Voluntas: InternationalJournal of Voluntary and Nonprofit Organizations, Vol. 9, No. 3, 1998
Social Origins of Civil Society: Explaining the
Nonprofit Sector Cross-Nationally
Lester M. Salamon1,3
and Helmut K. Anheier2
Recent research has usefully documented the contribution that nonprofit
organizations make to "social capital" and to the economic and political
development it seems to foster. Because of a gross lack of basic comparative
data, however, the question of what it is that allows such organizations to
develop remains far from settled. This article seeks to remedy this by testing
five existing theories of the nonprofit sector against data assembled on eight
countries as part of the Johns Hopkins Comparative Nonprofit Sector Project.
The five theories are: (a) government failure/market failure theory; (b)
supply-side theory; (c) trust theories; (d) welfare state theory; and (e)
interdependence theory. The article finds none of these theories adequate to
explain the variations among countries in either the size, the composition, or
the financing of the nonprofit sector. On this basis it suggests a new theoretical
approach to explaining patterns of nonprofit development among countries—
the "social origins" approach—which focuses on broader social, political, and
economic relationships. Using this theory, the article identifies four "routes"
of third-sector development (the liberal, the social democratic, the corporatist,
and the statist), each associated with a particular constellation of class
relationships and pattern of state-society relations. The article then tests this
theory against the eight-country data and finds that it helps make sense of
anomalies left unexplained by the prevailing theories.
KEY WORDS: nonprofit theory; social origins theory; comparative methodology; cross-national
comparisons.
1
Center for Civil Society Studies, Institute for Policy Studies, The Johns Hopkins University,
Baltimore, Maryland.
2
Center for Social Research and Instruction, Department of Sociology, Rutgers University,
New Brunswick, New Jersey, and Institute for Policy Studies, The Johns Hopkins University,
Baltimore, Maryland.
3
Correspondence should be directed to Lester M. Salamon, Center for Civil Society Studies,
Institute for Policy Studies, The Johns Hopkins University, Wyman Park Building, 5th Floor,
3400 North Charles Street, Baltimore, Maryland 21218; e-mail: lsalamon@jhunfac.hcf.jhu.edu
213
0957-8765/98/0900-0213$15.00/l © 1998 International Society for Third-Sector Research and The Johns Hopkins University
INTRODUCTION
Recent research has documented the vital importance that "social capital,"
and therefore the nonprofit associations thought to produce it, play in
both political and economic life (Putnam, 1993; Fukuyama, 1995). Still far
from settled, however, is the question of what it is that encourages, or allows,
such associations to develop. To be sure, theories about this abound. Almost
totally lacking, however, have been the data to put these theories to meaningful
empirical test. Even the most basic information about the scale, structure,
and composition of the nonprofit sector has been unavailable in most
countries, and what data have been available have been, as one scholar recently
put it, "hardly compatible or comparable" (Seibel, 1990, p. 43). The
result has been an open season for nonprofit theory building. With no real
possibility to subject such theories to systematic test, the only real constraint
on their development has been surface plausibility,the ability to formulate
a logical "story" explaining how the theory might work. As a consequence,
theories have proliferated in number, and often imaginativeness, without demonstrable
improvement in actual explanatory power.
The recent completion of Phase I of the Johns Hopkins Comparative
Nonprofit Sector Project has changed this situation fundamentally,creating
at least an initial cross-national database on the nonprofit sector (Salamon
and Anheier, 1996; Salamon et al., 1996). For the purposes of this project,
data were collected on the basic dimensions of the nonprofit sector in eight
countries that differ markedly with respect to a number of factors thought
to affect the role of the nonprofit sector, such as culture, religion, degree
of heterogeneity, and level of government social welfare spending. Included,
as reflected in Table I, were countries in North America, Western
Europe, Central Europe, and Asia.4
The resulting database creates a first
opportunity to subject nonprofit-sector theories to more serious testing than
has been possible heretofore, and thus to shed new light on the factors
that seem to account for the growth and development of this sector in an
assortment of countries throughout the world.
The purpose of this article is to present the results of such a testing.
To do so, the discussion first identifies the major findings of this cross-national
research, with particular emphasis on the variations in nonprofit size,
composition, and finance that emerge. It then outlines the six bodies of
4
The eight countries covered by this research were the United States, the United Kingdom,
France, Germany, Italy, Sweden, Hungary, and Japan. Because of its transitional character
we decided to drop Hungary from the analysis reported here. In addition to these eight
countries, five developing nations—Brazil, Ghana, Egypt, Thailand, and India—were also
included in the project, but resource limitations made it impossible to generate a full
empirical picture of the nonprofit sector in these countries.
214 Salamon and Anheier
Social Origins Theory
theory that can potentially shed light on these variations are identified.
Four of these come from the nonprofit-sector field and two emerge from
other theoretical traditions. These theories are then "tested" against the
cross-national data we have assembled. A concluding section, finally, outlines
some of the implications that seem to flow from this analysis both
for our conceptual understanding of the nonprofit sector and for further
research and theory-testing.
Given the limited number of countries for which we have solid empirical
data, serious caveats must be entered about the extent to which the conclusions
reached here are in any sense final. Nevertheless, the data seem to
suggest that many of the prevailing nonprofit theories, though useful as heuristic
devices, are too sweeping and one-dimensional, or have too restrictive
"boundary conditions," to account adequately for the tremendous complexity
of cross-national experience in this field. Instead, we suggest the usefulness
of what we term the "social origins" approach, which explicitly acknowledges
what one author has termed the nonprofit sector's "embeddedness" in
broader social, political, and economic realities (Seibel, 1990). Given this embeddedness,
the explanation of the shape of the nonprofit sector must be
found in these broader realities. This makes the study of nonprofit organizations
far more complex, but it also usefully integrates the field more fully
into the study of other aspects of social relations, where it clearly belongs.
WHAT'S TO EXPLAIN?
To understand this tentative conclusion, it is necessary to begin by
clarifying what it is that we expect nonprofit theories to explain. This requires
at least a brief summary of the general approach we pursued and
the major findings that emerged.
215
Table I. Country Coverage
Country
Sweden
France
Italy
Germany
United Kingdom
United States
Japan
Government Social Welfare
Spending as % of GDP
36
29
23
23
20
13
12
Key Dimensions of Interest
Two issues had to be resolved in order to assess the scope and structure
of nonprofit activity in the countries we targeted for scrutiny: first,
the basic unit of analysis to use; and second, the dimensions of this unit on
which to focus.
The Unit of Analysis
So far as the basic unit of analysis is concerned, the fundamental issue
in cross-national nonprofit-sector research is whether it is meaningful to
talk about a distinctive "sector" here at all, and, if so, how it should be
defined. Skeptics contend that no such entity really exists, that at best there
are particular types of nonprofit organizations that may exist in different
national settings, but that cross-national comparisons at the sector-wide
level are misleading or impossible. Even those who concede the existence
of a nonprofit "sector" in particular national settings often remain skeptical
that such an entity can be found elsewhere with sufficient conceptual similarity
to sustain meaningful cross-national comparisons. Cross-national nonprofit
research, not to mention theory-testing, in this view, is a fool's errand,
with little hope of meaningful results.
To cope with this problem, we worked with the national researchers
included in our study to identify the common features of a set of institutions
that we could confidently locate in all 13 project countries and to which
we could reasonably refer, following United Nations usage, as a distinctive
"nonprofit sector"5
(United Nations, 1993). As detailed more fully elsewhere
(Salamon and Anheier, 1992a, 1996), five crucial structural or operational
features emerged as the defining features of the set of entities
encompassed within this sector. In particular, such entities had to be:
• Organizations, i.e., institutionalized to some meaningful extent;
• Private, i.e., institutionally separate from government;
• Non-profit-distributing, i.e., not returning profits generated to their
owners or directors;
• Self-governing, i.e., equipped to control their own activities;
• Voluntary, i.e., involving some meaningful degree of voluntary par-
ticipation.
We then developed a classification scheme to differentiate the many
types of institutions that share these characteristics and validated this
5
Our definition of this sector differs markedly from the one used in the System of National
Accounts, however. For further detail, see Salamon and Anheier (1996).
