Contents lists available atScienceDirect
Journal of Accounting and Public Policy
journal homepage:www.elsevier.com/locate/jaccpubpolFull length article
Motivations behind users
’ participation in the standard-setting
process: Focus on
financial analysts
☆
Alessandra Allini
a,⁎, Massimo Aria
b, Riccardo Macchioni
c, Claudia Zagaria
caDepartment of Economics, Management, Institutions, Università degli Studi di Napoli“Federico II”, Via Cintia Montesantangelo, 81026 Naples (NA),
Italy
bDepartment of Economics and Statistics, Università degli Studi di Napoli“Federico II”, Via Cintia Montesantangelo, 81026 Naples (NA), Italy cDepartment of Economics, Università degli Studi della Campania“L. Vanvitelli”, Corso Gran Priorato di Malta, 81043 Capua, CE, Italy
A B S T R A C T
The aim of this study is to investigate what motivatesfinancial analysts to participate in the accounting standard-setting process. We focus onfinancial analysts because they are an im-portant group of thefinancial statements users. The paper employs the meso-level approach used by Durocher et al. (2007) that integrates the macro domain’s focus on the standard setters with the micro domain’s focus on individuals and thus it links the characteristics of due process for standard setting with users’ attitudes. We develop a survey for the Chartered Financial Analysts Institute (CFA), which is one of the largest associations of investment professionals in the world, and collected data through computer-assisted Web interviews. We use a structural equation model with PLS to test our hypotheses. Our mainfindings confirm that a combination of micro and macro domains explains the frequency offinancial analysts’ participation in the standard setting process. This investigation, thus, deepens our understanding of motivations behind ana-lysts’ involvement in the accounting standard-setting process and delivers both theoretical con-tributions and practical insights.
1. Introduction
The International Accounting Standards Setting Bodies (IASB) develops accounting principles and standards through an inter-national consultation process– known as the due process – which involves participation by interested individuals and organizations from around the world, especially those who are concerned about the users of these standards (IFRS Foundation, 2016). In recent years, issues pertaining to the users’ participation in the standard-setting process have attracted increasing attention. The users’ participation has been considered an indicator of the board’s legitimacy (e.g.Larson, 2007; Durocher et al., 2007; Bamber and McMeeking, 2016), which typically refers to acceptance of regulators’ claim that they act in accordance with social values of the users’ groups (Richardson and Eberlein, 2011). Thus, lack of users’ involvement or participation in this process might lead to the
https://doi.org/10.1016/j.jaccpubpol.2018.04.002
☆The authors would like to acknowledge Vincent Papa (Director Financial Reporting Policy at CFA Institute) for his constant help in collecting and discussing
evidence, and want to give him special thanks. We are grateful to Professor Sylvain Durocher (University of Ottawa) for his valuable advices received on the earlier version of this paper. A special thank goes to Professor Bikki Jaggi (Rutgers University). His suggestions were extremely useful in each step of the research, and allow the authors to develop and improve the paper according to the anonymous referees’ comments. The authors acknowledge the feedback gathered during the 40th Annual Conference of the European Accounting Association (Valencia, 10-12 May 2017). Finally, the authors are especially appreciative of the efforts of the Editors, Martin Loeb and Lawrence Gordon, and two anonymous reviewers whose input has greatly enhanced the study.
⁎Corresponding author.
E-mail addresses:[email protected](A. Allini),[email protected](M. Aria),[email protected](R. Macchioni),
[email protected](C. Zagaria).
0278-4254/ © 2018 Elsevier Inc. All rights reserved.
neglect of important aspects of the standards, which will result in their lower quality. Among different constituents, financial analysts form an important group of users because they are sophisticated users of thefinancial statements who have a significant influence on other users through earnings forecasts and advices.
The standard-setting organizations have worked hard to avoid reduction in legitimacy and enhance transparency in the due process and, more recently, they have increasingly claimed the importance of proactive users’ inputs (e.g.Bamber and McMeeking, 2016; IFRSFoundation, 2016), suggesting that users are no longer considered a rhetorical group as previous research contended (Young, 2006). Nevertheless, some scholars stillfind that the due process continues to be only a ritual procedure, which is limited to creating an impression of transparency (e.g.Ram and Newberry, 2013; Camfferman and Zeff, 2017).Wingard et al. (2016)document that the IASB procedures lack substantive engagement or participation because these opportunities are primarily reserved for powerful stakeholders who play an important role in the governance structure and in the national standard-setting process.Pelger and Spieb (2017)reveal that users’ orientation in the consultation activities is predominantly formal in nature and confirm that the actual practice of the due process raises concerns about the fairness in procedures followed by the standard setting organizations.
The above arguments make this research very timely because it sheds light on what factors can increase users’ participation and allows opening up the‘black box’ on how the standard setters might enhance more participation by the users (Evans et al., 2005; Pelger, 2016).
Therefore, despite the importance of users’ input, the existing studies largely provide results on users’ participation by linking it to the users’ positions and their effect on standard-setters’ decisions, but we still know very little about users’ motivation to participate in the standard-setting process (e.g.Durocher, 2009; Georgiou, 2010; Durocher and Fortin, 2011).
The researchfindings on the users’ positions provide conflicting results (e.g.Puro, 1985; McKee et al., 1991). Some studies suggest that the standard setters are receptive to the preparers’ preferences (Brown and Feroz, 1992) or users’ preferences (Saemann, 1999). Other studies consider that the geographical, institutional and cultural factors play an important role and they attempt to create various tools to coordinate their use in the due process, such as white papers, comment letters, and meetings, whereas another group of studies explain the determinants of users’ contributions to the due process (e.g.,Georgiou, 2010; Jorissen et al., 2012; Hansen, 2011; Larson et al., 2011; Larson and Herz, 2013; Morley, 2016). Very few researchers analyse motivations affecting the users’ participation or evaluate specific user categories with unique attitudes and needs (Schalow, 1995; Tandy and Wilburn, 1996; Durocher et al., 2007; Durocher and Fortin, 2011).
Durocher (2009) and Georgiou (2010)especially point out that accounting research has largely neglected particularfinancial statement users; thus, more work is needed tofill this gap. This study focusses on financial analysts as a specific user group for the following reasons.
Financial analysts are one of the primary users of accounting information because they are representatives of the investment community for which the reporting of corporate information is intended (e.g.Orens and Lybaert, 2010; Brown et al., 2015). They play a critical role in performing analyses and extensively use the financial statement information to provide forecasts, re-commendations and target price outputs aboutfirms (Schipper, 1991; Epstein and Palepu, 1999). Financial analysts’ goal to obtain standards that provide relevant information for investment decisions implies that analysts have an interest in contributing to the due process because it represents a means of shaping the formation of accounting standards towards higher quality (Orens and Lybaert, 2010; Jorissen et al., 2012). However, the standard setters have only recently recognized the important role played byfinancial analysts in the due process. The Boards have intensified efforts to mobilize them because they are supposed to possess knowledge essential to meeting the information needs of users of thefinancial statements (IASB and FASB website). The scholars have also called for more research on these issues to support the standard-setter organizations and they have emphasized that a deeper examination of this group would be welcome because it would meet the need for inclusion of thefinancial statement users in the due process (e.g. Morley, 2016).
The main focus of this study is to investigatefinancial analysts’ motivation to participate in the standard-setting process, where participation refers to the actions that interested parties adopt to influence the rule-making body (Schalow, 1995). We rely on the framework developed byDurocher et al. (2007), which has been developed to explain, more generally, users’ participation in the due process, and we apply this framework to examinefinancial analysts’ participation as a specific category of users.
