Why hospitals hire tobacco lobbyists: conflicts of interest among lobbyists’ clients

Abstract

Lobbyists sometimes represent clients with seemingly adversarial policy interests. We seek to explain the occurrence of such ostensible conflicts of interest. In hiring lobbyists, interests encounter a tradeoff between access and agency. Although some lobbyists promise access to lawmakers, they may not lobby as contracted. Interests hire seemingly conflicted lobbyists more often when access is costlier and reputational risks are smaller. We examine the hiring of tobacco lobbyists by health interests, given the possibilities for shirking and reputational damage. We find that institutions such as hospitals hire tobacco lobbyists regularly and more often than membership-based health groups. Intergroup competition for access and lobby laws, especially anti-conflict laws, affect the use of tobacco lobbyists independent of rates of multi-client lobbying. Conflicts are more common today than ever but interests can protect themselves somewhat from opportunistic agents. Our findings also suggest that reforms can improve the representation of client interests.

In the context of lobbying, ostensible conflicts of interest occur when individual lobbyists represent clients with seemingly adversarial or contradictory policy interests. For example, health-related groups may hire lobbyists who also represent tobacco interests. Such conflicts pose risks to clients since lobbyists may fail to represent all clients faithfully and clients may suffer reputational losses. In this study, we propose that not all ostensible conflicts pose harm and that, rather, they provide insight into how organized interests trade agency for access while hiring lobbyists. Organized interests must balance their individual agency and reputations with the abilities of various lobbyists to achieve access. These decisions occur across political systems with various lobby laws that can affect the balance of agency with access. Upon examining how often healthcare interests active in the American states hire tobacco lobbyists, we find evidence that both group- and system-level factors influence how often groups trade agency for access.

From our perspective, interest groups desire both to preserve agency over how they are represented and also to influence legislative outcomes. In acting as agents and achieving access (which is required for influence), lobbyists may undermine the agency of their clients. Although particular lobbyists may ensure that clients achieve access, the lobbyists may shirk. For example, lobbyists may not lobby as often as contracted, lobby against the interests of their clients, or prioritize the interests of better-paying clients (see Lowery and Marchetti 2012). How do organized interests seeking political influence adapt to such risks? We propose that groups preserve agency over their representation by hiring lobbyists carefully. In particular, although some lobbyists may be able to achieve access, they also present greater risks of harm (shirking, reputational losses) to their clients’ interests. We propose that lobbyists who represent clients with seemingly adversarial interests pose greater risks of harm than those who do not and that different kinds of clients have different levels of tolerance for such conflicts. Ostensible conflicts may not result in harm, but clients reduce the possibility of harm by not hiring lobbyists who represent traditional opponents. Our theory suggests that groups that are more vulnerable to reputational losses are more likely to avoid such conflicts and that more intense competition for access leads all groups to trade agency for access (and thereby hire ostensibly conflicted lobbyists) more often. Also, some lobby laws help improve clients’ abilities to evaluate levels of risk associated with lobbyists.

To test our expectations, we examine how often health-related interests hire tobacco lobbyists. Tobacco lobbyists are paid well, and their clients’ economic and political activities are a subject of public concern (Givel and Glantz 2001; Rotman et al. 2022). As a result, it is particularly risky for health interests to hire tobacco lobbyists since these lobbyists may prioritize the representation of their better-paying clients, and because there may be reputational losses associated with hiring tobacco lobbyists. Moreover, given that the use of tobacco-based products is associated with poor health outcomes, the hiring of tobacco lobbyists by health interests may lead to actual conflicts or when lobbyists represent clients with opposite preferences. Yet, health interests have a long history of hiring tobacco lobbyists (Goldstein and Bearman 1996). We build a dataset consisting of more than 9,000 health interests that are registered to lobby state legislators in the United States. The interests registered to lobby in either 1989 or 2009. Generally, in accordance with our expectations, we find that health interests with members or donors united by non-economic causes were less likely to hire tobacco lobbyists than health interests who either lacked members or whose members joined primarily for economic or professional gain. Additional test results based on data from 2018 show that these trends are robust to the inclusion of controls for lobbyist pay levels, a potentially important confounder. We also find that health groups generally hired more tobacco lobbyists where there were more interests per legislator and that transparency and anti-conflict laws were correlated with such hires in opposite directions. These results are robust to the inclusion of a control for the total numbers of tobacco interests registered to lobby in each state.

