Effect of outsourcing public sector audits on cost-efficiency
We would like to acknowledge the helpful advice and comments from Jean Bedard, Allen Craswell, Dominic Gasbarro, Ted Mock, Peter Roebuck, Roger Simnett, Donald Stokes, Terry Walter, and participants at various research workshops and research forums, including the International Symposium on Audit Research, the Mid-Year Auditing section Meeting of the American Accounting Association, the Asian–Pacific Conference on International Accounting Issues, the University of Technology Sydney/University of Sydney Summer School, the seminar series of the University of New South Wales and the University of Western Australia. This research was undertaken while the fourth author was employed at Edith Cowan University and the Australian National University. We are grateful to the Office of the Auditor-General in Western Australia for providing data for and feedback to this project.
Abstract
This study compares the cost-efficiency of ‘in-house’ and outsourced to private sector audit supplier arrangements to deliver financial audits in the public sector by examining audit cost-efficiency within the context of the public sector arrangement at one state in Australia (Western Australia). The results for 178 public agencies show that outsourced audits are, in general, more costly than in-house audits, but this result is conditional on the type and size of public agency. Specifically, outsourced audits are more costly than in-house audits for small statutory authority audits, whereas for specialist audits (i.e. hospitals) and large and complex statutory authority audits, the in-house supply is equally efficient as the outsourced service.
1. Introduction
The purpose of this study is to compare the cost-efficiency of ‘in-house’ versus outsourcing to private sector supplier arrangements to deliver financial audits in a public sector environment. This study is motivated from the recent trend to create a more market-based, competitive environment in the public sector audit market.1 This trend has increased the presence and role of contestable models, with greater involvement from private sector suppliers, in the public sector audit market. Although the traditional sole provision of public sector audits by government auditors has been questioned and criticized, particularly in regards to efficiency, no attempt has been made to measure this performance criterion and compare this criterion to various models of public sector audit delivery.
We seek to extend prior literature that examines the operations of private sector suppliers in the public sector audit market. One strand of prior research has examined the role of private sector suppliers and their impact on audit quality in government audits (e.g. Rubin, 1988; Copley and Doucet, 1993; O’Keefe et al., 1994a; Raman and Wilson, 1994; Brown and Raghunandan, 1995; Jensen and Payne, 2005a; Lowensohn et al., 2007). These studies highlight the importance of examining audit quality in the public sector when alternative suppliers are allowed to conduct audits in this sector. Although some evidence suggests that there is demand for high-quality audits at the local government level in the USA (Copley, 1989; Ward et al. 1994; Thorne et al., 2001; Jensen and Payne, 2005a), reports from the National Commission on Fraudulent Financial Reporting (1987) and the General Accounting Office (1985, 1986) indicate that private sector suppliers were providing ‘substandard’ audits to the public sector. Procurement practices play an important role in the public sector to ensure that non-government suppliers deliver the desired audit quality (Copley and Doucet, 1993; Raman and Wilson, 1994; Jensen and Payne, 2005a). Another strand of research has focused on developing audit fee and production cost models in the public sector, motivated by the need to develop models that capture the unique characteristics of the public sector environment (e.g. Rubin, 1988; Copley, 1989; O’Keefe et al., 1994a; Ward et al., 1994; Deis and Giroux, 1996). This study relies on the audit fee and production literature to develop a public sector audit cost model to compare the efficiency of public and private sector auditors in their undertaking of public sector audits – an issue that has received limited attention despite increasing interest in the literature to examine production efficiency in auditing (see Hackenbrack and Knechel, 1997; Knechel and Payne, 2001; Dopuch et al., 2003).
The results from our study should be of interest to public policy-makers in evaluating decisions regarding contestability and the most efficient form of audit delivery in the public sector audit market. The results can also assist public sector auditors to benchmark their performance by comparing it to the private sector contractors’ cost performance. To date, comparison of performance has normally been conducted on a qualitative basis. Finally, this study contributes to the line of inquiry that examines the difference between government auditors and private sector auditors in the public sector audit market by examining the relative cost-efficiency between the two types of suppliers.
2. Institutional setting of public sector audit market in Western Australia
2.1. Institutional environment of the research site
The Western Australian Financial Administration and Audit Act 1985 (FAAA) and the Treasurer's Instructions govern the conduct, operations and funding of the Auditor-General and the Office of the Auditor-General for Western Australia (OAG WA). The State Parliament is specified as the principal client (FAAA, s. 71). The Auditor-General is empowered by the FAAA to audit all types of public sector agencies (i.e. departments, statutory authorities, hospitals, government business enterprises and government-owned corporations), and the accounts of any person or institution in receipt of a specific purpose grant or advance if requested by the Treasurer (FAAA, s. 78). The OAG's tasks for financial statement audits consist of forming opinions, as required by Sec. 93 of the FAAA, in relation to controls and financial statements, and reporting audit findings and significant control weaknesses.