TheDurocher et al.’s model (2007)has the merit of integrating the macro domain’s focus on the standard setters and the micro domain’s focus on individuals, and thus it links the characteristics of due process for standard setting with users’ attitudes (Klein et al., 1999). Other theoretical frameworks explain constituents’ participation from an isolated perspective, such as the political nature of the standard-setting process and perceived legitimacy of the board. In addition, these papers devote very little attention to individual attitudes and expectations.
In contrast, the framework developed byDurocher et al. (2007)is considered particularly suitable for the current study because it allows understanding more in depth the expected factors affecting participation by employing idiosyncratic combination of ap-proaches that link the characteristics of the due process with the personal attitudes of a group of users. These authors stressed that it is necessary to test this framework with a sample offinancial statement users, which is something that has been neglected in other approaches.
We formulate our hypotheses according to the above-mentioned framework. We develop a questionnaire to obtain data for analyses. The target respondents are members of the Certified Financial Analysts (CFA) Institute, a prestigious global association of investment professionals representing investor interests infinancial reporting proposed standards published by the International Accounting Standards Board (IASB) and the US Financial Accounting Standards Board (FASB). Data are collected through computer-assisted Web interviews, and a structural equation model using the PLS approach (SEM-PLS) is used to estimate the model and test our hypotheses. Post-survey interviews were also conducted with three randomly selected CFA analysts.
The mainfindings suggest that a combination of micro and macro domains results in greater financial analysts’ participation in the standard setting process. In particular, analysts’ perceptions of the standard-setters’ characteristics in terms of their instrumental, symbolic, and systemic power, personal attitudes of analysts and legitimacy, are considered to be main reasons to encourage analysts to participate in the due process.
The moral legitimacy, however, does not represent a valid motivation.
This study is expected to provide important contributions. From a theoretical perspective, it adds knowledge to a relatively unexplored strand of literature pertaining to the motivations behind user’s participation in the standard-setting debate. It also contributes to the existing literature, because Durocher et al.’s framework has been developed based on a qualitative research approach with the use of interviews, but it has not been empirically tested with a group of users. This research contributes to the literature by quantitatively assessing the model through survey data and by examiningfinancial analysts’ actual perceptions about the characteristics of the standard-setting process and the relationships between these characteristics, personal behaviours and le-gitimacy perceptions. Thus, the results offer a deeper understanding of motivations of a primary group of the financial statement users, i.e.financial analysts, that has been neglected in previous research.
Moreover, the study has relevant policy implications at the international level because the results help the standard setters to foster greater integration of analysts’ opinions into the accounting development activities. Therefore, our findings are relevant to enriching the knowledge of boards on key factors to increase cooperation and coordination of analysts in the due process to con-tribute to make the standards of higher quality and relevance.
The remainder of the paper is structured as follows. The second section reviews the existing literature, and the third describes the theoretical framework and formulates the hypotheses. The fourth section explains the research design. Section Five describes the statistical analysis, and Section Six discusses thefindings. The last section concludes the study and highlights implications for further research.
2. Literature review
The issue of users’ participation and reasons for involvement in the accounting standard-setting process have been receiving an increasing level of attention in recent years (e.g.,Weetman et al., 1996; Cooper and Robson, 2006; Young 2006; Durocher et al., 2007; Hansen, 2011; Larson et al., 2011; Jorissen et al., 2012; Alvarez et al., 2014). It has been argued that a strong connection exists between the development of standards processes andfinancial statement users (Young, 2006) because these users act as a guide in shaping thefinancial accounting standards. According toBarth (2000), an imbalance in their participation might decrease our understanding of how the standard setters make decisions.
Many theoretical perspectives have been employed to understand the reasons for users’ participation. However, empirical in-vestigations remain scarce (e.g.Durocher et al., 2007; Durocher and Fortin, 2011).
Within the economic theory of democracy, user involvement relates to economic motivations, in which the cost-benefit relationship is the main driver of participation (Downs, 1957). This approach, however, was primarily conceived for preparers, who, as opposed to users, are more likely to be active participants because of their homogeneous economic interests. In particular,Olson (1965)showed that under certain conditions, an individual will not contribute to a shared action to obtain a public good, even when that action would be beneficial for the entire group. In another research project,Lindahl (1987)adopted the same framework to extend the focus on lobbying by preparers and accountingfirms but did not analyse users, because of the lack of previous empirical work for ex-pectation matching. Other authors addressed the subject of academic community participation in the due process (e.g.,Alvarez et al., 2014).Tandy and Wilburn (1996)documented the academic community’s low level of participation, their lack of understanding or interest in this process and their low expectations in terms of their ability to affect standard-setters’ decisions. Moreover, the lack of academic rewards and the absence of a relationship between the issues at hand and teaching/research are cited.Larson et al. (2011) documented other reasons, such as language barriers, in conjunction with short comment periods, which might hinder academic participation by non-English speakers.
More recently, based onSutton’s predictions (1984), Georgiou (2010)analysed the participation of investment management associations and showed that the major factor inhibiting practitioners from participating is the cost of lobbying, not complacency that the standard setter is‘on their side’ and safeguarding their interests. Conversely,Elbannan and McKinley (2006)developed a the-oretical framework that specified the conditions under which corporations are likely to resist financial reporting standards.
The second strand of studies adopts the position of influencing group approach, questioning how strong alliances affect the standard-setting process (Haring, 1979). Empirical results, however, appear to be mixed. Although Puro and McKee (1985) and McKee et al. (1991)suggested that a coalition between accounting standard setters and their corporate clients existed, andMezias and Chung (1989)documented a form of collusion or consensus among accounting standard setters,Hussein and Ketz (1980) and Brown (1981)did notfind this. Investigating the specific reasons behind the users’ participation, an early study conducted by Weetman et al. (1996)found that the motivations behind users’ inactivity included lack of time and that informal meetings with standard setters was the main approach used to participate. Later,Van Lent (1997)analysed the lobbying activities of various interest groups, including those of report users, and revealed that lobbying was not intensive. More recently, the study byBamber and McMeeking (2016), who investigated whether the IASB still favoured the Big Four accountingfirms in the due process, found that thesefirms were statistically significantly less influential than other groups.
By adopting a meso-level approach, a more in-depth explanation of the motivations surrounding users’ participation in the standard-setting process has been proposed byDurocher et al. (2007). Their model integrates the macro domain’s focus on orga-nization with the micro domain’s focus on individuals and groups (Klein et al., 1999). The proposed framework suggests that user
participation is explained by their perceptions about the board’s characteristics. More recently,Durocher and Fortin (2011)used the expectancy theory to explain practitioners’ intentions to participate in the standard-setting process and found that valence is a determinant of the attractiveness of becoming involved, whereas the expectancy forces are determinants of practitioners’ behavioural intentions to participate.
This review exhibits a relatively small stream of the accounting literature that addresses the subject of user involvement; few studies have analysed the reasons of a specific category of user.Durocher et al. (2007), despite providing an in-depth understanding, does not explain how the proposed integrated model works with respect to an individual category of user. Some reasons, in fact, might prevail or might not be confirmed among a given type of user. Furthermore, some considerations might act differently or offset others for a specific category of user. It is likely that users have diverse needs, and their individual motivations to participate might vary depending upon their attitudes, incentives, professional positions or social roles (Saemann, 1999).
