Introduction
The empirical literature on the congruence of the public’s policy preferences and those of political actors has brought out the general assumption that elected politicians are expected to be responsive to (changing) public preferences due to the threat of electoral sanctions (Hobolt and Klemmensen, Reference Hobolt and Klemmensen2005). Several studies have confirmed that policy adjusts over time according to (changing) public demands, both through the mechanism of elections and the rational anticipation of decisionmakers (Page and Shapiro, Reference Page and Shapiro1983, Reference Page and Shapiro1992; Stimson et al., Reference Stimson, Mackuen and Erikson1995; Wlezien, Reference Wlezien1996; Erikson et al., Reference Erikson, MacKuen and Stimson2002; Hakhverdian, Reference Hakhverdian2010; Bartle et al., Reference Bartle, Dellepiane-Avellaneda and Stimson2011). A parallel finding in this literature is the relevance of issue salience in the actual implementation of public preferences (Burstein, Reference Burstein2003). Normative democratic theory has also emphasized this parallel side to positional representation: political representatives should not only represent the citizens’ policy preferences, but also their issue priorities (Jones and Baumgartner, Reference Jones and Baumgartner2004).
The strand of research investigating the congruence between the public’s issue priorities and those of presidents, governments and parliaments, was defined by Bevan and Jennings (Reference Bevan and Jennings2014) as dynamic agenda representation. A general finding of this literature is an emphasis on the connection between the public’s policy priorities and those of their political representatives, as laid out in speeches or legislative activities (Jones and Baumgartner, Reference Jones and Baumgartner2004; Hobolt and Klemmensen, Reference Hobolt and Klemmensen2005, Reference Hobolt and Klemmensen2008; Jones et al., Reference Jones, Larsen-Price and Wilkerson2009; Bonafont and Palau, Reference Bonafont and Palau2011; Arnold and Franklin, Reference Arnold and Franklin2012; Lindeboom, Reference Lindeboom2012; Bevan and Jennings, Reference Bevan and Jennings2014).
Building on this literature, the present article investigates Italian dynamic agenda representation – the transmission of the public’s policy priorities to the Italian political system’s legislative priorities. The investigation involves multiple access points to the legislative arena, namely the parliament’s policy agendas – divided according to the majority and opposition parties – and those of the government. Biannual data on the Most Important Problem (MIP) from regular Eurobarometer polls are used to assess the public’s policy priorities. A new data set coding all the Italian bills according to the rules of the Comparative Agendas Project (CAP) is used to build the policy agendas of all members of parliament (MPs), of majority and opposition parties, and those of the government.
The major contribution of this article is investigating the connection between the Italian public’s priorities and the legislation introduced by MPs and the government across multiple policy areas; to the author’s knowledge, this is the first time the topic has been addressed, especially with the inclusion of original data on bills. Until now, few studies have investigated the opinion–policy nexus in Italy, mostly due to a lack of consistent public opinion surveys measuring Italian public preferences and priorities over a sufficiently long timespan. Recently, investigating ‘the relationship between the policy preferences of citizens, parties and governments, and electoral behaviour in Italy’, Bellucci and Pellegata (Reference Bellucci and Pellegata2017: 24) found that the Italian policy mood is positively related to electoral behaviour, with a congruence between public opinion and electoral shifts. Other works focussed on single policy areas like foreign policy or crime and justice (see, for instance, Putnam and Penniman, Reference Putnam and Penniman1977; Bellucci and Isernia, Reference Bellucci and Isernia1999; Isernia et al., Reference Isernia, Juhasz and Rattinger2002; Diamanti, Reference Diamanti2006; Isernia, Reference Isernia2008), finding that public opinion preferences tend to be associated with some form of government responsiveness. Memoli (Reference Memoli2013) analysed the responsiveness of the Italian political system, looking at trust in political institutions and government popularity; he found that citizens’ confidence in their political representatives has consistently declined in the last 30 years of Italian politics. Regarding the relationship with policy priorities, the only available study, that of Russo and Cavalieri (Reference Russo and Verzichelli2016), showed a strong correlation between public priorities concerning economic issues and Italian parties’ parliamentary questions. By adding Italy to several existing studies (involving, for instance, Canada, Denmark, Spain, the Netherlands, the United Kingdom, and the United States), this article is also noteworthy for comparative politics. It offers new findings on a parliamentary country in which executives have weak agenda-setting powers, a characteristic traditionally associated with a low degree of responsiveness (Wlezien and Soroka, Reference Wlezien and Soroka2007; Hobolt and Klemmensen, Reference Hobolt and Klemmensen2008).
The results from time series cross-sectional (TSCS) analyses across 10 policy issues and 20 semesters (2003–13) suggest a persistent link between the public agenda and prioritization of legislation by the Italian MPs, majority parties, and governments. The same does not hold true for the opposition, whose agenda is never related to public opinion.
