Reclaiming the “Third Sector” from “Civil Society”: A New Agenda for Development Studies

Civil society is one of the most widely used—and widely maligned—concepts in development studies. In this paper, we argue that much confusion regarding civil society stems from the omnibus nature of its conceptualization. We consider civil society to be an omnibus concept because it has been imbued with several distinct meanings—a normative meaning (civil society as civilized), a functional meaning (civil society as democratizing), and a structural meaning (civil society as a third sector). Using the example of humanitarian NGOs, we demonstrate how the omnibus nature of civil society resists systematization and requires scholars to make problematic assumptions when designing empirical research. As a solution, we propose replacing “civil society” in empirical research with the structural “third-sector” concept. This move narrows the gap between the actors that scholars study and the theoretical construct that they are supposed to represent; it brings the third sector into conceptual alignment with our understanding of the first and second sectors (themarket and the state); and it improves our efforts to compare findings across cases and build generalized theories. It also enables scholars to consider questions of power, resources, and influence when studying development NGOs—questions that are difficult to ask when notions of “civil society” are defined as actors that understand, represent, and advocate on behalf of their “constituents.” We conclude that “civil society” as a concept should be maintained for theoretical analyses of what makes society civil but that empirical studies of development are best served by a third-sector approach.

On the other hand, scholars have prolifically critiqued civil society actors, typically for shortcomings in these same areas: failing to mobilize communities (Biekart ; Silliman ; Kamat , ; Henderson ), prioritizing funders' needs over the needs of their "constituencies" (Hulme and Edwards ; Fox and Brown ), promoting "uncivil" social values or even civil unrest (Pearce ; Caple James ), and, more generally, failing to make a measurable dent in the poverty and inequality endemic to developing nations (Easterly ).These critics typically share advocates' understanding of an ideal civil society, but they argue that real-world civil society actors often fail to achieve the positive outcomes that the theory predicts will emerge from a "healthy" or "vibrant" civil society.
Scholarly attempts to define civil society have done little to mitigate the confusion.To give just a few examples, civil society has been conceptualized as the organized expression of a society's values (Castells ), a sphere of freedom in which individuals come together for debate and advocacy (Cohen and Arato ), a zone where individuals associate and groups shape norms and articulate purposes (Post and Rosenblum ), a system of networks linking individuals in their pursuit of influence over community and political affairs (Skocpol and Fiorina ), a network of associations that institutionalizes problem-solving discourses (Habermas ), a force for democratic change (Jordan ), or a sphere of ideas, values, institutions, networks, and individuals that are based upon civility (Anheier ).Although distinct, these definitions often tend to combine ideas about actors, functions, and positive normative values into a single conceptual package.
Efforts to operationalize civil society only compound the confusion.Since an "organized expression of values" or a "sphere of freedom" is difficult to measure, scholars have regularly approximated "civil society" by analyzing the actors who purportedly express those values or inhabit those spheres-so-called civil society organizations.The actors chosen as proxies for "civil society" have been strikingly diverse, ranging from bowling leagues (Putnam ), to labor unions (Fitzsimmons and Anner ), to revolutionary insurgencies (C.Kumar ), to national-level counts of transnational nongovernmental organizations (NGOs) (Smith and Wiest ).Few scholars consider the potential distance between the organizations studied and the theoretical concept they are supposed to represent.
In short, civil society is central to development scholarship, but there remains remarkably little consensus about what it is, and little guidance about how to effectively and systematically analyze its relationship to development outcomes.In this paper, we aim to reclaim the clearest and most useful elements of the concept of civil society for studies of development.At the heart of our argument is a critique of the present-day status of "civil society" as an "omnibus concept."By omnibus concept, we mean that contemporary scholarship has imbued the single concept of civil society with several distinct meanings-specifically a normative meaning (civil society as civilized), a functional meaning (civil society as democratizing), and a structural meaning (civil society as a third sector).
The omnibus nature of "civil society" is problematic for several reasons.Because it collapses what ought to be several distinct variables (i.e., structural actors, the "space" they occupy, normative values, functional expectations) into a single concept, it often obscures variation that is important to study.Collapsing multiple meanings into a single concept also opens scholars to critiques of tautology; for example, the proposed democratization function of civil society may predetermine both which organizations or associations are selected for analysis (e.g., those engaged in democratizing activities) and which organizational outcomes are analyzed (e.g., the promotion of democracy).Meanwhile, organizations that are assumed to be nondemocratizing-like development organizations with a religious rather than a secular mission-are frequently left unanalyzed (Bush ).The omnibus nature of the civil society concept also predisposes scholars to evaluate civil society actors on a relatively narrow set of potential outcomes (e.g., their effects on democracy or civility), leaving largely unexplored these same actors' effects on other outcomes that are also relevant to development scholarship (e.g., their effects on local-level inequality, mobilization, competition, migration, or violence, inter alia).And at a fundamental level, the omnibus nature of "civil society" limits its clarity and specificity, making it difficult to apply the concept uniformly across diverse cases, and limiting our ability to build collective insights as a scholarly community.
Our efforts to reclaim the utility of "civil society" for development scholarship proceed across four sections.In the first section, we trace the long historical development of the civil society concept from the ancient Greeks to present-day scholarship to determine the roots of its omnibus nature.We find that, like heirlooms in an attic, the various meanings of civil society have accumulated over time, with particular meanings being added during specific historical periods.We further identify the s and s as a critical moment in this intellectual history, a moment when three distinct threads of meaning-a normative meaning, a functional meaning, and a structural meaning-were jointly revived in the specific empirical context of democratization movements in Latin America and Eastern Europe.Although the resulting omnibus concept reasonably captured the empirical reality of that particular political moment, we argue that its utility has declined as scholars have turned their attention to other civil society actors in later historical periods.Contexts changed, and civil society actors themselves transformed, tending toward more professionalized, less participatory organizational forms (Skocpol , ; Walker ).Thus we document how civil society became an omnibus concept, while simultaneously illustrating why that omnibus concept cannot adequately account for the contemporary reality of civil society actors.
In the second section, we elaborate our critique of the civil society concept, delineating three pitfalls that scholars tend to face because of its omnibus nature.We conclude that a structural, or "third-sector," approach avoids these pitfalls and provides the most useful analytical tool for empirical studies of development.We define the third sector as a sector of organized human action composed of collective actors beyond the family and distinct from the state and the market.This concept captures all of the actors conventionally referred to as civil society, in addition to the many nonstate, nonmarket actors that are often excluded from civil society analyses.Importantly, our third-sector concept strips these actors of any assumed normative value or functional outcome.As a result, the third-sector concept substantially narrows the gap between the actors that scholars study and the theoretical construct that these actors are supposed to represent.It opens up to analysis variations in actors' functional effects and normative orientations, rather than incorporating functions and norms into the sector's definition.It also opens to analysis new questions about the many additional ways in which third-sector actors shape development outcomes, moving beyond civil society scholarship's traditionally narrow focus on democracy and civility.And because a structural conceptualization of these actors is cleaner to define and operationalize, it becomes possible to apply the concept uniformly across diverse cases, thus allowing studies of third-sector organizations to build upon one another and to expand our collective knowledge about the relationship between these actors and development outcomes.
In the third section we illustrate the utility of our proposed conceptual reorientation by applying it to the case of humanitarian NGOs.We choose to focus on humanitarian NGOs because they represent a type of third-sector actor that has become central to development scholarship over the past several decades.Through a series of short vignettes, we demonstrate how using the existing omnibus definition of civil society predisposes us to miss complex, counterintuitive, or novel dynamics introduced by humanitarian NGOs, especially in relation to questions of power and resources in development.We then illustrate how our proposed reorientation to the third sector would allow scholars to better investigate the complex and multifaceted ways in which humanitarian NGOs are shaping development outcomes.
In the fourth and final section we summarize our main arguments and discuss their implications for development studies more broadly.Our central conclusion is that scholars interested in how contemporary actors outside the state and the market affect development processes should replace analyses of "civil society" with analyses of the "third sector." 2 By freeing scholarship from the normative and functional baggage that, after centuries of development, will not be easily divorced from the concept of civil society, this revised terminology makes it easier for scholars to study how actors, actions, and outcomes may change over time and across the sector.It also brings into the conversation many relevant players (like religious associations) that are often excluded from studies limited to functionally or normatively defined "civil society organizations."Most centrally, third-sector terminology motivates analyses of how development or advocacy organizations, which are often powerful, resource-rich players operating in resource-poor environments, can dramatically reshape the political and social dynamics of those environments, sometimes in unexpected ways.Such arguments about power, resources, and influence are difficult to make when organizations are by definition considered "civil" entities that understand, represent, and advocate on behalf of their "constituents." Ultimately, our quest to regain the utility of a civil society concept has led us to advocate for the concept's retirement from empirical analyses of development.To be clear, we do not advocate a similar retirement of "civil society" from social theory.The concept has a long, rich theoretical history, especially when it has been used to understand political transformations or to articulate normative ideals of development.However, given the numerous problems associated with using the concept in empirical studies, we argue that "civil society" is best reserved for political theorists who study whether and how society is "civil." Contemporary scholars of civil society have inherited a body of scholarship rich with ideas and observations about associational life.However, as many scholars have pointed out, the concept of civil society continues to defy clear definition or operationalization despite its rich history (Van Rooy ).In this section, we describe how the civil society concept has evolved over time to reach its contemporary incarnation, which we argue simultaneously incorporates the following three dimensions: A normative dimension: Civil society is defined by its civilized character.Different authors have different ideas about what constitutes civility, but the common thread is that civil society (and the associations it encompasses) carries a positive normative valence and promotes a more enlightened society: for example, a society characterized by virtues like solidarity, equity, or justice.
