Studying the influence of national parliaments in EU affairs: reconnecting empirical research and the Principal-Agent approach

ABSTRACT The literature on national parliaments and the EU has built up its knowledge on the strength and activities of domestic parliaments in the European Union. While these studies were of outmost importance to understand how and why parliaments and MPs act on EU affairs, we still know little on the influence of domestic parliaments on the EU policy of their executive and on EU policy-making more generally. The article first presents the current gaps in knowledge as well as the challenges to study the influence of parliaments in EU affairs and ends by presenting a research strategy which advocates ‘reconnecting’ empirical works on national parliaments and the EU with the Principal-Agent model. This research agenda could foster new theoretical and empirical studies with important implications for the debate on the (re)parliamentarisation of the European Union as well as on EU negotiations.


Introduction
The literature on the role of national parliaments in the European Union (EU) has greatly evolved over the last 25 years and has built up its knowledge on the strength and activities of national parliaments in the EU (Auel 2016). In the last years, scholars have also started to reflect on the determinants of an effective parliamentary behaviour in EU affairs and questioned whether parliamentary scrutiny had an influence on executive actors (e.g. Christiansen and Fromage 2019; Miklin 2014; Smeets and de Ruiter 2019;Sprungk 2010). Indeed, as Rozenberg and Hefftler argue, forthcoming research, 'to convincingly address the debate on de-parliamentarisation, has to focus on the impact of parliaments on the EU policy of their governments, and on EU legislation more generally' (Rozenberg and Hefftler 2015, 25).
Following the call of Rozenberg and Hefftler (2015), the question this research note seeks to address is the following: does parliamentary activity taking place at the domestic level have an influence on the EU policy of domestic and of European executive actors? Phrased differently, does parliamentary activity translate into parliamentary influence? Influence is understood as the ability of domestic parliaments to shape the position of domestic executives and of European-level executive actors in line with their preferences (Dür 2008;Helboe Pedersen 2013). Since 'influence can be caused by one's own intervention or by the decision-makers' anticipation' (Arts and Verschuren 1999, 413; see also. Russell and Gover 2019), but also by actors outside of the traditional legislative-executive relationship, the assessment of influence is a challenging task, especially in complex (EU) decision-making processes. While no solution is perfect in addressing this question, the challenges associated in measuring parliamentary influence should not stop scholars continuing to take up the challenge.
To this end, the research note introduces a research strategy advocating the use of the Principal-Agent approach as a heuristic device to empirically study the determinants of domestic parliaments' influence on EU affairs. While the use of the Principal-Agent in legislative studies is not new in itself, the Principal-Agent literature has been increasingly fragmented across (sub)disciplines of the social sciences in the last years (Delreux and Adriaensen 2018). Therefore, one of the novelties of this research note is that it tries to 'reconnect' empirical works on national parliaments and the EU with the most recent developments of the model, and in particular applications of the latter in the field of EU studies.
The next section provides an overview of the most prominent research agendas on national parliaments and the EU over the last 25 years. In particular, it discusses the institutional and behavioural Europeanisation of parliaments and concludes by reviewing the current studies focusing on the influence of parliaments in EU affairs. On this background, section 3 presents the main challenges scholars face when analysing the influence of national parliaments in EU affairs. The next section introduces the research strategybased on the Principal-Agent modelto study the determinants of national parliaments' influence on the EU policy of domestic and of European executive actors. The article ends with a short conclusion.

years of research on national parliaments and the EU
The institutional Europeanisation of national parliaments As a first step, scholars focused on the institutional Europeanisation of national parliaments. The most visible outcome from these studies is the development of rankings evaluating the (formal) strength of national parliaments in EU affairs (Auel, Rozenberg, and Tacea 2015;Karlas 2011;Raunio 2005;Saalfeld 2005;Winzen 2012). To explain the variation in the strength of parliaments, both QCA studies of Raunio (before the enlargement, 2005) and of Karlas (after the enlargement, 2011) outlined the importance of the domestic strength of parliaments independently of integration. Raunio (2005) also outlined that the combination of a powerful parliament and of a Eurosceptical electorate were sufficient conditions to explain tighter levels of formal powers. While this first set of studies was important to understand how and why parliaments adapted to European integration, most studies did not (yet) enquire about the actual use of their formal powers.

