Accountancy Business and the Public Interest 2013 AUDIT QUALITY POST SARBANES-OXLEY ACT

Transcription

1 AUDIT QUALITY POST SARBANES-OXLEY ACT by Alireza Dorestani (corresponding author) College of Business and Management Department of Accounting, Business Law and Finance Northeastern Illinois University 5500 North St. Louis Avenue, Chicago, Illinois Phone: (773) , Fax: (773) and Ahmed Jir MSA Alumnus College of Business and Management Department of Accounting, Business Law and Finance Northeastern Illinois University 5500 North St. Louis Avenue, Chicago, Illinois Phone: (773) , Fax: (773) ABSTRACT Since the quality of audit provided by any particular auditor is not observable by audit clients, certain auditor related variables have been used as proxies to gauge audit quality. DeAngelo (1981) argues that audit quality depends on many variables including auditor size. DeAngelo s argument is based on the premise that large audit firms have more at stake to lose than smaller audit firms if they provide poor quality audit. To this end, previous studies on earnings management have used auditor related proxies such as auditor type, client importance, and auditor tenure to test the association between audit quality and earnings management. These studies have found that auditor size has a mitigating effect on the level of discretionary accruals reported by clients of large audit firms. However, after the passage of the Sarbanes Oxley Act (SOX) in 2002 all public audit firms whether small or large undergo similar and stringent requirements by the Public Company Accounting Oversight Board (PCAOB) imposed by the United States Securities and Exchange Commission (SEC). Because the PCAOB requirements are applied irrespective of auditor size, we speculate and show that the superior audit quality of large audit firms has diminished post SOX. In particular, contrary to findings of prior research, our results point to the diminishing importance of auditor size as a proxy for audit quality in the new PCAOB era, post SOX. Key Words: Audit Quality; Auditor Size; Auditor Tenure; Earnings Management; Sarbanes-Oxley Act; PCAOB 119

2 I. INTRODUCTION he need for auditors to give opinion on the integrity of corporate financial reports is mainly driven by the Agency Theory. Agency theory posits the presence of information asymmetry between those who prepare financial reports and those who use them (Jensen and Meckling, 1976). Becker (1998) mentions that auditing reduces the information asymmetry through the examination of financial statement integrity by a third party-the auditor. However, Becker points out that there are other mechanisms such as strong corporate governance and other internal factors that can be used to constrain management behavior. Since auditors are seen as the main factor in reducing the information asymmetry inherent in corporate financial reporting, the issue of audit quality has gained prominence in auditing literature. More importantly, the users of audit reports are also faced with a similar information asymmetry conundrum how to ascertain the quality of audit reports. DeAngelo (1981) defines audit quality as the joint probability of auditor s ability to: (1) detect the misstatement and (2) stand reporting the detected misstatement of financial reporting. The former is a measure of auditor s competence and the latter is a measure of independence. Explicit in this definition is that the audit quality is sacrificed when any of these two conditions are not met. The fact that audit quality (i.e. whether the aforementioned two conditions were met) is not observable by the average users of audited financial statements, the market in general has generated a wide range of discussions and attempts to find the best measures of audit quality. The yard stick used to gauge audit quality can best be summarized into two categories: Supply side/direct measures and demand side/indirect measures. The direct measure of audit quality or supply side is derived from the internal activities of audit firms audit staff, audit manuals and programs, employee training programs, and adherence to AICPA and PCAOB quality control criteria. The high cost of this measure of audit quality and the reluctance of public accounting firms to allow academicians to have access to their audit files have caused researchers to use the other and less reliable measure of audit quality, demand side audit quality. Due to information asymmetry inherent in the demand side of the audit quality, researchers use proxies to gauge the input parameters 120

