effects of government support of nonprofit institutions on...
TRANSCRIPT
ORI GIN AL PA PER
Effects of Government Support of NonprofitInstitutions on Aggregate Private Philanthropy:Evidence from 40 Countries
S. Wojciech Sokolowski
� International Society for Third-Sector Research and The John’s Hopkins University 2012
Abstract This paper examines the effects of aggregate government payments to
nonprofit organizations on aggregate private philanthropy. Four behavioral models
of private philanthropic giving are proposed to formulate four hypotheses about
those effects: no net effect (null hypothesis), crowding in (positive effect), crowding
out (negative effect), and ‘‘philanthropic flight’’ or displacement (negative effect
across different subsectors). These hypotheses were tested against the evidence from
40 countries collected as a part of a larger research project aimed to document the
scale and finances of the nonprofit sector. The data show that, on the balance,
government payments to nonprofit institutions (NPIs) have a positive effect on
aggregate philanthropic donations to nonprofits, as stipulated by the crowding in
hypothesis, but a field level analysis revealed evidence of ‘‘philanthropic flight’’ or
displacement from ‘‘service’’ to ‘‘expressive’’ activities by government payments to
‘‘service’’ NPIs. Due to the limitations of the data, these results indicate empirical
plausibility of the hypothesized effects rather than their incidence. The findings
demonstrate the complexity of the relationship between government funding and
philanthropic donations to nonprofits, which depends on the goals of the actors
(donors and recipients) and institutional settings mediating the transaction costs of
difference sources of nonprofit support.
Resume Cet article examine les effets de l’ensemble des subventions gouvern-
ementales en faveur des organisations sans but lucratif sur la philanthropie privee a
titre global. Quatre modeles comportementaux de dons philanthropiques prives sont
proposes afin de formuler quatre hypotheses quant a ces effets : pas d’effet net
(hypothese nulle), rassemblement (effet positif), dispersion (effet negatif)
et « evasion philanthropique » ou deplacement (effet negatif a travers differents
sous-secteurs). Ces hypotheses ont ete testees sur des elements issus de 40 pays et
S. W. Sokolowski (&)
Johns Hopkins University, Baltimore, MD, USA
e-mail: [email protected]
123
Voluntas
DOI 10.1007/s11266-011-9258-5
collectes dans le cadre d’un vaste projet de recherche visant a documenter la portee
et les finances du secteur sans but lucratif. Les donnees indiquent que tout bien
considere, les subventions gouvernementales en faveur des OBL ont un effet positif
sur l’ensemble des dons philanthropiques a ces organisations, tel qu’enonce par
l’hypothese de rassemblement, mais une analyse de terrain a mis en evidence
une « evasion philanthropique » ou un deplacement entre le « service » vers des
activites « expressives » par les subventions gouvernementales pour « assurer les
services » des OBL. En raison des donnees limitees, ces resultats indiquent une
plausibilite empirique des effets hypothetiques plutot que leur incidence. Les con-
clusions soulignent la complexite de la relation entre le financement gouverne-
mental et les dons philanthropiques aux organisations sans but lucratif, laquelle
depend des objectifs des acteurs (donateurs et beneficiaires) et des structures in-
stitutionnelles servant d’intermediaire pour les frais de transaction des differentes
sources de soutien du secteur sans but lucratif.
Zusammenfassung Der vorliegende Beitrag untersucht die Auswirkungen der
Gesamtheit staatlicher Zahlungen an Nonprofit-Organisationen auf die gesamte
private Philanthropie. Es werden vier Verhaltensmodelle zu privaten philanthrop-
ischen Spenden zur Formulierung von vier Hypothesen uber diese Auswirkungen
vorgeschlagen: ausbleibender Nettoeffekt (Null-Hypothese), Crowding-in bzw.
Verstarkungseffekt (positiver Effekt), Crowding-out bzw. Verdrangungseffekt
(negativer Effekt) und ,,philanthropische Flucht‘‘oder Entfernung (negativer Effekt
auf verschiedene Teilsektoren). Diese Hypothesen wurden anhand von Nachweisen
getestet, welche in 40 Landern im Rahmen eines großeren Forschungsprojekts zur
Dokumentation der Tragweite und Finanzen des Nonprofit-Sektors gesammelt
wurden. Laut den erfassten Daten wirken sich staatliche Zahlungen an Nonprofit-
Einrichtungen entsprechend der Crowding-in-Hypothese insgesamt positiv auf die
Gesamtheit philanthropischer Spenden an Nonprofit-Organisationen aus. Eine
Feldebenen-Analyse hingegen erbringt den Nachweis, dass staatliche Zahlungen an
Nonprofit-,,Dienstleistungseinrichtungen‘‘eine ,,philanthropischen Flucht‘‘oder
Entfernung von ,,Dienstleistungsaktivitaten‘‘hin zu ,,Ausdrucksaktivitaten‘‘nach
sich ziehen. Aufgrund der begrenzten Daten weisen die Ergebnisse auf eine em-
pirische Plausibilitat der angenommenen Auswirkungen hin, nicht jedoch auf
ein tatsachliches Eintreten. Die Erkenntnisse stellen die Komplexitat der
Beziehung zwischen staatlicher Finanzierung und philanthropischen Spenden an
Nonprofit-Organisationen dar, die von den Zielen der Akteure (Spender und
Spendenempfanger) und dem organisatorischen Aufbau zur Verhandlung der
Transaktionskosten verschiedener Quellen zur Unterstutzung von Nonprofit-
Organisationen abhangt.
Resumen Este documento examina los efectos de los pagos gubernamentales
acumulados a las organizaciones sin animo de lucro en la filantropıa privada acu-
mulada. Se proponen cuatro modelos del comportamiento de donacion filantropica
privada para formular cuatro hipotesis sobre dichos efectos: ningun efecto neto
(hipotesis nula), atraccion (efecto positivo), exclusion (efecto negativo) y ‘‘vuelo
filantropico’’ o desplazamiento (efecto negativo en diferentes subsectores). Estas
Voluntas
123
hipotesis fueron sometidas a prueba frente a las evidencias de 40 paıses recopiladas
como parte de un proyecto de investigacion mas amplio que tenıa como objetivo
documentar la escala y las finanzas del sector de las organizaciones sin animo de
lucro. Los datos muestran que, sobre la balanza, los pagos gubernamentales a las
instituciones sin animo de lucro (NPI, del ingles Non-Profit Institutions) tienen un
efecto positivo sobre las donaciones filantropicas acumuladas, segun lo estipulado
por la hipotesis de atraccion, pero un analisis a nivel de campo revelo pruebas del
‘‘vuelo filantropico’’ o desplazamiento de las actividades de ‘‘servicio’’ a ‘‘expresi-
vas’’ por los pagos gubernamentales a ‘‘servicio’’ de las NPI. Debido a las limitac-
iones de los datos, estos resultados indican la plausibilidad empırica de los efectos de
las hipotesis en lugar de su incidencia. Los hallazgos demuestran la complejidad de la
relacion entre la financiacion gubernamental y las donaciones filantropicas a las
organizaciones sin animo de lucro, que depende de las metas de los actores (donantes
y receptores) y de las configuraciones institucionales que intervienen en los costes de
transaccion de diferentes fuentes de apoyo sin animo de lucro.