The behavioural Europeanisation of national parliaments
As a second step, scholars started to discuss the relevance of evaluating the strength of national parliaments' performance solely based on institutional and formal criteria (Auel 2007). These interrogations gave birth to a second research agenda focusing on the behavioural Europeanisation of national parliaments and aiming at understanding how and why national parliaments use their formal powers. A first set of studies highlights the importance of understanding the MPs' role orientations and attitudes towards European integration (Goetz and Meyer-Sahling 2008;Katz and Wessels 1999). These studies outline the need to address parliamentary involvement not only from a rationalist perspective, but in considering the motivations MPs have when they perceive their own roles (see. Rozenberg 2009Rozenberg , 2012Blomgren and Rozenberg 2012;Wessels 2005).
A second set of studies focuses on party politics and in particular, on the role of opposition and majority political groups (Auel 2007;Finke and Dannwolf 2013;Finke and Herbel 2015;Holzhacker 2005;Saalfeld 2005;Sprungk 2010). We learn from these studies that opposition parties are stronger in minority governments' systems, closely followed by parliaments with coalition governments (Holzhacker 2005). Scholars also identified the existence of a disagreement with the government's position, a 'weak' government in terms of seat share (Finke and Herbel 2015, 508) or the presence of strong EP delegation (Finke and Dannwolf 2013, 738) as factors stimulating the scrutiny of opposition parties. That being said, while some studies outlined the existence of an 'opposition deficit' (Rauh and de Wilde 2018), this finding has recently been nuanced by Karlsson and Persson (2020) who observed more opposition than previously assumed.
A third set of studies deals with the communicative function of national parliaments. These works investigate the links between MPs and voters, along with the (de)politicisation of the EU in national parliaments (Neuhold and Rosen 2019). Recent analyses of parliamentary debates have challenged the beliefs that the role of the plenary is limited and that a 'depoliticisation' process of EU policies in national legislatures is preferred by political actors (Auel and Raunio 2014;Miklin 2014;Rauh and de Wilde 2018;Wendler 2016). In particular, a great share of studies focused on salient political eventssuch as the Eurozone crisis, European Semesters, Banking Union or Brexitand discussed their consequences on domestic parliamentary involvement (see. for instance, Auel and Hoïng 2015;Crum 2018;Kreilinger 2018;Meissner 2019;Smeets and de Ruiter 2019). Overall, these studies have shown that national parliaments are increasingly discussing EU issues, which suggests a movement towards a politicisation of the EU in national legislature.
A fourth set of studies focuses on specific scrutiny instruments (at the exception of parliamentary debates, see supra). For instance, some authors discussed the use of parliamentary questions (Rozenberg and Martin 2011) while others analysed specific forms of (collective) scrutiny, such as the use of resolutions and subsidiarity opinions. For example, Hoerner (2017) found that resolutions were mostly used as a position-taking tool rather than as a controlling mechanism. Others have focused on the activity of national parliaments in the early warning mechanism (EWM) (Gattermann and Hefftler 2015;Rozenberg 2017). Overall, while some studies are critical and outline that national parliaments face many challenges to act effectively within the EWM (de Wilde and Raunio 2018), other studies nuance this picture and highlight the long-term influence (Cooper 2019) and 'signaling' role of subsidiarity opinion (Van Gruisen and Huysmans 2020).
Finally, a last set of studies investigate the relationship between domestic parliaments and specific EU (executive) actors, such as the European Council, the European Commission (beyond the EWM) and, to a lesser extent, EU agencies. For instance, Hefftler et al. (2013) analysed the scrutiny conducted by national parliaments over the European Council and Euro summits and found that parliaments eventually 'learn to fight back'. Crum and Oleart (2020), on their side, examined direct relations between the Commission and national parliaments (i.e. visits of Commissioners). Finally, some studies also discuss the parliamentary scrutiny of EU agencies, notably Europol, Eurojust and Frontex (Wills and Vermeulen 2011). This second research agenda made it possible to open the 'black box' of parliaments by focusing on the use of scrutiny mechanisms by parliamentary actors and on their relationship with EU (executive) actors and institutions. We learned that the variation in the use of formal powers by MPs is best explained by a mix of institutional capacities and motivational incentives, completed by sociological incentives (e.g. parliamentary culture or attitudes towards European integration) (Auel, Rozenberg, and Tacea 2015;Auel and Christiansen 2015). Yet, most of these studies took the formal strength of parliaments (institutional adaptations) or their degree of activity (behavioural adaptations) as their dependent variable, hence making it difficult to assess the influence of parliaments on domestic and EU level executive actors.

