Linear and Nonlinear Causality between Corn Cash and Futures Prices

Abstract This study investigates linear and nonlinear causality between the daily Chicago Board of Trade corn futures price series and each of seven regional cash series from Iowa, Illinois, Indiana, Ohio, Minnesota, Nebraska, and Kansas for January 2006–March 2011. Empirical results suggest bidirectional linear causality between cash and futures prices under a bivariate vector autoregressive model in differences (Bi-VAR-Diff) which consists of the futures and one of the seven cash series that are not cointegrated, and unidirectional linear causality from futures to cash prices under an octavariate vector error correction model (Octa-VECM) which consists of the futures and all of the seven cash series that are cointegrated. With linear relationships among prices removed using the Bi-VAR-Diff or Octa-VECM filtering, nonlinear causality is tested through a bivariate vector autoregressive model in levels (Bi-VAR-Lev) on residuals associated with the futures and a cash series from the linear models, and is found to be unidirectional from the futures market to Illinois, Indiana, and Ohio in general. Finally, a GARCH-BEKK specification is added to a Bi-VAR-Diff or the Octa-VECM to obtain residuals for the nonlinear causality test using a Bi-VAR-Lev, and the futures market leadership against Illinois, Indiana, and Ohio is still identified.