Repeatedly meeting-beating analyst forecasts and audit fees.

Rickling, Maria ; Rama, Dasaratha V. ; Raghunandan, K. 等

I. INTRODUCTION

In this paper we examine the association between repeatedly meeting-or-beating analysts' earnings forecasts (MBE) and audit fees. Motivation for this paper comes from the well-documented role of MBE in influencing managers' actions and the importance assigned to MBE by investors, and the associated risk for firms and their auditors when firms miss analysts' forecasts.

Analysts' forecasts serve as a proxy for the market's expectations and are one of three benchmarks that managers seek to meet (Degeorge et al., 1999). (1) Many prior studies have noted that companies have incentives to meet or beat analysts' forecasts of earnings per share (EPS) because investors and others care about such performance. For example, Bartov et al. (2002) and Brown and Caylor (2005) find that firms that consistently meet or beat analysts' forecasts tend to have a higher equity return premium. Jiang (2008) finds that firms beating earnings forecasts are more likely to receive an increase in bond ratings and a smaller initial bond yield spread. Further, prior research also finds that the rewards are higher for firms that consistently meet or beat forecasts; Kasznik and McNichols (2002) find that the market assigns higher values to firms that consistently meet or beat analysts' estimates while Duarte et al. (2008) and Brown et al. (2009) report lower cost of capital for firms that consistently meet or beat analysts' forecasts.

Given the strong incentives of managers to meet or beat analysts' forecasts, many prior studies use MBE as a proxy for earnings management (e.g., Bartov et al., 2002; Bauman and Shaw, 2006; Burgstahler and Eames, 2006), and the SEC considers repeatedly meeting-beating analysts' forecasts as a factor in its regulatory actions (Bryan-Low, 2002; SEC, 2009). If auditors view managers of firms that repeatedly MBE as being more aggressive and/or more frequently engaging in earnings management, and hence assign higher inherent risk to such firms, then audit fees should be higher for such firms (either due to increased work, or due to increased risk premium). Conversely, Kasznik and McNichols (2002, p. 730) note that investors perceive that "firms that consistently meet expectations are less risky than those that do not" (emphasis added). If auditors share such perceptions, then audit fees should be lower for firms that repeatedly MBE.

Negative earnings surprises lead to sharp declines in stock price and negative publicity for the firm (Brown et al., 1987; Matsumoto, 2002). Such declines in stock prices can, in turn, lead to shareholder lawsuits and auditors are often named as defendants in such lawsuits (Alexander, 1991; Tucker, 1991). Prior research suggests that litigation risk is positively related to audit fees (Simunic and Stein, 1996; Bell et al., 2001). Hence, firms with more occurrences of missing analysts' forecasts (i.e., fewer occurrences of MBE) may be viewed as opening themselves to litigation more often and hence viewed as more risky, which in turn can result in higher audit fees. However, if auditors view managers of firms that miss analysts' forecasts on multiple occasions as being not aggressive and hence assign lower inherent risk to such firms, then audit fees should be lower for such firms.

Thus, ultimately, the association between repeatedly meeting-beating analysts' forecasts and audit fees is an empirical issue. In this study, we examine if audit fees are higher or lower for firms that repeatedly meet or beat analysts' forecasts. We use data from 791 non-financial firms that have analyst forecast data for all 12 quarters during the period from 2005 to 2007 and find that, after controlling for other factors usually included in audit fee models, audit fees are lower for firms that repeatedly meet or beat analysts' forecasts. This evidence indicates that auditors view firms that repeatedly meet or beat earnings forecasts as less risky, consistent with the suggestion in Kasznik and McNichols (2002). An alternative, but not mutually exclusive, inference from the evidence is that firms that miss analysts' forecasts on multiple occasions are viewed as more risky by auditors.

