Printer Friendly

Auditor industry specialization, client bargaining power, and audit pricing.

INTRODUCTION

The purpose of this study is to examine audit pricing by Big 6 auditors, and the joint effects of industry specialization and client bargaining power on audit fees. Drawing on Porter's (1985) analysis of corporate strategy, auditor industry specialization is viewed as a differentiation strategy whose purpose is to provide auditors with a sustainable competitive advantage over nonspecialist auditors. Porter (1985, 14) explains differentiation as follows:
   In a differentiation strategy, a firm seeks to be unique along some
   dimensions that are widely valued by buyers. It seeks one or more
   attributes that many buyers in the industry perceive as important,
   and uniquely positions itself to meet those needs. It is rewarded
   for its uniqueness with a premium price.


While differentiators can potentially charge a premium if buyers value their services, the ability to do so depends on the bargaining power of clients (Porter 1985, 9). Prior research on auditor industry specialization has ignored the role of bargaining power in the pricing process and simply assumed that industry specialists have the ability to set higher prices.

There has been relatively little research on industry specialization and audit fees. There are two published studies that report an industry specialization premium: Craswell et al. (1995), who use Australian data from 1987, and DeFond et al. (2000), who use 1992 data from Hong Kong. There are also two published U.S. studies, both of which fail to find a premium for industry specialization: Palmrose (1986), who uses a sample of utilities, and Pearson and Trompeter (1994), who use a sample of insurance companies. More recently, Ferguson and Stokes (2002), who use Australian data from the 1990s, do not find strong support for industry-specialist premiums. Conversely, Ferguson et al. (2003) re-examine the same 1998 data as that of Ferguson and Stokes (2002) and do find evidence of industry-specialist premiums when city-level measures of specialization are considered. The U.S. studies are narrowly focused on two regulated industries, whereas Craswell et al. (1995), DeFond et al. (2000), Ferguson and Stokes (2002), and Ferguson et al. (2003) test a broader cross-section of industries. However, the Australian and Hong Kong audit markets are much smaller than the U.S. market, which makes the notion of industry specialization (whether measured on a national or city level) more problematic given the relatively small industry clienteles in these countries. In addition, three of these six studies, including both U.S. studies, use data from the mid-1980s and market conditions appear to have changed considerably since that time. (1) For all of these reasons, the prior findings in these studies may not generalize to the current U.S. audit market.

Our empirical tests use more recent U.S. data, and the results are consistent with predictions from the corporate strategy literature. Using a sample of 651 publicly listed companies audited by Big 6 auditors, we find that fees are higher for Big 6 industry specialists. However, these results only hold for the lower half of company size in the study (assets < $123 million). In contrast, for larger-sized companies, we find no evidence of a fee premium for auditor industry specialization, and evidence that audit fees actually decrease as a company's size increases relative to their auditor's clientele in that particular industry. Together these results are consistent with a differentiation premium for industry specialization, but only for smaller clients having low bargaining power with their auditors. When bargaining power is greater, there is no specialization premium and evidence of a fee discount.

The remainder of the study is organized as follows. First we use Porter's (1985) corporate strategy framework to analyze industry specialization and audit pricing. Next the empirical design and sample is introduced followed by empirical results and sensitivity analyses. The study concludes with a discussion of the results and the limitations of the study.

CORPORATE STRATEGY, INDUSTRY SPECIALIZATION, AND AUDIT PRICING

In the early 1990s, large accounting firms began to change their organizational structures, which, at that time, were built around the traditional product lines of auditing, taxation, and consulting. The firms restructured their activities around broad industry sectors (Public Accounting Report 1993, 1995), and began marketing their expertise around industry specializations (De Belde 1997; Hogan and Jeter 1999; Solomon et al. 1999). Today, the Big 4 firms continue to promote their industry-based expertise. This is apparent from the way firms characterize themselves on their web pages. For example, the KPMG website at http://www.kpmg.com states:
   Specialization by sector is fundamental to our approach. We believe
   that we cannot truly add value for our clients without a thorough
   understanding of their industry throughout the world. This is why we
   invest in continuously improving our knowledge of the industries we
   serve.


The KPMG website goes on to list nine broad industry specializations: banking and insurance, industrial, automotive, chemicals, pharmaceuticals, consumer markets, electronics, communications, energy and natural resources. Industry focus is also evident in the business-risk model of auditing developed in the 1990s, which builds on knowledge of the client's industry and business processes (Bell et al. 1997).

Why did an emphasis on industry specialization emerge in the 1990s? Following the deregulation of the U.S. audit market in the late 1970s, accounting firms initially faced increased competition (Maher et al. 1992). With increased competition came the need for strategies to achieve competitive advantage. Porter's (1985) analysis of corporate strategy is helpful in understanding this market development and the responses of accounting firms.

Following deregulation in the late 1970s, each of the (then) Big 8 accounting firms initially attempted to increase their market share by taking clients from other Big 8 firms. This strategy was of course a zero-sum game, leading to price wars and declining industry profitability during the 1980s (Wall Street Journal 1985a, 1985b, 1987; Work 1985). (2) If auditing is viewed as homogeneous, and switching costs are relatively low, then accounting firms are likely to engage in price competition to gain market share. The strategy literature views product differentiation as a way to avoid such head-on price competition. If an accounting firm can differentiate its product from other accounting firms, and if buyers value the differentiation, then firms can potentially earn a fee premium on their differentiated services. Porter (1985, 130) says, "A successful differentiator finds ways of creating value for buyers that yields a price premium in excess of the extra cost." A differentiated product also makes it costlier for buyers to switch suppliers, and DeAngelo's (1981) transaction cost analysis shows that increased switching costs also allow sellers to charge higher prices. To sum up, the intense price competition following deregulation in the late 1970s led to lower industry profits in the 1980s, and this was followed in the 1990s with a move toward product differentiation based on industry expertise.

While a differentiation strategy can potentially create a competitive advantage, it will succeed only if it creates value for buyers. There is evidence that clients value their auditor's industry expertise. Carcello et al. (1992) report a survey of Fortune 1000 controllers who indicate that their auditor's industry knowledge/expertise is a primary attribute of overall perceived audit quality. However, a differentiation strategy, per se, will not necessarily result in a price premium. Porter (1985, 9) notes that "[t]he crucial question in determining profitability is whether firms can capture the value they create for buyers." A major factor affecting the willingness of clients to pay a premium is the relative bargaining power of auditors and their clients.

