Introduction
The assumption that representatives are interested in getting reelected provides a solid basis for the study of legislative behavior. Reelection-seeking parliamentarians are indeed expected to act in view of maximizing their support in subsequent elections (Mayhew, Reference Mayhew1974), while voters are expected to take the incumbents’ performance into account when casting their vote (Ferejohn, Reference Ferejohn1986). There have been numerous publications documenting the impact of constituency-oriented activities on reelection (Norris, Reference Norris1997; Wolak, Reference Wolak2017; Chiru, Reference Chiru2018). There is however less real-world evidence concerning the effectiveness of MPs’ personal vote-seeking strategies through work inside the parliament (Papp and Russo, Reference Papp and Russo2018). The purpose of this article is specifically to bridge this gap by asking whether MPs’ behaviors and performance within the parliamentary arena are related to their chances of staying in office.
More precisely, our ambition is twofold. First, we question the plausibility of the hypothesis that incumbents’ electoral performances are related to their past parliamentary activities. How likely is it that MPs’ electoral results would depend not only on their party and ideological stances, but also on how active these MPs were during the previous term? We hypothesize that, all things being equal, hard-working parliamentarians receive more votes than the less active ones. This hypothesis is tested in one parliament where the electoral impact of parliamentary work is a priori uncertain, namely the French National Assembly. French MPs are elected through a candidate-centered electoral system, which theoretically allows voters to sanction or reward MPs individually. On the other hand, the National Assembly is usually seen as a weak institution where individual work has a limited policy impact. Besides, the results of the legislative elections are generally interpreted as a mere confirmation of the presidential elections, not as a retrospective evaluation of the assembly’s record (Dupoirier and Sauger, Reference Dupoirier and Sauger2010). These contradictory characteristics make the French National Assembly a good test case to ask if, and to what extent, MPs can be held accountable, individually, for their parliamentary performance.
Second, this article seeks to contribute to the study of parliamentary representation by questioning what parliamentary activities matter for the reelection prospects of incumbents and under what conditions. The emerging literature on the electoral impact of parliamentary work tends to restrain itself to one indicator (bill initiation) or to one type of activity (legislative work). However, the functions of parliaments – from representation to control through deliberation and legislation – are manifold and, consequently, the diversity of MPs’ activities is potentially unlimited. It is therefore necessary to investigate the effects of various activities and behaviors for individual careers as well as the effectiveness of various individual vote-seeking strategies. Concretely, this article tests the electoral consequences of a plurality of indicators reflecting the great variety of parliamentary functions and activities.
Our empirical analysis focuses on the reelection of French MPs for the 2007 legislative elections. The French case constitutes an excellent benchmark for the electoral impact of parliamentary activities because of its conflicting institutional characteristics, in particular a weak parliament, a candidate-centered electoral system, and the timing of the election. We bring evidence of the electoral benefits associated with hard work in the parliament, but we also uncover the contrasting effects of parliamentary activities. Not all activities are equally rewarding and, contrary to our expectations, leadership positions within the assembly do not increase incumbents’ reelection probability. Our empirical analysis of French MPs’ in-parliament records and electoral performances gives a nuanced picture of how electoral accountability mechanisms work in an executive-dominated democracy.
The remainder of the article is organized as follows. The next section reviews the literature on the electoral consequences of MPs’ personal vote-seeking efforts; it also discusses the specificities of French political institutions. The third section introduces the data and methodology, while the fourth section describes the empirical findings. The article concludes in the fifth section with a discussion of the significance of this research.
Theoretical framework
The consensus in legislative studies is that MPs act to maximize their reelection chances (Mayhew, Reference Mayhew1974). However, whether the voters reward the incumbents’ efforts within the parliamentary arena remains unclear (André et al., Reference André, Depauw, Shugart, Martin, Saalfeld and Strøm2014: 242); this question relates to the debate on personal vote.