II. BACKGROUND

A. Meeting or Beating Earnings Forecasts

The significant economic consequences of MBE have led management to adopt earnings management practices to achieve such meet-beat goals (e.g., Bartov et al., 2002; Bauman and Shaw, 2006; Burgstahler and Eames, 2006). Graham et al. (2005) corroborate these economic implications through a survey of CFOs. They report that more than eighty percent of CFOs agree that meeting and beating earnings benchmarks helps "maintain or increase the stock price" and "build credibility with capital markets" (Graham et al., 2005, p. 5).

Such practices have led to the Securities and Exchange Commission (SEC) using patterns of MBE to identify possible cases of accounting irregularities (Bryan-Low, 2002). For instance, in August of 2009, General Electric (GE) agreed to pay a $50 million penalty to settle charges stemming from a risk-based investigation made by the SEC alleging that GE consistently met or exceeded financial analysts' consensus EPS expectations (SEC, 2009), and that on at least four separate occasions GE's executives approved improper application of accounting standards so as to maintain the pattern of MBE.

B. Audit Fees and Client Risk

The risk of auditor litigation increases as client-specific risk increases, and litigation risk is a fundamental consideration during the fee-setting process of an audit engagement (Clarkson and Simunic, 1994; Simunic and Stein, 1996). Such association between litigation risk and auditors' decisions, including client acceptance/resignation and audit fees, has been corroborated in several empirical studies (e.g., Simunic and Stein, 1996; Krishnan and Krishnan, 1997; Bell et al., 2001). The collective evidence from such studies implies that auditors attempt to mitigate litigation risk through increased work, or by incorporating a risk premium. The increase in audit effort in turn increases the audit fees charged to the client.

Under the audit risk model, the auditor's risk is a joint function of inherent risk, control risk, and detection risk. The former two are client specific factors, while detection risk is under the control of the auditor. Inherent risk is influenced by various characteristics of the client organization, such as management's attitude toward financial reporting and earnings management. The audit risk model suggests that, ceteris paribus, auditors can offset higher levels of inherent risk by decreasing detection risk. Auditors can reduce detection risk by increasing the quality and/or quantity of the audit services provided. Such an increase in audit effort results in an increase in the fees charged for the audit services. Many studies, since Simunic (1980), have shown that audit fees are higher when client-related risk is higher (see Hay et al. [2006] for a detailed summary of such research).

III. DEVELOPMENT OF HYPOTHESIS

Empirical evidence suggests that managers use earnings management practices to meet or beat analysts' forecasts due to the significant economic advantages of such meet-or-beat behavior (e.g., Bartov et al., 2002; Bauman and Shaw, 2006; Burgstahler and Eames, 2006). Further, the SEC uses a pattern of meeting or beating analysts' estimates to identify possible cases of accounting irregularities (Bryan-Low, 2002). Therefore, auditors may view managements of companies that consistently meet or beat analyst forecasts as more aggressive and, hence, view such clients as having higher inherent risk. This in turn should lead to lower detection risk resulting in higher audit fees.

However, Kasznik and McNichols (2002) find that abnormal annual returns are positively associated with firms at least meeting analysts' estimates, and that the market assigns a higher value to firms that meet estimates consistently. Kasznik and McNichols (2002, p. 730) note that the pattern of results suggest "...investors' perceptions that firms that consistently meet expectations are less risky than those that do not" (emphasis added). (2) If auditors also view firms that repeatedly MBE as less risky, then audit fees should be lower for such firms.

Similarly, given the association between missing analysts' forecasts and shareholder litigation and auditors' incentives to avoid being named as defendants in such lawsuits, firms that repeatedly miss analysts' forecasts may be viewed as more likely to be targets of litigation and, hence, more risky audit clients. This latter view suggests that audit fees should be higher for clients that repeatedly miss analysts' forecasts (i.e., audit fees should be lower for firms that repeatedly MBE). However, firms that repeatedly miss earnings forecasts may be viewed as less aggressive and hence viewed as having lower risk; this suggests that audit fees should be lower for such firms (or, audit fees should be higher for firms that repeatedly MBE).

