One Tick and You re Out: The Effects of the. Master Lever on Senators Positions

Transcription

1 One Tick and You re Out: The Effects of the Master Lever on Senators Positions Olga Gorelkina University of Liverpool, Liverpool Ioanna Grypari Max Planck Institute, Bonn This Version: December 2016 Abstract This paper accounts for the effects of the master lever (ML), aka the straight-ticket voting option, on elected US senators from 1960 till The ML, still present in some states, allows voters to select a specific party for all elections listed on a ballot, as opposed to filling out each office individually. Introducing it leads to an increase in the number of partisan votes, and thus changes the groups of voters targeted by parties and shifts the positions of senatorial candidates. Theoretically, we examine this change in tradeoffs by building a model of pre-election competition. Empirically, we use a triple-difference estimator to account for selection into treatment and find that, controlling for party trends, the ML has led to a right-wing shift of senatorial positions; an effect that is larger for the Republican party. We use the theory to explain how the political climate, as observed in the data, has implied the specific result. Keywords: Ballot Design, Elections, Political Positions, U.S. Senate - JEL: D72, K16, N42 The authors thank Ying Chen, Paola Conconi, Mira Frick, Martin Hellwig, Johannes Hörner, Anna Kochanova, Rida Laraki, Derek Stemple, Stan Veuger, Maurizio Zanardi and participants of the MPI workshop; as well as Darya Babushkina and Georg Treuter for excellent research assistance, and the Cowles Foundation at Yale University for hosting Gorelkina in University of Liverpool Management School, Chatham Street, Liverpool L69 7ZH, UK, olga@liv.ac.uk. Max Planck Institute for Research on Collective Goods, Bonn 53113, Germany, grypari@coll.mpg.de. 1

2 1 Introduction The Senate is one of the two chambers of the legislative branch in the US. Each state is represented by two senators, who serve staggered six-year terms. In other words, in every midterm or Presidential election at most one Senate seat is up for a vote per state. In these years other offices are also contested, so that a voter is faced with a list of elections on the ballot, that differs across states. The master lever (ML) is an option that offers voters the possibility to vote for the candidates of a specific party for all elections listed on a ballot. For example, by selecting the Democratic party at the ML (or straight-ticket) option, the voter elects all Democratic party candidates on the ballot without needing to go through it office by office. 1 The existence of this option (a state-level decision) has been heavily debated in politics and the media and the number of states that offer it has been decreasing over time. The main argument used against the ML is that since in its presence voters are more likely to vote by the partisanship of a candidate as opposed to by her positions, this leads to a less informed electorate and a lack of accountability for the politicians. However, another side to this issue, and the focus of this paper, is that this increase in partisanship votes redefines the groups of voters targeted by political candidates. As partisans are more likely to vote through the ML, parties have an incentive to change their platforms so as to satisfy those in the electorate that are interested in the platforms of candidates, as opposed to only their party affiliation. Theoretically, we examine this change in tradeoffs through a pre-election competition model that allows for multiple issue dimensions. Empirically, we identify the effects of the straight-ticket option using a triple-difference estimator that controls for 1 a. The online Appendix includes a sample ballot from Alabama for the 2010 midterm election that offers a ML. b. For the non-partisan elections on the ballot (those whose candidates are not affiliated to a party) voting only through the ML counts as a non-vote. d. The voters have the option to select a party through the ML and still fill out the rest of the offices individually, if they so wish. 2

3 selection into ML status, and find that it is a significant predictor of elected senators positions, using data from 1960 to It leads to a right-wing shift in positions, that is larger for Republican senators. Lastly, we use the theory to explain how the political climate, as we estimate it from the data, has implied the specific outcome. The contribution of this paper is three-fold. Firstly, the forces that we identify can be due to any other exogenous change in partisanship votes in an electorate, be it from a change in ballot design, or another determinant. Therefore, our focus is beyond solely that of the ML. Second, the straight-ticket option is one that is technically easy to implement, in the sense that it is only an added box on a ballot, so that our findings can thus be considered from a policy perspective. In the Conclusion, we briefly discuss the implications of our results in terms of Congressional gridlock. 2 Thirdly, we provide a first connection between the literature on the determinants of the positions of elected officials, and that on the importance of ballot design on voting outcomes. The variation in ballot format across the US due to the ML provides for a natural experiment that links voting technology to political and economic outcomes. We start by building a static model of elections in a multi-dimensional issue space. For each office contested, parties nominate candidates to maximize their vote share which they discount by the distance of their candidate s position from the party bliss point. The latter represents the party s willingness to be united in order to be able to pass more policies, as well as an incentive to maintain ideological purity and consistency over time. Voters then elect a candidate for each office by making two choices. First, they decide whether to use the ML. If they do, they save themselves the cognitive cost of going through the entire ballot, but they cannot fine-tune their choices per office. 3 Second, if a voter has decided to use the ML, she selects a party, otherwise she votes individually for each office on the ballot. We model voters payoffs 2 As this is not the goal of the paper, we do not attempt to evaluate the ML, but merely discuss different aspects of the option that arise from our findings. 3 As they would be choosing one party for all offices, as opposed to a candidate per office. 3

