Research Updates You will participate in a series of “research updates” throughout the semester. We will be periodically reading the most up-to-date research within state politics, and you must review the research question, methods, and findings in your own words, along with providing original insights of your own. You will follow the CREATE method to help you do this. 1 Consider Read Elucidate the hypothesis Analyze and interpret the data Think of the next Experiment Your paper must begin with your evaluation of the paper, where you include a statement that captures your original thoughts about it. “This is a good paper” is not acceptable. “This paper by Hanson (2020) is a positive development for our understanding of state legislative committee systems because of its original and thorough data collection effort” is acceptable. Throughout the rest of the paper, you should back up your statement by including specific evidence from the article, in addition to providing an overview of the paper and its contribution to the literature. Be sure to include all components of the CREATE model within your paper. All papers must be three full pages with 12 point, times new roman font and one inch margins (double-spaced). You must attach a memo showing your CREATE notes. This does not count towards your overall page count. You will complete three research updates worth 15% of your final grade. Notes on mechanics: Do not include block quotes. I know they fill up the pages, but ultimately they’re a lazy way of summarizing a journal article. Do not tell me about Bayesian Information Criterion, pooled event history analysis, negative binomial regressions, or any other methods that I know you don’t use or know about. I want you to tell me, in accessible terms, how the results speak to the authors’ expectations. Include page numbers at the top of each page. 829408 research-article2019 SPAXXX10.1177/1532440019829408State Politics & Policy QuarterlyStrickland Article America’s Crowded Statehouses: Measuring and Explaining Lobbying in the U.S. States State Politics & Policy Quarterly 2019, Vol. 19(3) 351­–374 © The Author(s) 2019 Article reuse guidelines: sagepub.com/journals-permissions https://doi.org/10.1177/1532440019829408 DOI: 10.1177/1532440019829408 journals.sagepub.com/home/spa James Strickland1 Abstract Across the United States over time, numbers of registered interest groups have continued to increase, but these populations mask the total amount of lobbying that is occurring within America’s statehouses. Among registered interests, average numbers of hired lobbyists have increased markedly since the late 1980s. This study both quantifies this increase and identifies a set of causal variables. Previous studies have proposed a variety of short-term, political and long-term, institutional factors that govern rates of lobbying. Using a new data set spanning multiple decades, I find that changes in lobbying can largely be ascribed to institutional variables, including the implementation of term limits and regulations on lobbying. Lobby regulations, oneparty dominance, and legislative expenditures also appear to play a role in determining rates of multiclient lobbying. Direct democracy and state spending do not affect the hiring of lobbyists by registered interest groups. Keywords campaign finance, political behavior, interest groups, parties and interest groups, lobbying, legislative professionalism, legislative politics, term limits Introduction Recent reforms to the federal tax code have renewed interest in the influence of lobbyists over national policy. According to reporters, the quick legislative pace of the proposal created a “frenzy” among thousands of registered lobbyists “desperate to preserve prized tax breaks” (Tankersley, Kaplan, and Vogel 2017). According to one 1University of Michigan, Ann Arbor, MI, USA Corresponding Author: James Manning Strickland, University of Michigan, 5700 Haven Hall, 505 S. State Street, Ann Arbor, MI 48109-1045, USA. Email: strickl@umich.edu 352 State Politics & Policy Quarterly 19(3) analysis, more than 6,000 lobbyists had reported that they lobbied on tax-related issues during 2017, which was more than half of all lobbyists registered in the nation’s capital that year (see Lincoln 2017). While lobbying activity on tax issues might have spiked, this sort of frenzy is not unprecedented. There have been similar bursts of activity on prior tax bills at both the federal (Birnbaum and Murray 1987) and state levels (see Brasher, Lowery, and Gray 1999). Such temporary bursts of lobbying activity present several challenges for scholars of interest representation. Scholars have yet to propose a method for measuring intensity of lobbying effort that is applicable to many different political systems, or develop a means for predicting when and where organized interests lobby more intensely. I seek to achieve both goals. I first argue that lobbying intensity, at an aggregated level, can be measured by the average number of lobbyists hired by each interest group. I then measure changes in lobbying intensity in all U.S. states over multiple decades. No other study has examined numbers of lobbyists to such an extent. I review what prior scholars have had to say about lobbying intensity, and I categorize these hypotheses into two competing camps: explanations oriented toward political factors that are focused more on bursts of lobbying activity, and institutional factors that help to explain long-term trends in lobbying. I then subject these different explanations to statistical analysis and find that institutional variables such as legislature expenditures, the implementation of term limits, and regulations on lobbying all affect how many lobbyists interest groups register. In additional tests, I find that two of these factors and one-party dominance also appear to change how often interest groups rely on multiclient lobbyists. My findings have implications for those interested in representation, policy change, and transparency. Especially since the publication of Schattschneider’s (1960) incisive critique of interest pluralism, scholars have been concerned about the influence of money on American politics and policymaking. Interest groups that are able and willing to spend lots of money can hire teams of lobbyists to target legislators in tandem, possibly affecting policy outcomes. At the same time, however, traditional membership groups that seek public goods can also call on members to storm legislative offices and make their desires known. While groups with more resources do not always get their way, and linking money spent on lobbying to policy outcomes has proven elusive (see Baumgartner et al. 2009, 190–214), lobbying is always an ongoing effort. If registered interest groups consistently maintain many lobbyists to maintain existing policies in some political systems, then this suggests that policy change might be more difficult to come by in those systems than in others (Olson 1982). Moreover, particular regulations of lobbying may not necessarily lead to more compliance on the part of lobbyists (see H. F. Thomas and LaPira 2017). The data set assembled for this project allows us to test some of these hypotheses. Measuring Lobbying While many scholars have commented on or examined the influence of interest groups on policy outcomes, there has been a shortage of research on group agents or lobbyists (see Lowery and Marchetti 2012). Scholars have tended to assume that lobbyists Strickland 353 successfully convey the interests of their clients without loss of specificity or “energy” (p. 140). This assumption is problematic since lobbyists may not serve the interests of their clients faithfully or exclusively. It is difficult for interest groups either to observe the efforts of their agents or judge their advocacy success (see Schiff et al. 2015, 226–29), and shirking may especially be problematic for the clients of small lobby firms (Whitesell, Schiff, and Lowery 2018). Prior studies have conflated numbers of registered interests with the overall level of interest-group activity. This is also problematic since groups may choose to dedicate more resources to their mobilization effort by hiring multiple lobbyists (Rosenthal 1993, 57). After all, it is lobbyists who fulfill the crucial function of linking their clients with policymakers. Prior studies of interest mobilization have conflated numbers and types of active interest groups with mobilizational intensity. For example, there being less “certainty” over how policy might change can lead to more groups registering to lobby (Lowery and Gray 1995, 12). Indeed, in states dominated by a single political party, there are fewer registered interests (see Strickland 2018a). Boehmke (2008) has also found that in U.S. states with direct democracy techniques, more citizens’ interests register. Residents of these states are also more likely to be members of such groups (Boehmke and Bowen 2010). Certainly, while numbers of interests were shown to fluctuate in response to these factors, these studies missed the second decision facing a group once it chooses to lobby: how many lobbyists to hire. Brasher, Lowery, and Gray (1999) did examine lobbyist numbers across two decades in two states, but the authors were chiefly concerned with how many interest groups those lobbyists represented. More recent studies have begun to explore the connections between lobbyists and their clients in more detail. Relying on a combination of interviews and lobby transparency data, Drutman (2015) argues that lobbyists are responsible for encouraging firms to support more lobbying efforts, leading to crowdedness and competition. LaPira, Thomas, and Baumgartner (2014) suggest that interest groups will hire different kinds of lobbyists based on how much monitoring of an issue is needed. LaPira and Thomas (2017) have also examined the number of “revolving door” lobbyists active in Washington and how their activities differ based on their prior government experience. These studies are important steps in the development of our understanding of how lobbyists affect the representation of their clients, but they prioritize (as subjects of theory and observation) the actions and characteristics of the lobbyists over those of the clients. The number of lobbyists a group hires should be interpreted as a function of how intensely the group seeks to engage in lobbying. Just as groups might pay lobbyists to spend additional time advocating for their interests, the number of lobbyists a group hires is also a result of how much lobbying the group seeks to support (once it chooses to lobby). For example, an interest group might choose to hire a “team” or firm of lobbyists, which is not uncommon in the U.S. states (see Rosenthal 1993, 57). This occurs often when legislatures consider controversial issues that pit teams of lobbyists against each other. Interest groups might also engage in “crowd lobbying” where groups of citizen-lobbyists register and storm legislative offices (see Lofland 1982). By contrast, a group may choose to hire a fraction of a lobbyist by retaining a multiclient professional on an hourly basis as needed (see Drutman 2015, 155–67). 