As already noted, there is greater competition when audits are undifferentiated. However, if differentiation through industry specialization is valued by clients, then auditors (sellers) are in a stronger bargaining position because there will be less competition. Bargaining power is also influenced by the relative economic importance of each party to the other. (3) A small client is likely to be less important to an auditor than a large client, causing bargaining power to be a function of client size. Larger clients, by virtue of their importance to auditors, will be able to negotiate their audit fees, while smaller clients are more likely to be price takers.

Client bargaining power is measured in the study in two ways: absolute size and relative size. Absolute size is important, and we expect large clients to have greater bargaining power than smaller clients simply because audit fees are greater, which makes the client more economically important to the auditor. Relative size is also important and is measured in terms of how large a client is relative to its auditor's industry clientele. If a single client represents a large portion of the auditor's total fees in a particular industry, then that client will be of greater importance to the auditor and will have greater bargaining power over the auditor. In the limit, an auditor's credible claim of industry expertise could be based on a single large client such as Microsoft in the software industry, or General Motors in the auto industry. These large clients, given their prominence relative to the auditor's other industry clients, would be in a stronger bargaining position with their auditors than smaller clients in the industry. Thus, as a company's size increases relative to its auditor's industry clientele, the bargaining power of the client should also increase.

Based on the above arguments, we make two predictions. First, we expect a significant fee premium for industry specialization, but only for clients that are smaller in absolute size. For clients that are larger in absolute size we expect no significant fee premium for industry specialization due to greater bargaining power. Second, relative size within industry is also expected to give clients greater bargaining power and to result in lower audit fees, ceteris paribus. However, the development of specializations by auditors could also develop an alternative economic effect of production economies. Willenborg (2002, 112) notes that specialization could be motivated by--and result in--a fee discount due to auditor production efficiency. (4) If there are production efficiencies by industry specialists, we would expect lower fees for all clients, both small and large. So if we observe a fee decrease from industry specialization, then this would be consistent with production economies, and if we observe a fee increase (as predicted above), then this would be consistent with auditor differentiation. If we observe no significant results, then this could mean one of two things: (1) that industry specialization has no effect on audit fees, one way or the other; or (2) that both effects are present (differentiation and production economies), but their effects neutralize each other.

RESEARCH METHOD, SAMPLE, AND DATA

Research Method

Audit fees are regressed on a set of variables that control for auditee size, complexity, and risk similar to those used in prior studies (Simunic 1980; Francis 1984). The OLS regression model is specified as follows:

LnFee = b0 + b1 LnASSETS + b2 SUBS + b3 SEGMENTS + b4 FOREIGN + b5 RECINV + b6 ROI + b7 LOSSES + b8 OPINION + b9 S1C49 + b10 LnTENURE + b11 COMMENT + error term.

where:

LnFee = natural log of audit fees ($ mil.);

LnASSETS = natural log of total assets ($ mil.);

SUBS = square root of number of subsidiaries;

SEGMENTS = number of business segments reported on Compustat;

FOREIGN = percentage of total assets that are foreign-based;

RECINV = percentage of total assets in receivables and inventories;

ROI = return on investment (net income divided by total assets);

LOSSES = 1 if loss reported in any of the past three years;

OPINION = 1 if audit report is modified for going concern;

SIC49 = 1 if observation is in the utility industry (SIC code 49);

LnTENURE = natural log of the number of years with the same auditor; and

COMMENT = 1 if a significant event affected the current year audit fee.

Audit fees are expected to be increasing in relation to auditee size (LnASSETS), complexity (SUBS, SEGMENTS, FOREIGN), and risk (RECINV, OPINION). Prior studies also report that fees are positively related to profitability (ROI, LOSSES), possibly due to the client's financial condition and ability to pay fees. Three additional control variables are included the model. SIC49 is coded 1 for utilities. Simunic (1980) and Palmrose (1986) note that audit fees are usually lower for utility companies. LnTENURE controls for the effect of auditor tenure on audit fees. COMMENT is a variable collected from the audit fee survey that indicates the presence of special circumstances that affected audit fees. Examples of special circumstances (which are coded 1) include: separate audit of a subsidiary, new accounting standard or accounting system, merger, significant internal auditor participation, or acquisition.

Prior studies find that audit fee models are sensitive to audit fee size, and this is the case in our sample as well (Francis and Stokes 1986; Craswell et al. 1995). We perform a Chow test to determine if the full sample parameters are consistent (as a set) across the upper and lower halves of company size. The median value of assets in the full sample ($123 million) is used to split the sample into upper and lower halves. The Chow test has a significant F-ratio at p < .01, rejecting a null hypothesis of no differences in the regression parameters between the upper and lower halves of the sample. As a result of the Chow test, our tests are based on separate estimations for "large" and "small" companies based on upper/lower halves of the sample.

Two experimental variables are added to the above audit fee model, SPECIALIST and POWER. Following Craswell and Taylor (1991) and Craswell et al. (1995), SPECIALIST is coded 1 if an auditor has 20 percent or more market share in an industry, and 0 otherwise. (5) Industry is defined as two-digit SIC classifications, and all companies in Compustat are used to determine which auditors are industry specialists. (6) An auditor's market share for an industry is calculated as the sum of sales of its individual clients in an industry, divided by the sum of sales for all companies in the industry. (7) In the regression model, SPECIALIST takes on a value of 1 if a company's auditor is a specialist (as defined above) in the company's industry.

The effect of absolute client size is implicitly tested by the research design used in the study in which models are estimated separately for smaller and larger companies. We predict that the parameter value for SPECIALIST will be lower for larger companies than it is for smaller companies, due to greater bargaining power of larger companies.

POWER measures how large a single client is relative to the auditor's total clientele in the industry. Client bargaining power is expected to increase as the value of POWER increases. POWER is measured for each company as the natural log of company sales divided by the sum of industry sales for all firms in the industry audited by the company's auditor. Audit fees are expected to be lower as POWER increases due to greater client bargaining power.

