What drives trust in regulatory agencies? Probing the relevance of governmental level and performance through a cross-national elite experiment on EU regulation

Abstract

Trust between constituent actors within the European Union (EU)’s multilevel regulatory regimes is decisive for regulatory success. Trust drives information flows, increases compliance, and improves cooperation within these regimes. Despite its importance, systematic knowledge regarding the drivers of trust within regulatory regimes is limited. This paper inquires whether trust in regulatory agencies is influenced by their affiliation with the national or EU governmental level, as well as by their performance. While existing literature predominantly focuses on why citizens place their trust in governments or regulatory agencies, this paper presents original insights regarding the formation of trust among elites within the regulatory regime, including politicians, ministerial officials, agency officials, interest groups, and regulated entities. We employ data obtained from a large-scale vignette experiment conducted in six countries involving 752 decision-makers from relevant organizations. The experimental results suggest that both public and private elite actors’ trust assessment of regulatory agencies does not hinge on cues associated with the governmental level, but rather depends on agency performance. Accordingly, belonging to the national or EU governmental level does not create a difference in trust assessment of regulatory agencies in itself. It, however, shows that particularly elite actors are rather sensitive in terms of the performance of a regulatory agency.

Introduction

The extent to which trust research addresses elite relations within European Union (EU) multilevel regulatory regimes is limited. This limitation is noteworthy given that trust-informed interactions between constituent actors across governmental levels are decisive for regulatory success. Alongside the greatly increasing scale and scope of regulations delegated to independent regulatory agencies (Braithwaite, 2000; Leisering & Mabbett, 2011; Levi-Faur, 2014; Levi-Faur et al., 2020), particularly at the EU level (Majone, 1994; Caporaso, 1996; Levi-Faur, 2005), an enhanced understanding of core actors’ (from politics, public administration and business) trust in these agencies is required (Six &Verhoest, 2017; Kappler & Schomaker, 2023).

Therefore, it is pertinent to examine the factors impacting regulatory core actors’ trust in regulatory agencies within a multilevel context. Building upon existing literature that predominantly focuses on citizens’ trust in governmental institutions in a multi-level context, two primary avenues of inquiry emerge. Trust is formed on the basis of either (a) a rational evaluation of the performance of institutions (e.g., Fitzgerald & Wolak, 2016), or (b) connotations linked to the governmental level to which institutions are affiliated (Muñoz et al., 2011; Muñoz, 2017). Some studies also identify interaction effects, indicating that high performance translates more directly to trust at positively connoted governmental levels (Harteveld et al., 2013). Applying this logic to the context of regime actors (as trustors, or those who trust) and regulatory agencies (as trustees, or those ones who are trusted), the constituent elements of the relevant regulatory regime with extensive experience in dealing with the regime and with one another, we expect that an elite notion of trust formation will be less affected by emotional attachments to the involved governmental level. Instead, it would be more accurately explained by a rationalist notion that emphasizes the importance of performance.

To investigate this issue, we conducted a vignette experiment where respondents were provided with contextual cues regarding the governmental level (either the EU or the national state) and regulatory performance (either high or low) of a regulatory agency. The respondents included public regime actors performing various functions in a regulatory regime including elected politicians, ministry officials, or agency officials, as well as private regime actors (regulatees) such as business or interest group representatives. This research design facilitated an examination of whether and how one or both factors affect trust formation in regulatory agencies. The empirical findings validate our theoretical expectations, demonstrating a substantial practical understanding of trust amongst elite actors. This is evident in the robust effects of performance in both cases – at the national as well as the EU-level regulator – with the level of government assuming less significance. Moreover, no interaction effects were observed between the levels of government and performance. These findings remain stable across both public and private respondents.

The paper proceeds as follows. The following section provides an overview of the field by discussing the concepts of regulatory regimes, regulatory agencies, trust, and their interrelationship. Subsequently, we examine various mechanisms that impact trust formation, focussing on the level of governmental action and regulatory performance and derive testable hypotheses. We then present our experimental research design and the sample. Lastly, we present our results, discuss our findings, and conclude with suggestions for further research.

Regulatory regimes, agencies and regime actors

Trust serves as a facilitator for social interaction by equipping one actor (trustor) with “positive expectations of the intentions or behavior of another” (trustee) (Rousseau et al., 1998: 395). Accordingly, there is typically an evaluation of the level of trust that can be placed in the other party, ”which informs a preparedness to be vulnerable that, in genuine cases of trust, leads to a risk-taking act” (Dietz, 2011: 215). Hence, trust is less a question of assent or dissent, but rather a finely nuanced assessment on a scale ranging from ‘no trust at all’ to ‘complete trust’ that, in turn, informs an acceptable degree of vulnerability. Therefore, the actions of the trustor hinge on the expected behavior of the trustee. Mollering (2006: 111) describes a minimal level of trust for overcoming “irreducible social vulnerability and uncertainty” as the necessary condition for any interaction.

Beyond the significance of the ‘leap of faith’, in general, higher levels of trust facilitate interaction with a reduced need for control and monitoring (lower transaction costs), based on a shared understanding of appropriate behavior within specific contexts (higher legitimacy and compliance), and a transparent sharing of information (leading to enhanced innovation and policy-making). Trust is ‘the glue that keeps the system together and […] the oil that lubricates the policy machine’ (van der Meer & Dekker 2011: 95). Hence, low levels or lack of trust may impede cooperation, render regulation inefficient and endanger compliance (Fukuyama, 1995; Six & Verhoest, 2017; Schomaker & Bauer, 2022). Exceedingly low levels of trust or outright distrust may evoke resistance, evasion, and dishonesty (Cook et al., 2005: 161).

