Fee competition among Big 4 auditors and audit quality

Abstract

Both the GAO (Public accounting firms: mandated study on consolidation and competition. GAO, Washington, 2003; Audits of public companies: continued concentration in audit market for large public companies does not call for immediate action. GAO, Washington, 2008) and the US Treasury (Advisory committee on the auditing profession: final report, 2008. http://www.tres.gov/offices/domestic-finance/acap/docs/final-report.pdf) have implied that the Big 4 dominated US audit market lacks competition. More recently, the PCAOB has expressed a somewhat different concern, i.e., that because audit committees may be primarily interested in negotiating a lower audit fee (rather than championing higher audit quality) for their clients, fee competition in the US audit market could pressure the incumbent auditor to compromise on audit quality (Doty in Keynote address: the reliability, role and relevance of the audit: a turning point, 2011. www.pcaobus.org). We utilize the notion of counterfactual fees chargeable by auditors to assess fee competition and investigate competing views on the relation between fee competition among Big 4 auditors and audit quality in US local audit markets. To operationalize fee competition at the client-level in the context of each local audit market, we compute a separate counterfactual audit fee that would be charged by every other Big 4 auditor for that particular engagement and use the minima of the counterfactuals. We validate our audit fee competition metric by showing a positive relation with the incumbent auditor’s switching risk. Collectively, our findings suggest that fee competition is useful as a mechanism for improving audit quality in the highly concentrated US audit market, albeit only in local audit markets where the incumbent auditor has below-median market power and only for higher quality clients. Overall, our findings speak to the interplay between fee competition and auditor incentives and are of potential interest to regulators such as the PCAOB concerned about competition in US audit markets.

This is a preview of subscription content, log in to check access.

Notes

  1. 1.

    We focus on the Big 4 firms because of their oligopolistic dominance of the highly concentrated US audit market and to avoid the confounding audit quality effects of Big 4 versus non-Big 4 auditors. Consistent with prior research, we use the terms “city,” “local audit market” and “CBSA” (Core Based Statistical Area) interchangeably to vary the exposition. The US Office of Management and Budget defines CBSA as an area surrounding an urban center of “at least 10,000 people and adjacent areas that are socio-economically tied to the urban center by commuting.”

  2. 2.

    As noted by DeFond and Zhang (2014), audit fee models have very high explanatory power (R-squares) and, consequently, estimates derived from audit fee models are reliable. In our study, our fee estimation model R-squares range between 75 and 79%.

  3. 3.

    As reported in Accounting Today, the commoditization of the Big 4 audit is part of the motivation behind the rapid growth of nonaudit services, albeit to nonaudit clients, at the Big 4 firms from 48 to 62% of total revenues between 2004 and 2013. PCAOB Chairman Doty (2013) notes that within 10 years for these firms audit fees are likely to amount to less than 20% of total revenues.

  4. 4.

    Cohen et al. (2010) suggest that CFOs, rather than audit committees, continue to hold power in the hiring of the auditor. Also, survey evidence of CFO beliefs suggests that earnings management is common and that approximately a fifth of public companies manage earnings in any given year (Dichev et al. 2013). Hence, it is not unreasonable to view at least some clients with a Big 4 auditor as reprobate or lower quality, i.e., as preferring a lower (rather than a higher) quality Big 4 audit. Relatedly, Doty (2013) notes that PCAOB inspections have identified and documented promises by auditors to clients to be their “trusted partner” and to support the client in obtaining a “desired outcome” in accounting matters. DeFond et al. (2000) report that in China higher audit quality is associated with loss of market share.

  5. 5.

    Our discussions with Big 4 auditors suggest that faced with audit fee competition from another Big 4 auditor, the incumbent may offer incentives that are often personnel related. In other words, given the high levels of staff turnover in public accounting, it can be frustrating for a client to have to answer the same questions repeatedly or otherwise “train” the auditor’s staff year after year. Hence, an offer of personnel-related incentives (such as more senior/experienced personnel assigned to the audit) may be sufficient to ward-off a client bid process with little or no fee discount.

  6. 6.

    “If we want the audit profession to compete on quality more than price, we’ve got to provide markets more information about the audit,” Doty (2013, p. 9). Proposed PCAOB initiatives such as a more explanatory audit report (opinion), identifying the engagement partner on an audit, disclosure of other firms participating in the audit are all intended to help investors better judge audit quality.

  7. 7.

    As noted previously (fn. 4), the incumbent Big 4 auditor may offer to assign its “A-team” (i.e., its more senior and experienced personnel) to the client’s audit. Given the high levels of staff turnover in public accounting, it can be frustrating for a client to have to answer the same questions repeatedly or otherwise “train” the auditor’s staff year after year. Hence, an offer of personnel-related incentives may be sufficient to ward-off a client bid process.

  8. 8.

