Printer Friendly

Management trade-offs of internal control and external auditor expertise.

INTRODUCTION

This study examines whether managers of organizations who are required to have independent audits substitute external auditor expertise for certain closely related internal control mechanisms in their overall control systems. Specifically, we examine whether managers of such organizations trade the control provided by hiring internal auditors and/or employees with accounting expertise for the control provided by working with external auditors with higher levels of industry expertise. By including proxies for the costs of the internal mechanisms in our analyses, we also examine whether these mechanisms are viewed as true economic substitutes, or whether such trade-offs simply represent subtle differences between organizations. Such questions stem from a relatively unexplored paradigm for examining the market for audit services. This paradigm views the external audit as only one of the many control mechanisms managers can choose from in assembling an organization's overall control system (Eilifsen et al. 2001; Knechel 2000). Most extant research examines the demand for external auditing as a function of agency costs alone. Only recently have researchers begun to focus on the impact other control mechanisms might have on the demand for external audits (Anderson et al. 1993; Yeoh and Jubb 2001).

Managers of organizations value the integrity of financial information. Accurate financial information allows them to make better decisions. Communicating this ability to outside parties to whom they are accountable can lead to better perceptions of their management skills and of the organizations they manage. To provide assurance regarding the integrity of financial information--both for themselves and for outside parties--managers establish and maintain control systems comprised of both internal and external mechanisms (Abernethy and Chua 1996; Eilifsen et al. 2001; Simons 1995). Such control systems are likely to reduce the costs associated with poor decisions (e.g., poor performance evaluations, missed investment opportunities), as well as the costs associated with information risk in general (e.g., cost of capital, government oversight). Although there are many different types of control mechanisms (e.g., internal auditors, professionally certified accounting personnel, external auditors, management, boards, audit committees, regulators), their purposes are often similar. It seems reasonable that managers might view some internal and external control mechanisms as substitutes. This notion has been assumed by some researchers (e.g., Francis and Wilson 1988; Simunic 1984; Sunder 1997), yet few studies have examined whether managers trade the control provided by internal mechanisms for the control provided by external mechanisms. And while many studies examine the demand for audit services, almost none include reasonable substitutes in their models as explanatory variables. (Barefield et al. [1993] is a notable exception.)

We posit that managers view external auditors' industry expertise as a substitute for certain internal control mechanisms. We test this using demand models for internal and external control while controlling for the existence and level of agency costs. Using a sample of cities, we examine whether the choice of external auditor is associated with the choices of (1) hiring accounting personnel with varying levels of accounting expertise, and (2) hiring internal auditors, and whether such choices are associated with the costs of the internal mechanisms. Using information obtained from publicly available sources, and from a questionnaire circulated to a sample of cities, we find evidence that managers who hire accounting personnel with relatively low levels of accounting expertise compensate by working with external auditors with relatively high levels of industry expertise. We also find evidence that managers who do not hire internal auditors compensate by working with external auditors with relatively high levels of industry expertise. These trade-offs appear to be linked to the costs of the internal mechanisms--as the cost to hire internal employees increases, managers are less likely to hire certified personnel and/or internal auditors and more likely to engage auditors with high levels of industry expertise.

This study makes three important contributions to our understanding of the market for audit services. First, we provide a framework for examining the demand for auditor industry expertise using agency theory and alternative measures of industry expertise. (1) This allows us to identify several significant determinants of the demand for industry expertise at both the audit-firm level and the individual-office level. Second, we are the first to include proxies for cost in an analysis of the relationship between control mechanisms. This allows us to test whether control mechanisms are true economic substitutes. Finally, we provide evidence that the market for auditing is much more complex than previous studies would suggest. Management's choice of auditor appears to be a function not only of an organization's agency relationships and their related agency costs, but also a function of the availability of other control mechanisms and their related out-of-pocket costs.

The remainder of this paper is presented as follows. The second section discusses the motivation for organizational control and the ability managers have to combine control mechanisms to attain their desired overall level of control. The role of the external audit is discussed, and hypotheses are presented. The third section develops a model of the demand for external control, which is used to test our hypotheses. The fourth section describes the data-gathering process. The fifth section presents results of the hypothesis tests and several supplemental analyses. A general discussion appears in the last section.

THEORY AND HYPOTHESES

The demand for control derives primarily from the presence of agency costs. Agency costs are generated by the conflicting interests and information asymmetries among parties within and without organizations (e.g., Jensen and Meckling 1976). Common relationships associated with agency costs include owner-manager, creditor-manager, and manager-employee. Agency costs created by such relationships can be reduced by either the principal (through monitoring) or the agent (through bonding) (Craswell et al. 1995). Monitoring and bonding mechanisms are part of an organization's overall control system.

Managers have many control mechanisms from which to choose as they regulate the overall level of control in their organizations (Abernethy and Chua 1996). Some of these are internal to the firm (e.g., budgeting, personnel policies, accounting controls, internal auditors, boards, audit committees). Studies have suggested that internal accounting controls may reduce agency costs (e.g., Barefield et al. 1993; Abdel-khalik 1993). Other mechanisms are external to the firm (e.g., independent audits, governmental oversight and regulations). Studies have suggested that independent audits may also reduce agency costs (e.g., Francis and Wilson 1988; Jensen and Meckling 1976; Watts and Zimmerman 1983).

An organization's control system is comprised of the internal and external control mechanisms selected by the manager (Eilifsen et al. 2001; Simons 1995).2 As rational decision makers, managers are likely to select the combination of internal and external control mechanisms that maximizes their profit or utility (Zimmerman 1977). This process is depicted formally by Simunic (1980, 1984):

(1) max [pi] = [PHI]([q.sub.i], [q.sub.e]) - [C.sub.i]([q.sub.i]) - [C.sub.e]([q.sub.e])

In this model, [pi] represents some notion of the manager's profit or utility (Zimmerman 1977). The term [PHI](*) represents the benefit function related to the reduction of agency costs by hiring different quantities of internal and external control; [q.sub.i] represents a vector of internal control mechanisms and their respective quantities; [C.sub.i] represents the cost at which these can be acquired in the marketplace; [q.sub.e] represents a vector of external control mechanisms and their respective quantities; [C.sub.e] represents the cost at which these can be acquired in the marketplace. In this scenario, managers maximize their profit/utility by continuing to hire additional units of internal and external control up to the point where the marginal cost of the next control increment is equal to the marginal benefit of that increment. The resulting quantities, ([q.sub.i]*, [q.sub.e]*) constitute the optimal control mix for the organization. However, there is no reason to believe that ([q.sub.i]*, [q.sub.e]*) are--or should be--the same for each organization, since each organization is likely to have a unique set of cost functions for internal and external control mechanisms. Instead, it is likely that managers can achieve the same level of overall control using different combinations of these mechanisms (Simunic 1984).

Only recently have researchers begun to examine empirically whether such trade-offs occur, specifically examining the trade-off of various internal control mechanisms with external auditor expertise. Using a sample of publicly traded companies, Anderson et al. (1993) find that certain firm characteristics--namely the amount of tangible versus intangible assets--are associated with different mixes of internal and external auditing, suggesting that different control mechanisms are attractive in different circumstances. Using a sample of public companies in Australia, Gul and Lynn (1998) find evidence suggesting that external audit quality and insider ownership appear to be substitutes. In a study similar to this one, Carey et al. (2000) find evidence of substitution between internal and external auditing using a sample of small businesses not required to have audits. However, the less restricted economic decisions of such managers may be peculiar to such settings. Indeed, using a sample of publicly traded companies in the U.S., Ettredge et al. (2000) find no evidence of substitution between internal and external auditing over time.

