The Principal–Agent Model, Accountability and Democratic Legitimacy

  • Gijs Jan Brandsma
  • Johan Adriaensen
Open Access
Part of the Palgrave Studies in European Union Politics book series (PSEUP)


This chapter explores the normative underpinnings of the principal–agent model. These are situated in Rousseau’s analysis of the representative democracy and Weber’s study of the bureaucracy. Whereas many of their arguments still maintain their value in present-day politics, the empirical reality in which these ideas developed has changed drastically. The rise of alternative forms of public contestation, and above all, the multi-level decision-making setting of the EU, begs the question for which contemporary debates the principal–agent model still holds moral sway. We argue that—while principal–agent analyses can benefit from an explicit normative debate—the model is too reductionist to provide conclusive answers to such debates.

1 Introduction

Most principal–agent scholars refrain from making normative statements: the model has mainly been used to inform a descriptive rather than a prescriptive debate. Scholars use the principal–agent model both as a heuristic framework and as an explanatory toolkit for a variety of purposes in situations of delegated power. It can be used to measure and explain the agent’s discretion, to explain the set-up and the workings of administrative control procedures and to explain the effectiveness of incentives for affecting the agent’s behaviour (Delreux and Adriaensen this volume). As Robert Elgie stated “[T]he principal–agent approach is value-free. By focusing on the exercise of control within institutional arrangements, principal–agent theory forces attention to focus on positive rather than normative analysis” (Elgie 2002: 187). Originating from the domain of organizational economics (Spence and Zeckhauser 1971), the contrast with normative political theory probably could not be bigger.

Yet, empirical studies that use the principal–agent model often allude to the normative conclusions that may be drawn. Many studies situate their relevance through their contribution in debates about accountability and (the adequacy of) representative democracy. Whereas some make the reference explicit (e.g. Reichert and Jungblut 2007), others address it tacitly in the introduction or refer to it in the concluding remarks (Gilardi 2007; Kassim and Menon 2003: 135; Thatcher and Sweet 2002: 18). Even in Elgie’s article, the analysis aimed to enrich a normative discussion on the alleged democratic deficit of the European Central Bank. The tension between the study of what is and what ought to be has long been documented, but the gap between both has rarely been so wide in the study of political science.

In a plea for a normative turn in political science, Gerring and Yesnowitz succinctly write: “The disciplinary segregation of empirical and normative theory is crippling as well as dishonest since both ventures rest on an implicit understanding of the other’s territory. The selection of a topic in empirical political science presumes a judgment of moral importance. Insofar as this is true, the author had best make these connections explicit. There is nothing to be gained, and potentially a great deal to be lost, by smuggling in normative assumptions through the back door. Inexplicit normative theorizing is apt to be slipshod and escapes counter-argument, for it rests beneath the surface” (Gerring and Yesnowitz 2006: 108).

While the use and limitations of the principal–agent model for empirical research have already been discussed in the introduction (Delreux and Adriaensen this volume), this chapter aims to explore the model’s adequacy for subsequent normative interpretations. The first step in such an exercise is to spell out the model’s normative underpinnings. We situate these in Rousseau’s analysis of the representative democracy and Weber’s study of the bureaucracy. Whereas many of their arguments still maintain their value in present-day politics, the empirical reality in which these ideas developed has changed drastically. The multi-level decision-making setting of the EU as well as the rise of alternative forms of public contestation begs the question for which contemporary debates the principal–agent model still holds moral sway.

Through this chapter, we aim to tackle this challenge. While the general argument draws on the wider principal–agent literature, the chapter progressively concentrates on the institutional density of the EU as an interesting test case for the developed arguments. The following section traces the principal–agent model back to the political theories of Rousseau and Weber. The third section contrasts these normative underpinnings with the current EU decision-making context. For this analysis, we draw extensively on empirical work in public administration as well as political science scholarship. In the conclusion, we reflect on the limitations of the principal–agent model in contributing to the study of EU politics.

2 The Normative Underpinnings of the Principal–Agent Model

Several defining features of the principal–agent model strongly resemble elements that are found in political theory, where they form the building blocks for a legitimacy-related argument. For instance, the core of the principal–agent model appears both (albeit put in slightly different words) in Weber’s notion of hierarchy and in Rousseau’s interpretation of democracy: the principal, primarily motivated by efficiency reasons, chooses to delegate power to an agent. In return, the principal can control the powers of the agent (Miller 2005; Bovens 2007: 463).

2.1 Rousseau: Representative Democracy

Rousseau interprets the implications of delegation of power in most radical terms. Sovereignty, he claims, lies with the people and consists of the general will. Since the general will cannot be represented by anything else than itself, popular representatives can only be the agents of the people and should take their cues from the general will. They do not represent the people or the general will; they merely carry out specific tasks as defined by the people, and they should not exercise any form of leadership. He does acknowledge, though, that consensus among the people in fact does not exist, which makes it hard to define the general will in practice. He solves this problem, however, by assuming unidimensional preference orderings for any given policy, of which the mean is taken as the general will. Hence, the people retain ultimate authority, reducing the role of political representatives to that of mere delegates. This element of Rousseau’s theory of democracy equals one of the core elements of the principal–agent model, namely that the people are the ultimate principal and hence have ultimate authority, and choose to delegate a specific set of tasks or powers to an agent.

