Party Competition in Partisan and Nonpartisan Governments

All of the existing evidence documenting a relationship between party competition and coalitional stability comes from formally partisan governments. Yet, despite the near universal use of partisan election systems at the state and national levels, approximately 70% of local governments use nonpartisan election systems instead (Svara Reference Svara2003). By definition, this means that party primaries are not a part of the electoral process and party affiliation is not listed on the ballot. In practice, however, nonpartisan elections also diminish or eliminate the role of party organizations in elections and government. Indeed, in many cases, party organizations are simply not present under nonpartisanship (Adrian Reference Adrian1959; Bledsoe and Welch Reference Bledsoe and Welch1987). Moreover, even when they are, they are typically less active and influential than their partisan counterparts, lack the type of collective responsibility that comes with having party labels on the ballot, and have no institutional advantages over other local interest groups (Adrian Reference Adrian1959; Anzia and Meeks Reference Anzia and Meeks2016; Cutright Reference Cutright1963; Hawley Reference Hawley1973; Pelissero and Krebs Reference Pelissero and Krebs1997). As Adrian (Reference Adrian1959, 458) concludes “in most nonpartisan elections, organizations identified with the major political parties are of little or no importance in determining the outcome and most candidates are not active participants in political party activities.”

These organizational differences across partisan and nonpartisan governments raise the question: with the increasing nationalization of subnational politics in the United States, will the trends related to competition documented in previous studies hold in nonpartisan contexts today? Indeed, though party organizations may not be formally part of the electoral and governing system in nonpartisan governments, with the strength of modern partisanship nationally and the increasingly ideological direction of politics locally, simply being in an area that is evenly split between the parties may be sufficient to make partisanship a salient feature of electoral competition. While this hypothesis contrasts with early work finding relatively disorganized, high-dimensional voting behavior in both the nonpartisan Confederate Congress and nonpartisan Nebraska legislature (Jenkins Reference Jenkins1999; Welch and Carlson Reference Welch and Carlson1973; Wright and Schaffner Reference Wright and Schaffner2002), it is consistent with both empirical evidence from more recent terms of the Nebraska legislature demonstrating an increase in partisan polarization (Masket and Shor Reference Masket and Shor2015) and theoretical work examining the conditions under which party in the electorate can translate into partisan outcomes under nonpartisanship (Krehbiel, Meirowitz, and Romer Reference Krehbiel, Meirowitz and Romer2005).

Whether we see similar patterns of elite behavior within partisan and nonpartisan governments, however, likely depends on the mechanism underlying competition and coalitional stability. If stability requires strong, active party organizations, then there is little reason to expect partisanship in the electorate to influence elite coalitions in nonpartisan contexts regardless of the level of competition; if, however, stability stems from the preferences of individual lawmakers, then we would expect to see similar levels of coalitional stability in both contexts. In the following two sections, I discuss these two potential mechanisms in further depth and explain how each yields differing predictions for partisan and nonpartisan contexts.

Partisan Electoral Competition and Party Organizations

The first mechanism through which competition might facilitate coalitional stability centers on the way in which competition interacts with formal party organizations. Specifically, when electoral competition represents either a credible threat to the majority party’s power or a realistic opportunity for the minority party to win office, party organizations have a greater incentive to invest in and leverage the electoral and legislative tools at their disposal to improve their members’ electoral prospects (Harmel and Hamm Reference Harmel and Hamm1986; Schlesinger Reference Schlesinger1985). As Schlesinger (Reference Schlesinger1985) argues, the more competitive the electoral context, the more valuable organization becomes for members of the party because they simply cannot rely on party identification to win office.

There are two critical pieces to this logic: first, that party organizations can leverage electoral and legislative tools to influence member behavior, and second, that the use and effectiveness of these tools varies with the threat of electoral loss. To the first point, there is a significant body of scholarship that demonstrates the significance of party organizations and party activity for member behavior. On the electoral side, for example, parties engage in a broad range of activities to support the election of their members, including fundraising, endorsements, turning out the vote, and developing and informing voters of the party brand (Cohen et al. Reference Cohen, Karol, Noel and Zaller2008; Hassell Reference Hassell2015; Key Reference Key1942; Snyder and Ting Reference Snyder and Ting2002). These activities, which scholarship on local politics suggests are not exclusive to national parties (Bledsoe and Welch Reference Bledsoe and Welch1987; Eldersveld Reference Eldersveld1995; Krebs Reference Krebs1998, Reference Krebs2005), can decrease the costs of running for office, help voters understand the differences among their choices at the polls, and minimize the need to cultivate a personal vote (Primo and Snyder Reference Primo and Snyder2010), making them quite valuable to members of the party seeking election. This value, in turn, creates the incentive for members to support the party’s agenda to ensure that the party continues to provide this type of assistance.