354 State Politics & Policy Quarterly 19(3) Being able to measure the number of lobbyists that interest groups hire allows scholars to capture how vigorously those groups are choosing to lobby, on average. If two political systems each contain roughly equal numbers of registered interest groups but with substantial differences in numbers of registered lobbyists, then this would suggest that groups chose to hire (and register) more agents in one system. Gray and Lowery (1996a, 255) have distinguished this temporary “intensity of lobbying effort” from the interest representation that they aimed to measure. To them, interest populations are different from the “gradations of engagement” that individual groups might employ depending on which issues become part of the “public policy agenda” (p. 7). One might imagine that if legislators in a state were to consider a particularly controversial proposal, then relevant interest groups might seek influence by crowding the statehouse with advocates. Such an event occurred in Florida in 1990 when interest groups registered more than 5,000 lobbyists in response to a fractured budget battle (Brasher, Lowery, and Gray 1999). In contrast to populations of interest groups, lobbyist-client pairings or dyads consist of the individual agreements that link groups with advocates (Hunter, Wilson, and Brunk 1991 referred to these dyads as contracts). For example, whereas there were more than 5,000 lobbyists registered in Florida in 1990, there might have been thousands more lobbyist-client dyads since many of those lobbyists might have been authorized to represent more than one client. Some clients might have also hired teams of lobbyists, thereby increasing numbers of lobbyist-client dyads even further. Most U.S. states provide lobbyist registration data granular enough to make measuring average rates of lobbying a straightforward task. Historically, the states have provided lists of registered lobbyists that included the individual names of their clients. Totals of lobbyists, clients, and lobbyist-client dyads can be calculated based on these lists. To demonstrate how lobbyist-client dyadic data can be used to measure lobbying intensity, let us turn to a sample of registered lobbyists and clients from Alaska for the year 2000. Figure 1 lists the first twenty lobbyist-client dyads from Alaska’s lobbyist list. There are 15 unique lobbyists and 17 unique clients. On average, each interest group hired 1.18 lobbyists. If these same clients had decided to hire more lobbyists, then this quotient would have increased. Likewise, if the clients had hired fewer lobbyists, then the average number of lobbyists hired per client would have decreased. Complete lists of lobbyist-client dyads from the U.S. states are always much longer than the sample presented in Figure 1. Figure 2 shows a box plot of interest group populations, lobbyist-group dyads, and dyads per 1,000 groups for nearly all U.S. states from 1989 and 2011. The box plot shows that there are differences between U.S. states not only in numbers of interest groups registered but also in how many lobbyists those groups hired. Whereas group populations have increased over time in the states, state-level totals of lobbyist-client dyads have increased even more. Importantly, the average number of lobbyists hired by groups has increased over time. The average number of lobbyists hired by each interest group is represented by the third and sixth boxes. It has been multiplied by 1,000. In 1989, across all U.S. states, interest groups hired an average of 1.8 total lobbyists. By 2011, across 47 states, interest groups hired an average of 2.1 total lobbyists. These statistics were calculated based on state-level averages.1 355 Strickland Lobbyist Client Don Kubley Global Seafoods North America, Inc. Jerome Selby Global Seafoods North America, Inc. Linda Anderson Alaska Visitors Association Linda Anderson Fairbanks Native Association Linda Anderson Fairbanks North Star Borough School District Linda Anderson Utility Services of Alaska Linda Anderson Fairbanks Memorial Hospital Linda Anderson Coalition of Alaskan Way of Life Sharon Anderson Alaska State Hospital and Nursing Home Association Willie Anderson NEA - Alaska Timothy M. Armstrong American Legion Department of Alaska Rod Arno Coalition of Alaskan Way of Life Tuckerman Babcock Matanuska Electric Association Judy Bakeberg City of Wrengell John Baker City of Wrengell John A. Barnes Marathon Oil Company Marie Becker Alaska Village Electric Cooperative David G. Bedford Southeast Alaska Seiners Association Joel Bennett Defenders of Wildlife Nancy Bennett American Cancer Society Figure 1. Alaska registered lobbyists in 2000 (excerpt). Limitations of the Measure While the average number of lobbyists hired by each interest group provides a measure of how intensely groups are lobbying within a state and year, there are other kinds of political mobilization that this measure does not capture. Many interest groups consist of dues-paying members. These groups might engage in inside lobbying infrequently, preferring instead to engage in outside, grassroots efforts. Such groups include public interest groups and labor unions, and such activities include coordinating protests or letter-writing campaigns (see Kollman 1998). My measure of lobbying mobilization does not reflect these outside activities since the measure relies only on lists of registered lobbyist-client dyads. Moreover, my measure does not consider any aspect of campaign finance activities. In the U.S. states, interest groups often (but not always) give to candidates via political action committees (Benz et al. 2011). Shifts in local policy agendas can sometimes spur more giving (Kirkland, Gray, and Lowery 2010). While my measure focuses exclusively on lobbying, grassroots efforts and campaign finance activities are forms of political mobilization separate from direct, inside lobbying. All these forms of mobilization should be interpreted as techniques that groups 356 State Politics & Policy Quarterly 19(3) Figure 2. Groups and lobbyist-group dyads for two years. rely on more or less often depending on political circumstances. Having a more direct measure of lobbying mobilization can help shed light on when groups prefer this tactic more often over outside efforts or campaign giving. Moreover, the average number of lobbyists hired per group masks differences in relationships between lobbyists and clients. Some lobbyists work for individual clients as full-time, in-house counsel. Others not only work on retainer for multiple clients but also work as lobbyists on a full-time basis. Still other lobbyists may not lobby on a full-time basis but instead have other jobs or fulfill other functions for their organizations. Indeed, Milbrath (1963, 117) found substantial differences in the activities that federal lobbyists focused on during each day. He also found that some lobbyists have more prominent roles in managing or leading their client organizations (pp. 145–61). In the U.S. states, contract lobbyists are quite different from in-house lobbyists in terms of hours spent lobbying and quantity of campaign donations given (see Gray and Lowery 1996b; Rosenthal 1993). My measure of overall lobby intensity is not intended to account for lobbyist-level differences in time spent lobbying. More granular data and measurements are needed to determine how often interest groups turn to different lobbyists, or even to other forms of political mobilization. Despite its limitations, my measure of lobby mobilization is intended to serve more as a proxy for the overall demand for lobbying within a state (see Dusso 2010; Leech et al. 2005), and it can be used to test existing proposals of how politics and institutions affect such demand. Prior studies of lobbying in the states have used imprecise measures. While the Energy-Stability-Area model of Gray and Lowery (1995) captures the dynamics of interest group populations, subsequent scholars have often conflated Strickland 357 interest populations with overall lobbying activity. This is problematic given that the hypothesized effects of various political and institutional factors should chiefly affect the lobbyists of groups. Groups are often credited with seeking access, for example, but it is individual lobbyists who must gain access and communicate with policymakers. In a political system where clients’ single-client lobbyists cannot achieve access to incumbents easily, for example, groups might be more likely to retain multiclient contractors who act as de facto gatekeepers. Since groups are no longer hiring their own singleclient advocates, my measure of overall lobby intensity would decrease and reflect the scarcity of access while interest populations would remain stable (provided that groups do not exit politics altogether). Hence, it is important to test such theories with data that would more precisely capture hypothesized effects. Scholars of interest groups have proposed several hypotheses about the effects of political and institutional factors on the mobilization of interest groups. Political factors refer to short-term events or circumstances that might energize groups into lobbying more intensely, or dissuade them from lobbying. For example, if legislatures are more evenly split between competing parties, then the resulting policy uncertainty might spur groups to hire more lobbyists as a form of insurance (LaPira and Thomas 2017, 52–58). Institutional factors consist of more long-term features that might affect the expected returns from lobbying, or the cost of lobbying and political access. If a state’s constitution allows for direct democracy techniques, then interest groups might focus less on direct, inside lobbying and instead shift towards outside techniques. Government spending, legislative staff capacity, legislator term limits, and formal regulations of lobbying have all also been proposed as possible catalysts or dampers of lobby activity. Politics, Institutions, and Lobbying Reduced policy stability can energize groups into hiring more lobbyists. For example, as in Florida in 1990, if interest groups perceive that policy is more likely to shift on issues that they care about, then not only will they be more likely to register lobbyists, but the average number of lobbyists hired per client should also increase. If policy is perceived to be stable, however, then there is little added benefit to crowding a statehouse with advocates. Strickland (2018a) found that more interest groups register in states with legislatures that are more evenly divided between political parties. Likewise, there are fewer groups with registered lobbyists in states with legislatures dominated by single parties. To determine whether partisan division affects lobby intensity, I incorporate a folded Ranney index into statistical models that predict lobby intensity (see Ranney 1976). The index ranges from 0 to 0.5, with higher values indicating greater one-party dominance within the legislature over the prior six years.2 If partisan competition encourages interest mobilization, then this variable ought to be negatively associated with the average number of lobbyists hired by each interest group. If a state’s constitution allows for direct democracy, then this might affect the strategies of registered interest groups. Boehmke (2002) found that there are roughly 17% more groups with registered lobbyists in states with direct democracy than in other states. The bulk of these additional groups tended to be citizen interests. On average, groups in these states were more likely to rely on outside lobbying tactics instead of on 358 State Politics & Policy Quarterly 19(3) direct, inside lobbying (Boehmke 2005, 120). Following Boehmke’s example, I include a dichotomous indicator within my models for whether a state allowed for direct democracy. It is possible that groups in these states rely less on direct lobbying and more on outside efforts such as coordinating protests and letter-writing campaigns (see Kollman 1998). Accordingly, I expect registered interest groups to engage in lobbying less frequently on average in states with direct democracy. State spending may energize interest groups into lobbying more intensely. If interest groups lobby to protect particularized benefits enshrined in state budgets, then they have more to lose in case the status quo is disrupted. Multiple Virginia School (i.e., public choice) scholars of lobbying have offered similar hypotheses. Using data from 1970, Mueller and Murrell (1986) found that there were more mobilized interests in countries with larger public sectors. Coughlin, Mueller, and Murrell (1990) further illustrated the logic and comparative statistics of interest-driven government. While these scholars expected interests to seek more state spending, Salisbury (1992, 85–86) instead expected public spending to attract more interest groups. Regardless of the causal direction, both theories suggest there will be more interest mobilization in states that spend more money. Those registered groups might also lobby more intensely. Once a government allocates more resources toward constituencies in various sectors (perhaps public education, health care, or safety services), then groups representing those constituencies