Professional Documents
Culture Documents
Bhagwan Chowdhry
University of California, Los Angeles
Brian Waters
University of Colorado, Boulder
We study joint financing between profit-motivated and socially motivated (impact) investors
and derive conditions under which impact investments improve social outcomes. When
project owners cannot commit to social objectives, impact investors hold financial claims to
counterbalance owners’ tendencies to overemphasize profits. Impact investors’ ownership
stakes are higher when the value of social output is higher, and pure nonprofit status may
be optimal for the highest valued social projects. We provide guidance about the design
of contingent social contracts, such as social impact bonds and social impact guarantees.
(JEL M14, O35, D86)
Received May 23, 2016; editorial decision April 28, 2018 by Editor Francesca Cornelli.
Authors have furnished an Internet Appendix, which is available on the Oxford University
Press Web site next to the link to the final published paper online.
We would like to thank Maitreesh Ghatak, Robert Gertner, Jesse Davis, Archishman Chakraborty, Ivo Welch,
Edward Van Wesep, Robert Dam, Kyle Matoba, and Francesca Cornelli; two anonymous referees; and seminar
participants at CU-Boulder Leeds School of Business, UCLA Anderson School of Management, the HKUST
Conference on the Impact of Responsible and Sustainable Investing, the Emerging Markets Finance Conference,
the Geneva Summit on Sustainable Finance, The University of Chicago Economics of Social Sector Organizations
Workshop, and the Midwest Finance Association annual meeting for their helpful insights and suggestions.
Supplementary data can be found on The Review of Financial Studies Web site. Send correspondence to Bhagwan
Chowdhry, Anderson School of Management, University of California, Los Angeles, 110 Westwood Plaza Suite
C-411, Los Angeles, CA 90095; telephone: (310)-825-5883. E-mail:bhagwan@anderson.ucla.edu.
© The Author(s) 2018. Published by Oxford University Press on behalf of The Society for Financial Studies.
All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
doi:10.1093/rfs/hhy068 Advance Access publication July 4, 2018
We model a project that produces both a monetary payoff and a social benefit
and consider settings in which there is a trade-off regarding which output to
emphasize. While the for-profit project owner is agnostic toward the project’s
social goals, a social impact investor, who values (and is willing to pay for)
the project’s expected social output, also contributes to the project’s cost of
investment. In doing so, the impact investor provides the for-profit owner a
subsidy for directing scarce resources and attention to the production of the
1 For example, the underlying project may be an initiative to improve employee health at a firm. Employee health
is improved through more expansive health insurance and greater emphasis on preventive care and healthy living.
865
well by a social impact investor. Because funds available for social good are
relatively scarce and socially minded investors face additional constraints in
their portfolio selection problem, the impact investor has a higher opportunity
cost of using capital than does the for-profit owner. In addition, the for-profit
owner employs a manager who allocates unobservable scarce attention between
the two production technologies. While the for-profit owner is agnostic toward
social output, the social impact investor and manager are, to some extent,
The two outputs from the project are (1) a monetary payoff (cost savings) to the firm via lower turnover and a
reduction in sick leave and (2) a social benefit that stems from improvements in long-term health outcomes. The
firm may stress greater sick patient care in order to reduce employee leave, but do little to affect overall health
outcomes or life expectancy. Conversely, the firm could provide nutrition counseling and subsidized gym access
to reduce the likelihood of diabetes and heart disease but do little to reduce employee sick days due to common
illnesses.
866
purchase equity in socially focused firms. Impact funds, which have an explicit
mandate to invest in sustainable and responsible opportunities, increased their
investments in public equity from 276 million dollars in 2013 to nearly 1.5
billion dollars in 2015 (GII 2016). We show, as well, that impact investors
overpay for financial claims (to subsidize for-profit owners) and, thus, earn
lower expected financial returns. This result is consistent with recent studies
that document that social impact investors pay a premium for holding socially
2 See Geczy, Stambaugh, and Levin (2005), Hong and Kacperczyk (2009), Chava (2014), and Hartzmark and
Sussman (2017).
867
We show, as well, that the social impact investor should hold the opposite
contract; a guarantee-against-social-failure contract which guarantees a large
financial payment when the project’s social value is small. We refer to this
security as a social impact guarantee. While financial contracting on social
outcomes in the for-profit sector is rare to date, our model shows that for-
profit owners in socially valuable firms should issue social impact guarantees to
impact investors so that the owners’ own residual claims provide the necessary
3 See Baron (2001), Bagnoli and Watts (2003), and Kotchen (2006) for settings in which profitability and social
good production are complements.
4 See also Hansmann (1996).
5 See also Tirole (2001) for an analysis of whether managerial incentives and control structures should be used to
promote stakeholder rights.
868
1. The Model
We consider a project which may produce both a monetary payoff and a social
benefit — a “social venture.” The project requires an upfront investment equal
to I > 0 supplied by an investor(s) and a single unit of labor supplied by a
manager. If the project is undertaken, the manager allocates an unobservable
scarce resource between two production technologies, one which aids in the
production of the social benefit and another that aids in the production of the
monetary payoff. For ease of exposition, we refer to the unobservable resource
as attention, and we denote by ab ≥ 0 the fraction of attention allocated to
the social technology and by ax ≥ 0 the fraction of attention allocated to the
traditional for-profit technology (where ab +ax = 1).8
The project may either “succeed” or “fail” in the production of the monetary
payoff. When the project succeeds, it generates a payoff x = X > 0, whereas
6 See also Baron (2008) for a setting in which corporate governance is used to further social goals.
7 Geczy, Stambaugh, and Levin (2005) provide similar evidence; investors that adhere to socially responsible
investments add an additional constraint to their portfolio selection problem and underperform relative to an
unconstrained portfolio by as much as 30 basis points per month.
8 The scarce resource also could be interpreted in the context of managerial effort or the distribution of resources
within firms’ internal capital markets.
869
when the project fails, its payoff is x = 0. The probability of a successful outcome
depends on the fraction of attention allocated to the for-profit technology and
is given by
Pr(x = X| ax ) = f (ax ), (1)
P r (b = 1| ab ) = g(ab ), (2)
Assumption 1.
