تخصص حسابرس، ممیزی کیفیت درک شده، و هزینه های حسابرسی در بازار حسابرسی دولت های محلی

Prior governmental research implies a positive relation between auditor specialization and audit quality, but the effect of specialization on audit fees is mixed. However, no single governmental study investigates the effect of auditor specialization on both audit quality and audit fees. Also, prior studies focus on either large- or small audit firms and often employ indirect proxies for audit quality. We study the effects of auditor specialization on perceived audit quality and audit fees. Our data represent both Big 5 and smaller audit firms and include three market-based measures of specialization. We survey 241 Florida local government finance directors and find that specialization is positively associated with perceived audit quality but not with audit fees. We also find that Big 5 auditors, often used as a proxy for higher audit quality in prior research, are not uniformly associated with increased perceived audit quality but consistently charge higher audit fees. Our results confirm a relation between measures of audit firm specialization and audit quality and raise questions regarding audit firm size and audit quality in the municipal sector. Our findings suggest that engaging specialized auditors may be good policy for many local governments.

Policy suggests that governments should endeavor to obtain high-quality audits at reasonable cost (Gauthier, 2005 and GAO, 1987), but the extent to which this objective is met in practice is unknown due to limited existing research findings. Prior studies of the effect of auditor specialization on governmental audit quality reveal a positive relation between specialization and quality; however these studies often omit segments of the audit market by focusing on only Big 5 firms or non-Big 5 firms and do not consider the effects of audit fees. While Elder (1997) argues that cost considerations are also relevant to the auditor appointment decision, the available evidence concerning the effect of auditor specialization on government audit fees is limited, inconsistent, and does not incorporate measures of audit quality. In sum, the extant literature provides incomplete guidance to government officials charged with optimizing auditor procurement decisions. We report a study of the effects of auditor specialization on perceived audit quality and audit fees in a competitive market in which specialized audit firms play a considerable role. Our study employs continuous, market-based measures of specialization rather than relying on ad hoc definitions of government audit “specialists” used in previous studies. We survey 241 finance directors of Florida local governments regarding their perceptions of the quality of the audit their government received. We also obtain data on audit fees and other audit-related control variables. We find that auditor specialization is positively related to perceived audit quality but unrelated to audit fees. In contrast, controlling for specialization, our results indicate that Big 5 auditors are not associated with differential perceived audit quality yet charge premium audit fees.1 Our findings substantiate Jensen and Payne’s suggestion (2005) that specialized auditors provide high-quality audits at reduced fee levels and are consistent with Copley’s observation (1991) that non-Big 5 firms provide high-quality service in the public sector. The policy implication of our study is that governments may be best served by retaining a specialized audit firm. The remainder of this paper is organized as follows. The next two sections provide the background for the study and research model development, while the fourth section addresses the research method. The last section of the paper presents the study’s results and our discussion and conclusions.

Table 2 presents descriptive statistics for the overall sample. Panel B reports those variables measured on a 7-point scale. The table shows that respondents generally have favorable views of the performance of their auditors, in that the means for the dependent variables and the independent variables are well above the scale midpoint of 3.5, although responses for most variables span the full range of values (1–7). Panel C reports that 12.9% of the respondents engaged a Big 5 audit firm. Panel D contains descriptive statistics for continuous and discrete study variables. This panel illustrates that the entities vary in size (i.e., audit fees, revenues), level of finance director involvement in the auditor selection process (INPUT) and financial statement preparation (DRAFT). Auditor tenure averaged 8.01 years, while ranging from 1 to 40 years. Descriptive statistics for the three specialization measures show considerable range in the number of government audit engagements that the various audit firms (or Big 5 offices) maintained either in 2003 or over the 10-year period (1993–2003), as well as the revenues audited by the firm offices. Table 3 provides auditor trend data for the Florida governmental audit market during the 10-year period between 1993 and 2003.14 Panel A reflects the substantial decrease in the presence of Big 5 firms in the Florida governmental audit market. Over the 10-year period, Big 5 audits of Florida governmental entities declined from 134 to 46, although the decrease was not uniform across the Big 5. Panel B illustrates that the composition of non-Big 5 audit providers also underwent significant change during that time. It shows that the total number of non-Big 5 firms in the market declined from 162 to 128 from 1993 to 2003. The greatest decline is noted for firms that performed only one governmental audit annually. Thus, a trend toward specialization emerges; over time, most Florida auditors either increased their portfolio of governmental