Modeling U.S. Soy-Based Markets with Directed Acyclic Graphs and Time Series Econometrics: Evaluating the U.S. Market Impacts of High Soy Meal Prices

This paper demonstrates the application of a recently developed methodology, the combination of directed acyclic graphs (DAGs) with Bernanke structural vector autoregression (VAR) models, to model a system of U.S. commodity-related and value-added markets. As an example, the paper applies this methodology to a monthly system of three U.S. soy-based markets: the soybean market upstream and the two downstream markets for soy meal soy oil. Analyses of results from simulating the model's impulse response function and of forecast error variance decompositions provide updated estimates of market elasticity parameters that drive these markets, and updated policy-relevant information on how these monthly markets run and dynamically interact. Results suggest how a positive shock in U.S. soy meal price dynamically influences the soybean market upstream and the soy oil market further downstream