Reporting on sustainable operations by third sector organizations: A signalling approach

Abstract Awareness of (un)sustainable operations is increasingly researched in the for-profit and government sectors, but little is known about the third sector. Still, these not-for-profit organizations are challenged by progressive accountability requirements and increasingly they seem to be responding to these demands through sustainability reporting. In this paper, we explore sustainability reporting by third sector organizations (TSOs) in the context of signalling theory; a useful theoretical lens to explore organizational reports as a signal to stakeholders. Using a document analysis, we explore twenty-three TSOs to discover what they are signalling, to whom, and whether different organizations send different messages.


INTRODUCTION
After decades of being the demanders of accountability to major companies, third sector organizations (TSOs), particularly non-governmental organizations (NGOs), have been increasingly challenged in terms of accountability (e.g. Goddard and Assad, 2006;Jepson, 2005;O'Dwyer and Unerman, 2008;Steffek and Hahn, 2010;O'Dwyer, 2006, 2010). Third sector organizations generally comprise of a set of organizations in between the market, the state and the community (Pestoff, 1998); including a wide range of non-profits, associations, NGOs, mutual benefit societies and some foundations, among others. In order to strengthen accountability, they need to 'commit to the same accountability standards they frequently demand from others'. 1 Accountability is a multi-dimensional concept, often regarded as good governance or virtuous behaviour. It is 'a relationship between an actor and a forum, in which the actor has an obligation to explain and to justify his or her conduct, the forum can pose questions and pass judgment, and the actor may face consequences' (Bovens, 2007: 450). In the case of TSOs, it is essential to recognize to whom are they accountable, i.e. which forums are the TSOs obligated to explain and justify their conduct to? O'Dwyer (2006, 2010) distinguish three main types of accountability approaches for NGOs that can easily be transposed to the TSOs in general. These are upward accountability, where the organizations are believed to be accountable to those who provide funds; downward accountability, to their beneficiaries and not only to those who provide funds; and finally, holistic accountability. Holistic accountability combines both upward and downward accountability 'ideally with the addition of multidirectional dimensions of accountability' (Unerman and O'Dwyer, 2010: 482) suggesting that organizations should be accountable to those upon whom they impact (Unerman and O'Dwyer, 2006). As Ebrahim (2005) alerts to accountability myopia, accountability should not be only towards donors, but instead, to a larger set of stakeholders. Indeed, TSOs face growing pressure to show their reliability as sustainable organizations, not only in terms of survival to continue serving their constituents (Weerawardena et al., 2010), but there seems to be an increasing pressure in terms of sustainability from environmental and social perspectives as well.
This idea of a broader spectrum of stakeholders to whom the organization is accountable resembles the notion of Corporate Social Responsibility (CSR) which encompasses the 'economic, legal, ethical, and discretionary expectations that society has of organizations at a given point in time' (Carroll, 1979: 500). Global CSR norms are relevant both within and beyond organizational boundaries, involving suppliers, partners and other stakeholders. They cross the private, public and non-profit sectors, and organizations are increasingly changing their daily operations to comply with the principles of the triple bottom line: Profit, People and Planet (Elkington, 1999). The topic of sustainable operations and supply chains is becoming mainstream literature (Corbett and Klassen, 2006;Corbett and Kleindorfer, 2003;Pagell and Wu, 2009). Internal (sustainable) operations include not only the daily operations performed by organizations, but also the procurement of new products and services to enable the organizations' activities. Similarly, the literature on sustainable procurement, which integrates environmental and social criteria next to the economic criteria, is growing and so is the number of scopes, labels and definitions in this area (see Carter, 2005;Seuring and Müller, 2008;Vachon and Klassen, 2006;Walker et al., 2008). For TSOs, sourcing of products, materials and services can be closely related to their missions (e.g. material for providing service or campaigning) or to internal operations (e.g. water, energy or office materials). Internal operations require procurement of services and materials and so have a major influence on specifications and quantities of purchases. Still, only a few examples in the literature include cases of non-profit organizations (see, e.g. Low and Davenport, 2009;Meehan and Bryde, 2011;Wiser et al., 2001;Zuo et al., 2009).
