6 reporting bias by his subordinates on earnings quality. If CFO’s can hardly be held accountable
for misreporting earnings, increasing the responsibility of the CEO becomes more beneficial, i.e., the CFO’s and CEO’s responsibility for the CFO’s bias in financial reports are substitutes
with regard to earnings quality. Finally, we address the role of compensation disclosure and transparency with regard to managerial incentives. We show that higher responsibility of the
CEO is beneficial if there is high uncertainty about the CFO’s incentives in the market. Such uncertainty potentially arises from limited transparency with regard to his financial incentives
and career opportunities. The remainder of the paper is organized as follows. Section 2 describes the model and the
assumptions underlying the analysis. Section 3 considers the optimal biasing strategies of the CFO and the CEO and Section 4 characterizes the rational expectations equilibrium. Section
5 presents comparative static results for earnings quality, and Section 6 concludes.
2. The Model
We consider a one-period reporting game with three strategic players: a risk-neutral Chief Financial Officer CFO and a risk-neutral Chief Executive Officer CEO of a firm and a
competitive, risk-neutral financial market. The firm’s fundamentals, v , are normally
distributed,
2
0,
σ
v
v N
. The timeline of the reporting game consists of three dates. First, the CFO of the firm engages
in earnings management and provides a potentially biased interim report to the CEO. Second, after observing the CFO’s report, the CEO may engage in earnings management and issues a
report to the financial market. Finally, the financial market uses the CEO’s report to adjust the market price of the firm.
Earnings Management The CFO privately observes the firm’s earnings,
e , that are correlated with the firm’s fundamentals,
ε = +
e v
, where
2
0,
ε
ε σ
N
is independently distributed noise. After observing the firm’s earnings, the CFO prepares an interim earnings report and submits the
report to the CEO. The interim report,
F
r , equals the sum of the firm’s earnings and a bias
F
b that the CFO introduces into his report,
F F
r e b
= + . The CFO bears a private cost when biasing the interim earnings report that represents expected
fines or reputational losses if earnings management is detected by the market. We assume that
7 the CFO’s cost of managing earnings is
2 2
2
γ ε
⋅ ⋅ +
F F
F
c b
, where
F
c is the publicly
known marginal cost of biasing and the normally distributed random variable
2
0,
ε σ
F F
N
captures idiosyncratic circumstances that influence the CFO’s misreporting costs, such as legal mis-reporting leeway Dye Sridhar 2004. While the CFO knows the realization of
ε
F
before choosing his bias level, it is unobservable by both the CEO and the market. Finally, the parameter
[0,1]
γ
∈
measures the responsibility of the CFO for the chosen reporting bias.
3
We explicitly allow for constellations in which the CFO is not fully accountable for the
consequences of his earnings management decision,
1
γ . There are several reasons why this
might be the case. For instance, managers are protected by limited liability and are only partially accountable for the distortion caused by misreporting. Furthermore, accounting
systems in general offer some leeway in interpreting standards. Lower values of γ might
represent accounting systems in which CFOs have sufficient discretion in choosing accounting rules. Then, accounting bias can be achieved within the legal scope defined by the standard
setters and there is only a small probability for fines or reputational losses.
4
Moreover, γ
might reflect problems with enforcement. Depending on the governance system, CFOs possibly get away cheaply. The probability of detection and the consequential fines are lower
than intended by regulation. In this sense, γ can be interpreted as an effective responsibility.
We capture increasing CFO’s responsibility by increasing γ . For example, the Sarbanes-
Oxley Act of 2002, Section 302, requires the CFO to certify the financial statements of the firm together with the CEO, thus increasing the personal accountability of the CFO.
The CEO privately observes the CFO’s earnings report, engages in earnings management and issues a final report to the financial market. The final report equals
= +
E F
E
r r
b , where
E
b is
the bias that the CEO introduces into the earnings report. We assume that the CEO’s cost of managing earnings is
2
2
κ ε
⋅ ⋅ +
+
E F
E E
c b
b
. Similarly to the CFO’s marginal cost of biasing,
E
c is publicly known and the normally distributed random variable
2
0,
ε σ
E E
N
reflects
3
The parameter restriction
[0,1] γ ∈
does not influence our results. Assuming that the CFO cannot be held fully accountable for his chosen reporting bias, we provide an intuitive reason for increasing the responsibility of the
CEO for the CFO’s misreporting. However, as we will show in Section 5, increasing the responsibility of the CEO does not necessarily improve earnings quality.
4
Ewert and Wagenhofer 2005 use a similar model and use this interpretation, referring to
γ
as tightness of the accounting system.