In addition, on the legislative side, political parties can also develop and leverage tools within the legislative chamber to improve the party’s electoral fortunes and hold individual members accountable for deviating from the party line. For example, a large body of scholarship argues that parties manipulate the legislative agenda for electoral gain, limiting issues that might divide the party from reaching the floor (Cox and McCubbins Reference Cox and McCubbins1993, Reference Cox and McCubbins2005). This form of agenda control is particularly effective in tandem with strong committees (Anzia and Jackman Reference Anzia and Jackman2013), which might explain why partisan local governments are significantly more likely to have institutionalized committee systems than their nonpartisan counterparts (Pelissero and Krebs Reference Pelissero and Krebs1997). Furthermore, parties can also use both selective benefits and targeted punishments to shape member behavior (Cann and Sidman Reference Cann and Sidman2011; Rohde Reference Rohde1991). For example, to incentivize cooperation, party leaders can offer positions of power or prestige, increased resources, or help getting a piece of legislation through the chamber. They can also, of course, strip a member of any of these perks or block the progress of important legislation as punishment. And indeed, the evidence suggests that these types of incentives, perhaps in combination with electoral organizing, do influence member behavior (Ansolabehere, Snyder, and Stewart Reference Ansolabehere, Snyder and Stewart2001; Snyder and Groseclose Reference Snyder and Groseclose2000).

While these electoral and legislative tools are certainly valuable for party organizations, there are two reasons to believe that their use and effectiveness will vary with the threat of electoral loss. First, because these types of organizing are costly, parties will only have an incentive to engage in them when they are challenged or have a realistic chance at winning office (Schlesinger Reference Schlesinger1985). For example, Harmel and Hamm (Reference Harmel and Hamm1986) document how the introduction of two-party competition in Texas led to growth in the organizational capacity of each party within the legislature, with each party hoping to increase the likelihood that they would ultimately win enough seats to hold a majority. As Representative Bob Bush explained at the time, “When Texas was a one-party state, it wasn’t necessary to identify yourself and what you stood for” to win election (Harmel and Hamm Reference Harmel and Hamm1986, 84). With competition, however, members needed to differentiate themselves from the out-party to win office and newly bolstered party organizations had both the incentives and resources to facilitate this process.Footnote ³

Second, changes in the level of competition alter the incentive structure for individual members of a legislature, decreasing the power of the carrots and sticks that parties wield. For example, given that parties exist—whether wholly or in part—to facilitate reelection, then absent strong, credible electoral competition, parties will become less significant for a member’s electoral prospects. As a result, there is less benefit to the member for adhering to the party line when it conflicts with their preferences because doing so is less likely to affect their electoral future. Indeed, while party-provided pork and committee power are certainly valuable, in a context where the electorate overwhelmingly favors one party they are unlikely to significantly improve a member’s electoral fortunes, particularly relative to the cost of adhering to the party line, and so members will have less of an incentive to sacrifice for them. Consider, for example, former New York City Councilmember Charles Barron. In 2010, he challenged fellow Democrat and Speaker Christine Quinn for leadership of the chamber, losing 48 to 1.Footnote ⁴ Quinn subsequently stripped Barron of his committee chairmanship and his $10,000 “lulu” stipend. Yet, given that he was an incumbent Democrat in a highly Democratic city, Barron had little reason to cooperate to win these perks back, and he instead continued to challenge the Democratic leadership and foster dissent within the party thereafter.Footnote ⁵

