PLEASE DO NOT QUOTE OR DISTRIBUTE WITHOUT THE PERMISSION OF THE
AUTHOR “AUDIT QUALITY, INFORMATION DYNAMICS
AND THE PARTNER EFFECT”
J.P. van Buuren
PhD Student
Nyenrode University, School of Accountancy
Breukelen, the Netherlands ABSTRACT
The main objective of this study is to assess the assumed homogeneity of audit quality between
and within large audit firms. To measure the audit quality differences, the Auditor Conservatism
Ratio (ACR) is developed, which draws on the value relevance of accounting data theory of
Feltham and Ohlson (1995). The results are based on a sample of 378 Dutch municipalities with
1043 yearly observations over the period 1999-2005. Results show that significant differences
exist in audit quality (i) between large audit firms, (ii) between audit partners within an audit firm
and (iii) even between audit partners of the same local audit office.
Key words: audit quality, information dynamics, audit partner effect, municipalities


An increasing body of auditing research literature suggests that within audit firms,
quality differences in audit services exist, e.g. price premiums are paid for higher quality services
(e.g. Francis, 2005; Ferguson et al. 2003). In this paper, theory is developed why audit quality is
likely to be dependent on the individual auditor’s financial reporting preferences. The applied
theory is based on differences in materiality assessment and risk appetite by individual audit
partners. Auditors are expected to develop audit strategies based on materiality and risk
allocation. Auditing literature suggests the existence of these audit strategies e.g. risk adverse
(conservative) auditors are expected to downsize risks in the financial statements, i.e. have lower
discretionary accruals (Becker et al., 1998; Francis et al. 1999). On the other hand, auditors may
develop specific risk allocation knowledge and are able to e.g. except higher than average accrual

1
Or assumes differences in professional judgments are sufficiently low and/or assumes that all large firms are able (to the same extent)
to manage potential differences to an acceptable low level though monitoring and peer reviews.
2
positions. Liu and Simunic (2005) suggest that profit sharing rules may lead to specialization in
specific industries or clients with certain risk profiles.
The audit quality definition in this paper draws on the value relevance of accounting
data approach, which is based on the linear information model (LIM) as developed by Feltham
and Ohlson (1995). It is reasoned that audits and information dynamics from LIM are closely
related: the better the audit, the clearer the information dynamics and the better the predictability-
property of accounting data becomes. The predictability property is essential for the relevance of
financial statements according to the financial reporting framework of the IASB (Framework,
par.15ff). Also IFAC auditing standards (ISA No. 200.44) consider the usefulness of accounting
data as central principle. Auditors should always consider how the applied reporting options by
the auditee influences [economic] decision making (IFAC Assurance Framework, par.47). The
audit quality definition is therefore defined as “the performance of an auditor to (i) deliver
appropriate professional opinions supported by necessary evidence and objective judgments and
(ii) let audited (financial) accounting statements have high value relevance”. This definition is
equal to the definition of European Federation of Accountants (FEE, 2006), but stresses the

in the value relevance literature of Feltham and Ohlson (1995). Furthermore, the auditor
conservatism ratio (ACR) is developed in section 3. In section 4, evidence is provided of audit
quality differences between and within large audit firms. The research is concluded in section 5.
2. Literature review
The purpose of this paper is to examine whether the quality of audit services provided by large
audit firms are homogenous. To answer this question, one should realize the main reason for
auditing research: the assumption that information asymmetry is not equal over all companies and
therefore quality-differentiated audits are demanded by both the company management and
4
investors (Titman and Trueman, 1986; Datar et al., 1991; DeFond, 1992). Beattie and Fearnley
(1995) recapitulate the differences in audit quality demands: (i) product differentiation hypotheses
that exist of (a) differential agency costs across clients and over time (DeAngelo, 1981) and (b)
signaling the credibility of financial statements through auditor choice and thus the honesty of the
management (Dopuch and Simunic, 1982) and (ii) different levels of insurance (deep pockets) of
audit firms.
Secondly, one should consider that different perspectives on audit quality should
result in different definitions. In their report on audit quality, the Institute of Chartered
Accountants in England and Wales (ICAEW, 2002) acknowledge the versatility of audit quality
perceptions as they state that “each stakeholder will give a different meaning to audit quality”.
Agency theory settings (Jensen and Meckling, 1976; Watts and Zimmerman, 1983)
are often used to motivate audit research. From the agency theory, three mean research
perspectives on audit quality can be distinguished:
a) Demand side - users of audited information; This perspective on audit quality focuses on
the perception of audit quality by users of audited information;
b) Demand side – providers of audited information; Following the economic rationale of
management, audit quality is considered the extent of marginal benefits the auditee has
received from the audit
2
. Simunic and Stein (1987) reason that the demand for product
differentiation of audit services is based on three characteristics: (i) contribution of the

misstatements”. However, these currently applied definitions seem to be principally focused on
the auditor’s risks, not on the auditor’s responsibilities regarding the usefulness of audited