Electoral rewards for personal vote cultivation
A broad literature suggests that the performance of candidates in legislative elections does not exclusively depend upon their partisan affiliation. The personal vote, defined as ‘that portion of a candidate’s electoral support which originates in his or her personal qualities, qualifications, activities, and record’ (Cain et al., Reference Cain, Ferejohn and Fiorina1987: 9), is a major factor affecting electoral outcomes, even more so in the context of a growing personalization of politics (McAllister, Reference McAllister, Dalton and Klingemann2007). Voters do not only vote for a party, they also express a preference for a person. In the French case, which is the focus of our study, voters’ evaluation of candidates’ personality and efficiency seems to affect their voting behavior (Brouard and Kerrouche, Reference Brouard and Kerrouche2013).
Beyond specific personal traits, individual parliamentary activities have been hypothesized to increase the support that incumbent legislators receive. These include the legislators’ home style (Fenno, Reference Fenno1978), their constituency service endeavors (Norton and Wood, Reference Norton and Wood1990), pork-barrel favors in committees (Cain et al., Reference Cain, Ferejohn and Fiorina1987), policy positions consonant with the constituency in roll-call votes (Carey, Reference Carey2009), and the choice of specific role models (Martin, Reference Martin2010). These factors of personal vote have more to do with incumbents’ efforts directly geared toward their voters and constituencies than with their overall efforts within the parliamentary arena. How the level and type of individual parliamentary work could enhance their reelection chances has not been the subject of much scrutiny (Volden and Wiseman, Reference Volden and Wiseman2014: 199–200).
Through individual parliamentary activities, MPs may attempt to cultivate a personal vote amongst their voters and gain additional support from their party. This hypothesis lies on the assumption that the amount of work in the parliament gives the voters an indication of the incumbents’ efforts to represent them. Although parliamentary activities have arguably little policy value, they are expected to allow MPs to gain visibility in their constituency and establish a reputation of being a responsive representative (Teuber, Reference Teuber2018). Showing one’s willingness to change legislation by introducing bills may indeed prove efficient in increasing one’s media presence and vote share (Bowler, Reference Bowler2010). The effect is amplified by the growing attention to parliamentary work and an increased number of parliamentary monitoring organizations benefiting from the development of the Internet (Edwards et al., Reference Edwards, de Kool and Van Ooijen2015).
There are many ways for MPs to cultivate a personal vote through parliamentary activity. The existing research remains limited and partially inconclusive. Because it tends to limit itself to one activity indicator, bill initiation generally speaking (Bowler, Reference Bowler2010), it is unable to establish what strategy, if any, is electorally successful. With the partial exceptions of Däubler et al. (Reference Däubler, Bräuninger and Brunner2016) and Bouteca et al. (Reference Bouteca, Smulders, Maddens, Devos and Wauters2019) for Belgian deputies and Navarro (Reference Navarro2010) for members of the European Parliament, there has been no systematic attempt to compare the effects of different types of parliamentary activities.
Since MPs face time and resource constraints, they may be forced to choose between alternative activities (Blomgren and Rozenberg, Reference Blomgren and Rozenberg2012). These patterns of resource allocation reflect differences in the efforts MPs put into the pursuit of their various goals (Strøm, Reference Strøm, Blomgren and Rozenberg2012). That MPs proposing very few (or no) bills are reelected does not necessarily mean that voters do not take the MPs’ records into consideration: it might be that these MPs are rewarded because they did something else that the voters value more than bill initiation. It is thus crucial to compare the effects of different dimensions of parliamentary work, which may correspond to competing vote-seeking strategies and could ‘compensate’ for each other.
The uncertainty surrounding parliamentary roles renders the question of their electoral impact particularly relevant. What tasks MPs have to carry out in the accomplishment of their mandate remains partially undetermined. Both citizens (Bengtsson and Wass, Reference Bengtsson and Wass2010) and parliamentarians (Hall, Reference Hall1998; Harden, Reference Harden2015) have multiple (and sometimes contradictory) expectations about what MPs should do. Although they do not necessarily converge, the views of voters and those of MPs are not unrelated. In the contexts of the US and the UK, the citizens’ representational preferences influence their approval of their representative: the citizens’ job expectations condition the effects of members’ legislative activities on their job approval (Grant and Rudolph, Reference Grant and Rudolph2004; Carman, Reference Carman2006). Taking France as a test case, the challenge for us is to assess the electoral consequences of different parliamentary practices.