216 Salamon and Anheier
scheme through the same collaborative process. Following United Nations
usage again, the basis of this classification was the "principal activity" of
the entity (for example, education, health). On this basis, 12 broad groupings
of nonprofit organizations were identified. This made it possible to
test various theories not only at the aggregate "sector" level but also at
the "subsector" level.6
Key Variables
So far as the variables are concerned, of particular interest were two
critical dimensions: first, the scale of nonprofit operations, as measured
principally by employment, volunteer time, and operating expenditures; and
second, the sources of nonprofit financial support, including "donative" or
contributed income, public-sector support, and revenue from the sale of
services or other products.7
Major Findings
Out of this work emerged a number of major conclusions about the
scope, structure, and financing of nonprofit activity in the countries we examined.
Most important for our purposes here were four of these conclusions
(for further detail on the conclusions of this research, see Salamon
and Anheier, 1996, 1997; Salamon et al., 1996).
First, and most fundamentally, the nonprofit sector as defined above
was found to be a major economic force in the countries we examined.
Thus, these organizations employed 11.9 million employees in the eight
countries. This represented 4.5% of the total labor force in these countries,
or close to one out of every 20jobs, and one out of every 8 service industry
jobs. In addition, these organizations attracted the energies of volunteers
whose time translates into the equivalent of close to 5 million additional
full-time employees.
Second, the size of the nonprofit sector varies significantly from place
to place. Thus, for example, nonprofit employment varied from a low of
6
Of these 12 groupings, data were collected on 10. The two excluded groupings were political
parties and religious congregations. These were excluded largely to keep the work manage-
able.
7
Several other facets of nonprofit operations could also have been examined, including the
number of organizations, the beneficiaries they serve, the composition of their governing
boards, the size of their memberships, their internal styles of operation, their adherence or
nonadherence to certain values, and many more. The variables we selected provide, however,
the most basic measures of the scope and structure of this sector and therefore seemed the
most fruitful to pursue at this stage of theory building.
Social OriginsTheory 217
218 Salamon and Anheier
0.8% of total national employment in Hungary to a high of 6.9% in the
United States, with Italy, Japan, and Sweden grouped toward the low end
of the spectrum, and Germany, the U.K., and France toward the higher
end (see Fig. 1).
While some of these variations seem understandable given conventional
theories, others seem counterintuitive. For example, as a share of
total employment, the nonprofit sector turns out to be larger in Sweden
than in Italy despite the existence of a full-fledged welfare state in the
former and the preservation of a substantial, church-run social welfare sector
in the latter. When volunteer staff are included, moreover, some interesting
changes occur. Most notably, Sweden vaults ahead of the other
European countries in the scale of its nonprofit sector. In particular, from
one of the smallest nonprofit sectors in Europe, albeit larger than the case
of Italy, it jumps to one of the largest. Evidently the nonprofit sector takes
a different form, and potentially assumes a different role, in Sweden as
opposed to the other countries we are examining, a point to which we will
return below.
Third, the nonprofit sector also varies in composition from place to
place, though the extent of this variation is somewhat "constrained." Specifically,
four components—education and research, health, social service,
and culture and recreation—seem to dominate the sector almost everywhere,
accounting together for at least 75% of sector expenditures in seven
of the eight countries, and in the only exception (Sweden) they account
Fig. 1. Nonprofit employment as percent of total employment, 1990.
for well over half. Nevertheless, considerable variation is also apparent
within this overall pattern. Education dominates the expenditures of the
nonprofit sector in Japan and the U.K., health in the U.S. and Germany,
social services in France and Italy, and recreation and culture in Sweden
and Hungary. This underlines the importance of testing alternative theories
not only at the sector level but also at the level of the subsectors to which
they can be expected to be most applicable.
Fourth, the revenue structure of this sector also varies among countries.
Overall, the major source of income for the nonprofit sector in our
eight countries was fees and service charges (49%), followed by government
(41%) and private giving (10%). At the aggregate level, this general pattern
holds for six of the eight countries, with the exceptions (Germany and
France) funded primarily by government (Table II). Even more significant
variations in revenue structure were evident among subsectors. Thus, for
example, earned income, though the dominant source of income overall,
is the dominant source of income for at least six countries in only 4 of 11
fields covered (Table II).
How can we explain these variations? More specifically, to what extent
do various theories of the nonprofit sector shed useful light on them? To
answer these questions, the next section examines available theories and
seeks to identify the explanatory variables and resulting hypotheses to
which they draw our attention. In a subsequent section we then "test" these
theories against the available evidence.
Table II. Patterns of Funding Nonprofit Organizations, by Field, by Number of Countries
Funding Patterns
Field
Culture and Recreation
Education Research
Health
Social Services
Environment
Development, Housing
Civic, Advocacy
Philanthropy
International
Business, Professional
Other
Total
Private Fees
and Payment
Dominant
8
4
2
1
6
5
4
6
1
8
3
6
Public-Sector
Payment
Dominant
0
4
5
6
2
2
4
2
3
0
0
2
Private
Donations
Dominant
0
0
1
1
0
1
0
0
4
0
1
0
Total
8
8
8
8
8
8
8
8
8
8
4
8
Source: The John Hopkins Comparative Nonprofit Sector Project.
Social Origins Theory 219
220 Salamon and Anheier
THE CANDIDATES: SIX THEORIES IN SEARCH OF VALIDATION
Broadly speaking, six bodies of theory seem especially relevant to the
question of the presence, scale, composition, and financing of nonprofit organizations
in different parts of the world. Four of these theories focus
squarely on the nonprofit sector field and two are borrowed from other fields
because of the potential contribution they can make to the subject at hand.
1. Government Failure/Market Failure Theory
Perhaps the dominant theoretical perspective in the nonprofit field until
relatively recently has been the one developed by economist Burton
Weisbrod to reconcile the persistence of nonprofit organizations with classical
economic theory (Weisbrod, 1977). The starting point for this "market
failure/government failure theory" is the inherent limitation, acknowledged
by classical economics, in the market's ability to supply sufficient quantities
of "public goods," i.e., goods that are available to all whether or not they
pay for them. In classical economics, this shortcoming of markets serves
as a major justification for government. But, Weisbrod points out that
where considerable differences of opinion exist about which public goods
to produce, public supply, in a democracy, will tend to reflect the preferences
of the median voter only, and considerable unsatisfied demand for
public goods will consequently persist. Such "government failure" is most
likely, therefore, the more heterogeneous the population. In such circumstances,
people will turn to nonprofit organizations to supply the public
goods they cannot secure through either the market or the state.
Three specific hypotheses flow from this line of theory. In the first
place, we would expect the size of the nonprofit sector in a country to vary
directly with the degree of heterogeneity in the population, as measured
in terms of religious or ethnic diversity. While this relationship could be
expected to hold at the aggregate level, it could be expected to apply with
particular force to the field of education, where concerns about particular
ethnic, religious, and cultural norms are typically most intense. Hence:
Hypothesis 1A1: The greater the diversity of a population, the larger
the nonprofit sector.
Hypothesis 1A2: The greater the diversity of the population, the larger
the education component of the nonprofit sector.
In the second place, this line of thinking would lead us to expect an
inverse relationship between the size of the nonprofit sector and the scale
of governmental provision of collective goods. This is so because this theory
predicts that the nonprofit sector is a response to failures of the state to
provide the kinds of collective goods that people want but that the market
is unable to provide. To the extent that the government is providing such
collective goods, the need for nonprofit provision would decline. Hence:
Hypothesis 1B1: The greater the level of government social welfare
spending, the smaller the nonprofit sector.
Hypothesis 1B2: The greater the level of government social welfare
spending, the smaller the nonprofit education sector.
Finally, the market failure/government failure theory would lead us to
expect that the nonprofit sector would be funded mostly out of private charitable
contributions. This is so because this theory views the nonprofit sector
as emerging from demands for public goods not being met by either the
market or the state. Under these circumstances, there would be no reason
to expect the resulting nonprofit organizations to be financed either through
market transactions or governmental subsidies. To the contrary, to the extent
such demands are satisfied beyond the realms of either government or the
market, the most likely source of support is voluntary contributions. Hence:
Hypothesis 1C1: The greater the diversity of the population, the
greater nonprofit reliance on private giving as a source of support.
Hypothesis 1C2: The greater the diversity of the population, the
greater the reliance of nonprofit education organizations on private
giving as a source of support.
2. Supply-Side Theory
Where the "market failure/government failure" theory focuses exclusively
on the unsatisfied demand for public goods left by failures of the
market and the state, a second body of theory treats this as a necessary
but not sufficient condition to explain the variations in nonprofit development.