A functional dimension: Civil society is defined by its effects, and specifically its effects on democracy.It mobilizes citizens, advocates on their behalf, generates social trust, and safeguards against government corruption or abuse.Civil society (and the associations it encompasses) in this sense helps generate democracy, consolidate democracy, and then safeguard democracy once it is in place.
A structural dimension: Synonymous with the third sector, the structural idea of civil society denotes a sector of organized human action composed of collective actors beyond the family and distinct from the state and the market.

Building the Omnibus Concept: From the Ancient Greeks to Modernization Theory
The normative definition of civil society has its roots in ancient Greece and Rome, where civility, or the normative ideas surrounding what it meant to be a civilized society, was central to the term's use.The Greeks used the phrase politike koinonia (political society/ community), which was translated into Latin as societas civilis (K.Kumar ).This original idea of civil society encompassed the definition of the polis, or the political order of a city-state.Aristotle argued that the polis was the most virtuous form of association because, in its right form, it allowed men to fulfill their telos as political animals.Thus for the ancient Greeks a civil society was a public ethical-political community of free and equal citizens under the rule of law (Cohen and Arato ).Under this Aristotelian idea of civil society, scholars did not distinguish between the state and society but instead theorized a cohesive sociopolitical and moral order that spanned all sectors of social and political life (Miller ).A civil society was a civilized society, and for the ancients this required the right kind of state. 3 Seventeenth-and eighteenth-century European thinkers continued to understand a civilized society as distinct from an uncivilized one, though each philosopher had a different idea about what that should encompass.Hobbes argued that civil society was created when the social contract gave birth to a Leviathan-run state and people exited the state of nature, thus permitting civility.Locke proposed that civil society emerged when citizens were given the right to liberty and property.Ferguson suggested that civil society became possible only when men actively participated in politics (Varty ; Anheier, Glasius, and Kaldor ).
Importantly, the seventeenth and eighteenth centuries were also critical for the expansion of the civil society concept to include the first seeds of a structural definition.Locke's and Montesquieu's writings introduced language that referred to the civil and the political-a precursor for the separation between government and society (K.Kumar ; Cohen and Arato ; Palmer ).This structural understanding of civil society developed further in the writings of Hegel and Marx.Writing in the early nineteenth century, Hegel defined civil society as an intermediate power between the individual and state, and is credited with building a theory that clearly distinguished between civil society and the state (Gordon ; Lively and Reeve ; Baynes ; Cohen and Arato ).Marx also understood civil society as separate from the state, but unlike Hegel he equated civil society with the economic realm of bourgeois society in which individuals pursued their class interests (K.Kumar ; Cohen and Arato ; Palmer ; Anheier et al. ).
Although relatively marginal in the context of his overall oeuvre, Antonio Gramsci's ideas about civil society are considered to be the predecessors of our modern conceptualizations of this sphere as a third sector alongside the state and the market.Like Hegel and Marx, Gramsci saw civil society as separate from the state, but he also distinguished civil society from economic activities.He understood civil society as being a temporary third sphere, between the market and the state, from which a communist revolution could be staged (Anheier et al. ).
Talcott Parsons further solidified the understanding of civil society as a structural "third sector."His work drew a distinction between the state, the economy, and the realm of associations, though he used the term societal community rather than civil society.In his writings, Parsons (:) suggested that "societal community" not only was separate from the state but also had influence over it: "The societal community was to be differentiated from government as its superior, legitimately entitled to control it."Between them, Gramsci and Parsons established the template for a structural third-sector model of civil society.
The development of the functional definition of civil society ran parallel to the development of the structural definition.Montesquieu took up Locke's idea of civil society as a separate sphere, but where Locke theorized civil society as an associational sphere separate from the state and apolitical in nature, Montesquieu conceptualized a civil society sphere that performed a key political function: it preserved the rule of law and reined in monarchies to prevent them from turning despotic (Taylor :).Thus Montesquieu is credited as the first political thinker to suggest that civil society, as an actor outside the state, had an important role to play vis-à-vis the state-namely to protect society from tyrannical monarchs.
Tocqueville's early nineteenth-century study of the United States further strengthened the theoretical link between associational life and democracy (Whitehead ).Tocqueville admired what he saw as Americans' unusual proclivity for participation in associational life and theorized that these activities helped protect US democracy against tyranny, especially given that a democracy, by definition, lacked a nobility to keep the state in check.Although Tocqueville never used the phrase civil society, his ideas about associations are foundational to scholars who employ the functional sense of the concept in contemporary research.
As these various meanings arose, many thinkers began to blend them, endorsing early versions of the omnibus concept.Locke and Montesquieu wove an emergent structural concept together with a normative conceptualization.Tocqueville took up all three conceptual threads, pointing to the practice of associating-participating in nonstate "civil associations" beyond the family-as a key civilizing practice that supported democracy, equality, and freedom (Tocqueville :-).
By the twentieth century civil society theorists were regularly combining structural, normative, and functional elements into the single concept of civil society.Parsons pioneered the structural third-sector concept because he understood civil society as associational life outside the state or the market, but his approach also included clear normative and functional elements.His theory of action conceptualized associational life as inherently solidary and normatively consensual (Howell and Pearce ).Because associations were created through consensus, he argued, the influence they exerted on others was based on mutual agreement more than control of resources (Warren ).As a result, Parsons expected civil society to pluralize power, safeguard a normatively ideal society, and provide an important check on state powers (Cohen and Arato ; Parsons ).Modernization theory also promoted an omnibus definition of civil society.Lipset (:-), for example, argued that "intermediary organizations and institutions" could act as sources of "countervailing power," especially by curbing states' powers, generating new ideas, and serving as a training ground in the "skills of politics" for the citizenry.And Habermas (), in his critique of modernization theory, nevertheless shared with his predecessors an understanding of civil society as a system of discussion-focused associational forms, like coffeehouses, that nurtured peer-based dialogue and debate and encouraged citizens to confront and critique the state, becoming a powerful antihegemonic force in society.

Empirical Applications of the Omnibus Concept: Overthrowing Dictatorships and Supporting Democracy
The s and s were a turning point for civil society scholarship.Before that time, civil society was discussed in largely theoretical terms, with few empirical studies to support claims of its normative and functional outcomes.Nevertheless, when civil society actorsincluding labor unions, peasant associations, religious leaders, student movements, rebel insurgencies, and human rights organizations, inter alia-forced the overthrow of a series of Latin American dictatorships, the power of civil society as a normative and democratizing force became a proven reality and generated renewed scholarly interest (Diamond ; Anheier et al. ).Around the same time, the term civil society came into use among Eastern European intellectuals engaged in struggles against the totalitarian control of the Soviet Union.In these contexts, both Latin American and Eastern European intellectuals saw civil society not only as a tool with which to fight back against an authoritarian state but as an end in itself-a sphere that should be protected from state intervention (Skapska ; Post and Rosenblum ; Anheier et al. ).Hence, the structural understanding of civil society-associations outside the state and the market-was associated with real-life social actors who also fit the functional understanding of civil society by opposing the repressive regime and thus were imbued with a normatively positive valence in the eyes of prodemocracy intellectuals.
The revival of the civil society concept by scholars of Latin America and Eastern Europe was followed by a similar resurgence in academic literature on Western societies.Robert Putnam's () Making Democracy Work was one of the first to operationalize the concept in a causal argument.Putnam argues that northern Italy's strong civic community explained its superior governmental performance in relation to southern Italy, which was characterized by weak levels of civic community and social capital.Although Putnam never uses the term civil society, his notions of "civic community" and "social capital" are important and widely acknowledged precursors to present-day civil society scholarship.
The consensus in the literature quickly became, in the words of Ernest Gellner: "No civil society, no democracy" (, quoted in Fukuyama :).Though scholars emphasize different features of the concept in their arguments, in general they agree that civil society promotes democracy because it mobilizes society, holds states accountable, fosters pluralistic and tolerant values, and represents community interests (Muller and Seligson ; Clark ; Krishna ; Finkel ; Hoogh and Stolle ; May and Milton ).Building on earlier theorists, researchers initially credited such societal-level democratic outcomes to civil society because of its associational characteristics: the very process of associating was expected to encourage dialogue and debate among individuals, facilitate the spread of knowledge about political events, and allow individuals to first determine, then powerfully express, their collective interests.Civil society organizations were also considered a training ground for creating democratic citizens: through associating, individuals could learn parliamentary procedure, gain experience with group-level elections and other forms of representational governance, and reinforce notions of good (participatory) citizenship (Putnam ; Skocpol ).Moreover, although this scholarship largely focused on civil society's connection to democracy, it also introduced the possibility that civil society could affect other outcomes.For example, Putnam (:) concludes that strong civic community fosters economic prosperity as well as democracy.In his words: "Economics does not predict civics, but civics does predict economics."

