Description
Many in the nonprofit sector view accounting-based performance measures to be overly
influential and counterproductive in the evaluation of charities and their leaders. The contention
is that such measures are imperfect and often biased, leading to dysfunctional
rationing of fundraising and administrative infrastructure. To examine these concerns
and the broader question of nonprofit executive incentives, we develop a model of nonprofit
executives who are concerned with influencing external perceptions.
Career concerns and accounting performance measures
in nonpro?t organizations
Anil Arya
?
, Brian Mittendorf
Ohio State University, United States
a b s t r a c t
Many in the nonpro?t sector view accounting-based performance measures to be overly
in?uential and counterproductive in the evaluation of charities and their leaders. The con-
tention is that such measures are imperfect and often biased, leading to dysfunctional
rationing of fundraising and administrative infrastructure. To examine these concerns
and the broader question of nonpro?t executive incentives, we develop a model of non-
pro?t executives who are concerned with in?uencing external perceptions. In doing so,
we demonstrate that accounting-based performance measures alter executive incentives
in critical ways. In particular, disclosure of the functional classi?cation of nonpro?ts’
expenses can reduce incentives to overinvest in fundraising and restore investments in
programs; at the same time, it also comes with the potential downside of undermining
key investments in long-term infrastructure.
Ó 2014 Elsevier Ltd. All rights reserved.
Introduction
In his now-viral TED talk (‘‘The Way We Think About
Charity is Dead Wrong’’), Dan Pallotta provided a voice to
many in the nonpro?t sector who view the reliance on
accounting-based performance measures for non-govern-
mental organizations (NGOs) to be detrimental to their
overall effectiveness. The viewpoint, elaborated upon in
books by Pallotta (2010), Stern (2013), and others, is that
by developing an imperfect picture of how much of a non-
pro?t’s resources are devoted to its mission, the functional
classi?cation of expenses has become a key focus of outsid-
ers’ evaluation of NGOs and their executives; this focus, in
turn, incentivizes executives to cut key investments in
fundraising and advertising. Despite the almost universal
acceptance of this view and concomitant disdain for
accounting performance measures in the nonpro?t com-
munity (even spawning the ‘‘Overhead Myth’’ movement
to discredit reliance on accounting measures), there has
been little or no formal analysis of the viewpoint.
In this paper, we present a parsimonious model of non-
pro?t executives’ incentives in order to examine the conse-
quences of reliance on accounting-based performance
measures. The model captures two key elements: (i) to
the extent that nonpro?t executives are driven by extrinsic
motivation, such incentives are typically not due to explicit
pay arrangements but rather an incentive to in?uence
external perceptions; and (ii) the primary accounting per-
formance measure that is critical and unique to nonpro?ts
is the functional classi?cation of expenses, which provides
an assessment of the portion of expenses attributable to
achieving an organization’s mission, i.e., the program
expenses.
Our setting provides a formal analysis of actions non-
pro?t executives are incentivized to undertake when at
least part of their motivation is driven by a desire to boost
market perceptions of their effectiveness. Consistent with
intuition, we show that the more the market values the
ability to generate revenues and/or the more executives
can bene?t fromadministrative perquisites, the more focus
executives place on efforts to generate revenues. Similarly,
the more the market values the ability to ef?ciently focushttp://dx.doi.org/10.1016/j.aos.2014.10.002
0361-3682/Ó 2014 Elsevier Ltd. All rights reserved.
?
Corresponding author.
Accounting, Organizations and Society 40 (2015) 1–12
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
resources toward a mission and/or the less executives can
bene?t from perquisites, the more emphasis they place on
using (rather than generating) resources effectively.
Interestingly, we also show that the more ef?cient an orga-
nization is in using resources, the less incentive an execu-
tive has in trying to generate such resources. Among other
things, this suggests that if left unchecked, the natural
order of things may lead to inherently inef?cient organiza-
tions being the ones generating the bulk of donations.
The model also presents a different view of observed
executive pay-to-performance sensitivities. That is, though
empirical observation of variation in nonpro?t executive
pay moving in concert with accounting metrics is often
interpreted as evidence of contractual performance pay
arrangements, implicit market forces may also explain
such variability. In particular, despite the lack of explicit
incentive pay, our model demonstrates how external
demands for nonpro?t executives will generate pay that
varies predictably with accounting outcomes. This mar-
ket-driven view of pay-to-performance sensitivity also
offers additional empirical implications. For one, we show
that the more the market places a premium on revenue-
generating ability relative to the ability to devote resources
to the mission, the more sensitive pay will be to revenue
and the less sensitive it will be to the measure of program
expenses. This means that in markets characterized by
strong presence of for-pro?t entities, for whom revenue
is at a premium (e.g., education; health care), the more
(less) pay should appear sensitive to revenues (program
expenses). Additionally, we show that the more precise
the accounting cost allocation exercise, the less sensitive
pay will be to revenue and the more sensitive it will be
to program expenses. In other words, the greater the
uncertainty about the functional allocation of expenses
(driven by, say, the use of SOP 98-2 in allocating joint
costs), the greater the pay sensitivity will be to revenue
and the less weight the market will place on program
expenses.
Returning to the fundamental criticism of how account-
ing estimates can distort incentives, our model of career
concerns of nonpro?t executives demonstrates that the
incentive to in?uence external perceptions in itself can dis-
tort decisions away from what may be in the best interest
of donors. In particular, in order to boost perceptions of
revenue-generating abilities, a nonpro?t executive may
overinvest in fundraising efforts. This, in turn, can lead to
a reduced focus on improving efforts to ef?ciently direct
resources to the organization’s mission. Interestingly, the
functional classi?cation of expenses stipulated by account-
ing rules can help mitigate each of these concerns.
By tracking expenditures that go into the revenue-gen-
eration process, the accounting measure helps sift out
what revenues are attributable to executive ability and
what are simply due to high fundraising spending. This
undercuts the incentive for the executive to overinvest in
fundraising as a means of trying to posture to a market-
place that values revenue generation. The attempt to
separate program expenses as part of functional expense
classi?cation also gives the marketplace a second measure
on which to evaluate executives – their ef?ciency at putt-
ing resources to use. The newfound emphasis on program
expenses incentivizes executives to put more energy into
cutting administrative bloat and effectively directing
resources toward the mission.
However, this desire to cut administrative costs to sig-
nal greater ef?ciency to markets comes with a downside
in that it also encourages executives to divert resources
from potentially useful long-term infrastructure spending.
In other words, since accounting does not provide a natural
distinction between wasteful and useful administrative
spending, it incentivizes executives to ‘‘trim the fat’’ even
in cases where some administrative spending is critical.
That is, an additional implication of our model is that it
provides a theoretical justi?cation for the so-called non-
pro?t starvation cycle (Gregory & Howard, 2009), where
efforts to trim administrative costs undermine long-term
viability.
Taken together, our results suggest a nuanced balance
between viewing accounting measures with utter suspi-
cion or with unadulterated faith is warranted. The func-
tional classi?cation of expenses shifts the executive’s
emphasis away from excessive fundraising while also
simultaneously forcing him to focus on the use of
resources. The latter entails a trade-off. When excessive
administrative spending is the norm in the absence of
functional classi?cation, the tight discipline introduced
by accounting metrics proves bene?cial. On the other
hand, cost cutting can be excessive when it discourages
spending on valuable infrastructure. The paper’s proposi-
tions succinctly capture these economic forces.
This paper lies at the nexus of the literatures on career
concerns incentives and accounting performance measure-
ment for NGOs. In terms of the ?rst literature stream, the
seminal analysis of career concerns incentives is
Holmstrom (1982, 1999). This initial analysis of implicit
career incentives was expanded by Dewatripont, Jewitt,
and Tirole (1999a, 1999b) who generalize the model to
examine, among other things, the allocation of effort across
tasks, complementarities between skill and effort, and
alternative information structures. The career concerns
framework has also generated insights about managerial
investment (Holmstrom & Ricart i Costa, 1986), informa-
tion acquisition incentives (Milbourn, Shockley, & Thakor,
2001), team dynamics (Auriol, Friebel, & Pechlivanos,
2002), job design (Kaarboe & Olsen, 2006), disclosure reg-
ulation (Autrey, Dikolli, & Newman, 2007), and perfor-
mance measure design (Arya, Frimor, & Mittendorf, 2010;
Autrey, Dikolli, & Newman, 2010).
Theoretical inquiry of career concerns is motivated by
circumstances wherein employees face limited explicit
incentive compensation but rely on salaries determined
by labor markets. Driven by both donor and regulatory
restrictions, such limited incentive pay is commonplace
among NGOs, making incentives to in?uence external per-
ceptions all the more important to examine in nonpro?ts.
Surprisingly, there is a dearth of theoretical study of
incentives among NGOs and none (to our knowledge) on
the incentives of nonpro?t executives to in?uence market-
place perceptions. That said, there has been substantial
empirical study of NGO behavior and the role of accounting
measures therein. In terms of executive pay incentives,
Baber, Daniel, and Roberts (2002) and Sedatole, Swaney,
2 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
Yetman, and Yetman (2013) each demonstrate, to different
degrees, that key accounting measures (in particular, reve-
nues and program expenses) are associated with pay vari-
ations among nonpro?t executives. Frumkin and Keating
(2010) demonstrate that executive pay is more heavily
in?uenced by pay levels at peer institutions than it is by
performance. In a sense, our study brings together these
two seemingly disparate views by theoretically examining
how the in?uence of outside pay opportunities can alter
executive behavior and induce variation in the observed
sensitivities between pay and nonpro?t performance
despite the absence of explicit incentive pay. To this end,
our study also relates to and complements the work of
Krishnan, Yetman, and Yetman (2006) and Krishnan and
Yetman (2011), who demonstrate that nonpro?ts are
incentivized to manage accounting performance measures
in order to manage external perceptions. Our model incor-
porates the potential for both bias and noise in accounting
measures and demonstrates the consequences for execu-
tive incentives, pay, and efforts.
Model
An executive of a nonpro?t organization is charged with
generating revenues and putting those revenues to use in
ful?lling the organization’s mission. The revenues raised
by the organization, R, depend on the extent of ?nancial
resources devoted toward fundraising and the executive’s
skill and effort in generating funds. In particular,
R = a
???
F
_
÷ h
R
÷ e
R
÷j, where F denotes fundraising expen-
ditures, a denotes inherent fundraising ef?ciency, h
R
denotes the executive’s skill in raising funds, e
R
, e
R
P0,
denotes the executive’s effort in revenue generation, and
j re?ects all other revenue-related factors.
The organization’s net revenues, R ÷ F, are split
between use for administration and for implementing the
nonpro?t’s programs. The executive has a measure of con-
trol over this split. In particular, the amount directed
toward programs is P = b[R ÷ F] + h
P
+ e
P1
+ e
P2
+ w. Here,
b ÷ (0, 1) denotes the default fraction of resources devoted
to programs, h
P
re?ects the executive’s skill in ef?ciently
utilizing resources for programs, e
P1
and e
P2
, e
Pi
P0, re?ect
executive efforts to divert resources to programs, and w
re?ects other factors. Higher program skills (h
P
) can re?ect
abilities to ?nd ef?ciencies and reduce redundant expendi-
tures but can also re?ect a higher degree of integrity, i.e., a
lower willingness to divert resources for personal bene?t
(as in Shleifer & Wolfenzon, 2002). The two effort terms
re?ect that the executive can take actions to shift resources
toward programs, and such resource shifts could either
divert from wasteful administrative spending (e
P1
) or could
reduce useful administrative spending (e
P2
).
To elaborate, administrative expenses, denoted by A,
equal the residual expenditures R ÷ F ÷ P. These adminis-
trative expenses consist of both useful and wasteful com-
ponents – some administrative spending, denoted A
I
, is
infrastructure spending critical to mission success while
the remainder represents wasteful spending. Denoting
the maximum infrastructure needs by I, infrastructure
spending is simply A
I
= I ÷ e
P2
, re?ecting that the execu-
tive can take effort to boost program spending in the
short-run at the expense of long-run mission success by
neglecting key infrastructure. The remaining administra-
tive funds, R ÷ F ÷ P ÷ A
I
, though unhelpful to the mission,
may yield some perquisites for the executive; executive
perquisites equal c[R ÷ F ÷ P ÷ A
I
], c ÷ (0, 1).
The prior belief of the executive’s skills, held by all, is
that h
R
and h
P
are independent and normally distributed,
with mean
h and variance r
2
h
(with precision denoted
s
h
= 1=r
2
h
). Similarly, j is normally distributed with mean
j and variance r
2
j
(with precision denoted s
j
= 1=r
2
j
); for
w, the corresponding mean, variance, and precision values
are denoted
w, r
2
w
, and s
w
, respectively.
As is often the case in nonpro?t organizations, the exec-
utive is not provided explicit incentive pay. However, mar-
ket demands for his skill may necessitate adjustments to
his future pay. The executive’s future wage, denoted w, is
determined by a competitive labor market. The market
wage is w = kE h
R
¦ ¦ ÷ [1 ÷ k[E h
P
¦ ¦, where E ¦ ¦ denotes mar-
ket expectations, and k (1 ÷ k) re?ects the degree to which
the market cares about the executive’s ability to generate
revenues (ef?ciently use revenues on programs). While
we model the executive’s incentive as one of market-
driven pay, this can be more broadly thought of as an
incentive to boost market perceptions – under this inter-
pretation, boosting market perceptions may be valued by
the executive even if it does not come with higher pay
since it can come with greater visibility or higher-pro?le
positions. Under either interpretation, the market’s expec-
tations rely on (observable) revenues and the functional
classi?cation of expenses. The classi?cation splits total
expenses into its three components: fundraising expenses
(F), program expenses (P), and administrative expenses
(A = R ÷ F ÷ P).
In choosing his efforts, the executive balances the
potential for future compensation and perquisites with
his personal cost of efforts, k
R
e
2
R
, k
P1
e
2
P1
, and k
P2
e
2
P2
. Given
this basic structure, we investigate how career concerns
affect nonpro?t executives’ efforts, and how these incen-
tives are in?uenced by the functional classi?cation of
expenses. Fig. 1 summarizes the sequence of events.
Results
First-best benchmark
Before deriving the equilibrium that arises when non-
pro?t executives are interested in external perceptions
and internal perks, we begin with a ?rst-best benchmark.
That is, what is the outcome if the executive is solely
focused on mission impact? In this setting, the entity’s
mission is enhanced both through current program expen-
ditures (P) and investments to boost the long-term impact
of such expenditures (A
I
). In particular, denote the overall
welfare, W, as the expected value of mission-based expen-
ditures (both programs and infrastructure) less effort
costs:
W = E¦P ÷ A
I
¦ ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
:
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 3
If the executive is concerned only with welfare, he
solves the following program in choosing fundraising
expenditures and efforts:
Max
F;e
R
;e
P1
;e
P2
E¦P ÷ A
I
¦ ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
= E¦b[a
???