According to this "supply-side" theory, a second condition is needed
for nonprofit organizations to emerge: namely, the presence of "social entrepreneurs,
people with an incentive to create nonprofit organizations to
meet such demand (James, 1987). The appearance of such individuals is
not random, moreover. It is most likely under particular circumstances. One
of the most common of these circumstances is the presence of religious
competition, where one or more religious bodies vie to attract adherents
by providing services such as health care or education.
Social Origins Theory 221
This line of theory would lead us to expect the nonprofit sector to be
most highly developed where religious competition is most intense.8
What is
more, this relationship should apply particularly strongly to the field of education
since this is where religious competition is often most evident. Hence:
Hypothesis 2A1: The greater the level of religious competition, the
larger the nonprofit sector.
Hypothesis 2A2: The greater the level of religious competition, the
larger the nonprofit education sector.
This supply-side theory also has implications for expected patterns of
nonprofit finance. In particular, like the market failure/government failure
theory on which it depends, the supply side theory assumes that charitably
inclined individuals have an opportunity to win adherents by founding nonprofit
institutions in fields where neither the market nor the state is providing
needed services or support. It follows that this line of theory would posit a
positive relationship between the potential supply of nonprofit entrepreneurs
and the extent of nonprofit reliance on private giving, and this relationship
can be expected to apply most forcefully in the education field. Hence:
Hypothesis 2B1: The greater the extent of religious competition, the
greater the nonprofit reliance on private giving as a source of support.
Hypothesis 2B2: The greater the extent of religious competition, the
greater the reliance of nonprofit education organizations on private
giving as a source of support.
3. Trust Theories
A third line of theoretical analysis finds the source of nonprofitactivity
in another form of market failure: "contract failure" arising from "information
asymmetries" often facing consumers when they lack the information
they need to judge the quality of the goods or services they are purchasing.
This can occur because the purchaser is not the same person as the consumer
(for example, the purchase of nursing home care by children for an
elderly parent), because the service in question is inherently complex and
difficult to assess, or for other reasons (Krashinsky, 1986). In such cases,
purchasers seek alternative bases for trust in the quality of the resulting
service. One such basis is the nonprofit sector because of the "nondistribu-
8
Proselytuing clergy are, of course, only one type of "social entrepreneurs" with an incentive
to create nonprofit organizations. Identifying alternative types without falling into tautology
problems is difficult, however.
222 Salamon and Anheier
tion constraint," the prohibition on nonprofit distribution of profits to owners,
which may be perceived as a sign of trustworthiness since those involved
in nonprofit organizations are less likely to be in a field solely for the money
(Hansmann, 1980, 1987).9
Several hypotheses flow from this line of thinking. In the first place,
we would expect the scale of the nonprofit sector to vary inversely with
the level of trust in the business sector in a society. The greater the level
of that trust, the more confident people will be to secure the services they
need through the market system and therefore the less they will feel obliged
to turn to the nonprofit sector. What is more, this will be particularly true
in the fields of health and social services, where information asymmetries
are likely to be most acute. Hence:
Hypothesis 3A1: The higher the level of trust in business in a society,
the smaller the nonprofit sector.
Hypothesis 3A2: The higher the level of trust in business in a society,
the smaller the nonprofit health and social services subsectors.
Because the services secured through the nonprofit sector according
to this theory would likely be purchased from the business sector if sufficient
trust were present, it follows that the financing of these services
through the nonprofit sector is likely to take a more commercial form as
well. This suggests that the absence of trust would be associated with a
nonprofit sector able to secure a larger share of its income from fees and
service charges. Hence:
Hypothesis 3B1: The lower the level of trust in business in a society,
the more likely the nonprofit sector is to be financed by fees and service
charges.
Hypothesis 3B2: The lower the level of trust in business in a society,
the more likely the nonprofit health and social services subsectors are
to be financed by fees and service charges.
4. Welfare State Theory
The preceding theories all take as given that the expansion of stateprovided
welfare services displaces nonprofit organizations, but they leave
unanswered the question of what accounts for the expansion of state-provided
social welfare services in the first place. To answer this question, it
9
A variant of this theory views nonprofit organizations as vehicles for affected stakeholders
to gain delivery of services to third parties (Ben-Ner and Gui, 1993).
Social Origins Theory 223
224 Salamon and Anheier
is necessary to resort to the substantial body of literature that has emerged
on the modern welfare state.
Generally speaking, this literature treats the nonprofit sector, at best,
as a residual category. In fact, conventional welfare state approaches all
but ignore the role of the nonprofit sector both historically and in relation
to current policy debates and developments (Flora and Heidenheimer,
1981; Offe, 1984; Hicks and Swank, 1992; Huber et al., 1993). To the extent
that they treat the nonprofit sector at all, these theories view it as an essentially
premodern mechanism for dealing with social problems that, like
the family, the Church, and the feudal lord, lose their function with the
growth of the market system and the expansion of the state's responsibilities
to deal with economically extraneous populations. Thus, for example,
Quadagno (1987, p. 112) points out that "as industrialization proceeds, it
creates new needs for public spending by reducing the functions of the
traditional family and by dislocating certain categories of individuals whose
labor becomes surplus—thevery young, the old, the sick, the disabled . . .
Because traditional societal institutions are unable to meet the needs of
these vulnerable individuals, the state expands more or less automatically."
In this line of thinking, the expansion of the state is an almost automatic
by-product of economic development (Flora and Alber, 1981). It was for
this reason that Titmuss (1974) labeled countries such as the United States
as "residual welfare states," since they had not yet established the level of
welfare guarantees that their level of economic development would suggest
and retained premodern forms of reliance both on market and nonprofit
solutions to social welfare problems.
Traditional welfare state theory would thus lead us to expect that the
greater the level of economic development, the more extensive the state
provision of social welfare services; and the more extensive the state provision
of social welfare services, the smaller the nonprofit sector. Hence:
Hypothesis 4A1: The higher the level of income per capita, the smaller
the nonprofit sector in a country.
Hypothesis 4A2:The higher the level of income per capita, the smaller
the health, social service and education component of the nonprofit
sector in a country.
5. Interdependence Theory
Both the welfare state theory and the market failure/governmentfailure
thesis that underlies the heterogeneity and supply-side theories take as given
that the relationship between the nonprofit sector and the state isfundamen-
tally one of conflict and competition. The persistence of a nonprofit sector,
in this view, is a by-product, at best, of inherent limitationsof the state; and,
at worst, of the state's inability to obliterate all bases of pluralism and diversity.
Depending on the political persuasion of the observer, the extensive presence
of the nonprofit sector is therefore seen as a boon to liberty or a denial
of basic social protections and hence a source of persistent inequality.
This "paradigm of conflict" (Salamon, 1995) is not the only way to view
the relationship between government and the nonprofit sector, however. Sideby-side
with the potential sources of conflict are important elements of potential
interdependence and partnership as well. Indeed, there are strong
reasons to expect the latter to dominate the relationship (Salamon, 1987a,b).
For one thing, nonprofit organizations are often active in a field before government
can be mobilized to respond. They therefore often develop expertise,
structures, and experience that governments can draw on in their own activities.
Beyond that, nonprofit organizations often mobilize the political support
needed to stimulate government involvement,and this support can often be
used to ensure a role for the nonprofit providers in the fields that government
is persuaded to enter. Finally,for all their advantages, nonprofit organizations
have their own significant limitations that constrain their ability to respond
to public problems. In addition to "market failure" and "government failure,"
in other words, there is "voluntary failure"—i.e., inherent limitations of the
voluntary sector as a mechanism for meeting public needs (Salamon, 1987a).
These include "philanthropic insufficiency," the difficulty voluntary organizations
have in generating the resources often needed to "scale up" their operations;
and "philanthropic paternalism," the inability these organizations
have in establishing rights to benefits and hence the difficulty they have in
fostering a true sense of self-reliance.
Under certain circumstances, therefore, close cooperative relationships
can be forged between the nonprofit sector and the state in addressing
public problems. This can occur where, for ideological or other reasons,
resistance to direct state action is strong yet demands arise for protections
from particular social or economic ills; or where support from those associated
with the nonprofit sector becomes crucial to enhance the role and
power of the state. Under either circumstance, we would expect a positive
relationship between government spending and the scope of the nonprofit
sector. What is more, since government is most likely to turn to the nonprofit
sector for assistance in delivering basic human services, we would
expect this relationship to hold most strongly in the fields of health and
social services. Hence:
Hypothesis 5A1: The greater the government social welfare spending,
the larger the nonprofit sector.
Social Origins Theory 225
Hypothesis 5A2: The greater the government social welfare spending,
the larger the nonprofit health and social services subsectors.