Two recent studies provide some evidence regarding the relationship between audit committees (another control mechanism) and external auditor industry expertise. Using a sample of public companies in Australia, Yeoh and Jubb (2001) find weak evidence of a negative association between the presence of an audit committee and external auditor industry expertise--consistent with substitution. However, both Yeoh and Jubb (200l) and Abbott and Parker (2000) find evidence of a positive association between audit committee independence and external auditor industry expertise. These conflicting results may arise because of the complex nature of audit committees as control mechanisms. Indeed, the presence of an audit committee is likely to alter the maximization problem in Equation (1) by shifting choices to members of the audit committee, who face different incentive structures than do managers. This is the explanation given by both Yeoh and Jubb (2001) and Abbott and Parker (2000).

Given the large number of potential internal and external monitoring mechanisms mentioned previously, we extend prior research by focusing our study on the substitution of external audit expertise for certain closely related internal control mechanisms. We concentrate on the two mechanisms perhaps most closely related to external auditing through their association with accounting information and accounting systems. The first of these is the hiring of accounting personnel with high levels of accounting expertise. Managers rely on accounting personnel to capture and report relevant financial information useful for decision making, to establish and follow internal control activities designed to protect the integrity of financial information and assets, and to prepare GAAP financial statements for external parties. Clearly, a manager can reduce the risks associated with these functions (e.g., poor decisions, lost assets, fraud, litigation, regulatory investigations, etc.) by hiring highly competent and/or experienced accounting personnel. However, expertise is often costly or unavailable. Since many organizations are required by law to have independent audits each year, some managers may find it more efficient--or even necessary--to hire accounting employees with relatively low levels of accounting expertise and then to compensate by hiring external auditors with higher levels of industry expertise. This provides our first hypothesis:

H1: Managers who hire employees with low levels of accounting expertise are more likely to hire an audit firm with high industry expertise than managers with employees with high levels of accounting expertise.

A second internal control mechanism managers are likely to exchange for more external control is the internal audit function (Abdel-khalik 1993; Ettredge et al. 2000). Although internal auditors perform many tasks that are unrelated to organizations' accounting information systems, many of their responsibilities are linked directly to the production and monitoring of accounting information (Moeller and Witt 1999). One of the primary responsibilities of internal auditors is to test, evaluate, and make recommendations regarding an organization's accounting system and its internal accounting controls. By doing so, internal auditors reduce the risk of fraud and protect assets from theft or loss. External auditors generally perform similar activities with similar benefits, particularly when they rely on an organization's internal controls. Indeed both internal- and external-auditing texts devote attention to the importance of coordination between internal and external auditors to prevent duplication of effort (e.g., Knechel 2000; Moeller and Witt 1999).

The argument made for accounting expertise can also be made for the internal audit function. Internal audit is often costly or unavailable, yet many organizations are required by law to have independent audits each year. Given these circumstances, some managers may find it efficient--or even necessary--to not hire internal auditors and then to compensate by hiring external auditors with higher levels of industry expertise. (3) This provides our second hypothesis:

H2: Managers who do not hire internal auditors are more likely to hire an audit firm with high industry expertise than managers who choose to hire internal auditors.

The arguments for our first two hypotheses suggest that the costs of internal control mechanisms may be a driving force behind any substitution effect. Importantly, in order to be economic substitutes, the hypothesized substitutions must relate to the costs of the mechanisms in question. If the internal control mechanisms and auditor expertise are economic substitutes, internal control costs should be negatively associated with the quantity of internal control demanded but positively associated with the quantity of external control demanded. This leads to our final hypotheses:

H3: Hiring of certified personnel and/or internal audit personnel is negatively associated with the costs of related internal control mechanisms.

H4: External auditor industry expertise is positively associated with the costs of related internal control mechanisms.

METHOD

We test these hypotheses using a model describing a manager's demand for external control in terms of agency costs and the existence of internal control mechanisms that may substitute for external control:

(2) External Control = [florin] (Agency Costs, Internal Control)

According to this model, the amount of external control a manager includes in his control mix is a function of the agency costs created by the organization's structure and by its relationships with outside parties (which raise the overall level of control demanded), as well as the level of control provided by internal mechanisms. One could argue that it is the overall level of control provided by both internal and external mechanisms that is a function of the organization's agency costs. However, we initially follow Simunic (1980, 1984) in assuming that managers generally follow the sequence of hiring first internal and then external control mechanisms. This is consistent with internal salaries being viewed as fixed and with the cost of audit quality beyond a minimum level being viewed as variable. This assumption is relaxed later on.

The Municipal Market

We apply this model to a sample of cities. (4) There are many advantages to using the municipal market to study economic questions such as ours. First, the municipal market makes up a significant part of the U.S. economy. There are over 36,000 cities/townships in the U.S., employing over 11 million people, holding over $800 billion in debt, and spending almost $1 trillion dollars annually (U.S. Census 2002). Second, the demand for control systems is strong in cities. Like managers in other types of organizations, elected officials and city managers demand control because of the agency relationships germane to their organizations (Abdel-khalik 1993; Raman and Wilson 1994; Zimmerman 1977). In the municipal world, the role of owners is played by voters, who elect officials whom they believe will act in their best interests. Elected officials (and their city managers) manage the municipal organization, and hire employees to carry out the day-to-day operations of the city--including providing management with information. Creditors play an integral part in financing the development of municipal infrastructure.

Third, municipal (governmental) accounting is a highly specialized type of accounting. Municipal accounting must be consistent with the governmental accounting standards issued by the GASB in addition to the accounting standards issued by the FASB. Many business schools have separate courses to teach governmental accounting. Indeed, because of the uniqueness of governmental accounting, most states require auditors of cities and other governmental entities to include governmental accounting/audit training as part of their annual CPE regimen. Professional certification for government finance officers is also a growing phenomenon. Hence, municipal expertise is likely to be a valuable characteristic for both municipal accountants and municipal auditors. Fourth, the universe of cities provides a great deal of variation in size and sophistication. This provides an excellent cross-section of organizational structures demanding different levels of quality in their accounting information, and different levels quality in firms they hire to audit this information. (5) This results in a market for municipal audit services that is not dominated by Big 6 (now Big 4) accounting firms. Indeed, the pool of audit firms performing municipal audits is much more heterogeneous than the pool of audit firms performing audits of public companies (Jensen and Payne 2002b). This allows us to separate audit quality effects from the brand name/reputation effects associated with Big 6 firms.

The Model

Hypotheses are tested using two analyses. The first analysis tests directly the relationship between internal control and external control. This is performed by estimating a demand model with external auditor industry expertise (AUDEXP) as the dependent variable and the two internal control mechanisms (CERT and IAUDIT) as explanatory variables, along with a number of control variables:

(3) AUDEXP = [[alpha].sub.o] + [[alpha].sub.1]MFI + [[alpha].sub.2]DEBT + [[alpha].sub.3]ISSUE + [[alpha].sub.4]SIZE + [[alpha].sub.5]MGR + [[alpha].sub.6]SINGLE+ [[alpha].sub.7]BIG6 + [[alpha].sub.8]CERT + [[alpha].sub.9]IAUDIT + [epsilon]

where:

AUDEXP = number of cities audited by the audit office/firm;

MFI = median family income for the city;

DEBT = total debt outstanding for the city;

ISSUE = 1 if the city issued bonds during subsequent three years, 0 otherwise; SIZE = total annual expenditure by the city;

MGR = 1 if the city had a city manager form of government, 0 otherwise;

SINGLE = 1 if the city was required to have a single audit, 0 otherwise;

BIG6 = 1 if the city hired a Big 6 audit firm to perform the audit, 0 otherwise;

CERT = 1 if employee preparing financial statements professionally certified, 0 otherwise; and

IAUDIT = 1 if city employed internal auditors, 0 otherwise.

