Conclusion. Opportunities and Challenges for the Principal–Agent Model in Studying the European Union

  • Johan Adriaensen
  • Tom Delreux
Open Access
Part of the Palgrave Studies in European Union Politics book series (PSEUP)


This chapter examines whether the principal–agent model will mature by accommodating the new empirical reality or whether it will become outdated as the practical challenges narrow the researcher’s focus to the simplest of hierarchical relations. We address this question by covering the three main components of the research process. We distinguish (1) the formulation of research questions; (2) the contribution one can deliver to the existing literature; and (3) the required methodology. Each section questions the opportunities that are to be gleaned from applying the principal–agent model to an ever-more complex setting, and confronts it with practical challenges that may emerge in the research process. In so doing, we derive guidelines that may help to ensure the contemporary relevance of the principal–agent model.

1 Introduction 1

This volume started from the observation that the principal–agent model faces a challenge if it wants to continue being a useful tool to grasp contemporary European politics. Decision-making involves a widening range of increasingly different stakeholders who often operate in a network—rather than in a hierarchical—setting. Acknowledging that most of such networks function in the shadow of hierarchy, we argued that there is still ample scope for meaningful use of the principal–agent model (Delreux and Adriaensen this volume). It does, however, imply a concerted two-stage effort from the researcher. First, the hierarchical and dyadic “principal–agent-proof” relationship has to be identified amidst a plethora of other relations. Second, once this relationship is detected, a principal–agent analysis can be conducted on the conditions that have led to the observed pattern of delegation (i.e. the politics of delegation) and/or the consequences this pattern has on the distribution of power between the principal and the agent (i.e. the politics of discretion).

In the Introduction, we proposed our two-stage approach in abstract terms. Yet the proof of the pudding is in the eating, namely the application of the principal–agent model to specific empirical cases that take the increased complexity of EU decision-making seriously. Each of this volume’s contributors sought to apply the principal–agent model in a wide range of policy contexts. The contributions laid bare that expanding the principal–agent model to ever more complex settings not only comes with opportunities but is also likely to confront researchers with important challenges. For that reason, drawing upon the findings of the preceding chapters as well as our own experience, this conclusion takes a prospective outlook and points at the opportunities but also the challenges one is likely to encounter when pursuing this route. It is driven by the following question: Will the principal–agent model mature by accommodating the new empirical reality or is it destined to become outdated as the practical challenges narrow the researcher’s focus to the simplest of hierarchical relations?

As this concluding chapter seeks to focus on the practical implementation of such a “new generation” principal–agent research project, we structure our discussion in three sections, corresponding to the main components of the research process. We distinguish (1) the formulation of research questions; (2) the contribution one can deliver to the existing literature; and (3) the methodology required to address the formulated questions. Each section questions the opportunities that are to be reaped from applying the principal–agent model to an ever more complex setting and confronts it with practical challenges that may emerge in the research process. The main opportunities and challenges are summarized in Table 1. On the basis of the preceding chapters, we conclude by arguing that the principal–agent model has the potential to remain a vital tool in analyses of EU politics.
Table 1

Opportunities and challenges for the principal–agent model




Research questions

Stimulus for the formulation of new research questions

- Need for in-depth understanding of the decision-making processes

- Risk of conceptual complacency

Contributions to the literature

a. Expansion of the empirical scope of application

a. Identification of the act of delegation

b. Identification of scope conditions for delegation/discretion

b. Limitations to generalization and theory-building

c. Conceptual refinement

c. Conceptual stretching and conceptual overload


Clearer fit with (more) advanced methods

- Identification of the relevant unit of observation

- “High tariff” methods

2 The Questions We Ask

A first opportunity derived from the principal–agent model’s use in, at first sight, “less hospitable” contexts is the model’s ability to problematize empirical observations in a way that advances our understanding of contemporary European politics. Theory-informed empirical research traditionally starts from the identification of a research problem for which a suitable theoretical lens is chosen that can provide a (partial) answer to the identified problem. Throughout this volume, we were repeatedly confronted with a different dynamic: applying the principal–agent model to a more complex decision-making context stimulated the formulation of new research questions that were previously hidden. The application of the model indeed enables scholars to discover political dynamics and phenomena in their data that are empirically and theoretically puzzling and that would have remained hidden otherwise. The answers to such new puzzles can either be within the remit of the principal–agent model or point towards the model’s limitations. In any case, the multitude of empirical and theoretical questions that emerged suggest the great potential of the model’s application in more complex settings.

