Theoretical considerations on corporate governance in the third sector in the agency theory perspective

Resumo Apresenta-se neste artigo uma revisão da literatura sobre governança corporativa em Organizações Não Governamentais (ONGs) sob a perspectiva analítica da teoria da agência, em especial alinhando os conceitos ao contexto do terceiro setor brasileiro e à relação de agência na qual o doador (principal) contrata a ONG (agente) para o desenvolvimento de projeto por meio da doação de recursos vinculados. Espera-se, com isso, contribuir para o melhor entendimento da relação entre ONGs e doadores e para o desenvolvimento de futuras pesquisas. Os fundamentos da teoria da agência são muitas vezes abordados de forma incompleta, quando a governança corporativa é sobreposta ao contexto do terceiro setor. Assim, foram extrapolados os pressupostos e conceitos para o ambiente do terceiro setor no Brasil. Como resultados podem-se destacar a proposição de que o pressuposto da incompatibilidade de horizontes de planejamento entre principal (doador) e agente (ONG) ocorre de forma inversa no terceiro setor ao assumido como regra geral para organizações empresariais e o reconhecimento de custos de agência relacionados às perdas residuais, dos quais decorrem problemas de agência de natureza propriedade x gestão também no ambiente do terceiro setor. Assim, a estruturação da governança corporativa no terceiro setor sob a perspectiva da teoria da agência permitiu formular relações que complementam as generalidades da teoria, contribuindo para o melhor entendimento da sua aplicação, bem como para a proposição de uma agenda de pesquisa.


INTRODUCTION
The topic of corporate governance -from this point on, only governance -has been extensively explored in studies on Non-Governmental Organizations (NGOs). To understand the relevance of governance in the context of the third sector, we must reflect on the potential impact on NGOs. The assumption is that some of them are different, due to a "good" governance. These NGOs, coeteris paribus, are more attractive to donors; thus, they receive more funds. In general, the discussion starts from the hypothesis that governance mechanisms affect organizational performance (HASNAN, MOHAMAD, ZAINUDDIN et al., 2016).
Literature review on governance in the third sector shows analyses from different theoretical perspectives, with the prevalence of agency theory (RENZ and ANDERSSON, 2014;MARQUES, RODY, CAMPOS et al., 2015).
Despite the undeniable contribution of previous studies, considerations on the alignment of the foundations of agency theory with the concepts of governance in the third sector are still shallow (YETMAN and YETMAN, 2012;ÁVILA and BERTERO, 2016). In general, these studies marginally address the assumptions of agency theory in the context of governance in the third sector, focusing only on their different functions and usefulness (HARRIS, PETROVITS and YETMAN, 2015).
Although there is abundant literature on governance in the third sector, the understanding of its relationships is incipient. Thus, we sought to fill this gap a little, by aligning the grounds of governance, in the light of agency theory, to the context of the Brazilian third sector, in order to increase the comprehension of the relationship between NGOs and donors.
Research on governance, in the Brazilian third sector environment, requires a cutout that favors the comparison of more homogeneous entities, among all bodies of this sector. This is necessary because it addresses comprehensively civil society's activities that do not fall into the so-called first and second sectors, such as political parties, private non-profit associations and foundations, unions, class councils, religious organizations, etc. (IBGE, 2018). Thus, at an abstract level, we chose as units of analysis private not-for-profit associations and foundations commonly called NGOs, since their specific attributes make them similar to each other and different from the rest (IPEA, 2018). In this context, we extrapolated the assumptions and concepts of agency theory to the Brazilian third sector environment, in order to discuss the existing conflicts between principal and agent, considering the agency relationship observed in hiring a NGO (agent) by the donor (principal), for the transfer of donations.

