Introduction

While the existence of associations and charitable organizations is not new, the global proliferation of these organizations and legal units has grown rapidly since the 1960s in what has been termed a “global associational revolution” (Salamon, 1994). Among many outcomes, the growth of the nonprofit, or third, sector has challenged the theories underpinning common understandings of economic and labor force activity and the systems that governments use to measure it. The study of the third sector was initially guided largely by classical economic theories (Salamon, 1995) which influenced thinking about the sector’s size, financing, and cross-national patterns. According to these early theories, nonprofits emerge in response to consumer preferences for collective goods left unsatisfied by both the market and the state; to situations where the normal market connections between consumers and purchasers do not operate and trust becomes necessary; and to the presence of religiously rooted popular sentiments of altruism and charity. These theories have helped reinforce prevailing, though misguided, assumptions at the time (Salamon et al., 2017): (a) that nonprofits are most plentiful where government social welfare provision is least generous; (b) that philanthropy is the dominant source of nonprofit revenue; (c) that nonprofits dominate industries such as nursing homes where consumers and purchasers diverge; and (d) that third sector size varies directly with the presence of religious traditions emphasizing altruism.

Systematic empirical research undertaken from the beginning of the 1990s on (Salamon & Anheier, 1996; Salamon et.al. 1999), however, has largely discredited many of these prior assumptions; and also began to address the parallel issue that explanations of variations in volunteer participation had been difficult to pin down (Lyons et.al. 1998; Butcher and Einolf 2017). In doing so, systematic comparative research cast serious doubts about the conceptual and methodological underpinnings of the third sector field (Salamon amd Anheier, 1998; Salamon et al., 2017). In this article, we discuss the methodological foundations of efforts over the past 30 years to improve comparative nonprofit sector research by overcoming the lack of systematic, reliable and accurate data. We see this as a history of science, in a way, of the cross-national, comparative nonprofit research field which was significantly fueled by the application of the methodology discussed here and still is. Creating a record of how comparative data were created and improved serves to preserve the process that was achieved.

The misperceptions about nonprofits, philanthropy, and volunteerism that prevailed before systematic comparative research was first undertaken also had their roots in inadequate data generated through the two official global statistical systems: the System of National Accounts (SNA) overseen by the UN Statistics Division (UNSD, 2003) has made nonprofits virtually invisible in official statistics; and the Labor Force Survey (LFS) system overseen by the International Labor Organization (ILO, 2011) had not distinguished volunteering from other forms of unpaid work (The major statistical systems discussed in this article are further detailed in Appendix 1). The SNA had been using an economic-financial definition that classified organizations primarily by dominant revenue sources (Salamon & Anheier, 1997): nonprofits with most revenues from fees for service and other earned income were classified as part of the business sector; nonprofits receiving most funds from government as part of the public sector; and philanthropic foundations earning investment income from their endowments as part of the financial sector. As a result, official statistics only showed the very small number of nonprofits which draw a majority of their funding from philanthropic gifts and grants as nonprofit institutions serving households (NPISH). With the sector thus appearing as small and irrelevant in official statistics, common misperceptions continued about the importance of philanthropy as revenue source and the relation of nonprofits to the welfare state.

Definitions and methods for measurement therefore have important consequences, not just because of the data that are reported out, but also for policy, administrative processes and regulatory relationships that organizations and associations have with the state. What is measured can impact filing and reporting requirements and these changes open opportunities for the sector to improve its experience in this regard. In India, for example, the decision to undertake a census of nonprofit institutions resulted not just in a comprehensive set of information about the sector nationally, it also resulted in transforming its filing and record-maintenance systems from a paper-based to a digital system (Government of India, 2012).

Over the last 30 years, major success has been achieved in mitigating these problems, the advancement of tools for conceptualizing and measuring the basic size, scope and scale of the global nonprofit, civil society, or third sector, and the volunteer workforce it mobilizes, and in the adoption of these tools by official statistical agencies (e.g., Salamon, 2010). Despite the progress that has been made, more information and training are needed to inform the third sector research community of these developments, to clarify the role researchers play in their implementation, and to explain the information that might be generated by their use (see Sokolowski, 2022). This article provides a step toward these goals.