Sample and Data

Survey questionnaires were mailed in 1994 and requested audit fee data for the fiscal 1993 audit for 3,047 companies that were selected from the 1993 Compustat database. The selection criteria includes having a Big 6 auditor and a SIC code below 6000, which excludes financial institutions from the study. Companies with SIC codes higher than 6000 are non-industrials, which have different financial statement reporting formats. The survey questionnaires requested total audit fees, auditor tenure, number of subsidiaries, and comments on any nonrecurring events and circumstances that affected total audit fees. Of the 3,047 companies surveyed, 774 responded, for response rate 25 percent, and 651 companies have the necessary Compustat data, for a useable response rate of 21 percent.

We examined two features of the sample to determine its representativeness of the underlying Compustat population of Big-6-audited companies with SIC codes under 6000: (1) company size, measured by sales; and (2) industry distribution, defined as one-digit SIC codes. Median sales of the sample are $134 million; this is very close to median sales of $136 million for the Compustat population. The sample (population) industry distributions are listed in Table 1. The largest difference is five percentage points for SIC 4. The differences are within two percentage points of the population distribution for the other four SIC codes. Thus, the sample appears to be broadly representative of the Compustat population, based on company size and distribution across industry sectors.

Big 6 industry specialists audit 34 percent of the sample (n = 224), and nonspecialist Big 6 auditors the other 66 percent of the sample (n = 427). Of the 39 SIC two-digit industries in the sample, a total of 8 industries with 21 observations have only one auditor specialist; 23 industries with 162 observations have two auditor specialists; and 8 industries with 41 observations have three auditor specialists. The total number of observations per Big 6 accounting firm, and the number of observations in specialist industries, are as follows: Arthur Andersen (137/62), Coopers & Lybrand (110/29), Deloitte Touche (119/41), Ernst & Young (126/31), KPMG (88/29) and Price Waterhouse (71/32).

Descriptive statistics are reported in Table 2 for the full sample, and for the upper and lower halves of the sample split at the median value of assets ($123 million). All of the variables except OPINION, COMMENT, and POWER are significantly different between the upper and lower halves of the sample at p < .05. The differences in means are generally in the direction of increasing audit fees for the upper half of the sample, i.e., larger in size (LnASSETS), greater complexity (SUBS, SEGMENTS, FOREIGN), and more profitable (ROI, LOSSES). The upper half of the sample also has significantly more utilities (SIC49) and longer auditor tenure (LnTENURE). With respect to the test variable SPECIALIST, the upper half is more likely to be audited by an industry specialist (42.6 percent) than the lower half (26.2 percent). The other test variable, POWER, is larger for the upper half (.076) than for the lower half (.064), but the difference in means is not significant.

If the two experimental variables, SPECIALIST and POWER, are highly correlated with each other or with the control variables, then there could be a multicollinearity problem. Correlation matrices for upper and lower samples are provided in Table 3. The Pearson product-moment correlation between SPECIALIST and POWER are insignificant (-0.075 for the upper half; -0.045 for the lower half). Correlations between SPECIALIST/POWER and the control variables are also low, indicating that multicollinearity is unlikely to be a problem in the OLS estimations. For example, in the upper half of the sample, the largest correlation coefficients are SPECIALIST and LnASSETS (.252), SPECIALIST and SIC49 (.245), and POWER and SIC49 (-.248). For the lower half of the sample, the largest correlation coefficients are POWER and OPINION (.160), POWER and LnASSETS (.147), and POWER and ROI (.102). In addition, the variance inflation factors are all under 2.4 in the regression models in Table 4, which further indicates that multicollinearity is not a problem in the model estimation (Greene 1999).

RESULTS

The OLS audit fee regression models are reported in Table 4. For completeness, a full sample estimation is reported along with the two subsamples representing the upper and lower halves of company size. The reported p-values for parameter t-statistics for the control variables are two-tailed probabilities. Because we have directional tests for the experimental variables the p-values for the POWER and SPECIALIST parameter t-statistics are one-tailed probabilities. All three models have F-ratios that are significant at p < .01. Adjusted [R.sup.2]s are .87 for the full sample, .83 for larger companies in the sample, and .57 for smaller companies in the sample. All of the control variables are significant at p < .05 except ROI and LOSSES in all three samples, OPINION in the upper half of the sample, and LnTENURE, SEGMENTS, and SIC49 in the lower half of the sample. Results for the experimental variables in the full sample are as follows: SPECIALIST has a positive sign but is not significant at conventional levels (p = .09), and POWER has a negative sign and is significant at p = .02.

Smaller Companies

For smaller companies, the variable for industry specialization, SPECIALIST, has a parameter value of +0.097, which is significant at p = .04. The magnitude of the specialist premium averages 10 percent using the estimation procedure described in Craswell et al. (1995, 307). While statistically significant the magnitude of the fee premium is lower than the 34 percent premium reported in Craswell et al. (1995), and the 29 percent premium reported in DeFond et al. (2000). Nevertheless, our results are consistent with an audit fee premium for industry specialization when the client is small and has low bargaining power. Craswell et al. (1995) reports a fee premium only for larger clients whereas we report a specialization premium only for smaller clients. The inconsistency in these results may be due to the large difference in firm size across the two studies. The median asset level of the sample in this study is 123 million versus 18.2 million in Craswell et al. (1995). The large companies in the Craswell et al. (1995) sample would be considered small companies in our sample. (8)

The variable POWER, which proxies for bargaining power, and which measures client size relative to the auditor's industry clientele, is not significant (p = .35) indicating that smaller companies are not able to negotiate lower fees. Note that POWER averages 6.4 percent for companies in the lower half of the sample, so these companies are small in both absolute size and relative size.

Larger Companies

For larger companies, the variable for industry specialization, SPECIALIST, is not significant (p = .44). However, the variable POWER, which proxies for client bargaining power, has a parameter value of -0.589 and is significant at p = .01. Thus, larger companies in the sample do not pay a premium for industry specialization, but they do have lower fees as their bargaining power increases. The variable POWER averages 7.6 percent for larger companies in the upper half of the sample. Multiplying this amount by the parameter value of -0.589 gives an average value of -0.045 for POWER in the regression model, and this represents an average fee reduction of 5 percent using the estimation procedure described in Craswell et al. (1995, 307). In sum, for the upper half of the sample, audit fees are thus affected by both the absolute size of companies (no specialist premium for larger companies) and by relative size vis-a-vis their auditor's industry clientele (a fee discount as POWER increases). These results are consistent with larger companies having greater bargaining power than smaller companies, and lower audit fees as a consequence.

