Manuscript accepted for publication in the Journal of European Public Policy, 10 July 2020.

1 Introduction

The extent to which the decision-making process is open to the participation of a broad range of voices and stakeholders has often been presented as an important component of the legitimacy of the regulatory governance in the EU (Judge and Thomson 2019). As a result, recent years have witnessed a sustained push to introduce a variety of consultative instruments such as open online consultations, stakeholder consultations or meetings with selected groups (Binderkrantz, Blom-Hansen, and Senninger 2020) designed to provide opportunities for organized groups and citizens to weigh in on the policy initiatives originated in Brussels.

An important recent body of scholarship has investigated how EU regulatory bodies involve stakeholders through consultative processes (Arras and Braun 2018; Beyers and Arras 2019; Binderkrantz, Blom-Hansen, and Senninger 2020; Quittkat 2011), what are the factors driving the use of consultations by European institutions, (Van Ballaert 2017), and the impact that these mechanisms have over the design of policies within the EU (Bunea and Thomson 2015).

For all we do know about European regulatory governance, there is nevertheless still a lot we do not know about how stakeholders engage with these consultative mechanisms and the origins of information that is fed to policymakers through them. A number of studies have studied “who” answers the call for public comments from EU authorities and investigated the number, type, and diversity of the groups that are involved in the EU policymaking process through various consultative mechanisms (Berkhout, Hanegraaff, and Braun 2017; Beyers and Arras 2019; Quittkat 2011; Rasmussen and Carroll 2013). One limitation with existing scholarship in this area is that it when it is not looking at population aggregates, it tends to examine interest groups individually and seperately. Yet stakeholders’ engagement in the design of public policy - European or otherwise - is rarely an individual endeavour. The process through which stakeholders select which consultations to engage with and develop a response is often the result of the interactions with other groups. As Mahoney (2007, 366) puts it, groups frequently “share tidbits of information after committee meetings, forward emails with talking points, hold conference calls to get allies up-to-speed on policy-making developments, send joint letters to policy-makers”.

In this paper we develop a novel text-as-data approach to investigate information exchange among the stakeholders which relies on detecting instances of text reuse among the comment letters submitted by different groups to the same policy proposal. Ideas and position papers float around interest group communities that are engaged in a policy space, and text reuse allows us to observe which groups is sharing text with whom.

We deploy this approach to analyze the characteristics and determinants of the communication networks among the groups that engage with EU consultative mechanisms. To do so, we build a new dataset comprising all the publicly available comments submitted online from interest groups to a wide a range of different policy initiatives by the European Commission. By applying the text reuse approach to this corpus of online comments, we are able to analyze informal communication networks that emerge among stakeholders engaged with the EU-level policymaking process and to answer the following questions: How do information flow across groups that responds to EU-level policy consultations? Which groups are engaged in the exchange of information regarding EU policies? And how do these informal communication networks differ from more formal lobbying coalitions?

Our analysis reveals that the way information flows across groups that engage with the EU-level policymaking process is in many ways distinctive from more formalized instances of coordination among the same actors. First, we show how information exchange networks connecting stakeholders in the EU tend to link fewer groups and have less staying power than formal lobbying coalitions but are also more likely to connect groups based in different countries. Then, we analyze statistically the conditions under which information exchange networks across stakeholders engaging with EU consultative processes are likely to emerge and compare them with formal lobbying coalitions. We find that the sharing of text is particularly associated with groups outside of the business community, although we don’t find support for the notion that this is a “weapon of the weak”. The likelihood of engaging in text reuse is also associated with the number of associational ties that a group has, thus providing support for the notion that trade associations and other groups act as important channels for the sharing of information among stakeholders. We conclude by suggesting further ways in which text reuse could be deployed to investigate the relationship among interest groups in the policymaking process.

2 Literature Review

An established body of literature has long recognized how a key way in which interest groups engage in the policymaking process is by working together. While this literature has primarily focused on the US context (Baumgartner et al. 2009; Hojnacki 1998), recent work has detailed how “coalitional activities” have become a prominent feature of the way interest groups engage in advocacy vis-à-vis EU institutions as well (Beyers and Braun 2014; Beyers and De Bruycker 2018; Klüver 2013; Mahoney 2008; Pijnenburg 1998; Sorurbakhsh 2016). This literature has found that, in the world of EU policymaking, the ties that groups establish with others play a key role in in determining their preferences towards EU policies (Bunea 2014), the information provided to policymakers (Chalmers 2013), and the access that they have to EU institutions (Beyers and Braun 2014).

While research to date has generated important insights into the way interest groups work together in the policymaking process, one of the main difficulties associated with studying this issue is the fact that such coordination often varies significantly in the level of formality and openness (Heaney and Strickland 2016). Most scholarship to date has often focused on formal lobbying coalitions. These are easy to identify since they involve public declarations of the alliance among different groups and in some cases “highly coordinated enterprises with logos, letterheads and secretariat” (Mahoney 2007, 366).

These are, however, only one way in which interest groups work together and coordinate their advocacy initiatives. Interaction among groups in the policy process often remain “informal and loose” (Mahoney 2007, 366), based on the occasional information sharing and private communications with other groups that mobilize around the same issues. Numerous scholars have argued that information regarding a given policy frequently flows within “communication networks that emerge from the interaction of interest group representatives”, either in person or through telephones or e-mail (Heaney 2014, 68).

