Seasonal_Workload_Compression_Affects

The Effect of Workload Compression on Audit Quality Dennis M. López-Acevedo* PhD Candidate Sam M. Walton College of Business University of Arkansas ABSTRACT: This paper investigates whether seasonal workload compression affects audit quality. Using 52,672 firm-year observations over the period 1990-2000, we estimate discretionary accruals (Jones 1991) as a proxy for earnings management and concomitant audit quality. We find that workload compressed audits display greater magnitudes of discretionary accruals. The results are consistent with compressed workloads impairing audit quality, thus giving client’s more freedom to manage earnings.  We also find that workload compressed firms with income-increasing (income-decreasing) earnings management are able to book more positive (negative) discretionary accruals.  The results of this study have implications for regulation that either compresses the audit process or alternatively spreads auditors’ workload over the course of the year. Recent examples of such regulation include the SEC’s accelerated filing requirements.   Key Words: Audit Quality, Earnings Management, Workload Compression, Discretionary Accruals Draft: November 15, 2005 *Please direct comments and suggestions to Dennis M. López-Acevedo, University of Arkansas, Department of Accounting, 401 Sam M. Walton College of Business, Fayetteville, AR 72701. Email: dlopez@uark.edu; Phone: 479-575-612; Fax 479-575-2863. This paper is based on my doctoral dissertation at the University of Arkansas. My doctoral dissertation committee consists of Carolyn M. Callahan (Chairperson), Gary F. Peters, and Gary Ferrier. I. Introduction This study investigates the effects of workload compression demands on the quality of audited financial statements. Recent high profile audit failures have highlighted the importance of the auditors’ oversight function in the financial reporting process. Using empirical data I find that audits performed under workload compression conditions are of lower audit quality when compared to audits performed under non-workload compression conditions. The results of this study demonstrate the importance of adopting measures that would help to keep auditors’ workloads evenly spread throughout year. Workload compression (WLC), better known by auditors as the “busy season,” occurs as a result of most companies having their fiscal years aligned with the calendar year. For many auditors, the busy season is plagued with long hours, fatigue, and demanding time budget constraints. DeAngelo (1981) defines audit quality as the probability that the auditor will both discover and report a breach in the client’s accounting system. Fatigue and tight time budgets, common conditions in workload compressed audits, can decrease auditors’ capacity to either discover or report any existing exceptions. This study represents one of the first attempts to investigate the effects of workload compression on audit quality from an empirical perspective. Experimental studies have shown that auditor burnout and budget constraints may lead auditors to perform lower quality audits (Alderman & Dietrick, 1982; Kelley and Margheim, 1990; Ragunathan, 1991; Willet and Page, 1996; Sweeney and Summers, 2002; Coram et al., 2004). In contrast to previous behavioral studies, the focus of this study will be on the effect of WLC on overall audit quality. I find evidence that workload compression affects audit quality across all levels of the audit firm staff. The proxy measure for audit quality, discretionary accruals, is estimated using the cross-sectional version of the Jones model (Lee and Mande, 2003; Myers et al., 2003; Heninger, 2001; Jones, 1991). Accrual based accounting requires extensive use of judgment and estimation. This creates an opportunity for management to manipulate earnings and thus affects earnings quality. Consistent with previous literature, earnings quality is used in this study to draw inferences about audit quality (Myers, 2003). I find evidence indicating that WLC firms exhibit lower levels of audit quality, as proxied by higher levels absolute discretionary accruals. I also find evidence indicating that workload compression promotes the existence of income-increasing accruals as well as income-decreasing discretionary accruals. Thus, auditors do not adjust audit rigor with respect to the direction of management discretions in workload compressed firms. The reminder of the paper is organized as follows. Section II presents background information and hypothesis development. Section III presents the sample selection and methodology. Section IV presents the empirical results. Section V presents the conclusions of this study. II. Background and Hypothesis Development Workload Compression and Audit Quality A great majority of domestic corporations close their fiscal year on or around December 31, creating what is known as the “busy season.” As shown in Figure 1, 58.12% of all Compustat firms (years 1951-2004) have a December year-end date. The high concentration of companies with fiscal years ending in December imposes a significant burden on auditors during the first calendar quarter of each year. Anecdotal evidence indicates that the intense demands of the busy season can diminish employee performance and lead to low morale, absenteeism and the high turnover rates experienced in the accounting industry. Academic research on the impact of workload compression on audit quality is very limited. There are a handful of studies providing evidence that WLC may have an adverse effect on auditor behavior. For example, using Sweeney and Summers (2002) find that busy season demands cause a significant escalation in public