AN EMPIRICAL EVALUATION OF CAPM PREDICTION BIAS AND MODEL ADEQUACY WITH PANEL EVIDENCE FROM THE INDIAN STOCK MARKET

Abstract This study provides a comprehensive empirical evaluation of the Capital Asset Pricing Model (CAPM) by examining the divergence between actual and predicted returns using panel data from Nifty 200 firms. While CAPM remains one of the most influential asset pricing models in financial economics, its empirical validity—particularly in emerging markets—has been widely debated. This research moves beyond traditional CAPM validation by incorporating prediction bias analysis, alpha persistence, and panel econometric techniques to assess model adequacy. Using Ordinary Least Squares (OLS) and panel data models, the findings indicate a statistically significant beta coefficient (β ≈ 0.9864, p < 0.001) and strong explanatory power (R² = 0.906), confirming the importance of systematic risk. However, the presence of statistically significant positive alpha (α = 0.1209, p < 0.05) reveals systematic prediction bias, suggesting that CAPM consistently underestimates returns. Panel data analysis further confirms the persistence of abnormal returns even after controlling for firm-specific heterogeneity. The study contributes to the asset pricing literature by demonstrating that CAPM is statistically robust yet economically incomplete in the Indian stock market. The findings have implications for investors, policymakers, and researchers, emphasizing the need for multi-factor and behavioral models in emerging markets.