Centre for Economic Policy Research

Discussion and Roundtables 91

---

Ernst-Ludwig von Thadden 
University of Lausanne
Ernst-Ludwig von Thadden, praising the Report for an impressive blend of 
practitioner and academic insights, focused his discussion on the audit side which
is mostly concerned with the problem of accounting, and less on the rating side.
As an example of what accounting is about, he quoted from Enron’s Risk
Management Manual: ‘Reported earnings follow the rules and principles of
accounting. The results do not always create measures consistent with underlying
economics. However, corporate management’s performance is generally measured
by accounting income, not underlying economics. Risk management strategies are
therefore directed at accounting rather than economic performance’. What are,
then, the incentives of accounting?
According to the Report, the recent accounting scandals have brought to the
fore three aspects. First comes the distinction between rules-based and principles-
based accounting. The former allows the auditor to hide behind formalities. In a
society that puts a strong emphasis on litigation, rules-based accounting may be
the best response of the profession. Second, the internal organization of audit
firms matters. The argument, not often made in this context, is that decentraliza-
tion forces local offices to accommodate local monopolists. This is highly similar
to what in banking is called concentration risk. The last component of scandals is
the governance of the firm-auditor relationship. Auditors who are hired and
remunerated by management instead of the board have the wrong loyalties. 
Von Thadden said that these points are well-taken but more interesting is that
the Report exonerates two of the most publicly accused culprits. The first is the
decay of ethical standards, and the second is the provision of non-audit 
consulting services by audit firms. He fully agreed with the report on the first
point, arguing that ethical decay is typically too easily invoked and too little
quantified. He pointed out, however, that the second issue might deserve more
attention. The authors exonerate service bundling because of the opportunity cost
of separating audit and non-audit services. Yet the authors are silent about the
potential costs arising from the bundling of these two services. In fact, the
‘dependency or conflict’ may depend precisely on the ‘question of what service is
performed’. So the two activities cannot be as easily separated as suggested by the
Report.
Von Thadden proposed to make his point with a little thought experiment of
what the auditing industry might be about. Consider a firm that may be either
well managed (IBM) or mismanaged (Enron). Call this state, 
with
proba(x=1)=q. An auditor exerts effort
to find out about x. Given their
effort, they learn about x through a signal
which is such that 
Once the auditor has observed s, they announce their audit that is either bad
or good,
. In the long run, the state of the firm, x, is revealed publicly. The
auditor’s objective is p(x,a), reflecting the potential trade-off between having a
good reputation and being on good terms with management. In particular, we
have
with p(0,g) and p(0,b) in between. p(1,b) is what you get if
the firm is good and you state that it is bad. This is the worst thing that can hap-
pen to an auditor. He is getting on bad terms with the management and he is los-
ing his reputation for being a good forecaster. On the other hand, p(1,g) is the best
that can happen to an auditor because he is having the right forecast and being
on good terms with the management.
p(0,g) and p(0,b) are in between. If the firm is bad and the auditor is making a
92 Conflicts of Interest in the Financial Services Industry
}
1
,
0
{
∈
x
]
1
,
2
/
1
[
∈
e
}
1
,
0
{
∈
s
e
x
s
proba
x
s
proba
=
=
=
=
=
=
)
0
0
(
)
1
1
(
}
,
{ g
b
a
∈
)
,
1
(
)
,
1
(
g
p
b
p
<<

---

good forecast, that is p(0, g), the auditor is on good terms with management but
since the forecast is wrong, that may be bad for his reputation. p(0,b) is the other
way around; the auditor makes the right forecast but the management does not
like it. How to solve this? 
The auditor has two possible strategies for her reporting: (strategy H being the
honest strategy):
Strategy H: a(0) = b, a(1) = g
Strategy A: a(0) = a(1) = g
A simple comparison shows that strategy H yields more than strategy A if and only
if:
(1-q)ep(0,b) + q(1-e)p(1,b) > (1-q)ep(0,g) + q(1-e)p(1,g)
This is likely to be the case if:
• p(0,g) and p(1,g) are small: gains from the firm-auditor relationship are not 
too large. 
• e is large: auditors monitor carefully.
• p(0,b) is large: high professional standards in auditing.
If these conditions do not hold, the auditor is likely to fall into the ‘Andersen
trap’, namely, provide favourable audit opinions when it is not justified by the
data. 
Of course, this is just a very simple example; but it is not grossly implausible
and describes a situation in which the ‘dependency or conflict’ depends precisely
on the ‘question of what service is performed’. Von Thadden’s overall conclusion
was that although there are reasons to believe that the Sarbanes-Oxley Act is a case
of premature and half-baked legislation, the argument against it in the 
present report needs to be strengthened.