3. Explanatory theory and hypotheses development
3.1. Explanatory theory
The explanatory theory proposed byDurocher et al. (2007)is particularly suitable for the purposes of our analysis because it allows us to understand greater (lower) participation of analysts in the standard-setting process from a broader and more compre-hensive perspective by referring to a meso-level approach, which jointly includes the focus on organizations, the focus on individuals and groups (Klein et al., 1999), and links these two perspectives. Therefore, considering that analysts’ opinions are strongly ad-vocated by standard-setter organizations, which are now trying to strategically mobilize them in their activities, the idiosyncratic combination of this perspective with the organization attributes leads to obtaining a more depth understanding of analysts’ in-dividual perceptions behind their greater participation in the due process.
Specifically,Durocher et al. (2007)consider three organizational theories:Hardy’s (1994)power framework, explaining the macro-level perspective;Vroom’s (1964)expectancy theory, focusing on the micro-level viewpoint; andSuchman’s (1995)typology of the forms of legitimacy, acting as a link between the macro level of the standard-setting process and the micro level of the specific determinants of participation by eachfinancial analyst.
Hardy’s multidimensional power framework (1994)pertains to the macro-level dimension of the standard-setting process, which is political in nature (Tandy and Wilburn, 1996; Zeff, 2002; Young, 2014).
Hardy (1994)defines the main characteristics of such a process, moving from a broad definition of power as a force that in-fluences the outcome. This definition encompasses three dimensions: instrumental power, i.e., the ability to intentionally secure preferred outcomes in the face of conflict among declared opponents (Dahl, 1957); symbolic power, i.e., the ability to secure preferred outcomes by preventing conflict from arising; and systemic power, i.e., the ability to create unconscious advantages or disadvantages for political actors.
Vroom’s expectancy theory (1964)relates to the micro-level dimension and is used to choose whether to participate in the standard-setting process. These behavioural alternatives are selected based on three determinants of user participation: valence, i.e., possible affective orientations towards certain outcomes (Van Eerde and Thierry 1996); instrumentality, i.e., the perception that participation could lead to obtaining the resultant outcomes; and expectancy, i.e., the perception that one has the ability to contribute adequately.
To link the descriptive categories provided by Hardy and the determinants of participation suggested byVroom (1964), the framework employsSuchman’s (1995)typology of legitimacy.Suchman (1995)refers to legitimacy in a broad sense as a generalized perception or assumption that the actions of an organization are desirable, proper or appropriate within some socially constructed system of norms, values, beliefs and definitions.
Thus, the author identifies three primary forms of legitimacy that encompass different dimensions: pragmatic, moral and cognitive. Pragmatic legitimacy concerns a utilitarian, discursive evaluation of the interested parties of an organization based on their self-interests. Moral legitimacy refers to a positive normative evaluation of an organization and its activities from the point of view of an audience’s socially constructed value system (Suchman, 1995). Cognitive legitimacy relates to assumptions that individuals can make about an organization and involves the simple acceptance of that organization based on taken-for-granted cultural accounts.
Analysing this co-existing form, rather than referring to legitimacy in general terms, is useful to assess more in depth the fairness perceived by analysts concerning the standard-setting process. Indeed, the legitimacy of the due process is not pre-existing or innate (Burlaud and Collasse, 2011); rather, it is enhanced when accounting boards do the right things for society (Bamber and McMeeking, 2016). Thus, Durocher et al.’s model uses Suchman’s typologies of legitimacy as a link to understand whether and to what extent the standard-setting process (macro level dimension under Hardy’s framework) is considered desirable, proper or appropriate by analysts (micro-level dimension under Vroom’s framework).
3.2. Hypotheses development
Durocher et al.’s framework is graphically represented inFig. 1, together with our expected hypotheses directions.
In line with the expectancy theory (Vroom, 1964), the model postulates that motivational forces and individual characteristics of behaviour (i.e., valence, instrumentality and expectancy) affect participation positively. Attractiveness, desirability and anticipated satisfaction with outcomes, and the perceived likelihood that participation in the setting process will lead to the desired benefits of participating are expected to increase user involvement.Baskerville (2005)used Vroom’s expectancy theory to explain the increased
constituent participation in the New Zealand standard-setting process and suggested that the increase in participation might stem from a belief on the part of constituents that there is a greater possibility of influencing the outcome when institutional changes occur. Financial analysts considerfinancial statements to be useful for their professional activities (Brown, 1997; Epstein and Palepu, 1999); therefore, they have a great interest in the outcome of the process, i.e., the development of high-quality accounting standards. Moreover, they perceive that their coalition– as a professional association – can have a degree of influence on the final content of the accounting principles. Moreover, their previous experience, personal background, and knowledge givefinancial analysts the ability to contribute, which might motivate them to actively join in the process.
All of the above-mentioned factors are likely to immediately explain the personal attitudes behindfinancial analysts’ interest in the process. Thus, we postulate the following:
H1. Personal behaviours (i.e., valence, instrumentality, and expectancy) positively influence financial analysts’ motivations towards greater participation in the standard-setting process.
The framework also suggests that cognitive legitimacy negatively affects participation (Durocher et al., 2007). The concept of cognitive legitimacy relates to the perceptions that people have about an organization. In other words, it involves the taken-for-granted acceptance, from a cultural perspective, of the role played by the standard setters (Suchman, 1995). Therefore, in the context of the standard-setting process, cognitive legitimacy refers to the perception that it is the natural role of accounting experts to set standards. Thus, the main assumption of the Durocher et al.’s framework is that standard setting has traditionally been considered an activity of the accounting profession and that accountants are considered the most qualified people to establish standards (Gaa, 1986; Hines, 1989). Undoubtedly, in recent years, the international standard-setter organizations have substantially changed how they operate; the most important change concerns refining the due process. In particular, through the enforcement of the three underlying principles of consultation defined in the new version of the Due Process Handbook 2016 – i.e., transparency, full and fair consultation and accountability– the IASB has started to increase the legitimacy of its activities within society by enhancing the mobilization of an increasing number of temporary and informal consultative and advisory committees. However, the board has experienced diversi-fication, advocating the participation of a non-technical audience in the due process, regardless of their specific accounting expertise. Although expertise apparently is shifting away from national professional organizations,Botzem (2014)notes that accountants are the most important force for board composition and activities.
Because the IASB is a central arena defining the body of knowledge in cross-border accounting, it might be expected that a lack of professional competence and specific accounting skills might not provide substantial insights on technical issues. Thus, it is likely that analysts do not operate as accountants, and their personal backgrounds might not reflect all accounting knowledge.Botzem and Quack (2009)contend that standard setting is expert driven, and the recent study ofPelger and Spieb (2017)documents a certain reluctance by some IASB board members to obtain feedback, particularly from analysts because they lack technical knowledge. This might imply that analysts perceive that they are not specialists in the accounting area, which is the task of accountants and, therefore, they do not have the skills to define the standards. Based on this argument, and following the Durocher et al.’s framework, we aim to test the following hypothesis:
H2. Cognitive legitimacy negatively influences financial analysts’ motivations towards greater participation in the standard-setting process.
According to the Durocher et al.’s framework, pragmatic legitimacy has an immediate effect on attitudes towards participation. In the accounting standard-settingfield, pragmatic legitimacy is conceived as a utilitarian assessment of behaviour of the standard setters that is performed by interested parties in relation to their self-interests. In other words, users can observe the activities of the board to assess the main practical effects of its decisions for themselves. Based onSuchman’s arguments (1995), three forms of pragmatic legitimacy occur: exchange, influence and dispositional.
