Abstract
Previous research has found that the items that are included in GAAP earnings but excluded from Street earnings to allow the firm to meet or beat analyst earnings forecasts (“MBF exclusions”) are more persistent than the other excluded items. In this study, I find that the difference in the levels of persistence between MBF and non-MBF exclusions declined after the introduction of Regulation G, which requires public companies that disclose non-GAAP earnings to also present GAAP earnings and a reconciliation of the two. Analysts underestimate the persistence of non-MBF exclusions, but the degree of this underestimation is lower in the post-regulation period. In contrast, there is little evidence to indicate that analysts underestimate the persistence of MBF exclusions in either time period. I also find strong (weak) evidence that investors underestimate the persistence of Street exclusions in the pre- (post-) regulation period. These results suggest that Regulation G constrains the practice of excluding recurring expenses from Street earnings to meet or beat analyst forecasts and helps analysts and investors to understand the persistence of Street exclusions.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
Notes
In this study, “Street earnings” refers to the non-GAAP earnings reported by analyst-tracking services (Bradshaw and Sloan 2002; Brown and Sivakumar 2003), and “pro forma earnings” refers to the non-GAAP earnings disclosed in a firm’s earnings announcements. Bhattacharya et al. (2003) and Johnson and Schwartz (2005) find that pro forma earnings and Street earnings are equal in most cases and that, when the two numbers are not equal, the difference tends to be small.
One extreme case reported in the media is that of Motorola Inc., which presented a non-GAAP earnings number that excluded certain nonrecurring items in its earnings release in 20 consecutive quarters from 1999 through 2003. See Drucker (2002) and Motorola’s earnings announcements from the second quarter of 2002 to the fourth quarter of 2003.
For most firms, the inclusion and exclusion decisions of analysts are made at the time the nonrecurring items are first disclosed. For some firms, however, the analyst decisions could be made as early as when the earnings forecasts are issued.
Analysts may also have incentives to issue optimistic earnings forecasts (Hong and Kubik 2003), in which case the changes in the association between Street exclusions and analyst earnings forecasts could be due to changes in analysts’ incentives.
If Reg G only deters managers from excluding the most manipulative items from Street earnings and the rest of the excluded items are easy to understand even without the reconciliation, then one may still find that analysts and investors are more able to understand Street exclusions after the adoption of Reg G than before, even though the reconciliation does not provide additional information about the (remaining) Street exclusions to the public.
Due to an overlap between accruals and Street exclusions, the coefficient on EX in Eq. 1 shows the persistence of the cash component of Street exclusions and the sum of the coefficients on EX and ACC shows the persistence of the noncash component of Street exclusions.
Recall that MBF exclusions are excluded expenses that allow the firm to meet or beat analyst forecasts (coded as negative values and denoted by EXMBF, as described below). For firms with nonzero MBF exclusions, the following relationships hold: GAAP earnings = STR + EXMBF < AF, and STR ≥ AF. Rewriting these relationships yields 0 ≤ STR − AF < −EXMBF, or 0 ≤ FE < −EXMBF, which indicates that forecast errors are always non-negative but less than Street exclusions (in magnitude). Technically, nonzero MBF exclusions can be viewed as including two parts, −EXMBF − FE, which is the amount of exclusions that is just enough to allow a firm to meet analyst forecasts, and FE, which is the amount of exclusions that is more than enough to allow a firm to meet analyst forecasts. In Sect. 5.5, I discuss the results after controlling for FE t in the regressions.
When CONTROL and IQ are not included in Eq. 4, the results from running an OLS regression for the equation and testing the statistical significance of β j (i.e., \(\alpha_{1}(\gamma_{j}-\gamma_{j}^{\rm I}))\) are asymptotically equivalent to those from jointly estimating Eqs. 1 and 4 and testing the significance of the constraint \(\gamma_{j}=\gamma_{j}^{\rm I}\) (j = 1, 2, 3, and 4). Both tests are illustrated by Mishkin (1983) and the latter (joint estimation) has been used in a number of accounting studies (e.g., Sloan 1996; Beaver and McNichols 2001; Bradshaw et al. 2001). The inferred value of \(\gamma_{j}^{\rm I}\) in Eq. 4 is asymptotically equal to the value of \(\gamma_{j}^{\rm I}\) obtained from the joint estimation, and the p-value for the coefficient \(\alpha_{1}(\gamma_{j}-\gamma_{j}^{\rm I})\) in Eq. 4 is asymptotically equal to the p-value for the test of \(\gamma_{j}=\gamma_{j}^{\rm I}\) in the joint estimation. By including the control variables, the model in Eq. 4 is not subject to the potential limitations of the Mishkin test as pointed out by Kraft et al. (2007).
The net cash flows from operations reported in the Compustat quarterly file (item #108) are cumulative (e.g., the reported number for the second quarter is the sum of the numbers in the first two quarters). In this study, the values of CFO have been adjusted so that they represent the amount for a single quarter.