Ultimately, our findings have implications for representation and reform. Lobbyists are the intermediaries who link organized interests with policymakers, so their efforts matter for how well clients are represented. Conflicts raise questions about the quality of representation that clients receive and which interest groups are most represented and influential in legislatures. Shirking by lobbyists, an example of harm to client interests, can affect the information that lawmakers receive from interests (as in Hall and Deardorff 2006). Moreover, conflicted lobbyists may choose to prioritize the interests of better-paying clients and, therefore, magnify existing inequalities in the influence that various interests exercise (Holyoke 2016; Schlozman and Tierney 1986). Given that more lobbyists now represent multiple clients than ever (Drutman 2015; Strickland and Crosson 2023), there is more potential for conflicts and harm to occur. Understanding the circumstances under which interests hire conflicted lobbyists can help reformers identify means for improving representation. Our findings suggest that large assemblies and anti-conflict laws minimize the occurrence of conflicts.

Despite its wide-ranging implications, our study provides novel contributions to scholars’ knowledge of interest groups. There are few studies on how lobbyists may shirk (see Holyoke 2016; Kersh 2000), and few studies suggest that groups work to preempt shirking (see Strickland 2020). Practically, our study is the first to explain the emergence of conflicts among lobbyists’ clients. Whereas Goldstein and Bearman (1996) measured how often health firms hired tobacco lobbyists, they did not attempt to explain variation in conflicts across group types or political systems. Our theory is novel, and our dataset is the first to examine conflicts over multiple decades. Finally, all the findings presented in our study are new: scholars have no other statistical evidence suggesting that purpose-driven groups avoid conflicts, that intergroup competition produces more conflicts, or that anti-conflict laws actually reduce the numbers of conflicts. In providing these contributions, our study allows scholars to speak on a topic that has received coverage in popular press outlets (see Ingersoll 2017; Lyden 2015). Moreover, the ongoing rise in multi-client advocacy ensures that the findings presented in this study will remain relevant for years to come.

Conflicts as sources of risk

Both lawyers and lobbyists are paid to represent others, and both have incentives to disserve their clients on occasion (see McMunigal 2001). Whereas lawyers often work in public and may be observed by their clients, however, lobbyists work in private and away from their clients. Moreover, whereas lawyers are required to represent their clients faithfully, and are prevented from representing clients with conflicts, by the American Bar Association, there are no professional associations that can punish lobbyists for not representing their clients faithfully. McMunigal argues that it is impossible to eliminate all risks of harm in the practice of the law. Therefore, clients must consider acceptable levels of risk and make accordingly strategic decisions over whom to hire. For example, joint representation of criminal defendants poses a risk that the attorney will represent better only one defendant, but the benefit of such an arrangement may lie in clients presenting a “unified front, saving money, and increasing access to a defendant’s counsel of choice” (69). Building on this idea, we believe that it is impossible for the clients of lobbyists to remove all risk of harm from their relationships with the agents. Rather, clients may manage risk by hiring particular agents. Ostensible conflicts of interest among the clients of registered lobbyists are useful for learning about how clients manage risk by trading agency for access.