2
Note that the marginal benefits of audit quality do not impair audit quality itself, as external audits are assumed a surrogate to internal
audits/control (Simunic, 1980) and it is assumed to be a rational equalization of costs and benefits.
6

financial statements. The auditor’s role concerning usefulness of audited statements is considered
in this paper.
3. Research model
3.1 Usefulness of financial statements
The objective of the audit of financial statements is to “enable the auditor to express an
opinion whether the financial statements are prepared, in all material respects, in accordance with
an applicable financial reporting framework” (ISA No. 200.2). An opinion is considered
appropriate when the tenor of the auditor’s opinion and the extent of “true and fair view
3
” of the
financial statements align. In extreme cases, this may include a true and fair override (par.
2;362:4 Dutch Civil Code), if a specific requirement of the Civil Code does not result in a ‘true
and fair view’. In the IASB Framework for financial statements, usefulness
4
is presented as a
central feature: “the objective of financial statements is to provide information about the financial
position, performance and changes in financial position of an entity that is useful for a wide range
of users in making economic decisions” (IASB Framework, par. 12). Information in financial
statements is thus considered useful, if it supports economic decision-making. IASB Framework
(par. 26) states: “To be useful, information must be to the decision-making needs of users.
Information has the quality of relevance when it influences economic decisions of users by
helping them evaluate past, current and future events or confirming or correcting their previous

response to reported earnings.
3.2 Value relevance of accounting data
Feltham and Ohlson (1995) have analyzed theoretically the relation between market valuations
and current accounting numbers by introducing the ‘dynamic linear information model’ (LIM).
LIM illustrates that accounting data captures relevant information, even under conservative
accounting (FO’95). Easton (2001) reports two kinds of value creation that are potentially not

4
More generally, the IFAC-handbook (2006) states that acceptable financial reporting frameworks should consider the
usefulness of the reported information for users, which exhibit normally the following five attributes (ISA No. 200.44, IFAC
Handbook, 2006): relevance, completeness, reliability, neutrality and understandability.
5
Improving the usefulness is in line with one of the primary objectives of the IFAC: “to contribute to the efficiency of the global
economy by improving confidence in the quality and reliability of financial reporting” (IFAC, Handbook, 2006).
8
mapped in the financial statements due to conservative financial reporting principles (realization
and prudence principles):
- Economic value added is communicated through non-accounting information about future
developments (e.g. directors report), but also through a true classification of transitory and
permanent earnings. Regarding to non-accounting information, the auditor should
consider reported developments and reality of current conditions as mapped in the
financial statements. The true classification (and disclosure) of permanent and transitory
earnings is a result of thorough understanding of the auditee’s business and circumstances
by the auditor;
- Accounting added value is created due to accounting principles e.g. non-valued profit
margins in inventory. Feltham and Ohlson (1996) have analyzed that overdepreciation
leads to a downward bias in earnings under the assumption of growth. This
overdepreciation is not problematic if earnings reports remain predictable: i.e. capital
markets should be able to estimate and value this bias. Easton and Pae (2004) report
(limited) evidence on this matter.

opinion. In this paper, it is reasoned that in this grey area, the value relevance of accounting data
depends partly on the auditor. In Figure 1, the relation between the value relevance of accounting
data and the auditor’s assessment of materiality is illustrated.
[Insert figure 1]
The curves in Figure 1 represent the assumed distribution functions of financial reports audited by
three archetypes of auditors: high quality, conservative and liberal. High quality auditors have
signed – on average – financial reports with high(er) value relevance of accounting data.
Conservative (liberal) auditors have signed financial reports with a systematic bias towards
undervaluing (overvaluing) the information dynamics of accounting data. A simple example of

