Assessing fuel tax revenues - Study on impact of reducing fuel taxes using artificial intelligence

Over time, Brazilian citizens have suffered with the recurring increases in fuel prices based on international price parity. The price fluctuation indirectly increases the ICMS tax - Tax on Operations related to the Circulation of Goods and on the Provision of Interstate and Intercity Transport and Communication Services - which is a tax revenues from states, considered their main source of revenues. If on one side, it benefits states, on the other, it negatively impacts the taxpayers and undermines some services, thus becoming a reverse problem, in which products and services are no longer sold or are replaced by lower-value products and consequently diminishing the state’s revenues. Therefore, the fluctuations in fuel final prices impact several production chains that directly influence the state’s revenues. Allied to this theme, there are frequent discussions about the change in the Brazilian taxation policy, which directly impacts the states’ revenues. However, little is discussed about the direct and indirect revenue projections caused by potential tax changes. In this context, this study aims to develop a type of ICMS tax collection simulator in the state of Rio de Janeiro. The developed tool uses public and historical fuel price data together with several deep learning models to generate an ICMS collection forecasting system. In this way, it is possible to modify fuel prices in order to predict potential impacts on ICMS tax collection. In this way, using the tool, it is possible to avoid a sudden drop in revenue, even from a tax increase. This study also highlighted the relationship among the collected revenues, the fuel’s price and consumption and the local economy activity indicator. Moreover it showed that keeping the fuel prices around a stable value over a year period results in no difference on state collection.