City Research Online

Abstract

This thesis investigates the asset pricing implications of different issues arising in the commodity futures market: the parameter uncertainty problem in commodity style integration, the role of investors’ sentiment and the limited attention effect in the commodity futures market.

We review the mainstream long-short strategies in commodity futures markets and the theories underpin them in the first chapter.

In the second chapter, we solve the parameter uncertainty problem arising in commodity style integration by utilizing a Bayesian framework. Commodity style integration is appealing because by relying on a composite signal that combines multiple commodity characteristics, the integrated portfolio ought to capture a larger premium consistently over time. A key decision that a style-integration investor face is which criteria or model to use for determining the style weights at each portfolio formation time. By adopting a Bayesian framework, it is allowed that the investor to account for parameters uncertainty. Focusing on the allocation problem of a commodity futures investor that seeks exposure to the hedging pressure, term structure, momentum, skewness, and basis momentum styles, we assess a Bayesian integrated portfolio versus the naive equal-weight integrated (EWI) portfolio and other sophisticated integration approaches. The results suggest a Bayesian optimization approach outperforms the others according to diverse performance criteria. The findings are robust to transaction costs, variants of the scoring schemes, longer ranking windows, and economic sub-periods analysis.

In the third chapter, we argue that the overall tone of recent news articles serves as a proxy for commodity futures investor sentiment. Studying a cross-section of commodities in the energy, agriculture, livestock, and metals sectors the findings indicate that media tone is able to predict subsequent commodity futures returns after controlling for well-known predictive signals such as hedging pressure, momentum, and roll yield inter alia. Sentiment-adjusted long-short portfolio allocation strategies significantly enhance the performance of traditional long-short commodity portfolios. Time-series and cross-sectional pricing tests suggest that the media tone has pricing ability over and above known commodity risk factors.

In the fourth chapter, we investigate the spillover effect of investors’ attention from the equity market to the commodity market. We argue that investors’ attention to a specific firm will spill to its related commodity futures market. This effect helps to construct a limited attention measure for commodity futures contracts. We show that the thus constructed measure is associated with higher returns of commodity futures in the future week, after adjusting for a battery of risk and characteristic benchmarks. Time-series and cross-sectional pricing tests suggest that the attention measure has pricing ability over and above known commodity risk factors.

Publication Type:	Thesis (Doctoral)
Subjects:	H Social Sciences > HG Finance
Departments:	Bayes Business School > Bayes Business School Doctoral Theses Bayes Business School > Finance Doctoral Theses

Export

Downloads