The OAG has the authority to contract qualified auditors to carry out financial audits on its behalf, with the OAG retaining responsibility for the audit opinions (FAAA, s. 82). Although there is no mandate to indicate that the OAG has to outsource the audits, it does so for a significant minority of audits to assist in benchmarking its own in-house performance. Discussions with the relevant staff (the then Auditor-General, three Deputy Auditor-Generals and several senior OAG staff) reveal that audits for departments are not outsourced because of political sensitivities. They added that the majority of statutory authority and hospital auditees are outsourced by rotation and not due to reasons of auditor expertise or technical competency. A small number of statutory authority audits, for which the OAG has no expertise, are outsourced indefinitely. This includes the major state-based insurance agency. Public sector agencies in this (and many other) jurisdictions do not select their external auditors.
Audit engagements are outsourced for a set period of time in Western Australia; usually 3 years, with the possibility of an extension in the contract for a further period of 1 or 2 years. Potential contractors are required to compete for public sector audits through a tender process where the audit agency invites prospective contractors to submit bids. Tenders are evaluated based on the types of expertise, audit methodology and audit plan to be used by the contractors in the engagement, in addition to their tender fees, total budgeted audit hours and a segregation of audit hours by rank level.2 This evaluation process ensures that both quality and price are considered when choosing a supplier.
2.2. Audit quality in the Western Australian public sector audit market
Prior studies have highlighted the importance of examining audit quality in the public sector when alternative audit suppliers (i.e. private sector auditors) are involved (e.g. O’Keefe et al., 1994a; Raman and Wilson, 1994; Brown and Raghunandan, 1995; Houghton and Jubb, 1998). In the public sector, audit quality is defined as compliance with professional standards for reporting and fieldwork (Copley and Doucet, 1993; O’Keefe et al., 1994a; Lowensohn and Reck, 2004). In Western Australia, these professional standards include the Australian Auditing Standards, which became operative for financial statement audits in both the public and private sectors on or after 1 July 1996. Financial audits in Western Australia cover compliance with legislation and attestation of performance indicators, in addition to financial reports.
Consistent with prior studies, the audit agency is assumed to deliver a fixed level of audit assurance (quality) at any given point in time because an audit firm's investments in knowledge are fixed costs and the firm would find it more cost-efficient to deliver audit services to clients who demand a level of audit quality that is consistent with those investments (O’Keefe et al. 1994b; O’Keefe and Westort, 1992). This study also assumes that the audit quality of outsourced audits is similar to that of in-house audits. The audit agency has incentives to ensure that outsourced audits meet the minimum level of audit quality given that the agency is responsible for the audit opinions. Discussions with several senior audit agency staff reveal that the audit agency implements quality control procedures in the tendering and evaluation processes. In addition, the agency reviews the contractors’ audit plans, handles significant issues with the contractors during the course of the audit, and reviews the contractors’ final work. The contractor is required to undertake further work, if necessary, to meet the audit agency's desired audit quality level.3
3. Research question
The relative cost-efficiency of outsourced and in-house audits is expected to be influenced by the audit suppliers’ operating environment, audit specialization and set-up costs.
3.1. The audit suppliers’ operating environment
The audit agency does not have to compete for public sector audits because legislation provides it with a monopoly on the supply of financial audits to public sector agencies. Contractors, in contrast, are subject to competitive pressures, through a competitive tender bidding process.4 Prior studies found that a greater competitive environment reduces the costs of providing services (e.g. Copley and Doucet, 1993; Bandyopadhyay and Kao, 2001; Cavalluzzo, 2002; Jensen and Payne, 2005b). Competition in the private sector market also ensures that contractors continually develop, refine and utilize new audit technologies to minimize costs and fees (Otley and Pierce, 1996; Houghton and Jubb, 1998).
However, it is unclear that contractors have a cost-efficiency advantage over the audit agency in conducting governmental audits. Due to reforms in public sector management and the increasing adoption of competitive structures in the public sector, there is some threat that the audit agency will lose its monopoly of public sector audits (see Joint Committee of Public Accounts, 1989, 1996; Price Waterhouse, 1995; Maddock et al. 1997). This threat prompts the audit agencies to adopt private sector's intellectual capital and customize private sector audit technologies for public sector audits to increase cost-efficiency (e.g. Price Waterhouse, 1995). These arguments suggest that the relative cost-efficiency of the audit agency and contractors might be similar.
3.2. Audit specialization
Audit specialization is another reason why it is unclear for contractors to have a cost-efficiency advantage over the audit agency in conducting governmental audits. Due to legislation, the audit agency has been the traditional supplier of audits in the public sector market. As a result, the agency is able to develop insight into the history and background of the market and the interdependence between public sector agencies, and at the micro level, develop and update client-specific knowledge continuously. This is important in achieving cost-efficiency because industry specialists achieve economies of scale as a result of having a large clientele in the public sector and a greater understanding of the agencies’ institutional and organizational context (Eichenseher and Danos, 1981; Neal and Riley, 2004).