9 Our setup therefore assumes that conditional on the chosen levels of attention, the success or failure of the social
technology is independent of the success or failure of the for-profit technology.
10 In our Online Appendix, we consider the alternative case in which the impact investor owns the project.
11 Equivalently, the for-profit owner is unwilling to pay for social output, whereas the impact investor and manager
are willing to pay ψs and ψm for a successful social outcome.
870
xi +ψi b
ki + , (3)
ρi
12 Regarding (a), despite recent growth in sustainable, responsible, and impact investments, the pool of dollars
directed toward social projects is still limited in comparison to the amount of funds commercially available
for investments that generate market returns (GII 2017). Regarding (b), in subsequent analysis, we show that
the impact investor must maintain a concentrated stake in a limited number of social ventures in order to
align incentives and increase social investment. As such, impact investors may be underdiversified relative to
profit-motivated investors.
13 This assumption is consistent with a contactual incompleteness. If the manager can walk away from the project
after accepting the upfront transfer, such transfers cannot be optimal.
14 We provide the explicit conditions on γ in Assumption 2 in the appendix.
15 Our contracting environment is consistent with settings featuring moral hazard and limited liability. See Innes
(1990).
871
share, and z ≥ 0 is the equity share retained by the for-profit owner (and in
which w +y +z ≤ 1).16
If the manager accepts the wage contract, she chooses a ∈ [0,1] to maximize
her expected payoff:
a ∈ arg max E xw +ψm b|a . (5)
a ∈[0,1]
For a given wage contract w, the manager allocates her attention across the
for-profit and social technologies in order to maximize her expected payoff.
When the wage contract is higher, the manager benefits relatively more from
the production of the monetary payoff and therefore shifts attention away from
social production in favor of greater attention toward financial returns.17
Since the function a(w) is differentiable and strictly decreasing, the inverse
function w(a) = a−1 (w), which maps social attention to the manager’s wage
contract, exists and is differentiable and strictly decreasing.
Lemma 2. The inverse function,
g (a)ψm
w(a) = a−1 (w) = , (6)
f (1−a)X
is a one-to-one continuous and differentiable function. The function w(a) is
strictly decreasing in a.
Following Holmstrom (1984), the first-order approach is valid for the manager’s
attention choice problem. The inverse function which maps social attention to
wages therefore follows from the manager’s first-order condition for optimal
attention. While the stated assumptions for the production functions g(·) and
f (·) guarantee a unique solution to the manager’s attention allocation problem
for a given w, we assume as well that the for-profit owner’s expected payoff is
concave in attention, and thus that the second-order condition for optimality is
satisfied.
16 Since x = 0 when the project fails to produce the monetary payoff, the manager’s wage and cash flow sharing rule
can be equivalently represented as debt contracts.
17 For example, what we have in mind is increasing managerial wages to mitigate the agency cost associated with
corporate social responsibility (CSR). This agency cost is due to firm managers shifting resources away from
profitable uses and toward nonpecuniary CSR activities that provide social benefit. See Cheng, Hong, and Shue
(2016) and Krüger (2015) for empirical studies examining CSR as an agency conflict.
872
873
participation constraint (7.2) binds. Then, for γ < γo , the optimal level of social
attention ao is defined implicitly by,18
dw
−f (1−ao )X(1−w(ao ))−f (1−ao )X = 0, (8)
da a=ao
In either case, the optimal wage contract must satisfy the manager’s incentive
compatibility constraint (7.1) and is therefore given by wo = w(ao ). Moreover,
ao is the lower bound on a; any optimal contract featuring the involvement of
the impact investor achieves an attention allocation at least as large as ao .
In our subsequent analysis of impact investing, we compare equilibrium
levels of social attention to the commercial benchmark ao . If equilibrium
attention exceeds the commercial benchmark, impact investing leads to greater
emphasis of social goals. Otherwise, impact investing is ineffective. While
the first-best level of social attention, which maximizes joint surplus among
the for-profit owner, the project manager, and the social impact investor, is
straightforward,
ψm ψb
−f (1−aF B )X +g (aF B ) + = 0, (10)
ρm ρb
our model does not provide clear predictions comparing equilibrium levels of
social attention to the first-best allocation aF B . To see this, consider the simplest
case in which the manager’s participation constraint is slack. Since attention is
not observable, the owner must provide the manager a cash flow claim in order
to reduce attention to social goals. Since paying the manager a higher wage
is costly for the for-profit owner, the owner prefers a level of social attention
above the first-best allocation in order to lower the manager’s wage, all else
equal. On the other hand, a high level of social attention is also costly for the
for-profit owner, since it requires the impact investor to hold a larger financial
claim (see Section 2). In this case, the owner prefers a level of social attention
below the first-best allocation, all else equal. Ultimately, it is unclear which
effect dominates.
Since the relationship between equilibrium levels of social attention and
the first-best allocation are unclear, we consider, as well, a setting in which
the for-profit owner can commit not to alter the manager’s equity wage after
18 To focus on projects for which there is a tension between the production of the monetary payoff and social benefit,
we assume that lim f (1−a) dw da is finite valued. The assumption ensures that the commercial benchmark level
a→1
of social attention is strictly less than 1.
874
the transfer ks is sunk. In the next section, commitment is limited, and the
optimal contract must be robust to renegotiation between the for-profit owner
and manager. While outside the scope of the current model, one can envision
the full-commitment setting as one in which the for-profit owner engages
in repeated interactions with social investors and must therefore maintain a
reputation for achieving sufficient social value. For the purposes of this paper,
the full-commitment model highlights the basic contracting frictions among the
where we have substituted for the security’s price ks from (11) into the for-
profit owner’s expected profit in (12). The following lemma characterizes the
profit-maximizing contract when the for-profit owner can credibly promise not
to renegotiate the manager’s wage contract (where we use the subscript v to
denote his “vow” to not renegotiate the contract).
