Benoît Sévi, LEMNA, Université de Nantes
En collaboration avec Marie Bessec, Yannick Le Pen
Résumé : In financial markets, it is common to distinguish between hedgers, who take positions in futures contracts to reduce their risk, and speculators, who engage in futures markets to benefit from a risk premium. The most standard view (Hull, 2015) commonly assumes that hedgers do not speculate, i.e. their positions are not influenced by market prices. The recent paper by Cheng and Xiong (2014) provides strong empirical evidence that hedgers indeed speculate in agricultural futures markets. Using CFTC data on positions by categories of traders, the authors show that ‘non-commercial’ traders (hedgers) indeed respond to price changes. Hedgers short more futures contracts in response to price increases and reduce their short position as the futures price falls.
This result, however, is in line with early equilibrium models of Hirshleifer (1988, 1991) (see also the recent model in Cheng et al. (2015)) which assumes price sensitivity for hedgers. It seems that the result in Cheng and Xiong (2014) supports the view à la Hirshleifer, namely that hedgers respond to price changes. In other words, hedgers somewhat speculate in agricultural futures markets. From these two competing theories, it appears that the issue of hedger’s sensitivity to commodity prices ultimately resembles an empirical question.
We answer this question highlighting an important limitation of the analysis in Cheng and Xiong (2014) about the frequency at which observations are sampled. From CFTC releases, the positions of traders are publicly available at the weekly frequency, while futures prices can easily be accessed on a daily basis (or even at a higher frequency). For these variables, Cheng and Xiong (2014) aggregate data to create monthly variables thereby losing much information making their results only moderately robust.
In this paper, we make use of mixed-data sampling (MIDAS) as developed in Ghysels et al. (2006, 2007) to use higher frequency variables (prices) as explanatory variables for lower frequency variables (positions of traders). Our empirical analysis investigate the response of ‘commercial’ and ‘non-commercial’ categories of traders to changes in futures prices for a set of commodities quoted on the ICE or CME.
We show that the responses of hedgers to market prices are not significant in energy markets but it is in non-energy markets. Moreover, the response is significantly different for increase vs. decrease in prices which points to a potential behavioral bias in the trading behavior of hedgers
as a group. We rely on the disposition effect that has been developed in the behavioral finance literature to explain our findings with respect to asymmetry.