The effect of busy season pressure and local office workload compression on audit quality

(1)

Amsterdam Business School

The effect of busy season pressure and local office

workload compression on audit quality

Name: Tjarko Tadema

Student number: 5980844

Date: 16 June 2014

Supervisor: dr. B.J. van Praag

MSc Accountancy & Control, specialization Accountancy Faculty of Economics and Business, University of Amsterdam

(2)

Abstract

This study investigates whether workload pressures, as proxied by the audit busy season (i.e., December fiscal year-end date) and local office location fees affect audit quality. Using a sample of 17,837 firm-year observations during the period 2003 – 2012, the results indicate that busy season companies show slightly lower magnitudes of abnormal discretionary accruals. Additional tests show that these associations are supported by the degree of local office workload compression. Prior experimental and survey research indicates that workload compression lead to dysfunctional behaviour and lower audit quality among individual auditors. The findings suggest that this compression does not negatively influence the audit quality at the local office engagement level. This inconsistency with prior literature could be the results of increased regulation and oversight. Furthermore companies could currently use methods to manage earnings other than accruals-based earnings. These results are important considering they can benefit to the discussion on audit quality.

(3)

Introduction

The European Commission (EC, 2010) issued a Green Paper entitled ‘Audit Policy: Lessons from a Crisis’ regarding the role of the auditor, auditor independence and the structure of the audit market in Europe. To improve audit quality, the EC followed up on the 2010 Green Paper by issuing a set of legislative proposals (COM, 2011) focusing on auditor independence. Within the US financial reporting environment a comparable banning on non-audit services was introduced by the Sarbanes-Oxley act (SOX) and has been effective as from 2002. Within this regulation several non-audit services have been banned to complement the auditor independence and increase audit quality. Over the years a significant amount of research is performed on audit quality in relation to auditor independence. From this research very inconsistent evidence has been obtained on this relation and therefore the influence of auditor independence on audit quality can be questioned. This study will focus on other variables within the audit environment that might affect audit quality. Specifically, it will aim on the complexity of the audit environment and the increasing pressures on the auditors due to more extensive legislation, high expectations from society and pressures on profitability.

The Centre of Audit Quality (CAQ) published a paper (2014), based on the insights from the Public Company Oversights Board (PCAOB) briefing paper presented at the May 15-16, 2013 Standing Advisory Group meeting. The CAQ developed a set of potential Audit Quality Indicators (AQIs) that will provide the greatest opportunity to enhance discussions between auditors and audit committees and the most benefit to audit committees in fulfilling their responsibility relative to the oversight of the audit (CAQ, 2014). One of the four identified AQI’s is defined as: “Engagement team knowledge, Experience and Workload”. Based on prior research it is noted that large offices will have greater engagement team knowledge and experience in detecting material problems in the financial

statements of clients (Francis & Yu, 2009). By implication, auditors in smaller Big 4 offices1 have

less experience and therefore develop less skill in detecting such problems. Sweeney and Summers (2002) note that a period can be identified with higher levels of workload at accounting firms known as the “busy season”, which is typically marked by an increase in stress and potential decrease in auditor performance. During the busy season, audit staff deal with frequent and demanding deadlines, causing conflict between work and other responsibilities, and little time for any leisure activities (Jones, Strand, & Wier, 2010). According to Lee (2007), the increase in capital market activity (e.g. IPOs, mergers and acquisitions, etc) and the economic crisis have caused additional focus on the accounting profession and regulations. Public accountants are expected to have a critical mind-set and

1_{The Big 4 accounting firms are identified as: PWC, KPMG, EY and Deloitte.}

(5)

to develop this they follow mandatory courses like the “Professioneel kritische instelling” prescribed by the NBA (Nederlands Beroepsorganisatie van Accountants, 2014) to meet this expectation. Finally, research has indicated that several factors like auditor burnout and time constraints can reduce audit quality (López & Peters, 2012). Based on the research as described above it is noted that society’s critical view on the audit profession and decreasing profitability due to the global financial crisis has resulted in additional pressure on the auditors. Despite the importance of such effects, there is little archival evidence that documents whether workload pressures ultimately affect the overall quality of an audit. The workload pressure of the busy season result from the need to deliver high quality audits with a large number of engagements within a limited time period. The potential impact of these pressures includes impaired critical mind-set and inappropriate responses to other stressful conditions by the audit team (DeZoort & Lord, 1997). This raises questions if the combination of these conditions will impact the quality of the audit, which increases the opportunities for manipulation of the financial reporting by the audit client.

The research includes 17,837 firm year observations during the period 2003-2012. To proxy for audit quality the abnormal accruals corrected for financial performance model is used (Kothari, Leone, & Wasley, 2005). In the remainder of this paper additional background information and the hypotheses development is presented, followed by a description of the sample description and descriptive statistics. After the empirical results and discussions, the conclusions and limitations are presented.

Background and hypotheses development

Busy Season, Workload compression and audit environment

Most U.S. publicly traded companies close their fiscal year in December, causing a significant increase in workload known to auditors as the busy season. As Figure 1 shows that 67.0 percent of all active Compustat US based companies during 2003–2012 had a fiscal year-end date of December.

In a recently published framework from the International Audit and Assurance Standards Board (IAASB) it is noted that the partners and staff of audit firms can deal with challenging timetables and therefore it is important to perform appropriate planning to ensure adequate resources are available (IAASB, 2014). Also stated in this framework is that audit engagement partners are usually accountable for the financial return on the audits they perform and, if audit fees are restricted by management, this may put pressure on the engagement team to change the nature and timing of audit procedures or reduce testing. These implications could potentially threaten the audit quality. The

(6)

former Public Oversight Board (POB, 2000) recognizes the potential impact of such challenges, arguing that time workload pressures can compromise the audit quality.

Figure 1 0% 10% 20% 30% 40% 50% 60% 70%

The enactment of SOX contributed to these challenges by increasing the testing requirements and reporting responsibilities of auditors. Subsequent to SOX, the Securities and Exchange committee (SEC) adopted rules requiring publicly traded registrants to accelerate the filing of their annual reports (SEC, 2002). A main objective of the rule was to increase the timeliness of accounting information; however, accelerated filing also compressed auditors’ workloads into a shorter busy season. Figure 2 shows that 67.0 percent of all active Compustat US based companies during 2003–2012 file their financial statements in January, February or March.

Figure 2 0% 5% 10% 15% 20% 25% 30% 35%

(7)

From prior research (Francis, Reichelt, & Wang, 2005) it has been argued that it may be more insightful to analyse specific offices of large accounting firms rather than the firm as a whole. This insight is based on the fact that individual audit engagements are managed from a local office and an engagement partner who is typically located in the same city as the client’s headquarters. In addition it is noted that the audits are mainly managed at a local office level given the availability and experience of local resources. The results from an analysis on a local office level can be significantly different from an analysis on firm level. As stated by Francis (2004): “a Big 4 accounting firm is not so big

when we shift to the office level of analysis”. Consider that, while Enron represented less than 2% of

Arthur Andersen’s national revenues from publicly listed clients, it was more than 35% of such

revenues in the Houston office (Francis, 2004). For an additional insight in the local office fees, Table

3 shows the fees allocation of the top ten cities of all active Compustat US based companies within

the sample.

Additional pressure on the auditor’s is imposed through the economic crisis. Based on research performed in Hungary (Pál, 2010) it is noted that auditors lose clients due to the crisis. In Hungary, as in many other countries, the audit obligation is based on a several financial performance indicators (Pál, 2010). As the crisis has caused a decrease in the company’s financial performance, many companies that were previously obligated to audit can now avoid this obligation. As a result, there is a

loss of clients (Pál, 2010). Also SOX imposes higher costs on all public company auditors by

demanding stricter compliance with auditing standards, increasing regulatory oversight, and by raising the penalties for auditor misconduct. From the PCAOB’s annual reports the following statement is obtained: ‘‘The overall objective of the Board’s enforcement program is to promote improvements in the quality of public company auditing by taking remedial and disciplinary measures’’ and ‘‘This remediation process is a cornerstone of the PCAOB’s supervisory model of oversight’’ (PCAOB, 2007 & 2008). The primary objective of the enforcement program appears to help auditors increase their audit quality, however this could lead to relative higher cost and consequently higher pressures for the auditors.