F
_
÷ h
R
÷ e
R
÷j ÷ F[ ÷ h
P
÷ e
P1
÷ w ÷ I¦
÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
: (1)
Solving (1) yields the ?rst-best benchmark, as summarized
in Lemma 1. (All proofs are provided in the appendix.)
Lemma 1. The welfare-maximizing equilibrium entails:
F
+
=
a
2
4
; e
+
R
=
b
2k
R
; e
+
P1
=
1
2k
P1
; e
+
P2
= 0; and
W
+
= [1 ÷ b[
h ÷ b
j ÷
w ÷ I ÷
1
4
a
2
b ÷
b
2
k
R
÷
1
k
P1
_ _
:
The ?rst-best outcome presented in the lemma offers
some intuitive properties. First, the level of fundraising
undertaken is increasing in the organization’s inherent
fundraising ef?ciency, a. The effort undertaken on boosting
revenues is increasing in the degree to which the organiza-
tion ef?ciently uses funds for its mission, b, and decreasing
in k
R
, the cost of revenue effort. Finally, the executive
undertakes efforts to trim wasteful spending but does not
trim infrastructure spending (e
+
P1
> 0 and e
+
P2
= 0). With
this intuitive benchmark in place for comparison, we next
consider the outcome when nonpro?t executives are
in?uenced by external perceptions.
Equilibrium market wages
While there is little doubt that nonpro?t executives are
motivated by a desire to improve society, they are also not
immune to the pressures to boost perceptions of their abil-
ity and performance. Such career concerns may even be
more pronounced in nonpro?t organizations than their
for-pro?t counterparts due to the notable absence of expli-
cit incentive pay arrangements. In such cases, a nonpro?t
executive’s actions are not solely determined by overall
welfare but also affected by a desire to in?uence external
perceptions and future pay potential. Following this frame-
work, we work backwards in the game to determine the
equilibrium, starting with a derivation of how the market
wage is determined.
The wage depends on market beliefs about the
executive’s revenue-generation skill as well as his skill in
ef?ciently utilizing the raised resources. Market beliefs,
in turn, can be conditioned on revenues (R) and, under
functional classi?cation, on the decomposition of expenses
into fundraising (F), program (P), and administration (A).
Since A simply equals R ÷ F ÷ P, it follows that market
beliefs are effectively based on R, F, and P. Conditioned
on this information set, and relegating technical details
to the appendix, the market’s conjecture about the execu-
tive’s revenue-skill equals:
E¦h
R
R; F; P¦ [ =
s
h
s
h
÷s
j
_ _
h ÷
s
j
s
h
÷s
j
_ _
R ÷a
???
F
_
÷ e
c
R
÷ j
_ _
_ _
:
(2)
In (2), e
c
R
represents the market’s conjecture of the execu-
tive’s revenue effort. The reason conjectures come into
play is that the market wage is not a reward for past efforts
or successes but instead a market-based inference of skill
parameters from observation of past performance. Thus,
in making inferences about h
R
, the market seeks to adjust
performance to remove parts attributable to past effort
and other non-persistent components. The market does
not observe effort, but instead relies on its belief about
what effort was – in equilibrium, such conjectures will
be correct.
In particular, the inference of revenue-skill in (2)
re?ects a weighted average of the market’s prior belief
and a signal of ability inferred from observed revenues.
This signal equals observed revenues less the portion
attributable to fundraising expenses, conjectured effort to
boost revenues, and other (transitory) factors. The weights
placed on the two components depend on the precision of
the prior belief (s
h
) relative to that of the observed
revenue-based signal (s
j
).
In a similar fashion, the market’s conjecture of skill in
ef?ciently directing resources towards programs is:
E¦h
P
R; F; P¦ [ =
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
P ÷b[R÷F[ ÷e
c
P1
÷e
c
P2
÷
w
_ _
_ _
:
(3)
In (3), e
c
P1
and e
c
P2
represent the external market’s conjec-
ture of efforts to divert resources from wasteful and useful
administrative spending, respectively. The market’s expec-
tation of skill in ef?ciently using resources is again a
weighted average of its prior and a signal of such ability
derived from observed program expenses. In this case,
the program expenses are adjusted for net revenues, con-
jectured cost-cutting efforts, and other transitory factors.
The weights depend on the relative precision of the prior
(s
h
) and the accounting-based estimate inherent in
functional classi?cation (s
w
).
Given (2) and (3), the executive’s market wage, after the
market has observed revenue and the functional
classi?cation of expenses, can be expressed as:
The
executive
exerts
revenue
effort e
R
and
fundraising
costs F.
Revenues are
realized.
The executive exerts
program efforts
e
P1
and e
P2
.
Functional expenses
are disclosed. The
labor market
determines the
executive's
continuation wage, w.
Fig. 1. Timing of events.
4 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
w=k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
R÷a
???
F
_
÷e
c
R
÷
j
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
P ÷b[R÷F[ ÷e
c
P1
÷e
c
P2
÷
w
_ _
_ _
:
(4)
The two parts of the wage are derived from the infer-
ences of ability to generate revenue weighted by k, and
the inference of ability to effectively deploy funds toward
the mission weighted by 1 ÷ k. Delaying consideration of
the executive’s effort choices for now, consider the impli-
cations of equilibrium market wage for observed pay-to-
performance sensitivities.
Take ?rst the relation between revenue and pay. From
the ?rst term in (4), higher revenues are indicative of
higher fundraising skill and are rewarded as such. This is
consistent with the empirical evidence in Baber et al.
(2002) and Sedatole et al. (2013) who each show a positive
relationship between pay and overall revenue for nonpro?t
executives. In this case, however, the relationship is not
evidence of contractual performance pay, but instead dri-
ven by market forces.
From the second term in (4), a second feature is also in
play: higher revenues also indicate that a given level of
program expenses is attributable to scale rather than exec-
utive skill at ef?ciently employing resources. Roughly, this
is consistent with the notion of a program expense ratio
(here it is difference rather than ratio), re?ecting that
higher revenues alone demand higher program expenses.
As such, to the degree the external market cares about
and can effectively isolate skills at ef?ciently using
resources, higher revenues can actually reduce pay.
These two forces mean that pay-to-performance sensi-
tivity of revenues is increasing in the extent to which the
market values the executive’s ability to generate revenue
(k) and decreasing in the informativeness of the functional
classi?cation of expenses (s
w
).
Proposition 1. Sensitivity of pay to revenue, @w/@R, is:
(i) increasing in labor market demand for revenue skills
(k); and
(ii) decreasing in the informativeness of functional expense
classi?cation (s
w
).
Part (i) of the proposition provides a notable empirical
implication. For some nonpro?ts, the ability to generate
revenues is far more important than their ef?cient use
when it comes to outside market wages. For example, it is
often discussed that the presence of for-pro?t counterparts
in health-care and education creates a greater outside
demand for revenue generation and less demand for ability
to achieve missions (i.e., higher k). This suggests that
such nonpro?ts should empirically see a higher pay-to-
performance sensitivity when it comes to revenues, this
despite the fact that there no explicit incentive contracts
in play.
Part (ii) of the proposition demonstrates the subtleties of
inferring the usefulness of accounting estimates based on
howthey are used. Since higher revenues indicate that pro-
gram expenses are less attributable to executive skill in
managing resources, the more the market relies on program
expenses to infer such skills, the less it will reward higher
revenues. This means that the more effective an accounting
estimate of cost-management (program expenses) is, the
less the market relies on another (revenues). So, a reduced
reliance on one accounting ?gure may actually indicate
more, not less, effective inference of skill.
Now consider how pay varies with observed estimates
of program expenses. Again, using (4) reveals the equilib-
rium pay-to-performance sensitivity.
Proposition 2. Sensitivity of pay to program expenses,
@w/@P, is:
(i) decreasing in labor market demand for revenue skills
(k); and
(ii) increasing in the informativeness of functional expense
classi?cation (s
w
).
The predictions again mirror empirical data but provide
a different interpretation due to the absence of incentive
pay: higher program expenses are associated with higher
pay. However, this relation is muted the more the market
cares about revenue-generation skill (k). This may provide
some explanation for the varied degrees to which Baber
et al. (2002) and Sedatole et al. (2013) observe market
rewards for program expenses. While higher program
expenses are rewarded, the degree of this reward varies
predictably with both market priorities (k) and the non-
pro?t’s accounting system (s
w
). Further, in contrast to
Proposition 1(II), Proposition 2(II) shows that more precise
accounting estimates of ef?ciency in utilizing resources
implies a more positive pay-to-performance relation when
it comes to program expenses.
Career concerns equilibrium
Noting the in?uence of outcomes on market wages as
derived above, and taking market conjectures of effort as
given, the executive has incentives to incur efforts in
order to in?uence perceptions. That is, knowing the out-
comes will be used to infer ability, the executive has
incentive to in?uence such outcomes. Though any such
efforts will be correctly inferred in equilibrium, the exec-
utive nonetheless cannot resist the temptation. In partic-
ular, working backwards in the game, the executive
chooses cost-cutting efforts to maximize expected wage,
less effort cost adjusted for perks from administrative
expenditures, as in (5):
Max
e
P1
;e
P2
k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
R÷a
???
F
_
÷e
c
R
÷
j
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
h÷e
P1
÷e
P2
÷e
c
P1
÷e
c
P2
_ _
_ _
÷k
R
e
2
R
÷k
P1
e
2
P1
÷k
P2
e
2
P2
÷c[(1÷b)(R÷F) ÷
h÷e
P1
÷
w÷I[:
(5)
From(5), the executive’s effort choices weighs three fac-
tors: (i) higher efforts increase program expenses and thus
perceptions of ef?ciency in utilizing resources; (ii) higher
efforts are personally costly; and (iii) higher effort to reduce
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 5
wasteful spending (e
P1
) also reduces perks. Taken together,
these features lead to the executive’s cost-reduction effort
choices. As it turns out, the choices that solve (5) are free
of the market’s conjectures, e
c
Pi
, ensuring they are also the
equilibrium choices. The equilibrium efforts aimed at
reducing administrative costs are:
e
P1
=
^
e
P1
= Max
[1 ÷ k[s
w
2k
P1
[s
h
÷s
w
[
÷
c
2k
P1
; 0
_ _
and
e
P2
=
^
e
P2
=
[1 ÷ k[s
w
2k
P2
[s
h
÷s
w
[
: (6)
The following proposition summarizes some key charac-
teristics of the equilibrium efforts to reduce administrative
spending.
Proposition 3. In the presence of career concerns,
(i) the (non-zero) effort to reduce administrative waste
(
^
e
P1
) is decreasing in both labor market demand for
revenue skills (k) and executive perk consumption (c).
(ii) the effort to reduce infrastructure spending (
^
e
P2
) is
decreasing in labor market demand for revenue skills
(k) but unaffected by executive perk consumption (c).
Intuitively, the proposition demonstrates that the more
the market cares about revenues (and less it cares about
cost-ef?ciency), the less effort the executive exerts to cut
administrative spending. It also shows that the more the
executive cares about consuming perks now rather than
in?uencing pay in the future, the less he will exert efforts
to reduce wasteful administrative spending.
Using these equilibrium efforts, we can now work back-
ward to determine the executive’s choice of fundraising
expenditure and revenue effort. The executive makes this
choice to maximize expected wage less effort cost adjusted
for perk consumption, as in (7):
Max
e
R
; F
k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
h÷e
R
÷e
c
R
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
h÷
^
e
P1
÷
^
e
P2
÷e
c
Pi
÷e
c
P2
_ _
_ _
÷k
R
e
2
R
÷k
P1
^
e
2
P1
÷k
P2
^
e
2
P2
÷c[(1÷b)(a
???
F
_
÷
h÷e
R
÷
j÷F)
÷
h÷
^
e
P1
÷
w÷I[: (7)
From (7), the executive’s revenue effort balances three
factors: (i) higher effort increases revenues and thus mar-
ket inferences of revenue ability; (ii) higher effort is costly;
and (iii) higher effort generates more revenues, a portion of
which are consumed as perks. While fundraising expendi-
tures, F, do not in?uence market perceptions (the market
observes them so removes them in the inference process),
they can in?uence the level of funds available to be con-
sumed as perks, making them another nontrivial choice.
These factors, taken together, determine the optimal reve-
nue and fundraising choices, each of which are indepen-
dent of market conjectures of effort and thus comprise
the equilibrium outcomes. The solutions to (7) are:
e
R
=
^
e
R
=
ks
j
2k
R
[s
h
÷s
j
[
÷
c[1 ÷ b[
2k
R
and F =
^
F =
a
2
4
(8)
The critical properties of revenue effort are presented in
Proposition 4.
Proposition 4. In the presence of career concerns, revenue
effort (^e
R
) is:
(i) increasing in labor market demand for revenue skills
(k);
(ii) increasing in executive perk consumption (c); and
(iii) decreasing in the organization’s inherent program ef?-
ciency (b).
Intuitively, (i) demonstrates that the more the market
cares about the executive’s ability to generate revenue,
the more incentive he has to exert revenue effort to (try
to) in?uence the market’s perception. Interestingly, and
in contrast to the case of cost-cutting efforts, (ii) demon-
strates that the desire to consume perks now also pro-
vides revenue effort incentives. The reason is that if the
executive gets to make use of a portion of revenues to
consume as perks (e.g., nicer of?ce, ?rst-class travel,
etc.), that alone provides him incentive to generate reve-
nues. The underlying force in (ii), however, is mitigated
by the organization’s inherent ef?ciency in using
resources (e.g., board oversight, restrictions on spending,
etc.), as con?rmed in (iii). That is, if the organization is
particularly ef?cient in using resources, the executive is
aware that fewer resources will become perks, and his
incentives to generate resources are muted. This compar-
ative static suggests that, absent intervention, organiza-
tions which are particularly poor at using resources may
ironically also be the ones that have executives who do
more to generate such revenues.
The incentives that underlie the career concerns equi-
librium derived herein are the desire to in?uence pay in
the future and the desire to consume perks in the present.
The desire to in?uence pay is best re?ected by k which cap-
tures the relative importance the market places on ability
to generate resources versus utilizing the resources,
whereas the incentive for current perquisite consumption
is captured by c. Fig. 2 demonstrates the consequences of
each incentive on equilibrium efforts.