Sensitivity Analysis and Robustness Cheeks

Several sensitivity checks were performed and are outlined in Table 5. Consistent with prior research, a 20 percent cutoff was used for the SPECIALIST variable results in Table 4. To test the sensitivity of our results to alternative SPECIALIST cutoffs, we re-estimated the models in Table 4 using cutoffs of 25 percent, 30 percent, and 35 percent. For smaller companies, SPECIALIST is positive and statistically significant (p-values are .05, .02, .02, respectively) and POWER remains insignificant. For larger companies, SPECIALIST remains insignificant and POWER is negative and significant (p-values are .01, .01, .01, respectively). This demonstrates that the results in Table 4 are not driven by the cutoff percentage for SPECIALIST.

Recall the model in Table 4, which includes POWER, a variable that measures client size relative to all firms in the industry audited by the company's auditor. We also tested another measure of relative client size, TOTPOWER, which measures client size relative to all firms audited by the company's auditor, regardless of industry. When TOTPOWER is added to the model in Table 4 it is positive and significant (p < .01) for larger and insignificant for smaller companies. More importantly, the results for SPECIALIST and POWER remain essentially the same despite the addition of TOTPOWER. For larger companies TOTPOWER is significantly positively correlated (.39) with LnASSETS and significantly positively correlated (.39) with LnFEE and seems to be picking up a client-size effect not fully captured by LnASSETS.

We also re-estimated the models in Table 4, deleting observations from the utilities industry (SIC code 49). Utilities are in a regulated industry that could make them qualitatively different from the other industries in the study. Utilities are also the largest single industry (n = 76) in the study, and represent 12 percent of the sample. Results of this re-estimation are as follows. For smaller companies, SPECIALIST is positive and statistically significant at p = .01, and POWER is insignificant. For larger companies, SPECIALIST is insignificant and POWER is negative and significant at p < .01. This estimation is consistent with the results reported in Table 4, and demonstrates that the results in Table 4 are not driven by observations from the utilities sector.

We also re-estimated the models in Table 4 deleting the 76 companies in the utilities industry, plus the deletion of another 35 companies from industries having less than 30 observations (per the underlying Compustat population). If an industry has less than 30 observations, then it may be difficult for an auditor to have a sufficient clientele to develop industry expertise. Craswell and Taylor (1991) and Craswell et al. (1995) use a similar screen to define industries with more plausible auditor specializations. These two deletions reduce the sample to n = 540 observations. For smaller companies, SPECIALIST is positive and statistically significant at p = .02, and POWER is insignificant. For larger companies, SPECIALIST is insignificant and POWER is negative and significant at p = .01. This estimation is consistent with the results reported in Table 4, and demonstrates that the results in Table 4 are not driven by utility companies or from small industries.

We also re-estimated the models in Table 4 using the test variables SPECIALIST and POWER one variable at a time rather than jointly as in Table 4. The purpose of this analysis is to verify that the insignificance of POWER in the smaller half of the sample, and the insignificance of SPECIALIST in the upper half of the sample, are not the result of collinearity between the two variables. The results of this analysis are consistent with the results in Table 4, which confirms the results for SPECIALIST and POWER.

DISCUSSION AND LIMITATIONS

Porter's (1985) analysis of competitive strategy and the role of differentiation in creating competitive advantage are used to frame our understanding of auditor industry specialization. The Big 6 accounting firms promote their industry specialization and expertise in the U.S. audit market, but because of data unavailability there has been little research on the impact of such specializations on audit fees. Two U.S. studies, both using data from regulated industries in the mid-1980s, report no evidence that industry specialization is a successful differentiation strategy leading to higher audit fees (Palmrose 1986; Pearson and Trompeter 1994). The primary evidence that specialization leads to an audit fee premium comes from studies of the Australian and Hong Kong audit markets (Craswell et al. 1995; DeFond et al. 2000; Ferguson et al. 2003). These two audit markets are considerably smaller than the U.S. market, and we have no way of knowing if the results generalize to the U.S. market.

An important insight from Porter's (1985) analysis is that differentiation, per se, does not necessarily lead to higher audit fees. Specifically, if clients have bargaining power, then the value created by differentiation would be captured by clients rather than accounting firms. We test for both the existence of higher fees from specialization, and for the effect of client bargaining power on audit fees. Our results are consistent with Porter's (1985) framework. We find evidence that industry specialization (differentiation) leads to higher audit fees, but only for a subset of the sample in which clients appear to have less bargaining power. Specifically, audit fees are higher for clients that are smaller in absolute size (assets < $123 million). For clients with assets > $123 million, there is no premium for industry specialization, and audit fees actually decrease as client size increases relative to the auditor's total industry clientele.

These findings raise questions about the efficacy of industry specialization as a differentiation strategy if a premium is earned only for smaller companies. Porter (1985, 153-160) articulates two broad paths to achieving successful differentiation. First, a firm can become unique in some aspect of performing its existing activities. In the case of accounting firms, auditors could acquire greater in-depth knowledge of particular industries, leading to a deeper understanding of clients and implicitly higher-quality audits, but leaving the basic audit production process unchanged. In this case, the cost would be in the form of human capital investments in industry-specific knowledge, and higher audit fees would be required to earn a return on these investments. The normal presumption is that differentiation is costly, and Porter (1985, 153) says, "Differentiation will lead to superior performance if the value perceived by the buyer exceeds the cost of differentiation." This is the basis for the standard assumption that industry specialization leads to higher audit fees (Craswell et al. 1995, 301).

The alternative path is that a firm may reconfigure its production process in some fundamental way that enhances its uniqueness. As a by-product, reconfiguring the production process could also lead to a more efficient production process. Thus, a reconfiguration could lead to both a differentiated product and to lower production costs. With regard to production efficiencies, Porter (1985, 159) says, "A firm with a sustainable cost advantage in performing the activities that lead to differentiation will enjoy much greater sustainability." In other words, if you can differentiate and lower costs, then you are more likely to create a sustainable competitive advantage. In the case of auditing, investments in industry-specific human capital could lead to the redesign of the audit process custom-tailored to specific industries (Eichenseher and Danos 1981). Such customization could create both a differentiated product (a higher-quality audit) and production efficiencies for the accounting firm. If a large client has bargaining power over the auditor, then the client could capture some of the production efficiencies through a lower audit fee. The net effect on fees will depend on whether the premium for differentiation dominates production efficiencies, or vice versa. Even if the net effect is a "fee discount" for larger clients due to their bargaining power, differentiation could still be a profitable strategy if the audit firm's production efficiencies exceed discounted audit fees.