Prior research

Frankel (2002) states that in contrast to the agency literature, which characterizes auditor bias as deliberate, the behavioural literature suggests that psychological heuristics unconsciously lead auditors to bias judgments. Although the cause of auditor bias differs in the agency and behavioural literatures (i.e., intentional distortion vs. cognitive bias), the hypothesized effect is the same; auditors are more likely to acquiesce to client pressure, including pressure to allow earnings management.

(8)

Although earnings management does not have to violate generally accepted accounting principles, firms that manage earnings are viewed as having lower quality earnings (Francis & Yu, 2009). In addition Francis & Yu (2009) argue that aggressive earnings management can result in materially misleading financial reporting.

Audit Quality

A conventional measuring method for audit quality used in prior studies2 is earnings

management using discretionary accruals as a proxy. This is considered an appropriate proxy as it is assumed that auditors should constrain earnings management (Kinney, Jr. Palmrose, & Scholz, 2004). When the auditor identifies material misstatements in the financial statements, the auditor should initially encourage the client to revise the financial statements. Normally the consideration of issuing a qualified opinion only occurs at the final stage of the audit process (Svantstrm, 2013). Evidence from prior research also indicate that clients of Big 4 audited companies have lower abnormal accruals which imply less aggressive earnings management behaviour and therefore higher earnings quality (Francis, 2004). Consistent with these findings, Nelson et al. (2002) report evidence from one Big 4 accounting firm, that auditors detect earnings management attempts (especially income-increasing attempts) and require clients to make appropriate adjustments.

In addition to earnings management behaviour, it has been investigated whether the audit quality in bigger offices is higher compared to smaller offices (Francis & Yu, 2009). This hypothesis is based on the knowledge that local practice offices are the primary decision-making unit within Big 4 auditing firms. The audit engagement is contracted and managed from the local offices and the audit report is issued on the local office letterhead. The engagement teams are typically based in specific practice offices and audit clients in the same geographic region, hence, their expertise and knowledge is both office and client-specific. This decentralized structure reduces information asymmetry and enables Big 4 auditors to develop better knowledge of their clients in a particular location (Francis & Yu, 2009). Also clients have greater knowledge of and confidence in the expertise of locally based personnel that actually perform the audit (Carcello, Hermanson, & McGrath, 1992). These statements assume that Big 4 firms are unable to achieve uniform audit quality across offices, and that a certain amount of client knowledge and overall audit expertise is office-specific (Francis, Reichelt, & Wang, 2005). Based on this, Francis (2009) argues that a large office has more “in house” experience in dealing with public companies and therefore more collective human capital in the local office. In addition Becker (1993) states that experience is an important aspect of human capital, and therefore

2_{Prior studies on audit quality include: Becker et al. (1998); (Francis, 2011) and Choi et al. (2010).}

(9)

concludes that a larger office with more engagement hours will have greater opportunities to acquire expertise in detecting material problems in the financial statements of public companies. In the research performed by Reynolds & Francis (2000) theyreport the first study on office-level results of US companies and auditors. They conclude that auditors treat relatively larger clients in offices more conservatively than smaller clients. This conclusion is based on the result that larger clients in offices have smaller abnormal accruals. Therefore increasing the average size of small audit firms may enhance audit quality. However, this will result in a higher number of clients for the auditor’s causing possible difficulties in serving their clients, due to factors such as human resource constraints. Therefore an increase in the average size of small audit firms might also damage the audit quality. It is considered an empirical question whether an increase in client concentration improves or erodes audit quality (DeFond & Lennox, 2011).

Conform Francis & Yu (2009) the specific office administering an audit engagement, is identified from the letterhead of the audit report filed with the SEC. An office’s aggregate audit fees is used each year to measure office size using all observations for the sample in the Audit Analytics database. Audit fees are directly related to engagement hours, and offices with higher fees will therefore have more hours of experience in the audits of SEC registrants.

Workload pressures

According to Pierce and Sweeney (2005), the audit environment is characterised by constant pressure for cost reduction and an absence of any direct measures of audit quality. Given that audit firms are labour intensive, their main cost is that of employee time. In addition Pierce and Sweeney (2005) found that auditors sometimes respond to workload compression by engaging in behaviours such as reducing sample sizes and prematurely signing-off tests, which could threaten the audit quality. The IAASB (2014) note that the structure of the larger audit firms is hierarchical and the make-up of many engagement teams for individual engagements reflects this structure. As a result, much of the detailed audit work is likely to be performed by staff members who are relatively inexperienced, also considering they have not completed their CPA exams. The pressure for starting accountants to pass their Certified Public Accountant (CPA) exams will cause additional conflict between work, family responsibilities and leisure activities (Carpenter & Hock, 2008). This also might

result in additional interpersonal stress (Figler, 1980). It is noted that employees can experience “job

burnout” when they encounter extended periods of excessive stress with little time for family or leisure activities. In addition the results of the research indicate that stress, mediated by job burnout and the effects on physiological well-being, have a negative impact on the job performance of the

(10)

audit staff (Jones, Strand, & Wier, 2010). Furthermore research has reported consistently high levels of quality threatening behaviour (QTB) by audit staff. QTB refers to specified behaviours which have the potential to reduce audit quality. DeZoort and Lord (1997) suggested that time deadline pressure may also be relevant in audit firms, describing time deadline pressure as acute time. Acute time pressure has been shown to be more harmful to an individual’s well-being than chronic time pressure (Pierce & Sweeney, 2004). Ultimately a consistent thread in previous studies is the significance of time pressure in explaining dysfunctional behaviour (Pierce & Sweeney, 2006).

Within the audit environment several controls on auditor behaviour have been implemented and include rules and regulations over the audit process and the audit supervision (Pierce & Sweeney, 2005). Subjective measures as intuition and judgment maintain part of the audit process and even from an experienced engagement partner these measures maintain difficult for the audit firm to control (Power, 2003). A control identified in the audit environment to ensure constant levels of quality, can be identified as supervision within the engagement team. The current structure of the audit teams comprises several functional levels to ensure sufficient knowledge and supervision is available on the audit engagements. The teams mainly consist of junior staff performing a significant part of the audit procedures and are supervised by senior staff on-site. However Pierce & Sweeney (2004) show that this reporting structure is not always present, where juniors report directly to managers that are largely removed from the on-site fieldwork and thus are not able to provide the required supervision. This can result in insufficient guidance and therefore is often limited as a form of control. Also due to the considerable time pressure seniors deal with in finishing their own work, little time may be available to supervise the juniors. This effect is often known as the senior-squeeze. Pierce & Sweeney (2004) report evidence of failure by audit seniors to report detected QTB at the junior level. These results further highlight the limitations of this form of control.