Keywords Philanthropy � Charity � Nonprofit funding �Government grants and payments � Crowding out � Crowding in
Introduction
This paper examines the relationship between the magnitude of public (government)
payments to nonprofit institutions (NPIs) and aggregate private donations (philan-
thropy) to these institutions. This relationship is often in the center of debates
concerning government policy toward NPIs. While some writers argue that
government support to NPIs discourages and displaces private participation and
philanthropy (for review see Andreoni 2004; Andreoni and Payne 2003; Simmons
and Emanuelle 2004; Steinberg 1985), others maintain that such support is essential
for the existence of NPIs that provide venues for individual participation and giving
(Borgonovi 2006; Kunemund and Rein 1999; Motel-Klingebiel et al. 2005; Salamon
and Sokolowski 2003). This paper addresses this issue by outlining several
alternative causal models of the relationship in question, and looking for evidence of
these models in macro-economic data on private and public support for NPIs in 40
countries, gathered as a part of a larger research project aimed to measure scope and
financing of NPIs cross-nationally (Salamon et al. 1999; 2004).
NPIs receive different kinds of payments from various sources. Those payments
can be classified by type and by the institutional sector where they originated
(Fig. 1). The three different types of payments include transfers, market sales, and
property income. Transfers (e.g., grants, gifts, or donations) entail payments for
which the payer does not receive anything of equivalent value in return. Market
sales entail payments for the market value of goods or services received by the
payer. Property income entails payments received for the use of property owned by
NPIs (e.g., dividends, interest, or rent). The three institutional sectors where
payments originate include government, households, and corporations, i.e., private
businesses (United Nations 2003, pp. 42–52).
Voluntas
123
In national accounting, government reimbursements for services rendered to
individuals (T2) are treated as transfers to households, which are then used to pay
for the received services (e.g., health care services paid by the Medicare). However,
for the purpose of this discussion, these transfers are treated as government
payments to NPIs, because the availability of these funds is a matter of public policy
rather than individual market decisions. In most developed countries, such
reimbursements account for most of government-originated funds received by
NPIs (e.g., about 88% in the US). The remainder of government payments consists
of government grants paid directly to NPIs and government contract payments (T1).
Private philanthropy includes donations of money or other assets given to NPIs
by households (T3) and private businesses (T4). Market purchases represent
payments by households (M1) or private businesses (M2) for goods or services
received from NPIs. Examples include ticket sales, hospital charges, tuition, and
membership dues1 (e.g., in a health or country club, and union or professional
association dues). Finally, the revenue that NPIs receive from their investments, or
rent for the use of their assets represents property income (P1). For the purpose of
Government
Households
Corporations
Nonprofit Sector
T1
T2 T2’
T3
T4
M1
M2 P1
Fig. 1 Model of financial flows to the nonprofit sector. T1 government transfers (grants) to- and contractswith-NPIs. T2 and T20 government reimbursements for services rendered to households (social benefittransfers) (T20 are treated as market sales to households by service producers and reported as such innational accounts (United Nations 2003, Sect. 4.7 and 4.27). T2 and T20 are not equal because some of thegovernment transfers to households T2 pay for services rendered by for-profit providers, such as privatemedical practices). T3 transfers from households (philanthropic gifts and donations). T4 transfers fromcorporations (philanthropic gifts, grants, donations). M1 household purchases of goods and services (incl.membership dues). M2 corporate purchases of goods and services (incl. membership dues). P1 propertyincome (rent, dividends, interest, etc)
1 This treatment of membership dues differs from that in national accounting, where dues paid to NPIs
serving businesses are viewed as market sales, but dues paid to NPIs serving households—as transfers
(philanthropic donations).
Voluntas
123
this discussion, market purchases and property income are treated as a ‘‘earned
income’’.
In sum, NPI revenue consists of three broadly defined types of resources:
government payments (T1 ? T20), private philanthropy (T3 ? T4), and earnedincome (M1 ? M2 ? P1), which on average account for 35, 15 and 50% of
nonprofit revenue in the 40 countries on which the data are available (although
significant variations in this distribution exist among countries). The relationship
between these types of nonprofit revenues is investigated in this paper. Specifically,
the effects of government payments on the aggregate private philanthropic
donations made to NPIs in 40 countries will be examined.
Macro-Economic Model of Philanthropic Support to NPIs
The most favorable scope condition for ‘‘crowding out’’ of one source of nonprofit
revenue by another is a single organization. The total amount of revenue a nonprofit
organization needs to generate is for the most part given. Since nonprofits do not
generate profits, their management has no incentive for producing more revenue
than it is needed to cover their organizations’ operating costs. The main concern of
the management is finding sources of income. Therefore, an increase of revenue
from one source will likely lead to decrease in the other sources, as the ‘‘crowding
out’’ argument holds.
This, however, does mean that an increase in government support must
automatically lead to a decrease in private philanthropy, because that increase can
be offset by a decrease in earned income instead, while the level of philanthropic
support remains constant. Reducing earned income rather than philanthropic
donations is more consistent with the charitable mission of a nonprofit organization.
So even under this most favorable for crowding out condition, the determination
which of the revenue sources is likely to be ‘‘crowded out’’ rests on the knowledge
of the motivation and decision making considerations of nonprofit managers in
pursing different revenue generation strategies.
At the macro level, however, crowding out is only one of several possible
outcomes. At the aggregate level revenue of the NPIs is not necessarily given
because of the possibility of new organizations being established, which leads to an
increase in aggregate operating cost and thus an increase of aggregate revenue
needed to cover that cost. Such growth is consistent with the so-called ‘‘Baumol
effect’’ stipulating the growth of cost in labor intensive industries (Baumol and
Bowen 1966), and it has been observed in several OECD countries, where the
nonprofit sector has been growing at faster rates than the rest of the national
economies (author’s analysis of the cross-national data assembled by the Johns
Hopkins Center for Civil Society Studies (CCSS)). A possible outcome of this
growth is an expansion of revenues from all sources, or ‘‘crowding in.’’
At this level two broadly defined groups of factors can possibly affect the level of
aggregate philanthropic donations and grants available to NPIs: general socio-
economic conditions that are favorable or adverse for philanthropic giving, such as
the level of personal income, the social value attributed to charity, or the social need
Voluntas
123
for charitable donations, and the transaction cost of strategies to obtain funding from
each of the three sources. Aggregate level of private philanthropy will generally be
higher in countries that have high levels of personal income and whose cultures put
a high value on philanthropic behavior than in countries lacking these attributes.
What is more, the existence of an acute need, such as dire poverty, income disparity,
or natural disasters often spurs altruistic impulses in the population, thus leading to
an increase in the aggregate level of philanthropic donations. Consequently, these
factors must be controlled for in some way when studying the effects of government
payments to NPIs on the aggregate level of private philanthropy.
Transaction cost of obtaining support from each of the three sources is the main
consideration affecting the recourse to any particular source. Generally, speaking,
the lower the transaction cost of any particular source vis a vis other sources (i.e.,
the lower the opportunity cost of that source), the more likely this source will be
preferred, and thus the more likely it will crowd out sources with higher
transaction costs. However, transaction costs are highly dependent on the
institutional environment in which organizations operate, and these different
conditions must be systematically taken into consideration when studying the
effects of changes in the availability of one revenue source on other sources. What
is more, transaction cost considerations can be mediated by non-economic
circumstances, such as organizational values, goals, or missions or legal
requirements. For example, in the US the legal qualification for the public
charity status is receiving a certain share of revenue (typically about 33%) from
‘‘public support’’ which is defined as a broad range of philanthropic sources (IRS
Publication 557, rev. October 2010).
