Is CEO certification credible?

Bhattacharya, Utpal ; Groznik, Peter ; Haslem, Bruce 等

IN THE WAKE OF THE COLLAPSE OF ENRON AND WorldCom, investors began to reevaluate the integrity of the financial statements issued by publicly listed firms. It became clear that there were systemic abuses of accounting standards at all levels--CEOs, corporate boards, and even the auditors that were supposed to prevent abuses. As a result, pressure mounted on the government to step in and restore the confidence that investors once held in the system of financial reporting.

In June of 2002, the Securities and Exchange Commission ordered the CEOs and CFOs of all firms with revenue greater than $1.2 billion to sign statements certifying the validity of their financial reports. Under the order, 688 firms whose financial years coincided with the calendar year were required to file their statements with the SEC by August 14, 2002.

The SEC justified the issuance of the order as follows:

 [T]he purpose of the Commission's investigation is to provide
 greater assurance to the Commission and to investors
 that persons have not violated, or are not currently violating,
 the provisions of the federal securities laws governing corporate
 issuers' financial reporting and accounting practices,
 and to aid the Commission in assessing whether it is necessary
 or appropriate in the public interest or for the protection
 of investors for the Commission to adopt or amend
 rules and regulations governing corporate issuers' reporting
 and accounting practices and/or for the Commission to recommend
 legislation to Congress concerning these matters.

Without awaiting the Commission's recommendation, Congress included in the Sarbanes-Oxley Act a provision making CEO and CFO certification mandatory for all publicly listed firms. The Sarbanes-Oxley Act, which was enacted in July 2002, put teeth in the requirement by making the penalty for willfully certifying false earnings reports punishable by a maximum penalty of 20 years in prison, a fine of $5 million, or both.

Clearly, regulators viewed certification as a valuable addition to the arsenal of corporate governance mechanisms. Yet, market professionals and market commentators greeted certification with a great deal of skepticism. Their view is represented by the cartoon on the cover of the August 17, 2002 issue of The Economist, which shows a CEO pledging, "I swear ... that, to the best of my knowledge (which is pretty poor and may be revised in the future), my company's accounts are (more or less) accurate. I have checked this with my auditors and directors who (I pay to) agree with me ..."

The reason for the skepticism is clear: Why would this certification be any more credible than the financial documents already submitted to the SEC and signed by CEOs who know that lying constitutes fraud? Economic theory suggests that certification is a credible signal about the quality of the firm only if the costs of a false signal are significantly greater for a CEO with lower earnings transparency than the benefits that could be gained by pretending to be a CEO with high earnings transparency. But what are those costs? And, even if the potential costs are high, what is the probability that a CEO who certifies an earnings report that he knows is incorrect will be found guilty and punished?

The shareholders' opinion The true value of CEO certification must ultimately be decided by the people who the rule was designed to protect: the shareholders. To gauge what shareholders think, financial economics tells us to document how they trade. Assuming that the event--certification or the lack thereof--will be reflected in traded asset prices, a careful analysis of trading behavior and the price reaction during events can reveal whether shareholders consider certification (or not certifying) to be good news, bad news, or no news at all.

A commonly used method to analyze investor behavior is an event study. Event studies are conducted by comparing the actual market return during an event to a benchmark return that would have been expected to occur in the absence of the event. This difference between the actual return and the expected return is called an abnormal return. If the event lasts more than one day, the abnormal returns are summed up over multiple days and the sum is called the cumulative abnormal return. If an event does affect pricing, abnormal returns will be significantly different from zero.

Of the 688 firms that were required to certify by August 14, 664 firms did so using the required form without qualifications. Of the remaining 24 firms, 15 filed their own form while nine firms did not file anything at all. We can thus examine the market's reaction to the 664 firms that certified without qualification, as well as the market's reaction to the 24 firms that did not.

Using three different empirical methods to measure abnormal returns, we get the same result: The market did not react to news of certifications. Furthermore, there was no significant difference in returns around the date of certification--even for the non-certifiers. In addition, we found that there was nothing unusual in either the volatility of returns or the volume of trade for either set of firms.

Why no response? Does the absence of a reaction mean the regulation was irrelevant? Not necessarily. There could be several possible reasons why we were unable to find a reaction to the news of whether a firm certified or not. First, there could be an empirical problem in that our sample size is too small. However, our sample is similar to or larger than the samples used in other studies testing for abnormal market returns, which suggests this explanation is unlikely. The second explanation may be that the market is not efficient and the news of certification was not incorporated into prices. But our sample is made up of the largest firms listed in the U.S. markets and the certification events were widely covered by all major U.S. news outlets, so this explanation also seems unlikely.