Changes in the audit environment arising from the implementation of SOX, which increases pressure on audit firms to maintain greater audit file documentation, may have impacted on the firms’ ability to detect the behaviours and underline the importance of continued research as the audit

environment changes (Pierce & Sweeney, 2006). According to Barrett (2005) there is more risk of

litigation for audit firms arising from failure of audit processes than the concern with documents obtained from the client. As a response to the busy season and increasing regulation and stricter audit standards, auditors respond by shifting their procedures to earlier interim periods and encourage their clients to implement systems that will be able to provide information and comfort over their systems and processes during the year. However, certain procedures cannot be performed until the end of the fiscal year period or shortly thereafter. In addition, some auditing standards emphasize the importance

(11)

of year-end evidence as a means to reduce audit risk (López & Peters, 2012). Therefore the busy season and workload compression remain important challenges for audit firms to ensure a high quality audit can be performed. Despite the apparent impact of these challenges, there is little archival evidence that show if workload compression ultimately affects the overall quality of an audit. In response, it is investigated whether the busy season, as well as the relative concentration of companies with the same fiscal year-end date in an auditor’s client portfolio (i.e. workload compression), is associated with indicators of lower audit quality.

Hypotheses

Based on the literature as described it is expected that workload pressures, negatively affect audit quality. For the research we measure workload pressures in two ways. In the first variation

BUSY is used as an indicator for the December year-end companies. The expectation is that BUSY will

have a positive coefficient, as this would indicate that the financial statements of December year-end companies carry higher levels of abnormal accruals and therefore lower levels of audit quality. The second variation is represented by OFFICE_WLP, a proxy for the level of workload compression of a local auditor office during the fiscal year-end month of a client. Higher values of OFFICE_WLP should be associated with higher concentrations in auditor workloads, considering the fees reflect the amount of audit effort expended on specific audit engagements (Akono, Hranaiova, Vinelli, & Stein, 2011). OFFICE_WLP should display a positive coefficient, such that audit quality decreases as the level of workload compression increases.

Sample selection and research methods

Sample

For the sample a cross-sectional sample of company-year observations from 2003-2012 is used. This period followed the implementation of the Sarbanes–Oxley Act of 2002. This period is characterized by increasing regulatory focus on auditing practices and the issuance of several new auditing standards. Following DeBoskey and Jiang (2012), the study concentrates on the post-SOX period for two reasons. First, recent research found evidence of more conservative management and auditor behaviour due to increased regulatory, investor, and media scrutiny and heightened legal liability. Therefore, including data from two distinct sample periods could potentially confound the analysis of the effect of auditor specialization on the financial reporting quality. Second, by focusing

(12)

on a more recent period, the findings from the research can be more useful to interested parties, such as regulators and investors.

The data is extracted from Compustat or Audit Analytics and the companies missing the necessary data are excluded to operationalize the regression model. In line with Lopez (2012) the financial institutions, utility companies and other highly regulated industries are excluded from the sample (standard industrial classification (SIC); 6000-6999 and 4000-4999), considering they face different regulatory and reporting requirements. To control for the dominant role and greater resource base of Big 4 firms, the sample is limited to companies audited by the Big 4 firms. Furthermore big four firms are organized as national partnerships that set firm-wide policies and provide technical support for their city based practice offices and therefore are better to compare (Francis & Yu, 2009). Office location and fee data reporting are collected from Audit Analytics; all other data come from Compustat. After merging the Audit Analytics sample with Compustat, the final sample consists of 17,837 firm-year observations for fiscal years 2003 through 2012, comprised of 2,606 different companies. The 17,837 firm-years in the sample are audited by 305 unique Big 4 offices, which are distributed as follows: EY (86), KPMG (78), Deloitte (71), and PWC (70). For the full period 2003-2012, the smallest office of each Big 4 firm has a single SEC registrant and less than $100 thousand in audit fees. The largest office of each Big 4 firm has over 229 clients and audit fees in excess of $145 million. The median number of clients per office ranges from 25 for KPMG to 39 for EY, and median fees per office range from $35 million for KPMG to $79 million for PWC.

Audit quality

Following Francis & Yu (2009), the ordinary least-squares (OLS) regression is used to

estimate the following performance-adjusted Jones model3 (1991) for the full Compustat sample by

fiscal-year and controlling for financial performance by adding net income: Equation 1 TA Ai,t−1= a0 + a1� 1 Ai,t−1� + a2� ∆REVi,t Ai,t−1 � + a3� PPEi,t Ai,t−1� + a4� NIi,t Ai,t−1� + ϵ

The error term in Equation 1 equals the unexplained level of accruals or the ‘discretionary accruals’. These discretionary accruals function as the proxy for accrual earnings management. In addition, ∆REV is the change in revenues and trade receivables; PPE is gross property, plant, and equipment;

3_{We use the cross-sectional version of the Jones (1991) model based on the primary SIC code of the companies as}

categorized in Appendix A

(13)

NI is the operating income after depreciation; i indicates the company; and t refers to the fiscal year. All variables are deflated by lagged total assets.

The next step is the calculation of the total accruals. An estimation method from prior literature is used (López & Peters, 2012). This method uses balance sheet accounts like current assets and liabilities to calculate the total accruals:

Equation 2

TA = (∆CAi,t− ∆CLi,t− Depi,t)/ Ai,t−1

TA is defined as total accruals, estimated as the change in non-cash current assets minus the change in current liabilities excluding the current portion of long term debt, minus depreciation and amortization, scaled by lagged total assets (López & Peters, 2012). It is tested if management earnings behaviour differ across the busy season and non-busy season companies. Specifically, the expectation is that the companies with a year ending during the busy season period will evidence higher earnings

management behaviour. In Table 1 the total accruals are presented for the companies within the sample

separated in the busy season and non-busy season companies. The average total accruals for the busy season companies are higher compared to the non-busy season companies. These results could provide an indication that the companies with a period ending during the busy season evidence higher earnings management behaviour (higher abnormal accruals) compared to the non-busy season companies.

Table 1 Total accruals

(December Year-End) (Non-December Year-End) Year T ot al accruals T ot al accruals

2003 0,0460 -0,0058 2004 0,0454 0,0199 2005 0,0656 -0,0181 2006 0,0446 -0,0074 2007 0,0248 -0,0231 2008 -0,0722 -0,0353 2009 0,0232 -0,0035 2010 0,0522 -0,0020 2011 0,0326 -0,0108 2012 0,0272 -0,0113 Average 0,0289 -0,0098

Consistent with prior literature4 the following dependent variable of interest is included in the

model: the absolute value of the abnormal component of a company’s discretionary accruals (ABS_DA). The variable is calculated as the absolute value of the abnormal component of a

4_{Discretionary accruals are commonly used to proxy for audit quality e.g. Becker et al. (1998), Choi et al. (2010), (Francis}

& Yu, 2009), (Francis, 2011), (López & Peters, 2012).

(14)

company’s total accruals, with a minimum value of 0.000 and a maximum winsorized value of 0.999 (Francis and Yu 2009).

Workload pressures

Two separate proxies for workload pressures are used. The first (BUSY) equals 1 if a company has a fiscal year-end date of December and 0 otherwise. The second is represented by OFFICE_WLC and is calculated through the audit fees charged to clients with the same fiscal year-end month in each local office divided by the sum of the total audit fees collected by the local office for the year.

Regression model development

The regression model intends to capture the effects of the busy season and auditor workload compression on audit quality. The following model is estimated using ordinary least-squares regression:

𝐴𝐵𝑆_𝐷𝐴 = 𝑎0+ 𝑎1(𝐵𝑈𝑆𝑌) + 𝑎2(𝑂𝐹𝐹𝐼𝐶𝐸_𝑊𝐿𝐶) + 𝑎3(𝑆𝑖𝑧𝑒) + 𝑎4(𝐷𝑅𝐸𝑉𝐸𝑁𝑈𝐸𝑆) + 𝑎5(𝐶𝐹𝑂)

+ 𝑎6(𝐿𝑂𝑆𝑆) + 𝑎7(𝐷𝐸𝐵𝑇) + 𝑎8(𝐺𝐶) + 𝑎9(𝐸𝐶) + 𝑎10(𝑇𝐸𝑁𝑈𝑅𝐸)

+ 𝑎11(𝐼𝑁𝐹𝐿𝑈𝐸𝑁𝐶𝐸) + 𝑎12(𝑌𝐸𝐴𝑅𝑡) + 𝑎13(𝑆𝐼𝐶)

(15)

Table 2 Variable Definitions

Variable Definition

ABS_DA = absolute value of abnormal accruals derived f rom the perf ormance adjusted accruals model in Equation 1, w ith a minimum value of 0.000 and a maximum w insorized value of 0.999;

BUSY = 1 if a company has a f iscal year-end date of December, and 0 otherw ise;

OFFICE_WLC =

relative level of w orkload compression of a local auditor of f ice during the f iscal yearend month of a client. For each month, w e add the audit f ees charged to clients w ith the same f iscal year-end month in each local of f ice; w e then divide each monthly sum by the total audit f ees collected by the local of f ice f or the year.

