See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/228273267

Voluntary Audits Versus Mandatory Audits

Article  in  The Accounting Review · August 2011

DOI: 10.2308/accr-10098

CITATIONS READS

66 971

2 authors:

Clive S. Lennox Jeffrey Pittman

Nanyang Technological University Memorial University of Newfoundland
56 PUBLICATIONS   2,442 CITATIONS    48 PUBLICATIONS   1,291 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Clive S. Lennox on 11 November 2014.

The user has requested enhancement of the downloaded file.

 Voluntary Audits versus Mandatory Audits*

Clive Lennox**
Nanyang Technological University, Singapore 639798
[email protected]

Jeffrey Pittman
Memorial University of Newfoundland, St. John’s, NL A1B 3X5, Canada
[email protected]

September 2009

* We appreciate helpful comments from Elisabeth Dedman, Omrane Guedhami, Gilles Hilary,
Siew-Hong Teoh, Nan Zhou, and seminar participants at Nanyang Technological University on
an earlier version of this paper.
** Corresponding author: [email protected].

1
 Voluntary Audits versus Mandatory Audits

Abstract
An important policy question is whether audits of financial statements
should be mandatory or voluntary. A major economic rationale for
mandatory auditing is that financial statement users enjoy benefits
from having access to reliable audited information for every company.
On the other hand, advocates for voluntary auditing argue that
imposing audits removes the signaling value stemming from
companies revealing whether they wish to be audited. To evaluate
these arguments empirically, we examine a setting in which
mandatory audits were replaced by voluntary audits for private UK
companies. We first demonstrate that the benefits from forcing
companies to be audited are modest because companies that do not
want to be audited choose low quality audit firms and pay lower audit
fees during the mandatory regime. Second, we show that there is an
important signaling benefit when companies are allowed to choose
whether they will be audited. After mandatory audits are replaced by
a voluntary regime, companies that retain their auditors enjoy
significant upgrades to their credit ratings. In contrast, companies that
dispense with being audited suffer downgrades to their credit ratings
since avoiding an audit sends a negative signal and their unaudited
financial statements are less credible.

2
1. Introduction

Requiring independent audits is an important policy mechanism available to

governments to regulate the supply of reliable accounting information to investors

(Barton and Waymire, 2004). Information is a public good and, as with all public goods,

there is a concern that too little would be supplied under private contracting.1

According to this market failure argument, companies should be compelled to have

their financial statements audited to ensure that outsiders can have access to reliable

information. In the other direction, the supporters of voluntary audits stress that the

mandatory requirement suppresses the signal that is conveyed when companies can

choose whether to be audited (Sunder, 2003). The purpose of this study is to provide

empirical evidence on the merits of these competing arguments by analyzing outcomes

stemming from voluntary versus mandatory audits.

External audits are imposed on all publicly traded companies in the US and UK,

although these countries differ in their policies towards private companies. In contrast

to the US where audits of private companies remain voluntary, they were mandatory in

the UK until recently.2 In 1994, the EU Fourth Directive permitted national governments

to abandon the requirement that small companies submit to an audit. A subsequent

1 Firms operating in an unregulated environment understandably focus on their own costs and
benefits without considering the socially optimal level of disclosure. Zingales (2009: 394) stresses
that firms disclose sub-optimally because firm-level benefits from disclosure are smaller than its
society-level benefits: “General Motors’ disclosure helps investors evaluate Ford, but GM will
never internalize this benefit.” Since the private and social values of information can diverge,
regulation is frequently justified on the grounds that this induces positive externalities (e.g., Dye,
1990; Admati and Pfleiderer, 2000; Lambert et al., 2007). However, other theory implies that,
rather than maximizing social welfare, mandating disclosure of accounting information can
generate negative externalities (e.g., Fishman and Hagerty, 1989). Leuz and Wysocki (2008)
survey this literature.
2 We do not examine the effect of the 1994 change in regulation because our data source provides

financial statement information for the most recent ten years only. Collis et al. (2004) recount the
specific amendments to UK legislation governing statutory audit exemptions.

3
amendment to the UK Companies Act further relaxed the eligibility thresholds,

culminating with private companies qualifying for the audit exemption in fiscal years

ending after January 30, 2004 if their sales did not exceed £5.6m and total assets did not

exceed £2.8m. Prior to this date, it had been permissible for private companies to avoid

having an audit if their sales did not exceed £1m and their total assets did not exceed

£1.4m. The shift in exemption eligibility thresholds in 2004 enables us to assemble a

sample of companies that were affected by the regime switch.3 Fewer than 6,000

companies were affected by the rule change whereas there are more than one million

private companies in the UK, implying that it is unlikely that the rule change had a

major impact on the degree of competition within the private company audit market as a

whole. Accordingly, we focus on the companies that were affected by the regime switch

rather than the entire audit market. For each affected company, audits were mandatory

in 2003 and voluntary afterward.

Our analysis proceeds in two steps. First, we evaluate whether companies’ audit

choices during the mandatory regime hinge on whether they genuinely want to be

audited. The motivation for this analysis is the uncertainty surrounding whether

companies not wanting to be audited can be compelled to undergo audits that are as

stringent as those supplied to companies that would purchase audits even under a

voluntary regime. Specifically, any companies that do not wish to be audited would

likely minimize their costs under a mandatory regime. Such companies are likely to be

“going through the motions” in terms of passively complying with the audit

requirement, reducing the benefit of making audits mandatory.

Consequently, we begin by examining companies’ audit choices during the final

year of the mandatory audit regime. To identify whether companies genuinely wanted

3 We refer to the voluntary audit regime as starting in 2004 since none of the companies in our

sample had year-ends falling between January 1 and January 29, 2004.

4
to be audited at this time, we observe their decisions on whether to remain audited in

the following year when audits became voluntary. We expect the companies not

wanting to be audited (as revealed in their subsequent decision to abandon an audit)

under the mandatory regime were intent on minimizing their audit costs, rather than

relying on auditing to provide strict external monitoring. Corroborating this intuition,

we find that these companies were significantly less likely to hire Big Four auditors and

they paid significantly lower audit fees during the last year of the mandatory regime. In

contrast, the companies that wanted to be audited were significantly more likely to

appoint high-quality auditors and they incurred significantly higher fees. These results

suggest that the assurance benefits of mandatory audits are lower for companies that do

not want to be audited relative to companies that would choose to be audited under a

voluntary regime. This significantly weakens the main argument in favor of mandatory

audits, which is based on the assumption that there are considerable assurance benefits

when companies that do not want to be audited are forced to undergo an audit.

Importantly, imposing audits prevents companies from signaling through their

decision on whether to appoint an auditor. The second part of our analysis involves

isolating whether the regime switch from mandatory to voluntary auditing yields

signaling benefits by enabling outsiders to observe companies’ choices on whether to be

audited. Melumad and Thoman (1990) provide a theoretical model in which companies

requiring loans voluntarily choose whether to subject themselves to an audit. In their

set-up, companies are composed of unobservable risk types and lenders draw rational

inferences about the company’s type, conditional on its decision on whether to purchase

an audit. These authors demonstrate that auditors can be hired purely for their

signaling value. In their model, companies elect to hire auditors even when audits have

no assurance value because lenders conclude that only the bad types of borrowers

would choose not to be audited. In other words, a company would rather hire an

5
uninformative auditor and be pooled with the good types than avoid an audit, which

divulges to lenders that they are a worse type. Costly auditing permits a separating

equilibrium in which the good types of borrowers appoint auditors, unlike the bad

types. It follows that the company’s decision to appoint (not appoint) an auditor sends a

positive (negative) signal about its credit risk.

Consistent with Melumad and Thoman’s (1990) intuition, critics of mandatory

auditing argue that audits should be voluntary in order to facilitate the signaling of

companies’ types. For example, Sunder (2003) writes:

“All things being equal, investors can logically conclude that the firms
that choose to have their financial reports audited by independent
auditors have nothing to hide from the investors; that the managers of
such firms are relatively more confident about the status, performance
and prospects of their business; and that they deserve the trust of the
investors. On the other hand, the investors may logically conclude that
the firms which choose not to have their financial reports audited by
independent auditors, even though they could have done so, are less
deserving of the investors’ trust and money. When we make
independent audit a statutory requirement, we shut the door on the
ability of the better managers to distinguish themselves from the less
competent managers in the eyes of the investors.”

Despite this argument, we are unaware of any prior empirical evidence on the signaling

role of voluntary audits. We primarily contribute to extant research by assessing

whether voluntary audits have signaling value separate from their assurance value.

Theory shows that companies are able to signal their types even when audits are

mandatory since companies retain the discretion to hire either low-quality or high-

quality auditors (Titman and Trueman, 1986; Datar et al., 1991). Analytical research

suggests that superior companies signal their types by spending more money on

appointing high-quality auditors. The bad types of borrowers are more likely to appoint

low-quality auditors when they are forced to purchase audits. The evidence in the first

part of our analysis indicates that the strength of companies’ preferences for an audit

6
would have been partly revealed during the mandatory regime since the companies that

were audited involuntarily were choosing non-Big Four auditors and were paying

significantly lower audit fees. We therefore expect that outsiders can rationally infer a

company’s likely preference for an audit even when audits are required for every

company.

The second part of our empirical analysis focuses on the company’s credit rating

given that our sample consists of small private companies which rely mainly on external

credit finance rather than outside equity. The credit rating is an important source of

information for creditors deciding whether to lend and pricing the terms of their loans.

Credit ratings are assigned to virtually all private companies in the UK by Qui Credit

Assessment Limited. If the ratings agency is able to infer companies’ types even when

all companies are required to be audited, we expect that the bad types of company that

do not want to be audited would suffer lower ratings. In contrast, the good types that

genuinely want to be audited would enjoy relatively high credit ratings even when they

are unable to signal their types through their decision on whether to be audited.

