Audit-planning judgments and client-employee compensation contracts.
Incentive contracting is often a necessary component of motivating productive employees (e.g., Kaplan and Atkinson 1998). Based on agency theory, incentive contracting is used to motivate the employee (the agent) to exert effort in performing his duties (Demski 1994). However, many performance measures used in incentive or bonus contracts can be easily manipulated allowing opportunistic employee behavior that may not be consistent with the best interests of the shareholders (the principal) (Healy 1985). Based on professional auditing guidelines (e.g., Canadian Institute of Chartered Accountants [CICA] 1997), when an auditor (the independent monitor) is hired to provide an opinion about the firm's financial statements, the auditor should be concerned about the performance measures used in the incentive contract, particularly if their manipulation could affect the firm's financial statements. Historically, financial performance indicators have typically been used in incentive contracts; however, recently, nonfinancial performance measures have become more common because they can be timely indicators of future financial performance (Banker et al. 2000; Rucci et al. 1998; Ittner et al. 1997). (1)
Not surprisingly, a common characteristic of the fraud found in some recently failed companies is that the interests of executives, seeking to maximize incentive pay, are vastly different from those of other stakeholders in the firm (e.g., see Byrne et al. 2002). Such failures have an obvious impact on the auditor's professional-indemnity insurance premiums, reputation, and, in the wake of the Enron collapse, potentially on the audit firm's future existence. Arguably, without incentive pay, client-employees have relatively less motivation to manipulate information provided to the auditors.
No previous work has explicitly studied this link between the type of client-employee incentive pay (i.e., based on financial or nonfinancial performance measures), client-generated information, and the audit performed. Prior studies show that auditors plan more procedures when a client's internal auditors receive incentive compensation (DeZoort et al. 2001) and when the client has explicit incentive to misstate financial results (Glover et al. 2000) However, this work does not distinguish between incentives based on different types of performance measures. Other related work examines how different types of information, i.e., financial or nonfinancial information, affect the work that auditors do. Specifically, auditors use financial information more than nonfinancial information in audit-planning tasks where financial statement assertions are tested (Dilla and Stone 1997). In addition, auditors emphasize financial information more than nonfinancial information in determining the overall scope of an audit (Cohen et al. 2000). Prior work on the effects of different types of information on the level of required audit procedures, however, does not include a link to different types of client-employee incentive pay.
The purpose of this research is to investigate experimentally whether variations in the types of performance measures used in client-employee (2) incentive compensation contracts will lead to variations in audit planning. Using auditing guidelines (CICA 1997), the bonus-plan hypothesis (Healy 1985), and attribution theory (DeZoort et al. 2001; Petty and Wegener 1998; Chaiken and Maheswaran 1994), we predict that variations in the performance measures used to determine client-employee bonuses will be associated with variations in audit-planning judgments. Specifically, the extent and scope of audit-planning judgments are expected to be greater when manager bonuses are based on financial rather than nonfinancial performance measures, which, in turn, are expected to be greater than when managers do not receive a bonus.
In the current study, 110 Canadian auditors at the senior level or higher from one Big 4 firm took on the role of the audit manager planning the upcoming audit of a hotel client. The independent variable, client compensation scheme, was incorporated into the case. The three compensation schemes are: (1) fixed salary only, (2) fixed salary plus bonus based on financial performance measures, and (3) fixed salary plus bonus based on nonfinancial performance measures. Participants were asked to make several audit-planning judgments.
The results of this study are as follows. We find that audit-planning judgments in general differ in magnitude with differences in client-employee compensation contracts. Specific types of client-employee bonuses, i.e., based on financial and nonfinancial performance measures, are associated with relatively higher audit-planning judgments, relative to the case of no bonuses. Our main result is that more extensive audit plans are made for a client who pays employees bonuses based on financial performance measures than for a client who pays employees bonuses based on nonfinancial performance measures.
This paper makes three contributions to the accounting literature. First, the study highlights that audit-planning judgments will vary in magnitude across different types of client-employee compensation contracts, such as contracts that incorporate bonuses based on either financial or nonfinancial performance measures. Hence, the study has possible implications for audit firms that emphasize an audit approach focused on nonfinancial performance measures; audit-firm training programs, for example, may need to consider how client-manipulated nonfinancial performance measures can provide misleading operational signals in detecting material misstatement of financial information.
Second, the paper adds to a small but important literature on the effect of different performance measures in the employee incentive contracts of large organizations. The differential incentive effect studied here is in the context of the work of external auditors, which has not previously been investigated in the literature. Third, this paper provides empirical evidence of the consistency between auditor behavior and required audit standards. It shows that auditors conduct work in accordance with general prescriptions regarding the potential for manipulation of bonus contracts.
This paper proceeds as follows. In the second section, we present background to the context of the study and the study's hypotheses. In the third section, we outline the experimental design. In the fourth section, we present the data analysis. In the fifth section, we summarize the findings of the paper and present directions for future research.
BACKGROUND AND HYPOTHESES DEVELOPMENT
Compensation schemes that incorporate bonuses based on financial performance measures are well established (Murphy 1998; Ittner et al. 1997). However, research evidence suggests that such measures can be manipulated for incentive purposes by affecting related accruals (Healy 1985; Holthausen et al. 1995; Guidry et al. 1999). Accordingly, incentives based on financial performance measures can influence managers to take opportunistic actions that conflict with the best interests of the owners of the firm.
Using predictions based on agency theory, several papers have explored the manipulation of incentives based on financial performance measures through strategic accruals management. The papers find, across firms, a correlation between earnings-based incentive structures and the level of reported earnings. (e.g., see Healy 1985; Holthausen et al. 1995; Guidry et al. 1999) An implication for auditing is that such manipulations might increase the risk of material misstatement in the financial statements and thus should affect the audit plan. Section 5135 (Appendix A 1.a.(i)) of the Canadian Institute of Chartered Accountants (CICA) handbook specifically lists the existence of bonus contracts based on financial results as a factor that might increase the risk of material misstatement. The handbook suggests that an increased risk should cause the auditor to question the internal control system, obtain more reliable evidence, and expand the extent of audit procedures performed. In the U.S., the recently issued SAS No. 99 explicitly identifies "significant portions of [management] compensation ... being contingent upon achieving aggressive targets for stock price, operating results, financial position, or cash flow" (3) as a risk factor relating to misstatements arising from fraudulent financial reporting.
