Fighting corruption with transparent organizations: Anti-corruption and functional deviance in organizational behavior
Thus every Part was full of Vice,
Yet the whole Mass a Paradice.
Bernard Mandeville, The fable of the bees
Compared with the large body of theoretical and empirical work on corruption, anti-corruption still is a somewhat neglected topic in contemporary social sciences. Particularly with regard to organization studies, there seems to be a knowledge gap on how anti-corruption programs affect the structures of formal organizations. Over the last two decades, social scientists have in fact investigated anti-corruption, but mainly by analyzing the worldwide spread of the transparency regime. These discussions were mainly reactions to ‘principal-agent’ models of corruption, corresponding incentive-based models of institutional design to curb corruption, and attempts to draw a political map to point to the institutional and ideological sources of this particular framing of corruption. Thus, the discourse on anti-corruption was dominated by economists and lawyers on the one hand and political scientists and social critics on the other.
In this paper, I want to argue, first, that all these approaches, although contributing in important ways to our understanding of anti-corruption programs, pay only relatively little attention to potential organizational side effects of such measures. Second, I want to show that the concept of ‘functional deviance’ in organizational behavior is a good starting point for bridging the gap between analyses of anti-corruption and organizational studies. Here, the argument is that organizations regularly rely on a certain degree of deviance to hide behavior that is formally forbidden but functional for the organization’s adaptability and flexibility. Third, I will present a case study as empirical evidence of how the introduction of transparent, i.e. accurately observable, organizational structures can lead to both enhanced intransparency and ineffectiveness. Fourth, I want to outline a theoretical discussion on some difficulties in distinguishing functional deviance and corruption in organizational behavior.
Anti-corruption in organizations: Mapping the field
Current social scientific debates about anti-corruption programs are, by and large, dominated by three disciplines: economics, law, and political science. Not surprisingly, the leading role is played by economics, in particular by the so-called ‘new institutional economics’ and ‘agency theory’, broadly associated with this approach. Without going deeper into the concept, ‘agency theory’ is a model for analyzing the relationship between two parties, where one party, the principal (p), for whatever reason hires another party, the agent (a), to act on her behalf. The theory then offers ways of modeling such a relationship under conditions of imperfect information (p does not per definition know if a does what she was hired for) and incongruent interests between p and a, both parties being inherently self-interested (Laffont and Martimort, 2002). First developed by Becker and Stigler (1974), the basic idea is to define corruption as an exchange relationship where an entrusted agent betrays the principal for personal gains (see also Banfield, 1975: 587-588; Rose-Ackerman, 1978: 7; Shleifer and Vishny, 1993).
In this model, corrupt behavior is understood as a question of incentives – agents will betray principals whenever they can and whenever the lack of effective institutional mechanisms inhibits a close monitoring of the agent’s behavior. This is the background of Robert Klitgaard’s (1988) famous formula of the causes of corruption: ‘corruption = monopoly + discretion – accountability’. From this follows that corruption is mainly a consequence of weak monitoring and that anti-corruption programs should entail an organizational design that is based on a rather strict notion of transparency, accountability, and compliance.
Over the last decades, this concept was prominently taken up by international organizations putting corruption on their agendas and designing institutional mechanisms to curb corruption. So e.g. the World Bank defines corruption as the ‘use of public office for private gain’ (Campos and Bhargava, 2007: 9) and Transparency International (TI) as ‘the abuse of entrusted power for private gain’. This development gave rise to countless descriptions of the history and implementation of anti-corruption regulation, usually with a focus on either the nation-state level or on the big international players like the WTO, the UN, the EU, the World Bank, the OECD, or Transparency International (TI) promoting certain ideals of ‘good governance’ (e.g. Nichols, 1999; Pieth et al., 2007; Salbu, 2000; Webb, 2005).
Some political scientists went a step beyond the description of anti-corruption regulation in arguing that international policies increasingly converge in creating an anti-corruption ‘regime’ or a global ‘transparency norm’ (e.g. Krasner, 1986; McCoy and Heckel, 2001; Wolf and Schmidt-Pfister, 2010). Without going into conceptual details, political scientists regard the emergence of the global anti-corruption movement as an unintended side effect of economic liberalization, globalization of trade and democratization, bringing together diverse international players in promoting anti-corruption legislations (e.g. Moroff, 2010). These and similar critical accounts would argue that though the rhetoric of international anti-corruption frames the issue as a manifest ‘social pandemic’, the actual reason for putting it on a global policy agenda was a coalition of diverse actors and the convergence of their interests – facilitating global business, safeguarding institutional legitimacy, and NGOs like TI creating demand for anti-corruption expertise (Gephart, 2009: 10ff; Krastev, 2004). The problem of corruption is, thus, actively constructed by an emerging ‘anti-corruption industry’ (Sampson, 2010a) that benefits from offering solutions.
The second skeptic branch is leftist and post-structuralist social criticism that sees the current anti-corruption wave as a scheme for weakening the position of public officials and welfare states by framing state monopolies as a major source of extortion and corruption (Hindess, 2005; Ivanov, 2007). From a similar position, the anti-corruption movement and its advocates are sometimes accused of pathologizing political corruption, while simultaneously normalizing political influence through pecuniary incentives, e.g. by think tanks and corporate interest groups (Bratsis, 2003). Thus, the anti-corruption industry allegedly pictures an idealized, modernist vision of a potentially ‘corruption-free’ and autonomous politics, while political reality in capitalist societies is inherently and structurally interwoven with private and business interests.
The third skeptical position is neither actor-based nor deconstructionist, but methodological and held by those who criticize the proliferation of ‘corruption perception indices’. At the core of the methodological debate lies the observation that corruption perception indices (such as TI’s ‘Corruption Perceptions Index’ – CPI), are, in one way or another, heavily biased and, in extreme cases, also create perverse incentives to downplay corruption in order to ‘look good’ in international corruption rankings (Galtung, 2006; Jansen, 2005; Johnston, 2001; Razafindrakoto and Roubaud, 2005; Sampson, 2010b).