To gauge whether companies genuinely wanted to be audited during the

mandatory regime, we examine their decisions on whether to be audited when audits

subsequently become voluntary. Consistent with our predictions, we find that

companies wanting to be audited received significantly higher ratings during the

mandatory regime in contrast to companies that are audited involuntarily, which were

penalized with lower ratings. These findings lend support to the argument that the

ratings agency was able to distinguish between companies’ types even when auditing

was mandatory. This in turn implies that the switch to a voluntary regime would not

necessarily provide an additional informative signal about companies’ types.

7
 To isolate whether voluntary audits did in fact generate an incrementally

informative signal, we examine the changes in credit ratings after the transition from

mandatory to voluntary audits. When a company voluntarily chooses to continue being

audited, we expect little or no change in audit assurance since the company is audited in

both regimes. However, such a company conveys a positive signal when it chooses to be

audited voluntarily since it is separated from the bad types that relinquish the audit.

Therefore, to the extent that a voluntary audit communicates an incremental signal of

the company’s superior type, we expect an increase in credit ratings for the companies

that elect to remain audited.4 For a company that chooses to no longer engage an

auditor, its decision signals not only that it is likely to be a bad type of borrower, but

also sacrifices the assurance that was previously provided under the mandatory regime.

For both reasons, we expect that credit ratings decline for companies that dispense with

an audit when this becomes permissible.

In regressions that control for changes in company characteristics and

macroeconomic conditions, we provide strong, robust evidence that credit ratings rise

(fall) for companies that continue (stop) being audited. Specifically, credit ratings

increase by approximately two percentage points when companies choose to remain

audited. We interpret this evidence as implying that these companies enjoy upgrades to

their credit ratings because their decision to remain audited conveys a positive signal

about their credit risk. Apparently, the level of audit assurance was stable for these

companies during the transition from mandatory to voluntary audits since we find that

their audit fees and choices of audit firm do not change following the regime switch. In

4 We do not assume that every unaudited company is a bad type or that every audited company

is a good type. Rather, our predictions are based on the theoretical premise that, everything else
constant, the good (bad) types have stronger (weaker) incentives to be audited (Titman and
Trueman, 1986; Melumad and Thoman, 1990; Datar et al., 1991). Accordingly, the decision on
whether to be audited sends a signal about the average type of company making that choice,
although the signal is not necessarily indicative of every company’s true type.

8
the other direction, we find that credit ratings drop by approximately four percentage

points when companies choose to stop being audited. For such companies, the financial

statements become less credible because they are no longer audited and the decision to

abandon the audit also conveys a negative signal about the company’s type.

Our evidence on credit ratings is summarized as follows. During the mandatory

regime, the companies that wanted to be audited receive credit ratings that were

approximately nine points higher than those that did not. After auditing becomes

voluntary, the companies that remain audited receive a further two point boost to their

ratings while the unaudited companies suffer a four-point penalty. Consequently, the

spread in credit ratings widens following the switch away from mandatory audits

because the ratings agency is able to better distinguish between the good and bad types

of borrowers. During the voluntary regime, the audited companies receive credit

ratings that are fifteen points higher than the unaudited companies. This is comprised

of the initial spread during the mandatory regime (i.e., nine points) plus the increase in

the spread when auditing becomes voluntary (i.e., six points). To provide some

perspective on the materiality of these results, the mean (median) credit rating over our

sample period is 67 (69) points on a 100-point scale. Collectively, our evidence suggests

that the switch to voluntary auditing had a first-order economic impact on credit

ratings.5

In their comprehensive literature reviews, Francis (2004) and Watkins et al.

(2004) outline the extensive prior theory and evidence that audit quality varies along

several dimensions. We complement this research by focusing on the more basic

5 In a sensitivity test, we dispel concern that auditor choice is spuriously responsible for our

evidence on the importance of legal regime (mandatory versus voluntary) to credit ratings by
verifying that our core results hold when we run the regressions on non-Big Four (Big Four)
clients separately.

9
question regarding the impact of allowing auditing to be optional rather than required. 6

To our knowledge, this is the first study to provide direct evidence on the signaling

benefits of voluntary auditing. Blackwell et al. (1998) provide evidence that lenders

charge lower interest rates to private companies whose financial statements are audited.7

Although their study indicates that there are benefits from being audited voluntarily, it

does not disentangle whether such benefits stem from greater assurance or from

signaling. Specifically, an audited company may enjoy a lower interest rate because its

financial statements are more credible or because its decision to purchase an audit sends

a positive signal about its credit risk. Thus, their finding that audited companies incur a

lower cost of borrowing does not necessarily imply that voluntary audits yield signaling

benefits. Likewise, there is evidence that lenders insist that companies supply audited

financial statements to obtain loans (Leftwich, 1983; Allee and Yohn, 2009), although this

could reflect either the assurance or signaling benefits of auditing. It is important to

separately isolate the signaling effect because theory holds that⎯even if there is zero

assurance value to an audit⎯companies would have incentives to appoint auditors

voluntarily due to the positive signal that rational lenders would infer (Melumad and

Thoman, 1990). In a major contribution to extant research, we provide the first direct

evidence supporting this theoretical prediction about the signaling value of voluntary

audits.

6 There is already fairly extensive evidence on the impact of voluntary audits, leading Francis et
al. (2008: 1) to highlight that: “the most robust finding across these studies is that private
[voluntary] audits are related to securing external debt financing.”
7 The demand for voluntary audits predates regulations requiring audits. Companies have been
voluntarily appointing independent auditors since at least the 13th century (Watts and
Zimmerman, 1983; Sunder, 2003), which implies that the private benefits of auditing often exceed
their costs. Bentson (1969) reports that 82 percent of US public companies purchased audits
shortly before the Securities Acts of 1933 and 1934, which introduced mandatory auditing of
financial statements. Chow (1982) examines the voluntary audit decision made by large US
public companies in 1926. Given the evolution in the institutions governing financial reporting,
Barton and Waymire (2004) caution against generalizing results from earlier times to the present.
Similarly, Coffee (1984) calls for research on contemporary regulation since so much has changed
since the 1933 and 1934 Acts were passed.

10
 Our evidence on signaling is also central to the debate about the relative merits

of voluntary and mandatory audits since both types of audits can yield assurance

benefits but only voluntary auditing facilitates signaling. Our findings provide

empirical support for the argument that the mandatory requirement suppresses the

signal that is conveyed when companies are allowed to choose whether to be audited

(Sunder, 2003). Moreover, our study indicates that companies not wanting to be audited

were only passively complying with the audit requirement under the mandatory

regime. These companies chose lower quality auditors and they were intent on reducing

the costs of the audit when they were audited involuntarily. It would appear that there

are limited assurance benefits arising from involuntary audits since it is difficult to force

such companies to undergo stringent auditing. This in turn weakens the main argument

in favor of mandatory auditing.

Finally, it is important to acknowledge that our study cannot provide a definitive

answer to policy-makers on whether audits should be voluntary or mandatory. One

reason is that companies focus on their own private costs and benefits when considering

whether to purchase an audit and we are not able to measure the external spillovers that

accrue to external users from having the accounts audited for every company. Another

reason is that our empirical analysis is restricted to small private companies and the

results may not generalize to companies that are publicly traded.

The remainder of this paper is organized as follows. Section 2 outlines prior

theory and evidence in developing our testable predictions. Section 3 covers our

research design, while Section 4 describes our empirical results. Section 5 concludes.

11
2. Hypotheses development

The purpose of this paper is to shed light on the role of different legal requirements

governing auditing. The rationale behind regulators imposing mandatory audits is that

companies have insufficient private incentives to voluntarily provide reliable financial

information.8 Reinforcing that mandatory auditing may be socially optimal, theory

implies that information underproduction can arise because of positive externalities

(e.g., Dye, 1990; Admati and Pfleiderer, 2000) and financial statements constitute a

public good that will be under-supplied in a free market (e.g., Gonedes and Dopuch,

1974; Beaver, 1998). On the other hand, it remains unclear whether private incentives for

the demand and supply of audits are insufficient, particularly as private markets for

other forms of certification services are ubiquitous in the economy (Jamal and Sunder,

2008). Moreover, voluntarily submitting to an audit enables companies to credibly

signal their types whereas mandatory audits deprive investors of this important

indicator (Sunder, 2003).

The arguments in favor of regulation rather than the free market were influential

in the UK, where private companies are required to make their financial statements

publicly available and, until recently, those statements also had to be independently

audited (Aranya, 1974; Dedman and Lennox, 2009). The rationale for requiring all

companies to be audited hinges on the necessary condition that mandatory audits

provide assurance benefits. However, some companies may be going through the

motions when complying with this regulation, which would translate into mandatory

audits delivering relatively low assurance.

8 Importantly, Dye (1990) and Leftwich (2004), among others, stress that advocates for mandatory

financial reporting do not argue that firms will supply no information unless compelled or that
managers’ incentives are irrelevant to reporting decisions. Indeed, the opportunity to reduce
agency (Jensen and Meckling, 1976), information (Botosan, 1997), and liquidity (Welker, 1995)
costs can convince managers to divulge high-quality information voluntarily.

12
 We begin by testing whether companies that did not want to be

audited⎯evident in their subsequent decision to avoid an audit when this becomes

permissible⎯were only passively complying with the mandatory audit provision.

Imposing audits precludes companies from signaling their types through the decision on

whether to appoint an auditor, although they can still partly reveal their type with their

decision on whether to choose a high-quality or low-quality auditor (Titman and

Trueman, 1986; Datar et al., 1991). The bad types of borrowers that are reluctant to be

audited will resist choosing high-quality auditors when they are forced to be audited.