Lemma 3. There exists a unique contract {(wv ,yv ,zv ),av } which maximizes
the for-profit owner’s expected payoff. The contract is characterized by wv =
w(av ), yv = 0, and zv = 1−wv .
875
1. For γ < γo , the level of social attention exceeds that in the commercial
benchmark, av > ao .
2. For γ ≥ γo , social attention equals the commercial benchmark, av = ao .
Lemma 3 shows that, when the for-profit owner can commit not to alter the
manager’s wage contract (perhaps in order to maintain his reputation as ethical
among potential future social investors), the profit-maximizing contract sets
the social impact investor’s equity claim to zero. Since the impact investor
discounts future payments at a higher rate than the for-profit owner, any
monetary payment promised to the impact investor at date 1 is inefficient.
Hence, the for-profit owner chooses y to be as small as possible, setting yv = 0.
While the impact investor holds none of the project’s equity, he still pays for
his expected value of social good through the upfront transfer ks = g(a)ψ
1+δ
s
, and it
is through this subsidy that the impact investor convinces the for-profit owner
to implement a higher level of social attention. We therefore interpret the
optimal contract in this setting with full commitment as a grant or a donation
provided by the impact investor to the for-profit owner. We now transition to
the more natural setting with limited commitment and show that grants and
donations are ineffective when commitment is limited. Instead, the impact
investor must hold a nonzero cash flow claim in order to affect the level of
attention directed toward social goals. Thus, Section 2 provides a theory of
impact investments in for-profit firms.
876
satisfies limited liability (w +z ≤ 1−y) and which makes both the for-profit
owner and manager better off (with at least one strictly better off):
E xz a = a w ≥ E xz a = a(w) ,
E xw +ψm b a = a w ≥ E xw +ψm b a = a(w) .
We refer to a contract which cannot be renegotiated by the for-profit owner and
To see this, observe that renegotiation increases the manager’s expected payoff
only if the new wage w is larger than the initial wage w. Under the initial
contract, the manager chooses a = a(w) to maximize her expected payoff:
f (1−a(w))Xw +g(a(w))ψm .
An increase in the wage contract therefore leads to an increase in the manager’s
payoff,
da
f (1−a(w))X + −f (1−a(w))Xw +g (a(w))ψm > 0,
dw
Equals zero
by envelope condition
while a decrease in the wage contract reduces the manager’s payoff. As such,
renegotiation is feasible only if the new contract increases the manager’s wage
and subsequently reduces the level of social attention.
877
y ≥ y(a), (14.4)
where we have again substituted for the upfront transfer ks from (11) into the
for-profit owner’s expected profit in (14) and where constraint (14.4) ensures
that the impact investor’s equity claim is large enough so that the contract is
renegotiation-proof.
Proposition 1. There exists a unique optimal contract {(wr ,yr ,zr ),ar }
which maximizes the for-profit owner’s expected payoff and which is
renegation-proof. The contract is characterized by wr = w(ar ), yr = y(ar ), and
zr = 1−wr −yr .
1. For ψs > ψs (and γ < γo ), the social impact investor holds a nonzero
equity claim, yr > 0, and social attention exceeds the benchmark level,
ar > ao . The unique threshold ψs is given by
dy
δf (1−ao )X da
a=ao
ψs = . (15)
g (ao )
2. Otherwise, yr = 0 and ar = ao .
19 Although we cannot distinguish between debt and equity in our model (since x = 0 when the project fails to
produce the monetary payoff), we conjecture that, in a more general model with outcomes XH > XL > 0, an
equity claim sold to the impact investor more strongly dulls the for-profit owner’s incentive to focus on monetary
payoffs than does an equally sized debt claim. As such, our model is consistent with the rapid growth in socially
targeted equity investments in recent years (GIIN 2016).
878
The first part of the proposition shows that the impact investor’s equity stake
is set just high enough to satisfy renegotiation-proofness: yr = y(ar ). Increasing
the impact investor’s claim y is costly for the for-profit owner, since the impact
investor has a higher cost of capital. To maximize his expected profit, the
for-profit owner therefore sets y as small as possible without violating
constraint (14.4).
In choosing the optimal level of social attention, the manager therefore faces
a trade-off between increasing the upfront subsidy from the impact investor
(higher y and higher a) and limiting the inefficiency that arises from the impact
investor’s equity stake (lower y and lower a). When the impact investor places
a sufficiently high value on a successful social outcome, that is, ψs > ψs , the
benefit from increasing the subsidy is large enough to offset the cost of a nonzero
equity claim. As such, the for-profit owner chooses ar > ao and yr > 0. However,
when the impact investor’s value for social output is limited, the effect on the
subsidy from an increase in a is small relative to the loss from selling equity to
the impact investor. Thus, for ψs ≤ ψs , the for-profit owner chooses yr = 0 and
settles for the commercial benchmark level of social attention ar = ao .
For the impact investor to hold a nonzero equity stake, not only must the
impact investor have a large enough value of social output ψs > ψs but also the
manager’s reservation payoff must be small enough γ < γo . Figure 3 illustrates
this point. When the manager’s reservation payoff is large, the manager requires
a large wage contract in order to participate in the social venture. As such, there
is no flexibility to reduce the manager’s wage and increase the level of attention
allocated to social goals. Importantly, the manager’s value for social output ψm
is positively related to the threshold γo . That is, all else equal, a larger value
of ψm increases the manager’s expected value of employment and raises the
879
dyr dyr
Proposition 1 shows that impact investors hold cash flow claims only in
those firms with significant social value. In this sense, our model provides a
theory of directed or exclusionary investment among socially minded investors.
While exclusionary investment campaigns are sometimes framed in the context
of protesting or screening certain companies with low social value, our model
suggests exclusionary investment also may arise as impact investors concentrate
equity positions only in those firms for which joint funding optimally improves
social outcomes. In addition, Corollary 1 shows that impact investors hold
larger equity stakes in more socially valuable firms. Since the size of the upfront
subsidy depends on the impact investor’s expected social value, an increase in
ψs raises the marginal benefit of an increase in social attention to the for-profit
owner. In response, the for-profit owner prefers a higher level of social output
which requires a larger equity sale in order to make the contract renegotiation-
proof. As such, the model predicts that impact investors hold a larger fraction
of firm equity in more socially valuable firms and that social output be higher
and firm profitability lower in such firms (since a is higher).