Ultimately, the association between MBE and audit fees is an empirical issue; this analysis seeks to provide some relevant empirical evidence. Thus, our hypothesis is (in the null form):

[H.sub.0]: Audit fees are not associated with the frequency of meeting or beating analyst earnings forecasts.

IV. METHOD

A. Model

We use the following regression model:

LNAFEE = [[beta].sub.0] + [[beta].sub.1] * LNTA + [[beta].sub.2] * RECINV + [[beta].sub.3] * SQRTSEG + [[beta].sub.4] * FOROPS + [[beta].sub.5] * LIQ + [[beta].sub.6] * LOSS + [[beta].sub.7] * ROA + [[beta].sub.8] * BM + [[beta].sub.9] * ICW + [[beta].sub.10] * GC + [[beta].sub.11] BIG4 + [[beta].sub.12] * AUDINIT + [[beta].sub.13] * NUM_MB + [[beta].sub.14-22] * (Industry Dummy Variables) + [epsilon]

where LNAFEE = natural logarithm of audit fees; LNTA = natural logarithm of total assets ($000); RECINV = receivables plus inventory, scaled by total assets; SQRTSEG = square root of the number of business segments per Compustat; FOROPS = 1 if the firm has foreign operations, 0 otherwise; LIQ = ratio of current assets divided by current liabilities; LOSS = 1 if the firm reports negative income for the year, 0 otherwise; ROA= return on assets, measured as net income, divided by TA; BM = book to market ratio, as of the end of the fiscal year; ICW = 1 if the firm discloses an internal control weakness, 0 otherwise; GC = 1 if the firm receives a going concern opinion, 0 otherwise; BIG4 = 1 if the firm's auditor is a Big 4 auditor, 0 otherwise; AUDINIT = 1 if it is the auditor's initial year on the engagement, 0 otherwise; NUM_MB = number of quarters during 2005 through 2007 that the actual reported earnings meets or exceeds the analysts' forecasts.

Starting from Simunic (1980), standard audit fee models use various measures of client size, complexity, financial health, risk, auditor type and tenure. Our model accordingly controls for client size (LNTA), complexity (RECINV, SQRTSEG, and FOROPS), financial health (LIQ, LOSS, and ROA), risk (BM, ICW, and GC) and auditor type and tenure (BIG4 and AUDINIT). We also control for industry membership and use the ten industry groups as defined by Dr. French. (3) All balance sheet based variables are measured as of the end of the fiscal year.

Analysts make their earnings forecasts throughout the year, making revisions as they receive new earnings-relevant information concerning their target firms. Consequently, forecasts issued closer to the earnings announcement date are based on a more rich set of information and thus tend to be more accurate than the preceding forecasts, ceteris paribus (Sinha et al., 1997). This positive association between most recent forecasts and forecast accuracy has been well documented in prior research (Crichfield et al., 1978; O'Brien, 1988; Brown, 1991; Sinha et al., 1997). Hence, following prior research (Bartov et al., 2002; Burgstahler and Eames, 2006; Koh et al., 2008), we use the most recent forecast issued prior to the earnings announcement date as the analyst forecast measure, and calculate the forecast error for the sample of firms as actual earnings per share less forecasted earnings per share. NUM_MB represents the number of quarters a firm's actual earnings meets-or- beats the analyst forecast for that quarter during the twelve quarters beginning January 2005 and ending December 2007.

B. Data

We begin by obtaining the universe of non-financial U.S. firms in the I/B/E/S database, having a fiscal year end of December 31, and having analyst forecast data for the 12 quarters beginning January 1, 2005 and ending December 31, 2007. We impose the "non-financial" condition to be consistent with prior research related to audit fees that has excluded firms in the financial sector given their unique regulatory and financial statement characteristics. The December 31 fiscal year end condition is imposed because of the significant changes that occurred during the time period of our examination. We use the 2007 fiscal year because, as another part of our analysis, we examined the association between repeatedly meeting or beating analysts' forecasts and subsequent restatements (we examined restatements for a period spanning three years after the fiscal year end).