Organizational Transformations and Conceptual Stagnations
Shortly after this resurgence of civil society scholarship, scholars began to note a dramatic shift in the associational aspects of civil society organizations.In the United States, for example, the early civic associations celebrated by scholars like Tocqueville () and Skocpol (), which were organized largely on the principle of face-to-face interactions among their members, became much less prominent between the late s and the early s (Skocpol ).They were replaced with a new crop of professionally managed organizations that either had few members or limited their members' participation to signing petitions or writing checks (Skocpol , ).By the s, this trend appears to have spread around the globe (Edwards ).Although still regularly termed "voluntary" organizations or "associations" by scholars, civil society organizations increasingly prioritize professional staffing, fund-raising, litigating, lobbying, and releasing public statements, eclipsing the more traditional emphasis on bringing together individuals for dialogue, debate, or participation (see also Walker ).The reasons for this organizational transformation of civil society actors are multiple, 4 and they probably reflect a growing consensus that policy goals are often more efficiently achieved by professional organizations with financial resources than through more traditional forms of associating (Soule and Olzac ).
Despite this widespread transformation in organizational forms, many contemporary scholars continue to define even low-membership "civil society organizations" as generators of government transparency, accountability, a society-wide consensus on civility, and the construction of democratic and progressive values.This line of research has been empirically supported by new scholarship examining "global civil society," which demonstrates that increasing numbers of ties between states and transnational NGOs are producing significant national-level changes in development outcomes like educational attainment, human rights, environmental protection, and scientific advances (Hafner-Burton and Tsutsui ; Schofer ; Schofer and Hironaka ; Schofer and Meyer ).Not surprisingly, scholars increasingly emphasize the democratizing effects of civil society organizations' advocacy work on behalf of their "constituents," whereas in the past they might instead have focused their analyses on associations' efforts to mobilize their "members." In sum, the concept of civil society experienced a revival in the s and s, when all three meanings were resurrected from various strands of the concept's theoretical heritage.In addition to acting as a force for democracy (reinforcing the functional understanding of civil society), civil society in these contexts was seen as importantly separate from the state (reinforcing a third-sector conceptualization).And because civil society actors were the heroes of these movements' efforts to overthrow authoritarian states, the concept simultaneously took on a positive normative valence; civil society organizations protected human rights, promoted solidarity, and represented the true interests of the people, not the powerful.The omnibus usage has endured in civil society scholarship since that moment, even as the historical context has changed and even as the civil society actors themselves have transformed.
Despite this rich intellectual history, scholars of civil society have not yet reached consensusor anything approaching consensus-on a systematized concept of civil society (Edwards ; Pearce ; Van Rooy ).Adcock and Collier () argue that social scientists must systematize concepts before using them to analyze the social and political world.A systematized concept-"a specific formulation of a concept used by a given scholar or group of scholars"-is distinguished from a background concept-the "broad constellation of meanings and understandings associated with a given concept"-by a process of deliberately defining the boundaries around a concept to clarify how empirical examples can be mapped neatly onto the conceptual framework (Adcock and Collier :).If we are to have a coherent conversation as social scientists and avoid talking past each other, we need to be clear that the language we use maps consistently onto the same set of ideas and that both language and ideas map consistently onto empirical reality.
Civil society has resisted systematization, we argue, because scholars theorize civil society as a space or network but have no option other than to operationalize that space or network by studying the organizations that occupy (or constitute) it.In other words, the normative and functional elements embedded in the concept of civil society can be empirically demonstrated only through a study of the actors within it.Yet analyzing a universe of actors that are theoretically defined by their composite functional or normative characteristics inevitably requires scholars to make problematic assumptions.
We suggest that analysts typically approach the operationalization of civil society in one of three ways, each of which requires its own set of assumptions.To begin, many scholars opt to limit their universe of so-called civil society organizations to a particular subset of actors thought to best represent the democratizing or civilizing aspects of the sector.For example, scholars might choose so-called social change organizations, human rights organizations, or advocacy organizations to serve as proxies for civil society, while excluding religious or development organizations from their analyses because, in the authors' determination, these latter organizations do not fulfill the functions or norms associated with civil society.Using actors' presumed functions to select a sample for study is of course not a problem in itself (although we often wish civil society scholars would be clearer in documenting the empirical criteria they use to choose their samples).Such processes of categorization and sampling are inherent to any analysis that examines a subcomponent of a broader group or sector.However, because civil society is defined by its collective democratic function or civilizing values, we can never know the full population of organizations contributing to that collective function or value.Indeed, an organization that meets a scholar's criteria for a civil society organization one month may not meet the criteria the next month because it may have adopted different projects or actions.Or, as we demonstrate with our vignettes below, a single organization's activities can be simultaneously democratizing and undemocratic.As a result, scholars often use conclusions about their chosen actors to inform broader claims about civil society, without providing any sense of how well these actors represent "civil society" more broadly, or how the relative size and influence of the chosen actors compare to the size and influence of countervailing organizations (e.g., nondemocratizing, non-social change organizations).
Additionally, scholars who study a functionally defined subset of civil society organizations, like human rights organizations, are primed to study only the effects for which the organization was chosen, such as its defense of human rights.Whether or how these civil society organizations influence other social processes-inequality, competition, mobilization, or migration, for example-is often left unexplored.
A second process that scholars use when operationalizing a functional or normative concept of civil society is to openly acknowledge that the full universe of civil society organizations is characterized by a diversity of norms and functions (e.g., it would include both organizations like the NAACP and organizations like the Ku Klux Klan) but to assume that the interaction between the various organizations nevertheless generates a positive normative or functional outcome for the sector as a whole.This expectation, that the process of associating will on average build democratic culture despite the diversity of organizations involved, can be traced to theorists like Parsons and Habermas.Yet a number of case studies call into question the assumed positive outcomes from associating.Berman's () study of Weimar Germany, for example, demonstrates how organized civil society led to the rise of fascism.Jamal () found that strong associational life in the West Bank reinforced clientelism and patronage.Walsh () found that, when people come together to "talk politics," they tend to reinforce the narrow political ideals they already hold, rather than reach an evolving consensus on civility.And McCarthy () found that prolife mobilization achieved legislative changes that were actually contrary to the prochoice opinions of the majority.In short, there is still little evidence to support the assumption that civil society organizations have a civilizing or democratizing effect because of their interactions (Hearn ; Fung ; Riley ).
In a third approach to operationalization, scholars openly acknowledge that the full universe of civil society organizations frequently fails to fulfill its democratizing or civilizing potential.But for these scholars the problem is not a faulty concept but rather a faulty civil society.They suggest that civil society organizations must be nurtured to generate a "healthy" or "vibrant" civil society.Because civil society is the only potential space in which regular folk can come together to exercise power over states and markets, fomenting a healthy civil society, understood as a democratizing and civilizing civil society, becomes a political project worthy of scholarly support.Stripping the civil society concept of its functional and normative elements, scholars argue, would be detrimental to that political project (Howell and Pearce ; Edwards ).In short, these supporters of the civil society concept suggest that scholars should work to match the empirical reality to the concept, rather than finding concepts that best define, articulate, and understand our empirical reality.
Given the problematic assumptions inherent in the above approaches to civil society research, we propose that scholars should retire the concept of civil society from empirical analyses and adopt instead the concept of the third sector.We define the third sector as a sector of organized human action composed of collective actors beyond the family and distinct from the state and the market.This is a purely structural definition; it has no normative or functional content.We suggest that substituting the third-sector concept for studies of civil society brings multiple analytical advantages.
First, the third sector is a concept that is readily systematized.It accurately represents the organizations being analyzed, and because it can be operationalized consistently across studies, it allows scholars to build knowledge about the sector as a whole, one study at a time.By purging analyses of the assumptions accompanying civil society organizations, a third-sector approach can lay the groundwork for a typology of all third-sector actors, comprehending the range, size, and strength of various subgroups within it.We can also begin to study similar groups over time and across societies and to better map the connections between organizations.How do third-sector organizations interact?When they do, what determines which organization has more power to meet its goals?By divorcing the actors, their actions and interactions, and their effects from any assumed function or norm, scholars can go further in understanding the nature of this sector and its multiple, changing, and likely contradictory effects on development.
Adopting a structural third-sector approach has the added benefit of consistency with the way we conceptualize the first and second sectors-that is, the state and the market.Both states and markets are defined as structural actors rather than as normatively or functionally defined actors.This approach acknowledges normative variation in structurally similar actors, allowing researchers to clearly identify actors on the basis of observable structural characteristics and then to further characterize them as having empirically derived "good" or "bad" qualities.Thus scholars write about "kleptocratic states," "benevolent monarchs," "extractive rulers," "hostile takeover acquisitions," "corrupt CEOs," and so on.An empirically defined adjective modifies a clear structural actor.Since we use structural definitions of markets and states and then modify them with empirically defined adjectives, there is a strong argument for treating the third sector in the same manner.