In addition, the impact investor’s equity stake is decreasing in the impact
investor’s added cost of capital δ. Recent studies suggest that the pool of socially
conscious capital has increased dramatically in recent years (Barber, Morse,
and Yasuda (2017)) and that the millennial generation is more pre-disposed to
investing in firms that focus on social impact. As the opportunity cost of socially
conscious capital declines, the model suggest impact investors will hold larger
ownership stakes in for-profit companies in the years to come.
We now complement the earlier analysis with a numerical example (in Figure
4) in order to highlight the relationship between the size of the monetary payoff,
X, and the threshold values ψs and γo . Panel 4(a) shows that ψs is increasing in
X. When firm profitability is high, it is optimal for the impact investor to hold
a nonzero equity claim and increase social attention above ao only for those
firms with immense social value. The analysis therefore suggests that we are
less likely to obverse impact investments in firms with very high profitability,
as measured for example using return-on-assets or return-on-equity. However,
20 In our Online Appendix, we further analyze the match between managers and projects along the dimension of
ψm .
880
(a) (b)
881
While Glaesar and Shleifer (2001) model a nonprofit as a firm in which profits
are distributed through the enjoyment of “perquisites” that are less valuable than
cash, an alternative interpretation is that firm profits are held by someone who
both has no control over the firm and values the cash flows less than the for-profit
owner, that is, the impact investor in our model.21
Corollary 2. Nonprofit status is never optimal.
21 What we have in mind is that the impact investor plows-back produced cash into additional social opportunities
rather than realizing a distribution of profit.
882
The first term in (16) captures the benefit of shifting attention away from social
883
social attention ar and not the manager’s wage or impact investor’s equity
claim given ar . When the for-profit owner chooses to operate as a for-profit
firm and sells a portion of the firm’s cash flows to the impact investor, the for-
profit owner’s expected payoff is equal to the value of his expected after-tax
equity claim plus the value of the upfront payment from the impact investor
(which reflects the tax paid on the impact investor’s equity claim) minus the
cost of investment,
22 In actuality, the firm owner would implement a different level of social attention when operating as a nonprofit
firm, which would further increase the value of nonprofit status. Thus, the condition we establish in (19) for
nonprofit status to be optimal is merely sufficient (but not necessary).
23 This directly follows from (1−w(a )) 1− τ
r δ(1−τ ) < 1 and lim ψs →∞ y(ar (ψs )) = 1.
884
(a) (b)
24 For example, the United Nations provides guidelines on integrating ESG benchmarks into executive compensation
contracts (PRI 2012).
885
25 For example, Impact Reporting & Investment Standards (IRIS) (http://iris.thegiin.org/) and GIIRS Ratings and
Analytics for Impact Investing (http://b-analytics.net/giirs-funds) provide independent evaluations of social value
similar to the role credit rating agencies play in providing default information on corporate bonds.
26 For the sake of brevity, the supporting analytic results are provided in the appendix.
27 Although also outside the scope of the current model, one can imagine the full-commitment setting in the context
of a repeated game, in which the for-profit owner uses the prospects of future impact investments in order to
commit today to the promised level of social attention.
886
problem is given by
y(1) y(0)
f (1−a)X g(a) z(1)+ +(1−g(a)) z(0)+
1+δ 1+δ
max (22)
{a,Ω(b)}
g(a)ψs
+ −I
1+δ
0 ≤ f (1−a)X[g(a)w(1)+(1−g(a))w(0)]+g(a)ψm −γ , (22.2)
1. The manager’s wage is larger when social output is low, wc (0) > wc (1).
2. The impact investor receives no cash flow claim, yc (b) = 0 for b ∈ {0,1}.
3. The for-profit owner retains any cash flows not paid to the manager
zc (b) = 1−wc (b) for b ∈ {0,1}.
28 When the manager’s outside option is instead large, the manager may over- or underemphasize social attention
depending on the size of the manager’s required wage contract. When the manager overemphasizes social goals,
she is compensated with a guarantee-against-social-failure contract, whereas when the manager underemphasizes
social goals, she is compensated with a “pay-for-social-success” contract featuring wc (0) < wc (1).
887
888
is slack (small γ ). To ensure that the for-profit owner does not renegotiate
the manager’s wage contract after receiving the upfront payment from the
impact investor, the impact investor must hold a nonzero financial claim to the
project’s produced cash. While selling financial claims to the impact investor is
costly, the benefit from increasing social attention and obtaining a larger upfront
payment is valuable when ψs is sufficiently large. In addition, Proposition 2
shows that the for-profit owner always retains some residual cash flow and thus
Corollary 3 shows that the impact investor is first compensated when the project
fails to produce the social benefit. By reducing the for-profit owner’s residual
claim when social output is low, this contract most cheaply mitigates the owner’s
preference to reduce attention to social goals after receiving the upfront payment
from the impact investor, and therefore allows the for-profit owner to (most
cheaply) commit to the desired level of social attention ex ante. When the for-
profit owner’s incentive for renegotiation is large enough, the owner may be
unable to commit to the desired level of social attention even when retaining
none of the project’s cash flow when social output is low. In these cases, the
impact investor is also compensated when social output is high.