Contractors have limited incentives to invest heavily in developing knowledge and expertise in the public sector industry.5 Reasons behind this include the current limited opportunities for private sector suppliers to enter the public sector audit market and the short-term contracting arrangements of between 3 and 5 years. This short-term contracting arrangement impacts on the contractors’ efficiency because it does not allow them sufficient time to develop substantial client-specific knowledge and, hence, some form of client specialization in the public sector. This limitation is compounded by the audit agency's tender rotation where the contract for a public sector agency is terminated at the end of the contract term and the conduct of financial audit for that agency is taken up again by the audit agency for another 3–4 years. Therefore, it can be argued that the audit agency possesses a greater knowledge and expertise of public sector audits relative to contractors and, for a given level of audit quality, is more cost-efficient than the contractors.
3.3. Set-up costs
Another factor that contributes to the audit agency's greater cost-efficiency lies in the contractors’ set-up costs. These costs are spread over a relatively short time period due to the high turnover of outsourced audits. Although contractors adopt the latest audit technology to manage more efficiently, they do not operate in a similar structural setting as the audit agency (i.e. long-term relationship with agencies) and, therefore, the resulting efficiencies are lost in the set-up costs. Since it is not viable for contractors to absorb these costs every time, they price these costs in their tender fees. Combined with a profit margin component in the tender fees and other factors, the set-up costs will result in outsourced audits being less cost-efficient when compared to in-house audits.
The overall combination of auditors’ operating environment and issues relating to audit specialization and set-up costs suggests that there are two competing arguments on the effect of type of audit arrangement on cost-efficiency in the public sector. Furthermore, the audit specialization and set-up costs arguments are complicated by the fact that while contractors have short-term arrangement with the auditees, they may develop long-term relationship with the audit agency. The conflicting arguments lead to the following research question:
RQ: Are the relative cost-efficiencies of outsourced and in-house audits affected by the audit suppliers’ operating environment, audit specialization and set-up costs?
4. Research design
This study relies on the following audit cost model:
Audit costs = f[size, complexity, risk (audit and business), internal control reliance, Big Five audit firm, industry, audit delivery]
4.1. Variable measurement: dependent variable
4.1.1. Audit costs
Audit costs are defined as the costs of issuing an audit opinion in the public sector. The unit of analysis is costs at the audit engagement level and is assumed to be more efficient if it incurs lower audit costs than another engagement in providing an audit opinion at a specific level of audit quality, for a given level of client-related characteristics.
For in-house audits, audit costs are comprised of the sum of billable labour hours times the standard billing rate of the audit agency staff at each rank level. The standard billing rate is based on the full cost of maintaining an auditor in the field: the auditor's salary, information technology infrastructure and office overhead, among other things, and out-of-pocket costs associated with the audit (e.g. travel, lodging and meals). For outsourced audits, audit costs are comprised of the actual audit fees charged by the contractors and the costs of supervision by the audit agency staff. The supervision costs reflect the additional tasks that are undertaken by the audit agency to ensure that outsourced audits meet the minimum level of audit quality required by the agency.6
4.2. Variable measurement: control variables
4.2.1. Agency size
Total expenditure is expected to be a suitable proxy for agency size. Larger monetary transaction requires more audit effort and activities to verify transactions and ensure that the transactions are properly accounted for and comply with statutes and regulations.
4.2.2. Agency complexity
Discussions with the audit agency staff indicate that a public sector agency's complexity can be proxied by its operational complexity in terms of the breadth and scope of functions and activities performed by the agency. The auditors-in-charge provide an ex post assessment of the agency overall complexity, on a five point Likert-type scale (see also O’Keefe et al., 1994b).7
4.2.3. Risk
This study uses three proxies for audit risk in the public sector. First, this study relies on the audit agency's assessment of the public sector agency's overall financial statement risk to measure audit risk. The overall risk measure is a categorical variable of high (1) or low (0) and was assessed during the planning stage of the audit. Consistent with prior studies, the type of audit opinion for the financial statements is used to measure audit risk (e.g. Rubin, 1988; Copley, 1989; Ward et al., 1994). The audit opinion for performance indicators is also measured because public sector auditors in Western Australia must express an opinion on the agency's performance indicators as part of the financial audit. For both opinions, a qualified audit opinion is coded as ‘1’ and an unqualified audit opinion is coded as ‘0’.8
A proxy for business risk in the public sector is the costs of providing advice to public sector agencies by the audit agency. This proxy captures the concept of political sensitivity in the public sector. The advice generally relates to accountability and performance of public sector agencies, which includes advice on the application of FAAA and Treasury Instructions, accounting standards, performance indicators, and Electronic Data Processing (EDP) audit on major computing issues (OAG WA, 1999). This proxy also includes the attendance of the audit agency staff at the agency's audit committee meetings. The advice provided by the OAG to the agencies is measured as the sum of billable labour hours times the standard billing rate of the OAG staff at each staff level.9
4.2.4. Internal control reliance
The level of reliance on internal control is measured by a three point scale; that is, ‘limited’ (coded as ‘1’), ‘moderate’ (coded as ‘2’) or ‘extensive’ (coded as ‘3’). Respondents recognized that ‘limited’ reliance also means ‘zero’ reliance on internal controls. The auditors-in-charge provided an ex post assessment of the level of reliance on internal controls.