    Prior research (e.g., Banker et al. 2003) on the audit industry production function pools data at the national level for both Big 4 and non-Big 4 audit firms, i.e., implicitly assumes that the production function is the same for all (Big 4 as well as non-Big 4) audit firms. By contrast, we examine the audit fee model for each Big 4 firm separately for each year over our study period (2005–2015). Put differently, we add to the degrees of freedom in our estimation of AFCOMP by allowing the fee models (Appendices 1 and 3) to be different for each Big 4 firm each year. The significant differences in the coefficients of the annual regressions for the Big 4 firms (reported in Appendix 1) justify this modeling choice consistent with the reasonable notion that these audit firms are not homogenous and that pricing models can vary by audit firm.

  9. 9.

    We are grateful to Elaine Mauldin for sharing the audit committee related data.

  10. 10.

    All variables are winsorized in the (1, 99%) range. To avoid sample attrition, mean values for the industry-year are used for missing control variables. Exclusion of client-years with such missing data does not change conclusions.

  11. 11.

    Significant correlations among the independent variables (other than the test variables AFCOMP_NEG and AFCOMP_POS) are not a concern since they do not affect any of our test results or interpretations, and are therefore not reported for brevity. Later in the study we report VIFs.

  12. 12.

    Recall that AFCOM_NEG implies absence of audit fee competition, i.e., for this particular audit engagement no other Big 4 auditor has a counterfactual fee that is lower than that of the incumbent Big 4 auditor.

  13. 13.

    Additionally, the correlation of AFCOMP with abnormal audit fees for AFCOMP > 0 is 0.0169 and for AFCOMP < 0 is 0.0354. This further confirms that our competition measure is fundamentally different from abnormal audit fees.

  14. 14.

    In Table 7, for lower quality clients, our test variable AFCOMP_POS is significant only for one audit quality proxy (DACC), i.e., significant with a positive sign in column 1, which appears to provide weak support for the PCAOB concern that audit fee competition may lower audit quality. However, this concern applies only for lower quality clients, i.e., clients where the relative power of the audit committee vis-à-vis the CFO is below median.

  15. 15.

    We utilize 1-digit (rather than 2-digit) SIC for practical reasons. With 2-digit SIC, there are over 50 industries, and 4 Big 4 auditors × 11 years × 50 industries implies over 2200 regressions with the 12,618 observations in our sample (see Table 1), i.e., an average of only 6 observations per regression. Even with the 1-digit SIC, for industries with SIC codes 0 and 9 we do not have enough observations to run by year and auditor; hence, for these 2 industries (51 + 48 = 99 observations) we run the regressions only by auditor.

  16. 16.

    One limitation of AFCOMP is that we do not know which audit office of the other Big 4 firm would perform the engagement if in fact the lowest counterfactual fee was accepted by the client. However, when we use the incumbent auditor’s attributes and the characteristics of the incumbent’s CBSA in estimating AFCOMP, sequential regression analysis shows that 96% of the explanatory power of model (3) comes from client characteristics and only 4% from audit office and audit market characteristics. Hence, the measurement error in AFCOMP is expected to be minimal. Moreover, any measurement error in AFCOMP is likely to bias the coefficient against being significant.

References

  1. Antle R, Gordon E, Narayanamoorthy G, Zhou L (2006) The joint determination of audit fees, non-audit fees, and abnormal accruals. Rev Quant Financ Acc 27:235–266

    Article  Google Scholar 

  2. Ashbaugh H, LaFond R, Mayhew B (2003) Do nonaudit services compromise auditor independence? Further evidence. Account Rev 78(3):611–639

    Article  Google Scholar 

  3. Asthana S (2017) Diversification by the audit offices in the US and its impact on audit quality. Rev Quant Financ Acc 48:1003–1030

    Article  Google Scholar 

  4. Asthana S, Boone J (2012) Abnormal audit fee and audit quality. Audit J Pract Theory 31(3):1–22

    Article  Google Scholar 

  5. Asthana S, Raman KK, Xu H (2015) US-listed foreign companies’ choice of a US-based vs. home country-based Big N principal auditor and the effect on audit fees and earnings quality. Account Horiz 29(3):631–666

    Article  Google Scholar 

  6. Ball R, Jayaraman S, Shivakumar L (2012) Audited financial reporting and voluntary disclosure as complements: a test of the confirmation hypothesis. J Account Econ 53:136–166

    Article  Google Scholar 

  7. Balsam S, Krishnan J, Yang J (2003) Auditor industry specialization and earnings quality. Audit J Pract Theory 22:71–97

    Article  Google Scholar 

  8. Banker R, Chang H, Cunningham R (2003) The public accounting industry production function. J Account Econ 35:255–281

    Article  Google Scholar 

  9. Baumol W, Panzar J, Willig R (1982) Contestable market and the theory of market structure. Harcourt Brace Jovanovich, Inc