Expected rewards of parliamentary activities in the French political context
France constitutes a pertinent case study because of the ambiguous expectations that its political institutions generate. The extent to which the individual activity of French MPs could affect their electoral fortune is not evident. There are three reasons for this: the voting system based on single-member districts, the weakness of the parliament, and the timing of the legislative election after the presidential one. We assume these features of French politics to have differentiated and opposite impacts on the personal vote.
Members of the National Assembly are elected through a single-member district system with a run-off election in every constituency where nobody wins a majority in the first round. This places France in an intermediary place for the degree of candidate-centeredness of its electoral system, which should in principle facilitate an effective control of what MPs do (Farrell and McAllister, Reference Farrell and McAllister2006). This system allows personal considerations to play a significant part in the voters’ decisions. The latter is not as restricted as with a first-past-the-post system where the voters have, so to say, more incentive to vote strategically; on the other hand, the vote is much more personalized than in a closed-list system.
The French political system has other features that should diminish the chances that MPs’ personal records have an effect on their electoral outcomes. The most obvious is the domination of the system by the executive. In France’s semipresidential system, Parliament’s powers are severely limited by the constitution and the National Assembly is considered one of the weakest in Europe (Kesselman, Reference Kesselman, Kesselman, Krieger and Joseph2007). The Cabinet, rather than Parliament, is in control of the legislative process and can require priority for bills it wishes to promote. This second institutional element should reduce the personal dimension for an MP’s reelection. Given the weakness of Parliament and consequently the lack of impact of MPs’ activities, the voters should not take into account MPs’ activities when they make their voting decision.
A third feature of French political life should also diminish the probability that the results of legislative elections vary according to MPs’ individual merits: the timing of the presidential and legislative elections. Since 2002, the latter have been held immediately after the presidential elections. With the reduction of the mandate of the President from 7 to 5 years in 2000, it has become highly unlikely that they would not be organized the same year with the former following the latter. For this reason, the legislative election can be viewed as a mere confirmation of the presidential election (Evans and Ivaldi, Reference Evans and Ivaldi2017), which weakens the hypothesis of an electoral accountability mechanism for individual MPs.
To sum up, whereas the voting system should enhance the personal accountability mechanism, the other two major institutional characteristics – the weakness of the parliament and the timing of the election – are expected to deter it. Because of these contradictory assumptions, the net final influence of the parliamentary activities of MPs on their reelection cannot be theoretically predicted and rather begs an empirical approach.
Data and methodology
We rely on the French legislative elections to test the relationship between MPs’ parliamentary activities and their electoral performance. In what follows, we discuss in detail the data selection, statistical methodology, measurement strategy, and choice of control variables.
The case of the legislative election of 2007
Our empirical work scrutinizes the legislative election of 2007 and the twelfth legislative term covering parliamentary activities from 2002 to 2007. In 2007, there were 554 incumbent MPs in metropolitan France.Footnote 1 Among them, 455 (82%) stood as candidates for a new mandate, of whom 379 (83%) were successful in being reelected. A large proportion of the incumbents – around one-fifth – had thus abandoned their mandate before the new election,Footnote 2 and the same proportion failed to be reelected. To be reelected therefore involves passing two consecutive tests successfully: to stand as a candidate and to receive enough votes in the electoral competition.Footnote 3
There are two reasons why we focus on a unique election (and parliamentary term). First, at a technical level, data produced by the National Assembly regarding the individual activities of MPs is much more detailed during the twelfth legislative term than during the periods preceding or following it. For example, after 2007, the services of the National Assembly stopped providing the number of reports by subcategory (either legislative or information reports) that each MP had written. The additional details available for the 2002–07 term enable us to gauge the effect of a greater variety of activities than would be possible with other terms. As a matter of fact, we will show that the different categories of reports do not have the same effect.
Second, the twelfth legislative term precedes the 2008 constitutional reform and the ensuing changes in the National Assembly Rules of Procedure that have significantly modified the workings of the assembly (Rozenberg, Reference Rozenberg2016). For example, the committees and the rapporteurs play a greater role in the legislative process at the expense of the plenary. Following the 2008 reform, the National Assembly has also gained the possibility to submit its bills to the Council of State and, consequently, the bills’ success rate has increased significantly. In other words, it is virtually impossible to standardize the measure of MPs’ individual activities across the legislatures preceding and following the 2008 constitutional revision, and this makes it impossible to include later periods in the same model.