4.2.5. Audit firm size
Consistent with prior studies, audit firm size or affiliation with top-tier audit group is a dichotomous variable and is proxied by the suppliers’ affiliation with a (then) Big Five audit firm (1) or otherwise (0) (e.g. Palmrose, 1986; Raman and Wilson, 1994; Lowensohn et al., 2007). These studies used audit firm size or top-tier affiliation to proxy for audit quality.
4.2.6. Industry
Client industry has been found to affect audit fees and audit hours (see O’Keefe et al., 1994b; Stein et al. 1994; Butterworth and Houghton, 1995; Hackenbrack and Knechel, 1997). Presumably, this is due to differences in audit risk across industry groupings or differences in audit requirements that require different amount of work in the audit (Butterworth and Houghton, 1995) and to differences in the nature of assets, complexity of information systems and the regulatory environment on the strength of control procedures (Stein et al., 1994).
In the context of this study, industry effects on audit costs can be operationalized using different types of agencies in the public sector; namely, statutory authorities and hospitals.10 Hospitals are expected to incur lower audit costs compared to statutory authorities because hospitals are less diverse and less complex in nature in terms of asset base, basic services, management structure, regulatory environment, and conformance to different set of legislation and control procedures. This leads to the expectation that hospital audits require less effort and time, which translates to lower audit costs. The industry measure is a dichotomous variable, with ‘Hospital’ (1) and others (0).
The variables described above are control variables in the audit costs model, designed to control for client-related characteristics. The test variable in the model, ‘audit delivery’, refers to the types of supplier for financial audits in the public sector. It is measured using a dichotomous variable, with outsourced audits (1) and in-house audits (0).
4.3. Data collection
Data were collected from three sources: (1) the OAG's internal records; (2) public sector agencies’ year-end 1998 annual reports; and (3) questionnaires. The year 1998 is important as it was a period of time when the outsourcing arrangement was no longer new and might be thought of as stable. It is also the most recent period for which a full set of data was available to the present authors. The population of financial statement audits for the OAG WA for the 1998 audit cycle is 314 audits (OAG WA, 1999), with 30 per cent of financial audits outsourced to private sector auditors. The sample consists of 178 financial audits after eliminating some audits from the population, as presented in Table 1.11
| Total financial statement audits for year-end 1998 | 314 |
|---|---|
| less: | |
| Treasurer's annual statement | 1 |
| Financial statement audits conducted under legislation other than FAAA | |
| Subsidiaries | 18 |
| Corporatized entities | 3 |
| Local cemetery boards (conducted under the Cemeteries Act) | 11 |
| Request audits | 10 |
| Agencies not active | 11 |
| Agencies that ceased operations during year-end 30 June 1998 and 1999 | 23 |
| Information relating to independent variables was not determinable due to non-response to the questionnaires | 14 |
| Department audits | 45 |
| Total sample | 178 |
4.4. Audit cost model
()where:
Lncost = natural logarithm of total audit costs
Lnexp = natural logarithm of total operating expenditure
Complexity = breadth and scope of agency's functions and activities
Risk = overall financial statement risk (1 = high, 0 = low)
FSOpinion = audit opinion for financial statement (1 = qualified opinion, otherwise 0)
PIOpinion = audit opinion for performance indicators (1 = qualified opinion, otherwise 0)
AdviceCost = total costs of advice provided to agencies by the OAG
ICReliance = internal control reliance on a scale from 1 (limited) to 3 (extensive)
Big5 = (1 = Big Five audit firms, otherwise 0)
Agency type = (1 = hospital, 0 = otherwise)