  10. Beasley M, Carcello J, Hermanson D, Neal T (2009) The audit committee oversight process. Contemp Account Res 26(1):65–122

    Article  Google Scholar 

  11. Beck M, Mauldin E (2014) Who’s really in charge? Audit committee versus CFO power and audit fees. Account Rev 89(6):2057–2085

    Article  Google Scholar 

  12. Beck M, Francis J, Gunn J (2013) Auditing and city-level human capital. Working paper, University of Missouri-Columbia, MO

  13. Bell T, Landsman W, Shackelford D (2001) Auditors’ perceived business risk and audit fees: analysis and evidence. J Account Res 39:35–43

    Article  Google Scholar 

  14. Bell T, Doogar R, Solomon I (2008) Audit labor usage and fees and business risk auditing. J Account Res 46:729–760

    Article  Google Scholar 

  15. Belsley DA, Kuh E, Welsch RE (1980) Regression diagnostics: identifying influential data and sources of collinearity. Wiley, New York

    Google Scholar 

  16. Boone J, Khurana IK, Raman KK (2012) Audit market concentration and auditor tolerance of earnings management. Contemp Account Res 29(4):1171–1203

    Article  Google Scholar 

  17. Boone J, Khurana IK, Raman KK (2015) Did the 2007 PCAOB disciplinary order against Deloitte impose actual costs on the firm or improve its audit quality? Account Rev 90(2):405–441

    Article  Google Scholar 

  18. Butler M, Leone A, Willenborg M (2004) An empirical analysis of auditor reporting and its association with abnormal accruals. J Account Econ 37(2):139–165

    Article  Google Scholar 

  19. Cabral L (2017) Introduction to industrial organization, 2nd edn. MIT Press, Cambridge

    Google Scholar 

  20. Chen L, Krishnan G, Pevzner M (2012) Pro forma disclosures, audit fees, and auditor resignations. J Account Public Policy 31:237–257

    Article  Google Scholar 

  21. Choi J, Kim J, Zang Y (2010) The association between audit quality and abnormal audit fees. Audit J Pract Theory 29(2):115–140

    Article  Google Scholar 

  22. Choi J, Kim J, Qui A, Zang Y (2012) Geographic proximity between auditor and client: how does it impact audit quality. Audit J Pract Theory 31(2):43–72

    Article  Google Scholar 

  23. Chow GC (1960) Tests of equality between sets of coefficients in two linear regressions. Econometrica 28(3):591–605

    Article  Google Scholar 

  24. Chung H, Kallapur S (2003) Client importance, nonaudit services, and abnormal accruals. Account Rev 78(4):931–956

    Article  Google Scholar 

  25. Cohen J, Krishnamurthy G, Wright A (2010) Corporate governance in the post-SOX era: auditors’ experiences. Contemp Account Res 27(3):751–786

    Article  Google Scholar 

  26. Dechow P, Ge W, Larson C (2011) Predicting material accounting misstatements. Contemp Account Res 28(1):17–82

    Article  Google Scholar 

  27. DeFond ML, Jiambalvo J (1994) Debt covenant violation and manipulation of accruals. J Account Econ 17(January):145–176

    Article  Google Scholar 

  28. DeFond M, Zhang J (2014) A review of archival auditing research. J Account Econ 58(2):275–326

    Article  Google Scholar 

  29. DeFond M, Wong T, Li S (2000) The impact of improved auditor independence on audit market concentration in China. J Account Econ 28:269–305

    Article  Google Scholar 

  30. DeFond ML, Raghunandan K, Subramanyam KR (2002) Do non-audit service fees impair auditor independence? Evidence from going concern audit opinions. J Account Res 40(4):1247–1274

    Article  Google Scholar 

  31. Demsetz H (1973) The market concentration doctrine. American Enterprise Institute, Stanford

    Google Scholar 

  32. Dichev ID, Graham JR, Harvey CR, Rajgopal S (2013) Earnings quality: evidence from the field. J Account Econ 56:1–33

    Article  Google Scholar 

  33. Doogar R, Sivadasan P, Solomon I (2015) Audit fee residuals: costs or rent? Rev Acc Stud 20(4):1247–1286

    Article  Google Scholar 

  34. Doty J (2011) Keynote address: the reliability, role and relevance of the audit: a turning point. May 5. www.pcaobus.org

  35. Doty J (2013) The role of the audit in the global economy. April 18. www.pcaobus.org

  36. Doyle J, Weili G, McVay S (2007) Accruals quality and internal control over financial reporting. Account Rev 82:1141–1170

