Institutional complexity in professional service firms and dysfunctional behaviors: Evidence from a multi-country study

Pruijssers, Jorien

doi:https://doi.org/10.59876/a-b96y-y9fr

In many fields that once adhered to a dominant institutional logic, we now observe institutional complexity, characterized by the coexistence of two or more institutional logics. This phenomenon is evident in various organizational fields worldwide, ranging from commercial banking to college textbook publishing, from health care to courts, from food cooperatives to accounting firms, from business schools to symphony orchestras, from microfinance organizations to multinational energy corporations, and from reinsurance trading to film production (Andersson & Liff, 2018; Battilana & Dorado, 2010; Dunn & Jones, 2010; Faulconbridge & Muzio, 2016; Høiland & Klemsdal; 2022; Marquis & Lounsbury, 2007; McPherson & Sauder, 2013; Smets et al.,2015; Svenningsen-Bertelem et al.,2018; Thornton, 2002; Thornton & Ocasio, 1999). Firms operating in such fields face multiple—sometimes competing or contradictory—institutional prescriptions and compliance pressures (Greenwood et al., 2011). Consequently, managing firm-environmental relations in these fields has become more intricate than when monolithic institutions exerted singular compliance pressures on firms.

While prior research used to consider institutional complexity as a temporary and problematic phenomenon (Reay & Hinings, 2005), recent studies recognize that it can be sustained and may even be desirable (Smets et al., 2015). Embracing institutional complexity enables organizations to strategically maintain their institutional legitimacy by balancing varying institutional demands (Greenwood et al., 2011; Smets et al., 2015). By keeping different parties satisfied and involved (Smets et al., 2015), organizations can stabilize associated resource flows (Jay, 2013). To reap such benefits, organizations worldwide seek a cooperative approach to differing institutional demands by combining certain organizational practices prescribed by the different logics at work, a process known as selective coupling (Mair et al., 2015; Pache & Santos, 2013; Reay & Hinings, 2009). The resulting sustained institutional complexity, therefore, refers to the durable and persistent influence of multiple logics conferred by the existence of practices embedded in different logics.

Institutional scholarship on institutional complexity has taught us much about which organizational practices organizations couple (Dunn & Jones, 2010; Pache & Santos, 2013) and somewhat less about the dynamics of this process—who does it and how (Smets et al., 2015). But we still know surprisingly little about the implications of institutional complexity internalized at the organizational level. We are only starting to understand how the resulting sustained complexity can affect individual behavior (Andersson & Liff, 2018; Ben Romdhane & Ben Slimane, 2018; Høiland & Klemsdal, 2022; Svenningsen-Berthelem et al., 2018). Institutional complexity being as widespread as it is (Greenwood et al., 2011), this is an important phenomenon to be studied. Institutionally complex organizations—those which internalize elements from different institutional logics by adopting sets of organizational practices that are consonant with those institutional logics (Binder, 2007; Smets et al., 2015; Voronov & Yorks, 2015)—provide the structural assurance that certain behaviors are legitimate within the bounds of one espoused logic even if they contravene the norms of another espoused logic (Muzio et al., 2016). I suggest that individuals may exploit multiple logics as resources or tools for personal gain (Andersson & Liff, 2018; Bevort & Suddaby, 2016; Høiland & Klemsdal, 2022; Perkmann, Philips & Greenwood, 2022), potentially fueling dysfunctional behaviors. I thus provide a more nuanced view of internalized institutional complexity.

The accounting industry is a natural laboratory for a field study on institutional complexity and dysfunctional behaviors. This field also offers an interesting international perspective due to the shared international standards and norms amongst accounting organizations, contributing to the homogenization of the field. Institutional complexity has been a fact of life in accounting for about 20 years as the professional logic durably coexists with the commercial logic (Malsch & Gendron, 2013; Suddaby, Cooper, & Greenwood, 2007; Suddaby et al., 2009). Accounting firms around the globe have managed to internalize and thus sustain institutional complexity by adopting practices from both logics (Cooper et al., 1996). This unique international setting allows to identify the behavioral implications of sustained complexity through organizational design without the noise that may exist in less well-established fields and organizations. This setting is even more interesting not only because of the highly normative nature of accountants’ behavior, but also because we may gain a new perspective on why accountants engage in behaviors that may violate the public trust (Suddaby et al., 2009). Dysfunctional behaviors are persistent in the accounting environment and may threaten the stability of the financial markets worldwide.

The results are based on survey responses collected from 247 accounting firms in Belgium, Germany, and the Netherlands. The findings show that perceived individual dysfunctional behaviors – decisions made aiming to maximize professionals’ own short-term interests rather than aiming to maximize the overall development of the firm (Coram & Robinson, 2017) – can be triggered by perceived institutional contradictions and be reinforced by perceived strategies of practice adoption, which can fuel intra-organizational tensions. Ultimately these dysfunctional behaviors can threaten audit quality and violate the public’s trust (Fiolleau, Libby & Thorne, 2018; Ishaque, 2021). In accounting, I observe two types of dysfunctional behaviors: Inappropriate partner behaviors which are “violations that have the potential to lead to an audit failure” (Carcello et al., 1996: 246) and reputation jeopardizing behavior which is the failure to exert sufficient effort to maintain the audit firm’s reputation (Greenwood & Empson, 2003).

