The Consequences of Legislative Term Limits for Policy Diffusion

Abstract

Policy diffusion scholarship has long sought to understand when lawmakers will imitate innovations adopted by other jurisdictions and when they actually invest the time and resources necessary to learn about potential policies. We develop the theoretical rationale that term limits will reduce the incentive and capacity of state legislatures to gather information about policies available from previous adoptions in other states. We hypothesize that this will decrease the importance of previous adopters when term-limited legislators consider policy innovations. A multilevel analysis of the diffusion of eighty-seven policies between 1960 and 2009 provides support for this expectation. Our findings provide insight into the way in which institutional features shape policy diffusion.

Keywords

policy diffusion term limits state politics

Introduction

The ways in which policy makers use the policies in other jurisdictions to inform their own adoption choices has been a central feature of the diffusion literature for decades. That work has concluded that the use of policy information approximates a continuum, with simple imitation of policy choices in other states at one end and careful replication of only effective innovations at the other. At the heart of this literature are assumptions about the incentives that policy makers have to gather information about potential innovations. For example, aspirations to be more similar to wealthier or more prominent jurisdictions may encourage states to simply imitate the policies that they have adopted, without paying much attention to the consequences of those policies. Alternatively, jurisdictions concerned with avoiding policy failure, and the political costs associated with such failure, will search for an alternative that has proven successful elsewhere (Berry and Baybeck 2005; Shipan and Volden 2008).

In addition to developing relatively convincing ways to distinguish simple imitation from more sophisticated policy learning, the scholarship on diffusion has also attempted to understand when lawmakers will be concerned with policy failure and, thus, invest the time and resources necessary for policy learning. One of the answers they have offered is that less professionalized legislatures have less capacity for learning (Shipan and Volden 2006). Very recent work also demonstrates that individual policy makers have less incentive to learn about policies that they are ideologically predisposed against (Butler et al. 2017).

We engage this body of work by examining the impact of legislative term limits, a prominent institutional feature of state governments, on lawmakers’ incentives and capacity for policy learning. Currently, fifteen states have legislative term limits, which were implemented between 1996 and 2010. This institutional reform was a largely a populist reaction to concerns over “career politicians” and, as such, was explicitly designed to shorten the actual and electoral time horizon of lawmakers. In addition, research suggests that legislative term limits have reduced legislative professionalism and the capacity of that institution relative to the executive in states where they have been implemented (Carey et al. 2006; Kousser 2005).

Drawing on these and similar observations from the literature, we develop the theoretical rationale that term limits will reduce the incentive and capacity of state legislatures to gather information about policies available from previous adoptions in other states. We hypothesize that this will decrease the importance of previous adopters when term-limited legislators consider policy innovations. A multilevel analysis of the diffusion of eighty-seven policies between 1960 and 2009 provides support for this expectation. The positive association between the number of previous adopters and the likelihood of policy adoption disappears in term-limited states. This result is robust to different conceptualizations of term limits and model specifications, which are presented along with the primary findings. We also show that the relationship between the number of previous adopters and term limits did not exist before the introduction of term limits.

This research adds to our growing understanding of the way in which institutional features structure the incentives for policy adoption, shaping policy diffusion across jurisdictions. As term limits remove the possibility of a career in the legislature, the incentive to invest time and resources into policy choices in the current institution are diminished, altering the way in which legislators search for policy ideas. Moreover, this research provides insight into the consequences of term limits. Considerable scholarship has investigated the expected and unexpected implications of term limits, revealing shifts in legislators’ career goals, institutional comparisons, and policy output. This paper advances our understanding of the way in which the relatively short time horizons established by term limits influence policy adoption.

Learning and Imitation in Policy Diffusion

The idea that state policy makers take cues from the actions of other states when considering a policy innovation grows from the earliest work on the diffusion of innovations among individuals, which suggested that the spread of something new is a social process dependent on communication among users and potential users (see Rogers 1995 for a review). Walker (1969) focused the discussion on governments and policy innovations, and suggested that jurisdictional decisions are driven by both internal state characteristics and information from other states, the latter providing a heuristic cognitive shortcut for policy makers considering an innovation. Later research suggests that this internal–external diffusion model has defined and continues to dominate the study of policy diffusion (Berry and Berry 1990).

Walker (1969) emphasized that state policy makers were most likely to imitate the policy choices of “similar” states, and as a proxy for this similarity he used geographic contiguity. The argument that neighboring states are likely to share relevant characteristics is an intuitive one, and the empirical research has consistently confirmed that a jurisdiction is more likely to adopt a policy innovation if a higher proportion of its neighbors has done so (see, for example, Berry and Berry 1990, 1992; Gray 1973; Karch 2007a; Mintrom 1997; Mintrom and Vergari 1996; Volden 2002). Scholars have also begun to look for other “peers” that states may choose to emulate when considering policies. In this vein, scholars have demonstrated that policy makers are likely to mimic states that share their political ideology, particularly for policies that are ideologically charged (Grossback, Nicholson-Crotty, and Peterson 2004; Volden 2006). Research has also demonstrated that states follow the policy example of the federal government (Gray 1973; Karch 2007a), which often serves to disseminate or amplify previous state-level innovations.

Evidence of policy contagion across peers, however defined, has often been labeled “policy learning.” Recently, however, scholars have attempted to be more precise about the mechanisms of diffusion, and consider whether diffusion actually entails learning. As an example, Boehmke and Whitmer (2004) argue that social learning is often conflated with economic competition as a motivation for state behavior in the diffusion process; they demonstrate that the former can explain initial adoption decision and the latter is more likely responsible for subsequent changes to policy. Similarly, research has suggested that what it often termed “learning” is simply an emulation of behavior in other jurisdictions, rather than a conscious search for information about the effectiveness of a policy (see, for example, Weyland 2005). Finally, formal work has suggested that what looks like diffusion due to policy learning could be the result of intrastate experimentation with different innovations (Volden, Ting, and Carpenter 2008).

