Examining Causal Chains Between Real US GDP and Energy Consumption Through Mixed Frequency Granger Causality Testing

This paper explores potential short-term causal relationships between energy consumption and real GDP in the United States from 1982-2006. Despite extensive study, there is no agreed upon consensus about the existence and nature of such a relationship. A primary issue in the literature is that existing studies are plagued by temporal aggregation issues due to inconsistency of data sampling (quarterly GDP and monthly energy consumption samples). We address this issue by conducting a Granger Test for Causality on rolling windows with parametrically bootstrapped residuals per Goncalves and Kilian (2004) in a bivariate Mixed Frequency Vector Autoregression (Ghysels, Hill, and Motegi 2014). We then compare the results to two standard VAR models with aggregated data, one with stock sampling and another with flow sampling, to highlight issues with aggregation. We conclude that there is more likely to be a causal chain from GDP to energy, but that major political events involving oil rich countries can create a simultaneously significant relationship.Bachelor of Art