This study makes two main contributions. First, it offers a refreshing perspective on the possible unintended behavioral consequences of sustained institutional complexity. Thus far, research has primarily been concerned with explaining how field- and societal-level logics reference macro-social orders (Friedland & Alford, 1991) and much less with analyzing the behavioral repercussions of these logics (Bevort & Suddaby, 2016; Empson, Cleaver, & Allen, 2013; Greenwood et al., 2014). As one of the few behavioral studies in the logics field, I reply to repeated calls for a stronger integration between behavioral and institutional research (Dunn & Jones, 2010; Greenwood et al., 2014; McPherson & Sauder, 2013; Smets et al., 2015; Voronov & Yorks, 2015). This study complements the nascent literature on the individual responses to institutional complexity (Anderson & Liff, 2018; Ben Romdhane & Ben Slimane, 2018; Bevort & Suddaby, 2016; Høiland & Klemsdal, 2022; Svenningsen-Berthelem et al., 2018) by demonstrating that actors in institutionally complex organizations can make the logics fit their own needs.

Second, I develop new insights for the literature on professional service firm (PSF) design and performance (Gabbioneta et al., 2014; Greenwood & Empson, 2003; Greenwood & Miller, 2010; Muzio et al., 2016). Despite the belief that PSF organizations are designed to foster professional integrity, I show that when institutional complexity is internalized at the organizational level, dysfunctional behaviors may result. In line with the work of Gabbioneta and colleagues (2013, 2014), my research supports the notion that professional institutional environments can themselves facilitate dysfunctional behaviors.

The remainder of the paper is structured as follows. First, I situate the study within the institutional complexity literature and motivate why institutional complexity may trigger possible unintended behavioral consequences. Second, I apply this theoretical framework to the accounting field and develop two hypotheses. The third section outlines the research design and data collection procedures, followed by a presentation of the results. Lastly, I discuss the findings.

Theoretical background

Institutional complexity

Institutional logics guide social action by providing “assumptions and values, usually implicit, about how to interpret organizational reality, and what constitutes appropriate behavior, and how to succeed” (Thornton, 2004: 70). Abiding by the cultural prescriptions of institutional logics provides organizations with external legitimacy and increases their chances of survival (Thornton et al., 2012). Most organizational fields do not face a single dominant institutional logic, but rather are confronted with multiple logics operating in concert (e.g., Greenwood et al., 2010). This gives rise to institutional complexity, with organizations facing multiple and possibly incompatible prescriptions (e.g., Kraatz & Block, 2008). Institutional theorists now recognize that institutional complexity has become the norm rather than the exception in nearly every organizational field around the globe (Greenwood et al., 2010).

In fields once dominated by a single monolithic logic, institutional complexity frequently develops at the field level due to changes in, for example, competition, client requirements, cost pressures or it originates within the organization due to factors like diversification strategies, the development of new practices, or an increasing scale of operations (Empson et al., 2013; Greenwood & Hinings, 1996). Increases in institutional complexity tend to follow the emergence of new logics or logics that were there only in latent and subdued forms. Hospitals, for example, used to face only the normative prescriptions of a professional logic, but now also face the demands of a commercial logic (Dunn & Jones, 2010). Mutual funds and British reinsurance trading organizations deal with market and community logics (Marquis & Lounsbury, 2007; Smets et al., 2015) and universities face professional, state, and market logics (Mair et al., 2015). Most organizations—professional organizations in particular—have now permanently espoused multiple, sometimes incompatible, institutional logics (Besharov & Smith, 2014; Greenwood et al., 2011; Jarzabkowski et al., 2013; Mair et al., 2015; Smets et al., 2015).

Benefits of sustaining institutional complexity

The enduring coexistence of competing logics creates the potential for internal tensions because the underlying assumptions, values, and cultural prescriptions of the logics sometimes conflict. While prior literature focused on strategies to eliminate complexity (e.g., Jarzabkowski et al., 2013; Kraatz & Block, 2008) or to symbolically deal with it through decoupling strategies (Meyer & Rowan, 1977; Westphal & Zajac, 2001), more recent studies suggest that organizations should reconcile and balance the demands of persisting multiple logics (e.g. Battilana & Dorado, 2010; Mair et al., 2015; Reay & Hinings, 2009; Smets et al., 2015) by selectively coupling (or combining) organizational practices from different institutional logics within an organization in order to seek a cooperative resolution (Mair et al., 2015; Pache & Santos, 2013; Reay & Hinings, 2009). Selective coupling would therefore strategically position the organization by internalizing institutional complexity in order to satisfy the variety of institutional demands and secure their endorsement (e.g., Pache & Santos, 2013).

Internalizing institutional complexity can indeed confer benefits. It allows an organization to draw from a broader repertoire of practices and to determine the most successful combination (Binder, 2007; Greenwood et al., 2011; Pache & Santos, 2013). Sustaining institutional complexity may also confer access to different resource pools (Jay, 2013). Tracey and colleagues (2011) showed that combining logics led to additional resources, which was beneficial for a social enterprise. Organizations may also benefit from enhanced innovation and important relationships in multiple domains when they internalize institutional complexity (Smets et al., 2015; Tracey et al., 2011). To derive these benefits, organizations should use different logics to govern different activities (Battilana & Dorado, 2010; Friedland & Alford, 1991).