The consequential effect of issue salience
A normative problem related to the representation process proposed by Jones and Baumgartner poses the following question: ‘How representative is a legislative action that matches the policy preferences of the public on a low priority issue but ignores high priority issues?’ (Reference Jones and Baumgartner2004: 2). Consistent with this, among the essential features of an ideal democracy, Robert Dahl listed the ‘[f]inal control of the agenda. The demos would have the exclusive opportunity to decide how (and if) its members chose which matters are to be placed on the agenda’ (Reference Dahl2006: 8–9). To function properly, a democratic political system needs not only to transform the desires of the electorate into policy outcomes, but also to represent its main policy priorities. Empirical findings support this argument showing that citizens are more satisfied with democracy when the political agenda of policymakers better represents their policy priorities (Reher, Reference Reher2015).
An extensive body of literature has focussed on the congruence between the policy preferences of political representatives and public opinion, that is, positional policy representation. This line of research has demonstrated a link between the public’s preferences and policymaking at the aggregate level (Page and Shapiro, Reference Page and Shapiro1983, Reference Page and Shapiro1992; Stimson et al., Reference Stimson, Mackuen and Erikson1995; Wlezien, Reference Wlezien1996; Erikson et al., Reference Erikson, MacKuen and Stimson2002; Hakhverdian, Reference Hakhverdian2010; Bartle et al., Reference Bartle, Dellepiane-Avellaneda and Stimson2011; Bellucci and Pellegata, Reference Bellucci and Pellegata2017). Moreover, comparable patterns of representation have been shown to hold across different national contexts (see Hobolt and Klemmensen, Reference Hobolt and Klemmensen2005, Reference Hobolt and Klemmensen2008; Soroka and Wlezien, Reference Soroka and Wlezien2010; Wlezien and Soroka, Reference Wlezien and Soroka2012). The general assumption emerging from this literature is that ‘elected politicians are expected to respond to public preferences due to the threat of electoral sanction’ (Hobolt and Klemmensen, Reference Hobolt and Klemmensen2005: 380). The implication is that policy adjusts over time in response to changes in public opinion, both through the mechanism of elections and policymakers’ rational anticipation. The only study of the Italian political system on government responsiveness to citizens’ policy preferences found support for the thesis that Italian governments take citizens’ preferences into consideration and tend to balance shifts in public opinion mood (Bellucci and Pellegatta, Reference Bellucci and Pellegata2017). This finding supports the idea of a thermostatic relationship (Soroka and Wlezien, Reference Soroka and Wlezien2010) between the Italian electorate and its political representatives.
Empirical evidence has also stressed that, for positional representation to occur, attention needs to be focussed on similar issues for the public and political institutions. For instance, Burstein’s (Reference Burstein2003) meta-analysis of the opinion–policy nexus found that policy is usually affected by opinion, but also, when salience is considered, the substantial effect of public opinion is more frequent (35). Consistently, the interplay between salience and preferences produces an increased effect on policymaking (Adams, Reference Adams2016). Plott (Reference Plott1991) summarized formal political theory contributions in the so-called fundamental equation of politics as
$$\matrix{ {{\rm Preferences}{\times}{\rm Institutions}\Rightarrow{\rm Outcomes}} \cr } $$
Here, preferences represent what individuals want in the aggregate (e.g. more spending on welfare), while institutions are the rules and practices that govern the process by which collective decisions are made. The equation implies that (political) outcomes depend both on voters’ preferences and the institutions aggregating and filtering them (Hinich and Munger, Reference Hinich and Munger1997: 17).
Based on the empirical and normative arguments presented above, the previous equation can be transformed in the following way:
$$\matrix{ {{\rm Preferences}{\times}{\rm Priorities}{\times}{\rm Institutions}\Rightarrow{\rm Outcomes}} \cr } $$
Here, priorities refer to the policy areas in which the public wants its political representatives to act, rather than registering the support for their action (Jennings and Wlezien, Reference Jennings and Wlezien2015). They interact both with preferences and institutions to produce specific outcomes, and a new principle emerges: if the priorities change, the outcomes can change, even if the institutions and preferences remain constant. This is an extension of Vliegenthart et al.’s argument that the allocation of decisionmakers’ attention ‘is determined by the interplay of preferences and information as well as institutions’ (Reference Vliegenthart, Walgrave, Baumgartner, Bevan, Breunig, Brouard, Bonafont, Grossman, Jennings, Mortensen, Palau, Sciarini and Tresch2013: 392). With issue attention being one of the main drivers of policymaking, the interaction between preferences and priorities is a determinant not only of attention allocation, but also policy outcomes. This creates an arena of competition between political parties in which they take different positions on the same issues or strategically emphasize different ones (Vliegenthart et al., Reference Vliegenthart, Walgrave, Baumgartner, Bevan, Breunig, Brouard, Bonafont, Grossman, Jennings, Mortensen, Palau, Sciarini and Tresch2013). This justifies an investigation of whether public opinion policy priorities influence the institutional policy agenda, with an effort to understand the agenda-setting power of Italian citizens in the legislative process.