The influence of national parliaments in EU affairs
While studying the influence of national parliaments in EU affairs has not been the main focus of the literature in the past, some scholars already took up the challenge. On this matter, a first set of (quantitative) studies focused on Council decision-making and on the EWM (see. Van Gruisen and Crombez 2019; Van Gruisen and Huysmans 2020; Hagemann, Bailer, and Herzog 2019). For instance, Hagemann, Bailer, and Herzog (2019, 646) found that 'when a national parliament has extensive competences to scrutinise and amend government policies, governments are more likely to oppose the majority in the EU Council and record their positions in formal policy statements'. The findings of Van Gruisen and Crombez (2019) on the Council of the EU are in line with Hagemann, Bailer, and Herzog (2019, 996): the stronger the parliament, the more likely governments are to vote 'No' or abstain or issue negative statements. Finally, analysing the EWM, Van Gruisen and Huysmans (2020) outlined that the adoption of reasoned opinions by national parliaments was a strong predictor of government opposition in EU negotiations. The authors also found that the Commission was more likely to withdraw a proposal if national parliaments submit reasoned opinions.
A second set of studies concentrates on salient and politicised issues, including Brexit (e.g. Christiansen and Fromage 2019), the Euro crisis and EU economic governance (e.g. Fromage and van den Brink 2018; Jančić 2016) or the 'contentiousness' of trade policies (Jančić 2017;Roederer-Rynning and Kallestrup 2017). The edited volume of Christiansen and Fromage (2019, see also the chapter of Smith 2019) outlined the major influence that the UK parliament had on the process and outcome of Brexit. However, in Belgium, Ireland, Poland and Spain, the role of parliaments was mainly limited to information gathering rather than policy-influencing. Another policy field that received a lot of attention relates to trade policies, and in particular the negotiations of the Comprehensive Economic Trade Agreement (CETA) and of the Trade and Investment Partnership (TTIP). The nature of the agreement (i.e. is it a mixed treaty?), access to information as well the relationship between the parliamentary majority and the government were identified as key factors to gain parliamentary influence. Finally, a last set of studies focused on the influence of parliaments on negotiations leading to the adoption of directives or regulations (Miklin 2009(Miklin , 2014Randour 2018;Sprungk 2010). For example, Miklin (2009) in his analysis of the Service Directive in Germany, Austria and Swedenfound that an Agent is more likely to push forward his own preference at the beginning of the legislative process rather than at the end, where representatives in the Council 'had to pay considerably more attention to factors like economic adaptation costs and institutional misfit' (Miklin 2009, 944) previously highlighted by the Principal. A second example is the study of Randour (2018) that examined the negotiation autonomy of the Austrian, Belgian and German federal executives in the Council of the EU vis-à-vis their domestic parliaments. The study outlines that executives enjoyed an important level of autonomy vis-à-vis their domestic parliaments, even in cases where parliaments were active scrutinizers. Overall, these studies paved the way to the study of parliamentary influence and helped in identifying key factors impacting on executives and on EU policy-making more generally. However, as it will be discussed in the next section, there are two main (empirical) problems when one wants to study the influence of national parliaments on the EU policy of their domestic executives and on EU-level executive actors.

Challenges to study the influence of national parliaments on domestic and European level executive actors
The first challenge is 'observational equivalence'. This problem 'arises when a perceived absence of (formal) control or conflict between Principal and Agent can be explained as either a case of extreme Agent autonomy or near perfect Principal control' (Damro 2007, 887;Delreux and Adriaensen 2017;Reykers and Beach 2017). As Saalfeld (2005) puts it, a lack of visible control in the parliament does not necessarily mean that parliaments abdicate in controlling their governments. Indeed, governing parties also rely on more informaland therefore less visiblestrategies to scrutinise their government (i.e. internal coordination within and across majority parties) (Auel and Benz 2005). In addition, executives can also anticipate the preferences of their Principal (Arts and Verschuren 1999, 413; see also. Russell and Gover 2019), therefore increasing the complexity to measure the influence of parliaments.
