Multiscale nonlinear Granger causality and time-varying effect analysis of the relationship between iron ore futures and spot prices

Abstract

Using the ICEEMDAN approach, the Granger causality test, and the TVP-VAR model, this research examined Granger causality and its time-varying characteristics of four types of iron ore daily spot and futures prices from January 2014 to July 2021. There are two major conclusions. (1) There exist several one-way nonlinear Granger causations from spot markets to futures markets or futures markets to another futures market, and the widespread two-way nonlinear Granger causations. Platts iron ore prices are more sensitive to new market information than others. The iron ore futures prices on the Singapore Exchange are also more sensitive to those on the Dalian Commodity Exchange. (2) The effect of one price on another has a concentrated influence, a brief duration, and a fast decrease in intensity. The effect's intensity, which shows periodic patterns, varies over time and is unstable in positive and negative directions. Therefore, we suggest that (1) nonlinear models are more reasonable for predicting iron ore prices; (2) better private decision-making and public intervention should be based on the multi-time scales variation features of causations between different prices; (3) principal iron ore consuming countries, such as China, should speed up the construction of more mature futures and spot trading markets for increasing price influence.