Objective Sustainability Assessment in the Digital Economy: An Information Entropy Measure of Transparency in Corporate Sustainability Reporting

Abstract

The Internet is now a central enabler for sharing sustainability information. Yet, such enablement is complicated through an exponentially increasing array of information. What is lacking in the digital economy are objective and transparent mechanisms to provide reliable assessments of the published sustainability information in a timely and efficient manner. In addressing such limitation, this research proposes an objective automated mechanism for measuring transparency in sustainability reporting using an information entropy-based approach. Through text-mining methods and expert validation, the study built a sustainability dictionary corpus and then applied the corpus for objectively assessing the relative entropy between the probability distributions of words in the sustainability dictionary and those in corporate reports. To demonstrate its effectiveness, the mechanism was empirically applied to compare sustainability reporting of organizations in the energy sector. Here, the research effectively compared cartels with non-cartels by assessing the sustainability reports of major OPEC (Organization of the Petroleum Exporting Countries) and non-OPEC producers spanning a three-year period and found consistent differences in transparency between the two groups. The findings demonstrate likely normative transparency pressures on disaffiliated producers for which cartels may be immune. The automated mechanism holds important theoretical and practical contributions to the field of sustainability as it provides a rapid and objective means for textual analysis of sustainability information, thus promoting transparency in sustainability reporting in the rapidly evolving digital economy.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 32 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.