3 of the auditing process, parameters such as: auditor size, tenure, expertise, client importance, the level of non-audit fees, audit fees, and the levels of discretionary accruals The demise of large corporations such as Enron and WorldCom heightened the audit quality debate in the United States of America. An article published in the Journal of Leadership and Organizational Studies in the summer of 2004 makes an interview with Senator Paul Sarbanes, one of the co-creators of the Sarbanes Oxley Act of 2002 (SOX). The Senator reflects on the hearings done in the United States Senate before the passage of SOX. The Senator asserts that hearings pinpointed to a raft of problems in corporate financial reporting that included, among other issues, impaired auditor independence and perceived conflict of interests. As a result, the United States Congress passed the Sarbanes Oxley Act in 2002 to increase the overall integrity of the financial statements of publically held firms. As such, the main motive of this paper is to examine the difference in audit quality between two periods: before and after the passage of SOX. Implicit in our findings is to test whether the presence of the new Public Company Accounting Oversight Board (PCAOB) requirements will diminish the importance of auditor size, client importance, and auditor tenure as proxies for audit quality. We expect and show that all public auditors are now faced with the same stringent PCAOB requirements. To this end and consistent with prior studies (e.g., Dechow, 1995), we use the levels of discretionary accruals to proxy for earnings management, while auditor size, client importance, and auditor tenure are used as proxies for audit quality. The rest of this paper is organized as follows: We first discuss the relevant literature pertaining to audit quality in section II, especially those that deal with the superiority of auditor size. We proceed with hypothesis development in section III. Then, we discuss our research design method and data collection in section IX, following by results in section X. Last, we highlight the key findings of our study and list our contributions and conclusions in the last section, section XI. 121

4 II. LITERATURE REVIEW Managerial discretion is an important component of accrual accounting. However, the presence of confounding factors such as agency costs cloud the accuracy of financial reporting. Becker et al. (1998) argues that managers have implicit and explicit incentives that are based on reported earnings. These include management compensation and debt agreements that may call for the attainment of specific earning targets that would then trigger the underlying incentives. As such, information asymmetry dictates that users of financial reports find ways to assuage the aforementioned concerns. One of the accepted proxies to gauge managements veracity is to examine total discretionary accruals. Further, since management can either manage earnings upwards or downwards, the direction of discretionary accruals is also important. Some of the existing research pinpoint that most auditor litigation emanates from positive earnings management. That is, the management of earnings through positive accruals. Heninger (2001) posits that there is a positive association between auditor litigation and income increasing accruals. Therefore, the direction of discretionary accruals is of critical importance in audit quality. Further, Johnson et al. (2002) provides additional evidence of the importance of the sign of accruals. In particular, the authors posit that when prior expectation of management s behavior exists, the sign of accruals can be used as a powerful benchmark. In her seminal paper, DeAngelo (1981) argues that larger auditors provide a higher audit quality than their smaller counterparts due to two main factors. First, revenues accrued from one client represent a small portion of the total quasi-rents accrued from all clients. Second, the possible loss of quasi-rents accrued from all other clients entices larger auditors to provide higher quality audits. That is, if larger auditors are exposed as providing lesser audit quality than promised or expected, then they potentially have more at stake to lose than their smaller counterparts. Similarly, research done by Francis (1999) lends credence to the superiority of larger audit firms in providing better audit quality. He argues that, larger auditors are in a better position to stand against aggressive and questionable accounting accruals. 122