6
Note that the auditor’s responsibility concerning usefulness of accounting data goes beyond the legal minimum (i.e. absence of
(observable) material misstatements/errors) and cannot be enforced under the current jurisdiction. However, the lack of legal action
possibilities against auditors concerning low usefulness of accounting data is considered not to reduce the concerning responsibilities
of auditors.
10

systematic bias is the impairment of goodwill: a conservative (liberal) auditor will have the
tendency to let goodwill bias downwards (upwards) in the grey area of acceptable values. In
contrary, high quality auditors will have, on average, the best estimates of the ‘true’ value of
goodwill.
Figure 1 suggests that the three types of auditors have different boundaries of
materiality. It is assumed that all boundaries of assessed materiality are still acceptable for the
auditing standards. Figure 1 illustrates that conservative auditors will not accept financial
statements beyond point M
C
(left tail), and will deliver an qualified opinion, but a liberal auditor
will still accept financial statements up to point M
L
(left tail)

audit risk varies accordingly;
- Auditors may invest in knowledge to improve the assessment of materiality and audit risk
allocation by specializing in certain industries or certain company profiles. As a result, the
auditors can develop effective and efficient audit strategies and offer more competing
audit prices than non-specialized auditors (Liu and Simunic, 2005);
- Audit risk appetite may also be dependent on the commercial strategy in competing
markets. As stated frequently by researchers (e.g. Becker et al., 1998), too high values of
accruals are more risky for reasons of litigation, than too low values of accruals. An
auditor may specialize in biasing accruals downwards, reducing audit risks and may be
able to cut back audit effort. As a result such an auditor may be able to offer very
competing audit prices, but is still able to make attractive profits.
The distribution of signed financial statements is considered dynamic. An auditor can change
commercial strategies and choose a certain specialization. Thus, to a certain degree, a high quality
auditor can choose to become a (more) liberal or conservative auditor over time, when it is
considered more attractive. However, the extent of auditor’s (technical) capabilities to
understand the true audit risks and the ability to communicate it effectively with management, are
considered important constraints on choosing the auditor’s business case. Please note that Figure
1 is just for illustrating differences between three (extreme) auditor’s archetypes. It is assumed
that in the ‘real world’ auditors are in between the three extremes.
In Figure 1, the shaded parts are the objective of my research as they deviate from the
“average auditor”. Please note that there is a large overlap between the three archetypes in which
the auditors deliver similar levels of audit quality. This overlap is explained by differences in (i)
12
the intrinsic preference of the auditee to depend on the auditor and (ii) the bargaining power of
the auditor. The intrinsic preference of the company to depend on the auditor is considered an
important attribute of the auditor effect: the more the client is inclined towards the auditor, the
stronger the auditor effect will be. The extent that clients depend on the auditor is assumed to be
randomly divided over the auditor client groups. On the other hand, the stronger the bargaining
power by the auditor, the more pronounced the auditor effect will be. Note that the bargaining
power is stronger if the auditor has better communicating skills. It is assumed that all auditors

statements and self-review threats. However, an auditor can be considered useful, if not only
material risks are communicated, but also a direction is given how to address the risk or error
appropriately: i.e. how management should address the risk or error in such a way that the auditor
is able to express an unqualified opinion
8
.
3.4 Auditor Conservatism Ratio
This research project examines whether the level of audit quality of services provided by large
audit firms is homogenous. As starting point for this research, the FO’95 framework is selected as
it focuses directly on value relevance of accounting data. As described in the previous section, it
is suggested that an auditor should affect the information dynamics of accounting data positively.
To measure the auditor’s effect, the following assumptions are made:
Assumption 1: High quality audits result in high quality information dynamics of accounting
data;
Assumption 2: High quality information dynamics lead to, on average, a better predictability
property of accounting data;

8
This is in line with the Code of Ethics (CoE, art. 290.168, IFAC Handbook, 2006) states that “the audit process involves an extensive
dialogue between the [audit] firm and audit client. During this process, management requests and receives input regarding such
matters as accounting principles and financial statement disclosures, …, and the methods used in determining the stated amounts of
assets and liabilities. Technical assistance of this nature and advice on accounting principles for financial statement audit clients are an
appropriate means to promote fair presentation of the financial statements. The provision of such advice does not generally threaten the
firm’s independence”.
14
Assumption 3: High audit quality is significantly affected by the individual auditor’s
assessment of significant estimates and the evaluation of the overall
presentation of the financial statements;
Assumption 4: All things equal, differences in individual auditor’s financial reporting
preferences affect the probability of profits and losses.