Although from a holistic accountability perspective, TSOs are accountable to a broad spectrum of stakeholders O'Dwyer, 2006, 2010); there is an information asymmetry between the organizations and their stakeholders in terms of organizational conduct. As it is put by signalling theory (Connelly et al., 2011;Spence, 1973), TSOs may decide to 'signal' their behaviour to the external world because of this information asymmetry, often through sustainability reporting. Literature on sustainability accounting and reporting has followed different theoretical paths throughout the years (Burritt and Schaltegger, 2010) and there has been an increase of empirical studies on sustainability reporting of companies (e.g. Branco andRodrigues, 2006, 2008;Kolk, 2003;Tate et al., 2010). Nevertheless, again there has been scant attention on sustainability reporting by TSOs (some exceptions include: Crespy and Miller, 2011;Mussari and Monfardini, 2010). Still, considering the up going trend 2 of sustainability reporting by TSOs, this becomes an increasingly important phenomenon to understand. The new NGO supplement 3 of the Global Reporting Initiative (GRI), which is tailor-made for not-for-profits, is an example of the reporting and accountability developments for TSOs, making GRI a customized framework to analyse how TSOs signal their sustainable practices to their stakeholders. This paper addresses these gaps in the literature by exploring sustainability reporting of TSOs through the lens of signalling theory and accountability perspectives. The purpose is to understand the signalling process behind the sustainability reporting by TSOs, exploring not only the signal itself, but also to whom it is being sent and whether the characteristics of the signaller play a role in the message. Our main contribution is twofold: (1) the exploration of sustainability reporting practices by TSOs, specifically in terms of operations and procurement practices and policies that go beyond traditional reporting standards; and (2) the combination of signalling theory with accountability perspectives in the specific context of TSOs. In the next section, we introduce signalling theory as a useful theoretical framework to understand how TSOs are transmitting their message about sustainable conduct, positing three main research questions which are explored using a documentary analysis of public reports from TSOs.

SIGNALLING THEORY AND SUSTAINABLE REPORTING BY TSOS
Signalling theory has its origins in the early work by Spence (1973) who studied job market signalling. Since then, management scholars have increasingly applied signalling theory 'to help explain the influence of information asymmetry in a wide array of research contexts' (Connelly et al., 2011: 40). This theory has mostly been applied to the contexts of finance (e.g. Certo, 2003;Gao et al., 2008), marketing (e.g. Basuroy et al., 2006;Kirmani and Rao, 2000), strategy (e.g. Campbell et al., 2001) and human resources (e.g. Spence, 1973;Suazo et al., 2009) disciplines. Few examples can be found of applications in the context of TSOs (e.g. Gugerty, 2009).
In general, the signalling is used to distinguish 'quality', which has been interpreted in several ways depending on the context of research. Following Connelly and colleagues (2011: 43), quality 'refers to the underlying, unobservable ability of the signaler to fulfill the needs or demand of an outsider observing the signal'. In our particular case, the information asymmetry arises from the fact that TSOs are aware of their sustainable conduct but external stakeholders are not. Hence, TSOs may choose to signal or not their behaviour to the external world as a response to accountability expectations of stakeholders or, for instance, to distinguish themselves from other TSOs; giving them increased legitimacy compared to their counterparts (analogous to Gugerty, 2009).
Signalling theory assumes the existence of three main elements in the signalling process. These are the signaller, the signal itself and the receiver of the signal (Connelly et al., 2011;Spence, 1973).
The signaller tends to be an insider who has some private information that provides him a 'privileged perspective regarding the underlying quality of some aspect of the individual, product, or organization' (Connelly et al., 2011: 44). Different signallers have been identified in the literature, ranging from individuals (e.g. Spence, 1973) to organizations (e.g. Gugerty, 2009). In the context of our research, TSOs are the signallers and can be classified in terms of their approach as: (1) 'campaigning', such as campaigning organizations and political parties; (2) 'mutual support', including trade unions, professional associations and cooperatives or (3) 'service-providing' organizations, which comprise of mostly voluntary organizations, housing associations and arts organizations (Handy, 1990;Hudson, 2009). The privileged information from the TSO in terms of the sustainable behaviour of the organization entitles it to decide whether to signal or not to the external world about its own behaviour.