The scenario described above is consistent with what we observe. Clients in the upper half of the sample (assets > $123 million) pay no premium for specialization and have lower overall fees as their size increases relative to their auditor's industry clientele. Smaller clients with less bargaining power are unable to capture these production efficiencies and instead must pay a premium for differentiated industry expertise. While logical and consistent with Porter's (1985) framework, our empirical results are only suggestive of this scenario. For example, there could be correlated omitted variables such as internal auditing. Specifically, it is possible that larger companies, instead of exerting bargaining power, are simply "lending" their internal audit staff to the outside auditor (Felix et al. 2001). (9) Ultimately, these possibilities can only be evaluated empirically with proprietary cost data from accounting firms to determine the effect of industry specialization on production efficiency and the profitability of a differentiation strategy based on industry specialization.

This study is not without limitations. First, and as mentioned earlier in the paper, we do not include city-level specialization data in our study. The work of Ferguson et al. (2003) and Francis and Stokes (1999), suggest that industry reputations of Big 6 audit firms vary from city to city. Unfortunately, collection of city-level industry specialization data for our U.S.-based study is cost prohibitive. Therefore, it is possible that the model results are misspecified with respect to a local effect of pricing industry specialization. However, it is likely that specialization has two components: one that relates to audit personnel and another that relates to audit best practice systems. Systems such as methodology, standards, training, etc., are more likely national-level attributes, whereas audit personnel is more of a local-level attribute. On the other hand, large national audit firms are able to reassign (on a short- or long-term basis) specialists that are needed for certain audit clients, so specialist audit personnel may also be a centralized (national) attribute.

Second, this research uses 1993 survey data instead of more recent, publicly available, data. While the data is "older," we believe this research does add to the literature. As the only U.S.-based, multi-industry audit fee study it provides a benchmark to which future industry-specialist audit research can be compared. In addition, future research could focus on the positive and negative effects of mandating future disclosures and whether the mandated disclosure of audit fees has affected the fee/specialization relation described herein.

To conclude, there has been much conjecture but little empirical research on the effect of industry specialization on audit fees. Using more recent data than prior published studies, we find evidence of a fee premium for specialization in the U.S. audit market. However, this effect is moderated by the client's bargaining power. Bargaining power has not been considered in prior studies, and our results indicate that it is an important factor in understanding the auditor-client relationship and the pricing of audit services.
TABLE 1
Sample and Population Industry Distributions

                                     Sample      Population

Company size measured by sales     134 million   136 million

Industry
 SIC 1                                  9%            9%
 SIC 2                                 20%           21%
 SIC 3                                 37%           39%
 SIC 4                                 20%           15%
 SIC 5                                 14%           16%

TABLE 2
1993 Descriptive Statistics for 651 Publicly Listed U.S. Companies
(Upper/Lower Halves of Split at Median Assets of $123 million)

                       Full Sample                     Lower Half
                        (n = 651)                       (n = 325)

Variable            Median     Mean      S.D.      Median   Mean   S.D.

Audit Fee ($mil)     0.140     0.514      1.371    0.070   0.086  0.060
Assets ($mil)      122.710  1,814.00   7,336.00   40.935   46.26  32.74
LnASSETS             4.810     5.143      2.083    3.715   3.479  0.960
SUBS                 2.000     2.707      3.163    1.410   1.364  1.101
SEGMENTS             1.000     1.630      1.111    1.000   1.292  0.697
FOREIGN              0.000     0.078      0.159    0.000   0.049  0.136
RECINV               0.310     0.318      0.221    0.365   0.363  0.237
ROI                  0.038    -0.009      0.213    0.035  -0.054  0.287
LOSSES               0.000     0.459      0.499    1.000   0.566  0.496
OPINION              0.000     0.055      0.229    0.000   0.065  0.246
SIC49                0.000     0.117      0.321    0.000   0.037  0.189
LnTENURE             2.398     2.406      0.954    1.946   1.979  0.821
COMMENT              0.000     0.194      0.395    0.000   0.182  0.386
SPECIALIST           0.000     0.344      0.475    0.000   0.262  0.440
POWER                0.032     0.070      0.114    0.024   0.064  0.117

                            Upper Half                Test of
                            (n = 326)              Difference in
                                                     Subsample
Variable            Median     Mean      S.D.          Means

Audit Fee ($mil)     0.330     0.940      1.842         *
Assets ($mil)      612.470  3,577.00  10,069.00         *
LnASSETS             6.420     6.802      1.492         *
SUBS                 3.000     4.046      3.899         *
SEGMENTS             1.000     1.966      1.325         *
FOREIGN              0.000     0.107      0.175         *
RECINV               0.240     0.274      0.195         *
ROI                  0.039     0.036      0.065         *
LOSSES               0.000     0.353      0.479         *
OPINION              0.000     0.046      0.210        n.s.
SIC49                0.000     0.196      0.398         *
LnTENURE             2.944     2.831      0.886         *
COMMENT              0.000     0.206      0.405        n.s.
SPECIALIST           0.000     0.426      0.495         *
POWER                0.037     0.076      0.110        n.s.

* t-test significant at < .05, or Chi-square test for categorical
variables.

Variable Definitions:

 Audit Fee = total audit fees in millions of dollars;

    Assets = total assets in millions of dollars;

  LnASSETS = natural log of total assets ($ mil.);

      SUBS = square root of number of subsidiaries;

  SEGMENTS = number of business segments reported on Compustat;

   FOREIGN = percentage of total assets that are foreign-based;

    RECINV = percentage of total assets in receivables and inventories;

       ROI = return on investment (net income divided by total assets);

    LOSSES = 1 if loss reported in any of the past three years;

   OPINION = 1 if audit report is modified for going concern;

     SIC49 = 1 if observation is in the utility industry (SIC code 49);

  LnTENURE = natural log of the number of years with the same auditor;

   COMMENT = 1 if a significant event affected the current year
             audit fee;

SPECIALIST = coded 1 if an auditor has 20 percent or more market share
             in an industry, 0 otherwise; and

     POWER = measured as the natural log of company sales, divided by
             the sum of logged sales for all firms in the industry
             audited by the company's auditor.

