Leaking, leak prevention, and decoupling in public administrations: the case of the European Commission

Abstract

The European Commission is frequently faced with leaks, much like other public administrations in Western democracies. While executive leaders often criticise the practice of leaking as an unwanted breach of confidentiality and secrecy, leak prevention is usually not taken seriously compared to the scale of the phenomenon. This article discusses leaking and leak prevention more broadly and analyses the efforts of the European Commission to prevent leaks. It finds that leaking and leak prevention were regularly discussed at the highest level of the EU’s executive between 2006 and 2015. However, few Commission officials have been sanctioned for leaking in that period, and mostly for leaks that appear unrelated to the substance of those discussed at the political level. This mismatch is explained by a decoupling of talk and action regarding leak prevention, allowing the European Commission and other public administrations to manage inconsistencies in competing internal and external demands for openness and confidentiality.

Keywords:

• European Commission
• secrecy
• leaks
• decoupling
• public administration

By virtue of their unvarnished nature, leaks have evolved into the realest of facts. This epistemological status has been fortified in recent years with a series of spectacular leaks that successfully reframed official narratives as grand fictions. (Herrman 2017)

In the past decade, a few spectacular megaleaks – from Wikileaks’ Cablegate to a series of leaks on the EU‒US Transatlantic Trade and Investment Partnership (TTIP) – have shifted global attention to matters of secrecy, transparency, and the role of whistleblowing (see for example the volume edited by Brevini et al. 2013). Beyond the exceptional nature of large-scale leaks such as the ones by Edward Snowden or Chelsea Manning, leaks are a common phenomenon in everyday politics (Reich and Barnoy 2016). For example, regular leaks dominated the first months of the new Trump administration (see Herrman 2017), not megaleaks, with leak investigations tripling under attorney general Jeff Sessions (New York Times 2017). The newly appointed Juncker Commission had similar experiences during its early days in 2014 and 2015.

Researchers have studied leaking of this more common type in various Western political systems, including in the USA (Alaimo 2016; Fenster 2014; Pozen 2013; Prémon 2010, 2012), the Netherlands (Bovens et al. 1995; de Jong and de Vries 2007), Israel (Reich 2008), Australia (Flynn 2006, 2011), Germany (Spörer-Wagner and Marcinkowski 2010), and the European Union (Herrmann 2015; Patz 2014, 2016). A key finding has been that, despite the frequency of leaks, public administrations most of the time do not prioritise leak prevention (Bovens et al. 1995; Fenster 2014: 316; Pozen 2013: 515). However, there is limited research tracing and explaining the dynamics of both successful or failed leak prevention, and matters of leaking still do not feature prominently in the ongoing debate on the politics and regulation of secrecy, confidentiality, and transparency in the EU (Abazi 2015; Curtin 2014; Galloway 2014; Hillebrandt and Novak 2016; Perera et al. 2014; Rosén 2015; see Rittberger and Goetz 2017 in this issue).

The central question for this article therefore is why the prevention of leaks by public administrations seems so unsuccessful. The argument made follows a political science and public administration conceptualisation of ‘leaking [as] an instrument in the political and administrative struggle for power’ (Bovens et al. 1995: 23) and as one influence strategy among others in bureaucratic politics (Prémon 2010: 22–5). It is hypothesised that lax leak prevention is part of the ‘organised hypocrisy’ inside public administrations, where talk and action become organisationally decoupled (Brunsson 2006). This builds on previous public administration research on the European Commission as the EU’s core executive (Hartlapp et al. 2013; Hustedt and Seyfried 2016; Kassim et al. 2013) and on the previous observation of decoupling in the European Commission (Boswell 2008). It is expected that the dynamics of leak prevention found in the Commission as a ‘normalised’ public administration (Wille 2013) are comparable to national public administrations. Studying the Commission therefore can provide comparable insights on how other public administrations are expected to approach leak prevention.

The article is structured as follows: After conceptualising leaks in the second section, and after presenting the theoretical argument on why public administrations employ decoupling when it comes to leak prevention in the third section, the fourth section discusses key findings on the dynamics of leaking and leak prevention in Western political systems. Then, building in particular on recent research on ethics management in the European Commission (Nastase 2017), the Commission is presented in the fifth section as a normalised public administration faced with conflicting demands as regards openness and confidentiality that are EU-specific to some degree but are not unique. The following section then provides an overview of the formal rules and measures put in place in the Commission to prevent leaking. The penultimate section presents novel data on decoupling between talk and action on leak prevention by contrasting the frequency and substance of discussions of the Commission on leaking and leak prevention (‘talk’) – covering the weekly meetings of the College of Commissioners, of the chefs de cabinet (CdCs), and of the directors-general (DGs) of the Commission in the period 2006–2015 – with data on the few disciplinary sanctions for leaking (‘action’) that can be found in the same period. In the conclusion, these observations are considered in view of the comparative study of leaking and leak prevention in public administrations.

Conceptualising leaking

There are various definitions and conceptualisations of what constitutes a ‘leak’. Bovens et al. (1995: 19) define leaking as ‘making confidential information public by office-holders on the basis of anonymity’. They distinguish leaking from whistleblowing, arguing that the former concerns ‘any type of information’ whereas the latter concerns only revelations of abuse. Whistleblowers may decide to go public by leaking to the press, but they may also only share information with internal or external accountability bodies without necessarily leaking, making whistleblowing an analytically separate category. In the public discourse around the megaleaks of recent years, both terms have often become interlinked, underlining the need for clearer conceptual distinction.