    Article  Google Scholar 

  37. Dye R (1991) Informationally motivated auditor replacement. J Account Econ 14:347–374

    Article  Google Scholar 

  38. Ettredge M, Fuerherm E, Li C (2014) Fee pressure and audit quality. Acc Organ Soc 39:247–263

    Article  Google Scholar 

  39. Ettredge M, Fuerherm E, Guo F, Li C (2017) Client pressure and auditor independence: evidence from the ‘‘Great Recession” of 2007–2009. J Account Public Policy 36(4):262–283

    Article  Google Scholar 

  40. Francis JR, Michas P (2013) The contagion effect of low-quality audits. Account Rev 88(2):521–552

    Article  Google Scholar 

  41. Francis JR, Yu M (2009) Big 4 office size and audit quality. Account Rev 84(5):1521–1552

    Article  Google Scholar 

  42. Francis JR, Pinnuck M, Watanabe O (2013) Auditor style and financial statement comparability. Account Rev 89(2):605–633

    Article  Google Scholar 

  43. Frankel R, Johnson M, Nelson K (2002) The relation between auditors’ fees for nonaudit services and earnings quality. Account Rev 77(Supplement):71–105

    Article  Google Scholar 

  44. Geiger MA, North DS (2006) Does hiring a new CFO change things? An investigation of changes in discretionary accruals. Account Rev 81(4):781–809

    Article  Google Scholar 

  45. Goldschmeid H, Mann H, Weston J (1974) Industrial concentration: the new learning. Columbia University Center for Law and Economic Studies, New York

    Google Scholar 

  46. Government Accountability Office (GAO) (2003) Public accounting firms: mandated study on consolidation and competition. GAO, Washington

    Google Scholar 

  47. Government Accountability Office (GAO) (2008) Audits of public companies: continued concentration in audit market for large public companies does not call for immediate action. GAO, Washington

    Google Scholar 

  48. Gul FA, Fung SYK, Jaggi B (2009) Earnings quality: some evidence on the role of auditor tenure and auditors’ industry expertise. J Account Econ 47:265–287

    Article  Google Scholar 

  49. Higgs J, Skantz T (2006) Audit and nonaudit fees and the market’s reaction to earnings announcements. Audit J Pract Theory 25(1):1–26

    Article  Google Scholar 

  50. Hope OK, Langli JC (2010) Auditor independence in a private firm and low litigation risk setting. Account Rev 85(2):573–605

    Article  Google Scholar 

  51. Hribar P, Collins D (2002) Errors in estimating accruals: implications for empirical research. J Account Res 40(1):105–134

    Article  Google Scholar 

  52. Hribar P, Nichols D (2007) The use of unsigned earnings quality measures in tests of earnings management. J Account Res. 45(5):1017–1053

    Article  Google Scholar 

  53. Hribar P, Kravet T, Wilson R (2014) A new measure of accounting quality. Rev Acc Stud 19:506–538

    Article  Google Scholar 

  54. Jaggi B, Mitra S, Hossain M (2015) Earnings quality, internal control weaknesses and industry-specialist audits. Rev Quant Financ Acc 45:1–32

    Article  Google Scholar 

  55. Jensen K, Kim JM, Yi H (2015) The geography of US auditors: information quality and monitoring costs by local versus non-local auditors. Rev Quant Financ Acc 44:513–549

    Article  Google Scholar 

  56. Jha A, Chen Y (2015) Audit fees and social capital. Account Rev 90(2):611–639

    Article  Google Scholar 

  57. Jones J (1991) Earnings management during import relief investigations. J Account Res 29(2):193–228

    Article  Google Scholar 

  58. Kaplan SE, Williams DD (2013) Do going concern audit reports protect auditors from litigation? A simultaneous equations approach. Account Rev 88:199–232

    Article  Google Scholar 

  59. Kinney W (2005) Twenty-five years of audit deregulation and re-regulation: what does it mean for 2005 and beyond? Audit J Pract Theory 24(Suppl):89–109

    Google Scholar 

  60. Kinney W, Libby R (2002) Discussion of the relation between auditors’ fees for nonaudit services and earnings management. Account Rev 77(Supplement):107–114

    Article  Google Scholar 

  61. Klein B, Leffler K (1981) The role of market forces in assuring contractual performance. J Polit Econ 89:615–641

    Article  Google Scholar 

  62. Kothari K, Leone A, Wasley C (2005) Performance matched discretionary accruals. J Account Econ 39(1):163–197

    Article  Google Scholar 

  63. Krishnan J, Wen Y, Zhao W (2011) Legal expertise on corporate audit committees and financial reporting quality. Account Rev 86(6):2099–2130

    Article  Google Scholar 

  64. Larcker D, Richardson S (2004) Fees paid to audit firms, accrual choices, and corporate governance. J Account Res 42(3):625–658

    Article  Google Scholar 

  65. Lawrence A, Minutti-Meza M, Zhang P (2011) Can Big 4 versus non-Big 4 differences in audit-quality proxies be attributed to client characteristics? Account Rev 86(1):259–286