The foregoing discussion suggests that while general socio-economic factors are
exogenous as far as the scope of this inquiry is concerned, earned income, which is
an alternative to private philanthropy source of nonprofit revenue is partially
endogenous, and so is government support. That means that the level of aggregate
earned income is at least in part affected by the aggregate level and transaction cost
of government support available to NPIs, as well as by general socio-economic
factors affecting philanthropic behavior. For example, nonprofit managers may
anticipate the general likelihood of obtaining support from philanthropy and
government and in anticipation adjust the price of their services, because this factor
is under their direct control (especially if the service is priced below its market
value, which gives some room for raising the price before being ‘‘priced out of the
market’’). What is more, government support to NPIs may be affected by general
socio-economic factors, especially social needs. The causal model guiding this
inquiry is shown in Fig. 2.
The foregoing discussion also makes it clear that understanding various
transaction cost considerations as well as possible motivation of both recipients
and donors of philanthropic support is essential for explaining the aggregate
level of private philanthropy, since different combinations of these factors may
lead to different funding strategies, and thus different levels of aggregate
philanthropic support that NPIs receive. The next section examines these factors
in more detail.
Voluntas
123
Behavioral Models of Philanthropic Donations to NPIs
The behavior of private philanthropic donors and managers of nonprofit organiza-
tions seeking and receiving philanthropic donations (NPI agents) can be explained
by taking into consideration two dimensions: their overall goal orientation and the
level of constraints or transaction costs that they face in achieving these goals. The
overall goal orientation of supporting NPIs by private donors may range from a
desire to advance worthy causes or values, to gaining social respect or recognition
among peers (Galaskiewicz 1985) or the so-called ‘‘warm glow giving’’ effect
(Andreoni 2004) and kindred intangible rewards, and to maximizing the social
impact (i.e., utility) of the resources at one’s disposal. The goals of NPI agents may
vary from attainment of a specific social mission to most effective procurement of
adequate resources for their organizations (Weisbrod 1998b). Following Max
Weber (1978, pp. 24–26), it is useful to conceptualize the orientation dimension as a
continuum between two polar extremes: the utilitarian orientation concerned
primarily with an effective execution rather than specific content (e.g., a web site
designer concerned with the most effective utilization of the screen space and data
flow regardless of the message that the site promotes) and the value attainment
orientation concerned primarily with achieving a particular substantive outcome
regardless of its cost. Of course, in real life, most human actions have both
orientations, albeit one may be more dominant than the other.
On the supply side, the bulk of constraints or transaction costs that donors (or
potential donors) face include efforts that need to be taken to obtain adequate
information to decide which nonprofit organization is most suitable to achieve the
supporter’s goals. Those efforts may vary depending on the legal and political
environment in which potential recipients operate. On the demand side (NPI
agents), transaction cost typically involves the amount of resources the organization
must spend to obtain adequate level of support or resources to implement its
Fig. 2 Macro-economic model of philanthropic donations to NPIs. CV control variable, TV test variable
Voluntas
123
program objectives. In most general terms, these involve efforts to monitor
availability of government grants and steps necessary to secure them, as well as
efforts to find potential donors and persuade them to donate money or assets.
These transactions cost vary considerably depending on the social, political and
legal environment in which the parties operate, as well as on individual
characteristics of these parties. Open and democratic societies with effective
governments and fair and efficient legal systems will have more trustworthy
information on NPIs available to potential donors, more reliable information about
different funding options, and more transparent ways of awarding and distributing
funds than societies that are undemocratic, corrupt or experiencing low levels of
trust in public institutions. The effectiveness of some activities may be more
difficult to judge than that of other (e.g., it is more difficult to measure outcomes of
social assistance than that of health care or education). Different NPIs may have
vastly different capacities of producing and disseminating information about their
operations, and donors may differ in their ability to find and process information
about potential recipients. Consequently, transaction costs for both, grant or
donations recipients (the demand side) and donors (the supply side) may vary quite
significantly, from very low to very high.
Although both, the orientation and the transaction cost, dimensions fall on a
continuum, for heuristic purposes it is useful to group them into two broadly defined
categories, defined by their polar extremes to highlight their meaning, which follows
the ‘‘ideal type’’ approach to social action proposed by Weber (1978, pp. 18–22).
The combinations of these categories can define four ‘‘ideal type’’ models of the
behavior this inquiry seeks to explain, i.e., giving donations to NPIs by private
parties (supply side) and seeking different forms of support for their organizations
by NPI agents (demand side). These four models are outlined in Table 1. Each of
these models stipulates different effects of government payments (transfers or third
party payments) on private support (property donations or volunteering) to nonprofit
organizations.
1. The efficiency maximization model implies that social actors focus mainly on
obtaining the maximum output or utility from the resources they control, and
face little or no transaction costs in that pursuit. This behavioral model
stipulates that donors (the supply side) have sufficient information about all
relevant aspects nonprofit organization operations, and sufficient capacity to
decide which recipient organization will utilize those resources most effectively
and efficiently. The mission of the recipient organization is of secondary
Table 1 Four ideal type
behavioral models of private
donors and NPI agents
Orientation
Utilitarian Value attainment
Transaction cost
Low 1. Efficiency maximization 3. Social solidarity
expression
High 2. Transaction cost avoidance 4. Strategic position
attainment
Voluntas
123
importance as long as it stays within the bounds of legitimate altruistic pursuits.
An example of this behavior on the demand side, is NPI leadership’s focus on
the most effective ways to procure resources needed to sustain the organization,
while paying less attention to how this funding may affect the organization’s
mission (Weisbrod 1998a, b).
2. The transaction cost avoidance model implies that social actors pursue the
maximum gain or utility from their resources, however they face significant
transaction costs in that pursuit. This combination of orientation and constraints
results in a behavior that minimizes transaction costs as much as possible
(Horne et al. 2005). On the supply side, this may entail donating money to
‘‘safe’’ NPIs, ones that have a well-established reputation or work for popular
causes (e.g., helping children or providing conventional health care). On the
demand side, this may involve efforts to obtain support from the ‘‘easiest’’
source (e.g., limiting funding strategies to church collections by a religiously
affiliated NPI, to supporter contribution by a community organization, or to
government grant by a politically connected service provider).
This model hypothesizes a condition that is opposite to that specified by the
contract failure theory of nonprofit organization (Hansmann 1987; Ben-Ner and
Van Hoomissen 1993). According to this theory, information asymmetry and
associated transaction costs of overcoming it pose a transaction cost to donating
money to NPIs, since efficiency-minded donors must have reasonable assurance
that their resources are not misappropriated, and monitoring organizational
behavior is often difficult to outsiders. However, the argument goes, the
nonprofit legal status of an organization from profiteering, and this provides
assurance that donations to this organization will not be misappropriated, thus
creating trust between the donor and the recipient.
The transaction cost avoidance model, on the other hand, refers to situations
when the nonprofit status alone is an insufficient basis of trust and does not
provide a adequate guarantees against misuses of donations. Such situations
may be a result of a wide array of factors ranging from concerted attacks on
nonprofits by politicians (e.g., the Russian government’s campaign in the early
2000s against foreign-based nonprofits and their local affiliates) or demagogues
(e.g., in sub-Saharan Africa nonprofits are often accused of various ‘‘conspir-
acies’’ targeting the local population), to being implicated in frauds or scandals,
and to the lack of cultural acceptance or perhaps unfamiliarity with this
organizational form. In such situations, the nonprofit status alone is likely to
provide an insufficient guarantee of trust (Sokolowski 2000), creating the need
for additional guarantees and thus increasing the transaction cost to donors,
which in turn may lead to transaction cost avoidance among some donors.