In sum, the focus on the twelfth legislative term enables us to avoid two major drawbacks: the use of poor data on MPs’ activities and the comparison of activities across different institutional contexts. That said, and beyond data availability, the findings on the 2002–07 legislature may have one additional advantage. While the 2008 reform left the main features of the French political system unchanged, it involved an increased autonomy of the parliament and more individual leeway for its members. If the connection between parliamentary activities and elections manages to pass the test of the 2002–07 term, it is likely to work in later terms as well.
Statistical methodology
Due to the specificities of the French electoral system, there are various possible measures of the incumbents’ electoral performance. In theory, an incumbent candidate can face four alternative scenarios: reelection or elimination in the first round of the election, or, if going to the second round, defeat or reelection in this final round. Consequently, the electoral system enables assessing the electoral impact of legislative activities in both rounds of the election, and it is possible to take into account the electoral results through either the percentage of votes received or the final election outcome. In view of keeping the methodology as simple and efficient as possible, the subsequent analysis relies on two distinct measures of the electoral performance. The first measure is the percentage of votes received in the first round of the election (model 1 hereafter): this gives a fair idea of the electoral support the incumbents enjoy and is more informative than their score in the second round, which only some incumbents reach. Second (model 2 hereafter), the impact of parliamentary activities on the final outcome of the election is modeled by a binary variable indicating whether or not the MP is eventually reelected.
Since the dependent variables are not of the same type, we need two different statistical methods. We estimate collected votes through an ordinary least-squares method of estimation due to the linearity of the variable. By contrast, the probability of being reelected, measured by a binary variable, implies the usage of a logit model and a maximum likelihood estimation method. Beyond different dependent variables and estimation methods, we rely on the same specification described below.
The statistical analysis could face a potential selection problem related to the fact that only a part of the incumbents run again. We account for this by implementing suitable and alternative statistical methods that are detailed in the additional materials and show that the results are not altered by a potential selection bias.
Measures of MPs’ activities
Following from the discussion above, our central hypothesis bears on the amount of work inside Parliament as a factor affecting electoral support for the incumbents. The aim is to assess which elements of parliamentary activity, if any, are related to electoral outcomes.
We therefore rely on a plurality of indicators reflecting the different means French MPs have to influence policy and to build up their personal reputation. Due to the diversification of the activities French MPs have to perform (Lazardeux, Reference Lazardeux2009), these means are varied (Kerrouche, Reference Kerrouche2006) and MPs display a great heterogeneity in their propensity to use them (Navarro et al., Reference Navarro, Vaillant and Wolff2012; François and Weill, Reference François and Weill2016). Our first criterion in establishing the list of indicators is to grasp the plurality of Parliament’s functions, namely linkage and representation, oversight and control, and policy-making (Kreppel, Reference Kreppel and Caramani2014). Both the voters and the parties can expect their representatives to perform – collectively and individually – these functions. Our second criterion is that the chosen indicators correspond to the most influential and visible aspects of parliamentary work, as least from the MPs’ viewpoints. Our indicators of parliamentary activity can be clustered into two categories: quantitative indicators of in-parliament activities, on the one hand, and committee membership and leadership positions, on the other hand (Table 1).
Table 1. Descriptive statistics of parliamentary activity
The indicators ‘per day’ are the count of each activity divided by the length (in days) of the MPs’ mandates.
The different types of questions to the government constitute the first indicators of parliamentary activity. Parliamentary questions do not necessarily serve the purpose of reelection (Lazardeux, Reference Lazardeux2005); however, about one-third of the MPs see them as a useful parliamentary instrument (Vigour, Reference Vigour2013). In fact, we must distinguish between written and oral questions, as they do not have the same purpose, policy impact, or visibility. Written questions (the first indicator) can be sent to the ministers at any time. Although they generally lack public visibility, MPs see them as a more useful tool than the oral questions (the second indicator), which are very short and can only be asked during the plenary sessions. The oral questions are covered live by the public television broadcaster, which guarantees a broad media coverage to this specific parliamentary activity despite their limited absolute number.