Resemblances between Rousseau’s work and the principal–agent model are similarly striking as regards the need for appropriate control structures to keep the agent in check. In Rousseau’s view, representative democracy comes with the danger that popular representatives will pursue private agendas rather than expressing the general will. This corresponds to the well-known agency slack issue in principal–agent models (Delreux and Adriaensen this volume). But while those models point towards various incentive-based options to control the agent, Rousseau merely points towards the problem and its dangers for the legitimacy of a political system without hinting at a solution. The more private interests dominate in representative democracy, so he argues, the more politically apathetic the people will become, and the political system will lose legitimacy. This may eventually lead to its collapse. According to Rousseau, sanctioning is not the panacea to solving the agency slack problem since it comes after the violation of the general will and does not necessarily reinstate it. The problems associated with delegating decision-making powers to agents, in Rousseau’s view, can best be addressed by reducing the geographical size of states so that democracy can be organized more locally, allowing for a more direct exercise of sovereign power. Rousseau therefore does not propose a system of controls, but he rather argues that we should avoid the problem of agency slack popping up in the first place by delegating as little as possible (Rousseau 1973 [1779]: 182–207).

Both Rousseau and the principal–agent model, however, stress the superiority of the principal over the agent. While for principal–agent models it defines the domain of application, namely hierarchical relations (Delreux and Adriaensen this volume), it holds moral sway for normative political theory. When interpreting principal–agent relationships from a Rousseauean normative perspective, the principal’s preferences should always prevail and the agent must follow these slavishly—even when they are ethically problematic (e.g. Philp 2009). If the agent chooses not to do so, he is following private preferences rather than those of his principal, and the political system loses legitimacy.

Later normative political theories have challenged the notion that the principal’s preferences must always be directly translated into policies. More in particular, they sought to define how political leadership could be theorized in a system of representative democracy, i.e. leadership exercised by the agent, while keeping the general idea intact of citizens being the ultimate principals. Burke, for instance, did so by putting the general interests first rather than those of his constituency. Being a political agent himself, he famously advocated in his own constituency that “parliament is a deliberative assembly of one nation, with one interest, that of the whole; where, not local purposes, not local prejudices ought to guide but the general good, resulting from the general reason of the whole” (Burke 1774, in Pitkin 1967: 171). Arguably, this line of behaviour increases the room for manoeuvre for political agents when comparing it to Rousseau’s notion of democracy, widening the agents’ role from mere delegates to trustees.

At present, conceptualizations of representation and its role in contemporary democracy are manifold, including the role of accountability to keep the power of the rulers in check. They stress the importance of responsiveness, leadership, the representation of interests as well as mirroring the demographics of the constituency (e.g. Przeworski et al. 1999; Mansbridge 1999, 2009; Powell 2004). The drawback of applying a principal–agent model to representative democracy is that it is based on a very specific and very narrow interpretation of democracy: it is unidirectional with voters having clear preferences and a clear set of controls available to them, and political representatives are simply the recipients of their demands and the subjects of their control (Andeweg and Thomassen 2005). It resonates well with Fritz Scharpf’s idea of input legitimacy—government by the people (Scharpf 1999). Lack of control by the principal in a representative democracy is thus quickly associated with a loss in input legitimacy.

Observations of democratic deficits that are based on the principal–agent model, thus, are implicitly based on a very reductionist notion of democracy: they presuppose that legitimacy can only result from the hierarchical involvement of the citizens which carry highest moral powers. Consultative, participatory, cooperative and deliberative notions of democracy are out of scope as well as representative notions of democracy that emphasize a strong role for political leadership.

2.2 Weber: The Independent Bureaucracy

In contrast to Rousseau’s emphasis on the moral superiority of the principal, Weber recognizes the need for an independent bureaucracy governed by a clear set of rules as a natural complement to a democratic political system. Whereas democracy may have increased the power of the people, it would not render equal treatment by the state unless the administration would become more independent. He argued that the “non-bureaucratic” administration would seek to exploit their position economically or socially as a compensation for their administrative functions (Weber 1978: 984). This observation also implied that bureaucratic rule often finds itself at odds with the idea of democracy. Giving into popular demands may not always serve the nation’s interest. At the same time, favouring the preferences of the administration over its people’s runs against the idea of democracy. In other words, the attainment of output legitimacy—i.e. effective solutions to common problems of the governed (Scharpf 1999)—may sometimes require a loss in input legitimacy.

