Introduction

The flow of politicians and public officials into the private sector has become a stable feature of modern politics. As an example, Alter-EU reports that six out of thirteen departing European Commissioners in 2009–2010 went into corporate or lobbying jobs and noted that similar patterns existed for high-level officials from the European Union (EU) institutions in general.Footnote 1 Both academics and commentators speculate that the significant financial transactions associated with the "revolving door" imply that special interests can secure political favors by hiring former political insiders. A smaller number of studies have for example examined the consequences of the revolving door for individual lobbyists. This research suggests that connections through the revolving door pay off for lobbyists, as the employers of revolvers obtain an economic premium (e.g., Blanes i Vidal et al. 2012; McCrain 2018) and see improved results in the stock market (Luechinger and Moser 2020). Not surprisingly, the labor market prospects for people with a background in politics are therefore quite good (McCrain 2018; Strickland 2020; Egerod and Tran 2023; Palmer and Schneer 2016, 2019). However, there remains a gap in our understanding of the situations in which individuals with political ties utilize their connections to advance private interests following their employment transitions from the public sector.

In this article, we aim to investigate whether political access plays a role in facilitating the ability of these individuals and their employers to solicit political favors and gain advantages through the revolving door. To achieve this, we systematically evaluate whether organized interestsFootnote 2 that employ former civil servants and politicians are subsequently granted greater access to policymakers.

We find limited evidence that all organized interests are able to use revolvers to obtain access to the political system as a whole. However, we do find that revolvers can facilitate access to subsets of meetings pertaining to the setting of the legislative agenda, although only for a short period of time. Importantly, prior research (e.g., Jones et al. 2009) has found that this is when institutional friction is still relatively low. Furthermore, we demonstrate that the positive access gained from hiring revolvers is concentrated within contract lobbying firms.

Our analysis uses a difference-in-differences (DiD) strategy to estimate the impact of the revolving door on access to policymakers with a new unique dataset from the EU. We start from a list of bureaucrats and politicians who retire from the EU Commission (Commission), and then cross-reference the names of their new employers with official records documenting the Commission's interactions with external interests. Additionally, we leverage a distinctive aspect of the EU system, wherein the Commission publishes concise descriptions of meeting topics. We explore this unique textual monthly data on Commission meetings to measure what organized interests discuss, and with which policymakers they discuss it. The design allows us to use highly granular behavioral data to examine how the propensity for an organized interest to gain meetings with the Commission changes after the revolver joins the organization.

The article's contribution is both methodological and empirical. It represents the first study to evaluate the impact of the revolving door in EU institutions using a difference-in-differences identification strategy to estimate how career transitions to the private sector affect access to meetings with policymakers. This helps shed light into consequences of the revolving door rarely studied, i.e., whether hiring revolvers is ultimately a benefit for obtaining access to policymakers and whether this effect differs for different types of organized interests and meetings. In addition, it potentially helps understand the related finding in the literature that contract lobbying firms that hire revolvers on average secure economic gains (e.g., Blanes i Vidal et al. 2012; McCrain 2018; Luechinger and Moser 2020). Hence, it suggests that part of the mechanism rendering revolvers valuable sources of income for these types of organized interests could be their enhanced capacity to access the political system regarding certain issues, enabling them to directly communicate their concerns to policymakers. While there are many types of “favors” organized interests can extract from hiring revolvers, access is one of the most important ones, since it is an important step toward influencing the political agenda (e.g., Bouwen 2004; Gross and Rasmussen 2015; Miller 2021).

Our findings offer grounds for both optimism and pessimism. On a positive note, they imply that the influence of revolvers, while significant, may be more restricted in its reach than feared by critics. We demonstrate that employing revolvers does not automatically ensure access to meetings across the entire policy-making process. By analyzing meeting agendas, we reveal a concentration of advantages in meetings concerning broader political agenda topics, rather than specific legislative matters. Nevertheless, the presence of varied effects among different actor types, where contract lobbying that possess less democratic legitimacy than other groups can obtain political favors such as high-level meetings by utilizing revolving doors, raises significant normative questions about the broader ramifications of this practice.

Theoretical framework: lobbying, political connections and information

A large literature investigates the role of political connections and expertise in lobbying (e.g., Blanes i Vidal et al. 2012; Bertrand et al. 2014; Bouwen 2004; LaPira and Thomas 2017; McCrain 2018; Shepherd and You 2020). We build on it to argue that organized interests that hire former civil servants and politicians should gain by subsequently being granted greater access to policymakers (H1). While access does not necessarily guarantee a group's ability to influence policy, it is often considered a crucial element in the process. This is the case, because—for access to happen—policymakers must spend their scarce time on meetings with organized interests, and they would not do so unless they expected to get something in return (Binderkrantz et al. 2015, 2017).