We provide a numerical example in Figure 6 to further illustrate optimal
contracts with contractible social output. Figure 6(a) depicts the manager’s
compensation contract wcr (b) and her equilibrium attention choice acr as
functions of ψs (when γ is sufficiently small that her participation constraint is
slack), while Figure 6(b) depicts the impact investor’s claim ycr (b). When ψs
is small, a higher level of social attention a increases the subsidy paid by the
impact investor only a small amount. This small increase is more than offset
by the loss due to selling a financial claim to the impact investor. As such, the
for-profit owner offers the wage contract that compels the manager to choose
the commercial benchmark level of social attention aco , and the impact investor
29 Like in Proposition 2, the corollary assumes ψ is sufficiently large and γ is sufficiently small.
s
889
(a) (b)
890
(in our model, the for-profit owner) should be more heavily compensated when
social output is high, whereas intrinsically socially motivated agents (in our
model, the impact investor) should be more heavily compensated when social
output is low. The economic tensions at play are therefore consistent with the
design of social impact bonds, which make greater repayments to private (for-
profit) investors in public works projects when these projects are successful
in obtaining their desired social goals. Moreover, our model suggests that for-
4. Concluding Discussion
We provide a model of joint financing by profit-motivated investors and social
impact investors in projects that produce both corporate profits and social good,
and we show that impact investors must hold financial claims in order to
incentivize profit-motivated owners to pursue social goals. When a project’s
potential social value is large enough, joint financing is mutually beneficial,
leading to an increase in the owner’s expected profit and an increase in the level
of social output enjoyed by impact investors.
Since it is costly for socially minded investors to hold financial claims, impact
investors invest only in those firms with significant social value. In addition,
we show that impact investments increase attention to social goals only when
the manager’s participation constraint is slack, which is most likely for firms
with sufficiently socially motivated managers. We argue that these results are
consistent with the common practice of targeting investments in companies that
are most likely to provide societal and/or environmental benefits and excluding
investments in those companies that do not. Moreover, we show that social
investors hold a larger fraction of firm equity in the most socially valuable firms.
When the model is extended to include corporate taxes, impact investments in
for-profit firms arise as an intermediate solution between pure for-profit status
and pure nonprofit status.
In practice, governments, in addition to socially minded investors, also
subsidize social activities within for-profit firms. To the extent that private social
capital is limited, the role for governments to aid in the achievement of social
good is vast. However, unlike private investors, the government need not hold
financial claims in order to incentivize profit-motivated agents to pursue social
goals, since the requisite reduction in owners’ profit motives can be obtained
simply through taxation.
Finally, we provide guidance into the design of contingent social contracts,
for example, social impact bonds and social impact guarantees. Our results
suggest that profit-motivated agents be compensated more heavily when
projects succeed in achieving predefined social objectives (like in social impact
bonds), while socially motivated agents be compensated more heavily when
891
Because f (·) and g(·) are twice-differentiable, continuous, and strictly concave functions, the
manager’s first-order condition is necessary and sufficient for determining the payoff-maximizing
attention allocation choice ã. For any w > 0, the manager’s payoff-maximizing choice ã is implicitly
defined by
0 = −f (1− ã)Xw +g (ã)ψm . (A2)
Define a(w) = ã as the equilibrium attention choice as a function of w. Because f (·) and g(·) are
twice-differentiable and continuous, a(w) is continuous and differentiable. By the implicit function
theorem, a(w) is strictly decreasing in w:
da ∂
−f (1−a)Xw +g (a)ψm
= − ∂w (A3)
dw ∂
(−f (1−a)Xw +g (a)ψm )
∂a a=ã
f (1− ã)X
= , (A4)
f (1− ã)Xw +g (ã)ψm
which is strictly negative. Therefore, a(w) is a one-to-one mapping between ã and w.
Proof of Lemma 2
By Lemma 1, a(w) : w → ã is one-to-one, continuously differentiable, and strictly decreasing
in w. By the inverse function theorem, there exists a function w(a) = a−1 (w) : ã → w that is one-
to-one, continuously differentiable, and strictly decreasing in ã. The explicit form of w(a) comes
from a rearrangement of (A2):
g (ã)ψm
w(ã) = = w. (A5)
f (1− ã)X
Furthermore, w(ã) is increasing in ψm .
Before proceeding with our the proof of Lemma 3, we provide the explicit assumptions that
guarantee that the set of contracts that satisfy the manager’s participation constraint and raise at
least I from investors is not empty.
Assumption 2 guarantees that the manager’s participation constraint can be satisfied, that is, if the
manager is given the project’s entire cash flow, w = 1, it is individually rational for her to participate.
The requirement on γ is a necessary condition for the existence of a nontrivial contract derived in
Lemma 3, but the condition is not enough to guarantee existence.
892
Assumption 3.
g(ã)ψs
I≤ , (A7)
1+δ
where ã ∈ arg max f (1−a)X +g(a)ψm .
a∈[0,1]
Assumption 3 guarantees that the project will be funded in equilibrium; that is, the project generates
sufficient enough social good so that the impact investor would be willing and able to cover the
Proof of Lemma 3
Before constructing a proof of Lemma 3, we provide two preliminary results.
Lemma 8. If the for-profit owner can commit to not renegotiate contracts, any optimal contract
never exhibits cash burning:
1 = wv +yv +zv . (A8)
Proof of Lemma 8
The proof is constructed as a proof by contradiction: There exists an optimal contract
{(w ,y ,z ),a } that exhibits cash burning, that is, 1 > w +y +z . Now consider an alternative
contract {(w ,y , ẑ), â} with 1 = w +y + ẑ. By Lemma 1, the alternative contract induces the same
level of equilibrium attention as the original contract,
â = a(w ) = a . (A9)
However, the alternative contract yields a strictly higher payoff for the for-profit owner:
y g(a )ψs y g(a )ψs
f (1−a )X ẑ+ + −I > f (1−a )X z + + −I. (A10)
1+δ 1+δ 1+δ 1+δ
Therefore, an optimal contract never exhibits cash burning.
Lemma 9. If the for-profit owner can commit to not renegotiate contracts, an optimal contract
features no cash repayment to the social impact investor:
yv = 0. (A11)
Proof of Lemma 9
The proof is constructed as a proof by contradiction: There exists an optimal contract
{(w ,y ,z ),a } for which the social impact investor receives a strictly positive fraction of the
project’s cash flow, that is, y > 0. Now consider an alternative contract {(w , ŷ, ẑ), â} with ŷ = 0.