The second element is the signal itself, i.e. what is communicated to the outsiders. Signals can include the education of a job applicant to signal his qualities (Spence, 1973), marketing variables to signal unobservable product quality (Kirmani and Rao, 2000) or a corporate strategy communication to influence financial analysts and public investors (Gao et al., 2008), just to name a few. Although the theory focuses primarily on the 'deliberate communication of positive information in an effort to convey positive organizational attributes', it is possible that even unintended, negative information may be signalled as well (Connelly et al., 2011: 44). In the context of our research, sustainability reporting is regarded as the signal that TSOs use to communicate their sustainable conduct to their stakeholders. Considering this, the first exploratory question is: RQ1: What are TSOs signalling in their sustainability reporting?
The third element is the receiver of the signal, like the 'outsiders who lack information about the organization in question but would like to receive this information' (Connelly et al., 2011: 45). Considering the notion of accountability and the idea that it involves the obligation to explain and justify organizational conduct to whom they are accountable (Bovens, 2007), signals may be sent to different stakeholders. This can be seen from the three accountability perspectives of TSOsupwards to funders, downwards to clients and holistically to a wider range of stakeholders O'Dwyer, 2006, 2010). A combination of signalling theory with accountability perspectives suggests that sustainability reporting may be used as a signal sent to multiple stakeholders, in line with a holistic perspective.
As noted earlier, depending on the judgment by whom the organization is held accountable to, the organization may face consequences (Bovens, 2007). This accountability notion is in line with the 'strategic effect of signalling', i.e. 'for signaling to take place, the signaler should benefit by some action from the receiver that the receiver would not otherwise have done (. . .); this usually involves the selection of the signaler in favor of some alternatives' (Connelly et al., 2011: 45). In the specific case of TSOs, this choice may refer to enhanced legitimacy of the organization or donations or any other type of support that otherwise would be provided to another organization. As noted by Connelly and colleagues (2011), a key point in signalling is that the receiver would somehow benefit from receiving the signal. In the case of sustainability reporting by TSOs, the reward may involve lower opportunity costs of supporting the signalling TSO rather than another organization. In case stakeholders do care for the sustainable conduct of the TSOs, they may benefit from being aware as this allows them to choose an organization that works in line with their principles. Our second research question addresses those intended receivers of sustainable reporting: RQ2: To whom are TSOs signalling in their sustainability reporting?
Considering the potential interplay among these three elements of signalling theory when combined with accountability perspectives of TSOs, characteristics of the signaller are worth exploring as well. First of all, it has been argued that different activities of international NGOs require different structures of accountability (Brown and Moore, 2001). Considering this, one can wonder if what TSOs decide to signal in their sustainability reports may vary according to the type of mission they have, i.e., whether they are service-providing, campaigning or mutual support organizations.
Besides the type of TSO, the size of the organization can play a role as well. Buying firms with larger purchasing volumes are more likely to be involved in green purchasing than smaller firms (Min and Galle, 2001). This might apply to TSOs as well depending on their income level. This is particularly important if we consider that despite the potential incentives to signal through sustainability reporting, it involves 'signalling costs' (Spence, 1973). In this research, costs include the ones associated with the good practices themselves and investments in systems that allow for data collection and reporting. Hence, income level may play a role as some organizations are in better position than others to absorb the signalling costs (Connelly et al., 2011).
Besides the organizational factors, the national context of the organization may influence their participation in global CSR frameworks -such as GRI -as predicted for companies and governments (Lim and Tsutsui, 2012). The country of origin may play a role since greater legitimacy of CSR principles at the national level, governmental transparency and the level of rationalized business environments are predictors of higher participation in global CSR frameworks (Lim and Tsutsui, 2012). In this case, we can explore these variables in the signalling process of TSOs. Hence, the following question is worth exploring: RQ3: Does what is signalled vary according to who signals?