In contrast, the good types that genuinely want to be audited are more likely to choose

high-quality audit firms when auditing is mandatory. The extensive empirical evidence

that the Big Four firms supply higher quality audits than do the smaller auditors (e.g.,

Francis, 2004) extends to the U.K. (e.g., Lennox, 1999; Peel and Roberts, 2003;

McMeeking et al., 2006; Clatworthy and Peel, 2007). Reflecting that the bad types of

borrower are less willing to bear the costs of a high quality audit, their lower demand

for audits translates into our first prediction (stated in the alternative form):

H1: During the mandatory audit regime, companies that do not want to be
audited are less likely to choose a Big Four auditor compared with
companies that do want to be audited.

Similar arguments apply to a company’s incentives on whether to pay for a high-

quality audit. A company that does not want to be audited will likely put pressure on

the auditor to cut costs in order to reduce the audit fee. In contrast, a company that

genuinely wants to be audited would be more willing to pay the high fee that ensues

when an auditor supplies more effort on the assurance engagement. We therefore

expect the companies that did not want to be audited were paying relatively low audit

fees during the mandatory regime, which provides our second hypothesis:

H2: During the mandatory audit regime, companies that do not want to be
audited pay lower audit fees compared with companies that do want to
be audited.

13
 Our next set of hypotheses concern the credit ratings assigned to private

companies during the mandatory and voluntary regimes. Private companies provide an

opportune setting for analyzing the assurance and signaling benefits of auditing given

that their information structure is typically poor relative to public companies. For

example, Brav (2009) holds that debt contracting is more sensitive to information for

private companies than for their public counterparts. Reinforcing that this testing

ground suits our inquiry, Fenn (2000) and Santos (2003) report that lenders demand

higher yields on private companies’ debt to compensate for the worse information

asymmetry that they suffer.9 This evidence squares with Graham et al.’s (2005) finding

from a survey of chief financial officers that, compared with public companies, private

companies are more inclined to manipulate earnings to preserve their credit ratings and

to avoid violating bond covenants, rendering their financial statements less informative

for the debt contracting process. It follows that the links between auditing and credit

ratings will be stronger in private companies that are lesser known, increasing the

power of our tests.

Analyzing credit ratings maps into our research questions that focus on the

benefits of auditing under the mandatory and voluntary regimes. Prior research shows

that information risk affects credit ratings and that reliable accounting numbers facilitate

debt contracting (e.g., Watts, 1977; Smith and Warner, 1979; Leftwich, 1983; Francis et al.,

2005; Yu, 2005). This is particularly relevant to UK private companies that rely heavily

on loan financing (Brav, 2009). In fact, prior research finds that companies relax

accounting-based covenants by managing their earnings through accounting changes

(Sweeney, 1994) and discretionary accruals (DeFond and Jiambalvo, 1994). Moreover,

9 Similarly, Pagano et al. (1998) report that firms’ borrowing costs fall after going public, which

they attribute to the greater accounting transparency that ensues. Indeed, Givoly et al. (2009)
report that public firms have more conservative earnings than private firms. Prior research
implies that debt contracting is partly behind the demand for conservative financial reporting;
e.g., Leftwich (1983) and Holthausen and Watts (2001).

14
recent research uses credit ratings to measure the perceived benefits of Big Four versus

non-Big Four audits (e.g., Mansi et al., 2004), including in private companies (e.g., Fortin

and Pittman, 2007).10

Outsiders can rationally infer the company’s type by observing its audit choices

according to theory. Naturally, the low risk borrowers are eager to signal their types

since they will be rewarded in the form of higher ratings for becoming better known. In

analytical models, a separating equilibrium prevails in which companies with

unfavorable private information choose low audit assurance while the good types

choose high assurance. The bad types are deterred from mimicking the good types

because it is too costly to appoint a high-quality auditor (e.g., Datar et al., 1991) or yields

inadequate benefits (e.g., Titman and Trueman, 1986). Accordingly, we expect that the

ratings agency is able to rationally infer whether a company genuinely wants to be

audited even when every company is required to be audited. In short, we predict that

credit ratings are lower for companies that do not want to be audited under the

mandatory regime since the credit rating agency recognizes that these companies are

more likely to be the bad types of borrowers:

H3: During the mandatory audit regime, companies that do not want to be
audited receive lower credit ratings compared with companies that do
want to be audited.

10 Auditors provide implicit insurance coverage to investors in the event of audit failure (Dye,
1993), although empirical research generally struggles with cleanly distinguishing between the
information and insurance roles that auditing plays in securities pricing. However, auditor
insurance protection is almost certainly trivial in our setting. St. Pierre and Anderson (1984) and
Palmrose (1987) find that civil lawsuits against auditors infrequently involve US private firms
despite that this country is a global outlier in terms of its high rate of auditor litigation (Francis,
2004). Similarly, Palmrose (1986) reports that the litigation risk premium embedded in audit fees
is higher for public companies than private companies. In the UK, legal standards for suing
auditors for issuing an unqualified opinion on materially deficient financial statements are
stricter (Seetharaman et al., 2002; La Porta et al., 2006); i.e., relative to the litigious environment in
US, auditor discipline is more lenient in the UK where it is more difficult for investors to sue the
auditor to recover losses. In fact, auditors only owe a duty of care to shareholders, not creditors,
under UK case law. Reinforcing the minor role of litigation there, the UK prohibits class-action
lawsuits.

15
 Finally, we analyze whether credit ratings impound that requiring audits ruins

the signaling value inherent in companies electing to subject their financial statements to

an external audit. Our identification strategy exploits the new UK legislation to provide

insight on whether companies enjoy higher credit ratings when they more fully reveal

their types by choosing to appoint an auditor under the voluntary regime. In the

absence of an audit requirement, a company can incur the cost of engaging an auditor to

ensure that lenders infer that it is more likely to be a good type. More formally,

Melamud and Thoman’s (1990) theory implies that the information conveyed with the

decision to hire an auditor vanishes when auditing becomes mandatory. A mandatory

audit regime brings uncertainty by preventing lenders from learning about a company’s

type by observing its strategic decision on whether to have an audit (Sunder, 2003); i.e.,

requiring audits suppresses this form of information.

In estimating the signaling value of voluntary audits, the benefits of audit

assurance are held constant for companies that are audited in both the mandatory and

voluntary regimes. On the other hand, there is a role for signaling under the voluntary

regime as a company that continues to be audited is able to signal that it is more likely to

be a good type. In this case, remaining audited under the voluntary regime transmits

positive news evident in higher ratings.

H4: Credit ratings increase for companies that switch from mandatory audits
to voluntary audits.

In contrast, companies that choose to stop being audited under the voluntary

regime forego any assurance benefits that existed when they were involuntarily audited

and their decision also conveys a negative signal because outsiders are able to more

clearly distinguish between them and the companies that want to be audited. These

dynamics are behind our final hypothesis:

16
 H5: Credit ratings fall for companies that switch from mandatory audits to no
audits.

Collectively, we expect that credit ratings are higher for companies that want to

be audited than for those that do not, and the spread in credit ratings between the two

types of company expands when audits become voluntary since this enables companies

to more fully reveal their types. Although requiring audits dilutes their signaling value,

we expect that companies still partly divulge their type through audit fees and auditor

choice when audits are mandatory.

3. Research Design

3.1 Final Year of the Mandatory Regime

To test the first three hypotheses, H1 to H3, our analysis initially focuses on the final year

of the mandatory regime. Companies that do not want to be audited may be simply

going through the motions when audits are required, which would be evident in these

companies choosing lower quality audit firms (H1) and paying lower audit fees (H2)

during the mandatory regime and then avoiding an audit during the voluntary regime.

In contrast, the companies that genuinely want to be audited are more likely to be

perceived as low-risk borrowers. If the ratings agency is able to discriminate between

the two types of companies even when they are both subject to an audit, we would

expect that the companies that want to be audited enjoy higher credit ratings (H3).

We test these three predictions by estimating the following cross-sectional

models for the final year of the mandatory audit regime (2003):

BIG4it = β0 + β1 VOL_AUDITit + β2 Xit + β3 INDUSTRYi + vit (1)

LAFit = α0 + α1 VOL_AUDITit + α2 Xit + α3 INDUSTRYi + uit (2)

RATING it = δ0 + δ1 VOL_AUDIT it + δ2 Xit + δ3 INDUSTRYi + wit (3)

17
Where:

BIG4it = one if company i chooses a Big Four audit firm in 2003, and zero if company i

chooses a non-Big Four audit firm in 2003.

LAFit = the log of company i’s audit fees in 2003.

RATINGit = company i’s credit rating in 2003 (the rating is scored by Qui Credit

Assessement Ltd. on a scale ranging from 0 to 100).

VOL_AUDITit = one if company i is voluntarily audited in 2004, and zero if company i is

unaudited in 2004.

Xit = a vector of time-varying control variables for company i (t = 2003).

INDUSTRYi = a vector of industry dummy variables for company i.

The coefficients on the VOL_AUDITit variable will be significantly positive in eqs. (1) to

(3) under H1 to H3.

3.2 Transition from the Mandatory to the Voluntary Regime

Next, we analyze the changes in credit ratings after the regulatory regime switches from

mandatory to voluntary audits. We expect that credit ratings will rise for companies

that remain audited when this becomes voluntary in 2004 since electing to have an audit

would serve as an incremental signal of the company’s good type (i.e., lower credit risk).

In comparison, abandoning an audit is both a signal that the company is a likely to be a

bad type of borrower and it removes the assurance value associated with having audited

financial statements. Both effects are expected to contribute towards lower credit

ratings. Accordingly, we examine the predictions in H4 and H5 by estimating a model

that explains how credit ratings change from the final year of the mandatory regime

18
(2003) to the initial year of the voluntary regime (2004). Re-writing the credit ratings

model in eq. (3) by taking changes gives the following:

ΔRATINGit = µ0 + µ1 VOL_AUDIT it + µ2 ΔXit + wit (4)

where

ΔRATINGit = the change in company i’s credit rating from the last year of the mandatory

audit regime (2003) to the first year of the voluntary audit regime (2004).

