Machine Learning-Based Predictive Analysis on Electronic Billing Machines to Value Added Tax Revenues Growth

Abstract

This paper explores the application of predictive analysis in assessing the impact of Electronic Billing Machines (EBMs) on Value Added Tax (VAT) revenue growth in Rwanda. EBMs represent a significant innovation in VAT revenue collection, yet negative taxpayer perceptions and a lack of understanding impede their full potential. Without predictive analysis, accurately projecting the growth in VAT revenues due to EBM implementation remains a challenge for the Rwanda Revenue Authority. Using machine learning techniques and historical data obtained from Rwanda Revenue Authority Annual reports, this research aims to develop a predictive system capable of forecasting the influence of EBMs on VAT revenue growth. Employing a descriptive survey design approach, secondary data analysis was conducted to gather information on VAT taxpayers, EBM users, VAT revenues, and other relevant variables. The study reveals a notable increase in EBM adoption among VAT payers over a six-year period, indicating successful governmental efforts. Time series analysis demonstrates a positive correlation between EBM usage and significant increases in both VAT and total tax revenues. The Seasonal Autoregressive Integrated Moving Average (SARIMA) model, specifically an ARIMA (1, 0, 0) model, is identified as suitable for forecasting VAT revenue growth due to its balance between accuracy and simplicity. The developed predictive system provides highly accurate forecasts of VAT revenue growth, facilitating informed fiscal policymaking and enhancing financial stability and revenue collection in Rwanda.