Hypotheses
Dynamic agenda representation is a complementary approach to the study of citizens’ positional representation that seeks to capture the relevance of salience in the policy process (Bevan and Jennings, Reference Bevan and Jennings2014). This approach to the study of responsiveness takes into consideration the general attention paid to issues (Jones and Baumgartner, Reference Jones and Baumgartner2004; Bonafont and Palau, Reference Bonafont and Palau2011; Lindeboom, Reference Lindeboom2012; Bevan and Jennings, Reference Bevan and Jennings2014), or what has also been described as rhetorical responsiveness (see Cohen, Reference Cohen1999; Hobolt and Klemmensen, Reference Hobolt and Klemmensen2005, Reference Hobolt and Klemmensen2008). Citizens, MPs, and governments face a vast range of policy issues and limited resources with which to process them (Jones and Baumgartner, Reference Jones and Baumgartner2005). The funnelling of social problems into political agendas has become more complex due to the relative stability of political actors’ capacity to attend to them and the significant expansion of the set of problems needing to be addressed in the last decades (Baumgartner and Jones, Reference Baumgartner and Jones2015). This growing imbalance has further emphasized the already substantive and consequential role the agenda-setting stage plays in the policy process.
Given that prioritization is a necessary step when resources are limited (Vliegenthart and Walgrave, Reference Vliegenthart and Walgrave2011), it seems fair to assume that political actors will strategically prioritize issues that can ensure them the best returns in terms of votes or policy outputs (Bertelli and John, Reference Bertelli and John2013). Brunner (Reference Brunner2012) argued that if policymakers are expected to respond to public policy preferences due to the threat of electoral sanction, then it seems fair to assume that they are also expected to represent public policy priorities. A further implication is that the policy agenda of MPs and the government will adjust to changes in public opinion over time, both indirectly, through the mechanism of elections, and directly, through policymakers’ rational anticipation. The model of dynamic agenda representation combines the promissory and anticipatory models of representation described by Mansbridge (Reference Mansbridge2003). On the one hand, voters delegate a mandate to governments that reflects their priorities (McDonald et al., Reference McDonald, Budge and Pennings2004; Pennings, Reference Pennings2005), either according to the policy priorities put forward by parties in their manifestos or during electoral campaign activities. On the other, the anticipatory model focusses on the incentives for political actors to please the voter in the next election with ‘anticipated reactions’ (Mansbridge, Reference Mansbridge2003: 520) due to their ‘fear’ of electoral punishment (Fiorina, Reference Fiorina1981). Concerning the electoral mandate relationship in Italy, Borghetto et al. (Reference Borghetto and Carammia2014), Borghetto and Carammia (Reference Borghetto and Giuliani2015), and Marangoni (Reference Marangoni2013) have emphasized the increased relationship between manifesto pledges and governments’ legislative output or prime ministers’ programmatic speeches in the Second Republic. On the second point, Bellucci (Reference Bellucci2012) has underlined the growing importance of valence politics, with judgements on performance and the economy being a decisive factor in voters’ choices, together with their long-term political predispositions. These findings justify the adoption of the dynamic agenda representation framework for an analysis of the Italian case.
Assuming that parties follow a rational course of action (or at least tend to do so) when pursuing any of their objectives, whether in terms of policy, office, or votes (Müller and Strøm, Reference Müller and Strøm1999), it makes more sense for them to take a position on the issues prioritized by voters. While parliamentary and government policy (and therefore legislative) activities do not necessarily have to signal policy positions, they can be used primarily to signal to voters that their concerns are being taken seriously and the issues they care about are being dealt with (Brunner, Reference Brunner2012). For politicians, a simple heuristic process to please voters between elections consists of following public opinion policy priorities and their relative changes when structuring their agenda. Indeed, public opinion polls represent an important source of information and drive politicians’ attention. The general hypothesis stemming from this reasoning related to the agenda space given to issues in public and political institutions can be outlined as follows:
Hypothesis 1 The public’s issue priorities are represented in the policy priorities of policymakers (MPs and the government).
