Akademik Veri Yönetim Sistemi
Borsa Istanbul Review, cilt.25, sa.6, ss.1359-1377, 2025 (SSCI, Scopus)
This study employs machine learning and explainable artificial intelligence to examine the impact of working capital strategies—aggressive, moderate, and conservative—on Tobin's Q and EBITDA (earnings before interest, taxes, depreciation, and amortization), identifying key financial indicators for each approach. When the LightGBM algorithm is run, the R2 values for Tobin's Q are 57 percent (aggressive), 40 percent (moderate), and 55 percent (conservative) and, for EBITDA, the R2 values were 43 percent, 60 percent, and 60 percent, respectively. SHAP-based analyses reveal that Tobin's Q is predominantly affected by macroeconomic variables, especially in aggressive and moderate strategies, while EBITDA is mainly determined by operational efficiency and liquidity indicators across all strategies. The findings indicate that advanced algorithms—such as random forest, LightGBM, and XGBoost, when paired with SHAP explainability—capture the complex dynamics of working capital management more effectively than traditional approaches. Practically, these insights can help firms optimize liquidity, profitability, and debt policies to enhance sustainable competitive advantage.