Abstract
A key factor in modern democracies’ legitimisation is the extent to which policies submitted for public approval before an election translate into material outcomes once a political party has won power. Current research finds no clear empirical evidence for partisanship in policy-making nor has any unified theory been offered or tested systematically. This article addresses that gap by offering a conditional approach to policy-making undertaken by parties in government. It suggests that partisan influence on policy depends on both office-holders’ capacity for implementing policies evoked during their electoral campaigns and on governing parties’ incentives to implement electoral promises. Data from French Agendas Project datasets is used to compare the contents of governing parties’ pre-election manifestos with legislation passed in France between 1981 and 2012. Panel negative binomial regressions on electoral and legislative agendas support the expected outcome, namely that issues featuring in governing parties’ electoral manifesto have had an impact on their subsequent legislative agendas, with the effect depending on both partisan capacities and incentives. Party programmes do matter in policy-making, albeit only under certain conditions.
A key factor in modern democracies’ legitimisation is the extent to which policies submitted for public approval before an election translate into material outcomes once a political party has won power. Political parties and their candidates are expected to aggregate sets of policy positions and commit to implementing them as policies. Yet, the ability of any political system to achieve elected representatives’ normative goals, or even popular sovereignty, may be undermined if the “programme-to-policy linkage” (Thomson, 2001) is too weak. This concern is not new. However, having become one of political science’s most researched topics, the problem is that empirical evidence in the field is at best contradictory.
The most influential approaches to public policy accord political parties and elections a role that is at best marginal (John, 2012). Policies change slowly if at all (Lindblom, 1959; Pierson, 2000) in the face of all the problems that constantly emerge (Adler and Wilkerson, 2013; Wildavsky, 1964). In the main, there is little room for mandate fulfillment, given policymakers’ cognitive limits and institutional friction (Baumgartner and Jones, 2005) and the presence of networks of experts, bureaucrats, interest groups and citizen organisations (Marsh and Rhodes, 1992). Current literature moreover suggests that governments’ room to manoeuver is vanishing (Boix, 2000) due to growing interdependence (Keohane and Nye, 1989), international capital mobility (Frieden and Rogowski, 1996) and demographic dynamics (Castles, 2004; Häusermann, 2010).
These challenges to the party-mandate conception of democracy have triggered an extensive literature investigating the nature of the “programme-to-policy linkages”. The debate has long revolved around the question of whether parties actually matter. Analyses have generated mixed findings. Some observed left-right differences with regards to budget increases (Blais et al., 1993; Cameron, 1978), government employment (Cusack et al., 1989), macro-economic intervention (Boix, 2000; Hibbs, 1977) and redistributive policies (Esping-Andersen, 1990; Huber et al., 1993). Others revealed unexpected effects, such as leftwing parties who deregulate and restrain spending (cf. Armingeon et al., 2016; Baumgartner et al., 2009; Häusermann, 2010). This ambivalence is confirmed by meta-analyses (Burstein and Linton, 2002; Imbeau et al., 2001).
Research on party manifestoes has attempted to assess the implementation of party mandates through a more granular operationalisation. The assumption that political parties represent stable interests and positions, common to much earlier literature, seems fragile given the changing nature of party constituencies and electoral strategies’ influence on policy direction (see Häusermann et al., 2013 for a review). Analyses of party manifestos and government spending priorities have found a strong relationship between the two (Budge and Hofferbert, 1990; Klingemann et al., 1994). This research strand has, however, been criticised for the limitations of its coding scheme, one more adapted for analysis of party positions rather than policy outputs (Guinaudeau, 2014: 268). Moreover, attention to an issue does not necessarily imply greater spending.
The present article argues that evicting party programmes from policy studies, altogether creates a risk of missing one possible determinant of policy decisions, namely elected officials’ incentive to commit to at least some electoral pledges since their party is going to have to communicate about policy enactment in order to get re-elected in the future. Faced with contrasting findings in the “do-parties-matter” literature and the empirical observation that many if not all pledges are in fact fulfilled, we show that a conditional approach seems better suited to understanding at what juncture party platforms actually matter (cf. Schmidt, 1996).