In a related argument, Volden (2006) argues that a desire to avoid policy failure gives both lawmakers and administrators incentives to replicate only successful policies. Authors empirically confirm that effective State Children’s Health Insurance Program innovations were more likely to diffuse. The key implication of this finding is that conclusions about policy learning are more robust when scholars find evidence that policy makers mimic rationally only those policies that work. Work on policy diffusion in the developing world, though rarely cited in studies of U.S. diffusion, similarly argues that, to conclude that learning is occurring, studies should find evidence that potential adopters examine both policy actions and outcomes in previously adopting jurisdictions, rather than simply the first (Meseguer 2005; Weyland 2005, 2007). Building on this insight, a number of recent studies have sought to refine the ways in which scholars can determine whether policy makers are actually gathering information about outcomes, and thus truly learning from the experience of other states, when considering an innovation (see, for example, Nicholson-Crotty and Carley 2015).

In addition to empirically distinguishing between imitation and learning, research has sought to understand the conditions and characteristics that predict whether lawmakers in a state will simply mimic a neighboring state or gather information about the effectiveness and political support for a policy adopted in other jurisdictions. The answers to this question have centered on the incentives of those lawmakers to spend time and resources gathering additional information about a policy and their capacity to do so.

In terms of incentives, scholars have focused on the type of policy, arguing that highly salient policies put pressure on legislators to adopt quickly and, perhaps, to forgo significant information about the innovations (Nicholson-Crotty 2009). Boushey (2010) further suggests that policy salience interacts with the political environment in the states and the presence of interest groups that facilitate the spread of innovations to explain the rapid uptake of certain policies. Turning to the state-specific incentives, Grossback, Nicholson-Crotty, and Peterson (2004) demonstrate that greater uncertainty over the “political fit” of a policy motivates lawmakers to move beyond geographic peers to gather additional information regarding the ideology of previous adopters. In a related argument, Gilardi (2010) argues that concern over electoral consequences is the primary motivation for expending the resources necessary to learn about a policy for right-leaning governments, while left-leaning governments are more motivated by concern over policy effectiveness. In very recent experimental work, Butler et al. (2017) find that city managers are less willing to invest time learning about policies that they are ideologically opposed to, but that this resistance can be partially overcome by evidence that co-partisans in other jurisdictions have adopted similar policies.

In addition to differing incentives, research suggests that the capacity to effectively gather information about a potential policy innovation influences adoption decisions. Here again scholars have examined policy characteristics, concluding that complex policies tax the capacity of more states and, therefore, diffuse more slowly (Nicholson-Crotty 2009; see also Boushey 2010). Others have focused on the characteristics of adopters, and a set of characteristics that fall under the umbrella of “legislative professionalism.” The features associated with more professional legislatures – higher – legislator salaries, larger staffs, and longer session lengths – increase members’ capacity to tackle complex issues and develop nuanced and detailed policy solutions (Bourdeaux and Chikoto 2008; Huber and Shipan 2002; Squire 1992). This logic translates easily to the type of search and processing skills necessary for policy learning. Indeed, Shipan and Volden (2006) demonstrate that legislators who have more staff resources to draw upon, and for whom policy making is more of a full-time job, are more able to distinguish effective and politically popular policies. Desmarais et al. (2015) do not find evidence that professionalism influences the prominence of states within diffusion networks, but they do find evidence that older proxies, such as population and wealth, influence the ways in which states send and receive policy information.

The Impact of Term Limits

The research reviewed above suggests that uncertainty about the potential electoral costs of an innovation, either because of a bad political “fit” or because of blame associated with failed policies, provides a powerful incentive for lawmakers to invest more resources gathering information from previous adopters. It is important to note that these results are premised on the assumption that lawmakers have to care about these electoral costs. Findings from existing work also intimate that lawmakers are more likely to engage in effective policy learning when they have the time and capacity to do so. In an effort to contribute to this literature, we make the argument that legislative term limits diminish the incentive and ability for learning among potential adopters because they mute the electoral consequences of bad policy and reduce the capacity to gather relevant information from previous adoptions.

First, given that they limit elected officials time horizons, term limits likely diminish the political risk associated with poor policy decisions, creating different incentives for term-limited legislators while in office. Research indicates that term limits reduce elected officials’ incentive to focus on the long-term consequences of policy decisions (see Berman 2004; Herron and Shotts 2006). For example, Besley and Case (1995) find support for their political reputation model, in which term limits weaken governors’ incentive to care about their political reputations, leading to different fiscal policy outcomes (also see Krause, Lewis, and Douglas 2013). Similarly, Cummins (2013) suggests that term-limited legislators might be less likely to make difficult financial decisions, finding that legislative term limits are associated with a decrease in state fiscal health.¹ Alt, Bueno de Mesquita, and Rose (2011) find that term-limited governors are associated with higher levels of taxes, spending, and borrowing costs and lower economic growth.

This research offers support for the idea that the political risk associated with poor policy choices is diminished under term limits, suggesting that policy learning from other states is less important under these circumstances. If legislators do not have to deal with the long-term political fallout from failed policy choices, then their incentive to hold off on adopting policies until after other states have experimented with them is reduced. Moreover, because term-limited legislators are more likely to have progressive ambitions than non–term-limited legislators (see Herrick and Thomas 2005; Steen 2006) and have shorter time windows to secure victories, term limits also create incentives to pass legislation quickly in order to have legislative accomplishments to advertise in their campaign for another office.² This further reduces the incentive to wait to learn about policy effectiveness from a number of previous adopters.