A Brief Contextualization of the Third Sector in Brazil
The introduction brought a reflection on governance in the third sector and the possibility of offering contributions for its understanding. To contribute to the defined objective, we present an overview of this sector in Brazil, which helps delimiting the object of analysis.
When addressing the third sector, we distinguish initially the three sectors of economic activities of the society: state activities refer to the first sector, represented by Public Administration; market activities, to the second sector, represented by forprofit companies; in turn, the third sector contemplates the activities of the civil society that do not fit in the other sectors. In another perspective, the first sector operates public resources for public purposes; the second sector, private resources for private purposes; and the third sector, public and private resources for public purposes (FERNANDES, 1994).
In 2016, non-profit entities in Brazil represented 9.5% of companies and other organizations, 6.8% of the total employed persons, 7.2% of those who received salaries, and 6.7% of wages and other compensations paid in the year (IBGE, 2018 In the IPEA (2018) survey, based on the five criteria above, several non-profit entities were removed from the list: schools' cash and similar; political parties; unions, federations and confederations; condominiums; registries; the S system; mediation and arbitration bodies; prior conciliation commissions; local councils, funds and consortia; and cemeteries and funeral homes. In 2016 there were 820,000 CSOs registered in the National Register of Legal Entities, of the Federal Revenue Service.
Third sector entities in Brazil originated from the ecclesial base communities, born in the 1960s, and from the bodies that emerged within the scope of the international system of cooperation for development, as of the 1970s, both with roots in a philanthropic tradition (FERNANDES, 1997).
In the 1990s, the New Public Administration (NAP) agenda strongly affected the conception of third sector entities, by promoting a paradigm shift in philanthropy (DANGINO, 2011). The traditional approach of almost granting social services to non-profit entities, in a non-competitive environment, gave way to an environment where donors do not assign contracts to third sector entities for what they are, but for what they can deliver (RYAN, 1999). This enhanced the external responsibility of third sector bodies, which increasingly were "compelled" to show performance in terms of efficiency and effectiveness (SILVA, 2010), and to introduce management models for this purpose (SMITH, 2010), especially to deal with the restrictions of the donation market (ALDASHEV and VERDIER, 2010).
In short, NAP fostered competition and pressure to "professionalize" management methods through the adoption of business principles from for-profit organizations (EIKENBERRY and KLUVER, 2004), since donations are based on donors' own values, and results are measured by their standards (POZORSKI, 2000). Therefore, NGOs started to adopt practices and procedures prevalent in organizations that constitute the donation market, that is, public and private companies, and business institutes. As NAP's agenda strongly influenced the government sector (GREILING and STÖTZER, 2015), and business institutes are created by companies, market logic prevails.
Hence, donors' pressures in the donation market can explain, to some extent, why non-profit organizations have assumed management models under the influence of the market (SALAMON, 1997), among them governance.
This brief characterization of the third sector in Brazil disclosed a very heterogeneous environment, leading us to an empirical delimitation in order to examine entities that are more homogeneous, such as associations and non-profit private foundations. It also suggests that motivation for the adoption of governance practices by third sector bodies stems, in part, from partnerships with private and state companies, business institutes, and other entities of the third sector (arising from the fundraising process), because they trigger organizational arrangements to coordinate activities that result from the partnerships. In addition, the intense competition for public and private resources forces the professionalization of management processes of third sector bodies, among them the adoption of governance practices, in search of higher efficiency, which leads them to migrate from a philanthropic model to a corporate model.