    Article  Google Scholar 

  66. Lennox C, Li B (2012) The consequences of protecting audit partners’ personal assets from the threat of liability. J Account Econ 54:154–173

    Article  Google Scholar 

  67. Magee R, Tseng M (1990) Audit pricing and independence. Account Rev 65(2):315–336

    Google Scholar 

  68. Menon K, Williams D (2004) Former audit partners and abnormal accruals. Account Rev 79(4):1095–1118

    Article  Google Scholar 

  69. Michas P (2011) The importance of audit profession development in emerging market countries. Account Rev 86(5):1731–1764

    Article  Google Scholar 

  70. Myers LA, Schmidt J, Wilkins M (2014) An investigation of recent changes in going concern reporting decisions among Big N and non-Big N auditors. Rev Quant Financ Acc 43:155–172

    Article  Google Scholar 

  71. Newton N, Wang D, Wilkins M (2013) Does a lack of choice lead to lower quality? Evidence from auditor competition and client restatements. Audit J Pract Theory 32(3):31–67

    Article  Google Scholar 

  72. Numan W, Willekens M (2012) An empirical test of spatial competition in the audit market. J Account Econ 53(1–2):450–465

    Article  Google Scholar 

  73. O’Keefe T, Simunic D, Stein M (1994) The production of audit services: evidence from a major public accounting firm. J Account Res 32:241–261

    Article  Google Scholar 

  74. Oster S (1999) Modern competitive analysis. Oxford University Press, New York

    Google Scholar 

  75. Prawitt DF, Smith J, Wood D (2009) Internal audit quality and earnings management. Account Rev 84(4):1255–1280

    Article  Google Scholar 

  76. Ruddock C, Taylor S, Taylor S (2006) Nonaudit services and earnings conservatism: is auditor independence impaired? Contemp Account Res 23(3):701–746

    Article  Google Scholar 

  77. Scherer F (1996) Industry structure, strategy, and public policy. Harper Collins, New York

    Google Scholar 

  78. Sheth J, Sisodia R (2002) The rule of three. The Free Press, Mankato

    Google Scholar 

  79. Shu S (2000) Auditor resignations: clientele effects and legal liability. J Account Econ 29:173–205

    Article  Google Scholar 

  80. Simunic D (1980) The pricing of audit services: theory and evidence. J Account Res 18(1):161–190

    Article  Google Scholar 

  81. Stiglitz J (1987) Competition and the number of firms in a market: are duopolies more competitive than atomistic markets. J Pol Eco 95(5):1041–1061

    Article  Google Scholar 

  82. US Treasury (2008) Advisory committee on the auditing profession: final report. October 6. http://www.tres.gov/offices/domestic-finance/acap/docs/final-report.pdf

  83. Wang C, Raghunandan K, McEwen R (2013) Non-timely 10-K filings and audit fees. Account Horiz 27(4):737–755

    Article  Google Scholar 

  84. Whisenant S, Sankaraguruswamy S, Raghunandan K (2003) Evidence on the joint determination of audit and non-audit fees. J Account Res 41(4):721–744

    Article  Google Scholar 

Download references

Acknowledgements

We thank the editor and two reviewers for their helpful comments and suggestions. We also thank Jong-Hag Choi, Clive Lennox, workshop participants at the University of Texas at San Antonio and the Chinese University of Hong Kong, and attendees at the AAA 2016 Annual Meeting (2016) and the Journal of Contemporary Accounting and Economics 2017 Symposium for their feedback and suggestions. K. K. Raman acknowledges support from the Ramsdell Endowed Chair for Accounting at The University of Texas at San Antonio.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Sharad Asthana.

Appendices

Appendix 1: Audit fee competition (AFCOMP) variable

To estimate AFCOMP (audit fee competition) among Big 4 auditors we estimate the following audit fee model (based on prior research, such as Jha and Chen 2015; Choi et al. 2010; Larcker and Richardson 2004; Simunic 1980; Wang et al. 2013; Whisenant et al. 2003; Antle et al. 2006):

