Integrating Macroeconomic Indicators into Machine Learning Models for Used Car Price Prediction

Abstract

Accurate forecasting of used car prices plays a critical role for buyers, dealers, financial institutions, and policymakers. This study advances existing approaches by explicitly incorporating macroeconomic variables—such as fuel prices, interest rates, and inflation—into machine learning frameworks. While conventional regression models and standalone tree-based methods often fail to capture nonlinear interactions, this paper applies ensemble techniques, including Bagging, AdaBoost, and XGBoost, to address these limitations. Using a multi-source dataset of vehicle listings enriched with macroeconomic indicators, we demonstrate that integrating economic signals improves predictive accuracy, with AdaBoost achieving the highest performance (R² = 0.91). Beyond statistical results, our findings highlight the role of external market forces in shaping resale values, offering actionable insights for inventory management, sustainable financing, and policy design.