It's about time: How recent advances in time series analysis techniques can enhance energy and climate research

Abstract

While an extensive body of social science research focuses on the interrelationships among economic development, energy systems, and the environment, studies involving the analysis of historical time series data are limited outside the field of economics. Along with the typical statistical challenges that accompany longitudinal data, time series data are likely to experience time-variation and bidirectionality in their relationships with one another. We address these issues and illustrate the utility of semiparametric, time-varying vector autoregressive (VAR) models for analyzing socioecological relationships. We analyze data for the United States from 1870 to 2017, and show how the relationships between GDP per capita, trade openness, fossil fuel intensity, and GHG emissions per capita have changed over time. We conclude by suggesting potential avenues for future research that use these methods.