Read an Excerpt
CHAPTER 1
Who Wants to Run?
If the people can choose only from among rascals, they are certain to choose a rascal. — V. O. Key, The Responsible Electorate
To summarize: it is a well-known fact that those people who must want to rule people are, ipso facto, those least suited to do it. — Douglas Adams, The Restaurant at the End of the Universe
The goal of this chapter is to convince you that, if you want to understand polarization in American politics, you need to ask the question: who wants to run?
American legislatures are in turmoil. The United States Congress is crippled by ideological disagreements and partisan rankling that make it difficult to craft new policies or to update obsolete ones. This book will not explain all the complex causes of this state of affairs, but it will explore a key factor — the willingness of individuals of varying ideologies to seek political office — that is largely absent from existing work on political polarization.
This chapter presents the overall argument in three parts. First, I document the rise of ideological polarization in the U.S. House and explain that we should care about it because it creates legislative dysfunction and appears to be contrary to voters' wishes.
Second, I discuss how existing research on legislative polarization focuses mainly on the preferences of voters rather than the behavior of candidates. I then present evidence that the preferences of voters cannot explain most of legislative polarization, or its rise. Even if voters selected the most moderate available candidates in every election since 1980, polarization in the U.S. House still would have risen roughly 80 percent as much as we have observed. To understand legislative polarization, we need to understand why the people running to become members of Congress have themselves polarized.
Third, and finally, I offer a theory, based closely on citizen-candidate models, for why some people become candidates while others do not and for how this decision contributes to polarization. The theory explains that the set of people who run for office will become more ideologically extreme when the costs of running for office are high and when the benefits of holding office are low — conditions that, I will argue in chapter 4, are increasingly met in our federal legislatures. The logic is as follows. Someone on the far left or far right will particularly dislike having a representative from the other end of the ideological spectrum and therefore will be more likely to be willing to bear the considerable costs of running for office to avoid this fate. Someone toward the middle of the ideological spectrum is not as far from extremists in either party; this person's disutility from an extremist representative is therefore not as large as it is for an extremist in one party facing an extremist representative from the other party. As a result, moderate citizens will be less willing to bear the costs of candidacy.
Rising Polarization in U.S. Legislatures
The primary focus of this book will be to explain why who runs for office has helped spur the well-documented rise in ideological polarization in our legislatures since 1980 (McCarty, Poole, and Rosenthal 2006). Figure 1.1 plots a common measure of polarization for the U.S. House (Poole and Rosenthal 2000), based on how differently members of the two parties cast roll-call votes, for the years 1879–2014. Higher values of this measure indicate greater ideological disagreement between the two parties. The pattern is obvious. Since 1980 (shaded in gray in the figure), the ideological gap between the two parties in the House has more than doubled, and it now exceeds previous all-time highs from the turn of the twentieth century.
Why is it worth explaining this recent rise? Legislative polarization is not just some statistical artifact from measuring roll-call voting behavior. Rising polarization means that the parties are systematically disagreeing more, whether due to ideological differences, partisan loyalty, or both. The result is a legislature that struggles to accomplish even the most basic tasks before it. Perhaps the clearest symptom of this issue is the budgeting process. A process once known as "Regular Order" is now "best described as 'Regular Disorder'" (McCarty 2014, 1), with Congress regularly failing to complete the budget process. Government shutdowns and brinksmanship are clear consequences of legislative polarization.
Perhaps more important, legislative polarization appears to be contrary to voters' wishes. There is considerable confusion in political science about how to study the links between elections and ideology. In chapters 2 and 3, I will explain this confusion and offer a framework for evaluating what voters prefer based on their actual choices in real elections. Applying this framework, I show that voters prefer more-moderate candidates in U.S. House elections, on average. I do not mean prefer in the sense that voters themselves necessarily espouse particular policy views that are moderate, or even that it's the candidates' positions per se that drive voters to support them — although these things may well be true. Instead, I mean it in the precise sense that, when given the choice to elect candidates who have taken more-moderate positions or to elect candidates who have taken more-extreme positions, voters tend to elect the candidates who offer more-moderate positions. If our legislatures are nevertheless populated with candidates whose positions are more extreme than those of the candidates voters seem to prefer, then our political institutions have failed to translate the desires of voters into legislative activity efficiently. Put simply, it is likely that voters would be better off — in the sense that they would change whom they elected to office — if more candidates who offer moderate platforms ran for office.