Finally, adopting a third-sector concept will quite simply generate new questions and better answers for scholars interested in how individuals associate outside the state and the market, and the consequences of those associations for development.At present, "civil society organizations" are overwhelmingly evaluated according to whether they accomplish what the intellectual history of the concept tells us they "should."This in itself is not a problem.For example, civil society actors have historically been defined as promoters of democracy, and there is of course utility in studying whether or how the introduction of new actors in this sector-like modern humanitarian NGOs-either inhibits or promotes democracy.However, as we demonstrate below, when scholars are primarily tuned to evaluate what has been studied before, they miss important new consequences brought about by changing actors or actions within the sector.Civil society scholars may be so focused on analyzing democratic actions or outcomes that they may overlook undemocratic actions or outcomes.Or they may be so focused on analyzing democratic outcomes that they fail to examine the many other potential, if unanticipated, effects that civil society actors could have on societies beyond democratizing ones.For example, civil society actors might have effects on local levels of inequality, ethnic divisions, youth migration, or technology usage, just to imagine a few.Most centrally, we suggest that the normative and functional baggage accompanying the notion of civil society has prevented scholars from adequately examining questions of how power and resources shape the actions, interactions, and outcomes of third-sector actors.
One effect of shifting from "civil society" to the "third sector" is that we move up Sartori's () ladder of abstraction.That is, by eliminating the normative and functional assumptions built into the omnibus civil society concept, we get a third-sector concept that describes a broader, more varied set of actors.It is less specific than the omnibus version of the civil society concept in that it encompasses more variation, so we lose some specificity.Detractors may argue that a structural third-sector concept does not give us much analytical leverage because it is very broad.Rather than looking for one subset of actors that best represents a narrowly defined civil society, we prioritize analyses of the great diversity of actors operating outside the state and the market.While this is true, we feel it is preferable to the present predicament: the civil society concept is treated in the contemporary literature as if it were fairly high up the ladder of abstraction-as high up as the state or the market-but in its omnibus form it is in fact much lower, specific enough that it fails to capture important variation in the broad empirical realities it purports to describe.Shifting to a thirdsector concept that sheds normative and functional assumptions corrects this problem-it better describes the more general idea of collective actors outside the state and the market, encompassing variation in function and normative value.To mitigate the higher level of abstraction and prevent "empirical vaporization" (Sartori ), the third-sector concept can be subdivided with modifiers and thus transformed into a medium-level category.This approach is common in work on the other two sectors.Thus, much as we have literature on authoritarian states and democratic states, perfect and imperfect markets, and so on, scholars using a "third-sector organizations (TSOs) with adjectives" approach can develop conceptual categories like "development TSOs," "grassroots TSOs," and so on (for a review of a similar approach in democratization studies, see Collier and Levitsky ). 5 In this section, we have outlined the advantages of adopting the third-sector approach over a more traditional civil society approach when conducting empirical research.In the next section we use the case of humanitarian NGOs to demonstrate how substituting a structural third-sector concept for the civil society concept can improve field analyses of nonstate, nonmarket actors.Specifically, we focus on three potential pitfalls: .Because "civil society" actors are imbued with normative values like altruism and solidarity with their constituents, scholars design research programs that too often fail to uncover instances of "uncivil" society-for example, instances of third-sector corruption or community disconnect.
.Because "civil society" actors are imbued with democratic functions, like advocating on behalf of constituents or serving as watchdogs for government performance, scholars design research programs that too often fail to uncover instances of the third sector's undemocratic actions or outcomes.
.Because "civil society" actors are imbued with both normative values and democratic functions, scholars design research programs tailored disproportionately to these two themes and too often fail to investigate the myriad other ways in which third-sector actors may affect outcomes.Significantly, they fail to investigate the effects of the power and resources that many third-sector actors bring to the typically resource-poor areas in which they operate.

T H E C A S E O F H U M A N I T A R I A N N G O S
When the notion of civil society made its resurgence in the s, its omnibus definition was relatively unproblematic because the multiple meanings it encompassed were remarkably congruent with the actors and contexts under study.By the s, however, when development scholars began to adopt the concept of civil society in earnest, the congruence between actors, activities, and functions loosened dramatically.The new face of civil society in development studies became the humanitarian NGO.Although humanitarian NGOs were empirically different from the protest-oriented actors typically analyzed in the s, they enthusiastically adopted both the title and the assumed functions of "civil society" as theorized in the academic literature.Specifically, humanitarian NGOs (and many scholars who analyzed them) claimed that their importance for development came not only from their developmental programs but also from their very existence.Practitioners often argued that because humanitarian NGOs constituted "civil society," they automatically strengthened democracy and promoted progressive social norms in developing nations (Ward ; Bano ).
In this section, we use the example of humanitarian NGOs to illustrate the problems associated with the existing omnibus definition of civil society.We begin by defending our claim that humanitarian NGOs do indeed constitute the new face of "civil society" in development studies.We then provide a brief synopsis of how, unlike the civil society actors studied in the s, humanitarian NGOs today fail to achieve congruence with the omnibus definition of civil society.Finally, we present a series of short vignettes to demonstrate how a "civil society" approach may confuse and limit scholarly analysis, whereas our proposed "third-sector" definition transforms the questions we ask, as well as the answers we may uncover, in our analyses of how the third sector matters for development.
Humanitarian NGOs: The New Face of "Civil Society" in Development Studies In the s, when the term civil society was making its resurgence in the social sciences, much of the developing world was rocked by debt crises and market crashes.These crises, in combination with Reagan-and Thatcher-era politics, heralded a new era of neoliberal policies in global development processes.The resulting "Washington Consensus" proposed that unleashing the free market by eliminating state interference was the best route to development.Eliminating state manipulation of the economy was expected to spur growth by opening developing markets to foreign investment, allowing supply and demand to develop each nation's comparative advantage, and by reducing states' opportunities for corruption by leaving business in the hands of private enterprise.Highly indebted governments were required by intergovernmental organizations like the International Monetary Fund to liberalize their economies, eliminate market protectionism, and significantly reduce their social service spending as conditions of gaining new loans or restructuring old ones (see Viterna and Robertson forthcoming for a review).While there is disagreement on whether such restructuring affected national-level economic growth, many scholars agree that the resulting cutbacks in state social service provision resulted in a significant increase in poverty and human suffering around the globe.
It was in this context that "civil society" became a central fixture in development scholarship.Scholars and practitioners concurred that civil society organizations were necessary to fill the service gap left by retreating states under structural adjustment, thus promoting "development with a human face" (Cornia, Jolly, and Stewart ).Specifically, civil society organizations were to provide services like education, health care, food provision, and so on when state provisions were insufficient or nonexistent.Importantly, however, civil society organizations were also deemed necessary for development because of their historically assumed civilizing and democratizing functions.As civilized organizations, NGOs were expected by definition to combat government corruption (a central concern implicitly supporting the neoliberal turn) by delivering services more honestly and effectively than states and by serving as a watchdog on government spending (Hamad, Swarts, and Ranniste Smart ).And as democratizing organizations, NGOs were expected, by definition, to advocate on behalf of citizens' rights and promote vibrant democracies (Bano ).
Calls for "more civil society" in the s departed from earlier civil society analyses primarily because of the newly central role of service provision.Humanitarian NGOs became the primary civil society agents delivering such basic human services to impoverished populations, including food, clean water, education, health care, and housing.Interestingly, the rationales for their assumption of these new service provision tasks were often expressed in terms of the traditional normative and functional notions of civil society: scholars came to argue that civil society organizations were actually better providers of social services than were states because civil society by definition was considered a repository for the collective will of the people, acting in solidarity with the constituents it served, and thus less prone to corruption and bureaucratic bloating and more capable of efficient and effective service provision to impoverished populations.
Humanitarian NGOs have existed for centuries (Keck and Sikkink ), but by all accounts their numbers exploded in the s (Hulme and Edwards ), 6 particularly in the global South.Humanitarian NGOs became the face of "civil society" in both development institutions and scholarly analyses.Calls to open the United Nations, the World Bank, and the Organisation for Economic Co-operation and Development (OECD) to greater input from "civil society" were met by institutionalizing both the consulting role and the service provision role of humanitarian NGOs (OECD ; World Bank ; United Nations ).Scholarly investigations operationalized "global civil society" as counts of international humanitarian NGOs, or some subset of them (Boli and Thomas ; Smith and Wiest ; Bush ).A quick read of scholarly articles in any development journal makes clear how the terms NGOs and civil society organizations are regularly used interchangeably (Howell and Pearce : and n. ).This is not to argue that scholars ceased studying other forms of civil society in development scholarship (such as women's groups, labor unions, or participatory budgeting programs; see, for example, Baiocchi, Heller, and Silva ; Agarwala ).Rather, we note that, from the s forward, the overwhelming majority of studies of civil society in development focused primarily on humanitarian NGOs (Pearce ).

