Parliament’s (lack of) agenda-setting power over the executive decision agenda: evidence from Belgium, France and Portugal

Abstract One of the functions of parliamentary questions in modern legislatures is to pressure executives to pay attention to specific issues. But can these questions effectively influence executive decisions? There is surprisingly little empirical research in this area. Adopting an agenda-setting perspective, this article examines the extent to which issue attention in oral parliamentary questions influences the issues addressed in the weekly meetings of the Council of Ministers in three countries (Belgium, France and Portugal). Our findings suggest that the agenda-setting power of parliaments vis-à-vis the executive is usually weak in the contexts studied here. In Belgium, we find evidence that the executive does pick up on issues debated in parliament but that the media seems to play a crucial role in focusing attention. These conclusions testify to the dominance of the executive power in many Western democracies. The findings also demonstrate that agenda-setting patterns are more complex than single-country studies often suggest, and that comparative research is the way forward.


INTRODUCTION
To what extent do legislatures influence the agenda of the executive? This is an important question for political scientists to answer as modern parliamentary democracies revolve around the confidence relationship between the legislative and executive branch. In theory, parliaments should be able to attract executives' attention to pressing issues, and to hold them accountable for their decisions. The various instruments they have at their disposal to do so range from simple recommendations to the more drastic vote of no-confidence, which can lead to the dismissal of the cabinet. But are these parliamentary tools effective in influencing executive agendas in practice?
This paper tackles the matter by focusing on one specific instrument: oral parliamentary questioning. In many countries, parliaments organize debate sessions where members of parliament (MPs) can submit questions and elicit an oral response from the government (Beyme, 2000). The way in which oral questions are used varies considerably. Firstly, procedural details can be very different from one country to another (Russo & Wiberg, 2010). Moreover, legislative scholars have pointed out that even when regulated by similar procedures, parliamentary questions are used to fulfil different functions (Wiberg, 1995, p. 181). They are mostly seen as oversight tools. Parties, notably when in opposition, use questions to control the conduct of government, i.e. to force ministers to justify their actions or disclose relevant details. 1 MPs may also ask locally oriented questions as a constituency service (see e.g. Borghetto, Santana-Pereira, & Freire, 2020;Martin, 2011;Russo, 2011) or use questions to gain publicity to advance their career, especially when there is significant media attention (Bailer, 2011). Recent research linking the increase in non-legislative activity to the growing importance of issue-competition in Western Europe has further argued that questions are widely used to challenge the government and press ministers to pay attention to a specific issue (Green-Pedersen, 2010).
In this paper, we focus on this latter function: the use of parliamentary questioning to influence the government agenda. Although the scholarly literature has started to focus on the agenda-setting potential of parliamentary questions, large-N research in this area is still lacking. This paper aims to fill this gap by taking an agenda-setting perspective (Baumgartner, Breunig, & Grossman, 2019;Green-Pedersen & Walgrave, 2014;. We focus on issue attention and test whether and to what extent parliamentary attention for specific issues affects the share of executive decisions on those same issues. A number of factors may affect the way in which the agenda-setting function is performed. This study addresses two key moderators: whether the party of the questioner is in the majority or the opposition, and the role of media. First, different incentives drive majority and opposition MPs to submit questions (Borghetto & Chaques-Bonafont, 2019;Vliegenthart & Walgrave, 2011). Questions sponsored by opposition parties generally attack the executive and attempt to influence an otherwise inaccessible cabinet agenda. In contrast, MPs from the majority have different channels at their disposal (e.g. direct contact with the cabinet through their party) to bring issues to the executive's attention. That is why majority-sponsored questions are often employed as credit-claiming devices rather than to put the government under pressure. This will be taken into account in our expectations and analysis.
Secondly, we test the extent to which media attention matters for the day-to-day agenda-setting relationship between legislative and executive. In particular, we expect the impact of parliamentary questions (by majority and opposition MPs alike) on the ministerial council decision agenda to be greater when the issue is also the focus of media attention. Constantly confronted with an extremely complex flow of problems that simultaneously call for executive attention, ministers in a mediatized environment are likely to use media coverage as a cue to decide which issues should be prioritized and discussed in their weekly meetings (Thesen, 2014).