These variables are described more fully in the next few sections. Negative coefficients on the test variables (CERT, IAUDIT) would be consistent with our hypotheses.

The second analysis investigates our third and fourth hypotheses and examines the role relative costs play in the demand for internal and external expertise using the following model:

(4) CTRL = [[alpha].sub.o] + [[alpha].sub.1]MFI + [[alpha].sub.2]DEBT + [[alpha].sub.3]ISSUE + [[alpha].sub.4]SIZE + [[alpha].sub.5]MGR + [[alpha].sub.6]SINGLE + [[alpha].sub.7]BIG6] + [[alpha].sub.8]ICCOST + [epsilon]

where:

CTRL = use of certified accounting personnel/internal audit staff; number of cities audited by the audit office/firm;

MFI = median family income for the city;

DEBT = total debt outstanding for the city;

ISSUE = 1 if the city issued bonds during subsequent three years, 0 otherwise; SIZE = total annual expenditure by the city;

MGR = 1 if the city had a city manager form of government, 0 otherwise;

SINGLE = 1 if the city was required to have a single audit, 0 otherwise;

BIG6 = 1 if the city hired a Big 6 audit firm to perform the audit, 0 otherwise; and

ICCOST = municipalities cost of living index provided by the American Chamber of Commerce Researchers Association (ACCRA).

The formulation of this model assumes that the organization selects between a pool of internal and external control options to enhance control. Therefore, each dependent variable (AUDEXP and CTRL) is treated similarly with respect to the determinants of the demand for control. Model 4 is also used to introduce a proxy for internal control costs (ICCOST). If the internal control mechanisms and auditor expertise are economic substitutes, internal control costs should be negatively associated with the quantity of internal control demanded but positively associated with the quantity of external control demanded.

Dependent Variables

The model assumes control can be implemented incrementally. This is easily visualized for internal control because of the myriad internal control mechanisms available to the organization. While there are certainly interactions among internal control mechanisms, a manager should generally be able to increase the overall level of control by implementing additional internal mechanisms. The same also may be true of external control mechanisms. However, organizations generally have only one external audit per year. Hence, we assume that managers can regulate the "quantity" of external control in their control mix by hiring different "quality" audit firms. Although auditor quality in not easily observable, managers can usually differentiate quality among audit firms using various proxies likely to be correlated with quality. One generally accepted proxy for auditor quality is audit firm market share (see footnote 6). Thus, managers may regulate the level of external control provided by the auditor by hiring audit firms with more/less industry expertise.

We use two measures of auditor industry expertise (AUDEXP) in our model, each of which emphasizes a different aspect of expertise. The first expertise variable is the total number of municipal audits performed in the sample year by the particular office of the firm performing the audit of the observed municipality (OFF_EXP). This variable emphasizes the industry expertise of the audit team, and ex ante is believed to be most associated with the "quality" used for control purposes (see Carcello et al. 1992). Use of this variable as the primary test variable is strongly motivated by recent research showing that expertise at the office level may be quite different than expertise at the firm level, and that audit firms who are specialists at one level may not be so at the other (Francis et al. 1999). The second expertise variable is the total number of municipal audits (within the same state) performed in the sample year by the particular firm performing the audit of the observed municipality (FIRM_EXP). (6) This variable recognizes that the firm itself may have collective expertise among its partners and managers (Salterio and Denham 1997). Such expertise is likely to be accessible to the audit team and may be used by audit firms in marketing themselves to potential clients.

Test Variables

The variable used to test H1 (CERT) indicates whether the person in charge of preparing the city's financial statements was professionally certified. Certification is a method of distinguishing oneself in an imperfect labor market as someone who has attained a relatively high level of competence and expertise. Hence, professionally certified employees are assumed to possess relatively high levels of competence and accounting expertise. Employees were considered certified if they were certified public accountants (CPA), certified management accountants (CMA), or certified government finance officers (CGFO). The variable used to test H2 (IAUDIT) indicates whether the city had internal audit personnel. Internal auditors contribute to an organization's overall level of control by providing (among other things) assurance regarding the effectiveness of an organization's accounting procedures and accounting controls.

To test H3 and H4, since actual personnel costs for both municipalities and audit firms are unavailable, reasonable proxies for such costs are needed. Since local cost of living indices are (by definition) confined to small geographic areas, we use the local cost of living index (ICCOST) provided by the American Chamber of Commerce Researchers Association (ACCRA) as a proxy for the average cost of hiring municipal accounting and internal auditing employees. We believe this to be a reasonable proxy given that people use such indices to aid in relocation decisions and/or to help with salary negotiations. Labor markets for municipal employees and markets for external audits do not cover the same geographic areas. While cities tend to compete against local businesses and organizations for their employees, audit firms compete for audits over a much larger area. The geographic market in which audit firms compete may encompass many local markets--even entire states or countries. This implies that the average cost of hiring municipal employees and the average cost of hiring external auditors are not necessarily correlated across cities. Therefore, we assume the cost of external audit labor is constant across the sample used for this analysis. (7)

Local labor costs are included in local cost of living indices. Consistent with our assumptions, previous research does not find the local cost of living index to be correlated with the cost of a municipal audit (see Hackenbrack et al. 2000, footnote 12). If our measure of internal control costs is reasonable, H3 and H4 predict that rising internal control costs should be associated with both lower demand for internal auditors and certified accounting personnel and with higher demand for external auditor industry expertise.

Control Variables

Control variables are included in the models to capture variation in the overall demand for control resulting from differences in agency costs. (8) Such variations are likely to affect the level of both internal and external control in the same direction. Several common relationships in municipal organizations increase agency costs and the demand for control. First, costs associated with the voter-official relationship are likely to increase as local voters perceive they have more at stake in the management of the municipality (Zimmerman 1977). In other words, as local voters perceive that they have increasing levels of personal investment in the municipality, they are likely to be more concerned that their elected officials properly manage municipal funds. Voter investment in a municipality is represented by the individual voters' tax burdens. Copley et al. (1995) use the percentage of municipal revenue received through local taxes as a proxy for local investment and find it to be positively correlated with auditor quality (also see Ward et al. 1994). However, we believe individual voters' tax burdens are more likely to be a function of individual incomes within the municipality rather than a function of municipal revenues. Studies have shown a person's income to be a good predictor of whether he or she will vote in a given election (e.g., Leighley and Nagler 1984). Therefore, we use a city's median family income (MFI) as a proxy for the level of voter-official agency costs. (9)

Second, agency costs associated with an official's relationship with creditors are expected to increase as the amount of debt increases (Zimmerman 1977). Several studies have documented the relationship between debt levels and the demand for audit quality (e.g., Francis and Wilson 1988; Barefield et al. 1993; Craswell et al. 1995; Copley et al. 1995). We control for agency costs associated with creditors using two variables. First, we include the city's debt per capita (DEBT) to control for agency costs associated with the debt existing at the time of the audit. Second, we include a variable indicating whether the city issued bonds within three years after the sample year (ISSUE). This variable focuses on the official's motivation to reduce the cost of debt prior to obtaining new funds.

Third, agency costs associated with the official-employee relationship are expected to increase as the municipal organization grows in size (Abdel-khalik 1993; Baber 1983; Evans and Patton 1987). Audit-demand studies consistently show a positive relationship between firm size and the demand for higher-quality audits (e.g., Palmrose 1984; Copley et al. 1995). We use the natural log of total municipal expenditures (SIZE) to control for the agency costs associated with this characteristic. Zimmerman (1977) also suggests that city managers demand more organizational control than elected mayors because city managers are accountable to small groups of elected city council members who can easily dismiss them, whereas mayors can be voted out of office only if relatively large numbers of voters become sufficiently discontented. Both Zimmerman (1977) and Copley et al. (1994) find that city manager forms of government are positively related to audit quality. Hence, we include a variable in the model to control for government type in each city (MGR).