Financial analysts can grant exchange legitimacy to the standard-setting process– which implies that they might grant support to the debate over accounting standards development– because they perceive that the results of this process are useful for them. Scholars contend that the usefulness and quality of accounting standards are indications of exchange legitimacy (e.g.Larson and Kenny, 2011). Thus, the usefulness offinancial information and accounting projects are main issues creating interest among analysts. From this perspective, exchange legitimacy positively affects personal behaviour of valence to obtain more useful standards.
Another form of pragmatic legitimacy, influence legitimacy, is granted by financial analysts because they might believe that the CFA organization is responsive to their interests. For example, the inclusion of some of their group’s representatives on the standard-setting board might increase analysts’ perception of their ability to affect the final content of the accounting principles. From this perspective, influence legitimacy will affect the personal behaviour of instrumentality. Empirically, many scholars have examined whether standard setters are more responsive to specific constituencies; however, results are inconclusive (e.g.Chee Chiu Kwok and Sharp, 2005; Georgiou, 2010; Bamber and McMeeking 2016).
Financial analysts also grant dispositional legitimacy to the standard-setting process to the extent to which they perceive that the accounting board is actually considering their views when making decisions concerning standards. From this perspective, disposi-tional legitimacy will affect the personal behaviour of valence to obtain standards that satisfy analysts’ needs.
All of the above-mentioned forms explaining pragmatic legitimacy are likely to enhance the attractiveness of participation. Thus, the following hypothesis is tested:
H3. Pragmatic legitimacy– in all its forms – positively influences the personal behaviours of greater financial analyst participation in the standard-setting process.
Moral legitimacy concerns a positive, normative assessment of the organization’s activity from the audience’s perspective (Suchman, 1995). In other words, this (primary) form of legitimacy encompasses four sub-forms of legitimacy1consequential, pro-cedural, structural and personal.
Consequential legitimacy is the notion that the board operates in the public interest, which might influence financial analysts’ decision to become involved in the process. For instance,financial analysts can evaluate whether the accounting board is operating in the right direction and in accordance with common rules and social values. Therefore, granting consequential legitimacy tofinancial analysts would in turn would affect their valence of participation.
Procedural legitimacy pertains to the proper characteristics of the due process, which implies the chance to submit one’s opinion to the public debate and believe that it will actually be considered. In other words, the procedural legitimacy of the consultative process stimulates participation in terms of instrumentality and expectancy for users who decide to join the process. Thus,Le Manh (2012), in investigating the IASB’s comprehensive income process, finds that the board was not responsive to the opinions expressed by con-stituents during that process, which could negatively affect the IASB’s legitimacy.
Structural legitimacy refers to the structural characteristics of an organization.Baylin et al. (1996) and Gorelik (1994)document that standard setters have historically revised their structures to achieve social acceptance. Iffinancial analysts believe that the structures of the IASB ensure the representation of all interested parties, they will be indirectly motivated to enhance their parti-cipation. Additionally, the creation of ad hoc study groups within the board might reassure analysts that they have influence, hence soliciting greater participation. In other words, the structural legitimacy granted to the process affects instrumentality, which in-fluences broader participation.
Personal legitimacy is the form of legitimacy that, apart from professional skills, relates to the charisma and personal characteristics of the standard-setting committee members. Consistent withRichardson and Eberlein (2011), the exhibition of technical competence is a condition for standard setters to gain legitimacy with their constituencies.Allen and Ramanna (2013)find that FASB members with background infinancial issues emphasize relevance more than reliability.Jiang et al. (2015)reveal that investors prioritize the decisions of some board members. Thus, perceptions of the characteristics of members of standard-setting committees could affect users’ opinions about the boards’ legitimacy. To this end, for instance, the independence of the board in its decision-making process will affect instrumentality, which might affect the participation of financial analysts.
All of the above-mentioned forms explaining moral legitimacy are likely to influence the personal behaviours of analysts. Thus, we hypothesize the following:
H4. Moral legitimacy– in all its forms – positively influences the personal behaviours of greater financial analysts’ participation in the standard-setting process.
Participation is also expected to be influenced by the perceived characteristics of the standard-setting process, which can be 1Durocher et al. (2007)added another form of moral legitimacy (legal) toSuchman’s (1995)typology, which is deeply addressed byJohnson and Solomons (1984).
addressed byHardy’s (1994)power framework, relying on various dimensions of power, namely instrumental, symbolic and systemic. With specific reference to instrumental power, the standard setter can benefit from different sources of power in terms of the authority to establish standards, money and business relationships, internal committees, and a consensus among various groups of participants, which will undoubtedly secure the outcomes expected byfinancial analysts.
From a general perspective, if standard setters are positively recognized in terms of their expected power to issue valuable accounting standards, users can grant exchange legitimacy (a form of pragmatic legitimacy) to the due process (Heidhues and Patel, 2012). More specifically, if financial analysts perceive that the promotion of professional judgement by the standard setting enhances the quality of corporate reporting, they are likely to grant exchange legitimacy to this promotion. Therefore, the model assumes the existence of a positive link between instrumental power and exchange legitimacy.
Instrumental power is also revealed in the strategies that users pursue to achieve their goals and in the results of those actions, such as the development and introduction of specific standards and principles. In other words, individuals or groups might use strategies to influence the final content of accounting standards, which are intended to be the outcome of a consultative process. For instance, European accounting directives provide the IASB the authority to establish standards in the European markets under certain conditions. Thus,financial analysts can promote intentional strategies, such as coalitions among themselves or between themselves and preparers, to mobilize sources of power to attain the expected outcomes. From this perspective, the model hypothesizes the existence of a positive link between instrumental power and influence legitimacy (as a form of pragmatic legitimacy).
Furthermore, instrumental power also results in attitudes of standard setters evaluated by constituents, such as “wise” or “trustworthy” (Suchman, 1995), and is related to the perception concerning the power of the board to react to users’ interests. Thus, if financial analysts perceive that the standard setters are making wise decisions consistent with their values and needs, they will grant dispositional legitimacy (a form of pragmatic legitimacy) to the due process, which in turn should affect their level of participation. From this perspective, the framework hypothesizes the existence of a positive link between instrumental power and dispositional legitimacy.
All of the above-mentioned forms explaining pragmatic legitimacy are likely to be influenced by the instrumental power. According to the model, thus, we formulate the following hypothesis:
H5. Instrumental power positively influences pragmatic legitimacy in all its forms.
A different dimension of power, which helps to clarify the main characteristics of the standard-setting process, is symbolic power. It refers to the intent of the standard setter to pursue the public interest by giving priority to users’ needs (Hardy, 1994).
In the accountingfield, because there are many (information) needs to be satisfied for several interested parties, one means of achieving the public interest is to prevent conflicts between actors from arising. In other words, the standard setters might use symbolic power to prevent conflict, which in the long term might lead to a situation in which stakeholders accept their processes as natural and unchangeable. More precisely, the belief that accounting standards are developed in the public interest might affect the consequential legitimacy (as a form of moral legitimacy) granted byfinancial analysts.
Thus, symbolic power views the due process as a ritual, supporting the perception that one input can have some degree of influence on the final content of standards (Fogarty, 1994). Procedural legitimacy (a form of moral legitimacy) appears when the organization embraces socially accepted techniques and procedures, and is established and maintained through open debates, transparency and opportunities for analysts to influence the process (Beisheim and Dingwerth, 2008). In the case of a standard setter, this type of legitimacy occurs if the standards are going through an open public debate and the decisions are adequately justified. To this end, the model hypothesizes the existence of a positive link between symbolic power and procedural legitimacy.