The Centrality of Service Provision for Humanitarian NGOs
Although scholarship on "civil society" often highlights NGOs' advocacy work, the reality is that most humanitarian NGOs engage in some sort of service provision in addition to, or instead of, advocacy work.The reasons for this are multifaceted and interconnected.Initially, the NGO boom of the s grew out of a specific humanitarian need for service provision.Over time, funders increasingly looked to fund evidence-based policies and practices with measurable data showing the gains of the project (Ebrahim ).Since service provision is much easier to measure than advocacy outcomes, most NGOs found that getting funding required incorporating some sort of service provision project into their tactical repertoires.Meanwhile, as NGOs learned from their successes and failures in the field, a new understanding of "best practices" promoted a "holistic" approach to development studies among NGOs (Saha and Jayawickrama ).For example, one NGO that aims to promote education in an East African nation advertises the following holistic approach on its website: the organization first brings brick-making machines to a village and employs local masons to make the bricks.The villagers themselves are asked to supply their labor to construct the new schoolhouses.The NGO employees work with local education leaders to develop a culturally appropriate but progressively valued new school curriculum.The NGO also hires and trains the new village schoolteachers.Finally, the NGO engages in advocacy with the central government, specifically promoting new government initiatives to increase the priority given to girls' education.In short, humanitarian NGOs adopting a holistic approach to development combine advocacy, service provision, community building, and cultural training toward the achievement of a single development goal-improving the educational system of a village.
Scholars of civil society and development have sometimes suggested that they can parse out advocacy NGOs-which they consider part of civil society-from service-providing NGOs-which they do not.Although data are sparse, the growing literature on NGO missions, best practices, and funding evaluations strongly suggests that such divisions are almost impossible to achieve with empirical data (Krause ; Glennerster and Takavarasha ).Many humanitarian NGOs engage in both advocacy and development actions, and the ratio of advocacy to service provision is likely to change dramatically over time within any one organization depending on the issues in the current political context, the projects that are currently being funded, and the needs articulated by constituents. 7 In sum, the "civil society organizations" most typically studied in development scholarship since the s are humanitarian NGOs.These organizations engage in significantly different actions from the civil society organizations studied in the s.Most centrally, they frequently combine advocacy work with service provision.Nevertheless, because these humanitarian organizations have been viewed as central civil society actors, they have been imbued with the functional and normative expectations of "civil society" as developed in earlier scholarly analyses.

Three Pitfalls of the Civil Society Concept
As both the face of "civil society" in contemporary development studies, and as actors that significantly depart from the empirical reality of past civil society analyses, humanitarian NGOs provide an ideal illustration of our three pitfalls.In this section, we illustrate each pitfall with a vignette constructed on the basis of information we received while doing fieldwork in or about developing-country contexts. 8After describing the humanitarian actor and actions under investigation, we outline how the omnibus nature of the civil society concept would predispose scholars to artificially narrow their research design.We then elaborate how a third-sector approach frees scholars from preexisting expectations and focuses scholarly attention on empirically grounded analyses of power and influence, thus enriching and extending our understanding of the complex ways in which third-sector actors shape development.
Pitfall #: Because "civil society" actors are imbued with normative values like altruism and solidarity with their constituents, scholars design research programs that too often fail to uncover instances of "uncivil" society-for example, instances of third-sector corruption or community disconnect.
A national-level women's organization in El Salvador did extensive advocacy work in the capital city, but it also received funding to develop a project of "women's empowerment" in rural zones.Well-trained "promoters" were to enter rural communities that had been traumatized by war and organize the local women into strong women's organizations.These organizations were to empower women economically by providing them with a small, cooperative business opportunity.The organizations were also to empower women through the organizing process itself.With the support of the promotora, rural women were to learn about the power of association, the need for advocacy, and the importance of solidarity.In addition, they were to gain experience in leadership positions that would ideally translate into holding elected positions in other arenas of social life.The promotoras helped the new women's organizations to elect their own leaders and collectively choose the business they would run.In this particular village, the women chose to invest in sewing machines and run a small sewing cooperative.The promotora visited the women's group once or twice a month to support their meetings, give small-business courses in marketing and finance, and provide leadership training.As the promotoras themselves liked to joke, their other important job was to provide the refrigerio, or snack, as the presence of a tasty treat seemed to dramatically increase meeting attendance among women who already had extensive work obligations in their daily lives.
The usual civil society approach to understanding how this NGO affects development would probably begin by interviews with the NGO workers themselves in the national headquarters.The scholar would then be likely to visit the community with the promotora and interview women in the organization, asking questions about how their lives had changed because of the new association.The scholar might even collect "objective" data on women's lives-Have their incomes increased?Have they participated in protests or advocacy events?What have they learned about leadership?Our scholar might also interview the (mostly male) village leaders to determine whether and how the new organization improved the power and standing of the women in their community.Our scholar would spend time listening to women talk about the difficulties associated with the project-the fact that there were too few machines for the number of women who wanted to use them, the frustration that machines would break and they had no technician to fix them, and the lack of money to buy cloth needed to sew products in the first place.In short, the scholar would critically evaluate whether the project accomplished what the NGO intended it to achieve.Because she would assume a norm of solidarity between the NGO and the community it served, the scholar would be unlikely to question the organization's philanthropic intentions toward the community it served and would question only the barriers to whether and how those intentions were fulfilled.
Imagine now that the researcher entering the village had set out to investigate the impact of a value-neutral "third-sector" organization instead of a value-laden "civil society" organization.Had the researcher intentionally purged normative values from her research design, she might not have taken "solidarity with the community" at face value.She might have thought to avoid associating herself with the NGO upon her entrance into the community.This perceived independence might in turn have allowed her to seek out villagers who chose not to participate in the new organization and investigate their reasons why.
Had the researcher adopted a healthy skepticism about the NGO's solidarity with the community, she would have perhaps come to realize that another women's organization had existed in this village for almost a decade prior to the NGO's arrival.This first women's organization had been founded toward the end of the war, when women were among the majority leading a resettlement from the refugee camps to establish this new "repopulation."Women had been central players in founding and planning the new village, and they had formed their women's organization with the specific purpose of keeping women's interests front and center in community affairs after men returned from war and retook local leadership roles.The original women's group had no funding, no resources, and no snacks at their meetings.But every year they elected a new group president from among their members, and every year the group's leadership managed to keep women's concerns on the agenda of the local village council, even over the objections of many of the elected male leaders.
Had the researcher adopted a healthy skepticism about the NGO's solidarity with the community, she might have asked why, out of all the potential rural communities in the zone, the humanitarian NGO chose to work in this community.She would have then found that the village was chosen not because women there needed organization but rather because one of the NGO workers had a family relation in the community.This family relation was considered an essential social tie that would help the NGO establish support within the village and ensure the project's success.
The new women's group drew members away from the established women's group, largely because of the material resources it offered in microfinance, but also because women liked the trainings and the snacks they received from attending the new group's meetings.The woman who was credited with bringing these new resources into the community-the one who had a family relation with the NGO-was elected the first president of the new group, despite her relative lack of political experience and her lack of interest in local-level political affairs.Although there was technically no reason why a woman couldn't be a member of both groups, the tensions that evolved between the two groups, plus the difficulty women had in carving out time to attend even two community meetings per month, let alone four, resulted in a sharp divide among the community's women.The original women's group continued to exist with a much smaller membership.Its remaining members saw the new group as being run by women who were interested only in personal financial gain and who didn't understand the value of working collectively toward the good of the community.The women who attended the new group felt judged by the women in the old group; they felt they shouldn't be penalized because they chose an organization that provided them with trainings and a few minutes of uninterrupted time on a sewing machine each week.
The local village council was only too glad to support the new women's group.In addition to bringing a few additional financial resources to the community, the new group never made demands on the local council.Moreover, when the old group lobbied the council for a policy or product to support women, the council now used the existence of the new group to deny their request.The council argued that the original women's organization must be out of touch with what most village women wanted, as evidenced by the exodus of so many women from the old group to the new.The new organization therefore not only created a powerful rift in the solidarity between women in this village but also significantly weakened women's collective political voice.
Krause () and Bob (), among others, have already demonstrated that the right kind of "community" is very valuable to NGOs.As a result, NGOs are more likely to choose a community that supports their organizational mission than a community that is most at need for their services.Our "third-sector" approach takes these normative critiques a step further.Systematically eliminating scholarly assumptions of altruism from our research design not only improves our understanding of which community is chosen by an NGO, but also of whether the NGO promotes solidarity and empowerment when working within the chosen community.Only by eliminating the normative expectations of community building and solidarity from our analysis of third-sector actors will scholars be freed to investigate the multiple potential outcomes of a women's empowerment project like the one above, whether those outcomes reflect increasing solidarity, increasing division, increasing competition, or something else entirely.
Pitfall #: Because "civil society" actors are imbued with democratic functions, like advocating on behalf of constituents or serving as watchdogs on government performance, scholars design research programs that too often fail to uncover instances of the third sector's undemocratic actions or outcomes.