3. The social solidarity model implies that donors who support values and goals of
an organization and do not face any significant constraints in making their
donations to such an organization are likely to support it financially, while the
efficiency of that organization is of secondary importance. This condition is
likely to exist in areas with high levels of public consensus about a high social
value of the services being delivered by NPIs, such as disease cure or
prevention, assistance to disadvantaged children, or disaster relief. Another
Voluntas
123
example is the behavior of cause-motivated individuals (e.g., people with strong
religious convictions) making donations to organizations espousing their
causes. Yet another example is ‘‘sympathy support,’’ or donating money to
organizations that are supported by others with whom the donors strongly
identify (e.g., spouses or relatives, friends, work colleagues, celebrities, or
community leaders). In such situations, the attainment of a particular socially
defined value (promotion of a point of view, or expression of social solidarity)
is the primary motive of making a donation, without paying much attention to
efficiency concerns regarding the use of donated assets.
4. The strategic position maximization model stipulates behavior arising from a
combination of strong value orientation with considerable obstacles (transaction
costs) in the pursuit of goals congruent with that value. This condition is likely
to emerge in social or political activities dealing with contentious matters, such
as issue advocacy, defense of civil rights, or expression of cultural, economic,
or political interests, rather than in service delivery. An example of this is the
situation faced by social movement activists pursuing a cause that is strongly
opposed by a competing social movement, or perhaps by a considerable
segment of the general public, which poses a serious constraint on achieving the
movement’s goals. Therefore, the movement’s activists try to maximize their
chances to prevail over these constraints by building capacity, mobilizing
potential supporters, or persuading others to their cause, the behavior known as
‘‘frame bridging’’ (Snow et al. 1986) in social movement literature. They often
do so by supporting organizations viewed as instrumental in expanding the
movement’s appeal, either because those organizations are connected to
potential supporters, or because they may bestow legitimacy on the movement
itself, thus increasing its popular appeal. Another example is strategic
marketing of products or services by businesses making donations to
organizations connected to a social group that includes likely consumers
(e.g., gay or lesbian communities, fundamentalist Christians, or ethnic
minorities).
The four behavioral models outlined above are not mutually exclusive. The same
person or organization may engage in different types of behavior under different
circumstances. These models, taken together, represent a heuristic device that
specifies and categorizes a range of possible behaviors of private donors and
nonprofit agents. As such, this device is useful for delineating a range of possible
effects of government payments to nonprofits on private donations to these
organizations, which is the focal point of this paper. More specifically, these
behavioral models allow formulating two different predictions about the nature of
these effects, as outlined below.
The efficiency maximization model implies a negative effect of government
payments to NPIs on private donations, an effect known in economic literature as
‘‘crowding out’’ (Andreoni 2004; Andreoni and Payne 2003; Simmons and
Emanuelle 2004; Steinberg 1985). The chief reason for such a prediction is the
assumption of diminishing marginal utility assumption that underlines this line of
thinking. According to this assumption, the utility of each unit of a product or
Voluntas
123
service decreases as the supply of that product or service increases. It follows that
the ‘‘marginal utility’’ of a philanthropic donation decreases as alternative funding
(government payments) becomes more available, so a donor whose main concern is
to maximize the ‘‘efficiency’’ of her donation would avoid giving to NPIs that
receive substantial government support.
A slightly different argument along these lines is represented by the ‘‘government
failure’’ theory of the nonprofit sector (Weisbrod 1977, 1980; Hansmann 1987),
which claims that general public consent is a necessary condition of public financing
of public goods. In the absence of that consent, public financing of and thus supply
of these goods falters, thus creating unmet demand for public goods. This unmet
demand, the argument goes, offers an opportunity for private nonprofits to fill in this
void in demand, which in turn encourages growth of privately funded charities.
While the diminishing marginal utility argument can be conceptualized as a ‘‘push’’
factor that forces private philanthropy out of the nonprofit activity receiving too
much government support, the ‘‘government failure’’ theory specifies a ‘‘pull’’
factor that attracts private philanthropy to activities that do not receive adequate
government funding.
There are two logical consequences of this model. First, large levels of aggregate
government payments may lead to an overall reduction of aggregate philanthropic
giving (subsequently called ‘‘crowding out.’’) Second, large level government
payments may lead to donors shifting their donations to those areas of nonprofit
activity that receive less government support, but the aggregate level of
philanthropic giving remains more or less constant (subsequently called ‘‘philan-
thropic flight’’ or displacement). Since government payments to nonprofit areas are
typically concentrated in the areas of substantial public policy interests, such as
education, health or social assistance, the displacement of private donations is likely
to occur to those nonprofit activities that receive little financial support from
government (e.g., arts and entertainment, human rights, environmental issues,
religion, etc).
The transaction cost avoidance model, on the other hand, predicts a positive
effect of government payments on private donations, an effect known in economic
literature as ‘‘path dependence’’ (Arthur 1994). According to this argument, if
transaction costs pose a substantial constraint in achieving an economic goal,
avoidance of that cost becomes an important consideration in economic decision
making and will lead to behavior taking advantage of the already existing solutions
(or infrastructure) to minimize transaction costs. An example of this type of
behavior is ‘‘crowding in’’ of industries to certain geographical areas where
newcomers can take advantage of the already existing infrastructure developed by
their predecessors and government (Krugman 1991).
Following this logic, we can expect that high transaction costs in philanthropic
transactions, for example, the difficulty that an efficiency-minded donor may
encounter in obtaining adequate information about prospective recipient organiza-
tions will lead her to follow the ‘‘well-established path’’ and give only to those
organizations that already receive funding from trusted sources. Since government
typically is a trusted source, as it routinely vets organizations before awarding them
any funding, such a donor may view organizations receiving public funds as
Voluntas
123
trustworthy, and select them as recipients of her charity. A logical consequence is
that government funding may act as a ‘‘pull’’ factor for private charity, resulting in a
crowding-in effect. Another possibility is that donors may simply do not know
whether a potential recipient organization receives any government funding (Horne
et al. 2005), and crowding in is a coincidental outcome resulting from the fact that
large, well-established and reputable organizations that attract private donors are
also more likely to receive substantial government funding.
Likewise, the social solidarity expression model predicts a positive relationship
between government funding of NPIs and private donations, especially in
democratic societies. The fundamental assumption here is similar to that underlying
the ‘‘government failure’’ theory discussed earlier, namely that public policy and
public funding in a democracy is contingent on the consensus of the governed. It
follows that there is a considerable congruence in public perceptions of what causes
are worthy of general public support, which includes both government funding and
philanthropic donations. An example of such behavior may be donors’ response to
government ‘‘matching grant’’ programs aimed to enhance public support of a
charitable cause (e.g., the National Public Radio or cancer research). Knowing that
their donations will result in increased public funding to a worthy cause, such
donors may decide to throw in their charitable donations. Another example is
response to natural disasters, which typically result in increases of both public and
private support to nonprofit organizations providing disaster relief.
This ‘‘jumping the bandwagon’’ (Kunemund and Rein 1999) scenario is likely to
result in a positive effect of government payments to NPIs on aggregate private
philanthropy. The net result is an overall increase of aggregate philanthropic giving,
although there is also a theoretical possibility of shifting of that philanthropic giving
from one activity area to another with little change in the total aggregate volume of
philanthropy. However, the empirical manifestation of this theoretical possibility is
virtually indistinguishable from the effect of ‘‘philanthropic flight’’ described
earlier, and the only way of distinguishing them is to know the exact motivation of
human actors that caused that shift. However, in the opinion of this writer the shift
of philanthropic giving from one subsector to another due to ‘‘jumping the
bandwagon’’ effect is not a very likely scenario, except under very special short
term circumstances, such as natural disasters, in which case its effect on aggregate
philanthropy is only temporary. By contrast, ‘‘philanthropic flight’’ due to crowding
out by government support in one subsector is more likely to be a long term effect.