These three approaches certainly do not cover all facets of critique countering the anti-corruption movement and a much broader review of critical approaches can be found in e.g. Wedel (2012). Yet, most of those approaches, as illuminating as they might be, pay relatively little attention to effects anti-corruption programs have on the level of organizational behavior. This aspect is also missing in the currently intensifying discourse on corruption in and by organizations (e.g. Ashforth and Anand, 2003; Ashforth et al., 2008; Pinto et al., 2008).
Now anti-corruption initiatives cover a wide range of measures, not all of them concerning organizational reform directly, e.g. the regulation of organizational inputs and outputs, inter-organizational relationships, ‘awareness’ programs, disclosure policies, staff rotation, or raising salaries (Rose-Ackerman and Truex, 2012). Where anti-corruption does directly refer to intra-organizational design, it usually rests on two pillars. First, creating external and independent auditing bodies like anti-corruption agencies and, ideally, furnishing these bodies with a legal mandate to regularly monitor the internal procedures in bureaucracies and punish wrongdoers as severely as possible (ibid.: 14ff). The second pillar is what Rose-Ackerman and Truex (2012: 22ff) and Rose-Ackerman (1999) call ‘program redesign’ to reduce corrupt opportunities. In line with Klitgaard’s (1988) above-mentioned formula, this aspect of anti-corruption is directly targeted at organizational design of public institutions and basically means reducing the discretion and monopoly of bureaucratic decision-making and stating rules for public officials as clearly as possible.
The background of this ‘program redesign’ approach is the fear that disclosure as such does not provide for corruption-free bureaucracies, mainly because information asymmetry in principal-agent relationships enables agents to manipulate disclosed data, to disclose only secondary data and, most importantly, to disclose huge amounts of records that can only be interpreted with the agents’ help. Thus, organizational transparency has to be linked with steady and independent monitoring by anti-corruption agencies to ensure everyday compliance with regulations. To do so, proper organizational behavior has to be formulated as unambiguously as possible, not only because, in Klitgaard’s terms, discretion is just a permission to break the rules, but also because anti-corruption watchdogs need absolutely clear signals for when to bark. Eventually, the principal-agent approach to organizational transparency comes down to organizational compliance (Rock, 2007). It thereby follows a general trend in western societies that substitutes trust in professional decision-making for distrust in discretion and trust in external control, standard-setting, formalized auditing, and punishment in cases of failing to meet predefined targets (Hood et al., 2005; Hood and Heald, 2006; Philp, 2009; Pollitt et al., 1999; Power, 1997).
Despite this rich theoretical background, anti-corruption has not caught much attention on the level of its actual organizational effects, mainly because it seems to be yet another piece in the story of changing visions of governance, generally associated with trends like ‘audit society’ (Power, 1997) or ‘governing at a distance’ (Rose and Miller, 1992). Another reason for the relative disinterest of organizational sociology for anti-corruption initiatives could be called the ‘neo-institutionalist reservation’. Several studies suggest that it is far easier to create an anti-corruption agency than to maintain its political independence and keep it from becoming a toothless token institution without actual power to implement noble legislation (e.g. Batory, 2012; Lawson, 2009; Wolf, 2010). Organizational sociologists, especially those influenced by neo-institutionalism, can easily infer from this that anti-corruption initiatives targeting organizational change are more of a normatively legitimized ‘talk’ than actual ‘action’ and, more often than not, a globally called-for fashion in organizational ‘window-dressing’ (e.g. Reisman, 1979).
Still largely missing are empirical studies of changing organizational micro-structures caused by anti-corruption programs and corresponding theoretical concepts that take these changes seriously. Before presenting an empirical case, I will outline a theoretical framework that sees the current anti-corruption movement mainly as a bureaucratic reform set that runs the risk of paralyzing organizations by inhibiting not only corrupt, but also ‘functionally deviant’ behavior.
Theoretical framework: The concept of functional deviance
The most striking aspect of the transparency-accountability-compliance nexus is that it assumes that such measures increase the efficiency or legitimacy of agent behavior. On the one hand, it is assumed that information disclosure and opening professional practice to public scrutiny increases trust in and legitimacy of public institutions (see for an extensive review Albu and Flyverbom, 2013: 9ff.). Here, it is not directly argued that transparency increases organizational performance as this discourse is mainly concerned with information flows and the citizens’ ‘right to know what is going on’. Hood (2007: 196) calls this the ‘populist-particularist’ version of transparency, where citizens observe officials (in their public roles), but where officials are carefully restricted in observing citizens.
On the other hand, and more closely tied to the principal-agent vision of transparency and anti-corruption, there is the ‘Benthamian’ or ‘bureaucratic’ (ibid.: 196f.) version of transparency. Here it is assumed that agents will behave better if being monitored and that this, in turn, would enhance organizational performance by inhibiting corruption, embezzlement, sloth, lavishness, goldbricking etc. This discourse envisions the matching of principal interest and agent behavior via formalized rules, standardized audits and measurable performance targets and assumes that rule-following will not only lead to better observability, but to enhanced organizational performance. What follows from this is a suspicious view on organizational informality and the assumption that it not only fosters self-interested concealment, but also organizational ineffectiveness (Bennis et al., 2008; Wehmeier and Raaz, 2012). From this point of view, enhanced transparency, compliance, and accountability have only short-term effects on efficiency during an adjustment period or if the outcomes to be accounted for are themselves flawed. In short, in the bureaucratic vision of organizational transparency it is argued that transparency, rule-following, and efficiency are mutually self-enforcing mechanisms.