We obtain audit fee data from the Audit Analytics database and relevant financial data from the Compustat annual database. We remove observations with less than 12 quarters of forecasts (we relax this condition as part of our sensitivity tests), as well as observations with missing data for audit fees or any of the independent variables. This process yields 791 firms in the final sample. (4)

Table 1 provides empirical evidence about the number of quarters in which firms meet or beat analyst forecasts. As seen in Table 1, only one firm in the sample did not meet or beat analysts' forecasts for any of the 12 quarters examined; only 13 firms had three or fewer MBE quarters, while 73 firms had MBE in all 12 quarters. The mean (median) number of MBE quarters for the 791 firms in our sample is 8.61 (9.00).

Table 2 displays descriptive data about the sample. The mean (median) total assets and audit fees for the companies in the sample are $7,049 million ($1,450 million) and $3,205 million ($1,560 million), respectively. On average, more than half of the sample firms have foreign operations. The mean (median) measure of liquidity, ratio of current assets to current liabilities, is 2.59 (1.82). On average, 20% of the sample firms reported negative earnings for 2007. The mean (median) book to market ratio for the sample firms is .44 (.38). Four percent of sample firms had an adverse internal control report, on average Big 4 firms audit 92% of the sample firms, and 3% have a new auditor during the year of analysis.

Analysis of (untabulated) Pearson correlations involving the explanatory variables indicates that LNTA has the highest correlation coefficients and is significantly correlated with LIQ, LOSS, ROA, and BIG4, with [rho] = -0.35, -0.34, 0.32, and 0.32, respectively. Only six of the correlations exceed 0.3 (four of which are the aforementioned correlations pertaining to firm size), suggesting multicollinearity is not a problem. This is later confirmed by a review of the variance inflation factor (VIF) scores from the regressions, none of which are high enough to create a cause for concern of multicollinearity (the highest VIF score is 2.68, well below the usual cutoff of 10 for multicollinearity problems).

IV. EMPIRICAL RESULTS

Table 3 presents the regression results. The overall regression is significant (F = 148.487, p < 0.001) with an adjusted [R.sup.2] of 0.804, in line with the reported [R.sup.2] of regression models utilized in extant audit fee studies. In general, the control variables have the expected signs and are significant; the only exceptions are that SQRTSEG, LOSS and GC are not statistically significant at conventional levels. The coefficient of NUM_MB is -0.020, significant at p = .018. The magnitude of the coefficient indicates that as a firm moves from the 25th percentile to the 75th percentile in terms of MBE (i.e., as MBE increases from 7 to 10), the audit fee decreases by about 5.8 percent.

A. Further Analyses

Since the focus of our analysis is on the association between firms repeatedly meeting or beating earnings forecasts and audit fees, we employ an alternative specification of the fee model. We create an indicator variable MBED that takes the value of 1 if the number of quarters that the firm has MBE equals or exceeds a threshold value, and 0 otherwise; we use various cutoffs for the threshold, starting from 6 to 11. In each instance, the coefficient of MBED is negative and significant, indicating that audit fees are lower for firms that repeatedly MBE.

Next, we delete sample firms that are audited by non-Big 4 auditors, and thus restrict the sample to clients of Big 4 firms. The results with this subsample are qualitatively similar to those presented in Table 3, and show that Big 4 auditors charge less to firms that more frequently meet or beat analysts' earnings forecasts.

B. Additional Tests

1. Expanded sample

We first relax the condition that analyst forecast data should be available for all 12 quarters starting from the first quarter of 2005 to the last quarter of 2007. This increases the sample size to 1,588 firms with all required data to estimate the fee model for the year ending December 31, 2007. With this expanded sample, we obtain substantively similar results. The coefficient of NUM_MB is negative and significant in the fee regression indicating that audit fees are lower for firms that repeatedly meet or beat analysts' forecasts.