accountant’s burnout, as measured by the Maslach Burnout Inventory (Maslach and Jackson 1981, 1986). The researchers investigate three different dimensions of the burnout construct – emotional exhaustion, depersonalization, and reduced personal accomplishment. Their study finds that by the end of the busy season subjects appear to experience significantly higher emotional exhaustion levels from their work and are more depersonalized in their approach to their job. There is also experimental and survey evidence indicating that time budget pressures lead auditors to engage in dysfunctional behavior or to perform substandard audit work. For example, some of the evidence indicates that time constraints are one of the major motivators of premature sign-offs (Alderman and Dietrick, 1982; Ragunathan, 1991; Kelley and Margheim, 1990). In a questionnaire survey of audit personnel of international accounting firms, Ragunathan (1991) finds that time limitations motivated 24 percent of identified premature sign-offs. Kelley and Margheim (1990) find evidence indicating that auditors underreport audit hours and engage in other audit quality reduction acts when under audit time pressure. The authors identify premature sign-offs, superficial reviews of client documents, reductions in the amount of work performed to unreasonable levels, and the acceptance of weak client explanations as audit quality reduction acts occurring in practice. Kelley and Margheim (1990) also find that there is an inverted U-shaped relationship between time budget pressures and the aforementioned dysfunctional behaviors. That is, greater amounts of pressure result in increased dysfunctional behavior until a point is reached when the time budget is simply unattainable and auditors do not feel as pressured to engage in dysfunctional acts to meet the budget. In a similar study Coram et al. (2004) examine the joint effects of time pressure and risk of misstatement on auditors’ propensity to deliver substandard audit work. In contrast to previous research, Cronam et al. (2004) study takes into consideration the auditor’s reaction to the probability of misstatement. They find evidence that auditors appear to accept doubtful audit evidence in the presence of time budget pressures, regardless of the level of misstatement risk. However, subjects only truncate the number of items in a sample when time budget pressure is high and risk of misstatement is low. Knechel and Payne (2001) find that the audit reports of busy season year-end companies are dated on average 17.34 days later than those of non-busy season year-end companies. This result provides some evidence that WLC clients receive divided attention, and thus additional days are necessary to complete busy season audits. This study draws upon the theory that auditor burnout and time budget constraints lead auditors to engage in audit quality reduction acts. Auditor burnout and time constraints are two conditions that are more likely to occur in WLC audits. In contrast to previous related studies, the focus of this study is not on auditor burnout or other dysfunctional behaviors that time budget pressures seem to induce in individual auditors. Using empirical data this study investigates the effects of WLC on the overall quality of the audit. This is a significant contribution to the literature for two reasons. First, this study represents one of the first attempts to investigate the effects of workload compression from an empirical perspective. Second, most previous studies use junior auditors as experimental subjects. The ultimate quality control checks in an audit take place at the later stages of the audit when working papers and other documentation are reviewed by senior auditors. Thus, this study provides evidence on whether workload compression affects audit quality across all levels of the audit firm staff. Audit quality is matter of concern to various stakeholders groups. For instance, the Committee of Sponsoring Organizations of the Treadway Commission (COSO) issued a report in 1987 that identifies causal factors that may lead to fraudulent financial reporting. In this report, better known as the Treadway Commission Report, COSO states that accounting firms should recognize and control individual pressures that potentially reduce audit quality. Tight reporting deadlines are identified as one of these pressures. The report states that: “…[firms should] relieve deadline pressures that may encourage auditors prematurely to stop pursing identified problems. These pressures are particularly troublesome because, as the Commission’s studies indicate, activities that result in fraudulent financial reporting typically occur near the end of a reporting period.” (p. 56) It is disconcerting that more than one decade after the Treadway Commission report, the Panel on Audit Effectiveness of the Public Oversight Board in 2000 reported a very similar view on the issue: “…[time] pressures can create an environment in which audit quality might be compromised if engagement team members, at any level, perceive that their individual performance is measured primarily by meeting time deadlines and budget estimates. These threats to audit quality frequently appear at or near the completion of the engagement in the form of client pressures on the engagement team to ‘finalize the audit’ and hurry the issue-resolution process.” (p.105) On September 5, 2002 the SEC adopted a rule requiring accelerated filing of quarterly and annual reports (SEC Release Number 33-8128). Under the new rules public companies have 60 days (instead of 90) after fiscal year-end to file their 10-K reports and 35 days (instead of 45) after the end of each quarter to file their 10-Q