METHODOLOGY
The empirical part of this paper is based on exploratory content analysis of secondary data with purposeful sampling, set up to investigate existing reporting practices in terms of internal operations. Theoretical sampling was used to facilitate theoretical generalization (Eisenhardt, 1989;Miles and Huberman, 1994). Based on the theoretical framework, we selected organizations that reported according to sustainability reporting standards, since they can be considered as exemplars which are most likely to provide insights by reporting on sustainable practices. Considering the exploratory nature of the paper, this is not a concern. Also, such reporting standards allow 'crosscompany and cross-industry comparisons based on its common reporting framework' (Waddock, 2008: 93), enhancing the chances that uniformed coding procedures can be applied to the whole sample. Still, for the aim of this paper, these guidelines only partially covered operations, especially in terms of procurement and an additional analysis of the reports was carried out.
From the list of all organizations that report using the GRI guidelines, we have selected the organizations in the non-profit/services category of the GRI listing of reports 4 published in 2009. From the initial list of thirty-two, we retained all retrievable reports of TSOs (and not public organizations or another type), published in Dutch, English, German, Portuguese and Spanish. The final sample consists of twenty-three organizations. Similar to Pollitt and Summa (1997), our analysis is confined to documents and we do not collect primary data, such as interviews or observations; we rather explore on the basis of publicly available reports which are there to provide transparency and signal the message from TSOs.
For each TSO, attributes were defined at the organizational level, covering the following areas: Classification by approach (Campaigning -campX, Mutual Support -supX, Service-providing -servX); Region (Europe; Asia; Africa; Latin America; Oceania); Income and GRI Reporting level, status and year. Also, following Lim and Tsutsui (2012), some country level codes were used such as Development level (based on the OECD), Democracy (based on the Polity IV project 2011 5 ), Transparency level (based on the Transparency International 6 ) and Secondary School (based on the World bank 7 ). Secondary school enrolments, the most common measure of the general level of education, are used to assess the capacity of citizens to understand CSR principles (Lim and Tsutsui, 2012).
In order to explore what is being signalled, for each TSO, we first collected data on the GRI Performance Indicators (PIs) from the G3.1 Content Index 8 that referred to Environmental and Social items, including labour practices and decent work, human rights, and society. Then, we checked the GRI grid in each report, which shows whether or not the organization reported on the PI, rather than the content reported. After collecting this information, we averaged the number of TSOs that had reported on each of the four dimensions and respective sub-dimensions (e.g. environmental dimension has nine sub-dimensions -materials, energy, water, etc. -but each of these has one or more PIs).
Besides this standard approach from GRI guidelines that provides a generic overview of the dimensions more often reported, we went into more detail in respect to internal operations and purchased materials or services. Hence, in order to explore sustainability reporting, the following codes were also analysed: Sustainable operations/procurement guidelines in place or planned; Report on internal operations (Water; Energy, batteries; Paper, ink, office materials; Travel, gas, petroleum; Infra-structure, buildings); Depth of report (Use; Evolution; Target); Procurement concerns (Environmental; Social: labour practice and decent work; Social: human rights; and Social: society; Other) and Procedures related to pollution (CO 2 emissions; Waste; Recycling/Reuse). Here we used a binary type of coding, but detailed information on these contents was also collected. These codes emerged from the analysis considering that they provided a framework of analysis of what TSOs are signalling specifically in terms of their operations and procurement.
The last category of codes was related to the intended receivers of the reports. To retrieve this information, we looked for stakeholders specifically addressed in the introduction and initial pages where the report is presented to the audience, as well as in the stakeholder section of the reports whenever that was the case.
Finally, the coding of the reports was double-checked by both authors whenever these were in English (ten); in the remaining cases they were checked twice by one author only due to language restrictions. Still, considering the discussion and agreement on the English reports, we believe this is not a major concern. Data were coded using the qualitative analysis software MAXqda; enhancing the potential to explore the analysis process.

Characterization of the sample
The TSOs differ in terms of organizational approach, income and GRI reporting, and are located in different regions of the world (Table 1). Considering the organizational approach followed, there are nine service-providing, nine mutual supporting and five campaigning organizations. Whenever more than one approach would fit the organizational mission, we considered the dominant approach. The income of these TSOs is highly varied, ranging from amounts until five million euros (seven), between five and fifty million euros (eleven) and above fifty million euros (five). In terms of GRI reporting, the dominant level is the lowest one, i.e. the C level which implies reporting on a minimum of ten PIs, including at least one from each of: Economic, Social and Environmental. Reports were mostly self-reported (seventeen), although in the remaining cases they were either checked by a third party (two) or GRI (three) and by both (one). Finally, most of the sample is composed of European organizations (twelve), followed by Latin America (five) and Oceania (four).