8 idiosyncratic circumstances that influence the CEO’s misreporting costs. The CEO privately
knows the realization of ε
E
. The variable
[0,1]
κ
∈
formalizes the idea that the CEO may be partly held accountable for the reporting bias introduced by the CFO.
5
As mentioned above, the Sarbanes-Oxley Act of 2002, Section 302, requires both the CFO and the CEO to certify
financial statements. It thus posits a joint responsibility of the CFO and the CEO for the appropriate presentation of the firm’s operations and financial conditions. This rule may
increase the CEO’s liability for the CFO’s reporting bias for several reasons. On the one hand, the enforcement agency may fail to disentangle the responsibility for the reporting bias and
partly attribute the decision to manage earnings to the CEO. On the other hand, the enforcement agency may deliberately postulate a shared responsibility of the CFO and CEO
for misreporting, since the CEO is the primarily responsible person for the activities of a firm. He decides on the adequacy of internal controls and should be held accountable for financial
misreporting by his subordinates, e.g., the CFO. CFO’s and CEO’s Utility
We assume that both the CFO’s and the CEO’s utilities depend on the market price of the firm. The interest in market price may arise, e.g., due to explicit compensation arrangements,
intention to sell equity holdings or repurchase shares of the firm, influencing the labor market’s perception of the managerial talent e.g., Feller and Schäfer 2017 or simply the desire to
increase the scope of control. The CFO’s and CEO’s utilities are linear in the market price
E
P r :
2 2
2
γ ε
= ⋅
− ⋅ ⋅
+
F F
F E
F F
c U
x P r
b
, and 1
2
2
κ ε
= ⋅
− ⋅ ⋅
+ +
E E
E E
F E
E
c U
x P r
b b
, 2
where
F
x
E
x is the CFO’s CEO’s incentive rate. We assume that while both the CEO and
the CFO know the actual incentive rate of the CFO, the market does not observe
F
x . Instead,
the market holds beliefs about the CFO’s interest in stock price and hence about his incentive to engage in earnings management. In particular, the market knows the a priori distribution of
5
Note that we do not impose any restrictions on the relation between
κ
and
γ
. For
1 κ γ
= =
, both parties are jointly responsible for the misreporting of the CFO. A centralization of responsibility is depicted by
1 κ =
and
γ =
when the CFO is not responsible for his biasing decision, but responsibility is fully allocated at the CEO level.
9
F
x which is normal with mean
µ
x
and variance
2
σ
x
. Note that the actual realization of
F
x might be negative. In fact, managers may have an interest in decreasing the stock price of
the firm if, e.g., they plan a management buyout, are about to repurchase shares or receive stock options. In addition, if the CFO has the ambition to become the CEO of the firm, he
might be interested in managing earning downward under the current CEO.
6
In our model, the variance
2
σ
x
of the incentive rate reflects the market’s uncertainty about the CFO’s incentives. In contrast, we assume that the CEO’s incentives,
E
x , are observable by all parties. This
assumption reflects that there is more public information about the CEO’s incentives than about the CFO’s motivation. Disclosure rules usually have a stronger focus on the CEO
compared to his subordinates. Furthermore, it is likely that CFOs are driven by career concerns to a greater extent than CEOs who already hold the top position within the firm. These implicit
incentives impede the ability of the market to infer the incentives of the CFO as opposed to the CEO.
Based on the CEO’s report, ,
E
r the market adjusts the price of the firm to the expected firm
value given all available information, [ | ]
=
E E
P r E v r
. The uncertainty about the CFO’s and CEO’s incentives to bias earnings reports prevents the market from perfectly inferring the bias
in the CEO’s report in equilibrium.
7
Figure 1 summarizes the timeline of the reporting game.
FIGURE 1 Timeline of the reporting game
6
If CEOs increasingly bear responsibility for firms’ financial reports, they must have accounting knowledge. This potentially improves CFOs’ claim to be a candidate for the CEO succession.
7
We combine two types of uncertainty in our model: the uncertainty about the CFO’s incentive weight on the firm’s stock price as in Fischer and Verrecchia 2000 and the uncertainty about the CEO’s and CFO’s cost of
manipulating earnings as in Dye and Sridhar 2004.
10 To obtain a tractable solution to our reporting game, we restrict the analysis to linear equilibria.
The equilibrium consists of the CFO’s and the CEO’s earnings reports and the market price reaction to the CEO’s report. We assume the following linear strategies:
ε
λ λ λ
λ ε =
+ ⋅ + ⋅ + ⋅
F e
x F
F
b e
x ,
3
ε
ε = + ⋅ + ⋅
+ ⋅
E r
F x
F E
b v
v r v x
v ,
4 α β
= + ⋅
E
P r
. 5
3. CFO’s and CEO’s Reporting Strategies