THE PRICING RELATIONSHIP OF AUDITS AND RELATED SERVICES IN MUNICIPAL GOVERNMENTS

Transcription

1 PUBLIC BUDGETING & FIN. MNGMT., 6(3), THE PRICING RELATIONSHIP OF AUDITS AND RELATED SERVICES IN MUNICIPAL GOVERNMENTS Marc A. Rubin Department of Accountancy Miami University Oxford, Ohio ABSTRACT For private sector organizations, Palmrose and Simunic document a positive relationship between fees for management advisory services (MAS) and fees for audits paid to incumbent auditors (1). In contrast, Abdel-Khalik provides evidence that the purchase of MAS does not affect audit fees. (2) This study, unlike previous studies, examines the relationship between the purchase of MAS and audit fees in government organizations. Results suggest a negative relationship between audit fees and fees paid for related audit services provided by incumbent auditors. In addition, bond issuance and consulting fees paid to incumbent auditors and all fees paid to nonincumbent auditors are found to be not significant. Copyright by PrAcademics Press

2 424 RUBIN INTRODUCTION Auditors' provision of tax, consulting and other services to their current clients raises the issue of whether joint production of audit and other services offers production efficiencies to auditors and fee savings to clients. The present study provides evidence relating to the joint supply of audits and other services provided by audit firms to their municipal government clients. Simunic, Palmrose and Abdel-Khalik examine the relationship between the pricing of audit services and the pricing of nonaudit services provided by public accounting firms to their private sector clients. (3) The results from these prior studies suggest either a positive relationship exists between fees paid for audit and nonaudit services (4) or no relationship exists between the pricing of MAS and audits. (5) This study differs from the previous studies because it focuses on factors that affect the relationship between audit and other fees paid by municipal governments to Big Eight audit firms. (6) Private sector organizations and municipal governments may differ in objectives, the nature of services, the type of organizational structure and the incentives of service providers. These potential differences of municipal governments may cause the relationship between audit fees and other fees paid to public accountants to differ from the results found for private sector organizations. In addition to organizational characteristics, the types of other services provided by audit firms to their municipal government clients may differ from the type of services provided to private sector companies. For example, municipal governments will not likely purchase tax services, as is common in the private sector. Also, municipal governments are frequently required to provide special audits for compliance purposes or need special reports required by statutes of other government jurisdictions. The pricing strategies auditors use for bidding on municipal government engagements may differ from pricing strategies used by auditors for potential clients in the private sector. For example, municipalities are likely to require audit bids to be fixed fee, whereas private sector companies will often use cost plus contracts. (7) In addition, many municipal governments have either a formal or informal policy to rotate their auditors on a periodic basis. (8) The rotation policy may affect the auditor in formulating both a multiperiod pricing strategy for audits as well as a pricing strategy for other nonaudit services. (9) The next section of the paper reviews the prior research in this area and specifies in greater detail the objectives of the study. The following three sections describe the sample, the empirical model, and the results,