VOL_AUDITit = one if company i is voluntarily audited in 2004, and zero if company i is

not audited in 2004.11

We expect that credit ratings increase for the companies that continue to be audited (H4)

and fall for the companies that switch from mandatory audits to no audits (H5). Jointly,

these two hypotheses generate the prediction that µ1 > 0.

To provide a separate test of H4, we estimate eq. (4) for the sub-sample of

companies that continue to be audited (i.e., VOL_AUDITit = 1):

ΔRATINGit = µ0 + µ1 + µ2 ΔXit + wit (4a)

Under H4, credit ratings increase for a company that remains audited because this

decision conveys a positive signal about its type (i.e., µ0 + µ1 > 0).

Similarly, we test H5 by estimating eq. (4) for the sub-sample of companies that

no longer undergo an audit (i.e., VOL_AUDITit = 0).

ΔRATINGit = µ0 + µ2 ΔXit + wit (4b)

Under H5, credit ratings decrease for the companies that dispense with an audit (i.e., µ0 <

0) because this decision both results in a loss of assurance and conveys a negative signal

about the company’s credit risk.

11 The VOL_AUDIT variable reflects both the decision to have an audit in 2004 (eq. 3) and the
it
change in audit status from 2003 to 2004 (eq. 4) since every company is audited during the final
year of the mandatory regime. It is therefore equivalent to a change variable in eq. (4).

19
3.3 Control Variables

Our set of control variables follows that used in recent research on credit risk, auditor

choice, and audit fees (e.g., Bharath et al., 2008; Fortin and Pittman, 2007):

AGE it = company i’s age in year t.

LTSit = log of total sales.

LTAit = log of total assets.

INTCOVit = interest expense divided by earnings before interest and taxation. The

INTCOVit ratio is capped at an upper bound of 2.00 to handle outliers and we assign a

value of 2.00 to the INTCOVit ratio if earnings are non-positive.

LIQUIDITYit = (current assets – inventory) divided by current liabilities.

LEVERAGEit = total liabilities divided by total assets.

We expect that insolvency risk is higher for companies that are younger (AGEit), smaller

(LTSit and LTAit), have lower interest coverage (INTCOVit), lower liquidity

(LIQUIDITYit), and more liabilities (LEVERAGEit). Besides explaining credit ratings,

prior research implies that these independent variables ⎯ particularly company size ⎯

are important determinants of the company’s choice of audit firm and audit fees.12

12 Some audit fee studies include a control for profitability or a dummy variable for losses.
However, in our study, profitability is very highly correlated with the interest coverage variable
which measures the ratio of the interest expense to earnings. Given this high correlation we do
not include the profitability and interest coverage variables in the same regressions. However, in
robustness tests, we find very similar results if we replace the interest coverage variable with
alternative measures of profitability.

20
3.4 Sample Formation and Description

The UK has long diverged from the US by stipulating that private companies have their

financial statements audited.13 However, in an effort to reduce the burden of regulation

on small private companies, the UK granted an audit exemption in 1994. It became

permissible after 1994 for private companies to avoid having an audit if their sales did

not exceed £1m and their total assets did not exceed £1.4m. A subsequent amendment

to the Companies Act relaxed these size thresholds, allowing more private companies to

qualify for the audit exemption. Specifically, companies with fiscal years ending after

January 30, 2004 were allowed to avoid an audit if their sales did not exceed £5.6m and

total assets did not exceed £2.8m. This shift in exemption eligibility thresholds enables

us to assemble a sample of companies that were affected by the regime switch in 2004.

We compile the sample from the Financial Analysis Made Easy (FAME)

database. All public companies must be audited regardless of their size, so we require

that each sample company is private. For companies belonging to a group, auditor

hiring decisions are routinely made by the ultimate owner rather than at the company

level, so we impose the restriction that the company is independent; i.e., it does not

belong to a corporate group. To ensure that each private company was required to

undergo an audit prior to January 30, 2004, we sample companies for which sales ≥ £1m

and total assets ≥ £1.4m in 2003. Next, we confine the sample to companies that

qualified for the audit exemption after January 30, 2004 (i.e., sales ≤ £5.6m and total

assets ≤ £2.8m). Certain types of regulated company⎯insurance companies, investment

13 More generally, this setting is unique in other ways, including that both private and public
companies must publicly file annual financial statements that follow the same accounting
standards in order to comply with UK disclosure regulations. Similarly, both types of companies
are subject to the same tax laws in the UK.

21
advisors, mortgage arrangers, trade unions, and employers associations⎯are required to

have an audit even if they fall within these size thresholds, so we exclude these

companies from the sample. Finally, we require that financial statement data are

available for both the year prior to January 30, 2004 and the next year. After applying

these data screens, we are left with 5,139 unique companies. By design, each company

in the sample was required to have an audit in 2003, but not in 2004. There are two

observations per company with the first pertaining to the final year of the mandatory

audit regime (2003) and the second to the initial year of the voluntary regime (2004).

3,440 (67%) of the 5,139 companies in the sample remain audited in the first year of the

voluntary regime while 1,699 companies (33%) choose to become unaudited once this

option becomes available.

3.5 Descriptive Statistics

The industry composition of the sample companies is shown in Table 1. There are 1,414

companies (26.3%) that operate in the business services sector and 1,221 (23%)

companies in the wholesale and retail trade sector. Other industries that are well

represented include construction (748 companies) and manufacturing (719 companies).

Our research design controls for differences between industries by specifying dummy

variables for the eight industry sectors in Table 1 that have more than twelve companies

(the remaining three industry sectors are captured in the regression intercept).

[INSERT TABLE 1 NEAR HERE]

The credit rating scores issued by Qui Credit Ltd. are provided in the FAME

database. Ratings are assigned on a numerical scale between 0 and 100 that quantifies

the agency’s assessment of the likelihood of corporate failure within the next 12 months.

22
The RATINGit variable corresponds to the company’s credit score with higher ratings

representing a lower perceived risk of financial failure. Doumpos and Pasiouris (2005)

provide evidence that Qui’s credit ratings are accurate indicators of default risk. In our

own analysis, we find that the credit scores are strongly associated with staple

accounting variables used in predicting insolvency risk (e.g., interest coverage).

Table 2 presents descriptive statistics for 2003 (the final year of the mandatory

regime) and 2004 (the first year of the voluntary regime). The motivation for this

analysis is to investigate whether there are important changes in the macroeconomic

environment or other time-varying factors that could confound the comparison of the

two regimes. The first row of Table 2 shows that the mean credit rating increases from

64.51 in 2003 to 68.37 in 2004. This ratings improvement, which is highly significant (t-

statistic = 9.32), occurs despite the potential loss of assurance that follows when

companies start to abandon audits in 2004. The ratings increase is consistent with a

general improvement in the economic environment between 2003 and 2004. Further, we

find a significant improvement in liquidity between 2003 and 2004 (t-statistic = 2.73),

although the changes in interest coverage, company size, and leverage are not

significant.

[INSERT TABLE 2 NEAR HERE]

Given the evidence in Table 2, we control for the general improvement in credit

ratings that affects our entire sample between 2003 and 2004 by subtracting the sample

mean values of credit ratings in each year. For example, we measure the deviation

between company i‘s credit rating in 2003 and the mean rating given to every other

company in 2003 (RATINGi2003 - RATING.2003). We similarly measure the cross-sectional

variation in credit ratings during 2004 using the variable RATINGi2004 - RATING.2004.

23
Finally, we calculate the change in credit ratings from 2003 to 2004 using the ratings that

are purged of the year effects. This change in credit ratings variable is (RATINGi2004 -

RATING.2004) – (RATINGi2003 – RATING.2003), which we label Δ(RATINGit – RATING.t).

Similarly, we purge all the control variables of any yearly effects following the same

approach; i.e., Δ(Xit – X .t). Thus, we modify the models of credit ratings changes in eqs.

(4), (4a) and (4b), which now become:

Δ(RATINGit – RATING.t) = µ0 + µ1 VOL_AUDITit + µ2 Δ(Xit – X .t) + wit (4’)

Δ(RATINGit – RATING.t) = µ0 + µ1 + µ2 Δ(Xit – X .t) + wit (4a’)

Δ(RATINGit – RATING.t) = µ0 + µ2 Δ(Xit – X .t) + wit (4b’)

4. Results

4.1 Univariate Evidence

In an initial pass at our research questions, we report in Panel A of Table 3 the mean

values of each variable during the mandatory regime. The BIG4it frequency is 8% in the

sub-sample of 3,440 companies that wanted to be audited, which is evident in their

decision to be voluntarily audited in the subsequent year (VOL_AUDIT it = 1). The BIG4it

frequency is significantly lower at 2% for the 1,699 companies that did not want to be

audited (VOL_AUDIT it = 0), consistent with the prediction in H1 that the companies

genuinely eager to be audited were significantly more likely to appoint Big Four audit

firms (t-statistic = -8.16).

[INSERT TABLE 3 NEAR HERE]

Similarly, the mean audit fee is £5,680 for the companies that wanted to be

audited compared with £4,270 for the companies that did not. These audit fees are

naturally lower than in prior studies since our sample consists of small private

24
companies (the average company has assets hardly exceeding £1 million according to

Table 2). More importantly for our purposes, we find that audit fees are significantly

lower for the companies that did not want to be audited (t-statistic = -11.87), supporting

the prediction in H2. The highly significant difference in the log of audit fees (LAFit)

between the two groups of companies (t-statistic = -13.32) reinforces this result.

Consistent with H3, it appears that the credit ratings agency perceived that the

companies wanting to be audited were significantly lower risks. The mean rating

assigned to these companies was 67.48 compared with 58.50 for the companies that did

not want to be audited; this nine-point difference is highly significant (t-statistic =

-13.64).