Fourth, officials may demand higher-quality auditors because of governmental regulations and oversight (nonlocal tax burden). For example, cities receiving significant funding from the federal government or any of its agencies are required by law to have a single audit in addition to their regular audit. (10) Because of this, the demand for external control may be higher for cities required to have a single audit. Hence, the model includes a variable indicating whether the municipality was required to have a single audit (SINGLE).

Finally, most previous studies have used the Big 6 versus non-Big 6 dichotomy to proxy for auditor quality. A strong argument can be made that on average, Big 6 firms are likely to have higher-quality auditors than non-Big 6 firms. To control for the demand for these other qualities that may be correlated with our expertise variables, we include a control variable for Big 6 audit firms in the model (BIG6). (11)

DATA

Data were gathered from several sources. We identified all 928 cities with populations greater than 5,000 in several southern states. We obtained copies of the financial statements for these cities for the fiscal years ending between January 1, 1992 and January 1, 1993. Complete or partial financial statements were obtained for 676 (73 percent) of the identified municipalities, either directly from the cities or from the Census Bureau. The audit firm and the specific office of the audit firm were identified by examining the independent auditor's report in each of these financial statements. This information was used to compute expertise variables. Office expertise was determined by counting the number of cities audited by the same office of the accounting firm. Firm expertise was determined by counting the number of cities audited by the same accounting firm. (12) Municipal expenditures and debt figures were also extracted from the financial statements.

We also circulated a questionnaire to all 928 cities to obtain (among other things) information about whether they employed internal audit personnel during the year, whether the person responsible for preparing the financial statements had any professional certifications, and whether the city issued bonds during the three years subsequent to 1992. We received 477 usable responses (51 percent response rate). Population and median family income figures were obtained from the Census Bureau's Internet site; government type was identified using the 1992 Municipal Yearbook. Cities required to have a single audit were identified by the Census Bureau. Eliminating observations with missing variables left 405 usable observations.

Finally, cost of living data for 1992 for 269 of the sample cities were obtained from the American Chamber of Commerce Researchers Association (ACCRA), which publishes quarterly indices for many cities and metropolitan areas. Local cost of living data is limited, and many of the smaller and/or rural towns and cities are not monitored by ACCRA. Hence, not only is the sample smaller for our cost analyses, but the average city size in the reduced sample is somewhat larger that that of the original sample.

RESULTS

Descriptive Data

Table 1 contains the distributions of auditor expertise for the 405 cities with usable observations. The majority of municipal audits are performed by auditors (either firm or office) who only audit one municipality. However, there are clearly some auditors having considerable municipal audit expertise, with 61 offices and 55 firms performing two or more municipal audits.

Table 2 contains descriptive data for the variables used in the models. Panel A shows a wide range of expertise levels for both offices and firms. Expertise levels for offices in the sample ranges from 1 to 7 municipal audits, with a mean of about 2. Expertise levels for firms in the sample ranged from 1 to 17 municipal audits, with a mean of about 4. There is a great deal of variability in the size of the cities in the sample. Total expenditures ranged from $1.1 million to $1.7 billion with a mean of $47.6 million. Total debt ranged from $0 to $3.3 billion with a mean of about $66 million. Panel B shows that while only 19 percent of the sample cities had internal auditors, 77 percent had professionally certified accounting personnel directly in charge of preparing the financial statements. Among the 405 municipalities were 274 CPAs, 50 CGFOs, and 7 CMAs.

Table 3 shows the correlation matrix for the variables used in the models. Some relationships are noteworthy. The two test variables (IAUDIT, CERT) are positively correlated, and as expected, BIG6 is positively correlated with both measures of expertise. The expertise variables themselves (OFF_EXP, FIRM_EXP) are positively correlated, and as predicted, average internal control costs (ICCOST) are negatively correlated with internal control mechanisms (CERT, IAUDIT) and positively correlated with external auditor expertise. Finally, the correlations between the expertise variables and CERT are significantly negative as predicted, however, the correlations between the expertise variables and IAUDIT are not. This suggests that if the predicted relationship exists between internal audit and external auditor expertise, it does so only after controlling for the effects of other variables.

Trade-Off Analysis

Results of the demand-model estimations are shown in Table 4. Models using both office expertise and firm expertise are highly significant, with adjusted [R.sup.2]s of .25 and .54, respectively. Control variables in both models are generally consistent with expectations. Hypothesis 1 predicts that managers who hire accounting personnel with relatively low levels of accounting expertise are more likely to hire external auditors with relatively high levels of industry expertise. This is supported in both models. The test variable, CERT, is significantly negative in each model at p < .01, indicating that municipalities without certified accounting personnel are significantly more likely to have external auditors with higher levels of municipal expertise--both at the office level and at the firm level.

Hypothesis 2 predicts that managers who do not hire internal auditors are more likely to hire external auditors with relatively high levels of industry expertise. This is also supported in both models. The test variable, IAUDIT, is significantly negative in each model at p < .05, indicating that municipalities without internal auditors are significantly more likely to have external auditors with higher levels of municipal expertise--both at the office level and at the firm level. (13)

Cost Analysis

Hypotheses 3 and 4 predict that costs play an important role in managers" decisions regarding which mechanisms to incorporate into their overall control systems. While there is little controversy that internal auditors and more qualified employees are costly control mechanisms, the argument is sometimes made that expert auditors may actually cost less because of the efficiencies they can provide (see Houghton et al. 2001). However other studies indicate a fee premium for expertise (e.g., Ward et al. 1994). (14)

Results of the cost analysis are shown in Table 5. Demand models for the two internal control mechanisms appear in the first two columns. These models were run as logistic regressions and do not contain BIG6 as a control variable, since BIG6 is used to control only for reputation effects that may be confounded with FIRM_EXP and OFF_EXP. (15) Both models are significant. While the demand for certified employees appears to be driven primarily by municipal size and government type, the demand for internal auditors appears to be driven primarily by municipal size and whether the city is in the process of raising capital through bond issues. As predicted in H3, the average cost of hiring municipal employees is significantly negatively associated with both internal mechanisms (p < .05). In other words, as it becomes more costly to hire internal auditors and certified employees, cities are less likely to hire them.

Demand models for external auditor expertise appear in the last two columns. Again, both models are significant, with adjusted [R.sup.2]s of .23 and .49. There are some differences in the effects of the control variables from the original analysis, perhaps because of the reduced sample size. However, as predicted, the average cost of hiring municipal employees is significantly positively associated with external auditors' industry expertise (p < .01). In other words, as it becomes more costly to hire internal auditors and certified employees, cities tend to hire external auditors with higher levels of industry expertise, even when their services are potentially more costly. These results support H4, and are consistent with managers (in cities) viewing external auditors' industry expertise as an economic substitute for internal accounting expertise and internal auditors. (16)

Sensitivity Analysis

Several additional analyses were performed to examine the robustness of these results. First, one can argue that industry expertise is viewed by the market as dichotomous rather than continuous. A firm either has or does not have industry expertise. One can also argue that additional audits in the same industry do not increase auditors' industry expertise by equal amounts. Expertise may increase at a decreasing rate; and large, relatively complex audits may provide more expertise that small, relatively simple audits. To address these issues the analyses were run as logistic regressions after classifying auditors as experts based on their performing a certain number of municipal audits. Results of these analyses were qualitatively the same as those shown in Tables 4 and 5 when the cutoff for office (firm) expertise was 3, 4, 5, or 6 (4, 5, 6, 7, 8, or 9) municipal audits. The analyses were then run using a measure of expertise that places more weight on large audits than on small audits. Expertise was defined as the log of total expenditures made by all other cities audited by the firm performing the observed audit. (17) Using this variable, a firm that audits one large city may appear to have more expertise than a firm that audits several small cities. Results of these analyses were qualitatively the same as those shown in Tables 4 and 5. Hence, our results appear to be robust to alternative measures of specialization used to operationalize expertise.