Both scholarly and popular accounts reveal instances when lobbyists harmed their clients’ interests because of conflicts of interest. Not all conflicts are the same, and the conflicts that have received the most scholarly and popular attention are those that involve the personal interests of lobbyists. These conflicts may occur when lobbyists represent anyone. For example, lobbyists may covertly advocate for legislation that harms their clients’ interests so that they may be paid to continue to lobby on behalf of the client (Gray and Lowery 1996a). This practice received substantial press coverage when Jack Abramoff, a prominent lobbyist active in Congress, was convicted of wire fraud and conspiracy (Lowery and Marchetti 2012). A more subtle form of this conflict occurs when lobbyists seek to preserve their relationships with lawmakers at the expense of their clients’ causes. Lobbyists commonly work to build professional relationships with lawmakers so as to achieve access on a regular basis (Ainsworth 1997), and they sometimes soften or reverse the stances of their clients so as to preserve their relationships with the officials (Holyoke 2016, 2022; Kersh 2000). There is always the risk of personal conflict whenever lobbyists are tasked with representing others.

In contrast, inter-client conflicts occur only in the context of multi-client lobbying. These conflicts occur when lobbyists represent sets of clients with seemingly adversarial interests. Aside from questions of how well lobbyists practically represent opposite interests, lobby clients may prefer their lobbyists not to represent groups with other policy preferences due to financial considerations. Clients may be concerned that adversarial interests pay their lobbyists more than they do, and therefore undermine their representation (see Givel and Glantz 2001; Kersh 2000). Conflicts between clients may also involve reputational harm. On occasion, the clients of lobbyists who represent seemingly adversarial interests are interviewed by journalists, and these clients sometimes prefer their names not to be associated with other kinds of interests (see Ingersoll 2017; Lyden 2015).

Explaining the emergence of conflicts

We propose that interest groups’ acceptance of risk may be predicted somewhat. In our view, groups generally face a tradeoff between preserving their agency and achieving access to officials. Organizations may maximize their agency by ensuring that their lobbyists (agents) faithfully represent them before officials. A means for ensuring agency is to hire a single-client, in-house lobbyist who personally values the client’s mission. Although hiring such a lobbyist would ensure that no inter-group conflicts of interest occur, that particular choice may entail a loss of access. In contrast, lobbyists with personal connections to lawmakers often advertise their connections and, as a result, represent various interests on a contractual basis (see Drutman 2015; Hirsch et al. 2023). The use of these lobbyists, unlike with single-client agents, may improve access but makes client organizations vulnerable to inter-client conflicts. If organizations looking to lobby indeed face a tradeoff between agency and access, then we expect to see particular patterns in the numbers of conflicts among different kinds of organizations active in different states. Importantly, although inter-group conflicts are possible only in the context of multi-client lobbying, we expect to find evidence for trade-offs even when holding constant the types of lobbyists organizations hire (or even when examining only the lobbyists of organizations that hired at least one multi-client lobbyist).

The nature of a group, which is often related to its mission, may matter for its propensity to hire a lobbyist who represents an adversarial interest. Numerous organizations that lobby consist of individual members who pay dues or provide other resources voluntarily. Moreover, whereas some membership-based organizations are formed for economic, occupational, or professional purposes (e.g., labor unions and trade associations), others advocate for public goods and are likewise known as public interest groups (Berry 1977). Prominent examples include the American Civil Liberties Union or Sierra Club. Importantly, whereas economic membership groups attract members due to the various material and solidary benefits; they provide members, supporters of public interest groups derive purposive or ideological benefits from participating (Clark and Wilson 1961; Cook 1984). Organizations that do not rely upon members at all are often private businesses or other institutions with narrow interests (Olson 1965).

Differences in sources of support have multiple implications for the emergence of conflicts or hiring of risky lobbyists. Unlike individual businesses or institutions such as hospitals and universities, membership-based organizations do not sell or produce private goods and services but rather rely on member or foundation support for their continued existence (see Gray and Lowery 2001; Walker 1983). This suggests that institutions not supported by members may not suffer the same reputational losses from conflicts as other organizations. In the case of organizations that depend on members, negative publicity related to an advocate may entail a loss of member or donor support (due to reduced confidence that member dues or donations are used effectively). Not all membership-based groups face the same prospect for losses, however. In the case of labor unions or trade associations, members may continue to support these organizations since membership may be mandatory (such as in closed shops or with bar associations) or due to the occupational benefits that these groups provide to members, which may be material or solidary. In the face of negative publicity regarding a conflict, purely purpose-driven interests cannot rely on members continuing to give. They must overcome the free-rider problem on an ongoing basis (Bevan 2013). ¹

