Legislative Civility, Gridlock, Polarization, and Productivity

Abstract

This article examines the relationship between legislative civility and legislative productivity in US state legislatures. The research employs data from the National Survey of State Legislative Lobbyists and from the State Policy Innovation and Diffusion (SPID) database. The former dataset is used to generate an overall civility index for each state as developed by Kettler et al. The SPID database allows one to measure the legislative productivity of a state legislature. Employing these data, negative binomial and Poisson regression models reveal that state legislatures rated as more civil by their own lobbyists produced significantly more pieces of noteworthy legislation than those legislative bodies rated as less civil. These results suggest that the quality of internal legislative dynamics matters for legislative productivity.

Introduction

This article examines the extent to which legislative civility produces important results in the legislative process. Specifically, the research reported here explores the role that greater civility among state legislators can play in fostering legislative productivity. We report evidence that legislatures which are rated as more civil by people who are highly knowledgeable about their operations produce more legislation on important topics of contemporary public policy. We also report findings indicating that state legislatures which are more civil are less prone to the ill effects of political polarization as it relates to legislative gridlock. We employ a dataset derived from a national survey of state legislative lobbyists conducted in 2018 and 2019. Survey respondents were asked to rate the degree to which they believe the legislators with whom they work in their respective states treat each other in a civil manner (Lovrich 2021). Over one thousand state legislative lobbyists from all 50 US states participated in the study, with many offering detailed comments and expressing concern over the decreasing levels of civility being displayed in the state legislatures in which they lobby. Many were former legislators and legislative staffers, many of whom reflected the widespread expression of concern over the state of civil discourse in Congress, state legislatures, local governments, school boards, public health districts, and even library boards across the country (Schuman 2016).

This study explores the question of whether greater civility among legislators is related to greater legislative productivity in the form of four specific outcomes – namely, passing new legislation, enacting greater proportions of bills submitted, passing larger numbers of major bills, and enacting a state budget in a timely manner. Our results show that civility is indeed linked to higher rates of policy adoption and lower levels of gridlock. These results suggest that internal legislative dynamics do matter for legislative productivity. This relationship remains even while considering the effects of other possible determinants shown to be related to legislative gridlock in prior research (e.g., Crosson 2019; Hicks and Smith 2009; Melusky 2020).

Civility, legislative productivity, and gridlock

The extent to which legislatures are productive has been the focus of considerable research, much of it examining the factors that inhibit the US Congress from passing legislation related to the important issues on the public agenda (e.g., Binder 2003; Mayhew 1991). Much of this research has examined the role that divided governments and partisan polarization have played in causing that gridlock in Congress (e.g., Coleman 1999; Howell et al. 2000; Jones 2001; Mayhew 1991). Similarly, additional research has been conducted with an eye to determining the role that divided state government, polarized state legislative parties, the presence of legislative term limits, and the legislatures’ gender compositions play in state legislative productivity and gridlock (e.g., Bernecker 2016; Bowling and Ferguson 2001; Hicks and Smith 2009; Melusky 2020). Collectively, this research has examined the relationship between these factors as they relate to productivity and gridlock as measured by things such as the total number of bills a state legislature enacts in a year or biennium, the percentage of bills introduced that were passed, the number of “important” bills passed, and whether a state’s budget bill was passed prior to the beginning of the states’ forthcoming fiscal year (e.g., Andersen, Lassen, and Westh Nielsen 2020; Bernecker 2016; Crosson 2019; Rogers 2005).

We add to this discussion of legislative productivity by examining the role civility plays in shaping legislative productivity and legislative gridlock, with the notion of “productivity” being reflected in the number of statutes being enacted into law, and the concept “gridlock” being reflected in draft legislation that failed to gain passage or was unduly delayed. Although civil discourse in the political arena has been a concern for a number of years, knowledge about the extent to which incivility hampers governance and harms legislative deliberation remains in its relative infancy. Some early positive steps have been taken toward enhancing this knowledge with the advent of some recent publications, such as an edited volume (Lovrich 2021) that reported findings derived from the National Survey of State Legislative Lobbyists conducted in 2018–2019. Along with Kettler, Fowler, and Witt (2021), these studies find that states with higher-rated levels of civility tend to have more professional legislatures, lack term limits for their members, have more nonwhite legislators serving, and feature less polarized political parties. Also, states with less competitive political parties and less economic inequality tend to have more civil legislatures (Kettler, Fowler, and Witt 2021; Schreckhise and Benjamin 2021; Schreckhise, Chand, and Benjamin 2021).

