An Analysis of Environmental and Economic Impacts of Fossil Fuel Production in the U.S. from 2001 to 2015

Abstract

The production and burning of fossil fuels is the primary contributor to CO₂ emissions for the U.S. We assess the impact of producing coal, crude oil, and natural gas on the environment and economic well-being by analyzing state-level data from 2001 to 2015. Our findings show that coal production has led to more CO₂ emissions and no significant benefit to economic well-being. Crude oil production has a non-significant impact on CO₂ emissions but is related to a lower poverty rate, a higher median household income, and a higher employment rate. Natural gas withdrawals have a positive impact on median household income. We discuss these findings in the context of current U.S. energy policies and then provide directions for future research.

KEYWORDS:

• CO₂ emissions
• economic well-being
• fossil fuels
• natural resources-environmental sociology

Notes

Notes

1 We check the robustness of results by analyzing the interaction effects for states that have production activities of all three forms of fossil fuel resources. The findings presented in Appendix 4 show that there is no interaction effect on CO₂ emissions and employment rate. However, producing all three fossil fuel resources together is related to less poverty rate and higher median household income. Because the magnitude of the significant coefficients is minor, we consider it might be more meaningful to focus on interpreting the individual impact of each fossil fuel resource separately.

2 The tests for multicollinearity within panels show no substantial problems. The VIF score is 1.01. The general criteria is that VIFs exceeding 4 warrants further investigation and VIFs exceeding 10 requires correction. To test for potential influential cases, we estimate the model by excluding each state in turn. The results are similar. We exclude each year in turn and the results are also similar. This indicates that there are no particularly influential cases. Moreover, we estimate the first-difference baseline model to test the spuriousness of the PW regression model estimates. The results are consistent and this analysis is not biased.