The existing literature has explored the reasons why groups engage in this type of exchange. Responding to an open consultation requires citizens, business groups, NGOs and other stakeholder to devote significant resources to monitor the policy agenda, understand the implications of proposed policy, and formulate a response By connecting with other groups and exchanging information, groups are able to lower the costs of responding to a consultation by drawing upon the expertise and resources of other groups ( Baumgartner et al. 2009; Hojnacki 1998; Junk 2019a). Moreover, while the pursuit of formal lobbying coalitions requires groups to compromise on a common text endorsed by all the coalition partners (Hojnacki 1998; Holyoke 2009), coordination based on the informal exchange of information allows groups to retain greater autonomy to pursue the preferred policy position by selectively using information from other groups (Mahoney 2007).

The existing literature has established that informal communication networks play a key role in the way groups engage with the policymaking process in the EU and elsewhere and called for a “better theoretical integration of the formal sides of collaboration (e.g., coalitions) with the informal sides (e.g., communication, trust)” (Heaney and Strickland 2016). For instance, we still lack clear answers to key questions such as: How do informal communication networks differ from more formal lobbying networks? Which groups are more likely to engage in information exchanges within informal networks rather than creating formal coalitions?

One of the main limitations in approaching these questions are the empirical mearures that scholars deploy. Three main empirical approaches are frequently found in this literature. First, different studies have mapped coalitional activities among interest groups by collecting information on existing formal memberships within different associations (Bunea 2014; Chalmers 2018; Chalmers and Young 2020). As Bunea (2014, 1232) acknowledges, however, “this estimation technique suffers… from the obvious neglect of informal communication or co-operation ties that most probably characterized the interactions of interest groups at the time of the event”. The mere presence of common associational ties between distinct groups does not necessarily imply that these will be “active” and that information will flow from the association to its members or between different members.

Second, a more direct way to capture instances when groups have worked together on an issue relies on the analysis of the patterns of co-signing of the same document by different groups. For instance Box-Steffensmeier and Christenson (2014) explore alliances by mapping when interest groups co-sign amicus curiae briefs before the US Supreme Court, while Pagliari (2018) maps co-signing of response letters to policy consultations by US financial regulatory agencies. According to Box-Steffensmeier and Christenson (2014) the advantage of this approach is that it is “culled from the actual, purposive and coordinated work of interest groups” and “it also comes close to a complete network of the population of interest”. Also, in this case, however, the focus on mapping public linkages between interest groups captured through the co-signing of documents misses informal and less visible information exchanges among different groups.

Third, the most common way in which existing studies of have tracked the existence of informal networking and coordination among groups in the EU (Beyers and De Bruycker 2018; Chalmers 2013; Mahoney 2007; Sorurbakhsh 2016) and elsewhere (Hojnacki 1998; Beyers and Braun 2014; Junk 2019b) relies on surveys or interviews of representatives from interest groups active on a given issue. This approach has the advantage of allowing researchers to capture actual channels of communication among groups on a given issue, including those that remain informal. However, surveying interest groups remains highly labour intensive (Junk 2019b), forcing researchers to focus on a single or a small number of policy issues while trying to generalize from those. It is also arguably subject to limitations such as “memory fatigue” due to the distance in time (Holyoke 2009), “staff turnover” within the organization and low response rates among the interview population (Bunea 2014, 123).

In the next section we outline a novel text-as-data approach that aims to capture informal channels of coordination across the entire population of interest groups active on a given policy space.

3 Method

This section outlines an approach to map the patterns of information exchange among groups by mapping instances of text reuse between the responses submitted by interest groups to the same policy consultation. This is certainly not the first study to rely on this type of textual data to study the relation among groups in the policymaking process. For instance, different studies have hand-coded the text of the responses of different groups to policy consultations of these responses against a coding scheme, for instance to identify those groups calling for a more stringent or weaker policy response (Yackee and Yackee 2006) . The resource intensive nature of manual coding and issues of reliability have also pushed scholars to explore other computer-aided approaches to content analysis. In her studies of coalitions in European politics, Klüver (2013) extracts the policy position of different interest groups based on the relative frequency of words used within and across the texts produced by different groups in response to a given policy consultation (Slapin and Proksch 2008). As Klüver (2013, 54) acknowledges, for groups to share the same policy position, “it is not necessary that interest groups formally cooperate or exchange information”. In other words, as Junk (2019b, 660) argues, existing approaches studying focusing on the text produced by different groups have captured the presence of “camps” or “sides” to an issue, but “they typically overlook active cooperation between these actors”.

In order to identify instances of “active” cooperation (Junk 2019b) whereas groups interact and exchange information in the pursuit of a common policy objective, we turn to a different approach and investigate patterns of text reuse across different submissions. As interest groups work together behind the scenes and share information regarding a policy or develop a coordinated response, we would expect some of the same text to be reused verbatim or with only limited changes in their respective responses to the same consultation. For instance, Figure 3.1 shows the first page of the letters submitted two letters submitted by two distinct groups in response to same policy.