I do not mean to suggest that polarization is the only problem in current American politics, or that reducing polarization would be a panacea. And I do not mean to suggest that previous, low polarization eras in American politics reflected any sort of political utopia — indeed, the 1950s and 1960s exhibited relatively low legislative polarization but presented enormous political challenges in their own right. Nevertheless, there are gains to reducing polarization from its current, unprecedented heights. Like all political problems, polarization implies trade-offs, and reformers must always consider these trade-offs when advocating for focusing on one issue over another. I will return to this issue in the book's conclusion.
Existing Accounts of Polarization Are Votercentric
Now that I have laid out the problem to be studied, the next question is how to study it. Political scientists have largely missed how important the question "who wants to run?" is for understanding legislative polarization because we have been so focused on votercentric accounts of polarization. This is not to say that political science does not study the choice to become a candidate. Indeed, large literatures in political science study political ambition — particularly as it relates to women in politics (e.g., Kanthak and Woon 2015; Lawless and Fox 2005, 2015; Preece and Stoddard 2015; Sanbonmatsu 2010; Thomsen 2015, 2017) and to the strategic decisions of career-minded politicians (e.g., Cox and Katz 1996; L. Fowler 1993; Maestas et al. 2006; Maisel and Stone 2014) — but they do not usually consider legislative polarization directly. The exception to this rule is the work of Thomsen (2017), which makes an argument similar and complementary to this book's argument, focusing on how party fit in the legislature discourages moderates from running for the U.S. House, thereby driving polarization. I will lean heavily on this account in chapter 4. Reviewing the literature, Thomsen observes that "candidate emergence has received only minimal attention from polarization scholars" (6).
Instead, political scientists have studied the polarization of American politics in two main ways. Institutional scholars have looked to real-world behavior in the legislature, documenting the way legislators of the two parties have grown farther apart, ideologically, and evaluating links between this change and a variety of structural factors such as income inequality (McCarty, Poole, and Rosenthal 2006; Voorheis, McCarty, and Shor 2015), redistricting (Carson et al. 2007; Eilperin 2007; Masket, Winburn, and Wright 2012; McCarty, Poole, and Rosenthal 2009; Theriault 2008), and primary elections (Aranson and Ordeshook 1972; Burden 2001; Brady, Han, and Pope 2007; Coleman 1971; Hill and Tausanovitch 2017; Hirano et al. 2010; McGhee et al. 2014; Owen and Grofman 2006; Pildes 2011).
Meanwhile, behavioral scholars have looked to individual voters, measuring their political attitudes in surveys to see how they have changed over time. But even on the fundamental question of whether voters have polarized, substantial disagreement remains. To pick perhaps the most famous of these disagreements, Fiorina, Abrams, and Pope (2005) and Fiorina and Abrams (2009) argue that only a small class of political elites has polarized over time, while voters themselves have not. Abramowitz (2011) strongly disagrees, arguing that voters of the two parties have sorted in their views on a range of issues and become more extreme. To date, the disagreement rages on.
Another thread of the behavioral literature suggests that the blind partisanship of voters allows parties to pursue the ideological goals of elites without electoral sanction. In a New York Times op-ed, Achen and Bartels write, "That is one key reason contemporary American politics is so polarized: The electoral penalty for candidates taking extreme positions is quite modest because voters in the political center do not reliably support the candidates closest to them on the issues." This is an argument they pursue in their recent book (Achen and Bartels 2016a). In chapter 3, I document a pronounced electoral advantage for moderate candidates when we study election returns rather than surveys — but the more important point is that this account, too, is focused on voters.
Each approach, the institutional and the behavioral, has its own strengths and weaknesses, but both have missed how the candidate supply — the set of people willing to run for office — contributes to polarization. When institutional scholars have observed changes in legislator behavior and attributed them to external forces, they have, implicitly or explicitly, assumed that these changes relate to changes in what their constituents want. When behavioral scholars have seen changes in voter opinion and connected them to the current state of legislative politics, they have, implicitly or explicitly, assumed that these changes are reflected in the behavior of voters' representatives. Both literatures study one side of the equation — either voters or legislators — and impute some kind of response to the other side. But if the set of people running for office has itself polarized, these assumptions are not necessarily met. Legislators might polarize even if voters do not want them to, because voters can only elect representatives from among the set of people who run for office.
Voter Preferences Not Enough to Explain Rising Polarization
Though there are good reasons to study voters in relation to polarization, ignoring candidates risks missing the bulk of legislative polarization. In fact, legislative polarization would still be high, and rising, even if voters always elected the most moderate possible candidate in every election cycle in every district.