The remainder of the paper is organized as follows. We start by presenting the existing literature on agenda-setting and its contributions to our understanding of the relationship between legislative and executive institutions, from which we derive specific expectations. We then explain how we tracked policy attention across media, executive and parliament and highlight some procedural differences between the countries under study. Finally, we carry out a time-series cross-sectional analysis and discuss the findings while stressing their implications from a comparative perspective.

THEORY
This study builds on a long tradition of research striving to understand the dynamics of issue attention on different political agendas. Indeed, building on the work of Baumgartner and Jones in the U.S. (2005,1993), scholars from a range of countries have studied the 'politics of attention' , in an attempt to shed light on how attention travels between various political, societal and media actors (see also Baumgartner et al., 2019;Green-Pedersen & Walgrave, 2014).
To what extent do parliamentary questions affect executive agendas? To our knowledge, this relationship has seldom been the focus of empirical research; this is surprising given the general agreement that agenda-setting is one of the main functions of parliamentary questioning (Green-Pedersen, 2010;Wiberg, 1995). Bartels (1996) explicitly looked at the relationship between Congress and the executive in the U.S. with regard to four issues that received attention throughout the 1990s. He shows that Congress exerted some influence over the executive, but only with regard to certain issues. Another study by Jensen, Proksch, and Slapin (2013) focused on the European Parliament, demonstrating that parliamentary questions about violations of EU law sometimes make the Commission take action against member states. This scarce evidence suggests that parliament can exert some agenda-setting influence on the executive. In this paper, we zoom in on the matter and aim to get a more fine-grained understanding of the extent to which parliament successfully fulfils its agenda-setting function vis-à-vis the executive.
The differential strategy of majority and opposition MPs on parliamentary questioning is a crucial aspect to address. Scholars have emphasized the association between parliamentary questions and the goals of opposition parties. The instrument of questioning, which was originally designed for parliament (the principal) to exert control over the government (the agent) (Saalfeld, 2000), corresponds well with the opposition parties' goal of demonstrating the incompetence of the government. Indeed, Bevan and John (2016) show how the opposition sets the tone during question time in the United Kingdom: questions from the opposition force the party leadership from the government to address issues, more than the other way around. In Denmark, the opposition's structural criticism on the government's lack of attention for the environment had some impact on government decisions (Seeberg, 2016). More generally, a longitudinal study of parliamentary questions in Denmark revealed that questions are essentially a tool at the disposal of opposition MPs. The opposition asks questions about issues that are advantageous to them, forcing MPs from government parties to address these matters (Green-Pedersen & Mortensen, 2010). Even though these studies do not directly measure the opposition's impact on ministers' (executive) attention, they suggest that the opposition may play an important and even crucial role in putting the government under pressure through parliamentary questioning.
The majority parties are expected to use parliamentary questioning differently since attacking their 'own' government ministers could threaten the stability of the coalition (De Winter & Dumont, 2006). This does not mean that they do not engage in this activity: MPs from majority parties in many countries ask just as many parliamentary questions as MPs from opposition parties. However, their motivation to do so is probably different. In their case, it might be more a matter of claiming credit for positive results or of trying to tackle negative evolutions themselves before the opposition has a chance to do so (Sevenans & Vliegenthart, 2016). 2 If we accept the assumption that majority parties have fewer incentives to use parliamentary questions as agenda-setting devices, the correlation between majority issue attention in parliament and executive issue attention during their weekly meeting should be weaker than in the case of opposition parties (Green-Pedersen & Mortensen, 2010).