The regression results are similar when EX is winsorized at the 1st and 99th percentiles of all values, although the coefficient for EX becomes larger (as the magnitude of the top and bottom one percent of EX become smaller).
The empirical results are similar when I simply use either December 2001 (when the SEC issued its warning) or November 2002 (when the SEC published a number of disclosure rules for comments) as a cut-off while allowing the quarter t and quarter t + k earnings announcement dates to fall within different regulatory regimes.
I also compute the Spearman correlation coefficients. The overall results are similar, but the correlations between Street exclusions and future Street earnings and forecast errors are weaker, and the correlation between Street exclusions and future abnormal returns is stronger, when compared with the Pearson correlation coefficients.
\(\gamma_{j}^{\rm I}\) in Eq. 4 (j = 1, 2, 3, and 4) is estimated using the following equation: \(\begin{aligned} CAR_{t+k}& =\alpha_{0}+\alpha_{1}[STR_{t+k}-\hbox{E}(STR_{t+k})]+ \alpha_{2}Control_{t+k}+u_{t+k}\\& =\alpha^\prime_{0}+\alpha_{1} STR_{t+k}-\alpha_{1}(\gamma_1^{\rm I}STR_{t}+\gamma_2^{\rm I} EX_{t}+\gamma_3^{\rm I} ACC_{t}+\gamma_4^{\rm I} GROW_{t}+\delta^{\rm I}\cdot IQ_{t})+\alpha_{2}Control_{t+k}+u_{t+k}.\\ \end{aligned}\) Whether investor expectation is different from zero is determined based on the statistical significance of the regression coefficient, \(-\alpha_{1}\gamma_{j}^{\rm I}\).
Kolev et al. (2008) examine the association between Street exclusions and future operating earnings (instead of future Street earnings) and find a significant decline in the association in the post-Reg G period.
As an additional test on investor mispricing, I regress CAR t+k (k = 1, 2, 3, and 4) on analyst forecast errors, Street exclusions, accruals, growth, firm size, book-to-market value of equity, and the security beta (all variables for quarter t), and dummy variables that represent the industry and the quarter. This model removes the contemporaneous variables in Eq. 4 (i.e., [STR t+k − E(STR t+k)] and EX t+k) and is similar to that used by Doyle et al. (2003). I run this regression for the pre- and post-Reg G samples separately and find a significant association between Street exclusions and abnormal returns around the subsequent earnings announcement date only in the pre-Reg G sample (results not tabulated).
In some cases, (γ − γI) is statistically insignificant yet greater than (γ − γA) in column (3), which is statistically significant. This is because the standard errors are much larger for (γ − γI) than for (γ − γA).
The median is determined based on all of the firms in the pre-Reg G sample that have a positive coefficient on Street exclusions and at least 20 quarters of data. All of the new firms in the post-Reg G sample are dropped. The results are qualitatively similar when the median is determined based on all of the firm-specific regression coefficients regardless of the sign, or when a minimum of 12 quarters of observations are required.
When I replicate the regression analyses for only the observations with excluded gains, I find no statistically significant association between excluded gains and future Street earnings in either the pre- or post-Reg G period.
The weaker evidence is due, at least partially, to the larger standard errors for the quarterly coefficients in the regression of future abnormal returns. Recall that (γ − γI) equals the regression coefficient, β, divided by α1 for each quarter. In a few of the quarterly regressions, the coefficient on Street exclusions is negative and α1 is very small, which yields a large negative value of (γ − γI).
The dependent variable in the long-window tests aggregates over k quarters and there is an overlap between observations. Therefore, I run the regression by quarter and examine the mean coefficient estimates. Similar to Doyle et al. (2003), I compute the t-statistics for the mean coefficients based on the Fama-MacBeth t-statistics with a Newey-West adjustment to account for possible serial correlations in the quarterly estimates.
References
Beaver, W., & McNichols, M. (2001). Do stock prices of property casualty insurers fully reflect information about earnings, accruals, cash flows, and development? Review of Accounting Studies, 6, 197–220.
Bhattacharya, N., Black, E., Christensen, T., & Larson, C. (2003). Assessing the relative informativeness and permanence of pro forma earnings and GAAP operating earnings. Journal of Accounting and Economics, 36, 285–319.
Bradshaw, M., Richardson, S., & Sloan, R. (2001). Do analysts and auditors use information in accruals? Journal of Accounting Research, 39(1), 45–74.
Bradshaw, M., & Sloan, R. (2002). GAAP versus The Street: An empirical assessment of two alternative definitions of earnings. Journal of Accounting Research, 40(1), 41–66.
Brown, L., & Sivakumar, K. (2003). Comparing the value relevance of two operating income measures. Review of Accounting Studies, 8, 561–572.
Burgstahler, D., Jiambalvo, J., & Shevlin, T. (2002). Do stock prices fully reflect the implications of special items for future earnings? Journal of Accounting Research, 40(3), 585–612.