Agency Theory
The influence of agency theory has historical reasons, involving scandals of mismanagement in large multinational companies, which led, in 1992, in the United Kingdom, to the Cadbury Commitee report, on the causes of several events related to the opportunistic management of companies.
The explanatory basis of agency theory assumed in this article starts with Jensen and Meckling's (1976) classic approachconsidered seminal in governance studies, despite the relevance of previous studies. After this paper, numerous empirical studies and new theoretical models were developed (EISENHARDT, 1989;GLAESER, 2003).
In this perspective, the firm is defined as a legal fiction that serves as a focus for a complex process where the conflicting objectives of individuals are brought to balance within a framework of contractual relationships. The firm, therefore, represents a way to integrate the conflicting objectives of the various participants efficiently, in a legal contractual context.
In other words, the firm behavior is similar to that of the market, being the result of a complex process of balance (JENSEN and MECKLING, 1976).
The model of human behavior that supports agency theory (resourceful, evaluative, maximizing model) was described by Jensen and Meckling (1994) under four assumptions: individuals (1) are evaluators (i.e., they care about several factors, such as prestige), are willing to make substitutions (trade-offs) and have transitory preferences; (2) they have unlimited needs (i.e., they relentlessly prefer a higher amount of goods that they value positively); (3) they are maximizers (i.e., they seek the greatest possible utility; however, satisfaction is limited by factors such as standards and restricted knowledge, which leads them to reflect on the costs of acquiring information and knowledge necessary to evaluate the alternatives); and (4) they are resourceful (i.e., they are creative and respond to environment changes, adapting themselves and creating new opportunities).
One of the fundamental assumptions of agency theory regards the existence of a conflicting relationship between the objectives of the parties that compose a set of contracts. Agency theory, in short, refers to the relationship between principal and agent, where a person (principal) hires another (agent) to carry out something that involves the delegation of decision-making and authority. This results in agency relationship, defined as: "[...] a contract under which one or more persons (the principal(s) employ another person (agent) to conduct on their behalf a service that involves the delegation of some decision-making power to the agent " (JENSEN and MECKLING, 1976, p. 308).
There are distinct interests between principal and agent, because each one has a different utility function. The basis of agency theory is the assumption that you cannot maximize a utility function that is not yours, because behavior results from the person's own set of preferences and goals. When the agent manages the principal's resources, having as reference the maximization of his utility function and not the principal's, this is an agency problem (JENSEN and MECKLING, 1976). In other words, it stems from the inconsistency between the agent's behavior required by the principal, and the agent's actual behavior.
According to these authors, in addition to the utilitarian and rational human nature, which leads the agent to maximize his utility function, the lack of a complete contract capable of ensuring that the agent aims to serve the principal's interests, also leads to misalignment between the interests of the principal and those of the agent. Fama and Jensen (1983) clarify that contracts also assign to agents the stages of the decision process. To this end, they divide the decision, so that it unfolds into four steps, namely: initiation, ratification, implementation, and monitoring. They explain that initiation and implementation must be combined in the function 'decision management'; and ratification and monitoring, in the function 'decision control, since such sets are, in general, performed by the same persons. For the authors, an efficient control system implies the separation between the initiation and implementation stages (decision management) and ratification and monitoring stages (decision control), because this will reduce or control agency problems.
These authors also suggest the delegation of decision-making functions. They discuss the complex character of organizations, since the specific knowledge necessary to make different decisions is spread among different agents. Therefore, delegation of decision-making functions to agents with more specific knowledge brings potential benefits for better decisions. However, the distribution of decisions to agents creates agency problems, given the rational nature of human behavior and the lack of a perfect contract, assumed by agency theory. Thus, they suggest dividing the management and control functions between different agents, as a mechanism to reduce these problems. Jensen and Meckling (1995) provided a more detailed explanation of the need to decentralize the decision-making process, based on specific and general knowledge. When knowledge is important for the decision-making process, there are advantages in combining decision-making authority and relevant specific knowledge. This combination can occur through the transfer of knowledge or the transfer of decision rights.
Given the impossibility of fully transferring specific knowledge to the decision maker, most decision rights should be delegated to those who have the specific knowledge. This implies developing control systems to reduce agency problems.
For Jensen and Meckling (1995), organizations solve these problems by establishing internal rules of the game that provide a system of division of decision rights for the organization's agents, and a control system that provides measurement and evaluation of performance, as well as one for reward and punishment. Jensen (1983) had previously proposed that contracts between agents and the legal entity should specify the rules within the organization, including these three dimensions: distribution of decision rights among agents, performance evaluation system, and reward system. Therefore, organizational efficiency stems from the combination of these two systems (JENSEN, 1983;JENSEN and MECKLING, 1995). However, it is impossible to eliminate agency problems, and the main one refers to agency costs in order to align the agent's interests with those of the principal. Jensen and Meckling (1976) define agency cost as: the sum of the expenses incurred with the principal's monitoring on the agent's activities; the agent's expenses for showing that his activities are not harmful to the principal; and the residual losses resulting from divergences between the agents' decisions and those that would maximize the principal's welfare, in spite of the monitoring by the principal and the alignment by the agent.
For Jensen and Meckling (1976), agency costs arise in any situation that involves a cooperative effort between two or more people, even if there is no clear relationship between the principal and the agent. Eisenhardt (1989) argues that agency costs occur in several relationships, even informal ones. The contract (by which the principal authorizes the agent to act on his behalf) is a metaphorical illustration of agency relationship, since it is present in both explicit and implicit agreements.
As a solution to relieve agency problems, Jensen and Meckling (1976) suggest audits, formal and informal monitoring and control systems, budget restrictions, compensation policies and incentive systems, contractual restrictions, etc. All these mechanisms involve agency costs. Fama (1980) observes that the capital market and the job market for high-level executives play a role of control mechanism against the opportunistic actions of agents. The degree of divergence in top executives' behavior, regarding the behavior of maximizing organizations' value, is limited by the existence of a job market for executives, which reflects their reputation (human capital), having as proxy the success or failure of the company where they work; and by a capital market that assigns a price to the company's negotiable rights. Jensen (1993) expands this argument by suggesting four control forces that operate in the organization that mitigate agency problems. In addition to internal mechanisms, which he calls internal control exercised by the Board of Directors, there are external mechanisms (linked to factors external to the organization): capital market; legal, political, and regulatory system; and the market for products and factors -which comprises the job market for top executives, as proposed by Fama (1980). In addition, ownership structure is a relevant factor for facing the main conflicts of interest and agency problems. In situations of dispersed ownership structure, the conflict of interest occurs between individual shareholders and executives; and in concentrated structures, between majority shareholders (who have control) and minority shareholders (dispersed). In the first case, hypothetically, dispersed shareholders have less incentive for monitoring and executives have more access to business information (information asymmetry), which allows them to make decisions in their favor, to the detriment of shareholders. In the second case, majority shareholders, with some ownership concentration (voting rights), achieve private benefits at the expense of the other investors. We assume coeteris paribus for both hypotheses.
There is empirical evidence to support these hypotheses. Shleifer and Vishny (1997) show that the presence of controlling shareholders increases the cost-benefit of monitoring, as they are both interested in maximizing value and in the voting power, in order to have their interests respected; and Bebchuk (1999), who showed a strong relationship between ownership structure, governance standards, and benefits of private control. Coffee's (2005) study on the influence of share concentration on the type of fraud illustrates this distinction. In business environments with a fragmented ownership structure, like Anglo-Saxon countries, the main problem is the manipulation of financial results by executives (due to information asymmetry), for increasing their personal compensation. On the other hand, in environments with a concentrated ownership structure, as Japan, Germany and emerging countries, the main problem is the achievement of private control benefits by majority shareholders. Thus, we can conclude that if the type of fraud is different in distinct ownership structures, agency problems tend to be different.
Theoretical considerations on corporate governance in the third sector in the agency theory perspective