2. Meeting-or-just-beating forecasts

Next, we use an alternative specification for the independent variable of interest. If a firm repeatedly meets or just barely beats analysts' forecasts that is arguably more persuasive evidence related to the importance of earnings benchmarks. Hence, we replace the NUM_MB with an alternative variable, NUM_JMB that takes the value of 1 if a firm meets or barely beats the analysts' forecast of earnings; specifically, we use a threshold of one cent. Hence, NUM_JMB increases by one for each quarter that the firm either meets or just beats the analyst forecast of EPS by one cent or less; if the firm beats the forecast by more than one cent (or, misses the forecast by any amount) then the value of NUMJMB does not increase. In this alternative specification, 1,212 of the 1,588 firms (76 percent) either meet or just beat analyst forecasts at least once during the 12 quarters. (5) The mean (median) number of quarters that a sample firm exactly met or exceeded analyst forecasts by no more than one cent is 2.15 (2).

We then replace NUM_MB by NUM_JMB in the fee regression model. The coefficient of NUM_JMB is negative and significant in the model indicating that there is a significant negative association between audit fees and the number of quarters a firm meets or just beats analyst forecasts by one cent.

Next, we calculate another variable, NUMJMISS that increases by one for each quarter that a firm just misses (by one cent or less) analysts' forecasts. The mean value of NUMJMISS for the sample of 1,588 firms is 0.64. We include NUMJMISS in the model (along with NUMJMB) and find that the coefficient of NUMJMISS is positive and significant (p < .10) indicating that there is a positive association between audit fees and the number of quarters a firm just misses such forecasts by no more than one cent.

We then repeat the above analysis using (a) two-cents and (b) five cents, instead of one cent, as the threshold. We obtain substantively similar results as discussed above: the results indicate that (a) there is a significant negative (positive) association between audit fees and the number of quarters a firm meets or just beats (misses) analyst forecasts by the specified threshold of two or five cents.

V. SUMMARY AND CONCLUSIONS

Analysts' earnings forecasts serve as important benchmarks for managers. Prior research shows that investors and others care about firms meeting-or- beating analysts' earnings forecasts (MBE) and that firms that consistently MBE tend to have a higher equity premium, lower cost of equity and higher bond ratings. Given such strong incentives, many prior studies use MBE as a proxy for earnings management and the SEC considers repeated MBE as a factor in its regulatory actions (Bryan-Low, 2002; SEC, 2009). However, there is little research about how such MBE behavior affects auditors' decisions.

If auditors view firms that repeatedly MBE as having higher inherent risk, then audit fees should be higher for such firms. Conversely, if auditors perceive, as Kasznik and McNichols (2002) suggest investors do, that firms that consistently MBE are less risky than those that do not do so, then audit fees should be lower for firms that repeatedly MBE. Alternatively, since negative earnings surprises lead to sharp declines in stock price which in turn lead to shareholder lawsuits, firms with more occurrences of missing analysts' forecasts (i.e., fewer occurrences of MBE) may be viewed as more risky, which in turn can result in higher audit fees. However, if auditors view managers of firms that repeatedly fail to MBE as being not aggressive and hence less risky, then audit fees should be lower for such firms.

In this study, we examine the association between repeated MBE and audit fees. Using data from 791 non-financial firms, we find that audit fees are lower for firms that repeatedly meet or beat analysts' forecasts. Additional analyses indicate that the frequency of meeting or just beating (just missing) analyst forecasts is negatively (positively) associated with audit fees. These results suggest auditors view firms that frequently MBE (miss) analyst forecasts as being less (more) risky engagements. The findings are consistent with both the suggestion of Kasznik and McNichols (2002) that repeated MBE firms may be viewed as less risky by investors and others, and the view that firms that miss analysts' forecasts on multiple occasions are viewed as more risky by auditors.