Consequently, only the ‘‘philanthropic flight’’ possibility, that is displacement due
to crowding out by government funding, is being considered here.
Finally, the strategic position maximization model predicts behaviors that result in
both, positive and negative, effects of government payments to NPIs on private
charity. These effects are likely to cancel each other out, leading to ‘‘no effect’’
prediction at the aggregate level (Horne et al. 2005; Borgonovi 2006). This type of
condition is more likely to exist in activities dealing with contentious social or
political issues rather than in more service-oriented fields. For example, adherents of
social movements on both sides of the abortion issue in the US are likely to engage in
philanthropic giving that is both congruent and incongruent with public support of
NPIs. Supporters of the pro-choice position are likely to donate to health clinic that
Voluntas
123
provide family planning services. Since such clinics, as health care providers, are
also likely to receive some government payments (e.g., under the Medicaid
program), this philanthropic giving behavior will produce a crowding in effect. On
the other hand, advocates of the ‘‘pro-life’’ position will most likely direct their
philanthropic giving to organizations that oppose abortion, such as churches and their
advocacy arms. Since such organizations typically receive little or no government
payments, this philanthropic behavior will produce a crowding-out effect. On the
aggregate level, however, these two effects are likely to cancel each other out.
This situation is mirrored on the demand side, as agents of nonprofit
organizations will likely pursue a mix of funding sources that maximizes their
own capacity, legitimacy, or political clout. They may use public support for
organizational capacity building, which in turn may enable to them to recruit
volunteers or raise funding from private donors. Alternatively, they may build their
private support base to boost their legitimacy and thus chances to receive public
support. This behavior results in a positive correlation between public and private
support. On the other hand, nonprofit managers may opt for private support and
avoid public support (or vice versa) to maintain their legitimacy with their
constituents, as it was for example the case of some AIDS/HIV advocacy
organizations (Lune and Oberstein 2001). This may result in a negative correlation
between these two forms of support. Again, on the aggregate level these two effects
cancel each other out.
In sum, the four behavioral models outlined above lead to the formulation of four
hypotheses of possible effects of aggregate government payments to nonprofits and
aggregate private charity:
H0 No effect. The aggregate level of government payments to NPIs has no net
effect on aggregate level private philanthropic donations (i.e., crowding out and
crowding in behaviors cancelling each other out).
H1 Crowding in effect: The higher the level of government payments to NPIs, the
higher the aggregate level of private philanthropic donations.
H2 Crowding out effect: The higher the level of government payments to NPIs,
the lower the aggregate level of private philanthropic donations.
H3 ‘‘Philanthropic flight’’ effect: The higher the level of government payments to
NPIs in one subsection of the nonprofit sector, the higher the aggregate level of
private philanthropy in other subsections (this represents a variant of the crowding
out argument stipulating crowding out of philanthropy from more heavily
government funded NPIs to the less government-funded ones).
These hypotheses are tested in the empirical part of this paper.
Data and Measurement
The data to test these two hypotheses were collected by the Johns Hopkins CCSS as
a part of a larger project mapping the size and financing of NPIs at the aggregate
Voluntas
123
national level in 43 different countries to date. The data set used in this analysis
contains observations on 40 countries on which suitable information was available.
Data for individual countries are for a single year, ranging from 1995 and 2007.
Using cross-sectional rather than longitudinal data for testing the effects of
changes in financial flows has obvious limitations. Nonetheless, different countries
in this dataset represent different stages of nonprofit development. For example, the
US represents ‘‘mature’’ and robust nonprofit sector with little longitudinal changes
in aggregate financing of its operations. On the other end of the spectrum are
Eastern European countries with newly emergent nonprofit sectors pursuing diverse
financing strategies (Sokolowski 2010). In the middle, there are Scandinavian
countries and to a somewhat lesser extent Western European countries in which the
roles and thus financing mechanisms underwent modifications during the past two
decades. Therefore, the individual country data can be viewed as proxies for
different stages in nonprofit sector development, and thus some approximation of
the longitudinal data.
Another limitation is that the countries covered by the CCSS data do not
constitute a probability sample in a statistical sense, but rather a sub-population of
countries. Therefore, sample statistics (such as the probability of sampling error)
typically used in statistical hypothesis testing are meaningless in this case (but
nonetheless reported). The only meaningful indicator of a ‘‘significance’’ of a
particular test variable is how much variance on the dependent variable it can
explain.
Due to these data limitations the empirical results presented in this section should
be interpreted as indicators of empirical plausibility of the hypotheses developed in
the previous section rather than a statistically rigorous falsification of mutually
exclusive explanations or indicators of the incidence of the hypothesized effects.
What is more, these results pertain only to aggregate levels of government funding
and philanthropic donations, and any attempt to apply these results to individual
organizations is the fallacy of composition.
The dependent variable, aggregate philanthropic donations, is measured by
the total value of private financial contributions to NPIs expressed as a share of the
country’s GDP (both in local currencies and current prices in the year for which the
data were collected). The advantages of this method of measuring aggregate
philanthropy are: (i) the control for the size of national economies, which obviously
impacts the total volume of all financial flows; (ii) cross national comparability
without the need of using problematic currency conversions; and (iii) relative
longitudinal stability, as there is little short term variations in the rates, which
eliminates the need for inflation adjustments due to different years of reporting.
The test variable, government payments to NPIs, is measured in a similar fashion.
For more information about the data and data collection methodology, see Salamon
et al. (2004).
In accordance with the macro-economic model of philanthropic donations
outlined in ‘‘Macro-Economic Model of Philanthropic Support to NPIs’’, there are
two set of control factors that must also be considered when studying the effects of
government payments on aggregate philanthropy. There first set is general socio-
economic factors that are likely to affect the overall aggregate supply of
Voluntas
123
philanthropic funds (CV1). The main, although not the only such factors include the
overall level of national wealth, the set of social values that are conducive or
unfavorable to charitable behavior, as well as evident need for philanthropic
assistance.
These three control factors are represented by several proxies. The proxy for
national wealth is per capita GDP converted to the US dollars. The proxy for
‘‘altruistic social values’’ is the amount of aggregate volunteering, measured as
volunteer time converted to full-time equivalent jobs and expressed as a share of the
total economically active population to control for cross-national differences in the
population size. The need for philanthropic assistance is represented by two proxies.
The first proxy is income inequality (measured by the Gini index) on the assumption
that the greater income inequality, the greater the need for philanthropic assistance.
It is so, because ceteris paribus higher poverty levels in countries with more
unequal distribution of wealth results in more people not being able to afford
services typically provided by nonprofits, such as education, health or social
assistance. Likewise, greater concentration of wealth results in greater availability
of disposable wealth that can be given to charity. What is more, the wealthy in such
societies may face greater social pressure to give money to charity to maintain their
social status and privileged position (Galaskiewicz 1985).
The second proxy is the aggregate amount of government social welfare
spending, expressed as a share of the GDP, on the assumption that the greater the
level of welfare spending the lower the need for private philanthropic assistance.
The effects of all these proxies for CV1 on aggregate philanthropy were tested
separately before introducing them to the model with the test variable. However,
their effect proved to be negligible, as demonstrated by the amount of variance they
explained. Per capita GDP and Gini account for only about 9% of the variance, and
adding volunteer input and social welfare spending not only does not explain any
additional variance, buy actually reduces the adjusted R square of the model (since
this statistic is a function of the number of variables in the equation). Consequently,
the latter two control variables were dropped for the sake of parsimony, and only per
capita GDP and Gini were used as indicators of general factors (CV1).