The present paper introduces a comprehensive model drawing on insights from the different aforementioned strands of literature. Its model acknowledges public policy literature’s depiction of considerable constraints, associating this with the hypothesis that partisan influence depends on parties’ capacity for getting their policies on the agenda. It remains that political party literature shows that office-holders are strategic agents seeking to optimise scarce resources to attract votes, take office and adopt policy – inferring that their influence also depends on the incentives they have to act on the issues highlighted during their electoral campaigns. We test our framework using extensive data on parties and policy-making in France.
The following section develops this conditional approach to partisan government. Subsequently, we present data and methods before analysing and discussing the findings in conclusion.
Capacities and Incentives: A Conditional Mandate Connection Between Programmes and Policies?
We combine different lessons from public policy and party politics literatures to develop testable hypotheses about conditions facilitating partisan influence. Public policy literature has shown that parties in office have relatively little leeway for triggering policy change. Government parties are unable to deal with the complete range of issues that arise during their campaigns. Having said that, literature on party politics claims that parties and their elected officials are generally strategic agents aware of the trade-offs between policy aims, getting re-elected and winning votes (Müller and Strøm, 1999). Incumbent parties usually seek re-election and must convey an attractive political vision while communicating about the policies enacted during their time in power. In turn, this creates incentives to implement partisan policies – and, in particular, to fulfill a fair proportion of all the campaign promises made.
Given the restrictions on governing parties’ room to manoeuver, they sometimes focus their efforts in those areas where they can expect the greatest electoral pay-off. In this case, the promised policy might be implemented at whatever point in the cycle is most favourable. The model proposed in this article therefore assumes that programme-to-policy linkages are most likely when parties have the ability (capacity) and/or encouragement (incentives) to be partisan in their policy-making.
Capacity
Literature contesting the concept of partisan influence mainly argues that governments lack the capacity to truly shape policy. This dimension, referring to the resources and constraints that define a governing party’s room to manoeuver, is often considered the only one that defines a conditional model of partisan influence on policy-making. A party can be expected to keep its promises if and only if it enjoys sufficient institutional power to implement them (Schmidt, 1996; Thomson et al., 2017). Moreover, the more complex decision-making becomes, the less governing parties have the capacity to set an agenda and enact policy (Tsebelis, 2002). The institutional component of capacity having been established, the present model also incorporates a second and possibly even more important (albeit less studied) component, one that involves issue-specific characteristics.
Institutional Capacity
Institutions do impact upon policy outputs and outcomes to the extent that they shape governing parties’ decision-making autonomy. The democratic mandate principle is based on the observation that in liberal (especially majoritarian) democracies, individuals in executive or ministerial roles hold great power and are key to the policy-making process. There are good reasons to expect that the political parties appointing these officials have the capacity to identify and define public problems at an early stage of the decision-making process. They therefore impel, frame, orient and veto policy change. Legislative politics studies have largely confirmed parties’ policy-making powers in parliamentary matters. Where basic models have predicted that policy decisions depend on the floor median position of the legislative chamber (Black, 1948) and can only expect to be marginally affected by changes in the governing party (Krehbiel, 1993), there have also been empirical observations of an off-median partisan bias in parliamentary voting behaviour and legislative outcomes (e.g. Richman, 2011). 1 This corpus finds that a government supported by a political majority does have the capacity to pass at least some of its policies, especially in areas like welfare policies (Korpi and Palme, 2003; Rueda, 2005).
The pivotal politics and veto-player models (Krehbiel, 1998; Tsebelis, 2002), on the other hand, stress that counter-majoritarian institutions dilute capacity. The greater the system of checks and balances and the less majoritarian the nature of the democracy, the harder it becomes for the main party in government to shape policy (see Blais et al., 1993; Schmidt, 1996 for the general argument). George Tsebelis (2002) has also argued that the need to form a government coalition spawns additional veto players, with coalition parties becoming “partisan veto players”. The shorter the distance between the political party extremes represented in cabinet, the more their policy-making capacity rises.