An organization’s ability to combine practices embedded in different logics gives it flexibility in responding to complexity and sustaining it. Organizations will thus not adopt practices in a uniform way, but along a continuum of conformity to the demands of the various logics (Quirke, 2013). We may therefore observe variation in organizational design within the same field which can give rise to perceived variations in institutional complexity across organizations within the field (Besharov & Smith, 2014; Lounsbury, 2007, 2008). While we can observe multiple logics peacefully coexisting within an organization, this does not mean that they always do (Battilana & Dorado, 2010; Pache & Santos, 2013; Smets et al., 2015; Tracey et al., 2011). Sustained institutional complexity can sometimes have unintended consequences.

The possible hidden cost of institutional complexity: dysfunctional behavior

While organizations can benefit from sustaining institutional complexity through organizational design, their people must sometimes navigate a world of institutional contradictions and contradictory organizational practices. Practices are the material vehicles of institutional logics that convey desired organizational values (Battilana & Dorado, 2010; Lounsbury, 2007), shape individual interests, and frame the opportunities and constraints for individual behavior. In institutionally complex organizations, actors need to interpret—that is, make sense of—both the practices and the internalized complexity (Bevort & Suddaby, 2016; Binder, 2007). In this way, they internalize the logics which will make certain behaviors relevant and desirable.

People will therefore not be unidimensional on one logic but will construct a repertoire of behaviors (Bevort & Suddaby, 2016; Binder, 2007), the choice depending on the individual’s experience and interests (Binder, 2007). This repertoire of behaviors are resources or tools individuals can use to pursue their own goals (Bevort & Suddaby, 2016; Høiland & Klemsdal, 2022; Svenningsen et al., 2018). In an institutionally complex environment, this repertoire will most likely include behaviors that can be contradictory. I surmise that having potentially incompatible behaviors to choose from in any particular situation can open the possibility for dysfunctional behavior—behavior “that contravenes normative expectations” (Muzio et al., 2016: 4). The structural assurance provided by the organization means that certain organizational practices embedded in one logic can endorse—as safe and legitimate—behavior that may be inconsistent with the norms of another logic (Grey, 2003; Muzio et al., 2016; Smets et al., 2012). Because certain behaviors are considered right in one logic and wrong in another, individuals can draw from multiple logics for strategic purposes while simultaneously justifying their behavior (Perkmann et al., 2022; Svenningsen et al., 2018).

While the literature usually blames one logic for dysfunctional behaviors, I believe that it is sustaining complexity that can sometimes spur on dysfunctional behaviors because it can expose individual actors to possibly internalized contradictions. For example, academics and regulators have blamed the commercialization of accounting for observed deficiencies in the audits of publicly listed firms (Wyatt, 2004). While the commercial logic will trigger commercial behavior, this type of behavior is not necessarily “wrong” if exercised within the bounds of professionalism. Therefore, rather than blaming a single logic, it could be the durable coexistence of several logics in organizations that can sometimes facilitate dysfunctional behaviors.

Setting and hypotheses

Institutional complexity in the accounting field

The traditional logic in accounting is the professional logic, which assumes that auditors are social trustees with clear fiduciary duties to the public (Malsch & Gendron, 2013). This logic is premised on the ideology of public service, the mythologies of independence and autonomy, and an indifference to commercial matters. It consists of clear professional norms and values (Suddaby et al., 2009) and durably coexists with a commercial logic, which is premised on the quest for short-term profitability (Malsch & Gendron, 2013), the generation of revenues (Suddaby et al., 2009), and the importance of value added (Covaleski et al., 2003). The commercial logic has been blamed for shifting the nature of audit services from a public good towards a commercial service.

Both logics are important in accounting worldwide. Accounting firms depend on the professional logic for their social acceptance and raison d’être and depend on the commercial logic for their economic survival and have therefore internalized both logics. Their prescriptions, however, make contradictory demands. Within the customary fixed-fee audit contract, for example, the professional logic prescribes independence, competence, and accuracy while the commercial logic vies for cost reductions and time savings. Accounting firms are therefore caught in a web of institutional complexity. How they handle the resulting internal tensions and associated organizational challenges is crucial for their social legitimacy, their access to critical resources, and ultimately their survival—not to mention the world economy from time to time.

Sustained institutional complexity through organizational design

An exploration of the literature on institutional logics in accounting organizations suggests the pervasive nature of institutional complexity and the drive of accounting organizations to gain and retain legitimacy from both institutional logics by organizing in similar ways across the globe (e.g., Malsch & Gendron, 2013; Suddaby et al., 2009). The most common strategy for sustaining institutional complexity in accounting firms is to balance and reconcile the demands of both logics by adopting organizational practices from each logic. Despite the clash between professional practices based on collegial or peer controls and the more business-like organizational control practices prescribed by the commercial logic (Anderson-Gough et al., 2001; Cooper et al., 1996), accounting firms carefully combine practices to attend to both institutional demands. To do so, accounting firms seek to implement the organizational practices prescribed by the different logics in different areas of firm governance. Each accounting firm will adopt the practices slightly differently according to whether it places both logics centrally, giving them equal representation, or prioritizes one over the other. This gives rise to variation in organizational design (Besharov & Smith, 2014). Next, I will present practices from both logics that have been implemented to manage institutional complexity.