Broader conceptualisations see leaks as a ‘violation of confidentiality (1), done intentionally to further one of [sic] more interests (2), involves information exchange (3) and is done anonymously (4)’ (de Jong and de Vries 2007: 217). Some argue that leaking may refer to disclosures that can be either unauthorised or actually authorised at higher level (Pozen 2013: 522). Others consider only unauthorised disclosures meant to disrupt decision-making procedures to be leaks in the narrow sense (see Prémon 2010: 15–16). This paper follows a wide definition of leaking as the non-official disclosure of information to persons outside an organisation. By focusing on ‘non-official’ disclosures, leaks are distinct from proactive transparency, i.e. disclosures that are officially going into the public domain, for instance by publication on official websites or following freedom of information requests. Also excluded is information officially distributed to a selected group of outsiders without publication of the information, for example during early informal, yet officially authorised, consultations. Defined as such, leaks do not contribute to government transparency, because leaked documents will not be considered in the public domain by the public body from which they have emerged; because they provide only selective access based on arbitrary decisions by the leaker; and because the authenticity or legitimacy of leaked documents may be contested and can undermine an open public discourse. Thus, even though the notion that leaks are necessarily negative remains contested (Bovens et al. 1995; de Jong and de Vries 2007), they clearly demonstrate where more transparency is required instead of systematically enhancing transparency in a given political system.

Explaining decoupling and organised hypocrisy in leak prevention

Leaks have always been an epiphenomenon of the struggle between secrecy and political control on one side, and between transparency and openness on the other. On one side, there is the ‘hierarchist worldview’ on information where the rules governing access to information are ‘dominated by the needs of established authority’ (Hood 2010: 999), on the other side there is a more ‘egalitarian’ world view in which transparency is seen ‘as a human right’ and as an instrument for ‘effective accountability’ (Hood 2010: 1001). Where those two value systems collide, leaks become more likely because demand for information will be higher than supply (Patz 2016). Following the arguments by Pozen (2013) and Fenster (2014), the expectation is, however, that leaks are not completely preventable, and nor should it be desirable for public administrations in democratic societies to completely ban leaks. The key questions are thus how public administrations manage those conflicting value systems, and how much effort those in favour of secrecy or in need of control by confidentiality are willing to invest in preventing leaks.

The central argument made in this paper is that, when large organisations such as public administrations are faced with conflicting value systems, such as between transparency and secrecy, or between public participation and political control, one expects to find organisational solutions to accommodate these inconsistencies (Brunsson 1986). Whereas some organisations that are dominated by talk and political disagreements, such as parliaments, have been designed for inconsistencies, other organisations dominated by a need for action are, in essence, built for consistency in order to achieve common outputs. Complex public administrations usually combine both talk and action, and they therefore have to organise internally in such a way that both worlds become reconciled (see Brunsson 1986: 182–3). This requires one or several forms of decoupling of talk and action inside such organisations, which is done by decoupling in time, by topic, by environments, and by organisational units (Brunsson 2006: 32–9; see also Brunsson 1986: 174–6).

Translating this to the case of leak prevention, one could formulate the following general hypothesis:

H_General: Under the condition of a value conflict between transparency (openness, public participation, inclusive decision-making, display of multiple views, etc.) and secrecy (control, limited participation, core-group decision-making, display of unity, etc.), public administrations will decouple talk and action on leak prevention.

Hypothesis H_General implies that, whenever public administrations are not faced with conflicting expectations regarding secrecy and transparency, one should find talk and action of leak prevention to match each other. Non-democratic political systems or closed societies that do not encourage open discourse and participation of interest groups, for example, may not face a strong value conflict in this regard. One would thus expect to find less decoupling than in open systems and much harsher leak prevention mechanisms. In most democratic systems, however, this value conflict should be visible among the general public and also inside public institutions. Whereas there will be a general understanding in open societies that some type of information held by public administrations (for example personal data or security-related information) should remain secret, other types of information will be expected to become public sooner or later. In the former case, one would expect talk and action on leak prevention to match inside public institutions. In the second case, it should be more likely to observe decoupling when public administrations find that they prefer to keep information confidential but are faced with legitimate demands for publicness.

