Three essays on the relationships between energy, economy and environment with heterogeneous panel data

Abstract

Energy, economy and environment (3E) are highly inter-connected as much as we call them ‘3E’. For the relationship between energy and economic output, energy is a driving force of economic output, since having been considered a major factor of production. In the energy-growth nexus theory, there is the conservation hypothesis that economic growth accelerates energy consumption. In case of energy-environment relationship, many international agencies have reported that the form of energy usage is the main driving force of environmental degradation, especially carbon emissions. Finally, the relationship between economy and environment also has been represented by numerous papers that cover an environmental economics hypothesis such as Kuznets curve.

This dissertation attempts to show one of the ways to analyze the relationship among 3E by conducting three 2E empirical analyses. First one is a causality between energy consumption and economic growth which focuses on investigating the relationship between energy consumption and economic growth considering the other factors of production. The second is the environmental Kuznets curve (EKC) hypothesis which estimates inverted U-shape curve relation between per capita income and per capita environmental degradation with a quadratic equation. The final one is called stochastic impacts by regression on population, affluence and technology (STIRPAT) analysis, which measure the impact of population, affluence, and technology on environmental variable based on decomposition analysis.

In three empirical analyses processes with panel data analysis, three methodological issues of panel data are addressed: 1) common factor restriction, 2) endogeneity from omitted variable problem and multicollinearity, 3) the choice of the null hypothesis, which is soundly related to cointegration analysis. The raw dataset consist of 28 OECD countries for annual period of 1990 to 2016. Common factors are measured how many factors exist by variable and extracted from raw data. Thus, the dataset does not suffer from common factor restriction.

The endogeneity problem within a stochastic approach is handled in the cointegration analysis. To overcome omitted variable problem, multiple panel cointegration tests are conducted. Combining the results of them, most appropriate cointegration analysis model is selected for three empirical research subjects. After then, to correct the endogeneity problem when occurring relevant variables are put in together as explanatory variables, fully modified OLS and dynamic OLS methods are adopted. The reason why two estimation methods are used is to check the robustness of estimated parameters.

In this dissertation, each of three empirical analyses is conducted to investigate each 2E relationship as covering econometric issues. For the causality between energy consumption and economic growth and the EKC analysis, it is covered whether the empirical models include fossil fuel usage and population variable as core explanatory variables. In STIRPAT analysis, the role of fossil fuel and renewable energy in the models are investigated. Consequently, the causality between energy consumption and economic growth and the EKC analysis without fossil fuel usage and population variable are more plausible than ones including them. In case of STIRPAT analysis, the estimated coefficient of fossil fuel usage and renewable energy consumption are statistically significant and intuitive in sign.

Summarizing the core empirical results, energy consumption positively affects economic growth which supports the growth hypothesis in the view of the causality between energy consumption and economic growth. The EKC hypothesis is empirically proven so environmental degradation will be stopped when the economic level of a country reaches a certain point. For the STIRPAT analysis of relationship between energy consumption patterns and carbon emissions, STIRPAT results show that fossil fuel usage and renewable energy consumption affect positively and negatively carbon emissions, respectively but a rather weak impact in case of renewable energy consumption.

Given that the growth hypothesis is adopted in the causality between energy consumption and economic growth and the EKC hypothesis is proven, and numerous existing papers have suggested the evidence that renewable energy consumption can accelerate economic growth, 100% of renewable energy deployment can be a nice solution. However, considering the results of STIRPAT analysis that renewable energy consumption has a relatively weak impact on carbon emissions, it is too early to imply excessive renewable energy encouraging policy. The gradual switching energy source is appropriate. This conclusion implies the applicability of separated three 2E analysis.