Figure 1.2 shows a set of simulations based closely on those developed by Bonica (2017). The first line, in solid black, plots the observed polarization in the U.S. House from 1980 through 2014, measured by scaling incumbents in terms of their ideology using the mix of campaign contributions they receive from donors (Bonica 2014). I will discuss this scaling technique, CFscores, in detail in chapter 3, where I will justify its validity as well as introduce alternate scaling methods that produce similar results. The second line, in dashed gray, plots how polarized the legislature would be if, in every election, House voters elected only the least-extreme available candidates, measured again using campaign contributions, defining least-extreme as the candidate scaled closest to the midpoint of the ideological scale. The gray line is therefore the hypothetical least amount of polarization we could see, given the set of people who have chosen to run for the House over this period. As figure 1.2 shows, even in this extreme hypothetical, we still see a very high degree of polarization. That is, even if voters always elected the most-moderate candidates they could, legislative polarization would still be very high in the U.S. House.
According to the simulation, roughly 20 percent of overall polarization in the House is due to choices voters make among the candidates who run for office. The other 80 percent of polarization exists no matter which candidates voters choose from among the existing pool — it is already baked into the set of people who run for office! Although the literature on polarization has focused on the choices voters make among candidates, the bulk of legislative polarization may not be due to these choices.
Voters may also contribute to polarization indirectly in ways the simulation does not capture. Some of the observed polarization could be due, for example, to moderates not running because they anticipate that voters will not support them if they do run. I will consider this possibility — and more generally, the degree to which voters support moderate candidates — in chapter 3, where I show that voters tend to elect moderates when moderates run for office. In any case, the takeaway here is not that voters' preferences do not matter but rather that who runs for office also matters a great deal for polarization.
How do we know that these simulations correctly measure candidate ideology, when they rely on campaign contributions to scale candidates? For example, what if donors are strategic, or what if the scalings are picking up partisanship rather than ideology? I will address these methodological issues in chapter 2, but I will note here that the simulation's conclusions are unchanged if we use alternative methods to measure candidate ideology. In appendix 1, I replicate this plot using DW-DIME, which uses the contributions in a machine-learning setup to predict roll-call ideology (Bonica forthcoming), an approach that addresses the issue of partisanship. Using DW-DIME, the simulation estimates that 86 percent of polarization is explained by the candidate pool rather than by which candidates voters choose from among the pool. I also replicate the plot using Hall-Snyder scores, an alternative technique that omits all contributions made to candidates after they become incumbents to ensure that the scalings are not driven by strategic donating. Again, we arrive at the same conclusion. See appendix 1 for this analysis.
Candidates Becoming More Polarized
If most of polarization is due to who runs, rather than due to whom voters choose for office, then the set of people who choose to run for the House must be polarizing over time. Figure 1.3 confirms this pattern. The graph shows the absolute distance between the average ideological position of each party for each year, separately for incumbents and nonincumbents or, as I call them, new candidates (both challengers and open-seat candidates). Candidate ideologies, again measured by the CFScores scaling methodology, are first residualized by district so that the resulting calculations reveal candidate divergence and not sorting across districts.
The two lines track each other well (r = 0.9). As incumbent polarization has risen, so too has the ideological polarization of new candidates for the House. Incumbent polarization is also consistently below that of new candidate polarization. The results in chapter 3 provide the likely explanation for this fact. Voters tend to select more-moderate candidates from among the pool — thus producing Democratic and Republican incumbents who are closer together, on average, than the overall candidate pools. Nonetheless, incumbent polarization remains high. The candidate supply does not give voters the opportunity to shrink the difference between the candidates of the two parties further.
Voters Can Select Moderates but Cannot Force Candidates to Move to the Middle
The simulation and data I just presented offer a clear view of how this book will think about elections and polarization — a view that is at odds with much of the literature on polarization. In my view, voters are presented with candidates of varying ideologies, they choose some for office, and the ideologies of the winning candidates determine the degree of legislative polarization. But what if candidates moderate their positions after winning office? If candidates moderate in office, then electing more-extreme candidates does not necessarily create legislative polarization, because extreme candidates can become moderate incumbents.
This possibility is at the heart of the so-called median voter theorem, as encapsulated in the Downsian model (Downs 1957). Downs's influential theory predicts that candidates will converge, offering the positions that the median voter prefers in order to secure a majority of votes. As a result, it is a votercentric model of the political process; the precise identities of candidates do not matter because any candidate can choose any position. The Downsian logic underpins much of the work on polarization — it provides the theoretical logic for why voter preferences might imply something about the behavior of representatives and for why the behavior of representatives might reflect something about voter preferences.
(Continues…)
Excerpted from "Who Wants to Run?"
by .
Copyright © 2019 The University of Chicago.
Excerpted by permission of The University of Chicago Press.
All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.