reports. As stated by the SEC, one of the main objectives of this change is to increase the timeliness of accounting information. However, accelerated filing also has the effect of further compressing auditors’ workload into a shorter busy season. The SEC received a total of 302 comments during the proposal stage of accelerated filing (SEC Release Number 33-8128). In general, the comments on the accelerated filing proposal SEC fell into two groups—one group supporting accelerated filing (6.6 percent of the commenters) and a second group opposing accelerated filing (93.4 percent of the commenters). In particular, 70.9 percent of the commenters express concerns about the accuracy and quality of financial reporting once the proposed deadlines are implemented. The full implementation of accelerated filing has been delayed by the SEC more than once in response to complaints by filers and their auditors. They complain that stricter internal control requirements of the Sarbanes-Oxley Act are placing substantial demands on personnel and systems key to prepare and file periodic reports. Audit Quality and Discretionary Accruals Audit quality is defined as the probability that auditors will both discover and report a breach in the client’s accounting system (DeAngelo, 1981). Management’s ability to manipulate the financial reporting process decreases as audit quality increases because auditors are more likely to identify and correct any existing manipulations (Becker et al., 1998). This study does not investigate audit quality in the traditional sense. That is, quality as auditor’s ability to comply with Generally Accepted Auditing Standards (GAAS). I investigate audit quality as auditor’s capacity to constrain extreme management’s reporting decisions. Following an approach similar to Myers et al. (2003) I investigate the relationship between earnings quality, as proxied by Jones 1991 discretionary accruals, and workload compression. Accrual based accounting requires extensive use of judgment and estimation. This creates an opportunity for management to manipulate earnings and thus affects earnings quality. Consistent with previous literature, earnings quality can be used to draw inferences about audit quality (Myers, 2003). For example, under the hypothesis that Big Six auditors are higher audit quality providers, Becker et al. (1998) find that clients of non-Big Six auditors report relatively more income-increasing discretionary accruals than the clients of Big Six auditors. In a related study, Lee and Mande (2003) examine the effects of regulatory changes on the audit quality of Big N auditors. In particular Lee and Mande (2003) investigate whether the Private Securities Litigation Reform Act (PSLRA) of 1995 affected audit quality by eliminating joint and several liability for auditors. PSLRA significantly decreased auditor litigation risk for Big N auditors since Big N auditors are often named to lawsuits mainly because they have “deep pockets.” Lee and Mande (2003) find that PSLRA induced a decrease in audit quality, as evidenced by a rise in income-increasing discretionary accruals for auditees of Big Six firms but not for auditees of non-Big Six firms. Using different measures of accruals as proxies for audit quality, Myers et al. (2003) find that longer auditor tenures are associated with higher audit quality. The researchers interpret this result as suggesting that longer auditor tenures, on average, result in auditors placing greater constraints on extreme management decisions in the financial performance reporting process. This study investigates whether audit quality varies as a function of workload compression. That is, auditors allow more earnings management, via discretionary accruals, on workload compressed audits. This discussion leads to the following testable research hypotheses, stated in the alternative: H1: Audits performed under WLC conditions exhibit greater magnitudes of discretionary accruals than audits performed under non-WLC conditions. This hypothesis investigates whether WLC affects the magnitude of discretionary accruals. Assuming that H1 holds true, it is necessary to investigate if there is a directional component to the effects of workload compression on audit quality. This would help to determine if workload compression only affects audit quality in firms with income-increasing earnings management, income-decreasing earnings management, or both. H2: WLC firms with income-increasing earnings management exhibit discretionary accruals that are more positive than the discretionary accruals of non-WLC firms. H3: WLC firms with income-decreasing earnings management exhibit discretionary accruals that are more negative than the discretionary accruals of non-WLC firms. The next section presents the sample selection and methodology of this study. III. SAMPLE SELECTION and Methodology Sample Selection This study uses a sample of cross-sectional observations. The initial sample includes all firm-years from 1990 to 2000 with total assets at least one million and all information necessary in the Compustat annual industrial and research files to estimate the cross-sectional Jones 1991 model. Observations from fiscal years 2001 or later are not included in the sample in order to avoid the possible confounding effects of the reporting changes generated by the Sarbanes-Oxley Act. The initial sample contains 88,562 firm-year observations. Financial institutions, utility companies, and other highly regulated industries (SIC codes 6000-6999 and 4000-4999) are eliminated because of their different regulatory and reporting requirements. Observations with fiscal year-end changes are excluded to eliminate possible confounding effects of earnings