will have an added incentive to mobilize both to protect themselves against losses and to seek additional resources. Not only will the groups become active but they will also seek to secure their gains through sustained increased lobbying activity (see Leech et al. 2005 for an examination of how changes in policy activity across issue domains affect lobbying intensity). Recent evidence suggests that this might not be a linear correlation. Drutman (2015, 68) found that the lobbying of a set of companies “level[ed] off” after reaching a threshold. Moreover, when examining lobby activities and government contracts, Ridge, Ingram, and Hill (2017) found that such lobbying returned diminishing benefits. I suspect that lobbying efforts yield state expenditures at a declining marginal rate, or that (equivalently) public spending will generate more lobbying at a declining marginal rate. If spending encourages lobbying, or even if lobbying encourages spending, then logged spending should be positively correlated with numbers of lobbyist-client dyads. General spending statistics were provided by Klarner (2015) and exclude utility, liquor store, or social insurance trust expenditures. Legislative institutions might also help predict lobbying intensity. Lobbyists often serve as sources of timely and relevant information for lawmakers (see Bauer, Pool, and Dexter 1963; Hall and Deardorff 2006). Berkman (2001) argues, however, that not all lobbying environments are similar in that some legislatures have permanent staff members who can also provide information to lawmakers. Such staff can substitute for lobbyists, thereby discouraging lobbying. Kattelman (2015) approached the relationship differently by expecting such staff to serve as additional access points for lobbyists, thereby increasing lobbying activity. Legislators in more professionalized assemblies would more effectively “capitalize on the information supply that groups provide” (p. 171). Even though both Berkman and Kattelman used Squire’s (2007) measure of legislative professionalism to test their theories, both of their theories Strickland 359 hinge on the presence of staff persons in the legislature. Accordingly, I incorporate the total dollars spent on each state’s legislature into my models, in terms of dollars per legislator. These figures were provided by Bowen and Greene (2014), who collected the data from the U.S. Census Bureau. Limitations on how long individuals may serve as legislators has the effect of increasing legislative turnover and reducing institutional knowledge and policy competence among incumbents (Carey, Niemi, and Powell 1998; Moncrief and Thompson 2001). As a result, term-limited legislators are perceived by lobbyists as having to rely more often on staff persons and outsiders (i.e., interest groups) for information. Mooney (2007) proposes that reductions in competence among legislators forces lobbyists to engage in educating incumbents more often. As a result, following the implementation of term limits, lobbyist populations grew more quickly in term-limited states than in other states. Such increases could have been the result of interest groups shifting from multiclient contract lobbyists to single-client in-house advocates whom amateur lawmakers regarded as more trustworthy. To determine whether term limitations truly increase lobbying within a state, I include in my models a dichotomous variable for when such limits went into effect. Formal regulations of lobbying might also affect the hiring of lobbyists. Strickland (2018a) showed that more interest groups were registered in U.S. states with more lobbyist registration criteria but that this effect was moderated by additional campaign finance regulations that affected only the activities of registered groups and lobbyists. His study examined interest populations and did not explore how registration criteria, campaign finance limits, and reporting requirements affect the hiring of lobbyists. In this study, I adopt a similar model specification to test for whether criteria encourage groups to register more lobbyists, and whether additional laws dampen this effect. When numbers of interest groups are held constant, the number of lobbyist-group dyads within a state should reflect the residual effect of these laws on lobbyist hiring rates. If the number of lobbyist-client dyads is depressed by lobby laws, then this would suggest that more laws encourage clients to hire multiclient contractors (instead of single-client, in-house lobbyists). I incorporate measures of different lobby laws into my models, as provided by Newmark (2005). As outlined in Table 1, Newmark’s scale of lobby laws consists of three types of laws: lobbyist registration criteria, prohibitions on their gift-giving and campaign finance activities, and reporting requirements. Registration criteria range in number from zero to seven. States received an additional point for each criterion they had on the books. Prohibited activities were numbered from zero to four, and reporting stringency was also measured between zero and seven. As with registration criteria, states received higher scores for each additional prohibition or reporting requirement. As Strickland (2018a) did, I use the three components of Newmark’s scale as separate independent variables. I expect criteria to be positively associated with lobbyist hiring rates (as least in terms of registered lobbyists), but prohibitions and reporting requirements might each dampen (interact with) this effect.Registered interest groups may prefer not to register additional lobbyists if doing so triggers prohibitions on their political activities, or required reporting of group and lobbyist activities. Table 2 provides summary statistics from each of my model covariates. 360 State Politics & Policy Quarterly 19(3) Table 1. Newmark’s Measure of Lobby Regulations. Definitions of lobbyists include: • Those seeking to lobby the legislature • Those seeking to lobby administrative agencies • Elected officials acting as lobbyists • Public employees acting as lobbyists • Compensation standards • Expenditure standards • Time standards Prohibited activities involving lobbyists include: • Making campaign contributions at any time • Making campaign contributions during legislative sessions • Making expenditures in excess of a certain dollar amount per official per year • Solicitation by officials or employees for contributions or gifts Reporting requirements for lobbyists include: • Semi-annual or more frequent reporting for lobbyists or their employers • Name of targeted legislation or administrative action • Expenditures benefiting public officials or employees • Compensation received, broken down by employer(s) or employee(s) • Total compensation received • Categories of expenditures made • Total expenditure made Source. Newmark (2005). Table 2. Summary Statistics for Model Covariates. Statistic Lobbyist-Client Dyads Registered Clients Registered Lobbyists One-Party Dominance Folded Ranney index. Initiative State State Spending In billions. Legislature Expenditures In thousands. Term Limits Lobby Definitions Lobby Prohibitions Reporting Requirements Minimum 25% Median 75% Max M SD 147 97 110 0.00 729 417 368 0.046 1,301 694 555 0.110 2,428 1,142 820 0.189 43,760 4,275 5,727 0.374 2,431 892 759 0.126 3,858 689 736 0.094 0 1 0 5.4 1 12 1 24 1 230 0.534 19.9 0.499 26.6 51 241 437 687 5,521 649 776 0 0 0 0 0 2 0 3 0 4 1 5 1 5 2 7 1 7 4 7 0.146 3.668 1.016 4.666 0.353 1.683 1.122 1.970 Controlling for Interest Populations When measuring mobilization using the number of registered lobbyist-group dyads within a state, one should control for the number of interest groups registered. The Strickland 361 number of registered groups in a state should be a leading predictor of how many lobbyist-group dyads there are. By holding the number of groups constant, one can isolate the effects of explanatory variables on the mobilizational intensity of the groups. Interest populations are typically correlated with the size of a state’s economy (Gray and Lowery 1996a). While it is possible to estimate rates of lobbyist hiring while holding economic output constant, this unnecessarily introduces statistical noise into regression models. Since all registered clients must have at least one registered lobbyist, controlling for client totals most efficiently isolates the effects of other variables on rates of lobbyist hiring. Controlling for group totals also results in more conservative estimates of effect sizes for other covariates. Since totals of registered interests have already been shown to fluctuate in response to political and institutional factors, controlling for their totals will already capture much of the effects of these factors. Any residual variance that my econometric models explain will, therefore, capture the effects of these factors on numbers of lobbyist-group dyads. Totals of lobbyists, clients, and lobbyist-client dyads were calculated based on more than 600 lists of registered lobbyists spanning all U.S. states between 1986 and 2013. The lists were generated by state authorities typically at the end of each legislative session. The lists were drawn from a variety of sources. These include state libraries and archives that the author visited in 29 U.S. states, online databases maintained by state authorities, a directory (i.e., Wilson 1990) consisting of state lobbyist lists from 1989, and lobbyist lists produced by the National Institute on Money in State Politics (which provided most lists from after 2005). Once the lists of lobbyists-client dyads were located in archives or elsewhere, they were transcribed into spreadsheets by research assistants. The spreadsheets allow for the straightforward calculation of dyad totals, and totals of unique lobbyist and client names. There are missing observations in the sample as many lobbyist lists could not be found. Lists from nearly all states were found, however, for 1989 and years after 2005. For nearly all states in my sample, lobbyists alone were tasked with registering their names and the names of their clients. The lists resemble the lobbyist-client dyads presented in Figure 1 but are significantly longer. The U.S. states were the first regimes in the world to require lobbyists to register and list client names (see Opheim 1991), and their political and institutional variation allows for statistical inference. Massachusetts was the first state to require lobbyists to register, beginning in 1891. Historically, lobbyists were responsible for registering their names and clients with state authorities. Such authorities included legislative clerks, secretaries of state, or ethics agencies (see Strickland 2018b for a longer history). Lobbyists would typically be required to sign their names in legislative docket books, and record the names of the firms, organizations, or interests they represented. In some states, lobbyists would also be required to provide expense statements. By 1975, all U.S. states required lobbyists to register. By the late 1980s, West Virginia and Arkansas were the last states to delegate lobbyist registration to staff persons or agencies outside of the legislature (C. S. Thomas 1998). When working with lists of lobbyists from the states, one must control for idiosyncratic registration procedures in some states. For recent years, there are a few states where lobbyist lists do not clearly indicate which lobbyists were authorized to 362 State Politics & Policy Quarterly 19(3) represent which clients. In these states, lobbyists were allowed to register as members of firms. Representatives of interest groups were also allowed to authorize firms to represent them. Hence, because these states provided only lobbyist-firm matches or client-firm matches, lobbyist-client dyads could be counted only if one assumed that all lobbyists members of firms were authorized to represent all clients associated with their firms. This was the case in California, New Jersey, New York, and Pennsylvania. As a result, lobbyist-client lists from these states provided by the Institute are particularly long and may contain some spurious lobbyist-client dyads. I include a dichotomous control for those four states within my econometric