A second hypothesis relates the governing status of parties to the size of the effect of public opinion. The expectation is that the opposition will better represent public priorities compared with majority coalition parties and the government. The reason for this is twofold: on the one hand, opposition parties face far fewer institutional and political constraints compared with the government and the parties supporting it. The legislative agendas of the majority – and even more so, the government – are driven by a great number of routine activities that need to be carried out each year (like defining budgets). On the other hand, the fact that all Italian cabinets are coalition governments, and considering that issue emphasis may have significant consequences for policy output, the distribution of attention is also determined by interparty agreements. Each party in the coalition will try to stress the issues it ‘owns’, that is, those for which it has a competitive advantage, thereby contributing to making the structure of government agenda more rigid. Given that it is not subject to these two formal and informal types of constraints, the opposition can perhaps afford to be relatively ‘irresponsible’ compared with the majority and the government, meaning that it can update its policy priorities more seamlessly following public mood. Furthermore, the way in which public opinion is measured, through the MIP, emphasizes critical issues. It is a survey question that highlights the problematic aspects of a political system, and thus, it lends itself to exploitation by the opposition, which can strategically prioritize issues that the executive and its parliamentary base may want to minimize or hinder (Vliegenthart et al., Reference Vliegenthart, Walgrave and Zicha2016: 295). Thus, the second hypothesis can be expressed as follows:
Hypothesis 2 Opposition parties better represent the priorities of the public compared with the government and majority parties.
Data and methods
Legislative agendas
To measure the Italian political actors’ policy agendas, legislative data are used; these comprise bills introduced by MPs and the government between 2003 and 2013. Legislative initiative is one of the main parliamentary activities, together with parliamentary questioning. In contrast to the latter, bills represent one of the most important ways in which the government and parties can effectively introduce, modify or terminate a public policy, proposing solutions to the problems that citizens deem important. The advantage of measuring decisionmakers’ policy agendas via their legislative initiatives is that legislation can be introduced almost seamlessly, with fewer institutional constraints (MPs or the government can present as many bills as they like), while parliamentary questions follow more restricted, formal rules on the allocation of time to MPs.
In Italy, MPs and ministers formulate almost all the laws that are presented and ultimately approved (Kreppel, Reference Kreppel2009). Probably the most striking feature of Italian legislative initiative is the huge number of bills proposed by the two main legislative actors – the parties and government – during each legislature. Previously, Di Palma (Reference Di Palma1977) noted that such behaviour results from the low constitutional barriers on legislative initiative and the political actors’ will to largely resort to these powers. Most of the bills proposed during the legislatures considered here (the second half of the 14th and all the 15th and 16th) were put forward by MPs. While the number of bills presented by executives has roughly remained constant over time, the share of MP proposals increased over the total period. Borghetto and Giuliani (Reference Borghetto, Visconti and Michieli2012) have shown that, of the huge number of bills proposed to the two chambers from 1987 to 2008 (more than 20,000), around one in three remained stuck at committee level without being explicitly aborted, and only around 7% were approved after the required readings in the two chambers. Beyond the filter that the legislative process applies to the total number of proposed bills, it is worth noting that most of the approved bills were initiated by governments, which showed a success rate of around 50%. In contrast, the major promoters of bills, MPs, had a very low success rate (around 15% of the total laws approved stemmed from parliamentary initiative). In any other European country, it is certainly an exception to the rule for private member bills to be more numerous than or even on the same level as government initiatives (Andeweg and Nijzink, Reference Andeweg and Nijzink1995; Capano and Giuliani, Reference Capano and Giuliani2001; Ström et al., Reference Ström, Müller and Bergman2003). This implies that in Italy, contrary to other European countries (Green-Pedersen, Reference Green-Pedersen2010), the parliament’s role has remained strongly focussed on legislative activities, which have maintained priority over the parliamentary control function. Moreover, the constant increase in the number of bills proposed, without any related growth in legislation approved, has been associated with a greater symbolic or propaganda value. Even if legislative actors are aware that most of the bills they propose will not bring about any real change, they think that it is worthwhile to introduce them anyway to represent local and microsectional interests (Di Palma, Reference Di Palma1977; Giuliani and Capano, Reference Giuliani and Capano2001; Giuliani, Reference Giuliani2005). As Brunner (Reference Brunner2012) argued, if party manifestos are intended as long-term instruments for signalling programmatic positions, then bills are one of the short-term tools available to MPs and governments to adjust their policy priorities between elections to better match those of the public.