Why employers should gain access from hiring revolvers

The reason revolvers should be able to help their private sector employers gain access to policymakers is that their careers in public service can be expected to have endowed them with a particular set of skills. We can broadly distinguish between the importance of expertise and connections, with the former being further sub-divided into procedural and substantive expertise (LaPira and Thomas 2017). First, in relation to expertise, Bouwen (2004) argues that policymakers interact with organized interests, because they are in an information deficit. In particular, when seeking to construct business regulation, policymakers have imperfect information about the cost structure facing the individual firm, and how a change in regulation would impact it (Grossman and Helpman 2001, chapter 3). Therefore, in order to construct a successful lobbying campaign that allows the organized interest to gain access to policymakers, they need to construct high-quality policy input (Bouwen 2004), delivered in a way that can easily be implemented in policy (Drutman 2015). In other words, information alone is typically insufficient; the suggestions need to be actionable policy advice (Drutman 2015). Revolvers can be helpful for organized interests that seek to obtain these assets. While many policymakers—particularly in top positions—are legislative generalists with little issue-specific expertise (Hall and Deardorff 2006), revolving door bureaucrats or legislative staffers have considerable expertise in specific policy-making areas (LaPira and Thomas 2017). Moreover, while many organized interests seeking to influence policy may already possess the requisite knowledge about their operations and how a regulatory change might impact them (Bouwen 2004), translating such knowledge into actionable policy is challenging (Drutman 2015). Revolvers, however, possess the expertise necessary to accomplish this task convincingly to policymakers (Strickland 2020). Second, connections facilitate access to policymakers. During their public service, revolvers will have established personal relationships with their colleagues, which will assist them in their work for private sector employers once they have transitioned into their new roles as lobbyists (Strickland 2020). A substantial and expanding body of literature suggests that lobbyists can utilize their social connections to policymakers as assets to secure access on behalf of their clients (Bertrand et al. 2014; Blanes i Vidal et al. 2012; Hirsch et al. 2019; McCrain 2018). However, connections and information are likely to complement each other. In particular, when choosing whom to meet with, policymakers face the problem that there are many groups that seek access, and the substantive expertise that they offer might be biased (Hall and Deardorff 2006). This places the policymaker in a conundrum: They need information to construct policy, but it is often too costly to validate the information offered by the organized interests. Repeated personal connections between lobbyists and policymakers are important to solve this problem—lobbyists must deliver honest information if they want to benefit from future access. By awarding some lobbyists a privileged insider status in return for information, policymakers can avoid problems with cheap talk (Groll and Ellis 2017). Revolvers play an important role because they have pre-existing social links to the current policymakers, making them more likely to be trusted both in terms of their honesty and the quality of their information. Hence, they are more likely to be granted the privileged insider status all lobbyists seek (Hirsch et al. 2019). This implies that policymakers do not have to validate the quality of information themselves—they can rely on the revolver for this (McCrain 2018). A similar way connections can play a role is that the policymaker is likely to know the biases of the lobbyists with whom they have had a professional relationship, and they can use this to “debias” the information they receive (Grossman and Helpman 2001, chapter 3). Lobbyists who lack personal relationships with policymakers cannot guarantee that they will only introduce them to organized interests with the most valuable information. The profit motive incentivizes them to represent as many clients as possible. However, due to their personal connections with policymakers, revolvers can more credibly commit to only introducing organized interests with high-quality information (Hirsch et al. 2019). In this way, political connections facilitate the transmission of substantive expertise between organized interests and policymakers.

While it has been theorized that revolvers might help their future employers access the political system (McCrain 2018), empirical research on the potential gains of hiring employers has focused on broader economic effects rather than political access (e.g., Blanes i Vidal et al. 2012; Luechinger and Moser 2020). Shepherd and You (2020) is an exception. However, they look at whether members of congress employing more soon-to-be revolvers grant more access to prospective employers before staff moves into the private sector rather than whether organized interests gain access after hiring revolvers, which is the focus of this study. Furthermore, akin to the majority of literature on the revolving door, their examination is centered on the USA.

