Statistical Learning as a Tool for Optimizing the Level of Excise Tax of Mineral Oils in Slovakia

Excise taxes are one of the oldest taxes in the world. Countries have used these taxes to ensure profits of public finances. However, countries also have used these taxes as an indirect tool for removing negative externalities which cause damage the environment. One of the most significant excise taxes is the excise tax on mineral oils. This tax generates circa 10 per cent of all taxes coming to the public budget of the Slovak Republic. In Slovakia, the tax burden is the highest in our region. The state can influence the final price of gasoline products; however, it does not want to lose tax profits. It is due to the fact that these taxes generate remarkable profit in the Slovakian budget. We believe that there is a space for decreasing this tariff what would cause the decrement of gas and diesel prices. In this paper authors suggest a way how to use statistical methods based on linear regression and neural networks for modelling the decrement of excise tax tariff on gasoline with the same tax profits. We also suggest methodology of identification and modelling of these factors influencing the excise tax revenue.