Organizational practices associated with the professional logic. Accounting firms have adopted practices embedded in the professional logic to guide audit practice. These are the values and beliefs best suited to deliver high-quality audits. They enforce professional conduct because they rely on collegial oversight (Cooper et al., 1996), which is believed to control the risk of unprofessional behavior by detecting and deterring it.

One such practice is the engagement quality control review (EQCR) or concurring partner review, which monitors audit quality by providing an objective review of an audit engagement by a partner not involved with the client (Epps & Messier, 2007). The goal of the engagement quality reviewer is that a partner independent from the engagement provides an objective evaluation of the audit conclusions reached by the engagement audit team before the date of the final report. The EQCR is deeply embedded in the professional logic because the International Federation of Accountants prescribes this practice. The International Standards on Quality Control, which 130 countries have adopted and enforced in their national audit standards, provides guidance on EQCR.

A second practice embedded in the professional logic is internal transparency, “the degree to which [partner] conduct and its consequences are perceptible to [other partners]” (Kaptein, 2008: 926). It is clear how internal transparency directly supports collegial monitoring, but it does more. The greater the internal transparency, the more the partners know about the ins and outs of the job, its ethical dilemmas, and the strengths and weaknesses of other partners, all of which can help them influence each other’s behavior.

Organizational practices associated with the commercial logic. To secure endorsement from actors promoting and defending the commercial logic, accounting firms adopted businesslike practices to guide partner compensation. Both anecdotal and empirical evidence suggest that accounting firms have replaced the traditional equal profit-sharing system (Morris & Pinnington, 1998) with an incentive and reward system that links compensation to performance (Coram & Robinson, 2017; Knechel, Niemi & Zerni, 2013, Vandenhaute, Hardies & Breesch, 2020). Accounting firms thus use commercial practices to govern activities relatively distant from activities governed by the professional logic.

Partner compensation today consists of a variable component dependent on individual performance evaluations (Coram & Robinson, 2017; Empson et al., 2013; Knechel et al., 2013). It is not only the adoption of the variable component, but also its relative weight - or importance - that affects audit firm partners’ incentives. Meeting the standards for performance evaluations now significantly affects their pay. Revenue-based performance evaluation, for example, allocates profit shares based on an individual’s revenue-generating success (Knechel et al., 2013). Partners are therefore motivated to focus on their personal contributions to the firm’s profits, their billable hours, and their ability to acquire and retain clients. In this way, the pursuit of material gain becomes a key incentive in accounting firms (Grey, 2003; Kornberger et al., 2011; Malsch & Gendron, 2013; Suddaby et al., 2009). Revenue-based performance evaluation is often heavily relied on in current profit-sharing schemes as elements such as fee generation or the acquisition of new clients are easier to measure than elements related to other aspects of performance such as quality control for instance (Coram & Robinson, 2017).

These commercial practices have been adopted not only to increase profitability, but also because other actors in the accounting field endorse and expect them. These practices have become institutionalized over time because of two main reasons (Suddaby et al., 2009). First, the nature of accounting work has changed due to market competition driving service diversification (audit and consulting services). Second, accounting organizations have also changed at a structural level to become more formalized and include more centralized decision-making, a stronger emphasis on strategic and financial planning, and the adoption of more formal organizational controls (Cooper et al., 1996).

The strength of presence of the professional and commercial practices will slightly vary between accounting firms according to how each firm responds to the institutional complexity (Besharov & Smith, 2014; Quirke, 2013). Figure 1 shows the conjectured positive association between the perceived presence of the professional and the commercial logics within accounting firms and their organizational response (that is, the strength of presence of the organizational practices consonant with each institutional logic). I therefore hypothesize:

H1: The perceived presence of the professional and the commercial logics within accounting organizations is associated with the perceived strength of consonant organizational practices.

Behavioral consequences of sustained institutional complexity

Partners face not only the internalized institutional influences at the organizational level, but also the direct signals of such organizational practices.

Behavioral effects of practices congruent with the professional logic. The practices embedded in the professional logic will prompt behaviors that foster audit quality, professional judgment, and the collegial monitoring necessary to keep up the firm’s reputation. Studies have established that the EQCR encourages professional behavior because it enhances risk assessments by engagement partners, leads to more testing, and reduces the focus on evidence confirming prior expectations (Ayers & Kaplan, 2003). Internal transparency helps make clear what the desired professional behaviors are and what happens—to the clients, the partners, and the firm—when they are violated (Kaptein, 2008, 2011).

Figure 1

**Conceptual model of the effect of the perceived presence of conflicting institutional logics on member perceived dysfunctional behavior**

Behavioral effects of practices congruent with the commercial logic. The practices embedded in the commercial logic trigger profit-maximizing behaviors. Both the importance of variable compensation and the revenue-based performance evaluation signal that the pursuit of material gain is important (Knechel et al., 2013). Given that partners have only one revenue source—their clients—they have only two ways to maximize their revenue: retaining existing clients and, even better, acquiring new ones.

Combined behavioral effects of sustained institutional complexity. By sustaining institutional complexity, accounting firm partners create a behavioral repertoire comprised of those professional and commercial behaviors that they can use to fit their needs. What are the combined effects? While the literature has suggested that the commercial logic alone is to blame for professional deficiencies that have taken place (Wyatt, 2004), I believe that the problem is the durable coexistence of the professional and the commercial logics. As reasoned above, partners have agency to strategically exploit the repertoire of available behaviors to make the logics fit their needs (Bevort & Suddaby, 2016) while the firm provides the structural assurance that certain behaviors are acceptable. Specifically, they may calculate (correctly) that certain unprofessional actions are very likely to increase their income and even be lauded. They may also calculate (correctly) that certain unprofessional actions are very likely to increase their income and much less likely to be discovered and get them in trouble.