In El Salvador, a municipality is a small geographic territory similar to a US county, which encompasses a number of different villages and is governed by a mayor and a municipal council.The central Salvadoran government charges each municipality with providing infrastructure, support, and opportunities for each of the villages it governs, and it provides municipalities with a per capita stipend for carrying out these tasks.The stipends, however, are nowhere near sufficient for meeting the tasks assigned to the municipal government, especially in more rural municipalities, where the population may be spread across as many as  villages.As a result, municipal mayors must regularly make politically savvy decisions about which villages will receive their strikingly limited resources.Perhaps unsurprisingly, Salvadoran villagers often feel that mayors prioritize service delivery to the communities where their constituents live, with the expectation that those constituents will continue to elect them to office.
Nevertheless, in one particular Salvadoran municipality, a local humanitarian NGO repeatedly challenged the mayor's tendencies toward selective resource distribution.Across several years, the NGO advocated for help from the mayor's office in providing potable water to a cluster of communities whose collective water source had run dry.Yet despite the continual watchdog and advocacy work of the NGO, the mayor's office maintained its official line that the dry village would have to wait for its water plan until its turn came up in the -village queue for limited municipal resources; its needs were not considered any more pressing than the needs of others.
Although the NGO's local-level advocacy failed to make a difference for the dry communities, its grant-writing activities paid off.A large transnational NGO awarded the local NGO with financing for a sustainable water project.The international humanitarian NGO provided technicians who located a sustainable water source and then paid for the pumps, filtration system, and pipes necessary to distribute clean water to several clustered communities.The villagers themselves provided the labor of digging ditches and laying the pipes that would bring clean water to their homes.
Concerned with the project's sustainability, the international NGO required the local government's collaboration in the clean water project before distributing any funds.The international NGO wanted assurance that the municipal government would maintain the pumps, the filtration system, and the pipes.They also wanted the local government to promise a sizable monthly collaboration to help the communities pay for the large electricity bill that the pumps would generate.Such NGO-government collaborations are often considered "best practices" by development practitioners.Once the local NGO secured the local government's participation, the resulting clean water project became a reality.Indeed, both the local government and the local NGO now frequently showcase the water project to demonstrate their adeptness at managing large development projects.
If scholars anticipate that NGOs, as "civil society," promote a more democratic distribution of resources for citizens by advocating for citizen rights and needs with local governments, then this case certainly provides data to support their expectations.A local government was being particularly partisan in its distribution of resources.The local NGO called out the government's perceived partiality and eventually managed to convince the local government to collaborate in a clean water project for an especially needy village.The local NGO also brought in desperately needed financial resources to stretch dramatically what the government could have accomplished on its own.The "civil society" scholar would conclude that the NGO, as a civil society agent, fulfilled its expected functional role; it engaged in advocacy and watchdog work on behalf of its community, and it compelled the local government to be fairer and more responsive.As a result, more than a thousand people gained access to clean water and its associated health benefits.
However, if a scholar were to instead define the local NGO as a third-sector actor, free of any inherent democratic function, then the scholar might ask questions aimed more specifically at analyzing the empirical reality of the situation than the expected reality of civil society.For example, the scholar might wonder how the local NGO incentivized the mayor's office to incur this project's large, recurring financial responsibility, especially given that the beneficiaries of the project were not his "constituents."And the scholar might investigate the several means available to this local NGO to wield influence over the mayor's office, rather than assume that influence was wielded through the NGO's democratic activities.
When we ask these questions without the answers implied by civil society, the inadequacy of advocacy becomes immediately obvious.Of all the tools that this local NGO could use to wield influence over the local government, advocacy would seem to be one of the least effectual.The local government had successfully ignored the NGO's advocacy work for years.If the government had refused to support the water project, it would certainly have angered the entirety of the relatively small population waiting to receive services (the NGO's constituents), and it perhaps would have unified them in protest activity against the mayoral office.Yet even in this extreme scenario, refusing the water project would probably have had little negative impact on the mayor's political career.The NGO's expectation that the local government commit a significant chunk of its scarce municipal resources to their water project every year, at the expense of all the other communities awaiting their turn in the -year-long queue, could be easily framed by the mayoral office as inherently unfair and as an inappropriate action for the municipal government to take.In short, given the political realities of the municipal power structure, there is no logical reason to expect that the NGO's advocacy would prove sufficient to incentivize the mayor to make a significant, long-term financial investment in the much-needed water project.
After ruling out advocacy, the scholar might ask, what would be the local NGO's most powerful tool of influence?She would probably conclude that the NGO's primary source of influence would be its control over new financial resources in a resource-poor region and its potential to gain more financial resources through future project proposals.Investigating these logical hypotheses might then lead the scholar to uncover how the NGO actually achieved the mayor's participation-by promising the mayor a sizable under-the-table donation for his political campaign and input about which communities would benefit from future projects won by the local NGO.
In this case, adopting a third-sector approach to development would transform our hypothetical scholar's initial interpretation of this situation.Had the scholar approached the investigation by first defining the NGO as a democratic agent of civil society, she would probably have concluded that the presence of the NGO in this particular community caused an increase in the government's responsiveness by serving as a watchdog against partisan favoritism and by advocating for more equitable resource distribution.In contrast, if the scholar had approached the investigation by defining the NGO as a third-sector actor devoid of expected functions, then she would probably have based her analyses on observed empirical realities instead of on the historical meanings artificially attached to a "civil society" actor.In this alternative approach, the correlation between NGO presence and government responsiveness would be explained by the finding that the local government wielded influence over local NGOs in addition to controlling its own resource distribution.The NGO employed a variety of actions to achieve its goals-some democratic, some antidemocratic-given the political realities of the situation in which it operated and the tools of influence it had available.
Pitfall #: Because "civil society" actors are imbued with both normative values and democratic functions, scholars design research programs tailored primarily to investigate these two themes and too often fail to investigate the myriad of other possible actions and outcomes by which third-sector actors may also affect development.Significantly, they often fail to investigate the effects of the power and resources that many third-sector actors bring to the typically resource-poor areas in which they operate.
In the discussions of the above two pitfalls, we have argued that notions of civil society predispose scholars to assume normative values and democratic functions on behalf of thirdsector actors, rather than leaving the presence of those norms and functions open for empirical analysis.In this, our third pitfall, we take the argument one step further.We argue that, by focusing on what these actors are "supposed to do," the notion of civil society predisposes scholars to miss the power and resource effects of third-sector actors.We highlight three possible alternative consequences here: reshaping local-level inequalities, restructuring mobilization processes, and replacing voluntarism with coerced participation.
Local-level inequalities: In El Salvador, two rural communities situated just a few miles apart had significantly different engagements with humanitarian NGOs.One successfully tapped into the burgeoning NGO community shortly after the civil conflict in El Salvador ended and was awarded with a series of snowballing "projects," including the extension of its local school from  grades to , the opportunity to earn a high school diploma, clean water, and scholarships to study at the university level.Over time, as the civil conflict receded further and further into history, humanitarian projects in rural El Salvador became increasingly scarce, but this village remained particularly adept at securing even scarce NGO opportunities, perhaps because it had proven itself a successful recipient of past projects and seemed a "sure deal" for success to NGOs with contemporary projects, or perhaps simply because of its growing network connections with the humanitarian sector.When visitors walk through this village, the inhabitants proudly share stories about the many grand things its universityeducated youth are becoming-teachers, lawyers, engineers, and even a mathematical physicist ("although none of us really know what that is," they chuckle).As the youth are landing scarce jobs in the region-for example, as the municipal engineer-the welfare of the community as a whole seems to be on the rise.
The other community did not succeed in tapping into humanitarian projects after the war.Its children continued to receive a third-grade education, continued to have health issues from a lack of clean water, and never dreamed of a high school diploma, let alone a college degree.As the youth in the first village were increasingly prepared for professional opportunities in the region, the youth in this village saw their relative educational attainment plummet, even though their actual educational attainment remained the same.When a visitor walks through this village, instead of hearing about youth getting professional jobs in the region, the visitor hears about the many youth who now live in the United States, working in construction or service jobs and sending money home as often as they can to their families.In sum, these two communities had been relatively similar for centuries, and both continue to celebrate the hard work and sacrifice of their youth.But in the two decades of transformed NGO opportunities after the end of the civil conflict, the youth in one community seem to have experienced dramatically improved life chances, and as a consequence of their improvements the youth in the other community have experienced relative declines in their educational attainment, health, and employment opportunities.
Social scientists have developed numerous theories about how policies, markets, and cultural processes shape local-level inequalities.But we have yet to investigate how inequalities may unfold when humanitarian NGOs provide services to one community but not another.The traditional "civil society" approach to analyzing the outcomes of the humanitarian NGO above would be to ask if it succeeded in improving the life chances of its constituents.In the third-sector approach, where NGOs are stripped of normative or functional expectations and are investigated as power holders in a resource-poor environment, a scholar might instead ask whether a major resource investment in one community might have consequences for the region more broadly.How might the presence of NGOs shape the broader political economy of the region?
Restructuring mobilization: Early in , El Salvador was devastated by two earthquakes.A few months later, a European humanitarian NGO decided to construct safe, temporary housing for one of the many destroyed communities.The NGO provided the materials and resources for new cinder block houses; the future owners of the houses provided the labor.The last step in the NGO's plan was to provide the community with a source of clean water-an uncommon luxury in many parts of rural El Salvador.Another European organization had already built a water tank for two nearby villages, and it was generating enough water to support a third community.Recognizing this, the NGO requested and received permission to connect their newly reconstructed village to the water tank.All of these efforts on the community's behalf culminated in a ceremony in which the village members, tired and proud after having completed the construction of their new cinder block homes, gathered around the tank to celebrate the connection of their community to its water supply.