3 PRICING RELATIONSHIP OF AUDITS 425 respectively. The final section provides a discussion of results and concluding comments. PRIOR RESEARCH AND OBJECTIVES Simunic provides an analysis of the effect that knowledge spillover between audit and nonaudit services offered by public accounting firms has on the pricing of those services to corporate clients. (10) Simunic develops a one-period model to analyze the relationship between audit and nonaudit fees. In the case where management advisory service knowledge reduces the marginal cost of auditing, he concludes that, "the empirical implications of the comparative solutions are unambiguous---consulting fees will increase, while total spending for audit substitutes (11) will decrease. However, the sign of the change in audit fees depends on the price elasticity of 'demand'..." (12) Simunic conducts a series of empirical tests on a sample of corporations to ascertain the relationship between audit and MAS fees. His results suggest a significant positive association between the two types of fees. He concludes that, "the observed relationship would arise if the production of auditing generates knowledge useful in MAS production and/or the production of MAS reduces the marginal cost of auditing and audit demand is elastic." (13) Palmrose replicates and extends Simunic's results. (14) Also using a sample of corporations, Palmrose found a positive relation between audit and nonaudit fees. In addition to the positive relation of audit and nonaudit fees for incumbent auditors, Palmrose also found a significant relationship between audit fees and nonaudit fees of nonincumbent firms for a small subset of her sample. This later result brings into question the validity of the joint-supply scenario suggested by Simunic. Abdel-Khalik uses a different methodology to examine the relationship between MAS and audit fees. (15) In order to correct for potential self-selection bias by firms electing to purchase both MAS and auditing from the same firm, Abdel-Khalik uses the Heckman-Lee two-stage method of switching regressions. He finds, "...the absence of cost (or benefit) accruing to clients for selecting the incumbent auditors to supply MAS...These results are consistent with expectations in that it would not be rational for clients to pay higher fees simply because they also pay the firms of their auditors additional sums of money for MAS." (16) Municipal governments may differ from corporate purchasers of audit and nonaudit services in significant ways. Most private sector research

4 426 RUBIN assumes that firms want to maximize profits, which suggests that costs for like products or services will be minimized. The profit maximization/cost minimization incentive is not as easily assumed for local government. Local government officials may only have an incentive to save on costs if the savings can translate into some benefit (direct or otherwise) the official can eventually realize. Thus, local government officials may not select the lowest cost provider of like services. Although this study will not directly test the incentives of local government officials, the potential difference in incentives may affect the empirical results. That is, the least costly provider of audits and related services, even if perceived to be of equal quality, may not be selected by local government officials because of a lack of incentive to do so. Thus, empirical results may not reflect available cost savings to local governments by having audits and related services provided by the same firm because the government may not have an incentive to capture the savings. Abdel-Khalik assumes that, "acquiring MAS is purely a discretionary activity." (17) Local governments are not likely to purchase the same type of nonaudit services purchased by the private sector (for example, municipalities do not purchase tax services). Many of the nonaudit services purchased by governments may be mandated and therefore not discretionary. Government purchase of nonaudit services are primarily one of three types. The first type are audit related services (ARS) which are those services that have a relatively direct relation to the provision of the financial audit, but are not considered necessary to the provision of the financial audit and are not included in the financial audit fee. Examples of these types of services include, compliance audits, audits for special jurisdictions or entities related to the government but not consolidated, and special reports required by overseeing bodies or by statutes. The second type of services are bond related services (BRS). BRS are services provided by an audit firm which are directly associated with the government's issuance of bonds. The third type of services are consulting services (CS). These services primarily include system design and implementation, cash management studies, and organizational studies. Another potentially significant difference between municipal governments and corporations which may impact the results of this study relate to the elasticity of demand for auditing. In part, the ambiguity in predicting the sign of the relationship between audit and MAS fees, according to Simunic, stems from the unknown elasticity of demand for units of auditing. (18) That is, corporations may substitute units (or the amount) of independent audit for units of internal audit (19) depending on the pricing difference. In contrast, for many municipal governments, internal audit is