4 from a specific candidate winning an election following Downs (1957) in that voters prefer candidates that are closer in their political positions. We introduce an added utility gain when a partisan voter elects a representative of her preferred party. 4 This implies that, facing two candidates with equivalent political positions, a partisan voter will strictly prefer the one who is affiliated with the party she supports. This setup creates two kinds of tradeoffs: 1. for the voters, of going through the ballot vs. using the ML and 2. for the parties, of picking a candidate that maximizes vote share vs. one that is more similar ideologically. In equilibrium, the first one is resolved so that partisan voters are more likely to use the ML than non-partisans (swings). The second tradeoff creates two ML effects on equilibrium senatorial positions. As introducing the straight-ticket option leads to an increase in partisan votes for both parties, they can afford to put forth candidates that are ideologically closer to the party - the party loyalty effect. This effect is always stronger for the party with the highest partisan support amongst the electorate. On the other hand, nonpartisan voters become more decisive as they vote only by position, so that political platforms will change to accommodate them better - the swing voter effect. If for example we are in a non-partisan state where swings are more likely to be right-wing, relative to the state s average, introducing the ML will lead to both parties putting forth more right-wing candidates to try to appeal to swing voters. The total effect of the straight-ticket option is determined by the level of partisanship in a state, and the distribution of political positions among partisan and swing voters. In order to identify it empirically, we employ the methodology of Ravallion et al. (2005) to account for selection into treatment. We separate US states into three groups: 1. participant stayers (ML always), 2. participant leavers (ML, then no ML) and 3. nonparticipants. We construct a triple-difference estimator (DDD) by taking 4 Note that we separate between partisanship and positions and allow for Republicans to be left-wing, for example. 4

5 the difference between the DD for stayers over nonparticipants and the DD for leavers over nonparticipants. This is equivalent to using the leavers as the treatment group. 5 Nonparticipant states are used in order to control for economy-wide changes that may affect stayers and leavers differently. The corresponding regression framework identifies the effect of the ML as long as: 1. there is no selection into abolishing the option and 2. there are no lasting effects on senatorial positions after it is removed. These conditions imply that leavers are trending similarly to stayers prior to ML removal and to nonparticipants afterwards. In Section 4.6, we show that these identifying assumptions are met. Controlling for state, Congress and party fixed effects and party trends, we find that the marginal effect of the ML is a right-wing shift in the positions of senators. Note that the inclusion of trends implies that we are not attempting to explain the increasing polarization in the Senate, 6 but rather we are looking at deviations from it. Across all specifications, the straight-ticket option explains more of the positions of the Republican party, whereas for Democrats, depending on the controls and senator sub-sample, it is not always economically significant. Our results are puzzling because first, there is nothing in the option itself that guarantees a larger benefit for the Republican party, and second, in our identification we also control for differences in selection bias into ML status by party. 7 Making a simplifying empirical assumption, we find that the distribution of variables present in the data is such that in most cases the party loyalty and swing voter effects are both positive for Republican senators (explaining the large positive ML effect), whereas for Democrats, they go in opposite directions. It is these forces implied by the political climate in our sample that seem to be behind our asymmetric findings. The rest of the paper is organized as follows. Section 2 summarizes the related 5 The treatment here being removing the ML. 6 See Section 2. 7 For this exposition, we assume that a right-wing (left-wing) shift in positions is beneficial to the Republican (Democratic) party. 5

6 literature and Section 3 presents the theoretical model and results. In Section 4 we discuss the data, identification strategy, our empirical findings and the mechanism behind them. Section 5 concludes and briefly considers the policy implications of our work. 2 Related Literature In this paper we account for the tradeoffs that senators, and the parties behind them, face in the presence of a ML, and document empirically its effects on the positions of elected US senators. As this has direct implications on the composition of Congress, our paper relates to the literature on political polarization. 8 Political polarization is defined as the gap between the positions of the Democratic and Republican parties at the mass and/or elite-level in some issue or ideological space. There is general agreement that political elites in the US are polarized. McCarty et al. (2006) show that the difference in mean positions between the two parties has continued to grow since the 1940s in both the House of Representatives and the Senate and that this increase is driven more by the Republican Party. There is an extensive literature on the causes of this phenomenon, 9 with a strong focus on the role of the voters. At the national level, there is no evidence of mass polarization and changes in the Congress composition have not been following changes in voter ideology, overall and across issues. 10 At the district-level, there is conflicting evidence on the importance of the electorate s preferences in predicting legislative behavior. Krasa and Polborn (2014) find that there is a stronger effect from politicians to voters, as opposed to the other way around, whereas Kirkland (2014) and McCarty et al. (2015) show that within state or district heterogeneity does lead to 8 To be clear, we do not aim to explain the increasing Congress polarization. Instead, we examine the importance of a common ballot characteristic on the type of senators that get elected. 9 See Barber and McCarty (2015) for a review. 10 E.g. Abramowitz (2010), Ansolabehere et al. (2006), Bafumi and Herron (2010), Fiorina et al. (2005), Fiorina and Abrams (2008), Tausanovitch and Warshaw (2013). 6