As noted by Goovarts (2022, 2), [p]olitical incivility can be defined as “politicians’ use of a disrespectful, impolite, or rude communication style.” Incivility can include uncivil behavior that violates social norms as they relate to interpersonal interactions (see also Muddiman 2017). More specifically, uncivil behavior refers to “discursive behaviors that represent the rejection of communication norms pertaining to considerate, courteous, and respectful discussion” (Hopp 2019, 206). As such, this article poses the specific research question: To what extent does incivility among legislators inhibit legislative productivity and enhance the danger of gridlock in state legislatures?

We suspect that less civil legislatures would be less productive and more prone to gridlock for several reasons. First, past research has found that uncivil behavior in the political arena by public figures weakens the public’s trust in political leaders (Goovaerts 2022; Mutz and Reeves 2005; Skytte 2021) and in the political system in general (Forgette and Morris 2006; Mutz and Reeves 2005). Such incivility-spawned mistrust even extends to such areas as news media credibility (Borah 2013; Prochazka, Weber, and Schweiger 2018; Naab et al. 2020) and public trust in the validity of science research (Chinn and Hart 2022). That same mistrust could translate into mistrust among politicians which would include those serving in the very state legislature from which the incivility emanates, including members of their own parties (see Druckman et al. 2018).

Second, incivility negatively affects the needed interpersonal dynamics necessary for legislative productivity. As Mutz (2007, 623) notes, “[w]hat is known about incivility – whether in political discourse or some other context – is that it heightens levels of arousal” (see also Mutz and Reeves 2005). This heightened sense of arousal manifests itself in emotional responses such as anger and sensitivity to slight (Valentino et al. 2011; Weber 2013), rendering an individual less willing to accept new information that contradicts their own views and less likely to compromise (Gervais 2019). Additionally, among the public, it has been found that over the course of the last four decades, Americans have expressed increasingly negative feelings toward members of the opposite party along with deepening loyalties to the ingroup of copartisans (Iyengar, Sood, and Lelkes 2012; Iyengar and Westwood 2015). More importantly, perceived uncivil behavior increases negative feelings toward members of the opposite party (Mutz 2007; Skytte 2021). To be sure, these findings have been limited to studies among the general public. However, these dynamics may very well be in active play among state legislators as well.

Third, enhanced civility allows for communication and policy making free of substantive or procedural issues ancillary to the policy at hand that might impede the search for, discovery of, and production of shared benefits or the facilitation of the process producing compromise, bargaining, and exchange of benefits within the issue or across issues. Those attributes would enhance the probability of agreement on legislation and hence increase the number of bills passed. This process has traditionally been labeled “logrolling,” and while not confined exclusively to the legislative arena, is most visible there (see Thrasher 2016). Along these lines, past research has found that incivility in the workplace is related to greater employee turnover, lower job satisfaction, and reduced organizational performance (e.g., Anjum et al. 2018; Estes and Wang 2008; Miner et al. 2019).

To be sure, as Loomis (2004, 2) noted, “the connection between civility and [legislative] deliberation seems straightforward. Comity would seem a necessary, if not sufficient, condition for deliberation on the major issues of the day.” Although declining levels of civility in American politics are a concern for many, it remains unclear to what extent incivility hampers the deliberative process and impairs governance in the 50 US state legislatures.

Greater interpersonal civility among legislators is hypothesized to lead to greater productivity because legislative bodies are themselves social entities that require interpersonal cooperation for their successful operation. When individuals elected to legislative offices can cooperate in their legislative work because of their positive personal relationships and interpersonal trust, this diffuse trust will elevate their legislature’s level of productivity. As Uslaner (2000, 35) noted in his observations of the relationship between the interrelated notions of civility, trust, and comity, and the productivity of the US Senate,

[s]trong friendship circles across party lines signify a legislature marked by trust. This is where civility turns into comity. Comity is more than being civil to others. It also involves reciprocity, which simply means that people must respect their promises and obligations to others. They must also recognize that another point of view is legitimate. Reciprocity involves respecting other people and their expertise. It means willingness to make deals and committing oneself to sticking with the agreement. Without reciprocity, you get either rule by a partisan majority or stalemate.

Thus, we suggest that members of more civil legislatures will be more likely to work in a collaborative fashion on issues on which they can agree. Relatedly, in less civil legislatures members are less likely to trust one another and less likely to engage in active collaboration. Much of the legislative process requires trust that members will keep their word when vote-trading deals are brokered. In this regard, Thurber (2000) found that declining levels of trust among US Senate Budget Committee members led to a decline in the quality of deliberations held on that committee. Interpersonal conflict resulting in uncivil behavior may lead to some members of those legislatures being unable to cooperate on otherwise consensual matters, and thus inhibit the passage of legislation.