Preliminary evidence corroborates the restrictive effect of counter-majoritarian institutions on partisan influence. The real effects of divided government (Alesina and Rosenthal, 1995; Blais et al., 1993) have been subject to debate and contested by some scholars (Conley, 2007; Mayhew, 1991). Others have looked more at the constraints on partisan policy-making when a government is divided (Baumgartner et al., 2014; Binder, 1999; Edwards et al., 1997).
Issue-related Capacity
Until now, conditional approaches to partisan influence in policy-making have examined institutional parameters only. This focus is striking, given that the constraints discussed in policy studies are usually not institutional but linked to issue-specific characteristics such as the inertia of past decisions and the cost and availability of technical solutions to any particular issue. The institutional setting may thus not be the only relevant dimension of partisan capacity – and perhaps not even the most important one. One meta-analysis has suggested, for instance, that partisan influence varies according to policy area and the type of instrument being considered – especially when cost-intensity is a factor (Imbeau et al., 2001: 7–8). This intuition has been generally downplayed in partisan influence studies, possibly due to the difficulty of operationalising many of the relevant variables on a comparative basis. Unsurprisingly, several case studies have demonstrated sector-level specificities in terms of the way in which elections and parties shape policies. Examples include green party influences on energy policy (Evrard, 2012), right-wing parties on immigration (Givens and Luedtke, 2005) and leftwing parties on taxation (Dennis et al., 2007). It remains that large-N aggregate analysis assessing partisan impact across policies has yet to be carried out.
The classic policy typology proposed by Theodor Lowi (1964) enables the identification of different opportunity costs associated with redistributive, distributive, constitutive or regulatory policies. 2 These types can be operationalised on a large scale and are therefore highly relevant. Political parties’ policy-making capacity may be more limited for redistributive (i.e. fiscal or social) policies due to the considerable budgetary trade-offs and, more generally, because of the high opportunity costs and societal veto players involved. Note along these lines many governments’ declining capacity in recent decades to make macro-economic policy (Schäfer and Streeck, 2013). Foreign policy and defence, characterised as they are by strong path-dependency (Imbeau, 1988; Thérien and Noël, 2000), might also be expected to afford governing parties limited room to manoeuver. Conversely, government parties may find it easier to change regulatory policies (such as environmental or civil right policies) that do not systematically imply public spending, that fall under the cabinet’s direct jurisdiction, and that tend to impact only specific types of social groups.
Incentives
Even where governments have the capacity to fulfill their electoral pledges, they may not always do so. Explanations for variations in this respect necessarily invoke governing parties’ incentives. Capacity relates to the ease with which policies can be enacted; incentives are shaped by the returns that political parties can get from said policies. Parties are not likely to fulfill an electoral pledge in the absence of an incentive. Instead, they may invest their limited resources in adopting those policies they expect to be the most rewarding.
Political parties and their elected officials are generally viewed as arbitrating between three types of incentives: policy, office, and votes. Most authors tend to agree on the primacy of vote seeking as this is the precondition for even taking office and passing policy (e.g. Müller and Strøm, 1999). Electoral incentives to implement electoral pledges include several factors that are not easily measurable over time or across a large number of cases. The main factors at this level include public support for policy (Page and Shapiro, 1983; Soroka and Wlezien, 2010); parties’ ties to specific issue publics (Henderson, 2014) and interest groups, and issue salience, which makes the policy environment more transparent and increases rewards for fulfilling promises or else sanctions for breaking them (Culpepper, 2011).