Ultimately, if electoral competition affects the influence and activity of party organizations—whether because individual members have less of an incentive to support the party’s agenda absent an electoral threat or because the party itself has less of an incentive to organize member behavior—then we should only see a relationship between competition and coalitional stability in governments that use partisan elections because those are the contexts where party organizations are embedded within the electoral and governing system. Thus, when elections are partisan and the electorate is evenly balanced between the parties, party coalitions should be more cohesive and voting behavior should be more one-dimensional. As the electorate becomes increasingly imbalanced, however, coalitions in these contexts should shift more frequently, meaning that voting behavior will be more multidimensional. In contrast, in nonpartisan governments, where organizations are either absent or significantly less active to begin with, elite behavior should be relatively disorganized regardless of the capacity for local party competition. Thus, any differences between elite behavior in partisan and nonpartisan contexts should grow smaller or disappear as the balance between the parties in the electorate grows increasingly lopsided.

Partisan Electoral Competition and Elite Preferences

The preceding logic suggests that—when the electorate is evenly balanced and competition between the parties is high—party organizations will have an incentive to develop and leverage electoral and legislative tools to facilitate the election of their members. This organizational effort, in turn, fuels greater cohesion within the party and thus more stability in legislative coalitions across conflicts. However, while a mechanism based on party organizations is certainly plausible, legislative stability might also arise as a function of how competition influences the preferences and priorities of the members who run and serve.

The logic for a preference-based mechanism is rooted in two key features of legislative behavior: first, strong, active party organizations—either in the legislature or as electoral-mobilizers—are not inherently necessary for partisan-based voting. As Krehbiel (Reference Krehbiel1993, Reference Krehbiel2007) has long argued, parsing the effect of party often suffers from a problem of observational equivalence, such that models based on the preferences of individual members can often explain outcomes interpreted as party influence. This is important because with the increasing nationalization of American politics (Hopkins Reference Hopkins2018), local governments are likely to be populated by ideological partisans regardless of the presence or activity of local party organizations. Indeed, scholars of local government have shown that partisan policy outcomes are common in contexts that use both partisan and nonpartisan elections alike (de Benedictis-Kessner and Warshaw Reference de Benedictis-Kessner and Warshaw2016; Einstein and Glick Reference Einstein and Glick2016; Einstein and Kogan Reference Einstein and Kogan2015). If these partisan-oriented preferences drive elite behavior, then we would expect voting within councils where the electorate is evenly balanced between the parties—regardless of whether they are formally partisan or nonpartisan—to be relatively organized, with stable ideological coalitions voting together over time.

Second, if elite preferences and voting behavior are oriented around partisanship, then—even in the absence of local party organizations—competition may still influence the relative salience of party and the need for elected officials to vote in an ideologically constrained manner. For example, when the electorate is evenly balanced, candidates will be able to better differentiate themselves during campaigns based on partisan affiliation and ideological compatibility such that these features will be more salient to voters and thus more important for voting decisions at the polls, even in nonpartisan contexts. This increase in the salience of party with competition is consistent with recent evidence that party competition appears to exacerbate polarization (Aldrich and Battista Reference Aldrich and Battista2002; Hinchliffe and Lee Reference Hinchliffe and Lee2016), making partisanship a more important factor for legislative behavior and representation (Bonica and Cox Reference Bonica and Cox2018; Carroll and Eichorst Reference Carroll and Eichorst2013).

In contrast, when the electorate strongly favors one party, politicians will need to develop electoral constituencies around other groups and interests.Footnote ⁶ As a result, partisan considerations may simply become less important for legislator decision making, leading to less stable coalitions within the legislative chamber regardless of electoral system. As Patterson (Reference Patterson1962, 200) explains, without “party as a reference group, the legislator is likely, consciously or unconsciously, to respond to different pressures in different voting areas.” In turn, members will deviate specifically in circumstances when the preferences of important electoral constituencies conflict with their partisanship. At the local level, specifically, where a variety of groups have been shown to influence the political process and where policy tends to be more allocational in nature (Trounstine Reference Trounstine2016), these types of cross-cutting constituencies may be particularly likely to form. Indeed, if the focus of debates is often on where resources are devoted or who benefits most from a service, then it creates the opportunity for coalitions to shift according to the specific locations or groups involved, particularly absent a clear partisan or ideological out-group.