By Lemma 8, the for-profit owner’s fractions of the cash flow under each contract are given by:
z = 1−w −y , (A12)
ẑ = 1−w , (A13)
implying ẑ > z . By Lemma 1, the alternative contract induces the same level of equilibrium attention
as the original contract:
â = a(w ) = a . (A14)
However, the alternative contract yields a strictly higher payoff for the for-profit owner:
y
f (1−a )X (1−y −w )+
g(a )ψs 1+δ
f (1−a )X(1−w )+ −I > , (A15)
1+δ g(a )ψs
+ −I
1+δ
893
because δ > 0. Therefore, if the for-profit owner can commit to not renegotiate contracts, an optimal
contract features no cash repayment to the impact investor.
We now proceed with the proof of Lemma 3. Since f (·) and g(·) are continuous functions, the
for-profit owner’s expected payoff is continuous in w, y, z, and a. Constraints 12.1–12.3 guarantee
the set of feasible contracts is closed and bounded. Furthermore, by Assumptions 2 and 3, the
contract {(w,y,z),a} = {(1,0,0),a(1)} satisfies constraints 12.1–12.3 and guarantees the for-profit
owner a nonnegative expected payoff. Thus, the set of feasible contracts is nonempty, closed, and
g(a)ψs
max f (1−a)X(1−w(a))+ −I. (A16)
a∈[0,1] 1+δ
Because the for-profit owner’s expected payoff in (A16) is concave by assumption, the for-profit
owner’s first-order condition is necessary and sufficient for determining the unique payoff-
maximizing attention allocation choice av . The optimal contract is given by {(w(av ),0,1−
w(av )),av }.
Now, suppose the manager’s participation constraint in (12.2) binds. The optimal level of social
attention is implicitly defined by
g (av )ψm
0 = f (1−av )X +g(av )ψm −γ , (A17)
f (1−av )X
(av )ψm
where we have made the substitution w(av ) = fg (1−a . Because f (·) and g(·) are twice-
v )X
differentiable, continuous, and strictly concave functions, there is a unique value av ∈ (0,1) that
satisfies (A17). The optimal contract is given by {(w(av ),0,1−w(av )),av }.
To complete the proof, we show that av > ao if γ < γo and av = ao if γ ≥ γo . Before proceeding,
we provide a preliminary result.
Proof of Lemma 10
The manager’s expected payoff with wage contract w is given by
da
d −f (1−a(w))Xw +g (a(w))ψm
(f (1−a(w))Xw +g(a(w))ψm ) = dw . (A19)
dw
+f (1−a(w))X
= f (1−a(w))X, (A20)
> 0. (A21)
894
We now proceed with completing the proof of Lemma 3. First, ignoring the manager’s
participation constraint momentarily, define the level of attention that maximizes the for-profit
owner’s payoff as â where â ∈ (0,1) is implicitly defined by
g (â)ψs dw
0 = −f (1− â)X(1−w(â))+ −f (1− â)X , (A22)
1+δ da a=â
where â > ao because g (·) > 0. If γ < γo there exists some equity wage contract w̃ < wo that satisfies
where ã = a(w). By Lemma 10, the manager’s expected date 1 utility is strictly increasing in w.
Therefore, renegotiation is not possible if it involves decreasing the manager’s wage, as the manager
would be worse off. Furthermore, by Lemma 1, the for-profit owner can only affect the manager’s
attention allocation choice by changing w. Thus, renegotiation is only possible if it involves strictly
increasing the manager’s wage.
Proof of Lemma 5
By assumption, the for-profit owner’s discounted expected date 1 payoff Up (w(a),y,z,a) is
twice continuously differentiable and concave in a. The contract {(w,y,z),a} will therefore be
renegotiated by the for-profit owner if and only if it is profitable for the for-profit owner to decrease
attention to social output by raising the manager’s wage (see Lemma 4),
∂Up dw ∂Up
+ < 0.
∂w da ∂a {(w,y,z),a}
895
renegotiation-proof contract directly follows from the proof of Lemma 3, which establishes the
existence of a unique optimal contract without renegotiation.
Constraint 14.1 guarantees that wr = w(ar ), and constraint 14.4 guarantees yr ≥ y(ar ). Since
increasing y is costly for δ > 0, the optimal renegotiation-proof contract sets y as small as possible,
yr = y(ar ). We now verify that yr = y(ar ) satisfies our limited liability constraints. Since dw
da < 0,
dy
we have y ≤ 1−w for all a. By construction y(ao ) = 0, and from Lemma 5, da > 0. Together, these
imply that y(a) ∈ [0,1−w] for all a ∈ [ao ,1]. A similar argument like in Lemma 3 establishes that
ar ≥ ao , and thus yr satisfies limited liability. Finally, Lemma 8 establishes that zr = 1−wr −yr .
896
which is strictly positive since P(a) is strictly concave in a by assumption. Thus, increasing ψs
raises the equilibrium level of social attention. Furthermore, since yr depends on ψs only through
dy
ar and da > 0,
dyr
> 0, (A32)
dψs
as well.
dyr
Next, we consider the comparative static of dδ : Taking the total derivative of (A30) with respect
dy
which is strictly negative since P(a) is strictly concave in a and da is positive valued for all a. Thus,
increasing δ decreases the equilibrium level of social attention. Furthermore, since yr depends on
dy
δ only through ar and da > 0,
dyr
< 0, (A34)
dδ
as well.
Proof of Corollary 2
From Proposition
1, the impact investor’s equity stake is given by yr = 1−w(ar )+
f (1−ar ) dw
f (1−a ) da a=a
which is the smallest financial claim needed to dissuade the for-profit owner
r r
from reneging on the promised level of social attention ar . Since dw da < 0 for all a ∈ (0,1), the
impact investor’s equity stake is strictly less than the entire residual cash flow claim, and the
for-profit owner therefore retains a positive equity stake, z = − ff(1−a r ) dw > 0.
(1−a ) da a=a r r
Proof of Lemma 6
For a given wage w(b) = {w(0),w(1)}, the manager’s problem in (20) is explicitly given by,
then ã = 0. In addition, if
then ã = 1. Finally, for any noncorner solution, the manager’s unique payoff-maximizing choice ã
is implicitly defined by
−f (1− ã)X(g(ã)w(1)+(1−g(ã))w(0))
0= (A36)
+f (1− ã)X g (ã)w(1)−g (ã)w(0) +g (ã)ψm .