The comparative statics re?ected in the left panel of
Fig. 2 point to a reasonable empirical test about different
types of nonpro?ts. Among nonpro?ts for whom executive
outside opportunities stress revenues (e.g., health care),
one would expect greater (lower) effort by executives in
generating (ef?ciently using) resources. The changed
incentives to cut costs manifests itself on two dimensions
in that an incentive to ef?ciently use resources also entails
some myopic cost-savings that reduce infrastructure.
The comparative statics in the right panel of Fig. 2 pro-
vide empirical implications about different types of execu-
tives. If one views the importance of current perquisites
relative to future pay as proxied by an executive’s age
(those early in their career are more concerned with future
pay), we would expect to see more (less) experienced
executives to be focused on revenue generation (resource
utilization). Interestingly, though, the temptation to cut
infrastructure spending to appear more ef?cient is
universal, i.e., applies equally to all stages of one’s career.
6 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
Using the equilibrium values in (6) and (8) together, the
career concerns equilibrium and associated welfare are
summarized in the following proposition.
Proposition 5. In the presence of career concerns, the
equilibrium with functional classi?cation of expenses entails:
^
F =
a
2
4
;
^
e
R
=
ks
j
2k
R
[s
h
÷s
j
[
÷
c[1 ÷ b[
2k
R
;
^
e
P1
=
[1÷k[s
w
2k
P1
[s
h
÷s
w
[
÷
c
2k
P1
if c<
[1÷k[s
w
s
h
÷s
w
; and
^
e
P1
=0; otherwise;
^
e
P2
=
[1÷k[s
w
2k
P2
[s
h
÷s
w
[
; and
^
W = W
+
÷
1
4
1
k
R
_
c(1 ÷ b) ÷ b ÷
ks
j
s
h
÷s
j
_ _
2
÷
1
k
P1
c ÷
s
h
÷ ks
w
s
h
÷s
w
_ _
2
÷
1
k
P2
[1 ÷ k[s
w
s
h
÷s
w
_ _
2
_
if
c <
[1 ÷ k[s
w
s
h
÷s
w
; and
^
W = W
+
÷
1
4
1
k
R
_
c(1 ÷ b) ÷ b ÷
ks
j
s
h
÷s
j
_ _
2
÷
1
k
P1
÷
1
k
P2
[1 ÷ k[s
w
s
h
÷s
w
_ _
2
_
; otherwise:
We conclude the results by examining if and how career
concern incentives are in?uenced by disclosure of the
functional classi?cation of expenses.
The role of functional cost classi?cation
There is no doubt that nonpro?t executives are driven
by an inherent desire to make a lasting impact on society.
That said, there is also little doubt that their incentives
move beyond intrinsic desires to extrinsic factors. Given
the notable absence of incentive pay in nonpro?ts, the
career concerns equilibrium derived above seeks to cap-
ture the two key extrinsic incentives faced by nonpro?t
executives: the desire to in?uence perceptions and future
pay, and the bene?t of current perquisite consumption.
It is these incentives many have in mind when they
question the ef?cacy of accounting rules that govern nonp-
ro?ts. The accounting rules stipulate the functional classi-
?cation of expenses between programs, fundraising, and
administration. The commonly expressed concern with
the accounting rules is that the functional classi?cation
comes with error (here, re?ected by r
2
w
> 0) and can be
biased (here, re?ected by
w –0 ). Further, the accounting
rules pool together all administrative costs, failing to
re?ect the distinction between those that are helpful (here,
A
I
) and those that are wasteful (here, A ÷ A
I
). These imper-
fections are believed to distort incentives of executives to
the detriment of the nonpro?ts’ missions. It is this question
we revisit in light of the career concerns equilibrium
derived herein. First, however, we derive the alternative
equilibrium, one arising in the absence of market reliance
on accounting cost classi?cation.
Relegating the details to the appendix, the next lemma
demonstrates the outcome without cost allocation.
Lemma 2. In the presence of career concerns, the equilibrium
without functional classi?cation of expenses entails:
~
F =
a
2
÷
aks
j
2c(1 ÷ b)(s
h
÷s
j
)
_ _
2
;
~
e
R
=
ks
j
2k
R
[s
h
÷s
j
[
÷
c[1 ÷ b[
2k
R
;
~
e
P1
=
~
e
P2
= 0; and
~
W = W
+
÷
1
4
1
k
R
_
c(1 ÷ b) ÷ b ÷
ks
j
s
h
÷s
j
_ _
2
÷
1
k
P1
÷
b
[c(1 ÷ b)[
2
aks
j
s
h
÷s
j
_ _
2
_
:
The ?rst thing to note in comparing Proposition 5 and
Lemma 2 is that the functional classi?cation of expenses
under accounting has no effect on revenue effort incen-
tives. Since the market relies only on observed revenues
(and not the estimate of their split) in inferring revenue-
generation skill, the effort levels are unaffected by either
the presence or the precision of the accounting estimates.
Fig. 2. In?uence of labor market and executive priorities on efforts.
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 7
The presence of the functional classi?cation, however,
does impact resource utilization effort. Since the market
gets no indication of resource utilization in the absence
of accounting estimates, market inferences do not generate
incentives to cut costs so as to boost program spending; in
fact, the extrinsic incentive is the opposite, since cost cuts
only reduce perks.
The ?nal choice that is at the executive’s discretion is
how much to expend on fundraising. Recall that with func-
tional expense classi?cation, the market observes fundrais-
ing expenditures and can discount them appropriately
when evaluating the realized revenues. Absent functional
classi?cation, however, the market only knows the reve-
nues generated and not how much of the nonpro?t’s own
resources were used in generating such resources. This cre-
ates an incentive to expend more on fundraising in order to
in?uence market perceptions of revenue-generation skill.
The market correctly anticipates such expenditures in
equilibrium, so it is, in a sense, an exercise in futility. These
ways in which accounting cost allocations can alter incen-
tives and their net effect on welfare are summarized in the
next proposition.
Proposition 6. Functional classi?cation of expenses leads to:
(i) higher helpful program (cost-cutting) effort e
P1
if and
only if c <
[1÷k[s
w
s
h
÷s
w
;
(ii) higher myopic program (cost-cutting) effort e
P2
;
(iii) equal revenue effort e
R
; and
(iv) lower fundraising expenditure F.
The proposition con?rms the potential upsides of func-
tional classi?cation: by shining a light on how money is
spent, the cost allocation exercise convinces executives to
shift a focus away from fundraising and wasteful adminis-
trative spending in favor of cost cutting. At the same time,
it also demonstrates the potential downsides of functional
expense classi?cation: by bringing attention to spending,
cost allocation lowers spending on key infrastructure since
it gets classi?ed as ‘‘overhead.’’ Jointly considering these
tradeoffs culminates in the following welfare comparison.
Proposition 7. Functional classi?cation of expenses leads to
higher welfare if and only if k
P2
> k
/
, where:
k
/
=
(1 ÷ k)s
w
s
h
÷s
w
_ _
2
b
[c(1 ÷ b)[
2
aks
j
s
h
÷s
j
_ _
2
_
÷
1
k
P1
2 ÷
(1 ÷ k)s
w
s
h
÷s
w
÷c
_ _
(1 ÷ k)s
w
s
h
÷s
w
÷c
_ __
÷1
if c <
[1 ÷ k[s
w
s
h
÷s
w
; and
k
/
=
(1 ÷ k)s
w
s
h
÷s
w
_ _
2
[c(1 ÷ b)[
2
b
_ _
s
h
÷s
j
aks
j
_ _
2
; otherwise:
As the proposition shows, the net welfare consequence
of functional classi?cation hinges on k
P2
, which re?ects the
executive’s relative dif?culty of cutting infrastructure
spending. The reason it proves critical is that if it is easy
to meet cost-cutting targets by reducing infrastructure
(k
P2
is small), functional classi?cation’s tendency to initiate
the ‘‘nonpro?t starvation cycle’’ is paramount. In that case,
since the program expense measure encourages the wrong
type of cost cutting, efforts to measure it are counterpro-
ductive. This characterization also means that concerns
about functional classi?cation undermining incentives to
invest in fundraising are less pressing. The reason is that
absent functional classi?cation, the incentive to invest in
fundraising is excessive – the executive cannot resist the
temptation to overinvest in fundraising in order to boost
revenues and the resulting impression of his ability to gen-
erate revenues. Functional classi?cation of expenses turns
off this lever for the executive, promoting a greater atten-
tion to ef?ciency. This feature, as well as the ability of func-
tional classi?cation to promote bene?cial cost-cutting
initiatives (reduction of administrative bloat) means that
despite its propensity to reduce useful infrastructure
spending, it can often still be bene?cial in toto.
Before concluding, we wish to acknowledge and discuss
the limitations of the career concerns framework
employed herein. While the standard career concerns
model provides a tractable and parsimonious way to
address incentives for multiple elements of effort and skill
among employees seeking to in?uence external percep-
tions, its linear framework also precludes examination of
interdependencies among such efforts and skills. Such lim-
itations mean that the present analysis excludes consider-
ation of circumstances where an executive’s diligent
efforts to ef?ciently use resources makes his task of raising
funds easier or vice-versa. To get a ?avor for how such
interdependencies between spending and fundraising can
alter the results, consider the paper’s analysis as before,
except that efforts to reduce spending and those to boost
fundraising are synergistic in that the cost of executives
efforts are reduced by ge
R
e
P1
, with g capturing the degree
of synergy. Further, in order to focus on how effort comple-
mentarities can affect effort choices, we set a = c = 0, i.e.,
fundraising is held ?xed and perquisite consumption by
the CEO is a non issue, and normalize b = 1. Finally, to
ensure interior solutions we presume 0 < g < 2
????????????
k
P1
k
R
_
,
with the limiting lower bound representing the original
setting. In this appended setting, the following proposition
shows the consequence of functional classi?cation on
career concerns efforts (closed form expressions are
presented in the appendix).
Proposition 8. With effort complementarity, functional clas-
si?cation of expenses yields:
(i) higher effort on all fronts:
^
e
R
(g) >
~
e
R
(g) > 0;
^
e
P1
(g) >
~
e
P1
(g) > 0;
and
^
e
P2
(g) >
~
e
P2
(g) = 0:
(ii) higher welfare if and only if k
P2
> k
/
(g), where
dk
/
(g)
dg
< 0.
Note from Proposition 8(i) that effort complementarity
introduces two additional features: efforts to cut wasteful
spending are positive even absent functional expenses
(~e
P1
(g) > 0), and functional expense classi?cation also
increases efforts to raise revenues (^e
R
(g) > ~e
R
(g)). Despite
8 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
the differences, the primary message that functional classi-
?cation of expenses increases efforts to in?uence percep-
tions, though such efforts can be either productive or
harmful, remains. As Proposition 8(ii) shows, this again
translates into a straightforward welfare comparison in
that the less the concern that the executive will reduce
infrastructure spending, the more attractive is functional
classi?cation from a welfare perspective. Further, the
greater the complementarity between these efforts, the
less prominent reduction in infrastructure spending
becomes and, thus, the more attractive functional classi?-
cation of expenses, as evidenced by
dk
/
(g)
dg
< 0.
The broader point here is not that introducing comple-
mentarities necessarily favors functional classi?cation of
expenses, but rather that it can add subtle additional con-
siderations. Such subtleties can be compounded if one con-
siders multi-period interactions. In that case, it seems
reasonable to presume that diligent cost-cutting in one year
can make for an easier fundraising pitch in a future year, i.e.,
a higher e
P1
can promote a higher future a. Alternatively,
one can argue that greater administrative spending in one
year can boost subsequent years’ ability to effectively use
resources, i.e., a lower e
P2
can promote a higher future b.
These considerations suggest that depending on the nature
of the interactions some efforts may prove more important
to welfare in the long-run than others. While our single-
period analysis does not fully capture these forces, perhaps
it gives a window into what these forces would mean. After
all, we showthat the more prominent career incentives are,
the greater the efforts to cut costs (both e
P1
and e
P2
are
higher under functional classi?cation). So, if cost-cutting
this year can boost fundraising next year, it points to
another upside of functional cost allocation; on the other
hand, if reduction in administrative infrastructure this year
proves particularly harmful in the long-run, our results sug-
gest a poignant downside of cost allocation.
Conclusion
Though nonpro?t executives are typically viewed as
driven more by intrinsic motivation than their for-pro?t
counterparts, there is little doubt that they are also moti-
vated to a degree by external perceptions and pay. For this
reason, many in the nonpro?t sector have condemned the
functional expense classi?cation embedded in accounting
rules for unduly creating myopic incentives for executive
behavior. The contention is that by creating imperfect
and often biased measures of how effectively nonpro?ts
have devoted resources to their mission, accounting state-
ments have led executives to engage in a ‘‘nonpro?t starva-
tion cycle’’, reducing investment in advertising and
infrastructure in order to in?uence external perceptions
of ef?ciency (see, e.g., Gregory & Howard, 2009).
This paper provides a formal model of incentives of
nonpro?t executives when they are concerned with exter-
nal perceptions. The model permits an examination of both
executive incentives to in?uence perceptions and the role
of accounting estimates therein. We show that even with
their inherent bias and noise, accounting estimates can
prove in?uential in executive pay and actions. Also
consistent with accounting detractors, we show that
functional classi?cation of expenses shifts resources away
from fundraising and administration toward immediate
program-based expenditures. Despite the emphasis shift
and the downsides it can bring, we also show that these
measures and the incentives they induce can improve
resource allocation and overall welfare.
Acknowledgements
We thank Ramji Balakrishnan, John Christensen, Peter
Christensen, John Fellingham, Hans Frimor, Jon Glover,
Robert Goex, Lisa Koonce, Eva Labro, Dan Pallotta, Christine
Petrovits, Jonathan Ross, Naomi Rothenberg, Doug Schroe-
der, Karen Sedatole, Jack Stecher, Austin Sudbury, Rick
Young, Chris Chapman (Editor) and two anonymous refer-
ees, and workshop participants in Arizona State University,
Binghamton University, Carnegie Mellon University,
University of Colorado, University of Iowa, Ohio State
University, and EIASM workshop on Accounting and
Economics. Anil Arya gratefully acknowledges assistance
from the John J. Gerlach Chair.
Appendix A
Proof of Lemma 1. The welfare measure is:
W = E¦P ÷ A
I
¦ ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
= E¦b[R ÷ F[ ÷ h
P
÷ e
P1
÷ e
P2
÷ w ÷ I ÷ e
P2
¦
÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
= E¦b[a
???
F
_
÷ h
R
÷ e
R
÷j ÷ F[ ÷ h
P
÷ e
P1
÷ w ÷ I¦
÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
= b[a
???