Often the puzzle already comes to light when mapping the principal–agent-proof hierarchical dyadic relation. For instance, in studying the role of the EEAS in EU foreign policy, Dijkstra (this volume) points to the peculiar phenomenon of non-exclusive delegation: a situation where the principals retained the authority to pursue their own policies alongside the agent that is the EEAS. From a straightforward rational choice perspective, we would expect either delegation (when expected benefits trump the expected costs) or non-delegation (if this is not the case). Applying the principal–agent model to the issue of non-exclusive delegation to the EEAS therefore reveals the need for a more refined explanation of the creation and maintenance of the EEAS as an agent.

In order to make the principal–agent model useful to identify new research questions, two challenges need to be overcome. First and foremost, a firm understanding of the decision-making processes in the targeted policy domain is required in order to identify and contextualize the principal–agent-proof relation amidst the wider network of actors. The mapping exercise that precedes the analysis of delegation and/or discretion often implies a large investment of time through a series of (exploratory) interviews, an extensive analysis of the legal and administrative framework that governs the policy sub-system, or even participatory observation. Surely, it helps if the complexity that is added to the model is of an incremental nature. Conceição-Heldt (this volume) could draw upon a rich literature that has studied the relation between Council anconclusion that this is not the cased Commission in trade negotiations through the principal–agent model before adding the role of the European Parliament as second (collective) principal. Irrespective of whether incrementally refining the existing literature is possible or not, a thorough understanding of EU decision-making in a given policy domain is a necessary but insufficient condition for the identification of compelling new research questions.

Second, researchers also need to be critical when conceptualizing the implications of such complexity for the principal–agent relationship. Conceptual complacency should be avoided, or its risk should at least be acknowledged in the theoretical reflection. Here, we refer to the risk that principal–agent scholars are not sufficiently self-critical when applying certain concepts. The principal–agent model’s main strength is derived from its reductionist perspective: it focuses on a dyadic, hierarchical contractual relationship. As a consequence, there is an ever-present risk that one is too quick in oversimplifying the underlying complexity or in interpreting the decision-making process imprecisely in order to let it fit with the classic principal–agent insights. For example, although they initially thought that the informal technical meetings in the context of trade policy-making are an additional control mechanism in the hands of the member states, Coremans and Kerremans (this volume) come to the conclusion that this is not the case, as it is not the Council (principal) that is organizing the meetings, but rather the Commission (agent). This observation challenged their initial conceptualization and led to the formulation of an original argument with potentially wide-ranging consequences: the agent has an incentive to actively manage information asymmetry. Similarly, Laloux (this volume) has demonstrated that an initial conceptualization of shadow rapporteurs in trilogue negotiations as control mechanisms of the EP is not the most adequate principal–agent understanding, as the shadow rapporteurs rather need to be considered as members of a collective agent. Helwig (this volume) offers a third example of this challenge. His puzzle originates from the observation that the lack of discretion for the High Representative is not the consequence of deliberate design by the principals as one would infer from the existing literature, but rather the result of interaction between multiple agents after the act of delegation has been designed. Discussing the politics of delegation leading up to the creation of the position of the High Representative, he discards this initial thesis and continues to show that the interaction between agents can also generate unintended consequences.