At the national level, some other prior research has indeed observed such a relationship between civility and legislative productivity. Dodd and Schraufnagel (2012) found that landmark legislation in the US Congress occurred during times when the combined effects of polarization and incivility were at moderate levels. Similarly, Schraufnagel (2005) found that longer delays in judicial nominees in the US Senate are related to greater interpersonal incivility among its members. At the local government level, Yuan and Schraufnagel (2020) observed that local government bodies in the US were viewed as less productive when members noted incivility among the members of their boards. The research reported here thus examines the extent to which state legislatures with greater civility are more likely to adopt new legislation than are those with more uncivil behavior.

Data and hypotheses ¹

Independent variable

For the primary independent variable, we measure civility as perceived civility. Data for this measure come from the National Survey of State Legislative Lobbyists. That survey was conducted by academic researchers at 12 US universities, drawing upon support from the National Institute for Civil Discourse at the University of Arizona, the Thomas S. Foley Institute for Public Policy and Public Service, the William Ruckleshaus Center, and the Division of Governmental Studies and Services at Washington State University. The lobbyists’ names and contact information (including email addresses) were collected from those secretaries of states offices which maintained lists of lobbyists registered with the state. In states without such listings, additional effort was expended to gather contact information through a variety of other strategies, such as searches of online legislative directories and internet searches. The survey was internet-based (Qualtrics platform) with follow-up mail surveys conducted in 2018–2019. ²

A US state-level civility index was created from the survey responses. We employed the method developed by Kettler, Fowler, and Witt (2021) calculating the “average” lobbyist’s rating of each respective state’s level of civility. The resulting index considers the scores which respondents gave for their own state legislature in terms of (1) the legislature’s quality of deliberation; (2) the prevailing level of civility in their state legislature; (3) the general level of civility in their state; and (4) the general level of civility among state legislators. The specific survey items are listed in Table 1.

Table 1. Civility index survey items

Multilevel-ordered regression models were estimated for the first three items with fixed effects for the individual-level variables and random effects for the state-level variables. ³ An intercept was then estimated for each state that reflects the differences for each state, controlling for the individual-level differences. This method considers individual-level demographic and political differences when producing the scores, generating “mean” ideal values for each state of the first three questions as deviations from the fourth survey question as a common point of comparison to ameliorate the potential for respondents from different states to interpret the first three questions differently. The result is the civility index which has a mean of near zero, the standard deviation of 0.38, and ranges from a minimum of −0.97 (Oklahoma) to a maximum of 0.76 (Maryland). Figure 1 displays a map of the score for each of the 50 states (see Kettler, Fowler, and Witt 2021, 10–11). Summary statistics for this and all other variables used in our analyses are displayed in Table A.1 in the Supplementary Material and the sources, coding, and directionality of each variable are set forth in Table A.2 in the Supplementary Material.

Figure 1. Civility index scores by state.

Note: Higher scores reflect higher rates of perceived civility among legislators.

Source: Values from the Kettler, Fowler, and Witt (2021) and National Survey of State Legislative Lobbyists (2022).

In the survey, many lobbyists commented on the connection between civility and legislative effectiveness in accomplishing the work of legislating in the public interest. Among the 1,200+ survey respondents, 72 were former state legislators and 288 were persons who served as legislatives aides or committee staffers prior to becoming lobbyists. Many “lobbyists” were serving as contract lobbyists or lobby firm employees, and many others were in-house lobbyists for private interests, nonprofit advocacy groups, or worked for public entities such as local governments, special districts, or state agencies.

Dependent variables and hypotheses

The following analysis presents four models predicting separate dependent variables, each of them used in previous research as a measure of legislative productivity (or conversely, legislative gridlock), along with corresponding control variables that have been employed in conjunction with those dependent variables. We provide a brief description of each of the dependent variables below with a further description provided as we present each model further below.

The first model examines the total number of bills passed by each of the state legislatures in 2018 and 2019, an approach that has been used in several prior studies of legislative productivity and innovation, including Bowling and Ferguson (2001), Crosson (2019), Gray and Lowery (1995), Hicks and Smith (2009), and Rogers (2005). The data for total counts of legislation passed are extracted from The Book of States. Using this as the primary dependent variable in our first model, we test our first hypothesis:

H₁: State legislatures with higher civility index scores will pass a larger number of bills than states with lower civility scores.

Similar to previous studies of legislative gridlock (i.e., Hicks and Smith 2009; Rogers 2005), our second model tests another hypothesis regarding the percentage of bills passed of those introduced by each of the state legislatures in 2018 and 2019. Data for the percentage of bills passed are likewise extracted from The Book of States.