The second challenge is data availability on EU negotiations. EU institutions are often described in the literature as opaque institutions, with only a limited amount of information available to the public. Still, according to Cross and Bølstad (2015), there is nowadays a tendency towards a higher level of transparency for EU decision-making processesat least for the Commission, the European Parliament and the Councilsince the implementation of the transparency regulation of 2001 (Regulation (EC) 1049/2001). Even though not all legislative records are publicly available (Hagemann and Franchino 2016), the progress made in the past year in terms of accessibility to EU documents could make it possible to (better) assess if the government followed parliament's wishes but also whether what the executive says in front of the parliament is what happened during the negotiations.
Linking empirical research on the influence of national parliaments and the Principal-Agent model The strategy presented in this research note advocates the use of the Principal-Agent approach as a heuristic device to empirically study the determinants of domestic parliaments' influence on EU affairs. As outlined by Rozenberg and Hefftler (2015:, 20), 'there has been a certain lack of connection between the empirical research on the issue and the conceptual approaches using agency theory'. Building on earlier works using the Principal-Agent approach, and following the call of Rozenberg and Hefftler (2015), the section discusses how forthcoming studies on parliaments and the EU could further develop and connect their empirical research with the Principal-Agent literature, while at the same time going beyond the analysis of the formal strength and activities of domestic parliaments. In particular, it reflects on the influence of the Principal (i.e. domestic parliaments) on (i) the EU policy of their domestic Agent (i.e. the domestic executive) acting in the EU Council or in the Council of the EU and (ii) on their (in)direct influence on supranational Agent(s) and EU policy-making more generally (e.g. the Commission, ECB or EU agencies).
The Principal-Agent model: major developments in EU studies and beyond Since the original development of the model in the field of economics in the 70s, the Principal-Agent model has increasingly been applied to international organisations , European (e.g. Pollack 2003) and legislative studies (see. below). More precisely, the pivotal work of Pollack (2003) and the rational neo-institutionalist turn in EU studies paved the way for a first batch of studies focusing on the 'politics of delegation' to the EU (i.e. how governments delegated power to EU actors, institutions or agencies) and in the EU (i.e. how delegation occurs between EU actors, institutions or agencies) (Delreux and Adriaensen 2017, 4). A second batch of studies quickly followed and analysed the 'politics of discretion', meaning the actual political dynamics taking place between an Agent and his Principal. More recently, these studies also started to emphasise that 'Agents matter' (Hawkins and Jacoby, 2006) and consequently, included the Agent's strategic behaviour as an integral part of analysis (Delreux and Adriaensen 2017).
The Principal-Agent model has also been applied to the analysis of delegation and accountability mechanisms in parliamentary democracies (Bergman 2000;Strom, Müller, and Bergman 2003). In legislative studies, the attention of scholars was mainly directed to the relationship between parliamentsacting as the Principaland their government, acting as the Agent. On this matter, already numerous studies examining executive-legislative relationship were embedded in new-institutionalist approaches, and in particular in rational-choice institutionalism and the Principal-Agent approach. For example, the 2015 West European Politics special issue on National parliaments and the EU (see. Auel and Christiansen 2015) is grounded in a neo-institutionalist perspective, underlying both the need to study the institutional capacity of parliaments and their motivational incentives (i.e. a mix of rational-choice and sociological institutionalism). Other works dealing with the EU-related activities of domestic parliaments were also relying on neo-institutionalist approaches, and in particular on the Principal-Agent model (e.g. Auel 2007;Auel and Benz 2005;de Ruiter, 2013;Miklin 2009Miklin , 2014Saalfeld 2005;Sprungk 2010).
Why should a new batch of (empirical) studies on domestic parliaments and the EU try to further develop and connect their work to the Principal-Agent literature? Overall, the model offers three concrete advantages for forthcoming empirical research on national parliaments and the EU. First, (1) rather than starting from a dependent variable that is situated on the 'Principal' side (i.e. the strength or activity of parliaments), recent developments of the Principal-Agent approach opened the way to the use of a dependent variable related to the 'Agent' side (i.e. the discretion, autonomy or deviation of an Agent). Consequently, (2) research adopting such a dependent variable starts from the perspective of the Agent, which de facto allows one to link the worlds of the Principal to the one of the Agent. Finally, with the 'Agent matters' turn in Principal-Agent studies, the model also has the potential to (3) embed empirical research on national parliaments and the EU in a broader heuristic device including both dependent and independent variables.