The challenge to avoid conceptual complacency should not be taken lightly. It is not because an empirical observation bears similarity to known mechanisms of control that one should just qualify it as such. Likewise, an actor exerting control over an agent should not automatically be conceptualized as a principal because delegation—and not control—is a necessary condition to consider an actor as a principal (Delreux and Adriaensen this volume). The problem with such loose use of the concepts is that it not only hollows out the heuristic and explanatory power of the principal–agent model, but that it would also obscure the many relevant questions principal–agent scholars could and should be asking.

In summary, applying the principal–agent model to complex decision-making settings has the potential to generate a series of research questions worthy of our attention. Yet such potential is only realized when thorough knowledge of the decision-making process is combined with conceptual rigour. If either is absent, the principal–agent model runs the risks of becoming a simplistic model that fails to observe let alone address the relevant questions.

3 Our Contributions to the Literature

While the principal–agent model is a useful heuristic tool for mapping the institutional setting in which decisions are being made, the ultimate objective—to study the politics of delegation and/or the politics of discretion—usually transcends a mere mapping exercise. Applying the principal–agent model to settings in which the hierarchy is embedded in a wider (network) structure has the benefit that it can contribute to innovation in the principal–agent literature in three ways: (1) expanding the model’s scope of application to other empirical cases; (2) discovering new conditions to explain the politics of delegation and discretion; and (3) refining existing concepts. The current section will show, however, that each kind of contribution to the literature also carries in itself a challenge that may undermine the aspired opportunities.

3.1 Expanding the principal–agent Model’s Scope of Application

A first opportunity of applying the principal–agent model to less obvious hierarchical relations is that it opens the empirical scope for new case studies in policy domains or on political processes that have currently escaped the attention from principal–agent scholars. The good news for future principal–agent projects is that there is still much uncharted territory as only a small sample of the population of hierarchical dyadic relations has been studied from this perspective.

The strong focus in the literature on the EU’s external relations and trade policy-making in particular stems from the relative simplicity of the principal–agent relation in these fields. The “external relations bias” in the empirical applications of the principal–agent model in EU studies is also reflected in the case studies presented in this book. Yet, several contributions have shown that, even in this area, the basic simplicity should not be taken for granted (Dijkstra this volume; Helwig this volume; Plank and Niemann this volume; Gastinger this volume; Conceição-Heldt this volume; Coremans and Kerremans this volume). Consequently, even in a policy domain so crowded with principal–agent analyses, further research is needed in order to focus on empirical complexities that were previously unaddressed. More importantly, for such studies, it may even be necessary to explore and integrate the empirical complexity into the model if a contribution to the literature is part of the research objectives. For instance, the “anti-agent bias” in the application of the principal–agent model to the EU’s trade negotiations (Elsig and Dupont 2012: 493) is but one such observation that stimulates further research on the Commission as opposed to the Council. Coremans and Kerremans (this volume) address this issue by highlighting the motives of the agent to report beyond the principal’s demands. So does Gastinger (this volume), who zooms in on the unitary actor assumption of the agent, as well as Plank and Niemann (this volume), who examine how the context in which the agent operates affects the latter’s discretion.

Beyond these incremental advances in the established principal–agent literature in the field of external relations, contributions to the literature are within reach by extending the model to new policy environments. In many cases, this requires greater conceptual and exploratory groundwork as the applicability of the principal–agent model is often less straightforward than in the trade domain. This does not only apply to the analysis of alternative policy domains (Plank and Niemann this volume) or policy processes (Laloux this volume) but also when scholars seek to integrate a “new” institutional actor such as the Council Secretariat (Reykers and Beach this volume), the European Council (Kroll this volume) or the European Parliament (Laloux this volume; Conceição-Heldt this volume) in the hierarchical relations that permeate the EU’s decision-making structure. Yet also recent institutional developments in the EU that are linked to normative discussions on democracy and legitimacy, such as the role of national parliaments or the Spitzenkandidaten system, seem to be promising empirical fields to unleash the principal–agent model on. Next to the model’s application to new policy environments, we can also imagine extensions to explain integration by stealth (Héritier 2015) or study recent phenomena such as differentiated and flexible integration (Leuffen et al. 2012), and—why not?—disintegration.