To do so, the discussion here unfolds in three sections. First, although the earlier stages have been documented elsewhere (Salamon, 2010; Salamon et al., 1999), we remind readers of the bottom-up research and consultative process undertaken in the development of these methods and conceptualizations, and the steps that were taken to encourage their institutionalization in official statistical systems. In this section, we update the information that has been previously provided to provide a full picture of the tools and resources available. Second, we discuss the theories that were generated and tested as a result of this work. A concluding section then identifies a number of barriers that still impede the full implementation of these systems and the lessons learned from past implementation experience.

The Analytical Process: A Bottom-Up Strategy with an Institutionalization Objective

Beginning in the late 1980s, the realization began to grow that the global statistical systems were not well equipped to collect and report on basic economic data on the third sector (Anheier, Rudney, and Salamon 1994). As noted above, the existing global statistical infrastructure did not enable a separate identification of, and reporting on, third sector institutions fostering assumptions at the time that the nonprofit sector represented only a relatively small part of the global economic system, was not of major economic consequence, and thus not worthy of much attention in the statistical systems of countries.

The now well-known Johns Hopkins Comparative Nonprofit Sector (CNP) Project (Salamon & Anheier, 1996; Salamon et al., 1999; 2004) sought to counter these assumptions and produce definitive economic proof that the sector could, first of all, be conceptualized as a distinct economic sector in a cross-national, comparative context, and second, was large enough to be deserving of at least the same degree of attention provided to other economic sectors, such as manufacturing, construction, shipping, or retail trade. To make meaningful progress on its three target tasks of conceptualizing, measuring, and testing theories about the third sector, the work adopted a consciously bottom-up strategy that engaged teams of researchers in a wide array of countries and religious traditions. Over the years the CNP project evolved from being a private research effort into an outward-facing public effort to develop global statistical standards and promote their adoption and institutionalization in the major global statistical systems. Though the work shifted focus over time, the projects’ criteria and bottom-up approach that guided the conceptualization and operationalization of measurement standards remained consistent.

At every phase, five key criteria guided the collaborative work to conceptualize third sector organizations and volunteer work. The resulting consensus conceptualizations embody:

  1. (i)

    Sufficient breadth to accommodate the sector’s great diversity;

  2. (ii)

    Sufficient clarity to differentiate the third sector from other types of institutions and behaviors, particularly corporate and government entities and leisure activities;

  3. (iii)

    Comparability, to facilitate cross-national comparisons;

  4. (iv)

    Operationalizability, to be objectively identifiable and not dependent on vague or subjective characteristics; and

  5. (v)

    Institutionalizability, to facilitate integration into international statistical systems.

These criteria in turn led to the identifaction of several operational features for the type of conceptualization research teams would be seeking:

  1. (i)

    Institutional focus: Given the enormous diversity of the third sector, a decision was made to focus initially on the sector’s basic organizational core: its nonprofit institutions (NPIs) and the volunteer activities associated with them. Later, when measurement protocols were clarified, the project embraced the measurement of individual activities and social economy institutions. Informal organizations are included conceptually, but their measurement has not yet been a priority.

  2. (ii)

    A common core conceptualization: Rather than an all-encompassing definition, a decision was made to adopt the approach that is standard in scientific taxonomies: to identify the narrowest array of “common core” features capable of embracing the widest array of in-scope entities.

  3. (iii)

    Use of proxies: to avoid overly subjective attributes, find objective features that effectively stand in for them (e.g. the nondistribution-of-profit in the case of NPIs as a stand-in for pursuit of “public purpose.”)

These criteria guided the work through its four major phases:

Phase I: The Johns Hopkins Comparative Nonprofit Sector Project

The Comparative Nonprofit Sector Project (CNP) was a major research effort mobilized by the Johns Hopkins Center for Civil Society Studies to conceptualize and then document the scale of the global nonprofit sector. Ultimately, over 40 countries were covered by this project, in each of which local research teams and advisory committees were recruited to identify the key components of this diverse sector. This process yielded a consensus common-core conceptualization of the nonprofit sector that then provided the framework for the data assembly process (Salamon & Anheier, 1997) that would ultimately be adopted by the United Nations. While far from perfect, this effort yielded the most comprehensive and systematically cross-national body of data on nonprofit organizations and volunteering ever assembled (Salamon et.al. 1999; 2004).