Adonai José Lacruz
Agency problem is widespread, since inducing an agent to behave as if he were maximizing the principal's welfare is a common issue in all organizations and in all cooperative efforts (GLAESER, 2003).

Corporate Governance in the Third Sector
When adopting the concepts of governance to the third sector, it is important to consider some distinctions and similarities. Thus, we assume in this article that the donor "hires" the NGO to develop projects of mutual interest, but the NGO may not operate in the best interest of the donor. Therefore, in the contract the donor delegates to the NGO a certain level of authority, to make decisions for the execution of project activities.
In this relationship, some concerns arise from agency conflicts, especially regarding that the actions of NGO's managers are aligned with the interests of the donor, given that both parties, in the principal-agent relationship, seek to maximize their utility function (JENSEN and MECKLING, 1976). Hence, it is necessary to create mechanisms that mitigate agency problems, since they overload agency costs, to a lesser or greater extent.

Assumptions underlying agency theory
Agency theory emphasizes that, by establishing a contract, agent and principal act according to rational behavior and motivated by their own interests. However, it does not rely exclusively on differences in motivation and goals between principal and agent, but also on assumptions of information asymmetry, distinct aversion levels to risk, and different planning horizons of the parties, which imply limited rationality of the economic agents (EISENHARDT, 1989).
Any NGO has information on requirements, restrictions, and results of the activities related to the project for which it received donation funds, which the donor may not know of (LACRUZ, MOURA and ROSA, 2019), and this configures an information asymmetry. This asymmetry creates conditions for opportunistic action, whether by recording inaccurate, incomplete, biased information, even involuntarily, or by fraud. Opportunism, in agency theory, can happen ex-ante (adverse selection) or ex-post (moral hazard) the time of hiring (WILLIAMSON, 1987). This is valid for third sector entities, which have ex-ante costs (structuring costs, safeguards, etc.) and ex-post costs (renegotiations, contractual amendments, monitoring compliance with agreements, assessment of partial accountability, etc.).
In addition, donors can be more risk tolerant than NGOs, since they can further diversify their donation portfolio, being indifferent to non-systematic risk. In turn, NGOs are expected to be risk averse, as they have more difficulty to diversify their project portfolio, and because receiving funds in donation and their credibility are linked to their contract performance -similar to the company's success/failure as a proxy for the reputation of top executives (FAMA, 1980). Thus, the agency relationship parties have different risk tolerance.
Finally, we also accept the assumption of incompatibilities in the planning horizon, but inversely, in the third sector, to that assumed as a rule for profit-making companies. We argue that transience does not regard the agent, but the principal.
The donor puts efforts in actions whose effects take place during the period of the donation contract, while the NGO has a permanent relationship with the firm, whose contract objects (i.e., the project) are sometimes related to long-term goals. In other words, donors are more concerned with the results of the firm during the term of the contract, while the NGO is more concerned with the results of its portfolio of present and future projects. Thus, the NGO tends to plan projects according to horizons that go beyond the limits of the donation contract, especially when such projects regard the institution's programs; on the other hand, donors are more interested in planning the project to which they donated (LACRUZ and CUNHA, 2018; LACRUZ, CUNHA, MOURA et al., 2019).
In addition, non-contractual performance evaluation is based on different metrics, due to a distinct planning horizon (LACRUZ, CUNHA, MOURA et al., 2019), as NGO's performance metrics take into account its collection of projects, while those of the donor may have short-term goals' bias. Hence, we assume that the firm has a finite life and that, in the third sector environment, the agent (NGO) considers the firm's value as an infinite series of actions, while the principal (donor) is largely limited to action flows in the interval of the contract.
In this context, projects in the third sector can be seen as temporary organizations (TONDOLO, BITENCOURT and TONDOLO, 2016;LACRUZ, CUNHA, MOURA et al., 2019). Based on the definition by Lundin and Söderholm (1995), where temporary organizations are decoupled from past and future organizational activities, temporality and the consequent decoupling can explain the inverse relationship, in the third sector, of the assumption of incompatibility of planning horizon between principal and agent.
Thus, because of agency relationship, agency problems arise, and so do the resulting agency costs to minimize them, as, for example, the elaboration of more complex contracts and the establishment of more intense monitoring mechanisms. In this chain, governance emerges to mitigate agency problems.