The third indicator corresponds to the number of ‘information reports’ produced by each MP. This type of report addresses a specific topic, for which Parliament wants to have in-depth information and for which one MP (or sometimes more than one) is appointed. As for the parliamentary questions, Table 1 displays a high level of interindividual variance.
The three last indicators of in-parliament activity are strictly related to the legislative process. Earlier studies treated the introduction of bills as a strategy for MPs to increase their reelection chances (Bowler, Reference Bowler2010; Däubler et al., Reference Däubler, Bräuninger and Brunner2016). In the French context, two categories of bills (propositions de loi) must be distinguished depending on how many MPs supported them: there is a distinct indicator for ‘single-author bills’ and another one for ‘co-signed bills’. The latter indicate how many bills introduced by other MPs have been co-signed by each MP. There are logically almost 10 times more co-signed bills than single-author bills.
The last indicator of legislative output – the number of legislative reports that each MP has been in charge of – also relates to the parliament’s law-making function. By contrast to the information reports, the scope of this type of report is more specific but it is directly a part of the legislative process. According to the MPs, to be a rapporteur is the single-most influential position in the law-making process (Vigour, Reference Vigour2013: 227).
For all these indicators, two types of measurements are possible.Footnote 4 Indeed, although most MPs were elected in 2002 and completed a full 5-year term,Footnote 5 some did not start their mandate at the beginning of the term.Footnote 6 To reflect these different mandate durations, all measures of activities are divided by the total number of days as an MP, as in Table 1. Otherwise, we cannot compare the MPs as they carry out their parliamentary activities over periods of different durations. The rest of our analysis relies exclusively on this daily measurement, which better reflects the real degree of involvement in Parliament. This strategy means that we assume voters to be sensitive to their MP’s productivity rather than to their absolute volume of activity (i.e., their production). Indeed, a ‘per day’ measurement consists of the volume of production given the time resources. In contrast, working on the ‘raw’ measurement of activities would suppose that voters do not take into consideration the time available for MPs to do their job. We discuss this point more thoroughly in the additional materials provided with this article, where we show that voters are more sensitive to productivity than production and that the estimation results differ when using the raw volume of MPs’ activities (see Appendix A3 in the supplementary material).
Another time-related issue concerns the timing of the MPs’ activities. Unfortunately, we are unable to discriminate precisely MPs’ production over time. We must therefore assume that there is no strategic manipulation of the indicators by MPs over time and that the parliamentary work is rather smooth all along the legislature. In the Appendix A4 of the additional materials, we examine this question carefully and carry out some additional statistical analyses that confirm the limited effect of time and make us confident about the validity of the assumption.
Beyond the parliamentary instruments available at the individual level, our indicators capture the offices that the MPs hold within Parliament: such offices enjoy a good public visibility and are hypothesized to influence the incumbents’ careers (Hermansen, Reference Hermansen2018). Committee membership in particular plays a positive role in the reelection of MPs in many countries (Golden and Picci, Reference Golden and Picci2015). French MPs judge that committee work is the most important tool at their disposal and they regard committee chairs as particularly influential (Vigour, Reference Vigour2013). We therefore control for MPs’ membership in one of the six standing committees covering a broad policy domain. Although there is no official hierarchy between these committees and it is impossible to formulate strong theoretical predictions about their impact, there are cues suggesting that they have an unequal prestige and influence on the policy-making process, with the committee of budget (which has a general supervisory role) and the committee of laws being the most influential (Türk, Reference Türk2005). We finally include a dummy variable for leadership positions within the standing committees as they can be expected to have a positive effect on incumbents’ visibility and electoral fortune despite their limited formal powers (Sieberer and Höhmann, Reference Sieberer and Höhmann2017). Table 1 shows that 8% of the MPs have a leadership role in a committee.