models. Also, beginning in Michigan in the early 1980s, lobbyists were no longer required to re-register for each legislative session. Michigan’s lobbyist list from 1989 (provided in Wilson 1990) likely includes numerous lobbyist-client dyads that were active during prior sessions. I include a dichotomous indicator also for this observation. More recent lists from Michigan provided by the Institute are significantly shorter and do not appear to contain outdated dyads.3 Estimation Method and Results My dependent variable is the total of lobbyist-client dyads registered within a U.S. state. Since this variable is a non-negative count of events, I estimated regression coefficients using the negative binomial variance function. This function helps account for overdispersion where model-conditional variances exceed model-conditional means (see King 1988; Long 1997, 230–41). Moreover, since I am working with multiple observations within each state, my observations are clustered into different political systems. This violates the least squares assumption of independent errors. As a result, I estimated models with standard errors clustered by state (see Primo, Jacobsmeier, and Milyo 2007). These models do not control for the different starting points of states or for national factors that might affect all states similarly. For eliminating the influence of those factors, I estimate models with fixed effects included for each state and year. These models provide more conservative estimates of effect sizes because they rely only on within-state changes (see Allison 2009). Effect sizes in these models may be artificially small for institutional variables that change slowly over time (see Beck and Katz 2001). Models 2 and 4 exclude observations from Nebraska’s nonpartisan Unicameral. Results We can draw several conclusions based on the results presented in Table 3. The first is that the number of registered interest groups within a state is a leading predictor of how many lobbyist-client dyads there are. This result is robust to the inclusion of state and year fixed effects, which forces my models to rely only on variation that occurs within states but over time. This significant correlation is unsurprising given that all lobbyists must have at least one client. Instead, the other coefficients that partially explain the residual heterogeneity are more interesting. They tell a complex story. 363 Strickland Table 3. Explaining Lobbying in the U.S. States. Model 1 Clustered SE Model 2 Clustered SE Model 3 Fixed Effects Model 4 Fixed Effects Interest Groups 1,000 One-Party Dominance 0.826*** (0.108) — 0.769*** (0.038) Initiative State 0.020 (0.075) 0.306*** (0.056) −0.184*** (0.057) −0.010 (0.073) 0.030 (0.046) 0.158 (0.081) −0.007 (0.026) −0.032 (0.017) −0.003 (0.008) 1.563 (0.820) −2.396 (0.163) 630 50 −4,734 9,495 0.833*** (0.100) −0.627** (0.286) 0.042 (0.076) 0.300*** (0.055) −0.200*** (0.056) −0.013 (0.077) 0.025 (0.045) 0.132 (0.075) −0.008 (0.027) −0.028 (0.016) −0.003 (0.008) 1.757** (0.827) −2.421 (0.152) 620 49 −4,658 9,344 0.763*** (0.039) −0.227 (0.137) 0.877*** (0.220) 0.015 (0.120) 0.100 (0.065) −0.120** (0.048) 0.087*** (0.026) 0.067 (0.054) 0.047** (0.020) −0.019 (0.012) −0.010 (0.005) 5.606*** (1.875) −3.399 (0.058) 620 49 −4,357 8,894 Logged State Spending (In $1,000s.) Legislature Expenditures Term Limits Lobby Definitions Lobby Prohibitions Reporting Requirements Definitions × Prohibitions Definitions × Reporting Constant ln(α) Observations No. of States Log Likelihood AIC — 0.916*** (0.218) 0.034 (0.119) 0.113 (0.065) −0.142*** (0.046) 0.094*** (0.025) 0.078 (0.053) 0.049** (0.020) −0.021 (0.012) −0.011 (0.005) 5.236*** (1.856) −3.407 (0.058) 630 50 −4,418 9,015 Note. Standard errors are in parentheses. AIC = Akaike information criterion. **p ≤ .05. ***p ≤ .01 on two-tailed tests. When models are allowed to use both across-state and within-state variation to calculate effect (coefficient) sizes, one-party dominance is negatively correlated with numbers of lobbyist-client dyads. Even though partisan dominance has been found to discourage interest groups from registering (see Gray and Lowery 1996a), there appears also to be some effect on the lobbying intensity of registered groups. In models with fixed effects, groups did not mobilize or de-mobilize in response to shifts in partisan dominance. Those models used only variation that occurs within states but other time. The absence of a discernible correlation Model 4 suggests that most heterogeneity in one-party dominance was found between states. The exclusion of client 364 State Politics & Policy Quarterly 19(3) totals from my fixed-effects models does not alter this result. The differences across models also suggest that party dominance may be correlated with a time-invariant confounder that is not included within the first two models. Models 3 and 4 also suggest that changing a state’s initiative status led to more lobbying on the part of registered interest groups. However, this result is due to a single outlier within a state.4 Re-estimating my models after excluding this outlier eliminates the statistical significance of initiative state status. Similar explanations may be applied to logged state spending and legislature expenditures. These variables were also significant predictors of lobby intensity only in models with clustered standard errors. In models with fixed effects, results do not provide evidence that increases in U.S. state spending led to more lobbying within states. This is in contrast to the results of Mueller and Murrell (1986), who identified more interest groups in nations with more state spending. My results also do not suggest that increases in legislative expenditures lead to less lobbying, as Berkman (2001) suggested. This does not mean that these factors have no effects on interest populations. State spending and legislature expenditures might still affect the totals of registered interest groups. If one excludes group totals from my models, then state spending becomes a positive and significant predictor of lobbyist-client pairings. This suggests that most of government spending’s effects on lobbying are through the entry or exit of additional groups, and not additional lobbyists. The exclusion of group totals from my models does not alter substantive results for legislature expenditures. In contrast to the expectations of Mooney (2007), states in which term limits took effect saw a statistically discernible drop in lobbyist hiring. This result persists even when group totals are excluded from my models with fixed effects. This result is likely not due to outliers since term limits went into effect in 16 states throughout my sample.5 This does not suggest that Mooney’s analysis is flawed. Whereas Mooney explored how lobbyist totals changed over time in states with term limits and in states without limits, my analysis takes into consideration fluctuations in both client and lobbyist totals. According to my results, interest groups in states with limits hired roughly 12.3% fewer lobbyists than groups in states without limits, on average. What might explain this counterintuitive relationship? In models not reported here, term limits are found to be a negative predictor of client totals. Yet, even among registered interests, lobbyist hiring is depressed as well. These trends suggest that term limits might create an environment that discourages lobbying in general. At the same time, however, the effects of term limits are not the same in every U.S. state. States vary in the severity of their term-limits laws (see Sarbaugh-Thompson 2010), and there is substantial variation in legislator turnover even among states without limits (see Moncrief, Niemi, and Powell 2004). Further exploration is needed to help identify the causal mechanism linking term limits with lobbying. My models also suggest that the implementation of some lobbyist laws affect rates of lobbyist hiring by interest groups. Strickland (2018a) found that more registration criteria get more interest groups to register but that this relationship is moderated by prohibitions on lobbyists giving gifts and campaign donations. Since group totals are already affected by lobby regulations, it is notable that these laws also affect the hiring Strickland 365 of lobbyists among the remaining groups who register. The addition of registration criteria within a state led to more lobbyists registering relative to group totals. The implementation of more reporting requirements also had a positive effect. It was expected that reporting requirements would dampen the influence of definitions in capturing more groups, but the (negative) interactive term between the two variables did not achieve traditional levels of statistical significance. Limits on gift-giving and campaign finance activities were not correlated with lobbyist-client dyads. Tests for Multiclient Lobbying While the initial set of regression models with fixed effects show that term limits and lobby laws are correlated with the hiring of lobbyists by interest groups, the measure of overall lobbying does not reflect the popularity of multiclient advocates over singleclient ones. Interest groups in some states rely more often on multiclient lobbyists than on single-client ones (see Strickland and Crosson 2016). These differences may prove problematic for measuring the overall level of lobbying. Dyad counts can be influenced by the share of clients hiring multiclient over single-client lobbyists, even when group counts are held constant. While dyad counts would decrease if interest groups shifted (on average) from single-client advocates to multiclient advocates, two states with similar numbers of groups could have similar dyad counts but with different numbers of lobbyists. This is possible so long as lobbyists in one state tend to represent one client each whereas those in the other state each represent multiple clients. Let us imagine two U.S. states with six clients each. In one state, there are 10 lobbyists who represent one client each (some clients hired two single-client lobbyists). As a result, there are 10 lobbyist-client dyads in all. In the second state, there are also six clients and 10 lobbyist-client dyads, but only six lobbyists. That is because four of the six lobbyists represent two clients each. Even though the interest groups in the first state have each hired only single-client lobbyists (with some of those groups hiring two such lobbyists), the state has the same number of dyads and clients as in the second state, where four of the six clients hired multiclient lobbyists. Hence, the dyad count within a state can mask the popularity of multiclient advocates, even when client totals are held constant. This is problematic for my measure since I argue that any group that hires a single-client lobbyist is mobilizing more intensely than a group that shares a multiclient lobbyist with other clients. In additional model specifications, I test for whether political and institutional variables are correlated with multiclient lobbying. When client and lobbyist counts are held constant, the number of lobbyist-client dyads in a state is a proxy for how much multiclient lobbying is occurring. If client and lobbyist totals are the same in two states, then there is more multiclient lobbying in the state with the greater number of dyads. If a political or institutional variable is correlated with multiclient lobbying (i.e., it predicts dyads in models with both lobbyist and client totals), then it may not affect overall lobbying so much as it affects the balance of multi- versus single-client lobbyists among groups. On the other hand, if a political or institutional variable is not correlated with multiclient lobbying, then the results presented for it in Table 3 are 366 State Politics & Policy Quarterly 19(3) Table 4. Tests for Multiclient Lobbying. Model 5 Clustered SE Interest Groups 1,000 Lobbyists 1,000 One-Party Dominance Initiative State Logged State Spending (In $1,000s.) Legislature Expenditures Term Limits Lobby Definitions Lobby Prohibitions Reporting Requirements Definitions × Prohibitions Definitions × Reporting Constant ln(α) Observations No. of States Log Likelihood AIC 0.619*** (0.090) 0.237** (0.102) — −0.010 (0.067) 0.298*** (0.056) −0.148*** (0.048) −0.016 (0.063) 0.050 (0.037) 0.186** (0.080) −0.002 (0.020) −0.039** (0.016) −0.003 (0.007) 1.584 (0.805) −2.550 (0.145) 630 50 −4,685 9,398 Model 6 Clustered SE Model 7 Fixed Effects Model 8 Fixed Effects 0.635*** (0.085) 0.224** (0.102) −0.418 (0.221) 0.006 (0.067) 0.296*** (0.056) −0.160*** (0.047) −0.015 (0.068) 0.045 (0.037) 0.166** (0.075) −0.003 (0.021) −0.035** (0.016) −0.002 (0.007) 1.698** (0.821) −2.557 (0.143) 620 49 −4,615 9,261 0.524*** (0.051) 0.294*** (0.042) — 0.953*** (0.211) 0.502*** (0.052) 0.310*** (0.043) −0.275** (0.131) 0.907*** (0.212) 0.072 (0.115) 0.227*** (0.066) −0.068 (0.045) 0.094*** (0.024) 0.083 (0.051) 0.049*** (0.019) −0.018 (0.011) −0.013** (0.005) 4.553*** (1.789) −3.487 (0.058) 630 50 −4,393 8,967 0.048 (0.115) 0.215*** (0.066) −0.033 (0.047) 0.084*** (0.025) 0.066 (0.052) 0.049** (0.019) −0.014 (0.011) −0.013** (0.005) 4.931*** (1.801) −3.487 (0.059) 620 49 −4,331 8,843 Note. Standard errors are in parentheses. AIC = Akaike information criterion. **p ≤ .05. ***p ≤ .01 on two-tailed tests. indeed reliable estimates of its effect on rates of overall lobbying. Checking for whether explanatory variables are correlated with multiclient lobbying is a necessary robustness check for determining sources of overall lobbying. I present additional model specifications in Table 4. With the exception of the introduction of lobbyist counts, the model specifications in Table 4 are the same as those in Table 3. The results presented in Table 4 provide insight into which political and institutional variables encourage multiclient lobbying. Both group and lobbyist populations are significant and positive predictors of lobbyist-client dyads. This is unsurprising Strickland 367 Figure 3. Predicted dyads for two reporting conditions. given that dyads consist of lobbyist and client pairings. One-party dominance is correlated with more multiclient lobbying. This might be because groups in divided states hire more multiclient lobbyists to appeal to legislators of both parties. In other words, divided government might benefit multiclient advocates by encouraging groups to hire firms with ties to both parties, as a form of hedging bets. Legislative expenditures is a significant predictor of multiclient lobbying, as well. This may be due to spending being correlated with another factor that might influence lobbying, such as legislative institutionalization (see Berry, Berkman, and Schneiderman 2000).6 Importantly, the implementation of term limits in a state was not correlated with more or less multiclient lobbying. Term limits did not compel groups to change their choice of lobbyist types. This re-affirms the results originally presented in Table 3. As for lobby laws, the interactive effect between registration criteria and reporting requirements achieves statistical significance in models with lobbyist counts. This suggests that the imposition of registration criteria and reporting requirements got existing groups to register more dyads with multiclient lobbyists in a state but that reporting requirements dampened this trend somewhat. Figure 3 shows this dampening effect under two reporting conditions when all variables (except definitions) are held at their means. The predicted dyad counts are based on the results presented in Model 8. In general, the results presented in Tables 3 and 4 suggest that the effects of political and institutional factors on lobbying are more complex than originally thought. Scholars of interest representation have proposed a variety of hypotheses for how these factors might influence lobbying. They have also used a variety of measurement scales for their explanatory variables. These variables, however, might affect any one 368 State Politics & Policy Quarterly 19(3) of at least three aspects of lobbying: interest populations, lobbyist-client dyad counts, and the incidence of multiclient lobbying. Whereas term limits depresses lobbyist hiring and is not correlated with multiclient lobbying, the effects of lobby laws on overall lobbyist hiring may be an artifact of groups shifting to multiclient lobbyists in response to new laws. Discussion and Conclusion This study was intended to improve our understanding of lobbying in the U.S. states. Prior studies of interest mobilization in the U.S. states have conflated interest populations with mobilization, or have explored variations in raw totals of lobbyists. To build on these studies, I proposed a simple measure for the average lobbying effort of interest groups in a state. At the group level, the number of lobbyists the group hires can be interpreted as a proxy for its intensity of lobbying effort. When aggregated to the level of states, the total number of lobbyist-client dyads is a proxy for lobby intensity (when client totals are held constant). This measure does not reflect other forms of political mobilization such as outside efforts or campaign finance activities, and it also does not measure the balance of single- versus multiclient lobbyists, but it advances our understanding of lobbying by allowing scholars to test existing hypotheses with greater precision. Do interest groups mobilize in response to short-term political factors, or are their rates of lobbying generally steady because of long-term institutional features? This study was conducted partly in response to prior studies of interest mobilization in the U.S. states. Others have found evidence that politics and institutions play some part in structuring interest representation. Gray and Lowery (1996a) found that partisan competition spurs more groups to register. In several studies, Boehmke (2002; 2005) found that there are more interest groups in states with direct democracy. Mueller and Murrell (1986) identified more groups in nations with more state spending. Lobbyist totals grew more quickly in states with term limits than in states without them (Mooney 2007). Berkman (2001) found there are fewer groups in U.S. states with more professional legislatures, and Strickland (2018a) found that gift-giving and campaign finance restrictions push down registration totals for groups. All these studies have contributed to our understanding of the mobilization of interest groups, but they focused more on the entry or exit of groups instead of on how many lobbyists, or what kinds of lobbyists, they hired. Interest populations are substantively meaningful, but they mask the true amount of lobbying that occurs within America’s crowded statehouses. By focusing on numbers of lobbyist-client dyads, I found that it is mostly institutional factors that shape the hiring rates of lobbyists. In Models 3 and 4, term limits and lobby laws have the strongest discernible correlations with the hiring of lobbyists. The implementation of term limits in a state was associated with decreased rates of lobbyist hiring, while registration criteria and reporting requirements were associated with more dyads. Once I controlled for lobbyist totals in Models 7 and 8, term limits were no longer a significant predictor of dyad counts, and reporting requirements were found to interact with registration criteria. Moreover, one-party dominance (a short-term, political factor) Strickland 369 and legislature expenditures became significant predictors of dyad counts. The differences in model results across tables were indicated shifts in the balance of single- versus multiclient lobbyists. If both client and lobbyist populations are held constant, then dyads increase in response to more multiclient lobbying. These findings present several additional questions for scholars of interest representation. To understand better the effects of both political and institutional factors on lobbying, it would be beneficial to examine rates of lobbying by individual sectors or guilds of interests. Legislative activity in specific policy domains may spur more lobbying on behalf of only some sectors of interest groups (see Dusso 2010; Gray, Lowery, and Fellowes 2005). As state economies and interest populations grow, there may also be unequal, long-term growth between sectors in their political mobilization (see Lowery, Gray, and Fellowes 2005). If sectors of interest groups vary the strength of their lobbying efforts in response to policy agenda shifts, then variables such as legislative gridlock and the granting of economic rents may partly be results of differential lobbying efforts across sectors of interests (see Gray and Lowery 1995; Olson 1982). Exploring how state-level partisan competition influences totals of lobbyist-client dyads may also be too broad of an approach for truly capturing the short-term effects of politics on lobbying. Such an approach surely misses granular differences in mobilization across policy domains or sectors. It might also be the case that types of interest groups vary in how often they employ single- or multiclient lobbyists. My measure of lobbying has limitations and might be most informative when used in conjunction with measures of other forms of political mobilization. Lobbying, as a form of mobilization, is distinct from outside grassroots efforts and campaign finance activities. For interest groups, lobbying should be seen as one tool in an arsenal of political tactics. Groups can vary their use of different tools for strategic purposes. Campaign donations may be used to bolster lobbying efforts (see Hall and Wayman 1990), or groups may shift toward outside lobbying tactics in response to a proposal’s status in the legislative process (see Hall and Reynolds 2012). Moreover, groups may choose to pay their lobbyists for more hours of representation. My measure of lobbying does not capture these metrics. Much work remains to be done on studying the political tactics of interest groups in the U.S. states and the determinants of their lobbying activities. Linking group-level accounts of lobby intensity to system-level changes in politics and institutions is a challenging task that requires granular lobbyist data. Nevertheless, this article was intended to be a step forward in advancing our understanding of interest representation in the U.S. states. I anticipate that others may build on my findings. 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. 370 State Politics & Policy Quarterly 19(3) Notes 1. 2. 3. 4. 5. 6. One outlier is excluded from the box plot. In 2011, there were 32,171 lobbyist-client dyads in New York. This state regularly contained more dyads than any other state. In 2010, New York had the longest lobbyist list in my sample at 43,760 dyads. Later on, I explain why New York’s lists are significantly longer than those from other states, and I include a control for this state (and a few others) in my statistical models. This measure does not capture polarization between parties. Gray, Lowery, and Fellowes (2015) concluded that polarization within legislatures did not strengthen the relationship between partisan competition and numbers of registered interests. The two coefficients for these idiosyncratic registration procedures are not reported here. Since models with fixed effects rely only on within-state changes to estimate effect sizes, the significance of initiative state status is the result of an outlier. Most states had adopted direct democracy prior to when my sample begins, in 1986. Throughout my sample, only two states (Kentucky and Mississippi) became initiative states. There are no observations from when Kentucky did not allow for direct democracy, and there is only one observation from Mississippi from before that state’s change in initiative status. Mississippi became an initiative state in 1992. Mississippi’s lobbyist list from 1989 is short compared with its lobbyist lists from after 1993 (when lists were published online). The significance of a state’s initiative status change hinges entirely on the one observation from Mississippi. Term limitations were approved in five more states but never went into effect in some due to legal challenges. In Oregon, term limits were briefly in effect between 1998 and 2003. As of 2018, term limits apply to legislators in 15 states. In additional model specifications, I reestimated my models using the raw count of permanent legislative staff persons in each state. These figures were provided by the National Conference of State Legislatures (2016). Such permanent staffers work for legislatures even when they are not in session. Unlike with legislature expenditures, raw staff counts were not significant predictors of dyads in any of the model specifications presented here. This suggests that expenditures (and certainly professionalism) measure much more than merely staff capacity. It also suggests that staff capacity has no effect on the lobbying strategies of registered interest groups. References Allison, Paul D. 2009. Fixed Effects Regression Models. Thousand Oaks: Sage. Bauer, Raymond A., Ithiel De Sola Poole, and Lewis Anthony Dexter. 