Other incentives may be easier to capture. Time, for instance, is a crucial factor most likely to affect a government’s motivation to act on promised issues (Canes-Wrone, 2006). Following elections, governing parties have an incentive to demonstrate their ability to deliver the policies promised during the campaign. The same applies at the other end of the electoral cycle, when the imminence of the next election may push them towards re-election-oriented policy-making (Alesina and Rosenthal, 1995; Tufte, 1978). The existence of electoral cycles in economic policies and outcomes (Franzese, 2002) seems to corroborate the idea that the distance to/from the next/last election affects how political parties behave.
As a consequence, the present article expects that partisan influence on policies will increase where a government enjoys broad institutional capacities; when the issue in question is unrelated to redistribution, foreign policy or defense; and when the electoral cycle is at its initial or final stage.
Data and Methods
The approach adopted here can be used to explain partisanship variations in both issue attention and direction terms. The present study is limited to the first of those dimensions. Its dataset is comprised of issues dealt with through legislation and/or highlighted in the manifesto of the party of the person who became prime minister following the previous legislative election, in France from 1981 through 2012.
Country Selection
It is always difficult to generalise from a case study, but there are several advantages to focusing on France in this way. First, holding constant certain factors such as the country’s institutional environment, history and political culture 3 enables analysis of real changes in government control patterns, headline issues and the electoral cycle. Second, French policy-making developments between 1981 and 2012 provide a perfect case study to test the validity of our conditional model. The 32 years covered feature substantial variations in all the variables of interest. The period includes eight legislative elections, starting with the first political alternation in France’s Fifth Republic and the Socialist Party’s victory in the 1981 presidential and legislative elections – followed by numerous subsequent alternations (1986, 1988, 1993, 2002), majorities of varying size (including minority governments between 1988 and 1993) and periods of divided government (1986–1988; 1993–1995; 1997–2002).
France also offers fertile ground for testing party organisations’ influence on policy-making. In this semi-presidential regime, political competition is mostly dominated by presidential elections, triggering considerable personalisation of politics and limiting political parties’ clout (Bélanger et al., 2012; Dupoirier and Sauger, 2010). Presidential candidates often distance themselves from their party’s manifesto to develop a profile as someone who is ‘beyond partisanship’ (Baumgartner et al., 2009). Once elected, presidents will try to put their own personal touch on policy. To this should be added France’s frequent characterisation in policy studies as a prime example of “technocratic” and “bureaucratic” policy-making, a place where incremental policy change is seen as being relatively disconnected from electoral competition (Culpepper et al., 2006; Kitschelt, 1986). Not to mention the thought that observing any sort of partisan influence on policy in France strongly suggests the likelihood of finding similar influences in other democratic regimes, ones usually characterised by a smaller distance between governing parties and the executive.
Measuring Parties’ Issue Priorities and Their Impact on Law
As aforementioned, the two main empirical strategies implemented to analyse programme-to-policy linkages both have limited capacity for answering the research question being asked here. Studies examining the impacts that political parties and their manifestos have had on very general policy output measures (i.e. level of public spending) or outcomes (i.e. macro-economic performance) have been criticised inter alia because of their dependent variable (e.g. Thomson et al., 2017). The alternative approach defended by the CPPG focuses on electoral manifestos’ policy content (i.e. pledges made). As a dependent variable, it also focuses on actual policy – not only spending but all policy outputs and outcomes related to electoral pledges. 4 The approach is useful for assessing what happens with electoral programmes but does not reveal whether such programmes shape policy to a significant extent. To achieve this, a whole set of adopted policies would have to be considered. At the very least, research must focus on those significant policy categories for which data can be systematically collected and coded in such a way as to enable the analysis party programmes’ impact over time.