Second, term limits likely reduce legislators’ capacity and willingness to gather relevant policy information from states that have previously adopted policies. There is considerable scholarship suggesting that term limits diminish the capacity of state legislatures. With the removal of experienced legislators, the institution loses expertise about the policy process and policy options (see Berman 2007). Furthermore, under term limits, the incentive to invest the time and effort necessary to gain policy expertise may be diminished, given legislators’ limited tenures in office. This is one rationale behind the finding that term-limited legislatures are institutionally weaker than the executive branch compared with their non–term-limited counterparts (see Baker and Hedge 2013; Carey et al. 2006; Kousser 2005; Miller, Nicholson-Crotty, and Nicholson-Crotty 2011); the argument suggests that term-limited legislators lack the capacity and the motivation to counter the executive branch effectively. Other research focuses more specifically on the way in which the reduced legislative capacity associated with term limits affects policy. For example, Lewis (2012), noting legislator inexperience as a possible causal mechanism, suggests that term limits might have a deleterious effect on a state’s fiscal policy performance, finding that term-limited legislatures are associated with lower bond ratings. Also, reflective of the consequences of diminished policy expertise, term-limited legislatures are associated with less policy innovation (Kousser 2005) and with shorter and less complex legislation, at least in less professionalized legislatures (Kousser 2006).

This is not to suggest that term-limited legislators are less interested in issues or passing legislation. Research indicates that term-limited legislators are actually more issue-oriented than their counterparts (Herrick and Thomas 2005) and that term-limited and non–term-limited legislatures are indistinguishable in the percentage of bills they pass (Kousser 2005). In addition, as noted above, given their higher levels of progressive ambition (see Herrick and Thomas 2005; Steen 2006), term-limited legislators have an incentive to get some quick legislative wins that might help them in their upcoming elections. Coupled with reduced political risk and capacity, we think that the push to establish a legislative record quickly may motivate term-limited legislators to copy the policy innovations of other states without waiting to learn from additional adopters.

Taken together, this evidence suggests that policy adoption might work differently in term-limited and non–term-limited states. Because term limits likely reduce the capacity and motivation of legislators to gather relevant policy information and curtail the political risk associated with adopting potentially poor policies, the value placed on policy information available from previous adoptions is likely diminished. Thus, legislators in term-limited states might be more willing to adopt policies without waiting to learn from the experience of a number of previous adopters. Existing literature suggests that the number of states that have adopted a policy allows non-adopters to acquire information about a policy and that the number of previous adopters should, therefore, be positively related to the likelihood of a state adopting a policy in a given year (see Karch et al. 2016; Makse and Volden 2011; Shipan and Volden 2008). Given the potentially reduced importance of policy information in term-limited states, we expect that this source of policy learning will be less influential in term-limited states.³ Specifically, we expect the following:

Policy Learning Hypothesis: The number of previously adopting states will have less of an effect on the likelihood of adoption in term-limited states than non–term-limited states.

Data and Method

To test these expectations, we use data on eighty-seven policies that were first adopted in 1960 or after. The policy data come from a dataset compiled by Boehmke and Skinner (2012a,b).⁴ The unit of analysis is the policy state year and the data end in 2009. States are dropped from the dataset once they have adopted a particular policy. For each policy, the data begin in the year in which the policy was first adopted, which in this dataset ranges from 1960 to 2007.

These eighty-seven policies allow us to test our theoretical expectations in a variety of policy areas. According to Boehmke and Skinner’s coding, these policies cover thirteen different topic areas, most prominently corrections (twenty-five policies) and health (twenty-two policies). Of course, while these policies cover a wide range of topics, we cannot entirely know if the eighty-seven policies that we consider in our analysis are representative. Nonetheless, we think that the diversity of policies included is valuable for the study of our question and the generalizability of the findings. The eighty-seven policies represent all policies from the Boehmke and Skinner dataset that were first adopted in 1960 or after. It is after that date that a measure of state-level political ideology is available, which allows us to include a measure of ideological distance between potential and previous adopters, one of the key variables highlighted in previous research on diffusion.⁵ As a robustness check, we also estimate the models discussed below in the sample of all policies that were adopted in 1979 or after, which allows us to include legislative professionalism, another validated predictor.

Dependent Variable

The dependent variable for this analysis captures whether a state has adopted a particular policy. Specifically, it is a binary indicator of whether a state adopted a given policy in a given year. Following previous studies of policy diffusion, we drop the first state(s) to adopt a particular policy.

Key Independent Variables

One of the primary independent variables in this analysis is whether a state had implemented term limits or not. We focus on implemented term limits instead of adopted term limits because previous research suggests that implemented term limits have the greatest effects on legislative behavior (see Carey et al. 2006; Kousser 2005; Miller, Nicholson-Crotty, and Nicholson-Crotty 2011). Although the presence of legislative term limits is sometimes measured with a dichotomous indicator, scholars have argued convincingly that doing so ignores significant differences in the scope of limits and the restrictions that they place on legislative service (Erler 2007; Moncrief et al. 2004; Sarbaugh-Thompson 2010). To avoid this, we employ measures of term limits that focus on legislator turnover, which was the primary expected consequence of limits. Measures focused on turnover are also most appropriate for a test of our argument that a short time horizon and lack of experience/capacity influences the degree to which term-limited legislators value policy information available from previous adoptions. Specifically, we use different versions of the term-limitedness measure developed by Sarbaugh-Thompson (2010). The first is a ratio of mandated turnover after term limits to turnover in the decade before they went into effect, with higher values reflecting more restrictive limits. The second weights this measure for the proportion of legislators serving in each chamber and the specific limits imposed on each chamber. As a robustness check, we also show models using the traditional dichotomous measure of term limits, the Sarbaugh-Thompson (2010) term-limitedness measure that adjusts for the fact that some term limits restrict only consecutive years of service and allow legislators to cycle between the upper and lower chambers of the legislature, which some legislators may do,⁶ and the percentage of turnover produced by term limits (see Cummins 2013).⁷ The term limits turnover data are available from 1983 to 2008.

A second key independent variable is the number of states that have previously adopted a particular policy. This variable is a count of the states that have adopted the policy in the previous years. Previous research on diffusion has validated this variable as a proxy for policy learning. We think that the number of previous adopters more closely reflects learning than imitation; unlike imitation, learning does not necessarily require only looking to states that have valued characteristics. A state can potentially learn from the experiences of any other state. However, imitation implies following only states that possess characteristics that policy makers view as worthy of imitation. As Berry and Berry (2014, 311) note,

Imitation occurs because policymakers in [state] A perceive [state] B as worthy of emulation, prompting A to adopt any policy that B adopts independently of any evaluation of the character of the policy or its effectiveness (Simmons, Dobbin, and Garrett 2006; Meseguer 2006; Karch 2007[b]).