Interestingly, a more balanced view comes from principal-agent approaches themselves. Here, it is conceded that transparency, accountability, and compliance obviously create costs in terms of time and financial resources and that an anti-corruption program is effective only if its costs do not exceed the costs caused by corrupt behavior itself. From this point of view, there exists a ‘socially optimal’ level of corruption. However, it was argued that a socially optimal level of corruption could be defined only for a fixed point in time, while, in the long run, corruption feeds on itself. This means that a moderate, socially optimal level of corruption will produce tolerance for that level and will slowly influence people’s behavior – more and more severe corrupt behavior will be tolerated, ultimately resulting in a sub-optimal level (see Elster, 1989: 268ff). Eventually, this calls for a zero-tolerance policy in anti-corruption programs and basically means that such programs may produce costs for organizations, but that these costs will per definition be lower than the long-term costs of corruption. For Philp this is more than a principalist twist of the debate, because it ‘undermine(s) the incentive-based account of principal-agent analysis’ (2009: 51). It leads to the assumption that organizational costs of anti-corruption can be disregarded altogether and turns the economic incentive model into a quasi-legal one.
The centerpiece of this disregard for organizational costs and unintended consequences of anti-corruption is a suspicious view on organizational discretion. Rose-Ackerman and Truex (2012: 22f) admit that e.g. limiting discretion to fight corruption might produce costs, but, surprisingly, they do not think of the costs for organizations. Instead, ‘the costs are the loss in discretion that might have been used for beneficial purposes to sort out the most deserving beneficiaries or to punish only the most harmful behavior’ (ibid.: 23).
That discretion in organizational decision-making can have important functions for organizations and not only for beneficiaries of organizations, and that, consequently, reducing discretion can induce side effects harmful for the affected organization, is, from this perspective, rarely considered as a practical issue of organizational reform. The quote stated above epitomizes the view that ideal, corruption-free bureaucracies should function like automated machines: formal rules and formal rules only! However, it has to be emphasized that the tendency to oversee organizational costs of limiting discretion and insisting on compliance is not a theoretical necessity in principal-agent models of corruption. It nevertheless creates very tangible side effects when practically implemented – as I will show below.
Envisioning ideal organizations as following rules au pied de lettre is a very demanding concept, to put it mildly. First, diminishing discretion demands from institutional designers not only to formulate procedural principles, but to anticipate in small detail every conceivable eventuality of cases, since every regulation gap will be filled with discretion and discretion with corruption. Second, diminished discretion requires agents to halt and to ‘escalate’ decision-making processes as soon as standard procedures do not apply, even at the cost of inevitable and expectable delay. Work-to-rule, as this mode is often called by its critics, is known and strategically used for paralyzing organizations – a handbook case in organizational sociology (e.g. Collinson and Ackroyd, 2005). But while work-to-rule is a frightening scenario for practitioners and only used as a radical last step in collective action, in the principal-agent vision of anti-corruption it became something of an ideal to strive for.
Despite its now popular principal-agent framing, the rule-bound and discretion-averse vision of organizational behavior is not new – and neither are critical voices. The insight that discretion and informality speed up decision-making and fill inevitable regulation gaps was e.g. the cornerstone of many widely-read studies in early functionalist sociology of organizations. From its beginning, most authors in this field questioned the empirical validity of rational bureaucratic organization models by arguing that strict hierarchies, clearly defined duties, procedures and goals, and top-down command lines only represent the formal side of an organization, complemented by an informal order that breaks official rules and questions automated rationalization in a significant way and on a regular basis. Gouldner’s (1954) concept of a ‘Janus-faced’ organization stands for this insight as much as do concepts like the ‘organizational paradox’ (Selznick, 1948) or the ‘organizational dilemma’ (Blau, 1955) of combining formal and social structures or, more generally, the concept of ‘bounded rationality’ (March and Simon, 1958). Similar points were made both for industrial relations (e.g. Burawoy, 1979; Dalton, 1959; Roethlisberger and Dickson, 1939) and for knowledge-intensive administrations (Blau, 1968; Blau and Scott, 1962). In short, all these approaches argued that even with the most precise planning, management cannot get rid of informal social relations in organizations and is in fact dependent on them to keep the formal order from petrifying into an inflexible mechanism.
In a radical version, it is not only discretion, i.e. the formally unregulated sphere of organizational behavior, that plays an important role in speeding up routines, but also outright deviation from the formal procedures that can have positive effects for an organization’s adaptability and efficiency. One of the first accounts of this idea was elaborated by Bensman and Gerver (1963), while extensive reviews of newer studies working with similar approaches can be found in Vaughan (1999) and in Greve et al. (2010).
The meanwhile huge amount of theoretical models and empirical evidence for this argument was first theoretically refined and radicalized in an early work of Niklas Luhmann (1964), who is nowadays usually associated with very abstract and general social theory, but actually started his academic career as an administration scientist. Like many sociologists of his time, Luhmann was fascinated by the concept of latent functions, i.e. the idea that unintended and unrecognized consequences of actions can contribute to the adjustment or adaptation of social units – persons, groups, or wider social and cultural systems (Merton, 1957: 60ff). Already observed by Emile Durkheim and George H. Mead, a classic example of such unrecognized adaptive consequences is that rule-breaking and crime in fact have the positive function of emotionally uniting an otherwise fragmented community against the criminal and thus reassuring the normative structure of the social system – an argument not for a law-and-order policy, but for de-moralizing deviance. Luhmann (1964: 304) starts from there, but radicalizes the concept in saying that it is not only the exposure and shared condemnation of the deviant act, but already the deviant act itself that can have adaptive functions for a social system, in this case an organization.
Luhmann called this kind of formally illegitimate, but productive behavior in organizations functional illegality or functional deviance. His basic idea is that formal organizations are dependent on a variety of actions that are hard to transform into formal rules, mainly because formal rule making is bound to fully explicable, mutually consistent, legally flawless and steady procedures. Insofar as organizations are confronted with highly complex environments, a tight ‘work-to-rule’ compliance regime would significantly restrict the organization’s adaptability to the various, inconsistent, and rapidly changing demands of multiple costumers or clients. In short, the argument is that behavior that deviates from the formal or legitimate order of a social system can be functional, where ‘functionality’ refers to the system’s adaptability to an ever-changing and incontrollable environment.
