Dual Mandates, Patronage, and Partisanship in the French Third Republic Alexandra Cirone⇤ Abstract How does simultaneously holding two elected offices a↵ect individual incentives for party investment in a new democracy? In this paper, I show that a pro-democratic institution initially designed to connect local and national politics – cumul des mandats,ortheabilitytoholdbothalocalandnational elected office – had negative e↵ects on the development of the nascent party system in France. Using a microlevel dataset of electoral outcomes, roll-call votes, and politician characteristics from the Third Republic (1870-1940), I show that holding a local office gave national legislators access to an external political network and patronage opportunities that served as a substitute for the benefits a party would typically provide. Using a regression discontinuity design, I find that legislators with a dual mandate were less likely to invest in early party institutions, as measured by partisan voting and support of key legislative reforms that strengthen parties. Yet I also establish that such negative e↵ects can be mitigated with specific institutional changes that give selective incentives to parties. This paper’s contribution is to estimate the causal e↵ect of multiple office-holding on partisan outcomes, as well as present original results using a clear and precise measure of party investment. ⇤London School of Economics and Columbia University. Email: [email protected], [email protected]. I thank John Huber, Sylvain Brouard, Andy Eggers, Nikhar Gaikwad, Simon Hix, Francesca Refsum Jenisius, Isabela Mares, Pavithra Suryanarayan, Janne Tukiainen, Stephane Wolton, Daniel Ziblatt, as well as seminar participants from NEWEPS, the Historical Working Group, CES Workshop on Comparative Historical Research in Europe, LSE, Oxford, and Columbia. 1 1 Introduction Strong parties are critical for the proper functioning of a newly established democracy.1 Yet the strength of a party depends on its ability to incentivize legislators – parties can distribute financial resources, legislative office, and a party label in exchange for loyal and cohesive behavior (Aldrich 1995, Schelsinger 1966). Politicians with independent electoral resources, however, pose a significant challenge for party organizations. In particular, individuals with access to local patronage – as opposed to personal wealth or reputation – are much harder to discipline. Party strength is also a function of the evolving institutional environment. Elites can significantly shape the path of an early party system with long-term consequences for its consolidation (Tavits 2009, Cappocia and Ziblatt 2010, Lupu and Riedl 2013). Therefore it is important to understand how independent political resources a↵ect legislative outcomes, in both the disciplining of politicians and the strengthening of early party institutions. This paper studies the consequences of a specific channel of independent resources for a politi- cian provided by multiple office-holding, more commonly called a dual mandate. This is when an individual holds both a local elected position and a national legislative post (for example, if a per- son is simultaneously a mayor and a legislator).2 A dual mandate ensures, among other things, a steady supply of resources essential for reelection, including control over local government funding and jobs, access to manpower and voter networks, and a regional reputation. E↵ectively creating an “elite within an elite” (Navarro, 2009), this institution could potentially advantage an individual politician at the expense of a party. Here I examine how dual mandates a↵ect two salient dimen- sions of early political development: party voting cohesion, and the support of institutional reforms that strengthen parties. In doing so, I test the hypothesis that politicians holding two offices are less likely to invest in partisan endeavors. The French Third Republic, during its formative years of party institutionalization between 1881 and 1919, provides an ideal test for this claim. The era was characterized by both weak party 1Strong and cohesive parties lead to greater accountability (Powell and Whitten 1993), party institutionalization (Mainwairing and Scully 2005), and even regime stability (Linz 2000, Carey 2009). 2Some countries allow for the accumulation of three or more offices, but for ease of exposition I focus solely on being simultaneously involved in two levels of government. 2 organizations and a strong legislature,3 and featured a pivotal subset of independently resourced elites – between 30-60% of deputies in the National Assembly held dual mandates, or cumul des mandats. This institution provides a clear and consistent measure of independent access to po- litical resources. In addition, early 20th century France featured a period of voting on legislative reforms that empowered party institutions, including legalizing parties, giving them control of the committee system, and electoral law reform. These reforms allow for a precise and novel measure of investment in party strength, in addition to standard measures of voting cohesion. Yet identification is a significant challenge in studying political behavior as a result of a dual mandate. Multiple office-holding is not randomly allocated across legislators, and those who suc- ceed in obtaining a local office are systematically di↵erent than those who do not (in terms of skill, interest, or other unobserved factors). Therefore I evaluate the e↵ect of dual mandates on parti- sanship in a quasi-experimental framework. Following a growing literature in American politics4 as well as Bach (2011) in France, I use a regression discontinuity design to estimate the di↵erence in legislative outcomes for a district that narrowly elects a candidate with a local office and a district that narrowly elects a candidate without a local office. I utilize the fact that France had single member districts, and restrict the analysis to head-to-head races, in which a candidate with a local office was competing against a candidate without. In this case, the election of a national deputy with an additional local office can be conceptualized “as if” randomly assigned in order to estimate the e↵ect of a dual mandate on legislative behavior. I find that holding two offices decreases the frequency with which the deputy votes with their cho- sen party, as measured using yearly roll call votes of redistributive bills. Dual mandate deputies are much more likely to vote against reforms that strengthened fledgling party institutions. Exploring alternative explanations for these findings, I also empirically dismiss those relating to absenteeism or incumbency e↵ects. Finally, I explore two potential mechanisms behind the negative relationship between dual mandates and party investment. I exploit the timing of committee reform to provide evidence of how independent resources a↵ect party discipline, and use the passage of the separation 3Also, universal manhood su↵rage had been achieved and the electoral law stayed constant, eliminating potentially competing historical explanations for institutionalization (Cox 1987). 4Lee 2008, Caughey and Sekhon 2011, Eggers et al. 2015, Hall 2015, among others. 3 of church and state law in 1905 to demonstrate the salience of constituency ties for deputies with dual mandates. Importantly, I also show that if parties have control over selective incentives within the legislature, they can potentially change the incentives of deputies with dual mandates. While France is most typically remembered for its frequency of dual mandates, the practice is common elsewhere – Germany, Belgium, Finland, Ireland, UK, Denmark, Italy, Luxembourg, Hungary, Slovakia, and Israel all allow for multiple office-holding, and even more countries permit the practice at subnational levels.5 Relatedly, the use of dual mandates speaks to an important debate in political science. Policy advocates argue they ensure local interests are represented nationally, foster higher turnout and accountability, and play a role in career professionalization; while opponents claim dual mandates increase legislative absenteeism and unfairly entrench local elites. Institutions that promote vertical linkages are often suggested for countries wishing to promote integration (Borchert 2011), and recently dual mandates were advocated as policy tool in order to improve turnout and confer legitimacy to the parliament of the European Union (Lelieveldt 2014). Meanwhile, qualitative studies from Hungary and Slovenia also suggest that dual mandates may be particularly influential in countries with weak party systems (Zagorc 2008, Varnagy 2012). The next section reviews the theoretical expectations behind multiple office-holding, while Sec- tion 3 discusses the context of dual mandates in the French Third Republic. Section 4 outlines the regression discontinuity design and the data; Section 5 presents results and robustness checks, Section 6 discusses mechanisms, and Section 7 concludes. 2 Theoretical Expectations Existing empirical research has mixed expectations regarding how access to independent resources a↵ects partisanship. On the one hand, a number of previous studies have shown that individual control of patronage or personal wealth leads to dissent in legislative voting (Moser 1999, Reuter 2010, Nalepa and Carroll 2013) or weakly institutionalized parties (Kirchheimer 1966, Shefter 1977, Samuels 1999, Tavits 2009). On the other hand, political assets aren’t necessarily anathema 5Historically, dual mandates have also