SIZE = natural log of a company’s total assets (in millions);

DREVENUES = percentage of change in a company’s sales revenue, w ith a minimum w insorized value of 1.00 and a maximum w insorized value of 2.00;

CFO = operating cash f low s def lated by lagged total assets;

LOSS = dummy variable that takes the value of 1 if operating income af ter depreciation is negative, and 0 otherw ise;

DEBT = company's total liabilities def lated by total assets;

GC = dummy variable that takes the value of 1 if a f irm receives a going-concern report in aspecif ic f iscal year, and 0 otherw ise;

EC = dummy variable that takes the value of 1 if the f iscal year is 2008 to 2012, and 0 otherw ise.

OFFICE_SIZE = natural log of the aggregated audit f ees (in millions) of a local of f ice

TENURE = dummy variable that takes the value of 1 if auditor tenure is three years or less, and 0 otherw ise; INFLUENCE = ratio of a company’s total f ees (i.e., audit f ees plus nonaudit f ees) relative to the aggregate annual f ees

generated by the local of f ice that audits the company;

Other

YEAR = f iscal year indicators; and

SIC = industry indicators, based on the SIC code classif ication of a company as included in Appendix A.

Variables of Interest

Company-Related

Auditor-Related

Company-related variables

Company-related variables are included in the model to capture the potential effects of managers’ different opportunities or incentives to manipulate earnings. From prior research it is noted that larger companies are more likely to have higher earnings quality (Becker, DeFond, Jiambalvo, & Subramanyam, 1998). Therefore the expectation is that client size (SIZE), measured as log of total assets (in $ millions), will have a negative association with discretionary accruals. In addition larger companies have fewer incentives to manage earnings to avoid litigation (Lang & Lundholm, 1993). According to Menon and Williams (2004), DREVENUES is positively associated with discretionary accruals. This variable controls for companies opportunities to manage accruals during periods of high growth. It is measured as the percentage of change in a company’s sales revenue and has a minimum winsorized value of -1.00 and a maximum winsorized value of 2.00 (Francis and Yu 2009). Operating cash flows (CFO) influence the discretionary accruals (Dechow, Sloan, & Sweeney, 1995). The variable is measured by operating cash flows deflated by lagged total assets and the expectation is that higher operating cash flows are associated with lower discretionary accruals.

(16)

Three variables are included in the model to control for the effects of debt and financial performance: LOSS, DEBT and EC. Companies with negative operating income (LOSS) have lower incentives to manage discretionary accruals and therefore control for audit quality differences between loss and profit companies (Choi, Qiu, Zang, & Kim, 2007), as well as for the incentive to engage in ‘‘big bath’’ accounting during years of poor financial performance. It is expected that LOSS will have a positive association with discretionary accruals and it is measured as an indicator that equals 1 if operating income after depreciation is negative and 0 otherwise. Sweeney (1994) argues that companies with higher debts (DEBT) have more incentives to manage discretionary accruals to comply with possible debt covenant agreements. Therefore the debt/asset ratio is added as a variable, to control for the amount of leverage. A positive association with discretionary accruals is expected and the variable is measured as the total liabilities deflated by total assets. In relation with debt and financial performance the variable (EC) is included, being a company’s fiscal year during the

economic crisis from 2007 to 2009 (Bloom, 2011). This global economic crisis is considered as the

most important since the Great Depression in 1929-33, and one of the most important characteristics of this crisis is the growing tendency of insolvency (Pál, 2010). In his research Pál (2010) examined the development of bankruptcy as well as the insolvency and liquidation procedures. He found that the insolvency index of companies increased by 36 percent in the first nine months of 2009 compared to the same period in 2008. Therefore it is expected that management will have more incentives to manage discretionary accruals to improve their solvency as a result of this crisis. EC (-) is a dichotomous variable that takes the value of 1 if the related year (t) falls during the global economic crisis from 2007 to 2009, and 0 otherwise. All variables discussed above relate to the companies or management incentives. In the following section the auditor’s incentives are considered.

Auditor-Related Factors

Reynold and Francis (2000) argue that auditors may report favourably in order to preserve influential clients, especially if the client is relative large to the size of the local engagement office. DeAngelo (1981) describes this proposition as economic bounding. Where you would expect that economic bonding implies the impairment of auditor independence, Reynolds and Francis (2000) show evidence of Big 4 auditors that report more conservatively for larger influential clients in the local engagement offices. Reynolds and Francis (2000) explain this through the auditor’s incentive to avoid costly litigation from inaccurate reporting by these clients. Following Francis and Yu (2009)

(17)

influential clients. INFLUENCE is measured as the ratio of the client’s fees for all services, to the sum of fees for all clients of the engagement office for a given year.

Two other important auditor characteristics are included in the model to ensure the results are not the consequence of correlated omitted auditor variables. First the model is controlled, as short

audit tenure is associated with lower earnings quality (Johnson, Khurana, & Reynolds, 2002). Auditor

tenure research examines if the length of the auditor-client relationship affects audit quality, and is motivated in part by calls for mandatory auditor rotation. The argument for rotation is that auditors can become captive to clients in long tenure situations. The counter-argument is that auditors have strong economic incentives to maintain their independence and internal mechanisms such as the rotation of engagement personnel are sufficient to maintain the skepticism and independence of auditors (Francis, 2004). In addition Meyers et al. (2003) find evidence that longer tenures enable auditors to place greater constraints on aggressive financial reporting. To ensure that the results are not confounded by systematic differences in auditor tenure across practice offices (Francis & Yu, 2009), the variable TENURE is included, which is coded 1 if tenure is three years or less, and 0 otherwise. Second, the local auditor’s OFFICE_SIZE is controlled. The coefficient on OFFICE_SIZE is expected to be negative if auditors in larger offices allow their clients less discretion over the use of

accruals to manage earnings (Francis & Yu, 2009). OFFICE_SIZE is measured by the natural log of

the aggregated audit fees for a local office. The auditor-related controls aim to capture auditors’ ability

to curtail aggressive reporting and perform an effective audit.

Other Factors

The regression model includes a set of fiscal year indicators (YEAR) to control for the possibility of temporal differences in the reporting environment of companies and their auditors. This variable is measure by the fiscal year-end of the companies. The model also includes a set of industry indicators (SIC) to control for potential industry-specific factors that could affect accrual reporting. This indicator is based on the primary companies SIC code categorized in Appendix A.

Descriptive statistics

Table 4 contains descriptive statistics. The mean value of BUSY in the first set of columns (n = 17,837) indicates that 68.5 percent of the observations come from companies with a fiscal year-end date of December. The mean value of OFFICE_WLC in the second set of columns (n = 12,220) shows that the average level of local office workload compression during December is 78.3 percent, and the third set of columns (n = 5,617) reveals that the average level of local office workload compression

(18)

during any other month is only 9.9 percent. Based on these results almost four-fifth of the local office workloads relate to their busy season clients. From an analysis on the abnormal accrual variable (ABS_DA) in Table 4, it is noted that busy season companies have slightly lower levels of abnormal accruals compared to the non-busy season companies (ABS_DA: 0.331 versus 0.338). In contrast with the hypotheses, these results offer preliminary evidence of higher audit quality among busy season companies. Furthermore the test of difference for ABS_DA is not considered significant (two tailed: 0,130). Based on the other variables descriptive analytics it is noted that busy season companies report higher changes in their revenues (DREVENUES), have lower net operating cash flows (CFO), are more likely to report a loss (LOSS), have lower levels of debt (DEBT), have their audits performed by larger auditor offices (OFFICE_SIZE) and more companies where subject to the economic crisis (EC). These differences are statistically significant (all p-values ≤ 0.01).