TABLE 3
Pearson Correlation Matrix

Panel A: Upper Half of Sample

            LnASSETS    SUBS  SEGMENTS  FOREIGN  RECINV     ROI

LnASSETS       1.000
SUBS           0.453   1.000
SEGMENTS       0.361   0.300     1.000
FOREIGN        0.184   0.405     0.107    1.000
RECINV        -0.284  -0.024    -0.077    0.166   1.000
ROI            0.041   0.070    -0.016   -0.006   0.040   1.000
LOSSES        -0.034   0.000     0.009    0.075   0.093  -0.413
OPINION        0.058   0.014    -0.050   -0.092  -0.080  -0.029
LnTENURE       0.369   0.232     0.219    0.042  -0.019   0.059
COMMENT        0.033   0.055     0.042    0.070  -0.065  -0.137
SIC49          0.123  -0.128     0.030   -0.291  -0.440  -0.017
SPECIALIST     0.252   0.032     0.045   -0.096  -0.165  -0.063
POWER          0.022  -0.028     0.000   -0.101   0.019   0.069

            LOSSES  OPINION  LnTENURE  COMMENT   SIC49

LnASSETS
SUBS
SEGMENTS
FOREIGN
RECINV
ROI
LOSSES       1.000
OPINION      0.052    1.000
LnTENURE    -0.087   -0.022     1.000
COMMENT     -0.085    0.069     0.038    1.000
SIC49       -0.252    0.076     0.084   -0.003   1.000
SPECIALIST   0.013   -0.041    -0.018   -0.039   0.245
POWER       -0.051   -0.006    -0.021   -0.007  -0.248

            SPECIALIST  POWER

LnASSETS
SUBS
SEGMENTS
FOREIGN
RECINV
ROI
LOSSES
OPINION
LnTENURE
COMMENT
SIC49
SPECIALIST       1.000
POWER           -0.075  1.000

Panel B: Lower Half of Sample

            LnASSETS    SUBS  SEGMENTS  FOREIGN  RECINV     ROI

LnASSETS       1.000
SUBS           0.295   1.000
SEGMENTS       0.144   0.232     1.000
FOREIGN        0.113   0.299     0.011    1.000
RECINV        -0.094   0.050     0.058    0.054   1.000
ROI            0.285   0.110     0.117   -0.002   0.196   1.000
LOSSES        -0.189  -0.072    -0.096    0.090  -0.018  -0.436
OPINION       -0.128  -0.039    -0.056    0.044  -0.018  -0.390
LnTENURE       0.228   0.119     0.144    0.039   0.050   0.180
COMMENT       -0.028   0.027     0.055   -0.017   0.032  -0.065
SIC49          0.143   0.077     0.082   -0.070  -0.190   0.046
SPECIALIST     0.004   0.007     0.102    0.037  -0.048   0.097
POWER          0.147   0.009     0.023   -0.073   0.020   0.102

            LOSSES  OPINION  LnTENURE  COMMENT   SIC49

LnASSETS
SUBS
SEGMENTS
FOREIGN
RECINV
ROI
LOSSES       1.000
OPINION      0.230    1.000
LnTENURE    -0.219   -0.034     1.000
COMMENT      0.026    0.071    -0.088    1.000
SIC49       -0.158   -0.051     0.114   -0.050   1.000
SPECIALIST  -0.101   -0.014     0.066    0.101   0.069
POWER       -0.090    0.160     0.047   -0.032  -0.067

            SPECIALIST   POWER

LnASSETS
SUBS
SEGMENTS
FOREIGN
RECINV
ROI
LOSSES
OPINION
LnTENURE
COMMENT
SIC49
SPECIALIST       1.000
POWER           -0.045   1.000

Correlations with absolute values greater than or equal to .10(.14) are
significantly different from zero at the .05 (.01) level.

Variable Definitions:

LnASSETS = natural log of total assets ($ mil);

SUBS = square root of number of subsidiaries;

SEGMENTS = number of business segments reported on Compustat;

FOREIGN = percentage of total assets that are foreign-based;

RECINV = percentage of total assets in receivables and inventories;

ROI = return on investment (net income divided by total assets);

LOSSES = 1 if loss reported in any of the past three years;

OPINION = 1 if audit report is modified for going concern;

LnTENURE = natural log of the number of years with the same auditor;

COMMENT = 1 if a significant event affected the current year audit fee;

SIC49 = 1 if observation is in the utility industry (SIC code 49);

SPECIALIST = 1 if an auditor has 20 percent or more market share in an
industry, 0 otherwise; and

POWER = measured as the natural log of company sales, divided by the
sum of logged sales for all firms in the industry audited by the
company's auditor.

TABLE 4
Audit Fee Regression Models
(Upper and Lower Halves Split at Median Assets of $123 million
Dependent Variable Is Natural Log of Audit Fees)

                            Full Sample
                             (n = 651)
                      (param/t-stat/p-value)

Control Variables
 Intercept          -4.987  -62.52        0.00
 LnASSETS            0.470   35.62        0.00
 SUBS                0.061    8.10        0.00
 SEGMENTS            0.078    4.22        0.00
 FOREIGN             1.081    8.42        0.00
 RECINV              0.715    7.79        0.00
 ROI                -0.188   -1.87        0.06
 LOSSES              0.034    0.81        0.42
 OPINION             0.214    2.55        0.01
 SIC49              -0.318   -4.73        0.00
 LnTENURE            0.071    3.12        0.00
 COMMENT             0.211    4.57        0.00
Test Variables
 SPECIALIST          0.053    1.33        0.09
 POWER              -0.344   -2.06        0.02
Adjusted [R.sup.2]           86.9%
F-ratio (p-value)            333.0 (.00)

                            Lower Half
                             (n = 325)
                      (param/t-stat/p-value)

