Corporate Risk Disclosure Practices and Their Impact on Investor Decision Making

Abstract

This research introduces a novel computational framework for analyzing corporate risk disclosure practices and their impact on investor decision-making, moving
beyond traditional content analysis by integrating natural language processing, network theory, and behavioral finance simulations. While existing literature examines
disclosure content and market reactions, our approach uniquely models the structural and semantic properties of risk disclosures as complex information networks,
where individual risk factors are nodes and their co-occurrence patterns create edges
with varying weights. We develop a proprietary corpus of 10-K filings from SP 500
companies (2018-2023) and apply a hybrid methodology combining transformerbased semantic embedding (BERT) with graph convolutional networks to extract
latent risk interdependencies that are not apparent through manual reading or
keyword counting. Our findings reveal that the network topology of risk disclosures—specifically, measures of centrality, clustering, and path length between risk
concepts—significantly predicts investor attention allocation, as measured by eyetracking experiments with professional investors, and subsequent trading behavior
in simulated markets. We identify a ’disclosure complexity paradox’: firms with
more interconnected and densely clustered risk narratives experience lower investor
comprehension but higher perceived managerial competence, leading to asymmetric market reactions. Furthermore, we demonstrate that machine learning models trained on network features outperform traditional sentiment and readability
metrics in forecasting abnormal returns following disclosure events. This research
contributes to the accounting, finance, and information science literatures by providing a new theoretical lens—the network theory of risk communication—and an
original analytical toolkit for assessing the informational quality and economic consequences of corporate transparency. The implications extend to regulatory policy,
suggesting that standard setters should consider mandating not just the presence of
risk factors, but also guidelines for their structural presentation to optimize investor
decision-making