7 more extreme politicians. Harden and Carsey (2012) show that voter preferences can predict senatorial positions only in homogeneous states and that the voters party affiliation is a much more important determinant of political polarization. In fact, there is a growing literature on the importance of party sorting (the increased correlation between voter positions and their party affiliation) and within-state party strength in determining Congress members positions, with the direction of causality debated. 11 McCarty et al. (2006) and Garand (2010), among others, have examined the importance of income inequality in determining political polarization. In general periods of higher income inequality are associated with more extreme legislatures, that are also more right-leaning (Voorheis et al. (2015)). A lot of attention has also been given to the interaction of mass media with the electorate. 12 Campante and Hojman (2013) show that broadcast TV has led to more extreme Congress members, but Prior (2013) finds no significant evidence of partisan media, in particular, influencing voter preferences. On the other hand, Snyder and Strömberg (2010) point to the importance of media coverage, with higher coverage leading to less ideologically extreme Congress members. Other factors that have been considered as a source of the increasing political polarization have been gerrymandering (redistricting), midterm vs Presidential elections, characteristics of the primaries and different elements of campaign financing. Engstrom (2013) accounts for the importance of redistricting on a variety of political outcomes (competitiveness of elections, partisan control, etc), but McCarty et al. (2009) find little evidence of a causal relationship between gerrymandering and ideological extremism, specifically. Different types of elections have been studied as well, with Halberstam and Montagnes (2015) finding that midterm elections are in fact associated with more extreme 11 See Krasa and Polborn (2014), Layman and Carsey (2002) and Levendusky (2009), among others. 12 See Strömberg (2015) for a review of the literature. 7

8 senators, whereas there is growing evidence that primary elections are weak in explaining polarization. 13 Lastly, in terms of campaign contributions, Barber (2016) finds that higher donation limits on PACs 14 lead to moderate legislators, and a larger number of donations from individuals to more extreme. The effect of ballot design on polarization has not been considered, however there has been extensive research on the way it can affect voting behavior. Different ballot characteristics that have been examined are ballot secrecy (e.g. Heckelman (1995)), the ordering of names (e.g. Chen et al. (2014)), and the office bloc vs. party column ballot form (e.g. Walker (1966)). In terms of the ML in particular, the literature shows that as expected, its presence reduces the number of split tickets (voting for different parties for different offices), with varying effects depending on the party and seat up for election. 15 Note also that when selecting the ML, all partisan elections on the ballot 16 are automatically voted on, and all non-partisan elections are counted as non-votes, unless a voter specifically chooses a candidate for these offices as well. Feig (2007), Feig (2009) and Kimball et al. (2002) do find that the ML decreases voter roll-off 17 and Bonneau and Loepp (2014) show that it decreases participation in non-partisan elections. In terms of voter errors, Herrnson et al. (2012) demonstrate that a straight-ticket option increases the occurrences of people not voting for the candidate they intend to, and Kimball and Kropf (2005) find that it reduces over-votes, i.e. cases when voters mark too many candidates. For our purposes, the most important take-away from this literature is that the ML is in fact used by voters, and it matters even for offices further up the ballot See Hirano et al. (2010), McGhee et al. (2014) and Barber and McCarty (2015). 14 Political Action Committees 15 See Kimball et al. (2002), McAllister and Darcy (1992) and Rusk (1970), among others. 16 Those were candidates are affiliated to a party. 17 Lack of votes for offices further down the ballot. 18 In terms of which voters actually use the ML, Feig (2007) and Feig (2009) show that blacks are more likely to use it, and in a lab experiment Lewkowicz (2007) finds that Democrats and Republicans are equally likely to select the ML, but strong, weak and non-partisans in order have 8

9 This paper brings together these two strands of literature. We do not examine directly the effects of the ML on electoral behavior, but we go one step further and account for the importance of this characteristic of ballot design on the positions and parties of elected senators. We contribute to the political polarization literature by offering one more significant determinant of positions, one that has direct policy implications. 3 Theory 3.1 Setup Fix a US state and an election period and let the offices listed on a ballot be indexed by k, k K {1, 2,...K}. µ {0, 1} indicates the availability of a ML, or straightticket option, so that µ = 1 when a ML is present and µ = 0, otherwise. We consider a multi-dimensional policy space P [ 1 2, 1 2] N, where N is the number of policy issues, such as economics, national defense, social issues, and so on. Three types of actors are positioned within P: voters, parties, and candidates. Each party j {R, D} has a bliss point denoted by a vector of issue positions Y j (Y j1, Y j2,...y jn ) P. (1) Without loss of generality, we label the positions so that the Democratic party s bliss point is to the left of the Republicans in every coordinate, Y Dn < Y Rn, n {1, 2...N}. decreasing probabilities of using it. 9