We also take the mass media into account. It is now widely accepted that news coverage is one of the most important political agenda-setters , affecting not only symbolic parliamentary agendas (Walgrave & Van Aelst, 2006) but also more substantive agendas such as government decisions (see e.g. Bartels, 1996;Walgrave & Vliegenthart, 2012). This is relevant to our study in two ways. First, we need media as a control variable as it is possible that the parliament and the executive are not just influencing each other but simultaneously responding to external events and the news coverage about these events. Second, we look at media as a possible moderator of effects, since the potential of parliamentarians to push issues onto the governmental agenda may depend on the extent to which they can also generate media attention for these issues. Indeed, politicians often respond to issues simply because of their media attention (Sevenans, 2018).
We summarize our expectations in 3 hypotheses:

H1 (agenda-setting): After MPs ask more parliamentary questions about an issue, the ministerial council also pays more attention to this issue. H2 (majority vs. opposition): The agenda-setting effect of parliamentary questions on the ministerial council is stronger for questions posed by opposition MPs than for questions asked by MPs from the majority. H3 (media moderation): The agenda-setting effect of parliamentary questions (by majority and opposition
MPs) on the ministerial council is greater when the media also pay more attention to the issue.

DATA AND METHODS
The following analysis relies on data regarding the distribution of policy attention in parliament, cabinet and the media in three countries: Belgium, France and Portugal. While these three countries are all 'old' , well-established European democracies, they differ in a number of ways that may affect legislative-executive relations. We see no immediate reason for parliament's agenda-setting power over the executive to differ between classic parliamentary systems (like Belgium) and semi-presidential systems (like France and Portugal). Arguably, it is the format of the governing coalitions that is more important. For the last three decades Belgium and France have been ruled by coalition governments; these are often more fragmented in Belgium (Deschouwer, 2012) than in France, which is usually led either by a conservative (Gaullist-led) coalition or a left-wing (Socialist-led) coalition (Elgie & Grossman, 2016). Even when one party holds an absolute majority, as in the case of France since 2017, it usually chooses to honor pre-electoral agreements and to set up a coalition government. Post-1974 Portugal has mostly seen the alternation in power of two parties, the socialists (PS) and the social-democrats (PSD, typically in a coalition with the Christian democrats of the CDS-PP). On the one hand, we know that coalition systems, as a means for coalition partners to keep tabs on each other's ministers, often produce stronger parliaments and committees (André, Depauw, & Martin, 2016). On the other hand, it seems plausible that cabinets with fewer parties (i.e. actors with veto power) as in Portugal can be quicker to react to external events and adapt their agenda (Tsebelis, 1995). Other possibly relevant differences include, for instance, the size of the majority (which also varies over time), the make-up and size of the government. Additionally, there are some procedural differences (see discussion below) and political actors in the three countries may take a different role vis-à-vis the media, which is relevant for our third hypothesis (see e.g. Vliegenthart et al., 2016). Unfortunately, our three-country design does not offer the necessary analytical leverage to pinpoint the exact causes of any distinctions we may (and will) find between countries, which is why we treat our comparative setup as a means to test the robustness of our results.

Parliamentary, executive and media agenda
Tracking policy attention over time in the three institutions (parliament, executive and media) entailed the topic coding of relevant comparable documents in the three countries using the methodology and codebook developed by the Comparative Agendas Project (CAP, www. comparativeagendas.net). The present analysis relies on datasets collected and coded by the Belgian, French 3 and Portuguese 4 CAP teams. Due to differences in data availability, the datasets cover different time periods: respectively from 1999 to 2008 for Belgium, from 1993 to 2013 for France and from 2007 to 2015 for Portugal. We rely on three distinct datasets, coded into 21 major topics 5 (see Figure 1 for the list of topics) 6 : • each front-page article of a major national newspaper to reconstruct the mass media agenda ( We chose oral and not written questions to track issue attention in parliament because we are interested in party issue competition strategies. Written questions usually give individual MPs some leeway to pursue individualized agendas, there is no ceiling to their number and they are not typically covered by the mainstream media, thus making them less suited for this project. Oral parliamentary questions in plenary sitting are subject to time constraints, namely a few hours either every week or every two weeks, and are more politically salient since they imply the presence of ministers in parliament and greater media coverage. Clearly, we do not assume