In this study, the dependent variable is the share of bills presented per policy area for each semester before and after the instrumentation of the public opinion surveys under consideration in the building of the main explanatory variable. To test the policymakers’ representative role, four different legislative agendas are considered, as follows:
∙ The government agenda, made up of all bills introduced by ministers alone or collectively;
∙ the parliament agenda, including all bills introduced by MPs, regardless of their party affiliation;
∙ the majority agenda, including only bills where the first sponsor is an MP belonging to parties supporting the government; and
∙ the opposition agenda, gathering all bills where the first signer is an MP of an opposition party.Footnote 1
The coding of bills was conducted using a semi-automatic strategy matching TESEOFootnote 2 keywords, provided by the Senate of the Republic for each bill, with the CAP policy topics. The CAP is an international research network of scholars who developed ‘an infrastructure for the systematic comparisons of public policy outcomes, processes, and institutional relations’ (Baumgartner, Reference Baumgartner2016: 64). It is based on the classification of a variety of data, such as party manifestos, media, laws, parliamentary hearings, and budgets, into 19 major (and 240 micro-) policy topics to study agenda-setting and policy dynamics (Baumgartner et al., Reference Baumgartner, Jones and Wilkerson2011). Here, after coding each of the available TESEO keywords into Italian CAP micro-topics (Borghetto and Carammia, Reference Borghetto and Carammia2010), each bill was classified based on its TESEO tags.Footnote 3 Given that some keywords were too generic, and it was not possible to assign them to one or another policy category, instead of being removed, they were manually coded based on the bill’s title and preamble. If a bill featured keywords related to the same policy category more than once, the unique and distinct policy topics were kept and redundant ones removed when building policy agendas. For instance, if a bill had two keywords associated with agriculture and three with public lands and waters, it was considered to cover both topics once. This practice was adopted to guarantee more consistency in the data set, given that, rather than following a specific pattern, the number of TESEO keywords attached to bills is erratic.
Data on bills were merged with contextual information from the Italian Law-making ArchiveFootnote 4 (Borghetto et al., Reference Borghetto, Carammia and Zucchini2012). Given their nature, and in line with previous literature, bills ratifying international treaties were excluded from the analysis. To match the policy areas analysed by the Eurobarometer surveys, the micro-topics of the Italian CAP were used to create a time series of issues that were as consistent as possible with those of public opinion.Footnote 5
Public agenda: the MIP in Italy
The main independent variable measures public policy agenda, that is, ‘the set of policy issues to which the public attends’ (Jones and Baumgartner, Reference Jones and Baumgartner2004: 3). To measure this, the same strategy used in previous works was adopted (Jones and Baumgartner, Reference Jones and Baumgartner2004; Hobolt and Klemmensen, Reference Hobolt and Klemmensen2005, Reference Hobolt and Klemmensen2008; Bonafont and Palau, Reference Bonafont and Palau2011; Lindeboom, Reference Lindeboom2012; Bevan and Jennings, Reference Bevan and Jennings2014), that is, analysing the MIP opinion polls. The survey question asks respondents to name the two MIPs facing their country at that specific moment. As Wlezien (Reference Wlezien2005: 555) noted, this measure brings a series of problems and imperfections, by confusing at least two different characteristics of salience, namely ‘the importance of issues and the degree to which issues are a problem’. Still, as Wlezien (Reference Wlezien2005) also suggested, while scholars could and should develop better indicators in future studies, for studies of the past, we must rely on proxies, and the MIP has proved to be one of the best. In fact, while Jennings and Wlezien (Reference Jennings and Wlezien2015) argued that it is problematic to use MIP answers as policy preferences, as this downplays the link between citizens and governments, they also contended that they are a good proxy for public policy priorities – and this is the function they serve here.
Among the available sources of MIP polls, the EurobarometerFootnote 6 offered the possibility of constructing the longest time series for Italy. The question has consistently been asked roughly once per semester from the end of 2003 to the present in the regular Eurobarometer surveys. Time series with aggregated policy priorities for the Italian public on 10 policy issuesFootnote 7 were created based on the 20 surveys from the decade analysed.Footnote 8 Table 1 lists the policy topics selected, along with a summary of their statistics for both public opinion and each of the four dependent variables.
Table 1 Descriptive statistics of policy topics
Mean and standard deviation in parentheses. Author’s calculations based on Eurobarometer data.
Control variables
For each of the four dependent variables considered – parliamentary, majority, opposition, and government agendas – factors with the potential to influence their relationship with public opinion are considered. In models related to parliamentary initiative, Mattarellum is a binary variable controlling for the electoral system. It is coded as 1 for semesters leading up to the 2006 elections and 0 otherwise. Giuliani and Capano (Reference Giuliani and Capano2001) argued that one of the reasons for the increase in the number of bills proposed by MPs in the Italian parliament could be related to the change in the electoral system. The adoption of the so-called Mattarellum – an electoral system that established a direct link with the constituency for 75% of elected representatives – is considered as having reinforced the already extensive propensity of Italian MPs to propose bills. According to Giuliani and Capano (Reference Giuliani and Capano2001), if an MP cannot accomplish any policy success for his/her constituency, at least he/she can demonstrate parliamentary activism to gain a higher chance of re-election. Therefore, this electoral system could provide incentives to increase the representation of public opinion priorities in comparison with the 2005 Calderoli electoral law (no. 270 of 21 December 2005), which consisted of a proportional rule with closed lists and the possibility for candidates to participate in multiple districts.