audit clients or exited the market.Table 4 presents bivariate correlations among the independent variables. Many of the independent variables captured from the questionnaire are highly correlated, in part due to the narrow range of those data items. The correlations mainly are between various audit quality attributes, where such correlation would be expected. In contrast, our measures of specialization are almost entirely uncorrelated with any independent variables except themselves. Potential problems of collinearity posed by high correlations are addressed later in the paper. We note that the correlation between B5 and LSPEC# is negative while B5 is positively correlated with SPECAVG. These correlations are further evidence that the Big 5 presence in Florida has diminished over time. B5 also is positively correlated with LSPEC$, indicating the Big 5 audit observations in our data are associated with larger governments.Table 5 displays the results of the perceived audit quality models estimated using ordinal regression.15 The χ2 statistic indicates that the model provides a significant improvement over the baseline intercept-only model. The Cox and Snell statistic, as well as the Nagelkerke Pseudo R-square, indicates that the model is capturing a great deal of the variance in audit quality.16 Control variables MGTLTR, MGRTIME, FIELDW, and LREV all are significant with the expected signs.17 LATEN is positive in all models and significant in the LSPEC# and LSPEC$ models, while INDEP is marginally significant in the LSPEC$ model only. Of the control variables, SKEPT and INPUT are not significant.LSPEC#, LSPEC$ and LSPECAVG are all positive and significant. Overall, these results suggest that auditor specialization is associated with higher perceived audit quality. On the other hand, after controlling for specialization, B5 is consistently negative and significant in the LSPEC$ and LSPECAVG models. We discuss this finding in more detail in a later section. 7.1. Analysis of audit fees We initially estimated our model of audit fees in Eq. (2) using OLS regression. (Descriptive statistics and correlations for the fee model variables are included in Table 2 and Table 4.) Diagnostic tests revealed no problems with collinearity (variance inflation factors), normality of residuals (Wilk-Shapiro) or outliers (Cook’s D); however White’s heteroscedasticity test revealed that the residuals exhibit heteroscedasticity.18 Therefore, we re-estimated the fee model using White’s heteroscedasticity-consistent standard errors and covariance. Table 6 presents the regression models using each of the three specialization variables. Although the three specialization measures are positively associated with perceived audit quality, none of the measures has a significant relation with audit fees. In contrast, the coefficient of B5 is positive and significant in all of the fee models. The adjusted R2 values of the fee models indicate that the model variables explain a considerable amount of the variation in audit fees.197.2. Additional analysis of audit firm size and specialization The specialization measures are associated with higher perceived audit quality; however, contrary to expectations, Big 5 auditors are associated with lower levels of perceived audit quality. Since the specialization measures include specialization for both Big 5 and non-Big 5 firms, we perform additional analysis to assess whether the effect of specialization on perceived audit quality and audit fees differs between Big 5 and non-Big 5 firms. We first add a Big 5 specialist interaction term to the LSPEC# models in Table 5 and Table 6 (B5 × LSPEC#).20 In the quality model, the Big 5 variable remains negative and is marginally significant (p = .086), LSPEC# remains positive and significant (p = .012), and the interaction term is also positive and marginally significant (p = .076). In the fee model, the interaction term is insignificant and the significance levels are unchanged for the remaining variables. We perform a similar analysis using a dichotomous specialization variable. Using a cut-off of 10 or more Florida governmental audits in 2003, eleven of the 31 Big 5 auditors are classified as specialists, and 115 of the 210 non-Big 5 auditors are considered specialists.21 In the quality model, B5 remains negative and significant (p = .038), while the dichotomous specialist and Big 5 specialist terms are positive and significant at the p ⩽ 10 level. In the fee model, the Big 5 variable remains positive and significant, and the dichotomous specialist variable and Big 5 specialist interaction term are not significant. Finally, we reperform the analyses in Table 5 and Table 6 limiting the observations to non-Big 5 firms. Each form of the specialist variable (i.e., LSPEC#, LSPEC$, LSPECAVG, or a dichotomous measure) is associated with significantly higher perceived audit quality but not associated with higher fees. Overall, these results indicate that both Big 5 and non-Big 5 specialists are associated with higher perceived audit quality, and the lower perceived Big 5 quality is attributable to non-specialist Big 5 auditors. Further, Big 5 specialists do not charge a fee premium over and above the Big 5 fee premium. 