Control Variables
 Intercept          -4.580  -37.41        0.00
 LnASSETS            0.349   12.87        0.00
 SUBS                0.159    6.86        0.00
 SEGMENTS            0.013    0.37        0.71
 FOREIGN             0.538    3.00        0.00
 RECINV              0.699    6.80        0.00
 ROI                -0.115   -1.15        0.25
 LOSSES              0.039    0.74        0.46
 OPINION             0.225    2.15        0.03
 SIC49               0.093    0.73        0.47
 LnTENURE            0.048    1.64        0.10
 COMMENT             0.226    3.74        0.00
Test Variables
 SPECIALIST          0.097    1.81        0.04
 POWER              -0.078   -0.38        0.35
Adjusted [R.sup.2]           56.7%
F-ratio (p-value)            33.59 (.00)

                             Upper Half
                              (n = 326)
                        (param/t-staVp-value)

Control Variables
 Intercept          -5.364  -33.18         0.00
 LnASSETS            0.534   22.61         0.00
 SUBS                0.039    4.71         0.00
 SEGMENTS            0.079    3.64         0.00
 FOREIGN             1.364    7.69         0.00
 RECINV              0.713    4.48         0.00
 ROI                 0.164    0.36         0.72
 LOSSES              0.044    0.69         0.49
 OPINION             0.191    1.48         0.14
 SIC49              -0.418   -4.89         0.00
 LnTENURE            0.085    2.58         0.01
 COMMENT             0.190    2.86         0.00
Test Variables
 SPECIALIST          0.009    0.16         0.44
 POWER              -0.589   -2.30         0.01
Adjusted [R.sup.2]           82.8%
F-ratio (p-value)            121.58 (.00)

Variable Definitions:

LnFee = natural log of audit fees ($ mil.);

LnASSETS = natural log of total assets ($ mil.);

SUBS = square root of number of subsidiaries;

SEGMENTS = number of business segments reported on Compustat;

FOREIGN = percentage of total assets that are foreign-based;

RECINV = percentage of total assets in receivables and inventories;

ROI = return on investment (net income divided by total assets);

LOSSES = 1 if loss reported in any of the past three years;

OPINION = 1 if audit report is modified for going concern;

SIC49 = 1 if observation is in the utility industry (SIC code 49);

LnTENURE = natural log of the number of years with the same auditor;

COMMENT = 1 if a significant event affected the current year audit fee;

SPECIALIST = 1 if an auditor has 20 percent or more market share in an
industry, 0 otherwise; and

POWER = measured as the natural log of company sales, divided by the
sum of logged sales for all firms in the industry audited by the
company's auditor.

TABLE 5
Results from Alternative Measures of Test Variables

                                   Variable
                         (Coefficient sign/p-value)

                                 Lower Half
                                 (n = 325)

                         SPECIALIST       POWER

Measure in Table 4:
  20% specialist cutoff    +/.04     Not significant
Alternative Measure:
  25% specialist cutoff    +/.05     Not significant
  30% specialist cutoff    +/.02     Not significant
  35% specialist cutoff    +/.02     Not significant

Add TOTPOWER               +/.04     Not significant
  Remove utilities from    +/.01     Not significant
   the sample
  Remove utilities and     +/.02     Not significant
   small industries

                                   Variable
                         (Coefficient sign/p-value)

                                   Upper Half
                                   (n = 325)

                           SPECIALIST       POWER

Measure in Table 4:
  20% specialist cutoff  Not significant    -/.01
Alternative Measure:
  25% specialist cutoff  Not significant    -/.01
  30% specialist cutoff  Not significant    -/.01
  35% specialist cutoff  Not significant    -/.01

Add TOTPOWER             Not significant    -/.01
  Remove utilities from  Not significant    -/.01
   the sample
  Remove utilities and   Not significant    -/.01
   small industries


We appreciate helpful comments from William Messier (the editor), Jane Mutchler (associate editor), and two anonymous referees. We also appreciate helpful comments from Mark Beasley, Michael Corballis, Gretchen Irwin, and Jilnaught Wong. We thank Lisa Bryant, Denise Jones, James Irving, and Nikhil Nath for their research assistance. The authors are grateful for the support of the University of Auckland, the University of Colorado Business Research Division, Colorado State University, and the University of Virginia's McIntire School of Commerce.

(1) For example, Ferguson and Stokes (2002) find that the specialist premium in the Australian audit market has declined subsequent to the period analyzed in Craswell et al. (1995).

(2) Annual surveys of audit fees show a steady decline in fees since the 1970s (Manufacturer's Alliance 2000).

(3) For related economic research on bargaining power, see Taylor (1995) for a theoretical analysis, and Fernandez and Ozler (1999) and Kauf (1999) for empirical studies in the banking and pharmaceutical industries, respectively.

(4) We thank the referee for clarifying this point. For a discussion of production economies, see also Craswell et al. (1995), Danos and Eichenseher (1986), and Eichenseher and Danos (1981).

(5) In the "Sensitivity and Robustness Checks" section, we test the sensitivity of our results to the 20 percent specialist cutoff.

(6) Audit firms do not publish specific information about how they form industry specializations. To the best of our knowledge, there is no theoretically based method in the literature to define industries. As a result, researchers have had to form reasonable proxies for industry specialization. Instead of presenting results based on our own ad hoc industry groupings, or based on other groupings such as those put forth by Barth et al. (1998) and Barth et al. (2001), we define industries consistent with that of the extant auditor specialization literature, which relies on standard industry classification (SIC) codes.

(7) Market share is measured, primarily, on a national level because collection of city-level data for a U.S.-based study is both cost prohibitive and impractical. Later in the "Discussion and Limitations" section, we describe the trade-offs of using national- versus city-level data.

(8) In an attempt to empirically validate this assertion, we ran our model on a smaller group of clients with a median asset level similar to that of Craswell et al. (1995). The upper group of companies in Craswell et al. (1995) had a positive and significant coefficient on SPECIALIST. Their upper group had median total assets of $63.8 million and all of the companies in the upper group had total assets of more than $18.2 million. Using our data, we ran the model in Table 4 on a subgroup of companies in our dataset similar to the upper group in Craswell et al. (1995). We constructed the subgroup by removing from our overall dataset very small companies and very large companies. We removed all companies with total assets less than $18.2 million and we removed enough large companies (starting with the largest company, second largest, and so on) such that the median total assets of the subgroup equaled $63.8 million. The resulting subgroup consisted of 288 companies. It is important to note that more than 80 percent of the 288 subgroup companies are classified in our study as smaller, lower half companies. Using the 288 subgroup of companies, the fee model shows a positive coefficient for SPECIALIST (significant at p < .01). Therefore, it seems reasonable to assume that the inconsistency in our results versus those of Craswell et al. (1995) is due to the large difference in firm size across the two studies.