We think that this is less likely to be captured in a simple count of previous adopters. We log the number of previous adopters to deal with the high variation across the large number of policies. To test our expectation regarding the differing importance of policy information for legislators in term-limited and non–term-limited states, we interact the number of states that have previously adopted a particular policy with the term limits variables. We expect the coefficients for the interactions to be negative and significant, indicating the diminished effect of information from previous adopters on the likelihood that term-limited states adopt a policy. In addition, we expect that the coefficient for the term limits variable will be positive and significant; given the interaction, this coefficient represents the effect of term limits when the number of previous adopters (logged) equals zero, which, given the transformation, is when the number of previous adopters actually equals one. We expect that term limits will have a positive effect on the likelihood of adoption when only one previous state has adopted a policy.

Control Variables

In the models, we account for a number of additional factors identified in the literature. We include a measure of ideological difference between the states that have previously adopted a policy and states that could adopt a policy. To capture this, we use the citizen ideology measure developed by Berry et al. (1998; Berry et al. 2010).⁸ We use the absolute difference between the average of ideology of all previous adopters and each state in a given year for a given policy. We also include a measure of whether the state’s neighbors adopted a particular policy. For a given year, this is the proportion of a state’s neighbors, defined as states with which it shares a border, that have adopted the particular policy.

We also account for additional legislative and political factors, including legislative professionalization, size of the legislature (logged), divided government, partisan competition, party control of the government, the availability of the initiative, and citizen ideology (Berry et al. 1998; Berry et al. 2010). We use Squire’s measure of professionalization, which includes legislator salary, days in session, and staff (Squire 1992, 2012, 2017).⁹ Legislature size is coded from the Book of the States. Divided government, partisan competition, and party control of the government (Republican control = 1, otherwise = 0) each come from Klarner’s (2013c) datasets. For party competition, we include the four-year version of a measure of party competition created by Ranney, as updated by Klarner (2013a). In the model considering term limits turnover, we also include a measure of non–term limits turnover (Cummins 2013).

In addition, we account for other state factors in the models, including per capita income, revenue growth, population size (logged), population density, and education levels.¹⁰ Per capita income, the general revenue measure used to calculate the revenue growth variable, and population size come from Klarner’s (2013b) dataset. We include the measure of population density to help capture the level of urbanization in the state. The U.S. Census Bureau’s definition of urbanization has changed over time, making it so that the urbanization measure is not comparable over our whole time period. The measure of population density is by decade and comes from the U.S. Census Bureau.¹¹ The measure of education level is the percentage of a state’s population twenty-five years or older with a bachelor’s degree or higher by decade; this measure comes from the U.S. Census Bureau.¹²

Modeling Approach

We have two non-nested levels in our dataset—states and policies. Given this, we estimate multilevel logistic regression models, with varying intercepts by state and policy. To account for duration dependence, we include three time splines in the models. Although most of our variables are available for our entire time period, legislative professionalization is only available starting in 1979. Thus, while many of our models consider all policies that were first adopted in 1960 or after (eighty-seven policies), excluding the measure of professionalization, we also look at models that include the measure of professionalization, considering all policies that were first adopted in 1979 or after (fifty-five policies). In addition, as noted above, the term limits turnover data range from 1983 to 2008.

Because we include policies starting in 1960, prior to the implementation of term limits in any state, we have the ability to consider whether the states that eventually instituted legislative term limits behave differently from those that did not in the period prior to the implementation of term limits in any state. In this test that is somewhat akin to a placebo test, the interaction effect that we expect after the implementation of term limits should not be significant. This will provide additional support for our assertion that it was the introduction of term limits, rather than some other unobserved characteristic, that reduced the importance of policy information.

Specifically, we model whether the effect of the number of previous state adopters is lower for the group of states that eventually adopted term limits than those that did not prior to the implementation of term limits. The data in this model range from 1960 to 1989. The first states adopted term limits in 1990; thus, we end the data prior to the introduction of term limits in any state to avoid any effects associated with term limits. We expect that the interaction in this model will not be significant, indicating that, prior to the introduction of term limits, the effect of the number of previous state adopters on policy adoption is not different for states that eventually instituted term limits and those that did not.

Results

The results of our primary analyses are presented in Table 1; these models explore the way in which term limits moderate the influence of previous adopters on policy adoption for policies that were first adopted in 1960 or after. The analysis in model 1 uses the weighted measure of term-limitedness, while model 2 presents the results of a model using the unweighted measure of term-limitedness.

Table 1.

Policy Diffusion and Term Limits.

	Model 1	Model 2
	Term-limitedness (weighted)	Term-limitedness (not weighted)
Ideological difference	−0.0126***	−0.0126***
Ideological difference	(0.00265)	(0.00266)
Neighbors adopt	1.657***	1.658***
Neighbors adopt	(0.0987)	(0.0970)
Number of previous state adopters (ln)	0.177***	0.178***
Number of previous state adopters (ln)	(0.0351)	(0.0350)
Term-limitedness (weighted)	0.728***
Term-limitedness (weighted)	(0.227)
Term-limitedness (weighted) × No. Previous Adopters (ln)	−0.255***
Term-limitedness (weighted) × No. Previous Adopters (ln)	(0.0820)
Term-limitedness (not weighted)		0.824***
Term-limitedness (not weighted)		(0.242)
Term-limitedness (not weighted) × No. of Previous Adopters (ln)		−0.290***
		(0.0869)
Citizen ideology	0.00244	0.00244
Citizen ideology	(0.00278)	(0.00268)
Citizen initiative	0.0320	0.0317
Citizen initiative	(0.0922)	(0.0927)
Population density	0.0279	0.0281
Population density	(0.0253)	(0.0240)
Percentage college degree or higher	0.0247	0.0247*
Percentage college degree or higher	(0.0152)	(0.0131)
Party competition	0.405	0.403
Party competition	(0.506)	(0.359)
Divided government	−0.0762	−0.0753
Divided government	(0.0914)	(0.0747)
Legislature size (ln)	−0.324**	−0.324**
Legislature size (ln)	(0.131)	(0.135)
Republican control of government	−0.0697	−0.0697
Republican control of government	(0.0928)	(0.0910)
Per capita income	−0.000323	−0.000465
Per capita income	(0.0234)	(0.0163)
Revenue growth	0.00280	0.00280
Revenue growth	(0.00224)	(0.00224)
Population (ln)	0.169***	0.168***
Population (ln)	(0.0498)	(0.0494)
Constant	−2.669***	−2.661***
Constant	(0.898)	(0.905)
Observations	63,377	63,377
Log likelihood	−8,228.44	−8,227.89
Wald χ²	812.93 (19)***	835.92 (19)***
AIC	16,500.89	16,499.77
BIC	16,700.14	16,699.02