Before analyzing their impact on incumbents’ reelection, we observe that the main independent variables are only weakly correlated with each other. Whereas the amount of bills proposed and questions asked, respectively oral and written, have the highest correlation coefficients, 0.3 and 0.4, all other coefficients remain far below 0.2. Generally speaking, the descriptive statistics point to huge disparities for the indicators of parliamentary output, probably echoing a high variance of investment made by MPs.
Control variables
A number of control variables are added to the main explanatory factors.Footnote 7 The first subset of control variables comprises personal characteristics describing the incumbent candidates, such as their age, gender, local roots,Footnote 8 tenure (measured by the number of past parliamentary mandates), local mandates, local executive responsibilities,Footnote 9 and ministerial experience; with the exception of tenure, all the variables are binary. The second subset corresponds to the dichotomous variables indicating MPs’ partisan affiliations (the largest group in Parliament, the right-wing UMP Union pour un Mouvement Populaire group, is the reference variable). The last subset of variables contains information about the constituency characteristics, in particular the degree of electoral competition. Here the first variable indicates the margin of victory of the incumbent in the previous election (in 2002).Footnote 10 The second variable is the money per registered voter spent by the incumbent during the electoral campaign. The next two variables serve to measure the competition intensity: the number of candidates in the first round of the election and the average campaign spending per registered voter of the other candidates. Finally, the unemployment rate in the department controls for the economic component of the vote.
Results
The empirical analysis broadly confirms the hypothesis of a relationship between the parliamentary activities and MPs’ careers. In detail, the impact of MPs’ personal parliamentary records varies depending on the type of parliamentary activity and on whether we consider the vote share or the reelection likelihood.
Effects of MPs’ activities on the first-round vote share
As our variables of interest are defined per day, it is more suitable to compare and comment on the estimates’ magnitude through elasticity that is computed at average observation. Moreover, we calculate the magnitude of the significant factor for an average MP who holds his or her mandate for 1144 days on average, in order to provide more intuitive results.
We begin with the votes in the first round of the election (model 1). The parliamentary questions – written and oral – have a highly significant and strong effect on the votes, but the effects diverge depending on the type of question. The written questions have a negative impact: a 1% increase in the written questions variable is associated with a decrease of 0.008% of the vote share. The impact of the oral questions is positive: an increase of 1% of the oral questions per day is associated with an increase of 0.03% of the vote share. As, on average, MPs ask 231 written questions and 6 oral questions during their mandate, an increase of 1 written (oral) question leads to a decrease (increase) of 0.002 (0.15) percentage point of an average MP’s share vote. This finding is consistent with earlier observations that written questions lack public visibility and are a way for MPs to obtain information, not to raise politicized matters or to foster electoral connection (Lazardeux, Reference Lazardeux2005). Electorally, the oral questions, which take place in the plenary sessions and are broadcasted on public television channels, are much more efficient.
Beyond the written and oral questions, the single-author bills have – unlike the co-signed bills – a strong impact on the votes obtained by incumbents. A 1% increase for this variable results in a 0.01% increase of the vote share. The average MP introduces 3.4 bills during his or her mandate; each additional bill is associated with a 0.2 percentage point increase in vote share. The voters clearly reward the MPs who can be identified as having taken the initiative for new laws. The overall magnitude of the MPs’ activities is illustrated by Figure 1, which shows the predicted vote share in the first round according to the three significant variables.
Figure 1. MPs’ activities and predicted 1st round votes. (a) Effect of oral questions. (b) Effect of written questions. (c) Effect of single-author bills per day. The continuous line represents the predicted votes and the dashed ones the 5% confidence intervals.
Contrary to our hypothesis, belonging to a committee’s bureau has no significant impact on the first-round vote share. Regarding committee membership, only the members of the foreign affairs committee benefit from an increased vote share compared to the members of the economy committee. These observations contradict the idea that internationally oriented activities are of little electoral value for the parliamentarians (Rozenberg, Reference Rozenberg2018).
In summary, the incumbents’ vote shares are correlated to their work in Parliament. However, only some activities influence the vote share and, even though these influences are statistically significant, their magnitudes are limited.
Effects of MPs’ activities on their reelection probability
Then, we focus on the final probability of reelection, which normally takes place in the second round of the legislative election, but sometimes occurs in the first one (model 2). The effect of parliamentary activities differs at this stage, compared to that of the first round.