Among comparable policy outputs, this article explores manifestos’ influence on laws rather than budgets. As noted by Louis Imbeau et al. (2001: 6), “do-parties-matter” studies of party influence on policy outcomes tend to be less conclusive, with growth, employment and inequality often portrayed as “much less susceptible to manipulation by governments than policy outputs”. The rising number of constraints and spread of austerity politics (Schäfer and Streeck, 2013) and regulatory states (Majone, 1997) is furthermore likely to dissuade political parties from promising higher spending. Instead, the expectation here is that they will focus on other forms of policy output. In the case of France, data collected by the Partipol project on the 485 electoral pledges made by governing parties over 1995–2012 has confirmed this expectation. Whereas more than half (51.3%) of the pledges in question implied the adoption of a law, only 30.3% led to actual spending. 5 Hence, the contention that laws offer the most meaningful policy indicator for assessing a “mandate connection”. This is particularly true in France, where laws have a strong policy purpose – contrary to other countries, such as the USA, where they often only provide a legal framework enabling the extension of existing programmes (Adler and Wilkerson, 2013).
The independent variable captures issue attention such as it materialises in the prime minister’s party manifesto. Clearly, political parties not only compete over positions but also over issue priorities (Budge and Farlie, 1983; Klingemann et al., 1994; Robertson, 1976). This means that issues emphasised by political parties during electoral campaigns vary depending on the party and point in time (e.g. Brouard et al., 2014; Green-Pedersen and Mortensen, 2014). Unfortunately, there is remarkably little empirical information whether partisan issues eventually reach the legislative agenda. Legislative agenda-setting studies (Döring, 2001; Huber, 1996) have suggested that there are reasons to expect partisan influence on policies pushed through a government or parliamentary agenda; portfolio holders and/or party leaders choose which bill to introduce and vote upon. Other works confirm that parties tend to stick to their issues at the initial stage of the legislative process, with the issue content of manifestos being correlated to the topics addressed through the introduction of bills (Bräuninger and Debus, 2009) or parliamentary questions (Green-Pedersen and Mortensen, 2010; Vliegenthart et al., 2013). It remains that this type of partisan influence has not been examined in terms of the laws actually enacted.
The present dataset has been constructed using the issue profile of party manifestos and legislation enacted between 1981 and 2012. 6 Each quasi-sentence of the electoral manifestos 7 and each adopted law has been coded by the type of policy issue in question, as per the Comparative Agendas Project coding scheme. 8 The resulting dataset, computed from data available on the CAP website, 9 has been arranged in the form of a panel cross-section, with 14 policy areas constituting cross-sectional units observed over 282 months of parliamentary session, yielding an N of 3.948 (14*282) policy area-months.
The data account for the policy issues referred to in the written sentences comprising the electoral manifestos or subsequent legislation but not for the substance thereof. In other words, congruence between independent and dependent variables does not necessarily imply that party pledges are being implemented. Having said that, acting in those areas that had been highlighted during the electoral campaign is an essential precondition for pledge fulfillment. The choice of which issues require political decision-making is often viewed as one of the two faces of power (Bachrach and Baratz, 1962). Similarly, adding a new issue to a legislative agenda is a crucial point in any policy change scenario (Kingdon, 1984). Until now, scholars have given extensive scrutiny to agendas at a number of different levels, including hearings, bills and laws (Baumgartner and Jones, 2005) without exploiting the partisan hypothesis in such a way as to account for the considerable variations observed over time and across agendas, issues and countries.
The dependent variable is the number of laws devoted to each topic every month. The main independent variable is the percentage of quasi-sentences devoted to the same topic in the prime minister’s party manifesto. Using the prime minister’s party (and not the presidential) manifestos makes sense from an institutional perspective; in France, policy change depends on support from the parliamentary majority, which is elected on policy proposals formulated in a prime minister’s manifesto. Using this source also offers the advantage of ensuring sources’ comparability over time, regardless of the concordance between the parliamentary and presidential majority.
