How large-cap exclusion affects alpha and factor stability in the Korean FF3 model

Purpose The Korean equity market is characterized by extreme concentration, with the top five stocks accounting for approximately 25–45 percent of total market capitalization. This study examines whether such concentration structurally distorts the Fama and French (1993) three-factor model (FF3) and assesses the impact of excluding large-cap stocks on coefficient estimates and explanatory power. I construct the original FF3 model along with three alternative versions, namely FF3_E1, FF3_E3 and FF3_E5, which sequentially exclude the top one, three and five firms by market capitalization. Using one-to-one correlations of estimated alphas and betas, the sum and difference of squared coefficients, and adjusted R2, I evaluate the robustness of the factor structure. The results show that even after excluding large caps, the FF3 model maintains highly stable coefficient estimates and slightly improved explanatory power, affirming its structural robustness. However, the downward shift in the benchmark return due to large-cap exclusion induces a systematic rise in estimated alphas. These findings confirm FF3’s validity in concentrated markets while highlighting the need for careful alpha interpretation.