Symbolic power is also expressed by the existence of committees and representatives that allows for the responsibility of the users. Such committees can be formed to establish the perception of structural legitimacy (a form of moral legitimacy). Indeed, feelings of being represented in the structure of the boards to develop standards positively affect the moral evaluation of the boards (i.e., structural legitimacy) granted byfinancial analysts. On these bases, the model postulates the existence of a positive link between symbolic power and structural legitimacy.
Finally, symbolic power of the standard setters is affected by expertise and independence of their members (Johnson and Solomons, 1984). This condition enhances boards’ legitimacy and, in particular, personal legitimacy (as another form of moral legitimacy) because the latter providesfinancial analysts with the confidence that their preferences would be considered in the final standards, ultimately affecting their level of participation. Thus, the model also hypothesizes a positive link between symbolic power and personal legitimacy.
All of the above-mentioned forms explaining moral legitimacy are likely to be influenced by the symbolic power. According to the model, we postulate the following:
H6. Symbolic power positively influences moral legitimacy in all its forms.
The perception thatfinancial analysts have of the standard-setting process relates to the so-called systemic power (Hardy, 1994) and reflects how the due process actually allows participation by various categories of users. Systemic power represents a non-intentional form of power that might lead to unintended and pervasive effects on financial analysts’ participation. More specifically, financial analysts acknowledge the cognitive legitimacy of accountants, recognizing that accountants can substantially contribute to accounting standards. Thus, analysts perceive themselves as being insufficiently able to provide input as accountants do. Clearly, this lack might well influence their willingness to engage in the due process. However, the standard-setters’ systemic power, well-re-cognized byfinancial analysts, is the main factor fostering them to assign a high level of cognitive legitimacy to accountants, as the previous discussion on accountants’ perceived broader ability to participate in the due process actually explains. Therefore, according
to the model, we formulate the following hypothesis: H7. Systemic power positively influences cognitive legitimacy. 4. Sampling design and questionnaire definition
To test the above-mentioned hypotheses, we undertake a study based on the sample respondents, who have knowledge or ex-perience in the accounting standard-setting process. The initial target respondents were members of the Certified Financial Analysts Institute (CFA), a prestigious global association of investment professionals representing investor interests concerningfinancial re-porting proposals published by the International Accounting Standards Board (IASB) and the US Financial Accounting Standards Board (FASB).
The target population respondents were selected by the Director of the CFA Institute’s policy arm, V. Papa, who included all financial analysts who were members of the user committees and working groups of the IASB, the FASB, and the European Financial Reporting Advisory Group (EFRAG). The population was generated from CFA databases using the specified SIC codes and job titles and contained 380 mailing list names. Furthermore, to limit potential selection bias and under-coverage concerning the web-based survey, we consider the following rules and controls (Bethlehem, 2010):
– frame population was defined starting from a clearly structured database – the CFA members who were users of financial statements and had expressed an interest in participating in the standard-setting processes;
– financial analysts were contacted through their own institutional e-mail address they use to periodically communicate with CFA; – financial analysts were previously contacted by the CFA Director, who illustrated for them the main objectives of the survey; – in the letter of invitation to participate in the survey, we asked the analysts to reply with either an affirmative or negative
response. We, thus, checked whether the e-mail message had been read;
– reminder e-mails were sent three times in the period from February to September 2015;
– a web-based survey system allowed the responders to retrieve their partial response so that they could start right where they left off the next time they clicked the survey link.
Power, legitimacy, attitudes and values were developed in line with the model developed byDurocher et al. (2007), and all items were measured using six-point Likert scales, in which“1” means completely disagree and “6” means completely agree. We chose the “forced choice” method of defining an even point scale, so a neutral choice was not available. Thus, the scale forces respondents to be more discriminating and eliminates any possible misinterpretation of a mid‐point (Zavala, 1965).
The study was preceded by double-exploratory, qualitative research that was conducted prior to the survey. To identify relevant questions, we initially developed a list of potential survey questions by converting the reasons based on the 2007 framework. Thus, we contacted an academic colleague who was familiar with this literature and asked him to suggest appropriate questions, and we also received feedback on the survey design from the CFA Director. Based on their suggestions, new questions have been developed for each construct.
We distributed the pilot survey to six randomly selected analysts with different socio-demographic and cultural backgrounds, which helped us to assess the reasonableness and presentation of our questions and the time required to complete the survey. This process helped reduce the possibility that we omitted fundamental questions, asked unimportant or ambiguous questions or designed a survey requiring too much time to complete.
As part of this effort, we created and administered one version of a pre-survey containing a list of 70 potential items on which the respondents were asked to select the common questions to explain the main motivations for their participation from the list of all previously obtained items. To avoid different interpretations, at the beginning of the survey, we included the definition of partici-pation as stated bySchalow (1995). At the same time, we also asked them to provide their interpretation of participation. What emerged is that terms such as activism, mobilization, participation, engagement or involvement have the same meaning for analysts, fully complying with the definition bySchalow (1995).
We also randomized the order of the questions presented within each construct, unless they had a natural location (e.g., never, once a year, twice a year).
We kept the most frequently mentioned reasons for thefinal version of the survey, which amounted to 47 items. A group of experts also evaluated possible overlapping among questions. This list of items was used as an initial set of variables to estimate the structural equation model. After the model assessment process, afinal subset of 22 indicators was identified. The description of this subset is reported inTable 1.
Some questions are formulated in a negative form, e.g., SBP_1“The accounting issues are already framed before the exposure-draft, making it difficult to influence the outcome of the process.” In these cases, high values on the scale indicate a negative opinion. For this reason, to perform the analysis, these indicators have been transformed by inverting the scale.
The dependent variable consists of“the frequency of participation in the accounting due process”, measured on a five-level ordinal scale from“never” (1) to “very frequently” (5).
Thus, to better interpret thefindings, to avoid bias, and develop a greater insight into the analysts’ world, we conducted post-survey interviews via phone calls with three randomly selected CFA analysts. Each interview lasted approximately 20 min. To limit misunderstandings, at the beginning, we clarified that the interviewees were to answer frankly and not to provide the answers they thought they ought to be making. At some points, interviewees were asked to provide examples or to phrase in their own words so
that their grasp of the issues could be confirmed. The interviews were held in March 2017, and further clarifications concerning some issues were obtained through e-mail correspondence.
4.1. Data collection and description
A computer-assisted Web interview survey (limesurvey 2.0 open-source platform, installed on our academic servers) was con-ducted to reach as many respondents as possible and retrieve as much information as possible within a short period of time. We informed the analysts that their responses would be held in strict confidence, that no individual responses would be reported and that the survey should take less than 15 min to complete. We receive a total of 55 responses, or a 15% response rate, which exceeds that of other accounting andfinance surveys administered to financial analysts via e-mail (e.g., 8.4% forGraham et al. (2005)and 10.9% for Brown et al. (2015)).
The characteristics of our sample appear inTables 2a, 2b and 2c.