10 Candidates are characterized by an office, k, the party they represent, j, and their own positions, y P. For each office, the pool of candidates is P and each party selects one candidate, y jk, to represent them and run for the seat (see (10) below). 19 There is a unit mass of voters, indexed by i, each with a bliss point given by x i, x i (x i1, x i2,...x in ) P. (2) Integrating over the mass of voters we obtain the average position in the state period X x i di P. (3) [0,1] Apart from their political positions, voters are characterized by their partisanship status. Let p i (j) denote the probability that voter i (whose position is x i ) is a partisan of party j. 20 The realization of the random variable is denoted by I P i ; I P i = 1 implies that the voter is a partisan, and I P i = 0 implies a non-partisan, or swing. 21 Assuming that everyone in the electorate can be a partisan of at most one party, we introduce the total probability of being a partisan 22 p i p i (j) + p i ( j). (4) Party j s partisan advantage in the state period is denoted p (j) = p (j) p ( j), where p (j) = [0,1] p i (j) di is the mass of party j partisans. p = [0,1] p idi = p (j) + p ( j) is the share of partisans. We also let X P n By analogy, denote the average position of partisans on issue n, 19 I.e. we assume that for each office, there are exactly two candidates running, one for the Republican and one for the Democratic party. 20 Alternatively, p i can be thought of as the mass of partisan voters within voter group i characterized by position x i. 21 From now on we will use non-partisans and swing voters interchangeably. 22 j = {R, D} /j. 10

11 X P n 1 p i x in di P. (5) p [0,1] A given voter s political positions and partisanship status are generally not independent, the covariance is given by (p i p) (x in X n ) di. In our analysis, we shall use its negative, namely the covariance between the voter s position on issue n and the likelihood of being swing, 23 σ n (p i p) (x in X n ) di. (6) If σ n is positive, then a non-partisan status is associated with a more right-wing position on issue n, as compared to the rest of the state. Similarly, σ n < 0 implies that the swing voters views on n tend to be to the left of the state s average. This variable plays an important role in the analysis of ML effects. Actions, Payoffs and Timing Our model of an election with a ML is a game between two parties and a mass of voters. The game proceeds according to the following timeline: t = 1 Party j chooses a candidate, y jk, to compete for seat k. The party derives utility from the share of votes it gets and incurs a loss if its candidate s positions differ from the party s bliss point. t = 2 Voter i decides whether to use the ML, if available; if she does not, she goes through the whole ballot incurring cost c i. t = 3 Voter i elects one candidate for each office, either indirectly (when using the ML) or directly (when going through the ballot). Conforming with the logic of backward induction, we analyze the game in reverse chronological order. 23 Here we use (1 p i ) (1 p) = (p i p). 11

12 t = 3 : Electing Candidates. If the voter uses the ML, she solves a single maximization problem for the entire ballot: max j {R,D} k=1,..k u ik (j) Ûi. (7) Let ĵ i {R, D} denote the problem s solution, so that ĵ i is the party that gives the maximum payoff to voter i in the election. If the voter goes through the ballot office by office, she solves a sequence of K distinct maximization problems: k=1,..k max u ik (j k ) Ui. (8) j k {R,D} Clearly, the more refined solution (j i1, j i2,...j ik ) {R, D}K yields greater utility to the voter: Ui 24 Ûi. We build upon the probabilistic voting framework of Lindbeck and Weibull (1987) to define the voter s utility in (7) and (8). n u ik (j) = ω n (x in y jn ) 2 + β k + ε ij, if i is a partisan of j, n ω n (x in y jn ) 2 + ε ij, otherwise. (9) The first component of (9), n ω n (x in y jn ) 2, is the disutility experienced by i if the candidate j s positions differ from i s bliss point, x i, where every issue n has weight ω n > 0. The second component is a partisanship bonus β k > 0, an extra payoff that the voter gets if the candidate from her partisanship party wins the election. 25 Thirdly, ε ij is a private preference shock, an advantage over the opponent j (ε ij = ε i, j ) that results from various factors such as presidential approval, 24 This is due to the restriction of the domain in (7) compared to (8). 25 The model predictions do not change if the utility function is modified so that electing a counterparty candidate yields a negative payoff to a partisan voter. 12

