Public Debt Forecasting in Kenya: Integrating Macroeconomic Variables with Arimax and Extreme Gradient Boosting

In recent years, public debt in Kenya has been increasing rapidly, raising concerns about its sustainability and economic implications. As of July 2024, the public debt was roughly 10.6 trillion Ksh, reflecting a 2.9 percent increase from June 2023 when it stood at 10.3 trillion Ksh (Statista, 2024). This trend highlights the urgent need for accurate forecasting to ensure effective debt management and economic stability. This study aims to forecast Kenya’s public debt by comparing a statistical model (ARIMAX) and a machine learning model (XGBoost). Macroeconomic indicators i.e. GDP, inflation, exchange rate and interest rate, are incorporated to analyze their impact on debt trends and evaluate the predictive performance of both ARIMAX and XGBoost. Additionally, scenario simulation is employed to assess how varying economic conditions affect public debt trajectories. The decision to compare ARIMAX and XGBoost is based on their distinct strengths and modeling assumptions. ARIMAX is effective in capturing linear relationships and temporal dependencies, while XGBoost is capable of detecting complex non-linear interactions and hidden patterns in the data. By comparing these two methods, the study aims to evaluate their respective forecasting performance, understand the influence of macroeconomic indicators on public debt, and determine which approach provides more accurate and policy-relevant predictions under various economic scenarios. The study utilizes historical data from 2001 to 2023, obtained from the Central Bank of Kenya (CBK) and Kenya National Bureau of Statistics (KNBS). By integrating statistical modeling, machine learning, and scenario simulation, this research aims to provide a comprehensive forecasting framework to support policymakers in making informed decisions for sustainable debt management.