Another potentially relevant factor affecting the distribution of attention across policy areas is the effective number of parties in parliament (ENPP) and cabinet (ENPC). While not a necessary condition, it may be sufficient for a parliament (or government) composed of more parties to cover a wider range of policy areas and interests, and it may do so in a more uniform way. Parliament and government fragmentation are measured using the ENPP and ENPC, following Laakso and Taagepera (Reference Laakso and Taagepera1979) and Blau (Reference Blau2008).Footnote 9
A further control relates to ideology. Following issue ownership theory (see Budge, Reference Budge2015 for a review), there may be general attributes of partisanship that influence which issues parties and governments prioritize and focus on in their parliamentary and government activities. Recent works on Italian policy agendas have revealed that ideology is associated with differences in the topics covered by parties in their parliamentary questions (Russo and Cavalieri, Reference Russo and Verzichelli2016), or by governments in their programmatic speeches (Borghetto et al., Reference Borghetto, Curini, Giuliani, Pellegata and Zucchini2017), or changes in public spending (Russo and Verzichelli, Reference Russo and Verzichelli2016). It is measured as the average ideology of parliament, of parties supporting governmentFootnote 10 and the opposition, and it is based on the left–right general variable of the Chapel Hill Expert Survey data set (Bakker et al., Reference Bakker, de Vries, Edwards, Hooghe, Jolly, Marks, Polk, Rovny, Steenbergen and Vachudova2015).
Methods
To test the salience hypothesis on the relationship between changes in the public agenda and Italian political representatives, and given the TSCS structure of the data, I recur to a series of dynamic panel modelsFootnote 11 estimated with feasible generalized least squares (FGLS) and Arellano (Reference Arellano1987) robust standard errorsFootnote 12 for within-group estimators. For each of the four legislative agendas, a dynamic model with the lagged dependent variable (LDV) and fixed effectsFootnote 13 for each policy area is reported.Footnote 14 We can fairly assume that each policy is structured differently due to its peculiarities: on the one hand, the topics may differ in terms of the issue space they attract in the long term, while on the other, there are intrinsic characteristics that give each policy arena a different structure (see Hood, Reference Hood1983; Page and Shapiro, Reference Page and Shapiro1983; Jones, Reference Jones1994; Burstein, Reference Burstein2003; Bevan and Jennings, Reference Bevan and Jennings2014). The traditional approach to dealing with serial correlation in political science has been to introduce an LDV on the right side of the ordinary least squares equation, especially when there are more time periods than units, as in the case at hand, where N=10 and T=19 (Beck and Katz, Reference Beck and Katz1995). While this procedure has been criticized because introducing an LDV as a regressor tends to bias coefficients (Achen, Reference Achen2000), ‘a regression with the proper lag structure produces the best estimates of the effect of the independent variable, so excluding LDVs is not the optimal approach’ (Wilkins, Reference Wilkins2017: 17).
The model takes the following form:
$$\eqalignno{ AgendaSpace_{{ti}} \,{\equals}\, & \alpha _{i} {\plus}\beta _{0} {\times}AgendaSpace_{{\left( {t{\minus}1} \right)i}} {\plus}\beta _{1} {\times}PublicOpinion_{{\left( {t{\minus}1} \right)i}} \cr & {\plus}\beta _{k} {\times}Controls_{{kt}} {\plus}{\epsilon}_{t} $$
Here, the levels of issue agenda space for bills (
$AgendaSpace_{{ti}} $
) of one of the four political actors (the parliament, majority, opposition, or government) is a function of its level in the previous semester to control for autocorrelation (
$AgendaSpace_{{(t{\minus}1)i}} $
), public opinion
$(PublicOpinion_{{\left( {t{\minus}1} \right)i}} $
) in the previous semester (t−1) for each specific policy issue i, plus k control variables that are group invariant, but that vary along the time dimension. The estimated panel models have fixed effects for each of the 10 policy areas considered, and thus, a different intercept α
i
is estimated for each topic. Along with the controls presented above, each model also controls for legislation introduced in the previous semester by other political actors (for instance, when the whole legislative agenda of parliament is the dependent variable, a lagged control is introduced for the government’s agenda).
Results
Table 2 reports the results for the two models explaining the dynamics of the legislative agenda of all MPs. In both models, public opinion is positively related to the parliament’s legislative agenda: on average, a 1% increase in public opinion at time t−1 is followed by a 0.09% increase in the share of bills introduced in parliament at time t. Even if it is a weak statistical relationship, the parliament’s legislative agenda is correlated with the public’s policy priorities, and the structure of the public policy agenda and that of parliament follow similar patterns. The effect is also less than 1, signalling that there is a sort of moderation in the translation of public opinion inputs into formal policy priorities. This can probably be explained in that certain legislative activities are routinely performed or accounted for by other factors, and the set of issues that parties face is more diverse than the issues measured by public opinion surveys. According to the model, the legislative agenda of the executive is not a good predictor of the parliamentary agenda in the next semester.