F
_
÷
h ÷ e
R
÷
j ÷ F[ ÷
h ÷ e
P1
÷
w
÷ I ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
: (A1)
The derivative of W with respect to F, e
R
, e
P1
, and e
P2
,
respectively, are:
@W
@F
= b
a
2
???
F
_ ÷1
_ _
;
@W
@e
R
= b ÷2k
R
e
R
;
@W
@e
P1
= 1 ÷2k
P1
e
P1
; and
@W
@e
P2
= ÷2k
P2
e
P2
: (A2)
From the last derivative in (A2), e
+
P2
= 0. Setting each of the
?rst three derivatives in (A2) equal to zero, and solving the
equations for F, e
R
, and e
P1
, yields the solution in Lemma 1.
Substituting this solution in (A1) yields the corresponding
welfare value. h
Proof of Proposition 1. 1We ?rst formally derive the
expression for wages presented in (4). Since A = R ÷ F ÷ P,
the information content associated with the market learn-
ing (R, F, P, A) is equivalent to it learning (R, F, P) and, thus,
we use this smaller set in characterizing the market’s
updated beliefs of the executive’s two skill dimensions
(and, thus, his wage). To conduct this updating, we make
use of a standard result (e.g., Greene, 1997, p. 90): if
X =
X
1
X
2
_ _
is multivariate normal with mean l =
l
1
l
2
_ _
,
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 9
covariance matrix R =
R
11
R
12
R
21
R
22
_ _
, and R
÷1
22
is the inverse
of R
22
(the pseudo inverse if R
22
is singular), then the con-
ditional mean of X
1
, given X
2
= x
2
, is:
E¦X
1
x
2
[ ¦ = l
1
÷ R
12
R
÷1
22
(x
2
÷l
2
): (A3)
With X
1
=
h
R
h
P
_ _
and X
2
=
R
F
P
_
_
_
_
, l
1
=
h
h
_ _
,
l
2
=
a
???
F
_
÷
h ÷ e
c
R
÷
j
F
b[a
???
F
_
÷
h ÷ e
c
R
÷
j ÷ F[ ÷
h ÷ e
c
P1
÷ e
c
P2
÷
w
_
_
_
_
,
R
12
=
r
2
h
0 br
2
h
0 0 r
2
h
_ _
, and
R
22
=
r
2
h
÷r
2
j
0 b(r
2
h
÷r
2
j
)
0 0 0
b(r
2
h
÷r
2
j
) 0 b
2
(r
2
h
÷r
2
j
) ÷r
2
h
÷r
2
w
_
_
_
_
; here, e
c
R
,
e
c
P1
, and e
c
P2
denote the market’s conjecture of the execu-
tive’s efforts. Substituting this in (A3) yields:
E¦X
1
x
2
[ ¦ =
E¦h
R
R; F; P¦ [
E¦h
P
R; F; P¦ [
_ _
=
r
2
j
r
2
h
÷r
2
j
_ _
h÷
r
2
h
r
2
h
÷r
2
j
_ _
R÷a
???
F
_
÷e
c
R
÷
j
_ _
r
2
w
r
2
h
÷r
2
w
_ _
h÷
r
2
h
r
2
h
÷r
2
w
_ _
P÷b(R÷F) ÷e
c
P1
÷e
c
P2
÷
w
_ _
_
¸
¸
_
_
¸
¸
_
:
(A4)
Using (A4), the wage is w = kE¦h
R
[R; F; P¦ ÷ [1 ÷ k[E¦h
P
[R; F; P¦. Substituting s
i
= 1=r
2
i
, i = h, j, w, this wage can be
expressed as:
w=k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
R÷a
???
F
_
÷e
c
R
÷ j
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
P÷b(R÷F) ÷e
c
P1
÷e
c
P2
÷
w
_ _
_ _
:
(A5)
From (A5), the sensitivity of pay to revenue is:
@w
@R
=
ks
j
s
h
÷s
j
÷
[1 ÷ k[bs
w
s
h
÷s
w
: (A6)
The proposition then follows by noting the sign of the
derivatives of (A6) with respect to k and s
w
, respectively,
as shown below:
@w=@R
@k
=
s
j
s
h
÷s
j
÷
bs
w
s
h
÷s
w
> 0 and
@w=@R
@s
w
= ÷
[1 ÷ k[bs
h
[s
h
÷s
w
[
2
< 0:
h
Proof of Proposition 2. From(A5), the sensitivity of pay to
program expenses is:
@w
@P
=
[1 ÷ k[s
w
s
h
÷s
w
: (A7)
The proposition then follows by noting the sign of the
derivatives of (A7) with respect to k and s
w
, respectively,
as shown below:
@w=@P
@k
= ÷
s
w
s
h
÷s
w
< 0 and
@w=@P
@s
w
=
[1 ÷ k[s
h
[s
h
÷s
w
[
2
> 0:
h
Proof of Proposition 3. Given e
R
, F, and R, and wage in
(A5), the executive chooses e
P1
and e
P2
to solve:
Max
e
P1
;e
P2
E
h
P
;w
w e
R
; F; R [ ¦ ¦ ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
÷cE
h
P
;w
R ÷ F ÷ P ÷ A
I
e
R
; F; R [ ¦ ¦: (A8)
Using (A5), P = b[R ÷ F] + h
P
+ e
P1
+ e
P2
+ w, and A
I
= I ÷ e
P2
,
(A8) can be written as:
Max
e
P1
;e
P2
k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
R÷a
???
F
_
÷e
c
R
÷
j
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
h÷e
P1
÷e
P2
÷e
c
P1
÷e
c
P2
_ _
_ _
÷k
R
e
2
R
÷k
P1
e
2
P1
÷k
P2
e
2
P2
÷c[(1÷b)(R÷F) ÷
h÷e
P1
÷
w÷I[:
(A9)
The derivative of (A9) with respect to e
P1
equals:
[
^
c ÷c[ ÷2k
P1
e
P1
; where
^
c =
[1 ÷ k[s
w
s
h
÷s
w
: (A10)
From (A10), if c _
^
c, the executive’s payoff is decreasing
in e
P1
, so e
P1
= 0; else, e
P1
is interior and is obtained by set-
ting (A10) equal to 0. That is,
^
e
P1
=
^
c ÷c
2k
P1
if c <
^
c; and
^
e
P1
= 0; otherwise:
(A11)
Also, from (A9), the ?rst-order condition with respect to e
P2
equals
^
c ÷2k
P2
e
P2
= 0. Solving this yields:
^
e
P2
=
^
c
2k
P2
: (A12)
Using (A11), and c <
^
c, the results then follow from the
sign of the following derivatives:
@
^
e
Pi
@k
=÷
s
w
2k
Pi
[s
h
÷s
w
[
0; and
@
^
e
R
@b
=÷
c
2k
R
0;
^
e
R
÷
~
e
R
= 0; and
^
F ÷
~
F = ÷
a
2
ks
j
[2c(1 ÷ b)(s
h
÷s
j
) ÷ ks
j
[
[2c(1 ÷ b)(s
h
÷s
j
)[
2
< 0: (A26)
(A26) proves Proposition 6. h
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 11
Proof of Proposition 7. Comparing the welfare expres-
sions in Proposition 5 and Lemma 2,
^
W ÷
~
W =
1
4
a
2
bk
2
s
2
j
(1 ÷ b)
2
c
2
(s
h
÷s
j
)
2
_
÷
(1 ÷ k)
2
s
2
w
k
P2
(s
h
÷s
w
)
2
÷
1
k
P1
2 ÷
(1 ÷ k)s
w
s
h
÷s
w
÷c
_ _
(1 ÷ k)s
w
s
h
÷s
w
÷c
_ __
if c <
[1 ÷ k[s
w
s
h
÷s
w
; and
^
W ÷
~
W =
1
4
a
2
bk
2
s
2
j
(1 ÷ b)
2
c
2
(s
h
÷s
j
)
2
÷
(1 ÷ k)
2
s
2
w
k
P2
(s
h
÷s
w
)
2
_ _
; otherwise:
(A27)
Setting
^
W ÷
~
W in (A27) equal to 0, and solving for k
P2
yields the value of k
/
. This proves Proposition 7. h
Proof of Proposition 8. With functional classi?cation of
expenses, the solution now entails ?rst solving (5) and,
then, (7) with the term ge
R
e
P1
added to both objective
functions (with a = c = 0 and b = 1). This yields the follow-
ing analog to Proposition 5:
^
e
R
(g) =
1
4k
P1
k
R
÷g
2
2k
sj
s
h
÷sj
_ _
k
P1
÷ (1 ÷ k)
s
w
s
h
÷s
w
_ _
g
_ _
;
^
e
P1
(g) =
1
4k
P1
k
R
÷g
2
k
sj
s
h
÷sj
_ _
g ÷2(1 ÷ k)
s
w
s
h
÷s
w
_ _
k
R
_ _
;
^
e
P2
(g) =
1
2k
P2
(1÷k)s
w
s
h
÷s
w
_ _
; and
^
W(g) = 2
h ÷
j ÷
w ÷ I ÷
^
e
R
÷
^
e
P1
÷ k
R
^
e
2
R
÷ k
P1
^
e
2
P1
÷ k
P2
^
e
2
P2
÷g
^
e
R
^
e
P1
: (A28)
In the absence of functional classi?cation of expenses, the
solution is obtained by solving (A20) and (A23) with the
term ge
R
e
P1
again added to both the objective functions.
This yields the following analog to Lemma 2:
~
e
R
(g) =
1
4k
P1
k
R
÷g
2
2k
sj
s
h
÷sj
_ _
k
P1
_ _
;
~
e
P1
(g) =
1
4k
P1
k
R
÷g
2
k
sj
s
h
÷sj
_ _
g
_ _
;
~
e
P2
(g) = 0; and
~
W(g) = 2
h ÷
j ÷
w ÷ I ÷
~
e
R
÷
~
e
P1
÷ k
R
~
e
2
R
÷ k
P1
~
e
2
P1
÷g
~
e
R
~
e
P1
: (A29)
Part (i) follows immediately by comparing the effort levels
in (A28) with the corresponding effort levels in (A29).
Turning to welfare comparison in part (ii),
^
W(g)÷
~
W(g)=
s
w
[1÷k[
4k
P2
[s
h
÷s
w
[[4k
P1
k
R
÷g
2
[
× 4k
P2
g 1÷
s
j
k
s
h
÷s
j
_ _ _ _
÷k
R
2÷
s
w
(1÷k)
s
h
÷s
w
_ __
÷
s
w
[1÷k[[4k
P1
k
R
÷g
2
[
s
h
÷s
w
_
:
(A30)
From (A30),
^
W(g) ÷
~
W(g) > 0 if and only if k
P2
> k
/
(g),
where:
k
/
(g) =
s
w
(1 ÷ k)(4k
P1
k
R
÷g
2
)
s
h
÷s
w
_ __
4g 1 ÷
s
j
k
s
h
÷s
j
_ _ _
÷4k
R
2 ÷
s
w
(1 ÷ k)
s
h
÷s
w
_ __
: (A31)
The numerator in the right-hand-side of (A31) is decreas-
ing in g while the denominator is increasing in g. Thus,
dk
/
(g)
dg
< 0, completing the proof of the proposition. h
References
Arya, A., Frimor, H., & Mittendorf, B. (2010). The bene?ts of aggregate
performance metrics in the presence of career concerns. Management
Science, 57, 1424–1437.
Auriol, E., Friebel, G., & Pechlivanos, L. (2002). Career concerns in teams.
Journal of Labor Economics, 20, 289–307.
Autrey, R., Dikolli, S., & Newman, P. (2007). Career concerns and
mandated disclosure. Journal of Accounting and Public Policy, 26,
527–554.
Autrey, R., Dikolli, S., & Newman, P. (2010). Performance measure
aggregation, career incentives, and explicit incentives. Journal of
Management Accounting Research, 22, 115–131.
Baber, W., Daniel, P., & Roberts, A. (2002). Compensation to managers of
charitable organizations: An empirical study of the role of accounting
measures of program activities. The Accounting Review, 77, 679–693.
Dewatripont, M., Jewitt, I., & Tirole, J. (1999a). The economics of career
concerns, Part 1: Comparing information structures. Review of
Economic Studies, 66, 183–198.
Dewatripont, M., Jewitt, I., & Tirole, J. (1999b). The economics of career
concerns, Part 2: Application to missions and accountability of
government agencies. Review of Economic Studies, 66, 199–217.
Frumkin, P., & Keating, K. (2010). The price of doing good: Executive
compensation in nonpro?t organizations. Policy and Society, 29,
269–282.
Greene, S. (1997). Econometric analysis. Upper Saddle River, NJ: Prentice
Hall.
Gregory, A., & Howard, D. (2009). The nonpro?t starvation cycle. Stanford
Social Innovation Review; Fall (pp. 49–53).
Holmstrom, B. (1982). Managerial incentives – A dynamic perspective. In
Essays in economics and management in Honor of Lars Wahlbeck.
Helsinki, Finland: Swedish School of Economics.
Holmstrom, B. (1999). Managerial incentive problems: A dynamic
perspective. Review of Economic Studies, 66, 169–182.
Holmstrom, B., & Ricart i Costa, J. (1986). Managerial incentives and
capital management. Quarterly Journal of Economics, 101, 835–860.
Kaarboe, O., & Olsen, T. (2006). Career concerns, monetary incentives, and
job design. Scandinavian Journal of Economics, 108, 299–316.
Krishnan, R., & Yetman, M. (2011). Institutional drivers of reporting
decision in nonpro?t hospitals. Journal of Accounting Research, 49,
1001–1039.
Krishnan, R., Yetman, M., & Yetman, R. (2006). Expense misreporting in
nonpro?t organizations. The Accounting Review, 81, 399–420.
Milbourn, T., Shockley, R., & Thakor, A. (2001). Managerial career concerns
and investments in information. RAND Journal of Economics, 32,
334–351.
Pallotta, D. (2010). Uncharitable: How restraints on nonpro?ts undermine
their potential. Tufts University Press.
Sedatole, K., Swaney, A., Yetman, M., and Yetman, R. (2013). Accounting-
based performance metrics and executive compensation in nonpro?t
organizations. Michigan State University working paper.
Shleifer, A., & Wolfenzon, D. (2002). Investor protection and equity
markets. Journal of Financial Economics, 66, 3–27.
Stern, K. (2013). With charity for all: Why charities are failing and a better
way to give. Doubleday.
12 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
doc_265634350.pdf
Many in the nonprofit sector view accounting-based performance measures to be overly
influential and counterproductive in the evaluation of charities and their leaders. The contention
is that such measures are imperfect and often biased, leading to dysfunctional
rationing of fundraising and administrative infrastructure. To examine these concerns
and the broader question of nonprofit executive incentives, we develop a model of nonprofit
executives who are concerned with influencing external perceptions.