$$\begin{aligned} {\text{LAFEE}} & =\upgamma_{0} + \,\upgamma_{1} {\text{LSIZE}} +\upgamma_{2} {\text{ROA}} +\upgamma_{3} {\text{CFFO}} +\upgamma_{4} {\text{B}}2{\text{M}} +\upgamma_{5} {\text{LEV}} +\upgamma_{6} {\text{CASSET}} +\upgamma_{7} {\text{INVRATIO}} \\ & \quad + \gamma_{8} {\text{SEGMENTS}} +\upgamma_{9} {\text{FOPS}} +\upgamma_{10} {\text{ACQUIRE}} +\upgamma_{11} {\text{FINANCE}} +\upgamma_{12} {\text{PINTAN}} +\upgamma_{13} {\text{REL}}\_{\text{AC}}\_{\text{CFO}} \\ & \quad +\upgamma_{14} {\text{FSCORE}} +\upgamma_{15} {\text{TENURE}} +\upgamma_{16} {\text{INDSP}} +\upgamma_{17} {\text{LOFFICE}} +\upgamma_{18} {\text{LDELAY}} +\upgamma_{19} {\text{LNAFEE}} \\ & \quad +\upgamma_{20} {\text{MODOP}} +\upgamma_{21} {\text{ICMW}} +\upgamma_{22} {\text{BUSY}} +\upgamma_{23} {\text{HERF}} +\upgamma_{24} {\text{DISTANCE}} +\upgamma_{25} {\text{LEADER}} +\upgamma_{26} {\text{LNUMEST}} \\ & \quad +\upgamma_{27} {\text{STDEST}} + error \\ \end{aligned}$$
(3)

All variables are defined in Appendix 2. The model is estimated separately for each Big 4 auditor, by year and by industry (1-digit SIC code).Footnote 15 Next, we use the model parameters to estimate the counterfactual audit fee that each Big 4 auditor (other than the incumbent) would charge, holding client and auditor characteristics constant (with the exception of TENURE which is set = 0) for the client in question. Audit fee (AFEE) is then calculated as the exponential of LAFEE. We then compare the incumbent Big 4 auditor’s AFEE with the counterfactual AFEE for each of the other three Big 4 auditors. Then AFCOMP = the incumbent’s AFEE minus the lowest counterfactual AFEE for any other Big 4 auditor, deflated by client total assets.Footnote 16 See Fig. 1 for more information. Then, AFCOMP_POS = AFCOMP for AFCOMP > 0, and 0 otherwise (see Panel A) and AFCOMP_NEG = AFCOMP for AFCOMP < 0, and 0 otherwise (see Panel B).

Fig. 1
figure1

Audit fee competition (AFCOMP) variable. a AFCOMP is positive (AFCOMP_POS). b AFCOMP is negative (AFCOMP_NEG)

Since it is not practical to present 352 regressions (4 Big 4 auditors × 11 years × 8 industries), in Appendix 3 we present the pooled versions of each of the four Big 4 firm regressions with fixed-effects for years and industries. We also compare the coefficients of each of the Big 4 firm regressions taken as a set as well as individually. The 6 Chow tests comparing the 4 sets of coefficients with each other show that all of them are significantly different at 1% level with the F values ranging from 3.89 to 5.37. We also compare the 27 individual coefficients (6 × 27 independent variables = 162 tests). Of these 162 tests, 124 are significant at 10% or better which suggests that the audit fee pricing models of the Big 4 are significantly different from each other, i.e., the audit fees charged by them for any individual client is likely to be significantly different from each other.

In Fig. 1, each Big 4 line represents the audit fee model (see Appendix 1) for each of the four Big 4 firms, including the incumbent Big 4 firm. Point XC on the X-axis represents the client-specific audit fee model vector of the independent variables in the fee model for client C for a particular year. For client C, the lowest counterfactual fee is from Big 4 firm 3. Hence, variable AFCOMP_POS (NEG) is equal to the incumbent’s audit fee minus the counterfactual audit fee of the Big 4 firm 3, deflated by client total assets and can be positive or negative (as depicted in the two panels).