Because government is viewed under this theory as a potential source
of financial, and not just political, support for the nonprofit sector, moreover,
it follows that if this theory holds, the government share of nonprofit
income also should be higher where overall government spending is higher.
Hence:
Hypothesis 5B1: The greater the government social welfare spending,
the higher the government share of nonprofit income.
Hypothesis 5B2: The greater the government social welfare spending,
the higher the government share of nonprofit health and social service
subsector income.
6. Social Origins Theory
Although the interdependence theory outlined above acknowledges
the possibility of a cooperative relationship between the nonprofit sector
and the state, it does not really specify the circumstances under which such
a relationship is most likely to emerge. Similarly, the other theories outlined
above assume a degree of flexibility in institutional choice that seems belied
by the historical record. Choices about whether to rely on market, nonprofit,
or state provision of key services are not simply made freely by consumers
in an open market as advocates of the economic theories seem to
assume. Rather, these choices are heavily constrained by prior patterns of
historical development that significantly shape the range of options available
at a given time and place.
This perspective on the dynamics of institutional choice is forcefully
reflected in the work of Barrington Moore Jr. (1966) on the "social origins"
of fascism and democracy and, more recently, in the work of G0sta EspingAndersen
(1990) on the origins of the modern welfare state. Central to
this argument is the notion that complex social phenomena—for example,
the emergence of the "welfare state" or "democracy"—cannot be easily
understood as the product of the unilinear extension of a single factor,
such as industrialization, diversity, or education. Rather, much more complex
interrelationships among social classes and social institutions are involved.
This is so, as Rueschemeyer et al. (1992, p. 5) have recently noted,
because such phenomena are, above all, matters of power. As such, they
reflect the balance of power among social classes, between state and society,
and even among nations.
226 Salamon and Anheier
Using this mode of analysis, Harrington Moore Jr. was able to discern
in the historical records of England, France, Germany, Japan, China, and
India three distinct "routes to the modern world"—democratic, fascist, and
communist—each of which could be attributed to a particular constellation
of relationships among landed elites, the rural peasantry, urban middleclass
elements, and the state. Thus, for example, according to Moore, the
emergence of democracy, and of what we would now term a vibrant "civil
society," is most likely where three preeminent factors are at work: first,
a royal absolutism held in rough check by strong landed elites; second, the
emergence of a vigorous and independent urban middle class to challenge
the power of the landed elements themselves; and third, a solution to the
agrarian problem that releases the mass of the population from the land
(Moore, 1966, pp. 413-432).
Esping-Andersen (1990) has employed a similar mode of analysis to
explain the appearance of three more or less distinct types of welfare "regimes":
(1) a "liberal" welfare state common in the Anglo-Saxon countries
and characterized by limited, means-tested assistance with strict entitlement
rules; (2) a "corporatist" welfare state more common on the continent of
Europe in which the state supplies welfare assistance but preserves many
of the status differences of premodern society; and (3) a "social democratic"
welfare state in the Nordic countries involving universalism and a
separation of welfare provision from the market system ("decommodification").
According to Esping-Andersen (1990, p. 110), the first of these took
shape as a result of the weakness of absolutism and the dominance of a
laissez faire-inspired bourgeoisie in the Anglo-Saxon countries. The "corporatist"
model, by contrast, emerged because of the extraordinary power
of the Church, the aristocracy, and the state in continental European capitalism.
Finally, the social welfare model emerged where working-class elements
were sufficiently strong, and where they managed to forge effective
alliances with the middle class.
While neither Moore nor Esping-Andersen applies his analysis to the
question of the appearance and growth of the nonprofit sector, there are
strong reasons to believe that the mode of analysis they utilize should have
considerable relevance to this question. As Seibel (1990, p. 46) has reminded
us, nonprofit organizations "are not only providers of goods and services but
important factors of social and political coordination." As a consequence,
they do not float freely in social space. Rather, they are firmly "embedded"
in prevailing social and economic structures, often serving as "the knotswithin
networks of elites with reputation, finance, and power." Work on the Johns
Hopkins Comparative Nonprofit Sector Project has made clear that such organizations
have deep historical roots in virtually every one of the societies
we examined. The interrelationships between this sector and the Church, to
Social Origins Theory 227
cite just one well-known example, have been especially pervasive, and nonprofit
institutions have therefore been fully caught up in the broader struggles
between the Church and secular authorities that animated European history
for much of the Reformation and Enlightenment, and that survived well into
the 20th century. Where Church authorities reached a rapprochement with
civil authorities, as was the case, for example, in Bismarck's Germany, voluntary
institutions could be expected to play a much stronger role even with
the growth of state-sponsored welfare. By contrast, where Church authorities
were domesticated, as was the case in Sweden and Italy, or overwhelmed as
in the case of France, we would expect Church-related welfare to be absorbed
by the state and voluntary institutions to play a much smaller role, particularly
in the provision of human services.
This line of argument suggests that the contours of third-sector development
go beyond the simple "large" vs. "small" dimensions of standard
empirical research. Subtler questions about the role of this set of institutions
in prevailing social structures are also important. At the very least,
it should be clear that there is more than one route toward the creation
of a sizable nonprofit sector. Indeed, it is possible to identify four more
or less distinct "models of nonprofit development," four types of "nonprofit
regimes," each characterized not only by a particular state role but also by
a particular position for the third sector; and, most importantly, each also
reflecting a particular constellation of social forces. Needless to say, these
nonprofit-sector regime types are heuristic devices intended to demarcate
broad tendencies. Significant variations can therefore exist among countries
that fall in any particular grouping.
Table III differentiates these regimes in terms of two key dimensions:
first, the extent of government social welfare spending; and second, the scale
of the nonprofit sector. Thus, in the liberal model, low government social
welfare spending is associated with a relatively large nonprofit sector.10
This
outcome is most likely where middle-class elements are clearly in the as-
228 Salamon and Anheier
Table III. Models of Third-Sector Regime
Government Social
Welfare Spending
Low
High
Nonprofit Scale
Low
Statist
Social Democratic
High
Liberal
Corporatist
10
Although we utilize some of Esping-Andersen's terms here, it should be clear that we use
them to refer to different types of state-nonprofit relationships rather than to different types
of social welfare and pension policies as is done by Esping-Andersen.
cendance, and where opposition either from traditional landed elites or
strong working-class movements has either never existed or been effectively
held at bay. This leads to significant ideological and political hostility to the
extension of government social welfare protections and a decided preference
for voluntary approaches instead. The upshot is a relatively limited level of
government social welfare spending and a sizable nonprofit sector.
At the opposite extreme is the social democratic model. In this model,
state-sponsored and state-delivered social welfare protections are quite extensive
and the room left for service-providing nonprofit organizations is
quite constrained. This type of model is most likely where working-class
elements are able to exert effective political power, albeit typically in alliance
with other social classes. While the upshot is a limited service-providing
nonprofit sector, however, it is not necessarily a limited nonprofit
sector overall, as some accounts would suggest. To the contrary, given the
political battles involved in the extension of state-provided welfare protections,
we can expect nonprofit organizations to still be quite active in such
societies, but with a significantly different role. This is a role not as service
providers but as vehicles for the expression of political, social, or even recreational
interests.
In between these two models are two additional ones, which have
tended to be overlooked in prevailing theories, but which may actually be
most pervasive. As reflected in Table III, both of these are characterized
by strong states. However, in one, which we can characterize as the corporatist
model, the state has either been forced or induced to make common
cause with nonprofit institutions so that nonprofit organizations function
as one of several "premodern" mechanisms that are deliberately preserved
by the state in its efforts to retain the support of key social elites while
preempting more radical demands for social welfare protections. This suggests
that the relationship between government and the nonprofit sector
may be more curvilinear than linear, with the scale of the nonprofit sector
decreasing with the growth of government social welfare spending in "liberal"
regimes but increasing with the growth of government social welfare
spending in "corporatist" regimes.
The fourth possible model is the statist model. In this model the state
retains the upper hand in a wide range of social policies, but not as the
instrument of an organized working class, as in the social democratic regimes.
Rather it exercises power on its own behalf, or on behalf of business
and economic elites, but with a fair degree of autonomy sustained by long
traditions of deference and a much more pliant religious order. In such
settings, limited government social welfare protection does not translate
into high levels of nonprofit action, as in the liberal regimes. Rather, both
Social Origins Theory 229
government social welfare protection and nonprofit activity remain highly
constrained.