The relevance of hiring revolvers in the European Union context

Despite the US-centered focus in existing research, we can expect the discussed assets of revolvers to be equally pertinent for benefiting future employers within other political contexts, such as the EU, which is the focus of our analysis. While not a traditional state, the EU bears resemblance to a political system in several crucial respects, particularly because its member states have delegated significant law-making competences to it, rendering it the primary source of regulation in numerous policy domains (Hix and Hoyland 2011). To examine whether hiring revolvers leads to access to policymakers, we specifically analyze meetings with one of the key EU institutions, the EU Commission. The Commission has the right of initiative, meaning that it is responsible for planning, preparing and proposing new EU legislation, which then proceeds to the European Parliament and the Council of Ministers. This privileged position in agenda setting renders access to the Commission particularly significant. Additionally, because Commission officials do not run for election, it offers an interesting opportunity to investigate the revolving door phenomenon in the absence of electoral pressure. The Commission is structured into distinct policy departments known as directorates-general (DGs), each overseen by a commissioner. Due to the vast scope of producing European legislation, the Commission faces resource constraints and must engage with external stakeholders to fulfill its legislative mandate (Bouwen 2004). This characteristic renders the EU context, especially the European Commission, particularly suitable for examining the theory. Existing work by Coen and Vannoni  (2016; 2020a; 2020b) of EU Business–Government relations—including the revolving door—demonstrates that, while career moves between the political and private sector are not as widespread in the EU as in the USA, they do occur. In support of this, recent survey evidence of hiring practices of organized interests lobbying the EU by Belli and Bursens (2023) shows that a little over a third of groups report having staff with a public sector background.

Recent literature has also started examining the consequences of such hires in an EU context. Crucially, Luechinger and Moser (2020) show that investors expect political connections to matter as they do in the USA: Firms that hire former EU Commissioners see improved stock market returns. Additionally, Belli and Beyers (2023) have recently published a study looking at whether those organizations that report having (any amount of) staff from the EU institutions in a survey experience a different likelihood of having meetings with the European Commission than those which do not. Their analysis provides important, suggestive evidence that the revolving door might also have political consequences for some types of organized interests in an EU context even if their data do not demonstrate longitudinal within-case variation that allows them to make causal claims.

As we will outline below, we complement existing work by linking granular behavioral data on movement of EU officials and politicians into the private sector with data of the meetings between the EU Commission and outside interests. This enables us to use a difference-in difference design to look at whether different types of interest organizations experience gains in access following their hires of EU revolvers (see below for a detailed discussion of the model). Additionally, we examine variation in the potential benefits of the revolving door among both the types of actors utilizing revolving door practices and the nature of meetings with the Commission. Consequently, our database of meeting records enables us to classify meeting content, thereby facilitating an investigation into differences in the impact of the revolving door on accessing different types of meetings, likely associated with distinct phases of the policy-making process.

Why gains from hiring revolvers might may vary across meeting content and actor types

We argue that access gains for organizations hiring revolvers may not be constant across meetings of different levels of “policy specificity.” We draw a theoretical distinction between meeting topics related to (a) broad policy discussions, (b) medium-broad discussions on specific economic sectors and (c) specific policies under negotiation or adopted. These categories highlight the spectrum from agenda setting to specific policy outcomes with broad discussions often aimed at introducing new topics to the agenda, while specific discussions usually occur in later policy stages, focusing on existing proposals or laws. We expect organized interests to capitalize more on hiring lobbyists for obtaining access to meetings involving broader discussions, likely to happen early on in policy processes, than to meetings involving more narrow discussions of policy outputs later in the policy cycle (H2).

First, during the early stages of the policy process, policymakers may demonstrate an increased demand for the assets typically associated with "revolvers" discussed in the previous sections. During this phase, the documented information deficit of policymakers (Bouwen 2004) and the demand for technical policy input are likely to be more pronounced, underscoring the significance of the expertise offered by revolvers in specific policy-making domains (LaPira and Thomas 2017). Using similar argumentation, the interest group literature has also convincingly argued and shown that exerting policy influence should generally be easier earlier than later in the policy process (e.g., Baumgartner et al. 2009a, b; Bevan and Rasmussen 2020; Olzak and Soule 2009; Soule and Olzak 2004). Second, apart from need for information, the literature on agenda setting also convincingly argues that different stages of the policy process exhibit varying levels of "institutional friction," which refers to formal structures that increase decision and transaction costs (Baumgartner et al. 2009a, b; Bevan and Jennings 2014; Jones and Baumgartner 2005; Jones et al. 2009). Early stages typically exhibit lower institutional friction compared to later stages, allowing decision-makers more flexibility to be responsive to public priorities, and potentially also give them greater latitude in addressing the concerns of organized interests. Lower friction may afford organized interests greater opportunities to benefit from having hired revolvers, as it may become easier to accommodate meeting requests compared to later stages of the policy process. In these later stages, policies may be more entrenched and locked in, and decision-makers may be compelled to prioritize listening to all relevant stakeholders involved in a given policy decision, regardless of whether the organizations in question offer specific connections and expertise through hired revolvers.