Dependent variable: State adopted policy or not (1, 0). Multilevel models with varying intercepts for state and policy. Three time splines included in the models, but not shown. AIC = Akaike information criterion; BIC = Bayesian information criterion.

p < .10. **p < .05. ***p < .01 (two-tailed test).

Before discussing the key variables of interest, it is important to note that predictors of adoption from previous scholarship perform largely as expected in both models. The ideological distance between previous and potential adopters is negatively associated with the probability of adoption. Alternatively, the proportion of neighboring states that adopted a policy is positively associated with that probability. Larger states, which previous research suggests have higher capacity, show a higher likelihood of adoption, while that probability is lower in states with larger legislatures where collective action problems are more pronounced.

Turning to the variables of interest, we see that, in both models, the number of states that have adopted a policy is positively and significantly associated with the likelihood that states that have not already done so will. It is important to remember, however, that in the presence of the interaction, this represents the impact of previous adopters in states that do not have term limits (or the term-limitedness score is 0). In addition, the coefficient for term-limitedness is positive and significant in both models, reflecting the relationship between term-limitedness and the probability of policy adoption when the number of previous adopters (logged) equals 0, which, given the transformation, is when the actual number of previous adopters is equal to 1. This is as expected, suggesting that term limits are positively related to adoption at the time when the opportunity for policy learning from other states is relatively limited. Also, as expected, both interaction terms are negative and significant, suggesting that the impact of previous adopters on the probability of adoption goes down significantly as effective turnover due to term limits increases.

To more fully consider the conditional relationship, we graph the predicted probability of policy adoption at different numbers of previous adopters (minimum [1] and maximum [49]) across the range of the measures of term-limitedness (Figure 1), with term-limitedness (not weighted) on the left and term-limitedness (weighted) on the right; the predicted probabilities are based on the estimates from models 1 and 2. As illustrated, the level of term-limitedness in a state conditions the association between the number of previous adopters and policy adoption. In both cases, the probability of adoption begins to increase as term-limitedness rises when the previous adopter number is at its lowest, while declining when the previous adopter number is at its highest. When term-limitedness is equal to 0, a high level of previous adopters has a positive effect; alternatively, when term-limitedness is greater than 0, the positive effect of previous adopters disappears (and actually reverses). At the highest level of term-limitedness (weighted), the probability of adopting a policy is 0.18 when only one state has previously adopted it, while the probability is 0.07 when 49 states have previously adopted a policy.¹³ The probabilities are reversed when term-limitedness equals 0; for this group, the predicted probability of adoption is 0.06 when previous adopters is held at one and 0.10 when previous adopters is held at 49.¹⁴ The probabilities for the unweighted measure follow the same pattern. In addition, the results suggest that the probability of adopting a policy after only one other state has done so is over 0.1 higher in states with the most restrictive limits compared with states where legislators are not term-limited (or the term-limitedness score is 0).

Figure 1.

Moderating impact of term limits on the influence of previous adopters (90% confidence intervals).

Alternate Model Specifications

These results are consistent with our expectations, but it is important to ensure, to the degree possible, that they are robust to other specifications and not a result of some unobserved difference between states that adopted term limits and those that did not. To that end, this section offers a series of sensitivity tests and a model somewhat akin to a placebo test in which we compare the states that eventually adopted term limits to those that did not prior to the introduction of term limits.

Table 2 presents five models using different measures of one of the primary independent variables, term limits, or different model specifications and samples. The first column substitutes a more traditional dichotomous the measure of term limits. The second uses a turnover-based measure of term-limitedness developed by Sarbaugh-Thompson (2010) that adjusts for the degree to which legislators are able to avoid limits by cycling back and forth between chambers, while the third column shows the model with a measure of actual term limit turnover. Columns 4 and 5 show analyses using the same measures of term-limitedness as the primary models but limit the sample of policies to those that were first adopted in 1979 or after. This allows for the inclusion of legislative professionalization, a well-validated predictor of policy learning, as a control. In the interest of concision, Table 2 includes only the coefficients for the key variables; the full models with the control variables are presented in Table 1 of the online appendix.

Table 2.

Alternate Models.