In defining a regulatory regime as ‘an institutional structure and assignment of responsibilities for carrying out regulatory actions’ (May, 2007: 9), it becomes apparent that regulatory regimes such as those governing food safety or data protection, require the cooperation of multiple interdependent actors to ensure effective functioning. Accordingly, we take Six and Verhoest (2017) as a starting point, who takes an actor-centered approach towards regulatory regimes. They commence with three fundamental types of actor – citizens, regulators, and regulatees – and then expand this basic framework to include various types of regulators (public and private) and different types of regulatees (private, third sector, and public) to offer a more realistic account of polycentric regulatory regimes. We apply an even broader perspective and include the actors involved in the design, negotiation and decision-making, implementation and enforcement of regulations, including political decision-makers, ministerial bureaucrats, regulators, regulatory intermediaries (such as licensing and certification bodies) as well as regulated entities and national interest groups. Excluding citizens from this perspective results from our elite approach, implying that we only include actors with substantive and direct influence (Hafner-Burton et al., 2013) over regulation. In a multilevel context such as the EU, actors in such regimes belong to and interact across various organizational layers, i.e., the EU, national, and subnational levels.

In line with the cooperative nature of the regime, the often-assumed command-and-control (top-down) approach to regulation, which involves distinct policy-makers, implementers and recipients of policy, does not accurately reflect regulatory reality in a pluralist (EU) multilevel context. Here, actors interact and cooperate both horizontally and vertically. For example, the quality of regulation relies heavily on (voluntary) upward flows of information from lower levels, including street-level bureaucrats, to the regulatory agency (e.g., Fisman & Khanna, 1999). As a further example, regulation can only be effective if regulatees comply. However, compliance is highly dependent on the perceived legitimacy of these agencies. Both voluntary flows of information and perceptions of legitimacy are largely trust-driven. Political decision-makers’ trust in the regulatory agencies is a fundamental factor in the delegation of regulatory tasks. Therefore, regulation is only as effective as the sum of individual regulatory interactions and trust – particularly in the regulatory agency – serves as the lubricant for well-functioning regulation.

Indeed, a growing body of literature has specifically addressed the role of citizens’ trust in regulation and regulatory agencies (Six, 2013; Rudolph & Riley, 2014; Grimmelikhuijsen & Knies, 2017; Six & Verhoest, 2017; Grimmelikhuijsen et al. 2021; Maman et al., 2023; Verhoest et al., 2024). However, trust between regime actors and regulatory agencies remains a niche in the relevant literature.

In addition, Six (2013) and Six and Verhoest (2017; Verhoest et al., 2024) demonstrate that this niche focuses on regulator-regulatee relationships, albeit with a strong focus on direct encounters between inspectors and regulatees. Regime actors other than regulatees are rarely considered in research on trust relationships between ministries and (regulatory) agencies, highlighting how trust may complement control (van Thiel & Yesilkagit, 2011; Oomsels, 2016; Bjurstrøm, 2020; Verhoest et al., 2024).

The dearth of literature on regulatory actors’ trust in regulatory agencies presents challenges, given that citizen-based findings are of limited value for understanding the functioning of a regulatory regime. Elite actors within the regulatory regime share the same professional environment as the regulatory agency itself. Therefore, even at the EU level, they have the opportunity to cultivate trust ‘through their professional contacts with EU colleagues, first-hand experience of working with the EU and knowledge of what capabilities are available’ (Persson et al., 2019: 643). Accordingly, we expect that the formation of trust by regulatory actors will be guided by specific logic.

Drivers of regulatory regime actor trust

Existing literature indicates two major explanatory factors when assessing trust in governmental bodies in a multilevel context: performance and governmental levels. Thus, we discuss the relevant literature with a specific focus on the potential application to the context of elite regulatory actors assessing trust in the regulatory agency.

Governmental level

In research on citizens’ trust in government institutions, both theoretical perspectives and empirical research emphasize the significance of the level of government for trust assessment (Kumlin, 2009; Muñoz, Torcal, & Bonet, 2011; Harteveld et al., 2013; Torcal & Christmann, 2019). One argument refers to proximity, with trustors being more likely to trust government institutions that are proximate to them, e.g., national governments (Jennings 1998; Denters 2002; Citrin & Stoker 2018). A second argument refers to identity (see for example Anderson, 1998). Harteveld et al. (2013) propose that trust is typically vested in actors and institutions perceived as belonging to the same system of identity, beliefs and shared values – namely national actors rather than EU ‘alien bureaucrats’. Similarly, based on community research and general sociological institutionalism (DiMaggio & Powell, 1983; Scott, 2010; Kappler, 2022), Offe (1999: 59) explores a perspective of general resentment towards the EU. Following these arguments, one may expect generally lower levels of trust in EU-level institutions, compared to national institutions.

When examining congruence and compensation theories on trust in the EU, it becomes apparent that trust in the various governmental levels is interconnected and mutually interdependent (Kumlin, 2009; Muñoz et al., 2011; Armingeon & Ceka, 2014). Levels of trust in EU institutions are mere derivations from trust into national institutions (Muñoz, 2017). Essentially, the congruence hypothesis expects a positive relationship between trust in national and EU institutions (Muñoz et al., 2011 and similarly in the ‘equal assessments’ model, for example, Anderson, 1998). Trust levels in national institutions, particularly, offer citizens a limited understanding of EU issues – a mere cue or proxy upon which to form attitudes towards EU institutions (Persson et al., 2019).