1963. American Business and Public Policy: The Politics of Foreign Trade. New York: Atherton Press. Baumgartner, Frank R., Jeffrey M. Berry, Marie Hojnacki, David C. Kimball, and Beth L. Leech. 2009. Lobbying and Policy Change: Who Wins, Who Loses, and Why. Chicago: University of Chicago Press. Beck, Nathaniel, and Jonathan N. Katz. 2001. “Throwing out the Baby with the Bath Water: A Comment on Green, Kim, and Yoon.” International Organization 55 (2): 487–95. Benz, Jennifer, Justin H. Kirkland, Virginia Gray, David Lowery, Jennifer Sykes, and Mary Deason. 2011. “Mediated Density: The Indirect Relationship between U.S. State Public Policy and PACs.” State Politics and Policy Quarterly 11 (4): 440-59. Berkman, Michael B. 2001. “Legislative Professionalism and the Demand for Groups: The Institutional Context of Interest Population Density.” Legislative Studies Quarterly 26 (4): 661–79. Strickland 371 Berry, William D., Michael B. Berkman, and Stuart Schneiderman. 2000. “Legislative Professionalism and Incumbent Reelection: The Development of Institutional Boundaries.” American Political Science Review 94 (4): 859–74. Birnbaum, Jeffrey H., and Alan S. Murray. 1987. Showdown at Gucci Gulch: Lawmakers, Lobbyists, and the Unlikely Triumph of Tax Reform. New York: Random House. Boehmke, Frederick J. 2002. “The Effect of Direct Democracy on the Size and Diversity of State Interest Group Populations.” The Journal of Politics 64:827–44. Boehmke, Frederick J. 2005. The Indirect Effect of Direct Legislation: How Institutions Shape Interest Group Systems. Columbus: Ohio State University Press. Boehmke, Frederick J. 2008. “The Initiative Process and the Dynamics of State Interest Group Populations.” State Politics & Policy Quarterly 8 (4): 362–83. Boehmke, Frederick J., and Daniel C. Bowen. 2010. “Direct Democracy and Individual Interest Group Membership.” Journal of Politics 72 (3): 659–71. Bowen, Daniel C., and Zachary Greene. 2014. “Should We Measure Professionalism with an Index? A Note on Theory and Practice in State Legislative Professionalism Research.” State Politics and Policy Quarterly 14 (3): 277–96. Brasher, Holly, David Lowery, and Virginia Gray. 1999. “State Lobby Registration Data: The Anomalous Case of Florida (and Minnesota Too!).” Legislative Studies Quarterly 24 (2): 303–14. Carey, John, Richard G. Niemi, and Lynda W. Powell. 1998. “The Effects of Term Limits on State Legislatures.” Legislative Studies Quarterly 23:271–300. Coughlin, Peter J., Dennis C. Mueller, and Peter Murrell. 1990. “Electoral Politics, Interest Groups, and the Size of Government.” Economic Inquiry 28:682–705. Drutman, Lee. 2015. The Business of America Is Lobbying: How Corporations Became Politicized and Politics Became More Corporate. Oxford: Oxford University Press. Dusso, Aaron. 2010. “Legislation, Political Context, and Interest Group Behavior.” Political Research Quarterly 63 (1): 55–67. Gray, Virginia, John Cluverius, Jeffrey J. Harden, Boris Shor, and David Lowery. 2015. “Party Competition, Party Polarization, and the Changing Demand for Lobbying in the American States.” American Politics Research 43 (2): 175–204. Gray, Virginia, and David Lowery. 1995. “Interest Representation and Democratic Gridlock.” Legislative Studies Quarterly 20 (4): 531–52. Gray, Virginia, and David Lowery. 1996a. The Population Ecology of Interest Representation: Lobbying Communities in the American States. Ann Arbor: University of Michigan Press. Gray, Virginia, and David Lowery. 1996b. “The World of Contract Lobbying.” Comparative State Politics 17:31–40. Gray, Virginia, David Lowery, and Matthew Fellowes. 2005. “Legislative Agendas and Interest Advocacy: Understanding the Demand Side of Lobbying.” American Politics Research 33 (3): 404–34. Hall, Richard T., and Alan V. Deardorff. 2006. “Lobbying as Legislative Subsidy.” American Political Science Review 100 (1): 69–84. Hall, Richard T., and Molly E. Reynolds. 2012. “Targeted Issue Advertising and Legislative Strategy: The inside Ends of outside Lobbying.” Journal of Politics 74 (3): 888–902. Hall, Richard T., and Frank W. Wayman. 1990. “Buying Time: Moneyed Interests and the Mobilization of Bias in Congressional Committees.” American Political Science Review 84 (3): 797–820. 372 State Politics & Policy Quarterly 19(3) Hunter, Kennith G., Laura Ann Wilson, and Gregory G. Brunk. 1991. “Societal Complexity and Interest-Group Lobbying in the American States.” Journal of Politics 53 (2): 488–503. Kattelman, Kyle T. 2015. “Legislative Professionalism and Group Concentration: The ESA Model Revisited.” Interest Groups & Advocacy 4 (2): 165–84. King, Gary. 1988. “Statistical Models for Political Science Event Counts: Bias in Conventional Procedures and Evidence for the Exponential Poisson Regression Model.” American Journal of Political Science 32:838–63. Kirkland, Justin H., Virginia Gray, and David Lowery. 2010. “Policy Agendas, Party Control, and PAC Contributions in the American States.” Business and Politics. 12 (4): 1-24. Klarner, Carl. 2015. “State Economic and Government Finance Data.” Version 1. Harvard Dataverse. doi:10.7910/DVN/CJBTGD. Kollman, Ken. 1998. Outside Lobbying: Public Opinion and Interest Group Strategies. Princeton: Princeton University Press. LaPira, Timothy M., and Herschel M. Thomas. 2017. Revolving Door Lobbying: Public Service, Private Influence, and the Unequal Representation of Interests. Lawrence: University Press of Kansas. LaPira, Timothy M., Herschel M. Thomas, and Frank R. Baumgartner. 2014. “The Two Worlds of Lobbying: Washington Lobbyists in the Core and on the Periphery.” Interest Groups & Advocacy 3 (3): 1–27. Leech, Beth L., Frank R. Baumgartner, Timothy M. La Pira, and Nicholas A. Semanko. 2005. “Drawing Lobbyists to Washington: Government Activity and the Demand for Advocacy.” Political Research Quarterly 58 (1): 19–30. Lincoln, Taylor. 2017. “Swamped: More Than Half the Members of Washington’s Lobbying Corps Have Plunged into the Tax Debate.” https://www.citizen.org/sites/default/files /swamped-tax-lobbying-report.pdf (accessed March 13, 2018). Lofland, John. 1982. “Crowd Lobbying: An Emerging Tactic of Interest Group Influence in California.” Davis: Institute of Governmental Affairs, University of California. Long, Scott. 1997. Regression Models for Categorical and Limited Dependent Variables. Thousand Oaks: Sage. Lowery, David, and Virginia Gray. 1995. “The Population Ecology of Gucci Gulch, or the Natural Regulation of Interest Group Numbers in the American States.” American Journal of Political Science 39 (1): 1–29. Lowery, David, Virginia Gray, and Matthew Fellowes. 2005. “Sisyphus Meets the Borg: Economic Scale and the Inequalities in Interest Representation.” Journal of Theoretical Politics 17:41–74. Lowery, David, and Kathleen Marchetti. 2012. “You Don’t Know Jack: Principals, Agents, and Lobbying.” Interest Groups and Advocacy 1 (2): 139–70. Milbrath, Lester W. 1963. The Washington Lobbyists. Chicago: Rand-McNally. Moncrief, Gary F., Richard G. Niemi, and Lynda W. Powell. 2004. “Time, Term Limits, and Turnover: Trends in Membership Stability in U.S. State Legislatures.” Legislative Studies Quarterly 29 (3): 357–81. Moncrief, Gary F., and Joel A. Thompson. 2001. “On the outside Looking In: Lobbyists’ Perspectives on the Effects of State Legislative Term Limits.” State Politics & Policy Quarterly 1 (4): 394–411. Mooney, Christopher Z. 2007. “Lobbyists and Interest Groups.” In Institutional Change in American Politics: The Case of Term Limits, eds. Karl T. Kurtz, Bruce Cain, and Richard G. Niemi. Pp. 119-133. Ann Arbor: University of Michigan Press. Strickland 373 Mueller, Dennis C., and Peter Murrell. 1986. “Interest Groups and the Size of Government.” Public Choice 48:125–45. National Conference of State Legislatures. 2016. “Size of State Legislative Staff.” National Conference of State Legislatures. http://www.ncsl.org/research/about-state-legislatures /staff-change-chart-1979-1988-1996-2003-2009.aspx (accessed December 11, 2017). Newmark, Adam J. 2005. “Measuring State Legislative Lobbying Regulation, 1990-2003.” State Politics & Policy Quarterly 5 (2): 182–91. Olson, Mancur. 1982. The Rise and Decline of Nations. New Haven: Yale University Press. Opheim, Cynthia. 1991. “Explaining the Differences in State Lobby Regulation.” Western Political Quarterly 44 (2): 405–21. Primo, David M., Matthew L. Jacobsmeier, and Jeffrey Milyo. 2007. “Estimating the Impact of State Policies and Institutions with Mixed-level Data.” State Politics & Policy Quarterly 7 (4): 446–59. Ranney, Austin J. 1976. “Parties in State Politics.” In Politics in the American states, eds. Herbert Jacob, and Kenneth Vines, 3rd ed. Boston: Little, Brown, 51–94. Ridge, Jason, Amy Ingram, and Aaron Hill. 2017. “Beyond Lobbying Expenditures: How Lobbying Breadth and Political Connectedness Affect Firm Outcomes.” Academy of Management Journal 60 (3): 1138–63. Rosenthal, Alan. 1993. The Third House: Lobbyists and Lobbying in the States. Washington: Congressional Quarterly Press. Salisbury, Robert H. 1992. Interests and Institutions: Substance and Structure in American Politics. Pittsburgh: University of Pittsburgh Press. Sarbaugh-Thompson, Marjorie. 2010. “Measuring ‘Term Limitedness’ in Cross Sectional Research.” State Politics & Policy Quarterly 10 (2): 199–217. Schattschneider, Elmer E. 1960. The Semisovereign People: A Realist’s View of Democracy in America. New York: Holt, Rinehart and Winston. Schiff, Eleanor L., Kim Seufer, Anne Whitesell, and David Lowery. 2015. “Agency Problems and Interest Representation: An Empirical Analysis of the Costs of Lobbying.” Interest Groups & Advocacy 4:225–48. Squire, Peverill. 2007. “Measuring Legislative Professionalism: The Squire Index Revisited.” State Politics & Policy Quarterly 7:211–27. Strickland, James M. 2018a. “A Paradox of Political Reform: Shadow Interests in the U.S. States.” American Politics Research. Published electronically July 31. doi:10.1177/15326 73X18788049 Strickland, James M. 2018b. “The Contingent Value of Relationships: The Supply and Demand for Revolving-Door Lobbyists.” Paper presented at the 89th Annual Conference of the Southern Political Science Association, New Orleans. https://papers.ssrn.com/sol3/papers .cfm?abstract_id=3099683 (accessed January 25, 2019). Strickland, James M., and Jesse M. Crosson. 2016. “K Street on Main: How Political Institutions Cultivate a Professional Lobbying Elite.” Paper presented at the 112th Annual Meeting of the American Political Science Association, Philadelphia. https://papers.ssrn.com/sol3/ papers.cfm?abstract_id=2838459 (accessed January 25, 2019). Tankersley, Jim, Thomas Kaplan, and Kenneth P. Vogel. 2017,. “Lobbying Frenzy Begins on Tax Bill.” New York Times, November 3. https://www.nytimes.com/2017/11/03/us/politics /republicans-tax-cut-obamacare-mandate-trump.html (accessed March 13, 2018). Thomas, Clive S. 1998. “Interest Group Regulation across the United States: Rationale, Development and Consequences.” Parliamentary Affairs 51 (4): 500–15. 374 State Politics & Policy Quarterly 19(3) Thomas, Herschel F., and Timothy M. LaPira. 2017. “How Many Lobbyists are in Washington? Shadow Lobbying and the Gray Market for Policy Advocacy.” Interest Groups and Advocacy 6:199–214. Whitesell, Anne, Eleanor Schiff, and David Lowery. 2018. “Agency Issues with Small Lobby Firms: A Test of Competing Hypotheses.” Interest Groups & Advocacy 7:82–103. Wilson, Robert, ed. 1990. American Lobbyists Directory: A Guide to the More than 65,000 Registered Federal and State Lobbyists and the Businesses, Organizations, and Other Concerns They Represent. Detroit: Gale Research. Author Biography James Strickland is a PhD candidate in the Department of Political Science at the University of Michigan. His research interests include state politics, legislatures, and lobbying. 