While legal definitions of organizations differ by country and are thus unsuitable for comparative purposes and the SNA’s economic approach was too limited, five common-core features were found to characterize the in-scope nonprofit sector institutions in the broad array of countries we examined (Salamon & Anheier, 1997; Salamon et.al. 2004). In particular, to be in-scope, these institutions had to be:

  • Organizations, i.e., to have some structure and regularity to their operations, regardless of whether they were formally or legally constituted;

  • Private, i.e., institutionally separate from the state and not “controlled by” it, even if they receive state financial support;

  • Non-profit-distributing, i.e., prohibited either by legislation, by-laws, or custom from distributing any profit they generate to their stakeholders. This effectively differentiates NPIs from for-profit enterprises;

  • Self-governing, i.e., fundamentally in control of their own missions and operations, and able to cease operations on their own authority; and

  • Noncompulsory, i.e., engaging people on the basis of freely given consent.

This conceptualization embraces a wide assortment of different entities. Moreover, the combination of the noncompulsory and non-profit-distributing features provided useful proxies suggesting that those involved in organizations with these features must have believed that the organizations served public purposes since the participants were involved voluntarily and the organizations were not allowed to distribute profits to them. While quite broad, this initial conceptualization nevertheless excluded so-called “social economy” institutions that were equally committed to a distinct public interest purpose (Evers & Laville, 2004). These institutions were initially excluded because they can distribute profits to their stakeholders, making them hard to distinguish from for-profit enterprises.

Phase 2: Penetrating the United Nations System of National Accounts (SNA)

Armed with the CNP findings, attention turned to reconfiguring the conceptual lens through which the SNA viewed nonprofit organizations and volunteering. Careful conceptualization of the third sector was a necessary precursor to systematic, comparative measurement. A key objective of the CNP effort was not simply to carry out a one-off research project but to institutionalize the measurement of the third sector into the SNA by demonstrating to official statistical agencies that it was feasible to objectively differentiate nonprofit organizations from other economic entities beyond differences in legal status and analyze their economic activity separately.

After reviewing the data produced by the CNP and comparing it to the data reported out through the SNA, the United Nations Statistics Division (UNSD) authorized the creation of a Handbook on Nonprofit Institutions in the System of National Accounts (UNSD, 2003) to guide national statistical offices in producing regular “satellite accounts” on NPIs and volunteering. This publication marked the first official identification of a distinct NPI sector embracing both market and non-market NPIs and volunteer work in official global statistics.

Despite the progress the publication of the NPI Handbook represented, important conceptual and methodological limitations remained (Salamon, 2010). These included, most notably, (1) the limitation of the NPI Handbook to “institutional units” within the “production boundary” of the economy, which made it difficult to include critical elements of civil society, such as social movements, citizenship, and participation; (2) a lack of guidance for the definition and measurement of volunteering, which when mentioned was limited to “formal” or “organization-based” volunteering; (3) coordination, but a lack of full integration, with international classification systems; and (4) restriction of measurement to nonprofit institutions without recognition of the social economy institutions that are key elements of the social sector in many countries. To address these shortcomings identified by Salamon (2010), a third and a fourth phase were launched.

Phase 3: Defining and Measuring Volunteer Work

While the publication of the 2003 NPI Handbook established the principle of officially valuing volunteer work in global economic data, it left open crucial operational issues about how to do this (Salamon et al., 2011). What was needed was the identification of characteristics that distinguish volunteering from paid and other forms of unpaid work so that statisticians could objectively identify this form of activity independent of local conventions, and include it in their accounting practices. To fill this gap, the International Labor Organization agreed to lead, and the United Nations Volunteers (UNV) Programme agreed to fund, a process for the development of a definition and a set of measurement protocols for volunteer work. The ILO commissioned a Technical Experts Group composed of labor statistics officials, volunteering researchers, and representatives from the UNV to establish the main criteria and measurement framework for a Manual on the Measurement of Volunteer Work that was officially issued by the ILO in 2011. The memos that document the entire deliberative process for the conceptualization of volunteering and the most feasible methods for measuring it in the maximum number of countries are available (CCSS 2010).