Governance mechanisms
The discussion on governance requires a specific analysis of the organizational nature of the object of analysis. Therefore, it is necessary to first align the concepts of governance, generally associated with business companies, to third sector entities, which in this study are associations and private non-profit foundations.

Incentive System
A first relevant distinction concerns the incentive system as a mechanism for aligning interests of principal and agent. The hypothesis of incentive systems granted to executives is that the form of compensation can induce behavior (FAMA, 1980;JENSEN and MECKLING, 1976). This seems to apply both to the environment of companies and third sector entities. However, there are more advantages for companies, whose incentive contracts can take a variety of forms, such as share distribution, stock options, or other types of variable compensation (FAMA, 1980). On the other hand, possibilities in the third sector are limited (compensation, in any way, linked to surplus would withdraw the constitutional benefit of tax immunities given to NGOs, according to art. 2º, § 3, II, of Act No. 10,101/2000). Jegers (2009) and Van Puyvelde, Caers, Du Bois et al. (2012) argue that this solution is not plausible for the third sector, due to the difficulty of measuring performance. For Mendonça and Machado (2004), there is no alignment through monetary benefits in the third sector, since there is no expectation from executives of this type of return, as their professional acting is motivated by their adherence to the cause, which Rose-Ackerman (1987) had already observed. Handy and Katz (1998) suggest that lower wages can generate a self-selection among non-profit managers, since this will attract professionals who are more committed to the organization's cause than to the salary. Jegers (2008) explained this phenomenon, based on the trade-off between compensation (financial health) and preferences (own preferences): "[...] the uselessness of a lower salary is compensated by the utility effect of doing something altruistic" (JEGERS, 2008, p. 45).
There is some empirical support for incentive systems in NGOs (BALSAM and HARRIS, 2014), which indicates that most organizational performance criteria are not statistically significant when related to managers' remuneration. However, there are biases that limit the generalization of results, since they are supported by cross-sectional analyses, while a longitudinal approach would be more appropriate for implying relationships between performance change and compensation change (JEGERS, 2009). In addition, they do not always consider the possibility of monetary rewards.
On the contrary, some authors argue that, although limited, alignment is possible through a monetary incentive system. The Brazilian legal system prohibits NGOs to distribute profits. However, it does not prevent them to establish a variable compensation. It is valid to assume that NGO's executives have other motivations, but it seems unlikely that monetary issues do not motivate them in current times. The study Salary survey, conducted by the Group of Institutes, Foundations and Companies (GIFE), showed that 75% of the organizations associated to the GIFE Network have a variable compensation program, and the most frequent is the performance bonus (GIFE, 2015). For all these reasons, limitations regarding compensation in the third sector may reduce the importance of incentive systems, but do not exclude it as an instrument of these systems.
Theoretical considerations on corporate governance in the third sector in the agency theory perspective Adonai José Lacruz