The results suggest some interesting avenues for future research. For example, to what extent does repeated MBE behavior affect other audit judgments, including those related to client acceptance or retention, audit planning and evidence collection, and audit opinions? Does the margin of MBE (meeting-or-just-beating or beating by a large margin) make any difference in such judgments?

REFERENCES

Alexander, J.C., 1991, "Do the Merits Matter? A Study of Settlements in Class Actions," Stanford Law Review, 43, 497-598.

Bartov, E., D. Givoly, and C. Hayn, 2002, "The Rewards to Meeting or Beating Earnings Expectations," Journal of Accounting & Economics, 33(2), 173-204.

Bauman, M., and K. Shaw, 2006, "Stock Option Compensation and the Likelihood of Meeting Analysts' Quarterly Earnings Targets," Review of Quantitative Finance and Accounting, 26, 301-19.

Bell, T.B., W.R. Landsman, and D.A. Shackelford, 2001, "Auditors' Perceived Business Risk and Audit Fees: Analysis and Evidence," Journal of Accounting Research, 15(1), 4-24.

Brown, L.D., 1991, "Forecast Selection When All Forecasts Are Not Equally Recent," International Journal of Forecasting, 7(3), 349-56.

Brown, L.D., and M.L. Caylor, 2005, "A Temporal Analysis of Quarterly Earnings Thresholds: Propensities and Valuation Consequences," The Accounting Review, 80(2), 423-440.

Brown, L.D., G.D. Richardson, and S.J. Schwager, 1987, "An Information Interpretation of Financial Analyst Superiority in Forecasting Earnings," Journal of Accounting Research, 25(1), 49-67.

Brown, S., S.A. Hillegeist, and K. Lo, 2009, "The Effect of Earnings Surprises on Information Asymmetry," Journal of Accounting & Economics, 47(3), 208-225.

Bryan-Low, C., 2002, "Meeting Expectations Used to Draw Favor, Now It Invites Scrutiny," The Wall Street Journal, Aug 5, C1.

Burgstahler, D., and M. Eames, 2006, "Management of Earnings and Analysts' Forecasts to Achieve Zero and Small Earnings Surprises," Journal of Business Finance & Accounting, 33(5-6), 633-652.

Chevis, G., S. Das, and K. Sivaramakrishnan, 2007, "Does It Pay to Consistently Meet Analysts' Earnings Expectations?" Working paper, Baylor University.

Clarkson, P. M., and D. A. Simunic, 1994, "The Association between Audit Quality, Retained Ownership, and Firm-Specific Risk in U.S. vs. Canadian IPO Markets," Journal of Accounting & Economics, 17, 207-28.

Crichfield, T., T. Dyckman, and J. Lakonishok, 1978, "An Evaluation of Security Analysts' Forecasts," The Accounting Review, 53(3), 651-68.

Dechow, P., S. Richardson, and I. Tuna, 2003, "Why Are Earnings Kinky?" Review of Accounting Studies, 8, 355-384.

Degeorge, F., J. Patel, and R. Zeckhauser, 1999, "Earnings Management to Exceed Thresholds," Journal of Business, 72(1), 1-33.

Duarte, J., X. Han, J. Harford, and L. Young, 2008, "Information Asymmetry, Information Dissemination and the Effect of Regulation FD on the Cost of Capital," Journal of Financial Economics, 87(1), 24-44.

Graham, J., C. Harvey, and S. Rajgopal, 2005, "The Economic Implications of Corporate Financial Reporting," Journal of Accounting and Economics, 40(1), 373.

Hay, D., W. Knechel, and N. Wong, 2006, "Audit Fees: A Meta-Analysis of the Effect of Demand and Supply Attributes," Contemporary Accounting Research, 23(1), 141-191.

Jiang, J, 2008, "Beating Earnings Benchmarks and the Cost of Debt," The Accounting Review, 83(2), 377-416.