Clarifying the conceptualization of volunteer work required distinguishing volunteering from leisure activities and household work, to specify the boundaries of direct as opposed to organization-based volunteering, and to find language for identifying volunteering without using the term “volunteering” itself. Volunteers are identified as: “all those of working age who, during a short reference period, performed any unpaid, noncompulsory activity to produce goods or provide services for others.” In addition to avoiding mention of the term “volunteering,” this conceptualization makes clear that in-scope volunteering is:

  • A form of work, to differentiate it from leisure activity and make clear it generates economic value alongside whatever social value it produces;

  • Unpaid, though volunteers may receive symbolic gifts and modest expense coverage. But participation in corporate volunteering programs during paid working time is not volunteering but is instead classified as corporate philanthropy;

  • Non-compulsory, i.e., involving a significant element of choice and not legally mandated (e.g., as an alternative to incarceration or required military service) or done by those under age. Volunteering in response to social or religious pressure is still in scope;

  • Done directly for persons in other households (direct volunteering) and through organizations (organization-based volunteering); and

  • Done only for persons outside one’s related family.

The adoption of this definition vastly expanded and clarified the previous CNP approach to conceptualizing and measuring volunteering, particularly the distinction between organization-based and direct volunteering. The recognition by the International Conference of Labor Statisticians of volunteering as an official form of unpaid work in 2013 opened the door for the integration of measurement approaches across the spectrum of tools made available by the ILO.Footnote 1

Developing strategies for the optimization of measuring in-scope volunteering at the national level required resolution of three key design issues: (1) Identifying an appropriate survey platform that would capture both direct and organization-based volunteering. The survey platform needed to be a household survey rather than an organizational one. Labor Force Surveys (LFSs) emerged as the preferred platform because they are conducted in nearly every country, are carried out at least annually, use skilled interviewers, tap super-large samples, and capture demographic data helpful in analyzing the demographic profile of volunteers, and responses to these surveys are usually required, helping to reduce the dreaded response-bias among volunteers (ILO, 2011, paras. 24, 25). Other platforms meet these criteria as well and can be used. Time-use surveys and stand-alone surveys can also provide useful data, though these are not widely implemented at regular intervals and are costly to implement, making them less useful for comparative purposes. The 2011 Manual thus proposed a module to be attached to a labor force survey, and this module has since been tested, updated, and adapted to other platforms by the ILO (Ganta, 2021; ILO, 2018, 2021).

The main features of the ILO measurement approach is that it is designed to gather the maximum amount of information with the fewest number of questions. Including questions about volunteering on other survey platforms enables the reporting of more than simple headcounts of how many respondents volunteer by connecting data on volunteering to the other demographic and socio-economic data being collected. Since LFSs already collected respondent demographic data, the volunteering module could focus its attention on multiple dimensions of each volunteering engagement and generate data on the number of volunteers, the hours volunteered, the type of work performed, the institutional setting, and the industry or field in which the volunteering occurred. Given these data items it becomes possible to compute the full-time equivalent (FTE) number of volunteer workers, the share that volunteers comprise of the third sector workforce and of the economically active population of the country, the economic value of volunteer work, the distribution of FTE volunteers among different fields, the volunteer share of each industry’s workforce, and the volunteer contribution to the country’s GDP (UNSD, 2018, paras. 4.35–38; ILO, 2011, paras. 5.11–17).

Phase 4: Bringing in the “Social Economy” in with Volunteering and Classification

Finally, to respond to growing interest in so-called “social economy” institutions (i.e., cooperatives, mutuals, and social enterprises), a major revision of the original 2003 NPI Handbook was undertaken to extend its reach to this broader set of institutions (Fig. 1). The key challenge here was how to differentiate social economy institutions from for-profit businesses in objective non-legal terms, since, unlike NPIs, they can distribute profits to their stakeholders. Working with 12 collaborating institutions and extensive practitioner input, a consensus approach to this challenge was developed and integrated into a revised UN handbook (UNSD, 2018). This was done by relaxing the total non-distribution-of-profit constraint applicable to NPIs and considering in-scope social economy units of newly termed “third or social economy (TSE) sector” that by law, organizational by-laws, or established practice distribute no more than 50 percent of their profits to members or other stakeholders, and that adhere to a “capital lock” requiring that any retained earnings be transferred to a similar social-purpose organization in the event of dissolution or conversion to for-profit status (Enjolras et al., 2018; Salamon & Sokolowski, 2016). This broadened conceptualization of in-scope institutional units was incorporated into the 2018 UN TSE Sector Handbook (UNSD, 2018).