Suddenly, the ceremony was interrupted by angry men bearing machetes; they were determined not to let the new village hook up to their water supply.It was soon discovered that these men were not from the two communities for whom the tank was built but rather from another village, which had been buying water rights from one of the original communities to receive water.Although they had not secured approval to connect to the tank, they believed that because the tank had been given to the two initial communities, these two communities should be able to sell the excess water to whomever they liked.With a fourth community now hooked up to the water tank, there would no longer be enough water to go around.In the end, after obtaining a series of legal documents, the NGO succeeded in detaching the third community (the village that had been buying water rights) from the tank so that the NGO's own village could have unfettered access to the clean water.
Civil society approaches regularly envision NGOs as vehicles that carry "the people's" demands to the state.Yet as demonstrated above, and indeed in all our vignettes, humanitarian NGOs today frequently pair advocacy work with service provision.As service providers to resource-poor zones, NGOs have quickly become local-level power players.And we suggest that NGOs, as new local-level power players, are just as likely to be the target of people's demands as the vehicle carrying people's demands to the state.
In the example of the water tank, we know little about why one community was chosen over another for housing and water.Did the winning community simply get lucky, or did it work to strategically engage NGO investments?How might questions of competition, instead of collaboration, determine the kinds of organizing in which communities engage when targeting NGO resources?
Social scientists have developed many theories about how local citizens, often in conjunction with "civil society organizations," may target states or firms with social movement activism.But social scientists have yet to theorize how citizens might target humanitarian NGOs to meet their demands.Like states, humanitarian NGOs are power holders that often provide needed social services, like clean water.Unlike states, humanitarian NGOs have no constitutional obligation to provide social services to their constituents.Given these parameters, how might citizens mobilize to solicit needed goods, services, and protections from a humanitarian NGO?What new tactics or narratives might they utilize?Whereas most social movement mobilization processes studied to date incentivize cooperation and collaboration to improve the number of people advocating for change, how might the goal of gaining limited NGO resources prioritize competition instead of cooperation?Whereas civil society scholarships envision NGOs as advocates carrying the people's demands to the state, a third-sector approach that eschews assumptions of solidarity and prioritizes questions of power and resources allows scholars to see when NGOs are also the target of activism, and it can inform new theories of the multiple ways in which impoverished communities can mobilize for scarce resources.
Coercive voluntarism: A large transnational humanitarian NGO embarked on a project to develop local-level political power in India.Their reasoning?When the people "on the ground" hold governments accountable, when people "on the ground" have influence over what governments do, then governments will create development policies that are geared toward helping those in need, rather than those with whom they have political or economic connections.The project design was intelligent: the international NGO would first create local-level political bodies and then bring  or  of those new village-level councils together to determine an agenda and to push that agenda collectively at the regional level.By bringing people together, the NGO argued, the project would build community, promote advocacy, reduce government corruption, and, as a result, build development sustainably by creating a beautiful synergy between the regional government and the local citizens, a synergy that would last even after the humanitarian NGO left the region.As a bonus, the humanitarian NGO required that  percent of all the local-level council positions it created be filled by thus promoting women's leadership in a society long resistant to such gender transformation.Organized villages were rewarded for their participation in the program with an influx of needed resources, including microfinance loans, a new community building, a health clinic, and educational improvements for the community.
By any measure, this humanitarian NGO has apparently achieved almost statelike power in the communities it serves.Like a state, it has the power to provide needed social services like education, health care, and infrastructure.And like a state, it has assumed the power to initiate and regulate local political councils that in turn are charged with formally regulating village life.
A typical civil society approach to analyzing the impact of this organization would evaluate how well the NGO does what it says it does-Are communities associating?Are governments responding?Are women empowered?In this instance, scholars would likely be quite pleased with the NGO's outcomes: this project achieved unprecedented community participation in politics, brought new attention by regional governments to local demands, and literally pulled thousands of women into local leadership positions for the first time in the region's history.
A third-sector approach, which eliminated assumptions of NGOs and communities working together in solidaristic harmony toward achieving a shared goal, would instead investigate how the multiple third-sector associations working together in this arrangement would be likely to each have their own understanding of the situation and their own goals to achieve.As Ann Swidler has shown (Swidler ; see also Tavory and Swidler ), the cultural clashes that occur when northern-based NGOs come into developing nations to create major changes in entrenched cultural norms require a series of negotiations so that all parties can function together-negotiations that may allow the project to go forward, but with watered-down objectives and outcomes.
Returning to the situation above, a scholar might ask: In a region with a centuries-long resistance to women's political participation, what caused the almost overnight transformation to having  percent of all local-level municipal councils filled by women?The answer, as the NGO itself acknowledges, is that participation is incentivized by resources.If the communities don't put women on the councils, then the NGO won't provide the community with needed resources.
In other fields, scholars have been careful to distinguish how forced participation has different causes and consequences from voluntaristic participation.For example, in studies of political insurgencies, scholars agree that insurgents prefer ideological adherents to those who join for material gain.When times are tight, those motivated by material gain will defect, while those ideologically committed to the cause will stay (Viterna ).While it is possible for coerced actors to become ideologically committed, the process by which this may happen, or the extent to which it may happen, has yet to be fully investigated or understood.
Civil society scholars have long argued that simple participation is an important outcome, and for the voluntaristic associations studied by earlier scholars this is probably true.But as "civil society" has transformed in development studies to include other types of associations-like humanitarian NGOs-these expectations must be questioned.Certainly humanitarian NGOs are still defined as "voluntary associations" in the literature, and they are still valued by scholars and practitioners alike because they promote participation, which is thought intrinsically valuable.But we have yet to determine if coercive participation has the same consequences as ideological participation, or if coercive participation ever becomes ideological.This raises a number of questions important to development scholars.Would the individuals in the above communities have come together to advocate for their needs in the absence of the incentives provided by the NGO? Are they participating because they believe in the cause or because they see participation as a requirement for accessing needed services?Did they set their own agenda, or are they promoting an agenda set by the NGO?Should we anticipate that top-down, reward-driven participation processes confer the same benefits as do more organic forms of organizing (i.e., creating a politicized citizenry)?Does the fact that a collective association was created from above instead of from below affect the kinds of community it can build or the kinds of members it can attract?If "activists" are created through the incentive of needed goods and services, will they remain activists when the goods and services go away?These questions are particularly relevant when incentivized participation forces new cultural relationships between actors-like purported gender transformations-as well as new participation.

L E S S O N S L E
In the s, civil society organizations were defined as democratic agents because they were citizen collectives that engaged in associating and advocating activities.Although these types of associations (social movements, labor unions) are still studied in present-day civil society scholarship, they have been joined by a new kind of "civil society actor"-the humanitarian NGO. 9 As we have argued, humanitarian NGOs became the face of civil society in the development scholarship of the s, and they remain the most prominent type of civil society organization evaluated in the present-day development literature.
Like their earlier third-sector counterparts, humanitarian NGOs are expected to function as democratizing and civilizing agents in developing societies.Yet the gap between these humanitarian NGOs and the functional and normative meanings of civil society is sometimes quite wide.Whereas the "civil society" organizations of the s were composed of citizens making demands on governments, the humanitarian NGOs studied in the present day are more likely to provide advocacy on behalf of citizens, now commonly referred to as their "constituents."Most centrally, humanitarian NGOs, unlike previously studied civil society organizations, overwhelmingly pair or even replace their advocacy work with service provision.
Our analysis suggests that the omnibus nature of the civil society concept has inhibited scholarly investigation into the many ways in which these newly prominent third-sector actors have influenced development.On the one hand, the meanings associated with "civil society" have encouraged scholars to accept as unproblematic notions that humanitarian NGOs are altruistic, prioritize community, eschew corruption, advocate on behalf of constituencies, and promote cultural change.On the other hand, the meanings associated with "civil society" have narrowed the kinds of questions that scholars ask to focus primarily on whether humanitarian NGOs lessen poverty and/or promote community power and government responsiveness.These are important questions, to be sure, but they are not the only questions development scholars should be asking.If scholars were to replace civil society analyses with a structural third-sector approach, focusing on the empirical reality of the actors, their actions, and their interactions, then the analytical focus would broaden to include the many potential ways that third-sector actors may affect development.
Most centrally, we suggest that the development literature has yet to seriously investigate the multiplicity of potential consequences produced by the irregularly distributed introduction of needed social services into resource-poor environments by humanitarian NGOs.As major providers of social services like education, health care, clean water, transportation, jobs, seed and fertilizer, and loans, humanitarian NGOs in essence provide what T. H. Marshall () would consider the "social rights of citizenship."And as organizations with advanced technical capacities and ties to transnational advocacy networks, their presence may also strengthen the political power wielded by some local citizens, perhaps at the expense of others.If we understand humanitarian NGOs as local-level power holders, then we must anticipate that their actions will have powerful effects on any of a large number of local-level development outcomes, including how new players mobilize to achieve their demands, whom citizens target with their demands, whether citizens are incentivized to build community or to compete for scarce resources, and whether social hierarchies of class, ethnicity, and gender are transformed, reinforced, or exacerbated.