Table 2 Effect of public opinion on the legislative agenda of all members of parliament (MPs)
Balanced panels: n=10, T=18, N=180.
ENPP=effective number of parties in parliament.
Dependent variable: agendas of all MPs 2003–13.
Feasible generalized least squares estimators with cross-section fixed effects and Arellano standard errors in parentheses.
Constants specific to the 10 cross-sections not reported.
†P<0.1; **P<0.05; ***P<0.01.
Table 3 reports the same models previously fitted, but it only considers bills proposed by MPs from the parliamentary majority. Model 1, the baseline, does not include the lagged agenda of the opposition or government (or a combination of the two), as do the next three models. The relationship between public opinion and the legislative agenda of parties supporting the government seems to be even stronger, with positive and significant coefficients in all four cases. Interestingly, from models 2 and 4, a relationship between the opposition’s agenda at time t–1 and that of the majority at time t emerges, while the relationship with the executive does not return significant results. This suggests a sort of dynamic game in place between the opposition and the majority, according to which, one responds to the priorities of the other when competing in the legislative arena.
Table 3 Effect of public opinion on the legislative agenda of majority members of parliament (MPs)
Balanced panels: n=10, T=18, N=180.
ENPP=effective number of parties in parliament.
Dependent variable: agenda of majority MPs 2003–13.
Feasible generalized least squares estimators with cross-section fixed effects and Arellano standard errors in parentheses.
Constants specific to the 10 cross-sections not reported.
†P<0.1; **P<0.05; ***P<0.01.
Table 4 reports the results for the agenda based on opposition MPs’ legislative proposals. For these MPs, the relationship put forward in the theoretical framework does not hold for any of the models, regardless of whether we control for the majority and/or government agenda. This result weakens the first hypothesis, but it also disconfirms the second one. While the expectation was that we would find a stronger relationship between public opinion and the opposition’s agenda due to less institutional and political constraints, this does not seem to be the case. A feasible explanation could lie in the fact that the opposition is much less homogeneous than the majority is, and therefore, pooling together all parties in opposition to the government may confuse the results. The only relevant factor explaining some of the variance of the dependent variable is the lagged agenda of the majority.
Table 4 Effect of public opinion on the legislative agenda of opposition members of parliament (MPs)
Balanced panels: n=10, T=18, N=180.
ENPP=effective number of parties in parliament.
Dependent variable: agenda of opposition MPs.
Feasible generalized least squares estimators with cross-section fixed effects and Arellano standard errors in parentheses.
Constants specific to the 10 cross-sections not reported.
†P<0.1; **P<0.05; ***P<0.01.
Table 5 reports the models explaining the legislative agenda of the five governments in power from 2003 to 2013 – Berlusconi II, III, and IV, Prodi II, and Monti. Again, the public’s attention to specific issues is always positively related to the legislative agenda of the executive. The rate of approval of bills introduced by the government – roughly 50% of its proposals are approved (Borghetto and Giuliani, Reference Borghetto, Visconti and Michieli2012) – hints that, when introducing bills, the members of cabinets do not have the same freedom as the MPs and parties do. Executives are held accountable by the public, by the majority in the legislature, by interests they represent and by supranational policy constraints (e.g. the European Union). Therefore, the institutional and political costs of introducing a bill in line with public opinion is higher for governments than it is for other actors; still, the relationship (at least for the period analysed) holds true, even if the models only explain a small proportion of the variance of the dependent variable. The negative relationship with the lagged agenda of majority MPs is of interest here. This coefficient is likely related to an agreed behaviour between parties supporting the majority and the cabinet to strategically cover different issues.
Table 5 Effect of public opinion on the government’s legislative agenda
Balanced panels: n=10, T=18, N=180.
ENPC=effective number of parties in cabinet; MPs=members of parliament.
Dependent variable: government legislative agenda.
Feasible generalized least squares estimators with cross-section fixed effects and Arellano standard errors in parentheses.
Constants specific to the 10 cross-sections not reported.
†P<0.1; **P<0.05; ***P<0.01.
In summary, the investigation into the first hypothesis seems to support the idea that public opinion policy priorities are related to how legislation is prioritized by the Italian political system. It has emerged that the structure of the public’s most significant problems tends to lead how attention is distributed in the agenda of the majority MPs, government, and all MPs pooled together. For the opposition, the relationship does not hold, which disproves the second hypothesis.
It seems worth making the results more robust by addressing ‘the elephant in the room’ – endogeneity. On the one hand, a third factor could be the real driver of both public opinion and legislative agendas; on the other, it may well be that public opinion policy priorities are counterfeited by political actors (Brooks, Reference Brooks1985, Reference Brooks1990). Addressing the first issue requires the availability of data sets on factors that could be behind shifts in both agendas, for instance, the mass media. Unfortunately, comparable data sets are not yet available for the Italian case, impeding a more thorough test of the influence of public opinion. Instead, one way of addressing the second issue consists of looking at reverse models and evaluating the autonomous nature of public opinion (at least with respect to legislative agendas). Table 6 presents four panel FGLS models, with fixed effects specific to each policy areaFootnote 15 regressing public opinion on the four types of legislative agendas. In none of the four cases did agendas built on bills return statistically significant coefficients. Thus, these tests seem to support the claims made in the previous section, given that they empirically establish temporal precedence.