Finally, we contend that the importance of hiring revolvers should not be uniform across all types of organized interests in securing access. This is because, beyond seeking expertise and connections, policymakers also interact with organized interests for their potential to provide legitimacy and contribute to political representation. The latter can be understood as a process where representatives claim to represent particular societal interests (Saward 2006). Organized interest faces the challenge that they are not directly elected but self-authorized representatives. However, provided they are authorized by and accountable to the affected constituency they claim to represent, they may be seen as contributing to democratic representation (Montanaro 2012). When policymakers are aware of the segments of society represented by a group, they also gain insight into the biases it may harbor. This insight is valuable for policymakers when assessing the quality of information (Grossman and Helpman 2001). Yet, the “representative potential” of different organized interests varies (see e.g. Eady and Rasmussen 2022; Giger and Klåuver 2016), meaning that they may not be equally dependent on hiring revolvers for getting access.

For certain organized interests, claims to represent either broader segments of interests or a large share of the potential constituency of a given type of interest are more likely to be sufficient for obtaining access. Interacting with broadly representative organized interests not only helps the policymakers assess the information offered by them—their representativeness carries legitimacy that will increase the value of meeting with them  (e.g. Dür and Mateo 2013; Junk 2019; Rasmussen and Otjes 2024; Rasmussen et al. 2018). NGOs are a good example: Many of them represent broader societal interests and are widely perceived as having a broader potential to representative society  (Flöthe and Rasmussen 2019; Rasmussen and Reher 2023; Willems and De Bruycker 2021). Policymakers need to engage in close dialogue with such organized interests to boost the legitimacy of their organization and policies, irrespective how many former ex-officials and politicians with technical expertise and political connections these organized interests have on their payroll. These NGOs still employ revolvers but may experience fewer direct benefits from doing so, as they might have a high likelihood of securing a seat at the table regardless.

Instead, the hiring of revolvers might have a more significant impact for organized interests with weaker "representative" and "legitimizing" potential. A prime example would be contract lobbyists—professional lobbyists employed by specific clients—many of whom lack a mission tied to broader societal concerns. Rather than representing constituencies associated with broader societal issues, they cater to specific interests. When policymakers consider granting access to contract lobbyists, concerns about enhancing legitimacy may carry less weight, while other factors, such as their level of professionalism and expertise, may exert a stronger influence. For contract lobbyists, employing former staffers and politicians from the EU apparatus may thus have a more significant impact on the political benefits they receive. Additionally, business associations and companies are likely to occupy a middle ground between NGOs and contract lobbyists. While they may not inherently claim to represent society broadly, they are expected to carry a degree of legitimacy, dependent on the proportion of any given sector they represent. In summary, we anticipate that the potential access effect of hiring revolvers should be more pronounced among contract lobbying firms than other types of organized interests, given their comparatively lower representative and legitimizing potential (H3).

Research design

Our analysis relies on a difference-in-differences design and longitudinal data from the EU that allow us trace how the likelihood of an organized interest of obtaining access to EU policymakers changes after they hire revolvers, who have previously been employed in the EU institutions. As we detail below, we extend our analysis beyond overall access by engaging in content coding of the topics of the meetings held between organized interests and policymakers.

Data

We rely on several data sources to measure our variables of interest. First, to estimate the effect of gaining a political connection by hiring a former Commission official or politician on access to meetings, we start by constructing a binary indicator of the year during which the organized interests hire revolvers. For this, we rely on work by Corporate Europe Observatory (CEO). CEO used a combination of desk research and available online documents provided by the Commission to construct a list of Commission officials and politicians that transferred to positions in the private sector. This list contains information on which organizations the 41 top EU officials and politicians in the dataset moved to. We collected data on the timing of the change in employment for the revolvers, and—in doing so—we also validated each case in the dataset. As we do not know when the revolver leaves the group again, we consider treatments during the first year only. That is, since we know which month the revolver arrived, we count a group as treated from the month of arrival and for twelve months into the future. If we were to extend the period beyond one year, and certain revolvers leave before others, our estimates would be biased. It is important to note that few Commission officials spin through the revolving door, and our results rely on relatively few revolvers. Because our data are very granular data, we still obtain a quite large sample size overall, allowing us to detect some effects. However, future research should probe the results further in larger samples with more treatment events.