H₂: State legislatures with higher civility index scores will pass a larger percentage of bills than states with lower civility scores.

Testing our third hypothesis, another model examines whether the state passed notable or needed legislation in 2017, somewhat similar to Mayhew’s (1991) measures of “important” legislation (see Bernecker 2016; Crosson 2019). The data for this dependent variable are extracted from the Boehmke et al. (2020) State Policy Innovation and Diffusion (SPID) Database.

H₃: State legislatures with higher civility index scores will pass a larger number of important pieces of legislation than states with lower civility scores.

The final model examines whether the state passed a budget “on time” (i.e., prior to the start of the state’s next fiscal year) in 2019 for the 2020 fiscal year as similarly employed by Andersen, Lassen, and Westh Nielsen (2020), Birkhead (2016), Klarner, Phillips, and Muckler (2012), and Melusky (2020). The data for on-time budgets are from the National Conference of State Legislatures. Thus, we present our fourth hypothesis:

H₄: State legislatures with higher civility index scores will be more likely to pass state budgets on time than states with lower civility scores.

Also included in our analyses are variables that past research has shown to be related to each of the dependent variables to represent legislative outcomes. We discuss these measures in greater detail below.

Results

To determine the relationship between legislative civility and legislative productivity, we first examine the link between the civility index and the total number of bills passed by state legislators, using a count of the total numbers of bills passed by each state legislature for the 2018–2019 period. Such a measure has been used by others (see, e.g., Bowling and Ferguson 2001; Crosson 2019; Gray and Lowery 1995; Hicks and Smith 2009; Rogers 2005). As Tsebelis notes, “policies are the principal outcome of a political system” (Tsebelis 2002, 6), and as such, this measure serves as an appropriate starting point for examining the civility-productivity link.

The analytical model includes other covariates that have been linked to the quantities of legislation passed by state legislatures. These specific variables of interest include: levels of legislative professionalization ⁴ as developed by Bowen and Greene (2014), state elite ideology, the presence of unified partisan state government, the size of the state’s population), whether the state’s voters can propose their own legislation via ballot initiatives, and the economic health and economic capacity of the state (Berry and Berry 1990; Boehmke and Skinner 2012; Gray 1973; LaCombe and Boehmke 2021; Mallinson 2020; Mallinson 2021; Rom, Peterson, and Scheve 1998; Walker 1969).

We also include in the first model a measure indicating whether legislators in either or both chambers in each state legislature have a legislative rule or constitutional provision limiting the number of bills introduced, either as a number that each legislator may introduce or a deadline for when they can be introduced. We do this because Squire (1998) found that state legislatures with such rules passed a higher percentage of bills. These rules are chamber-specific, and they vary from state to state. For example, Florida’s House Rule 5.3 permits a member to introduce only six bills during each legislative session. Some states impose the limit only after a certain period in the session has passed, like Arizona’s House Rule 8(c) which only imposes a cap on bill introduction after the fourth day of the session. Each state was given a value of “2” if both chambers had some type of limit on bill introductions, “1” if such a rule existed for only one chamber, but not the other, and “0” if neither chamber had such a rule.

The results for the first model are presented in Table 2. We employed a negative binomial model due to the discrete and nonnegative nature of our dependent variable, which is heavily skewed. The over-dispersed nature of these count data can render OLS and Poisson model results that are biased and inconsistent (Hilbe 2011; Paternoster and Brame 1997), potentially resulting in Type II errors (Payne et al. 2018). We report the incident rate ratios (IRR) which indicate the extent to which a one-unit change in the independent variable leads to a percentage change in the dependent variable, reflected in the extent to which the value of IRR is greater or less than 1.0 (Piza 2012).

Table 2. Total number of bills enacted in state legislatures, 2018–2019

Note. The total number of bills enacted in 2018 and 2019 is the dependent variable in this negative binomial model.

* p < 0.05.

** p < 0.01.

*** p < 0.001.

The results reported in Table 2 reveal a statistically significant relationship between the civility index scores and the number of laws adopted by states. Specifically, as one moves from the state with the lowest civility score (−0.97) to a state at the mean civility score value (0.0), one moves roughly 1 unit on the civility index. This one unit of change is thus reflected in a roughly 64% increase in the number of laws passed by state legislatures in 2018–2019, ceteris paribus. Additionally, states with more bills introduced, more liberal states, and larger states passed more legislation. The predicted values of legislative productivity in Table 2 are presented graphically in Figure 2 across the range of civility index scores.