Luhmann’s approach, especially from today’s often formalist perspective, is very radical in its permissiveness towards deviant behavior in organizations. First, and against some approaches listed by Vaughan (1999), for Luhmann the emergence of functional deviance is not an effect of specific, e.g. markedly strict or overly hierarchical, formal structures. Instead, he believed that every formal organizational order, as long as it strives for consistency and stipulation, will sooner or later begin to lag behind a changing environment that is more complex and dynamic than the organizational system trying to adapt. Functional deviance is, therefore, an essential and inevitable feature of formal organizations with largely unpredictable environments. This insight was partly due to theoretical considerations and partly meant as a critique of the ‘human relations’ (HR) movement in organization studies (e.g. Mayo, 1933). HR discovered informal and formally deviant behavior as a resource for management, which could raise productivity e.g. by allowing informal group cooperation, upward communication, and appreciative leadership. Yet, from Luhmann’s perspective, functional deviance cannot be integrated into the formal order after becoming aware of its positive functions. On the contrary: it derives its functions exactly from being latent and loses the function of keeping an organization adaptive when becoming part of the relatively inflexible manifest (read: formal) arrangement. A formally allowed-for functional deviance would no longer be deviant, but would turn into a plain formal structure, ready to by circumvented by new forms of deviance as soon as it proved to be as inflexible as the previous formal order.
Second, and as another aspect of its latent structure, functional deviance does not necessarily have an intentional base, secretly ‘designed’ or explicitly ‘agreed upon’ by members of an organization to compensate for the failures and pitfalls of the formal order. Though deviance is heavily dependent on personal trust and, in cases of unplanned disclosure, on elaborate forms of justification (e.g. Ashforth and Anand, 2003), functional deviance usually comes in the form of well-rehearsed and tacitly granted routines. The concept does not need to conceptualize organizational members as altruistic or, in whatever kind, emotionally attached to their organization’s adaptability. In fact, most examples Luhmann has in mind when speaking of functional deviance are primarily of use for deviating members themselves and also, though necessarily unacknowledged, for the organization at large.
One example he gives are ‘shortened communication lines’, where routine cases can be very quickly executed by deviating from some parts of the official procedure and, if necessary, ‘polishing’ them afterwards to sustain the impression of a correct handling (Luhmann, 1964: 190ff and 272ff). Another example is the production of ‘sugarcoated’ portrayals of organizations by their spokespersons (ibid.: 108ff.). Here, the spokesperson’s actual function can only be performed by latently deviating from the official one: formally designed to ‘communicate’ with the environment, the actual job is to deceive it and to shield the organization from external attacks on its image. This function is absolutely vital for an organization, but at the same time it is deviant, as it is hard to formalize it by officially forcing the spokespersons to gloss over the mess, i.e. to lie. The last example presented here is what Luhmann (1964: 220ff) calls ‘boundary positions’: persons with a relatively low status in the organizational hierarchy, but with a very sensitive intuition for the demands of the environment. The work experience of such persons, e.g. sales agents, helps organizations to identify upcoming problems much faster than by relying on formal superiors whose official job description entails screening the environment for alarm signals. Relying on boundary roles makes organizations more sensitive for changing demands of the environment, but, in turning them upside-down, deviates from and thereby questions the legitimacy of formal hierarchies and official duties.
In all these examples, Luhmann describes win-win situations: members, for whatever reason, break paralyzing rules and the organization can let them do so without needlessly broaching the issue and without shamefully confessing that the formal order has regularly been mocked. Thus, functional deviance is a much more common and a much less demanding arrangement than e.g. ‘productive resistance’ (Courpasson et al., 2012), where members break obsolete rules to intentionally save their beloved company. This is not to say that in the concept of functional deviance agency does not play a role – functionally deviant organizational members have to make a decision: play to the rules or bypassing them. However, a deviant act can be classified as functional for an organization even if the deviant actors act on purely self-interested grounds. So even if functionally deviant behavior rests on decisions, these decisions need not (but, of course, can) rest on ethical or altruistic motives in enhancing organizational flexibility. The functionality of a deviant act is a latent and emergent social structure, not deducible from a motivational base.
From this perspective, fighting corruption with transparency-, accountability-, and compliance-based programs is primarily problematic because, if successful, runs the risk of binding the affected organizations to a strict formal operating mode, discouraging not only deviant (read: corrupt), but also functionally deviant behavior. Put differently, the problem with anti-corruption programs based on principal-agent models is that they do not differentiate between corruption and functional deviance but frame all informal, i.e. intransparent, behavior as potentially corrupt.
Notwithstanding some rare exemptions (e.g. Hollinger and Clark, 1982), the concept of functional deviance was, in its narrow sense, neither theoretically, nor methodologically developed further after the 1960s, mainly because functionalism lost its role as the leading paradigm in the social sciences from the 1970s onwards. Luhmann and his students did develop organizational sociology further (e.g. Baecker, 1999; Luhmann, 2000), but left the strictly functionalist paradigm and replaced it with a focus on autopoietic decision-making. In this concept, however, questions of rule-following, deviance, formality, and informality do not play the prominent roles they used to do in a functionalist framework, though some researchers made considerable efforts in explaining limits, paradoxes, and the informality of organizational decision-making within Luhmann’s later perspective (Knudsen, 2006; Knudsen, 2012).
Apart from the paradigmatic shift in Luhmann’s own writing, another reason for choosing to work with his earlier, decidedly functionalist approach is that it fits well with current research that focuses on unintended side effects of making organizations transparent. Here it was argued, from the perspective of organizational anthropology, that making organizations transparent via standardized audits undermines trust in experts, measures only measurable formal procedures (and delegitimizes informality), and thus forces organizations to adapt to accounting and auditing techniques that focus on monitoring single performance variables, while not enhancing the efficiency of the organization as a whole (Strathern, 2000; Tsoukas, 1997). Thus, attempts to make organizations more transparent, accountable, and auditable are no neutral tools for increasing visibility, but actually have a performative aspect in changing organizations affected by such measures (Albu and Flyverbom, 2013; Garsten and Lindh de Montoya, 2008). More specifically, studies in critical accounting (Power, 1997; Power, 2003) and critical studies in public administration (Hood, 2007; Hood, 2010; Hood and Heald, 2006; O’Neill, 2002; O’Neill, 2006) have shown that making organizations transparent via auditing and accounting leads to elaborate forms of ‘blame-avoiding’ and reactant behavior. Classic examples are moving delicate decisions to intransparent venues, scapegoating, avoidance of record-keeping, defensive reporting, ‘gaming’ of reported performance numbers and similar (for an overview see e.g. Hood, 2007: 202ff).
