What Causes Legislative Cartelization?
Chapter Five: Party Labels and Legislative Cartelization
Introduction
The previous chapter investigated several factors that contribute to variation in majority roll rates and also determine the extent of legislative cartelization by definition (since many of them were institutions affecting agenda control). The key factors, such as legislative professionalism (i.e. legislators’ available resources) and veto players, are institutional features that help or hinder the majority’s ability to control the agenda. Some legislative chambers have these institutions and others do not. Upon knowing what factors enhance and inhibit cartelization, it is natural to ask why certain chambers have such institutions while others do not. Such is the subject of this final chapter.
Table 5.1 below shows the distribution of professionalism across states, with the bottom quartile being labeled “less professional,” the middle fifty percent being labeled “somewhat professional,” and the top quartile being labeled “highly professional.” Recall that the Squire professionalization index aggregates legislators’ salary, benefits, staff, resources, and available time (length of legislative session) (Squire 2007). Recall as well that Chapter Four showed that legislative professionalism is one of the most significant factors affecting majority roll rates and its only basis therein could be agenda control. The question is, why do states like New Hampshire and New Mexico (the lowest professionalism scores in 1996) institutionally limit the ability of majorities in their legislatures to cartelize while states like California and Michigan (the highest professionalism scores in 1996) institutionally promote legislative cartelization?
Table 5.1: State professionalism categorizedLess professional / Somewhat professional / Highly professional
Alabama / Arizona / North Carolina / Alaska
Arkansas / Colorado / Nebraska / California
Indiana / Connecticut / Nevada / Florida
Kansas / Delaware / Oklahoma / Hawaii
Kentucky / Georgia / Oregon / Illinois
Maine / Iowa / Rhode Island / Massachusetts
Montana / Idaho / South Carolina / Michigan
New Hampshire / Louisiana / Tennessee / New Jersey
New Mexico / Maryland / Virginia / New York
North Dakota / Minnesota / Vermont / Ohio
South Dakota / Missouri / Washington / Pennsylvania
Utah / Mississippi / West Virginia / Texas
Wyoming / Wisconsin
This chapter presents a theoretical framework for understanding why some of the U.S. state legislatures have institutions that enable cartelization while others do not. The argument (presented visually in Figure 5.2) is that in polities where party labels become less prominent in public political discourse, party brand names are less important for legislator reelection, which in turn makes it more difficult for majority party leaders to control their parties and cartelize the agenda. Additionally, as party label prominence decreases, state legislatures will have fewer institutions with which majorities can control the agenda.
The following sections discuss the three general ways in which party label prominence in a state affects the ability of majority parties to form cartels. First, as party label prominence decreases, the connection between rank and file legislators and party leaders weakens, leading to agenda control difficulties for majority leaders. Second, as party label prominence decreases, legislatures have fewer cartel-enabling institutions – which makes cartelization more difficult. Third, as party label prominence decreases, legislators are granted fewer resources, leading to weaker connections between rank and file legislators and leaders, which in turn leads to less agenda control. After the theoretical discussion, I present empirical evidence in support of the third item – the connection between party label prominence and legislators’ granted resources. The final section summarizes the theory and empirical results.
Defining “Party Label Prominence”
This chapter uses the expression “party label prominence” many times and so it is necessary to establish a definition thereof. In this discussion, the prominence of party labels in a given state refers to the prominence of political parties in state politics in general. For clarity, it is helpful to imagine a continuum (Figure 5.1 below) where one extreme has closed-list PR systems (not actually found in the U.S. but helpful for illustrative purposes) and the other extreme has no political parties at all. For example, there are some subnational elections in the U.S. where candidates’ party affiliations are not on the ballot, though it is often the case that said candidates’ affiliations are public knowledge.
This continuum therefore ranges from politics dominated by individual personalities sans parties to party-dominated politics sans personalities (closed-list PR systems). Of course, this chapter is only examines U.S. legislatures and so there is almost no variance in electoral system across cases but the point here is to show what it means for party to matter less or more. As one moves from left to right on the continuum, party labels matter less and less for legislators’ reelection.
We should also expect that party label prominence is tied to the prominence of party labels in people’s minds when they think about politics; states where party labels are less prominent are states where people think more about the candidates than the parties with which the candidates are affiliated. Because of this, party label prominence in a given state is likely to also be observable in that state’s news media. Specifically, if party labels are more prominent in people’s minds when they think about or discuss politics, then it is also likely that political news stories will discuss party labels more than they would otherwise. Using the same logic, in states where party labels are less important, people think about party labels less often and news stories are therefore likely to discuss party labels less frequently.