This situation is most apparent when studying the delegation to independent agencies. When the principals need to signal a credible commitment, it is in their interest to delegate full authority with only limited means of control (Majone 2001). A common example is the delegation of monetary policy to a central bank. In order to avoid hyper-inflation, it is required that the policy is removed from the opportunistic caprices of a political class seeking popular support and delegated to agents who are supposed to act independently. An alternative example is the provision of state support to infant industries. The political class is readily able to provide the carrot of financial support, but it faces much greater difficulty to wield the stick of removing such support if the industry fails to mature. Also in these cases, a certain degree of agency autonomy can ensure policies in the state’s interest (Rodrik 2004).

Whether we deem principal control or agent autonomy desirable depends in great length on the object of delegation and the subjective position of the author. When delegating representational rights, Rousseau’s perspective carries more moral sway, whereas those tasks that require a certain degree of bureaucratic independence find greater affinity in Weber’s thinking. In Elgie’s discussion of the democratic deficit of the ECB, normative arguments to support agent independence (maintain low inflation) as well as control (democratic accountability) could be provided. A positivist use of the principal–agent model seeks to understand the existing balance of power between a principal and an agent. But how should we interpret this observation? Ignoring for a moment that many principal–agent studies refrain from making explicit what they consider a lot or a little control, there is a bigger challenge that requires our attention. A brief allusion that limited control can create democratic deficits implies that scholars take the unrealistic assumption that more control is always desirable, thereby neglecting Weber’s argument. Lack of interpretation of the findings leaves the societal relevance up in the air. “The reader is left to intuit. This lends the practice of normative theorizing a subterranean cast. It is present but unaccounted for” (Gerring and Yesnowitz 2006: 107).

3 Contemporary Relevance of Principal–Agent Research for Normative Debates

The normative underpinnings of the principal–agent model can thus be summarized as follows. The principal may, but does not have to, delegate powers to an agent. Both the agent and the powers to be delegated need to be specified. But since the principal has moral superiority, he is the ultimate judge over the agent’s behaviour. He may install a variety of controls that in the end allow him to sanction any misbehaviour of the agent, whose pursuit of private interests are curtailed since he rationally tries to avoid being sanctioned. If the agent does not behave according to the principal’s preferences, we have a situation of agency loss which is considered a bad thing because the moral superiority of the principal has been impaired.

We have already discussed that normative assessments on legitimacy that exclusively rely on the principal–agent model are implicitly grounded in a very specific normative notion of democracy. But there is more to it. While empirical research on EU politics and administration has shown considerable merit in applying the principal–agent model, the broader institutional context in which specific principal–agent relationships are embedded calls for legitimacy-related normative assessments to be made only with the utmost care. We present three illustrations that support our call for caution: the logics of delegation, the principal’s payoff for exercising control and the accumulation of principal–agent relationships in complex webs of delegation.

3.1 The Logics of Delegation

Each of the logics of delegation described above finds a clear counterpart in Rousseau stressing the moral superiority of the principal and Weber emphasizing the need for a certain degree of independence. Despite the apparent contradiction between the two logics of delegation, there are only few cases of delegation where only one logic will be at work. Franchino studied six hundred cases of secondary legislation in the European Union. He differentiated a logic of efficiency and one of credible commitment and found that in most cases there is a mixture of both logics at stake. Moreover, he showed that cases of delegation that were conceived as guided by credible commitment were often accompanied by a lot of control (Franchino 2002). To explain this conundrum, he argued that it is necessary to differentiate between the types of constraints imposed on the agent. Indeed, not all forms of control compromise the agent’s abilities to commit credibly. From a normative perspective, this creates an additional layer of complexity in asserting whether too much or too little control has been installed or exercised. It implies that, in addition to an assessment of the underlying logic of delegation, a profound analysis of the installed mechanisms of control is required in order to assess whether the balance of power between principal and agent is desirable.

Up to now, we have only dealt with delegation according to a rationalist logic. An alternative explanation would point to principals being guided by a logic of appropriateness (e.g. McNamarra 2002). There is no reason to assume that the European Union is impervious to the latest trends in public administration such as the delegation to executive agencies to name but one. It is not hard to imagine that some agencies were created in response to such a paradigm. This poses a challenge for the utility of the principal–agent model to address normative questions. What if the principals did not make a deliberate trade-off between control and independence? On which grounds can we then make a normative assessment of the chosen institutional design?

3.2 The Principal’s Payoff for Holding Agents Accountable

In general, accountability has been conceptualized in the literature in two ways. One—predominantly American—interpretation of accountability is that it is a virtue. Governments can “be accountable”, essentially by being transparent and responsive. The other—predominantly continental European—interpretation is that it is a mechanism to keep the behaviour of public authorities in check. It is this second meaning that is found in studies of EU accountability. Accountability as a mechanism includes the transfer of information from the agent to the principal, the possibility for the principal to ask further questions, to pass judgement and to impose positive or negative consequences on the agent (adapted from Bovens 2007). Hence, it includes monitoring efforts as well as incentives for the agent to behave according to the principal’s demands as otherwise he might risk being faced with sanctions, or not to obtain rewards. Applied to the principal–agent model, accountability thus is synonymous to the observation of control over the agent, and it seeks to keep the behaviour of the agent in line with the preferences of the principal.