What the concept of functional deviance adds to this strand of organizational science is that it focuses on organizational benefits of deviating behavior, i.e. on constellations where reactance to techniques of organizational transparency is not only of use for deviant individuals, but also latently for the organization’s flexibility under conditions of rigid formal structures caused by compliance-based regulation. Functional deviance can be used as a theoretical model for explaining why transparency and compliance can indeed yield results in cases of individual deviance, but have counterproductive side effects in cases of deviance accepted, tolerated, and ordered by organizations.
With this background, functional deviance is a fruitful theoretical model for interpreting empirical data on unintended organizational side effects of anti-corruption (e.g. Lennerfors, 2012). In the following section, I will follow this path and present a case on how, at first sight successful, compliance- and transparency-based anti-corruption efforts limited functional deviance in organizations and what organizational structures resulted from this.
Limiting functional deviance
Anti-corruption models that distrust discretion will be explicitly disapproving of deviance – functional or not. If functional deviance is defined as deviations from formal organizational rules that increase the flexibility and adaptability of organizations, three classes of unintended side effects can be expected from attempts to eradicate it. Following Hirschman (1991) and Hood (2007: 202ff) unintended side effects of making organizations transparent can be classified as futile, jeopardizing, and perverse. ‘Futility’ refers to merely ceremonial changes in re-structuring organizations to meet transparency demands. In the case at hand, this would mean anti-corruption as ‘window-dressing’ or ‘tokenism’. ‘Jeopardizing’ outcomes are those in which organizational structures are effectively changed (in the case at hand: made transparent for external observers), but at the cost of other important organizational values. Side effects can, finally, be classified as ‘perverse’ when they effectively undermine the intended outcome of organizational transparency.
Ruling out constellations of ‘tokenism’ or ‘futility’ (as they can, at least in principle, be tackled by ever stricter operational control), jeopardizing side effects of transparency-based anti-corruption arise when making organizations transparent effectively overshadows the value of organizational efficiency and flexibility. Here, anti-corruption can be expected to have structurally similar side effects as they have been described for accounting and auditing from a critical perspective. For one thing, it can be expected that inhibiting functional deviance can indeed force organizational members to strictly comply with formal regulation, but that this kind of compliance severely slows down decision-making processes (Pollitt et al., 1999). Transparency-based anti-corruption would turn to an end in itself and force organizations to invest considerable resources in internal monitoring – creating organizational legitimacy but not efficiency (Power, 2003). Another jeopardizing side effect likely to occur is an effective erosion of flexibility in organizational decision-making. Here it was argued that strict and encompassing operational control can indeed make organizations transparent for external observers, but at the cost of undermining the value of context-specific and case-sensitive judgment in the administrative process. Instead of creatively dealing with complex, risky and uncertain cases, administrators begin to embrace a box-ticking mentality towards decision-making (O’Neill, 2002; O’Neill, 2006) – leading to weak management and a fear to decide on issues not covered by formalized protocol.
The second, or ‘perverse’, class of side effects of transparency-based anti-corruption programs is a reaction to jeopardizing effects and is likely to occur when organizations are forced into a strict compliance and surveillance regime but are simultaneously expected to decide efficiently, quickly, and case-sensitively. In such cases, organizational members are forced to withdraw into the least observable parts of the organization – i.e. into small, trust-based, and like-minded cliques where (functionally) deviant behavior can still be pursued, but now in absolute secrecy. Such unintended side effects of making organizations transparent are perverse insofar, as they reinforce exactly those structures transparency-based reforms try to overcome: organizational intransparency.
Yet, the effects of such strict transparency-, accountability-, and compliance-based anti-corruption programs have rarely been investigated empirically. As stated above, this is due to the (perhaps qualified) conjecture that such programs are mostly mere ‘window dressing’ or do not address organizational structures at all. But there is also an obvious methodological issue: violations of formal orders in organizational behavior (functional or dysfunctional) are utterly delicate issues for those involved in it and, therefore, extremely hard to observe directly. To exemplify the hypothesis that the implementation of principal-agent models of anti-corruption inhibit functional deviance, cases are needed that describe intra-organizational processes of redesign and use an empirical framework that goes beyond merely analyzing changes in the formal organizational structure (like document analysis).
The only study I am aware of that meets both criteria is an extensive, qualitative interview-based study by Anechiarico and Jacobs (1996) who analyzed the effects a very strict anti-corruption policy had on local public administration units in New York City. The authors describe a very radical transparency- and accountability-based anti-corruption regime that did not primarily rely on visible compliance departments and regular audits, but rather on undercover investigations. A central role was played by the ‘Department of Investigations’ (DOI), a public anti-corruption agency that originally started as a plain comptroller’s office, entitled to detect financial misconduct and organizational shortcomings in local public administration in an ex-post fashion. From its beginning in the 1930s until the early 1970s, the effectiveness of the DOI’s work was largely dependent on the leeway and resources it was granted by the Mayor’s Office: some mayors used it as a most welcome driving force in anti-corruption crusades, others cut the DOI’s budget to keep political machines in place.
Since the mid-1970s, however, and partly as a reaction to the Watergate scandal, the DOI’s mandate was steadily widened and the agency was slowly entitled to executive powers. It was turned into a law-enforcement agency, primarily staffed not by accountants or administration experts, but by personnel largely coming from special police forces (ibid.: 75ff). From now on, the main fields of activity where not accounts and recommendations for institutional reform, but undercover and ‘sting’ operations, assignments of informants, and wiretapping to fight corrupt behavior in the city’s local bureaucracy.