Although the predictions generated by this model are harder to convert
into testable propositions, such propositions are possible. Thus, for example,
this theory would lead us to expect that in liberal and social
democratic regimes the size of the nonprofit sector will vary inversely with
the scale of government social welfare spending whereas in corporatist and
statist regimes this relationship will be direct, with nonprofit organizations
growing as government grows. This is so because government and the third
sector are perceived as alternatives in the liberal and social welfare models
but as partners in the corporatist model.
Hypothesis 6A1: Where middle-class elements are strong and neither
landed elements nor the working class is in a position to mount an
effective challenge, a liberal regime is likely to emerge characterized
by limited state-provided welfare services and a relatively large nonprofit
sector.
Hypothesis 6A2: Where working-class elements mount an effective
challenge to middle-class power, a social democratic regime is likely to
emerge characterized by extensive governmental social welfare services
and a relatively small nonprofit sector.
Hypothesis 6A3: Where landed elements remain strong and confront
pressures from both middle-class and working-class elements, the prospects
are good for a corporatist outcome featuring sizable government
social welfare activity along with a relatively large private nonprofit
sector.
Hypothesis 6A4: Where conservative elements remain in the ascendance
and in control of the apparatus of the state, a statist solution is
most likely,with both government social welfare spending and the nonprofit
sector limited.
The social origins theory also has implications for the financial base
of the sector. In both liberal and social democratic regimes, the nonprofit
sector is likely to be financed more heavily by private charitable contributions.
This is so because in both regimes government and private funding
are perceived as alternative mechanisms for meeting public needs, leaving
the nonprofit sector to be funded, if at all, out of essentially private sources.
In the corporatist model, by contrast, extensive cooperation between government
and the nonprofit sector is anticipated, so governmental sources
are likely to be far more important. Finally, in the statist model, neither
230 Salamon and Anheier
private philanthropy nor state involvement is anticipated, leaving the nonprofit
sector to secure support, if at all, through essentially commercial activities.
Hence:
Hypothesis 6B: Where the social conditions conducive to either a liberal
or social democratic regime prevail, private giving is likely to play
a disproportionately large role in the funding base of the nonprofit
sector. Where the social conditions conducive to the corporatist regime
prevail, government support is likely to play a disproportionately large
role. And in circumstances conducive to statist regimes, fees and service
charges are likely to be more prominent.
APPLYING THE THEORIES: TOWARD AN EXPLANATION OF
NONPROFIT SIZE ANDFINANCES
How effectively do these theories account for the patterns of thirdsector
growth and composition reflected in our data? To what extent are
the factors identified in these theories systematically related to cross-national
differences in the size, composition, and revenue base of the nonprofit
sector? Although our data do not make it possible to answer these
questions definitively, they do allow at least a start in this direction.
Approach
To make such a start, we converted the various theories into operational
terms and tested them against the available data using a combination
of basic regression analysis and "scattergrams" designed to display the resulting
relationships in visual form. To ensure that we were measuring the
various variables in comparable units, we converted the values into z-scores,
which essentially measure the extent to which a country deviates from the
mean along a particular dimension.11
To avoid having our results affected
by variations in the overall size or structure of different national economies,
moreover, we used as our measure of nonprofit-sector size the share of
service-sector employment that nonprofit organizations represent in the dif-
11
More precisely, the z-score measures the number of standard deviations that a particular
country varies from the mean on a particular variable, e.g., nonprofit-sector size [i.e., z-score
= (observed value - 7-country average)/7-country standard deviation]. Positive scores
indicate a location above the mean, and negative scores indicate that a country's score ranks
below the mean.
Social OriginsTheory 231
ferent countries. This variable ranges from a low of 5.2% in Sweden to a
high of 15.4% in the United States, a quite considerable spread.12
Where
applicable, however, we also tested the hypotheses with volunteer employment
included. We report these results where they add significantlyto those
we obtained using paid employment only.
Findings I: PrevailingTheories
The central conclusion that emerges from this testing is that none of
the standard nonprofit theories seem adequate to account for the observed
variations in nonprofit scale or nonprofit finance. This general conclusion
holds true, moreover, whether we focus on the nonprofit sector as a whole
or on the particular subsectors where we might expect these theories to
work best. At the same time, however, several of these theories offer useful
insights into some of the basic dynamics that may be at work.
Support for this general conclusion can be found in Tables IV and V.
These tables report the results of testing the extent to which the prevailing
market failure/government failure, supply side, trust, welfare state, and interdependence
theories identified earlier account for cross-national variations
in the size and financing of the nonprofit sector, respectively. Each
table reports, for each theory, the hypotheses to which it gives rise, the
indicators used for the independent variables in these hypotheses, the direction
of the relationship hypothesized by the theory, the actual direction
observed in the data, and the extent of the variance accounted for by the
indicated variable.13
Market Failure/Government Failure Theory
As these tables show, the market failure/governmentfailure theory finds
perhaps the most consistent support in the data. Of sixhypotheses associated
with this theory, five find at least partial support, at least in the sense that
the observed relationship is in the direction hypothesized by the theory. In
at least three cases, moreover, the theory seems to explain a meaningful
share of the variance that exists. Thus, the greater the degree of diversity
of a country's population, which we measured in terms of an index of eth-
12
For the purpose of this discussion, the "service sector" was defined following International
Standard Industrial Classificationusage as the following industries: Major Group 8 (Finance,
Insurance, Business Services, and Real Estate) and Major Group 9 (Public Administration,
Community, Social, and Personal Social Services).
13
The figure reported here is the adjusted R2
.
232 Salamon and Anheier
nolinguistic diversity,14
the more likely a country's nonprofit sector is to be
supported by private giving.As Table V shows, this variable seems to explain
over 50% of the variation in the role that private givingplays in the financing
of a country's nonprofit sector, and this seems to hold both at the aggregate
level and in the education field where we expected it to have its greatest
explanatory power. This is consistent with this theory's emphasis on government
and market failures as the principal source of nonprofit development,
with the implication that the greater the degree of heterogeneity in a population,
the more likely there are to be unsatisfied demands for collective
goods that groups of people will seek to remedy through private charitable
activity. Similarly, there seems empirical support in these data for the proposition
that the greater the level of government social welfare provision, the
smaller the nonprofit sector that is needed (Hypothesis 1B1, Table IV),
which is another implication of this theory.
Although there is considerable support for the market failure/government
failure theory in these data, there are also significant ambiguities.
Thus, as Table IV shows, while the size of the nonprofit sector is positively
related to the level of heterogeneity, as this theory predicts, the relationship
fails to account for any measurable amount of the cross-national variation
in nonprofit size. Worse yet, in the education field, where we would expect
this theory to apply with special force, it turns out not to apply at all: Far
from increasing with the level of heterogeneity, as predicted, the size of
the nonprofit education subsector actually declines, as shown by the negative
coefficient for Hypothesis 1A2 in Table IV, and the negative slope of
the regression line in Fig. 2 What is more, this negative relationship is
fairly robust, accounting for 8% of the observed variance in the size of the
education subsector.
Similar ambiguities are apparent in the other test of the government
failure/market failure theory, which focuses on the level of government social
welfare spending. Thus, while the predicted negative relationship between
government social welfare spending and the size of the nonprofit
sector (Hypotheses 1B1 and 1B2) holds at both the sector level and in the
14
The index used here is a fractionalization index, which provides a way of measuring the
composition of a population in terms of specific attributes such as language groups, religions,
or ethnicity. The index is calculated as follows:
l-(E(x)/(Ex)2
)
where x is the size of the ethnolinguistic group as a percent of the total population. A fractionalization
index approaching 0 indicates that all people in a population belong to the
same group. By contrast, a fractionalization index approaching 1 indicates that the population
falls into many different groups. The index used here was computed from data available
in the Britannica World Data series (1991-1993).
Social Origins Theory 233
234 Salamon and Anheier
Social Origins Theory 235
236 Salamon and Anheier
Fig. 2. Ethnolinguistic heterogeneity and size of nonprofit education component
(numbers shown are standard deviations from the mean).
field of education, the relationship is curiously stronger at the sector level,
where it accounts for a substantial 22% of the variation in nonprofit size
among countries, than at the subsector level, where it accounts for almost
none of the variation.
Part of the explanation for these ambiguous results is apparent in Fig.