The second set of controls stipulated by the macro-economic model in ‘‘Macro-
Economic Model of Philanthropic Support to NPIs’’ is the aggregate level of earned
income (CV2). Since changes in the level of government payments to NPIs will
likely result in corresponding changes in the aggregate level of philanthropy or the
aggregate level of earned income (or both), the latter must be controlled for before
any effect on the former can be claimed.
Table 2 shows the descriptive statistics of all variables used in the empirical part
of this investigation.
Empirical Results
The hypotheses were tested by OLS regression modeling in which the dependent
measures were regressed first on the two sets of control measures (controlling for
general propensity toward philanthropic behavior and for the level of earned
Voluntas
123
income), and then on the control measures and the test variable (government
payments). This is known as the ‘‘nested model’’ approach, which allows a
systematic examination of the variance explained by each of these sets of variables.
This approach is appropriate for causal model outlined in Fig. 2 (Pedhazur 1982,
pp. 177–180). As already mentioned, the data set at hand does not represent a
probability sample in any true sense, but rather a sub-population of countries.
Therefore, the statistical significance statistic is not very meaningful in this context
(although it is reported in the empirical results). A more meaningful indicator of
‘‘significance’’ of the test variable is the sign of the standardized regression
coefficient and its value, and the amount of variance (adjusted R square) explained
by each of the subsequently introduced to the model sets of variables. The purpose
of this analysis is not to test the strength of the effects of control (CV1 and CV2)
and test (TV) variables on aggregate philanthropy, but only to determine the
existence and direction of causal relations.
The ‘‘null hypothesis’’ claiming no net effect (H0) is confirmed if the test
variable (government payments) does not improve the variance of the model with
the control variables only. The ‘‘crowding in’’ hypothesis (H1) is confirmed, if the
test variable improves the variance of the model with the control variables only, and
the sign of its regression coefficient is positive. The ‘‘crowding out’’ hypothesis
(H2) is confirmed if adding the test variable to the model with controls only
improves the variance explained in the model, and the sign of its regression
coefficient is negative. Finally, the ‘‘philanthropic flight’’ hypothesis (H3) is
confirmed if adding the test variable representing government payment in one
subsector improves the variance explained by the model with controls only for
another subsector.
Table 2 List and descriptive statistics of variables
Variable Mean Std. deviation
Philanthropic donations to all NPIs as % of GDP 0.0056 0.0048
Philanthropic donations to ‘‘service: NPIs as % of GDP 0.0019 0.0020
Philanthropic donations to ‘‘expressive’’ NPIs as % of GDP 0.0037 0.0031
Gov’t payments to all NPIs as % of GDP 0.0205 0.0233
Gov’t payments to ‘‘service’’ NPIs as % of GDP 0.0169 0.0218
Gov’t payments to ‘‘expressive’’ NPIs as % of GDP 0.0036 0.0029
Earned income of all NPIs as % of GDP 0.0228 0.0175
Earned income of ‘‘service’’ NPIs as % of GDP 0.0107 0.0103
Earned income of ‘‘expressive’’ NPIs as % of GDP 0.0121 0.0099
Per capita GDP (ppp US$) 24151.7 14099.6
Government social welfare spending as % of GDPa 0.172 0.093
Gini coefficient 36.7 9.4
FTE volunteers as % of economically active populationa 0.0071 0.0085
Number of observations 40
a Not included in subsequent models due to low explained variance
Voluntas
123
Table 3 shows the results of the test of the ‘‘crowding in’’ hypothesis.
Model 1, which consists of control variables representing general propensity for
philanthropic behavior explains about 9% of cross-national variance in the
aggregate level of private donations to NPIs, as it has been already discussed in
‘‘Data and Measurement’’. As expected, the effects of both controls are positive,
indicating that higher levels of personal income and greater social inequality tend to
increase the aggregate level of philanthropy. Adding the second control, the level of
earned income (Model 2) boosts the explained variance to 15%.
Adding the test variable, government payments to all NPIs (Model 3) to the
regression equation increases explained variance to over 31%, a considerable
improvement over the model with controls only. The sign of regression coefficient
is positive. This is consistent with H1 (crowding in) and inconsistent with H2
(crowding out). However, given the highly aggregated nature of the data at hand, it
is possible that changes in the levels of dependent and test variables occurred within
different subsectors, and thus are not necessarily related.
To address this concern, the second analysis was run at the field of activity level.
The data were divided into fields that tend to attract high levels of government
support (education, health and social assistance) and all other fields (arts and
recreation, environment, housing and community development, civic activities,
religion, labor unions and professional associations) that typically receive much
lower, if any, government support. Following Salamon et al. (2004), the first group
was labeled ‘‘service activities’’ and the second group—‘‘expressive activities.’’
There were two separate analyses for each of these types of activities that mirror the
analysis conducted for the nonprofit sector as a whole. Table 4 shows the results for
‘‘service activities,’’ while Table 5 shows results for ‘‘expressive activities.’’
The results obtained for the ‘‘service activities’’ (Table 4) pretty much mirror
those obtained for the nonprofit sector as a whole. The improvement of explained
variance by adding government payments (TV) is even better than in the sector as a
whole, over 37% versus 11% explained by the model with controls alone. This is
again consistent with the crowding in hypothesis (H1). For the ‘‘expressive
activities,’’ however, the results are very different. The model with controls only
(Model 2) explains over 22% of the variance, and the inclusion of the test variable
Table 3 Effect of government payments on aggregate private donations to all NPIs
Variable Model 1 Model 2 Model 3
Beta Signif. Beta Signif. Beta Signif.
Per capita GDP (ppp US$) 0.43 0.03 0.23 0.27 -0.01 0.96
Gini coefficient 0.37 0.06 0.28 0.15 0.27 0.12
Earned income of all NPIs as % of GDP 0.33 0.06 0.19 0.24
Gov’t payments to all NPIs as % of GDP 0.52 0.00
Adjusted R2 0.086 0.151 0.315
N 40 40 40
Voluntas
123
(Model 3) results is a lower adjusted R square (which is a function of the number of
variables in the equation). This is consistent with the ‘‘no effect’’ hypothesis (H0).
Finally, the ‘‘philanthropic flight’’ hypothesis (H3), which is a variant of the
crowding out argument, was tested by examining the effect of government funding
in ‘‘service’’ fields on aggregate philanthropy in the ‘‘expressive’’ fields (Table 6).
Adding the test variable to the model with controls only increases the share of
explained variance from 22 to 37%, a non-trivial improvement. This is a sharp
contrast to the results presented in Table 5, in which government funding in
‘‘expressive’’ fields and no observable effect on the level of aggregate philanthropy
in these fields. These results are consistent with the ‘‘philanthropic flight’’
hypothesis (H3).
Discussion
These regression results, limited as they are, show a clear support for the crowding
in effect of government payments to NPIs on private donations. This effect was
Table 4 Effect of government payments on aggregate private donations to NPIs in ‘‘service’’ activities
Variable Model 1 Model 2 Model 3
Beta Signif. Beta Signif. Beta Signif.
Per capita GDP (ppp US$) 0.27 0.18 0.18 0.39 -0.11 0.56
Gini coefficient 0.34 0.09 0.28 0.17 0.28 0.12
Earned income of ‘‘service’’
NPIs as % of GDP
0.19 0.28 -0.02 0.90
Gov’t payments to ‘‘service’’
NPIs as % of GDP
0.65 0.00
Adjusted R2 0.079 0.109 0.373
N 40 40 40
Table 5 Effect of government payments on aggregate private donations to NPIs in ‘‘expressive’’
activities
Variable Model 1 Model 2 Model 3
Beta Signif. Beta Signif. Beta Signif.