Empirical research investigating the actual workings of EU accountability processes has been flourishing mostly in the last decade, and answers to the question to what degree this principal–agent-inspired logic holds true in practice are rather mixed. One of the earliest insights is that the principal may in practice be monitoring the agent in two different ways. One is police patrolling , which refers to direct and extensive monitoring by the principal. In the EU system, the various information requirements from the Commission to the EP are a case in point (Brandsma 2013), as is the comitology system (Pollack 2003). The other is fire alarms , which means that the principal relies on third parties such as interest groups to monitor the agent. The principal may therefore receive much less information on the agent’s behaviour and might assess his behaviour differently than the interest group would do, but for the principal it dramatically reduces transaction costs for monitoring (McCubbins and Schwartz 1984; Lupia and McCubbins 1994). This distinction between monitoring regimes can be easily integrated into the principal–agent model as it is the principal who calls the shots, and the resulting degree of accountability towards the principal is contingent on the quality of monitoring.

Recently, a number of studies on EU governance have seen the light which argue that accountability shortages might not be caused by shirking agents, but rather by failing principals (Schillemans and Busuioc 2015). Brandsma, for instance, notes that this often applies to member state policy experts who attend comitology committees (Brandsma 2010). Even with appropriate control structures in place and agents duly reporting their activities to the principal, the latter regularly ignores the agents’ input or starts asking further questions that do not relate to the matter at hand. This may result from a workload problem: the principal may have more tasks than merely checking upon his agents, and accountability is simply not prioritized, especially in a professional context where he trusts the agent to do the right thing (e.g. Brandsma 2010).

In a political context, this workload problem also features prominently as legislatures are constitutionally both obliged to legislate as well as to hold the executive accountable. It is a well-known fact that control by legislators tends to be rather limited due to high transaction costs (Huber and Shipan 2000). Passing new legislation tends to be prioritized over accountability (Andeweg 2007), and moreover in coalition government systems, any parliamentary sanction needs majority approval while the government is dominant over parliament. The government usually consists of high-ranking party officials who are mostly considered to be too prominent politicians to fail. In the European Parliament, which does not feature a coalition government system, empirical studies show that similar priority issues surface on a number of occasions. To give a few examples, legislative evaluations—if they receive any parliamentary attention at all—are mostly not used as a basis for political accountability, but for pushing a new legislative agenda (Zwaan et al. 2016). The EP asks very few questions about the activities of EU agencies on the basis of the information it receives from them (Font and Pérez Durán 2016). Parliamentary attention for comitology or delegated acts, except for some exceptionally salient issues such as car emissions, generally remains limited. The process of digesting information on those is delegated to the EP’s administrators, who still need to rely on the attention of the MEPs. Only when the EP has significant control powers is it relatively more likely to spend more resources on holding the executive to account, and it can even be quite successful in that regard (Brandsma 2016), but still the number of actual sanctions imposed is very low (Kaeding and Stack 2015). This problem is not limited to legislators but can be extended to the context of national administrations involved in Council decision-making (Adriaensen 2016). We are aware that these studies offer only some snapshots; other information-gathering and accountability mechanisms are at work at the same time. But the point here is that the availability of information does not equal accountability. Information asymmetries can only be remedied, and principals can only decide whether or not to sanction the agent’s behaviour, when they actively process the information that is available (Naurin 2006).

Hence, we are stuck with a puzzle that has normative implications. Why is it that principals choose to delegate powers to agents, take care to install control procedures, and seem to forget about them once they are in place? The principal–agent model would suggest a few answers to this question: the principal already knows that the agent is doing the right thing because of a very specific mandate, or the agent is so constrained by various control procedures that the principal does not need to care about actively exercising control, or the powers delegated to the agent are not salient enough to the principal which makes the costs of exercising control outweigh the gains. In those cases, one might positively assess the arrangements in place for apparently they do keep the agent in check to a sufficient degree.

But an alternative interpretation might be that the currency of politics is not the pursuit of preferences, as is assumed by the principal–agent model as well as by Rousseauen democracy, but rather reputation (Busuioc and Lodge 2016). To put things bluntly, this interpretation holds that principals and agents primarily care about the soundness of their own reputation, and any other goal, such as pursuing preferences, is subordinate to it. Replacing preferences by reputation turns the principal–agent model upside down. It offers the room to agents to tacitly expand their mandate as long as their principals will allow them to do so, it explains why sometimes agency slack of an otherwise credible agent goes by unsanctioned, and it factors in that agents may in practice play a leading role rather than the principal. It goes against the normative underpinnings of the principal–agent model which is all about preference-based delegation: Weberists and Rousseauists would most certainly deplore it.

3.3 Complex Webs of Principal–Agent Relationships

The workings of principal–agent relationships can be theorized with relative ease when there exists a singular chain of delegation, such as in parliamentary democracies where citizens feature as the ultimate principal and the bureaucracy as the ultimate agent, with parliament and government in between (Strøm et al. 2006). Arguably, the European Union’s political system is quite far from this ideal-typical model of parliamentary democracy. We identify three issues that inhibit inferring clear-cut diagnoses on the democratic legitimacy of the European Union from empirical findings derived from principal–agent analyses. They are (1) the absence of a single ultimate principal in EU politics with, as a consequence, multiple co-existing chains of delegation; (2) the co-existence of alternative mechanisms of accountability; and (3) the occurrence of reversed delegation.