Reconnecting the Principal-Agent model and empirical research on national parliaments
Following the call of Maher, Billiet, and Hodson (2009) in applying the Principal-Agent model 'liberally' yet with some 'care', the first step in a Principal-Agent analysis is to map the Principal and the Agent and to identify the delegation act (Delreux and Adriaensen 2018). When analysing the parliament-executive relationship, we are facing a situation of collective Agent and Principal (see. Figure 1). On the one hand, the Principal can be understood as the parliamentary majority. However, as Agent of the voters acting as ultimate Principal, 'even opposition parties (have to) hold certain rights vis-à-vis governments' (Sprungk 2010, 5). Opposition parties ensure that the government justifies its policy proposals via public deliberation but cannot directly block a decision nor remove the government from office. The Principal can therefore also be conceptualised as the parliament as a whole. Finally, to evaluate the influence of parliamentary scrutiny on executives from federal Member Statesespecially in policy fields with shared competences -, it is essential to evaluate the combined actions of multiple parliaments within one Member State, even though subnational parliaments only have an indirect effect on federal governments. Therefore, the Principal can also be defined as the parliamentary system, understood as a collective Principal composed of both subnational and national parliaments (Randour 2018).
On the other hand, executive power in the EU is best described as highly fragmented, as there is no single and comprehensive EU government (Crum and Curtin 2015; Curtin 2014). Actually, executive actors can be found at the domestic and at the European levels. At the domestic level, the domestic Agent can be the head of State, the head of the Government (i.e. prime minister), a specific minister, the government or the executive (thus including national representatives negotiating in COREPER or at working group levels) acting and pooling sovereignty together with the other Member States in the European Council or in the Council of the EU (dotted arrows in Figure 1). Second, at the European level, there also exist several supranational Agents, notably the European Commission, the European Central Bank (ECB), the European External Action Service (EEAS) or EU agencies (e.g. Frontex, EFSA, etc.). Both domestic and supranational Agents can be scrutinised by the Principal (dashed arrows in Figure 1) allowing domestic-level legislative actors to influence the EU policy of their executive. While there is an exclusive delegation (full arrows in Figure 1) between the domestic Principal and the domestic Agent, this is not the case regarding supranational Agent(s). Indeed, the delegation between the domestic Principal(s) and supranational Agent(s) is done via the domestic Agent. This situation does not mean that the Principal(s) cannot scrutinise supranational Agent(s). In fact, the Principal can also control the actions of EU-level executive actors via their domestic Agent but can also favour a direct engagement with supranational Agent (e.g. EWM, visit of Commissioners in domestic parliaments).

A dependent variable situated on the Agent's side: the negotiation autonomy of executives
While looking at empirical research in EU studies using the Principal-Agent model (e.g. Conceição-Heldt 2011, 2017; Delreux and Adriaensen 2017, 2018; Delreux and Kerremans 2010; Pollack 2003; Laloux 2021), we can find numerous studies taking a dependent variable situated on the Agent side, such as the autonomy or discretion of an Agent acting on behalf of another (collective) actor. As explained previously, these variables offer the advantage to de facto include both the actions of the Principal and of the Agent in the analysis.
More precisely, the Principal-Agent literature makes a distinction between the notions of discretion and autonomy ). While 'discretion grants the leeway to accomplish Principal-determined goals in the way seen fit by the Agent, autonomy is the range of action available to the Agent, including the ability to set policy goals' (Niemann and Huigens 2011, 426). In other words, the concept of discretion measures the 'room for manoeuvre' an Agent has to realise the Principal's wishes, but it only focuses on the delegation setting (i.e. institutional strength) and the control of the Principal (i.e. activity). In short, the concept of discretion focuses on the formal prerogatives of domestic parliaments (i.e. their institutional Europeanisation) and on their control activities (i.e. behavioural Europeanisation). The concept of autonomy is more encompassing as it takes into account the whole set of actions of the Agent, including the possibility to set his policy preferences, to strategically use his Principal for forthcoming negotiations or to weaken the incentives of controls the Principal has.