13 differences in personality traits, perceived competence, etc. 26 ε ij is uniform on [ 1 2, 1 2 and independent of (x i, p i ). 27 ] t = 2 : Voter s Choice to Use the ML. The voter decides whether to use the option by comparing the cost, c i > 0, and the benefit U i Ûi of making a better choice by going through the entire ballot. Here c i represents the cognitive effort and opportunity cost associated with solving K separate decision problems. 28 t = 1 : Party s Choice of Candidate. We model the party s problem as a tradeoff between attracting votes and satisfying its own policy agenda (ideological purity). The party solves the following optimization problem: max y j P { V j n γ n (Y jn y jn ) 2 }, (10) where we have dropped the office subscript k; V j is the share of votes for candidate j, and γ n > The first term of the maximization program reflects the driving force of political competition. The second term corresponds to the cost of being represented by a candidate whose views diverge from those of the party. In a wider interpretation, the term captures any force that may deter optimal candidates from converging to the 26 In a model where parties maximize their winning probability, as opposed to vote share, we would also introduce an aggregate candidate-specific shock η j. 27 For a discussion of the framework with a continuum of random variables that are conditionally independent we refer the reader to Hammond and Sun (2008). 28 There is a long literature on choice fatigue. Most recently Augenblick and Nicholson (2016) show that agents incur a cost when filling out an entire ballot. 29 We can think of the party s { global election problem, i.e. the problem where the party cares about all seats k K: max y k π ke i Pr (j k i j k ) k,n γ jkn (Y jn y jkn ) 2} with some weights jkn π k, γ jkn. Due to the additive separability of the said global election problem in y jkn, we can focus directly on (10). Note that for our purposes, i.e. the senatorial election, the K 1 other offices are only important in that they create a tradeoff for the voter between going through the ballot and using the ML. In principle, as long as all optimization programs are independent across offices, they could be different than the one stated here. For example, US state ballots may also include small local offices (e.g. District Attorney), bill proposals (e.g. Proposition 8 in California), and other elections that may be susceptible to different dynamics than the ones we have here. 13

14 midpoint as implied by the standard Downs (1957) model. We assume that parties can freely select any candidate, y j P. Alternatively, their choice may be constrained to a candidate that would be able to pass through primary elections first, resulting in a position y j confined to a subset of the policy space, P. We do not study this possibility here. 30 Moreover, in practice, when selecting a senatorial candidate a party may also account for the congress member in power (of the Senate seat that is not up for election). We abstract from this consideration as well Assumptions. We normalize the issue weights n ω n = n γ n = 1 and set the partisanship utility payoff β < 1, to guarantee that the solutions are interior. We assume α n 2 γn ω n 1, for all n, implying that there is not too much disagreement between the parties and the voters on the issues relative importance. 33 To simplify the exposition, we also assume that the cost c i of going through the ballot satisfies the following double inequality: Ui IP ( ) ( ) Ûi c i U IP i =1 i Ûi. i =0 This implies that the voter uses the ML if and only if she is a partisan. Qualitatively, our results do not depend on this assumption. In the online Appendix, we consider an alternative set-up where c i is not constrained to the above interval and is distributed independently across the population of voters. In that framework, the partisans are only more likely to use the ML than the swing voters and the effects of the option are the same as the ones reported in Section In fact, Hirano et al. (2010) and McGhee et al. (2014) find little evidence of primaries affecting the polarization (and thus the positions) of elected officials. 31 In the robustness checks of the empirical exercise we include the party and position of the senator in power as controls. They are insignificant in all specifications. 32 In the case of incumbent senators the expected shares of votes would be higher than of any other senator of the policy space. This would make the party more likely to nominate an incumbent, but the main tradeoff would remain. 33 Note that if parties and voters weigh issues equally then α n = 2 > 1. 14

15 3.2 Master Lever Effects The change in ML status observed in the data provides an exogenous variation in µ {0, 1}. 34 In this section we study the model s solutions with and without the ML and deduce its effects on three outcomes: candidates platforms (Proposition 1), their vote shares (Proposition 2), and the expected platform of the election winner (Proposition 3) Candidates Platforms We start by characterizing the optimal platform derived as a solution to the threestage game. Note that since the choice set of candidates is unconstrained (i.e. it is the whole policy space, P), the party s choice of an optimal candidate is equivalent to simply picking an optimal position. Proposition 1. The optimal position for the candidate of party j on issue n, y jn, is a convex combination of the average voter position in the state X n and the party s bliss point Y jn, with a drift proportional to the swing-position covariance σ n (6). y jn = 1 µp 1 µp + α n X n + α n 1 µp + α n Y jn + µσ n 1 µp + α n (11) for all n, where α n 2γ n /ω n. Introducing a ML increases the weight of the party s bliss point, Y jn, and the effect of the swing-position covariance, σ n. The statement follows directly from 11, derived in the Appendix (see Proof of Proposition 1 in Appendix A.1). Recall that introducing a straight-ticket option results to all partisan voters using the it, and all swings voting by position. Thus, the effects of the ML above are due to 34 Empirically, we solve the problem of selection into ML status using a triple-difference estimator. 15