	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	Dichotomous term limits	Term-limitedness (recycling)	Term limits turnover	Term-limitedness (weight) w/professionalization	Term-limitedness (no weight) w/professionalization	Prior to term limits (1960–1989)
No. of previous adopters (ln)	0.180***	0.176***	0.256***	0.291***	0.292***	0.0323
No. of previous adopters (ln)	(0.0352)	(0.0354)	(0.0522)	(0.0478)	(0.0477)	(0.0700)
Term limits	0.984***
Term limits	(0.256)
TL × No. of Previous Adopters (ln)	–0.318 ***
TL × No. of Previous Adopters (ln)	(0.0919)
Term-limitedness (recycling)		0.677***
Term-limitedness (recycling)		(0.231)
Term-limitedness (recycling) × No. of Previous Adopters (ln)		–0.253 ***
		(0.0832)
TL turnover			0.0229**
TL turnover			(0.0102)
TL turnover × No. of Previous Adopters (ln)			–0.00956 **
TL turnover × No. of Previous Adopters (ln)			(0.00383)
Non-TL turnover			−0.00446
Non-TL turnover			(0.00443)
Term-limitedness (weighted)				0.651***
Term-limitedness (weighted)				(0.241)
Term-limitedness (weighted) × No. of Previous Adopters (ln)				–0.273 ***
Term-limitedness (weighted) × No. of Previous Adopters (ln)				(0.0876)
Professionalization				−0.922*	−0.929*
Professionalization				(0.495)	(0.534)
Term-limitedness (not weighted)					0.742***
Term-limitedness (not weighted)					(0.254)
Term-limitedness (not weighted) × No. of Previous Adopters (ln)					–0.310 ***
					(0.0936)
Term Limits—prior						−0.143
Term Limits—prior						(0.194)
TL—prior × No. of Previous Adopters (ln)						0.0473
TL—prior × No. of Previous Adopters (ln)						(0.0712)
Constant	−2.632***	−2.692***	−6.423***	−4.876***	−4.871***	−3.219***
Constant	(0.916)	(0.872)	(1.252)	(1.065)	(1.040)	(1.147)
Observations	63,377	63,377	21,508	28,625	28,625	23,301
Log likelihood	−8,226.56	−8,229.19	−4,007.23	−4,773.14	−4,772.60	−2,757.70
Wald χ²	823.41(19)***	809.12(19)***	399.72(20)***	481.80(20)***	475.51(20)***	216.10(19)***
AIC	16,497.13	16,502.37	8,060.47	9,592.29	9,591.19	5,559.39
BIC	16,696.38	16,701.62	8,243.92	9,782.32	9,781.22	5,736.63

Dependent variable: State adopted policy or not (1, 0). Multilevel models with varying intercepts for state and policy. The results for the key interactions are bolded. This table includes only the coefficients for the key variables; the full models with the control variables are presented in Table 1 of the online appendix. The Cummins (2013) data used for the term limit turnover model (model 3) range from 1983 to 2008. The other models start with policies first adopted in 1960 (models 1 and 2) or 1979 (models 4 and 5) and end in 2009, except for model 6 (1960–1989). AIC = Akaike information criterion; BIC = Bayesian information criterion.

p < .10. **p < .05. ***p < .01 (two-tailed test).

The most important thing to take away from these alternative specifications and samples is that our results remain unchanged across them. Regardless of the measure of term limits used or the inclusion of legislative professionalization, previous adoptions has a significantly larger impact in states that do not have term limits or those that have less severe term limits. This provides us with considerable confidence that our primary result is not the result of operational choices or omitted variables problems.

As a final test, the last column of Table 2 presents something akin to a placebo test designed to determine if some unobserved difference between states that eventually adopt or fail to adopt term limits is actually driving the results. To construct the test, we restrict the period of study to 1960–1989; the first states adopted term limits in 1990. In this model, the states that eventually adopted and implemented term limits are coded as 1 on an indicator of (eventual) term limits, while those that never did are coded as a 0.¹⁵ We then interact that indicator with the measure of previous adopters. If the interaction term were negative and significant, it would indicate that some characteristic shared by states that eventually adopted and implemented term limits influenced their receptivity to information from previous adopters, rather than the term limits themselves, as we have suggested. As the results in Table 2 suggest, however, this is not the case. The interaction is not statistically significant and is substantively quite close to 0, suggesting that it is not unmeasured state characteristics associated with the eventual adoption and implementation of term limits that are actually driving our primary results.

Conclusions and Implications

One important line of research in the diffusion literature assesses the conditions that lead legislators to invest the time and resources required for policy learning and the passage of effective policy (see Butler et al. 2017; Shipan and Volden 2006). Based on analyses of eighty-seven policies that diffused over almost fifty years, we conclude that legislative term limits are one of the factors that diminish the value placed on policy learning.

Taken together, our results support the idea that term limits reduce the incentive for legislators to invest in policy learning. This consequence follows from the relatively short time horizons of term-limited legislators. The political risk associated with poor policy choices is diminished under term limits, because legislators are unlikely to have to answer directly for policy failures. Moreover, career-minded legislators in term-limited states need fast legislative wins, if they are to get credit for new policies, which provides an incentive to adopt policies without waiting to learn from the experience of previous adopters. These results are robust to numerous changes in specification and in different samples of policies and diffusion periods.

The results of this paper speak to the way in which institutional features can structure lawmakers’ incentives to invest in their policy-making roles. Term limits (effectively) remove the possibility of a career in a particular legislature. In doing so, they reduce the incentive to focus on the long-term implications of policy choices and to build policy knowledge in a jurisdiction, while encouraging quick policy output. Thus, the number of previous adopters and the policy experience that such information contains is less meaningful for policy adoption in term-limited states. By shaping legislators’ policy incentives, term limits have altered the policy diffusion landscape, turning term-limited states into early adopters despite the fact many have characteristics that would traditionally lead us to expect them to be laggards in the diffusion process.

The speed with which term-limited legislators adopt policy may also have significant implications for more general diffusion patterns. More explicitly, it may have an impact on the speed with which policies diffuse and the shape of the well-documented S-shaped diffusion curve. In the Boehmke and Skinner data that we use for this analysis, the average number of adopters in the years 1966 to 1986 for policies that diffused during that period was 1.26, while it was 1.93 for policies that diffused between 1986 and 2006. This empirical observation is consistent with assertions by some scholars that the number of policies that diffuse rapidly, rather than in the traditional S-shaped pattern, has increased over time. Our results suggest that the adoption of term limits in multiple states may have contributed to this change. States that would have historically adopted more slowly are now moving more quickly; for other states that still rely on information from previous adopters, this change may also increase the speed with which they adopt, relative to historical patterns.

Finally, while considerable research has highlighted the implications of term limits for policy outputs, the results of this paper raise further questions about the long-term effectiveness of policies adopted in term-limited states. Legislators in term-limited states could be in a perpetual state of adopting new policy solutions to “fix” the problems generated by their predecessors. From a political standpoint, this might be acceptable, perhaps even advantageous. However, from a governing standpoint, this might have negative ramifications. This is, however, an empirical question that could be considered in future research.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

Supplemental Material

Replication data for this article are available on Susan Miller’s website: .