How to operationalise these dependent variables? In practice, the distinction between discretion and autonomy is based on Pollack who operationalises discretion as follows: discretion 'can be measured by first examining the extent of delegation in any given act, but then taking into account all of the possible constraints that Principals might impose upon their Agents. The net of these delegation and constraint ratios, then, provides a measure of discretion' (Pollack 2003, 28). Based on this quote, discretion can be operationalised as the net of the delegation win-set and of the control mechanisms activated by the Principal (discretion = delegationcontrols).
In order to measure the autonomy of an Agent, it is necessary to go one step further and to enquire about how the executive reacted to these control mechanisms. To do so, it is required to unpack both (i) the origin and purpose of the parliamentary scrutiny and (ii) the (re)actions of domestic and/or of EU-level executive actors during EU negotiations. 1 As a first step, if one wants to understand the (re)actions of executive actors vis-à-vis domestic parliaments, it is needed to go beyond the analysis of the sole quantity of parliamentary scrutiny. This can be done by investigating who is at the origin of the parliamentary scrutiny (i.e. is it an initiative of MPs or was it motivated by a demand of the executive?) and what its purpose is (is it aiming at position-taking, controlling or supporting the executive, communicating with citizens?). As a second step, it also requires to unpack how the Agent(s) behave at the European and domestic levels. More precisely, it is first necessary to identify the initial negotiation position of the domestic Agent at the start of the EU negotiation as well as the main evolutions of the latter during the decisionmaking process (i.e. a focus on the European level). Second, it is needed to investigate how the Agent acted at the domestic level, and in particular vis-à-vis the domestic parliaments (i.e. a focus on the domestic level). Scholars should pay attention to the extent and the kind of information that the Agent shares with the Principal, whether the Agent justifies his negotiation position in front ofor in reaction tothe parliament and finally, the kind of argumentative frames that are used by the Agent.
Finally, the last step consists in analysing whether and how the scrutiny conducted by the domestic Principal translates into an evolution in the domestic Agent's negotiation position (i.e. linking the domestic and EU levels), while taking into consideration what happened within the collective Agent and during the EU negotiations. Indeed, the influence of a domestic parliament needs to be understood in the broader context of (a) pressures within the Agent (e.g. divergent positions across ministerial departments) and (b) pressures within the (European) Council. The next section investigates potential independent variables related to both the Principal and Agent sides in greater detail.

Independent variables: internal, external and issue related factors
The Principal-Agent model also allows for the identification of independent variables. Indeed, to measure the discretion of an Agent, one has to look at three types of variables: (1) internal variables related to the relationship between the Principal and the Agent; (2) variables related to the issue negotiated and finally, (3) external variables, such as the negotiation environment in which the Agent evolves.
The first set of variables are linked to two key assumptions of the Principal-Agent model: goal conflicts (i.e. between the Principal and the Agent; within the (collective) Principal; within the (collective) Agent) and informational asymmetry (i.e. the degree of information of the Principal and how the Agent manages information flows to the Principal). For instance, it can be expected that in the case of heterogeneous preferences between the Principal and the Agent, the later will enjoy less negotiation autonomy (Dür and Elsig 2011;Pollack 2003;Waterman and Meier 1998). In addition, conflicts can also take place within the collective Agent (Gastinger, 2017). In the case of national parliaments and the EU, it can be assumed that visible conflicts within the Agent (e.g. during the executive coordination process for example) will provide a stimulus to the Principal to activate their control mechanisms, which can lead to a decrease in the negotiation autonomy of the latter. Finally, in the case of a disagreement between Principals, the domestic Agent will enjoy more negotiation autonomy as the Agent will be able to use disunity within the collective Principal at his own advantage (Conceição-Heldt 2011Niemann and Huigens 2011). Now turning to informational asymmetry, the prevalent hypothesis in Principal-Agent literature is that informational asymmetry benefits the Agent (Pollack 2003;Delreux and Adriaensen 2017). However, informational asymmetry is not always about an Agent's bad-will, as it can also become detrimental to the Agent under certain conditions (Coremans and Kerremans 2017).