However, the DOI’s activities were not only planned to be about detection, but also about the prevention of corruption in local administration. Therefore, the so-called ‘Inspectors-General’ (IG) system was introduced (ibid.: 84ff). Originally, IGs were members of city agencies, entrusted with investigating charges of corruption raised against their administrative unit, while the DOI supervised the agencies’ efforts in handling such cases. Though this was the first attempt to install anti-corruption officers at the agency-level, it soon became clear that IGs were in fact pawns of their agency’s heads. IGs became the prime example of anti-corruption as ‘window-dressing’: charged with investigating corruption, they actually investigated little more than absenteeism and sleeping on the job. So it was decided that IGs, to act independently, should be responsible only to the DOI. As planned, the IGs could act as spies: they had unrestricted access to all meetings and records of their agency, they could start undercover investigations and they were allowed to launch proactive investigations to test the integrity of their colleagues. Most importantly, they could do all that without the agreement of the agency’s head. Almost needless to say, the power of IGs to monitor was accompanied by increasingly comprehensive criminal laws ‘backed by threats of sanctions, including jail, fines, and job and pension forfeiture’ (ibid.: 62) and a far-reaching whistleblower-protection policy.
The anti-corruption program in NYC involved several other measures that cannot be illustrated in detail here. An additional aspect of preventing corruption that should be mentioned, however, is the introduction of very strict conflict-of-interest regulation in the local administration in NYC (ibid.: 45ff). To prevent nepotism or favoritism in procurement processes, officials were required to disclose in every detail their personal and their family’s income, property, and involvement in private business of whatever kind. Conflict-of-interest laws especially targeted non-governmental employment opportunities, which had become extremely restricted the tighter the DOI’s grip on local administration got.
Altogether, the effect of this anti-corruption reform movement was a ‘panopticon-like’ environment, where public officials should and in fact did assume being permanently observed by an invisible force in order to enhance compliance with formal regulations. Referring to Foucault’s approach (1995), the authors argue that the DOI proved to be an effective agency exactly because officials internalized its monitoring techniques. This is why the measures could go beyond mere ‘window dressing’: the IG systems made monitoring simultaneously pervasive and invisible, while more and more sanctions were designed to punish wrongdoers. Officials began to play strictly to the rules, simply because they did not know if they were watched and what rules they would break if not following protocol as strictly as possible. Anechiarico and Jacobs argue that while the internal structure of the DOI proved to be an extraordinarily effective tool to ensure individual compliance, ensuring compliance did not result in measurably less corruption. What could be observed, however, was that the effectively imposed ‘work-to-rule’ regime resulted in several unintended side effects, causing a loss in organizational efficiency.
First of all, the anti-corruption program caused severe delays in the decision-making process. This is mainly because functional deviance in the form of ‘shortened communication lines’ was effectively discouraged. Even the most routine decisions required multiple approvals to avoid any possibility of fraud, conflict of interest, or thievery (Anechiarico and Jacobs, 1996: 175f). No intuition or discretion was allowed. Also, this regime fostered a climate of caution, fear, and self-preservation. Knowing that every decision not covered by protocol could be interpreted as potentially corrupt, decision-makers followed rules as strictly as possible to exclude even the slightest suspicion of misconduct.
Second, the anti-corruption regime in NYC led to what Anechiarico and Jacobs (1996: 177) call ‘inadequate authority’. This means that the diverse monitoring bodies in NYC required agency heads to get approval even for secondary decisions, thus losing a considerable amount of power. The IG’s far-reaching monitoring assignments, combined with a strong whistleblower protection policy, went as far as creating competing power centers in an agency, leading to confusions about who had the final say in strategic decisions – the IG or the agency head. All these measures effectively undermined the agency head’s autonomy and authority, and resulted in a very defensive way of managing public administrations. Instead of flexibly and creatively adapting to ever-changing demands and to take risks, upper and middle-level bureaucrats were mostly dealing with impression management. They constantly had to ask themselves how a decision looked for the IG and if it could be interpreted as an attempt to act corruptly. In such a situation it was rational not to cut through ‘red tape’ but to reluctantly stick to obsolete rules (ibid.: 178f). In fact, the authors describe how the anti-corruption project in NYC led to the well-known bureaucratic phenomenon of ‘goal displacement’: instead of rationalizing administration, limiting discretion in fact created over-bureaucratization as the bureaucracy became ‘absorbed in monitoring itself’ (ibid.: 180). The monitoring process, though only a means in creating compliance, became the actual end of administrative processes.
Beyond massive delays in decision-making, goal displacement and defensive management, another harmful side effect of the anti-corruption project in NYC was poor morale and adverse selection in public administration (ibid.: 181ff). As the principal-agent model envisions a corruption-free administration to work like an automaton, bureaucrats are envisioned as cogs in a machinery. As an effect, especially well-trained professionals turned away from an administration that discouraged creative solutions, enforced strict compliance, and deliberately left no space for professional judgment. Furthermore, financial disclosure regulations forced officials to give an account of very private data. Finally, conflict-of-interest regulation and minute accounting of funds and employee time effectively erected barriers between organizations, especially between public and private organizations. All these aspects taken together created a climate that treated officials as potential criminals. Thus, the better-educated strata of administrators increasingly left public administration in search for jobs in quasi-public authorities still not under the aegis of the anti-corruption panopticon. Eventually, local administration in NYC had to rely on officials who could not get a job anywhere else.
The most striking side effect of NYC’s anti-corruption regime, however, was increased intransparency. On one occasion, the authors describe how city officials, in fear of wiretapping, used pay phones for official communication (ibid.: 89f). The crucial point here is not that officials used pay phones to arrange corrupt transactions, but to get the job done swiftly and to avoid charges by the DOI for unknowingly violating protocol. At least this anecdote should be worrisome for transparency-, accountability, and compliance-based anti-corruption programs. It is one thing to downplay efficiency and resilience in favor of compliance, but where measures to raise transparency and compliance lead to less transparency, the model becomes self-defeating in creating the above mentioned ‘perverse’ side effects. In this case, transparency-oriented anti-corruption created exactly those effects it claimed to eradicate.