Table 6 Effects of legislative agendas on public opinion
Balanced panels: n=10, T=18, N=180.
MPs=members of parliament.
Dependent variable: public opinion policy agenda.
Feasible generalized least squares estimators with cross-section fixed effects and Arellano standard errors in parentheses.
Constants specific to the 10 cross-sections not reported.
†P<0.1; **P<0.05; ***P<0.01.
Conclusion
The objective of this article was to shed light on a representative problem – the relationship between legislative initiative and aggregated public opinion policy priorities in Italy. The empirical test involved the Italian political system across three legislatures and five governments in the decade spanning from 2003 to 2013. Rather than focussing on positional representation, it addressed the covariation of the public agenda with that of institutions, contributing to the literature on policy agenda representation. The salience hypothesis, tested based on the assumption that MPs, parties and governments react strategically to the ebbs and flows of public opinion to gain (or preserve) votes, considered whether issues sharing higher prioritization among the public are also favoured in parliament and government agendas. The findings support the literature on dynamic agenda representation measured through legislative data, suggesting a link between citizens’ policy priorities and legislation introduced in the Italian political system.
A second hypothesis argued in favour of a stronger link between the agenda of the opposition and that of the public, due to the lower institutional and political constraints guiding opposition activity compared with the majority and government. The analysis showed that public opinion policy problems do not have any significant relationship with the opposition’s agenda. These results could stem from the fact that this study pooled all opposition parties together, regardless of their coalition potential. This may have counteracted any eventual significant relationship. An alternative hypothesis could be that cabinets and their supporting MPs have a comparative advantage as ‘first movers’ compared with their opposition counterparts. Having access to the machinery of government, they may be able to spot issues of public concern more quickly and deliver policy signals to constituents swiftly. In contrast, the opposition may have an incentive to implement a different strategy by addressing niche issues to outflank the government.
The empirical evidence presented runs against some previous studies that have emphasized the detachment between the Italian public and its political representatives (Russo and Verzichelli, Reference Russo and Cavalieri2012; Memoli, Reference Memoli2013). Instead, the results are more compatible with another line of research that has stressed voters’ increased democratic accountability, as portrayed by Bellucci (Reference Bellucci2012), Bellucci and Pellegata (Reference Bellucci and Pellegata2017), and Russo and Cavalieri (Reference Russo and Verzichelli2016). Citizens delegate the responsibility of ruling according to a mandate to the government. Knowing that they will be held accountable, parties (and the government) also compete by emphasizing those issues with the potential to grant them a strategic advantage in the eyes of voters, thereby moving their preferences and priorities towards those of the public. The positive relationship between executives’ legislative activities and public priorities goes hand in hand with the increasing majoritarian nature characterizing Italy’s so-called Second Republic (Capano and Giuliani, Reference Capano and Giuliani2001; Marangoni, Reference Marangoni2013; Cotta and Marangoni, Reference Cotta and Marangoni2015). Indeed, parliamentary systems are expected to show a weak link between government policies and public opinion when majoritarian dynamics are absent (Wlezien and Soroka, Reference Wlezien and Soroka2007).
The overall picture emerging of the Italian political system does not seem to be as dire as is sometimes thought, at least in terms of policy agenda representation. Given the intricacies of delivering substantive policy goods, one way to survive in government and parliament is to constantly update priorities following the public. It is true that the political dynamics typical of the years under investigation have already mutated with the entry onto the scene of the Five Star Movement and the end of a bipolar competition.
Finally, the limitations of this study are worth noting. A source of problems lies in the structure of the categories considered. The MIP question found in the Eurobarometer offers respondents a limited set of alternatives compared with those available to MPs and cabinet members. Therefore, the match (or lack thereof) between policy categories could be an artefact of selection issues. Still, so far, no better solution has become available in the literature on the nexus between public opinion and policymakers’ priorities. Future works should try to address these issues, with a view to providing a more complete picture of policy agenda representation in Italy.
Acknowledgements
The author would like to thank the editors and the anonymous reviewers for their constructive comments and suggestions, Paolo Bellucci, Luca Bernardi, and Oriol Sabaté Domingo with whom the author discussed the paper in its preliminary stage.
Financial Support
The research received no grants from public, commercial, or non-profit funding agency.
Data
The replication data set is available at http://thedata.harvard.edu/dvn/dv/ipsr-risp
Conflicts of Interest
None.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/ipo.2018.4