5 PRICING RELATIONSHIP OF AUDITS 427 not well developed and may not be a viable substitute for external audits. (20) Thus, for most municipal governments the units of independent audit are likely to be less elastic relative to private sector firms and demand is not likely to increase with a reduction in the price per unit. If the provision of nonaudit services creates a cost savings in government audits, and if the cost savings can be appropriated by the government, audit fees will likely be lower (than under similar conditions in corporations). The differences found between municipal government and corporate purchasers of audit services may be reflected in the pricing of those services. Expectations regarding the pricing relationship between the previously described types of nonaudit services and audits are as follows: Although the elasticity of demand for audit services is unknown, it is more likely in local governments to observe a negative pricing or fee relationship between audit related services (ARS) and audit fees. Since audit related services are closely tied to the audit, spillover knowledge is likely to occur when a municipality purchases both services from its auditor. The spillover knowledge should either reduce audit setup costs (21) or reduce the marginal cost of auditing, either way reducing the total audit fee. (22) The relation of either bond related services (BRS) or management consulting (MC) to audit fees is less obvious since the existence of spillover knowledge between these services and auditing may not be as likely. Palmrose, contrary to expectations, found the relationship between audit fees and nonaudit fees of nonincumbent auditors significant in large companies. (23) She suggests that her results do not support the notion of joint-supply as an explanation of the relationship between audit and nonaudit fees. Palmrose offers potential reasons for her results including the possibility that nonaudit fees, regardless to whom they are paid, may signal an accounting system problem causing higher audit fees. Although Palmrose's results are unexpected, it is difficult to predict whether a similar relationship exists in municipal governments, thus there is no a priori expectation regarding the relationship between audit fees and nonaudit fees of nonincumbent auditors. SAMPLE A survey questionnaire was sent to all cities with populations over 50,000 in the United States. Responses were received from 272 of the 449 cities. 124 responses were not used because the city did not use a Big Eight

6 428 RUBIN auditor (these cities either used a state auditor or used a non-big Eight audit firm). An additional 62 responses could not be used because of incomplete information. (24) The remaining 86 usable responses are cities who gave complete information regarding the variables of interest and use a Big Eight audit firm for their 1982 fiscal audit. Similar to the criteria used by both Simunic and Palmrose for selecting corporations, (25) only municipalities using Big Eight auditors were included. The reason for including only Big Eight auditors is to provide greater control for potential quality differences between audit firms. The assumption is that Big Eight firms provide approximately equal quality services. Restricting attention to Big Eight firms also allows more comparability with the previous studies. The information regarding fees paid for audit services is for the 1982 fiscal year. A survey questionnaire was used to collect the audit and nonaudit fee data as well as additional information considered important in capturing the audit environment in a municipal government. The information regarding nonaudit fees was collected for the period from two years prior to the audit to one year after the audit to capture any relatively small timing differences in the provision of nonaudit services. Spillover knowledge between nonaudit services and audits is not necessarily restricted to a single period and the billing for such services may not reflect when the knowledge is obtained. Nonaudit service fees paid to incumbent auditors includes only those fees paid to the auditor of the fiscal 1982 financial statements. If an auditor switch took place during the period of interest, fees paid to the previous incumbent auditor were not considered as paid to the current incumbent. Nonaudit fees paid to nonincumbents consist of fees paid only to Big Eight firms who were not involved with the audit over the period of interest. Forty percent of the municipalities in the sample purchased some form of nonaudit services from their Big Eight auditor. Of the sample cities, 16 percent purchased audit related services (ARS), 17 percent purchased bond related services (BRS), and 16 percent purchased consulting services

7 PRICING RELATIONSHIP OF AUDITS 429 TABLE 1 Data on Purchases of Nonaudit Services by Municipalities n=86 Number of purchasers of audit related services (ARS) from incumbent audit firms 14 Mean fees for ARS purchased from incumbent audit firms $34,981 Mean audit fee for purchasers of ARS from incumbent audit firm $51, Number of purchasers of bond related services (BRS) from incumbent audit firms 15 Mean fees for BRS purchased from incumbent audit firms $12,103 Mean audit fee for purchasers of BRS from incumbent audit firms $54, Number of purchasers of consulting services (CS) from incumbent audit firms 14 Mean fees for CS purchased from incumbent audit firm $23,133 Mean audit fee for purchasers of CS from incumbent audit firms $49, Number of purchasers of nonaudit services from nonincumbent audit firms 16 Mean fees for nonaudit services purchased from nonincumbent audit firms $146,041 * Mean audit fee for purchasers of nonaudit services from nonincumbent audit firms $66,621 * This figure reflects very large purchases of consulting services by two large cities. (CS). (26) Nineteen percent of the municipalities purchased nonaudit services from auditors other than the incumbent. These percentages are considerably lower than the percentage of companies purchasing nonaudit services in the Simunic and Palmrose studies. In Simunic's total sample, 33 percent of the firms purchased MAS services and in the Palmrose study, 87 percent of the