7.3. Discussion of findings In this study, perceived audit quality is positively associated with the degree of audit firm specialization. This finding is robust to the choice of specialization metric, that is, between cross-sectional measures (LSPEC# and LSPEC$) or a longitudinal one (LSPECAVG). This result strongly supports the suggestion that audit quality increases with specialization (Gramling and Stone, 2001 and Neal and Riley, 2004). Further, our results show that the quality advantage offered by specialized auditors does not entail a higher cost to the auditee. This is consistent with the notion that specialization yields economies of scale (Neal and Riley, 2004, Mayhew and Wilkins, 2003 and Chan, 1999). Indeed, it is reasonable to assume that increased competition within the Florida local government market (following the 1993 elimination of the long-standing bidding restriction) gives impetus to auditors to pass the economic benefits of economies of scale along to auditees by charging “nonpremium” fees. The clear implication of these results is that engagement of a specialized audit firm is a desirable governance decision. While more-specialized audit firms do not charge higher audit fees, Big 5 firms were found to charge premium audit fees. The finding of a Big 5 fee premium is consistent with almost all prior government audit fee research, as well as audit effort and model fee results reported by Johnson et al. (2003). One potential explanation is that, at least to some extent, certain governments are willing to pay a fee premium for a “brand name” auditor, as suggested by Firth (1993) and Bandyopadhyay and Kao (2001). Additionally, it is possible that the municipal bond market rewards engagement of Big 5 auditors, as does the corporate bond market (Mansi et al., 2004). Noting that our Big 5 observations tend to be larger governments, it may be that some of the fee premium reflects complexities of larger governments that are not fully proxied for by our size variable. It is also possible that, simply speaking, only the largest and more costly audit firms have sufficient resources to perform audits of the largest governments, such that these governments are “locked in” to paying higher audit fees as a consequence of their size. In our sample, it appears that lesser perceived quality is associated with Big 5 audit firms that do not specialize in governmental audits. This may reflect the prevailing view that large firms are generally not enthusiastic about government auditing. An AICPA task force (1987, p. 45) noted, “There is a perception that the CPAs consider government clients secondary to the private sector.” Declines in Big 5 service quality may also be attributable to the decline in audit fees documented by Jensen and Payne (2005) following the repeal of a restriction on competitive bidding in the Florida market. Finally, we cannot discount the possibility that governments engaging Big 5 auditors have unmet higher expectations of their auditors. The market changes noted between 1993 and 2003 are notable and have resulted in higher audit quality and lower audit fees for Florida local governmental entities. It is important to note that non-Big 5 firms can seemingly establish niche markets. This is consistent with Cullinan’s (1998) study of the multi-employer pension audit market, where he finds that non-Big 6 firms have a dominant market position. He suggests that smaller CPA firms can establish a “brand name” in a niche assurance service market to distinguish their expertise. In some cases, these “boutique” firms may have the ability to command higher fees than other firms in the market based on their industry expertise. 7.4. Conclusion We are aware of no prior study that investigates the influence of auditor specialization on both audit quality and audit fees. Our study finds a positive association between (1) the degree to which audit firms specialize in government auditing and (2) audit quality as perceived by the finance officers of Florida local governments. In contrast, we find no relation between specialization and the level of audit fees. As such, we replicate in a single study the findings of separate prior research concerning improved audit quality (Lowensohn and Reck, 2004) and lower audit fees (Jensen and Payne, 2005) in the Florida government audit market. Our findings strongly suggest that, as a matter of policy, local governments might be well served by engaging specialized auditors. It is widely presumed that Big 5 audit firms charge higher fees to reflect superior quality (e.g., Firth, 1993). However, while our study finds higher fees for Big 5 auditors, the results indicate that Big 5 auditors that do not specialize in governmental audits are not associated with higher perceived audit quality. This is an especially interesting finding given that Big 5 auditors have often been used as proxies for audit quality in both public and private sector research. We suggest further study of actual and perceived audit quality for the current Big 4 audits in the municipal sector to assess whether these results are unique to the Florida market. Following the landmark Sarbanes–Oxley Act, the remaining Big 4 auditing firms are more selective in the clients and markets they serve, creating opportunities for other auditors to fill the “vacuum” left as the largest firms exit certain segments of the audit market. Future research should investigate whether the government audit market outside of Florida also manifests an increased presence of non-Big 4 specialized auditors. We acknowledge factors that may limit our study’s results and their generalizability to other samples. All respondents represent the State of Florida because our measures of specialization are confined to that state. As a result, our findings as they may apply in other states should be interpreted cautiously. Second, use of questionnaire methodology poses potential problems such as “errors of severity/leniency”, a “halo effect”, understandability, and respondent truthfulness. However, the survey was adapted from prior research and respondent confidentiality was assured to enhance the likelihood of truthful response. Subject to the potential limitations noted, this study suggests that governments may find it advantageous to appoint specialist auditors. If a Big 4 firm is selected, a government is likely to pay higher fees, and a Big 4 specialist is preferable to a non-specialist.