While these univariate results are consistent with all three hypotheses, it is

important to caution that they do not control for the observable differences between the

two types of company. For example, according to Panel A of Table 2, the companies that

wanted to be audited are significantly older (t-statistic = -5.11) and larger (t-statistics =

-4.21, -12.80). In addition, these companies had worse interest coverage (t-statistic =

-3.50), although they also had greater liquidity (t-statistic = -1.87) and lower leverage (t-

statistic = 3.63).

Panel B of Table 3 reports the mean values during the first year of the voluntary

audit regime. Interestingly, we find that the spread in credit ratings between the

companies that choose to be audited and those that do not is larger in 2004. The mean

rating to the voluntarily-audited companies is 73.38 compared with 58.25 for the

companies that chose not to have their financial statements audited. The difference in

the ratings assigned to these two types of company is more than fifteen points during

the voluntary regime (73.38 minus 58.25) compared with only nine points in the

25
previous year (67.48 minus 58.50). This wider spread is our first piece of evidence that

the regime switch affected the ratings issued to the two groups of companies.

It is important to consider the alternative explanation that the changes in credit

ratings were different for the two sets of companies because they experienced significant

differences in their performance between 2003 and 2004. The results in Panels C and D

of Table 3, however, clearly dispel this concern. In these panels, we compare the mean

values of the ratings and control variables between 2003 and 2004, after purging the

variables of their yearly mean values (i.e., these panels report the mean values of

RATINGit – RATING.t and Xit – X .t). Panel C provides no evidence of significant changes

in performance or operating characteristics between 2003 and 2004 for the 1,699

companies that did not want to be audited. In fact, the mean values are statistically

insignificant between these two years for every control variable. In contrast, Panel C

shows that credit ratings fell dramatically between 2003 and 2004 for the companies that

avoid an audit. Their mean credit rating was 6.01 points below the mean in 2003

whereas it was 10.13 points below the mean in 2004. Therefore, consistent with H5, these

companies suffered a relative downgrade of just over four points in their credit ratings.

In Panel D, we report descriptive statistics for 2003 and 2004, focusing on the

3,440 companies that remain audited following the regime switch. The first two rows

reveal that the audit fees paid by these companies did not change significantly and their

choices of audit firm also did not change. Therefore, there was no apparent change in

audit assurance between 2003 and 2004 for the companies that were audited in both

years. Nevertheless, we expect that their credit ratings would improve over time since

voluntarily submitting to an audit signals their superior types. Supporting this

argument, Panel D shows that credit ratings are 5.00 points above the mean in 2004, but

26
only 2.97 points above the mean in 2003 for these companies. In short, consistent with

H4, the companies that voluntarily continued with an audit enjoyed a two-point ratings

upgrade. Again, we find no evidence that changes in company characteristics explain

this change in credit ratings. Panel D shows that these companies exhibit no significant

changes in interest coverage, size, liquidity, or leverage between 2003 and 2004.

Collectively, the results in Panels C and D strongly suggest that the changes in credit

ratings between 2003 and 2004 are attributable to the regime switch, rather than changes

in company characteristics.

4.2 Auditor choice and audit fees during the mandatory audit regime

Companies that do not want to be audited may be simply going through the motions

when they are forced to be audited during the mandatory regime. This would be

evident in these companies attempting to minimize audit costs during the mandatory

regime and then avoiding an audit during the voluntary regime. In contrast, companies

genuinely eager to improve financial reporting credibility would accept incurring the

higher fees that accompany a higher-quality audit. We provide evidence on the

predictions in H1 and H2 by estimating models on the determinants of auditor choice

(eq. 1) and audit fees (eq. 2) during the final year of the mandatory audit regime.14

The results for the auditor choice model are shown in Col. (1) of Table 4.

Consistent with H1, the VOL_AUDITit coefficient is positive and highly significant (t-stat.

= 5.94). That is, the companies that wanted to be audited were more likely to appoint

Big Four audit firms during the mandatory regime compared with the companies that

14 Although prior research finds that Big Four auditors charge higher fees, the pair-wise
correlation between auditor choice and audit fees is only 0.14 in our sample, suggesting that these
variables reflect different underlying constructs.

27
were passively complying with the audit requirement. The results for the audit fee

model are shown in Col. (2) of Table 4. The VOL_AUDITit coefficient is again positive

and highly significant (t-stat. = 7.20). This supports the prediction in H2 that companies

not wanting to be audited resort to paying lower audit fees. In contrast, the companies

that wanted to be audited had a greater demand for assurance services evident in their

higher audit fees. Overall, our findings are consistent with the theoretical prediction

that companies’ types can be partially signaled through their audit choices even when

auditing is mandatory (Titman and Trueman, 1986; Datar et al., 1991). However, it

remains to be resolved whether the switch to voluntary auditing helped to more fully

distinguish companies’ types.

[INSERT TABLE 4 NEAR HERE]

The results for the control variables are consistent with prior studies on the

determinants of auditor choice and audit fees. The auditor choice model shows that Big

Four clients tend to be larger, while they also have higher liquidity and greater interest

coverage. Larger and older companies also pay significantly higher audit fees. The

INTCOVit variable has a significantly positive coefficient in the audit fee model,

implying that fees are higher when a company’s earnings are low relative to its interest

expense. This could reflect that audit fees are higher when companies are in financial

distress to compensate for increased audit risk.15

15 The R2 in the audit fee model is only 21%, which is lower than in prior studies because the
companies in our sample are nearly homogeneous in terms of their size. Most of the R2 in prior
audit fee studies comes from the cross-sectional variation in company size which is low in our
study because every company must meet the size thresholds for mandatory audits in 2003 and
voluntary audits in 2004. Moreover, it is important to note that the R2 statistics are not
comparable across studies that examine very different samples (Gu, 2007).

28
4.3 Credit ratings during the mandatory and voluntary audit regimes

We report the estimation results for the model of credit ratings during the mandatory

regime (eq. 3) in Col. (1) of Table 5. In this regression, the VOL_AUDITit coefficient is

positive and highly significant (t-stat. = 13.41), consistent with the prediction in H3 that

the credit rating agency perceives that the companies that want to be audited have lower

credit risk. The coefficient estimate on the VOL_AUDITit variable is 7.68 indicating that

credit ratings are nearly eight points higher for the companies that wanted to be audited.

The magnitude of this multivariate coefficient is very similar to the univariate result

reported in Panel A of Table 3, where the spread in ratings between the two groups of

companies amounted to nine points. The similarity in results between the univariate

and multivariate tests is comforting as it suggests that the results are unlikely to be

affected by extraneous independent variables.

Overall, we conclude that there are systematic differences in the credit ratings of

voluntary and no audit companies even when auditing is mandatory. In particular, the

credit rating agency considers the companies that wanted to be audited to be the

superior types of borrowers in that they receive higher credit ratings. This is despite the

fact that such companies are unable to fully reveal their types at this time through the

signaling mechanism of choosing to be voluntarily audited.

[INSERT TABLE 5 NEAR HERE]

Next, we report the results for the model of credit ratings during the voluntary

audit regime (Col. (2) of Table 5). In this regression, the coefficient on the VOL_AUDITit

variable is 12.94 and it is again significantly greater than zero (t-statistic = 29.55). This

finding implies that voluntarily-audited companies received credit ratings that are

nearly thirteen points higher compared with the no audit case. Again, this magnitude

29
reinforces the univariate result (Panel B, Table 3), where it was found that credit ratings

are fifteen points higher for the companies that are voluntarily audited. Further, Table 5

confirms that there is an increase in the credit ratings spread between the two types of

company after the change in the regulation. This is consistent with the joint prediction

in H4 and H5, that credit ratings increase (decrease) for companies that want (do not

want) to be audited between 2003 and 2004. We report separate tests of these two

hypotheses in the next section.

The results for the control variables are generally consistent with prior research

that examines the determinants of private company credit ratings; e.g., Fortin and

Pittman (2007). In particular, companies have better credit ratings when they are older

(AGEit) and larger (LTSit, LTAit). In addition, companies with lower interest coverage

(INTCOVit), higher leverage (LEVERAGEit), and lower liquidity (LIQUIDITYit) attract

worse ratings.

4.4 Changes in credit ratings after auditing becomes voluntary

The first model in Table 6 presents estimation results for eq. (4’) where the dependent

variable (Δ(RATINGit – RATING.t)) captures the change in credit ratings as companies

adjust to the voluntary regime (2004) from the mandatory regime (2003). As explained

in Section 3.5, we subtract the yearly sample means for each variable in order to control

for time-varying factors such as the macroeconomic environment. Table 6 does not

include a variable for the change in company age since every company in the sample

grows older by one year from 2003 to 2004; i.e., the change in age is the same for every

company.

30
 In Model 1, the VOL_AUDITit variable gauges the change in credit rating

between 2003 and 2004 for companies that remain audited compared with companies

that are no longer audited. The coefficient on this variable is positive and highly

significant (t-statistic = 10.61), lending support to the joint prediction in H4 and H5 that

there is a relative decline in credit ratings for companies that switch from mandatory

audits to no audits, compared with companies that continue to be audited. Reflecting

its economic materiality, the VOL_AUDITit coefficient is 5.84 in Col. (1), which implies

that credit ratings increase (decrease) by nearly six points when companies continue

being audited rather than avoid the audit. This magnitude is very similar to the increase

in the credit ratings spread estimated in Table 5, where the difference in the

VOL_AUDITit coefficients is 5.26 (= 12.94 – 7.68).

[INSERT TABLE 6 NEAR HERE]

Model 2 in Table 6 presents estimation results for the 3,440 companies that

switch from mandatory audits to voluntary audits (eq. 4a’). The intercept in this model

is estimated to be 1.93 and it is significantly greater than zero (t-statistic = 5.99). This

implies that, between 2003 and 2004, credit ratings increase by nearly two points for the

companies that continue to be audited following the regime switch. (This finding can

also be calculated directly from model 1 since 1.93 = -3.91 + 5.84.) Consistent with H4,

we conclude that companies transmit a positive signal about their credit risk by

voluntarily continuing with an audit. This finding is important because it supports the

argument that voluntary auditing permits a signaling role that is absent when audits are

required by law. To our knowledge, this study is the first to isolate the signaling effect

stemming from allowing audits to be voluntary.