3, which shows the location of the countries we are examining in terms of
the two key variables of interest to this theory: the nonprofit share of service
employment and the extent of population diversity. As this figure shows,
four countries (the United States, Sweden, Italy, and, to a lesser extent, Japan)
lie far away from the values predicted by the regression line. In particular,
the U.S. nonprofit sector is larger than would be predicted by its
relative level of heterogeneity, and both Italy's and Sweden's are relatively
smaller. What is more, while France, Germany, and the United Kingdom
are much closer to the regression line, at least one—France—nevertheless
performs differently from the way this theory would predict. In particular,
though having a lower level of ethnolinguistic heterogeneity than the United
Kingdom, France nevertheless has a larger nonprofit sector, contrary to what
the market failure/government failure theory would predict.
In short, while this theory captures some of the realities at work, it hardly
captures them all. To the contrary, the central relationship predicted by this
theory—between the level of heterogeneity and the size of the nonprofit sector—finds
little empirical support and is in fact refuted by the evidence in
the field to which it is supposed to apply most powerfully. What is more, the
Social Origins Theory 237
Fig. 3. Ethnolinguistic heterogeneity and nonprofit sector size (numbers shown are
standard deviations from the mean).
theory leaves a variety of unexplained outliers that seem to defy its central
thrust.
Supply-Side Theory
To some extent, these anomalies are addressed by the "supply-side theory,"
which emphasizes the need for a sufficient supply of "moral entrepreneurs"
with an incentive to create nonprofit organizations as an additional
precondition for translating unsatisfied demands for collective goods into
the formation of such organizations. Because this theory argues that such
entrepreneurs are likely to come forward where religious groups are competing
for adherents, we used the degree of religious diversity in a country
as an indicator of the extent to which this supply-side theory would apply.
As Tables IV and V show, a strong positive relationship does seem to exist
between the degree of religious diversity in a country and the size of the
country's nonprofit sector (Hypothesis 2A1, Table IV). In addition, a robust
positive relationship exists between such diversity and the extent of nonprofit
reliance on private giving (Hypothesis 2B1, Table V).
Curiously, however, in the education field, where we would expect this
theory to apply with special force because it is commonly one of the most
crucial "battlegrounds" of the religious rivalries thought to produce the re-
quisite moral entrepreneurs, this relationship actually breaks down. Indeed,
far from increasing, the scale of the nonprofit education sector declines as
the level of religious diversityincreases, just the reverse of what this theory
would predict (Hypothesis 2A2, Table IV). Similarly, no significant relationship
exists in the education field between the degree of religious diversity
and the extent of nonprofit reliance on private giving, again contrary
to what this theory would predict (Hypothesis 2B2, Table V).
These results suggest that while religious competition may work handin-hand
with ethno-linguistic diversity to stimulate nonprofit growth, its independent
effects seem weak and somewhat inconsistent, applying most
powerfully at the aggregate level but failing to have much effect in the
field where it is supposed to operate with special force.
Trust and Welfare State Theories
Even these modest results outdistance the performance of the trust
and welfare state theories, however. So far as the former is concerned, we
found no measurable relationship between the relative degree of trust in
nonprofits as opposed to business in a country and either the size of the
nonprofit sector or the extent of nonprofit reliance on private fees to finance
their activities (Hypotheses 3A1, 3A2, 3B1, 3B2).15
What is more,
not only do these relationships not hold at the aggregate level, but also
they do not hold in the subfields of health, education, and social services
where we would particularly expect them to apply.
Similarly, the data summarized in Tables IV and V provide little support
for the "welfare state theories," which posit a negative relationship
between the size of the nonprofit sector and the levels of industrialization
and economic development in a country on the grounds that such developments
create both the need and the opportunity for extensive public systems
of social aid and thereby displace nonprofit institutions, especially in
the fields of health, education, and social services (Hypotheses 4A1 and
4A2). As it turns out, countries with similar levels of per capita income
have widely divergent levels of nonprofit employment. What is more, the
15
The level of trust in nonprofit organizations in a country was measured with the aid of data
available from the World Values Survey (World Values Study Group, 1994). This population
survey measured trust in major business firms and different types of public institutions on
a 4-point Likert scale. On the assumption that respondents used more or less the same
underlying scale in assessing trust levels for institutions such as corporations, government,
or churches, we used the difference between the reported trust in major corporations and
the reported average trust levels for all other institutions as our measure of the level of
trust in nonprofit organizations. To obtain national scores, we averaged individual scores
for all respondents in each of the seven countries.
238 Salamon and Anheier
Social Origins Theory 239
relationship is not much stronger in the fields of health, social services,
and education. Finally, this picture does not change when we take account
of volunteer labor as well. Evidently, the concept embodied in the welfare
state theories does not provide much help in explaining variations in nonprofit-sector
strength.
Interdependence Theory
The situation with respect to the "interdependence theory" is more
complicated. As already indicated in the discussion of the heterogeneity
theory above, the overall thesis embodied in this model—thatthe relationship
between government and the nonprofit sector may be more cooperative
than competitive, at least in some fields, so the scale of the nonprofit
sector may increase with the growth of government social welfare spending,
not decrease—does not find firm support in the data. The relationship between
nonprofit-sector size and the level of government welfare spending
turns out to be negative for both the nonprofit sector as a whole and for
the combined fields of education, health, and social services (Hypotheses
5A1 and 5A2, Table IV).
Fig. 4. Government social welfare spending and nonprofit sector scale (numbers
shown are standard deviations from the mean).
Although this model does not hold across the board, it does seem to
explain some of the anomalies that limit the explanatory power of the market
failure/government failure theory discussed above. This is evident in
Fig. 4, which shows the distribution of countries in terms of the size of
their nonprofit sectors and the levels of government social welfare spending.
As this figure shows, while this relationship is negative overall, there
are some notable outliers. In particular, both Germany and France display
much larger nonprofit sectors than their levels of government social welfare
spending would predict under the government failure/market failure theory.
The interdependence theory explains why this might be so: Governments
in these countries, far from displacing nonprofit organizations, have chosen
to forge partnerships with them instead and are actually helping to underwrite
and extend nonprofit operations.
Further support for this line of thinking is evident in the data on nonprofit
finance. At least in the human service sphere where the interdependence
theory is supposed to work most forcefully, the positive relationship
that this theory posits between government social welfare spending and the
government share of nonprofit revenue seems to hold, suggesting the presence
of an interdependent relationship between government and the nonprofit
sector in this sphere, as Hypothesis 5B2 predicted.
In short, these findings suggest some significant complementarity between
the heterogeneity theory and the interdependence theory. The
strengths and weaknesses of the one seem to be the mirror image of those
of the other. Instead of a single pattern, multiple patterns of nonprofit
development seem to exist. But why does one pattern apply in some cases
and another in other cases? To answer this question, we turn to the "social
origins" theory discussed above.
Table VI. Test of Social Origins Model of Nonprofit Sector
Government Social
Welfare Spending
Low
High
Nonprofit Scale
Low
Statist
Japan
Social Democratic
Sweden
Italy
High
Liberal
U.S.
U.K.
Corporatist
Germany
France
Source: Johns Hopkins Comparative Nonprofit Sector Project.
240 Salamon and Anheier
Findings II: Social Origins Theory
The "social origins" theory provides a bridge between the economy and
simplicity of the economic theories we have been considering and the complex
and often inconclusive world of comparative historical work. It does so by
positing a finite set of more or less distinct "models" or "regimes" that can
be traced to an identifiable, and predictable, set of social circumstances—
precisely what our analysis of alternative theories suggests is needed.16
As reflected in Table VI, our data give considerable support to this
social origins approach. As it turns out, all four of the nonprofit "regime
types" reflected in this theory turn out to be represented among the countries
we studied, and the explanations seem consistent with what the "social
origins" theory would suggest. In the first place, two of the countries, the
United States and the United Kingdom, fall into the category of what we
termed liberal nonprofit regimes. In both, government social welfare spending
is relatively low and the size of the nonprofit sector is relatively large.
In both, moreover, the social conditions that the social origins theory posits
should be associated with this pattern are very much inevidence—namely
a sizable urban middle class that effectively disrupted (or, in the case of
the United States, never really confronted) a landed upper class while holding
urban working-class elements at bay. The American middle class was
much more successful at this than its British counterparts, however, in part
because it never really had an entrenched landed elite to unseat, and in
part because ethnic and racial diversity kept the working classes more
highly splintered. Political support for a state-centered welfare system was
therefore far more muted in the American setting than in the British, where
pro-state attitudes fostered by the government's wartime successes and
working-class mobilization led to the establishment of certain key features
of a classic social democratic model, particularly in the field of health care
(Kendall and Knapp, 1996). The United States is thus a "purer" form of
the liberal model whereas the United Kingdom is more a mixture of the
liberal and the social democratic.