For such decoupling in leak prevention, Brunsson’s model suggests multiple options (see above). H_General can thus be adapted by specifying the dependent variable (DV). Each of these three DVs has different implications for the empirical observations that can be made:

•	Decoupling in time (DVtime): This implies that discussions on leaks and leak prevention at time t1 are not followed by observable actions at time t1+x. This could mean a discussion about the need to put in place stricter rules, but not actually strengthening the rules; putting in place stricter rules, but not increasing the number of investigations or convictions; complaining about a leak, but never prosecuting anyone for it.
•	Decoupling by topic (DVtopic): When it comes to leak prevention, decoupling talk and action on leaks by topic could be found where public discussions about leaking and leak prevention concern highly political issues but actual leak prosecutions (‘actions’) happen mostly in those cases that are legally simple and clear-cut but not politically salient. Whereas Brunsson (1986, 2006) also distinguishes a decoupling by environment, the environment can be seen as a function of topics. Decoupling of talk and action would then be found in relation to all topics where public administrations are faced with a high level of value conflict between secrecy and transparency in their environment. Decoupling offers talk on leak prevention to those in the environment who favour secrecy in a certain policy domain, while non-action is offered for those in favour of transparency.
•	Decoupling by organisational units (DVunit): Within complex public administrations, organisational decoupling in relation to leak prevention can be found when the highest levels are complaining publicly about the need to prevent leaks but are leaving the actual actions on leak prevention to lower levels of the administration, such as investigative or ethics departments. Leaks that are clearly coming from the political level as part of bureaucratic politics (see Bovens et al. 1995) then may not actually be investigated but simply discussed to pretend to the public that they were an issue, whereas the internal investigation units are more concerned with leaks by officials without political roles, cases that may never even be discussed at the political level.

The previous literature on leaking and leak prevention suggests that all three types of decoupling may be interrelated. Separating them analytically still allows for a better understanding of how the collision of different value systems with regard to secrecy and transparency is managed in the practice of leak prevention. Empirically, such observations of decoupling imply finding, first, that the legal and practical measures to prevent leaks match neither the quantity or quality of political complaints about leaking nor the overall scope of leaking as a phenomenon. They also imply, second, that the prosecution of leaks does not happen in those cases that are the most debated, and that most prosecutions should instead be found where the value conflict between publishing information and keeping it secret is weakest. However, there is a risk of finding false positives of decoupling, for instance when high-profile leak investigations do not lead to the identification of the actual leaker(s), and therefore cannot lead to an observable prosecution. Similarly, when suspected leakers are not prosecuted but face more subtle measures, such as being excluded from the respective information flows in the future, this type of social sanction may not be observable even when the case is politically debated. Nevertheless, previous research on leaking and leak prevention presented in the next section suggests that decoupling should be considered a default option in most Western public administrations.

Leaking and leak prevention in Western public administrations

Estimating the real number of leaks in a given political system is challenging (Reich 2008). It is estimated, for example, that between 2.3% and 20% of all news items in the US are based on leaks (Reich and Barnoy 2016: 888). In the Netherlands, leaking occurs in all major policy areas (de Jong and de Vries, 2007), and even leaking of national security information was found to be ‘a routine daily occurrence’ in the US (Willard Report 1982). This suggests that leaking far exceeds a few headline-grabbing cases of whistleblowing to the public or practices of revealing state secrets to enemies. Instead, officials regularly try to support their institutional and policy preferences by leaking (Bovens et al. 1995: 24). Where leaks created public debates during political or administrative negotiations in Germany, these debates mostly reflect already existing and well-known conflicts among actors (Spörer-Wagner and Marcinkowski 2010). This is why leaks may affect the dynamics of decision-making much more than the substance (Prémon 2010: 292–3). A key goal may be to try to discredit an opponent by making an unpopular position public, or by revealing an early positioning of an actor to lock in this position against future revisions (Bovens et al. 1995: 26–8), thus shifting power balances.

Most leaking happens orally (Reich and Barnoy 2016) and in relatively stable networks based on regular interaction between policy-makers and the receivers of leaks (de Jong and de Vries 2007; Patz 2014). Traditionally, leaks were thought to come mainly from senior officials, those with good access to information, political interests of their own, and regular contact with outside actors, for example in Israel (Reich 2008) or the US (Pozen 2013: 529–30). Qualitative studies indicate, however, that regular civil servants in the USA leak as much as political appointees (Alaimo 2016), while journalists in Australia report their preference for leaks from the middle ranks because information from these levels is further removed from the political spin but more interesting than technical information from lower levels (Flynn 2006). Recently, there seems to be a trend that leaks are increasingly coming from less senior officials. Technology and the ability of junior officials to access and copy politically sensitive or confidential information more easily are seen as the drivers of this change (Reich and Barnoy 2016).

Given that ‘[n]o bureaucracy is comfortable with unauthorized disclosures’ since they expose real or constructed ‘organizational deviance’ (Flynn 2006: 269), leak prevention should address all administrative levels, and needs to take technological change into account. It should not just be a matter of protecting highly sensitive or classified information and not just a matter of administrative ethics but also an important means of political control. Consequently, preventing all leaks would require a large-scale effort. The question thus is why and when leak prevention actually becomes important for public administrations. This question is particularly interesting in democratic systems such as most Western democracies where public discourse and participation of interest groups are key elements of the democratic process, both of which may be supported by leaks. These values of openness, however, can and will clash with values of professional confidentiality, the protection of legitimate secrets, or the interests of political and administrative leaders who try to steer complex decision-making processes.