Appendix 2: Variable definitions

Variable Definition
Dependent variables
DACC Performance adjusted discretionary accruals calculated using the modified Jones model (DeFond and Jiambalvo 1994) controlling for concurrent performance based on 2-digit SIC code and year (Kothari et al. 2005), deflated by beginning of fiscal year total assets. We use the difference between net income and cash from operations as our measure of total accruals (Hribar and Collins 2002). The higher the DACC, the lower the quality of audited earnings and the lower the implied audit quality
MBEX A dichotomous variable equal to 1 if the firm meets or beats the earnings expectation (proxied by the most recent median consensus analyst forecast available on IBES file) by one cent or less; 0 otherwise. The higher the probability of MBEX, the lower the quality of audited earnings and the lower the implied audit quality
PROFIT Dichotomous variable equal to 1 if the net income before extraordinary items and cumulative effect of accounting changes deflated by lagged total assets is between 0 and 5%; 0 otherwise (Francis and Yu 2009). The higher the probability of PROFIT, the lower the quality of audited earnings and the lower the implied audit quality
GCOPN Dichotomous variable equal to 1 if the client receives a going concern opinion in the current year; 0 otherwise. The higher the probability of GCOPN, the higher the implied audit quality
Test variables
AFCOMP Audit fee competition among Big 4 auditors measured as the incumbent Big 4 firm’s actual audit fee minus the lowest counterfactual fee from another Big 4 firm (which may or may not have an office in the local audit market) deflated by the client’s total assets. The counterfactual fee is based on cross-sectional audit fee regressions run by year, by industry (1 digit SIC code), and by each Big 4 auditor, controlling for client-specific, local audit office-specific, and local audit market-specific factors
AFCOMP_POS Equals AFCOMP for AFCOMP > 0 and 0 otherwise
AFCOMP_NEG Equals AFCOMP for AFCOMP < 0 and 0 otherwise
Control variables
ACQUIRE A dichotomous variable equal to 1 if the client is involved in acquisition activities during the year; and 0 otherwise
B2M Book-to-market equity ratio at the end of the fiscal year
BUSY Dummy variable equal to 1 for December 31st fiscal-year-end clients; 0 otherwise
CASSET Ratio of current assets to total assets
CFFO Cash flow from operations divided by total assets
DISTANCE Spatial distance metric (based on Numan and Willekens 2012) defined as the absolute fee market share difference between the incumbent Big 4 auditor and closest Big 4 competitor in the local industry audit market fiscal year. A local industry audit market consists of all companies within a two-digit SIC code in a CBSA. The lower the metric, the higher the competition in the local industry audit market.
EGROWTH Annual growth rate of net income before extraordinary items and cumulative effect of accounting changes
FINANCE A dichotomous variable equal to 1 if number of outstanding shares increased by at least 10% or long-term debt increased by at least 20% during the year (Geiger and North 2006); and 0 otherwise
FOPS Proportion of a client’s total income from foreign (non-US) operations
FSCORE Fraud Score calculated using Dechow et al. (2011) methodology on page 55 (Table 7, Panel A, Model 1)
HERF Herfindahl index (concentration measure) for local industry audit market fiscal year, where a local industry audit market consists of all public clients within a two-digit SIC group in a CBSA. Defined as Σ[s/S]2, where “s” is the sum of audit fees of the Big 4 audit office from all clients within the 2-digit SIC industry, and “S” is the total audit fees of all Big 4 auditors in the CBSA from all clients in that industry (Numan and Willekens 2012)
ICMW Number of material internal control weaknesses reported in Audit Analytics
INDSP Measure of industry specialization, defined as a dichotomous variable equal to 1 when an audit firm has a fee market share of at least 30% in an audit market, 0 otherwise. An audit market is defined as a two-digit SIC industry in a CBSA (Numan and Willekens 2012)
INVRATIO Inventory deflated by total assets
LAFEE Natural log of audit fee during the current fiscal year
LDELAY Natural log of 1 plus the number of calendar days from fiscal year-end to the date of the audit report
LEADER An indicator variable equal to1 when an audit firm has the largest fee market share in an audit market, 0 otherwise. An audit market is defined as a two-digit SIC industry in a CBSA (Numan and Willekens 2012)
LEV Long term debt plus debt in current liabilities deflated by average total assets (Lawrence et al. 2011)
LNAFEE Natural log of non-audit fees during the current fiscal year
LNUMEST Natural log of the number of analysts’ forecasts
LOFFICE Natural log of total annual audit fees of the local office of the incumbent Big 4 auditor (Francis and Yu 2009)
LSIZE Natural log of the client’s total assets (in millions of dollars)
MODOP A dichotomous variable equal to 1 if the audit opinion is modified (different from the standard three paragraph report); 0 otherwise
PINTAN Proportion of intangible assets to total assets
REL_AC_CFO Power of the audit committee relative to CFO as defined on page 2065 of Beck and Mauldin (2014)
ROA Net income before extraordinary items and cumulative effect of accounting changes deflated by total assets
SDCFFO Standard deviation of cash flow from operations deflated by total assets, calculated over the current and prior four years
SDEARN Standard deviation of earnings deflated by total assets, calculated over the current and prior 4 years
SDSALES Standard deviation of sales deflated by total assets, calculated over the current and prior 4 years
SEGMENTS Number of segments reported in Compustat segment file
STDEST Standard deviation of analysts’ earnings forecasts
TENURE Number of years the client has been with the current auditor

Appendix 3: Big 4 audit fee models (dependent variable = LAFEE)