that every oral question is meant to attract executive attention, as some questions are merely asked to obtain information. Even if only some questions manage to attract executive attention, this should be visible in our analysis. The coded documents in the three countries are largely comparable, but it is useful to point out some procedural differences across our cases. With regards to oral parliamentary questions, Belgian MPs may ask questions to the government (Prime Minister and ministers) at least once a week (Rule 124 of the Standing Orders), and they must notify the subject of the question one day in advance. In France (art. 133 of the internal rules) the "questions to the government" 7 (Prime Minister and ministers) are also held on a weekly basis (art. 48 of the 1958 Constitution), and do not have to be sent in advance. The Portuguese Rules of Procedures provide for regular sessions of oral questions (with the Prime Minister) every other week 8 . Unlike Belgian and French questions, some of the Portuguese questions are not really initiated by MPs; instead, question time starts with the Prime Minister delivering a speech, followed by questions relating to the topics raised in the speech. In addition, just as in Belgium and France, MPs ask their own questions followed by the PM's impromptu answers. To conclude, there are differences in the timing of plenary sessions (weekly vs. two-weekly), the target (all ministers vs. only Prime Minister) and the initiation of questions (submitted in advance vs. on the spot vs. partly reactive) that we take into account when analyzing the data and interpreting our results.
The press releases of Council of Ministers are more comparable: they appear weekly and report all decisions communicated to the media by the Prime Minister and ministers at the end of their meeting (Borghetto & Belchior, 2020). As the minutes of the meetings are not available, these are the best documents to track cabinet attention. They differ in length and format in the three countries but have the same status and include at least all bill proposals, ministerial decrees, administrative appointments and regulations, as well as foreign policy decisions.

Dependent and independent variables
To compute attention for issue i in week t we decided to use the share of questions, the share of executive decisions and the share of newspaper articles falling into each of the 21 major topics in that week. In order to test H2, oral parliamentary questions were divided according to the position of the questioner's party in either the majority or the opposition and we averaged issue attention for parties belonging to the two camps. Table 1 reports the overall share of questions in the two countries. It shows that majority-sponsored questions are a minority in Portugal (16% of the total), but they make up almost half of questions in Belgium (42%) and an even higher proportion in France (62%). This is the result of rules (e.g. in Belgium, each party has a fixed maximum number of questions it can ask every week) and the format of the government coalition. Majority questions make little sense in Portugal's single-party governments.  These results alert us to two relevant aspects of oral questioning. First, there is a permanent back and forth between majority and opposition; and it responds to current events, which means that the two groups will often talk about the same issues. Second, Portuguese question time tends to deal with more general questions (e.g. about the state of the economy, macroeconomics, or the functioning of the administrative apparatus, government operations) than its equivalents. This is due to the fact that only the Prime Minister participates in Portuguese question time, which limits the degree of technicality of questions (he/she is likely to defer to the Transport ministers if asked, for example, about the state of the railway service). As far as the agenda of the Council of Ministers is concerned, unsurprisingly "government operations" is prominent in all countries. The remaining topics rank differently in the three cases, although attention is quite fairly distributed among them. Before moving on to testing our assumptions, we offer some further insight into the data by elaborating briefly on the diversity of topics addressed in any given week. Agendasetting research has taught us that agendas are characterized by differing levels of friction:

Continue
institutional and cognitive limits will restrict the responsiveness of agendas to outside stimuli.
Friction creates a bottleneck of attention, limiting the number of problems dealt with at any given time . Those limits may be more or less important depending on the agenda. While some agendas may be rather "open" and thus highly responsive to changes occurring in the real world, with opportunities to address multiple topics simultaneously, others may be much more restricted. For our purposes, we can expect media agendas to be characterized by lower levels of friction than the political agendas.
The political and, in particular, the government agenda may be understood as a huge filtering device. It may achieve a great many things through parallel processing, but studies show that government officials are subject to bounded rationality, preventing them from reacting proportionally to changing circumstances (Jones, Larsen-Price, & Wilkerson, 2009, p. 282).
Source: Elaborated by the authors.