Attribution theory also provides a framework for our hypotheses. Attribution theory suggests that when information sources have incentive to bias, the evaluator will assess their message as less persuasive due to the source's incentive to bias, and thus will rely less on the source (see Eagly and Chaiken  and Jaspars et al.  for reviews of relevant research and Glover et al.  and Hirst  for related auditing research). Relating this to the incentive compensation scenario, given (as stated above) that client-employees are prone to manipulating bonus-related information, auditors will be skeptical of the information they provide, perceiving its relation to compensation incentives. Thus, the audit plan is expected to be impacted by the client-employee compensation contract.
The Audit Plan
An important feature of the audit-planning process, relevant to this study, is that the audit plan and procedures to be performed rely in part on the client's control system and in part on specific audit risks (Robertson and Smieliauskas 1998). Section 5135 Appendix A of the CICA handbook recommends that auditors consider the nature of employee incentive contracts in assessing the risk of material misstatement in the client's financial statements. Similar recommendations are found in the International Standards on Auditing (AICPA 2003, AU Section 8240, paragraphs .34 to .38 and .78) and the U.S. Auditing Standards (AICPA, 2003, AU Section 316 paragraph .36). Increased risk of material misstatement caused by the presence of employee incentive contracts places increased responsibility on the auditor due to the CICA requirement to obtain more evidence and expand the extent of audit procedures performed.
A CICA (1995) publication, Guidance on Control, also indicates that compensation scheme can be an integral part of a client's internal control system. For example, one criterion in assessing control is the implementation of human resource policies consistent with achievement of the organization's objectives. The detailed explanation of this criterion in the CICA publication reveals that it requires an assessment of whether the reward systems are consistent with the organization's objectives and whether it includes incentives based on both financial and nonfinancial performance measures (CICA 1995, 16).
This study tests three hypotheses. The first hypothesis focuses on the effect of a bonus plan based on financial performance measures on the audit compared to a fixed salary compensation contract. Consistent with a direct application of Section 5135 of the CICA Handbook, all four phases of audit planning (business risk, audit risk, internal control, and audit procedures) should be affected when a client pays employees bonuses based on financial performance measures, relative to a similar client who pays employees a fixed wage only. Specifically, auditors are expected to adapt their audit plan for the increased risk by increasing the amount of audit procedures required.
This prediction is also bolstered by the bonus-plan hypothesis (Healy 1985), which predicts that employees will engage in opportunistic activities that manipulate earnings to maximize a bonus. Such actions increase the risk that financial statements will be materially misstated, undermine the internal control system, and require higher levels of audit planning. Furthermore, attribution theory (DeZoort et al. 2001; Eagly and Chaiken 1993; Jaspars et al. 1983) suggests auditors will rely less on the client-provided information when they have self-serving incentives. Accordingly, the audit plan will be affected by whether there exists a client-employee bonus scheme.
Overall, the first hypothesis can be formally expressed as:
H1: Audit-planning judgments will be greater when client-employee compensation contracts are comprised of a fixed salary plus a bonus based on financial performance rather than comprised of a fixed salary only.
The second hypothesis pertains to the relative level of planning required to audit client-employee bonus contracts based on nonfinancial performance measures compared to contracts based on a fixed salary only. With bonus pay, employees also may be motivated to manipulate the performance measure. Manipulating nonfinancial performance measures, however, does not directly imply financial statements are misstated. For example, a customer satisfaction measure that is inflated because of manipulation by employees does not necessarily mean the firm's financial statements are misstated.
Notwithstanding, we identify three reasons why we expect auditors to increase their audit-planning judgments more when the client uses a nonfinancial performance measure than the case where the client pays employees based on a fixed salary. First, attribution theory would suggest that the potential manipulation of a nonfinancial measure means the auditor will rely relatively less on that measure, which in turn will impact audit planning. Second, auditing standards recommend that auditors consider incentives when planning the audit and assessing risk. Third, client-employees may take actions that maximize a bonus based on nonfinancial performance measures and such actions may indirectly increase the risk that the financial statements are materially misstated. (4)
Overall, similar to H1, we predict:
H2: Audit-planning judgments will be greater when client-employee compensation contracts are comprised of a fixed salary plus a bonus based on nonfinancial performance rather than comprised of a fixed salary only.
The third hypothesis examines the effect of client-employee bonuses based on the different types of performance measures (i.e., financial or nonfinancial). (5) Although attribution theory, auditing guidelines, and the bonus-plan hypothesis from positive accounting theory all suggest that incentive compensation schemes will affect audit planning, we predict that the extent and scope of audit planning will be greater when bonuses are based on financial performance measures than when bonuses are based on nonfinancial performance measures. Specifically, audit risk will be assessed as higher, internal controls will be assessed as weaker and a greater level of audit procedures will be required.
We have two reasons for this prediction. First, previous research finds that when there is a greater opportunity for the information to be biased, evaluators are more sensitive to source attributes, and thus more motivated to attend to them (DeZoort et al. 2001; Petty and Wegener 1998; Chaiken and Maheswaran 1994). Therefore, in the current study, when bonuses are based on nonfinancial performance measures that are independently gathered, the opportunity to bias the measures should be reduced. Second, the implications of manipulations of nonfinancial measures may be less direct than manipulations of financial performance measures. That is, manipulation of a financial performance measure clearly increases the risk that the financial statements are misstated, whereas manipulation of a nonfinancial measure does not necessarily translate into materially misstated financial statements. (6) Formally stated:
H3: Audit-planning judgments will be greater when client-employee compensation contracts that include bonuses are based on financial, rather than nonfinancial, performance measures.
RESEARCH DESIGN AND METHOD
Participants and Administration
Participants in the experiment comprised 110 Canadian auditors at least at the senior level, with, on average, 2.5 years experience from one Big 4 firm. The participants were attending off-site senior training sessions in Canada. (7) One of the researchers was present at all sessions. All participation was confidential, anonymous, and voluntary. However, a lottery draw for $200 at each training session was used to motivate the auditors to participate in the study. Auditors at least at the senior level were selected to participate in this study due to their heavy involvement and responsibility of planning upcoming audits.
At each session, participants were randomly assigned to one of the three treatment groups. The researcher then gave a brief introduction containing generalized information only. The researcher then instructed participants to read the instructions, complete the instrument without consulting their peers, and not to change their responses after completing each part. Any questions participants had were responded to on an individual basis. The instrument took between 30-35 minutes to complete.
