Application of Information Theory to Detect Changes in The Evolution of Fossil and Non-Fossil Fuel Consumption in The United States

When it comes to electricity generation, fossil fuels have traditionally been the primary energy source. The share of non-fossil fuels has grown significantly, however. From 1960 to 2015, the shares of renewable energy and... [ view full abstract ]

When it comes to electricity generation, fossil fuels have traditionally been the primary energy source. The share of non-fossil fuels has grown significantly, however. From 1960 to 2015, the shares of renewable energy and nuclear energy have increased from 6% to 10% and from 0.01% to 9% respectively in the US (source: Energy Information Administration (EIA)). Accordingly, the share of fossil fuels has decreased from 95% to 81%. Using Fisher Information, the main objective of this work is to track overall trends in state-wide electricity generation by energy source since 1960 and to detect the possible presence of regime shifts that could have destabilized the energy market. As an information theoretical concept, Fisher Information (FI) is particularly apt at tracking overall trends in complex and multivariate systems. Initially developed in 1922 by Ronald Fisher, FI is a measure of information content in data. Specifically, for this work, FI can provide information about the stability of the electricity generation system based on observed data. FI can be: (1) constant (and non-zero) over time, suggesting a system is evolving in a stable manner; (2) steadily decreasing, suggesting a certain loss of functionality in a system that tends to become more disorganized; (3) steadily increasing, suggesting a system is seeking stability; (4) sharp decrease, suggesting a fundamental regime shift in the system.

Using EIA data, we track the evolution of FI from 1960 to 2014 for three categories of energy sources: (1) fossil fuel (coal, petroleum and natural gas), (2) non-fossil fuel (nuclear, solar, wind, biomass, geo-thermal and hydroelectric), and (3) both fossil and non-fossil fuel. Overall, we found that the fossil fuel category is expectedly more stable than the non-fossil fuel category. Nonetheless, we clearly detect regime shifts in the 2000s for fossil fuels in different states, including California, Indiana, North Carolina and New Jersey. Although, we are not able to detect any regime shifts for non-fossil fuel at the moment (because it is still too unstable), we detect trajectory changes in many states, which may be a precursor to a regime shift. We then identify whether certain policies and national and world events are correlated with these patterns, and determine their effectiveness to displace fossils fuels as the preferred energy source for electricity generation.