Chaney, P., Hogan, C., & Jeter, D. (1999). The effect of reporting restructuring charges on analysts’ forecast revisions and errors. Journal of Accounting and Economics, 27, 261–284.
Cohen, D., Hann, R., & Ogneva, M. (2007). Another look at GAAP versus The Street: An empirical assessment of measurement error bias. Review of Accounting Studies, 12, 271–303.
Doyle J., Lundholm, R., & Soliman, M. (2003). The predictive value of expenses excluded from pro forma earnings. Review of Accounting Studies, 8, 145–174.
Doyle, J., & Soliman, M. (2005). Do managers define ‘Street’ earnings to meet or beat analyst forecasts? Working paper, University of Utah.
Drucker, J. (2002, October 15). Motorola’s profit: “Special” again? The Wall Street Journal, C1.
Elliott, W. (2006). Are investors influenced by pro forma emphasis and reconciliations in earnings announcements? The Accounting Review, 81(1), 113–133.
Entwistle, G., Feltham, G., & Mbagwu, C. (2006). Financial reporting regulation and the reporting of pro forma earnings. Accounting Horizons, 20(1), 39–55.
Fairfield, P., Whisenant, J., & Yohn, T. (2003). Accrued earnings and growth: Implications for future profitability and market mispricing. The Accounting Review, 78(1), 353–371.
Frankel, R., McVay, S., & Soliman, M. (2006). Street earnings and board independence. Working paper, Washington University.
Frankel, R., & Roychowdhury, S. (2004). Testing the clientele effect: An explanation for Non-GAAP earnings adjustments used to compute I/B/E/S earnings. Working paper, Massachusetts Institute of Technology.
Frederickson, J., & Miller, J. (2004). The effects of pro forma earnings disclosures on analysts’ and nonprofessional investors’ equity valuation judgments. The Accounting Review, 79(3), 667–686.
Gu, Z., & Chen, T. (2004). Analysts’ treatment of nonrecurring items in Street earnings. Journal of Accounting and Economics, 38, 129–170.
Heflin, F., & Hsu, C. (2007). The impact of the SEC’s regulation of non-GAAP disclosures. Working paper, Florida State University.
Hong, H., & Kubik, J. (2003). Analyzing the analysts: Career concerns and biased earnings forecasts. Journal of Finance, 58(1), 313–351.
Johnson, W., & Schwartz, W. (2005). Are investors misled by “pro forma” earnings? Contemporary Accounting Research, 22(4), 915–964.
Kolev, K., Marquardt, C., & McVay, S. (2008). SEC scrutiny and the evolution of non-GAAP reporting. The Accounting Review, 83(1), 157–184.
Kraft, A., Leone, A., & Wasley, C. (2007). Regression-based tests of the market pricing of accounting numbers: The Mishkin test and ordinary least squares. Journal of Accounting Research, 45(5), 1081–1114.
Lougee, B., & Marquardt, C. (2004). Earnings informativeness and strategic disclosure: An empirical examination of “pro forma” earnings. The Accounting Review, 79(3), 769–795.
Marques, A. (2006). SEC interventions and the frequency and usefulness of non-GAAP financial measures. Review of Accounting Studies, 11, 549–574.
McVay, S. (2006). Earnings management using classification shifting: An examination of core earnings and special items. The Accounting Review, 81(3), 501–531.
Mikhail, M., Walther, B., & Willis, R. (1997). Do security analysts improve their performance with experience? Journal of Accounting Research, 37(3), 131–157.
Mikhail, M., Walther, B., & Willis, R. (2003). The effect of experience on security analyst underreaction. Journal of Accounting and Economics, 35, 101–116.
Mishkin, F. (1983). A rational expectations approach to macroeconometrics: Testing policy ineffectiveness and efficient-markets models. Chicago, IL: The University of Chicago Press.
Plitch, P. (2002, January 22). Firms in many sectors give earnings a pro forma makeover, survey finds. Dow Jones Newswires.
Securities and Exchange Commission. (2001). Cautionary advice regarding the use of “pro forma” financial information, Release No. 33-8039.
Sloan, R. (1996). Do stock prices fully reflect information in accruals and cash flows about future earnings? The Accounting Review, 71(3), 289–315.
Acknowledgments
I thank Gary Biddle, Kevin Chen, Jong-Hag Choi, Jim Frederickson, Gilles Hilary, Hyunkoo Lee, Maureen McNichols (the editor), Taychang Wang, T. J. Wong, two anonymous reviewers, and workshop participants at the Hong Kong University of Science and Technology, National Taipei University, National Taiwan University, and the Shanghai University of Finance and Economics for providing helpful comments.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Chen, CY. Do analysts and investors fully understand the persistence of the items excluded from Street earnings?. Rev Account Stud 15, 32–69 (2010). https://doi.org/10.1007/s11142-008-9079-y
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11142-008-9079-y