management opportunities created by these changes. Following Myers, et al. (2003), firms in the top or bottom 1 percent of the annual distribution of cash flows from operations or Jones 1991 model residuals are deleted to remove outliers. In addition, firm-year observations without enough information to operationalize study variables are also eliminated. There are 52,672 firm-year observations in the final sample. A total of 9,440 different companies are represented in the sample. Table 1 presents the industry distribution for both WLC and non-WLC firms. A total of 31,468 (59.7 percent) firm-year observations belong to the WLC group and 21,204 (40.3 percent) firm-year observations belong to the non-WLC group. There are 48 industry categories represented in the sample. Tests of the difference in sample proportions by industry groups revealed that the proportion of WLC firm is higher than the proportion of non-WLC firms for 14 industry groups; lower for 14 industry groups; and the same for 20 industry groups (α = .01). Discretionary Accruals Estimation Accounting researchers have developed several different models intended to measure the direction and extent of the earnings management activities of a company. The most sophisticated of these models separate discretionary accruals from non-discretionary accruals. Discretionary accruals (DA), the proxy measure of audit quality in this study, are estimated using the cross-sectional version of Jones 1991 model. There is empirical evidence demonstrating that the cross-sectional Jones model and the cross-sectional Modified Jones model perform better than their time-series counterparts in detecting earnings management (Bartov et al., 2000). An additional advantage of the cross-sectional version of the model is that it has less restrictive data requirements than its time-series counterpart. An estimate of total accruals is first obtained in order to estimate discretionary accruals using the Jones 1991 model. Total accruals are then regressed on variables that are proxies for normal accruals. Discretionary accruals are thus the unexplained or the residual components of total accruals. In this study total accruals are estimated using the cash flows approach developed by Hribar and Collins (2002). Hribar and Collins (2002) find that using the alternative balance sheet approach to estimate total accruals can lead to measurement errors in accrual estimates. Total accruals (TA) under the cash flows approach are estimated as follows: TAit = EBXIit – CFOit (1) where TA = total accruals EBXI = earnings before extraordinary items and discontinued operations (Compustat # 123) CFO = cash flows from operations (Compustat #308 – Compustat #124) The subscripts, i and t, index firm and year, respectively. The Jones 1991 model estimates discretionary accruals after controlling for changes in the firm’s economic environment. Non-discretionary accruals are estimated as follows: NDAit = α1 (1 / Ait-1) + α2 (ΔREVit / Ai,t-1) + α3 (PPEit / Ai,t-1) (2) where NDA = non-discretionary accruals A = total assets ΔREV = change in revenues PPE = gross property plant and equipment α1, α2, and α3 are firm specific parameters The subscripts, i and t, are as in equation (1). Estimates of the firm specific parameters, α1, α2, and α3, are obtained from the following regression: TAt / At-1 = a1(1 / At-1) + a2(ΔREVt / At-1) + a3(PPEt/A t-1) + εit (3) where a1, a2, and a3 denote the ordinary least square estimates of α1, α2, and α3. All other variables are as defined in equation (2). Since the models are estimated cross-sectionally, a1, a2, and a3 are estimated for each year-industry group. Year-industry groups are formed using the first two digits of the SIC code of the firms included in the sample. Year-industry groups with less than 10 firm-year observations are eliminated. An estimate of the discretionary component of total accruals is captured by the error term of equation (3). Thus, discretionary accruals can be expressed as: ^DAit = ^ TAit – ^NDAit (4) where ^DA = discretionary accruals ^TA = total accruals ^NDA = non-discretionary accruals ^TA and ^NDA is as computed in equation (1) and equation (2), respectively. Regression Model Development The regression model equation for the main investigation is as presented below: DA = α + β1WLC + β2TENURE + β3AUD_SIZE + β4SIZE + β5∆NI + β6∆ASSETS + β7LOSS + β8 Z_SCORE + Σ YEAR + ε (5) where DA = Jones 1991 model discretionary accruals WLC = 1 if the firm has a December or January year-end date; 0 otherwise TENURE = number of years the audit has been performed by the same auditor AUD_SIZE = 1 if auditing firm is a Big N auditor; 0 otherwise SIZE = log if total asset ∆NI = percentage of change in net income ∆ASSETS = percentage of change in total assets LOSS = 1 of the firm reported a loss during the year; 0 otherwise Z_SCORE = Altman’s Z score Σ YEAR = matrix of year dummies Workload compression (WLC), the independent variable of interest, is operationalized as a dichotomous variable that takes a value of 1 if the observation comes from a WLC firms and zero otherwise. Following Knechel and Payne (2001), WLC audits are defined as audits performed for companies whose fiscal years end in December or January. WLC is expected to have a positive coefficient, indicating that financial statements of workload compressed audits carry higher levels of discretionary accruals, which translates into lower quality audits. In addition to the WLC dummy variable, there are two groups of control variables in the regression model. These two groups are composed of auditor-related factors and client-related factors that can affect the direction and extent of the earnings management activities of a company. Auditor-related