The first issue is that in many situations, multiple co-existing chains of delegation can be identified. This is not typical for the EU since many states, and primarily presidential states, also feature multiple chains of delegation which co-exist and in practice may compete. From a normative perspective, those multiple chains are all of equal importance since they all start with the state’s citizens as the ultimate principals.

But in the EU’s political system, applying the principal–agent model comes with a major normative problem because the multitude of co-existing chains of delegation do not begin with the same set of principals. The chains that involve the European Parliament begin with the European electorate at large, while those that involve the member states begin with the citizenries of each respective member state. Both these chains may legitimately claim to channel their respective principals’ preferences, resulting in a clash of intergovernmental and supranational logics. An abundance of research has investigated to what degree one could argue that either of these logics is more appealing in a normative sense, in particular research investigating the possible emergence of a European demos or public sphere (Van de Steeg 2012; Risse 2015) or the continuing appeal of the intergovernmental logic for democratic legitimation of the European Union (Moravscik 2002; Mair and Thomassen 2010).

Situations where both logics intersect are many. They usually emerge where a supranational and an intergovernmental principal delegate powers to the same agent. The investiture procedure of the Commission is a case in point, with the European Parliament and the Council both attempting to be decisive in the selection of candidates and tempting to tie the Commission’s hands to policy commitments as well as to procedural commitments during the selection process (Magnette 2001; Brandsma 2013). The same applies to the legislative process where the Council and the European Parliament jointly agree on a legislative text, but executive powers which are delegated by the Commission are controlled by the member states alone, or independently by the Council and the European Parliament rather than jointly (Christiansen and Dobbels 2013; Brandsma and Blom-Hansen 2012). Similar difficulties can be observed in the realm of foreign policy, where the High Representative of the Union for Foreign Affairs and Security Policy is part of both the European Council and the Commission. Obviously, these are only three examples out of a much larger set of similar occurrences. A large body of literature has tried to identify whether the balance of power is tipped towards intergovernmental or supranational institutions, but the normative problem is more fundamental: if the EU’s citizenry does not equal the sum of its component parts, which of the two logics should dominate? Even though principal–agent analysis in the EU enabled us to move beyond the deadlock in the debate between supranationalism and intergovernmentalism (Pollack 1997), this debate does sneak in through the back door when normative assessments of the legitimacy of the EU are made, also when principal–agent analysis is used for empirical backup.

A second issue is that in a networked governance setting, there are a wide range of alternative channels through which an agent can be held to account. In other words, an exclusive focus on the hierarchy is overly reductionist and will lead to biased assessments. To best illustrate this argument, we will focus on delegation to independent agencies in general and the ECB in particular. Usually, the independence of central banks is constitutionally safeguarded, making it very hard if not impossible for the legislator to change the terms of delegation or to exercise strong controls. This also applies to the ECB. With regulatory agencies, such constitutional safeguards do not exist as they are usually put in place through a normal legislative act, but here too they can be endowed with significant powers. For such independent bodies, several observers feared a redistribution of power which might not be mirrored by an appropriate redistribution of accountability (Shapiro 1997; Flinders 2004).

There are, however, three reasons why this fear is not vindicated in most situations. First, some authors assume that agencies must have significant regulatory powers by definition, whereas in practice this is not the case. Taking the EU agencies as an example, most of them are only supposed to supply information to the Commission but do not regulate (Busuioc 2013; Groenleer 2009). Second, seemingly independent agencies might in fact be very dependent because of dense webs of accountability. They can have management boards, client panels, obligations vis-à-vis the court of auditors, be obliged to send annual reports to political bodies, et cetera (Schillemans 2008; Busuioc 2013). Moreover, there can also be strong ex ante controls in the form of appointment procedures for the agencies’ executives, an approval procedure for annual work plans, or perhaps most strongly, the principal can keep the power of the purse. Third, the principal may have made a conscious decision to endow regulators with independence. The most prominent advocate of this perspective is Majone (2001), who argued that regulators should be independent so as to provide more credible long-term policies as it is hard for political principals, who face re-election every few years, to overturn these. The agency, hence, is a trustee rather than a delegate (Majone 2001).

However, even in the case of a trustee relationship, legitimate outcomes can be achieved. Through informal interactions with legislative actors and societal stakeholders, Piattoni has argued that the ECB can gain a form of “throughput” legitimacy (Piattoni 2013). By grace of its deliberative quality, the independent administration can thus achieve policy outcomes that are considered legitimate (Risse and Kleine 2007: 72–74). It is possible to reconcile the benefits of independence with legitimacy if we widen the scope of analysis beyond the hierarchical relation. In summary, the principal–agent model is an overly reductionist tool to make normative statements about the legitimacy of decision-making if those decisions are made by independent agents who have been delegated powers following the logic of credible commitment.