Audit delivery = (1 = outsourced, 0 = in-house)
ε = error term
5. Empirical results
5.1. Descriptive statistics
Table 2 shows the distribution for agency type and the in-house versus outsourced subsamples. Table 3 presents the descriptive statistics for the independent and dependent variables. To test for univariate differences between (i) the in-house and outsourced and (ii) statutory authority and hospital subsamples, the Mann–Whitney U and χ2-tests are used for continuous and dichotomous variables, respectively.12 Outsourced audits are significantly more complex and more likely to be associated with qualified opinion for performance indicators compared to in-house audits. Statutory authority audits are significantly different in all aspects except for qualified audit opinion compared to hospital audits. Table 4 presents a matrix of Spearman (rank-order) correlation coefficient between the independent and dependent variables.13
| Agency type | Total sample | In-house | Outsourced |
|---|---|---|---|
| Statutory authority | 120 (67%) | 84 (78%) | 36 (51%) |
| Hospital | 58 (33%) | 24 (22%) | 34 (49%) |
| Total | 178 (100%) | 108 (100%) | 70 (100%) |
| Total sample (n = 178) | Subsamples | ||||
|---|---|---|---|---|---|
| Audit delivery | Agency type | ||||
| In-house (n = 108) | Outsourced (n = 70) | Statutory authority (n = 120) | Hospital (n = 58) | ||
| Audit costs ($A) | |||||
| Mean | 15 785 | 13 532 | 19 260 | 19 608 | 7 876 |
| Median | 8 408 | 7 814 | 9 718 | 11 272 | 5 229 |
| Standard deviation | 23 814 | 14 710 | 33 151 | 27 940 | 6 065 |
| Log audit costs | |||||
| Mean | 9.164 | 9.086 | 9.285 | 9.363 | 8.753 |
| Median | 9.037 | 8.964 | 9.181 | 9.330 | 8.562 |
| Standard deviation | 0.937 | 0.907 | 0.974 | 0.995 | 0.6325 |
| Significance | 0.166 | 0.000 | |||
| Total operating expenditure ($A) | |||||
| Mean | 33 191 180 | 20 003 639 | 53 537 671 | 46 383 480 | 5 896 776 |
| Median | 3 465 000 | 3 670 000 | 3 298 500 | 5 257 500 | 1 549 500 |
| Standard deviation | 121 589 340 | 56 537 069 | 179 652 730 | 146 315 100 | 9 461 247 |
| Log total operating expenditure | |||||
| Mean | 8.334 | 8.204 | 8.535 | 8.602 | 7.781 |
| Median | 8.150 | 8.208 | 8.101 | 8.567 | 7.346 |
| Standard deviation | 1.933 | 1.859 | 2.040 | 2.114 | 1.345 |
| Significance | 0.265 | 0.008 | |||
| Level of complexity (1–5) | |||||
| Mean | 2.666 | 2.551 | 2.843 | 2.554 | 2.897 |
| Median | 3.000 | 3.000 | 3.000 | 2.000 | 3.000 |
| Standard deviation | 0.670 | 0.700 | 0.581 | 0.764 | 0.307 |
| Significance | 0.004 | 0.001 | |||
| Total advice to agency ($) | |||||
| Mean | 1 242 | 1 397 | 1 004 | 1 680 | 337 |
| Median | 355 | 490 | 276 | 575 | 113 |
| Standard deviation | 3 132 | 2 708 | 3 702 | 3 723 | 537 |
| Significance | 0.416 | 0.007 | |||
| Internal control reliance (1–3) | |||||
| Mean | 1.719 | 1.694 | 1.757 | 1.633 | 1.900 |
| Median | 2.000 | 2.000 | 2.000 | 2.000 | 2.000 |
| Standard deviation | 0.542 | 0.555 | 0.523 | 0.607 | 0.307 |
| Significance | 0.452 | 0.002 | |||
| Overall high risk of financial statement | |||||
| Percentage | 24.72 | 25.0 | 24.3 | 35.8 | 1.7 |
| Significance | 0.914 | 0.000 | |||
| Qualified financial statement audit opinion | |||||
| Percentage | 2.81 | 2.8 | 2.9 | 2.5 | 3.4 |
| Significance | 0.975 | 0.720 | |||
| Qualified performance indicator audit opinion | |||||
| Percentage | 19.10 | 12.0 | 30.0 | 3.3 | 51.7 |
| Significance | 0.003 | 0.000 | |||
| Big Five (percentage) | 6.18 | — | 15.71 | 9.2 | — |
| Outsourced audits (percentage) | 39.33 | — | — | 30.00 | 59.00 |
| Lncost | Lnexp | Complexity | Risk | FSOpinion | PIOpinion | AdviceCost | ICReliance | Big5 | Agency type | |
|---|---|---|---|---|---|---|---|---|---|---|
| Lnexp | 0.835*** | |||||||||
| Complexity | 0.350*** | 0.387*** | ||||||||
| Risk | 0.437*** | 0.346*** | 0.187 | |||||||
| FSOpinion | 0.090 | 0.080 | 0.029 | 0.060 | ||||||
| PIOpinion | –0.056 | –0.097 | 0.186 | –0.212*** | 0.177 | |||||
| AdviceCost | 0.606*** | 0.600*** | 0.238*** | 0.376*** | 0.032 | –0.102 | ||||