factors include auditor tenure and auditor size. Auditor tenure (TENURE) is the number of years that an auditor remains with the same client firm, beginning with year 1985. Values for this variable range from 1 to 16, with audit firm mergers considered a continuance of the previous auditor (Myers et al., 2003; Krishnan and Krishnan, 1997; Stice, 1991). Empirical evidence indicates that auditors with shorter tenures commit more errors and experience higher litigation risks than other auditors (St. Pierre and Anderson 1984). Myers et al. (2003) find evidence suggesting that longer auditor tenures allow auditors to place greater constraints on extreme management decisions in the reporting of financial performance. On the other hand, independence may be impaired when an auditor is retained by a client for too long. For instance, some studies have found evidence indicating that clients get more reporting flexibility as the length of auditor tenure increases (Knapp 1991; Deis and Giroux, 1992). The actual relationship between tenure and audit quality remains an unsolved empirical issue. Thus, I make no prediction on the sign of the coefficient for TENURE. Auditor size (AUD_SIZE) is an dummy variable that takes a value of 1 if the auditor is a Big-N firm and 0 otherwise. Empirical evidence indicates that Big-N auditors are better audit quality providers (Craswell et al., 1995; Francis and Wilson, 1998). For example, Becker et al. (1998) find that clients of non-Big-N firms report discretionary accruals that are, on average, 1.5 to 2.1 percent of total assets higher than the discretionary accruals reported by clients of Big-N auditors. Consistent with previous research, AUD_SIZE is expected to have a positive coefficient. Client-related controls include client size, client industry, incentives to manage earnings, incentives to manage firm growth, incentives to take a “big bath,” and incentives to use aggressive accounting to avoid violation of debt covenants. Client size (SIZE) is operationalized as the log of total assets. This variable, besides controlling for the potential effects of size, also surrogates for possible omitted variables (Becker et al., 1998; Davidson and Neu, 1993). The expected coefficient for SIZE is negative because larger firms have incentives to present less discretionary accruals to avoid litigation (Lang and Lundholm, 1993). Many forms of management compensation contracts create incentives to manage earnings by being dependent on the firm’s net income. The percentage change in net income (∆NI) from the previous to current year controls for management incentives to manage net income (Lee & Mande, 2003). The percentage change in total assets (∆ASSETS) is used to control for managers’ incentives to manage firm growth (Lee and Mande, 2003). Previous research has shown that discretionary accrual models do not perform well for firms with extreme financial performance (Kothari et al., 2005; Hribar and Collins 2002; Dechow et al. 1995). The variable ∆ASSETS also controls for firms with extreme asset growth. Firms experiencing poor financial performance are prone to manage discretionary accruals opportunistically (Lee and Mande, 2003). A dummy variable (LOSS) controls for client incentives to take a “big-bath” during years of poor financial performance (Lee & Mande, 2003). LOSS takes a value of 1 if the company reports losses during the year, and zero otherwise. Altman’s Z-score (Z_SCORE) controls for management’s incentives to use aggressive accounting to avoid debt covenant violations (Sweeney, 1994). In addition, Z_SCORE partially controls for auditor litigation risk since companies in financial distress carry higher levels of litigation risk (Krishnan & Krishnan, 1997; Stice 1991; Lee & Mande, 2003). A matrix of year dummies (YEAR) was included to control for calendar year and general environmental changes such as changes in technology. IV. Empirical Results Table 2 presents descriptive statistics about the sample. There are three panels in this table. Panel A presents descriptive statistics for all the observations in the final sample (n=52,672). This panel shows that 59.7 percent of the observations come from workload compressed firms. Similarly, the mean value for AUD_SIZE indicates that 83.5 percent of the observations were audited by a Big N auditor. Panel B presents descriptive statistics for WLC and non-WLC firms. The discretionary accruals of WLC and non-WLC firms are statistically different at the .001 level. Idiosyncratic differences in the accruals of firms with a December year-end date could be suspected as a confounding factor in this study. For instance, retailers with a December year-end are expected to carry proportionally higher balances of accounts receivable. However, WLC firms carry lower balances of discretionary accruals in average, partially indicating that any possible idiosyncratic differences in the accruals of WLC firms only make the results of this study more conservative. Panel B also shows that WLC firms are is significantly larger in average than non-WLC firms, highlighting the importance of including SIZE as a control variable in this study. The proportion of firms audited by Big N auditors is 85.8 and 80.1 percent for WLC and non-WLC firms, respectively. These proportions are statistically different at the .001 level. However, this result should not be interpreted as an indication of an imbalance in the proportion of WLC and non-WLC firms audited by a Big N auditor since the different in the proportions is less than 5 percentage points. Panel C shows descriptive statistics for sub-samples created using the signed values of discretionary accruals. As shown in this portion of the table, the