Discussion: Functional deviance and organizational corruption
To sum up, the strict and panopticon-like anti-corruption policy in NYC created the expected ‘jeopardizing’ and ‘perverse’ effects exactly because the costs of deviant behavior were so high that it was avoided even if having positive effects in terms of flexibility and resilience – deviance was avoided even if being functional.
However, when using concepts like functional deviance to interpret such empirical cases, one has to be very cautious about the well-known pitfalls and limits of functionalist models in sociology. First, there is always the danger of crediting every kind of deviant behavior with some function for stabilizing a social system in an ex-post fashion. Thus, it is absolutely crucial to predefine expected functions as clearly as possible and to use empirical material that can, at least potentially, falsify a functionalist hypothesis. Admittedly, ‘flexibility’, ‘resilience’, and ‘efficiency’ are only first steps in defining the functionality of a deviant act, but more precise concepts are necessary.
Second, one should avoid the also well-known problem of Parsonian structural functionalism (Parsons, 1951) that worked with a fixed set of functions needed to sustain a social system and then inferred that a social system would fail to reproduce if not all functions were fulfilled properly. Instead, it seems more useful to work with the dynamic concept of functional equivalents (e.g. Gresov and Drazin, 1997). With Anechiarico and Jacobs it can be argued that bureaucracies do not simply stop to work when functional deviance is inhibited. Their pay phone example rather indicates that if there is no generalized trust in organizational discretion, this trust is sought in small, like-minded organizational cliques who support and ‘cover’ their members in cases of by-passing paralyzing regulation and who, due to panopticon-like transparency, have to do that in absolute secrecy (Dalton, 1959; Tichy, 1973). Thus, trust in closed, i.e. utterly intransparent, groups can be a functional equivalent for trust in discretion. A hypothesis could be that organizations bound to an extremely strict compliance regime would become simultaneously very transparent and very intransparent, at least if they are expected to make decisions and not only to demonstrate their devotion to formal correctness. Insofar, the ‘perverse’ class of side effects, in this case the intransparent decision-making processes in transparent organizations, need not be regarded as purely negative. In fact, they re-flexibilize organizations bound to inflexible compliance regimes – they are, in Hood’s (2007) terms, an antidote to the class of ‘jeopardizing’ side effects.
Re-flexibilization through intransparency is, however, a risky issue. If transparency regimes force organizational members into functionally deviant cliques, there also grows potential for corruption pure and simple, i.e. deviance for private gains only, to emerge in these closed and intransparent groups. In the pay phone case described by Anechiarico and Jacobs, bureaucrats circumvented protocol to get their job done properly without having to fear surveillance and blaming. But protocol can, obviously, also be circumvented because of psychological factors like frustration and without any latent contribution to the organization’s flexibility. Here, a growing body of empirical and experimental research shows that ‘when social-control agents push individuals into […] deviance by intrusive checking, they cause the individual to develop a deviant identity; that is, it causes them to think of themselves and act as wrongdoers’ (Greve et al., 2010: 81). It can be expected that social-control agents, such as the IGs in the NYC case, can turn functional deviance into corruption, i.e. into deviance from which only the wrongdoers themselves profit, and no longer their host organizations. This outcome is the more probable the more intrusive social-control agents behave. Such intrusiveness is fostered when accountability is understood in an egalitarian way (Hood, 2010: 998ff), i.e. where organizational members are not only held accountable by their designated superiors or visible auditing bodies, but where, as described for the NYC case, the process of accountability is, in a panopticon-like operational control, dispersed across the organization. The intrusiveness of social control agents can also be reinforced e.g. by using governance, risk, and compliance (GRC) software (e.g. Bamberger, 2010).
Having said that, it should be stressed that the concept of functional deviance does not presuppose an active interest in the organization’s flexibility. Thus, even deviance out of frustration can unintentionally break the deadlock of too rigid control as much as thoroughly well-meant functional deviance can, in the long run, turn into self-interested corruption when conducted in small, intransparent, and conspiring groups. What makes deviance functional is, therefore, not the motivational structure of deviant members, but its overall system-wide effects on an organization’s adaptability.
Third, both organizational sociology and the anti-corruption project should, more clearly than now, distinguish between organizational corruption and functional deviance. The model of functional deviance is tightly coupled with the phenomenon of corrupt organizations, i.e. the insight that corrupt activities can be undertaken not for the wrongdoer’s personal gain, but for the organization’s benefit – mostly with superior toleration, involvement, or enforcement (Pinto et al., 2008). But while this discussion offers good arguments for distinguishing corrupt organizations from organization-wide activities of self-interested corrupt individuals, it does not clarify the relationship between organizational corruption and functional deviance. Here, it is only suggested that organizational corruption is a case of ‘organizational citizenship behavior’ (OCB), i.e. ‘extrarole behaviors undertaken by individual employees for the benefit of colleagues or the organization as a whole’ (ibid.: 690) that has ‘gone bad’.
With Greve et al. (2010: 80ff) it can be argued that the line between functional deviance and organizational corruption is not a substantial one, but rather actively constructed by social-control agents, that it varies over time, and leads to inconsistency and arbitrariness in regulation – in short, incentives for hidden by-passes, clique-building, and, ultimately, intransparency. It would not be much of an exaggeration to argue that organizational corruption is functional deviance made bad – both manifestly and deliberately by legal ‘work-to-rule’ regulations and latently by its side effects. Elaborating on this point, the theoretical and empirical arguments regarding functional deviance seem to verify the hypothesis that strong social-control mechanisms might, as in the case of the daunted NYC officials, keep individuals from deviating, but ‘when misconduct is instead on behalf of the organization and at its behest, stronger organizational control leads to more misconduct’ (ibid.: 92). Individual compliance does not seem to add up to organizational compliance – quite on the contrary.