This brings us to the main challenge related to the expansion of principal–agent’s scope of application. As we argued earlier, the principal–agent model can only be applied if two necessary conditions are fulfilled: the acceptance of actors’ rationality and the clear identification of an act of delegation (Delreux and Adriaensen this volume). In other words, these two conditions determine the absolute delineation of the model’s scope of application, yet they do not come without any difficulties. Particularly, the question whether a contractual relation is established might be more challenging than initially expected. The contributions in this volume have revealed that the contracts can range from Treaty articles (Helwig this volume), binding negotiation mandates (Conceição-Heldt this volume; Gastinger this volume), up to rules of procedure (Laloux this volume; Reykers and Beach this volume) and even informal practice (Kroll this volume). Whereas the concept “orchestration ”—defined as a mode of governance which consists of the mobilization of an intermediary actor on a voluntary basis in pursuit of a joint governance goal—was proposed to differentiate the soft type of contracts from the hard, legally binding, variants (Abbott et al. 2015; Blauberger and Rittberger 2015), it remains unspecified “how hard” a contract should be before the principal–agent model becomes an appropriate analytical model. Therefore, it continues to be important that principal–agent scholars are explicit on what their act of delegation is and, in the case of the softer variants, justify why such a soft contract can be considered as establishing a principal–agent-proof relationship.

3.2 New Explanatory Variables and the Challenge for Generalization and Theory-Building

A second opportunity that emerged from the chapters is the identification of new conditions that explain the politics of delegation and discretion. Empirical research applying the principal–agent model thus has the potential to reveal new explanatory variables that travel beyond the studied case(s) and contribute to the broader principal–agent literature. When we study more complex decision-making structures through a principal–agent lens, it is more likely that the explanations we find for the politics of delegation and discretion go beyond the established literature. Various contributions uncovered such new explanatory variables: Conceição-Heldt (this volume) points towards the effect of preference divergence amongst a multiplicity of principals on agent discretion, Helwig (this volume) draws attention to interaction between rivalling agents as an explanatory factor for (limited) discretion, whereas Laloux (this volume) argues that the design of a collective agent is explained by a desire within the collective principal to mirror its proper dividing lines. Likewise, reflecting on the occurrence of non-exclusive delegation, Dijkstra (this volume) points at the need for an analysis at the level of individual principals and the usefulness of exploring the relevance of a temporal dimension.

Identifying additional conditions to explain the origin and consequences of principal–agent relations clearly adds to our knowledge but it also fuels a potential challenge for the model. As the list of potential explanations grows, our ability to arrive at more generalizable findings diminishes. Idiosyncratic explanations are more likely to occur as we continue to specify the particular nature of the principal–agent relation we are studying. The object of delegation, the nature of the (complex) principal and agent, the available control mechanisms, the position within a broader chain of delegation, and the interaction with other actors outside the principal–agent relation may all form part of the conditions that explain the occurrence of a specific form of delegation (or degree of discretion). This multitude of conditions one needs to account for pushes the researcher towards more in-depth case-study analyses (see Sect. 3.3) but may also reduce the propensity for the principal–agent model to generate findings that easily traverse to other empirical cases within or outside the EU (Reykers and Beach this volume). The observation of non-exclusive delegation (Dijkstra this volume) or agents managing information asymmetry (Coremans and Kerremans this volume) are more common phenomena than the counterintuitive case of delegation to a collective agent (Laloux this volume). Likewise, finding relations in the EU or in other political systems that resemble the one between the Council and the European Council (Kroll this volume) or between the Commission and the High Representative (Helwig this volume; Plank and Niemann this volume) will be quite challenging.