Kasznik, R., and M.F. McNichols, 2002, "Does Meeting Earnings Expectations Matter? Evidence from Analyst Forecast Revisions and Share Prices," Journal of Accounting Research, 40(3), 727-59.

Koh, K., D. Matsumoto, and S. Rajgopal, 2008, "Meeting or Beating Analyst Expectations in the Post-Scandals World: Changes in Stock Market Rewards and Managerial Actions," Contemporary Accounting Research, 25(4), 1067-98.

Krishnan J., and J. Krishnan, 1997, "Litigation Risk and Auditor Resignations," The Accounting Review, 72, 539-560.

Matsumoto, D.A., 2002, "Management's Incentives to Avoid Negative Earnings Surprises," The Accounting Review, 77(3), 483-514.

O'Brien, P., 1988, "Analysts' Forecasts as Earnings Expectations," Journal of Accounting & Economics, 10(1), 53-83.

Securities and Exchange Commission (SEC), 1999, SEC Staff Accounting Bulletin: No. 99 - Materiality, Washington, D.C., SEC.

Securities and Exchange Commission (SEC), 2009, SEC Charges General Electric with Accounting Fraud, 2009-178, available at: http://www.sec.gov/news/press/2009/ 2009-178.htm.

Simunic, D.A., 1980, "The Pricing of Audit Services: Theory and Evidence," Journal of Accounting Research, 18(1), 161-90.

Simunic, D.A., and M. Stein, 1996, "Impact of Litigation Risk on Audit Pricing: A Review of the Economics and the Evidence," Auditing: A Journal of Practice & Theory, 15, 119-34.

Sinha, P., L. Brown, and S. Das, 1997, "A Re-Examination of Financial Analysts' Differential Earnings Forecast Accuracy," Contemporary Accounting Research, 14(1), 1- 42.

Tucker, W., 1991, "Shakedown?" Forbes, August 19, 98.

ENDNOTES

(1.) The other two benchmarks are reporting (a) a small profit rather than a small loss, and (b) a small increase in EPS rather than a small decrease in EPS. The importance of earnings thresholds is also recognized by regulators (e.g., SEC, 1999). Our focus on analysts' forecasts is consistent with the findings from prior studies. For example, Dechow et al. (2003, p. 357) state that "meeting analysts' consensus forecasts is becoming the more important hurdle" while Brown and Caylor (2005, p. 424) state that the analyst forecast threshold "is the most important threshold in every year, 1996-2002."

(2.) Chevis et al. (2007) find that MBE firms are more highly valued in the market than "non-meet" firms and that this valuation premium is positively associated with the firm's "meet" horizon.

(3.) The classifications are available at http://mba.tuck.dartmouth.edu/pages/faculty/ ken. french/datalibrary. html.

(4.) Firms in the manufacturing sector constitute nearly half (49.9%) of the sample; this is followed by firms in the services sector (19.1%), and the transportation sector (14.8%).

(5.) By way of contrast, only 695 firms (44%) just missed analyst forecasts by one cent or less at least once during the same period.

Maria Rickling [a], Dasaratha V. Rama [b], and K. Raghunandan [c]

[a] Assistant Professor, M.E. Rinker, Sr. Institute of Tax and Accountancy Stetson University, Deland, FL 32723 mricklin@stetson.edu

[b] Professor, Florida International University ramad@fiu.edu

[c] Professor, Florida International University raghu@fiu.edu

Table 1
Frequency distribution of firms meeting or beating analysts'
EPS forecasts

# of Quarters of Meet-or-Beat # of Firms

0 1
1 1
2 5
3 6
4 19
5 41
6 67
7 94
8 124
9 136
10 116
11 108
12 73
Total: 791

Note: The sample includes all non/financial U.S. firms in the I/B/E/
S database having a fiscal year end of December 31, and having (a)
analyst forecast data for the 12 quarters beginning January 1, 2005
and ending December 31, 2007 in I/B/E/S, (b) audit fee data in the
Audit Analytics database and (c) relevant financial data in the
Compustat annual database.