There is anecdotal evidence that sources of support for health groups matter in their hiring decisions. When leaders of different health interests were asked about their lobbyists’ representation of tobacco interests, responses varied. Particularly, when leaders of a state licensing board for doctors, a hospital association, and an obstetrics association were all asked about their lobbyists’ work for tobacco interests, all the leaders indicated that the conflicts were of little concern (Ingersoll 2017). When the leader of the Rainbow Health Initiative in Minneapolis, a healthcare advocacy group that solicits donations, however, was informed that her lobbyist also represented a tobacco company, she expressed concern and promptly cancelled the lobbyist’s contract. The leader indicated that the “fact that [the lobbyist] represents a tobacco company, now that [she was] aware of it… [would] be problematic…” (Ingersoll 2017). In addition to providing health services, the Initiative engaged in fundraising activities and received funding as a result of the 1998 Tobacco Master Settlement Agreement. Upon being interviewed by journalists, other health groups dissolved their ties with tobacco lobbyists, including hospitals and pharmacies on occasion (see Gehrke 2009; Pallarito 1993; Wahba 2015).

More generally, if groups are choosy over whom to hire due to possible reputational losses, then their sources of support should be correlated with the numbers of conflicted lobbyists they hire. In the case of businesses and institutions, or associations of these entities, such as chambers of commerce, conflicts among lobbyists represent little existential threat to the organizations. Such conflicts pose more serious threats to organizations that recruit autonomous individuals as members or that raise funds (than to other organizations). Whereas those with economic or occupational missions may rely on members to join or contribute for various reasons, public interests are more vulnerable than other interests to possible reputational losses since their members join or donate for purposive benefit. After all, public interests compete for limited numbers of members and donors (Gray and Lowery 1996b).

H ₁: Organizations without members hire risky lobbyists (i.e., those with potential conflicts) more often than membership groups, ceteris paribus.

H ₂: Among membership groups, economic groups hire risky lobbyists more often than non-economic groups, ceteris paribus.

Institutional and legal contexts

We also propose that the institutional environments in which organized interests seek influence may matter for the hiring of lobbyists with conflicts, independent of variations in sources of support for groups. Recall that organized interests seek influence over policy via lobbyists. Personal access or meeting time is required for lobbyists to convey information to lawmakers and, ultimately, influence the content of laws (Wright 1996). Unfortunately for most interests, lawmakers cannot ensure equal access to all lobbyists, and lobbyists accordingly compete by building relationships or screening clients for lawmakers (see Grose et al. 2022; Hirsch et al. 2023). As the numbers of organized interests have increased over time but the numbers of legislators have remained steady, various lobbyists have come to acquire numerous clients by advertising their personal connections to legislators (Drutman 2015). In our perspective, when there are more interests competing for the limited attention of legislators, the interests are more prone to trade the agency that single-client lobbyists deliver for the access that multi-client advocates provide. Even when controlling for numbers of multi-client lobbyists hired or alternatively excluding groups that hire only single-client advocates, however, we expect to find that intergroup competition spurs groups to neglect agency in favor of access. More competition or when there are more interests per legislator improves the abilities of lobbyists with connections to attract clients, including tobacco lobbyists with connections. In other words, interests hire lobbyists with ostensible conflicts more often under conditions of increased competition for access to lawmakers.

Little research links group competition with legislature size but the theoretical accounts provide arguments relevant to our hypothesis. Powell (2012) finds that legislators raise more campaign funds on average in states with more registered lobbyists and fewer legislators, and Kattelman (2015) argues that legislators and staff persons are themselves points of access for interested groups. Apart from theoretical arguments, there is some anecdotal evidence that competition among groups has led to potential conflicts. When asked about his tobacco lobbyist, a director for a large health insurance company indicated that “[w]e simply hire the [lobbyists] we feel that are most effective in carrying our message to the legislative bodies” (Pallarito 1993).

