Green influence, economic realities, and political landscapes : understanding carbon tax determinants in Europe using machine learning

Abstract

This research develops classification trees, a machine learning method, to interpret and predict the implementation and variation of carbon taxes in Europe. As our planet’s temperature increases and the effects of climate change intensify, an increasing number of governments are implementing carbon pricing measures to reduce emissions. Carbon taxes are an interesting mechanism, as they are simple to introduce, generate revenue, and have been proven to reduce emissions. Despite these merits, as of 2020, only 21 countries worldwide have implemented carbon taxes, with 15 of those situated in Europe, and many of these taxes feature comparatively low rates. To better understand this intricate landscape, the classification trees are utilized to investigate two hypotheses concerning the effects of cabinet ideology and the influence of green parties on carbon tax implementation and price variation. By incorporating ideological features into the models, this research recognizes that carbon taxes do not exist in a political vacuum, and that the advice of economists and environmentalists only goes so far. The two models related to the price variation demonstrated the importance of the green parties performance in the election, although their presence in cabinet does not prevent price decreases. Interestingly, no cabinet that experienced a substantial vote share for green parties and did not have very low government debt decrease their carbon taxes. The models also indicated that being a Right-wing cabinet does not prevent price increases, and that earmark revenues are associated with carbon tax increases. On the other hand, the decision tree model for the tax’s implementation demonstrated the importance of the Left-Right cleavage when the GDP per capita (PPP) is lower than 39 thousand dollars. In wealthier countries, the creation of a carbon tax was associated with a strong liberal democratic system, the absence of a crippling debt burden, the occurrence of a natural disaster during the cabinet's tenure, and not being completely relying on fossil fuels for energy consumption. The ideological features were deemed relevant to the three models, further demonstrating the importance of political considerations in the realm of climate policy research.