Second, most audit demand studies use the Big 6 versus non-Big 6 dichotomy to proxy for auditor quality (e.g., Craswell et al. 1995; Simunic 1980). Although Big 6 firms performed only 22 percent of the audits in the sample, and although the models include a control variable for Big 6 firms, we are concerned about the possible confounding of expertise and other Big 6 characteristics such as size and reputation. To test for this, the models were first run after eliminating Big 6 audits from the sample and after dropping BIG6 as a control variable. The remaining models tested our hypotheses using only non-Big 6 firms. Results using the reduced sample were qualitatively similar to the results using the full sample. Second, the trade-off model was run as a logistic regression using BIG6 as the dependent variable and after dropping BIG6 as a control variable. This model was highly significant ([X.sup.2] = 104.1, p < .0001), and the variables MFI, SIZE, and MGR continued to be significant as predicted. However, neither CERT nor IA UDIT were significant at conventional levels, suggesting that managers (of cities) do not compensate for a lack of certified accounting personnel and/or internal auditors simply by hiring Big 6 auditors. Together, these results suggest that the results in Tables 4 and 5 are not driven by the presence of Big 6 firms, nor does there appear to be a confounding of expertise with other Big 6 characteristics.

Finally, as noted earlier, our model follows Simunic (1980) in assuming that the sequence managers generally follow in assembling their optimal control mix is to hire internal mechanisms and then external mechanisms. We believe this to be appropriate. However, the optimization problem suggests that the quantities of internal and external control mechanisms included in an organization's control mix may be determined simultaneously. Therefore, we also performed simultaneous-equation analyses (seemingly unrelated regression) (18) using external auditor expertise and the two internal control mechanisms as dependent variables in the three separate demand equations in Tables 4 and 5. Results of these analyses are shown in Table 6. When office expertise is examined, results are qualitatively similar to the original analysis. Coefficients on CERT and IAUDIT are both significantly negative (p < .01). However, when firm expertise is examined, the simultaneous estimation leads to some differences. DEBT is no longer significant and SIZE becomes so. More importantly, the coefficient on CERT continues to be significantly negative (p < .01); however, the coefficient on IAUDIT (although negative) is no longer significant at conventional levels (one-tailed p = .23). Hence, while the results using simultaneous equations are similar to those of the original analyses, they are somewhat stronger when expertise is measured at the office level rather than the firm level.

CONCLUSION

In this study, we examined whether managers of organizations required to have independent audits trade off the control provided by certain internal control mechanisms for the control provided by external control mechanisms (auditors). Consistent with our hypotheses, it appears that managers who hire accounting personnel with relatively low levels of expertise or who choose not to hire internal audit personnel are significantly more likely to work with external auditors with relatively high levels of industry expertise at the office and firm level. These trade-offs appear to be motivated by increases in the costs of the internal control mechanisms. In other words, the results are consistent with managers viewing external auditor industry expertise as an economic substitute for internal accounting expertise and internal auditing.

These are important results, particularly because of the implications regarding the demand for high-quality auditors. Organizations required to have external audits not only purchase higher-quality audits to reduce agency costs, but they also appear to do so because it is more efficient than implementing additional internal control mechanisms. Of course the significance of this study must be tempered by acknowledging some limitations. Of primary concern is the fact that we use cities to test our hypotheses. Although doing so provides several advantages over using other types of organizations, it limits the generalizability of our results. Municipalities may indeed be considerably different than large publicly traded corporations. Nevertheless, municipalities also may be considerably similar to smaller public corporations, private corporations, and nonprofit organizations. And the underlying motivations for managers to hire control mechanisms are similar in most organizations. While we believe these other types of organizations also exchange internal and external control mechanisms when assembling their overall control systems, whether this is true remains an avenue for future research.

We must also acknowledge our somewhat crude measures of industry expertise and internal audit. While it is likely that auditors performing multiple audits in the same industry will increase their level of expertise in that industry, using the number of audits performed as a direct measure of expertise may be noisy. It is possible that our measure of expertise reflects other correlated audit-firm or market characteristics. While we have provided sensitivity analysis to address these concerns, future research is needed to determine more fully the best proxies for industry expertise. We also acknowledge that the responsibilities of internal auditors are extremely broad and may vary significantly across organizations. Our use of an indicator variable to show whether a city had internal auditors does not allow us to examine such variations. Future research should also examine the role internal auditor characteristics plays in organizational control systems.

In conclusion, this study makes an important contribution to our understanding of the demand for external auditing. From a broad perspective, external audits should not be viewed as separate commodities, but instead as mechanisms to be incorporated as part of an organization's overall control system. Under this perspective, one ought not study the demand for external audit services without considering the availability of--and possible effects of--internal control mechanisms that may act as substitutes for the external audit. In addition, practical implications of this perspective are that in an increasingly competitive market for audit services, successful audit firms increasingly must be able to differentiate their products and their products' quality. This may help to explain why audit firms continue to specialize and tend to focus on industry groups within their organizations.
TABLE 1
Distributions of Expertise Levels for CPA Firms and Offices (a)

Municipal Audits               Municipal Audits
per Office (b)     Frequency     per Firm (c)     Frequency

       1             230              1             190
       2              70              2              62
       3              44              3              25
       4              19              4              20
       5              20             5-7             22
      >5              22             8-10            32
                                     >10             54

Total               405 (a)         Total          405 (a)

(a) The total number of cities for whom the audit firm and office
were identifiable was 676. This represents the value for the
405 observations used in the analyses for Tables 2, 3, and 4.

(b) Number of municipal audits performed by a particular office of
an audit firm.

(c) Number of municipal audits performed by a particular audit firm
within a state.

TABLE 2
Variable Definitions, Sources of Data, and Descriptive Statistics

Panel A: Continuous Variables (n = 405)

                                                          Mean/Median
                                                           (Standard
Variable Name and Definition               Data Source    Deviation)

OFF EXP: the number of municipal audits    Financial       2.01/1.00
the particular office performed during     Statements        -1.52
1992-1993

FIRM EXP: the number of municipal audits   Financial       4.12/2.00
the audit firm performed during 1992-1993  Statements        -4.73
(within the same state)

SIZE (a): 1992 total expenditures of the   Financial    $47,580/$12,662
municipality (in thousands)                Statements     ($132,017)

DEBT (b): total debt outstanding for the   Financial    $65,996/$10,429
municipality at the end of 1992            Statements     ($272,743)
(in thousands)

MFI: median family income for the          Census       $32,074/$30,144
municipality                               Bureau          ($10,824)

ICCOST: cost of living index               ACCRA           99.6/98.8
                                                             (6.2)

Variable Name and Definition               Range

OFF EXP: the number of municipal audits    1-7
the particular office performed during
1992-1993

FIRM EXP: the number of municipal audits   1-17
the audit firm performed during 1992-1993
(within the same state)

SIZE (a): 1992 total expenditures of the   $1,092-$1,745,471
municipality (in thousands)

DEBT (b): total debt outstanding for the   $0-$3,261,930
municipality at the end of 1992
(in thousands)