875695 research-article2019 APRXXX10.1177/1532673X19875695American Politics ResearchHolyoke and Cummins Article Interest Group and Political Party Influence on Growth in State Spending and Debt Thomas T. Holyoke1 American Politics Research 1­–23 © The Author(s) 2019 Article reuse guidelines: sagepub.com/journals-permissions https://doi.org/10.1177/1532673X19875695 DOI: 10.1177/1532673X19875695 journals.sagepub.com/home/apr and Jeff Cummins1 Abstract Does more lobbying by more interest groups, especially groups representing a state’s largest business sector, lead to greater spending and debt? Or does the blame really rest with state lawmakers and their political parties, which compete to attract and retain the allegiance of these powerful organized interests so they can win control of state government? We test this question with data on annual state budgets from 2006 to 2015, the number of interest groups in each state for those years, the size of the constituencies in different economic and social sectors these groups potentially represent, and the degree of competition between the political parties. Our results reveal that while there is a positive interest group effect on spending, the effect becomes negative as parties compete more for control of the state. As the gatekeepers, lawmakers and their parties, more than interest groups, are ultimately responsible for a state’s fiscal condition. Keywords interest groups, state budget, state debt, political parties, state legislatures The conventional wisdom is that government budgets increase in response to political demands, but demands from whom and under what circumstances is not always clear. The need for constituent support has often led elected 1California State University, Fresno, USA Corresponding Author: Thomas T. Holyoke, California State University, Fresno, 2225 East San Ramon, M/S MF19, Fresno, CA 93740-8029, USA. Email: tholyoke@csufresno.edu 2 American Politics Research 00(0) officials to promise new policies and financial benefits in return for enough votes to comfortably secure reelection, but not all constituents are equal when it comes to their political appeal. Those mobilized as interest groups may have greater value to legislators because they are more politically active and, at the behest of group leaders, direct their votes for or against candidates. Elected lawmakers, organized into parties competing to control state governments, therefore, have a powerful incentive to reach out to these organized constituencies, increasing the financial benefits they already enjoy, or promising whole new regimes of benefits and spending in exchange for votes. The unintended result is increased overall spending by the state and, consequently, greater debt due to interest group influence. Interest group influence over a government’s ability to deliver services has been studied, often in response to Olson’s (1982) dire prediction that too many demands from too many interests will retard the state’s ability to support economic growth, ultimately resulting in paralysis. Whether this happens because interest groups drove governments into debt by pressuring lawmakers to overspend, though, has at best received only modest attention. As elected officials make the actual spending decisions, we develop a model of group influence that is partially conditioned on political party competition, testing whether interest group advocacy directly, or indirectly through the parties, drives spending and debt with data on groups, parties, and budgets in the American states from 2006 to 2015. We find that more organized interests can drive up spending, but, surprisingly, their influence decreases as more parties compete to control state legislatures. This, we argue, is a significant change in the way we understand interest group influence. Interest groups may be players in spending politics, but lawmakers and their parties are the ones ultimately responsible for runaway budgets and crushing debts. The Politics of State Spending What drives government spending has drawn no shortage of scholarly attention. Studying state and national government spending trends, scholars have identified a number of factors influencing growth in public budgets. Often, their work came as contributions to debates about potential structural changes in national and state constitutions and governing institutions. For instance, in the 1980s, there was a flurry of research on the effects of line-item vetoes and debt limits on state spending because Congress was considering them as well (Abney & Lauth, 1997; Bails & Tieslau, 2000; Holtz-Eakin, 1988). When the term-limits debate heated up in the 1990s, researchers started exploring whether adopting them would reduce spending at both national and state Holyoke and Cummins 3 levels (Erler, 2007; Johnson & Crain, 2004; Payne, 1991; Reed, Schansberg, Wilbanks, & Zhu, 1998). Scholars also started investigating the effects of interstate competition when “race to the bottom” concerns emerged over changes in public safetynet spending, finding that some states were reducing benefits to deter potential recipients from relocating there from neighboring states (e.g., Bailey & Rom, 2004). While such concerns primarily arose in welfare policy, Bailey, Rom, & Taylor (2004) also found evidence of it in higher education spending. Surprisingly, though, Volden (2002) found that states were willing to increase their benefit levels if other states did so first. Further work also found that internal budgetary trade-offs matter at least as much as interstate competition, with increased spending in one policy area forcing reductions in others (Berry & Lowery, 1990; Garand & Hendrick, 1991; Nicholson-Crotty, Theobald, & Wood, 2006). Other research in both welfare and education policy attributes much of the rises and falls in spending to ideological clashes and political competition, especially between the two major parties for control of state governments (Barrilleaux, Holbrook, & Langer, 2002; Berkman & Plutzer, 2004; Dye, 1984; Hwang & Gray, 1991; Poterba, 1996). Recent research has found that both parties benefit at the polls from higher overall spending (Cummins & Holyoke, 2018), so recurring competition between them may drive it even higher, perhaps accounting for some of the cyclical pattern of occasional bursts of state spending uncovered by Jones et al. (2009). It is to these partisan and electoral explanations for increases in spending and debt (a consequence of overspending) that we hope to contribute. Targeting budget rewards toward key constituencies is a long-practiced means of gaining and retaining electoral support (Arnold, 1990; Fiorina, 1989). Yet, while bringing home money for all kinds of things from big dams (McCool, 1994) to higher education dollars (Balla, Lawrence, Maltzman, & Sigelman, 2002) can produce good results at the polls, the obligation on available budget dollars and levels of long-term debt can be significant. Even the demands of conservative, antitax interests, such as Tea Party organizations, can lead to greater debt. Unfortunately, for state treasuries, short-term needs to attract key voting constituencies often trump any long-term concerns lawmakers may have. California Democrats aggressively expand public assistance programs to satisfy progressive constituencies, whereas Kansas Republicans cut taxes across the board to appeal to conservatives, both jeopardizing their states’ financial well-being. That parties use spending to appeal to motivated constituencies suggests they are trying to attract the support of organized interest groups. What might be the consequences? Perhaps the closest scholars have come to looking for 4 American Politics Research 00(0) an interest group-driven effect on overall government performance has come from testing Mancur Olson’s (1982) argument that excessive interest group influence will lock up public sector investment in new industries and technologies to such a degree that a nation’s productive economic growth is paralyzed. Empirical support for Olson’s rather apocalyptic hypothesis in studies of democratic nations (e.g., Coates & Heckelman, 2003; Horgos & Zimmermann, 2009) and the American states (Ambrosius, 1989; Crain & Lee, 1999; Dye, 1980; Gray & Lowery, 1988) has been decidedly mixed (Heckelman, 2007), though most of this work focused on how states stimulate economic activity rather than spending and debt.1 However, with Bacot and Dawes’s (1996) and Newmark and Witko’s (2007) finding that interest group advocacy led to more spending on state environmental programs, we argue it is worth testing for a general group effect on spending and debt in the American states. Not only do state budgets and debts vary considerably, so do the size and diversity of state interest group populations (Holyoke, 2019; Nownes & DeAlejandro, 2009; Strickland, 2019). Interest Groups, Parties, Spending, and Debt We start by assuming that interest group lobbyists want to obtain public resources for their members, or otherwise enact policies providing them with benefits, regardless of whether these members are individuals, businesses, or other kinds of organizations. The benefits may be targeted appropriations, guaranteed spending formulas, tax breaks for conservative interests, or all of the above. Because lobbyists cannot directly manipulate public budgets themselves, they must convince state legislators to do it for them. What is the connection? The literature cited above emphasizes the influence of party competition on state policy as the elected officials embodying these parties use the tools available to them to advance their collective goal of dominating state government (Dye, 1980; Morehouse, 1981). Scholars have shown how important it is to a party’s electoral fortunes to gain and retain the loyalty of coalitions (e.g., Bawn et al., 2012; Brown, 1995; Herrnson, 2009), and others like Heaney (2010) argue that interest group members’ benefits rise and fall depending on whether their patron party is in power. Thus, group influence on government spending may run through party competition to win elections and control government.2 Specifically, legislators want to be reelected to office and serve in the majority party controlling the state legislature and governor’s office (Hershey, 2012; Lee, 2016). Organized interests represent specific constituencies that are politically motivated and likely to vote or provide the financial resources that party legislators believe is essential for achieving reelection and majority Holyoke and Cummins 5 control (Hansen, 1991). Legislators, therefore, offer tax and spending benefits in exchange for the votes and campaign support of group members as they form electoral coalitions, or at least promise it if they are in the minority party hoping to become the majority. Interest group influence on state spending and debt is, therefore, conditional on receiving party support. To identify testable hypotheses, we consider four scenarios in Figure 1. Interest groups tend to focus only on policy areas relevant to their members, so each scenario (which we admit cannot fully represent every policy area) regards the dispositions of several interest groups relative to the two major parties in a policy area.3 There is no meaningful party competition in Scenario 1; party P1 has a firm majority and, consequently, controls all spending. Groups IG1 and IG2 are close to the majority party and likely enjoy some spending and tax benefits in return for their electoral support. However, because P1 is in control, and because these two groups’ members are unlikely to give their leaders the significant flexibility needed to align with right-leaning, but powerless, P2, the ruling party does not need to spend heavily to keep the loyalty of IG1 and IG2. If they are the only groups lobbying, there is no reason to expect increased spending. Yet, there are two other groups, IG3 and IG4, which are center-right, suggesting some diversity among the positions of