Before moving on to testing our assumptions, we offer some further insight into the data by elaborating briefly on the diversity of topics addressed in any given week. Agenda-setting research has taught us that agendas are characterized by differing levels of friction: institutional and cognitive limits will restrict the responsiveness of agendas to outside stimuli. Friction creates a bottleneck of attention, limiting the number of problems dealt with at any given time . Those limits may be more or less important depending on the agenda. While some agendas may be rather "open" and thus highly responsive to changes occurring in the real world, with opportunities to address multiple topics simultaneously, others may be much more restricted. For our purposes, we can expect media agendas to be characterized by lower levels of friction than the political agendas.
The political and, in particular, the government agenda may be understood as a huge filtering device. It may achieve a great many things through parallel processing, but studies show that government officials are subject to bounded rationality, preventing them from reacting proportionally to changing circumstances (Jones, Larsen-Price, & Wilkerson, 2009, p. 282).  Source: Elaborated by the authors. Figure 2 presents the relative distribution (density plots) of attention for each agenda. To be coherent with the rest of the analysis below, we split the parliamentary questions into majority and opposition questions. The plots show the relative share of attention by topic by week. For the purpose of comparison with the actual distribution, we also plot the normal distribution (plain black lines): insofar as changes in the real world are random, a frictionless distribution should be normal. The more skewed the distribution, the more the agenda should exhibit signs of friction.
As expected, the media agenda features the most "normal" distribution in all three countries. Unsurprisingly, newspapers are specialists of parallel processing and try to keep up with changes in the real world. While this does not mean that news is not subject to friction (cf. Boydstun, 2013), there is certainly less friction than for parliamentary questions or the executive agenda.
The order of the two other agendas differs in the three countries. Although this is partially due to the lower number of observations in the other agendas, the differences are still interesting. In France there is a much higher share of issues on the parliamentary agenda that receive no weekly attention and very few issues receiving exceptionally high levels of attention. The executive agenda features a strongly "leptokurtic" distribution with a very high zero peak and fat tails (cf. . This is different for Belgium and Portugal where the parliamentary questions are more skewed than the executive agenda.

Model
For the purpose of the following multivariate analyses and in line with our three hypotheses, we analyze the effect of the majority (Majority PQ(lag)) and opposition (Opposition PQ(lag)) parliamentary agendas on the executive agendas (Executive) in the three countries, controlling for and in interaction with the media agenda (Media(lag)). By including all three independent variables simultaneously in the models, we control for many of the confounding variables that may affect both party agendas. Since the number of documents varies across weeks, we measured issue attention as the proportion of documents dedicated to a specific issue out of the total in that week. We use a weak understanding of causation as "temporal precedence". In other words, when we test the "effect" of the opposition agenda on the executive agenda, we refer, ceteris paribus, to the correlation between the share of attention (across all topics) in opposition parliamentary questions in week t -1 and executive decisions in week t. 9 The following analyses rely on time-series models with fixed effects on the level of issues. 10 Datasets were assembled in panel-data form, in which panels are the 21 major topics and time is the week when the cabinet met. To account for autocorrelation in the dependent variable, we also included the share of executive issue emphasis in the previous week (Executive(lag)). Our model specifications are summarized by the following equations:

Executive = Media(lag) + Majority PQ(lag) + Opposition PQ(lag) + Executive(lag) + major topics Executive = Media(lag) + Media(lag) * Majority PQ(lag) + Majority PQ(lag) + Opposition PQ(lag) + Executive(lag) + major topics Executive = Media(lag) + Media(lag)* Opposition PQ(lag) + Opposition PQ(lag) + Majority PQ(lag) + Executive(lag) + major topics
9 As often seen in agenda-setting research (see e.g. Vliegenthart & Walgrave, 2011) to avoid excessive zeros, weeks without executive decisions (e.g. during ministerial recess) are left out of the dataset, resulting in a 'shortened year' . If no question time debate occurred in t-1, we decided to use data from the most recent debate up to a month away from the executive meeting. 10 Hausman tests recommend the use of fixed over random issue effects. Tables 2 and Table 3 present the results for the stacked and country-specific models respectively, testing the three hypotheses on the effect of majority and opposition issue attention on the executive decision agenda. Whereas the stacked model, which includes country dummies, serves to test our hypotheses in general, the separate models test whether the results are robust across countries. Model 1 in each table reports the results from the model containing only main effects, which allow H1 (agenda-setting) and H2 (majority vs. opposition) to be tested, while models 2 and 3 introduce interaction effects between media and, respectively, majority-and opposition-sponsored oral questions (always lagged version) to test H3 (media moderation). Before discussing results, we note that, in general, executive attention is strongly path dependent as indicated by the positive and statistically significant coefficients for its corresponding lags.

