This study examines the relationship between corporate cash holdings and firm performance within the IT industry, which is characterized by intense competition and rapid technological advancements. We propose an integrated framework that combines principal component analysis (PCA), machine learning (ML) algorithms, and Shapley additive explanation (SHAP) values to estimate and interpret model outcomes. Based on 21,051 corporate financial statement data items from 2004 and 2023, the empirical evidence supports an inverted U-shaped relationship between cash holdings and profitability, suggesting that holding either too little or too much cash is suboptimal. Among the tested models, the random forest demonstrates the highest explanatory power (R²) and the lowest prediction errors (RMSE), outperforming the traditional ordinary least squares (OLS) regression by explaining 47% more variance. Our findings provide practical implications for researchers and stakeholders interested in enhancing corporate risk management and performance.

Management with high organization capital, which can be seen as an indicator of superior internal governance, can be expected to exhibit a preference for cash reserves to safeguard assets and mitigate the risk of underinvestment. However, external parties may see high cash reserves as a risk factor for the agency problem. Strong external governance can mitigate the preference of management with high organization capital for cash reserves. The empirical analyses show a positive association between the organization capital of U.S.-listed firms and their cash reserves when controlling for multiple variables. Furthermore, through employing the hostile takeover threat index, I reveal the disciplinary effects of strong external governance. This study adds to the existing literature investigating corporate governance that is useful in corporate decision making.