Informational asymmetry can also be linked to the characteristics of the issue negotiated. In particular, the complexity of the file negotiated can have a detrimental effect on the Principal-Agent relationship. As Laloux (2021) outlines, gaining knowledge on a complex file is (even more) costly for the Principal and this leads to an increase in the degree of information asymmetry. In the case of a complex file, it is therefore even more difficult for domestic parliaments to judge the work of domestic and supranational Agents. Another factor influencing the autonomy of negotiation of an Agent is the saliency of the issues negotiated (see Delreux 2011;de Ruiter, 2013;De Wilde 2009;Saalfeld 2005). This is because domestic parliaments want to control their executive for issues regarded as important for them or their electorate. However, this is not always the case. Randour (2018), studying the negotiations leading to the adoption of the 2014 CO2 cars regulation (Regulation 333/2014), finds that the saliency of an issue can as well decrease the influence of parliaments on their domestic executive. If there is a lot at stake in a negotiation, coupled with hard bargaining between Member States, domestic parliaments may prefer not to intervene in the negotiation to let their executive negotiate the best possible deal at the European level.
The last set of variables discussed in this paper is situated on the Agent's side. While these variables are not directly experienced by the Principal, they can also have an effect on the Principal-Agent relationship. For example, Delreux (2011) analysed the role of the European Commission in international environmental negotiations and found that the compellingness of the external environmentunderstood as the political pressure on the Agent not to jeopardise the negotiationsimpacted on an Agent's discretion. When looking at the parliament-executive relationship, such a situation could happen, for example, in the case of the negotiation of a European Treaty or of a trade agreement. Finally, another external variable to the Principal-Agent relationship that could be investigated is the impact of (visible) conflicts within EU institutions (i.e. the EU Council and the Council of the EU) and across EU institutions (i.e. the Commission, the European Parliament and the Council). In such cases, one could ask: 'can national parliaments help ministers to gain influence in Council negotiations?' (Rozenberg and Hefftler 2015, 26). Indeed, domestic parliaments may also decide to signal their position via the use of resolutions (Hoerner 2017)to express their support for their executive negotiation strategy.
Research methods: fostering the use of small to medium N research design To study the influence of domestic parliaments in EU affairs using the Principal-Agent model, the research note advocates the use of 'small to medium N' (comparative) analyses, as studying the influence of parliaments in EU affairs requires to have an indepth knowledge of the cases. In particular, the use of process tracing, (comparative) case studies or qualitative comparative analysis (QCA) should be encouraged in the future, even though the degree of external validity of these studies is lower compared with 'large N' studies.
Yet, while these research methods come with advantages (i.e. it focuses on a small number of cases and makes it possible to cover almost all explanatory factors, the internal validity is also high), it also suffers from several shortcomings (i.e. the external validity is lower) (Dür 2008). That being said, I agree with Reykers and Beach (2017:, 275) when they explain that 'a more explicit focus on tracing causal mechanisms would allow Principal-Agent analysts to open up the black box of causality', and consequently, would also help in facing the challenge of observational equivalence. Another promising research method would be to use QCA, as it allows for systematic cross-case comparison while simultaneously doing justice to the case complexity (Rihoux and Ragin 2009).
Actually, a promising way to address the question of the influence of parliaments on EU affairs would be to foster the use of mixed method designs, combining for example, 'medium N' QCA analyses (i.e. to analyse the combination of causal relations leading to influence) with 'small N' qualitative research, and in particular process-tracing. However, it has to be acknowledged that these methodsalso called 'high tariff methods'imply an important investment of the researcher as it necessitates an indepth knowledge of the domestic parliamentary scrutiny, the domestic executive behaviour during EU negotiations as well as the main dynamics of the EU policy-making.

Data availability: what can be used to study the influence of parliaments in EU affairs
What kind of data could be used to study the influence of domestic parliaments? In addition to the information available on the websites of (sub)national parliaments (i.e. parliamentary questions, plenary debates, committee reports, (de)briefing of executives), there exist nowadays several sources of information that could help to (better) assess executive behaviour during EU negotiations and whether the government followed parliaments' wishes. Regarding the executive side, in addition to the output of EU negotiations, scholars could analyse the voting behaviour and policy statements of governments in the EU Council (see. Van Gruisen and Crombez 2019 and of Hagemann, Bailer, and Herzog 2019); fully exploit the opportunity of the 2001 transparency regulation (Regulation (EC) 1049/2001) to access limité (but publicly available) documents of the Council such as 'comments from Member States' or 'Presidency compromise' as well as documents originating from the European Commission and of the European Parliament (i.e. committee reports, written explanation of votes made by MEPs etc.); and finally, the press releases and press points of executive actors before or after Eurogroup or European Council meetings.