Empirically, if this type of preference-based mechanism underlies differences in the stability of legislative coalitions, it should be observable by comparing across governments that use partisan and nonpartisan election systems. Specifically, because nonpartisan governments are significantly less likely to have strong party organizations active within local politics, if—regardless of the election system—legislative behavior becomes more multidimensional as the electorate increasingly leans toward one party, then it suggests that the relationship is rooted more in the preferences of individual council members than party organizations. In contrast, if this relationship is only present in contexts that use partisan elections, then it would suggest that the mechanism is rooted more in features of the institutional environment in partisan election systems, such as the presence and activity of strong party organizations.

Ideological Extremity in Partisan and Nonpartisan Councils

If differences in preferences across legislators were driving the results in the previous section, we would expect to see more ideologically extreme local council members in competitive partisan contexts. To test this hypothesis, Figure 4 depicts the bivariate relationship between partisan imbalance in local presidential voting and the ideological extremity of council member ideal points for those serving in nonpartisan and partisan governments, respectively. To calculate ideological extremity, the first row uses the ideal points from the first estimated dimension and the second row uses the ideal points from the second dimension. As Figure 4 shows, for partisan governments, there is essentially no relationship between the relative skew in local presidential voting and ideological extremity, regardless of the ideological dimension, such that council members in more one-party dominant contexts are nearly identical to their counterparts in two-party contexts, on average. Moreover, as the right two panels show, this flat relationship is also present in nonpartisan governments, with little apparent difference across the two institutional contexts. This similarity in trends implies that the ideological distribution of council members in partisan and nonpartisan governments is similar and casts doubt on the hypothesis that ideologically extreme preferences underlie the higher levels of coalitional stability found in competitive, institutionally partisan governments.

Figure 4.

Two-Party Competition and the Ideological Extremity of Council Members in Two Dimensions

To account for factors that might be confounding the simple bivariate relationship, Table 2 presents results from a series of regressions in which the absolute value of each council member’s first and second dimension ideal points are used as the dependent variables. As with the analysis in the previous section, I include an interaction term between the use of partisan elections and local partisan imbalance to estimate the trends separately for partisan and nonpartisan governments. The first and third columns model this relationship hierarchically, with covariates at both the individual council member and city level. The second and fourth columns omit the group-level intercepts and instead cluster the standard errors by city. Modeling the relationship in one of these two ways is necessary to account for correlation among council members responding from the same city. For all specifications, I include a set of legislator-specific covariates, including sex, race, and party. In addition, I also include all of the city-level covariates used in the previous section, with the exception of council size and the total number of contested votes, which were used to account for error in measuring dimensionality.

Table 2.

Relationship between Local Party Competition and Ideological Extremity

Note: ⁺p < 0.1; ^*p < 0.05; ^**p < 0.01.

Consistent with the raw data, Table 2 shows little evidence of a difference in the relationship between the balance between the parties in the electorate and ideological extremity across partisan and nonpartisan governments. This is true for both ideological dimensions and regardless of the model specification. Figure 5 illustrates this point by depicting the marginal difference in ideological extremity between council members elected to partisan and nonpartisan governments across all levels of partisan imbalance. While there is some evidence of an increasing trend, such that elected officials in partisan governments are more extreme than nonpartisan officials as the electorate increasingly favors one party, the largest differences are at the highest values of partisan imbalance, where there are fewer observations and the variance is substantial.

Figure 5.

Marginal Difference in Ideological Extremity between Council Members from Partisan and Nonpartisan Governments across Levels of Partisan Imbalance

Of course, a lack of a statistically significant relationship is not itself evidence of no relationship, so in Online Appendix D.3, I use the TOST method of equivalence testing to generate bounds for the differences in ideological extremity (Rainey Reference Rainey2014). For the most competitive of contexts—where we would expect to see a large difference if preferences were driving the results in the previous section—we can reject differences greater than 0.05 and 0.06 in magnitude for the first and second dimension, respectively. These magnitudes represent bounds of approximately one-third and one-half of a standard deviation. Given that these are the maximum differences, it suggests that—even if there are some differences in extreme preferences across council members in partisan and nonpartisan governments—they are likely to be substantively small.

Taken together, the evidence in this section suggests that the types of politicians who run and win at the local level tend to have similarly extreme preferences regardless of institutional context. This similarity across partisan and nonpartisan governments suggests that differences in preferences cannot explain the differences in dimensionality in the previous section because there are not any particularly large differences in preferences to begin with.