Measuring Financial System Coherence

Abstract

Public firms produce multiple simultaneous narratives: filings for regulators, earnings calls for investors, product claims for customers, and strategic statements for the market. This project proposes that the geometric distance between these narratives can serve as a measurable indicator of organizational coherence.

Using public filings, earnings-call transcripts, and market data, the paper constructs stakeholder-specific narrative distributions and measures their divergence using Wasserstein distances. The project defines three candidate metrics: Internal Alignment, measuring the distance among stakeholder narratives; External Alignment, measuring the distance between the firm narrative and the market environment; and the Index of Disbelief, measuring divergence between self-description and external attribution.

The hypothesis is that rising narrative divergence provides leading signals of financial distress, narrative fraud, succession breakdown, or failed integration before traditional indicators fully register the problem.

Fields and Methods

financial economics, NLP for financial disclosures, accounting, organizational theory, optimal transport.

• EDGAR and transcript pipeline design
• stakeholder narrative distribution construction
• Wasserstein distance measurement
• baseline comparison against sentiment and topic models
• predictive validation against distress, fraud, succession, and M&A outcomes

Collaborator Profile

Accounting researchers, financial NLP specialists, empirical finance scholars, econometricians, organizational theorists, and optimal-transport researchers.

Validation Needed

• Build the EDGAR and transcript pipeline.
• Define stakeholder-specific narrative distributions.
• Compare coherence measures against sentiment, readability, and topic-model baselines.
• Validate against distress, fraud, succession, and M&A outcome sets.

Publication Posture

This page is a collaboration abstract, not a peer-reviewed finding. The research question is public so qualified domain scholars can evaluate the hypothesis, methods, and evidence needed to develop it.