8 430 RUBIN corporations purchased nonaudit services from either incumbent or nonincumbent auditors. Additional data on the purchase of nonaudit services by the sample municipalities are contained in Table 1. MODEL A regression model was developed in order to ascertain whether there exists a relationship between audit and nonaudit fees. The variables used in the regression model, in addition to the nonaudit fee variables, are measures found in prior studies to significantly relate to municipal audit fees. (27) The form of the regression model is as follows: ln(audit fee) = b 0 + b 1 ln(population)+ b 2 ln(debt per capita) + b 3 Bond rating + b 4 Service index + b 5 ln(audit reports) + b 6 CAFR - b 7 Tenure-Bid Cities + b 8 Tenure-Non-bid Cities + b 9 Busy season - b 10 Finance Officer tenure + b 11 ln(ars) + b 12 ln(brs) + b 13 ln(cs) + b 14 ln(nonaudit fees- nonincumbent) + u where the control variables are defined as: ln (Population) = ln (Debt per capita) = Bond Rating= natural log of the municipality's population according to the 1980 census, natural log of the total debt of the city divided by population, indication of whether the municipality's general obligation bond rating according to Moody's Investor Service is noninvestment quality (Baa rating or below) (1= noninvestment grade), (28) Service index = the number of non-ordinary services offered by a municipality to its constituents, ln (Audit reports) = CAFR = natural log of the number of separate audit reports provided by the auditor, indication of whether the municipality's financial report was a comprehensive annual financial report (CAFR) as opposed to a general purpose financial statement (GPFS) (1=CAFR),

9 PRICING RELATIONSHIP OF AUDITS 431 Tenure-Bid Cities = Tenure-Non-bid Cities = the current auditor's tenure time if the city regularly solicits audit bids, the current auditor's tenure time if the city does not regularly solicit audit bids, Busy season = indication of whether the municipality's audit was performed during the auditor's busy season defined as November through March (1=Busy season), and Finance Officer tenure = the tenure time of the municipality's chief financial officer. Population is used to control for auditee size; debt per capita and bond rating are used to control for auditor loss exposure; service index is used to control for entity complexity; the number of audit reports and whether the report is CAFR are used to control for report complexity; the auditor tenure variables control for differences in the auditor selection process used by cities; and busy season and finance officer tenure control for specific aspects of the auditor production process. Data regarding the first two variables were obtained from the US Bureau of Census; the bond rating was obtained from Moody's; the remaining data were collected via the questionnaire except for finance officer tenure, which was collected from the Municipal Year Book. (29) The test variables are defined as follows: ln (ARS) = ln (BRS) = ln (CS) = ln (nonaudit fees-nonincumbent)= natural log of the fees paid by a municipality to their current auditor during the period for audit related services, natural log of the fees paid by a municipality to their current auditor during the period for services related to bond issuance, natural log of the fees paid by a municipality to their current auditor during the period for consulting services, natural log of the fees paid by a municipality to Big Eight auditors beside their current auditor during the period , and

10 432 RUBIN u = error term. Data on the test variables were collected from the questionnaire. Based upon the results of previous research, (30) a positive relationship is expected between each of the control variables and audit fees except for Tenure-Bid cities and Finance Officer tenure for which a negative coefficient is expected. Since the elasticity of demand for audit services is unknown, there is no expectation regarding the nature or direction of the relationship between ARS and audit fees. The association of BRS and CS should be similar to ARS if knowledge spillover takes place, but for both of these services the potential for knowledge spillover may not be as great as in the ARS case. Similar to reasoning of Palmrose, (31) there is no expected relation between services purchased from non-incumbent auditors and audit fees. RESULTS Table 2 provides the means or sample percentages for each of the variables used in the regression model. Regression results for the 86 sample cities are presented in Table 3. As noted in Table 3, the regression is significant. (32) The control variables, except for auditor tenure for bid cities, have the expected sign. Significance of the control variables varies. Procedures were used to determine whether the heteroscedasticity or multicollinearity assumptions of the regression model were violated. (33) No significant deviations were noted from the regression model assumptions. In addition, diagnostic procedures did not indicate any observations causing significant influence on the regression estimates. (34) For the test variables, the ln (ARS) variable is significant with a negative coefficient. The remaining test variables ln(brs), ln(cs), and ln(nonaudit fees-nonincumbents) are not significant, although the signs on the ln(brs) and ln(cs) coefficients are negative. In order to determine whether purchases of nonaudit services by themselves, without regard to the amount of the purchases, are associated with the audit fees, an alternative coding for the test variables is used. The alternative coding utilizes a series of indicator variables to capture whether or not the city made a purchase of each of the types of nonaudit services. The significance level of the model, as well as the test variables, changes only slightly with the alternative coding (see Table 4).