Career concerns and accounting performance measures
in nonpro?t organizations
Anil Arya
?
, Brian Mittendorf
Ohio State University, United States
a b s t r a c t
Many in the nonpro?t sector view accounting-based performance measures to be overly
in?uential and counterproductive in the evaluation of charities and their leaders. The con-
tention is that such measures are imperfect and often biased, leading to dysfunctional
rationing of fundraising and administrative infrastructure. To examine these concerns
and the broader question of nonpro?t executive incentives, we develop a model of non-
pro?t executives who are concerned with in?uencing external perceptions. In doing so,
we demonstrate that accounting-based performance measures alter executive incentives
in critical ways. In particular, disclosure of the functional classi?cation of nonpro?ts’
expenses can reduce incentives to overinvest in fundraising and restore investments in
programs; at the same time, it also comes with the potential downside of undermining
key investments in long-term infrastructure.
Ó 2014 Elsevier Ltd. All rights reserved.
Introduction
In his now-viral TED talk (‘‘The Way We Think About
Charity is Dead Wrong’’), Dan Pallotta provided a voice to
many in the nonpro?t sector who view the reliance on
accounting-based performance measures for non-govern-
mental organizations (NGOs) to be detrimental to their
overall effectiveness. The viewpoint, elaborated upon in
books by Pallotta (2010), Stern (2013), and others, is that
by developing an imperfect picture of how much of a non-
pro?t’s resources are devoted to its mission, the functional
classi?cation of expenses has become a key focus of outsid-
ers’ evaluation of NGOs and their executives; this focus, in
turn, incentivizes executives to cut key investments in
fundraising and advertising. Despite the almost universal
acceptance of this view and concomitant disdain for
accounting performance measures in the nonpro?t com-
munity (even spawning the ‘‘Overhead Myth’’ movement
to discredit reliance on accounting measures), there has
been little or no formal analysis of the viewpoint.
In this paper, we present a parsimonious model of non-
pro?t executives’ incentives in order to examine the conse-
quences of reliance on accounting-based performance
measures. The model captures two key elements: (i) to
the extent that nonpro?t executives are driven by extrinsic
motivation, such incentives are typically not due to explicit
pay arrangements but rather an incentive to in?uence
external perceptions; and (ii) the primary accounting per-
formance measure that is critical and unique to nonpro?ts
is the functional classi?cation of expenses, which provides
an assessment of the portion of expenses attributable to
achieving an organization’s mission, i.e., the program
expenses.
Our setting provides a formal analysis of actions non-
pro?t executives are incentivized to undertake when at
least part of their motivation is driven by a desire to boost
market perceptions of their effectiveness. Consistent with
intuition, we show that the more the market values the
ability to generate revenues and/or the more executives
can bene?t fromadministrative perquisites, the more focus
executives place on efforts to generate revenues. Similarly,
the more the market values the ability to ef?ciently focushttp://dx.doi.org/10.1016/j.aos.2014.10.002
0361-3682/Ó 2014 Elsevier Ltd. All rights reserved.
?
Corresponding author.
Accounting, Organizations and Society 40 (2015) 1–12
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
resources toward a mission and/or the less executives can
bene?t from perquisites, the more emphasis they place on
using (rather than generating) resources effectively.
Interestingly, we also show that the more ef?cient an orga-
nization is in using resources, the less incentive an execu-
tive has in trying to generate such resources. Among other
things, this suggests that if left unchecked, the natural
order of things may lead to inherently inef?cient organiza-
tions being the ones generating the bulk of donations.
The model also presents a different view of observed
executive pay-to-performance sensitivities. That is, though
empirical observation of variation in nonpro?t executive
pay moving in concert with accounting metrics is often
interpreted as evidence of contractual performance pay
arrangements, implicit market forces may also explain
such variability. In particular, despite the lack of explicit
incentive pay, our model demonstrates how external
demands for nonpro?t executives will generate pay that
varies predictably with accounting outcomes. This mar-
ket-driven view of pay-to-performance sensitivity also
offers additional empirical implications. For one, we show
that the more the market places a premium on revenue-
generating ability relative to the ability to devote resources
to the mission, the more sensitive pay will be to revenue
and the less sensitive it will be to the measure of program
expenses. This means that in markets characterized by
strong presence of for-pro?t entities, for whom revenue
is at a premium (e.g., education; health care), the more
(less) pay should appear sensitive to revenues (program
expenses). Additionally, we show that the more precise
the accounting cost allocation exercise, the less sensitive
pay will be to revenue and the more sensitive it will be
to program expenses. In other words, the greater the
uncertainty about the functional allocation of expenses
(driven by, say, the use of SOP 98-2 in allocating joint
costs), the greater the pay sensitivity will be to revenue
and the less weight the market will place on program
expenses.
Returning to the fundamental criticism of how account-
ing estimates can distort incentives, our model of career
concerns of nonpro?t executives demonstrates that the
incentive to in?uence external perceptions in itself can dis-
tort decisions away from what may be in the best interest
of donors. In particular, in order to boost perceptions of
revenue-generating abilities, a nonpro?t executive may
overinvest in fundraising efforts. This, in turn, can lead to
a reduced focus on improving efforts to ef?ciently direct
resources to the organization’s mission. Interestingly, the
functional classi?cation of expenses stipulated by account-
ing rules can help mitigate each of these concerns.
By tracking expenditures that go into the revenue-gen-
eration process, the accounting measure helps sift out
what revenues are attributable to executive ability and
what are simply due to high fundraising spending. This
undercuts the incentive for the executive to overinvest in
fundraising as a means of trying to posture to a market-
place that values revenue generation. The attempt to
separate program expenses as part of functional expense
classi?cation also gives the marketplace a second measure
on which to evaluate executives – their ef?ciency at putt-
ing resources to use. The newfound emphasis on program
expenses incentivizes executives to put more energy into
cutting administrative bloat and effectively directing
resources toward the mission.
However, this desire to cut administrative costs to sig-
nal greater ef?ciency to markets comes with a downside
in that it also encourages executives to divert resources
from potentially useful long-term infrastructure spending.
In other words, since accounting does not provide a natural
distinction between wasteful and useful administrative
spending, it incentivizes executives to ‘‘trim the fat’’ even
in cases where some administrative spending is critical.
That is, an additional implication of our model is that it
provides a theoretical justi?cation for the so-called non-
pro?t starvation cycle (Gregory & Howard, 2009), where
efforts to trim administrative costs undermine long-term
viability.
Taken together, our results suggest a nuanced balance
between viewing accounting measures with utter suspi-
cion or with unadulterated faith is warranted. The func-
tional classi?cation of expenses shifts the executive’s
emphasis away from excessive fundraising while also
simultaneously forcing him to focus on the use of
resources. The latter entails a trade-off. When excessive
administrative spending is the norm in the absence of
functional classi?cation, the tight discipline introduced
by accounting metrics proves bene?cial. On the other
hand, cost cutting can be excessive when it discourages
spending on valuable infrastructure. The paper’s proposi-
tions succinctly capture these economic forces.
This paper lies at the nexus of the literatures on career
concerns incentives and accounting performance measure-
ment for NGOs. In terms of the ?rst literature stream, the
seminal analysis of career concerns incentives is
Holmstrom (1982, 1999). This initial analysis of implicit
career incentives was expanded by Dewatripont, Jewitt,
and Tirole (1999a, 1999b) who generalize the model to
examine, among other things, the allocation of effort across
tasks, complementarities between skill and effort, and
alternative information structures. The career concerns
framework has also generated insights about managerial
investment (Holmstrom & Ricart i Costa, 1986), informa-
tion acquisition incentives (Milbourn, Shockley, & Thakor,
2001), team dynamics (Auriol, Friebel, & Pechlivanos,
2002), job design (Kaarboe & Olsen, 2006), disclosure reg-
ulation (Autrey, Dikolli, & Newman, 2007), and perfor-
mance measure design (Arya, Frimor, & Mittendorf, 2010;
Autrey, Dikolli, & Newman, 2010).
Theoretical inquiry of career concerns is motivated by
circumstances wherein employees face limited explicit
incentive compensation but rely on salaries determined
by labor markets. Driven by both donor and regulatory
restrictions, such limited incentive pay is commonplace
among NGOs, making incentives to in?uence external per-
ceptions all the more important to examine in nonpro?ts.
Surprisingly, there is a dearth of theoretical study of
incentives among NGOs and none (to our knowledge) on
the incentives of nonpro?t executives to in?uence market-
place perceptions. That said, there has been substantial
empirical study of NGO behavior and the role of accounting
measures therein. In terms of executive pay incentives,
Baber, Daniel, and Roberts (2002) and Sedatole, Swaney,
2 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
Yetman, and Yetman (2013) each demonstrate, to different
degrees, that key accounting measures (in particular, reve-
nues and program expenses) are associated with pay vari-
ations among nonpro?t executives. Frumkin and Keating
(2010) demonstrate that executive pay is more heavily
in?uenced by pay levels at peer institutions than it is by
performance. In a sense, our study brings together these
two seemingly disparate views by theoretically examining
how the in?uence of outside pay opportunities can alter
executive behavior and induce variation in the observed
sensitivities between pay and nonpro?t performance
despite the absence of explicit incentive pay. To this end,
our study also relates to and complements the work of
Krishnan, Yetman, and Yetman (2006) and Krishnan and
Yetman (2011), who demonstrate that nonpro?ts are
incentivized to manage accounting performance measures
in order to manage external perceptions. Our model incor-
porates the potential for both bias and noise in accounting
measures and demonstrates the consequences for execu-
tive incentives, pay, and efforts.
Model
An executive of a nonpro?t organization is charged with
generating revenues and putting those revenues to use in
ful?lling the organization’s mission. The revenues raised
by the organization, R, depend on the extent of ?nancial
resources devoted toward fundraising and the executive’s
skill and effort in generating funds. In particular,
R = a
???
F
_
÷ h
R
÷ e
R
÷j, where F denotes fundraising expen-
ditures, a denotes inherent fundraising ef?ciency, h
R
denotes the executive’s skill in raising funds, e
R
, e
R
P0,
denotes the executive’s effort in revenue generation, and
j re?ects all other revenue-related factors.
The organization’s net revenues, R ÷ F, are split
between use for administration and for implementing the
nonpro?t’s programs. The executive has a measure of con-
trol over this split. In particular, the amount directed
toward programs is P = b[R ÷ F] + h
P
+ e
P1
+ e
P2
+ w. Here,
b ÷ (0, 1) denotes the default fraction of resources devoted
to programs, h
P
re?ects the executive’s skill in ef?ciently
utilizing resources for programs, e
P1
and e
P2
, e
Pi
P0, re?ect
executive efforts to divert resources to programs, and w
re?ects other factors. Higher program skills (h
P
) can re?ect
abilities to ?nd ef?ciencies and reduce redundant expendi-
tures but can also re?ect a higher degree of integrity, i.e., a
lower willingness to divert resources for personal bene?t
(as in Shleifer & Wolfenzon, 2002). The two effort terms
re?ect that the executive can take actions to shift resources
toward programs, and such resource shifts could either
divert from wasteful administrative spending (e
P1
) or could
reduce useful administrative spending (e
P2
).
To elaborate, administrative expenses, denoted by A,
equal the residual expenditures R ÷ F ÷ P. These adminis-
trative expenses consist of both useful and wasteful com-
ponents – some administrative spending, denoted A
I
, is
infrastructure spending critical to mission success while
the remainder represents wasteful spending. Denoting
the maximum infrastructure needs by I, infrastructure
spending is simply A
I
= I ÷ e
P2
, re?ecting that the execu-
tive can take effort to boost program spending in the
short-run at the expense of long-run mission success by
neglecting key infrastructure. The remaining administra-
tive funds, R ÷ F ÷ P ÷ A
I
, though unhelpful to the mission,
may yield some perquisites for the executive; executive
perquisites equal c[R ÷ F ÷ P ÷ A
I
], c ÷ (0, 1).
The prior belief of the executive’s skills, held by all, is
that h
R
and h
P
are independent and normally distributed,
with mean
h and variance r
2
h
(with precision denoted
s
h
= 1=r
2
h
). Similarly, j is normally distributed with mean
j and variance r
2
j
(with precision denoted s
j
= 1=r
2
j
); for
w, the corresponding mean, variance, and precision values
are denoted
w, r
2
w
, and s
w
, respectively.
As is often the case in nonpro?t organizations, the exec-
utive is not provided explicit incentive pay. However, mar-
ket demands for his skill may necessitate adjustments to
his future pay. The executive’s future wage, denoted w, is
determined by a competitive labor market. The market
wage is w = kE h
R
¦ ¦ ÷ [1 ÷ k[E h
P
¦ ¦, where E ¦ ¦ denotes mar-
ket expectations, and k (1 ÷ k) re?ects the degree to which
the market cares about the executive’s ability to generate
revenues (ef?ciently use revenues on programs). While
we model the executive’s incentive as one of market-
driven pay, this can be more broadly thought of as an
incentive to boost market perceptions – under this inter-
pretation, boosting market perceptions may be valued by
the executive even if it does not come with higher pay
since it can come with greater visibility or higher-pro?le
positions. Under either interpretation, the market’s expec-
tations rely on (observable) revenues and the functional
classi?cation of expenses. The classi?cation splits total
expenses into its three components: fundraising expenses
(F), program expenses (P), and administrative expenses
(A = R ÷ F ÷ P).
In choosing his efforts, the executive balances the
potential for future compensation and perquisites with
his personal cost of efforts, k
R
e
2
R
, k
P1
e
2
P1
, and k
P2
e
2
P2
. Given
this basic structure, we investigate how career concerns
affect nonpro?t executives’ efforts, and how these incen-
tives are in?uenced by the functional classi?cation of
expenses. Fig. 1 summarizes the sequence of events.
Results
First-best benchmark
Before deriving the equilibrium that arises when non-
pro?t executives are interested in external perceptions
and internal perks, we begin with a ?rst-best benchmark.
That is, what is the outcome if the executive is solely
focused on mission impact? In this setting, the entity’s
mission is enhanced both through current program expen-
ditures (P) and investments to boost the long-term impact
of such expenditures (A
I
). In particular, denote the overall
welfare, W, as the expected value of mission-based expen-
ditures (both programs and infrastructure) less effort
costs:
W = E¦P ÷ A
I
¦ ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
:
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 3
If the executive is concerned only with welfare, he
solves the following program in choosing fundraising
expenditures and efforts:
Max
F;e
R
;e
P1
;e
P2
E¦P ÷ A
I
¦ ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
= E¦b[a
???