31
 In Model 3, we report the changes in credit ratings for the 1,699 companies that

choose not to be audited when this becomes permissible in 2004 (eq. 4b’). Consistent

with H5, the companies that switch from mandatory audits to no audits experience an

average downgrade of 3.91 points to their credit ratings. This drop in credit ratings is

significantly different from zero (t-statistic = -8.81), corroborating the intuition that there

is both a negative signal and a loss of assurance for the companies that cease to be

audited. It is beyond the scope of this study to estimate the fraction of the 3.91 fall in

credit ratings that reflects the negative signal versus the loss of audit assurance since

these effects coincide.

4.5 Additional Analysis on Big Four and Non-Big Four clients

In 2003, 4,835 sample companies were audited by non-Big Four firms compared with

only 304 companies that were audited by the Big Four. These descriptive statistics

reflect that all of the companies in our sample are very small, translating into relatively

few companies needing to approach the Big Four firms to undertake their audits. In

2004, the number of Big Four audits falls by 18.4% from 304 to 248, while the number of

non-Big Four audits drops 34.0% from 4,835 to 3,192. Accordingly, consistent with the

evidence in support of H1, the regime switch had a bigger impact on the number of

audits conducted by the non-Big Four firms than those by the Big Four.

Given that Big Four audits are relatively scarce in our sample, we repeat the tests

of H2 to H5 by re-estimating the audit fee and credit ratings models after excluding the

companies that were audited by Big Four firms in 2003. Consistent with H2, the audit

fee model continues to show that the clients that did not want to be audited during the

mandatory regime paid significantly lower audit fees. The models in Table 5 reveal that

32
these companies also received significantly lower credit ratings compared with those

that wanted to be audited, which supports H3. Further, the credit ratings spread

between the two types of company increased from nearly eight points to thirteen points

between 2003 and 2004. Finally, the ratings change models (Table 6) reveal that the non-

Big Four clients enjoyed a statistically significant two point increase in their credit

ratings if they continued being audited in 2004 (H4), while the non-Big Four clients that

stopped being audited suffered a significant four point drop (H5). Predictably, the

statistical significance and the economic magnitude of these results are very similar to

our main tabulated specifications given that the vast majority of audits in our sample are

performed by the non-Big Four firms.

We also re-estimate the audit fee and credit ratings models for the sub-sample of

304 companies that were audited by Big Four firms in 2003. In Table 4, the

VOL_AUDITit coefficient is just 0.02 (compared with 0.13 in the full sample) and it is not

statistically different from zero. This is perhaps unsurprising given that the companies

that chose Big Four audits had already revealed their preference for high-quality

auditing. Thus, our results for H2 are significant only for the companies that hired non-

Big Four firms during the mandatory regime. The credit ratings models in Table 5

continue to provide support for H3. Specifically, the clients that wanted to be audited

received credit ratings that were nearly seven points higher during the mandatory

regime (t-statistic = 1.61) and fourteen points higher during the voluntary regime (t-

statistic = 5.21). Finally, the models in Table 6 reveal that credit ratings increased for the

268 Big Four clients that chose to remain audited in 2004 (t-statistic = 4.08), while there

was an insignificant drop in credit ratings for the 36 Big Four clients that were no longer

audited (t-statistic = -0.48).

33
5. Conclusions

The paucity of evidence on the implications of forced versus voluntary auditing

motivates our research on economic outcomes surrounding legislation that rescinds

mandatory audits for small private companies in the UK. We analyze two main

arguments. First, we expect the companies that did not want to be audited were

privately contracting for a relatively low level of audit assurance during the mandatory

regime. Second, we argue that there is a signaling benefit from allowing audits to be

voluntary since the company’s decision on whether to be audited conveys valuable

information to outsiders about its type. To test these predictions, we exploit a natural

experiment in which audits were required for companies with fiscal years ending before

January 30, 2004 and voluntary afterward. We examine both the audit choices that these

companies made when audits were mandatory and the consequences of the change in

the regulatory regime.

Two major insights stem from our empirical analysis. First, the companies that

did not want to be audited were significantly less likely to appoint Big Four audit firms

and they paid significantly lower fees during the mandatory regime relative to the

companies that wanted to be audited. These results suggest that such companies were

passively complying with the audit requirement and they were subject to less strict

monitoring than the companies that genuinely wanted to be audited. The main policy

rationale for mandatory auditing is that external stakeholders obtain significant

assurance benefits when companies that would not voluntarily choose an audit are

forced into it. Our results suggest that these benefits are likely to be modest since the

companies that do not want to be audited privately contract for low levels of audit

assurance when audits are legally required.

34
 Our second major finding is that the move away from mandatory auditing

engenders an important role for signaling, in which the low-risk companies conveyed

their favorable borrowing characteristics by continuing with an audit. Given recent

research that auditors play a major role as information intermediaries in debt

contracting, credit ratings suit our analysis of the signaling impact of the change in audit

regime on the rating agency’s perception of borrower risk. The companies that remain

audited enjoyed significantly higher credit ratings following the regime change even

though these companies were audited in both 2003 and 2004 and the assurance value of

their audits apparently did not change. We attribute this upgrade in credit ratings to the

positive signal that voluntarily submitting to an audit sends to outsiders. Importantly,

the decision to be audited voluntarily conveys information that is incremental to the

signal that would occur in a mandatory regime, where companies turn to alternative

mechanisms ⎯ namely, appointing Big Four auditors and paying higher audit fees ⎯

which can partly reveal their types. This supports the main argument in favor of

voluntary auditing, which is that imposing audits prevents companies from more fully

revealing their types.

35
References

Admati, A. and P. Pfeiderer. 2000. “Forcing firms to talk: Financial disclosure

regulation and externalities.” Review of Financial Studies 13: 479-515.

Allee, K. D. and T. L. Yohn. 2009. “The demand for financial statements in an

unregulated environment: An examination of the production and use of financial
statements by privately held small businesses.” The Accounting Review 84: 1-25.

Aranya, N., 1974. “The influence of pressure groups on financial statements in Britain.”
Abacus June, 3–12. Reprinted in Parker, R.H. and T.A. Lee, eds. “The Evolution
of Company Financial Reporting” (Thomas Nelson and Sons, Ltd., 1979), 265-
274.

Barton, J. and G. Waymire. 2004. “Investor protection under unregulated financial

reporting.” Journal of Accounting and Economics 38: 65-116.

Beaver, W. H. 1998. Financial Reporting: An Accounting Revolution. Prentice-Hall,

Engelwood Cliffs, NJ.

Bentson, G. 1969. “The value of SEC’s accounting disclosure requirements.” The

Accounting Review 54: 515-532.

Bharath, S. T., J. Sunder, and S. V. Sunder. 2008. “Accounting quality and debt
contracting.” The Accounting Review 83: 1-28.

Blackwell, D., T. Noland, and D. Winters. 1998. “The value of auditor assurance:
Evidence from loan pricing.” Journal of Accounting Research 36: 57-70.

Botosan, C. A. 1997. “Disclosure level and the cost of equity capital.” The Accounting
Review 72: 323–349.

Brav, O. 2009. “How does access to the public capital market affect firms’ capital
structure?” Journal of Finance 64: 263-308.

Chow, C. W. 1982. “The demand for external auditing: Size, debt and ownership
influences.” The Accounting Review 57: 272-291.

Clatworthy, M. A. and M. J. Peel. 2007. “The effect of corporate status on external audit
fees: Evidence from the UK.” Journal of Business Finance & Accounting 34: 169-201.

Coffee, J. 1984. “Market failure and the economic case for a mandatory disclosure
system.” Virginia Law Review 70, 717-753.

36
Collis, J., R. Jarvis, and L. Skerratt. 2004. “The demand for the audit in small companies
in the UK.” Accounting and Business Research 34: 87-100.

Datar, S. M., G. A. Feltham, and J. S. Hughes. 1991. “The role of audits and audit
quality in valuing new issues.” Journal of Accounting and Economics 14: 3-49.

Dedman, E., and C. Lennox. 2009. “Perceived competition, profitability, and the
withholding of information about sales and the cost of sales.” Journal of
Accounting and Economics, forthcoming.

DeFond, M. and J. Jiambalvo. 1994. “Debt covenant violation and manipulation of

accruals.” Journal of Accounting and Economics 17: 145-176.

Doumpos, M. and F. Pasiouris. 2005. “Developing and testing models for replicating
credit ratings: A multicriteria approach.” Computational Economics 25: 327–341.

Dye, R. A. 1990. “Mandatory versus voluntary disclosures: The case of financial and
real externalities.” The Accounting Review 65: 1-24.

Dye, R. A. 1993. “Auditing standards, legal liability, and auditor wealth.” Journal of
Political Economy 101: 887-914.

Fenn, G. W. 2000. “Speed of issuance and the adequacy of disclosure in the 144A high-
yield debt market.” Journal of Financial Economics 56: 383-405.

Fishman, M. and K. Hagerty. 1989. “Disclosure decisions by firms and the competition
for price efficiency.” Journal of Finance 44: 633-646.

Fortin, S. and J. A. Pittman. 2007. “The role of auditor choice in debt pricing in private
firms.” Contemporary Accounting Research 24: 859-896.

Francis, J. R. 2004. “What do we know about audit quality?” The British Accounting
Review 36: 345-368.

Francis, J. R., I. K. Khurana, X. Martin, and R. Pereira. 2008. “Private audits in

developing and developed countries.” Working paper: University of Missouri-
Columbia.

