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.

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).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.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

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 X_C 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.