Figure 2. The relationship between legislative civility and the predicted number of bills enacted, 2018–2019.

Note: Shaded area reflects 95% confidence intervals. Estimates are obtained from Table 2.

We also employed a second measure of legislative productivity, namely the percentage of bills introduced that were passed by each legislature for the 2018–2019 period. Such an approach has been used in past research to examine such things as the effect of divided government on gridlock, with “gridlock” being associated with the failure of proposed legislation (Coleman 1999; Hicks and Smith 2009; Howell et al. 2000; Rogers 2005). As Edwards (1997, 547) and his colleagues note when examining the effects of divided government on legislative activity, “[i]f our concern is whether or not divided government obstructs the passage of legislation, we ought to investigate legislation that was proposed that did not pass” [emphasis original]. We employ the same logic here as it relates to legislative civility, including the percentage of bills as the dependent variable.

We also take into direct account the total number of bills introduced. This is because it is possible that state legislatures may very well have a maximum capacity for legislation each year; thus, some states may eventually pass a smaller percentage only because more bills were initially introduced. Indeed, considerable variation exists across the states in the number of bills introduced for the two years we examine. For example, Ohio’s legislature saw a total of 1,832 bills introduced in 2018 and 2019, while New Jersey’s legislators introduced more than 10 times that number for the same two-year period.

Table 3 reveals that states with legislatures rated as more civil produce a higher percentage of the bills originally introduced. This is the case even when taking into the account the extent to which a state’s political parties are ideologically polarized or whether the state’s legislative chambers impose bill introduction limits, as well as other variables that have been found in past research to have a relationship with a legislature’s bill passage rate (Bowling and Ferguson 2001; Gray and Lowery 1995; Hicks 2015; Hicks and Smith 2009; Jones 2001; Rogers 2005; Squire 1998). Regardless, our model in Table 3 reveals that a one-unit change in the civility index is related to a 17% increase in the percentage of bills being passed in 2018–2019.

Table 3. Percentage of bills enacted, 2018–2019

Note. The percentage of bills introduced that were enacted in each state in 2018–2019 is the dependent variable in this OLS regression model. Values are unstandardized coefficients.

* p < 0.05.

** p < 0.01.

*** p < 0.001.

To be sure, measures of the absolute volume of legislation passed and the percentage of bills passed obscure one important fact: not all legislation is of equal importance. Each year, states collectively pass thousands of bills, some of which are important, while others are quite less so (e.g., noncontroversial “technical bills” submitted by state agencies to address administrative needs). As such, employing only measures of the total number of bills passed and the percentage that passed may obscure the reality that some legislatures may be more effective at enacting a plethora of legislation that has little effect, while other states may pass relatively very few pieces which have profound ramifications. Faced with a similar concern, David Mayhew (1991) in his study of the effect of divided government on Congressional productivity included in his tally of enacted statutes bills only those mentioned in post-congressional wrap-up news articles appearing in the New York Times and the Washington Post, or which were subject to later noteworthy historical examination. Others, such as Howell et al. (2000), expanded their tally to include somewhat similarly the number of pages CQ Almanac dedicated to these statutes as an additional measure of each law’s importance.

Clearly, employing similar methods to indicate the importance of laws at the state level would be impossible. The newspapers mentioned above do not produce similar end of year wrap-ups for all 50 states. Examining other state-specific major newspapers that may write end-of-session articles in every state would yield inconsistent data since the myriad different newspapers in each of the 50 states would likely employ very different standards for what constitutes “important” legislation. Thus, we are confronted with finding another source of data on legislation that could be deemed “noteworthy.” To be sure, Crosson (2019) took a slightly different approach; he examined whether a state had passed a law or set of laws stemming from, and essentially mandated by, the passage of the federal Affordable Care Act in 2010, such as state bills creating state health care exchanges or expanding state Medicaid coverage (see also Bernecker 2016 for a similar approach). Unfortunately, during the period our study examines, we were unable to find a similar federal law that triggered similar essential state legislative implementation activity.

However, the dataset developed by Boehmke and his colleagues (Boehmke et al. 2018) in their SPID dataset allows us to identify the number of “important” new laws adopted by each state. Their scores include additional measures of policy adoption recorded by other researchers such as laws related to health policy by Silver and Macinko (2013), abortion policy by Kreitzer (2015), elections legislation by Biggers and Hanmer (2015) as well as others (e.g., Boushey 2016; Warshaw and Caughey 2015). They also include state laws passed that were based on model state laws drafted by the Uniform Law Commission such as the model state Administrative Procedure Act and the model state Uniform Commercial Code. To be sure, it is admittedly impossible to determine whether the laws included in this set include all possible laws that could be deemed “important,” but they do include a very broad range of state laws – 728 laws, in total – that are sufficiently noteworthy to be examined by past researchers or proffered by the Uniform Law Commission and frequently are passed in multiple states.