A third empirical phenomenon which complicates normative interpretation is what we dub reversed delegation. The constitutional fabric of the European Union is such that the member states may take up a variety of guises in this system: as political “masters” in the European Council, as co-legislators in the Council of the EU and as executive agents in their own right. Under those guises, it regularly happens that the principal delegates power to an agent, who in turn delegates back to the principal who takes on another guise. Ordinary policy processes nearly always include instances of reversed delegation. European law mostly needs to be implemented by the member states, but (unless for exceptional circumstances) can only be initiated by the Commission. Legislative initiatives include a proposal by the Commission that specifies which specific executive powers the Council and the European Parliament should delegate to the Commission, which is subsequently empowered to flesh out the details of these powers—if so agreed by both legislative bodies. Thus, the Council and the European Parliament are principals, but they can only act on a proposal made by the agent. When the Commission adopts executive measures, the process is reversed: then the Commission acts as a principal when it decides on the specific policy contents. The member states, as agents, need to implement. In principal–agent terms, this is a messy picture: one of the principals delegates powers to itself, but it can only act on the proposal of an agent-in-between who is also empowered to add further details. The problem becomes apparent in the case of agency slack (Blom-Hansen 2005): in principle, the Commission should hold the member states accountable for implementation. But if European law is not properly implemented by member states, the Commission often lacks the political backup to effectively make the member states comply since one of the Commission’s principals is the Council, which again represents the (shirking) member states. This problem makes that the Commission carefully picks its battles when it drags member states to Court: the Commission often chooses not to play a hard game when its position is supported by only a few member states (Steunenberg 2010).

This back and forth between principals and agents points to a normative challenge: Where does superiority actually lie in cases of reversed delegation? The principal’s superiority is a core component of the principal–agent model, and as argued earlier in this chapter, this assumes a unidirectional form of politics. When delegation loops back to a principal who becomes an agent two steps down the line, hierarchical accountability no longer enhances control: the immediate principal of the ultimate agent is dependent on its own principal, which represents the ultimate agent in another guise.

4 Conclusion

The principal–agent model has a strong track record in political science, including studies into EU politics and administration. It presents an analytical toolkit which has proven to be very useful for systematically describing situations of delegated decision-making, as well as for explaining the behaviour of principals and agents, be it of individual or institutional kind. As with any theory in the social sciences, some bits of the empirical reality can be better explained by the model than others, but nonetheless it provides crucial insights into the workings of EU governance. In fact, some peculiarities of EU governance, such as “reversed delegation” as described before, can only be elucidated parsimoniously when applying the principal–agent model.

But in such an application, we should not forget that it is intended to be an empirical model, not a normative one. Even though it is remarkably easy to apply the core tenets of the principal–agent model in normative assessments on the legitimacy of a politico-administrative arrangement, we argue that this should only be done with the utmost care. The model essentially treats politics as being unidirectional: the principal has preferences and is able to delegate powers to an agent. The principal holds moral superiority and hence is able to apply sanctions to the agent’s behaviour. The agent is rational and hence is responsive to control incentives. These may be useful assumptions for explaining what is, but assessing what ought to be is a different ball game.

As we have discussed in this contribution, these assumptions provide only a narrow Rousseauean or Weberian framework for assessing the legitimacy of a political system. Alternative sources of legitimacy, such as political leadership exercised by the agent, are not in scope. But when it is used in a normative sense, it proves remarkably hard to distinguish right from wrong in complex situations of delegation. It does not specify which principal should hold moral superiority when there are multiple and competing ones who ultimately represent different but partly overlapping constituencies. It does not provide a normative reason why principals should delegate powers back to themselves via an intermediary agent who may add a few things, and how many control powers this agent may have in the first place given that it is only an intermediary actor. It does not include any normative reason why the principal should, at times, delegate extensive powers which are accompanied by a very weak control scheme. All these elements, however, are part of the constitutional fabric of the European Union. These can be empirically investigated by the principal–agent model and surprising findings may well be exposed, but empirical findings that do not exactly correspond to the mechanisms underlying the principal–agent model do not necessarily point to a legitimacy problem.

Our chapter should therefore not be mistaken for concluding that the European Union is inherently undemocratic. We argue that the findings from principal–agent analysis can often inform a normative debate but is insufficient to be used as a normative benchmark. Scholars should try to resist the temptation of assessing the democratic legitimacy of European Union decision-making solely on the basis of empirical findings stemming from the principal–agent model when they do not define the criteria on which they base their normative conclusions, or inter alia use the principal–agent model itself for normative purposes. For that matter, we hope to have provided useful building blocks for making carefully informed decisions on the normative applicability of the principal–agent model for European politics.


