Hidden in plain sight: Unraveling compensation disclosure bloat with generative AI and its impact on executive compensation

Whether compensation contract design reflects efficient contracting or rent extraction is an ongoing debate in academic research and public discourse. We contribute to this debate by examining whether textual bloat in compensation contract disclosures is associated with excess CEO compensation. We construct a measure of bloat, defined as irrelevant, boilerplate, and redundant content, by summarizing firms’ Compensation Discussion and Analysis sections with a large language model for a sample of S&P 1500 firms during 2011–2018. In line with our hypotheses, we find a positive association between bloat and excess CEO compensation. We find no empirical evidence that governance characteristics explain the magnitude of bloat in firms’ compensation disclosures. Our findings suggest that bloated disclosures can be used as an instrument to obscure compensation levels that are unrelated to the economics of the firm.