The compensation hypothesis suggests a negative relationship between trust in national and EU institutions. For example, in the case of low trust into national actors, a trustor tends to place trust in EU-level actors because they provide ‘them with alternative source of – potentially better – governance’ (Harteveld et al. 2013: 549). Hence, the EU and its actors serve as a lifebuoy, with the EU being perceived as a savior capable of solving problems that the nation state has been unable to address.

We contend that these arguments do not hold in the case of elite actors within regulatory regimes. Not only do different mechanisms come into play, but concepts such as identity and proximity also operate in unique ways. Indeed, Persson et al. (2019), although focusing on a different empirical context, examined national officials’ levels of trust in EU institutions and found that their trust is not primarily based on connotations and general sentiments associated with governmental levels. Similarly, we contend that dynamics of trust differ substantively when the trustor is an elite regime actor rather than a citizen. We observe significant disparities in terms of knowledge, (direct) experience, identity, and perception of similarity.

Elite regime actors are deeply immersed in and directly impacted by the agency’s work on a daily basis, actively participating in the regulatory process in various capacities (Hoffmann-Lange, 2018). They gain extensive information about the regulatory agency from multiple sources and are likely to actively seek out additional information. This allows for a more nuanced and less biased understanding of the agency’s actions and conduct. Thus, they possess an ‘awareness and understanding of what is taking place’ (Harteveld et al., 2013: 548) within EU institutions, even without or only limited interaction with certain institutions. Moreover, being part of the same regime and having insights into the overall functioning provides key actors with information about how it operates.

Building upon Anderson’s (1998) argument regarding EU complexity, elite actors whose raison d’être is to formulate and shape regulation within a specific (sub-)sector, can reasonably be assumed to possess detailed knowledge of collaborations, their purposes, and outcomes within the regulatory regime. Accordingly, regime actors’ access to information and understanding of regulatory practice is likely to be extensive, thus leading to a more accurate evaluation of trust mirroring the regulatory reality. Therefore, one can reasonably expect a relationship based on interaction and experience between the regime’s constituent actors, particularly with the regulatory agency. From this perspective, trust can be assumed to be developed through involvement, interaction and real experience (Harteveld et al., 2013). This aligns with the concept of identity-based trust (Lewicki & Bunker, 1996), which argues that individuals from a specific social category have an understanding of shared norms that is largely built through frequent interaction among members, leading to a general feeling of closeness and perceived (behavioral) similarity (Brewer, 1981).

In general, the logic of (perceived) similarity and shared identity operates differently for elite actors within a regulatory regime. For these actors, an alignment of interests, beliefs and values may not necessarily be rooted in their relationship with the national government but rather among like-minded peers (Gulbrandsen, 2007), which could extend to other countries or the EU level. Thus, it is less likely (compared to citizens) that trust is primarily tied to the national level due to emotional attachment alone. This is consistent with the concept of epistemic communities, where professionals from various disciplines and backgrounds produce policy-relevant knowledge about complex technical issues, sharing ”normative and principled beliefs” (Haas, 1992: 3; also Dunlop, 2011). Through ‘common professional socialization and culture’ (Mavrot & Sager, 2018: 394) these communities develop joint social identities that can be ‘supposed to be stronger when they have a long history and when policy actors have regular contact with others sharing the same social identity’ (Hornung et al., 2019: 214). Consequently, it can be assumed that within the regulatory regime where these features are present, the governmental level no longer assumes importance. Emerging communities span the boundaries of the national state, operating vertically in the multilevel system of the EU (Mavrot & Sager, 2018).

By delving into these considerations, it is plausible to assume that regulatory regime elites base their trust more substantially on information and experience rather than solely on a sense of attachment to a specific governmental level. Additionally, trust may stem from (perceived) proximity and shared identity, which due to the nature of the regulatory regimes under scrutiny depends less on the governmental level and more on the embeddedness of the elite actors.

Performance matters for trust

Their interaction with other actors within regulatory regimes, involvement in regulatory processes, and eventual experience with regulatory agencies provide both public and private actors within such regimes with specific insights into the agencies’ functioning. This includes aspects such as the staff and other inputs they utilize, their procedures and processes, and the quality of their regulatory decisions, instruments, and other outputs, offering perspectives on different aspects of the agency’s performance. Due to regime actors’ direct experience with the regulatory agency and their knowledge of the regulatory regime and the agencies’ performance, we expect that their assessment of trust in the regulatory agency will not be contingent upon the level of government but will align with the second perspective on trust formation in multilevel systems. This perspective suggests that trust in government institutions depends on a rational evaluation of their performance, irrespective of the government level at which they operate (Muñoz, 2017; Persson et al. 2019).

In general, the nexus between regulatory performance and trust can be approached through a virtuous circle (as suggested, for example, by social capital literature, see Fukuyama, 1995; Rothstein & Stolle, 2008; Bäck & Kestilä, 2009). As such, trust can drive performance. Trust can be the fuel for, for example, cooperation between actors, compliance and voluntary information flows that is the prerequisite for proper policy-making and highly performative regulation (Bélanger, 2017; Thomassen et al., 2017). The other way around, both outcome-oriented and procedural performance can drive trust. The quality of regulation sets the very foundation for trust to occur in the first place by, first, establishing the right setting for cooperation and voluntary flows to flourish and, second, by securing the trustee’s interests to be taken into account, to act according to rules and norms that are deemed appropriate and, finally, to be capable to do the job. All of these aspects decrease vulnerability a trustor faces. While the causal relationship ‘trust leads to performance’ - has received reasonable attention (La Porta et al., 1997; Peters, 2009), the opposite relation, ‘performance leads to trust’, is less frequently addressed. Despite being quite intuitive and well established, the idea of performance differences being decisive for trust levels remains a topic rarely discussed, particularly in the context of regulation (Huseby, 2000; Kumlin & Haugsgjerd, 2017).