In line with the arguments made in the previous section about the decoupling of talk and action, previous research suggests that decoupling does indeed happen as a result of value conflicts, although empirical studies are rare. For the US case, even when there are thousands of leaks of classified information, there have been only ‘roughly a dozen criminal prosecutions’ for national security-related leaks (Pozen 2013: 534). The number of publicly known leak cases in the US followed up by administrative sanctions is also low compared to the scale of the phenomenon (Pozen 2013: 540). Pozen deconstructed previously existing arguments that these observations relate to the lack of capacity of the government or agencies to detect and sanction leakers. Instead, he shows how ‘leaking is a heterogeneous activity that occurs in a repeat-play environment’ (Pozen 2013: 546). Political leaders complain about leaks, but at the same time they know that they will make use of the practice at a later stage. Hard prosecution in one case might limit their own ability for targeted leaks, or undermine their credibility when they do not go after leaks that are in their own interest. Combining the rhetoric of denouncing leaking while not following up is a key strategic choice to reflect this contradictory situation. In addition, leaks are also means ‘through which the executive branch speaks to itself’ (Pozen 2013: 577) and to those responsible for oversight (Pozen 2013: 582). Thus ‘a policy of permissive neglect towards leaking may … be efficiency-enhancing relative to the baseline of colossal official secrecy [because] overclassification threatens to stifle important decisional inputs and cause policy sclerosis’ (Pozen 2013: 577).

Seeing that previous research shows that leaking is frequent and that complaints about leaks are common but that leak prevention is rare in Western systems, there is a strong indication that decoupling is a default option. However, detailed empirical research beyond the US and the Netherlands is missing. The following sections thus present novel evidence based on a systematic study of talk and action on leaking and leak prevention in the European Commission as a normalised public administration. These findings add up to a comparative understanding of leak prevention dynamics, in particular in political systems in which there is a visible value conflict between openness and confidentiality.

The European Commission: a normalised bureaucracy also when it comes to leaking?

The European Commission is a complex public administration. It possesses the monopoly on proposing new EU legislation, including a monopoly to propose the budget (Goetz and Patz 2016), and also exercises a wide range of political, administrative, and supervisory functions, for example acting as quasi-judicial supervisor in EU competition policy. The Commission is led by a political bureaucracy comprising the Commission President, several vice-presidents, and individual commissioners, collectively forming the ‘College’ with one member per EU member state. Each of the 28 members of the College has their respective cabinet headed by a chef de cabinet. The main administrative structure of the Commission is based on relatively independent line departments called ‘directorates-general’ (and ‘services’), centrally coordinated by the Secretariat-General of the Commission (see Hartlapp et al. 2013 for details).

In recent years, the European Commission has been studied increasingly as a ‘normalised’ public administration (Wille 2013). This means that horizontal (Hustedt and Seyfried 2017) and vertical coordination (Kassim et al. 2017), bureaucratic ethics and ethics management (Nastase 2014, 2017), or administrative reforms in reaction to outside demands and shifting values (Moodie 2016) receive detailed attention. Perspectives and concepts that are used to study core executives and ministerial bureaucracies in most Western political systems are applied to the Commission with increasing detail.

As an international public administration (see discussions in Bauer et al. 2017), a key question therefore has been whether and how the Commission distinguishes itself from national public administrations, for example by its organisation according to nationality. The observation of normalisation implies that, even though the Commission is headed by a College with 28 members nominated by national governments, it is not expected to behave significantly differently than national executives; and that portfolio and functional logics are expected to drive most of the relevant dynamics, including when it comes to leaking. Research has indeed shown that nationality matters very little at the bureaucratic level and, when it comes to the political level of the Commission, much less so than in the intergovernmental European Council (Egeberg 2012). This has been confirmed with regard to values and ethics inside the Commission based on interviews confronting EU officials with a vignette that reflected ‘the inherent tensions between the principles of transparency and confidentiality’ (Nastase 2017: 94) in officials’ interaction with lobbyists. The differences in views were not guided by nationality. Instead, preferences for higher levels of confidentiality were rather related to functions requiring more confidentiality, such as supervisory tasks in the directorate-general dealing with competition cases (Nastase 2017: 108). And while ‘everyone’ interviewed condemned leaking, leaks were seen by officials as ‘a fact of life in the Commission, and the vignette triggered recollections of such occurrences in all DGs’ (Nastase 2017: 100). This confirms both a certain normality of the Commission and the existence of a value conflict, while providing first hints to decoupling between talk against leaking and the prevalence of leaks.

Nevertheless, despite this appearance of normality of the European Commission, there are strong indications that the Commission has been particularly leaky for a long time (Grønbech-Jensen 1998: 191). In the early 1990s, British officials working in the EU considered leaks to be ‘elevated almost to the status of accepted operating procedure’ in the EU. The reason seemed to be that Commission officials needed to be open to the outside world because ‘the Community’s effectiveness depend[ed] as much on engineering consent as on designing technically efficient regulations’. Secrecy was possible to ensure the technical regulation among experts, but openness through leaks ensured public and interest group support (Christoph 1993: 528).

One explanation for the observation of a perceived higher number of leaks in the case of the European Commission could be that, in its transition from a quasi-diplomatic and technocratic international public administration towards a normalised bureaucracy in an ever more politicised supranational political system, the Commission was faced with a stronger value conflict between secrecy and transparency than more established public administrations. Increased leaking was the outlet for this conflict. The introduction of Regulation 1049/2001 on access to EU documents (European Union 2001) alongside the formulation of wide-ranging rules on EU Classified Information (EUCI) (Galloway 2014) at a time when the European Commission also designed the European Transparency Initiative (Cini 2008) in the early 2000s only strengthened this conflict into formalised rules. As a result, the long existing demand for professional confidentiality (see below) stands in opposition to more formalised transparent and participatory approaches, resulting in value conflicts on when to be open and when to favour confidentiality, conflicts visible at the level of individual Commission officials (Nastase 2017: 97–100).

