Kevin Rasmussen and Dr. Theodore Christensen, SOAIS
On December 3, 1999 the Securities and Exchange Commission (SEC) issued an important accounting regulation, Staff Accounting Bulletin 101 (SAB 101), in response to concerns that many publicly traded companies were recognizing revenues inappropriately. Prior to the release of SAB 101 many investors and Wall Street analysts suspected that companies had the ability to manipulate their reported earnings numbers by inflating their sales figures. This new regulation is a direct signal that the SEC felt companies were too liberal in their policies for recording revenues since the purpose of SAB 101 is to provide clear instructions and restrictions on when companies can consider revenues to be “earned”. I predict that as a result of this regulation, the earnings of many publicly traded companies have declined since the conservative standards of SAB 101 were imposed. Because the SEC announced this bulletin so abruptly, the corporate community was taken by surprise. Therefore, I expect dramatic changes in revenues of adopting companies immediately following the implementation of SAB 101, but more importantly, I predict differences in investor behavior. The specific research question I investigate is, “Has SAB 101 increased investors’ perceptions of the quality of earnings?” If investors are indeed more confident and trusting of reported earnings numbers following the implementation of SAB 101, then this regulation adds significant value to capital market participants. If investors’ perceptions, on the other hand, are not affected, then the regulation proves to be a significant waste of time and money for companies who file with the SEC that have been forced to comply with this regulation. I hypothesize that SAB 101 has rendered earnings numbers more credible or believable to investors. Therefore, I expect to find a stronger positive correlation between stock returns and earnings of companies affected by SAB 101 following its implementation.
Researchers in accounting measure investor perceptions by examining changes in stock price following significant events like earnings announcements. Finance theory suggests that stock markets efficiently incorporate all relevant information about companies’ future prospects fairly rapidly. Prior research shows that when companies announce their quarterly earnings results in a public press release, there is generally a significant stock price reaction if there is a “surprise” relative to what investors expected. Most publicly traded companies have security analysts who carefully follow everything the company does and make regular forecasts of upcoming earnings. Therefore, when companies surprise investors by beating or falling short of analysts’ forecasts, stock prices usually go up or down depending on the direction and magnitude of the “earnings surprise.” My research question really focuses on the extent to which investors “believe” the reported earnings number and whether earnings figures reported in public press releases are “more believable” after companies adopt the strict provisions of SAB 101. In other words, after the SEC took away companies’ ability to manipulate revenues, did investors perceive their earnings to be more credible?
In order to answer my research question, I gathered the following data: the company name, ticker symbol, the date the company adopted SAB 101, and the quarter of the adoption date. To gather this adoption sample data, I used LexisNexis, a database that provides full-text access to national and international business news and earnings announcements. In LexisNexis, I searched for any companies that mentioned implementing or adopting SAB 101. I began with 740 earnings announcements and, after weeding out those announcements that were irrelevant or duplicates, I narrowed the sample to approximately 342 individual companies. Next, I used Standard & Poor’s COMPUSTAT, a resource for in-depth financial information of publicly traded companies, to collect all quarterly earnings announcement dates from 1996 to 2003 for each of the 342 companies. Finally, I gathered stock returns data from the CRSP database and analysts’ forecasts from the IBES database. All of these databases are available online to BYU faculty and students through Wharton Research Data Services (WRDS).
The first test of whether investors perceive earnings to be more credible following the adoption of SAB 101 is a simple test of whether stock returns around earnings announcements are higher in the post-adoption period. Since some earnings press releases report good news (by beating analysts’ forecasts) and others report bad news (by missing analysts’ earnings expectations), a simple T-test comparing pre- and post-adoption stock returns around earnings announcement dates would likely provide no evidence because negative (bad news) stock returns could cancel out positive (good news) stock returns. One way to avoid this issue is to simply measure the “magnitude” of abnormal stock returns by taking the absolute value. In order to do this, my first test uses the absolute value of abnormal stock returns in the three-day window surrounding quarterly public news announcement dates. A simple T-test in Excel examines whether the magnitude of stock price reactions to sample companies’ quarter earnings announcements is higher in the post-adoption period. The results indicate that the magnitude of abnormal stock returns is significantly higher in the post-adoption period (t > 6.00, which is highly statistically significant). This evidence is consistent with my hypothesis that investors believe reported earnings of adopting companies are more credible after the implementation of the more strict revenue recognition policies of SAB 101.
While the T-test provides evidence that supports my hypothesis, this result is not conclusive. For example, what if earnings surprises are systematically higher in the post-adoption period? While this explanation isn’t very likely, my faculty advisor suggested that a more credible test of my hypothesis would require me to control for the earnings surprise (i.e., the degree to which companies exceeded or missed analysts’ earnings expectations). Therefore, my next test involves a regression of abnormal stock returns around the earnings announcement date on the earnings surprise (actual reported earnings per share minus the consensus earnings per share forecast). Two separate regression analyses, first using only pre-SAB 101 adoption data (1996 to 1999) and then using only post-adoption data (2002 to 2003), provide results that are very supportive of my hypothesis. The key measures from the regressions to test my hypothesis are the adjusted-R2, the coefficient on the earnings surprise, the coefficient’s t-statistic, and the p-value of the t-statistic. The adjusted-R2 explains the percentage of the dependent variable (the cumulative abnormal stock return) that can be explained by the independent variable (the earnings surprise). In accounting research, adjusted-R2 values in these types of regressions tend to be very low (i.e. 1% – 5%). However, it is possible to compare them across the pre- and post-adoption periods to examine whether the correlation between earnings surprises and stock returns increased in the post-adoption period. In the pre-adoption period, the adjusted-R2 is 1.25% and in the post-adoption period it increases to 3.53%. This suggests that earnings surprises have a higher correlation with earnings announcement period stock returns in the post-adoption period. In addition, the coefficient on the earnings surprise is higher in the post-adoption period. The regressions indicate that in the pre-adoption period, for a $1 change in the earnings surprise variable there is a $1.048 change in the company’s stock price; and, in the post-adoption period, for every $1 change in the surprise variable there is a $1.353 change in stock price. Thus, there is a stronger investor reaction for a given level of earnings surprise in the post-adoption period, suggesting that investors are more confident in earnings numbers after companies adopt the new rules imposed by SAB 101. In addition to the magnitude of the coefficient, the statistical significance of the coefficient is also important. A t-statistic above 1.9 and a p-value below 5% is generally considered to be significant. The t-statistic in the pre-adoption period is 5.47 and the t-statistic in the post-adoption period is 6.17. Both are highly statistically significant. However, the t-statistic is higher in the post-adoption period, consistent with my hypothesis.
My final comprehensive test pools both pre- and post-adoption data into a single regression with an indicator variable to measure the change in the surprise slope coefficient in the post-adoption period. This final test shows that the investor response to earnings surprises is significantly higher in the post adoption period (t = 1.89, p < 6%). This result is consistent with the idea that investors perceive earnings to be more credible (and react more forcefully to quarterly earnings announcements of sample firms) after the adoption of SAB 101. Overall, the data suggest that investors react more vigorously to earnings announcements in the post-adoption period. In other words, investors believe earnings numbers more after firms change their revenue recognition policies. I plan to team up with my faculty advisor and another BYU faculty member to perform more detailed analyses. Our goal is to publish these results in a top academic accounting journal. Therefore, this project has been a very rewarding learning process for me that will hopefully result in a published research paper. I very much appreciate the financial support of ORCA that has made this research experience possible for me. It is one of the highlights of my undergraduate educational experience.