MFI: median family income for the          $12,818-$84,064
municipality

ICCOST: cost of living index               89.1-114.5

Panel B: Categorical Variables (n = 405)

                                                          Percentage
                                                           (Number)
                                                            where
Variable Name and Definition                Data Source   Value = 1

BIG6: (0,1) valued as 1 if the auditor       Financial     22 (89)
was a Big 6 firm; or 0 if it was not        Statements

ISSUE: (0,1) valued as 1 if the            Questionnaire   69 (280)
municipality issued bonds during the
three years after the observed audit; or
0 if it did not

MGR: (0,1) valued as 1 if the                Municipal     69 (279)
municipality had a city manager hired by     Yearbook
a city council; or 0 if it had an elected
mayor

SINGLE: (0,1) valued as 1 if the             Financial     61 (246)
municipality had a single audit during      Statements
1992; or 0 if it did not

CERT (0,1) valued as 1 if the person at    Questionnaire   77 (310)
the municipality who was primarily
responsible for preparing the 1992
financial statements held a professional
certification (CPA, CMA, CGFO); or 0 if
it did not

IAUDIT: (0,1) valued as  if the            Questionnaire   19 (77)
municipality employed internal audit
personnel during 1992; or 0 if it did not

(a) Models were run using the natural log of this variable.

(b) Models were run using per-capita debt because of the high
correlation between total expenditures and total debt (r = .86.)

TABLE 3
Correlation Matrix for Model Variables (a)

          OFF_    FIRM_
          EXP     EXP     MFI     DEBT    ISSUE   SIZE

FIRM_EXP   .66 *
MFI        .31 *   .24 *
DEBT       .12 *   .13 *  -.00
ISSUE      .08     .08     .07    .14 *
SIZE       .08     .14 *  -.01    .20 *    .18 *
MGR        .19 *   .25 *   .03    .02      .09    -.03
SINGLE     .01     .09    -.19 *  .12 *    .14 *   .21 *
BIG6       .42 *   .73 *   .26 *  .14 *    .14 *   .32 *
CERT      -.16 *  -.11 *  -.07    .06     -.03     .11 *
IAUDIT     .04     .14 *  -.02    .21 *    .11 *   .44 *
ICCOST     .35 *   .39 *   .30 *  .00      .01     .12 *

          MGR     SINGLE  BIG6    CERT    IAUDIT

FIRM_EXP
MFI
DEBT
ISSUE
SIZE
MGR
SINGLE     .04
BIG6       .20 *   .13 *
CERT      -.08     .07    -.03
IAUDIT     .03     .21 *   .29 *   .10 *
ICCOST     .09    -.04     .32 *  -.14 *  -.02

* p-value < .05 for Pearson correlation coefficient.

(a) Variable descriptions are found in Table 2.

TABLE 4
Regression Results for Trade-Off Analysis (n = 405)

Model: AUDEXP = [[alpha].sub.0] + [[alpha].sub.1]MFI + [[alpha].sub.2]
       DEBT + [[alpha].sub.3]ISSUE + [[alpha].sub.4]SIZE +
       [[alpha].sub.5]MGR + [[alpha].sub.6]SINGLE + [[alpha].sub.7]
       BIG6 + [[alpha].sub.8]CERT + [[alpha].sub.9]IAUDIT + [epsilon]

                               Estimated Coefficients
                               (t-statistic) (a) OLS
                               Estimation (n = 405)

                    Predicted  AUDEXP =    AUDEXP =
Variable (b)          Sign     OFF_EXP     FIRM_EXP

Intercept                      -0.32         1.26
                               (0.46)       (0.74)
MFI                     +       0.03 ***     0.03 **
                               (4.48)       (1.72)
DEBT                    +       0.05 **      0.10 *
                               (1.69)       (1.45)
ISSUE                   +      -0.11        -0.43
                               (0.73)       (1.16)
SIZE                    +       0.15 **      0.05
                               (1.89)       (0.27)
MGR                     +       0.32 **      1.00 ***
                               (2.22)       (2.85)
SINGLE                  +      -0.06         0.18
                               (0.37)       (0.48)
BIG6                    +       1.16 ***     7.96 ***
                               (6.09)      (17.28)
CERT                    -      -0.46 ***    -0.91 ***
                               (2.93)       (2.37)
IAUDIT                  -      -0.42 **     -0.86 **
                               (2.07)       (1.75)
Model F-statistic              15.73 ***    54.51 ***
Adjusted [R.sup.2]               .25          .54

***, **, and * significant at p < .01, .05, and .10, respectively.

(a) All tests are one-tailed tests due to the directional hypotheses.

(b) Variable descriptions are found in Table 2.

TABLE 5
Regression Results for Cost Analysis
(n = 269)

Model: CTRL = [[alpha].sub.0] + [[alpha].sub.1]MFI + [[alpha].sub.2]
       DEBT + [[alpha].sub.3]ISSUE + [[alpha].sub.4]SIZE +
       [[alpha].sub.5]MGR + [[alpha].sub.6]SINGLE + [[[alpha].sub.7]
       BIG6] + [[alpha].sub.8]ICCOST + [epsilon]

                            Estimated Coefficients ([c.sup.2] or
                                      t-statistic) (a)

                           Logistic Regression      OLS Regression

                           CTRL =     CTRL =      CTRL =     CTRL =
                            CERT      IAUDIT      OFF_EXP   FIRM_EXP
Variable (c)              (n = 262)  (n = 267)   (n = 269)  (n = 269)

Intercept                  4.17 *    -10.95 ***  -6.60 ***  -12.20 ***
                          (2.71)      (9.85)     (3.97)      (3.05)
MFI                       -0.01       -0.01       0.02 ***    0.01
                          (0.23)      (0.07)     (2.82)      (0.50)
DEBT                       0.17        0.15       0.03        0.06
                          (0.74)      (0.76)     (1.09)      (0.78)
ISSUE                     -0.24       -1.70 ***  -0.21       -0.54
                          (0.45)     (10.22)     (1.01)      (1.09)
SIZE                       0.30 *      1.69 ***   0.19 **     0.09
                          (3.45)     (41.75)     (2.16)      (0.40)
MGR                       -0.68 *     -0.52       0.45 **     1.34 ***
                          (3.43)      (1.22)     (2.32)      (2.87)
SINGLE                     0.09       -0.23      -0.18       -0.41
                          (0.06)      (0.19)     (0.88)      (0.82)
BIG6 (b)                                          0.61 **     6.40 ***
                                                 (2.54)     (10.99)
ICCOST (a)                -0.05 **    -0.06 **    0.06 ***    0.13 ***
                          (4.44)      (3.16)     (3.79)      (3.65)
Model
  [chi square]-statistic  16.41 **   109.33 ***
Model F-statistic                                10.81 ***   33.64 ***
Adjusted [R.sup.2]                                 .23         .49

***, **, and * significant at p < .01, .05, and .10, respectively.

(a) All tests are two-tailed tests in these models except for ICCOST
whose tests are one-tailed due to the directionality, of the hypothesis
test.

(b) BIG6 is included only in the demand models for external expertise
as its purpose is to control for reputation/brand name effects that may
be confounded with firm or office expertise.

(c) Variable descriptions are found in Table 2.