Francis, J., R. LaFond, P. Olsson, and K. Schipper. 2005. “The market pricing of accruals
quality.” Journal of Accounting and Economics 39: 295-327.

37
Givoly, D., C. Hayn, and S. Katz. 2009. “Does public ownership of equity improve
earnings quality?” Working paper: Penn State University, UCLA, and Harvard
University.

Gonedes, N. J. and N. Dopuch. 1974. “Capital market equilibrium, information

production, selecting accounting techniques: Theoretical framework and review
of empirical work.” Journal of Accounting Research 12: 48-129.

Graham, J. R., C. R. Harvey, and S. Rajgopal. 2005. “The economic implications of

corporate financial reporting.” Journal of Accounting and Economics 40: 3-73.

Gu, Z. 2007. “Across-sample incomparability of R2s and additional evidence on value

relevance changes over time.” Journal of Business Finance and Accounting 34: 1073–
1098.

Holthausen, R. and R. Watts. 2001. “The relevance of the value-relevance literature for
accounting standard setting.” Journal of Accounting and Economics 31: 3-75.

Jamal, K. and S. Sunder. 2008. “Are regulatory mandate and independence necessary
for audit quality?” Working paper, University of Alberta and Yale University.

Jensen, M. C. and W. H. Meckling. 1976. “Theory of the firm: Managerial behavior,

agency costs, and ownership structure.” Journal of Financial Economics 3: 305-360.

Lambert, R., C. Leuz, and R. Verrecchia. 2007. “Accounting information, disclosure,

and the cost of capital.” Journal of Accounting Research 45: 385-420.

La Porta, R., F. Lopez-de-Silanes, and A. Shleifer. 2006. “What works in securities

laws?” Journal of Finance 61: 1-32.

Leftwich, R. 1983. “Accounting information and private markets: Evidence from

private lending agreements.” The Accounting Review 58: 23-42.

Leftwich, R. 2004. “Discussion of: Investor protection under unregulated financial

reporting.” Journal of Accounting and Economics 38: 117-128.

Lennox, C. 1999. “Are large auditors more accurate than small auditors?” Accounting
and Business Research 29: 217-227.

Leuz, C. and P. Wysocki. 2008. “Economic consequences of financial reporting and

disclosure regulation: A review and suggestions for future research.” Working
paper: University of Chicago.

38
Mansi, S. A., W. F. Maxwell, and D. P. Miller. 2004. “Does audit quality and tenure
matter to investors? Evidence from the bond market.” Journal of Accounting
Research 42: 755-793.

Melumad, N. and L. Thoman. 1990. “On auditors and the courts in an adverse selection
setting.” Journal of Accounting Research 28: 77-120.

Pagano, M., Panetta, F. and L. Zingales. 1998. “Why do companies go public? An

empirical analysis.” Journal of Finance 53: 27-64.

Palmrose, Z-V. 1986. “Audit fees and auditor size: Further evidence.” Journal of
Accounting Research 24: 97-110.

Palmrose, Z-V. 1987. “Who got sued?” Journal of Accountancy 183: 67-69.

Peel, M. J. and R. Roberts. 2003. “Audit fee determinants and auditor premiums:
Evidence from the micro-firm sub-market.” Accounting and Business Research 33:
207-233.

Pittman, J. A. and S. Fortin. 2004. “Auditor choice and the cost of debt capital for newly
public firms.” Journal of Accounting and Economics 37: 113-136.

Santos, J. A. C. 2003. “Why firm access to the bond market differs over the business
cycle: A theory and some evidence.” Working paper: Federal Reserve Bank of
New York.

Seetharaman, A., F. Gul, and S. Lynn. 2002. “Litigation risk and audit fees: Evidence
from UK firms cross-listed on US market.” Journal of Accounting and Economics
33: 91-115.

Smith, C. W. and J. B. Warner. 1979. “On financial contracts and optimal capital
structure: An analysis of bond covenants.” Journal of Financial Economics 7: 117-
161.

Sunder, S. 2003. “Rethinking the structure of accounting and auditing.” Yale ICF
Working paper No. 03-17. Yale School of Management.

Sweeney, A. 1994. “Debt covenant violations and managers’ responses.” Journal of

Accounting and Economics 17: 281-308.

St. Pierre, K. and J. A. Anderson. 1984. “An analysis of the factors associated with
lawsuits against public accountants.” The Accounting Review 59: 242-263.

39
Titman, S. and B. Trueman. 1986. “Information quality and the valuation of new
issues.” Journal of Accounting and Economics 8: 159-172.

Watkins, A. L., W. Hillison, and S. Morecroft. 2004. “Audit quality: A synthesis of

theory and empirical evidence.” Journal of Accounting Literature 23: 153-193.

Watts, R. L. 1977. “Corporate financial statements: a product of the market and political
processes.” Australian Journal of Management 53-75.

Watts, R. and J. Zimmerman. 1983. “Agency problems, auditing, and the theory of the
firm: Some evidence.” Journal of Law and Economics 26: 613-634.

Welker, M. 1995. “Disclosure policy, information asymmetry and liquidity in equity

markets.” Contemporary Accounting Research 11: 801-828.

Yu, F. 2005. “Accounting transparency and the term structure of credit spreads.”
Journal of Financial Economics 75: 53-84.

Zingales, L. 2009. “The future of securities regulation.” Journal of Accounting Research

37: 391-425.

40
 Table 1
Industry composition of the sample companies.
UK SIC code Industry Companies
0001–0999 Agriculture, hunting, forestry, and fishing 79
1000–1499 Mining 12
1500–3999 Manufacturing 719
4000–4499 Electricity, gas, and water supply 3
4500–4999 Construction 748
5000–5499 Wholesale and retail trade 1,221
5500–5999 Hotels and restaurants 148
6000–6499 Transport, storage, and communication 338
7000–7499 Business services 1,414
7500–7999 Public services and defense 3
8000–9999 Other service activities 454
Total 5,139

41
 Table 2
Variable means during the final year of the mandatory audit regime (2003) and
the first year of the voluntary audit regime (2004).
Mandatory audit regime (t = 2003) Voluntary audit regime (t = 2004) Differences in means
Observations Mean Observations Mean
RATINGit 5,139 64.51 5,139 68.37 t-stat. = 9.32***
AGEit 5,139 17.99 5,139 18.99 t-stat. = 3.09***
INTCOVit 5,139 0.48 5,139 0.46 t-stat. = -1.03
Salesit (£000) 5,139 2,124.93 5,139 2,077.30 t-stat. = -1.52
LTSit 5,139 7.36 5,139 7.33 t-stat. = -1.31
Assetsit (£000) 5,139 1,172.69 5,139 1,173.06 t-stat. = 0.02
LTAit 5,139 6.82 5,139 6.84 t-stat. = 1.07
LIQUIDITYit 5,139 1.48 5,139 1.58 t-stat. = 2.73***
LEVERAGEit 5,139 0.70 5,139 0.69 t-stat. = -0.52

***, **, * = statistically significant at the 1%, 5%, 10% levels (two-tailed).
Variable definitions
RATINGit = the credit score (from 1 to 100 where a higher score implies a better rating) for company i in year t. AGEit = the age of
company i in year t. INTCOVit = interest expense divided by earnings before interest and taxation (the INTCOVit ratio is capped at
2.00 and we assign a value of 2.00 if earnings before interest and taxation is negative). LTSit = log of total sales (£000). LTAit = log of
total assets (£000). LIQUIDITYit = quick ratio ((current assets – inventory)/current liabilities). LEVERAGEit = total liabilities divided
by total assets. LAFit = log of audit fees (£000). BIG4it = one if the company is audited by a Big Four audit firm, zero otherwise.

42
 Table 3
Descriptive statistics after partitioning the sample by the company’s decision on whether to be audited (VOL_AUDITit) and the
prevailing audit regime (mandatory audits in 2003, voluntary audits in 2004).
Panel A: The final year of the mandatory audit regime (t = 2003)
Companies wanting to be audited Companies not wanting to be audited Differences
(VOL_AUDITit = 1) (VOL_AUDITit = 0) in means
Observations Mean Observations Mean
BIG4it 3,440 0.08 1,699 0.02 t-stat. = -8.16***
Audit feesit (£000) 3,440 5.68 1,699 4.27 t-stat. = -11.87***
LAFit 3,440 1.52 1,699 1.26 t-stat. = -13.32***
RATINGit 3,440 67.48 1,699 58.50 t-stat. = -13.64***
AGEit 3,440 18.81 1,699 16.32 t-stat. = -5.11***
INTCOVit 3,440 0.50 1,699 0.42 t-stat. = -3.50***
LTSit 3,440 7.40 1,699 7.28 t-stat. = -4.21***
LTAit 3,440 6.91 1,699 6.64 t-stat. = -12.80***
LIQUIDITYit 3,440 1.51 1,699 1.41 t-stat. = -1.87*
LEVERAGEit 3,440 0.68 1,699 0.74 t-stat. = 3.63***

Panel B: The first year of the voluntary audit regime (t = 2004)

Voluntarily audited companies Unaudited companies Differences
(VOL_AUDITit = 1) (VOL_AUDITit = 0) in means
Observations Mean Observations Mean
RATINGit 3,440 73.38 1,699 58.25 t-stat. = -28.49***
AGEit 3,440 19.81 1,699 17.32 t-stat. = -5.12***
INTCOVit 3,440 0.48 1,699 0.43 t-stat. = -2.25***
LTSit 3,440 7.39 1,699 7.23 t-stat. = -5.47***
LTAit 3,440 6.93 1,699 6.64 t-stat. = -13.37***
LIQUIDITYit 3,440 1.62 1,699 1.51 t-stat. = -1.78*

43
LEVERAGEit 3,440 0.67 1,699 0.75 t-stat. = 4.46***
Table 3 (cont.)

Panel C: Companies not wanting to be audited (VOL_AUDITit = 0)