Ownership structure
Another important distinction relates to ownership structure, which in the third sector regards the concentration of resources by donors. There are NGOs with a prevalence of resources from dispersed donors, and others with a concentration of resources from a few donors. As in the business environment, in NGOs with a prevalence of resources from dispersed donors, there may be less incentive for donors' monitoring than in entities with a concentration of resources from few donors. In other words, the ownership structure can moderate the incentive to principal's monitoring in NGOs (ANDRÉS-ALONSO, CRUZ and ROMERO-MERINO, 2006;YETMAN and YETMAN, 2012).
Another aspect related to ownership structure concerns property rights: decision right and right to resource flow. By definition, the "non-profit" condition excludes the "profit distribution" component. Glaeser (2003) and Fama and Jensen (1983) propose that there are no residual rights in NGOs. This could lead to assuming that it is more appropriate to consider the separation between control (principal) and management (agent) than between ownership (principal) and management (agent). However, extending the meaning of the term 'residual right', associates (associations) or founders (foundations) are holders of residual rights that should necessarily be reinvested in NGO's social purpose, since such rights can be understood as something beyond the appropriation of profit/surplus. Hence, there are agency costs related to residual losses, from which arise agency problems regarding property x management, also in NGOs.

Board of directors
The Board of Directors represents the collegiate deliberation body, responsible for keeping the entity's strategic direction through decision-making. It is important to highlight that the Board of Directors is not a mandatory body for NGOs in Brazil (Act No. 10,406/2002, chapter II), contrary to what the Public Companies Law requires. In addition, it is an unpaid body, and the rules for structuring the Board of Directors vary for each status (art. 54, V, and 55 of Act No. 10,406/2002), with no legal standard, which allows it, for example, to be composed by only two members, or elect lifelong members. All of this reduces the role of governance mechanisms of the Board of Directors, when compared to companies.
For Jensen (1993), regarding companies' corporate governance, some reasons for the boards' lack of commitment are their size, composition and independence. For the third sector, we can add the absence of remuneration and rules for its structuring (MINDLIN, 2009). The topic 'Board of Directors' is perhaps the most addressed in studies on governance in the third sector (RENZ and ANDERSSON, 2014). They regard its composition (ANDRÉS-ALONSO, AZOFRA-PALENZUELA and ROMERO-MERINO, 2010), its behavior (BALDUCK, VAN ROSSEM and BUELENS, 2010), or propose indexes of good governance practices (BOARDSOURCE, 2017), among other aspects. The results, however, are not convergent.
Often, board members are elected because they facilitate fundraising and establish relevant contacts for the NGO (MINDLIN, 2009). But Andrés-Alonso, Azofra-Palenzuela and Romero-Merino (2010) did not identify a statistically significant association between the percentage of members of the Board of Directors with experience in fundraising, and allocation or administrative efficiency. Harris, Petrovits and Yetman (2015), taking the Board of Directors as a construct, identified a positive association with the amount of donations received. In addition, Callen, Klein and Tinkelman (2003) showed a statistically significant association between the presence of a major donor and organizational efficiency. They suggest that the presence of major donors on the board could be a signal to other donors about the "quality" of the organization, which would lead to an increase in the volume of donations.

Fiscal Council
The Fiscal Council is a supervisory body for accounting and financial management. It is not a mandatory body in Brazil, either for companies or for non-profit organizations. Regarding companies, it does not exist in more advanced markets, such as the United States, United Kingdom, Germany, and France, where the audit committee is responsible for monitoring the organization's accounting and financial management (SILVEIRA, 2010).
In Brazil, if it were installed, companies already have their constitution and competence defined by law (Act No. 6,404/1976). For non-profit entities, which do not have their own legal rules, the entity's statute would govern it.
The presence of a Fiscal Council represents an additional line of defense of the principal's interest -although overloaded with agency costs. Fontes and Bronstein (2016) identified the presence of the Fiscal Council in 93.7% of the 157 Brazilian NGOs that comprised their sample. The Brazilian Institute of Corporate Governance, in the Best Practices Guide for Third Sector Organizations: Associations and Foundations, recommends as a good governance practice the provision of the Fiscal Council in the bylaws (IBGC, 2016). In other words, by considering the donor as the principal in the agency relationship, it is assumed that the Fiscal Council will act in his defense, since its actions are directed to the inspection of management's acts, in particular by checking the compliance to legal and statutory obligations (LAMB, 2002;IBGC, 2016), among them the terms of the donation contracts. In other words, the presence of the Fiscal Council can mitigate the agency conflict resulting from the contractual relationship established between the donor (principal) and the NGO (agent) in the donation process. Glaeser (2003) and Mendonça and Machado (2004) argue that the donation market in the third sector is equivalent, as an external governance mechanism, to the capital market for publicly traded companies. The capital market reflects, through the value of shares on stock exchanges, the company's current performance and the expectations of investors regarding future performance. This is an incentive for majority shareholders -in companies with concentrated capital -or boards of directors -in those of dispersed capital -to monitor managers' actions. Therefore, this is an incentive for managers to increase stock value. These authors acknowledge that, in the donation market, there is no possibility of hostile acquisitions, since there are no shareholders. However, they argue in favor of two elements that regulate the donation market: (i) the freedom of donors to invest their resources in NGOs that they consider most appropriate; and (ii) the risk of a NGO losing reputation, in the case of inefficient or improper management.