To measure the access that an organization gains to the Commission, we rely on Transparency International’s data gathering effort. The organization collected publicly available information on meetings from the websites of Commission DGs and combined it into a single file, including details about the name of the organization that is meeting with the Commission and the subject they discuss. This information provides a potent method for assessing access, aligning closely with the theoretical definition of access (Binderkrantz et al. 2017) and drawing on higher-quality data than typically accessible to researchers (Miller 2021). We rely on two measures of access in our main specification. First, to construct a measure of access at the intensive margin, we use the number of monthly meetings that an organization has with top-level staff in the Commission within a four-year period (between late 2015 and 2018). In our analyses, we log transform this measure.Footnote 3 This captures the frequency with which the group meets with policymakers. Second, to construct a measure of access at the extensive margin, we use a binary measure of whether an organization has a meeting with the Commission in each month in our data. This captures the likelihood of gaining any form of access.

Figure 1 presents an overview of the types of revolvers that previously worked in the Commission, with which the Commission held meetings as well as a distribution of the total number of meetings and the top 16 actors that obtained access. According to panel C, Companies and lobbying firms account by far for the largest number of meetings between organizations and the Commission, while, e.g., NGOs and Think tanks account for a considerably lower share. The same applies to trade unions and business associations, who are included in the “other category” along with for example law firms and public entities.

Fig. 1
figure 1

Meetings and (types of) organized interests that hired Commission Revolvers

Full size image

Figure 2 instead shows the types of actors employing revolvers in our dataset irrespective the number of meetings held. Importantly, individual firms and commercial lobbying companies employ far more revolvers than other types of actors. However, trade associations—traditionally, the main vehicle for business power—do not employ more revolvers than for example think tanks and NGOs. This is interesting, as it contrasts the findings from the US federal level in LaPira and Thomas (2017, p 116–117) that both companies and trade associations hire revolvers more than most other types of actors. It also contrasts Strickland’s work (2020) on lobbying in the US states, which finds that while companies, other types of private interests and government reform groups hire the former legislators,Footnote 4 public interests in general hire quite a large number as well.

Fig. 2
figure 2

Number of Revolvers Hired by Actor Type

Full size image

While this gives us a sense about broad patterns of access and hires, it leaves an obvious remaining question: What do the organized interests use this access for? Is the goal to influence specific policy outcomes or to put new issues on the EU agenda? Crucially, given that the Commission registers the topics of the meetings, we can construct a robust proxy measure for our variable of interest: whether the organized interest's objective in the meeting is to influence a specific policy outcome or to raise awareness of an issue within the EU agenda.

We categorize the subjects of meetings in a simple scheme capturing whether the meeting is related to (a) very broad policy discussions that are not currently covered by a policy proposal or active legislation, (b) discussions of moderate to broad scope regarding a particular sector of the economy, or (c) specific policies and issues either in preparation or already are in force. These categories of meetings match the expectations outlined previously. Table 1 presents examples of meeting subjects for each type of meeting we code.

Importantly, this measure of “policy specificity” captures agenda setting and specific policy outcomes and issues as two extreme cases. First, how closely a meeting’s subject relates to an actual policy proposal will naturally be related to the stage of the policy process. Very abstract meetings about subjects where the EU has not formulated a policy yet will typically have the goal of putting a new topic on the EU’s agenda, or deciding what should be discussed before an actual proposal is developed. In contrast, meetings of higher specificity, which concern specific proposals or active legislation, are more likely to concern later stages of the policy processes. A more detailed codebook is found in online Appendix A (Table 1).

Table 1 Examples of Different Meeting Types
Full size table

Our final dataset is based on 58 organized interests hiring 32 revolvers during our time frame. The final dataset includes 849 organized interest months, where 831 meetings were held. We emphasize that this is a limited dataset, and while the results we present are informative, future research should probe them further by collecting larger datasets. Table 2 presents descriptive statistics split according to whether the revolver had a past in the Commission or another EU institution. According to Panel A of Table 2, the number of meetings for organizations hiring Commission revolvers averages to approximately 0.7 each month, while the probability of having a meeting in a given month is about one in three. Approximately one-tenth of all organized interest months are treated with a connection as a benchmark. To provide a benchmark, panel B shows similar descriptives on revolvers from other EU institutions. Interestingly, revolvers from other institutions have fewer Commission meetings—the probability of gaining a meeting in any month is approximately 20 percent. Yet, they hire the same proportion of revolvers—in approximately one in ten months, the organized interests have a revolver appointed.