In the Introduction, we stated the aspiration that—by thoroughly engaging with the EU’s complex decision-making system—principal–agent analyses of the EU could deliver a contribution to the wider principal–agent literature beyond the EU studies. A latent challenge we face, however, is that the peculiarities of the EU system we lauded as providing an ideal breeding ground for theoretical innovation may very well jeopardize this ambition. Still, as one of the largest polities in the world, one can easily argue that any improvement in our knowledge of the institutional politics behind EU decision-making is relevant in and of itself. In short, whether informing the broader theoretical debate or advancing our understanding of the EU’s political system is our aspiration, there remains a clear potential for updating the principal–agent model in EU studies.

3.3 Conceptual Refinement , Conceptual Stretching and Conceptual Overload

A third and final contribution to the literature concerns the refinement of the concepts used in principalagent analyses. As we uncover an increasing variety in patterns of delegation, it seems inevitable that concepts originally conceived for a specific context (e.g. the US Congressional politics literature) lose some of their analytical potency. For instance, with the advancement of agent-centric analyses, the distinction between discretion and autonomy has become ever more mystified (see also Conceição-Heldt this volume). With control instruments having a largely conditional and even latent effect—and considering that there is an important distinction between the availability and the activation of control mechanisms—it is hardly tenable to treat discretion as something fixed in the act of delegation. Moreover, the many actions by the agent, which not necessarily imply self-serving strategic behaviour (Coremans and Kerremans this volume; Plank and Niemann this volume; see also the discussion on interest- vs structure-induced discretion in Delreux and Adriaensen this volume), are difficult to study empirically through the existing terminology.

The main challenge related to the quest for conceptual refinement is that a balance needs to be found between stretching existing concepts to the extent that they regain their functionality on the one hand, and refining or even creating ever more new concepts to emphasize their distinctiveness on the other hand. In other words, the risks of conceptual stretching and conceptual overload have to be balanced as caution is due both when stretching existing concepts and when creating new ones.

Conceptual stretching occurs when the researcher relaxes the set of assumptions that define a concept—something known as the concept’s intension—to increase the range of empirical cases where the concept holds sway, also known as the concept’s extension (Goertz 2006). The core assumptions of the canonical principal–agent model have been increasingly relaxed as the model made its ways into political science (Delreux and Adriaensen this volume; Miller 2005). The unitary actor assumption of the principal is but one such example (Conceição-Heldt this volume; Laloux this volume). At the same time, conceptual stretching fuels the need for a further qualification of the broadened concept. Thus, distinctions between collective, multiple, hybrid or complex principals were introduced in the principal–agent literature to address the broad variety of actors that could now be considered a principal. Alternatively, notions of control were widened to include oversight (informal) administrative procedures and even constructed dimensions such as learning effects by the principal or the anticipation of control (Kroll this volume). Here too, a qualification of the concept was considered necessary. In other words, key principal–agent concepts can be stretched to broaden their extension. But as their distinctive analytical power then diminishes, new concepts are introduced to further qualify the stretched concept.

This brings us to the second challenge: the risk of conceptual overload . The increasing use of the principal–agent model has created a large—and probably overcrowded—toolbox of concepts and terminology. In some cases, an observation can be explained through a qualification of several existing concepts. Dijkstra’s account of member states performing tasks that are also delegated to the EEAS is labelled as a case of “non-exclusive delegation” of authority (Dijkstra this volume). Making a similar observation in the domain of migration policy, Menz (2015) instead uses the notion of “principal slippage”, whereas Sobol (2016) refers to “pathological delegation”. Yet others may still interpret it as an extreme form of control: the partial revocation of the act of delegation as it were. The plurality of views can be considered a testament to the flexibility of the model. At the same time, it creates a fragmented literature that lacks a common vocabulary. The risk of the principal–agent model becoming largely self-referential can only be countered through a common reference framework, which is likely to facilitate communication across disciplines, policy fields and political systems. The distinction between the politics of delegation and the politics of discretion used in this book may be a first step to provide some structure in an ever-growing literature.