The inconsistency of value systems became obvious recently when the new Juncker Commission was faced with leaks from its first days in office (Corporate Europe Observatory 2014). In response, Jean-Claude Juncker introduced, in early 2015, ‘particularly stringent methods … used to guarantee the confidentiality’ when preparing the Commission’s Communication on the Stability and Growth Pact (European Commission 2015a: 25). However, his fellow commissioners complained about the negative effects of such political measures, voicing

general regret [in the College of Commissioners] that the measures taken to ensure the confidentiality of this complex and sensitive draft communication had resulted in the Members of the Commission and their chefs de cabinet being given insufficient time, in their view, to examine the proposal. (European Commission 2015a: 21)

Whereas this complaint appears to be about internal consultation only, previous research has shown that leaking from the Commission happens at the stage when internal consultations increase (Patz, 2014, 2016, 2017), suggesting that this is also the time when different parts of the Commission try to involve their respective stakeholders, if necessary through leaks. And these leaks can be highly consequential for the policy process: the leak of an early draft of a Commission proposal for the future of the EU Common Fisheries Policy led to a heavy lobbying process and pressure from interested stakeholders, resulting in significant changes before the final draft regulation was officially published (Patz 2014). It is therefore understandable that the Commission leadership but also departmental leaders have an interest in discouraging leaks through rules and guidelines that prioritise confidentiality over openness, not just in areas of classified information, but also wherever they want to exercise political control.

Leak prevention measures in the European Commission

The protection of confidentiality has been an established rule at the EU level for a long time, without ever stopping leaking. EU officials, including Commission staff, have been bound by the obligation not to leak information since the first EU staff regulations. The key rule has remained, in substance, almost the same over the past 55 years:

An official shall exercise the greatest discretion with regard to all facts and information coming to his knowledge in the course of or in connection with the performance of his duties; he shall not disclose to any unauthorised person any document or information, in any manner whatsoever, not already made public. (EU Staff Regulations 1962: §17, first sentence)

This rule, slightly amended but still §17 in today’s version of the EU Staff Regulations (2017), is implemented further by a range of rules and guidelines. Based on a request for access to documents answered by the Commission’s Directorate-General for Human Resources, Unit E3 ‘Ethics & Ombudsman’, in early 2017, the rules currently in place inside the Commission comprise Commission Decision 2015/443 of 13 March 2015 on ‘Security in the Commission’ (European Commission 2015b), in particular Article 9 on ‘Security measures regarding information’, and Commission Decision 2015/444 of 13 March 2015 on the security rules for protecting EU classified information (EUCI), in particular Article 8 on ‘Breaches of security and compromise of EUCI’ (European Commission 2015c). These are legally binding rules, which foresee special internal investigations whenever there is the suspicion that classified or sensitive documents have been leaked. Furthermore, there are three sets of guidelines for how to prevent unauthorised disclosure, including ‘The practical guide to Staff Ethics and conduct’, which contains details on how to interpret §17 of the staff regulations; a two-page leaflet titled ‘Security Provisions: How to protect the Commission’s sensitive information’; as well as further ‘Guidelines for staff on the use of the Commission’s information and communication technology (ICT) services’. The latter specify that email traffic and telephone contacts of Commission staff are monitored and that this information can be used in internal investigations.

In addition to these rules and guidelines directed at individual staff members, various political measures have been implemented to increase confidentiality inside the Commission. These include reducing the circle of people involved in critical decision-making on the EU’s long-term budget (Goetz and Patz 2016: 1049); holding sensitive sessions of the weekly chef de cabinet meetings where fines for member states were decided only shortly before the main Commission meeting so that there was no time to leak results before the official announcement (European Commission 2010b: 8); shortening consultation periods for the development of crucial policies (European Commission 2015a); or delaying the start of interservice consultations until such a time when leaks were expected to receive less media and member state attention (European Commission 2012a). In the new Juncker Commission, the Commission’s Secretary-General and the President’s chef de cabinet also sent a joint note in 2016 ‘on measures aimed at preventing leakage of information and documents’ (European Commission 2016), coined a new anti-leak strategy by the press (EurActiv 2016), to underline the priority that the new Commission leadership attached to leak prevention.

And yet these political measures only indicate that the effectiveness of the formal procedures in place seems limited, showing that much is talk and little is action. This was confirmed in the spring of 2017, when President Juncker’s chef de cabinet and the Secretary-General of the Commission reportedly sent out another note, asking that certain types of documents had to be marked more clearly as ‘preliminary’, basically anticipating leaks and mitigating negative public reactions to leaked documents in advance by clarifying the status of these documents in case they were leaked (Politico Europe 2017) – which thus seems the standard expectation.