Table 2 Descriptive Statistics
Full size table

Identification

To identify the effect of gaining connections on access to Commission policymakers, we rely on a DiD design. In the model, the "difference" refers to the change in outcomes before and after the intervention, and the "differences" refer to the contrasts in these changes between the treatment and control groups. By comparing these differences, the DiD model attempts to isolate the causal effect of the treatment from other factors that may influence the outcome. In this way, the design is well suited to estimate the effect of hiring revolvers on access. One main reason is that it allows us to control for all time-constant variables that could threaten identification. The DiD relies on the assumption that trends would have evolved in parallel if the newly connected organized interest had not hired a revolver. To increase the plausibility of this assumption, we add two features to the typical design, which make our particular DiD design even more preferable to alternative design choices. First, we only compare organized interests that are treated now to organized interests that will be treated in the near future. That is, we exclude all organized interests that never hire revolvers. This is a powerful design choice, since it is not random which organized interests choose to become connected. Excluding never-connected organized interests eliminates bias stemming from specific group types opting to employ revolvers.Footnote 5 As a second way of dealing with unobserved shocks, we make use of the high granularity of our data. Because we track meetings and hirings within years, we can control away all confounders that change from year to year. While real-world factors change between months, most of the confounders we could hope to collect data on would be on the yearly level. One important example includes organizational strategies which will tend to change relatively slowly. This strategy purges our estimates of all such unobservables.

We estimate the DiD through OLSFootnote 6 regression models:

$${Access}_{gmy}=\beta connec{t}_{gmy}+{\gamma }_{g}+{\delta }_{m}+{I}_{y}*{\gamma }_{g}+{\epsilon }_{gmy}$$

Access denotes one of our dependent variables capturing meetings and different types of meetings with the Commission, and g represents the organized interest, m represents the month, and y is the year. connect is our indicator of whether the organized interest hires a Commission revolver in a given year–month period, which makes β our parameter of interest—the effect of hiring a revolver on the organized interest’s access. \(\gamma\) represents an organized interest fixed effect, while \(\delta\) is set of year–month (calendar-time) fixed effects. The organized interest fixed effect removes all time-invariant factors (e.g., group type), while the time fixed effect deals with homogeneous monthly shocks to the system. The inclusion of these fixed effects makes this a difference-in-differences model (Goodman-Bacon 2021). Finally, the interaction between dummies for the year and the organized interest (\(I\) and \(\gamma\)) allows for unobserved shocks to each organized interest every year. This is quite powerful as it holds constant any confounder that changes on an annual basis.Footnote 7Footnote 8 Fixed effects can be problematic estimator of DiD (Goodman-Bacon 2021). In online Appendix D, we show that our results are not biased by issues arising from staggered treatment and heterogeneous effects over time.

Results

Table 2 presents our main results for both the probability of obtaining meetings (columns 1 and 2) and the number of meetings (columns 3 and 4) for revolvers from the EU Commission. Surprisingly, we do not uncover strong evidence that hiring these revolvers increases access in general. This holds for both the extensive and intensive margins. In column 1, we find that the probability of obtaining a meeting increases by 5 percentage points after the revolver arrives. However, the standard error is of the approximately the same size as the coefficient, and the estimate is not statistically significant. In column 2, to guard against the potential that groups that lobby more are more likely to both gain meetings and hire revolvers, we include the log of groups’ spending on Brussels lobbying as a control. Since most groups do not update their entries in the Transparency Register, and spending remains more or less constant over time, we include it in an interaction with time fixed effects. The results remain statistically insignificant. In columns three and four, we examine the logged number of meetings using the same specifications. The specification in column four, including the interactive control for logged spending, is significant at the ten percent level. This could happen by chance and should not be considered robust evidence that hiring revolvers increases meetings with the Commission in general.

In sum, while the coefficients are all large, they only meet common standards for statistical significance in one case. The lack of significance in support for hypothesis 1 could be driven by low power, since we have a relatively small number of organizations. However, we have estimated less restrictive models without interactions between group and year fixed effects, which yield the same results. Additionally, in online Appendix J, we include groups that never hire a revolver. This increases our sample size noticeably and helps with potential issues of low power. However, the results remain statistically insignificant. In online Appendix I, we also run a specification with linear time trends rather than an interaction between group and year fixed effects. This does not change the results. Finally, since the monthly data are likely to be quite noisy, we also aggregate it to a quarterly panel in online Appendix J and estimate our main set of regressions. This does not change the results either. Overall, we thus find no compelling evidence for an average effect on all types of meetings—neither concerning the probability of obtaining a meeting (the extensive margin) nor the frequency of meetings (the intensive margin).