11 PRICING RELATIONSHIP OF AUDITS 433 TABLE 2 Description of Variables n = 86 Variable Mean or Percentage Range Standard Deviation Audit fee $57,122 $12,400-$561,000 $66,381 Population (000) Debt per capita $784 $0-$3192 $641 Percentage of cities with noninvestment grade bond rating 2% 0-1 Service index Number of audit reports Percentage of cities issuing a CAFR (versus a GPFS) 77% 0-1 Tenure--Bid cities 2.33 years 1-15 years 2.7 years Tenure--Non-bid cities 2.23 years 1-15 years 3.9 years Percentage of cities with busy season audits 42% 0-1 Finance officer tenure 5.79 years 1-15 years 5.0 years ARS * $5,695 $0-$117,000 $18,556 BRS * $2,111 $0-$100,000 $11,040 CS * $3,766 $0-$69,000 $11,199 Nonaudit fees to nonincumbent auditors * $27,172 $0-$1,046,000 $133,045 * The means for these services are based on the entire sample, including both cities which purchased and did not purchase these services. Table 1 contains the means for only the cities purchasing these services.

12 434 RUBIN TABLE 3 Results of Regressions Analysis Dependent Variable = Audit fee n=86 R 2 =.59 Adjusted R 2 =.51 F statistic = 7.41 (p =.0001) Variable Expected Sign Coefficient t-statistic Prob > t Intercept ln (Population) ln (Debt per capita) Bond Rating Service Index ln (Audit reports) CAFR Tenure--Bid cities Tenure--Non-bid cities Busy season Finance officer tenure ln (ARS) +/ ln (BRS) +/ ln (CS) +/ ln (nonaudit fees--nonincumbents) +/

13 PRICING RELATIONSHIP OF AUDITS 435 In order to insure that the results regarding the significant relationship between ARS fees and audit fees is not due to "low-balling" on initial audits, (35) a variable was added to the regression model to control for municipalities which were using a new auditor for the fiscal 1982 audit (the year for which audit fees were collected). Regression results provide evidence that initial audits of municipalities are not priced significantly less than continuing audits. Therefore, the purchase of ARS is not capturing the effects of "low-balling" on initial audits. (36) DISCUSSION AND CONCLUSIONS The results of this study differ from prior research on private sector firms. (37) The relationship between municipal audit fees and a specific type of nonaudit service fees (ARS) is negative. In addition, there is no significant relationship between nonaudit fees of nonincumbent auditors and audit fees. The results regarding municipalities suggest that a "knowledge spillover" (38) may exist between ARS and audit services which allows for more efficient provision of services. Since purchasers of ARS have systematically lower audit fees, it appears that at least some portion of the cost savings is being passed along to the municipality. This result also appears reasonable in light of the results of Rubin (39) which suggest that the market in which municipal audits are purchased is competitive. The difference in this study's results and those in found in previous studies may be due to the likely scenario of relatively inelastic demand for external auditing services in cities. That is, since cities use very little internal auditing to substitute for external auditing, if the marginal price for units of external audit decrease, the demand for more units of auditing does not substantially increase. Thus, a decrease in the marginal cost of auditing, if passed through to the city, will lead to lower audit fees. Further, if the knowledge spillover decreases audit setup costs, again the city may obtain lower audit fees if they capture the savings. The negative relation between ARS and audit fees also suggests that, in at least some instances, municipal officials may have an incentive to contain costs for purchased services. (40) The lack of significance between the BRS and CS variables and audit fees suggests that there may not be substantial knowledge spillover between these services, or if there is knowledge spillover, any resulting benefits are