A promising—yet currently fairly underused—way to avoid the creation of new concepts is to mobilize existing knowledge on the topic of one’s study from related theoretical and empirical traditions. Most contributors in this volume were cognizant of the earlier applications in the area of US politics as well as the applications found in the area of international relations, but applications in organization theory, public administration and economics were less prominent in the cited references. While we do not wish to advocate an uncritical transfer of findings from one context to the next, in order to push the literature forward it is of the utmost importance that principal–agent scholars are aware of the developments that take place in tangent fields of the model—and mobilize these developments when appropriate. Doing so will not only foster a stronger linkage between the often separated traditions, it will also avoid lengthening the principal–agent dictionary which already acts as a barrier to certain audiences.

4 The Methods We Employ

As the analytical and theoretical designs grow more refined to address the complexity of the empirical reality they are studying, so too the methods need to adjust accordingly. Yet, the main opportunity for principal–agent scholars is that they can already benefit from well-established appropriate methods. To begin with, when mapping the hierarchical relation amidst a wider institutional framework, a graphical representation as suggested in this volume can be a helpful tool. And when subsequently studying the politics of delegation and the politics of discretion, methods that can cope with the large range of conditions are called for.

Here, process-tracing proves to be a viable option to overcome the problem of over-determination as it is particularly useful to lay bare the causal mechanisms that lead to a specific effect (or discard those that were not at work) (Reykers and Beach this volume; Kroll this volume). At the same time, process-tracing is useful to cope with the problem of observational equivalence that plagues much of the politics of discretion (Damro 2007; Weingast and Moran 1983). Through the intimacy with a specific case, researchers can—more convincingly—identify the relevant mechanisms driving delegation and discretion, discarding alternative explanations along the way (Reykers and Beach this volume). Likewise, the unintended consequences of a particular design of the principal–agent relationship can be singled out and linked to the balance of power between the principal and the agent.

In order to deal with the risk of the principal–agent model becoming restricted to idiosyncratic explanations and the underlying wild growth of alternative explanations (see Sect. 3.2), comparative research designs and methods can be helpful to identify the conditions that explain a particular pattern of delegation or a certain degree of discretion. Particularly in the event that a specific condition for delegation or discretion has already been well established (e.g. through process-tracing) and one is more concerned with its external validity, comparisons will lead to more progress in the principal–agent field. The units of analysis for such comparisons can be different policy initiatives (Gastinger this volume; Conceição-Heldt this volume) but also the various delegated tasks (Helwig this volume) or stages in time (Laloux this volume). In short, the methodological tools to address most of the challenges we highlighted above are available and they seem to be particularly fruitful when combined in a mixed-method design where for instance the cross-case comparison allows for identifying conditions and the within-case process-tracing for unravelling the causal mechanisms.

For many research designs in the EU studies, medium-N comparisons, using for instance configurational analyses such as qualitative comparative analysis (QCA) (Delreux 2009), are a fruitful venue for identifying necessary and sufficient conditions. Yet also large-N comparative designs, where a large number of units of analysis are quantitatively examined, do not have to be excluded a priori from the methods menu—although their application in principal–agent analyses on the EU is rare (Franchino 2005). Such statistical analyses would of course require standardized measurement tools for systematically quantifying delegation, control and/or discretion, which does not come without problems either.

Nevertheless, two specific challenges regarding methodology persist that warrant our attention. A first methodological challenge concerns identifying the appropriate level at which the analysis needs to be conducted. Gastinger (this volume) shows that disaggregating an actor in the principal–agent relation into its sub-units does not always add further explanatory power, yet breaking open the black box of the agent proved to be useful in Laloux’s analysis (this volume). The question on the level of analysis is not only pertinent regarding the actors, also the policy process or the tasks delegated to an agent can be cut up into an infinitesimal number of parts. It remains to be seen whether the additional explanatory power warrants the increase in complexity. In other words, the researcher has to weigh the costs of increased complexity in terms of the analytical demands against the benefits in explanatory power. On the one hand, a parsimonious model not only clarifies the argument, but also widens the possibility that the studied setting may be observed elsewhere enhancing the external validity of the claimed relations (see Sect. 3.2). On the other hand, a practical, methodological advantage of disaggregating units of analysis is that the potential number of observations increases, which opens the possibility for the appealing comparative research we advocated above.