These indications of decoupling will be investigated in more detail in the next section, where talk about leaks and action against leakers during the period from 2006 to 2015 are contrasted. In addition to quantitative observation, it is demonstrated that there seems to be no substantive link between debates about politically sensitive leaks at the top of the Commission and the actual anti-leak investigations that lead to sanctions for Commission officials. Instead, leak investigations seem most robust in cases which leave little doubt that rules beyond confidentiality had been breached, reducing the degree of value conflict in these cases.

Decoupling of anti-leak talk and actions in the European Commission between 2006 and 2015

The narrative account of leaking and of various anti-leak measures in the European Commission in the previous section has demonstrated that the underlying conditions for decoupling of talk and action on leak prevention in the European Commission formulated in H_General are met. There is a persistent value conflict between secrecy and transparency that has existed over the past decades, and this conflict is clearly reflected in the current legal framework of the EU. This conflict has resulted in a regular stream of leaks and thus is met with a realisation inside the Commission that leaks are part of the regular function of the organisation. In this section, systematic indications of decoupling along the three dimensions – time (DV_time), topic (DV_topic), and organisational units (DV_unit) – are presented.

To conduct this analysis, Commission documents that reflect talk and action on leak prevention have been analysed, covering the past decade from 2006 to 2015. To measure talk, all public meeting minutes of the weekly meetings of the College of Commissioners for this period have been searched for instances during which discussions about leaks have been recorded. Since these minutes are public by default, any recording of such discussions can be seen as speech that is ultimately meant to be seen by the public. This analysis is complemented with a quantitative assessment of the amount of discussion referring to leak or anti-leak measures in the weekly meetings of directors-general and of the weekly meetings of chefs de cabinet.¹

Action is measured by instances in which Commission officials have been sanctioned for leaking in the period 2006–2015 as identified in annual reports of the Commission’s Investigation and Disciplinary Office (IDOC).² Finding decoupling would mean that there is a disconnect between talk at time t₁ and action at time t_1+x (DV_time); that the talk is about different types of leaks rather than actions that lead to anti-leak sanctions (DV_topic); and that talk and action differ by organisation units (DV_unit). Disproving the hypothesis H_General would mean finding the opposite, i.e. consistency over time, topics, and organisational units, as well as action that matches the actual number of leaks.

The first observation is that, during the 2006–2015 period, the public minutes of the Commission’s College weekly meetings report only 14 instances when leaks were explicitly discussed. Only in 12 of those instances did this concern leaks coming from the European Commission (see Appendix 1 Table A1 for the meeting dates and document numbers).³ This suggests that leaking is rather discussed internally. This is confirmed by a request for access to the documents containing the minutes of the weekly meetings of Commission chefs de cabinet and of directors-general. For the latter, the Commission itself identified eight meetings in the period 2006–2015 during which leaks and anti-leak measures were discussed. In the meetings of chefs de cabinet, leaks seem to be a slightly more frequent topic, with 27 meetings over 10 years in which references to leaks can be found (see overview in Figure 1). In particular, in the first year of a new Commission – 2010 was the first year of the Barroso II Commission, and 2015 the first year of the Juncker Commission – there was an increased number of discussions on leaking at the level of chef de cabinet.

Figure 1. Number of weekly directors-general, chefs de cabinet and College meetings (2006–2015) during which leaks or anti-leak measures were discussed. Sources: Data on College meetings from public minutes; DGs’ and CdCs’ meetings identified by the Commission itself, based on an access to documents request asking for access to all minutes that referred to leaks or anti-leak measures. Own compilation (see Appendix 1 Table A1 for details).

Looking at the substance of the College discussions, most cases relate to concrete policy-related leaks and none explicitly mentions leaks of classified information. Only in a few cases are the leaks discussed in the context of wider working methods of the Commission. One-third of the cases concerned issues that are sensitive to certain member states, such as progress reports on Bulgaria and Romania (2007), recommendations of the Commission to member states on their economic policies in the context of the European Semester (2012) or reports on Cyprus (2012) and on Greece (2012). One-third of cases concerned leaks on more general policy issues such as carbon leakage measures (2010), banking supervision (2012), the preparation of the EU budget (2012) or gender equality on company boards (2012). Two discussions related to leaking in competition cases (2010, 2012) and the other two to leaks in the early days of the Juncker Commission (2014, 2015). The particular spike in cases in 2012 is thus related to the politically sensitive nature of anti-crisis measures and leaks in this context. Interestingly, the WikiLeaks leaks in 2010 and the National Security Agency (NSA) leaks in 2013 did not lead to specific public discussions in the regular College meetings, nor did they lead to more debate about leaks in the College in those years. The public minutes confirm that the increase in discussion at the level of chef de cabinet in 2014–2015 is most likely linked to the urge of the Juncker Commission to react to early-day leaks.