However, literature exists on how the overall performance of government institutions or the quality of specific government services (see Kampen et al., 2006) drives citizens’ trust. Regarding the relevance of perceived performance for citizen trust in regulatory agencies, studies indicate positive effects (Murphy et al., 2014; Rudolph & Riley, 2014) stemming from perceived performance in general, or from more specific aspects of their input (Riley et al., 2018), process (Walker & Hills, 2014; Maman et al., 2023), or output-related performance (Driedger et al., 2014; Rudolph & Riley, 2014). Along these lines, we conceptualize the performance of the regulatory agency in line with both the general literature on performance management and the more specific literature on regulatory performance. We propose to understand performance as the emanation of a production process (see Ammons et al., 1999; Pollitt & Bouckaert, 2009; Van Dooren et al., 2015). In this perspective, performance encompasses the level and quality of input (such as resources, like finances and staff), throughput (activities, processes, and procedures), and outputs (regulatory decisions). Accordingly, as established in the literature on regulatory performance (see Coglianese, 2012), the performance of regulatory agencies refers to the extent of their success in acquiring and efficiently using appropriate resources, conducting activities and procedures appropriately, making high-quality regulatory decisions aligned with the agencies’ goals.

Actor heterogeneity

Since a regulatory regime cannot be considered as a homogeneous entity, a disaggregated understanding of regulatory elites can be beneficial (Mathieu et al., 2017; Maggetti & Ewert, 2018). Differences between public and private actors in terms of position, interests, and roles can substantively impact the perception of performance and emotional attachments to the governmental level, thus influencing mechanisms of trust assessment.

The bipolar conceptualization of public and private actors as either rule-makers or rule-takers (and cost bearers), and pro-social or for-profit, depicts the two actor groups as natural antagonists (Büthe, 2010). Following these lines, one may argue that the perception of what constitutes performance may vary. While some may argue that private actors are output-driven rather than concerned with in input and throughput, literature, particularly on responsiveness and inclusiveness, makes a clear case for regulatees placing importance on input and procedural aspects as well (Beyers & Arras, 2021; Blakelock & Turnpenny, 2022). Hence, we adopt an inclusive notion of elites that captures the entire regulatory landscape from a comprehensive bird’s-eye perspective. For both actor groups, the perception of the regulatory agencies’ performance can be assumed to be refined, allowing for a nuanced view of various performance aspects. Beyond the perception of performance, the disproportionate burden of regulation borne by public and private actors can lead to differences. While private actors bear the costs of regulation, those on the regulator side are affected to a lesser (direct) extent (Levi-Faur, 2011). Accordingly, poorly designed regulation directly impacts regulatees – potentially up to a severely detrimental level. Therefore, it is reasonable to expect private actors to be more sensitive to changes in regulatory performance compared to public actors.

Similarly, the positioning of public actors at the regulatory core and private actors at the regulatory periphery may lead to varying degrees of interaction density impacting both the knowledge level of regulatory processes and the perception of belonging to epistemic communities. Private actors, such as regulated companies, indeed, participate in regulation but are typically expected to have their core business operations situated outside - or at the periphery - of regulatory regimes, firmly anchored in their home countries (Reich et al., 2000; Kappler, 2022). Another group of private actors includes national business associations, which may engage more extensively at the EU level compared to their constituent companies. However, according to organizational literature, associations’ autonomy is limited, and they largely pursue the same interests as their member companies. Their close connection to the constituent companies positions them as an aggregate of private corporate actors and they can therefore be assumed to be part of the same community (Beger, 2002; Li, 2023). In contrast, public actors exclusively work in regulation and interact with core regulatory actors at both the EU and the national levels on a daily basis. Hence they can be considered ‘full members’ of the regulatory community, based on shared beliefs and a common identity that transcends governmental levels. The notion of ‘elite unity’ is manifested in various ways: (1) as consensus on significant values or desired properties of society, (2) subjective feelings or perceptions of belonging to the same community, (3) perceptions across different elite groups of being complementary and mutually dependent upon each other, and (4) instances of collective action and cooperation (Gulbrandsen, 2007: 194).

Hypotheses

Following the discussion on the factors influencing trust assessment, we expect regulatory core actors to assess trust levels in the regulatory agency in a specific manner. Specifically, we hypothesize that regime actors are less likely to rely on emotional attachments or to extend trust to specific governmental levels.

Hypothesis 1: Among elite regime actors, the level of governmental action itself does not impact trust levels in the regulatory agency. ¹

Additionally, we expect elite regime actors to base trust evaluation more rationally on the assessment of regulatory performance ² (Harteveld et al., 2013; Muñoz, 2017), irrespective of the level of government to which the agency is affiliated.

Hypothesis 2: Among elite regime actors, higher regulatory performance (in terms of input, throughput, and output) itself increases trust levels in the regulatory agency.