A second methodological challenge is that the methods appropriate to deal with the uncovered complexity are what Hay (2016) termed “high tariff” methods. They imply a large investment of time and resources from the researcher and, therefore, should only be applied if the expected findings warrant such an investment. This not only applies to process-tracing (Reykers and Beach this volume) but also to comparative research methods, which necessitate a standardized measurement of independent and dependent variables. Again, a trade-off between expected added value and the availability of time and resources will determine how researchers are going to deal with this challenge.

5 The Future of principal–agent Analyses in EU Studies

Despite the apparent challenges, it is not only desirable but also necessary for the principal–agent model to adapt to a changing empirical context, albeit always within the remit of our necessary conditions (rationality and an act of delegation). Inability in the principal–agent community to reap the opportunities in terms of the questions asked and potential contributions to the literature would restrict the principal–agent model’s applicability to a minor set of cases at best, or even make it obsolete at worst. The theoretical, conceptual and methodological challenges we identified in the previous sections can be overcome by a greater investment in conceptual thinking, data-gathering and data analysis. They are mostly a plea towards greater reflection on the choice of research questions that warrant our scarce intellectual and material resources.

Under these conditions, the principal–agent model is likely to remain a vital tool not only to explain further (dis)integration (politics of delegation) or to understand who is driving policy-making in the EU (politics of discretion), but also as a starting point for normative reflection on the many debates that are emerging in a post-Brexit climate. Calls to restructure the EU or rethink the balance of power between the national and EU level are resonating with an ever-growing Eurosceptic audience. Still, caution is due as the model’s overly reductionist focus impairs its potential for normative analyses (Brandsma and Adriaensen this volume). However, seeing that hierarchical relations are bound to remain the clearest accountability mechanisms for a critical public that mostly evaluates EU politics through a national reference framework, such partial normative analyses are preferable to the inexplicit normative theorizing that too often takes place at the moment. The principal–agent model and the lessons drawn from this volume may be a useful starting point to inform a debate that is likely to continue for the EU’s lifetime.


  1. 1.

    We thank Gijs-Jan Brandsma, Evelyn Coremans, Markus Gastinger, Daniela Kroll, Thomas Laloux and Yf Reykers for their constructive comments on earlier drafts of this conclusion.