5 Further, Francis posits that firms with larger accruals can hire a large audit firm as a signal that they are not managing earnings through accruals. The view that firms use large auditors to signal higher earnings quality is supported by prior research. Carpenter and Strawser (1971) report that firms acquire a big auditing firm prior to going public. In some cases the underwriters require such a change to signal to the securities market that the reported financial position of the new entrant was verified by a reputable auditor. Furthermore, the academic literature has consent that the importance of the client to the auditor has a negative impact on audit quality. Chung and Kallapur (2003) find evidence of positive association between the auditor independence impairment and non-audit fee to total audit fee ratios. In essence, as the ratio of client fees increases, so does the risk of auditor objectivity impairment. This assertion is also supported by DeAngelo (1981), when she describes the auditor client relationship as a bilateral monopoly that incentivizes the auditor to shirk on their professional objectivity. This implies that the perennial pressure on audit firms to recruit or retain an audit client as well as management goal of meeting earnings expectations incentivizes both parties and can lead to goal congruence that will affect audit quality. Chambers (2011) provides additional evidence to the above stipulation when he posits that when the audit fee from one client represents a significant proportion of the firm s total revenue, the audit firm may be reluctant to qualify the client questionable reporting choices. This, however, runs into loggerheads with the brand reputation hypothesis put forth by DeAngelo (1981). That is, big size audit firms have more reputation to protect than just to maintain or raise their short term revenues from a single large client. As another proxy for audit quality, the length of the auditor client relationship has raised the alarm bells in the regulatory community in the United States especially after the demise of high profiled companies that received no going concern audit opinions. Johnson et al. (2002) mentions that during such periods, one of the solutions offered as a panacea to auditor tenure problem is the mandatory audit firm rotation. However, the authors are quick to point out about doubt raised on the effectiveness of mandatory 123

6 auditor rotation. Further, in an empirical investigation the authors uncover that short auditor-client relationships are associated with lower-quality of financial reports, but the same does not hold true for longer audit client tenures. This essentially undermines mandatory audit firm rotations and by proxy supports the notion that longer tenures provide better audit quality. III. HYPOTHESES DEVELOPMENT Since the main purpose of this study is to examine the association between the more stringent PCAOB requirements and audit quality before and after SOX, we have made a comparison of pre and post PCAOB data and speculate that auditor size would be less negatively associated with the discretionary accruals in the post PCAOB era. Although previous literature supports the superiority of larger auditors in providing high audit quality, we expect that this will not be true in the post PCAOB period due to the increased scrutiny of public accounting companies such as the annual or tri-annual PCAOB inspections, and mandatory audit partner rotations. In short, we expect to observe the same audit quality independent of auditor size in post SOX era. That is, we expect to observe a convergence of audit quality among audit firms independent of their sizes. Therefore, H1: Auditor size as a proxy for audit quality is diminished in importance in the post SOX era. Further, according to the academic literature, another important determinant of audit quality is client importance. Research indicates that client importance has a detrimental effect on audit quality. The more important a client is to a particular auditor, the less likely the auditor is to curtail client discretionary accruals. However, we expect that the auditor independence rules established by the PCAOB coupled with the PCAOB annual and tri-annual inspections of audit conduct for some public accounting firms will lessen this detrimental effect. As such, we hypothesize: 124

7 H2: Audit client importance as a proxy for audit quality is diminished in importance in the post SOX era. Prior studies on audit quality literature generally do not support the link between poor audit quality as measured through the levels of discretionary accruals and auditor tenure. In fact, the study by Johnson et al. (2002) finds short auditor tenures as more problematic than longer tenures. Therefore, we hypothesize that auditor tenure will be more negatively associated with the levels of discretionary accruals post PCAOB. That is: H3: Audit tenure as a proxy for audit quality is more negatively associated with discretionary accruals in the post SOX era. IV. METHOD AND DATA COLLECTION Method: Different models are used in the literature to detect earnings management, but as Dechow et al. (1995) argue, the modified version of the Jones model (1991) exhibits the most power in detecting earnings management. Therefore, in this study we first use the following modified Jones model to calculate total accruals.. TA it = CA it - CL it - Cash it + STD it Where: TA it = Total accrual of firm i in year t CA it = Change in current assets of firm i in year t CL it = Change in current liabilities of firm i in year t Cash it Change in cash of firm i in year t STD it = Change in current portion of debt of firm i in year t Then we run the following multivariable linear regression model, using panel data, to estimate the model s coefficients: TA it /A it-1 = α 0i [1/ A it-1 ] + α 1i [( REV it - REC it )/ A it-1 ] + α 2i [PPE it / A it-1 ] + ε it Where: TA it = Total accruals of firm i in year t, REV it = Change in revenues of firm i between years t and t-1, REC it = Change in receivables of firm i in year t, PPE it = The level of property, plant, and equipment of firm i in year t, 125