F
_
÷ h
R
÷ e
R
÷j ÷ F[ ÷ h
P
÷ e
P1
÷ w ÷ I¦
÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
: (1)
Solving (1) yields the ?rst-best benchmark, as summarized
in Lemma 1. (All proofs are provided in the appendix.)
Lemma 1. The welfare-maximizing equilibrium entails:
F
+
=
a
2
4
; e
+
R
=
b
2k
R
; e
+
P1
=
1
2k
P1
; e
+
P2
= 0; and
W
+
= [1 ÷ b[
h ÷ b
j ÷
w ÷ I ÷
1
4
a
2
b ÷
b
2
k
R
÷
1
k
P1
_ _
:
The ?rst-best outcome presented in the lemma offers
some intuitive properties. First, the level of fundraising
undertaken is increasing in the organization’s inherent
fundraising ef?ciency, a. The effort undertaken on boosting
revenues is increasing in the degree to which the organiza-
tion ef?ciently uses funds for its mission, b, and decreasing
in k
R
, the cost of revenue effort. Finally, the executive
undertakes efforts to trim wasteful spending but does not
trim infrastructure spending (e
+
P1
> 0 and e
+
P2
= 0). With
this intuitive benchmark in place for comparison, we next
consider the outcome when nonpro?t executives are
in?uenced by external perceptions.
Equilibrium market wages
While there is little doubt that nonpro?t executives are
motivated by a desire to improve society, they are also not
immune to the pressures to boost perceptions of their abil-
ity and performance. Such career concerns may even be
more pronounced in nonpro?t organizations than their
for-pro?t counterparts due to the notable absence of expli-
cit incentive pay arrangements. In such cases, a nonpro?t
executive’s actions are not solely determined by overall
welfare but also affected by a desire to in?uence external
perceptions and future pay potential. Following this frame-
work, we work backwards in the game to determine the
equilibrium, starting with a derivation of how the market
wage is determined.
The wage depends on market beliefs about the
executive’s revenue-generation skill as well as his skill in
ef?ciently utilizing the raised resources. Market beliefs,
in turn, can be conditioned on revenues (R) and, under
functional classi?cation, on the decomposition of expenses
into fundraising (F), program (P), and administration (A).
Since A simply equals R ÷ F ÷ P, it follows that market
beliefs are effectively based on R, F, and P. Conditioned
on this information set, and relegating technical details
to the appendix, the market’s conjecture about the execu-
tive’s revenue-skill equals:
E¦h
R
R; F; P¦ [ =
s
h
s
h
÷s
j
_ _
h ÷
s
j
s
h
÷s
j
_ _
R ÷a
???
F
_
÷ e
c
R
÷ j
_ _
_ _
:
(2)
In (2), e
c
R
represents the market’s conjecture of the execu-
tive’s revenue effort. The reason conjectures come into
play is that the market wage is not a reward for past efforts
or successes but instead a market-based inference of skill
parameters from observation of past performance. Thus,
in making inferences about h
R
, the market seeks to adjust
performance to remove parts attributable to past effort
and other non-persistent components. The market does
not observe effort, but instead relies on its belief about
what effort was – in equilibrium, such conjectures will
be correct.
In particular, the inference of revenue-skill in (2)
re?ects a weighted average of the market’s prior belief
and a signal of ability inferred from observed revenues.
This signal equals observed revenues less the portion
attributable to fundraising expenses, conjectured effort to
boost revenues, and other (transitory) factors. The weights
placed on the two components depend on the precision of
the prior belief (s
h
) relative to that of the observed
revenue-based signal (s
j
).
In a similar fashion, the market’s conjecture of skill in
ef?ciently directing resources towards programs is:
E¦h
P
R; F; P¦ [ =
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
P ÷b[R÷F[ ÷e
c
P1
÷e
c
P2
÷
w
_ _
_ _
:
(3)
In (3), e
c
P1
and e
c
P2
represent the external market’s conjec-
ture of efforts to divert resources from wasteful and useful
administrative spending, respectively. The market’s expec-
tation of skill in ef?ciently using resources is again a
weighted average of its prior and a signal of such ability
derived from observed program expenses. In this case,
the program expenses are adjusted for net revenues, con-
jectured cost-cutting efforts, and other transitory factors.
The weights depend on the relative precision of the prior
(s
h
) and the accounting-based estimate inherent in
functional classi?cation (s
w
).
Given (2) and (3), the executive’s market wage, after the
market has observed revenue and the functional
classi?cation of expenses, can be expressed as:
The
executive
exerts
revenue
effort e
R
and
fundraising
costs F.
Revenues are
realized.
The executive exerts
program efforts
e
P1
and e
P2
.
Functional expenses
are disclosed. The
labor market
determines the
executive's
continuation wage, w.
Fig. 1. Timing of events.
4 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
w=k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
R÷a
???
F
_
÷e
c
R
÷
j
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
P ÷b[R÷F[ ÷e
c
P1
÷e
c
P2
÷
w
_ _
_ _
:
(4)
The two parts of the wage are derived from the infer-
ences of ability to generate revenue weighted by k, and
the inference of ability to effectively deploy funds toward
the mission weighted by 1 ÷ k. Delaying consideration of
the executive’s effort choices for now, consider the impli-
cations of equilibrium market wage for observed pay-to-
performance sensitivities.
Take ?rst the relation between revenue and pay. From
the ?rst term in (4), higher revenues are indicative of
higher fundraising skill and are rewarded as such. This is
consistent with the empirical evidence in Baber et al.
(2002) and Sedatole et al. (2013) who each show a positive
relationship between pay and overall revenue for nonpro?t
executives. In this case, however, the relationship is not
evidence of contractual performance pay, but instead dri-
ven by market forces.
From the second term in (4), a second feature is also in
play: higher revenues also indicate that a given level of
program expenses is attributable to scale rather than exec-
utive skill at ef?ciently employing resources. Roughly, this
is consistent with the notion of a program expense ratio
(here it is difference rather than ratio), re?ecting that
higher revenues alone demand higher program expenses.
As such, to the degree the external market cares about
and can effectively isolate skills at ef?ciently using
resources, higher revenues can actually reduce pay.
These two forces mean that pay-to-performance sensi-
tivity of revenues is increasing in the extent to which the
market values the executive’s ability to generate revenue
(k) and decreasing in the informativeness of the functional
classi?cation of expenses (s
w
).
Proposition 1. Sensitivity of pay to revenue, @w/@R, is:
(i) increasing in labor market demand for revenue skills
(k); and
(ii) decreasing in the informativeness of functional expense
classi?cation (s
w
).
Part (i) of the proposition provides a notable empirical
implication. For some nonpro?ts, the ability to generate
revenues is far more important than their ef?cient use
when it comes to outside market wages. For example, it is
often discussed that the presence of for-pro?t counterparts
in health-care and education creates a greater outside
demand for revenue generation and less demand for ability
to achieve missions (i.e., higher k). This suggests that
such nonpro?ts should empirically see a higher pay-to-
performance sensitivity when it comes to revenues, this
despite the fact that there no explicit incentive contracts
in play.
Part (ii) of the proposition demonstrates the subtleties of
inferring the usefulness of accounting estimates based on
howthey are used. Since higher revenues indicate that pro-
gram expenses are less attributable to executive skill in
managing resources, the more the market relies on program
expenses to infer such skills, the less it will reward higher
revenues. This means that the more effective an accounting
estimate of cost-management (program expenses) is, the
less the market relies on another (revenues). So, a reduced
reliance on one accounting ?gure may actually indicate
more, not less, effective inference of skill.
Now consider how pay varies with observed estimates
of program expenses. Again, using (4) reveals the equilib-
rium pay-to-performance sensitivity.
Proposition 2. Sensitivity of pay to program expenses,
@w/@P, is:
(i) decreasing in labor market demand for revenue skills
(k); and
(ii) increasing in the informativeness of functional expense
classi?cation (s
w
).
The predictions again mirror empirical data but provide
a different interpretation due to the absence of incentive
pay: higher program expenses are associated with higher
pay. However, this relation is muted the more the market
cares about revenue-generation skill (k). This may provide
some explanation for the varied degrees to which Baber
et al. (2002) and Sedatole et al. (2013) observe market
rewards for program expenses. While higher program
expenses are rewarded, the degree of this reward varies
predictably with both market priorities (k) and the non-
pro?t’s accounting system (s
w
). Further, in contrast to
Proposition 1(II), Proposition 2(II) shows that more precise
accounting estimates of ef?ciency in utilizing resources
implies a more positive pay-to-performance relation when
it comes to program expenses.
Career concerns equilibrium
Noting the in?uence of outcomes on market wages as
derived above, and taking market conjectures of effort as
given, the executive has incentives to incur efforts in
order to in?uence perceptions. That is, knowing the out-
comes will be used to infer ability, the executive has
incentive to in?uence such outcomes. Though any such
efforts will be correctly inferred in equilibrium, the exec-
utive nonetheless cannot resist the temptation. In partic-
ular, working backwards in the game, the executive
chooses cost-cutting efforts to maximize expected wage,
less effort cost adjusted for perks from administrative
expenditures, as in (5):
Max
e
P1
;e
P2
k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
R÷a
???
F
_
÷e
c
R
÷
j
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
h÷e
P1
÷e
P2
÷e
c
P1
÷e
c
P2
_ _
_ _
÷k
R
e
2
R
÷k
P1
e
2
P1
÷k
P2
e
2
P2
÷c[(1÷b)(R÷F) ÷
h÷e
P1
÷
w÷I[:
(5)
From(5), the executive’s effort choices weighs three fac-
tors: (i) higher efforts increase program expenses and thus
perceptions of ef?ciency in utilizing resources; (ii) higher
efforts are personally costly; and (iii) higher effort to reduce
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 5
wasteful spending (e
P1
) also reduces perks. Taken together,
these features lead to the executive’s cost-reduction effort
choices. As it turns out, the choices that solve (5) are free
of the market’s conjectures, e
c
Pi
, ensuring they are also the
equilibrium choices. The equilibrium efforts aimed at
reducing administrative costs are:
e
P1
=
^
e
P1
= Max
[1 ÷ k[s
w
2k
P1
[s
h
÷s
w
[
÷
c
2k
P1
; 0
_ _
and
e
P2
=
^
e
P2
=
[1 ÷ k[s
w
2k
P2
[s
h
÷s
w
[
: (6)
The following proposition summarizes some key charac-
teristics of the equilibrium efforts to reduce administrative
spending.
Proposition 3. In the presence of career concerns,
(i) the (non-zero) effort to reduce administrative waste
(
^
e
P1
) is decreasing in both labor market demand for
revenue skills (k) and executive perk consumption (c).
(ii) the effort to reduce infrastructure spending (
^
e
P2
) is
decreasing in labor market demand for revenue skills
(k) but unaffected by executive perk consumption (c).
Intuitively, the proposition demonstrates that the more
the market cares about revenues (and less it cares about
cost-ef?ciency), the less effort the executive exerts to cut
administrative spending. It also shows that the more the
executive cares about consuming perks now rather than
in?uencing pay in the future, the less he will exert efforts
to reduce wasteful administrative spending.
Using these equilibrium efforts, we can now work back-
ward to determine the executive’s choice of fundraising
expenditure and revenue effort. The executive makes this
choice to maximize expected wage less effort cost adjusted
for perk consumption, as in (7):
Max
e
R
; F
k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
h÷e
R
÷e
c
R
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
h÷
^
e
P1
÷
^
e
P2
÷e
c
Pi
÷e
c
P2
_ _
_ _
÷k
R
e
2
R
÷k
P1
^
e
2
P1
÷k
P2
^
e
2
P2
÷c[(1÷b)(a
???
F
_
÷
h÷e
R
÷
j÷F)
÷
h÷
^
e
P1
÷
w÷I[: (7)
From (7), the executive’s revenue effort balances three
factors: (i) higher effort increases revenues and thus mar-
ket inferences of revenue ability; (ii) higher effort is costly;
and (iii) higher effort generates more revenues, a portion of
which are consumed as perks. While fundraising expendi-
tures, F, do not in?uence market perceptions (the market
observes them so removes them in the inference process),
they can in?uence the level of funds available to be con-
sumed as perks, making them another nontrivial choice.
These factors, taken together, determine the optimal reve-
nue and fundraising choices, each of which are indepen-
dent of market conjectures of effort and thus comprise
the equilibrium outcomes. The solutions to (7) are:
e
R
=
^
e
R
=
ks
j
2k
R
[s
h
÷s
j
[
÷
c[1 ÷ b[
2k
R
and F =
^
F =
a
2
4
(8)
The critical properties of revenue effort are presented in
Proposition 4.
Proposition 4. In the presence of career concerns, revenue
effort (^e
R
) is:
(i) increasing in labor market demand for revenue skills
(k);
(ii) increasing in executive perk consumption (c); and
(iii) decreasing in the organization’s inherent program ef?-
ciency (b).
Intuitively, (i) demonstrates that the more the market
cares about the executive’s ability to generate revenue,
the more incentive he has to exert revenue effort to (try
to) in?uence the market’s perception. Interestingly, and
in contrast to the case of cost-cutting efforts, (ii) demon-
strates that the desire to consume perks now also pro-
vides revenue effort incentives. The reason is that if the
executive gets to make use of a portion of revenues to
consume as perks (e.g., nicer of?ce, ?rst-class travel,
etc.), that alone provides him incentive to generate reve-
nues. The underlying force in (ii), however, is mitigated
by the organization’s inherent ef?ciency in using
resources (e.g., board oversight, restrictions on spending,
etc.), as con?rmed in (iii). That is, if the organization is
particularly ef?cient in using resources, the executive is
aware that fewer resources will become perks, and his
incentives to generate resources are muted. This compar-
ative static suggests that, absent intervention, organiza-
tions which are particularly poor at using resources may
ironically also be the ones that have executives who do
more to generate such revenues.
The incentives that underlie the career concerns equi-
librium derived herein are the desire to in?uence pay in
the future and the desire to consume perks in the present.
The desire to in?uence pay is best re?ected by k which cap-
tures the relative importance the market places on ability
to generate resources versus utilizing the resources,
whereas the incentive for current perquisite consumption
is captured by c. Fig. 2 demonstrates the consequences of
each incentive on equilibrium efforts.
The comparative statics re?ected in the left panel of
Fig. 2 point to a reasonable empirical test about different
types of nonpro?ts. Among nonpro?ts for whom executive
outside opportunities stress revenues (e.g., health care),
one would expect greater (lower) effort by executives in
generating (ef?ciently using) resources. The changed
incentives to cut costs manifests itself on two dimensions
in that an incentive to ef?ciently use resources also entails
some myopic cost-savings that reduce infrastructure.