Donation Market
Regarding the freedom of donors to invest their resources, the argument seems more valid in a market of untied donations (without restrictions), where it prevails. In a tied donation market (with restrictions), the donor's discretion is usually limited by an instrument of public call for project support. In analogy to the capital market, it is composed by several "stock exchanges", represented by public notices for project support, shaping a market that is absolutely fragmented and without standards that allow a priori comparison among entities in Brazil. This is similar to the different levels of governance, such as those of [B] 3 , the former BM& FBOVESPA in Brazil. Thus, in this study, we divide the donation market in tied and untied.
It seems that Mendonça and Machado's (2004) second argument -about the risk of the entity's loss of reputation, in case of inefficient or improper management -is better aligned with the dimension 'market of factors and products' (ignored by the authors) than with the dimension 'capital market'. Jensen (1993) addressed both dimensions, by discussing market role as a mechanism for external control of organizations, and so did Fama (1980), on the separation between organization's ownership and control.
In relation to the influence of governance on the decision about institutional donors' contributions to NGOs, Kitching (2009) found a positive association between third sector entities audited by one of the big five and the volume of donations received. Aggarwal, Evans and Nanda (2012) showed a positive relationship between the relevance of the entity's Board of Directors and the volume of donations. Saxton, Neely and Guo (2014) identified a positive relationship between transparency, measured by disclosure on the web, and donations. Harris, Petrovits and Yetman (2015), in turn, showed a positive relationship between the "quality" of governance, composed of seven dimensions (board, management, policies, access, audit, executive compensation, and minutes), and the volume of donations.
In addition to this discussion, Andrés-Alonso, Cruz and Romero-Merino (2006), analyzing the influence of the participation of public institutional donors on the efficiency of Spanish NGOs, identified that the presence of an active institutional donor is positively related to an entity's higher technical and allocation efficiency. Yetman and Yetman (2012) showed that NGOs with more restrictions imposed by their donors report more accurate financial information. Therefore, this indicates the donor's influence on monitoring the entities' actions, in addition to the role of the donations market as a governance mechanism.
Finally, in order to deal with the restrictions of the donation market, NGOs are increasingly adopting approaches from the private for-profit market, leading to what Salamon (1997) called "marketization" of the non-profit sector. The logic of this argument, under the assumptions that (i) NGOs need resources to survive, and (ii) there are requirements they must meet in order to receive donations, is that NGOs are driven to operate according to those that control the resources in the donation market. Thus, they adopt practices and procedures that prevail in organizations that constitute the donation market, among them governance practices (LACRUZ, CUNHA, MOURA et al., 2019).
In Brazil, Tenório (1999) was one of the first to discuss the topic, pointing that the "professionalization" of third sector entities aimed more at their organizational survival than at a greater effectiveness of their actions as social agents; hence, besides the market spectrum, rationality submits the third sector to the market logic. Silva and Aguiar (2011) assessed the effect of pressures from the institutional environment on the evaluation practices of the activities of third sector entities, concluding that the dynamics established between agents, institutions, and NGOs are guided mostly by instrumental rationality.