16 voters being diverted from position voting. This implies, on the one hand, that the candidate s position has a smaller effect on the voters behavior and thus the party can choose a more loyal candidate. On the other hand, swing voters play a greater role in position voting. Therefore, the party has to pay increased attention to these voters preferences. 35 We start by discussing these effects as they appear in (11). Party Loyalty Effect. To pin down the first effect, we focus on a state where the voter s partisanship status and her position on issue n are uncorrelated (σ n = 0). 36 In this case, the optimal senator position is a convex combination of the average voter and the party bliss points. In the presence of the ML, the party can afford to choose a candidate whose views on the issue are closer to those of the party. Whether this leads to more extreme or more moderate candidates depends on the position of the average voter as compared to the party bliss point. 37 In moderate states, where Y Dn < X n < Y Rn, the party loyalty effect leads to both parties nominating more extreme candidates. However, in an extreme right-wing state, Y Dn < Y Rn < X n, the ML leads to the Republican party choosing a more moderate candidate. The same is true for the Democratic party in an extreme left-wing state, X n < Y Dn < Y Rn. Swing Voter Effect. Now let us drop the assumption of zero covariance and suppose we are in a state with few partisan voters, so that the party loyalty effect is small. In this case, introducing the ML forces both parties to follow the direction of the swing voter. The reasoning is as follows. Assume that σ n > 0 so that holding more left-wing views on issue n is associated with being a partisan and, therefore, with using the straight-ticket option. Since then the ML attracts left-wing voters, 35 Expression (11) clearly demonstrates that the ML reinforces the effect of partisanship. Thus the ML effects in this model can be also seen as the effects of increased levels of partisanship. 36 As an illustration, consider any 0-symmetric distribution of positions x in and let partisanship p i be an even function of the position, i.e. p i (x in ) = p i ( x in ). 37 For the purposes of the exposition, we say that a Republican is more moderate when she moves to the left, and more extreme when she moves to the right; inversely for a Democrat. 16

17 the average position of those who go through the ballot, and judge the candidates by their political positions, shifts to the right. Hence, the optimal candidate s position must satisfy a more right-wing voter when the option is introduced, so that the swing voters become more decisive. Therefore, when σ n > 0 the swing voter effect is also positive and vice versa for σ n < 0. In the former case, this would imply a more extreme Republican candidate and a more moderate Democrat. 38 Total ML Effect. The direction of the total effect of the ML depends on the relationship between the party loyalty and swing voter effects, which may concur or counteract each other. It is determined by a single inequality. We have that introducing the ML results in a right shift of y jn if and only if α n α n + 1 Y jn + 1 α n + 1 X n > Xn P. (12) This implies that, if both the party and the average voter are more right-wing than the average partisan voter, the straight-ticket option leads to a more right-wing candidate. If both are more left-wing than the partisan voter, then the ML results in a left shift. When neither is true, the total effect depends on the partisan voter s position relative (RHS) to a convex combination between the party and the average voter (LHS). The party loyalty, swing voter and total effects are treated formally in Appendix A.1 Lemmas 1, 2, and 3, respectively Effect on Vote Share While both parties choose their candidates to maximize the share of votes they get, one of the parties will have an advantage due to the average partisanship and distribution of voters positions in the state. The ML has a differential effect on the 38 Since σ n can be viewed as a proxy for heterogeneity in an asymmetric state (X n 0), the swing voter effect is larger if the voters in a state hold more extreme views. 17

18 relative importance of these determinants of election success. Proposition 2. The Republican vote share increases in (i) the Republican partisan advantage, p(r) p(d), (ii) the swing-position covariance σ n, n only if the ML is available, and (iii) the average voter bliss point X n, n. Symmetrically, the Democratic vote share decreases in (i) p(r) p(d), (ii) σ n, n only if ML is available, and (iii) X n, n. The ML increases the effects of (i) and decreases the effect of (iii) on the distribution of votes between the parties. Note first, that the voters positions and partisanship have the obvious effect on the parties success, so that the greater the party support in the state and the closer the party is to the average voter the higher its vote share ((i) and (iii) of the proposition, respectively). (ii) is also straight-forward. In this setup, σ n is a determinant of the election outcome only if the ML is present. In that case, it benefits the party that follows the direction of the swing voters; the Republican if positive, the Democrats if negative. The more interesting result is that the ML has a differential effect on these determinants. While it reinforces the role of partisanship and the covariance effect, it devalues the advantage of position proximity between the state s average voter and the party. Suppose, for example, that there is an exogenous uniform right shift in all voters positions, without changing their partisanship status. The claim of the proposition is that the Republican party would benefit more from such a right shift if the ML is absent. The intuition for the result is as follows. The ML makes both the state partisanship and the swing voters more decisive in the election, while the position of the average voter becomes less important, since fewer voters elect by position. Notice that the voter s positions across issues are substitutes with respect to the party vote share. Suppose that New Jersey and Connecticut are equivalent except 18