Regarding the relationship between the level of government and regulatory performance, we question the proposed interaction effect wherein the ‘relevance of a rational evaluation of an object by its own merits is probably conditional upon one’s emotional attachment’ (Harteveld et al. 2013, 547). Following Harteveld et al. (2013), and applying this to the mechanisms related to regulatory level and performance, one could expect that heightened performance substantively impacts trust levels in the national context and less so in the EU context. Conversely, since ‘love blinds’ (Harteveld et al. 2013: 547), emotional attachment may render performance more or less irrelevant in the national context. However, considering our suggestion that the governmental level – if at all – has minimal impact in an elite environment, we therefore do not expect any such interaction effect.

Hypothesis 3: The effects of regulatory level and regulatory performance on core regulatory actors’ assessment of trust are independent of each other.

As outlined, we anticipate that private actors will be more sensitive to differences in both performance and governmental levels. Firstly, private actors are conceptualized as rule-takers and cost-bearers with relatively low impact on policy-making and implementation. This is exacerbated by their limited access to information on various aspects of regulatory performance compared to regulatory actors. Secondly, being part of the corporate community in their home country rather than the EU regulatory sphere, traditional borders and, thus, varying emotional attachments to governmental levels may be evident.

Therefore, we hypothesize that public and private actors assess trust in core regulatory actors differently:

Hypothesis 4: Governmental level and regulatory performance have a stronger impact on the trust assessment of private actors compared to public actors.

Methodological considerations

To test our hypotheses, we designed a vignette experiment that was integrated, along with other questions, into a large-scale survey as part of an international comparative research project. The experimental design allows for establishing a clear causality in which variation in the governmental level and regulatory performance affect trust levels in the regulatory agency. This methodological choice, particularly benefits the scrutiny of the performance effect on trust which is a difficult endeavor because of the circular nature between the two. The respondents are active regime actors in three sectors with distinct characteristics (food safety, data protection, and finance) ³ across six countries with diverse politico-administrative traditions (Belgium, Denmark, Germany, the Netherlands, Poland, and Spain). These countries represent relevant variation among EU members states in terms of state structure (federal vs. unitary) and levels of (political and generalized) trust ranging from low to high (Maman et al., 2020). ⁴ We selected sectors that are politically and socially important for EU citizens and are all part of the broader realm of consumer protection. Furthermore, we aimed to capture regulatory variation in terms of factors such as the degree of centralization, maturity, salience of regulated risk, and the extent to which the regulated risk is transversal or industry-specific. To identify survey participants, we employed an occupational model of elites (Kertzer & Renshon, 2022) and targeted individuals in decision-making positions within parts of the organization directly related to specific subsectors under scrutiny (see footnote 3). Our overall objective was to reach individuals with both responsibility and expertise in the given sector and/or subsector. Respondents were selected based on organizational charts and positions within the top two levels of the organization, excluding the Human Resources department. The respondents represent nine types of actors comprising the regulatory regimes: ministries, regulatory agencies, parliamentarians, executive bodies, regulatory intermediaries (certification bodies), arbitration bodies, regulated organizations, business and other interest groups. In the case of the parliament, all politicians serving on parliamentary commissions responsible for the relevant sectors were surveyed. All respondents, except companies and related national business associations – categorized as private actors - are considered public actors.

The experiment presents various versions of a hypothetical scenario from a sector not covered by the survey population, namely energy regulation. This choice of a different sector allows us to utilize the same experiment for all respondents, ensuring a sufficient number of observations in each treatment group. Moreover, employing a completely different sector helps to avoid both (a) potential negative reactions from respondents who might feel offended and (b) associations between the hypothetical scenario described in the experiment with the respondents’ actual work, which could bias the results. As outlined above, we anticipate that the actors in the regulatory regimes can perceive regulation from a bird’s-eye perspective. We assume that this generic and comprehensive understanding of regulation functions reasonably well regardless of the specific regulatory action and sector, as procedures and principles are somewhat similar. In our case, all sectors fall within regulatory environments in the broader realm of consumer protection. Furthermore, energy policy was not particularly prominent on the political agenda when the experiment was conducted in 2020, thereby minimizing potential spillover effects from the broader societal context.

The experimental design

The survey experiment employs a 2 × 2 between-subjects factorial design. The first treatment differentiates between the level of the regulatory body and distinguishes between EU and national levels. The second treatment differentiates between improvement in regulatory performance and continued poor performance of the regulatory body.

Figure 1 depicts the procedure of the experiment. Firstly, respondents received an instructional text informing them about the general purpose of the experiment, data handling procedures, voluntary participation, guaranteed anonymity, and the fictional nature of the experiment. ⁵ In the second step, respondents were randomly assigned four vignettes (see an example vignette in the online Appendix), each containing unique combinations of regulatory performance and regulatory level. Every vignette portrays a scenario in which a regulatory body at either the EU level or the national level faces criticism from an independent evaluation committee, addressing various performance elements (input, throughput, and output). The scenarios depicted in the vignettes were then subject to either ‘improvement’ or ‘no improvement’, as determined by a re-evaluation conducted by the committee a few years later. Finally, all respondents received post-test questions regarding their trust in the regulatory agency.

Figure 1. Processing of experiment.

Source: Own compilation.

Variables

Treatment variables follow the factorial design introduced above. Each treatment is represented by binomial variables. The regulatory level distinguishes between the EU level and the national level. Therefore, using contextual cues, the agency is described as either a national- or EU-level agency. Regulatory performance is operationalized according to the theoretical argument formulated earlier. Performance is built on a triad of (a) input, which we approach by the agency’s ability to recruit highly qualified staff; (b) throughput, described irregularities in procurement procedures and (c) output, entailing the quality of regulatory decisions, such as an overly strict application of rules that does not consider different companies’ characteristics. The treatment variable distinguishes between high performance, indicating performance improvement, and low performance, indicating non-improvement.

