Quantifying the Quality of Integrity: CEOs, Auditors and Outcomes

“The lady doth protest too much, methinks,” observes Gertrude in Shakespeare’s Hamlet, implying that too much explanation can be a sign of insincerity. Research by Shane Dikolli, Bank of America Associate Professor in Business Administration, has found that when it comes to CEOs, the same logic applies. Analyzing some 30,000 letters to shareholders, Dikolli and several colleagues found that excessive explanations were associated with higher auditing costs and suggested lower trustworthiness of the CEO.

“CEOs who over explain are more likely to mislead investors, as well as say they are going to do something and then not follow through,” Dikolli says. “And so the auditor would have to do more work to make sure that the financial statements are fairly stated.” What’s more, the researchers found that over explanation is also associated with lower performance, as that lack of integrity presumably filters down into lower ranks of an organization, damaging trust among customers and suppliers.

The CEO Behavioral Integrity Index

Since they first published the results of their research, Dikolli and his colleagues — Thomas Keusch of Institut Européen d'Administration des Affaires (INSEAD), William Mayew of Duke, and Thomas Steffen of Yale — have garnered a good deal of attention for the paper, which establishes a CEO behavioral integrity (BI) index based on corporate communications. Their innovative technique provides one of the first reliable measures of executive trustworthiness.

Last year, after applying their index to German firms, they found the index would have been able to predict the massive accounting scandal of Wirecard, a company called the “Enron of Germany.” This June, Dikolli presented to the Foundation for Auditing Research in the Netherlands, explaining to some 150 auditors how linguistic analysis can provide new insights into accounting improprieties. “I wanted them to see that there are important consequences to integrity or the lack thereof, which they probably already intuitively knew, but which there hasn’t been any systematic evidence of in the past,” Dikolli says.

While it’s long been surmised that lack of integrity in a CEO can have negative repercussions on a firm, the quality has been notoriously difficult to measure — often based on unreliable employee surveys. “Even defining ‘integrity’ was an important starting point,” Dikolli says. “With all previous definitions of the term, people have been able to poke holes in them.” He and his colleagues built on a definition of “behavioral integrity” developed in the organizational behavior literature: the perception of whether a CEO’s behavior lives up to their past promises. “A third party’s perception of whether the CEO is going to honor their word directly affects how much they trust them,” Dikolli says.

With that definition in hand, they turned to linguistics as an innovative way to measure that disconnect, reasoning that CEOs were more likely to “protest too much” with long communications in cases in which they were trying to justify past actions. In particular, the researchers measured the frequency of “causation words” such as because, therefore and hence in the CEO’s annual letter to shareholders, words linguists have identified as markers of excessive explanation. They controlled for company characteristics, such as the size of a firm and amount of leverage, as well as characteristics of the CEO, such as narcissism, also revealed by linguistic analysis. “What’s left over is the unexplained abnormal level of explanation,” Dikolli says.

They then used this analysis to develop an index they dubbed the CEO behavioral integrity index. The measure identified that lower-integrity CEOs were linked to higher auditing fees, providing evidence that untrustworthiness has consequences, forcing auditors to do more diligence — and bill more hours — for unreliable CEOs.

While originally Dikolli and his colleagues weren’t planning on measuring firm performance, they found that a higher CEO BI was also associated with better financial performance of a firm. Dikolli surmises the increased performance is due to a “culture of integrity” within an organization. “We know from other work that if the tone at the top is strong and a CEO has high integrity, then it will pervade throughout the organization,” Dikolli says. “If creditors and customers can’t trust what the firm does because of weak corporate culture, they won’t get as many sales.”

Building on the Research

Since Dikolli and his co-authors published the paper, other researchers have used it to find even more consequences of CEO untrustworthiness. For example, one paper used CEO BI to examine loan rates and found that higher BI was associated with lower interest rates, presumably due to higher confidence in repayment.

Last year, Dikolli and his colleagues applied their principles to German firms, and they repeated the analysis with firms in the DAX 30, a stock market index featuring Germany’s 30 largest firms trading on the Frankfurt Stock Exchange. When the researchers analyzed English-language shareholder letters, they found the financial services company Wirecard to be “off the charts” in terms of its poor score, says Dikolli. “They got a negative score for two years — and we rarely got negative scores.” A few months before, in June 2020, news had broken that 1.9 billion euros were missing from the company’s balance sheet, a titanic case of fraud and misrepresentation.

While Wirecard may be an outlier in the extent of its financial improprieties, Dikolli believes that the CEO BI index could be a valuable tool in predicting negative outcomes that don’t normally show up in typical audits. For example, CEOs sometimes deliberately backdate stock options in order to reap more profits, an unscrupulous practice derided as “lucky stock grants” that lies outside the jurisdiction of auditors. Examining CEO integrity may help differentiate cases in which CEOs are manipulating equity grant award dates, rather than cases in which they are legitimately lucky in their timing.

Another potential application, says Dikolli, is in potentially identifying cases in which executives game the system by opportunistically earning profits on well-timed stock trades of companies they manage. “Our measure may be able to predict some of these bad outcomes in which an auditor is not there to prevent them,” says Dikolli, who believes it would be possible to automate the index using machine learning.

While the researchers only had access to public communications by CEOs, Dikolli stresses that these techniques could become even more powerful when applied to the vast amount of privately available documents auditors run across in a typical audit. “There may be ways they could explore that data to learn more about the organization and the behavioral characteristics of the CEOs they don’t otherwise know about,” Dikolli says.

Such flags could help them identify cases in which the CEO is “protesting too much” privately as well as publicly, and where they may need to take more care to ensure corporate integrity.

Shane Dikolli co-authored “CEO Behavioral Integrity, Auditor Responses and Firm Outcomes,” which appeared in The Accounting Review, with Thomas Keusch of INSEAD, William J. Mayew of Duke University Fuqua School of Business and Thomas D. Steffen of Yale University School of Management.