In accounting firms worldwide, two types of dysfunctional behaviors are prevalent. Dysfunctional behaviors are not necessarily fraudulent, yet consist of decisions directed at maximizing professionals’ own short-term interests rather than maximizing the overall development of the firm (Coram & Robinson, 2017). Ultimately these behaviors can threaten audit quality and violate the public’s trust (Fiolleau et al., 2018; Ishaque, 2021). First, partners may engage in inappropriate behaviors. These are behaviors which are “violations that have the potential to lead to an audit failure” (Carcello et al., 1996: 246). This is no small problem: the national authority of the Dutch financial markets has pointed out that inappropriate partner behaviors account for 45 percent of the faults found in controlled Dutch public interest audits. Gathering insufficient audit evidence or performing an inadequate workpaper review (Carcello et al., 1996), for example, have been uncovered in the Dutch audit field and are behaviors that are diametrically opposed to the public function of the auditor as an independent gatekeeper.

Secondly, partners may engage in behaviors that jeopardize the firm’s reputation. Reputation jeopardizing behavior is defined as failing to exert sufficient effort to maintain the firm’s reputation (Greenwood & Empson, 2003; Hitt et al., 2006). Reputation is a very valuable asset to an accounting firm, enabling it to charge higher fees, attract new clients, and hire the brightest professionals (Greenwood & Empson, 2003; Greenwood et al., 2005). The cost to a firm of losing its reputation can be greater than the benefits of having it. Once it became publicly known that a handful of Arthur Andersen (AA) partners had acquiesced to Enron’s demands for fraudulent financial reporting, AA lost its reputation and consequently failed and, in fact, the entire accounting profession suffered a large loss in public confidence (Wyatt, 2004). Figure 1 summarizes the abovementioned associations. I therefore hypothesize:

H2: The perceived strength of sustained institutional complexity increases the propensity to engage in dysfunctional behaviors at the individual level.

Methods

Data collection and sampling

I relied on survey techniques to collect the data, not only because the data needed to test the hypotheses are not publicly available, but also because I am interested in perceptions. A survey enables me to understand field-level effects (as opposed to the single-firm ethnographies typical of institutional research). A Web-based survey was sent to all registered certified public accountants (CPAs) in the German, Dutch, and Belgian professional associations. The sampling frame consisted of the CPAs’ e-mail addresses available at the public registries of the national professional associations. I needed partners’ perceptions, because partners are the key organizational decision-makers, they are the target audience of organizational practices, and their behavior has a potentially large societal impact. The first page of the survey therefore filtered out CPAs not working for an accounting firm. The next page screened the hierarchical level: partner or salaried auditor. 819 CPAs completed the survey, 414 of whom were accounting firm partners (207 German, 111 Dutch, and 96 Belgian). The others were salaried professionals and were presented with a different set of questions intended for another study. The respondents were partners of the Big Four firms in all three countries and of 235 other public accounting firms. I was interested in firm-level variables, so in the case of multiple respondents per firm, I aggregated their responses to the firm level by averaging them; intraclass correlation coefficients (ICC, 2, 1) provided statistical justification. The final sample therefore consisted of 247 accounting firms.

The three countries in the study are similar on four counts. First, about 85 percent of the publicly listed firms are audited by the Big Four audit firms (Barnier, 2010). Second, the mid-tier audit market serving non-listed firms has a strong national presence[1]. Third, the accounting profession in all three countries is formally monitored by independent oversight bodies. Fourth, all accounting firms in the sample are subject to code law and similar levels of regulatory enforcement, making these countries comparable at an institutional level. Thus, aggregating the data from Germany, the Netherlands, and Belgium should not be an issue.

Instrument development

The survey consisted of eight psychometric scales. I used two existing scales validated in prior research and developed six, following canonical scale-development procedures (Hinkin, 1998). I used literature reviews and expert interviews to generate a pool of theoretically precise items. Content validity was assured by developing theoretically precise items. After item reduction, I used between five and eight items per construct in order to minimize response biases without compromising construct validity (see Appendix). The items for each scale were presented in a random order to avoid order effects.

Given the large number of items and the need to use structural equation modeling, I use a partial disaggregation strategy to link the indicators to their corresponding constructs (Little et al., 2002). Parceling is preferred to the use of single indicators not only because single indicators often have lower reliability and smaller common-to-unique variances, but also because parcels reduce the number of parameters to be estimated (Little et al., 2002). I created three parcels for each latent variable, based on the domain representativeness approach.

All items were measured on seven-point Likert scales. The two dependent variables capture dysfunctional behaviors in the accounting setting. Given the highly sensitive nature of these variables, prior studies in the accounting setting have concluded that there is likely to be substantial underreporting of dysfunctional behavior (Carcello et al., 1996). Therefore, a more valid measure consists of asking respondents not to rate their own behavior but to rate the observed behavior of the other partners. Inappropriate partner behaviors (α: 0.93) is an existing five-item scale (Carcello et al., 1996: 246). It consists of items like “The amount of audit evidence was insufficient” and “An inadequate partner review of a critical area of the engagement work papers was performed.” Reputation jeopardizing behavior (α: 0.74) is a six-item scale capturing the degree of reputation-jeopardizing behaviors in accounting firms. It consisted of items like “Partners in my firm put their personal interests before the collective reputation” and “In my firm, partners treat clients in a way that could damage the collective reputation”.