14 436 RUBIN TABLE 4 Results of Regressions Analysis Dependent Variable = Audit fee n=86 R 2 =.59 Adjusted R 2 =.51 F statistic = 7.35 (p =.0001) Variable Expected Sign Coefficient t-statistic Prob > t Intercept ln (Population) ln (Debt per capita) Bond Rating Service Index ln (Audit reports) CAFR Tenure--Bid cities Tenure--Non-bid cities Busy season Finance officer tenure ARS (0,1) +/ BRS (0,1) +/ CS (0,1) +/ Nonaudit fees--nonincumbents (0,1) +/

15 PRICING RELATIONSHIP OF AUDITS 437 not being passed along to auditees. The insignificance of the coefficient for nonaudit fees of nonincumbent auditors provides evidence that simply purchasing nonaudit services does not directly affect audit fees. Therefore, the scenario suggested by Palmrose that purchases of any consulting services may indicate firms which need more audit services, (41) does not appear to apply to municipalities. (42) Empirical evidence in both the public and private sectors tends to support the notion of efficiency in the joint supply of audit and nonaudit services, although the fee relationships may differ because of the differences found in the two types of organizations and their respective environments. Results of this study suggest that for some municipalities the officials responsible for letting the contract may have an incentive to control costs. The relationships found between audit and nonaudit fees also provides insights regarding the perception of independence of firms providing both audit and nonaudit services. (43) Since audits and ARS are apparently less expensive if jointly provided by a single supplier, municipal officials need to consider whether independence appears to be compromised if a single auditor is used. Since audits and ARS are similar services, it is likely that constituents will perceive that the provision of these services by a single firm does not compromise the independence of auditors. For the cases of BRS and CS, since there does not appear to be any economic incentive to have the same firm provide both audits and these services, city officials can avoid the independence issue by contracting with a different firm to provide these services. REFERENCES 1. Palmrose, Zoe-Vonna. "The Effect of Nonaudit Services on the Pricing of Audit Services: Further Evidence." Journal of Accounting Research 24 (Autumn 1986): ; and Simunic, Dan A. "Auditing, Consulting, and Auditor Independence." Journal of Accounting Research 22 (Autumn 1984): Abdel-Khalik, A. Rashad. "The jointness of audit fees and demand for MAS: A self-selection analysis." Contemporary Accounting Research (1990): Siminic, op. cit.; Palmrose, op. cit; and Abdel-Khalik, Ibid. 4. Palmrose, op. cit.; and Simunic, op. cit.

16 438 RUBIN 5. Abdel-Khalik, op. cit. 6. At the time this study was conducted there existed eight large international public accounting firms. Through mergers, there now exists six large firms. 7. Many local governments are not allowed to engage in deficit spending. Fixed fee contracts allow government officials to reduce the uncertainty surrounding the amount to be spent for auditing services in a particular budget cycle. 8. Results of a survey conducted by the author suggest that rotation policies are common in local governments to provide financial statement users with the perception of greater independence between auditor and government officials. (Rubin, Marc A. "Contracting for Municipal Audits." The Government Accountants Journal 37 (Summer 1988): 25-29). 9. For example, a municipality may change its auditor at short, regular, intervals. This type of contracting process would not necessarily allow the auditor to gain efficiencies otherwise available over a long-term relationship. Therefore, the discounting of initial year audits may be relatively small or insignificant, since the ability of the auditor to earn quasi-rents is curtailed. (DeAngelo, Linda E. "Auditor Independence, 'Low-Balling', and Disclosure Regulation." Journal of Accounting and Economics 3 (August 1981): ). 10. Simunic (op. cit.) states that spillovers or externalities are benefits which arise from interdependencies or interactions between the production functions for two or more services. For example, an audit organization may acquire information about a client (such as their exposure to specific internal control risks) during the provision of an audit. This information may then be utilized by the audit organization if contracted for a consulting engagement (such as the design and installation of a new accounting system). 11. Audit substitutes include various types of internal controls. 12. Simunic, op. cit., p In the case where MAS knowledge reduces the setup costs of auditing, Simunic concludes that audit fees will be reduced. 13. Simunic, op. cit., p. 698.