Our dependent variable in this analysis consists of a count of the total number of laws in this dataset passed by each of the states in the years 2016 and 2017. This includes a total of 105 instances of laws passed by 43 states during that two-year period, reflecting in each instance laws comparable to those that had been passed by an average of 19 other states by the end of 2017. By adopting this approach our analyses are not directed toward examining which states were early adopters of specific policies (i.e., “innovators”), nor do we examine patterns of diffusion of new policies across states. Instead, we examine the number of new pieces of important legislation as identified by Boehmke et al. (2018) as adopted by state legislatures during those years, the most recent years available.

We include as a control variable in this model a count of the number of laws among the 728 pieces of important legislation included in the SPID which each state had already adopted prior to 2016. We do this because if a state had adopted fewer of these laws prior to 2016, then there would remain a larger number in the set of potential laws the state could pass in 2016 and 2017. We also consider the other factors which Crosson (2019) included in his study of effective predictors of passage of the important post-Affordable Care Act state legislation and which Bernecker (2016) found to be linked with the passage of state welfare reform laws. As with the model presented in Table 2, the count and skewed nature of our dependent variable in this model warrants the use of a negative binomial statistical model. Our findings from this model are set forth in Table 4, similarly employing incident rate ratios.

Table 4. Number of “important” bills passed, 2016–2017

Note. The total number of bills enacted in 2016 and 2017 identified by Boehmke et al. is the dependent variable in this negative binomial model.

* p < 0.05.

** p < 0.01.

*** p < 0.001.

Even when considering a state’s degree of legislative party polarization along with the other factors found by Bernecker (2016) and Crosson (2019) to predict the likelihood that important state legislation passes, civility plays a very prominent role. As one moves from the state with the lowest rated civility to the state with the median civility index score (i.e., moving one unit of value of the civility scale), ceteris paribus, the state with the higher score produced 97% more of this type of legislation than the state with the lowest score. In fact, this is the only predictor in this model that reaches the standard levels of statistical significance.

Studies of gridlock at the state level have also explored the area of budgets (e.g., Andersen, Lassen, and Westh Nielsen 2020; Melusky 2020). As partial shutdowns of the federal government garnered national attention in 2013, scholars looked to US states to examine what factors determine when similar budget impasses occur at the state level. To be sure, not all state budget impasses result in government shutdowns. Like Congress, some state legislatures may pass continuing resolutions or take other measures to prevent the shuttering of state government office doors (NCSL 2010). Nonetheless, a state legislature’s failure to produce a new budget by the beginning of the start of the new fiscal year brings uncertainty about paychecks for state government workers and payments to state government contractors. Despite these problems, state budget impasses do occur. In fact, Klarner, Phillips, and Muckler (2012, 998) found that in the years between 1961 and 2006, over 15% of all state budgets were passed after the beginning of their respective fiscal years. This state of affairs obtained even though all 50 states, either through state statute or via a state constitutional mandate, require that legislation authorizing a new budget be passed by the legislature and signed by the governor before the start of that next fiscal year (Kirkland and Phillips 2018). Thus, annual, or biennial budget legislation certainly can be considered another type of “important” legislation.

Examinations of the passage of on-time budgets provide an additional benefit when examining states’ capacity for legislative productivity. This is because it avoids the problem of measuring the demand for policy change, or as Kirkland and Phillips (2018, 179) deem “the denominator problem.” That is, it is difficult (if even possible) to measure the demand for potential policy change across all policy areas, either in the form of political support for this change or in the form of genuine need for change in some objective sense. However, states must pass budgets each year, or in a few cases with biennial budget cycles, every two years.

To explore whether more civil legislatures are more likely to pass their budgets on time, we created a dichotomous variable indicating whether the state legislature had passed (and the governor signed) an annual state appropriations bill prior to the start of the succeeding year fiscal year 2020 (usually on July 1). The dichotomous variable created is coded “1” if the budget was enacted prior to the fiscal year’s start date, 0 if it was not (see Melusky 2020 for a similar approach).