Two constructs serve as independent variables and are proxies for the institutional logics in accounting. Professional logic (α: 0.67) is a seven-item scale inspired by individual-level scales of professional commitment (Suddaby et al., 2007) and organization-level professionalism items developed by Sorensen and Sorensen (1972) and Shafer (2002). I define it as the relative identification of the firm with the goals of the profession. It consists of items like “What is best for the welfare of my firm is also best for the welfare of the profession” and “My firm values professional standards over profit considerations”. Commercial logic (α: 0.78) is a six-item scale assessing the degree to which respondents perceived that their firm was primarily concerned with profits. It included items like “My firm is primarily concerned with generating profits” and “The bottom-line orientation of my firm is the generation of profits”. I also included a higher-order latent variable, complexity, which consisted of both professional orientation and commercial orientation.

Four constructs serve as mediating variables capturing the extent of certain practices in the accounting firm. Two of those constructs tap into practices embedded in the professional logic. Quality of engagement quality control reviews (α: 0.82) is an eight-item scale based on Gibbins and Trotman’s (2002) and Epps and Messier’s (2007) work. The EQCR has been associated with increased audit quality given certain conditions (Gibbins & Trotman, 2002): reviewers should complete a thorough and all-encompassing review (Agoglia et al., 2003), and should then proceed to face-to-face reviews (Brazel et al., 2004). Respondents were asked to rate the extent to which reviewers in their firm reflected those qualities by asking whether reviewers “provide clear and complete feedback” and “have a good understanding of the client and industry.” Internal transparency (α: 0.85) is a six-item scale developed by Kaptein (2008). It included items like “Management is aware of the type of incidents and professional misconduct that occur in the immediate working environment of partners” and “If a partner does something which is not permitted, I or another colleague will find out about it.”

Two other constructs measured practices embedded in the commercial logic. Importance of variable compensation (α: 0.80) is a five-item scale assessing the importance of the variable component of the compensation package. It consists of items like “Variable compensation is a substantial amount of total partner pay”. Revenue-based performance evaluation (α: 0.95) is a seven-item scale assessing the degree to which profit shares are awarded based on revenue generation. Respondents were asked to indicate to what extent “acquisition of clients” and “declarable hours” were important for the allocation of profit shares.

The survey was developed in English, then translated into Dutch for participants in the Netherlands, German for participants in Germany, and Dutch, French, and German for participants in Belgium. The researcher (Dutch national) translated the English instrument to Dutch. The Dutch instrument was then reviewed by two other Dutch academics not involved in the project. Translators made the French and German versions, each of which was then reviewed by two bilingual nationals active in the financial sector for accuracy and consistency.

Data validity and reliability

Confirmatory factor analyses established convergent validity: indicators loaded positively and significantly on their latent variables. This conclusion is bolstered by the values of the average variance extracted (AVE) for the latent constructs, which range between 0.4 and 0.8. I then assessed discriminant validity by comparing the fit of an unconstrained measurement model to that of a constrained measurement model. All possible pairwise comparisons were estimated and all constrained models demonstrated a poorer fit. Hence, discriminant validity is established, corroborated by a multicollinearity analysis which indicated that all variance inflation factors of possible pairwise comparisons were appropriate. Construct reliability was confirmed by appropriate Cronbach alphas and by composite reliabilities ranging between 0.7 and 0.9.

There is a risk of common method variance (CMV) because all the data are extracted from the same source (Podsakoff et al., 2003: 879). As an ex-ante precaution, I randomized the item orders of different scales. To avoid “capitalizing” on any individual’s response style, I ensured that the relationships I tested through the study vary in directionality. I also used post-hoc statistical techniques to assess CMV. First, the Harman’s single factor test rendered an eight-factor solution with the first factor accounting for only 21.9 percent of the variance (Podsakoff et al., 2003). This test provides diagnostic evidence of the absence of CMV but does not provide any evidence that accounts for method effects. Therefore, following Podsakoff and colleagues (2003), I followed the trait-and-method procedures. This presented a better fit to the data than the full measurement model (Δ χ² = 82.56; Δd.f. = 36; p < 0.001), attesting to the presence of CMV. To establish whether CMV was a real threat, I estimated the amount of variance explained by the common method factor. This proved to be 10 percent, well below the critical cut-off point of 25 percent. Therefore, I conclude that the results cannot be explained by CMV.