The Pearson correlation coefficients in Table 5 indicate that the highest correlation value is for

BUSY and OFFICE_WLC at 88.9 percent. This is not an unexpected condition considering 78.3 percent of the observations have a fiscal year-end date of December (see Table 4). This aligns busy season companies with auditors with higher levels of workload compression. The correlations between ABS_DA and the two independent variables of interest, BUSY and OFFICE_WLC, are negative. These results do not support the research propositions. The relation ABS_DA and BUSY however is not significant according to the t-test for equality of means.

Empirical results

In Table 6 the results from the OLS regression have been presented. All models are significant (p-value ≤ 0.01) and have adjusted R² values ranging between 7.41 and 10.56 percent. The estimated coefficient for BUSY in Model 1 is negative and significant (Coeff. = -0.009, p-value ≤ 0.10), indicating that busy season companies report lower abnormal accruals than non-busy season companies. This is inconsistent with the expectation of lower audit quality among busy season companies. The estimated coefficient for OFFICE_WLC in Model 2 is also negative and significant (Coeff. = -0.017, p-value ≤ 0.01), indicating that abnormal accruals decrease with the concentration of companies with the same fiscal year-end month in an auditor’s client portfolio. Again, this is in contrast with the expectation of a negative association between local office workload compression and audit quality.

It must be mentioned that there is a potential multicollinearity issue between BUSY and

(19)

season versus non-busy season companies, following the approach of prior studies (López & Peters, 2012). The estimated coefficient for OFFICE_WLC in Model 3 remains negative and significant

(Coeff. = -0.041, p-value ≤ 0.05). Again this is inconsistent with the view that audit quality decreases

as local office workload compression increases. Model 4 does not support an association between workload compression and audit quality for non-busy season clients. However, the lack of significance in this model may be the result of low average levels of local office workload compression among non-busy season clients, as presented by OFFICE_WLC in Table 4. To summarize the results from Table 6, the magnitude of abnormal accruals reported by busy season companies decreases with the level of auditor workload compression. The control variables: SIZE,

DREVENUES, CFO, DEBT and INFLUENCE all show coefficients in the expected directions, with

(20)

Table 3 – Annual top ten city audit fees

Fi scal Ye ar Ne w York C h i cago S an Jose Hou ston Boston Ph i l ade l ph i a Dal l as Los An ge l e s S an Fran ci sco Atl an ta

2003 296.575.659 127.016.908 100.502.332 83.204.427 72.165.579 80.527.555 67.152.340 59.649.807 51.570.986 43.620.410 2004 491.960.270 213.331.645 153.940.884 169.967.122 133.371.212 152.476.230 131.945.887 103.231.422 79.269.586 80.275.811 2005 573.915.345 263.766.907 211.519.048 194.994.405 168.743.754 159.107.519 134.207.529 119.285.070 127.234.792 107.156.810 2006 572.312.429 303.691.513 262.394.462 274.357.024 196.371.003 159.548.934 183.187.197 123.838.593 120.267.220 115.516.643 2007 502.610.296 288.444.792 285.011.373 252.697.376 221.913.543 158.089.203 153.969.252 129.505.203 134.545.075 116.525.011 2008 508.162.667 292.369.921 291.797.544 268.294.673 235.189.326 167.103.352 154.755.950 164.273.161 127.636.973 124.940.800 2009 468.063.534 301.594.896 267.521.881 259.831.342 218.677.510 154.835.681 150.641.659 134.044.374 114.704.183 121.395.234 2010 474.239.956 312.657.998 274.164.762 280.057.050 217.544.196 151.393.342 152.074.559 139.439.900 124.553.116 144.068.694 2011 493.906.636 320.047.423 292.233.945 306.020.073 238.518.433 164.115.732 154.375.615 130.731.875 134.608.927 141.667.228 2012 563.473.910 325.435.973 332.135.686 334.647.294 251.507.261 166.977.816 161.973.691 134.468.646 138.936.028 144.835.672 Gran d Total 4.945.220.702 2.748.357.976 2.471.221.917 2.424.070.786 1.954.001.817 1.514.175.364 1.444.283.679 1.238.468.051 1.153.326.886 1.140.002.313

Table 4 Descriptive statistics

Variable M ean M edian SD M ean M edian SD M ean M edian SD t

ABS_DA 0,333 0,234 0,002 0,331 0,231 0,003 0,338 0,242 0,004 -1,51 BUSY 0,685 1,000 0,003 1,000 1,000 0,000 0,000 0,000 0,000 ∞ OFFICE_WLC 0,568 0,710 0,003 0,783 0,821 0,001 0,099 0,041 0,002 259,68*** SIZE 2,938 2,924 0,006 2,948 2,926 0,007 2,916 2,921 0,011 2,44** Δ REVENUES 1,115 1,084 0,002 1,128 1,090 0,003 1,089 1,072 0,003 8,51*** CFO 0,073 0,096 0,002 0,064 0,093 0,003 0,092 0,104 0,002 -6,63*** LOSS 0,051 0,000 0,002 0,065 0,000 0,002 0,020 0,000 0,002 12,65*** DEBT 0,231 0,199 0,001 0,223 0,188 0,002 0,247 0,223 0,002 -7,70*** EC 0,295 0,000 0,003 0,299 0,000 0,004 0,288 0,000 0,006 1,49 TENURE 0,783 1,000 0,003 0,787 1,000 0,004 0,773 1,000 0,006 2,14** INFLUENCE 0,042 0,011 0,001 0,041 0,011 0,001 0,043 0,012 0,001 -0,84 OFFICE_SIZE 8,009 8,076 0,004 8,036 8,092 0,005 7,949 8,010 0,008 9,25***

*, **, *** Denote significance at the 0.10, 0.05, and 0.01 levels (two-tailed) resp ectively ª Rep orted test scores of difference in samp le means.

Test of Differenceª

n = 17837 n = 12220 n = 5617

Busy Season Comp anies Non-Busy Season Comp anies All Observations (December Year-End) (Non-December Year-End)

(21)

Table 5 – Correlation table

SIZE DEBT CFO DREVENUES T ENURE LOSS OFFICE_WLC INFLUENCE EC ABS_DA BUSY OFFICE_SIZE Pearson Correlat ion 1 -,111** ,284** -,017* ,131** -,284** ,086** ,227** ,045** -,247** ,018* ,164** Sig. (2-t ailed) ,000 ,000 ,021 ,000 ,000 ,000 ,000 ,000 ,000 ,015 ,000 Pearson Correlat ion -,111** 1 -,090** -,033** -,044** ,014 -,063** ,007 ,008 ,139** -,058** ,041** Sig. (2-t ailed) ,000 ,000 ,000 ,000 ,062 ,000 ,360 ,273 ,000 ,000 ,000 Pearson Correlat ion ,284** _-,090** ₁ _,106** _,069** _-,424** _-,031** _,035** _,015 _-,070** _-,050** _,018*

