One slope does not fit all: Investigating Heterogeneity in Cointegration and Panel Methods in the Energy Intensity Literature

Abstract: This thesis aims to analyze the methods of existing studies on the relationship be- tween energy intensity and economic variables. I study panel data from 42 countries to examine the cointegration between energy intensity and GDP, capital stock, pop- ulation, and CO2 emissions as well as estimate their relationship with a pooled common correlated effects (CCEP) estimator. Then I investigate the heterogeneity by testing for cointegration and estimating coefficients for each country separately. The heterogeneity analysis shows that the results for cointegration and regression estimates on a panel data level do not cohere with the test results and coefficient estimates for the countries separately. The results imply that previous studies which have not accounted for the cointegration vector and slope coefficient heterogeneity may not have robust results and could explain why the earlier literature presents such conflicting results.