We also include other covariates linked in past research to the probability a state adopts an on-time budget. These include measures of the size of the state budget, whether the state experienced a budget surplus during the previous fiscal year, and whether the state’s constitution required the passage of a balanced budget (as opposed to merely being required in statute). Like Melusky, who found that a legislature’s gender balance was a significant predictor of whether a legislature passed an on-time budget, we include a variable indicating the percentage of the state legislature who were women. We also take into account the degree of polarization between the legislative parties, whether control of the state government was exercised by one political party, the length of the legislative session (Birkhead 2016; Kirkland and Phillips 2018; Klarner, Phillips, and Muckler 2012; Melusky 2020), each of these variables have been found to offer additional predictive value in previous studies. This model also includes measures of whether each state’s governor can exercise a line-item veto, the states’ populations, and the states’ per capita GDPs.

To be sure, unlike our previous models, the following model considers the role of the governor in the legislative process. We do this by measuring not only whether the budget was passed by the state legislature, but also whether the governor signed the legislation approving the budget. We do this because, as with Melusky (2020), we consider it important to factor in the role that governors’ potential power play via line-item vetoes on budgetary items. Such gubernatorial powers might prompt legislatures to pass on-time budgets. If a governor is granted considerable authority to veto parts of a budget as passed by the legislature, the debate could be less intense and shorter in the legislature by virtue of the fact the side wanting to include something would be doing so with the knowledge their attempts could come to naught should the governor feel inclined to veto that portion of the budget.

Surprisingly, Model 5.1 in Table 5 reveals that states with constitutional requirements for passing a balanced budget prior to the beginning of the next fiscal year were less likely to do so. As suspected, however, states with governors possessing the power to execute line-item vetoes were more likely to pass a budget on time, as were states with lengthier legislative sessions. Considering these and the other covariates in the model, we once again find that a state’s civility score serves as a significant predictor. States that scored higher were more likely to pass budgets on time than states with lower scores. ⁵ Additionally, more polarized state legislatures are less likely to pass an on-time budget, as well.

Table 5. On-time budgets

Note. Whether the state enacted a budget for FY20 prior to the beginning of that fiscal year for the state is the dependent variable in each of these logit models.

* p < 0.05.

** p < 0.01.

*** p < 0.001.

We also explored the possibility that polarization may interact with civility – in particular, we suspect that higher levels of legislative civility could moderate the effect that polarization might have on legislative adoption, since past research has shown that state legislatures with more ideologically polarized political parties are less likely to pass budgets on time (Kirkland and Phillips 2018) and produce less legislation, in general (Bowling and Ferguson 2001; Jones 2001). Model 5.2 in Table 5 tests this possibility by including an interaction term for those two variables. The interaction term yielded is both positive and statistically significant. ⁶

To get a better sense of the meaning of the positive interaction term between the civility index and polarization, we display in Figure 3 the predicted probability of the passage of an on-time budget at three estimated values for the polarization variable: one standard deviation below the mean, the mean, and one standard deviation above the mean. The probability of passing a budget on time is then plotted across the range of civility score values. At the left-hand side of the x-axis where states are rated as being very uncivil, there is considerable variation among these three groupings. Highly polarized states are relatively unlikely to pass their budget on time (less than 14% on time). Moderately polarized states (i.e., those at the mean of the polarization variable) found in the middle of the y-axis are more successful at passing on-time budgets (at roughly 50% on time). Finally, states that are rated the most civil are even more likely to pass a timely budget (75% of the time). Conversely, on the right-hand side of the x-axis where the civility scores are at their highest, polarization plays no role in on-time budget passage. In other words, all states with higher rated values for civility are equally likely to pass their budget on time (at a rate approaching 100% on time), regardless of how polarized their state legislative parties may be. This finding suggests that the negative effect of polarization on legislative productivity, insofar as it involves passing state budgets, can be mitigated by higher degrees of civility with the state legislature.

Figure 3. The relationship between legislative civility, polarization, and the probability of the passage of a budget prior to the start of FY2020.

Note: Shaded area reflects 95% confidence intervals. Estimates are obtained from Model 5.2 in Table 5.

Discussion and conclusion

To what extent are more civil legislatures able to produce more new legislation? Our findings suggest that there is indeed a positive relationship between legislative statutory productivity and civility, confirming H₁ – namely, state legislatures that lobbyists rate as more civil do indeed pass more pieces of legislation. The model in Table 1 estimates that the most civil legislature passed more than 64% more legislation than the least civil ones in 2018 and 2019, while Model 2 estimates that the most civil legislature passed a higher proportion of bills that were introduced, providing support for H₂. Perhaps even more importantly, the results reported here suggest that our measure of civility would seem to be the best predictor in our models of total legislative productivity. Only three other state-level variables included in our first two analytical models – a state’s population and the total number of bill introductions in the first model, and the presence of limits on bill introductions in the second – offer any additional significant predictive power.