Whereas three types of activities correlate significantly to the first-round results, only two have a significant impact on the final outcome of the election. Both types of parliamentary questions lose their explanatory power. Surprisingly, the number of legislative reports per day becomes significant and positive, with an average elasticity of 0.02, meaning that a 1% increase in legislative report production implies a 0.02% increase of the probability to be reelected. The single-author bills variable keeps its positive and significant impact. A 1% increase in bills per day leads to a growth of 0.02% of the reelection probability. As previously, we illustrate the impacts of MPs’ activities on the predicted probability in Figure 2. As MPs produce on average 3.4 single-author bills during their mandate (which lasts 1144 days on average), we observe that their probability of being reelected is, on average, about 0.84 and, if they double their production, their reelection probability reaches 0.86. The ratio is similar for the legislative report, starting from the average production, a twice increase shift in the reelection probability from 0.84 to 0.86.
Figure 2. MPs’ activities and reelection probability. (a) Effect of single-author bills per day. (b) Effect of legislative reports per day. The continuous line represents the predicted probabilities and the dashed ones the 5% confidence intervals.
Turning to committees, we observe that the members of the committees on ‘law’, ‘culture and education’, and ‘national defense’ have a lower probability to be reelected than the members of committees on ‘economy’ and ‘foreign affairs’. There is, however, no clear underlying logic in this finding, as it is difficult to distinguish a rationale based on the MPs, the parties, or the voters behind it. As before, the administration of a committee does not increase the probability of reelection. While this type of leadership position can contribute to boost the career of some MPs, they are usually allocated following a seniority rule (Rozenberg, Reference Rozenberg2016: 29); MPs tend to reach such positions just before they decide to retire.
From this, we conclude that the reelection probability of the incumbents is also correlated to their work. While both first-round vote share and overall reelection probability are related to single-authored bills, the significance of other activities differs between the two election stages.
Ancillary estimation: successful bills
Finally, we conducted one additional analysis to assess the impact of successful bills (Table 3). The inclusion into our specifications of a dummy variable indicating that the MP had at least one single-author or co-signed bill successfully turned into law strengthens our findings. We therefore shift the perspective about the factor: from MPs’ activities (with the measure of law proposed) to MPs’ success (with the variable of laws passed). All 554 MPs in the dataset had co-signed at least one bill, but only 4% managed to validate a law. For the 344 MPs single-authoring a bill, the success rate reaches 6.4%. We rely on a binary variable that takes the value 1 if one or more of these legislative bills have been successful and 0 in all other cases. The detailed outcomes could be obtained upon request.
Table 2. Influence of ‘per day’ measure of parliamentary output on MPs’ vote shares (first round) and reelection probability
Number of observations = 455; ***, **, and * respectively mean different from zero at the 1%, 5%, and 10% threshold.
The SEs are corrected by the cluster method to take into account unobserved heterogeneity associated with political groups.
This additional factor has a positive and strongly significant effect (at least at 10% threshold) on both the proportion of votes in the first round and the reelection probability. Figure 3 shows the magnitude of the relationships. The predicted vote share of MPs without any successful bill is 43%, while with a successful bill it reaches 45%. The increase of probability is larger: from 0.8 without to 0.9 with. Since it is common in France for a law to be labeled by its initiator’s name, the statistical effect of this additional variable suggests that incumbents are rewarded for having a law named after them. The effect may be twofold: on top of the personal visibility given by a law named after oneself, the successful bills also give an indication of the more general efficiency and political skills of their initiators, which the voters want to reward.
Figure 3. Successful bills and MPs’ electoral fortune. (a) Effect on predicted votes (first round). (b) Effect on predicted probability of reelection.
Additional comments
Our findings offer new insights on the electoral accountability of French MPs. Most indicators of activity are significant at some point and the impact is almost always positive, meaning that hard-working MPs are rewarded. The only exception is with the written questions. The latter have a negative effect on the percentage of votes in the first round. The written questions probably lack the public visibility and prestige that could make them electorally worthy, and they are easily accessible to all MPs. They seem to be left to the MPs short of the personal contacts that could save them the effort and the formality of a written procedure (Lazardeux, Reference Lazardeux2005).