Supplemental materials for this article are available with the manuscript on the Political Research Quarterly (PRQ) website.

References

Alt

James

Bueno de Mesquita

Ethan

Rose

Shanna

. 2011. “Disentangling Accountability and Competence in Elections: Evidence from U.S. Term Limits.” The Journal of Politics 73 (1): 171–86.

Baker

Travis J.

Hedge

David M.

2013. “Term Limits and Legislative-Executive Conflict in the American States.” Legislative Studies Quarterly 38 (2): 237–58.

Berman

David R.

2004. Effects of Legislative Term Limits in Arizona. Joint project on term limits. Denver, CO: National Conference of State Legislatures.

Berman

David R.

2007. “Legislative Climate.” In Institutional Change in American Politics: The Case of Term Limits, edited by Kurtz

Karl T.

Cain

Bruce

Niemi

Richard G.

, 107–18. Ann Arbor: University of Michigan Press.

Berry

Frances Stokes

Berry

William D.

2014. “Innovation and Diffusion Models in Policy Research.” In Theories of the Policy Process, p. 307–360. edited by Sabatier

Paul A.

Weible

Christopher M.

, 3rd ed. Boulder: Westview Press.

Berry

William D.

Baybeck

Brady

. 2005. “Using Geographic Information Systems to Study Interstate Competition.” American Political Science Review 99 (4): 505–19.

Berry

William D.

Berry

Frances Stokes

. 1990. “State Lottery Adoptions as Policy Innovations: An Event History Analysis.” American Political Science Review 84:395–415.

Berry

William D.

Berry

Frances Stokes

. 1992. “Tax Innovation in the States: Capitalizing on Political Opportunity.” American Journal of Political Science 36:715–42.

Berry

William D.

Ringquist

Evan J.

Fording

Richard C.

Hanson

Russell

. 1998. “Measuring citizen government ideology.” American Journal of Political Science 42: 327–348.

10.

Berry

William D.

Fording

Richard C.

Ringquist

Evan J.

Hanson

Russell L.

Klarner

Carl E.

2010. “Measuring Citizen and Government Ideology in the U.S. States: A Re-appraisal.” State Politics & Policy Quarterly 10 (2): 117–35.

11.

Besley

Timothy

Case

Anne

. 1995. “Does Electoral Accountability Affect Economic Policy Choices? Evidence from Gubernatorial Term Limits.” The Quarterly Journal of Economics 110:769–98.

12.

Boehmke

Frederick J.

Skinner

Paul

. 2012a. “State Policy Innovativeness Revisited.” State Politics & Policy Quarterly 12 (3): 303–29

13.

Boehmke

Frederick J.

Skinner

Paul

. 2012b. “State Policy Innovativeness Scores V1.1.” hdl:1902.1/18507, Harvard Dataverse, V4.

14.

Boehmke

Frederick J.

Witmer

Richard

. 2004. “Disentangling Diffusion: The Effects of Social Learning and Economic Competition on State Policy Innovation and Expansion.” Political Research Quarterly 57 (1): 39–51.

15.

Bourdeaux

Carolyn

Chikoto

Grace

. 2008. “Legislative Influences on Performance Management Reform.” Public Administration Review 68: 253–65.

16.

Boushey

Graeme

. 2010. Policy Diffusion Dynamics in America. Cambridge: Cambridge University Press.

17.

Butler

Daniel M.

Volden

Craig

Dynes

Adam M.

Shor

Boris

. 2017. “Ideology, Learning, and Policy Diffusion: Experimental Evidence.” American Journal of Political Science 61: 37–49. doi:10.1111/ajps.12213.

18.

Carey

John M.

Niemi

Richard G.

Powell

Lynda W.

Moncrief

Gary F.

2006. “The Effects of Term Limits on State Legislatures: A New Survey of the 50 States.” Legislative Studies Quarterly 31 (1): 105–34.

19.

Cummins

Jeff

. 2013. “The Effects of Legislative Term Limits on State Fiscal Conditions.” American Politics Research 41 (3): 417–42.

20.

Desmarais

Bruce A.

Harden

Jeffrey J.

Boehmke

Frederick J.

2015. “Persistent Policy Pathways: Inferring Diffusion Networks in the American States.” American Political Science Review 109(02): 392–406.

21.

Enns

Peter K.

Koch

Julianna

. 2013. “Public Opinion in the U.S. States: 1956 to 2010.” State Politics & Policy Quarterly 13 (3): 349–72.

22.

Erler

Abbie H.

2007. “Legislative Term Limits and State Spending.” Public Choice 133:479–94.

23.

Gilardi

Fabrizio

. 2010. “Who Learns from What in Policy Diffusion Processes?”. American Journal of Political Science 54: 650–666.

24.

Gray

Virginia

. 1973. “Innovation in the States: A Diffusion Study.” American Political Science Review 67:1174–85.

25.

Grossback

Lawrence

Nicholson-Crotty

Sean

Peterson

David A. M.

2004. “Ideology and Learning in Policy Diffusion.” American Politics Research 32 (5): 521–45.

26.

Herrick

Rebekah

Thomas

Sue

. 2005. “Do Term Limits Make a Difference? Ambition and Motivations among U.S. State Legislators.” American Politics Research 33 (5): 726–47.

27.

Herron

Michael C.

Shotts

Kenneth W.

2006. “Term Limits and Pork.” Legislative Studies Quarterly 31 (3): 383–403.

28.

Huber

John D.

Shipan

Charles R.

2002. Deliberate Discretion? The Institutional Foundations of Bureaucratic Autonomy. New York: Cambridge University Press.

29.

Karch

Andrew

. 2007a. Democratic Laboratories: Policy Diffusion among the American States. Ann Arbor: University of Michigan Press.

30.

Karch

Andrew

. 2007b. “Emerging Issues and Future Directions in State Policy Diffusion Research.” State Politics & Policy Quarterly 7:54–80.

31.