(9) we appreciate this comment from the referee.

REFERENCES

Barth, M. E., W. H. Beaver, and W. R. Landsman. 1998. Relative valuation roles of equity book value and net income as a function of financial health. Journal of Accounting & Economics 25: 1-34.

--., D. P. Cram, and K. K. Nelson. 2001. Accruals and the prediction of future cash flows. The Accounting Review 76: 27-58.

Bell, T., F. Marrs, I. Solomon, and H. Thomas. 1997. Auditing Organizations Through a Strategic-Systems Lens. New York, NY: KPMG.

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

Craswell, A., and S. Taylor. 1991. The market structure of auditing in Australia: The role of industry specialization. Research in Accounting Regulation 5: 55-77.

--, J. R. Francis, and S. L. Taylor. 1995. Auditor brand name reputations and industry specializations. Journal of Accounting and Economics 20: 297-322.

Danos, P., and J. Eichenseher. 1986. Long-term trends toward seller concentration in the U.S. audit market. The Accounting Review 66: 633-650.

DeAngelo, L. 1981. Auditor independence, "low bailing," and disclosure regulation. Journal of Accounting and Economics 3:113-127.

De Belde, I. 1997. An exploratory investigation of industry specialization of large accounting firms. International Journal of Accounting 32: 337-355.

DeFond, M., J. Francis, and T. J. Wong. 2000. Auditor industry specialization and market segmentation: Evidence from Hong Kong. Auditing: A Journal of Practice & Theory 19: 49-66.

Eichenseher, J., and P. Danos. 1981. The analysis of industry-specific auditor concentration: Towards an explanatory model. The Accounting Review 56: 479-492.

Felix, W. L. Jr., A. A. Gramling, and M. J. Maletta. 2001. The contribution of internal audit as a determinant of external audit fees and factors influencing this contribution. Journal of Accounting Research 39:513-534.

Ferguson, A., and D. Stokes. 2002. Brand name audit pricing, industry specialization and leadership premiums post-Big 8 and Big 6 mergers. Contemporary Accounting Research 19: 77-110.

--., J. Francis, and D. Stokes. 2003. The effects of firm-wide and office-level industry expertise on audit pricing. The Accounting Review 78:429-488.

Fernandez, R., and S. Ozler. 1999. Debt concentration and bargaining power: Large banks, small banks, and secondary market prices. International Economic Review 11: 333-355.

Francis, J. 1984. The effect of audit firm size on audit prices: A study of the Australian market. Journal of Accounting and Economics 6: 133-151.

--, and D. Stokes. 1986. Audit prices, product differentiation, and scale economies: Further evidence from the Australian audit market. Journal of Accounting Research 24: 383-393.

--, and --. 1999. City markets as a unit of analysis in audit research and the re-examination of Big 6 market shares. Abacus 35: 185-206.

Greene, W. 1999. Econometric Analysis. 4th edition. New York, NY: Prentice Hall. Hogan, C., and D. Jeter. 1999. Industry specialization by auditors. Auditing: A Journal of Practice & Theory 18: 1-17.

Kauf, T. 1999. Price discrimination and bargaining power in the U.S. vaccine market: Implications for childhood immunization policy. The Quarterly Review of Economics and Finance 39: 249-265.

Maher, M., P. Tiessen, R. Colson, and A. Broman. 1992. Competition and audit fees. The Accounting Review 67: 199-211. Manufacturers Alliance. 2000. Survey of General Audit, 2000. Arlington, VA: Manufacturers Alliance.

Palmrose, Z-V. 1986. Audit fees and auditor size: Further evidence. Journal of Accounting Research 24: 97-110.

Pearson, T., and G. Trompeter. 1994. Competition in the market for audit services: The effect of supplier concentration on audit fees. Contemporary Accounting Research 11:115-135.

Porter, M. 1985. Competitive Advantage: Creating and Sustaining Superior Performance. New York, NY: The Free Press.

Public Accounting Report. 1993. BDO re-engineers. Public Accounting Report (November 15): 1, 4.

--.1995. New PW chair advances industry alignment. Public Accounting Report (June 30): 1, 4.

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

Solomon, I., M. Shields, and O. R. Whittington. 1999. What do industry-specialist auditors know? Journal of Accounting Research 38 (Spring): 191-208.

Taylor, C. 1995. The long side of the market and the short end of the stick: Bargaining power and price formation in buyers', sellers', and balanced markets. The Quarterly Journal of Economics 110: 837-855.

Wall Street Journal. 1985a. Audit fees fall as CPA firms jockey for bids. (January 28): 33.

--. 1985b. CPA firms diversity, cut fees, steal clients in battle for business. September 20): 1, 24.

--. 1987. Marketing battles erupt among business-hungry CPA firms. (May 21): 1.

Willenborg, M. 2002. Discussion of: Brand name audit pricing, industry specialization, and leadership premiums post-Big 8 and Big 6 mergers. Contemporary Accounting Research 19: 111-115.

Work, C. 1985. Accounting's bottom line: Big trouble. U.S. News & Worm Report (October 21): 58.

Jeffrey R. Casterella is an Assistant Professor at Colorado State University, Jere R. Francis is a Professor at the University of Missouri--Columbia, Barry L. Lewis is a Professor at Colorado State University, and Paul L. Walker is an Associate Professor at the University of Virginia.

Submitted: November 2001

Accepted: October 2003
COPYRIGHT 2004 American Accounting Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2004 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Casterella, Jeffrey R.; Francis, Jere R.; Lewis, Barry L.; Walker, Paul L.
Publication:Auditing: A Journal of Practice & Theory
Date:Mar 1, 2004
Words:8904
Previous Article:An empirical analysis of voluntarily supplied client-auditor realignment reasons.
Next Article:Comparing audit team effectiveness via alternative modes of computer-mediated communication.
Topics:

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters