Fool's gold or value for money? The link between abnormal audit fees, audit firm type, fair‐value disclosures, and market valuation

• Author: Ulrike Stefani,Ulf Mohrmann,Jan Riepe
• Published date: 01 July 2019
• DOI: http://doi.org/10.1111/ijau.12155
• Date: 01 July 2019

ORIGINAL ARTICLE

Fool's gold or value for money? The link between abnormal

audit fees, audit firm type, fair‐value disclosures, and market

valuation

Ulf Mohrmann

1

|Jan Riepe

2

|Ulrike Stefani

1

1

Department of Economics, University of

Konstanz, Konstanz, Germany

2

Department of Banking, Eberhard‐Karls‐

University Tuebingen, Tuebingen, Germany

Correspondence

Ulf Mohrmann, Chair of Accounting,

Department of Economics, University of

Konstanz, Box 142, 78457 Konstanz,

Germany.

Email: ulf.mohrmann@uni‐konstanz.de

We analyze whether the audit firm type and abnormal audit fees are associated with

the market valuation of banks' fair‐value assets. Our results indicate that different

auditor types use different strategies when auditing fair values. First, we show that

Big 4 auditors restrict the Level 3 valuations to the most illiquid assets. Thus, those

banks audited by a Big 4 auditor have a lower proportion of Level 3 assets. Second,

the market discount on the Level 3 assets is higher for the banks audited by a Big 4

auditor than for those audited by a non‐Big 4 auditor. Third, the discount on the

Level 3 portfolios of banks with non‐Big 4 auditors is higher if the unexpected

audit fees are high. Thus, non‐Big 4 auditors seem to allow more high‐uncertainty

valuations but charge a risk premium. We find similar effects for industry‐specialist

auditors.

KEYWORDS

audit fees, audit firm size, banks, fair‐value accounting

1|INTRODUCTION

The idea of fair‐value accounting (FVA) is to increase the relevance of

the information contained in financial statements and, consequently,

to reduce the information asymmetries between the firm's managers

and outside parties. However, applying mark‐to‐model (so‐called

Level 3 valuations) instead of using market prices opens up the

opportunity to report high book values although the underlying asset

values decrease (Kothari & Lester, 2012; Milbradt, 2012). Accordingly,

Laux and Leuz (2010) argued that the FVA loses several of its

favorable features when market prices are not available.

The concerns surrounding opportunistic mark‐to‐model valua-

tions might be alleviated by the statutory auditor because his/her

role is to assure the appropriate classification of an asset within

the fair‐value hierarchy and to confirm the quality and soundness

of the valuation models used for the assets contained in Levels 2

and 3 of the fair‐value hierarchy. Thus, the quality of the audit

should affect the link between firms' market values and their book

values. In this vein, Song, Thomas, and Yi (2010) showed that the

market discount on Level 3 assets is less pronounced in those firms

with a higher corporate governance score. In this paper, we focus on

the auditors' strategies to deal with the valuation risk inherent in

mark‐to‐model valuations. These strategies are important for evalu-

ating the auditors' impact on FVA.

Auditors might increase the quality of FVA through two channels.

First, to restrict overvaluations, auditors might enforce that more rep-

resentative assumptions are used in the valuation models. This

strategy would directly reduce the information risk of Level 3 fair

values and, hence, would increase the reliability of Level 3 fair values.

However, it is unclear whether this is a realistic option, given the com-

plexity of the models (Bratten, Gaynor, McDaniel, Montague, & Sierra,

2013; Christensen, Glover, & Wood, 2012; Martin, Rich, & Wilks,

2006). Second, a high‐quality auditor might improve the precision of

the classification of assets as either Level 2 or Level 3 by requiring

the usage of market inputs whenever possible (Level 2) and by

restricting the Level 3 category to the most illiquid assets. This

strategy might indirectly increase the reliability of the fair‐value mea-

surements if it facilitates the identification of the risk related to Level 3

fair values. Moreover, a more restrictive classification would reduce

auditor exposure to audit risk. Botosan, Carrizosa, and Huffman

Received: 22 December 2016 Revised: 27 November 2018 Accepted: 24 February 2019

DOI: 10.1111/ijau.12155

Int J Audit. 2019;23:181–203. © 2019 John Wiley & Sons Ltdwileyonlinelibrary.com/journal/ijau 181

(2012) presented anecdotal evidence that auditors use this approach.

Alternatively, auditors might accept the higher audit risk contained in

Level 3 valuations and charge a risk premium as compensation. This

option differs from the first two strategies because it does not affect

the quality of the FVA.

Based on a sample of 363 US banks and 1,073 bank‐year observa-

tions from 2008 to 2011, we explore the (combined) effect of auditor

type and abnormal audit fees (which are observable and relevant char-

acteristics of the audit) on the credibility of discretionary asset valua-

tions. The banking sector provides an ideal setting for analyzing the

association between the banks' fair‐value assets and their market

values because FVA is more important than in other industries. Thus,

the financial sector contains a large cross‐sectional variation in the

exposure to different valuation inputs (i.e., Level 1, Level 2, and

Level 3).

We focus on two observable characteristics of the audit that

should affect the association between banks' fair‐value assets and

their market values. First, we analyze the effect of the auditor type

as measured by the Big 4/non‐Big 4 dichotomy. The auditor size is

often seen as a proxy for audit quality (Becker, DeFond, Jiambalvo,

& Subramanyam, 1998; Francis, Maydew, & Sparks, 1999). Moreover,

the fact that the relative bargaining power of the Big 4 auditors is

higher than that of the non‐Big 4 auditors might affect their strategy

choice. Second, we explore the role of the abnormal audit fees that

capture either (1) the additional effort of the auditor (Blankley, Hurtt,

& MacGregor, 2012; Eshleman & Guo, 2014; Krishnan, Sami, & Zhang,

2005; Mitra, Deis, & Hossain, 2009), (2) a potential risk premium

charged by the audit firm (see Hay, Knechel, & Wong, 2006, and the

literature cited therein), or (3) the level of auditor–client bonding

(Asthana & Boone, 2012; Choi, Kim, & Zang, 2010). If auditors

improve the reliability of the valuation models by exerting more audit

effort, the abnormal audit fees should reflect higher quality. In that

sense, the higher price for the audit is valuable for the investors. If,

however, auditors just charge a risk premium or if higher fees indicate

bonding, the higher price charged for the audit does not provide

additional benefits for the investors because it does not indicate an

increased reliability of the fair‐value measurement. Given that—

according to prior research (e.g., Higgs & Skantz, 2006)—investors

perceive positive abnormal audit fees as an indication for high

audit quality, these additional audit fees are misleading, and thus only

“fool's gold.”

To account for the endogenous formation of the auditor–client

pairs, we run all our tests for the full sample and for a reduced sample

based on propensity score matching. We find qualitatively similar

results for both samples. Specifically, we find a significantly negative

link between the Big 4 indicator and the proportion of Level 3 assets;

that is, banks with Big 4 auditors have lower shares of Level 3 assets.

Turning to Tobin's Q, we find that the proportion of Level 3 assets is

negatively related to the banks' market valuations. The Big 4 indicator

has a positive main effect, indicating that audits conducted by the

Big 4 are regarded as audits of higher quality. The discount on the

Level 3 assets is more pronounced for the Big 4 auditors. However,

this higher discount does not mean that Big 4 auditors provide audits

of lower quality. Combined with our first result (i.e., banks with Big 4

auditors have lower shares of Level 3 assets), the Big 4 auditors

appear to restrict the usage of the Level 3 valuations to the most illiq-

uid assets and enforce the usage of the Level 2 category instead,

which has, on average, superior information quality. In doing so, they

allow a better identification of the information risk included in the

financial statements. If we evaluate the total fair‐value portfolio (for

which the classification decision is irrelevant), we find a higher valua-

tion for the banks with Big 4 auditors, which is again an indication

that the perceived audit quality is higher for the Big 4. The non‐Big 4

auditors, in contrast, allow the Level 3 valuations to a larger extent

and, as a consequence, the assets' riskiness is, on average, lower.

We find a significantly negative association between the interaction

of the abnormal audit fees and the Level 3 assets and Tobin's Q,

whereas the coefficient on the interaction term is nonsignificant in

the Big 4 subsample. That is, the investors' risk assessments of the

Level 3 assets is positively associated with the abnormal audit fees

if a non‐Big 4 auditor audits the bank, which favors the interpretation

of the abnormal audit fees as a risk premium. In the robustness

checks, we use the auditors' industry specializations as our proxy

for auditor type and find results that are similar to those for the Big 4

auditors.

Our results are also economically significant. The Big 4 auditors

restrict the Level 3 holdings by around one third of the Level 3 hold-

ings of the banks audited by a non‐Big 4. Moreover, an increase in

the abnormal audit fees of one standard deviation leads to a Level 3

discount that is 42% higher than for the average bank. Although the

book value of the Level 3 assets is small in magnitude, Level 3 assets

severely affect the market value of equity due to the banks' high lever-

age ratios. The median ratio of the Level 3 assets to the market value

is 20%.

Our findings are in line with the argument that audit firms with

different characteristics apply different strategies: The high‐quality

auditors (i.e., Big 4 auditors) limit their audit risk by restricting the

use of the Level 3 valuations, and, therefore, the reported proportion

of these valuations reflects the assets' riskiness. In contrast, the non‐

Big 4 auditors seem to be unable or unwilling to constrain the usage

of the Level 3 category (i.e., the Level 3 assets also contain assets that

could be classified as Level 2, but they would probably have a lower

value if measured using market inputs). These auditors, rather, charge

additional audit fees as a compensation for the higher risk contained in

the potentially overreported values of the Level 3 assets. This results

in a negative combined effect of abnormal audit fees and the Level 3

assets for the banks audited by a non‐Big 4 auditor. Supplementary

analyses reinforce the validity of this interpretation over possible

alternative explanations, like high‐quality auditors avoiding accepting

those clients with high levels of Level 3 fair values.

Our paper contributes to the literature in several ways. First, our

results indicate that different auditor types use different strategies

when auditing banks' fair‐value portfolios. In doing so, we contribute

to the literature on the determinants of the use of Level 3 fair values.

Botosan et al. (2012) presented early evidence on different frictions

(e.g., inadequate understanding of the requirements of Statement of

182 MOHRMANN ET AL.