Considering the country level, TSOs originate mainly from developed countries (seventeen) and all of them are democracies. In terms of transparency, there is a clear division among developed and other countries, except for Chile that has a high level of transparency, although not being considered a developed country. In terms of secondary education, all the countries present high gross enrolment rates except for Costa Rica. Considering these results, for the purpose of further analysis, only the development and transparency levels will be considered.
What are TSOs signalling in their sustainability reporting?
According to the reports, there are different levels of reporting by the TSOs for the various PIs of the GRI guidelines in terms of Environmental and Social items. Each of these items has multiple PIs and for each, we have identified whether or not the TSOs reported on them (reading directly from the grid in the report). Considering the data as a whole, the most commonly reported indicators are related to social issues including labour practices and decent work, followed by society, and then environmental issues. The conscience that organizations have to account for sustainability issues in their operations is present in statements such as these: Regardless of the type of work we do, if we Although important to understand the signals, the GRI grid analysis gives an idea of the extent to which the organizations were able to fill in the form but does not say much about the content nor necessarily the importance of each dimension to the organization. Hence, regarding the information reported, a more-detailed content analysis was developed to explore reporting practices in terms of operations and procurement that go beyond the GRI checklist. Table 2 provides a cross-code analysis that can enhance our understanding of the signals.
First of all, we have considered internal operations and found that all the organizations reported some type of sustainable guidelines in place or in development (ranging from informal to formal with a policy or a department). Mostly, reporting is related to consumption of resources such as energy (twenty-two), travel (eighteen), paper (seventeen) or water (fifteen). Considering the different spending categories or the procedures related to pollution, we looked for three levels of depth of report. Hence, we checked the extent to which the organizations reported only the use for the reported year or if an evolution was presented in figures or at least referred to as an increase/decrease. As a sign of stronger commitment, we looked for organizations that established targets for those items. Considering that in some cases, the organizations presented different reporting depth for the different categories; the same organization could have been coded as having reported use, evolution and target. The reports referred mainly to the use of resources in the reported year (eleven) or some evolution from the past (sixteen) but to a lesser extent there were targets established (nine). Still, we could see that among those that reported evolution and not only use, they did so, sometimes, for only one or two categories (four), or up to four or five categories (nine). This reveals that even when monitoring systems are in place for at least one year, the basic reporting is normally related to energy and water consumption. In most of the cases, where evolution was presented there was a positive progress, but there were also cases where a negative progression was presented. In any case, some organizations provided explanations for the figures, such as changes in the number of trips, activities developed and other operational needs. Finally, among the nine organizations that reported targets in at least one category, eight of them also reported evolution suggesting a relationship between defining targets and being aware of the evolution. Yet, again the level of detail in the targets is varied, since some organizations quantify the targets, whereas others only give a qualitative indication (e.g. green energy is a target or indicating the aim of minimizing the use of a certain resource). Secondly, disclosure on procurement was used to examine what is reported on the sustainability impact of the TSO in the supply chain, but here the extent of reporting is more limited. There were fourteen organizations that reported either formal or informal guidelines followed or at least in development, in terms of procurement practices. These range from references to informal purchasing criteria to formal procurement policies. Some excerpts illustrate the increasing importance devoted to sustainable procurement issues: When we spend money on our suppliers, we have two considerations in mind. We want value for money, and we want our suppliers to meet basic ethical standards. We have therefore been reviewing our procurement policies and procedures, giving particular attention to sustainable procurement principles, including ethical, environmental and economic considerations. (Camp3: 65); and In our purchasing of goods and services, we have to balance cost-effectiveness with our responsibility to promote sustainable livelihoods and to minimise any negative impacts in our supply chain. This involves influencing companies to source their products ethically, promoting Fair Trade, and campaigning on labour rights and climate change (Serv6: 17) Other TSOs also recognize that their behaviour as a client might influence others in the supply chain: 'Our Policy of Sustainable procurement, to