| ICReliance | 0.270*** | 0.327*** | 0.677*** | 0.147 | –0.035 | 0.112 | 0.171 | |||
| Big5 | 0.290*** | 0.283*** | 0.257*** | 0.286*** | –0.044 | –0.125 | 0.225*** | 0.166 | ||
| Agency type | –0.321*** | –0.225*** | 0.306*** | –0.371*** | 0.027 | 0.577*** | –0.309*** | 0.259*** | –0.178 | |
| Audit delivery | 0.078 | 0.038 | 0.229*** | –0.008 | 0.002 | 0.223*** | –0.121 | 0.061 | 0.319*** | 0.275*** |
- *** Denotes significance at p < 0.01 (two-tailed). Lncost is natural logarithm of total audit costs; Lnexp is natural logarithm of total operating expenditure; Complexity is breadth and scope of agency's functions and activities; Risk is overall financial statement risk (1 = high, 0 = low); FSOpinion is audit opinion for financial statement (1 = qualified opinion, otherwise 0); PIOpinion is audit opinion for performance indicators (1 = qualified opinion, otherwise 0); AdviceCost is total costs of advice provided to agencies by the Office of the Auditor-General; ICReliance is internal control reliance on a scale from 1 (limited) to 3 (extensive); Big5 takes the value 1 if Big Five audit firms, and 0 otherwise; Agency type takes the value 1 if hospital, and 0 otherwise; Audit delivery takes the value 1 if outsourced, and 0 if in-house).
5.2. Results of regression analysis
The research question addresses whether the relative cost-efficiencies of outsourced and in-house audits are affected by the audit suppliers’ operating environment, audit specialization and set-up costs. Panel A of Table 5 shows the results of the multiple regression analysis for the total sample.
| Panel A: Total sample | Panel B: Subsamples | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| Standard coefficient | p-value | Agency type | Size | ||||||||
| Statutory authority | Hospital | Large | Small | ||||||||
| Standard coefficient | p-value | Standard coefficient | p-value | Standard coefficient | p-value | Standard coefficient | p-value | ||||
| (Constant) | 0.000 | 0.000 | 0.000 | 0.000 | 0.000 | ||||||
| Lnexp | + | 0.693 | 0.000 | 0.598 | 0.000 | 0.661 | 0.000 | 0.644 | 0.000 | 0.613 | 0.000 |
| Complexity | + | 0.101 | 0.026 | 0.175 | 0.004 | –0.019 | 0.439 | 0.132 | 0.061 | 0.307 | 0.015 |
| Risk | + | 0.131 | 0.001 | 0.136 | 0.002 | 0.009 | 0.451 | 0.081 | 0.127 | 0.370 | 0.000 |
| FSOpinion | + | –0.007 | 0.424 | –0.028 | 0.267 | 0.039 | 0.300 | –0.052 | 0.223 | 0.000 | 0.498 |
| PIOpinion | + | 0.108 | 0.007 | 0.041 | 0.204 | 0.156 | 0.019 | 0.068 | 0.240 | 0.200 | 0.011 |
| AdviceCost | + | 0.072 | 0.045 | 0.084 | 0.064 | 0.132 | 0.075 | 0.098 | 0.122 | 0.097 | 0.117 |
| ICReliance | – | 0.011 | 0.404 | 0.069 | 0.101 | –0.153 | 0.120 | 0.036 | 0.322 | 0.075 | 0.273 |
| Big5 | + | 0.000 | 0.498 | –0.030 | 0.275 | — | — | –0.022 | 0.398 | 0.118 | 0.063 |
| Agency type | – | –0.211 | 0.000 | — | — | — | — | –0.103 | 0.138 | –0.593 | 0.001 |
| Audit delivery | ? | 0.063 | 0.102 | 0.114 | 0.022 | –0.082 | 0.304 | 0.050 | 0.513 | 0.172 | 0.037 |
| Adjusted R2 | 0.80 | 0.81 | 0.74 | 0.66 | 0.56 | ||||||
| F-statistic | 72.903 (p < 0.001) | 55.814 (p < 0.001) | 21.610 (p < 0.001) | 18.350 (p < 0.001) | 12.316 (p < 0.001) | ||||||
| Sample size | 178 | 120 | 58 | 89 | 89 | ||||||
- a A one-tailed test is used for significance except for the test variable Audit delivery. Lncost is natural logarithm of total audit costs; Lnexp is natural logarithm of total operating expenditure; Complexity is breadth and scope of agency's functions and activities; Risk is overall financial statement risk (1 = high, 0 = low); FSOpinion is audit opinion for financial statement (1 = qualified opinion, otherwise 0); PIOpinion is audit opinion for performance indicators (1 = qualified opinion, otherwise 0); AdviceCost is total costs of advice provided to agencies by the Office of the Auditor-General; ICReliance is internal control reliance on a scale from 1 (limited) to 3 (extensive); Big5 takes the value 1 if Big Five audit firms, and 0 otherwise; Agency type takes the value 1 if hospital, and 0 otherwise; Audit delivery takes the value 1 if outsourced, and 0 if in-house).