TABLE 6
Simultaneous Estimation of the Demand for Auditor Expertise
Using Seemingly Unrelated Regression (n = 260)

Model A (c): AUDEXP = [[alpha].sub.0] + [[alpha].sub.1]MFI +
             [[alpha].sub.2]DEBT + [[alpha].sub.3]ISSUE +
             [[alpha].sub.4]SIZE + [[alpha].sub.5]MGR +
             [[alpha].sub.6]SINGLE + [[alpha].sub.7]BIG6 +
             [[alpha].sub.8]CERT + [[alpha].sub.9]IAUDIT + [epsilon]

                             Estimated Coefficients
                               (t-statistic) (a)

Variable (b)      Predicted  AUDEXP =    AUDEXP =
                    Sign      OFF_EXP    FIRM_EXP

Intercept                    -1.71        0.92
                             (1.86)      (0.40)
MFI                   +       0.03 ***    0.03 *
                             (3.59)      (1.30)
DEBT                  +       0.05 *      0.08
                             (1.52       (1.05)
ISSUE                 +      -0.40       -0.91
                             (1.87)      (1.75)
SIZE                  +       0.41 ***    0.33 *
                             (3.95)      (1.33)
MGR                   +       0.33 **     1.13 ***
                             (1.67)      (2.36)
SINGLE                +      -0.15       -0.22
                             (0.70)      (0.44)
BIG6                  +       0.67 ***    6.58 ***
                             (2.85)     (11.42)
CERT                  -      -1.44       -3.11 ***
                             (6.74)      (6.01)
IAUDIT                -      -0.73 ***   -0.50
                             (2.64)      (0.75)
System [R.sup.2]               .35         .42

***, **, and * significant at p < .01, .05, and .10, respectively.

(a) All tests are one-tailed tests due to the directional hypotheses.

(b) Variable descriptions are found in Table 2.

(c) The two other simultaneously estimated equations are based on the
demand models in Table 5:

Model B: CERT = [[alpha].sub.0] + [[alpha].sub.1]MFI + [[alpha].sub.2]
         DEBT + [[alpha].sub.3]ISSUE + [[alpha].sub.4]SIZE +
         [[alpha].sub.5]MGR + [[alpha].sub.6]SINGLE + [[alpha].sub.8]
         ICCOST + [[alpha].sub.9]IAUDIT + [[alpha].sub.10]AUDEXP +
         [epsilon]

Model C: IAUDIT = [[alpha].sub.0] + [[alpha].sub.1]MFI +
         [[alpha].sub.2]DEBT + [[alpha].sub.3]ISSUE + [[alpha].sub.4]
         SIZE + [[alpha].sub.5]MGR + [[alpha].sub.6]SINGLE +
         [[alpha].sub.8]ICCOST + [[alpha].sub.9]CERT + [[alpha].sub.10]
         AUDEXP + [epsilon]


We thank the participants of the 2000 American Accounting Association Annual Meeting, workshop participants at the University of Oklahoma, Audrey Gramling. Robert Knechel, Brian Mayhew, Kenneth Reynolds, the associate editor, and two anonymous reviewers for their valuable comments and suggestions.

(1) Significant prior research utilizes audit firm market share or specialization measures to proxy for auditor expertise. Two prior works specifically link auditor industry expertise and audit quality (e.g., Deis and Giroux 1992; O'Keefe et al. 1994). The ability to directly establish this causation is still an empirical question, yet to date this still provides the best empirical measure of auditor expertise. We follow prior research and utilize measures of market share to surrogate for firm industry expertise. (See Gramling and Stone 2001 for an extensive review of the literature.) Additionally, we ensure that our results hold for multiple measures of specialization (see footnote 6).

(2) Some control mechanisms are imposed on organizations by society. For example, many organizations are required to have external audits. Due to possible interactions among control mechanisms, such impositions likely change the marginal costs and benefits of other available control mechanisms, and ultimately may change the optimal set of control mechanisms in a given situation.

(3) These arguments are also valid in the other direction. Organizations that can afford to hire internal auditors or personnel with accounting expertise may not need as much control provided by external auditors.

(4) Cities' involvement with external auditors is not unlike that of other organizations. For most cities, a group of top managers (the city council) is responsible for selecting the auditor, and there appears to be surprisingly little emphasis on hiring local firms to do the audit (Gabhart and Miller 1984). About 40 percent of cities have audit committees that play a role in auditor selection, and less than 10 percent have strict auditor rotation requirements (Jansen and Payne 2002a). While these statistics are probably less likely to hold for smaller cities, we have limited our study to cities with populations of over 5,000 in order to avoid any nuances associated with very small cities.

(5) Hackenbrack et al. (2000) show that as many as 40 percent of the cities examined produced financial statements of such high quality that they were given awards by the Government Finance Officers Association. Further, in response to the increasing demand for municipal accounting information, the GASB recently issued standards requiring cities to produce financial statements similar to those produced in the private sector. Hackenbrack et al. (2000) also show that cities demand auditors at all quality levels, with 48 percent of the cities examined hiring auditors with industry expertise (some with very high levels of expertise) and about 20 percent hiring Big 5 accounting firms.

(6) This variable was aggregated by state to prevent national/international firm from appearing to have extreme levels of expertise simply because of their size. Qualitatively similar results were produced using market share statistics based on the log of total expenditures for each audit firm by state and in total. This is a slight modification of the model extensively utilized in audit firm market concentration research (Gramling and Stone 2001) that utilizes total revenues. In municipal settings, total expenditures are essentially equivalent to total revenues and are easier to capture from the financial statements.

(7) In other words, as the average cost to hire municipal employees increases (decreases), cities should be less (more) likely to hire certified accounting employees and internal auditors. However, there should be no similar direct impact on the likelihood of hiring auditors with high levels of industry expertise because the average cost to hire external auditors may not be affected. This assumption is consistent with the fact that the local cost of living index in our sample of cities ranged from 89.1-114.5, whereas a review of a Big 6 firm's billing rates at the time of the study revealed little variation across offices.

(8) Because of the link between agency costs and inherent risk, proxies for agency costs are likely to also cover inherent risk. However, additional proxies for inherent risk (e.g., bond rating, audit delay, qualified opinions, and municipal growth) were also examined. Results using these models were similar, and explanatory power was not increased. In most cases, the additional variables were not statistically significant.

(9) The variable used by Copley et al. (1995) is available only for cities with populations greater than 50,000, whereas most of the cities in our study have populations less than 50,000. We compared the two variables for a sample of medium to large cities provided by the Census Bureau and found them to be highly correlated (r = .58).

(l0) Governmental and nonprofit entities receiving more than $100,000 (changed to $300,000 in 1996) are required to obtain a Single Audit in addition to the normal GAAS audit. A Single Audit requires procedures and reports in addition to those required by GAAS, including tests of the controls over federal financial assistance funds, tests of compliance with general requirements for all federal financial assistance programs, and tests of compliance with requirements specific to major federal financial assistance programs. It also mandates separate reports regarding (among others) the financial statements, internal controls, compliance with federal laws and regulations, and compliance with the specifications of major federal financial assistance programs.

(11) As noted in the supplemental analysis section that follows later, removal of the BIG6 variable does not qualitatively affect our reported results for the hypothesized variables of interest to this study.

(12) Since our expertise variables are computed using only 676 of the 928 municipalities (73 percent), some understatement in our expertise variables may exist. We believe any such understatement creates bias against our hypotheses.

(13) Results using the log of expertise are qualitatively similar.

(14) Extant research provides inconsistent evidence on fee premiums for auditor expertise. Therefore we examined audit fees for the cities in our sample. Fee models included size, debt, government type, and our expertise variables. In each case, the explanatory power of the model was high (R2 > .65) and the coefficient on expertise was significantly positive. Hence, for our sample, external auditor industry expertise appears to have received a fee premium.

(15) Including BIG6 in the Logistic regression or omitting BIG6 from the OLS regression does not alter the results.

(16) Using ACCRA's cost of labor index provides virtually identical results for H3 and H4.