all groups lobbying this policy. If P2 had a chance of becoming the majority, then a bidding war might occur, perhaps resulting in IG3 and IG4 giving their loyalties to the aggressive minority party. Yet, that might not increase spending because regardless of which party won, only two groups would be rewarded with spending benefits. Although P2 is unlikely to win in this scenario, P1 still wishes to strengthen its legislative majority and may promise more spending benefits to attract IG3 and IG4, whose loyalties are up for grabs because P2 cannot make credible promises. This is consistent with Morehouse’s (1981) argument that when interest groups are weak vis-à-vis parties, they will try to accommodate themselves to the ruling party. It also suggests that greater diversity among more interest groups may lead to greater spending and debt. We cannot test all of these spatial implications, but we can state our first hypothesis: Hypothesis 1 (H1): Absent significant party competition, greater policy diversity among interest groups leads to greater state spending and, consequently, debt. The effect of party competition on interest group influence and spending emerges in Scenarios 2 and 3 in Figure 1. Unlike Scenario 1 where P1 is the only game in town, here both parties compete for control of state government. The collective positions of the four interest groups are the same as in Scenario 6 American Politics Research 00(0) Scenario 1: Ruling party holds the loyalty of IG1 and IG2, spends to attract IG3 and IG4, so spending increases P1 P2 IG1 IG2 IG3 IG4 Groups attracted by P1’s spending promises Scenario 2: Parties compete for government control, so, despite promises, fewer groups receive spending benefits P1 P2 IG1 IG3 IG2 IG4 Scenario 3: Parties compete and more group diversity, so spending demands rise, but party competition limits it P1 IG1 P2 IG2 IG3 IG5 IG4 IG6 Scenario 4: Competing parties spend to attract an interest group dominating the state’s major employing industry P1 IG1 Party spending promises IG2 P2 IG3 IG4 Groups unable to get attention may shift Figure 1. Four scenarios of interest group and party interaction. 1, but now minority party P2 can make credible spending promises in exchange for group support. Majority P1 may promise benefits to attract IG3 and IG4, and the minority may promise more to attract Groups 1 and 2, but if both parties have realistic chances to become, or remain, the majority, IG1 and IG2, are likely to remain loyal to their proximate ally, P1, and the other two groups will Holyoke and Cummins 7 be loyal to P2. As only one party can win, only two groups receive benefits, so spending is less likely to rise in Scenario 2 than in 1. The parties are still competing in Scenario 3, but now we add two more interest groups, IG5 and IG6, which increases the overall policy diversity of all of the organizations seeking benefits. Collectively, this weakens their hands vis-à-vis the parties. P1 likely holds the loyalty of IG1 and IG2 with only a little extra spending over Scenario 2 as it is highly unlikely the groups’ members would let their leaders support P2. Indeed, given their proximity, they might support P1 regardless of any spending promises. The same is true for the minority party and IG5 and IG6. The only real targets the parties might compete for with spending promises are IG3 and IG4. The result is that more interest groups supporting a greater diversity of positions on policy depresses their collective influence on spending when the parties compete and might even lead to less spending and debt. So, a conditional hypothesis: Hypothesis 2 (H2): More interest group policy diversity lessens their influence on spending and debt (in H1) when there is greater party competition. Interest group competition, however, is complicated. Not only is it often not zero-sum, Gray and Lowery (1996) found that groups also compete to attract members, or at least contributors. If the groups in either Scenario 2 or 3 are competing for a small, fixed set of potential members, group leaders must promise to deliver more spending benefits to convince them to overcome collective action barriers and join. This, in turn, may mean trying to attract the support of the dominant party (as in Scenario 1) or of the other party if it has a chance to become the majority. Yet, the more an interest group strives to recruit members, but the fewer there are to compete for, the less attractive any single group is to the parties because it will have fewer members whose support they can offer. So, in the aggregate, the more interest groups there are competing for members, the less spending the parties will promise for their support. Furthermore, when the parties compete, as in Scenarios 2 and 3, they have less incentive to promise more spending to groups fighting for members. So, Hypothesis 3 (H3): Greater competition between more interest groups for members depresses their influence over spending and debt, especially when the parties are competing for government control. This also suggests another hypothesis, that if one interest group dominates the largest population of members in a policy area, or is a state’s biggest 8 American Politics Research 00(0) employer, it will be the most tempting target for the spending promises of each party, especially if both parties are competing to be the majority. In Figure 1’s Scenario 4, IG3 has a much larger membership than the other groups. Not only does this make it significantly more attractive to the two competing parties, who will make enormous spending commitments to attract its voter members, and perhaps even shift their preferred policy positions closer to IG3, the other interest groups may also shift their positions toward IG3 to be attractive to the parties (though we do not test for this). So our final hypothesis is as follows: Hypothesis 4 (H4): The fewer interest groups there are representing a state’s largest industry, the greater the increase in spending and debt, which will be even greater if the parties are competing for control. Research Design We test these hypotheses with data on public spending, interest group populations, party competition, and long-term debt in the American states. Our first dependent variable is the percentage change in annual state spending from 2003 to 2015, meaning the amount of money state legislatures appropriate each year as part of their annual budget process or otherwise spend through automatic funding formulas.4 These data come from the Fiscal Survey of the States published annually by the National Association of State Budget Officers.5 Our second dependent variable is each state’s long-term debt for the same years, which comes from the U.S. Census Bureau. After standardizing both fiscal variables to year 2000 dollar values, we find that California spent the most at more than US$84 billion in 2007, and Wyoming the least at US$417 million in 2004. To get a sense of trends in spending and debt over this time-period, which encompassed the Great Recession, we convert both variables into z scores to make them comparable, average them for all 50 states, and graph the results in Figure 2. The trend shows spending increasing as vibrant economies put more money into state coffers, then a significant dip when the recession hit the public sector in 2009 (lagging behind the private sector slump in 2008), and finally a recovery. We see a similar trend in debt, which, unsurprisingly, continued to increase after spending fell as states tried to cope with demands, only leveling off in 2011. California and New York achieved the dubious honor of having the most debt, whereas Wyoming carried the least. Rather than use z scores in the analysis below, we divide spending and debt by state population for per capita spending and debt measures, and then calculate the percentage change from 1 year to the next.6 Holyoke and Cummins 9 Figure 2. Trends in average state spending, state debt, and the number of lobbying organizations (converted into z scores). We obtained interest group data from the National Institute for Money in State Politics for 2006 through 2015, a source used in other research (e.g., Gray, Cluverius, Harden, Shor, & Lowery, 2015; Witko & Newmark, 2005).7 These data include traditional member-based groups, as well as businesses and nonprofits lobbying state government, representing different sections of business and society.8 To get a sense as to whether there is a link between interest group numbers and state spending, we add the annual average number of lobbying organizations (as z scores) in Figure 2. For several years, the number of groups appears to track spending, though, interestingly, growth in group numbers lagged state spending as the recession passed. The correlation of spending and organizations is .79 (p < .005).9 The relationship with state debt is less clear as debt increased at the very time the number of groups diminished, though the correlation of .76 (p < .005) suggests a link. Measuring interest group policy diversity requires us to assume there are differences in the policy outcomes preferred by the interest groups lobbying each policy area. As we cannot measure the difference of one group’s preference from another, we simply assume there is a single, ordinal difference between every group lobbying in a well-defined area of policy. More groups, therefore, mean more differences in position diversity, seen in Figure 1’s Scenario 3. We start constructing our measure by identifying policy areas using the system developed by the U.S. Department of Commerce, called the 10 American Politics Research 00(0) North American Industrial Classification System (NAICS; used in interest group research by W. L. Hansen, Mitchell, & Drope, 2005). As detailed in our Supplemental Appendix, we sort groups into NAICS policy areas such as “agriculture,” “communication,” “entertainment,” “energy and natural resources,” “finance, insurance, and real estate,” “leisure,” “general business including manufacturing,” “education,” “social service,” “health care,” and “transportation.”10 Health policy tends to have the most groups, and entertainment policy the least. To create an index of state group policy diversity each year, we use this equation where g refers to groups indexed by j in policy area i where the set of all issue areas is represented as P:  2  gj ∑   2 i =1   P   - g j     2      i P The equation calculates the number of differences between an interest group and every other group j in policy sector i.11 It then sums the results and divides by all policy areas for a single, annual state score. The distortion from squaring g is reduced by taking the square-root. Because we predicted that the interest group effect is conditioned on the level of party competition for government control, we capture this using the 4-year-folded Ranney Index of state party competition.12 H3 regards group competition for members. As our interest group data are already organized by the NAICS codes, we also obtained Occupational Employment Statistics data from the U.S. Department of Labor, which lists the number of people employed in dozens of tightly defined categories for every state and every year. We aggregated these employment categories into larger, more general groupings matching the NAICS (see the Supplemental Appendix for details).13 Then, we divided the total number of groups in each NAICS category for that state and that year by the number of people employed in the corresponding labor category. Larger ratio values mean more groups are competing for fewer potential members, which H3 predicts to have a negative effect on spending and debt, especially when party competition is more intense. We predict in H4 that interest groups dominating a state’s major industry are especially influential on spending and debt. We, therefore, found the total number of people employed in each industry for each year using the occupation codes and identified the one employing the most. Unsurprisingly, in 70% of our 500 cases, it turned out to be “general business including manufacturing,” with Holyoke and Cummins 11 “agriculture” in 30% of cases that primarily included rural states such as Montana, West Virginia, and the D...
Purchase answer to see full attachment