Sig. (2-t ailed) ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,052 ,000 ,000 ,014 Pearson Correlat ion -,017* -,033** ,106** 1 ,059** -,031** ,056** -,020** -,130** ,116** ,064** ,002 Sig. (2-t ailed) ,021 ,000 ,000 ,000 ,000 ,000 ,006 ,000 ,000 ,000 ,782 Pearson Correlat ion ,131** -,044** ,069** ,059** 1 -,023** ,023** ,036** ,250** -,045** ,016* -,033** Sig. (2-t ailed) ,000 ,000 ,000 ,000 ,002 ,002 ,000 ,000 ,000 ,032 ,000 Pearson Correlat ion -,284** _,014 _-,424** _-,031** _-,023** ₁ _,070** _-,059** _,011 _,049** _,094** _-,005

Sig. (2-t ailed) ,000 ,062 ,000 ,000 ,002 ,000 ,000 ,151 ,000 ,000 ,472 Pearson Correlat ion ,086** -,063** -,031** ,056** ,023** ,070** 1 ,120** -,006 -,032** ,889** ,025** Sig. (2-t ailed) ,000 ,000 ,000 ,000 ,002 ,000 ,000 ,438 ,000 ,000 ,001 Pearson Correlat ion ,227** _,007 _,035** _-,020** _,036** _-,059** _,120** ₁ _-,006 _-,060** _-,006 _-,507**

Sig. (2-t ailed) ,000 ,360 ,000 ,006 ,000 ,000 ,000 ,413 ,000 ,403 ,000 Pearson Correlat ion ,045** _,008 _,015 _-,130** _,250** _,011 _-,006 _-,006 ₁ _,003 _,011 _,070**

Sig. (2-t ailed) ,000 ,273 ,052 ,000 ,000 ,151 ,438 ,413 ,737 ,135 ,000 Pearson Correlat ion -,247** ,139** -,070** ,116** -,045** ,049** -,032** -,060** ,003 1 -,011 -,019** Sig. (2-t ailed) ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,000 ,737 ,130 ,010 Pearson Correlat ion ,018* _-,058** _-,050** _,064** _,016* _,094** _,889** _-,006 _,011 _-,011 ₁ _,069**

Sig. (2-t ailed) ,015 ,000 ,000 ,000 ,032 ,000 ,000 ,403 ,135 ,130 ,000 Pearson Correlat ion ,164** _,041** _,018* _,002 _-,033** _-,005 _,025** _-,507** _,070** _-,019** _,069** ₁

Sig. (2-t ailed) ,000 ,000 ,014 ,782 ,000 ,472 ,001 ,000 ,000 ,010 ,000 BUSY

OFFICE_SIZE

**. Correlation is significant at the 0.01 level (2-tailed).*. Correlation is significant at the 0.05 level (2-tailed). T ENURE LOSS OFFICE_WLC INFLUENCE EC ABS_DA DREVENUES

C orre l ati ons

SIZE DEBT CFO

(22)

Extended Analysis: Going-Concern Opinions

The primary tests show that local office workload compression does not lead to a decrease in higher audit quality in the form of decreased financial reporting quality. Thus it appears that busy accountants allow less discretionary accruals in the annual reporting. These results are based on ABS_DA that show low significance in the primary model, and are inconsistent with the hypotheses. To support these results further study is performed. Based on prior research (López & Peters, 2012) it is noted that it is important to consider whether workload pressures are also associated with extreme audit opinion outcomes, such as going-concern decisions. If there is more pressure on the auditor, they are less likely to be able to identify going-concern problems and issue going-concern reports. Hence, it is predicted that workload pressures are negatively associated with the probability of issuing going-concern opinions. The tests are extended to investigate the potential association between workload compression and the likelihood of issuing a going-concern opinion by estimating the following logistic regression model:

𝐺𝐶 = 𝑎0+ 𝑎1(𝑂𝐹𝐹𝐼𝐶𝐸_𝑊𝐿𝐶) + 𝑎2(𝑆𝑖𝑧𝑒) + 𝑎3(𝐷𝑅𝐸𝑉𝐸𝑁𝑈𝐸𝑆) + 𝑎4(𝐶𝐹𝑂) + 𝑎5(𝐿𝑂𝑆𝑆)

+ 𝑎6(𝐷𝐸𝐵𝑇) + 𝑎7(𝐸𝐶) + 𝑎8(𝑇𝐸𝑁𝑈𝑅𝐸) + 𝑎9(𝐼𝑁𝐹𝐿𝑈𝐸𝑁𝐶𝐸)

+ 𝑎10(𝑂𝐹𝐹𝐼𝐶𝐸_𝑆𝐼𝑍𝐸) + 𝑎11(𝑌𝐸𝐴𝑅𝑡) + 𝑎12(𝑆𝐼𝐶)

A negative relation is expected of GC with OFFICE_WLC, where GC is an indicator that equals 1 if a company receives a going-concern opinion, and 0 otherwise. The going concern firm-year data for the sample is extracted from Audit Analytics using the Company Identification Code (CIK). Consistent with the primary model, control variables are included for the financial health of the company and auditor characteristics as defined in Table 2. In line with the earlier tests, the model is estimated using subsamples of busy season and non-busy season companies. In order to identify companies with a higher risk for a going-concern opinion, the sample will partition financially distressed companies (i.e., LOSS = 1). The logistic regression results in Table 7 are significant as a whole (p-value ≤ 0.01 for all models) and have pseudo R² values ranging between 11.75 and 17.78 percent. The estimated coefficient for OFFICE_WLC in Model 1 is positive and significant (Coeff. = 0.021, p-value ≤ 0.05). Also for Model 2 this coefficient is positive and significant (Coeff. = 0.194, p-value ≤ 0.01). The results of the extended analysis is consistent with the prior results and show a higher likelihood of a going concern opinion among December year-end companies whose audits are performed by auditors with greater levels of local office workload compression. The model suggests that work load compression increases the probability of a GC qualification. It is noted that the estimated coefficient

(23)

for OFFICE_WLC in Model 4 is negative, but not statistically significant. These results support the main tests and imply that workload compression positively affects audit quality.

Sensitivity Tests and Additional Robustness Checks

Abnormal Accrual Definitions

For the primary results it is investigated whether these are sensitive to different variations of the discretionary accruals model. The regressions in Table 6 are estimated using the ‘‘raw’’ value of discretionary accruals (n = 17,837). The estimated coefficient for OFFICE_WLC is -0.067 which is in line with the primary results, however the results are insignificant. In addition the cash flows approach is used to estimate the Jones (1991) model (i.e., TA equals earnings before interest and taxes minus operating cash flows) (n = 17,837). The estimated coefficient for OFFICE_WLC is -0.452 with a p-value below 0.01. This result is in line with the primary results.

Audit Fees and Workload Compression Proxy

By using audit fees the audit related elements as audit production costs and efforts are captured. However using audit fees could also capture other elements, such as audit fee premiums or competitive market pressures (López & Peters, 2012). Through using a sample of audits performed by the Big 4 firms the results are partially controlled for these elements. In addition the sensitivity of the results are empirically assessed by using other weighting factors for estimating the OFFICE_WLC variable, as total fees (i.e., audit fees plus non-audit fees), net sales audited, and total assets audited. These weighting factors have distinct design trade-offs, similar to audit fees. For example, companies operating in capital-intensive industries could be over- or under-represented in an estimation of the regression model with OFFICE_WLC based on total assets (López & Peters, 2012). Also companies with significant off-balance sheet assets could have the same issue. The results based on the other weighting factors however are insignificant except for the total assets audited. The estimated regression coefficients for total OFFICE_WLC based on total assets audited is 0.000 with a p-value below 0.01.