RESULTS AND DISCUSSION
H1, proposing the existence of an agenda-setting effect of parliamentary questioning on the executive agenda, and H2, focusing on the influence of question time by opposition parties (as compared to majority parties), are not validated by our findings. Coefficients for Majority PQ (lag) and Opposition PQ (lag) in Table 2, model 1, are not statistically significant at conventional levels. This finding reveals that simply addressing the cabinet on the floor normally has no direct effect on government reaction. Neither majority nor opposition questions exert the expected agenda-setting effects. The result holds in each of the three countries (see Table 3) and is, to a large extent, confirmed if we run the model using different lag specifications (t -2 and t -3 see Annexes B and C). Parliamentary questions appear to be a surprisingly unsuccessful instrument to attract the Council of Ministers' attention to issues. This does not of course mean that they cannot exert any pressure on the government to pay attention to these issues. For example, governments that feel urged to respond to parliamentary attacks may do so by means of general declarations (e.g. in the press), rather than through making rapid decisions. Indeed, the cabinet agenda is usually set well in advance and a fair share of its content is devoted to administrative decisions, so it may not always be very receptive to the introduction of new topics following parliamentary pressure. Alternatively, it may react more slowly and the time between a question and a typical government response may vary, making it hard to find such dynamics with the type of lagged analyses proposed here. 11 That said, it remains remarkable that despite using the conventional and well-established method to detect agenda-setting influences, there appears to be absolutely no immediate responsiveness from the ministerial council to parliament.
As an exception, we find that issues emphasized in parliamentary questions have a chance of making it on next weeks' Ministerial Council agenda in Belgium, but only when the media also decides to pay attention to the specific issue (confirming H3 for the Belgian case). This finding holds for majority and opposition questions alike. It is the questions in combination with the media that makes them compelling from an agenda-setting perspective, at least in Belgium. Ultimately, question time debates are mainly about current events, but not all events make it to the front page of major 11 As a robustness test, Annex B and Annex C present our models, but computed using a different time lag specification, respectively t -2 and t -3 . But for a few country-specific effects (the impact of Majority PQ (lag t -2 ) is positive and significant for Portugal and the impact of Majority PQ (lag t -3 ) is negative and significant for France), the remaining results are largely comparable and confirm that parliament agenda-setting is remarkably weak. newspapers. When they hit mass media and are picked up in nationally televised parliamentary debates, the pressure increases and it is difficult for the government to remain silent.
In the other two countries, however, this media moderation does not occur and it does not affect the agenda of the Ministerial Council even when media and parliament jointly focus on issues (rejecting H3 for France and Portugal). Indeed, the positive and significant coefficient of the interaction effect in the stacked analysis (model 3, Table 2) appears to be driven by the Belgian data. In France, the results are insignificant. In Portugal, contrary to expectations, the effect of the interaction between opposition questioning and media attention is even negative and significant.  We can only speculate on the reasons behind these diverging results. The difference between Belgium and France, we believe, has to do with political actors' sensitivity to mass media. A seven-country comparison showed that MPs in France are a notable exception in Western Europe in that they are not very responsive to the media when determining their issue priorities, and their initiatives seldom make it into the media . Moreover, we know that the media and parliamentary attention tend to reinforce each other when they jointly focus their attention on certain issues. Therefore, MPs are weak in terms of agenda-setting capacity when they are on their own, both in France and Belgium; however, in Belgium, they sometimes get the support of the media and therefore succeed in putting pressure on the executive. This amplification mechanism seldom occurs in France.