Besides, the coefficients of most control variables have the expected sign, although some of them are not statistically significant (Table 2). A candidate’s party logically has an impact on both the first-round votes and the probability to be reelected. An increase in the intensity of the electoral competition – measured by the number of candidates and the campaign spending of the other candidates – reduces both the vote share in the first round and the reelection likelihood. Contrary to what is often asserted in the literature (Costa and Kerrouche, Reference Costa and Kerrouche2007), holding a local mandate does not improve the electoral results. One other surprising outcome concerns candidates’ local roots. To be a local candidate – that is, to be born in the department of the constituency – has no effect on the votes received in the first round and it diminishes the reelection probability: the coefficient is significant and has a negative sign in the estimation of reelection probability. The explanation may lie in the restrictive definition of the variable or in an electoral gain for national personalities supported by the newly elected President.
Table 3. Influence of parliamentary output on MPs’ fortune, including passed laws
Number of observations = 455; ***, **, and * respectively mean different from zero at the 1%, 5%, and 10% threshold.
The SEs are corrected by the cluster method to take into account unobserved heterogeneity associated with political groups.
Conclusion
This article has analyzed the effectiveness of work inside Parliament as a vote-seeking strategy. Taking the French legislative elections of 2007 as a test case, we hypothesized that the incumbents’ electoral results depend on their activities in the parliamentary arena. Our findings confirm this hypothesis: Even when controlling for a number of contextual and personal factors, an incumbent’s parliamentary record has a significant impact on what happens afterwards. Almost all indicators of work in Parliament are significant at some point, and the effect is almost always positive, meaning that hard-working MPs are rewarded for what they did in Parliament.
This finding could well alter the way academics – as well as political actors – assess the impact of parliamentary activities, even more so in the French-specific context characterized by a weak parliament. The observed correlation between parliamentary activities and electoral outcomes brings support to the recent trend in the literature on the French Parliament to recognize a more significant role of the parliament and the parliamentary work that has been done before (Brouard et al., Reference Brouard, Costa, Kerrouche, Cole, Meunier and Tiberj2013a). On the practical level, MPs may well have to revise their vote-seeking strategies. To this day, MPs who think that they are responsible for their reelection tend to be more locally oriented (Brouard et al., Reference Brouard, Costa, Kerrouche and Schnatterer2013b: 153). We demonstrated that a greater activity at the national level also had the potential to boost one’s chances to remain an MP, in the legislative term immediately preceding the 2008 constitutional reform. In the aftermath of this reform, which aimed to enhance the role of the parliament, it is likely that MPs have even more opportunities to increase their reelection chances by focusing on internal parliamentary work.
The statistical analysis also confirms that one may not rely on one single indicator of activity – in particular, bills – as a proxy for a wider measure of parliamentary work. Each distinct parliamentary task has its own logic and may bring differentiated electoral returns.
Our results contribute to a better understanding of fundamental accountability mechanisms in democratic systems (Pitkin, Reference Pitkin1967; Manin, Reference Manin1997). They show that, in the case under investigation, retrospective voting does not depend on the performance of incumbent majorities and governments, such as their economic performance (Healy and Malhotra, Reference Healy and Malhotra2013). To some extent, the personal vote in legislative elections can be understood as a retrospective evaluation of the parliamentarians’ overall activity within the legislative arena, and not only of their efforts to serve the specific interests of their constituencies. Of course, the electoral impact of legislative work is not massive, but it is real, and it probably hinges on the type of political and electoral system mediating the relationship between parliamentarians and voters in which the parliamentarians and voters evolve. Further comparative research is thus necessary to better explain the circumstances in which voters have the capacity and the willingness to hold their representatives accountable.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1755773919000274
Acknowledgements
The authors thank Zach Bastik, David Bounie, Thomas Däubler, Michael Holmes, Emiliano Grossman, as well as the participants of the workshop ‘Science Po Quanti’ at the Lille Catholic University (1/7/2016) and the journal’s referees for their helpful comments and suggestions. The usual disclaimers apply. This work benefitted from ANR funding (grant number ANR-18-CE26-0004).