be implemented soon, will formalize our search to influence positively through our choices' (Camp1: 28); and The reason 'Camp2' chose to focus on the development of a procurement policy is that procurement is a unique opportunity for any organization to promote concrete change on a broader scale, to make a real difference and scale up the positive effects of responsible business outside of its own operations for the benefit society as a whole (Camp2: 30) Contrarily to the results in general operations obtained through the GRI grid where some social dimensions are dominant; for procurement, the environment is the major area of concern. Although in some cases procurement guidelines were not mentioned explicitly, a generalized concern about reducing consumption of items such as water, electricity and travelling was reported. These organizations are concerned about future procurement volumes and its environmental impact. The social dimensions are much less commonly mentioned in relation to procurement and suppliers and in some cases, organizations mentioned that considering the nature of their activities and where they were located, issues like human rights were not a concern. A note should be made to the fact that the organizations that revealed some concerns about certain social issues, mostly had or at least planned to have sustainable procurement practices or guidelines. For instance, 'Serv4' is aware that the purchasing decisions that it makes can have human rights implications for people who produce goods. We consider these issues and choose to buy fair trade items in some cases (. . .)'Serv4' will continue to evaluate the purchasing decisions that it makes to become more aware of its human rights impact, and to change purchasing behaviours where possible (Serv4: 12).
Several procedures related to the reduction of pollution were identified, including data on CO 2 emissions, waste and recycling (and reuse). CO 2 emissions was one of the most common and detailed issues in the reports (nineteen). Often, we could find references about the calculations of emissions and ways to account for them. In general, these reports were accompanied by the energy and travel information. Procedures to deal with waste (fifteen) and recycling of materials (fourteen) such as paper, plastic or glass were also common.
Finally, some examples of sustainable procurement traced in the reports included the acquisition of recycled paper, fair trade products, products with eco-certification or green power, to name a few. Other measures in place to prevent future procurement included energy-and water-saving equipment, investment in more efficient appliances, e-books, online seminars and similar procedures to reduce the demand. In terms of selecting suppliers, some TSOs opted to buy from travel companies offering a service to offset carbon emissions or preferential purchases of products made by disabled people and small and medium enterprises. Statements like these reinforce this idea: We will not discriminate against local and smaller suppliers in our procurement process and give them preferential treatment wherever possible (Camp3: 65).
To whom are TSOs signalling in their sustainability reporting?
The targets of the reports are sometimes specifically expressed in the reports, as these examples show: 'in reporting to our key stakeholders, particularly our members on our CSR' (Sup6); or 'this report is presented with the aim of informing our partners, funders, general public and other stakeholders' (Sup7). Also, some recognized that different groups of stakeholders have different information needs from the report (Sup8) and even had an online version in addition to the report, where readers could access more-detailed information in function with the different profiles of stakeholders (Sup4). Although such discourse is not always present, these examples reinforce the idea that these reports serve as signals to certain intended receivers.
Considering that we focus on intended receivers, these results apply only to those cases where more or less the organization explicitly addresses the report to specific stakeholders. Hence, among the stakeholders, most commonly addressed are employees or staff (65%), followed by donors, sponsors, funders or supporters (48%), partners (48%), customers, users or clients (39%), members or affiliates (35%), and communities and society in general (26%). The remaining stakeholders such as suppliers, volunteers, governments appeared in the form of intended receivers in less than 25% of the reports.
Also, it is worth noting two aspects. On the one hand, all these reports were online, increasing the number of potential readers of these reports. On the other hand, the fact that the reports are mostly in the native language of the country of origin may limit the access to stakeholders outside the country (e.g. as is the case of TSOs that act at the international level). Twenty reports are in an official language of the country, even if it is not the most spoken language as is the case of English in South Africa. Only in the remaining cases, the reports are in English, although it is not an official language of the country (Costa Rica, The Netherlands and the Republic of Korea).
Does what is signalled vary according to who signals?
In order to address this question and considering the limited sample, we have explored potential patterns in terms of both organizational and country level variables. Since there are no extreme patterns, we can just report what our sample reveals.