Final year of the mandatory audit First year of the voluntary audit Differences
regime (t = 2003) regime (t = 2004) in means
Observations Mean Observations Mean
RATINGit – RATING.t 1,699 -6.01 1,699 -10.13 t-stat. = -6.49***
AGEit – AGE.t 1,699 -1.67 1,699 -1.67 t-stat. = 0.00
INTCOVit – INTCOV.t 1,699 -0.05 1,699 -0.03 t-stat. = 0.77
LTSit – LTS.t 1,699 -0.08 1,699 -0.11 t-stat. = 0.90
LTAit – LTA.t 1,699 -0.19 1,699 -0.20 t-stat. = 0.36
LIQUIDITYit – LIQUIDITY.t 1,699 -0.07 1,699 -0.07 t-stat. = 0.05
LEVERAGEit – LEVERAGE.t 1,699 0.04 1,699 0.05 t-stat. = -0.49

Panel D: Companies wanting to be audited (VOL_AUDITit = 1)

Final year of the mandatory audit First year of the voluntary audit Differences
regime (t = 2003) regime (t = 2004) in means
Observations Mean Observations Mean
LAFit 3,440 1.52 3,440 1.53 t-stat. = 1.17
BIG4it 3,440 0.08 3,440 0.07 t-stat. = -0.92
RATINGit – RATING.t 3,440 2.97 3,440 5.00 t-stat. = 4.02***
AGEit – AGE.t 3,440 0.82 3,440 0.82 t-stat. = 0.00
INTCOVit – INTCOV.t 3,440 0.03 3,440 0.02 t-stat. = -0.51
LTSit – LTS.t 3,440 0.04 3,440 0.05 t-stat. = 0.56
LTAit – LTA.t 3,440 0.09 3,440 0.10 t-stat. = 0.27
LIQUIDITYit – LIQUIDITY.t 3,440 0.03 3,440 0.04 t-stat. = 0.04
LEVERAGEit – LEVERAGE.t 3,440 -0.02 3,440 -0.02 t-stat. = -0.41

44
***, **, * = statistically significant at the 1%, 5%, 10% levels (two-tailed).
Table 3 (cont.)

Variable definitions
VOL_AUDITit = one if company i chooses to be audited during the voluntary audit regime, and zero if company i chooses not to be
audited during the voluntary audit regime. BIG4it = one if the company is audited by a Big Four audit firm, zero otherwise. BIG4.t =
the mean value of BIG4it across all companies in year t. LAFit = log of audit fees (£000). LAF.t = the mean value of LAFit across all
companies in year t. RATINGit = the credit rating score (from 1 to 100 where a higher score implies a better rating) for company i in
year t. RATING.t = the mean value of RATINGit across all companies in year t. AGEit = the age of company i in year t. AGE.t = the
mean value of AGEit across all companies in year t. INTCOVit = interest expense divided by earnings before interest and taxation (the
INTCOVit ratio is capped at 2.00 and we assign a value of 2.00 if earnings before interest and taxation is negative). INTCOV.t = the
mean value of INTCOVit across all companies in year t. LTSit = log of total sales (£000). LTS.t = the mean value of LTSit across all
companies in year t. LTAit = log of total assets (£000). LTA.t = the mean value of LTAit across all companies in year t. LIQUIDITYit =
quick ratio ((current assets – inventory)/current liabilities). LIQUIDITY.t = the mean value of LIQUIDITYit across all companies in
year t. LEVERAGEit = total liabilities divided by total assets. LEVERAGE.t = the mean value of LEVERAGEit across all companies in
year t.

45
 Table 4
Auditor choice and audit fees during the final year of the mandatory audit regime (2003).

Auditor choice model Audit fees model

Dep. var. = BIG4it LAFit
Model 1 Model 2
Coefft. z-stat. Coefft. t-stat.
VOL_AUDITit 1.10 5.94*** 0.13 7.20***
AGEit -0.00 -1.00 0.01 5.64***
INTCOVit 0.29 3.87*** 0.13 11.55***
LTSit 0.15 2.35** 0.25 21.83***
LTAit 0.77 7.62*** 0.19 14.84***
LIQUIDITYit 0.07 2.82*** 0.01 1.74*
LEVERAGEit 0.52 5.59*** 0.03 1.45
BIG4it 0.20 5.35***

Industry dummy variables? YES YES

R2 / pseudo R2 10.5% 23.5%
Observations 5,139 5,139

Industry dummy variables are included (see Table 1) but the coefficients and the intercept are
not tabulated.
***, **, * = statistically significant at the 1%, 5%, 10% levels (two-tailed). The t-statistics and z-
statistics are reported in parentheses with standard errors that are adjusted for
heteroskedasticity.
Variable definitions
BIG4it = one if the company is audited by a Big Four audit firm, zero otherwise. LAFit = log of
audit fees (£000). VOL_AUDITit = one if company i chooses to be audited during the voluntary
audit regime, and zero if company i chooses not to be audited during the voluntary audit
regime. AGEit = the age of company i in year t. INTCOVit = interest expense divided by earnings
before interest and taxation (the INTCOVit ratio is capped at 2.00 and we assign a value of 2.00 if
earnings before interest and taxation is negative). LTSit = log of total sales (£000). LTAit = log of
total assets (£000). LIQUIDITYit = quick ratio ((current assets – inventory)/current liabilities).
LEVERAGEit = total liabilities divided by total assets.

46
 Table 5
Credit ratings during the final year of the mandatory audit regime (2003) and the first year of
the voluntary audit regime (2004).
Final year of the mandatory First year of the voluntary
regime (t = 2003) regime (t = 2004)
Dep. var. = RATINGit RATINGit
Model 1 Model 2
Coefft. t-stat. Coefft. t-stat.
VOL_AUDIT it 7.68 13.41*** 12.94 29.55***
AGEit 0.25 14.57*** 0.29 21.04***
INTCOVit -10.21 -26.49*** -8.11 -25.82***
LTSit 0.48 1.46 1.82 7.30***
LTAit 2.43 5.85*** 3.79 11.08***
LIQUIDITYit 1.52 8.28*** 0.28 2.51**
LEVERAGEit -6.73 -10.33*** -3.16 -6.83***
Industry dummy variables? YES YES
R2 / pseudo R2 27.4% 23.6%
Observations 5,139 5,139

Industry dummy variables are included (see Table 1) but the coefficients and the intercept are
not tabulated.
***, **, * = statistically significant at the 1%, 5%, 10% levels (two-tailed). The t-statistics and z-
statistics are reported in parentheses with standard errors that are adjusted for
heteroskedasticity.
Variable definitions
RATINGit = the credit rating score (from 1 to 100 where a higher score implies a better rating)
for company i in year t. VOL_AUDITit = one if company i chooses to be audited during the
voluntary audit regime, and zero if company i chooses not to be audited during the voluntary
audit regime. AGEit = the age of company i in year t. INTCOVit = interest expense divided by
earnings before interest and taxation (the INTCOVit ratio is capped at 2.00 and we assign a value
of 2.00 if earnings before interest and taxation is negative). LTSit = log of total sales (£000).
LTAit = log of total assets (£000). LIQUIDITYit = quick ratio ((current assets – inventory)/current
liabilities). LEVERAGEit = total liabilities divided by total assets.

47
 Table 6
Changes in credit ratings when auditing switches from being mandatory to voluntary (2003–2004).
Companies that switch from Companies that switch from
Full sample mandatory audits to voluntary audits mandatory audits to no audits
Dep. var. = Δ(RATINGit – RATING.t) Δ(RATINGit – RATING.t) Δ(RATINGit – RATING.t)
Model 1 Model 2 Model 3
Coefft. t-stat. Coefft. t-stat. Coefft. t-stat.
Intercept -3.91 -8.76*** 1.93 5.99*** -3.91 -8.81***
VOL_AUDITit 5.84 10.61***
Δ(INTCOVit – INTCOV.t) -8.16 -23.76*** -7.90 -19.07*** -8.81 -14.33***
Δ(LTSit – LTS.t) -0.04 -0.06 0.34 0.41 -1.01 -1.01
Δ(LTAit – LTA.t) 2.84 2.91*** 2.24 1.81* 3.90 2.35**
Δ(LIQUIDITYit – LIQUIDITY.t) -0.11 -0.48 -0.01 -0.04 -0.36 -0.68
Δ(LEVERAGEit – LEVERAGE.t) -2.58 -2.84*** -1.99 -1.50 -3.34 -2.93***

R2 / pseudo R2 13.8% 11.5% 13.4%

Observations 5,139 3,440 1,699

***, **, * = statistically significant at the 1%, 5%, 10% levels (two-tailed). The t-statistics and z-statistics are reported in parentheses with
standard errors that are adjusted for heteroskedasticity.
Variable definitions:
VOL_AUDITit = one if company i chooses to be audited during the voluntary audit regime, and zero if company i chooses not to be audited
during the voluntary audit regime. RATINGit = the credit rating score (from 1 to 100 where a higher score implies a better rating) for
company i in year t. RATING.t = the mean value of RATINGit across all companies in year t. AGEit = the age of company i in year t. AGE.t =
the mean value of AGEit across all companies in year t. INTCOVit = interest expense divided by earnings before interest and taxation (the
INTCOVit ratio is capped at 2.00 and we assign a value of 2.00 if earnings before interest and taxation is negative). INTCOV.t = the mean
value of INTCOVit across all companies in year t. LTSit = log of total sales (£000). LTS.t = the mean value of LTSit across all companies in year
t. LTAit = log of total assets (£000). LTA.t = the mean value of LTAit across all companies in year t. LIQUIDITYit = quick ratio ((current assets
– inventory)/current liabilities). LIQUIDITY.t = the mean value of LIQUIDITYit across all companies in year t. LEVERAGEit = total liabilities

48
View publication stats

divided by total assets. LEVERAGE.t = the mean value of LEVERAGEit across all companies in year t.

49