Our dependent variable, trust is assessed using a well-accepted and well-tested item (Luo, 2005; Kääriäinen & Sirén, 2012: 281; OECD, 2017). Trust evaluation is based on respondents’ answers to a general question regarding how much they trust the regulatory agency on an eleven-point Likert scale ranging from ‘not at all’ to ‘completely’. This generic approach to trust reflects the general willingness to cooperate with the agency, even without personal acquaintance. Additionally, for robustness checks, we operationalized an alternative dependent trust variable proposed by Mayer et al. (1995), considering trustworthiness based on trustee’s perceived ability, benevolence, and integrity - i.e. ABI-aspects - as a multi-dimensional concept (Lewicki & Bunker, 1996; Six & Verhoest, 2017; Grimmelikhuijsen et al., 2020).

To address the internal heterogeneity of the regulatory regime, we consider actor types, specifically respondents’ group affiliations representing either public or consumer interests (labeled as ‘public actors’ below) or the regulatee interests (‘private actors’). Accordingly, regulatees and related national interest groups are assigned to the latter group, while all other actors are grouped under the former.

Furthermore, we incorporate a variable to control for potential (distorting) transpositions of the image the respondents hold from her - mostly gained as citizens - ‘real’ national energy regulator over the evaluation of trust in the experiment. To achieve this, we use import-export data from 2016 to 2018 obtained from the International Energy Agency. We categorize countries into two groups: those demonstrating good performance if they are net exporters and those exhibiting bad performance if they are net importers (IEA, 2023). Additionally, to account for country-specific effects that may influence citizen’s trust in local institutions, we introduce a dichotomous variable that categorizes respondents based on whether they hail from a high-trust or low-trust country, utilizing the latest European Social Survey data (2023).

Analysis and findings

Sample

In total, 752 elite regime actors participated in the experiment, marking it as the largest dataset covering the inner circle of regulatory regimes in the EU. Consistent with the comprehensive regulatory regime approach outlined earlier, we included a diverse range of regime actors. Our dataset comprised executive bodies, regulatees, regulatory agencies and business associations, each representing between 17 − 22 % of respondents. Additionally, legislative bodies and regulatory intermediaries accounted for 7 − 10 % of respondents. Ombudsmen, trade unions, and consumer associations followed representing approximately 3–5 %. The respondents originated from Belgium (24.92 %), Spain (11.38 %), Denmark (17.98 %), Germany (22.54 %), the Netherlands (9.75 %), and Poland (13.43 %). In terms of gender, roughly 62 % identified as male, 37 % as female, and 1 % as other. The age distribution showed a significant concentration of respondents between 36 and 65 years with peaks in the ‘46–55 years’ and ‘56–65 years’ age groups. Regarding respondents’ affiliation with either the ‘public sector’ or the ‘private sector’, 65.7 % identified as belonging to the former, while 34.3 % identified with the latter. Thus, our sample effectively captures the internal heterogeneity of regulatory regimes. Regarding the individual treatment groups, balance tests revealed no substantive differences concerning the variables discussed.

Do performance and governmental levels matter?

We employ a between-subjects design to address the following questions: (1) Is there a difference in average trust levels based on differences in the governmental level (main effect, independent of differences in performance)? (2) Is there a difference in average trust levels based on performance differences (main effect, independent of differences in governmental level)? (3) Is there an interaction effect between performance differences and governmental levels?

To approach questions (1) and (2), we utilize independent-samples t-tests to determine whether trust levels in the regulatory agency differ between two groups (e.g., national or EU governmental level). Robustness checks are conducted using Mann-Whitney U tests and are mentioned only in cases of deviation.

In the subsequent analysis, data are presented as mean ± standard deviation. Overall, 381 respondents assessed trust in the regulatory agency at the EU level, while 370 assessed it at the national level. We found homogeneity of variances for trust levels in both the EU and national groups, as assessed by Levene’s test for equality of variances (p = 0.091). Trust levels were normally distributed for both governmental levels, as confirmed by visual inspection of Normal Q-Q plots. Within our sample, mean trust levels were slightly higher in the national group (5.84 ± 2.089) than in the EU one (5.68 ± 1.906), with a difference of 0.158 (95% CI, −0.1283 to 0.444). However, the difference in mean trust levels is very small and not statistically significant, t(749) = 1.084, p = 0.279, d = 0.08.

Thus, Hypothesis 1 can be accepted. The governmental level, if it plays a role at all, has only a minor influence when regime actors assess trust in the regulatory agency within our sample. Concerning the broader population, we cannot observe differences in trust assessment among regime actors that are related to the governmental level.

Following the same rationale, we examined whether variations in performance (high and low) impact regime actors’ assessment of trust in the agency. A total of 379 respondents received treatment indicating an improvement in regulatory performance, while 372 respondents received the treatment indicating constant poor performance by the regulatory agency. We identified homogeneity of variances for trust levels for the EU and the national groups, as assessed by Levene’s test for equality of variances (p = 0.474). Trust levels were normally distributed for both national and EU governmental levels, as assessed by visual inspection of Normal Q-Q plots. Mean trust levels were higher in the high-performance group (6.32 ± 1.882) compared to the low-performance group (5.18 ± 1.950), representing a difference of 1.144 (95% CI, 0.87 to 1.42). The difference in mean trust levels was statistically significant, t(749) = 8.183, p < 0.005, d = 0.6.

