Essays on Textual Analysis of Corporate Disclosures

This dissertation consists of two studies. In the first study, I examine the direction and incidence of tone management at different stages over the CEO’s tenure and differentially for high-ability CEOs. From a textual analysis of the 10-K filings of US nonfinancial firms during 1993-2010, I compute abnormal tone as the residual term from a regression of the tone of words on its determinants, and then use it as a measure of strategic tone management – a linguistic tool often employed to influence the perceptions of investors and analysts.

I find that, on average, CEOs manage the tone downwards (more pessimistic) in the early years of their tenure than in the later years, and this relation is more pronounced for firms in high-litigation industries. Furthermore, the proportion of forward-looking sentences to total sentences in the 10-K filing is significantly lower in the early years. Since optimistic disclosures increase the likelihood of shareholder litigation (Rogers et al. 2011), these results suggest that new CEOs are concerned about litigation more in their early years, when the market is still assessing their ability. Interestingly, more able CEOs engage in upward tone management (more optimistic), more so in their early years than the later years. Finally, while departing CEOs also manage tone downwards (on average), I find that it is only the CEOs of balance sheet constrained firms (proxy for the limits to accruals management) who adopt a more optimistic tone in their final year. Overall, my findings suggest that it is important to study tone in addition to other tools (such as, earnings manipulation) to better understand CEOs’ impression management strategy at different stages of their tenure.

In the second study, I along with my collaborators construct a weighted word-count based measure to capture the quantity of a firm's 10-K narrative R&D-related disclosures, and document a persistent & significant (statistical and economic) negative association with subsequent firm profitability (ROA) during 1993-2006. These results stand in contrast to prior literature on 10-K narrative disclosures, across disciplines, where such disclosures have been found to convey meaningful information (via positive association) about current and future firm fundamentals. Also, the observed negative bias remains significant even after accounting for alternate explanations in this context. We argue that the unique characteristics of the R&D disclosure environment make it difficult for managers to develop skilled intuitive judgments about the outcome of their firms’ R&D investments, which in turn could adversely affect the accuracy and credibility of these disclosures. The assertion that the worst R&D performers are also
the most biased then seems to explain the negative association. The disclosure-type and features of the environment are thus important considerations in this area.