19 that X CT econ = X NJ econ+0.1 and X CT soc = X NJ soc 0.1, meaning that the average Connecticut voter is more right-wing in economic issues, but more left-wing in social, than the New Jersey voter. Then, under the assumption that both issues are weighted equally, the probability of a Democrat winning the state is exactly the same Compound Effect: Expected Position of the Elected Senator Knowing the optimal positions of candidates and their corresponding vote shares we can evaluate the expected position of the election winner. Let y n = V j y jn + (1 V j ) y jn. (13) The expected platform of the election winner, y, is a convex combination of the endogenous positions of the Republican and Democratic candidates, where the weights are their respective vote shares. 39 Proposition 3. The expected position of the elected senator on issue n, y n, increases in (i) the Republican partisan advantage, p(r) p(d), (ii) the swing-position covariance in all issues, σ m, m, only if the ML is present, (iii) the average voter bliss point in issue n, X n, and (iv) the average voter bliss point in all other issues X m, m n. The ML decreases the effect of (iii) and increases all other effects on y n. This finding sheds light on the interplay of the results of propositions 1 and 2. The election winner is expected to be more right-wing the more republican the state is, and the more right-wing the swings and the average voter are. Now, since issues are substitutes when it comes to parties vote shares there are spillovers from voters positions on one issue on the expected senator s position on another issue. For instance, a state that is right-wing in economic issues, but left-wing in social issues 39 In a model with aggregate noise, vote shares represent the winning probabilities; in this case (13) is the mathematical expectation of the election winner s position. 19

20 can expect a senator that is right-wing in social issues as well, if the weight of social issues in voters utilities is small enough. Moreover, fixing n, introducing the ML leads to a stronger effect of partisanship, and the positions of swings, and those of the average voter on other issues, m n. For example, let n be economic issues and take Texas, a Republican state and California, a Democratic one. 40 Suppose also that the Texan electorate is more right-wing in social issues than the Californian one, X T X soc > X CA soc. First note that Texas is more likely to elect a right-wing senator in economic issues, irrespective of the ML (due to (iv), the spillover effect). Introducing the option leads to an even higher y T X econ, since the ML reinforces the effects of partisanship and the spillover effect between the issues. 41 Now, shifting the focus to social issues and assuming away any difference in partisanship between the two states, we have that California is expected to elect a more left-wing senator than Texas ((iii) in the proposition). However, introducing the ML in both states implies that the expected senator positions in social issues will be driven closer. This is because the X soc becomes less influential in the election outcome, as we show in Proposition 2. Furthermore, the ML brings swing voters to the forefront, in the following sense. Suppose Georgia and Alabama are equivalent except for the swing voter covariance in any issue, and Georgian swing voters tend to be more right-wing than the Alabama ones. Before the ML is introduced, the expected senator position in any issue is the same for both states, but with a ML, the Georgian senator is more right-wing in expectation than the Alabama one. To conclude the theory section, we have shown that a straight-ticket option changes the relative importance of positions vs partisanship in a state and thus affects the types of voters targeted by senatorial candidates. In the presence of a ML, 40 In terms of their partisanship levels. 41 Note that the argument assumes that all else is equal in both states. 20

21 partisanship becomes more significant in that it is a more important determinant of vote shares and may thus allow senatorial candidates to offer platforms closer to the parties they represent. In terms of the political positions of the electorate, the average voter loses significance, whereas the swing voters and their positions become more decisive. 4 Empirics In this section, we identify the effect of the ML on an elected senator s position and party and use the theory to account for the mechanism behind our findings. We focus only on elected senators as this is what we can observe in the data, and throughout we discuss any issues arising from the lack of information on the losing candidate s platform. 4.1 Data We use data from the 87th ( ) to the 112th ( ) Congresses and their preceding elections. Our main dependent variable is taken from Poole and Rosenthal s DW-NOMINATE scores. They use a senator s (roll-call) voting history to summarize her position into a two-dimensional vector for each congress. 42 The first dimension is the one that explains most of the variation in votes and the second, minor one is perpendicular to the first and set to explain the rest of the variation (Carroll et al. (2009)). We use only the former and it corresponds to y j in the theory, where we no longer need the issue subscript n as all issues are covered in that one position See Poole and Rosenthal (2015). DW stands for dynamic weighted and allows for cross-congress comparisons. The scores are from around -1 to around 1 (some observations outside the interval), and we have converted them from 0 to Poole and Rosenthal (2007) find that after 1978 the first dimension is sufficient in explaining Congress member behavior. If we do use the second dimension, which captures deviations from a senator s main bliss point, the effect of the ML on it is insignificant. Poole and Rosenthal (2007) also study the types of issues that cause these deviations in different years, so in practice the main dimension that we use does not capture every single issue completely before

22 Moreover, we construct the following senator variables using information on the US Senate website and the CQ Press Guide to US Elections: their party, year(s) of election, years in Senate, whether they were an incumbent, appointed into office (i.e. not elected) or did not complete a full term. For each senator we also note the position and characteristics of the senator in power (SIP), i.e. the one was not up for election in the same year. State-level data on the average voter position, X in the theory, is taken from Enns and Koch (2013). 44 They combine select questions of all available public opinion polls into a two-dimensional dynamic state policy mood on the size and scope of government. This is in contrast with our variable for senators that sums up positions on all issues. However, as voter beliefs on the role of the government have direct implications on most Senate policies, we assume that the two policy spaces are comparable. 45 As with the DW-NOMINATE scores, the policy mood has a major and a (orthogonal) minor component, which is weak in explaining voter opinions and which we do not use. Enns and Koch (2013) also create state-level party identification variables (fractions of self-declared Democrat, Republican and non-partisan voters) that we use both as controls and as a way to classify states into red (mostly Republicans), blue (mostly Democrats), swing (mostly non-partisans) and purple (almost equal and high numbers of both Democrat and Republican partisans). Appendix A.2 gives the precise empirical definition of this partisanship classification. We also create a positional classification which separates states into extreme left-wing, moderate and extreme right-wing, depending on the position of the average voter with respect to the parties bliss points (see Appendix A.2). We define the latter as the median DW-NOMINATE score of all elected members of each party for a specific Congress It varies from 0 (left-wing) to 100 (right-wing). 45 This assumption is not essential for the identification. 46 We use these classifications in Section 4.5 when employing the theory to understand our empirical findings. We point out that a state can move across classes from one election to another, and also 22