Next, to delve further into this puzzling null result, we investigate the type of meetings organized interests do gain access to. According to Table 3, we find support for our expectation in hypothesis 2 that organized interests are more likely capitalize on hiring lobbyists for obtaining access to meetings involving broader discussions, which are likely to occur early on in policy processes, than for meetings involving narrower discussions of policy outputs later in the cycle. Column one of Table 3 shows that hiring a revolver is associated with an increase in the probability of getting at least one meeting for agenda setting purposes by approximately 15 percentage points. This estimate is statistically significant—the estimate is more than three times the size of its standard error, and we can confidently reject the null of no effect. Importantly, we estimate a noisy drop in the number of meetings held about specific policies (column three) and no change in meetings of a medium level specificity related to a certain economic sector (column two). Both these estimates are statistically insignificant at conventional levels, but they are still highly indicative of an important pattern. They show why we see no average effect: The effect is heterogeneous and concentrated on agenda-setting meetings.

Table 3 Hiring Former Commission Employees and Commission Meetings
Full size table

These are important findings suggesting that hiring lobbyists primarily leads to increased access to meetings with broader discussions. That is, they gain access to meetings early in the policy cycle but are less utilized to gain access to the more detailed stages of the policy-making process. This reinforces the notion that the role of former politicians and top bureaucrats is not primarily to influence the outcomes of very specific policy processes (e.g., the decision of who gets a specific procurement contract). Rather, they leverage their connections with the goal of steering the direction of the broader policy agenda. An alternative explanation for these findings is that organized interests that hire revolvers might want to hide the content of their meetings with the Commission. If this were driving our results, we would expect the description of the meeting content to be shorter and less linguistically diverse after a revolver is hired. In online Appendix C, we show that this is not the case (Table 4).

Table 4 Hiring Former EU Official or Politician and Meeting Content
Full size table

Finally, we examine whether the positive effect of the revolving door varies between organized interests representing different types of constituents. To do so, we allow for different effects depending on the type of organized interest and use a binary indicator of any meeting with the Commission as our dependent variable. We distinguish between lobbying firms, companies, and NGOs, which are the actor types for which we have a sufficient number of meetings to run comparisons. Because we encounter issues of perfect collinearity when including interactions with each of the group type indicators, we estimate separate models. Each of these models includes an interaction between one of the group types and the revolving door treatment. Figure 3 shows that in line with our expectations in hypothesis 3, the potential access effect of hiring revolvers is more pronounced among contract lobbying firms, which enjoy lower representative and legitimizing potential than other types of organized interests. They see large gains in meetings when they hire revolvers, with an increase in meetings of approximately 18 percentage points. In contrast, the increases observed in the two other types of organized interests examined are smaller and not statistically significant.

Fig. 3
figure 3

Effects are Driven by Corporations and Lobby Firms. Note Results show the marginal effect of hiring a Commission revolver. The dependent variable is a binary indicator of at least one meeting for each type of organized interest. Estimates are from three separate models, each with one interaction by type of organized interest. Models include fixed effects for organized interest, month and an interaction between organized interest and year. Confidence intervals are cluster robust 90 percent (black) and 95 percent (gray). Think tanks are left out, since coefficients could not be estimated for them and lobbying firms simultaneously

Full size image

Additional analyses and robustness checks

We conduct a wide range of robustness tests. First, we test for the presence of differential pre-trends, finding no evidence of this (online Appendix E), and show that the effect on agenda-setting meetings is very short-lived (online Appendix F). To guard our estimates against the possibility that organized interests pursue other influence-seeking strategies simultaneously with hiring revolvers, we show that organized interests do not increase lobbying activity when they hire revolvers (online Appendix G). We also investigate effect heterogeneity by showing that hiring revolvers from the EU’s other institutions (the Parliament and the Permanent Representation) is only noisily associated with access to the Commission (online Appendix H).

Conclusion

A large literature investigates the revolving door between business and politics and has documented that hiring revolvers might be associated with economic gains (e.g., Blanes i Vidal et al. 2012; McCrain 2018). However, we know much less about whether organized interests can gain political favors by hiring revolvers in the form of obtaining access to policymakers. We have gathered a unique dataset on the revolving door within the EU and employ a difference-in-differences design to examine whether various types of organized interests benefit from hiring revolvers and the subsequent access they gain to different types of high-level meetings with the European Commission.

Overall, our findings indicate that political access could be a component of the mechanism enabling revolvers and their employers to attain certain economic benefits associated with the revolving door in prior research. However, our results also suggest that the impact of revolvers on access to the EU Commission is transient, smaller than could be expected from existing work, and highly contingent. Most importantly, the benefits they derive from hiring revolvers do not extend to gaining access to all types of meetings, but rather to those focused on agenda setting, which involve discussions on broad policy topics not currently addressed by a policy proposal or active legislation. This aligns with our expectation that the expertise provided by hiring revolvers should be more readily utilized in the early stages of the policy process when institutional friction is low.