H ₃: Interest groups hire risky lobbyists more often in political systems where there is greater competition for personal access to lawmakers, ceteris paribus.

The legal environments in which organized interests seek influence may also matter for the hiring of lobbyists with conflicts. Two kinds of laws should matter specifically for conflicts among registered lobbyists. The first kind consists of laws that require lobbyists to disclose potential conflicts to clients before providing representation services. ² Such laws are uncommon but may be found in the United States. The state of Texas, for example, requires lobbyists to inform clients about potential conflicts and report conflicts to state authorities. Lobbyists report dozens of actual or potential (out of caution) conflicts each year (Stiles 2010). Although Pennsylvania also requires lobbyists to report conflicts to state authorities, most states with anti-conflict laws merely require lobbyists simply to notify clients and receive written authorization for continued services. As of 2023, lobby laws in seven states contain anti-conflict provisions: Colorado, Idaho, New Jersey, Pennsylvania, Texas, Utah, and Washington. Whereas lobby laws approved by voters in Idaho and Washington in the early 1970s contained these provisions, Colorado is the latest state to implement such a law. All the laws allow lobbyists to represent adversarial interests provided that the interests are notified first. Anti-conflict laws may be correlated negatively with inter-client conflicts given that clients may wish to minimize the potential for actual conflicts (i.e., when lobbyists do indeed express conflicting policy preferences to lawmakers).

Another aspect of the legal environments in which groups lobby is reporting requirements. In the United States and elsewhere, lobbyists are required to report the contents of their conversations with officials, amounts of money spent on lobbying (including compensation), contributions to candidates, or even gifts to officials (Newmark 2017). Such requirements may affect the emergence of ostensible conflicts by giving clients greater abilities to monitor their agents for shirking or actual conflicts. Organized interests active in Europe regularly make use of lobby registers and other sources of lobby information (Crepaz 2020), and there is no reason to believe that interests active elsewhere do not examine lobby registers and reports equally or possibly more often. Although lobby registers identify which clients each lobbyist represents (and thereby provides at least a blunt level of protection from actual conflicts), it is the supplemental reporting requirements that may help organized interests to evaluate the relationships that exist between their lobbyists and other clients. Compensation and expense reports, for example, may be used to determine how much of a lobbyist’s income or expenses are related to any particular client. As a result, conflicts may be more numerous in political systems with more reporting requirements than elsewhere. ³

H ₄: Interest groups hire risky lobbyists less often in political systems where there are anti-conflict laws for lobbyists, ceteris paribus.

H ₅: Interest groups hire risky lobbyists more often in political systems where there are greater reporting requirements for lobbyists, ceteris paribus.

We note that lobbyists themselves may avoid conflicts. Ethical lobbyists may vet clients to minimize the risk of conflicts occurring. One lobbyist in Colorado, for example, refused to take on clients whose interests could conflict with those of his existing clients (Rosenthal 1993). Lobbyists may even form professional associations that require members to take oaths of ethical conduct. Although there are several professional associations for lobbyists in the United States, membership is not mandatory for any lobbyists (see Bernstein 1991). Generally, we have no reason to believe that lobbyists in some political systems are intrinsically more or less likely to avoid conflicts than lobbyists elsewhere, absent the effects of local conflict and reporting laws. Also, as will be demonstrated, the effects of state context may be held constant with the use of statistical controls.

Venues and measurement

To find evidence for our expectations, we choose to examine trends in lobbying in the American states. There is significant variation in group mobilization and political environments among the states, and such variation is required for testing our hypotheses empirically. Importantly, the states also require lobbyists to register and indicate the identities of their clients. The earliest registration requirements emerged during the Progressive Era and gradually spread to other states. By 1976, all states required lobbyists to register (Strickland 2021). Other venues in which lobbying occurs, such as Congress, executive-branch agencies, or even governments beyond the United States, do not provide sufficient variation in explanatory variables or information about lobby activities sufficient for identifying ostensible conflicts (see Chari et al. 2019).