(17) Firms performing no other audits were assigned a zero value for expertise.

(18) Results using two-stage least squares were qualitatively similar.

REFERENCES

Abbott, L. J., and S. Parker. 2000. Auditor selection and audit committee characteristics. Auditing: A Journal of Practice & Theory 19 (Fall): 47-66.

Abdel-khalik, A. R. 1993. Why do private companies demand auditing? A case for organizational loss of control. Journal of Accounting, Auditing, and Finance 8 (Winter): 31-52.

Abernethy, M. A., and W. F. Chua. 1996. A field study of control system redesign: The impact of institutional processes on strategic choice. Contemporary Accounting Research 13 (Fail): 569-606.

Anderson, D., J. R. Francis, and D. J. Stokes. 1993. Auditing, directorships and the demand for monitoring. Journal of Accounting and Public Policy 12 (Winter): 353-375.

Baber, W. R. 1983. Toward understanding the role of auditing in the public sector. Journal of Accounting and Economics 5 (December): 213-227.

Barefield, R. M., J. J. Gaver, and T. B. O'Keefe. 1993. Additional evidence on the economics of attest: Extending results from the audit market to the market for compilations and reviews. Auditing: A Journal of Practice & Theory 12 (Spring): 74-87.

Carcello, J. V., R. H. Hermanson, and N. T. McGrath. 1992. Audit quality attributes: The perceptions of audit partners, preparers, and financial statement users. Auditing: A Journal of Practice & Theory 11 (Spring): 1-15.

Carey, P., R. Simnett, and G. Tanewski. 2000. Voluntary demand for internal and external auditing by family businesses. Auditing: A Journal of Practice & Theory 19 (Supplement): 37-52.

Copley, P. A., M. S. Doucet, and K. M. Gaver. 1994. A simultaneous equations analysis of quality control review outcomes and engagement fees for audits of recipients of federal financial assistance. The Accounting Review 69 (January): 244-256.

-- J. J. Gaver, and --. 1995. Simultaneous estimation of the supply and demand of differentiated audits: Evidence for the municipal audit market. Journal of Accounting Research 33 (Spring): 137-155.

Craswell, A. T., J. R. Francis, and S. L. Taylor. 1995. Auditor brand name reputations and industry specialization. Journal of Accounting and Economics 20 (December): 297-322.

Deis, D., and G. Giroux. 1992. Determinants of audit quality in the public sector. The Accounting Review 67 (July): 462-479.

Eilifsen, A., W. R. Knechel, and P. Wallage. 2001. Application of the business risk audit model: A field study. Accounting Horizons 15 (September): 193-207.

Ettredge, M., M. Reed, and M. Stone. 2000. An examination of substitution among monitoring devices: The case of internal and external audit expenditures. Review of Quantitative Finance and Accounting 15: 57-80.

Evans, J. H., and J. M. Patton. 1987. Signaling and monitoring in public sector accounting. Journal of Accounting Research 25 (Supplement): 130-155.

Francis, J. R., and E. R. Wilson. 1988. Auditor changes: A joint test of theories relating to agency costs and auditor differentiation. The Accounting Review 63 (October): 663-682.

--, D. J. Stokes, and D. Anderson. 1999. City markets as a unit of analysis in audit research and the re-examination of Big 6 market shares. Abacus 35 (June): 185-206.

Gabhart, D. R. L., and G. Miller. 1984. Survey of local governments' auditor selection practices. Governmental Finance (March): 13-17.

Gramling, A. A., and D. N. Stone. 2001. Audit firm industry expertise: A review and synthesis of the archival literature. Journal of Accounting Literature 20: 1-29.

Gul, F. A., and S. G. Lynn. 1998. Audit quality and debt monitoring as substitutes for management (insider) ownership: Evidence from informativeness of accounting earnings. Working paper, City University of Hong Kong.

Hackenbrack, K., K. L. Jensen, and J. L. Payne. 2000. The effect of a bidding restriction on the audit services market. Journal of Accounting Research 38 (Autumn): 355-374.

Houghton, K. A., C. Dolley, C. A. Jubb, and G. S. Monroe. 2001. Auditor industry specialization and the existence of "premium" and "discount" suppliers: Fee and production function effects. Working paper, University of Melbourne.

Jensen, K. L., and J. L. Payne. 2002a. Audit procurement and the demand for auditor industry specialization. Working paper, University of Oklahoma.

--, and --. 2002b. The effect of the introduction of competitive bidding in the market for audit services. Working paper, University of Oklahoma.

Jensen, M. C., and W. H. Meckling. 1976. Theory of the firm: Managerial behavior, agency costs, and ownership structure. Journal of Financial Economics 3: 305-360.

Knechel, W. R. 2000. Auditing: Assurance and Risk. Second edition. Cincinnati, OH: South-Western College Publishing.

Leighley, J. E., and J. Nagler. 1984. Individual and systematic influences on turnout: Who votes? The Journal of Politics 54:718-739.

Moeller, R., and H. Witt. 1999. Brink's Modern Internal Auditing. Fifth edition. New York, NY: John Wiley & Sons Publishing.

O'Keefe, T. B., R. D. King, and K. M. Gaver. 1994. Audit fees, industry specialization, and compliance with GAAS reporting standards. Auditing: A Journal of Practice & Theory 13 (Fall): 41-55.

Palmrose, Z-V. 1984. The demand for quality-differentiated audit services in an agency-cost setting: An empirical investigation. In Sixth Symposium on Auditing Research, edited by A. R. Abdel-khalik, and I. Solomon, 229-252. Champaign, IL: University of Illinois.

Raman, K. K., and E. R. Wilson. 1994. Governmental audit procurement practices and seasoned bond prices. The Accounting Review 69 (October): 517-538.

Salterio, S., and R. Denham. 1997. Accounting consultation units: An organizational memory analysis. Contemporary Accounting Research 14 (Winter): 669-691.

Simons, R. 1995. Levers of Control. Boston, MA: Harvard Business School Press.

Simunic, D. A. 1980. The pricing of audit services: Theory and evidence, Journal of Accounting Research 18 (Spring): 161-190.

--. 1984. Auditing, consulting, and auditor independence. Journal of Accounting Research 22 (Autumn): 679-702.

Sunder, S. 1997. Theory of Accounting and Control. Cincinnati, OH: South-Western Publishing. U.S. Census Bureau. 2002. Federal, state and local governments. Available at http://www.census.gov.

Ward, D. D., R. J. Elder, and S. C. Kattelus. 1994. Further evidence on the determinants of municipal audit fees. The Accounting Review 69 (April): 399-411.

Watts, R. L., and J. L. Zimmerman. 1983. Agency problems, auditing, and the theory of the firm: Some evidence. Journal of Law and Economics 26 (October): 613-633.

Yeoh, E., and C. A. Jubb. 2001. Governance and audit quality: Is there an association? Working paper, University of Melbourne.

Zimmerman, J. L. 1977. The municipal accounting maze: An analysis of political incentives. Journal of Accounting Research 15 (Supplement): 107-144.

Submitted: January 2002

Accepted: March 2003

Kevan L. Jensen and Jeff L. Payne are both Assistant Professors at the University of Oklahoma.
COPYRIGHT 2003 American Accounting Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Jensen, Kevan L.; Payne, Jeff L.
Publication:Auditing: A Journal of Practice & Theory
Date:Sep 1, 2003
Words:10961
Previous Article:Auditor industry specialization and earnings quality.
Next Article:Audit qualification and timing of earnings announcements: evidence from China.
Topics:

Terms of use | Privacy policy | Copyright © 2021 Farlex, Inc. | Feedback | For webmasters