Fig. 1
figure 1

Conceptualization of the Third Sector: Nonprofit and related institutions Based on Salamon & Sokolowski 2016

Full size image

The revision of the SNA in 2008 provided the opportunity to update the International Standard Industrial Classification (ISIC) system, which at the time did not provide sufficient detail on the activities of third sector organizations. This basic shortcoming of ISIC had already in the 1990s led to the creation of the International Classification of Nonprofit Organizations (ICNPO) (Salamon & Anheier, 1997). The revised version of ISIC (ISIC Rev. 4) adopted much of the detail presented in the ICNPO, making it possible to report out on the fields of TSE activity that are of interest to those concerned with the third sector.

The implementation of the new TSE Sector Handbook will generate the following data:

  • The number of TSE sector institutions, by type of institution and major field of activity;

  • The number of TSE sector workers, both paid and volunteer;

  • The “value added” by TSE sector organizations;

  • The value of volunteer work, both direct and organization-based, by field;

  • TSE sector operating expenditures;

  • Sources of TSE sector revenue, including philanthropy, fees, and government support; and

  • The size and distribution of foundation grants.

All of the information above is broken out by type of unit and by industry or field of activity, the latter using a classification structure that permits comparison to for-profit institutions in the same fields (Table 1). In short, the UN TSE Handbook lays the groundwork for generating cross-nationally comparable data of the third sector in different countries, thus creating a foundation for systematic analysis and theorizing of variations.

Table 1 Allocation of the TSE institutions and volunteer work to institutional sectors in the core SNA
Full size table

Analysis and Theorizing

Conceptualization and measurement are not, of course, ends in themselves. Rather, they are crucial steps in a broader process of analysis and theory-building, often revealing unexpected variances for which new explanatory theories have to be found. The range of puzzles and resulting theorizing that the CNP project, broadly conceptualized, have begun to stimulate are already rich, and promise to grow even more robust as further implementation of the UN TSE Sector Handbook and volunteer measurement tools proceeds to produce additional data and as other researchers tap into its results.

The discussion here illustrates this point with just one example—the challenge of explaining the varied patterns of third sector development that the data generated has so far brought to light. As previously noted, the need for such theoretical advances was already clear in this field when the work was launched since much of the available theorizing was essentially deductive.

The major outcome for the CNP project has been the development of the social origins theory of civil society development (Salamon & Anheier, 1998). As detailed fully in Explaining Civil Society Development: A Social Origins Approach (Salamon et al., 2017), the data resulting from the implementation of the CNP project in 41 countries enabled us to, first, determine if the observed features of the third sector fell into any distinctively different empirical patterns; and second, to assess what factors could explain the patterns that existed. As it turned out, five such patterns of civil society evolution emerged from our data as shown in Table 2.

Table 2 Patterns of power relations and their hypothesized effects on the civil society sector
Full size table

Significantly, of the 41 countries on which data were available, 26, or 63%, fell into one, and only one, of these empirically-identified patterns, and an additional 7, or 17%, were at least on the border of one of the patterns. This meant that 80% of the cases under examination fell squarely, or nearly squarely, into one of the identified civil society patterns. Furthermore, our 41 countries were widely dispersed among the identified patterns.

We then examined whether the structures of social and economic power identified in the social origins theory could explain the different patterns of third sector development evident in our data. Our testing provided strong confirmation of the social origins theory’s explanation of the factors shaping the patterns of third sector development identified in our data (Salamon et. al. 2017).

The development and testing of the social origins theory provides confirmation of the theoretical contributions to the third sector field that the conceptual and measurement work outlined here is capable of making. The work on the social origins theory is not complete, however. It awaits further testing and development following the generation of a wider variety of robust national data (e.g., Anheier et al., 2020). We anticipate further clarification of the status of borderline countries: does that status represent a transition from one pattern to another? Or does it suggest a new pattern yet-to-be identified? What information will be gleaned from the 150 + countries where no comparative nonprofit data exist as of yet? What new puzzles will emerge and what new patterns will be identified?