Experiment Design and Materials
A 3 x 1 factorial design experiment was used in the study. The independent variable, client-employee compensation contract is manipulated between participants. The three client-employee compensation contracts are: (1) the fixed salary contract only, (2) a contract comprising a fixed salary plus a bonus based on nonfinancial performance measures, and (3) a contract comprising a fixed salary plus a bonus based on a financial performance measure.
The research instrument contained three distinct parts (hereafter referred to as parts one, two, and three). (8) Part one of the instrument contained a case concerning the audit of a large, successful, urban high-end hotel. Participants were asked to assume the role of the audit manager planning the upcoming audit of the hotel. (9) Case facts varied only with respect to the independent variable, i.e., the client-employee compensation contract used to remunerate the hotel's general and departmental managers. In all other respects, case facts were identical. Information in the case included the background of the hotel (clientele, organization chart, positive relationship of results to the industry) as well as the audit history (client for 10 years, good auditor-client relations, strong internal controls). (10)
In the condition where a bonus was based on financial performance, manager compensation was comprised of a fixed salary plus a percentage of departmental net income. In the condition where a bonus was based on nonfinancial performance measures, manager compensation was comprised of a fixed salary plus a variable amount based on the departmental level of client satisfaction, client complaints, percent of return customers, and client use of other services. The compensation contract for the fixed salary condition is self-explanatory. (11)
Part two of the instrument contained a variety of questions about the case and about the participant's planned audit approach for this client (the Appendix lists the responses to these questions). The responses to these questions were used to operationalize the dependent variable, audit-planning judgments. Specifically, auditors were asked to make the following audit-planning judgments for the upcoming audit of this hotel client: (1) assess audit risk, (2) assess the internal control system, (3) estimate the level of internal control procedures, and (4) estimate the level of substantive audit procedures required for the current year's audit (the Appendix presents the four questions as specifically viewed by the participants).
The participants were asked to make their assessments relative to a typical audit. For each participant, we measured each of the four judgments on an 11-point scale. To test the hypotheses, we compared across treatment groups (i.e., financial, nonfinancial, and fixed salary) the average participants' response to each of the four audit-planning judgments. Accordingly, our objective was to determine whether the magnitude of audit planning, as measured by the four judgments, was significantly different for the different compensation contracts. Specifically, we were interested in determining whether audit-planning judgments, as measured (separately, on an 11-point scale) by an audit risk assessment, an internal control assessment, an estimate of internal control audit procedures, and an estimate of substantive audit procedures was of a different magnitude for the different types of compensation contracts..
Part three of the instrument contained demographic questions including experience in auditing compensation contracts, a manipulation check, which required participants to select the client compensation scheme employed in the case from a list of various compensation schemes, and general questions about compensation, internal control, and audit risks and the relationship between each of these. Specifically, we asked each participant how the compensation scheme in the case influenced his or her previous assessments of risk, internal control, and planned level of audit procedures. These latter assessments were used to corroborate and further explain our findings from testing the main dependent variables.
ANALYSIS AND DISCUSSION
In this section, we present some descriptive statistics as well as discuss and analyze the results. The manipulation check resulted in greater than 90 percent of participants correctly responding to a question regarding the type of compensation contract in the case. (12) The descriptive statistics are presented next, followed by the detailed results from testing the hypotheses.
Table 1 presents the descriptive statistics of the data. The participants were about 26.16 years old with on average 2.47 years of audit experience. (13) Participants were aware of strategic systems auditing, however only 35 percent had strategic systems auditing experience. (14) The descriptive statistics also indicate that a large majority of participants (87 percent) believe that compensation schemes influence internal control.
Generally, there were no significant differences across the three groups, in terms of demographic variables or general perceptions. However, participants in the condition with a bonus based on financial performance were more likely to believe that the compensation scheme would decrease internal control, whereas participants in the condition with a bonus based on nonfinancial performance measures and the fixed salary condition were more likely to believe that the compensation scheme would increase internal control. Finally, participants report that bonuses based on nonfinancial performance measures are about half as common in practice as bonuses based on financial performance measures.
Tests of Hypotheses
We used Multivariate Analysis of Variance to test the overall impact of compensation scheme on audit planning. The dependent variable, audit-planning judgments, was represented by assessments of audit risk, internal control, estimated level of internal control, and substantive procedures. (15) Compensation scheme served as the independent variable. All dependent variables were measured on 11-point scales.
Hypothesis 1 predicts that the extent and scope of audit planning would be greater when client compensation contracts incorporated bonuses based on financial performance measures than when client compensation contracts were based on fixed salary only. Per Table 2, our results support the hypothesis (Wilks' Lambda = 0.888, p < 0.01). We find that auditors assess a client with employee compensation contracts that incorporate bonuses based on financial performance measures as being riskier (4.94 versus 3.58; F = 9.58; p = 0.001) and having a weaker internal control system (5.83 versus 6.85; F = 5.05; p = 0.01) than a client with a fixed salary only compensation contract.
We also found a significant difference between the planned level of substantive audit procedures estimated to audit a client whose compensation contract incorporates a fixed salary plus a bonus based on financial performance measures (4.94) and a client whose compensation contract incorporates fixed salary contracts only (3.76) (F = 6.42; p =0.006). However, the planned level of internal control procedures did not significantly differ. Consistent with this finding, past research has found that auditors tend not to rely on internal control and thus do not adjust internal control tests when faced with riskier clients or weaker internal control systems (Mock and Wright 1993, 1999; Waller 1993; Bedard 1989).
To corroborate our findings, we compared the participants' responses to questions inquiring about the influence of the compensation contract on the participants' planning judgments in the condition that incorporated bonuses based on financial performance measures to the condition that incorporated a fixed salary only. These questions were asked subsequent to all questions relating to the dependent variables and the manipulation check. We found, as predicted, that contracts incorporating bonuses based on financial performance measures had a significantly stronger influence on the participants when they were making their assessments of risk, internal control, and the level of audit procedures required to perform the audit than compensation contracts that did not incorporate bonuses (p < 0.01, see responses in Table 1, Panel B). These results are consistent with the arguments used to develop H1. That is, the audit-planning judgments will be affected by the presence of client-employee bonus contracts based on financial performance measures because of the risk of material misstatement arising from opportunistic employees. Auditors are more sensitive to the compensation contracts involving bonuses since client-employees have the motivation and the opportunity to take self-interested actions that are not necessarily in the best interests of the client.