The Portuguese results are harder to explain. The negative and significant coefficient of the interaction effect suggests that, when media and opposition MPs jointly focus on an issue, the executive subsequently pays significantly less attention to that issue. One explanation might lie in the more reactive nature of oral questions in Portugal (compared to Belgium or France). As we explained above, a part of question time is devoted to questions that are asked in reaction to the Prime Minister's opening speech. These questions allow MPs to hold the government accountable for what it has done (ex-post control), rather than serving to attract the executive's attention to specific issues (ex-ante control through agenda-setting). If it is precisely this type of question that is often picked up by the media, it seems logical that the executive responds less: the executive debate on the topics underlying these questions has often just been closed. Alternatively, it might be that the more general nature of questions in Portugal (compared to the issue-specific nature of French and Belgian questions), which stems from the fact that only the Prime Minister is present during question time, makes them less well-suited for an immediate executive response.

CONCLUSION
This study set out to cast light on the relevant but, to date, relatively underexplored relationship between parliamentary control instruments and the conduct of government by focusing on the agenda-setting function of parliamentary questions. Our analysis aimed to test the extent to which parliamentary questions are successful in attracting ministerial attention to issues, and in which circumstances. We did this through a large-N research design studying the effect of issue attention in oral parliamentary questions on the decision agenda of the Council of Ministers in three countries, Belgium, France and Portugal.
Our findings indicate that the effectiveness of oral parliamentary questions on the floor as agenda-setting tool is weak in the three countries under study here. There is no evidence of a direct transmission of issue attention from the parliamentary to the executive arena. We suggest that the most convincing explanation for this finding is that our analysis is limited to short term cabinet reactions and, unless the circumstances make decisions unavoidable (e.g. natural disaster or international crisis), cabinets probably prefer to first release a statement and then choose the best timing to deliver a decision. Nevertheless, this finding is confirmed across the three countries under study and we believe it is relevant: as the executive postpones its reaction, attention probably fades away so that it becomes easier for ministers to refrain from tackling problems at all. The fact that parliament cannot really enforce quick executive decisions is, therefore, another indication of the strong dominance of the executive power in many Western democracies.
The situation may be different in instances where there is a simultaneous presence of media attention on the same issues. In Belgium, the combination of media and parliamentary attention on issues can spark executive decision-making. This result speaks directly to previous research on the close link between parliamentary questions and mass media (e.g. Vliegenthart et al., 2016). The observation that "for government, there is no equivalent to the legal right of silence" (Norton, 1993, p. 112) holds especially true when the issue is newsworthy and politically relevant. For a number of possible reasons that cannot be disentangled here, however, media fail to play such an amplifying agenda-setting role in France or Portugal.
Overall, the present analysis is but a first step in the pursuit of a greater understanding of the agenda-setting function of parliamentary questions. In particular, our study suggests (but does not theorize or test) that levels of transmission may be structured and constrained by institutional or historical forms of friction specific to each agenda and each country. A substantial literature dwelled on the importance of institutional design for legislative behavior (André et al., 2016;Wiberg, 1995).
Differences in government constellation (e.g. single-party vs. multi-party governments), the strength of parliament, and specificities of the media system all affect the incentive structures faced by political actors functioning in these institutions. And this is likely to bear implications for the degree to which parliament can exert agenda-setting pressure on the executive.
Our small n prevented us from exploring these institutional differences ourselves, which is an obvious limitation of the study. In this article, we started from the generic expectation that, irrespective of institutional differences, parliaments should have some agenda-setting impact on executives. Yet what we found is that this impact, if any, is strongly context-specific, which warrants caution when generalizing the results from our study. Covering a larger number of countries would be desirable in future studies, notably to examine effects of specific institutional setups on the agenda-setting capacity of opposition MPs and the media.