We use lobby registers from the states to detect ostensible (i.e., not actual) conflicts or when interests hire lobbyists who pose a greater risk of harm. We are interested in detecting the trading of agency for access by interests generally, and not actual instances in which lobbyists represented opposing sides. When registering in the states, lobbyists identify their clients in registration documents that are legally binding (i.e., lobbyists may be punished or sanctioned for providing false or inaccurate information, including for lobbying for any client without registering to represent the client). Even without sanctions, lobbyists have incentives to comply with registration and reporting laws (see Crepaz 2020), and registration costs are typically low. The lobby registers reveal, in a structured manner, the entities that each lobbyist represents during each legislative session. We believe that these qualities make lobby registers a superior means of detecting ostensible conflicts over interviews with lobbyists who may seek to obfuscate their relationships or interviews with clients who may be unaware of conflicts. ⁴

To detect instances in which interest groups hired lobbyists who posed a greater risk of harm to their interests, we examine health-related interests that hired at least one lobbyist who also represented a tobacco group during a legislative session. These advocates are riskier hires for health interests than other possible lobbyists for three reasons. Since the agents are paid well by tobacco-related clients, they may prioritize the interests of these clients (Givel and Glantz 2001). Second, because tobacco lobby efforts are a matter of public interest, being associated with tobacco interests may undermine the reputation of a health interest (Rotman et al. 2022). Third, since the use of tobacco products is associated with poor health outcomes and tax revenues or punitive judgments related to tobacco products are often earmarked specifically for medical or preventative programs, actual conflicts involving opposite policy stances may emerge among the clients of lobbyists who represent both health and tobacco interests. Yet, numerous healthcare firms already employ tobacco lobbyists, so there is presumable variance in hiring across types of interests and states. For example, Goldstein and Bearman (1996) examined lists of registered lobbyists from the states and found that there were 303 healthcare organizations that had hired lobbyists with tobacco interests. In fact, among the 450 individuals registered to represent tobacco interests, 220 were also registered to represent a healthcare organization. The number of healthcare firms that hired tobacco lobbyists ranged in value from 0 in several states to 25 in Michigan. Moreover, health lobby efforts in the states have received substantial academic attention (e.g., Gray et al. 2013; Lowery and Gray 2007), so their efforts in particular are of interest to a wide audience.

Before testing our hypotheses using statistical regressions, we first produce descriptive statistics that show how often healthcare interests hire tobacco lobbyists. These statistics help to demonstrate that there is considerable variation both over time and across states in the hiring of lobbyists with ostensible conflicts. We generate these statistics using both lobby registers and secondary sources. Since the institutional and legal environments in which lobbyists work have changed over time, we collected lobby lists spanning multiple decades. Lists from around 1973 and 1989 were gathered primarily from Reitman and Bettelheim (1973), Marquis Academic Media (1975), and Wilson (1990), which are all compilations of lobby registration lists published by state authorities. Lists from 2009 were gathered primarily from the National Institute on Money in Politics, a non-partisan research organization that collects lobby and campaign finance data from the national and state governments. Where possible, original lists from archives or libraries were consulted and prioritized.

From the lists, we identified all healthcare interest groups that are registered to lobby. Following Goldstein and Bearman (1996), healthcare firms include individual hospitals and hospital associations; heart, lung, and cancer societies; state health departments; medical societies and licensure boards; healthcare occupational associations; pharmaceutical companies and pharmacies; chiropractic, dental, optometric, and psychiatry interests; ambulance associations; and healthcare and health insurance corporations. Some organizations appeared in more than one state or year. For every healthcare group, we used the lists to determine both how many lobbyists it hired and how many of those lobbyists also represented a tobacco interest. These interests included the major companies (e.g., Altria, Lorillard, Philip Morris, R.J. Reynolds, Swisher, Swedish Match); tobacco advocacy groups such as the Center for Indoor Air Research, National Smokers Alliance, and Tobacco Institute; and various tobacco retailers, wholesalers, or associations (e.g., Cigar Association). The only pro-smoker grassroots organization to appear in the lists, Smokers’ Rights of New Mexico, was not coded as a tobacco interest group since there is no evidence it received support from the tobacco industry (see Lait 1989). Our dataset is the most comprehensive compiled in terms of health and tobacco lobbying, and related ostensible conflicts.