Our second hypothesis, which predicts that client-employee bonuses based on nonfinancial performance measures will increase audit-planning judgments, is not fully supported (Wilks' Lambda = 0.972 and not significant). We find that auditors do not assess clients that pay employees bonuses based on nonfinancial measures as riskier than clients without client-employee bonus contracts. In addition, the planned level of internal control procedures does not significantly differ between conditions. However, the internal control system is assessed as weaker (6.14 versus 6.85; F = 2.50; p = 0.06) and the planned amount of substantive procedures is higher (4.39 versus 3.76; F = 1.66; p = 0.10). The insignificant difference in internal control procedures can again be viewed in the context of past research that suggests auditors typically do not adjust the internal control procedures within their audits (Mock and Wright 1993, 1999; Waller 1993; Bedard 1989).
In follow-up analyses on the influence of the compensation scheme on the dependent variable responses, we find that the auditors' perceptions of the level of influence of the compensation scheme when making the audit-planning judgments (risk, internal control, and planned level of audit procedures assessments) was significantly higher for clients with bonuses based on nonfinancial performance measures than for clients with fixed salaries only (p < 0.05, see responses in Table 1, Panel B). These results are consistent with attribution theory, as it appears that auditors are sensitive to the potentially biased information they may receive when client-employee bonuses are based on nonfinancial performance measures. What is of interest, however, is that auditors do not find the audit to be more risky. This may be due to auditors' belief that it would be difficult for client management manipulation of information relating to nonfinancial performance measures to result in misstated financial statements.
The third hypothesis is marginally supported (Wilks' Lambda = 0.943, p < 0.10). Here, consistent with predictions, the results indicate that the audit is perceived to be significantly riskier when a client's compensation contracts incorporate bonuses based on financial performance measures (4.94), as compared to when a client's compensation contracts incorporate bonuses based on nonfinancial performance measures (4.00) (F = 4.74; p = 0.002, see Table 2). The planned amount of substantive procedures are significantly higher when clients use financial performance measures on which to base bonus payments to employees (4.94) as compared to nonfinancial performance measures (4.39) (F = 1.66; p = 0.10). However, there is no difference in the assessments of the strength of the internal control system, and consistent with past research (Mock and Wright 1993, 1999; Wailer 1993; Bedard 1989), no difference in the planned amount of internal control audit procedures. These results are consistent with attribution theory; the auditors' judgments suggest that they perceive that client-employees can manipulate bonuses based on financial performance measures more easily, resulting in riskier audits than the case of client-employee bonuses based on nonfinancial performance measures.
We found similar results when participants were asked about the influence of the compensation scheme on their audit-planning judgments. In support of H3, participants for whom the compensation scheme included a bonus based on financial performance measures perceived the influence of the compensation scheme on their risk assessment and level of audit procedures estimates to be significantly higher than those for whom the compensation scheme included a bonus based on nonfinancial performance measures (p < 0.0001, see responses in Table 1, Panel B). However, they did not perceive that their internal control assessments were differentially influenced by the compensation contract. These results suggest that auditors perceive that the manipulation of financial performance measures is more severe than the manipulation of nonfinancial performance measures. (16) This may be due to auditors' inexperience with nonfinancial performance measures.
Audit Risk as a Mediating Variable
Given that auditing methodology suggests that audit planning should begin with an initial risk assessment and that this risk assessment should affect the subsequent audit-planning decisions, we also analyzed our data to determine if the audit risk assessment is a mediating variable. (17) More specifically, does type of compensation contract affect audit risk assessments, which then in turn impact or mediate the other audit-planning judgments. A variable can be said to function as a mediator if it accounts for the relationship between the independent variable (type of compensation scheme) and the dependent variable (audit-planning judgments) (see Baron and Kenny 1986). To establish mediation, three conditions must hold: (1) type of compensation contract must be shown to affect the audit risk judgment, (2) type of compensation contract must be shown to affect audit-planning judgments, and (3) type of compensation contract must diminish in significance when audit risk judgment is included as a covariate in the model.
Our supplementary analysis indicates that audit risk does function as a mediating variable. The first two conditions are shown to hold in Table 2. The mediating variable, audit risk assessment, is affected by type of compensation scheme (F = 5.03, p < 0.001), thus condition (1) is met. The dependent variables, internal control assessment (F = 2.66, p < 0.10), and substantive procedures assessment (F = 3.21, p < 0.05) are affected by type of compensation contract, thus condition (2) is met. (18)
To determine if the third condition holds, we performed a Multivariate Analysis of Covariance, with audit risk as a covariate. This would establish the link between the mediating variable and the remaining dependent variables. The results of the Multivariate Analysis of Covariance (see Table 3), indicates that audit risk is significant (p < 0.001) in each of the individual Analyses of Variances for internal control assessment, and the substantive procedures assessment, but that these other audit-planning judgment variables are no longer significant. Thus, audit risk is functioning as a mediating variable. To interpret these results, the type of compensation scheme impacts audit-planning judgments, however, one of those judgments, audit risk, also impacts the remaining audit-planning judgments and, when viewed as a mediator, reduces the significant effect that type of client-employee compensation has on the remaining audit-planning judgments. These results are thus indicative that audit risk may be the primary reason why the type of client-employee compensation contract is associated with audit-planning judgments.
Corroboration of Dependent Variables
To further corroborate the dependent variables, we coded responses to the question, "What are the factors that would influence your risk assessment for the upcoming audit of The Lincoln Hotel?" as follows. (19) First, if the response contained a reference to the firm's compensation system, then it received one point. Second, if the response also contained a reference to a manipulation of the firm's either financial or nonfinancial performance so that the manager could increase their compensation, then the response received another point. If the response contained no explicit statement about the firm's compensation system, then it did not receive any points.
The results were as follows. For respondents in the group with client-employees having a bonus based on financial performance, the mean score was 1.472. For respondents in the group where client-employees were paid a bonus based on nonfinancial performance, the mean score was 1.052. For respondents where client-employees were paid fixed salaries, the mean score was 0.389. All group means were significantly different from each other (p < 0.02 for all mean comparisons). This evidence supports the view that auditors were concerned most by client-employee contracts that contained, in order, bonuses based on financial measures, bonuses based on nonfinancial measures, and fixed salaries only.