Such a pattern remains when we turn our attention to more significant pieces of legislation. The model presented in Table 4 (testing H₃) reveals that states with more civil legislatures passed significantly more pieces of the “important” state legislation identified in the Boehmke et al. (2018) SPID dataset than less civil ones. Such a relationship between civility and legislative productivity is also evident when examining the probability that a state passed their FY 2020 budget on time (see Model 5.1), providing evidence for H₄. Additionally, our findings suggest that civility and polarization interact in a way to combat both partisan polarization and legislative gridlock – namely, that states with legislatures rated as being more civil are more likely to pass these on-time budgets. This is true regardless of the extent to which their state legislative parties are ideologically polarized. Highly polarized, yet civil, legislatures are just as likely to pass their budgets on time than less polarized and less civil ones.

It is worth noting that our findings may generate additional questions requiring attention. First, our measure of civility does not locate the locus of the incivility documented in the national survey of state legislative lobbyists. Although one might assume uncivil behavior is more likely to occur between individuals on opposite sides of the ideological spectrum in different parties, the Kettler, Fowler, and Witt (2021) measure we employ does not reveal this. In other words, however unlikely it may seem, it very well could be the case that the perceived incivility is not just occurring between Democrats and Republicans but within legislative party caucuses as well. A good number of the comments made by state legislative lobbyists working in one-party dominant settings report that the incivility within dominant parties is a major factor inhibiting effective caucus deliberations (Benjamin et al. 2022). At the very least, even if one can assume that the bulk of uncivil legislative behavior is directed across the aisle, some of it also occurs within party caucuses.

To this point, one of the themes of lobbyists’ comments in the National Survey of State Legislative Lobbyists was that the proportion of “true believers” has grown and the presence of “practical problem solvers” has ebbed in the state legislatures in which they ply their craft -- leading to incivility and diminished prospects for “getting to yes” in the terminology of alternative dispute resolution (ADR). Additionally, in one state, it was noted that the “real contest for influence is among factions in the GOP wherein the uncompromising nut cases must be fought tooth and nail.” In ADR the observance of ground rules to which all parties pledge faithfulness is essential to dispute resolution. It is the routine violation of these fundamental ground rules (such as truthfulness) which characterizes much of the incivility being witnessed in the contemporary state legislative processes according to many of the lobbyists surveyed. It is worth noting, however, that we found that the correlation between the civility index and the degree of partisan competition in the state legislatures is mildly negative; that is, state legislatures dominated by a single party see only slightly higher civility scores than states that have roughly equal numbers of Democrats and Republicans in their legislative chambers. ⁷

A couple of caveats are in order regarding our findings. First, because of the relatively small number of cases in our analyses and the number of covariates included in our models, the possibility exists that the stability of the coefficients in these models is less than optimal. To be sure, the fact that our “sample” consists of the total population of American states, issues related to sampling distributions are less of a concern. Nonetheless, we examined the stability of our core independent variable by running our models using different combinations of the other covariates and examining whether the value of coefficient for civility index changed when used in these different combinations. The direction of the civility index remained consistently positive; however, the extent to which the coefficients for this variable reached standard levels of statistical significance was less consistent. ⁸

Second, although the models presented in this article provide evidence of a link between legislative productivity and legislative civility, we cannot discern the direction of the casual arrow with absolute certainty. That is, we cannot say lower levels of civility cause lower levels of legislative productivity. It is possible, if perhaps unlikely, that the causal direction runs the opposite of what we have suggested, and that more productive legislative sessions lead to more civil relations among legislators. Perhaps, buoyed by a productive legislative session, legislators are “nicer” to each other than they would otherwise be. ⁹ Moreover, this could occur as an iterative process, that is, when one important bill passes, relations among legislators improve, opening the door for future collaboration. Conversely, when an important bill fails or a budget bill is delayed, relations sour, and further collaboration is impeded, causing additional legislation to founder.

Regardless of these caveats, the analyses and results presented here show that a demonstrable relationship between legislative civility and legislative productivity does exist. Even when a number of other factors linked to legislative productivity and gridlock are taken directly into account, we find evidence that legislatures that produce more (and more important) legislation are rated by those individuals who work with them closely as being a more civil body. Such findings not only provide insight into the legislative dynamics of states but offer insight to the productivity of legislatures in general. Future work should feature comparative cases studies of the most productive state legislatures and the least productive state legislatures, noting the extent to which civility appears as a key element of discrimination between these two groups along with other factors of interest. Doing so can provide a richer and more nuanced understanding of the relationship between legislative civility and legislative productivity.