To be sure, the vast literature on media and public opinion gives us reason to believe that the media may affect people’s thoughts about politics rather than public opinion dictating what goes on in the media (Zaller 1992, 1996; Kinder 1998; Druckman and Parkin 2005; Baum and Jamison 2006; Berinsky and Kinder 2006; Gilens, Vavreck, and Cohen 2007). For this study’s purposes, it does not matter which of these is true. Regardless of whether party label prominence originates with the media or with public opinion, the degree of party label prominence in the media should reflect the degree of party label prominence in a given polity.
To summarize, party label prominence can be thought of in three different ways that are all intimately connected. First, party label prominence may be measured by how important party labels are for legislator reelection. Second, party label prominence may be measured by how prominent party labels are in people’s minds when they think about politics. Third and finally, party label prominence may be measured by the amount of discussion devoted to party labels in political news stories. The third measure turns out to be the easiest to utilize and a later section uses it to demonstrate the connection between party label prominence and the amount of resources given to legislators in a given state. Before that though, the next section puts forth a three-pronged theory to explain the connection between party label prominence in a state and the extent to which that state’s legislature is cartelized.
Theory: Party Label Prominence and Legislative Cartelization
This section discusses a theory on the connection between the prominence of party labels in a given state and the degree to which that state’s legislature is cartelized. The complete theory is illustrated in Figure 5.2 and its three parts, marked by the three rightward-facing arrows on the top-left, are discussed in the following three subsections.
Item One: Party Label Prominence and Leader-Member Legislative Relations
One of the central tenets of Cartel Theory is that the party brand name must be important. Specifically, the reputation of a member’s party must affect both the member’s personal probability of reelection and the party’s probability of securing a majority (Cox and McCubbins 2005), which in turn leads to the need for members to delegate powers to their leaders and for those leaders then to control the agenda.
If party labels are less prominent in a given state, then party brand names will have less effect on legislators’ chances of reelection than they would in other states. If this is the case, then rank and file legislators will be less responsive to appeals from their party leaders because they will have less to gain electorally from their parties. Additionally, there will be less need for them to solve collective action problems as a party and therefore they will have fewer incentives to delegate agenda-setting powers to party leaders, both of which are important Cartel Theory assumptions. For majority party leaders, having fewer delegated powers and a less obedient rank and file makes it more difficult not only to directly control what goes on the agenda but also to know how all their members will vote on a given bill. In sum, as party brand name importance decreases, majority party leaders will have less control over the agenda and will be less certain about which bills they can safely allow onto the agenda – i.e. bills where the party will not be rolled.
Item Two: Party Label Prominence and Legislative Institutions
States where party labels are less prominent should provide their legislatures with fewer cartel-enabling institutions than other states. Similar to the logic connecting party label prominence to legislator resources, this argument is based on the idea that states where party labels are less prominent want their legislatures to be free from majority control. After all, if party labels in general are less prominent, then the “majority party” label is also less prominent. Since these states do not wish to enable majority control of the chamber, they are less likely to add institutions to their legislatures that would promote such control. Examples of such institutions would be veto players mentioned in Chapter Four – such as agenda-controlling Rules Committees and Speakers – as well as committee bill-killing powers; such legislatures should also be more likely to add committee recall powers to the floor. As Chapter Four established, these kinds of cartel-enabling institutions are connected to majority roll rates and therefore are important contributors to agenda control. In sum, in states where party labels are less prominent we are more likely to see legislative institutions that inhibit majority control and less likely to see institutions that promote it.
Item Three: Party Label Prominence and Legislators’ Resources
States where party labels are less prominent should allocate fewer resources to their state legislators. “Resources” here refers to legislative session length – i.e. legislators’ available time for legislative activities – as well as legislators’ salaries and staffs, all three of which are the central components to Squire’s legislative professionalism index (Squire 2007). The logic is that states where party is less important want to have more simple district-oriented politics and less “nasty” partisan politics. This leads to legislators in these states acting less as groups with collective policy goals and more as individual representatives servicing their districts. Since these states do not want their legislators to be collective teams of policy producers, there is less of a need to grant them resources they can use to do things as teams.[1]
When all legislators have fewer resources, party leaders have more difficulty gathering information about members’ preferences on bills, which in turn makes it more difficult for these leaders to keep potentially divisive bills off the agenda. Indeed, existing research suggests that less professional legislatures delegate fewer powers to party leaders (Clucas 2007), which in turn makes it more difficult for those leaders to control the agenda.
