I recently published an article about the effects of US state legislative term limits on legislators’ behavior. The findings of the article suggest that in the presence of term limits, state legislators in more professionalized legislatures spend less time on constituency service and more time fundraising with their caucus. In the article, I argued that since term limited legislators do not rely on their constituents to maintain a long term career in politics, they are less likely to spend time on constituency service than their peers in non-term limited legislatures. Instead, term limited legislators spend more time on fundraising since it helps them get re-elected in the short term (as they cannot rely as much on incumbency advantage to get re-elected) and it helps them get elected to other positions after they are term limited out (as they will need resources to win open seats or win seats against incumbents).
The findings of my article may suggest that since the long term political career needs of term limited legislators are different than non-term limited legislators, state legislative political parties may function differently within term limits than without them. In the absence of term limits, legislative parties can fulfill their electoral role by helping their members get re-elected in their districts as incumbents. Since incumbents often easily win re-election, and do not have to spend a great deal of resources to win re-election, legislative parties may not have to focus as much time on fundraising. They may focus more on satisfying the partisan and fiscal needs of their members’ districts. However, in the presence of term limits, if legislative parties want to help their members maintain their seats in the short run and stay in politics in the long run, they may need to spend more time on helping their members raise money to run for re-election and election to other offices.
If the role of legislative parties is different under term limits versus under no term limits, that may suggest that legislators focus on different criteria when choosing their party leaders in each situation, as different types of leaders may be better at fulfilling different roles. Without term limits, legislators may be concerned about the ability of their party leaders to deliver on the policy promises they made to their constituents. So when choosing party leaders, these legislators may care about legislators’ ideological positions when choosing their leaders. However, under term limits, legislators may care less about delivering on policy promises, because once again, they rely less on their constituents to maintain a long term political career. Instead, these legislators may care more about the ability of party leaders to raise money so they can use the money to get elected in the short run and long run.
In order to determine if non-term limited legislators focused more on ideology when choosing party leaders than term limited legislators, I analyzed data from the Shor-McCarty Individual State Legislator Ideology Dataset. I used this data to determine if a legislator’s ideology played a role in whether they were selected as the Speaker of their chamber. In my analysis, I only focused on the year 2007, and only included legislators who were members of the majority party in their chamber.
Using the Shor-McCarty data, I created two ideological variables. First, I created a variable that measured the absolute distance of a legislator’s ideal point from the ideal point of the median member of their party (to capture whether the median member of the party was more likely to be selected as party leader, as per the median voter theorem). Then I created a variable that measured how extreme a legislator was to capture whether more extremely partisan legislators were more or less likely to be selected as party leaders. For Republican legislators, the variable equals the legislator’s ideal point minus the the ideal point of the median member of their party. For Democratic legislators, it equals the ideal point of the median member of their party minus the legislator’s ideal point. By calculating these variables this way, larger values indicate that the legislator is highly partisan while smaller values indicate that the legislator’s preferences are closer to the median of the chamber.
Next, to determine if a legislator was a party leader, I collected data on whether a legislator was Speaker of their respective chamber (coded a 0 for non-Speaker and 1 for Speaker). To control for tenure, I also collected data on how many terms each legislator served in their chamber. I created a ratio variable which was equal to the number of years a legislator served divided by the longest serving legislator in their chamber’s number of terms. I expected legislators who served longer terms would be more likely to be selected as the Speaker. I also collected data on whether the state the legislator served in had term limits.
Using the aforementioned data, I ran four logit models and four rare event logit models, with Speaker as the dependent variable. I ran four rare event logit models since the most of the observations for the Speaker variable were coded as 0. The results of the models are in the table below.
The results do not provide support for the median voter theorem (for the party) when it comes to selecting party leaders. However, the results do suggest that extremism plays a role in selecting party leaders under no term limits. According to the results, without term limits, the more extreme a legislator is, the more likely they are to be selected as Speaker. This may be in hopes that the speaker will introduce more partisan legislation. However, under term limits, partisanship plays no role in selecting the Speaker.
If partisanship does not seem to play a role in selecting party leaders under term limits, then other factors may matter. As mentioned earlier, term limited legislators care more about fundraising than non-term limited legislators, so when term limited legislators select party leaders, they may focus more on the ability of the leader to raise money. It would have been interesting to demonstrate such a relationship in this post, but I have to collect a lot of data to run the analysis. Instead, I am going to be pursuing this topic as a long term project, with the intention of possibly of turning the idea into another publication.