Table 2
Descriptive statistics (n = 791)

Variable Mean Std. 25th Median 75th
 Dev. percentile percentile

Audit Fees 3,205 4,482 847 1,560 3,394
(in $000s)

Total Assets 7,049 20,535 437 1,450 4,806
(in $millions)

LNAFEE 14.39 1.05 13.65 14.26 15.04
LNTA 7.32 1.75 6.08 7.28 8.48
RECINV 0.21 0.15 0.08 0.19 0.30
SQRTSEG 1.30 0.68 1.00 1.00 1.73
FOROPS 0.64 0.48 0.00 1.00 1.00
LIQ 2.59 2.88 1.24 1.82 2.92
LOSS 0.20 0.40 0.00 0.00 0.00
ROA 0.02 0.17 0.02 0.05 0.09
BM 0.44 0.37 0.23 0.38 0.59
ICW 0.04 0.20 0.00 0.00 0.00
GC 0.00 0.04 0.00 0.00 0.00
BIG4 0.92 0.28 1.00 1.00 1.00
AUDINIT 0.03 0.18 0.00 0.00 0.00
NUM_MB 8.61 2.23 7.00 9.00 10.00

Note: The variables are defined as follows: LNAFEE = natural
logarithm of audit fees; LNTA = natural logarithm of total assets
($000); RECINV= receivables plus inventory, scaled by total assets;
SQRTSEG= square root of the number of business segments per
Compustat; FOROPS = 1 if the firm has foreign operations, 0
otherwise; LIQ = ratio of current assets divided by current
liabilities; LOSS= 1 if the firm reports negative income for the
year, 0 otherwise; ROA= return on assets, measured as net income,
divided by TA; BM = book to market ratio, as of the end of the fiscal
year; ICW = 1 if the firm discloses an internal control weakness, 0
otherwise; GC = 1 if the firm receives a going concern opinion, 0
otherwise; BIG4 = 1 if the firm's auditor is a Big 4 auditor, 0
otherwise; AUDINIT = 1 if it is the auditor's initial year on the
engagement, 0 otherwise; NUM_MB = number of quarters during 2005
through 2007 that the actual reported earnings meets or exceeds the
analysts' forecasts.

Table 3
Regression results

Model: LNAFEE = [[beta].sub.0] +
[[beta].sub.1] * LNTA + [[beta].sub.2] * RECINV + [[beta].sub.3] *
SQRTSEG + [[beta].sub.4] * FOROPS + [[beta].sub.5] * LIQ +
[[beta].sub.6] * LOSS + [[beta].sub.7] * ROA + [[beta].sub.8] * BM +
[[beta].sub.9] * ICW + [[beta].sub.10] * GC + [[beta].sub.11] * BIG4
+ [[beta].sub.12] * AUDINIT + [[beta].sub.13] * NUM_MB +
[[beta].sub.14-22] * (Industry Dummy Variables) + [epsilon]

Variable Predicted sign Coefficient p-value

Intercept 10.096 0.000
LNTA + 0.524 0.000
RECINV + 0.971 0.000
SQRTSEG + 0.027 0.280
FOROPS + 0.266 0.000
LIQ - -0.022 0.001
LOSS + 0.053 0.377
ROA - -0.340 0.034
BM - -0.106 0.027
ICW + 0.321 0.000
GC + 0.107 0.845
BIG4 + 0.148 0.023
AUDINIT ? -0.271 0.004
NUM_MB ? -0.020 0.018
Industry Variables Yes
n = 791 Adj. [R.sup.2] = F-stat = 148.487 P < .001
 0.804

Note: The variables are defined as in Table 2. p-values are
one-tailed, except for AUDINIT and NUM_MB.