(24)

Table 6 Ordinary Least-Squares Regression of Abnormal Accruals

Pred

Variable Sign β t-stat β t-stat β t-stat β t-stat

Int ercept 0,718 0,42 0,806 0,47 -1,529 -0,74 6,614 2,16** BUSY + -0,009 -1,94* OFFICE_WLC + -0,017 -2,78*** -0,041 -2,43** 0,003 0,12 SIZE - -0,090 -28,55*** -0,089 -28,45*** -0,097 -25,12*** -0,072 -12,81*** Δ REVENUES + 0,130 17,07*** 0,131 17,12*** 0,125 14,8*** 0,147 8,16*** CFO - -0,010 -1,12 -0,010 -1,13 -0,014 -1,45 0,037 1,38 LOSS + -0,046 -4,18*** -0,046 -4,15*** -0,058 -4,92*** 0,021 0,66 DEBT + 0,186 16,40*** 0,186 16,35*** 0,175 13,84*** 0,220 8,48*** EC - 0,023 4,56*** 0,022 4,52*** 0,022 3,69*** 0,024 2,6*** T ENURE + -0,019 -3,04*** -0,019 -3,05*** -0,008 -1,12 -0,042 -3,96*** INFLUENCE + 0,002 0,1 0,010 0,41 0,031 1,04 -0,039 -0,75 OFFICE_SIZE - 0,007 1,58 0,008 1,68* 0,011 1,91* 0,003 0,33 YEAR SICC n = Adjusted R2 F-Statistic

*, **, *** Denot e significance at t he 0.10, 0.05, and 0.01 levels, respect ively. T he variable definit ions are in T able 2

(<0.001) (<0.001) (<0.001) (<0.001)

156,39 156,75 121,21 38,43

9,46% 9,49% 10,56% 7,41%

17837 17837 12220 5617

(included) (included) (included) (included) (included) (included) (included) (included)

All Obs. All Obs. Busy Season Non-Busy Season Model 1 Model 2 Model 3 Model 4

(25)

Table 7 Logistic Regression of Going-Concern Tests

Variable β t-stat β t-stat β t-stat β t-stat

Intercept 4,319 4,22*** 0,330 0,04 2,269 1,90* -47,539 -1,84* OFFICE_WLC 0,021 2,48** 0,194 2,64*** 0,004 0,37 -0,153 -1,40 SIZE -0,011 -5,72*** -0,101 -5,56*** -0,013 -5,72*** -0,116 -1,50 Δ REVENUES -0,012 -2,79*** -0,013 -0,86 -0,012 -1,64* 0,004 0,08 CFO -0,100 -21,18*** -0,047 -3,49*** -0,136 -13,02*** -0,026 -0,45 LOSS 0,019 3,21*** -0,001 -0,06 DEBT 0,167 26,65*** 0,103 5,22*** 0,148 14,64*** 0,172 1,60 EC 0,020 6,67*** 0,069 3,00*** 0,004 1,04 0,008 0,11 TENURE -0,036 -9,8*** -0,090 -3,12*** -0,016 -3,86*** -0,085 -0,96 OFFICE_SIZE 0,023 1,59*** 0,674 3,87*** -0,020 -1,00 -0,415 -0,50 INFLUENCE 0,008 2,84 0,076 3,60*** -0,004 -1,39 -0,112 -1,63 YEAR SICC n = Adjusted R2 F-Statistic

*, **, *** Denote significance at the 0.10, 0.05, and 0.01 levels, respectively. All other variable definitions are in Table 2

GC = 1 If a company receives a going concern opinion, and 0 otherwise

(<0.001) (<0.001) (<0.001) (=0.001)

Variable defintion:

13,83% 17,78% 11,75% 17,51%

164,40 16,57 63,29 3,18

(included) (included) (included) (included) 12220 793 5617 114 All Obs. LOSS = 1 All Obs. LOSS = 1

(included) (included) (included) (included) Busy season Subsample Non - Busy Season Subsample Model 1 Model 2 Model 3 Model 4

Definition of the Busy Season period

From the descriptive statistics as presented in Table 4, it is noted that the majority of sampled companies have a fiscal year-end date of December, resulting in the predefined busy season period. This variable of interest is included in the model to capture this effect. As described by López and Peters (2012), this period of heightened pressure can also impact the audits of companies with a January fiscal year-end date by auditor fatigue or delays due to overruns by December year-end clients. In addition, companies with a November fiscal year-end date could also be impacted when the auditors start their interim testing or hard-close procedures focusing on their December year-end clients. To investigate the sensitivity of the results to these elements, estimation is performed for the primary regression model using alternative versions of BUSY. To capture the elements as described the model is estimated using the following alternative busy season periods: December–January, November–December, and November–January. The variable adjusted for the November-December period is not significant, however the estimated coefficient for the other two periods range from 0.045 to 0.057 with p-values below 0.01. In general the regression results are similar to those from Model 3 in the primary regression analyses.

(26)

Supplemental Analysis

In line with López and Peters (2012), the results from the supplemental abnormal accrual model are present in Table 8, focusing on the auditor-related interactions to identify any possible moderation of these characteristics on the relation between workload compression and audit quality. The dependent variable included is ABS_DA. These results are split for the busy season companies and the non-busy season companies. Consistent with Model 4 from the primary results, the estimated coefficient for OFFICE_WLC for the busy season subsample in Model 1 is negative, however insignificant. For the non-busy season companies, a significant interaction is reported between workload compression and the companies influence on the local auditor office through the size of their relative audit fee, INFLUENCExOFFICE_WLC (Coeff. = -0.343, p-value ≤ 0.01). Thus, for non-busy season companies, workload compression appears to be a significant audit quality determinant only for companies with relative high audit fees. This could be the result of improved auditor performance for clients with relative high audit fees, despite the conflicting workload compression experienced by their auditors.

Table 8 Supplemental Analysis of Moderating Effects

Variable Pred. Sign β t-stat β t-stat Intercept -1,426 -0,69 6,691 0,03** OFFICE_WLC + -0,324 -1,19 0,282 0,45 Auditor-Related Controls: TENURE + 0,019 0,61 -0,040 -3,34*** OFFICE_SIZE - -0,021 -0,74 0,014 1,43 INFLUENCE + -0,101 -0,50 0,197 1,63 Auditor-Related Interactions: TENURE x OFFICE_WLC ? -0,035 -0,89 -0,031 -0,55 OFFICE_SIZE x OFFICE_WLC ? 0,039 1,15 -0,031 -0,68 INFLUENCE x OFFICE_WLC ? 0,149 0,69 -0,343 -2,18** Company-Related Controls: SIZE - -0,096 -24,48*** -0,077 -12,48*** Δ REVENUES + 0,125 14,82*** 0,147 8,18*** CFO - -0,014 -1,48 0,039 1,45 LOSS + -0,058 -4,93*** 0,020 0,65 DEBT + 0,175 13,81*** 0,216 8,31*** EC - 0,022 3,73*** 0,023 2,48** Other Controls YEAR ? SICC ? n = Adjusted R2 F-Statistic

*, **, *** Denote significance at the 0.10, 0.05, and 0.01 levels, respectively. The variable definitions are in Table 2

97,10 31,12

(<0.001) (<0.001) Model 1 Model 2

10,55% 7,45%

(included) (included) Busy Season Non-Busy Season

(included) (included) 12220 5617

(27)

Conclusions and limitations

The busy season and the associated workload demands have been challenging for public accounting firms throughout the years. Prior research indicates that the associated pressures can lead to quality threatening behavior at the individual auditor level, and the increasing regulatory scrutiny and compliance with stricter auditing standards do not reduce these pressures. However audit firms have strong incentives to maintain high quality performance because of the possible reputational costs associated with audit failures. As a response the regulators and audit firms have developed control mechanisms in order to maintain the overall quality of the financial audit process. Therefore it is investigated whether the busy season pressure and local office auditor compression as documented in prior experimental research, ultimately affect the audit quality. In contrast to prior literature and the hypotheses, the results show that busy season companies and the concentration of busy season engagements in the client portfolio of a local auditor office are not associated with a decrease in audit quality. Specifically, the results show that busy season companies exhibit lower levels of absolute abnormal accruals. In addition, the magnitude of the abnormal accruals reported by busy season companies decreases with the level of local office workload compression. An explanation can be that this inconsistency with prior literature is the result of the increased regulations and oversight after the passage of SOX and the global economic crisis. Furthermore there can be a shift in earnings management behaviour from accruals-based to real earnings management as identified by Cohen et al. (2008). Therefore further research can be performed on this subject. Based on the results the conclusion can be drawn that workload compression is an important determinant of performance at the local office engagement level, however the results do not indicate that workload pressures negatively impact the audit quality. The findings are supported by alternative tests of abnormal accruals, measures of local office workload compression, and a variety of the busy season period.

