Title: Media influences on agricultural commodity pricing
Authors: Xinquan Zhou - Dublin City University (Ireland) [presenting]
Guillaume Bagnarosa - ESC Rennes School of Business (France)
Jagadish Dandu - Zayed University (United Arab Emirates)
Michael Dowling - Dublin City University (Ireland)
Abstract: We apply textual machine learning to 290,271 business news articles related to corn markets (2009-2020), to model the impact of news on corn futures pricing. Our approach allows the identification of seven distinct topics of corn news that well-describe the typical range of news coverage. We identify four topics on the fundamentals of corn markets around crop progress, weather impacts, exports, and USDA reports. We also identify three further topics on the relationship with wheat markets, soybean and biofuel markets, as well as financial market news related to corn. We first discuss these topics, and then show how the integration of news variables for each topic allows improved understanding of corn returns and price volatility. We demonstrate that news about financial markets, soybean-biofuels, crop progress, and exports, significantly contributes to explaining corn price dynamics. Our volatility analysis also demonstrates that soybean-biofuel and weather news, especially, contributes to the level of uncertainty in corn pricing. Our study demonstrates the value of refined analysis of news flow in agricultural commodity markets.