonlinear relationships. Since many financial and economic data exhibit significant nonlinear features, nonlinear causality tests should be included in the analysis. Otherwise, important characteristics of the investigated relationships, which potentially could be exploited to build an effective predictor, might be overlooked.

In the research futures contracts from the Chicago Mercantile Exchange (CME) Group for three grains: corn, soybean, wheat and two meat commodities: live cattle and lean hogs are analysed. First, the linear Granger causality test is applied and the lack of linear causality relationships or the existence of only weak ones is demonstrated. Therefore, the outcomes of this study confirm the results published in other studies. Next, two nonlinear causality tests – the Hiemstra and Jones test and the Diks and Panchenko test are applied. They are used not only for the raw data but also the data filtered with the use of vector autoregressive (VAR) and Baba, Engle, Kromer, Kraft (BEKK) models in order to determine the nature of these relationships. Strong nonlinear causal relations between grain and meat prices are discovered. They have not been documented in the literature of the subject yet. It should be noted, that the exposed relationships are of different patterns and features. In particular, in some cases the detected nonlinearities arise from the second-moment dependencies. Yet, nonlinearities of a different type are also found. Some pairs of the commodities seem to be linked more closely than the others. The strongest nonlinear relationships exist between wheat and meat prices. On the other hand, the results indicate the lack of causality or very weak causal relationships between soybean and meat prices. Most of the discovered nonlinear relationships are bidirectional i.e. if a certain grain price is the Granger-cause of a meat price, then this meat price is the Granger-cause of the grain price at the same time.

Since, the consequence of the presence of causal relationships is the ability to predict time series, the obtained results show that it is possible to predict meat prices based on grain prices, and vice versa.