proportion of WLC firms remains at approximately 60 percent after using the signed values of discretionary accruals to partition the sample. This panel also shows that there is enough statistical evidence to conclude that, in average, Z_SCORE is lower for firms with negative discretionary accruals. Similarly, the proportion of firms with negative discretionary accruals and a loss is significantly higher (49.2 percent) than the proportion of firms with positive discretionary accruals and a loss (28.1 percent). These proportions are statistically different at the .001 level. In sum, firms with positive discretionary accruals seem to be in a relatively better financial condition than firms with negative discretionary accruals. Table 3 presents the Pearson correlations among the independent and dependent variables in the regression model. The coefficients are estimated using all available firm-year observations in the sample. Most the coefficients are significant at the .01 alpha level and all correlations are below .39. Table 4 presents regression results on the impact of workload compression on discretionary accruals. There are three regressions presented in this table, all significant at the .001 level. The first regression is on the absolute value of discretionary accruals. This regression provides evidence on the effect of workload compression on the magnitude of management discretions. The second and third regressions are the on signed values (positive and negative) of discretionary accruals. These two regressions provide evidence on the effect of workload compression on the direction of management discretions. The coefficient for WLC on the absolute value of discretionary accruals is positive and significant indicating that, on average, workload compressed audits display greater magnitudes of discretionary accruals. This result provides evidence that workload compression decreases audit quality. However, to investigate the possible directional effects of workload compression on management discretions it is necessary create a sample partition based on the sign of discretionary accruals. This sample partition helps to determine if WLC firms with income-increasing/income-decreasing earnings management are able to book more positive/negative discretionary accruals than non-WLC firms. Positive discretionary accruals can be used by management to boost the financial performance of a company. On the other hand, negative discretionary accruals can be used to smooth financial performance or to create “cookie jar” reserves that allow managers to increase earnings in the future. The second and third regressions in Table 4 were estimated using only firm-year observations with positive and negative discretionary accruals. These regressions test H2 and H3, respectively. The third regression model in this table is estimated using the absolute value of negative discretionary accruals for ease of comparison and interpretation. The regressions show that the estimated coefficients for WLC are positive and significant for both models, indicating that workload compression promotes the existence of income-increasing accruals as well as income-decreasing discretionary accruals. When taken in conjunction, these results provide evidence that auditors do not adjust audit rigor with respect to the direction of management discretions in workload compressed clients. The importance of this finding is highlighted by the fact that income-increasing accruals carry higher levels of auditor litigation risk than income-decreasing accruals (Lee and Mande, 2003). In sum, the results in Table 4 suggest that workload compression companies present discretionary accruals that are of greater magnitude and may affect income in either direction. Table 5 presents regression results when industry dummies are included in the regression model. The dummies are generated using the first two digits of the SIC code of the company and test whether the results of this study are sensitive to industry affiliation. The significance and direction of the estimated coefficient for the WLC dummy remains unchanged after industry affiliation is taken into consideration. As an additional sensitivity test, I re-estimate the models using the Modified Jones 1991 model and/or the balance sheet approach to estimate total accruals. The significance and interpretation of the results remain unchanged, providing evidence that findings of this study are not sensitive to different specifications of the Jones model. V. Conclusions This study investigates the effects of workload compression demands on the quality of audited financial statements. Three different measures of discretionary accruals are used to proxy for audit quality: absolute value of discretionary accruals, positive discretionary accruals, and negative discretionary accruals. I find that audits performed under workload compression conditions are of lower audit quality when compared to audits performed under non-workload compression conditions. Specifically, I find evidence indicating that WLC firms exhibit lower levels of audit quality, as proxied by higher levels absolute discretionary accruals. I also find evidence indicating that workload compression promotes the existence of income-increasing accruals as well as income-decreasing discretionary accruals. Thus, auditors do not adjust audit rigor with respect to the direction of management discretions in workload compressed firms. This study represents a significant contribution to the auditing literature for two reasons. First, it represents one of the first attempts to investigate the effects of workload compression from an empirical perspective. Second, most previous studies use junior auditors as