  1. Adriaensen, J. (2016). National administrations in EU trade policy: Maintaining the capacity to control. London: Palgrave MacMillan.CrossRefGoogle Scholar
  2. Andeweg, R. (2007). A comment on Auel, Benz, and Maurer. In B. Kohler-Koch & B. Rittberger (Eds.), Debating the democratic legitimacy of the European Union (pp. 102–110). Plymouth: Rowman and Littlefield.Google Scholar
  3. Andeweg, R., & Thomassen, J. (2005). Modes of political representation: Toward a new typology. Legislative Studies Quarterly, 30(4), 507–528.Google Scholar
  4. Blom-Hansen, J. (2005). Principals, agents, and the implementation of EU cohesion policy. Journal of European Public Policy, 12(4), 624–648.Google Scholar
  5. Bovens, M. (2007). Analysing and assessing accountability: A conceptual framework. European Law Journal, 13(4), 447–468.Google Scholar
  6. Brandsma, G.J. (2010). Accounting for input in comitology committees: An uncomfortable silence. Journal of European Public Policy, 17(4), 487–505.Google Scholar
  7. Brandsma, G.J., & Blom Hansen, J. (2012). Negotiating the post‐Lisbon comitology system: Institutional battles over delegated decision‐making. Journal of Common Market Studies, 50(6), 939–957.Google Scholar
  8. Brandsma, G.J. (2013). Bending the rules: Arrangements for sharing technical and political information between the EU institutions. European Integration Online Papers, 17(8).Google Scholar
  9. Brandsma, G.J. (2016). Holding the European Commission to account: The promise of delegated acts. International Review of Administrative Sciences, 82(4), 656–673. Google Scholar
  10. Busuioc, M. (2013). European agencies: Law and practices of accountability. Oxford: Oxford University Press.CrossRefGoogle Scholar
  11. Busuioc, E., & Lodge, M. (2016). The reputational basis of public accountability. Governance, 29(2), 247–263.Google Scholar
  12. Christiansen, T., & Dobbels, M. (2013). Non‐legislative rule making after the Lisbon treaty: Implementing the new system of comitology and delegated acts. European Law Journal, 19(1), 42–56.Google Scholar
  13. Delreux, T., & Adriaensen, J. (2017). Introduction. Use and limitations of the principal–agent model in studying the European Union. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 1–34). London: Palgrave MacMillan.Google Scholar
  14. Elgie, R. (2002). The politics of the European Central Bank: principal–agent theory and the democratic deficit. Journal of European Public Policy, 9(2), 186–200.Google Scholar
  15. Flinders, M. (2004). Distributed public governance in the European Union. Journal of European Public Policy, 11(3), 520–544.Google Scholar
  16. Font, N., & Pérez Durán, I. (2016). The European Parliament oversight of EU agencies through written questions. Journal of European Public Policy, 23(9), 1349–1366. Google Scholar
  17. Franchino, F. (2002). Efficiency or credibility? Testing the two logics of delegation to the European Commission. Journal of European Public Policy, 9(5), 677–694.Google Scholar
  18. Gerring, J., & Yesnowitz, J. (2006). A normative turn in political science? Polity, 38(1), 101–133.CrossRefGoogle Scholar
  19. Gilardi, F. (2007). The Same, but different: Central Banks, regulatory agencies, and the politics of delegation to independent authorities. Comparative European Politics, 5(3), 303–327.CrossRefGoogle Scholar
  20. Groenleer, M. (2009). The autonomy of European Union agencies: A comparative study of institutional development. Delft: Eburon Uitgeverij BV.Google Scholar
  21. Huber, J., & Shipan, C. (2000). The costs of control: Legislators, agencies, and transaction costs. Legislative Studies Quarterly, 25(1), 25–52.Google Scholar
  22. Kaeding, M., & Stack, K. (2015). Legislative scrutiny? The political economy and practice of legislative vetoes in the European Union. Journal of Common Market Studies, 53(6), 1268–1284.CrossRefGoogle Scholar
  23. Kassim, H., & Menon, A. (2003). The principal–agent approach and the study of the European Union: Promise unfulfilled? Journal of European Public Policy, 10(1), 121–139.CrossRefGoogle Scholar
  24. Lupia A., & McCubbins, M. (1994). Designing bureaucratic accountability. Law and Contemporary Problems, 57(1), 91–126.Google Scholar
  25. Mair, P., & Thomassen, J. (2010). Political representation and government in the European Union. Journal of European Public Policy, 17(1), 20–35.Google Scholar
  26. Magnette, P. (2001). Appointing and censuring the European Commission: The adaptation of parliamentary institutions to the community context. European Law Journal, 7(3), 292–310.Google Scholar
  27. Majone, G. (2001). Two logics of delegation agency and fiduciary relations in EU governance. European Union Politics, 2(1), 103–122.CrossRefGoogle Scholar
  28. Mansbridge, J. (1999). Should blacks represent blacks and women represent women? A contingent “yes”. The Journal of Politics, 61(3), 628–657.Google Scholar