897
Proof of Lemma 7
Before constructing a proof of Lemma 7, we rely on Lemma 8 and Lemma 9, located in the
proof of Lemma 3. By Lemma 8, the optimal contract features,
yc (b) = 0. (A38)
Proof of Lemma 11
We begin by considering the direct effects of both w(0) and w(1) on the for-profit owner’s
expected payoff. When renegotiation is not possible, the for-profit owner’s expected net payoff
with contractible social output is given by
g(a)ψs
P = f (1−a)X(1−g(a)w(1)−(1−g(a))w(0))+ −I, (A41)
1+δ
and the direct effect of changing w(0) on the net payoff is given by
∂P
= −f (1−a)X(1−g(a)), (A42)
∂w(0)
and the direct effect of changing w(1) on the net payoff is given by
∂P
= −f (1−a)Xg(a). (A43)
∂w(1)
∂a
The manager’s first-order condition in (A36) is used to calculate ∂w(b) ,
∂a
= −f (1−a)X(1−g(a))−f (1−a)Xg (a), (A44)
∂w(0)
∂a
= −f (1−a)Xg(a)+f (1−a)Xg (a). (A45)
∂w(1)
Using the preceding analysis, the marginal social-attention-to-cost ratios are,
∂a
∂w(0) −f (1−a)X(1−g(a))−f (1−a)Xg (a)
= , (A46)
∂P(a) −f (1−a)X(1−g(a))
∂w(0)
898
f (1−a) g (a)
= + > 0, (A47)
f (1−a) (1−g(a))
∂a
∂w(1) −f (1−a)Xg(a)+f (1−a)Xg (a)
= , (A48)
∂P(a) −f (1−a)Xg(a)
∂w(1)
f (1−a) g (a)
= − . (A49)
f (1−a) g(a)
∂a/∂w(0) ∂a/∂w(1)
> . (A50)
∂P/∂w(0) ∂P/∂w(1)
Lemma 11 shows, roughly speaking, that for a marginal change in the manager’s social attention,
the cheapest means to do so (i.e, the smallest impact on the for-profit owner’s expected payoff) is
via w(0).
We complete the proof of Lemma 7 by using a proof by contradiction; the optimal wage contract
features wc (0) > wc (1). Suppose not: If γ is sufficiently small such that the manager’s participation
constraint is slack, there exists an optimal wage contract featuring w (0) < w (1). Using Lemma
11, there must exists an alternative wage contract with ŵ(0) > ŵ(1) that induces the same level of
effort and yields a higher expected payoff for the for-profit owner. Therefore, the wage contract
{w (0),w (1)} cannot be optimal. Thus, when γ is sufficiently small and the manager’s participation
constraint is slack, the optimal contract features wc (0) > wc (1).
0 ≤ f (1−a)X[g(a)w(1)+(1−g(a))w(0)]+g(a)ψm −γ (A51.2)
To begin, we rely on Lemma 8 located in the proof of Lemma 3. By Lemma 8, an optimal contract
features,
1 = w(b)+z(b). (A52)
Moreover, since f (·) and g(·) are continuous functions, the for-profit owner’s expected payoff is
continuous in w(b), z(b), and a. Constraints A51.1-A51.3 guarantee the set of feasible contracts is
closed and bounded. Furthermore, by Assumption 2, Assumption 3, and Lemma 6, the cash flow
sharing rule w(b) = 1 (and equilibrium attention a(1,1)) satisfies constraints A51.1–A51.3. Thus,
the set of feasible contracts is nonempty, closed, and bounded, and by the extreme value theorem,
a maximum exists.
Finally, it is straightforward to show, like in the proof of Lemma 7, that a unique contract exists
and the contract elicits an attention allocation aco and the contract is characterized by wco (0) >
wco (1). Suppose not: If γ is sufficiently small such that the manager’s participation constraint
is slack, there exists an optimal wage contract featuring w (0) < w (1). Using Lemma 11, there
must exists an alternative wage contract with ŵ(0) > ŵ(1) that induces the same level of effort and
yields a higher expected payoff for the for-profit owner. Therefore, the wage contract {w (0),w (1)}
899
cannot be optimal. Thus, when γ is sufficiently small and the manager’s participation constraint is
slack, the optimal contract features wco (0) > wco (1) and aco < 1.30
Proof of Proposition 2
Before constructing a proof of Proposition 2, we rely on Lemma 8, located in the proof of
Lemma 3. By Lemma 8, an optimal contract features,
g(aco )ψs
f (1−aco )X(g(aco )(1−wco (1))+(1−g(aco ))(1−wco (0)))+ . (A54)
1+δ
Recall, the commercial benchmark level of attention, aco , is strictly less than 1. Furthermore,
wco (0) > 0. For γ sufficiently small, the manager’s participation constraint is slack and wco (0)
may be lowered, holding fixed wco (1), without violating the manager’s participation constraint.
As such, consider an alternative wage contract in which ŵ(0) = wco (0)− for some small > 0
and ŵ(1) = wco (1). Relying on Lemma 11, the alternative wage contract elicits a level of attention
â > aco . Furthermore, to ensure that the contract is renegotiation-proof set ŷ(0) = 1− ŵ(0) and
ŷ(1) = 1− ŵ(1) so that the for-profit owner receives no repayment. Under the alternative contract,
the for-profit owner’s profit is given by
30 To ensure that the commercial benchmark level of social attention is strictly less than 1, we require an assumption
dw(0)
similar to that used in Section 1: lim f (1−a) da(w(0),w(1)) is finite valued.
a→1 w(1)=0
900
equal to aco . Instead, for a sufficiently large ψs , the optimal level of social attention strictly exceeds
aco . Moreover, the impact investor, like in Proposition 1, must hold a nonzero financial claim if
aco > aco . If not, the for-profit owner will renegotiate the manager’s wage contract to the commercial
benchmark wage contract after the impact investor’s upfront payment is sunk.