a
= (number of profits
jt;a
)/ total observations
jt;a
(1)
Where company (or municipal) j at moment t ∈{1, …, n} and company (or municipal) j ∈
auditor client group a {1, , l} and ACR ranges from [0,1]. Note that in case of municipalities,
profits are interpreted as an exploitation surplus. The auditor client groups can be pooled at the
office level and at the firm level. ACR
a
represents the extent an auditor signs financial statements
that have a systematic bias towards conservatism or liberalism:
- If ACR
a
= 0, than the auditor is considered “extreme liberal”;
- If ACR
a
=1, than the auditor is considered “extreme conservative”;
- If ACR
a
= (0.5 + δ), the auditor has no systematic bias, i.e. considered a high quality
auditor:
o δ=0 in case of municipalities and non-profit organizations that attempt to
maintain current wealth. If, in case of municipalities δ is significantly different
from zero, i.e. ACR
a
is significantly different from 0.5, than a systematic audit
bias is assumed.
o In case of companies, δ is unknown, but being profit-driven organizations, δ will

a
is expected to be less pronounced: only if the auditor is intending to deliver an other than
unqualified opinion, the management is willing to make adjustments. However, in case of high
auditor dependency, the auditee is open for, and expects suggestions from the auditor: technical
assistance and advice is considered common and the bargaining power of the auditor is higher:
the auditee is likely to make adjustments if the auditor suggests it. In case of high auditor
dependency, ACR
a
is expected to capture the auditor’s effect more clearly. The extent of auditor
dependency is expected to be distributed randomly over the auditor client groups.

9
Under the assumption of comparable client groups.
10
The reason that the number of underestimates is used to determine ACR in stead of e.g. the average deviation per auditor group
relates to avoid dominance of observations with large losses and profits.

17
The intrinsic level of bargaining power of auditors is also expected to differ between
auditors, because auditors will differ in their communication skills. The better the communication
skills, the stronger the bargaining power will be and the more pronounced the auditor effect.
An example of ACR
In this section, an example is described that illustrates the relation between differences in
accounting choices and ACR. Imagine three auditors with different financial reporting
preferences: liberal, conservative and high quality. Assume that these three auditors have exact
equal client groups of sufficient size, that –on average- only differ in accounting method with
regard to the recording of inventory. The differences in accounting choices concerning inventory
are illustrated in table 1, based on IAS No. 2.
[Insert table 1]
At first glance, the determination of cost prices for inventory valuation seems rather

value develops solely under the condition of growth. The development of accounting added value
is shown in Figure 2.
[Insert Figure 2]
In the example as illustrated in Figure 2, the inventory has increased with 1% over the ten-year
period, resulting in a difference in the cumulative earnings of:
• conservative: € 5,085 and positive accounting added value of € 15
• HQ: € 5,100 and no accounting added value
• Liberal: € 5,115 and negative accounting added value of € 15
19
In case of no growth over this period, the cumulative earnings will be equal for all
groups, i.e. € 5,100 and in case of a decrease of inventory of 1%, the cumulative earnings of the
conservative (liberal) group will be € 5,115 (€ 5,085).
In Figure 2, the development of earnings and accounting added value is illustrated.
The growth of inventory varies over the years, ranging from -7% (year 4) to + 7% (year 2). From
figure 2, it is clear that the earnings of the liberal client group are more volatile than the other
groups: in case of growth: higher profits and in case of decrease: larger losses. However, note that
the development of positive, respectively negative accounting added value is equal over the
conservative respectively liberal client group. The added value of the HQ-client group is nil, as it
is assumed to approach the true value. Finally, observe that the distribution of profits and losses is
different over the client groups:
- conservative: 9 profits, 1 loss Æ ACR = 0.9 (9/10)
- HQ: 8 profits, 2 losses Æ ACR = 0.8 (8/10)
- Liberal: 6 profits, 4 losses Æ ACR = 0.6 (6/10)
This simple example illustrates the consequence of different valuation choices concerning
inventory and supports the previously mentioned theory that auditor conservatism
11
will result in
different distributions of profits and losses (assumption 4).
Unit of analysis - partner level
The unit of analysis to measure the auditor effect is preferably the partner level, for the following

From the above arguments, it follows that the audit firms have a decentralized
organizational structure. Also in academic literature, the decentralized structure is described
(Francis et al., 1999; Reynolds and Francis, 2000; Craswell et al, 2002) and suggests that in