Political, legal and regulatory system
The political, legal and regulatory system is an external governance mechanism that limits legal security. A generic legal context with little effective control, like that of the third sector in Brazil, is far from creating an effective mechanism of behavior within organizations. However, in order to benefit from tax incentives, distinctions, and fund receipt as a donation from the public administration, third sector entities must meet certain additional requirements, and this brings some effectiveness, as a mechanism of external control, to their political, legal and regulatory system.
Regarding tax incentives, the Federal Constitution considers education and social assistance institutions tax-free (art. 150, VI, c), and social welfare charities exempt from social contributions (art. 195, § 7). For entities to be tax-free, the National Tax Code (CTN) establishes the following requirements in its art. 14: non-distribution of profits, prohibition of revenue shipment abroad, and accounting books in perfect order and clarity (there are specific regulations on taxes within the jurisdiction of the Union, States, and Municipalities, such as the legislation on income tax, which expands the requirements for access to immunity). Entities that do not comply with these requirements may have their immunity suspended (art. 14, § 1 of CTN). In order to have access to social contributions' exemption, there are also requirements (art. 29 and 30 of Act No. 12,101/2009).
In addition, NGOs can claim certain titles, qualifications and certifications from the State, with emphasis on qualification as a Social Organization (OS), established by Act No. 9,637/1998; qualification as a Civil Society Organization of Public Interest (OSCIP), established by Act No. 9,790/1999; and the Certificate of Social Assistance Charity (CEBAS), created by Act No. 12,101/2009. These distinctions grant benefits to third sector entities and, in return, impose additional obligations.
Finally, we emphasize that partnerships involving the transfer of resources to third sector entities by the public administration result in additional obligations, such as the election of an executive board (Art. 34, V of Act No. 13.019/2014), which becomes an external governance mechanism within the political, legal and regulatory system.

CONCLUSIONS
This study contributes to the field of agency theory applied to relationships in the third sector, since the structuring of governance in NGOs, from the perspective of this theory, allowed formulating relationships that complement its generalities, leading to a better understanding of its application in this context. Throughout this study, we aligned the assumptions and concepts of agency theory to the third sector environment. We highlight that the assumption of incompatibility of planning horizons between principal (donor) and agent (NGO) takes place inversely to that assumed as a rule for companies. We understand that the donor is more concerned with the results of the firm during the term of the donation contract (that is, the results of the project that he sponsors), and the NGO is more concerned with the results of its portfolio of present and future projects. In other words, we propose that the agent (NGO) considers the value of the firm as an infinite series of actions, and the principal (donor), the flows of actions in the interval of the contract.
We also highlight the proposition of agency costs related to residual losses in the third sector environment, from which stems agency problems of the nature 'property X management'. In addition, the division of the donation market in tied and untied donations.
In view of the above, a research agenda emerged, on how the donation market manifests itself as an external control mechanism in NGOs governance. If we consider the following assumptions: (1) Different time horizons in the third sector imply donor's transience towards the NGO, as its actions are limited to the contractual relationship; (2) NGOs develop their actions through projects (DIALLO and THUILLIER, 2004;MARQUES, RODY, CAMPOS et al., 2015), whose resources are raised in the tied donation market; and (3) the marketization process (SALAMON, 1997) assigns to NGOs practices that prevail in the organizations that make up the donation market. Then we can ask: given the donor's transience towards the NGO, does the donation market act, in addition to the legal, political, and regulatory system, by reconfiguring governance in these organizations? In other words, does the donation market mediate the relationship between the legal, political, and regulatory system and governance in NGOs?
Although the governance topic has been extensively explored in NGOs studies (GAZLEY and NICHOLSON-CROTTY, 2018), especially regarding the dimensions 'Board of Directors' (REDDY, LOCKE and FAUZI, 2013) and 'transparency' (HASKI-LEVENTHAL and FOOT, 2016), there are still unexplored connections. These are the role of the project tied to the donation as an antecedent, moderator, or mediator of the influence of governance on donations' receipt, and the potential moderating role of NGO's qualification (as OSCIP), in the relationship between governance and donations' receipt in a low regulation environment such as Brazil. Likewise, there are differences of operation regarding the legal nature of associations and foundations, since the latter are supervised by the Foundations' Curatorship, a body of the Public Prosecutor's Office of each municipality (art. 66 of Act no. 10.406/2002). Hence, one could examine if governance is different between third sector entities of distinct legal nature.
It is important to record the difficulty for carrying out research based on secondary data in Brazil (MARQUES, RODY, REINA et al., 2015). A promising project in the country is FUNDATA (http://www.fundata.org.br), which keeps a database on nonprofit civil society organizations (foundations), but is not available for download. Another initiative is the OSCs Map (http:// mapaosc.ipea.gov.br), whose data, until now, are limited to the relationships of third sector entities with the federal government -this database is available for download. This type of initiative is relevant for strengthening the sector, as it sheds light on it, serving as a guide for society.