Besides these various sources of information, scholars should also collect first-hand information by conducting (case-specific) semi-structured interviews. This could be done in a two-step process, as it is often difficult to identify the key actors of a negotiation. First, I would suggest starting to make interviews with 'resource persons' who witnessed the negotiations but were not directly involved (e.g. representation of parliaments in Brussels and officials acting in Monday Morning Meetings; parliamentary clerks in (sub)national parliaments) and actors who are outside of the parliament-executive relationship. Starting with interviews of actors who were not directly involved in the negotiations or who were outside of the parliament-executive relationship makes it possible to apply the 'referred' sampling technique by asking the interviewees to suggest resource people whom can be interviewed (especially in the Council of the EU, in Permanent Representations or in the European Commission). It also allows one to develop a fine-grained knowledge of the context and of the key moments in the policy-making processes analysed.
As an alternative/complement to this first step, and in order to identify influential players and interesting case study, scholars could also rely on the 'attributed influence method'. It consists on the assessment of experts and on peers-evaluation to identify who is considered to be an influential parliamentarian. Starting from the expertise of MPs allows (1) to go beyond the analysis of 'visible' activity of parliamentarians and also include less visible parliamentary work (i.e. informal strategies of MPs) and (2) to link the measurement of formal strength and activities of MPs with their perceived influence (i.e. is there a relation between being an active scrutiniser and the perception of being an influential player?). While this approach has the advantages to be relatively simple and to focus on a larger number of cases, it also faces serious shortcomings: (1) the answer rate of such survey is generally really low (see. Bailer 2014); (2) the measure is based on subjective evaluation of influence rather than actual influence; and (3) there is also potential problems of rater bias (e.g. between members of the same party, governing coalition or gender). One way to increase the validity of such an approach would be to control for rater effect, to design a short and to the point survey (i.e. to increase the answer rate) but also to run in parallel a survey among civil servant (e.g. clerks of parliaments, officials participating in the Monday Morning Meetings) and executive actors aiming at collecting their (subjective) evaluation of parliament's influence. The triangulation of the collected data based on MPs, civil servants and of the national executives' subjective evaluations should allow to produce more reliable results than just contacting MPs.
The second step would then consist in meeting individuals from the Agent's and the Principal's side. Relying on the previously collected information, these interviews could be really fruitful as the interviewer, knowing the case, can concentrate on the main questions he/she is interested in. The combination of these various sources of information has an important potential to unpack executives' and parliaments' behaviours in EU affairs.

Conclusion
Future research needs to (continue) conceptualising and considering more systematically the relationship between the parliament and their executive in order to assess if and how parliamentary scrutiny translates into parliamentary influence. To this end, this research note advocates reconnecting empirical works focusing on national parliaments with the Principal-Agent modeland in particular to the recent developments of the approach in EU studies. In addition, by favouring 'small to medium N' research, it becomes possible to trace back the conditions and causal mechanisms allowing parliaments to have an influence on the EU policy of their government and on EU policy-making more generally. It also makes it possible to include both formal and informal scrutiny strategies and to make justice to the complexity of the cases, hence dealing to some extent with the observational equivalence challenge and the question of data availability. Note 1. The remainder of this paragraph is mainly dealing with how domestic Agent's reactions can be operationalised vis-à-vis their domestic Principal. Therefore, it only tackles indirectly whether the proposed procedure also applies for supranational Agent. Indeed, in the case of supranational Agent(s), the reader must bear in mind that there is no exclusive delegation between the domestic Principal (i.e., domestic parliaments) and the supranational Agent, as the latter are not directly accountable to domestic parliaments. While some studies already focus on the relation between national parliaments and the European Commissionthanks to the EWM and political dialogue mechanism and, on occasional basis, hearings of Commissioners in domestic parliaments -, less is known on other types of supranational Agents, such as EU agencies. Further research should be conducted on this matter.