Based onTable 2a, almost 60% of respondents have participated in the IASB standard-setting process, and 13% have participated in both the IASB and FASB processes. Respondents had high levels of expertise, typically having more than 10 years of professional experience (91%). The majority of selectedfinancial analysts had graduated with a degree in accounting and finance studies (76.4%). More importantly, approximately 49% reported systematic (frequent or very frequent) participation in the standard-creation process, whereas almost 50% reported moderate (rare or occasional) participation. Finally, the main two forms of participation were con-sultative groups (72.7%) and roundtable meetings (65.5%). Surprisingly, this lastfinding appears to disregard the growing body of research on the use of comment letters as a typical informal participation method by constituents in the due process (e.g.Georgiou, 2010; Dobler and Knospe, 2016). Consistent withPelger and Spieb (2017), one possible explanation could be that consultative groups might offer greater potential for user engagement because they refer to the early stage of the standard-setting process. Interviewees confirmed that dedicated committees, answers to surveys and questionnaires are also employed as tools to obtain their influence.
Analysing the frequency of participation by graduation (Table 2b) raises the question of how are there no relevant differences between respondents who graduated in accounting,finance and other fields. Therefore, summing the “frequently” and “very fre-quently” categories, the percentages are 50.0% against 46.2%, respectively.
Table 2creports the distribution of the frequency of participation by the respondent profile. The experts with a profile “equity sell side” show the highest participation rate in the standard-setting processes (frequently or very frequently, 72.8%), whereas the respondents with a profile “credit analyst” or “equity buy side” have a lower participation rate (frequently or very frequently, 43.8% and 42.8%, respectively). Furthermore, almost one-third of credit analysts affirm having participated in standard-setting processes
Table 1
The relevant indicators (questions) used in the model. Construct Indicator Description Block I– Hardy’s power framework
Instrumental Power IP_1 Your critical resources enabling you to participate in this process is adequate IP_2 Your structural position enabling you to participate in this process is adequate
IP_3 Your individual or group characteristics enabling you to participate in this process are adequate IP_4 You are able to form coalition with other groups to influence the standard setting process
Symbolic Power SBP_1 The accounting issues are already framed before the exposure-draft. making it difficult to influence the outcome of the process
SBP_2 The characteristics of the due process make it probable that your views will be considered if you participate SBP_3 The importance of consultative committees on the standard setting process is relevant
SBP_4 Personal characteristics of standard setting committee members (perceived expertise. independence. lack of bias) in standard setting process are relevant
Systemic power PA_1 The power exerted by your association in the standard-setting process ensures a right consideration of your need PA_2 The independence of the standard setter and the unbiased committee members positively affect your participation Block II– Schuman’s Legitimacy typology
Pragmatic Legitimacy PL_1 Standard-setting institutions develop accounting standards that producefinancial statements useful to users PL_2 My group’s needs and preferences are considered in the IASB/FASB standard setting process
Moral Legitimacy ML_1 The public interest (as defined by IFAC 2012) is a genuine concern into the standard setting process
ML_2 The IASB/FASB standard setting process is characterized by an open public debate and adequate justification for final decision
ML_3 IASB/FASB members are experts ML_4 IASB/FASB members are independent
Cognitive Legitimacy CL_1 Accounting experts play a key role in the establishment of accounting standards Block III– Vroom’s Expectancy theory
Personal behaviours DT_1 Interest in the standard setting issues
DT_2 Other groups form coalitions to influence the standard-setting process DT_3 Participating in the process takes a lot of time
DT_4 My own participation would influence the outcomes of the standard setting process Block IV– Frequency of Participation
Participation to due process PT_1 Frequency of participation to the accounting due process*
(31.8%).
5. Statistical analysis
5.1. Structural equation modelling with PLS
To test our hypotheses, we used structural equation modelling (SEM). SEM is a family of statistical methods designed to test a conceptual or theoretical model (Kaplan, 2007). Some common SEM methods include confirmatory factor analysis, path analysis and latent growth modelling (Kline, 2011). The term“structural equation model” most commonly refers to a combination of two things: a measurement model, or outer model, which defines latent variables (or constructs) using one or more observed variables (items), and a structural regression model, or inner model, which links these latent variables together. The parts of a structural equation model are
Table 2a
Demographic data for the respondents.
Variables Responses Frequencies %
Due process used – IASB 33 60.0
– FASB 9 16.4
– Both 13 23.6
Professional experience – Less than 5 years 2 3.6
– From 5 to 10 years 3 5.5
– More than 10 years 50 90.9
Gender – Female 9 16.4
– Male 46 83.6
Graduation – Accounting and Finance 42 76.4
– Other 13 23.6
Profile of respondent – Credit analyst 16 29.1
– Equity sell side 11 20.0
– Equity buy side 28 50.9
Participation to due process – Never 0 0.0
– Rarely 13 23.6
– Occasionally 15 27.3
– Frequently 15 27.3
– Very frequently 12 21.8
Form of Participation (multiple responses) – White papers 17 30.9
– Consultative groups 40 72.7 – Fieldworks 4 7.3 – Round-table meetings 36 65.5 – Discussion forums 27 49.1 – Other 11 20.0 Table 2b
Participation to standard-setting process by graduation of the respondents.
Graduation Total
Accounting and Finance Others
Participation to due process Rarely 26.2% 15.4% 23.6%
Occasionally 23.8% 38.5% 27.3%
Frequently 31.0% 15.4% 27.3%
Very Frequently 19.0% 30.8% 21.8%
Total 100.0% 100.0% 100.0%
Table 2c
Participation to standard-setting process by profile of the respondents.
Profile of respondent Total
Credit analyst Equity sell side Equity buy side
Participation to due process Rarely 31.3% 9.1% 25.0% 23.6%
Occasionally 25.0% 18.2% 32.1% 27.3%
Frequently 25.0% 45.5% 21.4% 27.3%
Very Frequently 18.8% 27.3% 21.4% 21.8%
linked with one another using a system of simultaneous regression equations.
AsGefen et al. (2000)note,“SEM has become de rigueur in validating instruments and testing the linkages between constructs.” One can further distinguish between two families of SEM techniques: covariance-based techniques, as represented by LISREL, and variance-based techniques, of which partial least squares path modelling (SEM-PLS or PLS-PM) is the most prominent representative (Henseler et al., 2009). SEM-PLS is used by a growing number of researchers from various disciplines, such as accounting (Lee et al., 2011), strategic management (e.g.Hulland, 1999), management information systems (e.g.Dibbern et al., 2004), e-business (e.g., Pavlou and Chai, 2002), organizational behaviour (e.g.,Higgins et al., 1992), marketing (e.g.,Reinartz et al., 2004) and consumer behaviour (e.g.,Fornell and Robinson, 1983).
The advantage of using SEM-PLS instead of a classical covariance-based estimation procedure is based on certain peculiar characteristics of SEM-PLS; the PLS algorithm allows the unrestricted computation of cause-effect relationship models that employ both reflective and formative measurement models (Diamantopoulos and Winklhofer, 2001). PLS can be used to estimate path models when sample sizes are small (Esposito Vinzi et al., 2007). PLS path models can be very complex (i.e., consist of many latent and manifest variables) without leading to estimation problems (Wold, 1985). Furthermore, PLS path modelling can be used when distributions are highly skewed or data are not normally distributed, because the algorithm has no distributional requirements.
PLS path modelling includes two types of outer models: reflective (Mode A) and formative (Mode B) measurement models. The selection of a certain outer mode is subject to theoretical reasoning.
The most commonly used latent variable measurement model is the reflective mode, in which co-variation among the measures is caused by– and therefore reflects – variation in the underlying latent factor. The direction of causality is derived from the construct of the indicators, and changes in the underlying construct are hypothesized to cause changes in the indicators; thus, the measures are referred to as reflective or effect indicators (Jarvis et al., 2003). Reflective indicators of a latent construct should be internally consistent, and because all of the measures are assumed to be equally valid indicators of the underlying construct, any two measures that are equally reliable are interchangeable. Typical examples of appropriate applications of the reflective indicator model include constructs such as attitudes, which usually are measured on multi-item scales with endpoints such as good-bad, like-dislike and favourable-unfavourable.