Univariate analyses

Because the sample draws on three different countries, a one-way ANOVA assesses the differences in means of the variables. Table 1 presents the descriptive statistics per country as well as the post hoc tests. First, with regards to the presence of the professional and commercial logics, German accountants (m = 5.21, s.d. = 0.73) report a lower score on the strength of the professional logic in their organizations compared to their Belgian (m = 5.63, s.d. = 0.78) and Dutch (m = 5.52, s.d. = 0.64) counterparts. Yet, as the overall score is high (m=5.38, s.d.=0.74), there is reasonable assurance that the majority of accounting professionals still report a high commitment of their firms to the values of professionalism. These findings are in line with Suddaby et al.’s (2009) study set in Canada. As expected, the commercial logic is well embedded in the accounting organizations of the three countries with an average score of 4.17 out of 7 (s.d. = 1.17). Belgian accounting firms (m = 3.39, s.d. = 1.19) score lower than German (m = 4.44, s.d. = 1.04) and Dutch (m = 4.29, s.d. = 1.11) accounting firms. Overall, the clear presence of both logics across countries provides reasonable assurance that institutional complexity is perceived as a pervasive reality in the field.

Second, while the importance of variable compensation presents no statistically significant differences between countries, there are slight differences in the strength of presence of the three other organizational practices. Dutch accountants (m = 5.69, s.d. = 0.76) report a stronger presence of internal transparency in their firms than their Belgian counterparts (m = 5.20, s.d. = 1.16). Engagement quality control reviews are also stronger in the Dutch accounting setting (m = 5.16, s.d. = 0.80) than in German accounting organizations (m = 4.85, s.d. = 1.01) and in Belgian accounting organizations (m = 4.64, s.d. = 1.03). With regards to the presence of revenue-based performance evaluation, Belgian accountants report a stronger presence in their firms (m = 4.06, s.d. = 1.20) compared to the Dutch (m=3.61, s.d. = 1.89) and the German (m = 3.21, s.d. = 1.33). This further attests to the notion that accounting organizations selectively combine the available organizational practices and that variation in design therefore exists.

Third, the perceived dysfunctional behaviors are present across the three countries albeit to a limited extent. This would bias against finding any results. Inappropriate partner behaviors are similar across Belgium, Germany and the Netherlands. German accountants report that their organizations present slightly more reputation jeopardizing behaviors (m = 2.97; s.d. = 0.97) than in the Netherlands (m = 2.61; s.d. = 0.85). This type of perceived dysfunctional behavior is similarly present in Belgian accounting organizations (m = 2.78, s.d. = 0.92).

Importantly, across countries, all participants consistently report the presence of the measured variables. This is consistent with prior research (Suddaby et al., 2009) and speaks towards the international standardization aims of the accounting profession.

Results

To empirically test the theoretical model, I run a structural equation model, using LISREL 10.20 with maximum likelihood estimation routines. Table 2 presents the reliability and validity coefficients, and correlations of the latent variables. The partially mediated theoretical model fits the data well (χ ² = 508; d.f. = 235; RMSEA = 0.068; CFI = 0.93; GFI = 0.85). The ratio of chi-square to degrees of freedom is 2.16, below the cut-off point of 3. This further supports the good fit of the data to the structural equations. Table 3 presents the structural parameter estimates for the hypothesized model. Figure 2 depicts the structural model, including path coefficients.

TABLE 2

**Reliability and correlation coefficients**

TABLE 3

**Structural equation modeling results**

The data provide ample support for the theoretical predictions. All but five paths are significant and in the hypothesized direction. Hypothesis 1 is fully supported. I establish that the perceived presence of the professional logic has positive and significant paths to practices like engagement quality control reviews (β = 0.39; t = 4.30) and internal transparency (β = 0.37; t = 4.23). I also establish that the perceived presence of the commercial logic has positive and significant paths associated with practices like revenue-based performance evaluation (β = 0.23; t = 3.06) and the importance of variable compensation (β = 0.35; t = 4.44). I therefore conclude that the perceived presence of both the professional logic and the commercial logic within accounting organizations is associated with the perceived strength of consonant organizational practices.

Hypothesis 2, which predicts a partial mediation effect of these perceived logics and practices on perceived dysfunctional behaviors, is partially supported. I find strong evidence that the perceived presence of institutional complexity is associated with dysfunctional behaviors. The perceived presence of both logics has a significant influence on both types of dysfunctional behavior. I find significant negative paths between the perceived presence of the professional logic and both (a) inappropriate partner behaviors (β = -0.29; t = -2.92) and (b) reputation jeopardizing behavior (β = -0.28; t = -2.61). I also find a significant positive path between the perceived presence of the commercial logic and reputation jeopardizing behavior (β = 0.22; t = 2.51). While the path between the perceived presence of the commercial logic and inappropriate partner behaviors is in the predicted direction (positive), the path is insignificant. I further establish that all but one organizational practice significantly affect dysfunctional behaviors. The professional practices seem to reduce dysfunctional behaviors, as engagement quality control reviews have a negative path to inappropriate partner behaviors (β = -0.14; t = -1.89) while their effect on reputation jeopardizing behavior is insignificant. An internal transparency policy is associated with a negative path to reputation jeopardizing behavior (β = -0.27; t = -3.41). These findings indicate that specific organizational practices affect dysfunctional behaviors differently. The behavioral effect of the practices linked to the commercial logic is partially supported. Revenue-based performance evaluation has a positive path to inappropriate partner behaviors (β = 0.16; t = 2.51) and to reputation jeopardizing behavior (β = 0.11; t = 1.63). The importance of variable compensation does not have significant paths to dysfunctional behaviors.