To observe whether talk and action were indeed decoupled between 2006 and 2015, the annual reports of IDOC for this period are analysed for reporting on cases referring to leaks or unauthorised disclosure of documents. Figure 2 shows that the annual reports cover only 11 cases in that period, and just four between 2006 and 2013, as many as in 2015 alone. Even though the overall figures are low, there seems to be no temporal link between public College debates on leaks and sanctions for officials, suggesting a decoupling of talk and action through time (DV_time). This is most obvious for 2012 when leaking was on the Commission agenda seven times and twice on the agenda of directors-general and chefs de cabinet, but only one official (a seconded national expert) was sanctioned in 2012 – even though it is unclear whether this case is connected to one of the seven cases discussed in College. Since no case can be identified for 2013, even the possibility that sanctions for some of the cases of 2012 were dragged into the next year can be ruled out. Nevertheless, the increase in the number of sanctions in 2014 and 2015 could indicate that talk against leaking might have turned into more action recently, in particular as there is a parallel increase in non-public discussions at the level of the chefs de cabinet (see Figure 1). This would coincide with the arrival of Martin Selmayr as chef de cabinet of Juncker, known in the Brussels bubble for his urge to control information flows.⁴

Figure 2. Number of Commission officials sanctioned per year. Source: Own compilation based on IDOC Annual Reports.

However, the analysis of the substance (DV_topic) of the cases to which the sanctions relate indicates very little connection to the leaks discussed at College level, including in 2014–2015. The hypothesis that there is decoupling can thus be upheld. The 2006 case and two of the 2014 cases relate to leaking of information during tender procedures, i.e. neither typical policy leaks nor leaks associated with breaches of political secrecy but rather ones that relate to fraudulent behaviour. Among all 11 cases, only the 2009 case relates to leaking to a news organisation. This particular case, however, was flagged because the official concerned received a payment for providing this information (IDOC 2009: 8). This is thus a case of passive corruption as much as a leak case. Two of the cases of policy leaking to interest groups also involved instances of undue influence, one official in 2011 receiving favours in return for leaking (IDOC 2011: 6) and one case in 2015 relating to an official who not only leaked information but also included the ‘industry stakeholder’ comments in a policy draft without informing the hierarchy (IDOC 2015: 8). Another of the 2015 cases related to a trainee forwarding information to outside recipients, also probably not a high-level leak discussed in the College or by the chefs de cabinet.

In sum, only a 2012 case of leak-related sanctions is closely related to what might have been discussed in one of the College debates. Overall, there is a clear decoupling by topic (DV_topic) between leaks discussed at the highest political level and the cases that are sanctioned based on the formal rules and investigated by the Commission’s internal ethics department. In about half of the cases with sanctions the leaks were linked to some other offence. This underlines how decoupling between talk and action seems to be weaker when there is less conflict about the underlying values as regards the rules violated.

The division of labour between political-level bodies – in particular the College and the chefs de cabinet – responsible for talk and the responsibility to act in concrete cases by the ethics and internal investigations department, IDOC, is also an indication of the organisational decoupling of talk and action when it comes to leaking and leak prevention (DV_unit). Further research is necessary to better understand how the political-level discussions and the formal internal investigations are linked, but access to relevant documents has so far been denied by the Commission.

In summary, leaks are a regular topic at the highest level of the Commission, especially when leaks ‘undermined the principle of collective responsibility and the confidentiality of the Commission’s deliberations’ (European Commission 2014: 20) and when they required additional ‘communication efforts […] in order to convey the right message’ (European Commission 2010a: 27). The temporal, substantive, and organisational decoupling of talk and action on leaks suggests that while, at some point, the political leadership may ask the bureaucracy ‘to review … problems and to propose improvements in working methods and investigations to establish how these documents were leaked’ (European Commission 2012b: 16), there is little empirical indication that these demands are translated into significant changes over time, at least not resulting in action that leads to a decrease in leaking or to an increase in the number of officials sanctioned for having leaked information.

Conclusion

This article has focused attention to the empirical reality of leaks and anti-leak measures as a representation of the struggle between secrecy and openness, in particular in Western public administrations. Building on previous efforts that explain the dynamics of leaks and of leak prevention in the executives of Western democracies, studying the European Commission has shown more systematically than prior research how talk and action on leak prevention are generally decoupled. The observation of organised hypocrisy, in particular in the US and Dutch cases – leaks and complaints about leaks are frequent but active leak prevention and sanctioning is selective – can thus be confirmed for the European Commission. The evidence presented in support of H_General is strongest for decoupling in time and in topic, but further research is required to understand how decoupling in organisational units is organised inside the Commission.

In general, the dynamics observed in the case of the European Commission confirm the image of a normalised public administration in which decoupling serves the purpose of organising conflicting value systems, both inside and outside the administration, and to ensure the effective functioning of an organisation that has to reconcile various goals and competing interests. One key observation in the case of the Commission, for which detailed data could be presented here, that speaks to this conclusion relates to a phenomenon that could also be observed in the case of the Trump administration: in the early months of a new political and administrative leadership, leaking and leak prevention seem to be particularly salient, as values shift from one leadership team to the next, and as the new leadership tries to gain control over their bureaucracies and to ensure that implementing their key priorities is not undermined by leaks. Future research could study these instances in more detail to better understand when and how new leadership tries to reduce decoupling as it is faced with early-day leaks, but ultimately may have to resort to organised hypocrisy as soon as it realises that its efforts are either in vain or lead to unwanted side effects such as increased friction and less efficiency inside its respective public administrations.