The social democratic pattern is more fully represented in our data by
the case of Sweden and, to a lesser extent, Italy. Both countries are characterized
by relatively high levels of government social welfare spending
and relatively small nonprofit sectors, at least as measured in terms of employment.
In both, moreover, the social conditions that the social origins
theory predict would lead to this pattern are in evidence. This is particularly
16
For a fuller explication of the analytic advantages of this "social origins" approach, see
Salamon (1970); on the foundations of comparative historical analysis, see the discussion
in Ragin (1987), particularly chapters 1 and 2.
Social OriginsTheory 241
true in the case of Sweden, where working-class political parties gained
extensive power earlier in this century and managed to push for extensive
state-guaranteed social welfare benefits in a context characterized by a
weakened, state-dominated Church and a limited monarchy. In Italy, the
same social outcome was produced through a slightly different route. With
Church-dominated social welfare institutions placed firmly under state control
beginning in the mid-19th century as part of the effort to achieve national
unification, the Fascist regime was able to move in the 1920s to
establish a state-centered system of social welfare protections that was then
extended by the democratic governments of the postwar era. The upshot
was a strong tradition of state-provided welfare assistance with little room
for an independent, nonprofit sector.
A small nonprofit sector in terms of employment does not necessarily
mean a small nonprofit sector overall in such regimes, however. Rather, the
nonprofit sector may simply perform a different function in social democratic
regimes—a predominantly advocacy and expressive role. This is certainly the
case in Sweden where a very substantial network of volunteer-based advocacy,
recreational and hobby organizations turns out to exist alongside a highly
developed welfare state. This may help explain our earlier finding that
whereas Sweden ranks at the low end of European countries when only paid
employment in the nonprofit sector is considered, it ranks at the high end
when account is taken as well of the involvement of volunteers.
Beyond these two widely accepted models, however, our data also validate
the existence of the two others identified in the social origins theory.
The first of these is the corporatist model represented here by Germany and
France. In both of these the state has either been forced or induced to make
common cause with nonprofit organizations, albeit for different historical reasons.
In Germany, the state, backed by powerful landed elements and in cooperation
with a relativelyweak urban middle class, responded to the threat
of worker radicalism by forging an agreement with the major churches beginning
in the latter 19th century to create a state-dominated social welfare
system that nevertheless maintained a sizable religious, and hence nonprofit,
presence. This agreement was ultimately embodied in the concept of "subsidiarity"
as the guiding principle of social policy.17
The upshot has been a
close working relationship between the state and voluntary organizations—
both secular and religious—and the resulting coexistence of extensive government
social welfare spending and a sizable nonprofit sector.
17
This concept essentially holds that principal responsibility for dealing with any social problem
lies first with the social unit closest to the problem, and that any state involvement should
operate with and through such local institutionsof neighborhood, church, and social group
(Anheier and Seibel, 1998).
242 Salamon and Anheier
In France, a rather different sequence led to a similar result. Following
the French Revolution, France effectively broke the back of Church-created
voluntary institutions and set a course toward state-sponsored social welfare.
However, political compromises made in the face of popular opposition
helped preserve the prominent role of Catholic schools in primary and
secondary education. At the same time, a strong sentiment of solidarity
sustained a network of friendly societies and associations through which
cultural and recreational interests were pursued, as in the social democratic
model described above. In the early 1980s, however, a Socialist government,
confronting severe resistance to the further extension of the classical social
democratic welfare state, found itself obliged to reach out to the associational
world for help, extending financial assistance to associations in the
process. The result has been the forging of a partnership between the state
and the voluntarysector that bears marked resemblance to that in the more
traditional corporatist regimes (Archambault, 1996).
Finally, the case of Japan fits the statist model, with low levels of government
social welfare spending accompanied by a relatively small nonprofit
sector. This reflects a tradition of state dominance established during the
Meiji Restoration of 1868 that, in the absence of effective urban middle-class
or working-class movements, has allowed the state apparatus to retain considerable
autonomous power. Combined with extensive corporate welfare,
the result has been a relatively low level of government social welfare protection
without a corresponding growth of independent nonprofit activity.
Not only does the social origins theory help explain the variations in
nonprofit size and the apparent anomalies in the relationship between the
growth of government and the growth of the nonprofit sector that are apTable
VII. Test of Social Origins Theory Predictions About
Nonprofit Revenue
Regime
Type
Liberal
Social
Democratic
Corporatist
Statist
Above-Average Source of Nonprofit Revenue
Predicted
Private philanthropy
Privatephilanthropy
Government
Fees
Actual (Ratio)a
Private philanthropy
(189%)
Fees
(122%)
Government (147%)
Fees (124%)
a
Ratio of share of nonprofit income from given source in specified
countries to average share in all countries examined.
Social OriginsTheory 243
parent in our data, but also this theory helps us account for the patterns of
nonprofit finance we have uncovered. Thus, of the four revenue-related predictions
embodied in the social origins theory as summarized in Hypothesis
6B, three find support in our data (see Table VII). In particular, as predicted,
private philanthropy is a considerably larger component of the funding
base of the nonprofit sector in the liberal regimes than for all the
countries studied, government is a considerably larger source in the corporatist
regimes, and fees are considerably larger in the statist regime. The
one deviation is the case of the social democratic regimes, where we predicted
that private philanthropy would be especially prominent but where
fee income is especially prominent instead. Interestingly, private giving does
turn out to be more prominent in these regimes than in any regimes other
than the liberal ones, thus confirming the theory even here, at least in part.
In addition to explaining patterns of nonprofit finance, the social origins
theory can help unravel apparent anomalies in the composition of the
nonprofit sector among countries. One of the curious features of the German
nonprofit sector is the substantial presence of nonprofit providers in
the fields of health and social services and their relative absence in the
field of primary and secondary education. In fact, education and research
account for only 12% of nonprofit operating expenditures in Germany versus
23% for the eight countries as a whole. How can we explain this curious
disparity?
The answer suggested by the social origins theory lies in the different
political contexts and constellations of power among stakeholders at the
time when the division of labor between the state and the nonprofit sector
was decided in these different fields. Education, for example, became a
central issue in the late 19th century when Bismarck, under the banner of
national unity and Prussian hegemony, fought the strong influence of the
Catholic Church over much of the rest of the country. The Church's control
over "its" schools was a major battleground in this struggle, which is known
among historians as the "culture struggle." As it turned out, the coalition
composed of the Catholic hierarchy (including the Vatican), agrarian elites
from southern Germany, and the representatives of the smaller kingdoms
and duchies proved unable to stem the combined, though fragile and diverse,
political force represented by Prussia, the emerging industrial class
linked to the state administration, and the landholders in the eastern parts
of the country eager to gain access to Western markets. Although the
Catholic Church lost many of its schools, a "compromise" was reached that
foreshadowed the fuller corporatist developments from the 1920s onward:
The Church was granted the right to teach catechism to Catholic students
in public schools and to maintain a presence in the formerly Catholic
schools, typically in matters of staff, curricula, and teacher training. This
244 Salamon and Anheier
compromise also put in place a financial relationship under which the state
compensates the Churches for services provided in public schools—a precursor
to today's church tax. In addition, the Church was permitted to continue
its more traditional social welfare functions.
Fifty years later, following World War II, a dramatically different situation
prevailed. By this time, Prussian-style state administration had been
discredited and had little support among the Allied powers and the emerging
political parties. Reflecting this, the Catholic Church and its allied
Christian Democratic Party managed to build into the design of a new postwar
health and social service regime for the new Federal Republic a system
based on the principle of "subsidiarity," which gives nonprofit providers a
privileged position vis-a-vis the state and obliges the state to provide them
financial support. The result is a dominating nonprofit presence in the
growing health and social service fields.
CONCLUSION
The analysis reported here thus casts considerable doubt on some of
the single-factor explanations that have dominated discussion of the nonprofit
sector in recent years. At the very least, it seems that such factors
apply only in certain circumstances and, even then, only to some components
of this complex set of institutions.
At the same time, it should be clear that the tests of these theories
offered here are preliminary at best. The number of observations is too
limited and the operationalization of some of the key variables too crude
to treat this as a final assessment of any of the theories we have explored.
Indeed, given the limitations of the data and the difficulty of converting
some of these theories into operational form, it is surprising how much
credence some of the theories gained from the empirical evidence we were
able to muster. This was particularly true of the market failure/government
failure theory and, to a lesser extent, of the interdependence theory. In
the process, the analysis reported here seems to validate the fruitfulness
of examining patterns of nonprofit development and finance cross-nationally
and thus for treating the nonprofit sector, and its major component
parts, as meaningful analytical categories.