Consequently, Hypothesis 2 can also be accepted. Regulatory performance substantively influences regime actors’ trust in the regulatory agency. Furthermore, controlling for various country-specific patterns, including citizen-related trust levels and experience with the ‘actual’ electricity sector, demonstrates stable effects of our independent variables of interest (see online Appendix).

Figure 2 illustrates the mean trust levels of regime actors in the regulatory agency (y-axis) at the basic level of further interest, that is per treatment group, each of which comprised roughly of 190 respondents. Upon initial visual inspection, the abovementioned findings are corroborated. Regardless of the governmental level differentiation, maintaining consistent performance measures appears to yield similar trust levels across treatment groups. Notably, at both EU and national governmental levels, substantive differences in trust levels are evident when performance disparities are considered.

Figure 2. Mean trust levels per treatment group ordered by governmental level and performance.

Source: Own compilation.

Considering the interaction effects between performance and the governmental levels on trust in the regulatory agency, we conducted two-way analysis of variance (ANOVA). Residual analysis was performed to test relevant assumptions. Data were found to be normally distributed, as assessed by visual inspection of Normal Q-Q plots. Homogeneity of variances was confirmed through Levene’s test for equality of variances (p = 0.090). Following the absence of a governmental level effect, we observed no statistically significant interaction between performance and governmental levels concerning trust level scores, F(1, 747), F = 0.285, p = 0.594, partial η² = 0.000.

Therefore, we can accept Hypothesis 3, as both performance and governmental levels do not significantly interact. In empirical terms, performance similarly influences trust levels across different levels of governmental action.

For robustness checks, all tests were replicated using the multi-dimensional variable of perceived trustworthiness (Six & Verhoest, 2017; Grimmelikhuijsen et al., 2020) as the dependent variable. All results were confirmed, indicating that general trust and ABI-based trust were very similar. This finding supports existing literature that sometimes treats the two terms synonymously.

Does trust assessment vary between public and private regime actors?

To account for the internal heterogeneity of the regulatory regime, we conducted two two-way ANOVAs to examine whether the effects of performance and governmental levels differ between public and private actor respondents. We found no statistically significant interaction between governmental level and actor type regarding trust levels in the regulatory agency, F(1,747) = 0.002, p = 0.967, η² = 0.000. Along similar lines, no significant interaction between performance differences and actor type were observed, F(1,747) = 0.049, p = 0.824, η² = 0.000. Thus, the mechanisms that translate performance differences and governmental level into trust levels appear to function equally among both public and private actors.

To assess whether public and private actors differ in their trust assessment of the regulator (rather than examining whether actor type interacts with treatments), we tested the main effect of actor type on trust. A statistically significant main effect of actor type on trust levels in the regulatory agency was found, F(1, 747) = 12.5079, p < 0.001, partial η² = 0.016. The marginal means for trust level in the regulatory agency were 5.938 ± 0.087 for those representing public and consumer interests, and 5.426 ± 0.115 for respondents representing regulatees’ interests – indicating a statistically significant mean difference of 0.512 (95 % CI, 0.210 to 0.769), p < 0.001. Therefore, there is a slightly higher overall assessment of trust in the regulatory agency among public actors compared to private actors, irrespective of the experimental treatment. However, since neither governmental level nor performance is moderated by actor type, we reject Hypothesis 4. Both independent variables of interest demonstrate stable effects across public and private actor respondents.

Discussion

Extant literature suggests that citizens’ trust in institutions within a multilevel context is largely explained by (a) governmental level (and its emotional connotations), (b) a rational evaluation of institutional performance, or (c) an interaction of both factors. However, our study reveals substantive differences in the assessment of trust in regulatory agencies by elites within regulatory regimes. We observe that the governmental level plays no role in the assessment of trust in regulatory agencies by regime elites. Hence, proximity appears to be either irrelevant in a regulatory elite environment or, more likely, the boundaries of ‘proximity’ have shifted alongside daily operations and functional lines. This implies that regime actors may perceive themselves as proximate to both national and EU-level regulatory agencies due to their interactions and connections within work-related and sectoral networks. Such a finding may indicate the emergence of a European identity that does not necessarily replace but rather complements national identity (Armingeon & Ceka, 2014). Following this logic of identity, elites’ attachment to the EU level may be sufficiently high (and comparable to their attachment to the national level) to render the level of government irrelevant as a driver for trust (Harteveld et al., 2013).

Disaggregating the regulatory regime, our findings suggest that trust mechanisms remain stable across both private actors representing regulatees’ interests and public actors representing public and consumer interests. This finding is surprising but may bolster the notion of a regulatory regime as a relatively cohesive professional and sectoral community, characterized to a substantive extent by a shared sense of proximity and collective identity. This shared identity may equalize or even surpass private actors’ perceived affiliation with the national level. Consequently, differences between public and private actors may become less important. Thus, rather than being viewed as ‘natural antagonists’ at either the core or the periphery, both actor types appear to function as ‘full members’(Gulbrandsen, 2007) of the regulatory community. This community transcends the boundaries of governmental levels and offers a nuanced perspective on regulatory performance. Nevertheless, it is noteworthy that pro-social values (implicitly associated with public actors as discussed by, for example, Büthe (2010)), and differences in actor types related to their roles within the regulatory regime (e.g., as suggested by Perry & Hondeghem, 2008), seem to impact trust assessment primarily by leading to generally higher levels of trust among public actors, regardless of performance or governmental levels. This finding may contribute to the ongoing debate surrounding the existence and role of horizontal and vertical epistemic communities that share beliefs and professional culture across the public-private divide within regulatory regimes in specific sectors of the EU (Dunlop, 2011; Cross, 2013; Mavrot & Sager, 2018).