This research investigates experimentally whether variations in the types of performance measures used in client-employee bonus contracts are associated with variations in audit-planning judgments. Theory and institutional evidence suggests that audit planning should be affected by audit-client use of employee bonus contracts. Prior work, however, has not examined whether audit-planning judgments will have different magnitudes depending on the type of client-employee compensation contract. The study tests three hypotheses. The first predicts that audit planning will be more extensive when client-employee compensation contracts incorporate bonuses based on financial performance measures than when clients use fixed-salary-only contracts. The second hypothesis makes the same prediction for the case of bonuses based on nonfinancial rather than financial performance measures. The third hypothesis predicts that more extensive audit planning will be required for clients using fixed salary plus employee bonuses based on financial performance than for clients using fixed salary plus employee bonuses based on nonfinancial performance.
Bonuses based on financial performance measures require the most extensive audit planning, followed by bonuses based on nonfinancial performance measures. Fixed salary compensation schemes require the least extensive audit planning. On balance, the results support the argument that client-employee compensation contracts that pay bonuses based on performance measures increase audit risk, which, in turn, weakens the internal control system, and requires higher levels of audit procedures.
We also find that clients with compensation contracts incorporating bonuses based on a financial performance measure are perceived as having greater audit risk and require significantly higher levels of audit procedures than clients with compensation contracts incorporating bonuses based on nonfinancial performance measures. Further research is needed to determine precisely why auditors might be relatively less concerned about client-employee bonuses based on nonfinancial performance measures.
One important limitation of this study is that it is difficult to distinguish whether the results are attributable to the type of performance measure, or, to the relative ease with which the performance measure could be manipulated. It might be argued that the nonfinancial performance measures included in the study, i.e., client satisfaction, client complaints, percent of return customers, and client use of other services, are relatively more objective and naturally provide less scope for manipulation than financial measures. It is conceivable that the results of our study might be sensitive to our choices of the nonfinancial performance measures. In other words, had we chosen nonfinancial performance measures that could be manipulated more easily, for example, self-reported customer satisfaction scores, then differences we found between financial and nonfinancial measures may have been mitigated. However, given that customer satisfaction does not directly affect the financial statements, it is unclear whether the differences would have been eliminated. Thus, a possibility for future research may be to investigate the effect of the ease with which performance measures used in client-employee bonus contracts can be manipulated.
Finally, this research separately examines the effects of client-employee bonuses based on financial performance measures and bonuses based on nonfinancial performance measures. Recent developments in managerial accounting suggest that bonus-based compensation is becoming more complex with the increasing popularity of the Balanced Scorecard (e.g., Ittner et al. 2003), a performance measurement tool that links a combination of financial and nonfinancial performance measures to the strategic objectives of the firm. Two associated issues arise. First, while the Balanced Scorecard may be a useful performance measurement tool, it does not necessarily follow that clients would use the Balanced Scorecard for contracting purposes. Rather, a client may choose to reward employees with a fixed salary or incorporate bonuses based on financial performance measures despite the implementation of a Balanced Scorecard measurement tool. Second, even if the Balanced Scorecard is used as a basis for compensation contracting, it is not clear what type of measures (e.g., financial or nonfinancial) will be weighted more heavily in the compensation contract.
Based on the results of this paper, one might contend that if a compensation contract incorporates bonuses based on the Balanced Scorecard, then a lower amount of audit procedures might be necessary than those required for a compensation contract that incorporates pure financial measures. In addition, a greater amount of audit procedures might be necessary than those required for a compensation contract that incorporates pure nonfinancial measures (i.e., because of the presence of financial measures in a potential Balanced Scorecard compensation contract). On the other hand, it is also possible that the presence of the Balanced Scorecard's m/x of financial and nonfinancial performance measures (with subjective weightings on each measure) creates even more uncertainty for the auditor and would require higher levels of audit procedures than the case where bonuses are based purely on financial performance measures. Further research is needed to disentangle such effects.
APPENDIX QUESTIONS USED FOR DEVELOPING MEASURES OF THE DEPENDENT VARIABLE: AUDIT-PLANNING JUDGMENTS
(All questions were measured on 11-point scales.)
Audit Risk Assessment
Based on the information you have been given, what is your preliminary audit risk assessment for The Lincoln Hotel?
Internal Control Assessment
Based on the information you have been given, what is your assessment of the internal control system?
Internal Control Testing
Please estimate the amount of internal control testing required for the current year's audit of The Lincoln Hotel
Substantive Audit Procedures
Please estimate the amount of substantive audit procedures required for the current year's audit of The Lincoln Hotel.