Measuring the Conditions Enabling the Linkage of Programmes with Policies
Our model implies that legislative priorities only derive from governing parties’ manifestos when certain conditions relating to partisan capacity and incentives are involved. As regards institutional capacity, dummy variables have been included in the dataset to reflect the status of the prime minister’s party (i.e. whether it has a parliamentary majority) and differentiate between united and divided governments. 10 An ideological distance or range measurement (Franzmann and Kaiser, 2006) has also been added to account for constraints relating to partisan veto players. To test for differences in policy-making capacities linked to issue specificities, a variable has been created identifying regulatory policies (coded 1) 11 expected to depend more on partisan influence than other policies. These include (re-)distributive measures (issues pertaining to macro-economic policy, finance, transport and welfare 12 ), foreign affairs and defense (coded 0). 13 Finally, electoral incentives have been accounted for through the electoral cycle. This variable is coded 1 for the 6 months following a general election, 2 for the 6 months preceding it, and zero otherwise. 14
Several additional variables have been controlled for as well. First, the cumulative legislative production, that is, the number of laws already adopted on a given issue during the current legislature, accounts for the lower likelihood that an issue will go onto a legislative agenda after having already been subject to legislation earlier in the legislature. Ljung-Box Q statistics found no significant autocorrelation in the residuals, removing the need to control for lagged legislative production in one and the same policy area. 15 Because of observable differences in legislative productivity between leftwing and right-wing French governments – the former adopt more laws, governmental political orientations have also been controlled for using a dummy coded 1 whenever the Socialist party led the government. Lastly, recognising that electoral programmes are not exogenous – and that their authors are likely to use all available information to anticipate their party’s capacity to implement its mandate – a control variable has been inserted to measure the difference between the party’s expected polling score published three months before the election and its main competitor’s score at the time. This indicator provides an approximation of whether parties expected to win or lose at the time their manifesto was being written.
Results
We use panel negative binominal regressions, reflecting the cross-sectional longitudinal structure of the data and the count nature of the dependent variable. 16 These regress the probability within any given month of having an additional law addressing any or each of the 14 issues in question, as well as the attention paid to the issue in the prime minister’s party election manifesto. Issue-specific effects have been accounted for by fixed effects. 17 Negative binomial regressions have been used to counteract count model over-dispersion violating Poisson distribution assumptions (Hilbe, 2011). Importantly, the conditional model implies an interactive specification (Brambor et al., 2006) enabling an estimate of the interaction of electoral priorities with, respectively, divided government, the majority status of the prime minister’s party, government range, issue type and advancement of the electoral cycle. The conditioning effect of institutional capacity has been examined first, followed by issue-specificities and electoral incentives.
Our first focus is on the institutional capacity’s conditional effect on the correspondence between electoral and legislative issues. As per the hypotheses presented in the literature review above, the models in Table 1 explore the influence of institutional factors – respectively whether the prime minister’s party has a parliamentary majority, if the government is divided and what ideological range is represented in the government. Controlling for these institutional configurations does not depict issue attention during electoral campaigns as having had a significant effect on adopted laws’ thematic profile. While past attempts to develop conditional models of policy-change induced by parties or elections focus on counter-majoritarian institutions, the present findings do thus not corroborate their relevance as conditioning variables. Nevertheless, the models in Table 1 display evidence of some direct – though non-significant – effects on legislative productivity. Holding a parliamentary majority facilitates the implementation of partisan legislation, but only very weakly. Divided governments appear – somewhat surprisingly, but in line with recent findings by Frank Baumgartner et al. (2014) – to have increased the volume of legislation (by 13%, model 1b). Note, however, that both factors did not increase the partisan nature thereof. In line with Tsebelis (2002), when the intra-government distance increases, the likelihood of a new legislation decreases as expected (model 1c). In sum, despite the general presupposition that institutional context affects government parties’ policy-making capacity, no such mediating effect was found here.
Effects of Partisan Priorities on Laws, According to Institutional Capacity: Negative Binomial Models.
Source: Agendas France.
%: percent change in expected count for one unit increase in X based on exp(ß). 18
Statistically significant: *p < .05; **p < .01.