It is also noteworthy that the present study could find little indication of leak cases inside the Commission related to classified or highly sensitive information, a domain that is of particular interest in the debate about secrecy in Europe (Curtin 2014; Galloway 2014). This seems in stark contrast to the USA. Access to documents about internal security investigations that could shed light on this aspect has been denied by the Commission so far, but since none of the IDOC investigations presented above mentioned classified information, there is reason to believe that leaking of classified documents is rare. This could be because the prevention measures against leaking of actual secrets may be more successful than in the US case; because of the Commission’s limited competencies in areas of high-level secrecy so that there is comparatively little classified information to leak; or because the Commission is more successful in designing an ethics system where officials dealing with sensitive information have less conflicting values and are thus intrinsically protecting sensitive or classified information (see Nastase 2017).

The present study therefore invites further efforts in understanding leaking and leak prevention inside EU institutions and across a variety of political systems. Under what conditions is leak prevention more successful in reducing the number of leaks? When and why is decoupling between talk and action on leak prevention reduced? The challenge for such comparative research is getting systematic access to documents and information that reflect those dynamics. Revealing information about how leaks are investigated and are dealt with inside public administrations is considered sensitive in itself by public administrations, as indicated by the negative responses and delays to freedom of information requests made to the European Commission. Research efforts that shed light on administrative practices regarding secrecy in general, and leaks in particular, may thus be most successful for political systems that are already relatively open. Developing methodological approaches that deal with this dilemma without having to resort to purely anecdotal evidence is therefore one key challenge for future research into leaking, leak prevention, and the politics of secrecy in Europe.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Ronny Patz is a postdoctoral research fellow in the Department of Political Science, LMU Munich. He is currently researching resource mobilisation by international public administrations in a project funded by the German Research Foundation (DFG). From 2012 to 2014 he worked for the EU office of Transparency International in Brussels on the transparency, integrity, and accountability of EU institutions. [[email protected]]

Acknowledgements

I thank the two anonymous reviewers, the participants at the workshop ‘Secret Politics and Politics of Secrecy in Europe’ on 2‒3 March 2017 in Munich, in particular the workshop hosts and editors of this special issue, Klaus H. Goetz and Berthold Rittberger, as well as Vigjilenca Abazi for commenting on earlier versions of this paper. I am grateful to the European Ombudsman and her office for supporting the access to documents requests made for this research.

Notes

1. The minutes of the weekly meetings of directors-general where leaks have been discussed have been obtained, but they contain very little information on the substance of the issue(s) debated. Access to the content of the chefs de cabinet meeting minutes has been granted only shortly before the publication of this article so that only quantitative aspects could be covered in Figure 1.

2. An additional request for access to documents identifying all investigations on leaks of sensitive or classified documents for the period 2006–2015 has been denied. By a letter of 6 October 2017 (C(2017)6818), the Commission confirms that there have been only six such investigations between March 2015 and March 2017. A complaint with the European Ombudsman is pending at the time of writing.

3. In one further case, the leakage of oil from a capsized ship was discussed, i.e. not an information leak.

4. Interpretation of the author based on informal accounts of two Brussels-based journalists in early 2017.

Appendix 1

Table A1. High-level Commission meetings (2006‒2015) at which leaks or leak prevention were discussed.

Table

Directors-general weekly meetings		Chefs de cabinet weekly meetings		Commission College weekly meetings
Meeting date	Document number	Meeting date	Document number	Meeting date	Document number
		30 January 2006	SEC(2006)1731
		20 February 2006	SEC(2006)1734
		13 March 2006	SEC(2006)1737
		22 January 2007	SEC(2007)1773
		29 February 2007	SEC(2007)1777
21 June 2007	SEC(2007)885	29 June 2007	SEC(2007)1795	27 June 2007	PV(2007)1794 final
		14 July 2008	SEC(2008)1839
		16 March 2009	SEC(2009)1867
10 December 2009	SEC(2009)1697	05 October 2009	SEC(2009)1889
		01 February 2010	SEC(2010)1904
03 June 2010	SEC(2010)676	31 May 2010	SEC(2010)1920	26 May 2010	PV(2010)1919 final
		05 July 2010	SEC(2010)1925
09 September 2010	SEC(2010)1052	13 September 2010	SEC(2010)1929
		15 November 2010	SEC(2010)1936	9 November 2010	PV(2010)1935 final
		06 December 2010	SEC(2010)1939
		31 January 2011	SEC(2011)1946
15 September 2011	SEC(2011)1074
				1 February 2012	PV(2012)1988 final
				14 February 2012	PV(2012)1990 final
		27 April 2012	RCC(2012)2000	25 April 2012	PV(2012)1999 final
				30 May 2012	PV(2012)2004 final
12 July 2012	SEC(2012)465
20 September 2012	SEC(2012)541	17 September 2012	RCC(2012)2016	11 September 2012	PV(2012)2015 final
				17 October 2012	PV(2012)2020 final
		10 June 2013	RCC(2013)2050
		13 January 2014	RCC(2014)2071
		10 November 2014	RCC(2014)2105	14 November 2012	PV(2012)2023 final
		15 December 2014	RCC(2014)2110	16 December 2014	PV(2014)2110 final
15 January 2015	SEC(2015)59	12 January 2015	RCC(2015)2111	13 January 2015	PV(2015)2111 final
		26 January 2015	RCC(2015)2113
		09 February 2015	RCC(2015)2115
		11 May 2015	RCC(2015)2126
		21 September 2015	RCC(2015)2139