AI DISCLOSURE QUALITY AND FIRM VALUATION: A SERIAL MEDIATION FRAMEWORK OF CORPORATE GOVERNANCE AND INFORMATION ASYMMETRY

Abstract

The rapid integration of artificial intelligence into corporate operations has intensified the challenge investor’s face in distinguishing between firms with genuine AI capabilities and those engaging in symbolic AI narratives. This paper develops a conceptual framework examining whether and how artificial intelligence (AI) disclosure quality translates into capital market benefits in the context of an emerging economy. Drawing on Signaling Theory and the voluntary disclosure literature, we propose a serial mediation model wherein corporate governance capability enhances firm valuation through the sequential pathway of AI disclosure quality and information asymmetry. We argue that high-quality AI disclosures characterized by specificity, balance, forward-looking orientation, verifiability, and readability credibly communicate firms' substantive AI capabilities and reduce information asymmetry. The framework introduces algorithmic opacity as a distinct manifestation of information asymmetry arising from the inherent characteristics of AI systems: black-box algorithms, proprietary training data, and unpredictable model behavior. We contend that traditional governance mechanisms, designed primarily for standardized financial reporting, are insufficient to address this technologically rooted information gap. Instead, disclosure quality serves as the critical mechanism through which governance becomes economically visible to capital markets. The framework, if supported empirically, would offer guidance for corporate boards, investors, and regulators in Pakistan's evolving AI governance landscape, particularly in light of the National AI Policy 2025 and the Securities and Exchange Commission of Pakistan's emerging IT governance disclosure requirements. We outline a comprehensive research agenda for empirical testing of the proposed framework using archival data from the Pakistan Stock Exchange, including detailed methodological guidance for constructing AI disclosure quality and corporate governance capability indices. To our knowledge, this is among the first conceptual frameworks integrating corporate governance, AI disclosure quality, information asymmetry, and firm valuation in an emerging market context. The framework contributes to the literature by proposing a serial mediation model linking governance, AI disclosure quality, information asymmetry, and firm valuation; conceptualizing algorithmic opacity as a new manifestation of information asymmetry; introducing a multidimensional AI disclosure quality construct tailored to emerging markets; and extending signaling theory to AI-related voluntary disclosure in an institutional context characterized by high information asymmetry and limited analyst coverage.