TABLE 1 Descriptive Statistics Panel A: Demographics Experimental Groups Client-Employee Compensation Scheme Types Bonus Bonus Based Based on on Non- Fixed- Financial financial Salary Description Scale Total Measures Measures Only Number of Participants 110 36 38 36 Male % 47 53 42 47 Female % 53 47 58 53 Age Years 26.16 26.17 26.34 25.97 Audit Experience Years 2.47 2.21 3.20 2.01 % Respondents: Senior % 96 92 95 100 Manager % 4 8 3 0 Partner % 1 0 3 0 with strategic % 35 42 39 25 auditing experience who use strategic % 0 0 0 0 auditing extensively using strategic % 4 3 5 3 auditing half the time rarely using % 35 42 39 25 strategic auditing Panel B: Experimental Question Responses Audit-Planning Assessments Based on Case Information: Audit Risk Assessment 1 to 11 4.17 4.94 (a) 4.00 (b) 3.58 (b) Internal Control 1 to 11 6.27 5.83 (a) 6.14 (a) 6.85 (b) Assessment Internal Control 1 to 11 6.12 6.44 (a) 5.87 (a) 6.04 (a) Tests Required Substantive 1 to 11 4.36 4.94 (a) 4.39 (b) 3.76 (c) Tests Required Internal Control 1 to 22 10.49 11.39 (a) 10.26 (b) 9.81 (b) and Substantive Tests Required Perceived influence of: Case compensation 1 to 11 6.18 8.33 (a) 5.63 (b) 4.58 (c) scheme on risk assessment Case compensation 1 to 11 5.18 7.40 (a) 4.91 (b) 3.24 (c) scheme on audit procedures required Case compensation 1 to 11 3.11 3.96 (a) 3.33 (a) 2.04 (b) scheme on internal controls Perceptions of the effect of compensation schemes on internal controls: Case compensation % Yes 35% 61% (a) 32% (b) 11% (c) scheme decreased internal control? Case compensation %_ Yes 30% 14% (a) 42% (b) 33% (a) scheme increased (b) internal control? (a,b,c) Items with the same letter are not significantly different. Items with different letters are significantly different at p<.10. TABLE 2 Multiple and Univariate Analysis of Variance for Audit-Planning Assessments Panel A: Overall MANOVA Df Wilks' Lambda 8 .875 Level of Audit-Planning Assessment Hyp Fixed vs Financial H1 4 .888 *** Fixed vs Nonfinancial H2 4 .972 Nonfinancial vs Financial H3 4 .943 * Panel B: Univariate ANOVAs for Audit Audit-Planning Assessments Risk Assessment (a) [SIGMA] Source Df Squares F Model 2 34.39 5.03 *** Error 107 362.14 Corrected Total 109 396.53 Internal Control Assessment (a) [SIGMA] Source Df Squares F Model 2 9.41 2.66 * Error 107 386.84 Corrected Total 109 406.24 Control Procedures Assessment (a) [SIGMA] Source Df Squares F Model 2 4.72 0.47 Error 107 535.20 Corrected Total 109 539.92 Substantive Procedures Assessment (a) [SIGMA] Source Df Squares F Model 2 28.39 3.21 ** Error 107 468.79 Corrected Total 109 497.18 Panel C: Treatment Means Level of Audit-Planning Assessment Mean Std Dev n Fixed Salary (b) (c) 3.58 1.58 36 Bonus based on Financial Measures (b) (c) 4.94 2.07 36 Bonus based on Nonfinancial Measures (b) (c) 4.00 1.84 38 Level of Audit-Planning Assessment Mean Std Dev n Fixed Salary (b) (c) 6.85 1.63 36 Bonus based on Financial Measures (b) (c) 5.83 2.04 36 Bonus based on Nonfinancial Measures (b) (c) 6.14 2.03 38 Level of Audit-Planning Assessment Mean Std Dev n Fixed Salary (b) (c) 6.04 2.42 36 Bonus based on Financial Measures (b) (c) 6.44 2.10 36 Bonus based on Nonfinancial Measures (b) (c) 5.87 2.23 38 Level of Audit-Planning Assessment Mean Std Dev n Fixed Salary (b) (c) 3.76 1.89 36 Bonus based on Financial Measures (b) (c) 4.94 2.01 36 Bonus based on Nonfinancial Measures (b) (c) 4.39 2.39 38 Panel D: Planned Contrasts Level of Audit-Planning Assessment Hyp F-value p-value Fixed vs Financial H1 9.60 0.001 Fixed vs Nonfinancial H2 0.94 0.17 Nonfinancial vs Financial H3 4.74 0.002 Level of Audit-Planning Assessment Hyp F-value p-value Fixed vs Financial H1 5.05 0.01 Fixed vs Nonfinancial H2 2.50 0.06 Nonfinancial vs Financial H3 0.50 0.24 Level of Audit-Planning Assessment Hyp F-value p-value Fixed vs Financial H1 0.38 0.27 Fixed vs Nonfinancial H2 0.11 0.37 Nonfinancial vs Financial H3 0.91 0.17 Level of Audit-Planning Assessment F-value p-value Fixed vs Financial 6.42 0.006 Fixed vs Nonfinancial 1.66 0.10 Nonfinancial vs Financial 1.66 0.10 *, **, *** significant at p < 0.10 (one-tailed); p < 0.05 (one-tailed); and p < 0.01(one-tailed), respectively. (a) The dependent variable, audit planning, was proxied by the following audit-planning decisions: audit risk, internal control, planned level of control procedures, and planned level of substantive procedures. The audit risk, internal control, planned level of internal control, and planned level of substantive procedures assessments were measured on 11-point scales. One-tailed tests were applied to all hypotheses. (b) The independent variable was type of compensation scheme: (1) fixed salary only; (2) bonus based on financial performance measures and (3) bonus based on nonfinancial performance measures. (c) Refer to Appendix A for additional descriptions of these measures. TABLE 3 Multiple Analysis of Covariance for Audit-Planning Assessments with Audit Risk as Covariate Panel A: Overall MANOVA Df Wilks' Lambda 8 0.958 Level of Audit-Planning Assessment Hyp Fixed vs Financial H1 4 .966 Fixed vs Nonfinancial H2 4 .981 Nonfinancial vs Financial H3 4 .985 Panel B: Univariate ANOVAs for Internal Control Audit-Planning Assessments Assessment (a) Source [SIGMA] Df Squares F Model 2 6.52 1.02 Error 1 50.43 15.74 *** Corrected Total 106 406.24 Panel C: Treatment Means Level of Audit-Planning Assessment Mean Std Dev n Fixed Salary (b) (c) 6.85 1.63 36 Bonus based on Financial Measures (b) (c) 5.83 2.04 36 Bonus based on Nonfinancial Measures (b) (c) 6.14 2.03 38 Panel D: Planned Contrasts Level of Audit-Planning Assessment Hyp F-value p-value Fixed vs Financial H1 1.33 0.13 Fixed vs Nonfinancial H2 1.71 0.10 Nonfinancial vs Financial H3 0.01 0.47 Panel A: Overall MANOVA Level of Audit-Planning Assessment Fixed vs Financial Fixed vs Nonfinancial Nonfinancial vs Financial Control Panel B: Univariate ANOVAs for Procedures Audit-Planning Assessments Assessment (a) Source [SIGMA] Df Squares F Model 2 4.99 0.49 Error 1 0.27 0.05 Corrected Total 106 534.93 Panel C: Treatment Means Level of Audit-Planning Assessment Mean Std Dev n Fixed Salary (b) (c) 6.04 2.42 36 Bonus based on Financial Measures (b) (c) 6.44 2.10 36 Bonus based on Nonfinancial Measures (b) (c) 5.87 2.23 38 Panel D: Planned Contrasts Level of Audit-Planning Assessment Hyp F-value p-value Fixed vs Financial H1 0.43 0.26 Fixed vs Nonfinancial H2 0.09 0.38 Nonfinancial vs Financial H3 0.96 0.17 Panel A: Overall MANOVA Level of Audit-Planning Assessment Fixed vs Financial Fixed vs Nonfinancial Nonfinancial vs Financial Substantive Panel B: Univariate ANOVAs for Procedures Audit-Planning Assessments Assessment (a) Source [SIGMA] Df Squares F Model 2 5.76 0.81 Error 1 95.29 26.79 *** Corrected Total 106 373.50 373.5 Panel C: Treatment Means Level of Audit-Planning Assessment Mean Std Dev n Fixed Salary (b) (c) 3.76 1.89 36 Bonus based on Financial Measures (b) (c) 4.94 2.01 36 Bonus based on Nonfinancial Measures (b) (c) 4.39 2.39 38 Panel D: Planned Contrasts Level of Audit-Planning Assessment Hyp F-value p-value Fixed vs Financial H1 1.47 0.11 Fixed vs Nonfinancial H2 0.90 0.17 Nonfinancial vs Financial H3 0.11 0.37 *** significant at p < 0.01 (one-tailed). (a) The dependent variable, audit planning, was proxied by the following audit-planning decisions: audit risk, internal control, planned level of control procedures, and planned level of substantive procedures. The audit risk, internal control, planned level of internal control, and planned level of substantive procedures assessments were measured on 11-point scales. One-tailed tests were applied to all hypotheses. (b) The independent variable was type of compensation scheme: (1) fixed salary only; (2) bonus based on financial performance measures and (3) bonus based on nonfinancial performance measures. (c) Refer to the Appendix for additional descriptions of these measures.
