Accounting Information Value Relevance for Equity Investment Decisions

Abstract

This research introduces a novel, cross-disciplinary framework that re-conceptualizes
the value relevance of accounting information for equity investment decisions by integrating principles from computational linguistics, behavioral finance, and complex
adaptive systems theory. Moving beyond traditional regression-based value relevance
studies, we propose that the relevance of accounting data is not a static property but an
emergent phenomenon shaped by the dynamic interplay between information structure,
investor cognition, and market ecology. We formulate three unconventional research
questions: (1) How does the syntactic and semantic complexity of financial disclosures,
analyzed as a linguistic corpus, modulate their predictive power for future equity returns? (2) To what extent do investor attention cycles, inferred from novel proxies,
act as a non-linear filter determining the temporal activation of accounting information’s relevance? (3) Can a bio-inspired, agent-based model of a market ecosystem,
where accounting signals serve as environmental nutrients, replicate observed patterns
of relevance decay and resurgence? Our methodology employs a tripartite approach: a
computational linguistic analysis of 10-K filings using entropy-based complexity metrics; the construction of an ’attention diffusion index’ from non-traditional data sources
including specialized financial forum discourse and search query volatility; and the development of an agent-based simulation where heterogeneous investors, modeled as
foraging agents with varying information processing heuristics, interact with a stream
of accounting ’signals’. Results from an empirical application to a longitudinal dataset
reveal that periods of high linguistic complexity in financial reporting correlate with
subsequent increased divergence in analyst forecasts and delayed market reaction, suggesting a ’digestion lag’ effect. The attention diffusion index demonstrates significant
predictive power for the short-term elasticity of earnings response coefficients. The
simulation model successfully generates endogenous cycles of value relevance, illustrating how the collective behavior of agents with bounded rationality can lead to the
episodic relevance of fundamental data. We conclude that the value relevance of accounting information is best understood as a path-dependent, ecology-driven process
1
rather than a deterministic input-output mapping. This paradigm shift suggests that
enhancing decision-usefulness may depend less on increasing information quantity and
more on optimizing its architectural design for cognitive accessibility and aligning its
dissemination with natural market attention rhythms. The study’s primary contribution is its original theoretical synthesis and its methodological departure from established practices, offering a new lens through which to examine the perennial question
of accounting’s role in investment decisions.