In the formative mode, changes in the measures are hypothesized to cause changes in the underlying constructs. This approach assumes that all measures have an effect on (or a causal link to) a single construct. The direction of causality starts with the indicators and moves to the latent construct; as a group, the indicators jointly determine the conceptual and empirical meaning of the construct.
5.2. PLS path modelling estimation and assessment
The analysis has been performed with the PLS-SEM Toolbox 2.4 (Aria, 2015), which was developed in the Matlab/Octave lan-guage and is freely available from the Matlab File Exchange repository.
We chose a structural model with a path-weighting scheme. The path-weighting scheme is the most adequate because it is the only scheme that considers the fact that the relationships between the latent variables are specified as being direct (as in our theoretical model) (Lohmoller, 1989).
The results of the model assessment are shown inTables 3–6, and the estimated path model is reported inFig. 2.
Circles inFig. 2represent latent variables (LV), and rectangles represent manifest variables (MV). Each arrow connecting a circle with a rectangle represents the relationship between a theoretical construct (latent variable) and one of its measurements (manifest variable). Arrows connecting two circles represent a dependence link between two LVs. Using a structural equation model with a PLS approach, the model assessment consists of two main steps: the assessment of the outer model and the assessment of the inner model.
5.3. Outer model assessment
All measurement blocks have been configured as reflective models (Mode A). In this case, measurement models must be assessed with respect to their reliability and validity. Usually, thefirst criterion checked is the internal consistency reliability. The traditional criterion for internal consistency is Cronbach’s α (Cronbach, 1951), which provides an estimate of reliability based on the indicator inter-correlations. In PLS path modelling, Cronbach’s α is characterized by a severe underestimation of the internal consistency reliability of a reflective construct. Hence, it is more appropriate to apply two different measures of composite reliability, Jöreskog’s
Table 3
Outer model assessment.
Constructs N. of indicators Cronbachα Jöreskog's rho (ρc) Dijkstra-Henseler's rho (ρA) Average Variance Extracted (AVE)
Instrumental power (IP) 4 0.860 0.876 0.906 0.709
Symbolic power (SBP) 4 0.633 0.712 0.787 0.493
Systemic power (PA) 2 0.594 0.746 0.792 0.658
Pragmatic legitimacy (PL) 2 0.551 0.691 0.808 0.680
Moral legitimacy (ML) 4 0.741 0.832 0.838 0.580
Cognitive legitimacy (CL) 1 1.000 1.000 1.000 1.000
Personal behaviours (DT) 4 0.617 0.756 0.714 0.502
rhoρc(Werts et al., 1974) and Dijkstra-Henseler’s rhoρa(Dijkstra and Henseler, 2015).
An internal consistency reliability value greater than 0.7 in the early stages of research and values greater than 0.8 or 0.9 in more advanced stages of research are considered satisfactory (Henseler et al., 2016), whereas a value less than 0.6 indicates a lack of reliability. The composite reliability and Cronbach’s α values (as reported inTable 3) confirm good internal consistency for all constructs.
Table 4
Discriminant validity. Squared bivariate correlations between constructs. Constructs Instrumental power Symbolic
power Systemic power Pragmatic legitimacy Moral legitimacy Cognitive legitimacy Personal behaviours Instrumental power 0.709 Symbolic power 0.101 0.493 Systemic power 0.111 0.327 0.658 Pragmatic legitimacy 0.096 0.364 0.155 0.680 Moral legitimacy 0.014 0.608* 0.303 0.363 0.580 Cognitive legitimacy 0.011 0.133 0.170 0.059 0.275 1.000 Personal behaviours 0.152 0.215 0.220 0.232 0.170 0.108 0.502 Participation 0.041 0.031 0.000 0.030 0.002 0.110 0.051
Fornell–Larcker criterion Satisfied Satisfied Satisfied Satisfied Not Satisfied Satisfied Satisfied Diagonal elements represent AVE.
* Indicates a squared correlation not satisfying FL criterion. Table 5
Discriminant validity. Heterotrait-Monotrait Ratio of Correlations (HTMT). Constructs Instrumental power Symbolic
power Systemic power Pragmatic legitimacy Moral legitimacy Cognitive legitimacy Personal behaviours Instrumental power Symbolic power 0.391 Systemic power 0.471 0.890 Pragmatic legitimacy 0.386 0.881 0.870 Moral legitimacy 0.132 0.879 0.890 0.900 Cognitive legitimacy 0.110 0.427 0.569 0.370 0.626 Personal behaviours 0.374 0.514 0.843 0.833 0.543 0.399 Participation 0.212 0.280 0.170 0.179 0.180 0.332 0.216 Table 6
Discriminant validity. Loadings and Cross loadings. Constructs Indicators Instrumental Power Symbolic Power Systemic Power Pragmatic Legitimacy Moral Legitimacy Cognitive Legitimacy Personal Behaviours Participation IP_1 0.782 0.152 0.166 0.200 −0.013 0.054 0.133 0.137 IP_2 0.915 0.318 0.309 0.294 0.159 0.118 0.377 0.250 IP_3 0.916 0.286 0.347 0.259 0.120 0.194 0.374 0.185 IP_4 0.740 0.266 0.255 0.257 0.096 −0.029 0.360 0.077 SBP_1 −0.034 0.454 0.115 0.199 0.378 0.038 0.047 −0.004 SBP_2 0.327 0.703 0.403 0.412 0.501 0.174 0.324 0.167 SBP_3 0.354 0.888 0.517 0.511 0.646 0.378 0.530 0.193 SBP_4 0.154 0.693 0.454 0.477 0.582 0.331 0.276 0.093 PL_1 0.070 0.420 0.329 0.730 0.569 0.317 0.356 −0.040 PL_2 0.369 0.547 0.324 0.910 0.455 0.131 0.423 0.256 ML_1 0.095 0.243 0.356 0.203 0.409 0.295 0.164 −0.194 ML_2 0.221 0.746 0.579 0.564 0.867 0.298 0.408 0.231 ML_3 0.022 0.605 0.421 0.547 0.877 0.624 0.402 −0.082 ML_4 0.002 0.605 0.277 0.386 0.794 0.370 0.192 0.001 CL_1 0.103 0.358 0.405 0.239 0.515 1.000 0.323 −0.326 DT_1 0.361 0.447 0.552 0.381 0.391 0.329 0.856 0.227 DT_2 0.110 −0.057 0.235 0.221 0.005 0.029 0.568 −0.017 DT_3 −0.257 −0.001 0.014 0.123 0.249 0.220 0.385 −0.007 DT_4 0.443 0.411 0.187 0.375 0.229 0.142 0.635 0.187 PA_1 0.344 0.478 0.888 0.279 0.420 0.384 0.327 −0.053 PA_2 0.158 0.439 0.725 0.378 0.483 0.256 0.458 0.072 PT_1 0.198 0.174 −0.003 0.171 0.041 −0.326 0.221 1.000
Some authors recommend eliminating reflective indicators that have a loading smaller than 0.4. Considering PLS’s characteristics, only if an indicator’s reliability is low and eliminating this indicator results in a substantial increase of composite reliability does it makes sense to discard this indicator. Following such a method, indicators were reduced to 22 (seeTable 3).