Additionally, it is worth discussing how these results should be interpreted in a causal sense. Our design ensures that most confounders are either differenced out by including group and calendar-time fixed effects or held constant by interacting year and group dummies. However, one factor we cannot exclude is whether the group has meetings with the EU Commission as its strategy. It is possible that when groups perceive these meetings as a strategic priority, they opt to enlist the help of revolvers to advance their objectives. This implies two possible interpretations of our results—either hiring revolvers leads to access (for some group types and for agenda-setting meetings), or the revolvers are an integral part of successful access-seeking strategies (for some group types and for agenda-setting meetings). The latter interpretation suggests that at the very least, from a causal perspective, revolvers are utilized by groups to secure access under specific circumstances, even though these groups might have gained access without the involvement of a revolver.

There are three potential theoretical explanations for these findings. First, it has previously been argued that information is particularly important for lobbying the EU Commission (Bouwen 2004). Although some revolvers have important substantive expertise, non-revolver lobbyists are able to build similar information capacities by working on certain issues and building expertise (Miller et al. 2024). Thus, one potential explanation for the overall null finding is simply that connections matter less in the context studied here and that—as argued in previous research—information is much more important when lobbying the EU Commission. Second, however, we also show that the arrival of a revolver does increase access among commercial lobbying firms. This speaks to the Ellis and Groll’s (2018) model, which shows that organized interests that engage in repeated interactions with policymakers will build the required level of trust needed to obtain insider status and preferential access to policymakers. According to this theory, lobbyists’ political connections matter, because policymakers come to trust the lobbyists they know. Importantly, the business model of commercial lobbying firms is quite often to rely on personal connections to obtain access on behalf of their clients. Therefore, we would conjecture that revolvers should be more important for commercial lobbying firms, because it allows them to “buy” the trust of policymakers by hiring someone they already have a relation to. Conversely, organized interests—that represent a well-defined constituency and have repeated interactions with policymakers—would build trust-based relations with policymakers at the organizational level and would not have to rely on individual lobbyists. While this is somewhat speculative, it is also consistent with Belli and Beyer’s (2023) finding that access is greater for what they refer to as "professionalized organizations," i.e., organizations where staff and organizational leadership dominate advocacy and policy work. This finding also aligns with prior research on revolvers as contract lobbyists (McCrain 2018). Overall, this supports our expectation that lobby firms—lacking broad representativeness, and a relation to policymakers at an organizational level—must rely on hiring revolvers to obtain meetings with policymakers, whereas other types of groups are able to amass the connections and trust needed for access without hiring revolvers.

A third explanation for the overall null finding is that the existing US-based literature showing that clients are willing to pay more to hire revolvers (Bertrand et al. 2014; Blanes i Vidal et al. 2012; McCrain 2018; Strickland 2023) is not driven by actual access. Clients may be willing to pay more to hire politically connected lobbyists, believing that they gain better access to policymakers. However, in reality, this might not be the case. This would be consistent with recent literature emphasizing the principal–agent problem between clients and their lobbyists, which shows that lobbyists sometimes choose to pursue goals that are at odds with the preferences of the groups that retain them (Ellis and Groll 2018; Holyoke 2022).

Turning to the normative implications, it may be a positive conclusion that revolvers do not seem to shape access to the EU Commission overall. This suggests that, in general, it is not possible to “buy” access by hiring former EU officials. However, we do find an effect for commercial lobbying firms—the part of the interest group system where the business model relies on personal relations (Ellis and Groll 2018). This could be cause for concern. These professional lobbying firms, in particular, can be seen as having weaker connections to broader society and lower democratic legitimacy compared to other organized interests. On the other hand, we also show that their access gains are generally short-lived. While this does not rule out that the most highly connected lobby firms exert disproportionate influence, at least it suggests that hiring a revolver does not buy such influence for a longer period of time.

While we have taken a significant step in assessing the effect of hiring revolvers on access in the EU in this article, there is scope for extending our analysis. Besides delving deeper into the three questions we outline above, this would include testing the generalizability of our results in larger samples with more treatment events and in other political systems. Future datasets would also benefit from collecting new and adding to existing datasets of the career paths of the revolving door lobbyists over longer time periods (see e.g. Blach-Ørsten et al. 2017; Blanes I Vidal et al. 2012; LaPira and Thomas 2017; Rasmussen et al. 2021; Selling and Svallfoss 2019) in order to develop more fine-grained measures of how the net balance of revolvers employed at any given point affects access. Finally, additional scrutiny of the practices for employing revolvers could help understand variation in the value of these staffers for obtaining access to different types of meetings.