experimental subjects. The ultimate quality control checks in an audit take place at the later stages of the audit when working papers and other documentation are reviewed by senior auditors. Thus, this study provides evidence that workload compression affects audit quality across all levels of the audit firm staff. The findings of this study highlight the need to adopt regulation that would evenly spread auditors’ workloads throughout the year. For example, new regulations should limit the number of firms with a December fiscal year-end date or increase the proportion of procedures that auditors are allowed or able to perform before the fiscal year-end date. Recent developments in the auditing industry (i.e., continuous auditing, new technology) might mitigate the effects of workload compression on audit quality in the future. References Alderman, C.W., J.W. Dietrick, 1982. Auditors’ perceptions of time budget pressures and premature sign-offs. Auditing: A Journal of Practice and Theory 1(2): 54-68. Altman, E. 1983. Corporate Financial Distress: A Complete Guide to Predicting, Avoiding, and Dealing with Bankruptcy. New York, NY: John Wiley & Sons, Inc. Bartov, E., F.A. Gul, and J.S.L. Tsui, 2000. Discretionary-accrual models and audit qualifications. Journal of Accounting and Economics 30(3): 421 - . Becker, C.L., M.L. DeFond, J. Jiambalvo, and K.R. Subramanyam, 1998. The effect of audit quality on earnings management. Contemporary Accounting Research 15(1): 4-24. Coram, P., J. Ng, and D.R. Woodliff, 2004. The effect of risk of misstatement on the propensity to commit reduced audit quality acts under time budget pressure. Auditing: A Journal of Practice and Theory, 23(2): 159-167. Craswell, A., J. Francis, and S. Taylor. 1995. Auditor brand name reputations and industry specializations. Journal of Accounting and Economics (December): 297-322. Davidson, R.A. and D. Neu, 1993. A note on the association between audit firm size and audit quality. Contemporary Accounting Research 9(2): 479-488. DeAngelo, L.E., 1981. Auditor independence, “low balling,” and disclosure regulation. Journal of Accounting and Economics 3 (August): 113-127. Dechow, P., R. Sloan, and A. Sweeney. 1995. Detecting earnings management. The Accounting Review 70 (April): 193-225. Deis, D. R. and G.A. Giroux. 1992. Determinants of audit quality in the public sector. The Accounting Review 67 (July): 462-479. Francis , J. and E. Wilson. 1998. Auditor changes: A joint test of theories relating to agency costs and auditor differentiation. The Accounting Review 63 (October): 663-682. Hribar, P. and D.W. Collins. 2002. Errors in estimating accruals: Implications for empirical research. Journal of Accounting Research 40 (1): 105-135. Jones, J. J. 1991. Earnings management during import relieve investigations. Journal of Accounting Research 29 (Autumn): 193-228. Kelley. T., and L. Margheim. 1990. The impact of time budget pressure, personality, and leadership variables on dysfunctional auditor behavior. Auditing: A Journal of Practice and Theory 9 (Spring): 21-41. Knapp, M. 1991. Factors that audit committee members use as surrogates for audit quality. Auditing: A Journal of Practice and Theory 10 (Spring): 35-52. Kothari, S.P., A.J. Leone, and C.E. Wesley. 2005. Performance matched discretionary accrual measures. Journal of Accounting and Economics 39: 163-197. Krishnan. J. and J. Krishnan, 1997. Litigation risk and auditor resignations. The Accounting Review 72(4): 539-560. Knechel, W. R., and J.L. Payne, 2001. Research notes: Additional evidence on audit report lag. Auditing: A Journal of Practice and Theory 20(1): 137-146. Lang, M. and R. Lundholm. 1993. Cross-sectional determinants of analyst ratings of corporate disclosures. Journal of Accounting Research 31 (Autumn): 246-271. Lee, H.Y. and V. Mande, 2003. The effect of the Private Securities Litigation Reform Act of 1995 on accounting discretion of client managers of Big 6 and Non-Big 6 auditors. Auditing: A Journal of Practice and Theory 22 (1): 93-108. Maslach, C. and S.E. Jackson. 1981. The measurement of experienced burnout. Journal of Occupational Behavior 2: 99-113. _____ and _____. 1986. Maslach Burnout Inventory Manual. Palo Alto, CA: Consulting Psychologists Press. Myers, J.N., L.A. Myers, and T.C. Omer, 2003. Exploring the term of the auditor-client relationship and the quality of earnings: A case for mandatory auditor rotation' The Accounting Review 78(3): 779-799. Panel on Audit Effectiveness Report and Recommendations, Public Oversight Board, New York August 31, 2000. Ragunathan, B. 1991. Premature signing-off of audit procedures: An analysis. Accounting Horizons 5 (2): 71-79. Report on the National Commission on Fraudulent Financial Reporting, Committee of Sponsoring Organizations of the Treadway Commission (COSO), October 1987. St. Pierre, K., and J. Anderson. 1984. An analysis of the factors associated with lawsuits against public accountants. The Accounting Review 59 (April): 242-263. Stice, J.D., 1991. Using financial and market information to identify pre-engagement factors associated with lawsuits against auditors. The Accounting Review 66(3): 516-533. Sweeney, A. 1994. The debt covenant violations and managers’ accounting responses. Journal of Accounting and Economics 17 (May): 281-308. Sweeney, J.T. and S.L. Summers, 2002. The effect of the busy season workload on public accountants’ job burnout. Behavioral Research in Accounting 14: 223-245. Willet, C., and M. Page, 1996. A survey of time budget pressure and irregular auditing practices among newly qualified UK Chartered Accountants. British Accounting Review 28: 101-120. United States Congress. 2002. The Sarbanes-Oxley Act of 2002. 