For the assessment of validity, two complementary aspects are usually considered: convergent validity and discriminant validity. Convergent validity refers to the fact that a set of indicators represents one and the same underlying construct.Fornell and Larcker (1981)suggest using the average variance extracted (AVE) as a criterion of convergent validity. An AVE value of at least 0.5 indicates sufficient convergent validity, meaning that a latent variable is able to explain more than half of the variance of its indicators on average. The results inTable 3confirm good convergent validity, with all values being close to or greater than 0.5.
Discriminant validity implies that the joint set of all indicators is expected not to be unidimensional. In PLS, this aspect is evaluated via three measures: the Fornell-Larcker criterion, the heterotrait-monotrait (HTMT) ratio of correlations and the cross-loadings. Thefirst of these measures states that a construct must explain more variance in its indicators than in the other latent variables. In other words, in statistical terms, the AVE of a construct should be greater than the latent variable’s highest squared correlation with any other latent variable. The second criterion measures the validity as the ratio of the heterotrait correlation (HT, the average correlations of indicators across constructs measuring different phenomena) and the monotrait correlations (MT, the correlations of indicators within the same construct) for each construct (Henseler et al., 2015). The authors suggested a ceiling value of 0.90. The third criterion implies that the loading of each indicator is expected to be greater than all of its cross-loadings (Chin, 1998).
Following the Fornell-Larker approach (Table 4), all constructs except“Moral legitimacy” satisfy the criterion, whereas for the HTMT test, no indicator has a value greater than the threshold (Table 5).
Examining the cross-loading results (Table 6), the criterion is met by all indicators. The construct that does not meet the Fornell-Larken criteria, however, is characterized by indicators whose cross-loadings are all satisfactory. The columns inTable 6show the links, measured as correlations (or loadings) between the items (questions) and their constructs. In particular, among personal behaviours, item DT1 (interest in standard-setting issues), item DT4 (influence on the outcomes), and items DT2 (use of coalitions) and DT3 (loss of time) have the highest relevant loadings (0.856, 0.635, 0.568 and 0.385, respectively), being fundamental elements offinancial analysts’ attitudes explaining their frequency of participation in the due process.
Considering the legitimacy dimensions, the pragmatic legitimacy construct is supported by two salient factors. The loadings refer to the usefulness of accounting standards (PL1, loading 0.730) and the focus by the board on the analysts’ preferences (PL2, loading 0.910). Moral legitimacy is primarily explained by the existence of open public debates and justifications for the decisions made by the standard setter (ML2 loads 0.867), by the expertise of the board members (ML3 loads 0.877), and by the independence of board members (ML4 loads 0.794). With respect to cognitive legitimacy, only item CL1 (role played by accountants expertise) supports this construct.
Thus, the main items explaining instrumental power concern the position of analysts (item IP2, loading 0.915 in Table 6), adequate individual/group characteristics (item IP3, loading 0.916), and access to resources (item IP1, loading 0.782). The salient determinants of symbolic power are the importance of consultative committees (SBP3, correlates 0.888) and the characteristics of the due process (item SBP2, correlates 0.703). For the systemic power construct, both the power of the organization (PA1, loading 0.888) and the independence of the board (PA2, loading 0.725) represent salient factors.
5.4. Inner model assessment
The essential criterion for this assessment is the coefficient of determination R2of the endogenous latent variables.Chin (1998) describes R2values of 0.67, 0.33 and 0.19 in PLS path models as substantial, moderate and weak, respectively. If certain inner path model structures explain an endogenous latent variable by using only a few (e.g., one or two) exogenous latent variables, a “mod-erate” R2value might be acceptable.
The individual path coefficients of the PLS structural model can be interpreted as linear bivariate correlation coefficients, which are equivalent to the standardized beta coefficients of ordinary least squares regressions. Structural paths, whose signs are in keeping with a priori postulated algebraic signs, provide a partial empirical validation of the theoretically assumed relationships between latent variables. Paths that possess an algebraic sign contrary to expectations do not support the a priori formed hypotheses.
To determine the statistical significance of the results, confidence intervals and p-values for the path coefficients are obtained through a bootstrap resampling technique (Tenenhaus et al., 2005).
Table 7shows the R2values for the endogenous latent variables. In particular, the relationship between symbolic power and moral legitimacy exhibits substantial strength.
Table 8 exhibits the p-values of the path coefficients. A p-value ≤ 0.05 or ≤ 0.001 implies that a coefficient is significantly different from 0. There is only one coefficient that shows links that are not significant, specifically, “Moral legitimacy → Personal behaviours”. The next section is devoted to discussing results.
6. Discussion offindings 6.1. Inner model results
Consistent with Durocher et al.’s framework, our analyses confirm the validity of financial analysts’ motivations to participate in the due process. With the exception of the expected relationship between moral legitimacy and personal behaviour offinancial analysts, thefindings support our hypotheses.
In particular,findings inTable 8show that the previously identified non-legitimacy attitude and internalized motivations are the substantial factors that explainfinancial analysts’ higher participation; the coefficient is positive and statistically significant (coef-ficient 0.375; p-value 1%). These results provide strong support to our H1. These findings suggest that involvement is enhanced when analysts’ participation results in the development of accounting standards that are adopted and meet users’ needs. Interviews in the post-survey period clarified that the development of accounting standards that focuses on users’ information needs represents a salient motivation. Therefore, thisfinding thus indicates that analysts are not complacent in working together with board. The respondents confirmed that there is generally greater interest when proposals by the boards are more operational rather than con-ceptual (a case in point is the standard IFRS 9).
Moreover, the ability to successfully influence the content of accounting standards is not the drive of analyst involvement. Rather the hope to influence the development of accounting standards and the expectation to participate adequately are the main reasons stimulating greater mobilization.
Consistent with our results, respondents revealed that CFA members regularly meet with standard setters' board members and senior staff, as well as securities market regulators to suggest improvements for standards that are not under review. Also, seminars get organized with portfolio managers and analysts to hear their views on potential improvements and use these insights in the standard-setting process.
Interestingly, the respondents also noted that the frequency of participation occasionally depended upon the phases of the due process. Though the recent study conducted byPelger and Spieb (2017)revealed that investors’ participation during the agenda consultation appeared largely formal, our study documents that the early stages of the standard development are considered highly
Table 7
Structural or Inner model assessment: R Squares.
Independent variable Dependent variable R Square Strength
Instrumental power - > Pragmatic legitimacy 0.096 Weak
Symbolic power - > Moral legitimacy 0.608 Substantial
Systemic power - > Cognitive legitimacy 0.170 Weak
Pragmatic legitimacy and Moral legitimacy - > Personal behaviours 0.255 Moderate Cognitive legitimacy and Personal behaviours - > Participation 0.236 Moderate
Table 8
Inner model assessment: path coefficients.
Research hypotheses Direct effects Path coefficient P-value
H1 Personal behaviours - > Participation 0.375 0.006**
H2 Cognitive legitimacy - > Participation −0.456 < 0.001**
H3 Pragmatic legitimacy - > Personal behaviours 0.365 0.016*
H4 Moral legitimacy - > Personal behaviours 0.193 0.112
H5 Instrumental power - > Pragmatic legitimacy 0.310 0.018*
H6 Symbolic power - > Moral legitimacy 0.780 < 0.001**
H7 Systemic power - > Cognitive legitimacy 0.412 0.002**
** Indicates that a path coefficient between the two construct is significant at 1%. * Indicates that a path coefficient between the two construct is significant at 5%.