  29. Mansbridge, J. (2009). A “selection model” of political representation. Journal of Political Philosophy, 17 (4), 369–398.Google Scholar
  30. McCubbins, M., & Schwartz, T. (1984). Congressional oversight overlooked: Police patrols versus fire alarms. American Journal of Political Science, 28(1), 165–179.Google Scholar
  31. McNamara, K. (2002). Rational fictions: Central Bank independence and the social logic of delegation. West European Politics, 25(1), 47–76.Google Scholar
  32. Miller, G. (2005). The political evolution of principal–agent models. Annual Review of Political Science, 8, 203–225.CrossRefGoogle Scholar
  33. Moravcsik, A. (2002). Reassessing legitimacy in the European Union. Journal of Common Market Studies, 40(4), 603–624.Google Scholar
  34. Naurin, D. (2006). Transparency, publicity, accountability—The missing links. Swiss Political Science Review, 12(3), 90–98.Google Scholar
  35. Philp, M. (2009). Delimiting democratic accountability. Political Studies, 57(1), 28–53.Google Scholar
  36. Piattoni, S. (2013). Representation as delegation: A basis for EU democracy? Journal of European Public Policy, 20(2), 224–242.CrossRefGoogle Scholar
  37. Pitkin, H. (1967). The concept of representation. Berkeley: University of California Press.Google Scholar
  38. Pollack, M. (1997). Delegation, agency, and agenda setting in the European Community. International Organization, 51(1), 99–134.Google Scholar
  39. Pollack, M. (2003). Control mechanism or deliberative democracy? Two images of comitology. Comparative Political Studies, 36(1–2), 125–155.CrossRefGoogle Scholar
  40. Powell, G. (2004). Political representation in comparative politics. Annual Review of Political Science, 7, 273–296.CrossRefGoogle Scholar
  41. Przeworski, A., Stokes, S., & Manin, B. (1999). Democracy, accountability, and representation (Vol. 2). Cambridge: Cambridge University Press.Google Scholar
  42. Reichert, M., & Jungblut, B. (2007). European Union external trade policy: Multilevel principal–agent relationships. Policy Studies Journal, 35(3), 395–418.Google Scholar
  43. Rodrik, D. (2004). Industrial policy for the twenty-first Century. Cambridge: Harvard University.Google Scholar
  44. Risse, T. (2015). A community of Europeans? Transnational identities and public spheres. Ithaca: Cornell University Press.Google Scholar
  45. Risse, T., & Kleine, M. (2007). Assessing the legitimacy of the EU’s treaty revision methods. Journal of Common Market Studies, 45(1), 69–80.Google Scholar
  46. Rousseau, J. (1973 [1779]). The social contract, or principles of political right. Book II. Translated by G. Cole. London: J.M. Dent & Sons.Google Scholar
  47. Scharpf, F. (1999). Governing in Europe: Effective and democratic?. Oxford: Oxford University Press.CrossRefGoogle Scholar
  48. Schillemans, T. (2008). Accountability in the shadow of hierarchy: The horizontal accountability of agencies. Public Organization Review, 8(2), 175–194.Google Scholar
  49. Schillemans, T., & Busuioc, M. (2015). Predicting public sector accountability: From agency drift to forum drift. Journal of Public Administration Research and Theory, 25(1), 191–215.Google Scholar
  50. Shapiro, M. (1997). The problems of independent agencies in the United States and the European Union. Journal of European Public Policy, 4(2), 276–277.Google Scholar
  51. Shepsle, A., & Bonchek, M. (1997). Analyzing politics: Rationality, behavior, and institutions. New York: W. W. Norton & Co.Google Scholar
  52. Spence, M., & Zeckhauser, R. (1971). Insurance, information, and individual action. The American Economic Review, 61, 380–387.Google Scholar
  53. Steunenberg, B. (2010). Is big brother watching? Commission oversight of the national implementation of EU directives. European Union Politics, 11(3), 359–380.CrossRefGoogle Scholar
  54. Strøm, K., Müller, W., & Bergman, T. (Eds.). (2006). Delegation and accountability in parliamentary democracies. Oxford: Oxford University Press.Google Scholar
  55. Thatcher, M., & Sweet, A. (2002). Theory and practice of delegation to non-majoritarian institutions. West European Politics, 25(1), 1–22.Google Scholar
  56. Van de Steeg, M. (2012). The making of a European public sphere: Media discourse and political contention. West European Politics, 35(2), 426–427.Google Scholar
  57. Weber, M. (1978). Economy & society: An outline of interpretive sociology. Edited by G. Roth and C. Wittich, Translated by E. Fishoff et al. Berkeley: University of California Press.Google Scholar
  58. Zwaan, P., van Voorst, S., & Mastenbroek, E. (2016). Ex post legislative evaluation in the European Union: Questioning the usage of evaluations as instruments of accountability. International Review of Administrative Sciences, 82(4), 674–693. Google Scholar

Copyright information

© The Author(s) 2017

Authors and Affiliations

  • Gijs Jan Brandsma
    • 1
  • Johan Adriaensen
    • 2
  1. 1.Utrecht UniversityUtrechtThe Netherlands
  2. 2.Maastricht UniversityMaastrichtThe Netherlands

Personalised recommendations