Two other explanations seem more plausible. The first is that certification was value-irrelevant, and whether or not a firm certified did not matter to investors when they decided how much a share of the firm was worth. The second explanation may be that certification was value-relevant, but the market had anticipated which firms would certify and which firms would not, so the information was incorporated into the price of the firm before the event occurred.

Through regression analysis, we found the second hypothesis to be very plausible because whether a firm would certify or not was very predictable. Among other shared characteristics, firms that did not certify were likely to be firms that restated their earnings in the previous year, were in financial distress (low cash flows), and were more likely to be audited by Arthur Andersen.

The predictability of earnings certification is not really surprising. Most firms were expected to certify their earnings numbers. Therefore, the market was not surprised when firms did certify. And the market was not surprised by those firms that did not certify, because many of them had already restated earnings in the past, had an auditor that was in court over its behavior, or had already been mentioned in the news like Enron or WorldCom. So, when they did not certify, it was not news.

Value relevance But predictability does not prove that certifications are value-relevant. For the certifications to be value relevant and predictable, the market must not only know who is likely to certify, but care enough about the outcome of certification that it changes the value that investors place on a firm's equity once it becomes apparent who is going to certify.

If certification were value relevant and predictable, two testable implications would follow. First, if the event is relevant but stock prices fully reflect the event, we should see evidence of information leakage before the event. The certifiers should exhibit a significant price increase from many days before the event until the event date, while the opposite should be true for the firms that did not certify. To be precise, we should see positive abnormal returns leading up to the event for certifiers, and negative abnormal returns in the period leading up to the event date for those firms that did not certify. However, we find no abnormal positive returns in the period leading up to certification for those firms that certified their earnings.

An argument could be made that the market simply expected that everyone would certify, so no abnormal returns would be expected. But that cannot be an explanation for the pre-event cumulative abnormal returns of the firms that did not certify. The pre-event abnormal returns for the non-certifiers were not negative during the period leading up to the confirmation of their non-certification. In fact, the abnormal returns actually rose significantly in the days before the certification deadline. Certification of earnings did not reward those firms that certified with higher market valuations, nor did the market punish those firms that failed to certify.

For the second test, suppose that certification did matter to the market, but it could not predict with certainty the firms that would or would not certify around the time of the event. Under such circumstances, abnormal returns should have predictable correlations with certain firm-specific corporate governance variables. If it were predictable which firms would or would not certify, and the event was value relevant, then news leakage during this period (measured as the pre-event abnormal returns) should have predictable correlations with certain firm-specific corporate governance variables. For example, if it is anticipated that only firms with good corporate governance will certify, then we should expect that measures of good corporate governance--like percentage of outside directors on the board--should be positively correlated with the abnormal returns at the time of or leading up to the event. But again, we do not find that result. We examined 24 variables that previous studies have shown to be good indicators regarding the quality of corporate governance, and almost none have a significant relationship with the abnormal returns. Therefore, the results of the empirical analysis leave one alternative remaining: The market did not consider certifications value-relevant.

Does that mean the market does not care about the integrity of the financial reports issued by firms? Certainly not. What it does indicate is that the market was able to separate the good firms from the bad firms without the aid of the required certification. That can be seen clearly by examining the returns of the two portfolios well before the SEC-ordered certification. First, as far back as April 2002, there are large, negative abnormal returns on average for those firms that would eventually fail to file the required certification. By June 27, the cumulative abnormal returns of the firms that certified were already higher than the cumulative abnormal returns of the firms that did not certify. That indicates the market had partially separated firms with good earnings transparency from firms with bad earnings transparency well before the SEC's order. Nor was there a significant change in the difference between the cumulative abnormal returns of the firms that certified and the cumulative abnormal returns of the firms that did not certify when the requirement was announced on June 27. That certainly implies the new regulations requiring certification neither helped nor hindered the ability of the market to further differentiate between the two types of firms.

Certainly, it may be too early to dismiss Sarbanes-Oxley as irrelevant. Besides the certification rules, it also makes CEOs liable for the internal controls of the firm and the executive's candor with those charged with auditing the firm. If Sarbanes-Oxley makes it easier for authorities to prosecute dishonest executives, the law may induce more honesty on the part of chief executives, which is certainly a worthwhile result. But it remains an empirical question whether the law will further enhance the ability of the market to differentiate between firms that are susceptible to accounting manipulation and those that are not. That is the true measure of whether regulation actually benefits the investor.

Utpal Bhattacharya is the LaSalle Bank Faculty Fellow and an associate professor of finance at Indiana University. He can be contacted by e-mail at ubhattac@indiana.edu.

Peter Groznik is an assistant professor of finance at the University of Ljubljana in Slovenia and a visiting professor of finance at Indiana University. He can be contacted by e-mail at pegrozni@indiana.edu.

Bruce Haslem is a doctoral candidate at Indiana University. He can be contacted by e-mail at bhaslem@indiana.edu.