Finally, and perhaps most importantly, the findings here suggest a fruitful
new line of analysisfor understanding the nonprofit sector at the global level,
a line that we have termed the "social origins" approach. This approach treats
the nonprofit sector not as an isolated phenomenon floating freely in social
space but as an integral part of a social system whose role and scale are a
by-product of a complex set of historical forces. What is more, it suggests
Social Origins Theory 245
that these forces are far from random or sui generis. Rather, distinctive patterns
are evident that can be analyzed and compared cross-nationally. Certain
circumstances are therefore more congenial to the blossoming of nonprofit
institutions than others, and the shape and character of the resulting nonprofit
sector is affected by the particular constellation of social forces that
gives rise to it. The social origins approach thus serves as a bridge between
the elegant simplicity of the economic models, which unfortunately turn out
to leave much of the important variation unexplained, and the dense detail
of traditional historical accounts, which make it extremely difficult to generalize
from place to place. In the process, it usefully integrates the study of
the nonprofit sector into the social analysis of societies more generally.
Whether this social origins approach will prove more effective than
the alternative theories for understanding the growth and development of
the nonprofit sector is too early to tell at this point. Our hope, however,
is that the exploratory attempt we have made to test available theories of
the nonprofit sector against solid empirical data will serve as both a stimulus
and a useful first step toward the more complete theory-testing that is
needed. Only in this way, we believe, will we come closer to understanding
what the true determinants of nonprofit growth and development really
are. Given the importance increasingly attached to this sector among both
scholars and practitioners alike, this would be a desirable outcome indeed.
ACKNOWLEDGMENTS
The authors wish to express their appreciation to Wojciech Sokolowski
for his invaluable assistance in data analysis and development of indicators
for use in this article. They also wish to thank the funders of the first phase
of the Johns Hopkins Comparative Nonprofit Sector Project and the Local
Associates whose collaboration made it possible to generate the data analyzed
here (Takayoshi Amenomori, Japan; Edith Archambault, France;
Gian Paolo Barbetta, Italy; Jeremy Kendall and Martin Knapp, United
Kingdom; Eva Kuti, Hungary; Tommy Lundstrom, Sweden; Pippo Ranci,
Italy; Wolfgang Seibel, Germany; Sven-Erik Sjostrand and Filip Wijkstrom,
Sweden; Tadashi Yamamoto, Japan). Responsibility for the analysis and
conclusions presented here, however, is ours alone.
REFERENCES
Anheier, H. K., and Seibel, W. (1998). The Nonprofit Sector in Germany, Manchester
University Press, Manchester.
246 Salamon and Anheier
Archambault, E. (1996). The Nonprofit Sector in France, Manchester University Press,
Manchester.
Ben-Ner, A., and Gui, B. (1993). Introduction. In Ben-Ner, A., and Gui, B. (eds.), The
Nonprofit Sector in the Mixed Economy, The University of Michigan Press, Ann Arbor,
Michigan.
Esping-Andersen, G. (1990). The Three Worlds of Welfare Capitalism, Princeton University
Press, Princeton, New Jersey.
Flora, P., and Alber, J. (1981). Modernization, democratization and the development of
welfare states in Western Europe. In Flora, P., and Heidenheimer, A. (eds.) The
Development of Welfare States in Europe and America, Transaction, New Brunswick, New
Jersey.
Flora, P., and Heidenheimer, A. (eds.) (1981). The Development of Welfare States in Europe
and America, Transaction, New Brunswick, New Jersey.
Fukuyama, F. (1995). Trust: The Social Virtues and the Creation of Prosperity, The Free Press,
New York.
Hansmann, H. (1980). The role of nonprofit enterprise. Yale Law Journal, 89, 839-901.
Hansmann, H. (1987). Economic theories of nonprofit organizations. In W. W. Powell (ed.),
The Nonprofit Sector: A Research Handbook, Yale University Press, New Haven,
Connecticut.
Hicks, A., and Swank, D. H. (1992). Politics, institutions, and welfare spending in industrialized
democracies. American Political Science Review, 86(3), 658-674.
Huber, E., Ragin, C, and Stephens, J. D. (1993). Social democracy, Christian democracy,
constitutional structures, and the welfare state. American Journal of Sociology, 99(3),
711-749.
James, E. (1987). The nonprofit sector in comparative perspective. In W. W. Powell (ed.),
The Nonprofit Sector: A Research Handbook, Yale University Press, New Haven,
Connecticut.
Kendall, J, and Knapp, M. (1996). The Voluntary Sector in the UK, Manchester University
Press, Manchester.
Krashinsky, M. (1986). Transaction costs and a theory of the nonprofit organization. In S.
Rose-Ackerman (ed.), The Economics of Nonprofit Institutions, Oxford University Press,
Oxford and New York.
Moore, B., Jr. (1966). Social Origins of Dictatorship and Democracy: Lord and Peasant in the
Making of the Modern World, Beacon Press, Boston, Massachusetts.
Offe, C. (1984). Contradictions of the Welfare State, MIT Press, Cambridge, Massachusetts.
Putnam, R. (1993). Making Democracy Work: Civic Traditions in Modem Italy, Princeton
University Press, Princeton, New Jersey.
Quadagno, J. (1987). Theories of the welfare state. Annual Review of Sociology 13, 109-128.
Ragin, C. C. (1987). The Comparative Method: Moving Beyond Qualitative and Quantitative
Strategies, University of California Press, Berkeley.
Rueschemeyer, D., Stephens, E., and Stephens, J. D. (1992). Capitalist Development and
Democracy, University of Chicago Press, Chicago, Illinois.
Salamon, L. M. (1970). Comparative history and the theory of modernization. World Politics,
XXIII(\), 83-103.
Salamon, L. M. (1987a). Partners in public service: The scope and theory of government-nonprofit
relations. In W. W. Powell (ed.), The Nonprofit Sector: A Research
Handbook, Yale University Press, New Haven, Connecticut.
Salamon, L. M. (1987b). Of market failure, voluntary failure, and third-party government:
Toward a theory of government-nonprofit relations in the modern welfare state. Journal
of Voluntary Action Research, 16(1-2), 29-49.
Salamon, L. M. (1992). America's Nonprofit Sector: A Primer, Foundation Center, New York.
Salamon, L. M. (1995). Partners in Public Service: Government-Nonprofit Relations in the
Modem Welfare State, The Johns Hopkins University Press, Baltimore, Maryland.
Salamon, L. M., and Anheier, H. K. (1992a). In search of the nonprofit sector. I: The question
of definitions. Voluntas, 3(2), 125-151.
Social Origins Theory 247
Salamon, L. M., and Anheier, H. K. (1992b). In search of the nonprofit sector. II: The problem
of classification. Voluntas, 3(3), 267-309.
Salamon, L. M., and Anheier, H. K. (1996). The Emerging Nonprofit Sector, Manchester
University Press, Manchester.
Salamon, L. M., and Anheier, H. K. (eds.) (1997). Defining the Nonprofit Sector: A
Cross-National Analysis, Manchester University Press, Manchester.
Salamon, L. M., Anheier, H. K., Sokolowski, W., and Associates (1996). The emerging sector:
A statistical supplement. Papers of the Johns Hopkins Comparative Nonprofit Sector Project,
The Johns Hopkins Institute for Policy Studies, Baltimore, Maryland.
Seibel, W. (1990). Government/third sector relationships in a comparative perspective: The
cases of France and West Germany. Voluntas, 1(1), 42-61.
Titmuss, R. (1974). Social Policy: An Introduction, Pantheon, New York.
United Nations (1993). The 1993 System of National Accounts, United Nations, New York.
U.S. Department of Commerce, Bureau of the Census (1994). Statistical Abstract of the United
States: 1994, U.S. Government Printing Office .Washington, D.C.
Weisbrod, B. (1977). The Voluntary Nonprofit Sector, Lexington Books, Lexington,
Massachusetts.
World Values Study Group (1994). World Values Survey, 1981-1984and 1990-1993 [Computer
File], ICPSR Version, Institute for Social Research, Ann Arbor.
248 Salamon and Anheier
Copyright of Voluntas: International Journal of Voluntary & Nonprofit Organizations is the property of
Springer Science & Business Media B.V. and its content may not be copied or emailed to multiple sites or
posted to a listserv without the copyright holder's express written permission. However, users may print,
download, or email articles for individual use.