107th Congress of the United States of America. H.R. 3763. Washington D.C. United States Securities and Exchange Commission. 2002a. Final Rule: Acceleration of Periodic Report Filing Dates and Disclosure Concerning Website Access to Reports. Release Nos. 33-8128; 34-46464, File No. S7-08-02. United States Securities and Exchange Commission. 2002b. Summary of Comments Relating to Proposed Amendments to Accelerate Periodic Report Filing Dates and Disclosure Concerning Website Access to Reports. Release No. 8089, File No. S7-08-02 (July 1). United States Securities and Exchange Commission. 2004. SEC Proposes to Postpone for One Year the Final Phase-In Period for Acceleration of Periodic Report Filing Dates. SEC News Digest (Issue 2004-165 August 26). Figure 1: Distribution of Fiscal Year-Ends for All Compustat Firms – Years 1951 to 2004 [pic] The results presented above are estimated using a pooled sample of all Compustat companies with fiscal years 1951 to 2004. Table 1: Industry Distribution for Sample Firms | | | |Workload | |Non-workload Compressed Firms | | | | |Compressed Firms | | | |Two-Digit | |Industry |Number of Firm-Year |Percent | |Number of Firm-Year|Percent | |SIC Code | | |Observations | | |Observations | | | | |Agriculture production crops |76 |0.24 | |76 |0.36 | |1 | | | | | | | | |10 |^ |Metal mining |601 |1.91 | |151 |0.71 | |13 |^ |Oil and gas extraction |1756 |5.58 | |608 |2.87 | |14 |^ |Mining and quarrying nonmetallic minerals |118 |0.37 | |48 |0.23 | |15 | |Building construction |54 |0.17 | |32 |0.15 | |16 | |Heavy construction other than building construction |133 |0.42 | |90 |0.42 | |17 | |Plumbing, heat, and air-conditioning |122 |0.39 | |94 |0.44 | |20 |+ |Food and kindred products |669 |2.13 | |766 |3.61 | |21 | |Tobacco products |31 |0.10 | |8 |0.04 | |22 |+ |Textile mill products |228 |0.72 | |216 |1.02 | |23 |+ |Apparel and other finished goods |348 |1.11 | |308 |1.45 | |24 |^ |Lumber and wood products excluding furniture |197 |0.63 | |92 |0.43 | |25 |+ |Furniture and fixtures |183 |0.58 | |209 |0.99 | |26 |^ |Paper and allied products |547 |1.74 | |169 |0.80 | |27 | |Printing, publishing and allied industries |534 |1.70 | |399 |1.88 | |28 |^ |Chemicals and pharmaceuticals |3227 |10.25 | |1502 |7.08 | |29 |^ |Petroleum refining and related industries |275 |0.87 | |30 |0.14 | |30 | |Rubber and miscellaneous plastic products |542 |1.72 | |313 |1.48 | |31 | |Leather and leather products |141 |0.45 | |85 |0.40 | |32 |^ |Stone, clay, glass, and concrete products |336 |1.07 | |126 |0.59 | |33 | |Primary metal industries |653 |2.08 | |406 |1.91 | |34 | |Fabricated metal products |652 |2.07 | |407 |1.92 | |35 |+ |Machinery and computer equipment |2428 |7.72 | |1980 |9.34 | |36 |+ |Electrical and electronic equipment |2487 |7.90 | |2521 |11.89 | |37 | |Transportation equipment |746 |2.37 | |506 |2.39 | |38 |+ |Scientific instruments, photo goods, watches |2362 |7.51 | |2015 |9.50 | |39 | |Miscellaneous manufacturing |484 |1.54 | |284 |1.34 | |50 |+ |Durable goods – wholesale |1004 |3.19 | |779 |3.67 | |51 |+ |Nondurable goods – wholesale |435 |1.38 | |489 |2.31 | |52 |^ |Building materials, hardware, garden - retail |159 |0.51 | |30 |0.14 | |53 |^ |General merchandise stores |383 |1.22 | |65 |0.31 | |54 | |Food stores |284 |0.90 | |203 |0.96 | |55 | |Auto dealers, gas stations |131 |0.42 | |74 |0.35 | |56 |^ |Apparel and accessories stores |469 |1.49 | |92 |0.43 | |57 |+ |Home furniture and equipment stores |187 |0.59 | |210 |0.99 | |58 |+ |Eating and drinking places |618 |1.96 | |528 |2.49 | |59 |^ |Miscellaneous retail |796 |2.53 | |424 |2.00 | |70 |^ |Hotels and lodging services |203 |0.65 | |74 |0.35 | |72 | |Personal services |89 |0.28 | |86 |0.41 | |73 | |Business services |4001 |12.71 | |2720 |12.83 | |75 |+ |Automobile repair and services |49 |0.16 | |57 |0.27 | |78 |+ |Motion pictures |157 |0.50 | |219 |1.03 | |79 | |Amusement and recreation services |415 |1.32 | |273 |1.29 | |80 |^ |Health services |913 |2.90 | |522 |2.46 | |82 | |Educational services |107 |0.34 | |86 |0.41 | |83 | |Social services |83 |0.26 | |49 |0.23 | |87 | |Engineering, research and management services |791 |2.51 | |460 |2.17 | |99 |+ |Non-classifiable establishments |264 |0.84 | |323 |1.52 | | | | | 31,468 |100.00 | |21,204 |100.00 | | | |Total | | | | | | ^ One-tailed test of difference in sample proportions significant at the .001; the proportion of workload compressed firms is higher than the proportion of non-WLC firms for this industry group. + One-tailed test of difference in sample proportions significant at the .001; the proportion of workload compressed firms is lower than the proportion of non-WLC firms for this industry group. Table 2: Descriptive Statistics Panel A – Descriptive Statistics for the Final Sample | |All Companies | | |(n=52,672) | |Variable |Mean |Median |Standard | | | | |Deviation | | | | | | |DA – Raw Value |0.005 |0.005 |0.181 | |DA – Absolute Value |0.113 |0.066 |0.141 | |WLC |0.597 |1.00 |0.490 | |TENURE |5.306 |4.00 |3.780 | |AUD_SIZE |0.835 |1.00 |0.371 | |SIZE |4.424 |4.283 |2.026 | |Total Assets |715.482 |72.441 |2694.65 | |∆NI |-0.570 |-0.014 |56.241 | |∆ASSETS |0.350 |0.077 |1.597 | |LOSS |0.381 |0.00 |0.486 | |ZSCORE |6.722 |3.370 |38.411 | | | | | | | | | | | Panel B – Descriptive Statistics for Workload Compressed Audits and Non-Workload Compressed Audits | |Workload Compressed Audits (n=31,468) |Non-workload Compressed Audits (n=21,204) |Test of Difference |Test of Difference | | | | |in Sub-sample Means |in Sub-sample | | | | | |Proportions | |Variable |Mean |Median |Standard |Mean |Median |Standard | | | | | | |Deviation | | |Deviation | | | | | | | | | | | | | |DA – Raw Values |0.003 |0.003 |0.184 |0.009 |0.008 |0.176 |