Per capita GDP (ppp US$) 0.49 0.01 0.26 0.19 0.21 0.32
Gini coefficient 0.35 0.06 0.28 0.12 0.29 0.11
Earned income of ‘‘expressive’’
NPIs as % of GDP
0.40 0.02 0.37 0.03
Gov’t payments to ‘‘expressive’’
NPIs as % of GDP
0.14 0.41
Adjusted R2 0.117 0.224 0.217
N 40 40 40
Voluntas
123
hypothesized from two different behavioral models, the transaction cost avoidance
model, stressing the effects of considerable transaction cost on decisions made by
efficiency conscious donors, and the social solidarity expression model, emphasiz-
ing the role of shared values in making philanthropic decisions. Of course, it is
impossible to deduce from these results which of these two behavioral models is
responsible for the observed results, it could be either or both. However, the scope
condition of the transaction cost avoidance model implies the existence of high
transaction costs of ‘efficient’ philanthropic giving, which is unlikely to hold in
most countries under investigation (e.g., the European Union member countries,
Australia, Canada, Israel, New Zealand, United States or Switzerland) where the
information on public charities can be obtained with relative ease. Therefore, it is
more likely that the social solidarity expression model may be at work here.
The ‘‘philanthropic flight’’ effect, or crowding out philanthropic donations from
fields where they have lower ‘‘marginal utility’’ due to government funding to those
where their ‘‘marginal utility’’ seems higher also received support from the data.
This hypothesis was derived from the efficiency maximization model. Government
funding in ‘‘service activities’’ has a substantial positive effect on private
philanthropy in ‘‘expressive activities’’ even though government payments in
‘‘expressive’’ activities have no effect on philanthropy. This suggests the ‘‘push’’
effect hypothesized from the diminishing marginal utility assumption underlying the
efficiency maximizing model. This is not necessarily detrimental to philanthropic
mission, in the opinion of this writer. It represents a situation of the basic human
needs, such as health care, education, or human services, being served by public
financing, and private philanthropy engaging in the pursuit of missions that further
enhance the quality of social life, such as arts and culture or various forms of civic
engagement.
On the other hand, the crowding out scenario underlying the efficiency
maximization model, and also implied by the ‘‘government failure’’ theory seem
unsupported by these data. Under this scenario one would expect a negative
correlation between government payments and aggregate philanthropy at either the
total nonprofit sector level, or at least in some of its subsets. Yet the data show a
positive effect at both the whole sector level and in the ‘‘service’’ fields, and no
Table 6 Effect of government payments to NPIs in ‘‘service’’ fields on aggregate private donations to
NPIs in ‘‘expressive’’ fields
Variable Model 1 Model 2 Model 3
Beta Signif. Beta Signif. Beta Signif.
Per capita GDP (ppp US$) 0.49 0.01 0.26 0.19 0.06 0.77
Gini coefficient 0.35 0.06 0.28 0.12 0.25 0.14
Earned income of ‘‘expressive’’ NPIs
as % of GDP
0.40 0.02 0.35 0.03
Gov’t payments to ‘‘service’’ NPIs as % of GDP 0.39 0.02
Adjusted R2 0.117 0.224 0.373
N 40 40 40
Voluntas
123
effect in the ‘‘expressive’’ fields. Of course, this does not mean that crowding out
does not occur on the organizational level, but there is no evidence of it in the
aggregate data.
The problem of applying the crowding out model to the macro-level of analysis
seems to be related to its underlying assumption of a fundamentally competitive
nature of government-nonprofit relation spelled in the ‘‘government failure’’ theory.
This theory is a practical application of the Hotelling’s (1929) principle of minimum
differentiation and Black’s (1948) median voter theorem stipulating that in a two-
party competition, the party whose program deviates from the ‘‘median vote’’
preference is likely to lose the election. Consequently, social programs that do not
have ‘‘median voter’’ support are unlikely to receive public funding, and thus must
be funded by philanthropy. However, the applicability of this theory is limited
mainly to a US-style two-party democracy where public funding of specific social
programs is highly contingent on partisan politics, and thus more likely to be
subjected to ‘‘median voter’’ preferences.
Yet most countries under investigation while nominally democracies, have
political systems that either give governments relatively broad freedoms from
popular demands (e.g., Japan, Kenya, Korea, Peru, the Philippines, Pakistan,
Tanzania and Uganda), or constitute what Lijphart (1999) calls the ‘‘consensus
democracies,’’ characterized by multi-party systems representing a wide range of
interests (e.g., Belgium, Denmark, France, Germany, Israel, Italy, the Netherlands,
Norway, Sweden, and Switzerland). In the first group, popular preferences had
relatively little effect on government funding priorities, which often are geared more
toward macro-economic development than meeting public consumption demands.
In the second group, the political model of governance developed a high level of
receptiveness to demands of diverse interest groups, which effectively eliminated
the ‘‘median voter’’ problem and resulted in public financing of collective goods
serving relatively narrow interest groups.
Consequently, a likely explanation of the lack of the observable ‘‘crowding out’’
effect in the aggregate country data lies in the fact that most of these countries do
not meet the scope condition necessary for this effect to occur, i.e., the ‘‘failure’’ of
government funding of services in the absence of majority support. Consequently,
‘‘crowding out’’ effect, while theoretically possible, is nonetheless unlikely to occur
on a scale discernible in aggregate national data due to institutional conditions
preventing it. It may, however, occur where institutional conditions warrant it (e.g.,
in the US).
This suggests that the level of government or private support to NPIs is mediated
by the political process (Gronbjerg 1987; Salamon 1987; Salamon and Anheier
1998; Sokolowski 2010). In ‘‘majoritarian’’ democracies (such as the US and other
English-speaking countries), the inverse relationship between aggregate government
spending on certain public goods and private charity may be more pronounced than
in ‘‘consensus democracies’’ (such as most of the European Union member states).
‘‘Consensus democracies’’ create more favorable conditions for political collabo-
ration among diverse interest groups, which leads to a greater differentiation of
collective goods receiving public funding, and a greater role of NPIs in the delivery
of those goods (Lijphart 1999).
Voluntas
123
These results suggest that government payments to NPIs do impact private
philanthropy, but that impact is not as unidirectional as some of the literature seems
to suggest. The government payments can have either positive or negative impact on
philanthropy or no effect at all, depending on the field of activity and socio-political
circumstances. This, in turn, suggests the necessity to consider different behavioral
models of philanthropic actors (donors and recipients), as stipulated by the heuristic
device proposed in this paper, to fully explain the impact of government funding on
private philanthropy and the nonprofit sector finances.
Conclusions
This paper proposed a theoretical framework for studying the relationship between
government payments to NPIs and private philanthropic giving. This framework,
grounded in Max Weber’s approach to social behavior, consists of four ‘‘ideal type’’
behavioral models, defined by intersecting two dimensions of purposive human
action: goal orientation (utilitarian vs. value attainment) and transaction cost level
(low vs. high) in achieving these goals.
These four models generated four hypotheses about the relationship between
government payments to NPIs and aggregate philanthropic support: no net effect,
crowding in, crowding out, and ‘‘philanthropic flight’’, which is a special case of the
crowding out argument. These hypotheses were tested against the evidence from 40
countries showing the aggregate levels of government payments and aggregate
private philanthropic donations to NPIs operating in ‘‘service’’ and ‘‘expressive’’
activity areas. The data show, on the balance, that government payments to NPIs
have a positive effect on aggregate philanthropic giving, as predicted by the
crowding in hypothesis. However, a field level analysis revealed evidence of
‘‘philanthropic flight,’’ or displacement, of private philanthropy from ‘‘service’’ to
‘‘expressive’’ activities by government payments to ‘‘service’’ NPIs. The crowding
out effect was not observed in the aggregate country data, most likely due to the fact
that most countries covered by this analysis have institutional features minimizing
the likelihood of this effect, and thus lie outside the scope condition of theories
stipulating this effect.