Starting with the GRI guidelines (Table 3), we found that in terms of organizational approach, campaigning TSOs in the sample tended to disclose more information on the various categories of environmental and social reporting. Considering the regions from where TSOs in our sample originate, there seems to be slightly more focus on environmental disclosure in Europe and Latin America, whereas human rights and society issues were slightly more commonly reported by the TSOs from Latin America and Oceania. Labour practices and decent work are generally reported in the three regions. The income level segmentation of TSOs is somehow intriguing. As expected, higher income seems to be linked to more reporting; the mid-income level is, however, the one that presents lower levels of reporting in the various categories. Next to this, there are country level variables. Considering the development level, the major difference is the reporting on social issues such as human rights and society which is predominant in the 'other' countries. In terms of transparency, there are no major differences. Still, reporting on corruption is almost double in the lower transparency level countries.
Regarding the other categories that we have explored on internal operations and procurement, there seems to be no clear pattern in terms of organizational or country differences in the sample. In terms of operations, all the organizations had sustainable guidelines in place or planned and the existence of sustainable procurement guidelines was similar in the three types of organizational approaches found in the sample. Also, these were more common in both Oceania and Europe, when compared to the organizations in Latin America. Higher income TSOs, originating from developed countries with higher transparency levels tended to have more often sustainable procurement practices in place or planned.
There are differences in the categories reported more often by these organizations, although again there is no clear pattern. For instance, water seems to be reported more often among service-providers, originating from Latin America, and in countries with lower development and transparency levels. When it comes to office materials, travelling, gas and petroleum, reporting is more common among the TSOs coming from developed and higher transparency level countries. In terms of the depth of report Table 3: GRI reporting on sustainability by organizational and country variables of internal operations, the organizations in the sample that tended to report more than just the use were mostly service-providers, European organizations and organizations with higher income levels. Finally, procedures related to pollution are common among the various segments analysed. In general, higher income TSOs in the sample reported more procedures on waste, recycling and reuse.

DISCUSSION AND CONCLUSION
These TSOs are disclosing information on sustainability of their operations which constitutes a first step in terms of transparency towards those whom they are accountable to. The first signal being sent is that they do care to the extent that they report on these issues (and the number of signallers is increasing), but per se having a report says little about how good their conduct is in terms of sustainability and how they are improving it. Going into detail on the signals being sent by these organizations enhances insights in current reporting developments. Despite the small sample, understanding contents of reports offers first insights in those recent developments.
The first research question explores the signals. Despite the standardized reporting framework, there are different levels of reporting and styles in how information is displayed and more importantly depth of information provided. The analysis of the GRI grid reveals that more frequently social issues related to labour practices are being reported in our sample, rather than the environmental dimension or other social dimensions. In general, all organizations do have some type of guidelines when it comes to their operations, whether they are formal policies or informal approaches. Contrarily to internal operations in general, fewer organizations do report specifically about procurement guidelines, and the role of environmental issues is much stronger compared to the social dimensions. Organizations reveal an effort to reduce consumption of supplies and this seems to happen even when they do not have any type of approach to sustainable procurement. Moreover, a tendency seems to be in place from merely reducing consumption of resources (demand management) to also including guidelines on procurement (e.g. supplier selection criteria).
We argue that these results may have different grounds that are worth exploring in future research. It may be the case that organizations may not feel the need to signal all aspects of their sustainable conduct to their stakeholders because: (1) they hardly see the environmental and human rights impacts of their operations, as they are nonproduction nor labour intensive industries; (2) stakeholders have specific expectations which are influencing the message being sent as the GRI framework encourages the organizations to consult their stakeholders during the process (e.g. Camp5 reduced the environmental section and reinforced the social section); (3) data on labour practices are easier to acquire compared to environmental related data and the need to balance signalling costs with the potential benefits can explain the importance of data-convenience; (4) GRI guidelines influence what is being reported and, for instance, currently there is no specific section about procurement or (5) organizations choose not to report on certain items. Organizations sometimes used materiality tests to justify why they would or would not report on certain issues. Nevertheless, this was not always the case and the analysis could not take this into account, indicating only weighted indicator frequencies for each of the four dimensions. 9 In fact, the absence of reporting on certain issues may itself be a signal. Third sector organizations may consciously decide not to report on some items because they are simply not relevant enough to be mentioned or because there are 'harmful' practices that they prefer not to report about. Although this may happen and it is important to distinguish the rhetoric behaviour from the substantive action (Campbell, 2007), there are cases where the TSOs recognize that certain parameters on sustainable operations had a negative evolution often linked with an explanation. Either way, the absence of a message may produce misleading signals to the stakeholders. The fact that only a minority of the reports is audited by a third party or by the GRI itself may facilitate the existence of misleading signals. Further research on the signals being sent could include in-depth qualitative research on the drivers and constraints of TSOs to develop sustainable practices and report them.