  1. Abbott, K., Genschel, P., Snidal, D., & Zangl, B. (2015). International organisations as orchestrators. Cambridge: Cambridge University Press.Google Scholar
  2. Blauberger, M., & Rittberger, B. (2015). Orchestrating policy implementation: EU governance through regulatory networks. In K. Abbott, P. Genschel, D. Snidal, & B. Zangl (Eds.), International organizations as orchestrators (pp. 39–65). Cambridge: Cambridge University Press.Google Scholar
  3. Brandsma, G., & Adriaensen, J. (2017). The principal–agent model, accountability and democratic legitimacy. In T. Delreux & J. Adriaensen (Eds.). The principal–agent model and the European Union (pp. 35–54). London: Palgrave MacMillan. Google Scholar
  4. Calvert, R., McCubbins, M., & Weingast, B. (1989). A theory of political control and agency discretion. American Journal of Political Science, 33(3), 588–611.CrossRefGoogle Scholar
  5. Conceicão-Heldt, E. (2017). Multiple principals’ preferences, types of control mechanisms, and agent’s discretion in trade negotiations. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 203–226). London: Palgrave MacMillan.Google Scholar
  6. Coremans, E., & Kerremans, B. (2017). Agents as information asymmetry managers in EU trade policy-making. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 227–253). London: Palgrave MacMillan.Google Scholar
  7. Damro, C. (2007). EU delegation and agency in international trade negotiations: A cautionary comparison. Journal of Common Market Studies, 45(4), 883–903.CrossRefGoogle Scholar
  8. Delreux, T. (2009). The EU negotiates multilateral environmental agreements: Explaining the agent’s discretion. Journal of European Public Policy, 16(5), 719–737.CrossRefGoogle Scholar
  9. 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
  10. Dijkstra, H. (2017). Non-exclusive delegation to the European External Action Service. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 55–81). London: Palgrave MacMillan.Google Scholar
  11. Elsig, M., & Dupont, C. (2012). European Union meets South Korea: Bureaucratic interests exporter discrimination and the negotiations of trade agreements. Journal of Common Market Studies, 50(3), 492–507.CrossRefGoogle Scholar
  12. Franchino, F. (2005). A formal model of delegation in the European Union. Journal of Theoretical Politics, 17(2), 217–247.CrossRefGoogle Scholar
  13. Gastinger, M. (2017). Effects of contestation within a collective agent in EU trade policy-making. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 181–202). London: Palgrave MacMillan.Google Scholar
  14. Goertz, G. (2006). Social science concepts: A user’s guide. Princeton: Princeton University Press.Google Scholar
  15. Hay, C. (2016). Process tracing: A laudable aim or a high-tariff methodology? New Political Economy, 3467, 1–5.Google Scholar
  16. Helwig, N. (2017). Agent interaction as a source of discretion for the EU high representative. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 105–129). London: Palgrave MacMillan.Google Scholar
  17. Héritier, A. (2015). Covert integration in the European Union. In J. Richardson & S. Mazey (Eds.), European Union: Power and policy-making (pp. 351–369). New York: Routledge.Google Scholar
  18. Kroll, D. A. (2017). Manifest and latent control on the council by the European council. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 157–180). London: Palgrave MacMillan.Google Scholar
  19. Laloux, T. (2017). Designing a collective agent for trilogues in the European parliament. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 83–103). London: Palgrave MacMillan.Google Scholar
  20. Leuffen, D., Rittberger, B., & Schimmelfennig, F. (2012). Differentiated integration. Explaining variation in the European Union. London: Palgrave Macmillan.Google Scholar
  21. Menz, G. (2015). The promise of the principal–agent approach for studying EU migration policy: The case of external migration control. Comparative European Politics, 13(3), 1–18.CrossRefGoogle Scholar
  22. Miller, G. (2005). The political evolution of principal–agent models. Annual Review of Political Science, 8(1), 203–225.CrossRefGoogle Scholar
  23. Plank, F., & Niemann, A. (2017). Impact of the agent’s environment on discretion in the field of EU conflict resolution. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 131–155). London: Palgrave MacMillan.Google Scholar
  24. Reykers, Y., & Beach, D. (2017). Process-tracing as a tool to analyse discretion. In T. Delreux & J. Adriaensen (Eds.), The principal–agent model and the European Union (pp. 255–281). London: Palgrave MacMillan.Google Scholar
  25. Sobol, M. (2016). Principal–agent analysis and pathological delegation: The (almost) untold story. Governance, 29(3), 335–350.CrossRefGoogle Scholar
  26. Weingast, B., & Moran, M. (1983). Bureaucratic discretion or congressional control? Regulatory policymaking by the federal trade commission. The Journal of Political Economy, 91(5), 765–800.CrossRefGoogle Scholar

Copyright information

© The Author(s) 2017

Authors and Affiliations

  • Johan Adriaensen
    • 1
  • Tom Delreux
    • 2
  1. 1.Maastricht UniversityMaastrichtThe Netherlands
  2. 2.Université catholique de LouvainLouvain-la-NeuveBelgium

Personalised recommendations