Table 1 reports the total number of states for each wave of observations, and the total numbers of healthcare and tobacco lobbyists and interest groups registered in those states. Some lobbyists or interests may have appeared in multiple states. The numbers should be interpreted as measures of total volume in lobbying or group mobilization. The table also provides the total number of instances in which a healthcare firm hired a lobbyist that also represented a tobacco firm, in the seventh column. To provide estimates for 1994, we collected information from Goldstein and Bearman (1996) and Lowery et al. (2015). Comparing the figures from 1989 and 1994 indicates that our coding method captured a comparable number of conflicts as the method of Goldstein and Bearman. From the table, since the early 1970s, both health and tobacco interests have increased steadily in number. Roughly 2 percent of the healthcare interests active in 49 states in the early 1970s hired one or more tobacco lobbyists. This figure increased significantly throughout the 1980s. By the end of that decade, nearly 10 percent of healthcare interests had hired tobacco lobbyists. Twenty years later, roughly 13 percent of healthcare interests had hired tobacco lobbyists.

Table 1. Healthcare and tobacco mobilization (state data)

Note: Some interests and lobbyists appeared in multiple states. Totals from 1994 taken from Goldstein and Bearman (1996) and Lowery et al. (2015).

There is significant interstate variation in numbers of health firms with conflicts within each wave of observations, and this variation is not illustrated in Table 1. Rather, to provide some insight into this variation, one can compare statistics within waves. The percentage of health firms in each state that hired at least one tobacco lobbyist in 1973 ranged from 0 (in 40 states) to 14.3 (in Nevada). By 2009, these percentages ranged from 0 (in only 3 states) to 35.3 (in Mississippi). These statistics do not account for differences in the kinds of health groups active in each state or differences in institutions and laws.

Explanatory variables

We are interested in determining if several explanatory variables or factors are correlated with numbers of tobacco lobbyists hired by healthcare interest groups. To test our hypotheses, we need to measure the explanatory variables that should be correlated with conflicts. To test our hypothesis regarding sources of group support, we built a dataset consisting of individual health interests active in each state. We classified each of the interests into one of three different categories. The first consists of individual businesses or institutions, or associations of institutions (as in Gray and Lowery 2001). Common organizations in this category include hospitals, insurance companies, pharmacies, pharmaceutical companies, and state medical or licensure boards. The second category consists of groups whose members or donors are autonomous individuals (as in Gerber 1999) who are united by some economic or professional purpose. Common organizations in this category include various labor unions and associations for particular occupations (e.g., anesthesiologists, chiropractors, and nurses). Finally, our third category consists of organizations whose members or donors are autonomous individuals not united in purpose by an economic or professional interest. Common entities in this category include the American Cancer Society, American Heart Association, and Arc of the United States. Eleven healthcare organizations were excluded from our analyses since their members or supporters included both institutions and autonomous individuals or because the nature of their membership could not be established.

To measure intergroup competition for personal access, we turn to the ratio of the total number of organized interests with lobbyists in a state to the number of incumbent lawmakers. The states vary drastically in terms of numbers of organized interests, with several states featuring a few hundred groups that lobby per year and others featuring several thousand (Gray and Lowery 1995). Numbers of legislators, however, do not vary nearly as much. Legislatures range in size from 49 senators in Nebraska’s Unicameral to 424 members in the New Hampshire legislature (Squire and Hamm 2005). The lack of variance in legislator totals suggests that there is significant variance in the average numbers of groups per legislator across the states. Indeed, during years in our last wave of observations, average numbers of groups per legislator ranged from 1.2 in New Hampshire to 23.6 in Florida.