The second research question explores the receivers of the signals. The idea that messages are being sent to specific receivers has been reinforced by several examples provided, together with the aforementioned stakeholder consultations. Although, the intended receivers of the report are not always explicit, the most explicitly mentioned are staff, donors, partners, clients, members and finally the communities or society. This is in line with the holistic accountability perspective that considers various stakeholders, when compared to the upward accountability perspective which focuses mainly on financial reporting to funders or the downward perspective that focuses on clients. The language in which the reports were written is also noteworthy as most organizations report in the national language. The only three TSOs whose reports were not in an official language of the country -but in English -were two campaigning organizations and a mutual support organization which were related to setting standards. These organizations potentially have a broader international stakeholder spectrum when compared to other organizations working at the national level. From here, we envision two avenues of research. First, accountability literature suggests that organizations have consequences from the signals (Bovens, 2007). Further research could investigate the receivers themselves to understand how they judge the signals and how this brings consequences to the TSOs. Second, voluntary disclosure may differ according to the stakeholders addressed (Campbell et al., 2001). A future study could explore the potential link between the signal and the receivers or even between the different signallers and the receivers of the signal.
The last research question explores a potential link between the signal and the signaller, i.e. whether the message being sent somehow varies according to the organizational and country attributes of the TSOs. Despite the limitations of the data to make these kinds of assertions, some ideas can be drawn from the results. First, the idea that campaigning TSOs in the sample tend to disclose more information on the various categories of environmental and social reporting seems consistent with the anticipated idea that different organizational approaches could have different accountability needs (based on Brown and Moore, 2001). Since, in general, campaigning organizations are among the TSOs that have a stronger watchdog function, they may feel a stronger need to justify their own conduct. Also, as expected, higher income seems to stimulate reporting considering signalling costs.
Finally, in terms of country level variables, literature suggested that national context could have an influence on the participation in global CSR frameworks (Lim and Tsutsui, 2012). In fact, most of the TSOs in the sample were from developed countries with high levels of transparency, particularly all from democratic countries and mostly with high levels in terms of secondary school. What we add to those results is that among those who report following the frameworks, there are potential differences according to the national context as well. Although not generalizable, results indicate that social issues such as human rights and society were predominant in reports of TSOs coming from developing countries and corruption issues in TSOs coming from countries with lower transparency levels. Organizations in these countries may need to signal that they are working on issues that are often taken-for-granted in developed countries.
This paper brings two main contributions to the existing literature that to our best knowledge have not yet been addressed. First, it explores empirical data on sustainable reporting practices of TSOs in a level of detail that goes beyond reporting standards. One major finding is that not only do these TSOs present quite different levels of reporting, but also that the dominant concern -various social issues and environmentdiffers significantly when it comes to daily operations or procurement. Second, it presents an analytic framework to study sustainability reporting in the context of TSOs that combines two streams of the literature: the signalling theory and the accountability literature. By combining these, results suggest that signals may vary according to the type of TSOs or even the country where it comes from. Also, in line with a holistic accountability perspective, sustainability reporting is a signal being sent to various stakeholders that have different accountability needs.
Despite its contributions, this paper reveals some limitations. First of all, it is selfreported data on organizational behaviour; still, this was precisely the signal we intended to study. Secondly, being purposeful and small, the sample has not allowed a mixed method study that would combine exploratory with explanatory research questions. Hence, our results are not intended to be generalized to the whole third sector but instead to reveal the adequacy of signalling theory in understanding the sustainability reporting behaviour of TSOs, considering their accountability requirements. Hopefully, this research and the increasing sustainability reporting trend by TSOs encourage further research.