11
Or more generally, at the individual client level: accounting conservatism, as auditor conservatism is considered a special case of
accounting conservatism.
12
This includes also the internal acquisition of new engagements, i.e. the ability of the local auditor to attract new engagements that are
received at the headquarters of the audit firm.
21

assessing issues of auditor independence, the focus should be on the individual audit partner,
local office or other unit that decides on the acceptance or continuance of engagements.
3.5 Research model
Why research on Dutch municipalities?
The research is done on the municipality market in the Netherlands. As of 1985, Dutch
municipalities use the accrual system for financial reporting. The main advantage of this study is
that Dutch municipalities are highly comparable and all have similar legal tasks and thus ACR’s
can be compared. Other advantages of this market are (i) sufficient large sample and public
availability of data, (ii) data can be standardized transparently (amounts per capita), (iii) highly
specialized audit market and auditors have sufficient (comparable) clients, (iv) dominated by
Big4 accounting firms and both IFAC and internal audit firm quality standards are unimpaired
applicable, (v) litigation risks are considered low, as I am not aware of any claim for
compensation that has been granted
13
. Thus, the auditor’s effect can be reasonably attributed to
the auditor and is not expected to be biased through liability risks. Therefore, the Dutch municipal
market is considered an appropriate and straightforward setting to test the “audit partner effect”.
Financial reporting standards of Dutch municipalities

based on basic bookkeeping principles, valid for all accrual based financial reporting systems. It
is considered that the FO’95 is sufficient general to be applicable for non-profit organizations and
governmental settings, for the following reasons:
- the information dynamics are considered intrinsic characteristics of accrual systems
regardless of applied settings: capital market, private companies or governmental settings;
- FO’95 illustrates how current equity is related to future equity if t→∞. The benchmark for
future book value of equity is today’s capital market value. Thus, the relationship between
book value and market value of equity is allowed to have a long time horizon, even to
infinity. However, a long time horizon is not required in FO’95: extreme stock models
have short time horizons to equal market values;
- In order to examine information dynamics and value relevance of accounting data, the
capital market benchmark is not necessary, if other appropriate benchmarks are available.
In the previous sections, it is motivated that predictability is considered an appropriate
23

benchmark. Municipalities have no market value and can be characterized as a “pure
stock model” for the following reasons:
i. municipalities are spending organizations and are in general not expected to have
projects with positive net present values: thus the book value of these projects
represents the fair value;
ii. National and local taxes are levied to fund operations. Other income has a more
incidental character and is related to real estate, (ancient) buildings, non-
developed (potential) building lands and valuable shares of utility companies
(electricity, gas, water supply)
15
. In other words, municipalities will face
exploitation deficits, unless sufficient taxes are raised. Therefore, the ‘result for
the year’ is considered irrelevant for the determining the “fair value” of the
municipality’s assets.
The prediction model is the equation of current year’s and next year’s total reserves (TR) per

The municipalities’ market for audit services is dominated by the Big4 audit firms. Recently, also
the small audit firm IPA-ACON has entered the municipality market, but has too few
observations to analyze and is therefore deleted from the sample.
An overview of the yearly observations per firm and the representativeness of the
sample for all Dutch municipalities is reported in table 2, Panel B. The market domination by the
Big4 audit firms is quite high. The total sample includes about 30% of the total filed financial
statements over the period 1999-2005. Depending on the year, up to 63% of the municipalities are
represented in the sample. The sample and total municipalities seem to be distributed reasonable
similarly with regard to size-classes. Further note the market domination of Deloitte of 74% in the
sample. This market position by Deloitte can be explained by the acquisition of VB group in
1998, which was the former (quasi) state auditor of the Dutch municipalities. VB group has been
privatized in 1985 and used to have a market share of almost 100% before 1985.
The number of municipalities and observations per audit partner are reported in table 5.
Only audit firms and audit partners are included that meet the threshold are included in the
analyses. The partner names are not given for reasons of privacy. From table 5, it is clear that the
threshold of at least 6 yearly observations concerning at least 3 municipalities is met

16
The objectives of the questionnaire were twofold: (1) adding missing financial statement data and (2) collect
information that was not included in the annual reports such as: number of official board-members, labour costs, auditors’
reports and audit fees. Only raw and objective data has been asked, no subjective elements were included. In autumn
2006, all 459 municipalities were invited telephonically to participate: 372 partly filled in questionnaires were sent. At the
end, a total of 150 usable questionnaires were received, which is a 40% response rate. The reasons for withdrawal were
principally due to unexpected pressure of work, illness and dismissal of officers.
The questionnaires resulted in a total of
335 additional yearly observations and 6 municipalities were added to the sample.