Asymmetric Forecasting of Tail Risk in Micro-Cap Stocks Using Inclusive Finance Factors: Evidence from Explainable Boosting Machines

<title>Abstract</title> <p>This study examines whether inclusive-finance factors contain asymmetric predictive information for Micro-Cap stock tail risk. Motivated by the financing sensitivity of Micro-Cap firms, the study focuses on five inclusive-finance dimensions: financing demand, financing supply, financing price, financing efficiency, and financing risk. Changes in each factor are decomposed into positive and negative components, and an Explainable Boosting Machine (EBM) is employed to compare their out-of-sample forecasting performance. The interpretable structure of EBM is further used to examine nonlinear predictive patterns through Partial Dependence analysis. The results show that the predictive information embedded in inclusive-finance factors is directionally asymmetric. Positive changes in financing demand, financing price, and financing risk provide stronger warning signals for tail risk, whereas negative changes in financing supply and financing efficiency better capture deterioration in the financing environment. The Partial Dependence results further indicate that the relationship between inclusive-finance factors and future tail risk is nonlinear, range-dependent, and directionally heterogeneous. These findings suggest that decomposing inclusive-finance changes into positive and negative components helps reveal predictive information that may be obscured by raw factor levels. Overall, EBM provides both forecasting evidence and interpretable insights for monitoring and managing Micro-Cap stock tail risk.</p>