we thank Urton Anderson, Keith Houghton, Steve Kaplan, Lisa Koonce, Theresa Libby, Steve Salterio, Charlene See, Kristy Towry, Arnie Wright, Alex Yen, two anonymous reviewers, and workshop participants at the 2001 American Accounting Association Annual Meeting in Atlanta, Georgia, the 2000 the Canadian Academic Accounting Association Annual Meeting in Halifax, Canada, and The University of Texas at Austin Summer Brownbag Workshop for their helpful comments. The authors are also grateful for the financial support of the Canadian Academic Accounting Association, the Curtin Business School Internal Research Grants, and the Australian Research Council Grants awarded by the University of Notre Dame, Australia.
(1) For the purposes of this paper, nonfinancial performance indicators or measures represent a quantitative statistic that excludes any reference to monetary metrics. Nonfinancial examples include the number of order backlogs, number of defective products, or machinery efficiency. On the other hand, a financial performance measure is defined for the purpose of this paper as a quantitative statistic that refers to a monetary metric. Examples include net income, quality costs, and return on investment.
(2) We consider a client's employees to include any of the client's executive or managerial employees who receive incentive compensation pay. We choose the specific case of divisional-manager client-employees for the case because we were developing unique case materials that exploited our access to the expertise of a Big 4 accounting firm partner in the hotel industry. Divisional managers best fit the case scenarios. A possible limitation of this study is that our auditors may have responded differently had the client-employees in the case materials been more senior than divisional-managers. However, during the experiment and in the experiment debriefing we did not detect any assessment by our auditors of the seniority of the client-employee.
(3) Codified in the U.S. Auditing Standards, AU Section 316, Appendix A.2, paragraph c.
(4) For example, a cheat-employee may know that current-period customer-satisfaction scores, on which incentives are paid, would increase if the employee issued coupons to customers for future discounts (or possibly frequent flyer bonus points) that were not recorded in the accounting system. Consequently, the current liabilities of the audit client would be understated. A further example relates to the collectability of accounts receivable. An employee whose bonus depends on the level of customer complaints may be motivated to misreport complaints or delay reporting customer complaints to the next reporting period. However, if the customer complaints are indicative of a probability that debtors will default, then the firm's current-period financial statements may understate accounts receivable. If a client-employee attempts to protect a bonus by "hiding" any customer complaints that (unbeknown to the employee) have potentially serious litigation consequences, then the financial statements may also be misstated because of an omission of a contingent liability.
(5) Despite the theoretical possibility that incentives based on only nonfinancial performance measures may be optimal for some firms (e.g., when such measures are perfectly noiseless), there is no evidence in the literature to suggest firms commonly use incentive contracts based only on nonfinancial performance measures. One of the experimental conditions in this study asks the respondent to assume that the client's employees are paid a fixed wage plus incentives based only on nonfinancial performance measures. This design choice was made to highlight differences in the treatment conditions. Ex post examination of written comments by the respondents showed no indication that the respondents were in any way confused by what might be argued as an "unusual" incentive contract structure. In addition, no respondent indicated that they treated the task any differently because the structure of the incentive contract they read in the case study differed from the structure of incentive contracts that they had normally audited in practice.
(6) If the nonfinancial measures were voluntarily disclosed, for example, in the financial statements, then the material misstatement risk is higher when there exists nonfinancial incentives relative to the case where the nonfinancial information was not disclosed. However, this risk is still unlikely to be as high as in the case where earnings (i.e., financial performance) are directly managed.
(7) The data were gathered during a training session comprising a review of strategic systems auditing. Accordingly, participants may have been sensitized to the importance of nonfinancial performance measures in responding to the research instrument. However, this sensitization was expected to bias results against finding support for H3.
(8) The research instrument is available upon request from the authors.
(9) We chose the audit manager role because both audit managers and seniors were expected to participate in the experiment and both typically perform tasks similar to the case task in practice. Level of experience did not impact the results.
(10) In developing this case, we consulted the marketing director at a large Australian hotel that is part of a worldwide chain, and a Big 4 Canadian audit partner who audits hotels and who suggested that our case seemed realistic.
(11) For the bonus based on either financial or nonfinancial performance measures, we did not include a target level of performance to be compared against the achieved performance in calculating the bonus. By including target benchmarks, we would have added a layer of complication to the case information that was not central to our study. Instead, we implicitly assumed the incentive weight was set in a way that essentially equated a bonus comprising a weight multiplied by the difference between actual and target performance, with a bonus comprising a smaller weight multiplied by the achieved performance. Thus, mathematically, the target level of performance in our setting was, in effect, zero.
(12) Results did not significantly differ when participants responding incorrectly to the manipulation check were excluded from the analysis.
(13) The results reported in the next subsection are qualitatively unaffected by gender, years of experience, and whether the participant was a partner or manager.