The comparative statics in the right panel of Fig. 2 pro-
vide empirical implications about different types of execu-
tives. If one views the importance of current perquisites
relative to future pay as proxied by an executive’s age
(those early in their career are more concerned with future
pay), we would expect to see more (less) experienced
executives to be focused on revenue generation (resource
utilization). Interestingly, though, the temptation to cut
infrastructure spending to appear more ef?cient is
universal, i.e., applies equally to all stages of one’s career.
6 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
Using the equilibrium values in (6) and (8) together, the
career concerns equilibrium and associated welfare are
summarized in the following proposition.
Proposition 5. In the presence of career concerns, the
equilibrium with functional classi?cation of expenses entails:
^
F =
a
2
4
;
^
e
R
=
ks
j
2k
R
[s
h
÷s
j
[
÷
c[1 ÷ b[
2k
R
;
^
e
P1
=
[1÷k[s
w
2k
P1
[s
h
÷s
w
[
÷
c
2k
P1
if c<
[1÷k[s
w
s
h
÷s
w
; and
^
e
P1
=0; otherwise;
^
e
P2
=
[1÷k[s
w
2k
P2
[s
h
÷s
w
[
; and
^
W = W
+
÷
1
4
1
k
R
_
c(1 ÷ b) ÷ b ÷
ks
j
s
h
÷s
j
_ _
2
÷
1
k
P1
c ÷
s
h
÷ ks
w
s
h
÷s
w
_ _
2
÷
1
k
P2
[1 ÷ k[s
w
s
h
÷s
w
_ _
2
_
if
c <
[1 ÷ k[s
w
s
h
÷s
w
; and
^
W = W
+
÷
1
4
1
k
R
_
c(1 ÷ b) ÷ b ÷
ks
j
s
h
÷s
j
_ _
2
÷
1
k
P1
÷
1
k
P2
[1 ÷ k[s
w
s
h
÷s
w
_ _
2
_
; otherwise:
We conclude the results by examining if and how career
concern incentives are in?uenced by disclosure of the
functional classi?cation of expenses.
The role of functional cost classi?cation
There is no doubt that nonpro?t executives are driven
by an inherent desire to make a lasting impact on society.
That said, there is also little doubt that their incentives
move beyond intrinsic desires to extrinsic factors. Given
the notable absence of incentive pay in nonpro?ts, the
career concerns equilibrium derived above seeks to cap-
ture the two key extrinsic incentives faced by nonpro?t
executives: the desire to in?uence perceptions and future
pay, and the bene?t of current perquisite consumption.
It is these incentives many have in mind when they
question the ef?cacy of accounting rules that govern nonp-
ro?ts. The accounting rules stipulate the functional classi-
?cation of expenses between programs, fundraising, and
administration. The commonly expressed concern with
the accounting rules is that the functional classi?cation
comes with error (here, re?ected by r
2
w
> 0) and can be
biased (here, re?ected by
w –0 ). Further, the accounting
rules pool together all administrative costs, failing to
re?ect the distinction between those that are helpful (here,
A
I
) and those that are wasteful (here, A ÷ A
I
). These imper-
fections are believed to distort incentives of executives to
the detriment of the nonpro?ts’ missions. It is this question
we revisit in light of the career concerns equilibrium
derived herein. First, however, we derive the alternative
equilibrium, one arising in the absence of market reliance
on accounting cost classi?cation.
Relegating the details to the appendix, the next lemma
demonstrates the outcome without cost allocation.
Lemma 2. In the presence of career concerns, the equilibrium
without functional classi?cation of expenses entails:
~
F =
a
2
÷
aks
j
2c(1 ÷ b)(s
h
÷s
j
)
_ _
2
;
~
e
R
=
ks
j
2k
R
[s
h
÷s
j
[
÷
c[1 ÷ b[
2k
R
;
~
e
P1
=
~
e
P2
= 0; and
~
W = W
+
÷
1
4
1
k
R
_
c(1 ÷ b) ÷ b ÷
ks
j
s
h
÷s
j
_ _
2
÷
1
k
P1
÷
b
[c(1 ÷ b)[
2
aks
j
s
h
÷s
j
_ _
2
_
:
The ?rst thing to note in comparing Proposition 5 and
Lemma 2 is that the functional classi?cation of expenses
under accounting has no effect on revenue effort incen-
tives. Since the market relies only on observed revenues
(and not the estimate of their split) in inferring revenue-
generation skill, the effort levels are unaffected by either
the presence or the precision of the accounting estimates.
Fig. 2. In?uence of labor market and executive priorities on efforts.
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 7
The presence of the functional classi?cation, however,
does impact resource utilization effort. Since the market
gets no indication of resource utilization in the absence
of accounting estimates, market inferences do not generate
incentives to cut costs so as to boost program spending; in
fact, the extrinsic incentive is the opposite, since cost cuts
only reduce perks.
The ?nal choice that is at the executive’s discretion is
how much to expend on fundraising. Recall that with func-
tional expense classi?cation, the market observes fundrais-
ing expenditures and can discount them appropriately
when evaluating the realized revenues. Absent functional
classi?cation, however, the market only knows the reve-
nues generated and not how much of the nonpro?t’s own
resources were used in generating such resources. This cre-
ates an incentive to expend more on fundraising in order to
in?uence market perceptions of revenue-generation skill.
The market correctly anticipates such expenditures in
equilibrium, so it is, in a sense, an exercise in futility. These
ways in which accounting cost allocations can alter incen-
tives and their net effect on welfare are summarized in the
next proposition.
Proposition 6. Functional classi?cation of expenses leads to:
(i) higher helpful program (cost-cutting) effort e
P1
if and
only if c <
[1÷k[s
w
s
h
÷s
w
;
(ii) higher myopic program (cost-cutting) effort e
P2
;
(iii) equal revenue effort e
R
; and
(iv) lower fundraising expenditure F.
The proposition con?rms the potential upsides of func-
tional classi?cation: by shining a light on how money is
spent, the cost allocation exercise convinces executives to
shift a focus away from fundraising and wasteful adminis-
trative spending in favor of cost cutting. At the same time,
it also demonstrates the potential downsides of functional
expense classi?cation: by bringing attention to spending,
cost allocation lowers spending on key infrastructure since
it gets classi?ed as ‘‘overhead.’’ Jointly considering these
tradeoffs culminates in the following welfare comparison.
Proposition 7. Functional classi?cation of expenses leads to
higher welfare if and only if k
P2
> k
/
, where:
k
/
=
(1 ÷ k)s
w
s
h
÷s
w
_ _
2
b
[c(1 ÷ b)[
2
aks
j
s
h
÷s
j
_ _
2
_
÷
1
k
P1
2 ÷
(1 ÷ k)s
w
s
h
÷s
w
÷c
_ _
(1 ÷ k)s
w
s
h
÷s
w
÷c
_ __
÷1
if c <
[1 ÷ k[s
w
s
h
÷s
w
; and
k
/
=
(1 ÷ k)s
w
s
h
÷s
w
_ _
2
[c(1 ÷ b)[
2
b
_ _
s
h
÷s
j
aks
j
_ _
2
; otherwise:
As the proposition shows, the net welfare consequence
of functional classi?cation hinges on k
P2
, which re?ects the
executive’s relative dif?culty of cutting infrastructure
spending. The reason it proves critical is that if it is easy
to meet cost-cutting targets by reducing infrastructure
(k
P2
is small), functional classi?cation’s tendency to initiate
the ‘‘nonpro?t starvation cycle’’ is paramount. In that case,
since the program expense measure encourages the wrong
type of cost cutting, efforts to measure it are counterpro-
ductive. This characterization also means that concerns
about functional classi?cation undermining incentives to
invest in fundraising are less pressing. The reason is that
absent functional classi?cation, the incentive to invest in
fundraising is excessive – the executive cannot resist the
temptation to overinvest in fundraising in order to boost
revenues and the resulting impression of his ability to gen-
erate revenues. Functional classi?cation of expenses turns
off this lever for the executive, promoting a greater atten-
tion to ef?ciency. This feature, as well as the ability of func-
tional classi?cation to promote bene?cial cost-cutting
initiatives (reduction of administrative bloat) means that
despite its propensity to reduce useful infrastructure
spending, it can often still be bene?cial in toto.
Before concluding, we wish to acknowledge and discuss
the limitations of the career concerns framework
employed herein. While the standard career concerns
model provides a tractable and parsimonious way to
address incentives for multiple elements of effort and skill
among employees seeking to in?uence external percep-
tions, its linear framework also precludes examination of
interdependencies among such efforts and skills. Such lim-
itations mean that the present analysis excludes consider-
ation of circumstances where an executive’s diligent
efforts to ef?ciently use resources makes his task of raising
funds easier or vice-versa. To get a ?avor for how such
interdependencies between spending and fundraising can
alter the results, consider the paper’s analysis as before,
except that efforts to reduce spending and those to boost
fundraising are synergistic in that the cost of executives
efforts are reduced by ge
R
e
P1
, with g capturing the degree
of synergy. Further, in order to focus on how effort comple-
mentarities can affect effort choices, we set a = c = 0, i.e.,
fundraising is held ?xed and perquisite consumption by
the CEO is a non issue, and normalize b = 1. Finally, to
ensure interior solutions we presume 0 < g < 2
????????????
k
P1
k
R
_
,
with the limiting lower bound representing the original
setting. In this appended setting, the following proposition
shows the consequence of functional classi?cation on
career concerns efforts (closed form expressions are
presented in the appendix).
Proposition 8. With effort complementarity, functional clas-
si?cation of expenses yields:
(i) higher effort on all fronts:
^
e
R
(g) >
~
e
R
(g) > 0;
^
e
P1
(g) >
~
e
P1
(g) > 0;
and
^
e
P2
(g) >
~
e
P2
(g) = 0:
(ii) higher welfare if and only if k
P2
> k
/
(g), where
dk
/
(g)
dg
< 0.
Note from Proposition 8(i) that effort complementarity
introduces two additional features: efforts to cut wasteful
spending are positive even absent functional expenses
(~e
P1
(g) > 0), and functional expense classi?cation also
increases efforts to raise revenues (^e
R
(g) > ~e
R
(g)). Despite
8 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
the differences, the primary message that functional classi-
?cation of expenses increases efforts to in?uence percep-
tions, though such efforts can be either productive or
harmful, remains. As Proposition 8(ii) shows, this again
translates into a straightforward welfare comparison in
that the less the concern that the executive will reduce
infrastructure spending, the more attractive is functional
classi?cation from a welfare perspective. Further, the
greater the complementarity between these efforts, the
less prominent reduction in infrastructure spending
becomes and, thus, the more attractive functional classi?-
cation of expenses, as evidenced by
dk
/
(g)
dg
< 0.
The broader point here is not that introducing comple-
mentarities necessarily favors functional classi?cation of
expenses, but rather that it can add subtle additional con-
siderations. Such subtleties can be compounded if one con-
siders multi-period interactions. In that case, it seems
reasonable to presume that diligent cost-cutting in one year
can make for an easier fundraising pitch in a future year, i.e.,
a higher e
P1
can promote a higher future a. Alternatively,
one can argue that greater administrative spending in one
year can boost subsequent years’ ability to effectively use
resources, i.e., a lower e
P2
can promote a higher future b.
These considerations suggest that depending on the nature
of the interactions some efforts may prove more important
to welfare in the long-run than others. While our single-
period analysis does not fully capture these forces, perhaps
it gives a window into what these forces would mean. After
all, we showthat the more prominent career incentives are,
the greater the efforts to cut costs (both e
P1
and e
P2
are
higher under functional classi?cation). So, if cost-cutting
this year can boost fundraising next year, it points to
another upside of functional cost allocation; on the other
hand, if reduction in administrative infrastructure this year
proves particularly harmful in the long-run, our results sug-
gest a poignant downside of cost allocation.
Conclusion
Though nonpro?t executives are typically viewed as
driven more by intrinsic motivation than their for-pro?t
counterparts, there is little doubt that they are also moti-
vated to a degree by external perceptions and pay. For this
reason, many in the nonpro?t sector have condemned the
functional expense classi?cation embedded in accounting
rules for unduly creating myopic incentives for executive
behavior. The contention is that by creating imperfect
and often biased measures of how effectively nonpro?ts
have devoted resources to their mission, accounting state-
ments have led executives to engage in a ‘‘nonpro?t starva-
tion cycle’’, reducing investment in advertising and
infrastructure in order to in?uence external perceptions
of ef?ciency (see, e.g., Gregory & Howard, 2009).
This paper provides a formal model of incentives of
nonpro?t executives when they are concerned with exter-
nal perceptions. The model permits an examination of both
executive incentives to in?uence perceptions and the role
of accounting estimates therein. We show that even with
their inherent bias and noise, accounting estimates can
prove in?uential in executive pay and actions. Also
consistent with accounting detractors, we show that
functional classi?cation of expenses shifts resources away
from fundraising and administration toward immediate
program-based expenditures. Despite the emphasis shift
and the downsides it can bring, we also show that these
measures and the incentives they induce can improve
resource allocation and overall welfare.
Acknowledgements
We thank Ramji Balakrishnan, John Christensen, Peter
Christensen, John Fellingham, Hans Frimor, Jon Glover,
Robert Goex, Lisa Koonce, Eva Labro, Dan Pallotta, Christine
Petrovits, Jonathan Ross, Naomi Rothenberg, Doug Schroe-
der, Karen Sedatole, Jack Stecher, Austin Sudbury, Rick
Young, Chris Chapman (Editor) and two anonymous refer-
ees, and workshop participants in Arizona State University,
Binghamton University, Carnegie Mellon University,
University of Colorado, University of Iowa, Ohio State
University, and EIASM workshop on Accounting and
Economics. Anil Arya gratefully acknowledges assistance
from the John J. Gerlach Chair.
Appendix A
Proof of Lemma 1. The welfare measure is:
W = E¦P ÷ A
I
¦ ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
= E¦b[R ÷ F[ ÷ h
P
÷ e
P1
÷ e
P2
÷ w ÷ I ÷ e
P2
¦
÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
= E¦b[a
???
F
_
÷ h
R
÷ e
R
÷j ÷ F[ ÷ h
P
÷ e
P1
÷ w ÷ I¦
÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
= b[a
???
F
_
÷
h ÷ e
R
÷
j ÷ F[ ÷
h ÷ e
P1
÷
w
÷ I ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
: (A1)
The derivative of W with respect to F, e
R
, e
P1
, and e
P2
,
respectively, are:
@W
@F
= b
a
2
???