Table 2 presents three models testing our conditional theory of partisan influence, looking first at the conditioning impact of issue types (Model 2), electoral cycles (Model 3) and simultaneously at both (Model 4). Models 2 and 4 confirm the importance of issues types, controlling for the institutional factors previously analysed. 19 The constitutive term for partisan attention shows that issues related to redistributive policies, defense and foreign affairs are not more likely to be a subject of legislation if they received greater attention in the prime minister party’s manifesto. The significant effect of the interaction term does, however, support the hypothesis of a significant programme-to-policy linkage where regulatory issues are involved. A 1% increase in the attention that a manifesto pays to a particular regulatory policy increases by almost 4.5% the likelihood that a law dealing with the same issue will be adopted – an effect that is likely to be decisive given the considerable variations in the attention devoted to different issues in party programmes. Not only do issue specificities alter governing parties’ ability to shape policies and mediate the programme-to-policy linkage, but controlling for them appears crucial in the identification of partisan influence.
Effects of Partisan Priorities on Adopted Laws, According to Electoral Cycle and Issue Type: Negative Binomial Models.
Source: Agendas France.
%: percent change in expected count for one unit increase in X based on exp(ß).
Statistically significant: *p < .05; **p < .01.
The conditional impact of electoral cycles is supported by models 3 and 4. As expected, laws mainly reflect issues highlighted during the main government party’s electoral campaign in the six first months after an election. At this stage of the electoral cycle, 1% more attention paid to an issue in the electoral programme goes hand-in-hand with a 3.6% greater likelihood that a law will later be passed in this area. By contrast, electoral priorities do not affect lawmaking during more anodyne legislative periods nor – and contrary to our expectations – towards the end of the electoral cycle. Data are congruent with the hypothesis of a variation of partisanship over an electoral cycle.
Figures 1 and 2 use estimations from Model 4 to plot electoral priorities’ conditional influence on the estimated likelihood that a particular policy issue becomes the subject of a law (average marginal effect). It does this by comparing regulatory and other issues (Figure 1) and routine and election periods (Figure 2). Findings confirm the importance of partisan attention, demonstrating how it interacts with types of policy – but also with the electoral cycle – to affect the contents of the laws being enacted in the expected way. The link between electoral and legislative attention is significant only for regulatory issues or at the beginning of an electoral cycle. These findings remain valid when controlling for traditional factors such as legislative productivity. The fact that ideological distance within a government decreases the number of laws – or that leftwing governments tend to adopt more laws – does not affect the significance and direction of the conditional partisan effect.

Electoral Priorities’ Average Marginal Effect on Legislative Attention, by Policy Type (at 95% Confidence Interval).

Electoral Priorities’ Average Marginal Effect on Legislative Attention, by Electoral Cycle Sequence (at 95% Confidence Interval).
To assess the robustness of these results, the models were replicated using alternative specifications of the main variables and some of the control variables (not shown). In particular, a variety of di- and trichotomic electoral cycle measurements were undertaken using different cut-offs between routine and election periods (ranging from three to nine months). Models were also run using an alternative measurement for divided government, restricting this to a divided executive (i.e. irrespective of whether the prime minister’s party controlled the Senate). A further robustness test analysed how a trichotomic variable affected different policy types (regulatory, (re-)distributive and foreign relation policies). None of these tests had any substantive effects on the conclusions.
Conclusion
Mixed empirical findings regarding the linkage between political parties and policy have cast doubt about how important parties and their programmes are to policy-making in representative democracies. Although scholars have called for a unified theory accounting for the ‘conditionality’ of mandates, such a theory has yet to be found. The present paper is a first attempt to conceptualise and test a theory capable of specifying the conditions conductive to partisan effects on policy outputs. In turn, this theory nurtures an expectation, namely that partisan effects are mediated by two conditions: capacities (both institutional and policy-related); and incentives to legislate.