To conclude the proof, we show that nonprofit status is never optimal, that is, ycr (0) < 1−wcr (0)
and/or ycr (1) < 1−wcr (1). The proof requires showing that the equilibrium a is continuous in y(b).
Proof of Lemma 12
Like in the proof of Proposition 1, denote the for-profit owner’s expected payoff, before the
impact investor’s transfer is sunk, as P(a,w(b),y(b),z(b)). Furthermore, denote the initial contract
consisting of the manager’s wage, impact investor cash flow claim, the for-profit owner’s cash flow
claim, and equilibrium attention choice as {(w(b),y(b),z(b)),a}b∈{0,1} . Now, like in the proof of
Lemma 5, denote the for-profit owner’s expected payoff, after the impact investor’s transfer is sunk
(and y(b) is fixed), as Up (a ,w (b),z (b);y(b)). The for-profit owner’s problem after the impact
investor’s transfer is sunk is given by
≥ f (1−a)X(g(a)w(1)+(1−g(a))w(0))+g(a)ψm , (A57.2)
0 ≤ w (b)+z (b) ≤ 1−y(b), and w (b) ≥ 0, z (b) ≥ 0, (A57.3)
where constraint A57.1 maps the wage w (b) to the manager’s optimal social attention choice a ,
constraint A57.2 is the manager’s individual rationality constraint in accepting wage w (b) in lieu
of the initial wage w(b), and constraint A57.3 represents the limited liability constraint.31
The for-profit owner’s problem may be generalized as follows: define χ ≡ {a ,w (b),z (b)}b∈{0,1}
and θ ≡ {y(b)}b∈{0,1} , then the for-profit owner’s objective function may be rewritten as Up (χ ;θ )
which is continuous in χ and θ . Furthermore, define C(θ ) : [0,1]2 [0,1]5 as a compact, vector-
valued, correspondence that maps the impact investor’s initial contract θ to constraints A57.1–
A57.3 and yields the constrained set of feasible contracts to choose from. Therefore, the for-profit
owner’s problem may be restated as,
Note, y(b) only affects limited liability in Constraint A57.3 and the constraint is linear in
y(b). Furthermore, a(w(0),w(1)) is continuous in w(b), and the manager’s individual rationality
constraint is linear in w(b) and continuous in a. As such, C(θ ) is both upper and lower
hemicontinuous. Consequently, by the theory of the maximum, the solution C ∗ (θ ) is continuous
31 Renegotiation-proofness requires that the solution to the for-profit owner’s problem outlined above is the same
as the initial contract P(a,w(b),y(b),z(b)).
901
902
However, by Proposition 2, {y (0),y (1)} = {0,0} cannot satisfy an optimal contract if acr > aco .
Next, consider y (1) > y (0). However, as shown above, y (1) > y (0) cannot be optimal. Thus,
ycr (0) > 0 if and only if wcr (0) < 1.
References
Bagnoli, M., and S. G. Watts. 2003. Selling to socially responsible consumers: Competition and the private
Barber, B. M., A. Morse, and A. Yasuda. 2017. Impact investing. Working Paper, University of California,
Berkeley.
Baron, D. P. 2001. Private politics, corporate social responsibility, and integrated strategy. Journal of Economic
& Management Strategy 10:7–45.
———. 2008. Managerial contracting and corporate social responsibility. Journal of Public Economics
92:268–88.
Besley, T., and M. Ghatak. 2005. Competition and incentives with motivated agents. American Economic
Review 95:616–36.
———. 2007. Retailing public goods: The economics of corporate social responsibility. Journal of Public
Economics 91:1645–63.
———. 2017. Profit with purpose? A theory of social enterprise. American Economic Journal: Economic
Policy 9:19–58.
Chava, S. 2014. Environmental externalities and cost of capital. Management Science 60:2223–47.
Cheng, I.-H., H. G. Hong, and K. Shue. 2016. Do managers do good with other peoples’ money? Working Paper,
Dartmouth College.
Glaeser, E. L., and A. Shleifer. 2001. Not-for-profit entrepreneurs. Journal of Public Economics 81:99–115.
Geczy, C. C., R. F. Stambaugh, and D. Levin. 2005. Investing in socially responsible mutual funds. Working
Paper, University of Pennsylvania, Wharton.
GIIN. 2016. Impact investing trends. New York: Global Impact Investing Network.
———. 2017. Annual impact investor survey 2017. New York: Global Impact Investing Network.
Hart, O., and L. Zingales. Forthcoming. Companies should maximize shareholder welfare not market value.
Journal of Law, Finance, and Accounting.
Hartzmark, S. M., and A. B. Sussman. 2017. Do investors value sustainability? A natural experiment examining
ranking and fund flows. Working Paper, University of Chicago.
Heinkel, R., A. Kraus, and J. Zechner. 2001. The effect of green investment on corporate behavior. Journal of
Financial and Quantitative Analysis 36:431–49.
Holmstrom, B. 1979. Moral hazard and observability. Bell Journal of Economics 10:74–91.
———. 1984. A note on the first-order approach in principal agent models. Mimeo.
Holmstrom, B., and P. Milgrom. 1991. Multitask principal-agent analyses: Incentive contracts, asset ownership,
and job design. Journal of Law, Economics, & Organization 7:24–52.
Hong, H., and M. Kacperczyk. 2009. The price of sin: The effects of social norms on markets. Journal of Financial
Economics 93:15–36.
Innes, R. D. 1990. Limited liability and incentive contracting with ex-ante action choices. Journal of Economic
Theory 52:45–67.
903
Jensen, M. C., and W. H. Meckling. 1976. Theory of the firm: Managerial behavior, agency costs and ownership
structure. Journal of Financial Economics 3:305–60.
Krüger, P. 2015. Corporate goodness and shareholder wealth. Journal of Financial Economics 115:304–29.
Kotchen, Matthew J. 2006. Green Markets and Private Provision of Public Goods. Journal of Political
Economy 114:816–834.
PRI Principles for Responsible Investment. 2012. Integrating ESG issues into executive pay. The United Nations
Principles for Responsible Investment.
904