Abstract

The appearance in the late 1990s of commodity index investing funds and the increasing popularity of commodities as an asset class has led numerous market participants and academics to blame index investing for creating an artificial demand for commodities and thus to inflate commodity prices. While the issue has been widely discussed and attracted a lot of attention from academics, previous research focused on correlation analysis and linear Granger causality tests to investigate whether commodity index investing had a significant impact on commodity prices. While linear Granger causality tests have been widely used in empirical finance and econometrics and are standard tools in an econometrician 19s toolbox, most recent research has focused on developing non-linear, non parametric tests. In this study I take a new approach to testing the hypothesis that commodity index investing Granger caused commodity futures prices to increase by using the test proposed by Diks and Panchenko (2006) which uses a non-linear non-parametric framework. I perform both the linear Granger causality and the Diks and Panchenko tests for 12 agricultural markets for which index commodity investing data is available from the CFTC, for the period 2006 to 2012. Overall, the empirical results provide limited evidence to support the hypothesis that index commodity investing caused a spike in commodity prices but highlight the importance of considering non-linear effects in empirical studies with financial time series.