Table 5 indicates that the coefficients for logarithm total expenditure, complexity, risk and audit opinion for performance indicators are positive and significant in the predicted directions. Audit qualification for financial statements is not significant at the 5 per cent level. This finding is similar to prior fee studies in the US public sector (see Rubin, 1988; Copley, 1989; Ward et al., 1994). The level of reliance on internal control does not significantly affect audit costs. This result is consistent with several prior studies (e.g. O’Keefe et al., 1994b; Hackenbrack and Knechel, 1997). The Big5 variable is not significantly associated with audit costs. This result provides support to the earlier discussion that Big Five audit firms have limited incentives to provide a higher audit quality than required by the audit agency and the agency selects the contractor that meets, rather than exceeds, its quality level. Alternatively, the direct application of audit fee premiums for the Big Five in the private sector does not automatically translate into this particular market. Big Five audit firms could not command a premium for their reputation in this market, due to the tendering policies of the audit agency. For the Agency type variable, the evidence suggests that hospital audits are associated with lower audit costs compared to statutory authorities.
The coefficient for the test variable, audit delivery, is not significantly associated with audit costs in the total sample. We performed further analysis by splitting the total sample by agency type (statutory authority and hospital) and by size to provide further insights into the results of the total sample. For the size subsample, we divided the total sample into large and small samples by splitting the total sample at the median, using total expenditure as a proxy for size. Results for the two subsamples in Panel B of Table 5 suggest that the audit agency (the in-house supplier) is more cost-efficient than contractors when undertaking statutory authority audits and small audits. There is no cost difference between the audit agency and contractors for hospital audits and large audits.
Given the results in Panel B of Table 5, there is a possibility that the audit agency's greater cost-efficiency in auditing statutory authority audits compared to contractors applies to small statutory authority audits only (i.e. the results in the statutory subsample are driven by small audits). To test this proposition, we split the statutory authority sample into large and small subsamples, once again splitting the sample at the median, using total expenditure as a proxy for size. Table 6 shows that the audit agency is more cost-efficient than contractors for small statutory authority audits only.
| Statutory authority | |||||
|---|---|---|---|---|---|
| Large | Small | ||||
| Standard coefficient | p-value | Standard coefficient | p-value | ||
| (Constant) | 0.000 | 0.000 | |||
| Lnexp | + | 0.553 | 0.000 | 0.510 | 0.000 |
| Complexity | + | 0.179 | 0.052 | 0.187 | 0.029 |
| Risk | + | 0.002 | 0.491 | 0.375 | 0.000 |
| FSOpinion | + | –0.120 | 0.127 | –0.052 | 0.283 |
| PIOpinion | + | 0.084 | 0.254 | 0.091 | 0.146 |
| AdviceCost | + | 0.166 | 0.095 | 0.074 | 0.209 |
| ICReliance | – | 0.019 | 0.421 | 0.079 | 0.190 |
| Big5 | + | 0.061 | 0.299 | 0.060 | 0.253 |
| Audit delivery | ? | –0.028 | 0.796 | 0.264 | 0.006 |
| Adjusted R2 | 0.62 | 0.63 | |||
| F-statistic | 11.779 (p < 0.001) | 12.321 (p < 0.001) | |||
| Sample size | 60 | 60 | |||
- a A one-tailed test is used for significance except for the test variable Audit delivery. Lncost is natural logarithm of total audit costs; Lnexp is natural logarithm of total operating expenditure; Complexity is breadth and scope of agency's functions and activities; Risk is overall financial statement risk (1 = high, 0 = low); FSOpinion is audit opinion for financial statement (1 = qualified opinion, otherwise 0); PIOpinion is audit opinion for performance indicators (1 = qualified opinion, otherwise 0); AdviceCost is total costs of advice provided to agencies by the Office of the Auditor-General; ICReliance is internal control reliance on a scale from 1 (limited) to 3 (extensive); Big5 takes the value 1 if Big Five audit firms, and 0 otherwise; Agency type takes the value 1 if hospital, and 0 otherwise; Audit delivery takes the value 1 if outsourced, and 0 if in-house).
We also performed a sensitivity analysis to observe the impact of audit agency's supervision activities, and hence costs, on the relative cost performance between the agency and contractors. We ran the same regression model, but removed supervision costs from our definition of audit costs, on the statutory authority sample. We found no difference in the audit agency's and contractors’ performance for both large and small subsamples. This result indicates that supervision costs were a significant portion of small statutory authority audits but have no impact on the large statutory authority audits. Therefore, it appears that the audit agency is as cost-efficient as outside suppliers in this market for audit services when the supervision costs are ignored. The supervision costs represent quality control costs incurred by the audit agency to ensure that the quality of assurance services supplied by external suppliers is at an appropriate level. Presumably, there is some duplication of these costs as the external supplier would also be incurring supervision costs for quality control purposes.