17 PRICING RELATIONSHIP OF AUDITS Palmrose, op. cit. 15. Abdel-Khalik, op. cit. 16. Abdel-Khalik, op. cit., p Abdel-Khalik, op. cit., p Simunic, op. cit. 19. Internal audit refers to audits conducted by direct employees of the organization being audited and are not employed by an independent audit organization. 20. A survey conducted by the author provided results indicating that municipal government internal audit departments have little, if any, effect on the external audit. Results are available on request from the author. 21. See Simunic (op. cit. p. 686] for an analysis of the effect that MAS knowledge has on audit setup costs. 22. Spillover knowledge also may occur in reverse. That is, the knowledge obtained in the audit reduces costs of performing audit related services. This direction of spillover also suggests a negative relationship between audit fees and fees for audit related services. 23. Palmrose, op. cit. 24. Upon analyzing the incomplete responses, there did not appear to be a systematic exclusion of information or a common reason for the lack of provision of information. 25. Simunic, op. cit.; and Palmrose, op. cit. 26. The purchase percentages of each type of service total greater than the 40 percent because some cities purchased more than one type of service. 27. Rubin, Marc A. "Municipal Audit Fee Determinants." The Accounting Review 63 (April 1988): For those cities without a general obligation bond rating, they were classified as having noninvestment grade debt if they had per capita debt higher than the mean of the sample. Otherwise they were classified as having investment grade debt.

18 440 RUBIN 29. International City Management Association. The Municipal Year Book: 1982, International City Management Association, Washington, D.C., Rubin, op. cit. 31. Palmrose, op. cit. 32. Regressions also were performed on various subsamples partitioned by city population. Due to the number of explanatory variables, the significance levels decreased substantially. The strongest association between ARS fees and audit fees is evident in the subsample of large cities (population over 85,000). 33. Procedures to ascertain if the disturbance terms are homoscedastic with respect to population include an inspection of the regression residuals. Both bivariate and multivariate correlations were examined to determine whether there is significant multicollinearity between explanatory variables. Procedures used to identify influential observations include those described in chapter 2 of Belsley, W. R., Kuh, E. and Welsch, R. E. Regression Diagnostics: Identifying Influential Data and Sources of Collinearity, John Wiley & Sons, New york, The only influential observations that were indicated in the diagnostic tests were for the bond rating variable, where only two of the sample cities had a bond rating below investment grade. 35. See DeAngelo, op. cit. 36. Introducing a variable to control for initial audits caused the ln(ars) variable to increase slightly in significance, while all variables have the predicted sign. The model R-square stays approximately the same as reported in Table 3. Full results are available from the author on request. 37. Palmrose, op. cit.; Simunic, op. cit.; and Abdel-Khalik, op. cit. 38. Simunic, op. cit. 39. Rubin, op. cit. 40. The negative relationship between ARS and audit fees also is consistent with a knowledge spillover from auditing to ARS.

19 PRICING RELATIONSHIP OF AUDITS Palmrose, op. cit. 42. Another explanation, which is not explored here, is that municipalities purchase significant consulting services from firms other than the Big Eight and these non-big Eight purchases may indicate "problem" audits. 43. Hillison, William and Kennelley, Michael. "The Economics of Nonaudit Services." Accounting Horizons 2 (September 1988):