Using data from the French Agendas Project, we tested whether policy types and/or electoral cycles mediate the way in which issue attention in a prime minister’s party manifesto affects the topics covered in the laws adopted. Findings reveal that congruence between electoral and legislative priorities does not depend upon institutional capacity, despite this variable having been the most widely discussed in past studies on legislative productivity, partisanship in policy-making or pledge fulfillment. However, support has been found for the other propositions contained in the conditional model. On one hand, government parties are more likely to legislate on the issues highlighted in their electoral programmes when they have the capacity to do so. This is particularly true where regulatory issues are involved, in contrast to redistributive, foreign and defense policies characterised by heavier constraints. Additionally, issue congruence between programmes and policies is also dependent on governments having incentives to implement a mandate, something shown here as having a significant effect during the first months of a term of office. This is a relatively surprising finding given the ostensible fading of “electoral business cycles” since the late 1970s (Shi and Svensson, 2006). It suggests that cycles today may have a greater effect on legislative attention than on spending decisions. Having said that, given the ebb and flow of real-world problems – and the reality of external shocks – the findings should not be interpreted as proof that parties in government are mere ideological agents. Instead, the lesson is that political parties could be depicted as strategic agents who do not implement policy platforms blindly but measure the constraints and incentives that weigh upon policy-making. Facing a huge number of real-world problems, governing parties prioritise the issues where they have the greatest capacities and incentives to act. In short, they behave like strategic agenda-setters (Bachrach and Baratz, 1962).
The findings have several implications for future studies. First, they strengthen the case for the “parties-do-matter” hypothesis (Blais et al., 1993; Budge and Hofferbert, 1990), albeit not necessarily in terms of spending. In a related fashion, the article also shows that elections still “matter”, too, contradicting the widespread scepticism among public policy scholars (Baumgartner and Jones, 2005; John, 2012). To be detectable, their effect requires a conditional approach, as well as an empirical strategy accounting for variations in electoral supply over time, based on empirical analyses of party manifestos rather than dummy variables for left, centre and right-wing parties. This seems crucial in a context of growing electoral volatility, changing ideology, and increased complexity of policy-making (Häusermann et al., 2013).
Scholars would be well advised to keep this in mind when studying policy, especially in those areas where party programmes are most likely to be influential. By so doing, they might shed new light on the way in which complex political systems and inertia occasionally create space for partisan policy-making. Second, in the French case at least, it has been demonstrated that institutions or institutional context wield much less influence than is commonly assumed, whereas policy characteristics matter significantly more. This should encourage scholars to focus more on partisanship and policy change factors, in a corpus that has mainly prioritised institutional explanations until now (see above). This article has already controlled for a long list of potential confounders but a number of others factors – including issue salience, exogenous shocks and party connections to issue publics or interest groups – are harder to measure and model. Investigating their conditional impact on the implementation of party mandates is a worthwhile avenue for further research drawing on complementary (and better adapted) small-N approaches.
Admittedly, the research design applied here has emphasized the internal validity of a study in France. France is clearly one of the least likely cases of partisan influence, as discussed earlier. It is therefore legitimate to question the findings’ generalisability. Institutional, political and policy-related factors shaping capacity and influence may vary considerably from one country to the next. The magnitude of partisanship in policy-making – and the associated dimensions of capacity and influence – may very well depend on context. The specific features of divided government in France (Baumgartner et al., 2014), where the President under cohabitation has a weaker influence on the lawmaking process, may explain why divided government does not have a non-trivial effect on the correspondence between electoral and legislative issues. Counter-majoritarian institutions could, for instance, be more important in Germany, a federal country with a stronger second chamber, and where less adversarial social dialogue between the country’s societal veto players attenuates any differences between regulatory and (re-)- distributive policies. Systematic investigation of the factors defining governing parties’ capacities and incentives to stick to their electoral priorities across different political systems offers an appealing avenue for future research. Comparative research agendas should help deepen our understanding of the influence of institutions on the party-policy link.
Footnotes
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This article would not have been possible without the financial support by the Agence nationale française de la recherche (National French Research Agency) to the Agendas France [grant n°ANR-08-Gouverner-055] and Partipol projects [grant n°ANR-13-JSH1-0002-01].