8 A it-1 = Total assets of firm i at the end of year t-1. Coefficients from the above regression are then used in the equation below to calculate non discretionary accrual of firm t in year t ( NDAit ): NDA it /A it-1 = α 0i [1/ A it-1 ]+ α 1i [( REV - REC)/ A it-1 ] + α 2i [PPE it / A it-1 ] Finally, the discretionary accrual (DA it )is calculated as the difference between total accrual and non discretionary accrual. That is: DA it = TA it - NDA it Then, the discretionary accrual is used as the dependent variable in the following multivariate regression model to test the hypotheses of this study: DA it = β 0 + β 1 AUD it + β 2 IMPRT it + β 3 TENURE it + ν it Where: DA it = The value of discretionary accruals of firm i in year t (as calculated using modified Jones model), AUD it = A proxy for audit quality which is equal to 1 if the audit firm is a big-4 auditor and 0 otherwise. IMPRT it = A proxy for audit quality which is measured by the ratio of an auditor client sales by the sales of all clients for a given auditor. TENURE it = A proxy for audit quality which is equal to 1 if the auditor client relationship started in year t-3 and 0 otherwise. Data Collection: We have used the Research Insights database (COMPUSTAT) to collect financial data for all active companies listed in the COMPUSTAT database for the period of and We have intentionally skipped the period because for many companies the effective year for being fully in compliance with SOX was year Following this procedure, we have started with 9780 firms with a total of close to 80,000 firm-year observations. Following prior studies (e.g., Becker, 1998), we have excluded financial institutions identified with SICs between because their capital structure and risks are not comparable with those of non-financial institutions. Further, consistent with Becker (1998), we have also excluded utility companies identified with SICs between because they have a different rationale for earnings management. Our goal in this study is to collect data for the same variables 126

9 from both and periods; therefore, we have eliminated all companies that do not have complete data in these two periods. After dropping data for financial and utility companies and eliminating outliers and cleaning data, our final sample consists of 2,338 companies with close to 20,000 firm-year observations. V. RESULTS Descriptive statistics in Tables 1 and 2 show that means of the three main audit quality variables are different in the two periods ( and ). The mean of AUD is higher in the pre-sox period than in the post-sox period. The means of TA (total accrual) and NDA (non-discretionary accrual) are significantly more negative in the pre-sox period compared to the post-sox period. The same is true for the mean of DA (discretionary accrual) which is significantly more negative in the pre-sox period than in the post SOX period. Table 3 shows the results obtained when comparing the means of positive and negative discretionary accruals in the pre-sox period ( ). The table shows that the mean of positive discretionary accrual is lower for firms audited by big 4 auditing firms than for firms audited by non-big 4 auditing firms and the difference in mean is significant. Lower mean indicates that firms audited by big 4 firms did not manage discretionary accruals upwards. This finding is consistent with that of Francis (1999) when he expects that both positive and negative discretionary accruals to be smaller for clients of large audit firms. Table 3 also provides similar results when comparing the means of negative discretionary accruals. As this table shows the mean of discretionary accruals for firms audited by big-4 firms is significantly less than the mean of discretionary accruals for firms audited by non-big4 auditing firms. As we mentioned earlier, these findings are consistent with those of Francis (1999) and are in line with the hypotheses of our study with respect to the pre-sox period. Table 4 shows the results from comparing means of positive and negative discretionary accruals in the post-sox period ( ). Although the mean of positive discretionary accruals is smaller for clients audited by big-4 auditing firms, the difference 127