Additionally, legislators from less professional chambers, where legislators are granted fewer resources, are less likely to be career-oriented and progressively ambitious (Berkman 1994; Squire 1992; Thompson and Moncrief 1992). Less ambitious legislators are less concerned with reelection, are therefore less likely to delegate powers to party leaders, and are more difficult for those party leaders to control. Thus, legislatures with fewer resources available to legislators should be more difficult to cartelize than legislatures that allocate many resources to their members.
In sum, when party labels are less prominent in state legislative elections, states delegate fewer resources to those legislators, which makes it more difficult for majority party leaders in those states to control the agenda. The next section presents an empirical model testing this connection between legislators’ resources and party label prominence in the U.S. states.
Empirical Analysis of Item Three: Party Label Prominence and Legislators’ Resources
The previous subsection claims that states where party labels are less prominent are likely to give their legislators fewer resources with which to do their jobs. This section tests that claim empirically.
My conceptual dependent variable is legislators’ resources, the empirical measure of which is an index of chamber professionalism (Squire 2007), which takes into account legislators’ salaries, staffs, and time resources. Squire has stated that professionalism scores do not change much over time and so this study uses the 1996 scores because they are the scores used in Chapter Four.
The independent variable of interest is party label prominence in a given state. My empirical measure thereof is the number of times party labels are mentioned in news stories in a given state during an election campaign year, the merits of which are discussed above. Specifically, I used Lexis Nexus to compile – for each state – all the news stories with the following terms from 1/1/08 through 11/15/08: “election,” “state legislature,” “(state name),” “Republican,” and “Democrat.” My goal was for search returns to only contain news stories about elections for state legislative offices. While it is very likely that not all the search returns were exact on this point, the searching process was the same for every state and so there should be no biases. I was then able to count the total number of times the words “republican” or “democrat” (plurals included) were mentioned in all the news stories from a given state. This number, the total number of times parties were mentioned in news stories, is my measure of party label prominence, the independent variable of interest.
I also include each state’s 2008 population as a control variable because states with large populations are likely to have more newspapers and more news stories, which in turn means they are likely to have more party mentionings. Another way to theoretically control for this phenomenon would be to include a variable for the total number of news stories in each state but there is such a high degree of correlation between the total number of news stories and the total number of party mentionings – 97.5% – that doing so would likely cause a multicollinearity problem in my statistical model, leading to unrealistically large standard errors.
Table 5.2 below shows the results of a linear regression with the variables discussed. Although the estimation results appear to be very small in absolute magnitude, it is important to recall that the dependent variable is always a decimal whereas party mentionings can be in the hundreds and thousands while state populations can be in the millions, so we expect coefficients of small absolute magnitude. More importantly, the results are very statistically significant – over 99% confidence – and the positive sign on the “party mentionings” variable indicates that indeed states where parties are mentioned more often in news stories tend to have higher professionalism scores. In conclusion, states where party labels are more prominent give more resources to their state legislators than states where party labels are less prominent.
Table 5.2 The Effect of Party Label Prominence on Legislators' ResourcesLinear Regression, DV = 1996 Squire professionalism score
Observations are U.S. states
Estimate / Robust SE / % Confidence
# of times parties are mentioned / 8.58 x 10-5 / 2.4 x 10-5 / 99%
2008 state population / 7.56 x 10-9 / 2.36 x 10-9 / 99%
constant / 0.089 / 0.012 / 99%
N = 49[2] / R2 = .54
Figure 5.3 below shows predicted levels of legislator resources for different levels of party label prominence, based on the results from Table 5.2. We can see that, on average, having more party mentioning is associated with having more legislator resources (95% confidence bands included). Most of the data occurs in the range of 500 - 2,000 party mentionings, which explains why the predictions are less certain at the edges of the graph. It is also important to note that the dependent variable, Squire professionalism score, ranges from near zero to about .600 in the dataset, so what appears to be a small range on the vertical axis in Figure 5.3 actually covers a great deal of the variance in the dependent variable.