Understanding Commodity Markets via Return Predictability

Commodity futures markets allow commodity producers and consumers to engage in hedging; one market participant may unload her unwanted price risk to someone who is more suitable to bear that risk. Hedging helps the economy function by allowing producers and consumers to conduct their businesses with greater certainty. Because commodities are central to manufacturing and industrial processes, their prices are also intimately related to the macroeconomy. Given the importance of commodities, researchers have long sought to understand these markets. Multiple theoretical explanations have been proposed to capture the behavior of commodity futures returns, but there is limited agreement on which theory providesthe best description of the real world. Researchers also cannot agree on which variables best capture expected return variation in commodity futures, a central question in understanding the behavior of these markets. In this proposal, we take a step towardsresolving the above issues. We seek to evaluate competing theories and characterize expected return variation in commodity markets through return predictability tests.  We set out to address three research questions. Our first question seeks to distinguish among different theoretical explanations of commodity returns. Existing papers tend to test the predictions of a single theory, which makes it difficult to assess the relativemerit of competing theories. We contribute to the literature by proposing a testing framework that allows us to jointly examine multiple competing theories, and we apply this framework to test four theories of commodity returns. Our second question relates to the measurement of commodity expected returns, calculated as the predictable component of future realized returns. Extant work typically uses a small number of predictors, which potentially leads to limited characterization of expected returns. Our tests are based on a large and diverse set of variables, and our work helps to resolve the problem of measuring commodity expected returns. Our final research question aims to uncover the drivers of average return differences across commodities. Our approach differs from current research in that we start from a large set of candidate variables, and we use a data-drive method to reduce the dimensionality and build an asset pricing model. The variables that explain the cross section of commodity returns can reveal the most important risks in commodity markets, paving the way for improved theoretical explanations.