10 is not as stark as in the pre-sox period. This is in line with our expectation that facing similar PCAOB requirement both auditors, big-4 and non big-4, will put forth comparable efforts and achieve similar audit quality levels. The difference of the mean of negative discretionary accruals shows a smaller mean for the firms audited by non-big 4 auditors. This is surprising and contrary to our expectation. This implies that non-big 4 clients accrued less negative discretionary accruals to manage their earnings. Correlation matrix in Tables 5 and 6 indicate significant correlation between our variables of interest in the two periods. In the pre-sox period, the audit quality variable TENURE measuring auditor tenure is significantly correlated (.25) at a one percent significance level with the variable AUD, while the variable IMPRT measuring client importance is correlated (-0.019) with AUD at a 10 percent significance level. In the post-sox period, TENURE and IMPRT are significantly correlated (-.141) at a one percent significance level with variable AUD. As these numbers indicate, the magnitude of correlations is not high and there are no other symptoms that indicate the possibility of multicolinearity in our study. Finally, Table 7 shows the regression results for directly testing the hypotheses of this study. This table shows that for pre-sox period the coefficient for the variable AUD ( ) is highly significant (p-value =.009) and negative, indicating that big-4 audit firms provided better quality audit compare to their smaller counterparts during pre SOX period. However, this significance does not hold during post-sox period. This finding is consistent with our expectation and hypothesis that when faced with more stringent PCAOB regime, auditor size will be less negatively associated with the total discretionary accruals. On the other hand, other audit quality variables, TENURE and IMPRT, show no sign of significance in either of the two periods. VI. CONCLUSIONS The main purpose of this study is to examine the impact of PCAOB requirements post SOX on accepted and existing audit quality parameters. Using discretionary accruals as a well established measure of audit quality, our study provides evidence showing that 128

11 the levels of both positive and negative discretionary accruals are significantly impacted by the size of the auditor in pre-sox period ( ) as evidenced in observations of the difference in the means of the discretionary accruals. The same result does not hold post-sox period, and even in the case of negative discretionary accruals the finding is surprising because it indicates that non-big 4 audit clients exhibit smaller means. Further, the correlation matrix shows significant correlations between the audit quality variables. In sum, as expected, this study provides additional evidence that auditor size was a significant factor in determining audit quality in the pre-sox period. However, the paper provides evidence that auditor size has not been a significant factor in the level of audit quality during post-sox period. Moreover, the results also show that auditor tenure and client importance have no mitigating effect on earnings management in either period. In sum, contrary to prior research, our results point to the diminishing importance of auditor size as a proxy for audit quality in the new PCAOB era. This finding supports our hypothesis and is of particular importance to audit clients of larger public accounting firms, as they can now purchase audit services at a lower cost from smaller but equally competent public accounting firms. Our study is expected to contribute to the auditing literature in many ways. First, we extend the test of demand side proxies of audit quality on discretionary accruals to a different data set than what were used in prior studies. We use two identical sets of data encompassing all active firms with all the variables of interest from the Research Insights (COMPUSTAT) database to collect financial data for and periods. Second, by looking at close to 20,000 firm-year observations, which is significantly larger than what are used in prior studies, we believe the power of our test is higher than that of prior studies. Third, we compare the efficacy of the stringent PCAOB requirements by comparing the effect of audit quality measures on discretionary accruals for the two periods and provide evidence on the effectiveness of requirements set by PCAOB on improvement of audit quality of firms known as non-big 4 audit firms. Fourth, we provide evidence to show that smaller audit firms are as good as big 4 when it comes to audit quality, so their clients can switch to smaller audit firms 129