F
_ ÷1
_ _
;
@W
@e
R
= b ÷2k
R
e
R
;
@W
@e
P1
= 1 ÷2k
P1
e
P1
; and
@W
@e
P2
= ÷2k
P2
e
P2
: (A2)
From the last derivative in (A2), e
+
P2
= 0. Setting each of the
?rst three derivatives in (A2) equal to zero, and solving the
equations for F, e
R
, and e
P1
, yields the solution in Lemma 1.
Substituting this solution in (A1) yields the corresponding
welfare value. h
Proof of Proposition 1. 1We ?rst formally derive the
expression for wages presented in (4). Since A = R ÷ F ÷ P,
the information content associated with the market learn-
ing (R, F, P, A) is equivalent to it learning (R, F, P) and, thus,
we use this smaller set in characterizing the market’s
updated beliefs of the executive’s two skill dimensions
(and, thus, his wage). To conduct this updating, we make
use of a standard result (e.g., Greene, 1997, p. 90): if
X =
X
1
X
2
_ _
is multivariate normal with mean l =
l
1
l
2
_ _
,
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 9
covariance matrix R =
R
11
R
12
R
21
R
22
_ _
, and R
÷1
22
is the inverse
of R
22
(the pseudo inverse if R
22
is singular), then the con-
ditional mean of X
1
, given X
2
= x
2
, is:
E¦X
1
x
2
[ ¦ = l
1
÷ R
12
R
÷1
22
(x
2
÷l
2
): (A3)
With X
1
=
h
R
h
P
_ _
and X
2
=
R
F
P
_
_
_
_
, l
1
=
h
h
_ _
,
l
2
=
a
???
F
_
÷
h ÷ e
c
R
÷
j
F
b[a
???
F
_
÷
h ÷ e
c
R
÷
j ÷ F[ ÷
h ÷ e
c
P1
÷ e
c
P2
÷
w
_
_
_
_
,
R
12
=
r
2
h
0 br
2
h
0 0 r
2
h
_ _
, and
R
22
=
r
2
h
÷r
2
j
0 b(r
2
h
÷r
2
j
)
0 0 0
b(r
2
h
÷r
2
j
) 0 b
2
(r
2
h
÷r
2
j
) ÷r
2
h
÷r
2
w
_
_
_
_
; here, e
c
R
,
e
c
P1
, and e
c
P2
denote the market’s conjecture of the execu-
tive’s efforts. Substituting this in (A3) yields:
E¦X
1
x
2
[ ¦ =
E¦h
R
R; F; P¦ [
E¦h
P
R; F; P¦ [
_ _
=
r
2
j
r
2
h
÷r
2
j
_ _
h÷
r
2
h
r
2
h
÷r
2
j
_ _
R÷a
???
F
_
÷e
c
R
÷
j
_ _
r
2
w
r
2
h
÷r
2
w
_ _
h÷
r
2
h
r
2
h
÷r
2
w
_ _
P÷b(R÷F) ÷e
c
P1
÷e
c
P2
÷
w
_ _
_
¸
¸
_
_
¸
¸
_
:
(A4)
Using (A4), the wage is w = kE¦h
R
[R; F; P¦ ÷ [1 ÷ k[E¦h
P
[R; F; P¦. Substituting s
i
= 1=r
2
i
, i = h, j, w, this wage can be
expressed as:
w=k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
R÷a
???
F
_
÷e
c
R
÷ j
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
P÷b(R÷F) ÷e
c
P1
÷e
c
P2
÷
w
_ _
_ _
:
(A5)
From (A5), the sensitivity of pay to revenue is:
@w
@R
=
ks
j
s
h
÷s
j
÷
[1 ÷ k[bs
w
s
h
÷s
w
: (A6)
The proposition then follows by noting the sign of the
derivatives of (A6) with respect to k and s
w
, respectively,
as shown below:
@w=@R
@k
=
s
j
s
h
÷s
j
÷
bs
w
s
h
÷s
w
> 0 and
@w=@R
@s
w
= ÷
[1 ÷ k[bs
h
[s
h
÷s
w
[
2
< 0:
h
Proof of Proposition 2. From(A5), the sensitivity of pay to
program expenses is:
@w
@P
=
[1 ÷ k[s
w
s
h
÷s
w
: (A7)
The proposition then follows by noting the sign of the
derivatives of (A7) with respect to k and s
w
, respectively,
as shown below:
@w=@P
@k
= ÷
s
w
s
h
÷s
w
< 0 and
@w=@P
@s
w
=
[1 ÷ k[s
h
[s
h
÷s
w
[
2
> 0:
h
Proof of Proposition 3. Given e
R
, F, and R, and wage in
(A5), the executive chooses e
P1
and e
P2
to solve:
Max
e
P1
;e
P2
E
h
P
;w
w e
R
; F; R [ ¦ ¦ ÷ k
R
e
2
R
÷ k
P1
e
2
P1
÷ k
P2
e
2
P2
÷cE
h
P
;w
R ÷ F ÷ P ÷ A
I
e
R
; F; R [ ¦ ¦: (A8)
Using (A5), P = b[R ÷ F] + h
P
+ e
P1
+ e
P2
+ w, and A
I
= I ÷ e
P2
,
(A8) can be written as:
Max
e
P1
;e
P2
k
s
h
s
h
÷s
j
_ _
h÷
s
j
s
h
÷s
j
_ _
R÷a
???
F
_
÷e
c
R
÷
j
_ _
_ _
÷[1÷k[
s
h
s
h
÷s
w
_ _
h÷
s
w
s
h
÷s
w
_ _
h÷e
P1
÷e
P2
÷e
c
P1
÷e
c
P2
_ _
_ _
÷k
R
e
2
R
÷k
P1
e
2
P1
÷k
P2
e
2
P2
÷c[(1÷b)(R÷F) ÷
h÷e
P1
÷
w÷I[:
(A9)
The derivative of (A9) with respect to e
P1
equals:
[
^
c ÷c[ ÷2k
P1
e
P1
; where
^
c =
[1 ÷ k[s
w
s
h
÷s
w
: (A10)
From (A10), if c _
^
c, the executive’s payoff is decreasing
in e
P1
, so e
P1
= 0; else, e
P1
is interior and is obtained by set-
ting (A10) equal to 0. That is,
^
e
P1
=
^
c ÷c
2k
P1
if c <
^
c; and
^
e
P1
= 0; otherwise:
(A11)
Also, from (A9), the ?rst-order condition with respect to e
P2
equals
^
c ÷2k
P2
e
P2
= 0. Solving this yields:
^
e
P2
=
^
c
2k
P2
: (A12)
Using (A11), and c <
^
c, the results then follow from the
sign of the following derivatives:
@
^
e
Pi
@k
=÷
s
w
2k
Pi
[s
h
÷s
w
[
0; and
@
^
e
R
@b
=÷
c
2k
R
0;
^
e
R
÷
~
e
R
= 0; and
^
F ÷
~
F = ÷
a
2
ks
j
[2c(1 ÷ b)(s
h
÷s
j
) ÷ ks
j
[
[2c(1 ÷ b)(s
h
÷s
j
)[
2
< 0: (A26)
(A26) proves Proposition 6. h
A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12 11
Proof of Proposition 7. Comparing the welfare expres-
sions in Proposition 5 and Lemma 2,
^
W ÷
~
W =
1
4
a
2
bk
2
s
2
j
(1 ÷ b)
2
c
2
(s
h
÷s
j
)
2
_
÷
(1 ÷ k)
2
s
2
w
k
P2
(s
h
÷s
w
)
2
÷
1
k
P1
2 ÷
(1 ÷ k)s
w
s
h
÷s
w
÷c
_ _
(1 ÷ k)s
w
s
h
÷s
w
÷c
_ __
if c <
[1 ÷ k[s
w
s
h
÷s
w
; and
^
W ÷
~
W =
1
4
a
2
bk
2
s
2
j
(1 ÷ b)
2
c
2
(s
h
÷s
j
)
2
÷
(1 ÷ k)
2
s
2
w
k
P2
(s
h
÷s
w
)
2
_ _
; otherwise:
(A27)
Setting
^
W ÷
~
W in (A27) equal to 0, and solving for k
P2
yields the value of k
/
. This proves Proposition 7. h
Proof of Proposition 8. With functional classi?cation of
expenses, the solution now entails ?rst solving (5) and,
then, (7) with the term ge
R
e
P1
added to both objective
functions (with a = c = 0 and b = 1). This yields the follow-
ing analog to Proposition 5:
^
e
R
(g) =
1
4k
P1
k
R
÷g
2
2k
sj
s
h
÷sj
_ _
k
P1
÷ (1 ÷ k)
s
w
s
h
÷s
w
_ _
g
_ _
;
^
e
P1
(g) =
1
4k
P1
k
R
÷g
2
k
sj
s
h
÷sj
_ _
g ÷2(1 ÷ k)
s
w
s
h
÷s
w
_ _
k
R
_ _
;
^
e
P2
(g) =
1
2k
P2
(1÷k)s
w
s
h
÷s
w
_ _
; and
^
W(g) = 2
h ÷
j ÷
w ÷ I ÷
^
e
R
÷
^
e
P1
÷ k
R
^
e
2
R
÷ k
P1
^
e
2
P1
÷ k
P2
^
e
2
P2
÷g
^
e
R
^
e
P1
: (A28)
In the absence of functional classi?cation of expenses, the
solution is obtained by solving (A20) and (A23) with the
term ge
R
e
P1
again added to both the objective functions.
This yields the following analog to Lemma 2:
~
e
R
(g) =
1
4k
P1
k
R
÷g
2
2k
sj
s
h
÷sj
_ _
k
P1
_ _
;
~
e
P1
(g) =
1
4k
P1
k
R
÷g
2
k
sj
s
h
÷sj
_ _
g
_ _
;
~
e
P2
(g) = 0; and
~
W(g) = 2
h ÷
j ÷
w ÷ I ÷
~
e
R
÷
~
e
P1
÷ k
R
~
e
2
R
÷ k
P1
~
e
2
P1
÷g
~
e
R
~
e
P1
: (A29)
Part (i) follows immediately by comparing the effort levels
in (A28) with the corresponding effort levels in (A29).
Turning to welfare comparison in part (ii),
^
W(g)÷
~
W(g)=
s
w
[1÷k[
4k
P2
[s
h
÷s
w
[[4k
P1
k
R
÷g
2
[
× 4k
P2
g 1÷
s
j
k
s
h
÷s
j
_ _ _ _
÷k
R
2÷
s
w
(1÷k)
s
h
÷s
w
_ __
÷
s
w
[1÷k[[4k
P1
k
R
÷g
2
[
s
h
÷s
w
_
:
(A30)
From (A30),
^
W(g) ÷
~
W(g) > 0 if and only if k
P2
> k
/
(g),
where:
k
/
(g) =
s
w
(1 ÷ k)(4k
P1
k
R
÷g
2
)
s
h
÷s
w
_ __
4g 1 ÷
s
j
k
s
h
÷s
j
_ _ _
÷4k
R
2 ÷
s
w
(1 ÷ k)
s
h
÷s
w
_ __
: (A31)
The numerator in the right-hand-side of (A31) is decreas-
ing in g while the denominator is increasing in g. Thus,
dk
/
(g)
dg
< 0, completing the proof of the proposition. h
References
Arya, A., Frimor, H., & Mittendorf, B. (2010). The bene?ts of aggregate
performance metrics in the presence of career concerns. Management
Science, 57, 1424–1437.
Auriol, E., Friebel, G., & Pechlivanos, L. (2002). Career concerns in teams.
Journal of Labor Economics, 20, 289–307.
Autrey, R., Dikolli, S., & Newman, P. (2007). Career concerns and
mandated disclosure. Journal of Accounting and Public Policy, 26,
527–554.
Autrey, R., Dikolli, S., & Newman, P. (2010). Performance measure
aggregation, career incentives, and explicit incentives. Journal of
Management Accounting Research, 22, 115–131.
Baber, W., Daniel, P., & Roberts, A. (2002). Compensation to managers of
charitable organizations: An empirical study of the role of accounting
measures of program activities. The Accounting Review, 77, 679–693.
Dewatripont, M., Jewitt, I., & Tirole, J. (1999a). The economics of career
concerns, Part 1: Comparing information structures. Review of
Economic Studies, 66, 183–198.
Dewatripont, M., Jewitt, I., & Tirole, J. (1999b). The economics of career
concerns, Part 2: Application to missions and accountability of
government agencies. Review of Economic Studies, 66, 199–217.
Frumkin, P., & Keating, K. (2010). The price of doing good: Executive
compensation in nonpro?t organizations. Policy and Society, 29,
269–282.
Greene, S. (1997). Econometric analysis. Upper Saddle River, NJ: Prentice
Hall.
Gregory, A., & Howard, D. (2009). The nonpro?t starvation cycle. Stanford
Social Innovation Review; Fall (pp. 49–53).
Holmstrom, B. (1982). Managerial incentives – A dynamic perspective. In
Essays in economics and management in Honor of Lars Wahlbeck.
Helsinki, Finland: Swedish School of Economics.
Holmstrom, B. (1999). Managerial incentive problems: A dynamic
perspective. Review of Economic Studies, 66, 169–182.
Holmstrom, B., & Ricart i Costa, J. (1986). Managerial incentives and
capital management. Quarterly Journal of Economics, 101, 835–860.
Kaarboe, O., & Olsen, T. (2006). Career concerns, monetary incentives, and
job design. Scandinavian Journal of Economics, 108, 299–316.
Krishnan, R., & Yetman, M. (2011). Institutional drivers of reporting
decision in nonpro?t hospitals. Journal of Accounting Research, 49,
1001–1039.
Krishnan, R., Yetman, M., & Yetman, R. (2006). Expense misreporting in
nonpro?t organizations. The Accounting Review, 81, 399–420.
Milbourn, T., Shockley, R., & Thakor, A. (2001). Managerial career concerns
and investments in information. RAND Journal of Economics, 32,
334–351.
Pallotta, D. (2010). Uncharitable: How restraints on nonpro?ts undermine
their potential. Tufts University Press.
Sedatole, K., Swaney, A., Yetman, M., and Yetman, R. (2013). Accounting-
based performance metrics and executive compensation in nonpro?t
organizations. Michigan State University working paper.
Shleifer, A., & Wolfenzon, D. (2002). Investor protection and equity
markets. Journal of Financial Economics, 66, 3–27.
Stern, K. (2013). With charity for all: Why charities are failing and a better
way to give. Doubleday.
12 A. Arya, B. Mittendorf / Accounting, Organizations and Society 40 (2015) 1–12
doc_265634350.pdf