Pursuing Implementation –Lessons Learned for Future Researchers

The conceptualization work and resulting official measurement tools identified above provide a clearer set of lenses for identifying nonprofit, civil society, social economy organizations and volunteering. They provide a set of rules established for separately identifying organizations in SNA data and volunteering via household surveys, and for reporting on them through regular “satellite accounts” (UNSD, 2018). This has opened a potential treasure chest of systematically cross-national data on the third sector in every country.

However, such implementation is far from automatic. So far, approximately 30 countries have generated such satellite accounts, in full or in part, at least on NPIs, with others broadening the focus to the full third sector institutions. The production of these satellite accounts resulted from the mobilization of the research community in these countries who encouraged national statistical agencies to implement these recommended satellite accounts, monitored and supported their implementation, and collaborated to report on the resulting information. The steps many researchers took to engage with national statistics offices and the lessons learned by their experience include that theyFootnote 2:

  1. (1)

    Identified what level of reporting is already available nationally. In many cases, a great deal of data is often already collected, but is simply not reported publicly and therefore not easy to find.

  2. (2)

    Established communication with national accounting offices and labor statistics offices to seek collaboration (Einarsson & Wijkström, 2019). Statistical officials face pressure to align their data to many updated standards and report on several different economic sectors. Where they have discretion to decide these matters, they are more likely to report on institutions for which there is a perceived demand for information and support for its production.

  3. (3)

    Supported the identification of organizations to be measured proactively. Statisticians are not typically familiar with the broad landscape of third sector legal definitions or activities. Researchers can play a key role in helping them to identify which local organizational types meet the international definitional criteria, and which are borderline. Statistical officials typically welcome support in identifying in-scope institutional types from experts in the field.

  4. (4)

    Helped design and test the survey questions to gather more information that reflect the nuanced understandings of these concepts on the ground (Guidi, Fonović, and Cappadozzi 2021). Civil society and researcher input are critical in ensuring that local nuances in terminology are properly used and understood in order gather the most accurate information.

  5. (5)

    Brought in and involved stakeholders to gather feedback, ideas, support, and funding. Doing so helped ensure that the resulting information was understood and supported by the communities it was intended to serve. Moreover, the engagement of stakeholders underscored the importance of implementation for national statistical offices, which have limited funding and many priorities.

The work has required persistence. Statistics agencies did not often respond to initial requests and when they did respond they were not initially enthusiastic. The development of these data is time-consuming. The learning curve for researchers and civil society stakeholders to understand the official systems, and for government statistical officials to understand the full scope of the TSE sector, was initially steep. Government budgets and policy documents planned years in advance proved to be barriers to implementation. It was difficult to identify who is responsible for making decisions to move forward. We learned that successful partnerships with government on this effort requires a focus on the long game. With time, however, trusting relationships between government, researchers, and third sector representatives can be built to successfully overcome these barriers to comprehensive nonprofit data collection.

Conclusion

The pattern of conceptualization, measurement, and theory-building outlined here constitutes substantial progress in the development of nonprofit studies, particularly from a comparative perspective. The institutionalization of these advances in the official SNA statistical system adds to their significance. The published satellite accounts to date demonstrate the tremendous opportunity that exists to grow the base of cross-nationally comparative empirical data on the third sector and thereby open the way for vastly expanded empirically grounded analysis of the variations in data.

For many years, the development of these data was achieved by a constellation of researchers, practitioners and funders, loosely coordinated by the team at the Johns Hopkins Center for Civil Society Studies. With its closing in January 2022, the future of comparative international data on the sector enters a new phase. It is our hope that the research community will actively engage with national statistics offices to urge them to develop TSE satellite accounts and to actively partner with them in doing so. We further hope third sector researchers will become familiar with SNA terminology and data systems in order to become better equipped to monitor the evolution of the guidelines at the UN level and work with the potential resulting data at the national level. The active engagement of the third sector research community has been crucial in the development of the international tools that are now available. The near-term future of solid, cross-national empirical research and empirically based theory-building in the third sector field may well depend, however, on the sustained engagement of the research community.