12 . Accountancy Business and the Public Interest 2013 without any loss of audit quality while saving on their audit fees. Fourth, our findings are in favor of supporting smaller and probably younger audit firms. Lastly, we have divided audit quality to demand side and supply side and discuss the importance of supply side or direct observation of audit quality. To the best of our knowledge these issues are not addressed in prior studies. References Becker, C. L., DeFond, M.L., Jiambalvo, J., and Subramanyam, K.R. (1998). The effect of audit quality on earnings management. Contemporary Accounting Research, 15, Carpenter, C. G. and Strawser, R. H. (1971). Displacement of auditors when clients go public. Journal of Accountancy, 131 (6), Chambers, D, Payne, J. (2011) Audit quality and accrual persistence: evidence from the pre- and post-sarbanes-oxley periods. Managerial Auditing Journal, 26(5), Chung, H., Kallapur, S. (2003). Client importance, nonaudit services, and abnormal accruals. The Accounting Review, 78 (4), DeAngelo, L. (1981). Auditor size and audit quality. Journal of Accounting and Economics, 3: Dechow, P., Sloan, R., & Sweeney, A. (1995). Detecting earnings management. The Accounting Review, 70(2), Francis, J. R., Maydew, E.L., and Sparks, H.C. (1999). The role of big 6 auditors In the credible reporting of accruals. Auditing: A Journal of Practice and Theory, 18, Heninger, W., (2001). The association between auditor litigation and abnormal accruals. The Accounting Review, 76(1), Johnson, V. E., Khurana, I. K. and Reynolds, J. K. (2002): Audit-firm tenure and the quality of financial reports. Contemporary Accounting Research, 19 (4): Jones, J. (1991). Earnings management during import relief investigations. Journal of Accounting Research, 29, Lucas, N. (2004). An Interview with United States Senator Paul S. Sarbanes. Journal of Leadership & Organizational Studies, Summer 2004, 11,

13 Table 1: Descriptive Statistics for period between Variable Obs Mean Std. Dev. 25th% tile Median 75th %tile Min Max AUD TENURE IMPRT E CA-CH CL_CH CASH_CH STD_CH DEP TA LAT_INV CH_REV PPE CH_REC SUM_CH NDA DA Table 2: Descriptive Statistics for period between Variable Obs Mean Std. Dev. 25th% tile Median 75th %tile Min Max AUD TENURE IMPRT E CA-CH CL_CH CASH_CH STD_CH DEP TA LAT_INV E CH_REV PPE CH_REC SUM_CH NDA DA

14 Table 3: Positive DA for : Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] pda_big pda_nbig diff Pr(T < t) = Negative DA for : Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] nda_big nda_nbig diff Pr(T < t) = Table 4: Positive DA for : Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] pda_big pda_nbig diff Pr(T < t) = Negative DA for : Variable Obs Mean Std. Err. Std. Dev. [95% Conf. Interval] nda_big nda_nbig diff Pr(T < t) =

15 Table 5: Correlation Matrix for Variables Used in Period AUD AUD TENURE IMPRT CA_CH CL_CH CASH_CH STD_CH DEP TA LAT_IN V CH_REV PPE CH_REC TENURE 0.25""" IMPRT * CA_CH CL_CH * -0.09*** CASH_CH *** STD_CH DEP 0.125*** 0.05*** 0.104*** 0.038*** *** *** TA LAT_INV ** -0.08*** ** -0.03*** 0.350*** -0.22*** CH_REV 0.028** *** 0.05*** 0.02** PPE 0.105*** 0.044*** 0.102*** *** *** -0.03** 0.385*** -0.27*** 0.480*** 0.298** CH_REC *** * 0.069*** 0.278** SUM_CH 0.033*** **** *** *** 0.067*** 0.02* 0.031*** ** * 0.07*** * ** Table 6: Correlation Matrix for Variables Used in Period AUD TENURE IMPRT CA_CH CL_CH CASH_CH STD_CH TA LAT_INV CH_REV PPE CH_REC AUD TENURE 0.215*** IMPRT *** -0.09*** CA_CH 0.035*** CL_CH *** CASH_CH *** 0.128*** STD_CH *** 0.378*** -0.27*** TA LAT_INV CH_REV PPE 0.021* 0.021* * 0.025** ** *** CH_REC * *** 0.093*** *** 0.156*** *** *** SUM_CH *** ** 0.040*** 0.047*** ** 0.023** *** 0.027** ** *** 133

16 Table 7: Regression Results BEFORE SOX ( ): After SOX ( ): DA Coef. Std. Err. z P>z Coef. Std. Err. z P>z AUD *** TENURE IMPRT _CONS Adj. R-squared = Adj. R-squared =