The findings are subject to certain limitations. The research focused on public companies in the auditor’s client portfolio, considering the composition of non-public companies in an auditor’s client portfolio is virtually unobservable. This composition could affect the estimation of the local office workload pressure proxies if their fiscal year-end dates differ compared to the public companies. This lack of available data for non-public companies limits the ability to measure the full extent of local office workload pressures. This limitation is common in prior research, that uses data from publicly traded companies to operationalize variables that depend on the relative importance of individual companies to the overall operations of their auditors (López & Peters, 2012). To control for this limitation data from audits performed by the Big 4 firms is used, considering these firms are more likely to have a more homogenous composition of public companies in their client portfolios. A

(28)

second limitation is that no observation is included relating to the interim testing or hard-close procedures performed by auditors. These procedures can result in a distribution of the auditor’s workload over time, which therefore may affect the local office workload compression. It must be noted that the auditing standards limit the possibility to distribute certain procedures over time. Assuming that auditors have the same opportunities to perform interim testing or hard-close procedures, the intercept of the models could be biased, but the estimated regression coefficients for the local office workload compression would not change. Future research on the impact of interim and hard-close procedures on audit quality would benefit current literature on this topic. In addition, the supplemental analysis suggests that the results could be subject to some environmental contexts. Therefore future research would benefit the understanding of these contexts and could help audit firms to improve their quality controls and staffing decisions.

(29)

References

Akono, J., Hranaiova, J., Vinelli, A., & Stein, M. (2011). Audit production and market: Evidence from publicly traded U.S. firms. 2011 PCAOB Academic Conference (April). Washington DC: PCAOB.

Barrett, M., Cooper, D., & Jamal, K. (2005). Globalization and the coordination of work in multi-national audits. Accounting, Organizations and Society(1), 1-24.

Becker, C., DeFond, M., Jiambalvo, J., & Subramanyam, K. (1998). The effect of audit quality on earnings management. Contemporary Accounting Research 15, 1-24.

Becker, G. (1993). Human Capital (3rd ed.). Chicago, IL: University of Chicago Press.

Bloom, R. (2011). The financial crisis inquiry report: analysis and commentary. The CPA Journal(81), 6-10. Carcello, J., Hermanson, R., & McGrath, N. (1992). Audit quality attributes: The perceptions of audit partners,

preparers, and financial statement users. A Journal of Practice and Theory(11), 1-15.

Carpenter, C., & Hock, C. (2008). The 150-hour requirements effect on the CPA exam. The CPA Journal, 62-64.

Centre for Audit Quality. (2014). Report from the working group: Audit Quality Indicators. IAG meeting. Choi, J., Kim, C., Kim, J., & Zang, Y. (2010). Audit office size, audit quality, and audit pricing. Auditing: A

Journal of Practice & Theory 29, 73-97.

Choi, J., Qiu, A., Zang, Y., & Kim, J. (2007). Auditor Locality, audit quality and audit pricing. AAA Anual

meeting. Chicago: Research Collection School Of Accountancy.

Cohen, D., Dey, A., & Lys, T. (2008). Real and accrual-based earnings management in the pre- and post-Sarbanes-Oxley periods. The Accounting Review(83), 757-787.

COM. (2011). Proposal for a Regulation of the European Parliament and of the council on specific

requirments regarding statutory audit of public-interest entities. Brussel: European Commission.

DeAngelo, L. (1981). Auditor size and audit quality. Journal of Accounting and Economics(3), 183-199. DeBoskey, D., & Jiang, W. (2012). Earnings management and auditor specialization in the post-sox era: An

examination of the banking industry. Journal of Bankings & Finance(36), 613-623.

Dechow, P., Sloan, R., & Sweeney, A. (1995). Detecting earnings management. The Accounting Review(70), 193-225.

DeFond, M., & Lennox, C. (2011). The effect of SOX on small auditor exits and audit quality. Journal of

Accounting and Economics(52), 21-40.

DeZoort, F., & Lord, A. (1997). A review and synthesis of pressure effects research in accounting. Journal of

Accounting Literature(16), 28-85.

EC. (2010). Audit policy: Lessons from the Crisis. Brussels: European Commission. Figler, H. (1980). Managing stress. Management Accouting, 22-28.

(30)

Francis, J. (2004). What do we know about audit quality? The British Accounting Review, 345-368. Francis, J. (2011). A framework for understanding and researching audit quality. Auditing: A Journal of

Practice & Theory 29, 247-260.

Francis, J., & Yu, M. (2009). Big 4 office size and audit quality. The Accounting Review(84), 1521-1552. Francis, J., Reichelt, K., & Wang, D. (2005). The pricing of national and city-specific reputations for industry

expertise in the U.S. audit market. The Accounting Review, 113-136.

Frankel, R., Johnson, M., & Nelson, K. (2002). The relation between Auditor's fees for Non-audit services and earnings management. The Accounting Review 77, 71-105.

IAASB. (2014). A framework for audit quality: Key elements that create an enviroment for audit quality. International Federation of Accountants.

Johnson, V., Khurana, I., & Reynolds, J. (2002). Audit firm tenure and the quality of accounting earnings.

Contemporary Accounting Research(19), 637-660.

Jones, A., Strand, C., & Wier, B. (2010). Healthy lifestyles as a coping mechanisme for role stress in public accounting. Behavioral Research in Accounting(22), 21-41.

Jones, J. (1991). Earnings management during important relief investigations. Journal of Accounting

Research(29), 193-228.

Kinney, W., Jr. Palmrose, Z., & Scholz, S. (2004). Auditor independence, non-audit services and restatements: was the US government right. Journal of Accounting Research 42, 561-588.

Kothari, S., Leone, A., & Wasley, C. (2005). Performance matched discretionary accrual measures. Journal of

Accounting and Economics(39), 163-197.

Lang, M., & Lundholm, R. (1993). Cross-sectional determinants of analyst ratings of corporate disclosures.

Journal of Accounting Research(3), 246-271.

Lee, J. (2007). Accounting assessment: Issues include qualified workers and industry regulation. Orange

County Business Journal, 30-33.

López, D., & Peters, G. (2012). The effect of workload compression on audit quality. A Journal of Practice &

Theory(31), 139-165.

Menon, K., & Williams, D. (2004). Former audit partners and abnormal accruals. The Accounting Review(79), 1095-1118.

Meyers, J., Meyers, L., & Omer, T. (2003). Exploring the term of auditor-client relationship and the quality of earnings: A case for mandatory auditor rotation? The Accounting Review(78), 779-799.

Nederlands Beroepsorganisatie van Accountants. (2014, June 15). www.nba.n. Retrieved June 15, 2014, from https://www.nba.nl/voor-leden/permanente-educatie/verplicht-pe-onderwerp-2012/verplicht-pe-onderwerp-professioneel-kritische-instelling-/

Nelson, M., Elliot, J., & Tarpley, R. (2002). Evidence from auditors about managers' and auditors' earnings management decisions. The Accounting Review(Supplement), 175-202.

The effect of busy season pressure and local office workload compression on audit quality