These results suggest the possibility of divergent effects of government payments
to NPIs on aggregate private philanthropy. The government payments can either
encourage private philanthropy in some activity fields or push it away to other fields
receiving less government support, or have little, if any, effect. This, in turn,
suggests the need to consider different behavioral models of philanthropic behavior,
as well as institutional settings in which this behavior takes place.
These findings imply that private giving is a complex phenomenon, shaped by a
multitude of factors, of which alternative forms of support received by NPIs is only
one. Other important factors include cultural norms, traditions and values, and
institutional settings in which philanthropic donors and NPIs operate, including
different models of democratic governance. The data at hand allow studying only
aggregate outcomes of philanthropic donors’ actions, and a micro-social or
individual level of analysis is needed to study the actual motives or behavioral
Voluntas
123
models underlying these outcomes. Therefore, these results should not be
interpreted as a ‘‘test’’ of any theory underlying this analysis, but rather as
evidence that all these theories and their underlying behavioral models deserve
equal consideration in further empirical studies of philanthropic behavior. On the
other hand, these aggregate results do dispel one myth, circulated by many
neoliberal social commentators, that government social spending is detrimental to
private philanthropy. The evidence from 40 countries covered by this study suggests
that it is not, and the opposite seems to be true.
References
Andreoni, J. (2004). Philanthropy. In G. Varet, S. C. Kolm, & J. M. Ythier (Eds.), Handbook of giving,reciprocity and altruism. Amsterdam: North Holland.
Andreoni, J., & Payne, A. A. (2003). Do government grants to private charities crowd out giving or fund
raising? American Economic Review, 93(3), 792–812.
Arthur, B. W. (1994). Increasing returns and path dependence in the economy. Ann Arbor, MI: Michigan
University Press.
Baumol, W. J., & Bowen, W. G. (1966). Performing arts: The economic dilemma. New York: The
Twentieth Century Fund.
Ben-Ner, A., & Van Hoomissen, T. (1993). Independent organizations in the mixed economy: A demand
and supply analysis. In A. Ben-Ner & B. Gui (Eds.), The independent sector in the mixed economy.
Ann Arbor, MI: The University of Michigan Press.
Black, D. (1948). On the rationale of group decision-making. Journal of Political Economy, 56, 23–34.
Borgonovi, F. (2006). Do public grants to American theaters crowd-out private donations? Public Choice,126, 429–451.
Galaskiewicz, J. (1985). Social organization of an urban grants economy: A study of businessphilanthropy and nonprofit organizations. Orlando, FL: Academic Press.
Gronbjerg, K. (1987). Patterns of institutional relations in the welfare state: Public mandates and the
nonprofit sector. Journal of Voluntary Action Research, 16, 64–80.
Hansmann, H. (1987). Economic theories of nonprofit organizations. In W. Powell (Ed.), The nonprofitsector: A research handbook (pp. 27–42). New Haven, CT: Yale University Press.
Horne, C., Johnson, J., & Van Slyke, D. M. (2005). Do charitable donors know enough—and care
enough—about government subsidies to affect private giving to nonprofit organizations? Nonprofitand Voluntary Sector Quarterly, 34(1), 136–149.
Hotelling, H. (1929). Stability in competition. Economic Journal, 39(153), 41–57.
Krugman, P. (1991). History and industry location: The case of the manufacturing belt. The AmericanEconomic Review, 81(2), 80–83.
Kunemund, H., & Rein, M. (1999). There is more to receiving than needing: Theoretical arguments and
empirical explorations of crowding in and crowding out. Ageing and Society, 19, 93–121.
Lijphart, A. (1999). Patterns of democracy: Government forms and performance in thirty six countries.
New Haven, CT: Yale University Press.
Lune, H., & Oberstein, H. (2001). Embedded systems: The case of HIV/AIDS nonprofit organizations in
New York City. International Journal of Voluntary and Non-Profit Organizations, 12(1), 17–33.
Motel-Klingebiel, A., Tesch-Roemer, C., & Von Kondratowitz, H. J. (2005). Welfare states do not crowd
out the family: Evidence fro mixed responsibility from comparative analyses. Ageing and Society,25, 863–882.
Pedhazur, E. J. (1982). Multiple regression in behavioral research. Fort Worth, TX: Holt, Rinehart and
Winston.
Salamon, L. (1987). Partners in public service: The scope and theory of government-nonprofit relations.
In W. Powell (Ed.), The independent sector: A research handbook (pp. 99–117). New Haven, CT:
Yale University Press.
Voluntas
123
Salamon, L., & Anheier, H. K. (1998). Social origins of civil society: Explaining the nonprofit sector
cross-nationally. International Journal of Voluntary and Non-Profit Organizations, 9(3), 213–248.
Salamon, L., Anheier, H., List, R., Toepler, S., Sokolowski, W., et al. (1999). Global civil society:Dimensions of the nonprofit sector. Baltimore: The Johns Hopkins Center for Civil Society Studies.
Salamon, L. M., & Wojciech Sokolowski, S. (2003). Institutional roots of volunteering: Toward a macro-
structural theory of individual voluntary action. In P. Dekker & L. Halman (Eds.), The values ofvolunteering: Cross-cultural perspectives (pp. 71–90). The Netherlands: Kluwer/Plenum Press.
Salamon, L., Sokolowski, W., et al. (2004). Global civil society: Dimensions of the nonprofit sector (Vol. 2).
West Hartford, CT: Kumarian Press.
Simmons, W., & Emanuele, R. (2004). Does government spending crowd out donations of time and
money? Public Finance Review, 32(5), 498–511.
Snow, D. A., Rochford, E. B., Worden, S. K., & Benford, R. D. (1986). Frame alignment process,
micromobilization, and movement participation. American Sociological Review, 51, 464–481.
Sokolowski, W. (2000). The discreet charm of the nonprofit form: Service professionals and the nonprofit
organizations in Poland 1989–1993. International Journal of Voluntary and Non-Profit Organiza-tions, 11(2), 141–159.
Sokolowski, W. (2010). Philanthropic leadership in totalitarian and communist societies. In K. A. Agard
(Ed.), Leadership in nonprofit organizations. Thousand Oaks, CA: Sage.
Steinberg, R. (1985). Empirical relations between government spending and charitable donations. Journalof Voluntary Action Research, 14, 54–64.
United Nations. (2003). Handbook on non-profit institutions in the system of national accounts. New
York: United Nations.
Weber, M. (1978). Economy and society. Berkeley, CA: University of California Press.
Weisbrod, B. (1977). The voluntary independent sector. Lexington, MA: Lexington Books.
Weisbrod, B. (1980). Private goods, collective goods: The role of the nonprofit sector. In K. Clarkson &
D. Martin (Eds.), The economics of nonproprietary organizations (pp. 130–179). Greenwich: JAI
Press.
Weisbrod, B. (Ed.). (1998a). To profit or not to profit: The commercial transformation of the nonprofitsector. Cambridge, NY: Cambridge University Press.
Weisbrod, B. (1998b). Modeling the nonprofit organization as a multiproduct firm: A framework for
choice. In B. Weisbrod (Ed.), To profit or not to profit: The commercial transformation of thenonprofit sector (pp. 47–64). Cambridge, NY: Cambridge University Press.
Voluntas
123