23 Note also the Democratic party used to be internally divided. In some southern states its members were representing significantly more right-wing views, as opposed to the North. After the passage of the Civil Rights Act of 1964 and the Voting Rights Act of 1965 the party division slowly dissipated. 47 Mapping this to the theory, it suggests potentially two sub-parties with two different bliss points. To allow for this possibility, we create a dummy variable southd that takes the value 1 when considering one of these originally deviant states. 48 Lastly, we construct our own data on the existence of the ML per state and Congress, which we compare with that of Klarner (2010) for corroboration Summary Statistics In the 26 Congresses that we examine, there are 894 senators up for election, 50 a total of 2620 observations (all senators, all Congresses) and 423 unique senators (excludes incumbents), belonging to either the Democratic or the Republican party. 51 The left graph of figure 1 depicts the decrease in the number of states with a ML over time, and the right one the evolution of the average position of elected senators per party. The distance between the two parties defines the polarization in the Senate, which has been increasing. 52 As it can be seen here, and consistent with previous findings, the Republican party seems to be the stronger driving force of the polarization. 53 On the left panel of figure 2, we show the evolution of the average senatorial posithat these classes are based on self-declared voter opinions and not on actual votes. 47 See McCarty et al. (1997) for an analysis. 48 a. They are Arkansas, Georgia, Kentucky, Louisiana, Mississippi, North Carolina, Oklahoma, South Carolina, Tennessee, Texas and Virginia. b. For convenience we will call non-southd states, northern. 49 In the contradicting cases we found sample ballots or other evidence to deal with the discrepancy. Note also that Washington, DC does not elect senators. 50 Before each Congress there is normally at most one seat up for election per state. 51 We have removed the very few observations of Conservative and Independent party senators. 52 There are different definitions of polarization used in the literature, we picked one for illustration. 53 See McCarty et al. (2006). 23

24 # States Year Average Position Year # States w/ ML DEM REP Notes - Left: Number of States with ML over time. Right: Evolution of average party position in the Senate, from left (0) to right-wing (100). Source: Data from Poole and Rosenthal (2015) Figure 1: ML Presence & Senate Polarization tion per party, separating states by whether or not they offer a ML on their ballots. 54 It is evident that for the Republican party a straight-ticket option is correlated with more right-wing senators (closer to 100), whereas for the Democratic party the relationship, if any, is not clear. The right graph displays the fraction of Democratic senators, by ML presence. Keeping in mind that there is a lot of variation across periods away from both fitted lines, we still see that in states without a ML there is not much change in the fraction of representatives per party over time, which is about equally split. For ML states, however, in early periods the Democratic party seems to be benefiting from its presence, whereas over time, as the set of states with the option decreases, there is stronger representation of the Republican party. On the flip side of the market, we have the voters. Figure 3 shows the evolution of self-declared positions and partisanship of voters by ML presence, on the left and right graphs respectively. Voters do not seem to be systematically different across types of states, with an exception towards the end of the time period Note that the sets of states with and without ML changes over time (figure 1). 55 Note that the remaining fraction of voters missing from the right graph are self-declared swings (non-partisans). 24

25 Average Position % Elected Senators Year Year DEM ML DEM no ML REP ML REP no ML DEM ML DEM no ML Notes - Left: Evolution of average party position in the Senate separating states by ML presence; positions vary from left (0) to right-wing (100). Right: Fraction of elected senators that are Democrats by ML presence. Source: Data from Poole and Rosenthal (2015) Figure 2: Positions and Party of Elected Senators by ML Presence Moreover, as we discuss in Section 4.1, we have created partisanship and positional classifications of states since the predictions of the theory may vary within each. 56 States can move across classes from one period to the next, and table 1 displays the number of states, in the whole time period, per class and ML. Note first that the latter is significantly present in all subcases. Second, the skewness observed across positional classes (more extreme left than extreme right-wing states) should be viewed with caution as the definitions of voter and senator positions, that determine this classification, do not perfectly align. Lastly, figure 4 includes scatter-plots depicting the correlation between the average voter position and different partisanship types across all states. As expected, a higher fraction of Democratic voters corresponds to a more left-wing average voter position (top right graph), and vice versa for the Republican voters (top left), although in this case the relationship is not as clear. In terms of partisan and swing voters, there is 56 Definitions in Appendix A.2. 25