C O N C L U S I O N
In this article, we built on existing critiques of the civil society literature, identifying the "omnibus" nature of the concept as a key source of confusion.To explain the origins of this quagmire, we analyzed the theoretical genealogy of the civil society concept and argued that its messiness derives directly from its particular intellectual history.We showed that three distinct traditions of meaning from the long theoretical heritage of the civil society concept-a normative meaning, a functional meaning, and a structural meaning-were jointly revived in the specific empirical context of democratization movements in Latin America and Eastern Europe during the s."Civil society" was reborn in that moment as an omnibus concept, encompassing all three historical meanings simultaneously.Although the concept reasonably captured the empirical reality in that particular political moment, we have argued that its utility declined when scholars turned their attention to other civil society actors in later historical periods-like contemporary humanitarian NGOs.The omnibus concept could no longer accommodate critical variations in civil society actors, activities, and outcomes.Nevertheless, despite repeated acknowledgment in the literature of this conceptual messiness, these various components have persisted to a significant degree in contemporary scholarship.Building on these critiques, we conclude that a pared-down structural concept-the third sector-is the most useful analytical category for empirical scholars and that the concept of civil society should be retired from empirical studies of development.
Some readers may wonder: Why is it not enough to simply avoid an omnibus usage of civil society?Indeed, taking care with the concept-using civil society only in a structural sense in empirical studies and defining and operationalizing it accordingly-gets us much of the way toward avoiding the pitfalls we have discussed.However, we go a step further and argue that the term imposes extra requirements of clarity, parsimony, and care that are nearly impossible to meet without adopting new terminology.Indeed, it is because civil society is typically meant to denote a political ideal-a space that permits the open exchange of ideas among diverse actors-as well as a set of actors that the concept has achieved such widespread usage in academia.Prominent scholars have argued that studies of "civil society" must remain central to our analyses despite the messiness of the concept, precisely because the political project itself (that is, developing the right kind of civil society) is important for social advancement (Pearce ; Edwards ).As a result of this usage in scholarship, policy, and activist circles, it has become practically impossible to divorce the normative and functional elements of "civil society" from its characteristic actors.Given its two-millennialong theoretical history, we do not advocate separating the political project from the terminology.Political theorists have done a lot of important thinking in exploring how societal actors might support norms of civility, and that theoretical heritage should not be left without a conceptual heir to continue the conversation.
We thus aim to split the current concept into two: a structural third-sector concept to be used to identify actors in empirical scholarship, and a normative civil society concept to be used by political theorists studying civility in society.
The third sector: A sector of organized human action composed of collective actors beyond the family and distinct from the state and the market.
Civil society: An idealized theoretical space in which societal actors come together for discussion and debate and produce shared normative values relating to our evolving ideas about civility.This conceptual reorientation both preserves a normative concept and a structural concept and distinguishes between them.Functions-the actions and outcomes of third-sector organizations-should be left primarily to causal analysis.This amounts to reclaiming a third-sector concept and retiring the concept of civil society from empirical scholarship.
We are aware that this proposal will be met with objections.The term civil society has long been in use, whereas the term third sector has emerged only recently (and particularly in European scholarship).One can argue that a significant shift in concepts and accompanying terminology will undermine meaningful continuity in scholarship on the issue.Most literature on NGOs to date refers to civil society; new scholarship that focuses on the third sector will face challenges in establishing and framing connections to past work.In short, theory is inertial and there are benefits to conceptual continuity.However, in this case the benefits of continuity come at an unacceptable cost.We have argued that no matter how careful individual researchers may be about disciplining the civil society concept to a well-specified structural definition in their own work, doing so en masse requires too much conceptual vigilance to prevent other elements of the omnibus concept from creeping into our collective analyses.As we have suggested, when functional and normative assumptions are built into the concept of third-sector actors, we risk missing important variation and we fail to allow the empirical reality of these organizations to drive our research questions.When we miss asking the right questions, we lose out on making the contributions we should be making as social scientists, in recognizing and explaining changes in the empirical reality of the third sector in developing countries.It is time to stop allowing civil society's conceptual history to drive our research agenda.The gains in clarity and analytical leverage will be well worth the growing pains involved in reclaiming the "third sector" from "civil society." In this article we have focused our analysis on the civil society concept as applied to studies of development.However, we believe that our conclusions may also hold for many other social science subfields that study civil society and its effects on social and political outcomes (e.g., democratization, social movements, ethnic conflict).Though we have not engaged directly with the way in which these literatures operationalize "civil society," we posit that their application of the same omnibus concept leads them into many of the same pitfalls that we have identified in the scholarship on development.Therefore, just as we argue that the third-sector concept is more useful for studies of development, we urge scholars writing in related subfields to consider the same conceptual shift in their empirical analyses of other social and political processes.
More broadly, there are lessons to be learned from this analysis about concept building and concept maintenance.First, when we build concepts from empirical reality (as all scholarship does), sometimes the problems with the concept become evident only when the empirical reality changes.As Sartori () argued, we want our concepts to travel to a meaningful degree beyond the context in which they are developed, and for this to be effective they must be high enough up the ladder of abstraction that they are generalizable.Since the omnibus concept of civil society was a good fit during its revival in s Latin America and Eastern Europe, it was not obvious that its bundled meanings might give it limited utility outside that context.A close look at how the concept fares in the very different context of contemporary humanitarian NGOs in developing countries, however, is revealing.Examining old concepts critically in new contexts gives us important information about the suitability of those concepts for analysis in a broad set of cases.
Second, foundational concepts like "civil society" that are used in the literature over long periods of time are at risk of what we might call theoretical creep.In general, social science theories frequently involve proposing connections between different kinds of variables (like associations, civility, and democracy).Theories recognize variation: they hypothesize a systematic relationship between variables, but they do not assume determinism or covering laws.The trouble comes when a well-known theory has been borne out in familiar contexts and has been accepted as generally convincing and well supported over an extended period of time.Under these conditions, the theory can become ossified into a single concept, in which the relationship between the relevant variables is treated as constant and determined.In that sense, what starts as a theory (third-sector organizations may produce democratic outcomes and reflect civility norms) can become a single concept (civil society is ipso facto democracy supporting and civil).It is a short step to move from a concept to a variable.Thus if we are not careful, the process of theoretical creep reduces multiple variables involved in a theory down to a single variable.
Both of these points highlight the importance of occasionally revisiting foundational concepts in social science to ensure that meaning aggregation and theoretical creep have not corroded our concepts.As we have shown here, such reexaminations can result in the productive disentangling of multiple meanings and can yield sharper, clearer analytical tools for empirical study.To keep our concepts well tailored to the evolving realities of the social and political world, we believe that taking the time to systematically clean house conceptually is worth the trouble.
parties?Or public universities?What about the increasing trend of NGOs partnering with states to jointly provide service provision?We recognize that the boundaries between state, market, and third sector may not always be clear, but we maintain that the term has utility nonetheless.As a concept that describes the shared characteristics of a set of similar social forces and organizations, it can be a highly useful analytical tool.Although occasionally these forces and organizations may share characteristics from other sectors, conceptually it still makes sense to maintain the distinction between these sectors.
.The number of transnational NGOs identified by scholars has grown from  in , to , by , to over , today (Hulme and Edwards :; Union of International Associations (, , ).Although there are no estimates on the number of national or bilateral NGOs in existence globally, a few country-level estimates make clear that the number is in the millions.For example, the US Department of State () estimates that the United States has . million NGOs; meanwhile the Indian government has estimated . million NGOs in its nation (Shukla ).
.Even when humanitarian NGOs do not provide material services, their advocacy actions still differ significantly from earlier conceptualizations of civil society.The humanitarian NGOs that proliferated in the s often advocated for citizens not just within nations but also from the broader transnational advocacy community (Keck and Sikkink ; Bob ).Bob and others have demonstrated how connections to powerful transnational actors, though not necessarily material goods, can launch some movements to prominence while leaving others with highly limited abilities to influence policy.
.Each vignette reflects a story that we uncovered or observed in fieldwork or while conducting secondary research, often related to us by well-positioned key informants.Because we have not conducted careful empirical analyses on any of these cases, we cannot attest to the factual content of our vignettes.Indeed, the second vignette is actually an amalgamation of several stories taken from the same zone and combined into one deliberately cohesive example.Nevertheless, since the point of the vignettes is to illustrate how traditional approaches to civil society predispose scholars to miss key development outcomes, the factual content of our cases becomes irrelevant.Our vignettes are valuable because they provide a logical illustration of what questions we miss when we allow the omnibus notion of civil society to influence our research design.
.The difference between the two types is not nearly as crisp as presented here, given the increasing tendency of social movements to formalize into more bureaucratic organizations.Certainly the teachers' unions calling for democratic overthrow of authoritarian governments of the s used strikingly different tactics from the present-day teachers' unions lobbying for salary increases.