The results from the subsamples indicate that the audit agency is more cost-efficient compared to contractors when undertaking small statutory authority audits. The results also show that there is no difference between the audit agency and contractors’ performance in hospital audits and large statutory authority audits. The combination of high set-up costs experienced by the contractors, due to client rotation, and the audit agency's expertise in the public sector are two reasons for the agency's greater cost-efficiency in auditing small statutory authority audits. However, for hospitals, the no difference result could be attributed to the contractor's expertise in auditing a similar agency type outside the public sector and the ability to transfer that expertise to the public sector. The contractors’ experience with hospital audits outside the public sector and the similarity in the hospital's organizational structure and operations in both sectors allow contractors to utilize and transfer their experience of hospital audits between both sectors. Therefore, the contractors possess similar auditing expertise as the audit agency for hospital audits. Although one could argue that there is no equivalent of statutory authorities in the private sector and, therefore, the contractors do not have an advantage over the OAG in terms of expertise for statutory authorities, the contractors have the technology and economies of scale when auditing large clients. These factors override the high set-up costs and their inexperience when auditing statutory authorities.
6. Concluding remarks
In many jurisdictions in the Western world, there are, periodically, moves towards a more competitive-based environment in the public sector with a push for a greater level of outsourcing of work with government agencies. Audit offices in some jurisdictions have not been exempt from this. This study provides empirical evidence that outsourced audits do not necessarily translate into lower cost audits when compared to in-house audits. In fact, this study suggests that outsourced audits are more costly than in-house audits for a specific agency type of a particular size (i.e. the small statutory authority audits). Assuming that audit quality is constant, there is no benefit in outsourcing as the government (i.e. audit agencies) can conduct financial audits as cost-efficiently as the contractors, and has the advantage of sidestepping discussions and debates on independence and accountability to parliament. Therefore, all small audits are best done in-house as there is little evidence to support efficiencies in outsourcing. The only possible benefit of outsourcing may be in differences in audit quality and the information acquired for benchmarking purposes, but this is not tested in the present study.
Given the results of this study, public policy-makers may wish to exercise further care when introducing competitive mechanisms, such as outsourcing in the public sector audit market. Deciding on the mix of in-house and outsourcing forms of audit delivery is important because the tendering process is costly and, as noted by Craswell (1997, p. 17), ‘the organization calling tenders incurs not only administrative costs but also risks discouraging tenderers for whom the probability of success is a function of the number of firms bidding’. This study points to the importance of evaluating the current audit arrangement and its impact on competition. For competition to have an effect on audit costs and quality, regulators need to review the current incentive structures, such as the procurement practices of the audit agencies, the length of the contract terms, and institutional factors that affect the suppliers’ incentives in the public sector audit market, so that suppliers are motivated to deliver low-cost and/or high-quality audits.
There are several limitations to this study. First, auditors’ perceptions of agency complexity and internal control reliance are measured using a single-item instrument and, as such, there is a potential for measurement error (Hair et al., 1998). Another limitation is the small sample size of the total sample and subsamples, which could result in either too little statistical power for the test to realistically identify significant results or too easily an ‘overfitting’ of the data such that the results are artificially good because they fit the sample very well (Hair et al., 1998). In addition, our sample is comprised only of public sector entities located in Western Australia, and data from other institutional settings in other jurisdictions might not yield similar results. Finally, there have been changes in private and public sector audit markets and auditing standards since our sample period, which might have an impact on audit costs. Therefore, caution is required in interpreting the findings.
The audit agency in Western Australia still maintains its policy towards outsourcing in that its total expenditure from 2003 to 2007 reflects usage of private sector audit resources, with percentages ranging from 17 to 29 per cent (OAG WA, 2007). The OAG in other jurisdictions also maintain their policy towards outsourcing (Audit Office of New South Wales, 2007; Queensland Audit Office, 2007; Victorian Auditor-General's Office, 2007). Although the ‘Big Five’ and ‘Andersen methodology’ are no longer applicable, the structure of the public sector audit market remains unchanged (i.e. procurement practices of the audit agencies, the length of the contract terms and institutional factors that affect the suppliers’ incentives in the public sector audit market), as evident from the information available in the OAGs’ Annual Reports and on the websites of the OAG in New South Wales, Victoria and Queensland, on the contracting processes. This study's relevance is to point out the impact of this audit market on competition and the importance of evaluating the market for competition to have an effect on audit costs and quality. The pressure to be cost-efficient in the public sector is still valid (due to the lingering influence of the new public management reforms) and alternative forms of carrying out public activities are still being considered.