Variables Deloitte EY KPMG PWC
Estimate p value Estimate p value Estimate p value Estimate p value
Intercept 8.7539*** < 0.0001 7.0932*** < 0.0001 7.8903*** < 0.0001 7.5844*** < 0.0001
Client-specific variables
LSIZE 0.5156*** < 0.0001 0.4393*** < 0.0001 0.4964*** < 0.0001 0.4668*** < 0.0001
ROA − 0.5378*** < 0.0001 − 0.5381*** < 0.0001 − 0.5740*** < 0.0001 − 0.4686*** < 0.0001
CFFO − 0.4594*** 0.0004 0.1910** 0.0303 0.0415 0.7082 − 0.0929 0.3631
B2M − 0.0888*** 0.0002 − 0.0093 0.6479 − 0.0734*** 0.0015 − 0.0247 0.2342
LEV 0.1357** 0.0357 0.1527*** 0.0015 0.2091*** 0.0005 0.2474*** < 0.0001
CASSET 0.6581*** < 0.0001 0.4565*** < 0.0001 0.9066*** < 0.0001 0.7136*** < 0.0001
INVRATIO − 0.1055 0.2466 0.1066 0.1551 0.1090 0.2460 0.1561* 0.0635
SEGMENTS 0.0205*** < 0.0001 0.0206*** < 0.0001 0.0282*** < 0.0001 0.0230*** < 0.0001
FOPS 0.2873*** < 0.0001 0.2962*** < 0.0001 0.3029*** < 0.0001 0.3101*** < 0.0001
ACQUIRE − 0.0020 0.9325 0.0033 0.8660 0.0592*** 0.0091 0.0553*** 0.0040
FINANCE 0.0267 0.2010 0.0188 0.2486 0.0165 0.3930 − 0.0089 0.5953
PINTAN 0.4407*** < 0.0001 0.4017*** < 0.0001 0.4440*** < 0.0001 0.1260** 0.0138
REL_AC_CFO 0.0175** 0.0124 − 0.0053 0.3129 − 0.0023 0.7178 − 0.0119 0.3290
FSCORE 0.0067 0.4265 0.0176* 0.0486 0.0026 0.7905 0.0059 0.4936
Auditor-specific variables
TENURE 0.0058*** < 0.0001 0.0044*** < 0.0001 − 0.0014 0.1881 0.0054*** < 0.0001
INDSP − 0.0203 0.4257 0.0609*** 0.0006 − 0.0314 0.2556 0.0779*** 0.0002
LOFFICE 0.0560*** < 0.0001 0.0914*** < 0.0001 0.0628*** < 0.0001 0.0624*** < 0.0001
LDELAY 0.0777* 0.0521 0.2530*** < 0.0001 0.1757*** < 0.0001 0.1983*** < 0.0001
LNAFEE 0.0361 < 0.0001 0.0751 < 0.0001 0.0628 < 0.0001 0.0771 < 0.0001
MODOP 0.0440* 0.0685 0.0798*** 0.0001 0.0558** 0.0178 0.1148*** < 0.0001
ICMW 0.1597*** < 0.0001 0.2268*** < 0.0001 0.1440*** < 0.0001 0.0787*** 0.0001
BUSY 0.0343* 0.0895 0.0540*** 0.0007 0.0021 0.9115 0.0883*** < 0.0001
Audit market − specific variables
HERF − 0.3125 0.4133 0.4207*** < 0.0001 − 0.2471 0.4018 0.2196*** 0.0079
DISTANCE 0.1057 0.2662 − 0.1298 0.2789 0.2419*** 0.0003 0.0910 0.1825
LEADER 0.0514** 0.0273 − 0.0128 0.4277 − 0.0212 0.3377 − 0.0272 0.1984
Information environment-specific variables
LNUMEST − 0.0908*** < 0.0001 − 0.0619*** < 0.0001 − 0.0605*** < 0.0001 − 0.0450*** < 0.0001
STDEST 0.0126 0.8706 0.1236** 0.0264 0.2650*** 0.0062 0.1791** 0.0113
Year dummies Yes Yes Yes Yes
Industry dummies Yes Yes Yes Yes
Observations 2611 4062 2701 3244
Adj-R square 0.7999 0.7945 0.8226 0.8496
F value 261.77 393.45 313.96 459.02
Probability > F < 0.0001 < 0.0001 < 0.0001 < 0.0001
  1. See Appendix 2 for variable definitions. ***, **, * imply two-tailed significance at 1, 5, 10% levels. Standard errors are corrected for clustering and heteroskedasticity by year and industry. For estimating audit fee competition among Big 4 auditors (variable AFCOMP), the regressions are run separately for each Big 4 auditor by year and by industry (see Appendix 1 for more information). Pooled regressions for each Big 4 auditor are shown here for presentation purposes only. Pairwise Chow (1960) tests of the coefficients reveal that the Big 4 pricing models are different from each other at 1% or better significance level (F values range from 3.89 to 5.37). Big 4 firms’ pricing models are proprietary and it is reasonable to expect different firms to put different weights on each of the various determinants of audit fees.

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Asthana, S., Khurana, I. & Raman, K.K. Fee competition among Big 4 auditors and audit quality. Rev Quant Finan Acc 52, 403–438 (2019). https://doi.org/10.1007/s11156-018-0714-9

Download citation

Keywords

  • Big 4 firms
  • Audit fee competition
  • Audit quality
  • PCAOB

JEL Classification

  • M41
  • M42