Forth, the concept of functional deviance does not specify on what organizational level a deviant act is allowed to occur to call it ‘functional’ for the organization. This leads to the interesting case where formally deviant behavior increases the adaptability and performance of an organization’s sub-unit, but does so at the expense of the adaptability and performance of the organization as a whole. Clandestine practices that smooth decision-making processes at the departmental level do not necessarily add up to the organization’s interests at large – an issue likely to, again, re-emerge on intra-departmental group levels. An even more difficult case seems to be functional deviance by corruption. This constellation emerges when an organization can pursue its goals only if its members deviate from formal rules for directly private gains. An example is ‘petty corruption’, where officials carry out at least parts their duties only because their personal income is augmented by corruption. As far as I can see, the ongoing discussion about distinguishing organizational from individual corruption does not take into account such subtleties, mainly because corruption is seen as being undertaken either for the benefit of organizations at large or for the benefits of individuals or groups of individuals. I leave it to further research to examine whether the ability to explain fuzzy cases between those two poles is an advantage or a downside of the concept of functional deviance.
Conclusion and outlook
The qualifications stated above show how functional deviance offers new insights in analyzing anti-corruption programs, while also being an under-theorized concept. Building on critical studies on accounting and auditing, the concept of functional deviance highlights constellations where strict compliance mechanisms are not only circumvented, evaded, or manipulated, but where such practices also flexibilize the rigidity of formal structures and, unintentionally, strengthen organizational adaptability. Transparency, compliance, standardized accountability and auditing obviously create incentives to outsmart control mechanisms. The concept of functional deviance goes a step further in arguing that such arcane behavior (the ‘perverse’ class of side effects of transparency) can sum up to an organization’s benefit in intransparently alleviating the inflexibility of compliance regimes. Thus, the conceptual innovation offered by the functional deviance approach is to focus on systemic and emergent effects of deviant organizational behavior instead of focusing on individual reactance and its motivational structure.
Further research in this field could concentrate on analyzing under what circumstances arcane, secret, and intransparent behavior does sum up to organizational adaptability and under what conditions it remains purely interest-maximizing, or corrupt behavior by individuals and organizational cliques. A deeper understanding of such processes can be delivered both by sociological and psychological micro-analyses of organizational group formation and by comparing deviant behavior in different organizational types confronted with compliance and transparency regulation – besides public administration obvious cases are the financial industry and the medical sector. From a theoretical perspective, further research could tie the concept of functional deviance with similar concepts such as Moore and Tumin’s (1949) concept of ‘social functions of ignorance’ or with concepts associated with organizational anthropology like the ‘performative’ role of transparency mechanisms (Albu and Flyverbom, 2013; Garsten and Lindh de Montoya, 2008).
Though it is not within the scope of this paper to offer policy solutions or a model for organizational reform, the concept of functional deviance can sensitize anti-corruption reformers for the unintended side effects of compliance and panoptic transparency. In practice, this could mean to acknowledge limits of operational control (Bernstein, 2012), i.e. to consider measures that allow for functional deviance but simultaneously keep it from becoming ‘privatized’ into corruption pure and simple. As already mentioned, this is socially a tricky task, as ‘allowed for’ functional deviance loses its latent structure and runs the risk of being perceived as a managerial containment mechanism, ready to be circumvented when proven too rigid. From this perspective, functional deviance is not a normative postulate that can be integrated into a formal organizational design.
Such difficulties in controlling deviance call for a renewed discussion on the bureaucratic ethos, i.e. a discussion that defends bureaucracy not only against entrepreneurial and post-modernist attacks (e.g. du Gay, 2000), but also against a principalist view that reduces organizational behavior to rule-following without being aware of well-documented side effects of this vision. One option is to argue that what makes a low-corruption bureaucracy (e.g. in terms of TI’s CPI) is not strict compliance, but the ability to keep functional deviance from turning into private rent-seeking. Currently, I am not aware of research working with such an approach. There is a growing interest in designing bureaucratic control mechanisms that rely on e.g. ‘stewardship’ (Segal, 2012) or ‘integrity’ (Paine, 1994) models instead of strict formal compliance only. But if the further path of organizational reform should stress and re-discover flexibility, adaptability and discretion, I doubt that this can be achieved by, again, making organizational members play to the rules, only this time via internalized ethical codes of conduct.
All this being said, what does the model of functional deviance add to the policy-oriented discussions on corruption and anti-corruption? Above all, it can make anti-corruption reformers aware of the fact that fighting corruption by trying to eradicate all deviations from formal organizational orders basically means throwing the baby out with the bathwater. In the best case, rigidity is evaded by hidden functional deviance and in the worst case, such arcane practices lead to corruption. Thus, anti-corruption reformers should acknowledge limits of compliance and limits of outlawing deviance. This is not a fatalist or a generally skeptical argument – corruption obviously can be tackled by institutional fine-tuning. The concept of functional deviance can contribute to this kind of fine-tuning with a sociologically informed perspective on organizational reform.
 As the aim of this paper is not to criticize this definition of corruption (for an elaborate critique see e.g. Philp, 1997), but to analyze the institutional visions and the organizational side effects that follow from it, offering my own definition of corruption would be beyond its scope.
 For the UN’s definition of corruption see e.g. United Nations (2004) see: http://www.transparency.org/whoweare/organisation/faqs_on_corruption#definecorruption.
 Here, a common-sense definition can be used, defining discretion as the right to decide according to personal judgment. However, in formal organizations this right is itself formally granted, i.e. based on formal regulation beyond the personal judgment of organizational members. Thus, a narrower definition of discretion in view of organizational behavior could be the formally unregulated sphere of organizational behavior, i.e. the formal right to base actions on personal judgment instead of prescribed protocol.
 A different approach that cannot be discussed here is anti-corruption via ‘integrity systems’, i.e. not to rely on external monitoring and compliance only, but to combine these mechanisms with value-based ethical self-control of public bureaucracies (e.g. Huberts et al., 2008; Menzel, 2005; Paine, 1994).
 The original German term Luhmann (1964: 304ff) used is ‘brauchbare Illegalität’, which literally means ‘useful illegality’. However, Luhmann’s examples mostly refer to violations of formal procedures in organizations and not to outright violations of a legal order. Thus, I prefer the term ‘functional deviance’, where deviance is broadly understood as the violation of a legitimate order.