The relative high homogeneity of trust assessment by both public and private actors may also be explained by the benefits accrued through regulation for regulated companies and their interest groups. For regulatees and their interest group, the performance of a regulatory agency serves as the primary criterion for assessing trust in that agency. Consistent with the findings of Six and Verhoest (2017), a well-functioning regulatory agency that garners trust is beneficial not only for ensuring their own compliance but also for fostering demand for their products and services among citizens/consumers. Trust in the regulator is a prerequisite for consumers to engage in market interactions in the first place.

Moreover, professionals within the regime seem to prioritize regulatory agencies’ efficacy in ‘getting the work done’. As our research indicates, regulatory performance plays a significant role when regime actors assess trust in the regulatory agency. Along these lines, performance and governmental levels do not exhibit significant interaction, indicating that performance similarly influences trust level at both EU and national levels of government. Consequently, our findings suggest a more rational evaluation of institutions at both national and EU levels. In such cases, actors place trust in institutions ‘based on performance in terms of policy of the institutions, or on procedural considerations related to perceived accountability or responsiveness’ (Muñoz, 2017, 77). This approach ‘assumes trust to be the consequence of subjects’ evaluations of an object by its merits’ (Harteveld et al., 2013, 3). A rational evaluation of this nature requires that actors are well-informed about the areas of competence and the performance of institutions, a criterion that can reasonably be assumed to apply to regime actors. Although we categorize regime actors’ evaluations of regulatory agency performance as rational, it is important to recognize that different regime actors may still vary regarding their interpretation of what constitutes good performance. This variation arises because the definition of good performance in terms of input, throughput, and output is socially constructed and dependent on the specific interest of regime actors (for example, regulatees may define good performance difference than consumer associations). However, despite this nuanced understanding, our analyses demonstrate that the assessment of regulatory performance is crucial in determining trust within a multilevel context, rather than being contingent on the governmental level at which the institution is situated.

Conclusion

Based on an experiment involving elite respondents from within regulatory regimes, we have found that while the government level does not play a substantive role in the assessment of trust among regulatory elites in regulatory agencies, agency performance clearly does matter. Moreover, performance similarly affects trust levels at both levels of governmental action, with no significant difference observed between the judgment of public and private actors within the regime.

Interpreting these findings with the necessary caution, we find some support for our underlying assumption that elites within a regulatory regime – regardless of their specific position in the regime – at least partially base their trust assessment on factors different from those of citizens. The implications of this finding are manifold. The relevance of performance, rather than the level of government, suggests a more rational evaluation of institutions, both at the national and EU levels, guiding elite’s trust formation. Additionally, it implies a perceived proximity of these groups to the EU level in terms of cross-border epistemic communities (Haas, 1992; Cross, 2013). Along these lines, the encouraging news may be that belonging to either the national or EU governmental level does not inherently lead to differences in trust assessment of regulatory agencies. However, it does indicate that elite actors are particularly sensitive to the performance of regulatory agencies.

Our findings also carry implications for regulatory practice: If there is significant disparity in experience and understanding of agency performance between regulatory elites and citizens, this could impact the relevance of direct communication by regulatory agencies with citizens regarding their performance and their ‘perceived value’ for citizens. Scholarship on bureaucratic reputation indicates that agencies use communication strategically to cultivate their unique reputations (Maor, 2020), and regulatory agencies are particularly sensitive to reputational threats related to their performance (Verhoest et al. 2021; Bach et al., 2022). Future research could further explore the interrelationship between regulatory communication, trust, and reputation (see Lee & van Ryzin (2020) on agency reputation and citizen trust). Despite the strong effect of performance, it is reassuring news considering the necessity of trust, as it can be directly influenced by the regulatory agency itself.

Our findings suggest avenues for further research. Sectoral and country differences, such as varying depths of EU embedding and different experiences with the national regimes, deserve more scrutiny. Given the relatively small target population with regulatory regimes, we would recommend to complement our approach by more qualitative research that may dig deeper into peculiarities that may contribute to explaining variation. In addition, the variation in factors generating trust among citizens and elites, especially regulatory actors, could be explored in other (experimental) studies to gain deeper insight into differences and similarities.

Furthermore, we cannot dismiss the possibility that the lack of a statistically significant effect of the governmental level treatment is partially due to the experimental design itself. The treatment effects may have been diluted to some extent by the complexity of the vignette (Brutger et al., 2022), which provided detailed information on performance dimensions, while mentioning the governmental level only at the outset. Therefore, future research should consider different experimental designs that explore the same theoretical argument (Mutz & Kim, 2020).

However, the roles of both governmental level and performance in the underlying trust formation processes and mechanisms remain black boxes and warrant further scrutiny. Along these lines, the implicit aspect of proximity could be further elucidated in future research: Currently, we only observe that trust levels (in absolute terms) appear to be unaffected by the governmental level. However, that does not indicate irrelevance of the governmental level in trust assessment per sé. By no means can we rule out that other relevant factors of trust assessment may interact with the governmental level. In addition, it would also be intriguing to assess the relative impact of changes in the various dimensions of performance (inputs, throughput, and output) of regulatory agencies on trust. Similarly, although we do not discern any disparity in the assessment between public and private actors within the regime, exploring regime composition and the specific roles of actors could be valuable. Lastly, while our findings primarily pertain to trust, delving into the formation of distrust among regulatory elites may also be worthy of further investigation.