In order to tease out the behavioral implications of sustained complexity, I run an additional model by adding a higher-order latent variable, complexity, to the structural model, which is presented in Table 4 and in Figure 3. The model fits the data well (χ² = 502; d.f. = 235; RMSEA = 0.068; CFI = 0.93; GFI = 0.86). Both the perceived presence of the professional logic (β = -0.47; t = -2.48) and the commercial logic (β = 0.29; t = 2.10) significantly load on complexity, which is significantly and positively related to inappropriate partner behaviors (β = 0.43; t = 3.84) and reputation jeopardizing behavior (β = 0.62; t = 3.59). The other paths remain significant and in the hypothesized direction as observed in the first model. I glean interesting insights from this model because it shows that the positive effects of the professional logic are nullified by the higher-order complexity variable. Additional analyses on a set of subsamples support the main findings.[2]

TABLE 4

**Structural equation modeling results including the higher-order complexity variable**

Discussion

The micro dynamics of institutional logics

Recent research has recognized that institutional complexity is a persistent condition in many fields (Besharov & Smith, 2014; Greenwood et al., 2011; Jarzabkowski et al., 2013) and has positioned complexity as a strategic opportunity rather than a constraint (Mair & Hehenberger, 2014; Mair et al., 2015). Institutional scholarship has recommended embracing institutional complexity in organizational design by balancing the prescriptions of the underlying logics at work (Pache & Santos, 2013; Smets et al., 2015). The literature has thus addressed the apparent disconnect between organizational reality and theory by shifting attention away from conflict and the ultimate dominance of one logic to the notion of the durable coexistence of multiple logics (Mair et al., 2015).

This study’s empirical efforts confirm earlier findings that organizations strategically seek to internalize and sustain institutional complexity by adopting practices consonant with different logics in different areas of firm governance (Smets et al., 2012; Thornton et al., 2012). In the accounting field, practices rooted in the professional logic are adopted to guide audit practice, while practices grounded in the commercial logic are adopted to motivate and reward partners. The impact of sustained institutional complexity on individual behavior is only beginning to be understood (Mair et al., 2015; Smets et al., 2015). Recent research has shown the crucial role of individual sensemaking and agency in interpreting and acting on institutional logics (Andersson & Liff, 2018; Ben Romdhane & Ben Slimane, 2018; Brevort & Suddaby, 2016; Høiland & Klemsdal, 2022; Perkmann et al., 2022). This study contributes to understanding how sustained institutional complexity can fuel dysfunctional behaviors. Complexity expands the range of available behaviors (Jarzabkowski et al., 2013; Smets et al., 2012), providing organizational actors with a toolkit of behaviors with which to enact specific logics in specific situations to meet their needs. The findings suggest that perceived institutional complexity can facilitate dysfunctional behaviors more than the perceived presence of logics separately. I thereby show that choice can be a problem. In a given situation, this broader repertoire can make it difficult to perceive the right behavioral response and easier to make a choice that qualifies as dysfunctional.

Figure 3

**Structural model Including the higher-order complexity variable**

Design challenges of professional service firms

Drawing on the literature on the design challenges of professional service firms (PSFs) (Greenwood & Miller, 2010; Greenwood et al., 2010), this study advances our understanding of the configuration of organizational practices in PSFs by highlighting the role of institutional complexity. Through a survey of accounting firms in different countries, I demonstrate that these firms have adopted organizational practices in response to varying institutional demands (Cooper et al., 1996). The findings suggest that the design of PSFs is crucial for their organizational performance and survival (Greenwood & Miller, 2010), as the right combination of practices can provide the foundation for professional and profitable behavior.

This study makes a particular contribution to the design challenges faced by PSFs in gatekeeping professions, such as accounting (Greenwood & Empson, 2003). The results also highlight the potential for organizational design to foster dysfunctional behavior (Gabbioneta et al., 2013). The findings reinforce Suddaby et al’s (2009) observation that the organizational context is important in shaping how professionals interpret and understand institutional logics and how this translates into behavior. The potential unintended behavioral consequences of institutional complexity in the gatekeeping accounting profession are real and can have global repercussions on the stability of the financial markets. The study therefore provides a more nuanced view on institutional complexity in the accounting profession and suggests that rather than the commercial logic alone, it may be the presence of dual logics that fuels dysfunctional behaviors.

Limitations and future research

Next to the usual limitations of survey research, the survey was relatively short in order to maximize the response rate. This kept me from fully capturing the organizational design and its variations as well as any country-specific differences (Lounsbury, 2007, 2008; Marquis & Lounsbury, 2008). To gain a comprehensive understanding of how institutional complexity affects organizational design, in-depth field studies are necessary. Such studies would also be insightful in understanding the drivers of observed variations in organizational design in institutionally complex fields. The literature would also benefit from further investigations into the stability of these designs over time. In addition, research could glean interesting insights from looking at the relative impact of institutional, organizational and cultural forces on individual behavior.

While this study focuses on dysfunctional behavior in the normative organizational context of accounting, investigating different fields and behaviors could provide more nuanced insights. The accounting context is characterized by severe conflicts of interests and independence concerns given the gatekeeping function of accountants in the financial markets worldwide. These specificities hinder drawing general conclusions from the study, leaving considerable opportunity for future research to extend its contributions. Future research could investigate the impact of institutional complexity in diverse organizational settings that are less regulated and not bound by professional norms, such as technology startups or social enterprises. In these contexts, multiple logics may be present, but not necessarily in the same way as in the normative context of accounting, resulting in potentially different effects on behavior.

Abstract

Résumé

Resumen