Karch

Andrew

Nicholson-Crotty

Sean C.

Woods

Neal D.

Bowman

Ann O’M.

2016. “Policy Diffusion and the Pro-innovation Bias.” Political Research Quarterly 69 (1): 83–95.

32.

Keele

Luke

Malhotra

Neil

McCubbins

Colin H.

2013. “Do Term Limits Restrain State Fiscal Policy? Approaches for Causal Inference in Assessing the Effects of Legislative Institutions.” Legislative Studies Quarterly 38 (3): 291–326.

33.

Klarner

Carl

. 2013a. “Other Scholars’ Competitiveness Measures.” hdl:1902.1/22519, Harvard Dataverse, V1.

34.

Klarner

Carl

. 2013b. “State Economic Data.” hdl:1902.1/20404, Harvard Dataverse, V1.

35.

Klarner

Carl

. 2013c. “State Partisan Balance Data, 1937-2011.” hdl:1902.1/20403, Harvard Dataverse, V1.

36.

Kousser

Thad

. 2005. Term Limits and the Dismantling of State Legislative Professionalism. New York: Cambridge University Press.

37.

Kousser

Thad

. 2006. “The Limited Impact of Term Limits: Contingent Effects on the Complexity and Breadth of Laws.” State Politics & Policy Quarterly 6 (4): 410–29.

38.

Krause

George A.

Lewis

David E.

Douglas

James W.

2013. “Politics Can Limit Policy Opportunism in Fiscal Institutions: Evidence from Official General Fund Revenue Forecasts in the American States.” Journal of Policy Analysis and Management 32:271–95.

39.

Lewis

Daniel C.

2012. “Legislative Term Limits and Fiscal Policy Performance.” Legislative Studies Quarterly 37 (3): 305–28.

40.

Makse

Todd

Volden

Craig

. 2011. “The Role of Policy Attributes in the Diffusion of Innovations.” The Journal of Politics 73 (1): 108–24.

41.

Meseguer

Covadonga

. 2005. “Policy Learning, Policy Diffusion, and the Making of a New Order.” The ANNALS of the American Academy of Political and Social Science 598 (1): 67–82.

42.

Meseguer

Covadonga

. 2006. “Learning and Economic Policy Choices.” European Journal of Political Economy 22: 156–78.

43.

Miller

Susan M.

Nicholson-Crotty

Jill

Nicholson-Crotty

Sean

. 2011. “Reexamining the Institutional Effects of Term Limits in U.S. State Legislatures.” Legislative Studies Quarterly 36: 71–97.

44.

Mintrom

Michael

. 1997. “Policy Entrepreneurs and the Diffusion of Innovation.” American Journal of Political Science 41 (3): 738–70.

45.

Mintrom

Michael

Vergari

Sandra

. 1996. “Advocacy coalitions, policy entrepreneurs, and policy change.” Policy Studies Journal, 24: 420–434.

46.

Moncrief

Gary F.

Niemi

Richard G.

Powell

Lynda W.

2004. “Time, Term Limits, and Turnover: Trends in Membership Stability in U.S. State Legislatures.” Legislative Studies Quarterly 29(3): 357–381.

47.

Nicholson-Crotty

Sean

Carley

Sanya

. 2015. “Effectiveness, Implementation, and Policy Diffusion Or “Can We Make That Work for Us?”. State Politics & Policy Quarterly 16: 78–97.

48.

Nicholson-Crotty

Sean

. 2009 “The Politics of Diffusion: Public Policy in the American States.” Journal of Politics 71: 192–205.

49.

Rogers

Everett. M.

1995. The Diffusion of Innovations. 4th ed. New York: Free Press.

50.

Sarbaugh-Thompson

Marjorie

. 2010. “Measuring ‘Term Limitedness’ in U.S. Multi-state Research.” State Politics & Policy Quarterly 10 (2): 199–217.

51.

Shipan

Charles R.

Volden

Craig

. 2006. “Bottom-Up Federalism: The Diffusion of Antismoking Policies from U.S. Cities to States.” American Journal of Political Science 50(4): 825–43.

52.

Shipan

Charles R.

Volden

Craig

. 2008. “The Mechanisms of Policy Diffusion.” American Journal of Political Science 52 (4): 840–57.

53.

Simmons

Beth A.

Dobbin

Frank

Garrett

Geoffrey

. 2006. “Introduction: The International Diffusion of Liberalism.” International Organization 60:781–810.

54.

Squire

Peverill

. 1992. “Legislative Professionalization and Membership Diversity in State Legislatures.” Legislative Studies Quarterly 17 (1): 69–79.

55.

Squire

Peverill

. 2012. The Evolution of American Legislatures: Colonies, Territories and States, 1619-2009. Ann Arbor: University of Michigan Press.

56.

Squire

Peverill

. 2017. “A Squire Index Update.” State Politics & Policy Quarterly 17:361–71.

57.

Steen

Jennifer A.

2006. “The Impact of State Legislative Term Limits on the Supply of Congressional Candidates.” State Politics & Policy Quarterly 6 (4): 430–47.

58.

Volden

Craig

. 2002. “The Politics of Competitive Federalism: A Race to the Bottom in Welfare Benefits?” American Journal of Political Science 46:352–63.

59.

Volden

Craig

. 2006. “States as Policy Laboratories: Emulating Success in the Children’s Health Insurance Program.” American Journal of Political Science 50 (2): 294–312.

60.

Volden

Craig

Ting

Michael M.

Carpenter

Daniel P.

2008. “A Formal Model of Learning and Policy Diffusion.” American Political Science Review 102 (3): 319–32.

61.

Walker

Jack L.

1969. “The Diffusion of Innovations among the American States.” American Political Science Review 63 (3): 880–99.

62.

Weyland

Kurt

. 2005. “Theories of Policy Diffusion: Lessons from Latin American Pension Reform.” World Politics 57 (2): 262–95.

63.

Weyland

Kurt

. 2007. Bounded Rationality and Policy Diffusion: Social Sector Reform in Latin America. Princeton: Princeton University Press.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.13 MB