(14) Supplementary analysis that includes strategic systems auditing experience as a control variable does not qualitatively change the results.
(15) Audit risk is significantly and positively correlated with substantive procedures (r = 0.45, p < 0.0001) and significantly and negatively correlated with internal control strength (r = -0.36, p < 0.0001); internal control strength is significantly and negatively correlated with substantive procedures (r = -0.43, p < 0.0001). Internal control procedures is significantly and negatively correlated with substantive procedures (r = -0.33, p = 0.0006).
(16) A majority of participants stated that incentives-based nonfinancial performance measures would not impact the financial statements as much as incentives-based on financial performance measures.
(17) We thank the reviewer who suggested that audit risk may affect the remaining audit-planning judgments.
(18) Type of compensation contract did not significantly affect the control procedures assessment, as was discussed in the Tests of Hypothesis section.
(19) More specifically, one of the authors coded the responses and at the time was aware of the treatment groups. A recoding of the responses by the same author five months later generated virtually identical results. The coding process required very little deliberation as the coding scheme was relatively objective and simple to apply.
American Institute of Certified Public Accountants (AICPA). 2003. AICPA Professional Standards. New York, NY: AICPA.
Banker, R. D., G. Potter, and D. Srinivasan. 2000. An empirical investigation of an incentive plan based on nonfinancial performance measures. The Accounting Review 75 (1): 65-92.
Baron, R. M., and D. A. Kenney. 1986. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology 51: 1173-1182.
Bedard, J. C. 1989. Archival investigation of audit program planning. Auditing: A Journal of Practice & Theory 9 (1): 57-71
Byrne, J. A., L. Lavelle, N. Byrne, M. Vickers, and A. Borrus. 2002. How to fix corporate governance. BusinessWeek (May 6): 68-76.
Canadian Institute of Chartered Accountants (CICA). 1995. Guidance on Control. Criteria of Control Board. Control and Governance--Number 1. Toronto, Canada: CICA.
--. 1997. CICA Handbook. Toronto, Canada: CICA.
Chaiken, S., and D. Maheswaran. 1994. Heuristic processing can bias systematic processing: Effects of source credibility, argument ambiguity, and task importance on attitude judgment. Journal of Personality and Social Psychology 66 (3): 460-473.
Cohen, J. R., G. Krishnamoorthy, and A. M. Wright. 2000. Evidence on the effect of financial and nonfinancial trends on analytical review. Auditing: A Journal of Practice & Theory 19 (1): 27-48.
Demski, J. 1994. Managerial Uses of Accounting Information. Norwell, MA: Kluwer Academic Publishers.
DeZoort, F. T., R. W. Houston, and M. F. Peters. 2001. The impact of internal auditor compensation and role on external auditors' planning judgments and decisions. Contemporary Accounting Research 18(2): 257-281.
Dilla, W. D., and D. N. Stone. 1997. Representations as decision aids: The asymmetric effects of words and numbers on auditors' inherent risk judgments. Decision Sciences (Summer): 709-743.
Eagly, A. H., and S. Chaiken. 1993. The Psychology of Attitudes. Fort Worth, TX: Harcourt Brace College Publishers.
Glover, S. M., J. Jiambalvo, and J. Kennedy. 2000. Analytical procedures and audit-planning decisions. Auditing: A Journal of Practice & Theory 19 (2): 27-45.
Guidry, F., A. J. Leone, and S. Rock. 1999. Earnings-based bonus plans and earnings management by business-unit managers. Journal of Accounting & Economics 26 (1-3): 113-142.
Healy, P. 1985. The effect of bonus schemes on accounting decisions. Journal of Accounting and Economics 7 (1/3): 85-107.
Hirst, D. E. 1994. Auditors' sensitivity to source reliability. Journal of Accounting Research 32 (1): 113-126.
Holthausen, R., D. F. Larcker, and R. G. Sloan. 1995. Annual bonus schemes and the manipulation of earnings. Journal of Accounting and Economics 19 (1): 29-74.
Ittner, C. D., D. F. Larcker, and M. V. Rajan. 1997. The choice of performance measures in annual bonus contracts. The Accounting Review 72 (2): 231-256.
--, --, and M. W. Meyer. 2003. Subjectivity and the weighting of performance measures: Evidence from a Balanced Scorecard. The Accounting Review 78: 725-758.
Jaspars, J., F. D. Fincham, and M. Hewstone. 1983. Attribution Theory and Research. Hinsdale, N J: Lawrence Erlbaum Associates.
Kaplan, R. S., and A. A. Atkinson. 1998. Advanced Management Accounting. Third edition. Englewood Cliffs, NJ: Prentice Hall.
Mock, T., and A. Wright. 1993. An exploratory study of auditor evidential planning judgments. Auditing: A Journal of Practice & Theory 12 (2): 39-61.
--, and --. 1999. Are audit program plans risk-adjusted? Auditing: A Journal of Practice & Theory 18(1): 55-74.
Murphy, K. J. 1998. Executive compensation. In Handbook of Labor Economics, edited by O. Ashenfelter, and D. Card. New York, NY: North-Holland.
Petty, R. E., and D. T. Wegener. 1998. Attitude change: Multiple roles for persuasion variables. In The Handbook of Social Psychology: Volume I, edited by D. T. Gilbert, S. T. Fiske, and G. Lindzey, 323390. Boston, MA: McGraw-Hill.
Robertson, J. C., and W. J. Smieliauskas. 1998. Auditing and Other Assurance Engagements. First Canadian edition. Toronto, Ontario: McGraw-Hill Ryerson Limited.
Rucci, A. J., S. P. Kim, and R. T. Quinn. 1998. The employee-customer-profit chain at Sears. Harvard Business Review (January-February): 82-97.
Waller, W. S. 1993. Auditors' assessments of inherent and control risk in field settings. The Accounting Review 68 (4): 783-803.
Shane S. Dikolli
The University of Texas at Austin
Susan A. McCracken
University of Toronto
Justin B. Walawski
The Institute of Professional Development
|Printer friendly Cite/link Email Feedback|
|Author:||Dikolli, Shane S.; McCracken, Susan A.; Walawski, Justin B.|
|Publication:||Behavioral Research in Accounting|
|Date:||Jan 1, 2004|
|Previous Article:||The role of cognitive and affective conflict in early implementation of activity-based cost management.|
|Next Article:||Clients' expectations on audit judgments: a note.|