DOI: 10.11648/j.ajam.20261405.20 ISSN: 2330-006X

A Machine Learning Approach to Assessing Tax Cuts and Economic Growth in Ghana

Isaac Ampofi, Daniel Arhinful, Alex Nti, Ernest Mawutor
Show PDF Cite
This study investigates the impact of tax cuts on Ghana's economic growth using a machine-learning approach applied to thirty-five years of macroeconomic data. The analysis focuses on the corporate income tax (CIT) rate, the personal income tax (PIT) rate, foreign direct investment (FDI), inflation, and the sectoral contributions of oil, gold, and cocoa to gross domestic product (GDP). Three feature-engineering strategies namely original, first-differenced, and ratio-transformed variables were used to train and compare an Extreme Gradient Boosting (XGBoost) regression model. The model trained on ratio-transformed features achieved the strongest predictive performance, with a coefficient of determination (R 2 ) of 0.9737, a mean absolute percentage error (MAPE) of 4.12%, and a mean percentage error (MPE) of 0.0986%, substantially outperforming the models trained on the original and first-differenced features. SHapley Additive exPlanations (SHAP) were then applied to interpret the model's predictions. The SHAP analysis shows that higher CIT and PIT rates are consistently associated with lower GDP growth, while FDI inflows and commodity export revenues (particularly gold and oil) are associated with higher growth. These findings suggest that reducing corporate and personal income tax rates could support economic growth in Ghana, but that the benefits of such cuts are likely to be realized only alongside complementary policies that improve investment efficiency, strengthen tax compliance, and add value to commodity exports.

More from our Archive