Disaggregating Social Disclosure: Volume, Sentiment, and Topic‐Specific Tone in Relation to Market Capitalization

ABSTRACT

As social dimensions of ESG disclosure receive increasing attention from investors and regulators, understanding how different aspects of social disclosure are associated with market valuation has become an underexplored issue. Existing studies, however, often rely on aggregate ESG measures and provide limited insights into how capital markets respond to disaggregated social information. This study investigates the association between granular dimensions of social disclosure and market capitalization among Japanese listed companies. We analyze disclosure volume, overall sentiment, and topic‐specific directional tone across six critical social topics: Employee Health, Customer Value, Human Rights, Human Capital, DE&I, and Community Coexistence. We estimate two‐way fixed effects models using 27,974 firm‐year observations from 3455 Japanese listed companies over 2017–2025, assess sector heterogeneity, and conduct robustness checks. In M1, the Topic‐Related Fragment Share coefficient is positive and statistically significant for Employee Health, DE&I, and Community Coexistence, negative and statistically significant for Customer Value, and statistically insignificant for Human Rights and Human Capital. Overall Sentiment is positive and statistically significant across all six M2 specifications. In M3, Positive Tone Intensity is positive and statistically significant for Employee Health and DE&I but negative and significant for Customer Value, Human Rights, and Human Capital, and negative but only marginally significant at the 10% level for Community Coexistence; Negative Tone Intensity is negative and statistically significant for Customer Value and DE&I, positive and statistically significant for Human Capital, and statistically insignificant for Employee Health, Human Rights, and Community Coexistence. These conditional associations vary across sectors and support topic‐ and direction‐specific interpretation rather than a uniform valuation effect.