Regarding the legal environments in which organized interests operate, we employ two measures. The first measure consists of a dichotomous indicator for whether a state had an anti-conflict law in effect. As described earlier, such laws require lobbyists to notify clients of conflicts and receive written authorization to provide continued representation. These laws were found in annual editions of the Lobbying, PACs, and Campaign Finance: 50 State Handbook, compiled by the State Capitol Group. The handbooks include chapters that summarize regulations on lobbying and campaign finance in each of the states. The second measure consists of an additive index of reporting requirements originally developed by Newmark (2005). In the index, states receive a point for every requirement that is in force: whether lobbying of administrative agencies requires registration, and whether expenditures benefitting public officials or employees, compensation received broken down by employer, total compensations received, categories of expenditures, total expenditures toward lobbying, and contributions received from others for lobbying purposes are all reportable (see Newmark 2017). States may receive scores ranging from 0 to 7 for this measure.

Ultimately, given the availability of observations of our dependent and explanatory variables, our dataset consists of 9,698 healthcare interests that registered to lobby in at least one state in either 1989 or 2009. Given that only 17 organizations hired tobacco lobbyists in the 1970s, and measures for lobby reporting standards are not readily available for that decade, observations from that wave were dropped. Every observation consists of a healthcare group that is registered to lobby in a particular state. Some groups registered in multiple states or during both observation waves, so their names appear more than once in the dataset. ⁵

Method and results

To test our hypotheses, we estimate a series of regression models that predict the total number of tobacco lobbyists hired by each healthcare group, with institutional interests being omitted as the base or reference category. The observations of the dependent variable range in value from 0 (for 88.5 percent of observations) to 23 lobbyists. Since our outcome variable is a count with overdispersion (i.e., the model-conditional mean exceeds the model-conditional median), we estimate a series of negative-binomial and zero-inflated negative-binomial regression models (see Long 1997). In the negative-binomial models, numbers of tobacco lobbyists hired are treated as distinct events, and we control for the total numbers of tobacco groups active in each state. However, given that some healthcare groups in our dataset registered to lobby in states in which no tobacco interests were present at all, observations for those groups (for the dependent variable) could not assume values greater than zero. The zero-inflated models each consist of two sets of coefficients: one set of traditional negative-binomial coefficients and another set of logistic coefficients. Whereas we control for total tobacco groups in our negative-binomial models, we use this variable to predict excessive zeroes in our zero-inflated models.

We include a small number of group- and state-level control variables in all regression models that are not related to our hypotheses. Goldstein and Bearman (1996) found that all of the 450 lobbyists active throughout the country in 1994 were multi-client advocates. This suggests that organizations that rely more often on multi-client lobbyists are more susceptible to hiring risky lobbyists in general. Indeed, Strickland (2020) finds that institutions hire these lobbyists more often than do other kinds of organizations. Therefore, to ensure that any correlations found between the hiring of risky lobbyists and membership types are not merely artifacts of omitted variables, we include the total numbers of multi-client lobbyists hired by each group in our regression specifications. We expect this variable is be a positive predictor of how many tobacco lobbyists each group hired. Moreover, in a second set of tests, we conduct similar regression analyses using only observations of groups that hired at least one multi-client lobbyist (i.e., by omitting all groups that relied solely upon single-client agents).

For state-level controls, we hold constant the effect of tobacco groups in the states. Presumably, the likelihood of hiring a tobacco lobbyist increases as more tobacco groups register to lobby in a state. Group numbers were calculated using the lobby registration lists. We also include an indicator for whether states had separate registration procedures for law firms. Finally, we include effects for states and periods. These effects help to capture state-level differences in the hiring of tobacco lobbyists that are not included in our models and the overall (national) difference in hiring trends between the two waves of observations. As a result, the coefficients for state-level variables (i.e., institutional and legal variables) are based on changes in hiring trends that occurred within states and over time and changes in local institutions or laws (Mummolo and Peterson 2018).

Our regression coefficients are included in Table 2. The first two models present negative-binomial regression coefficients, while the last two models present the results of zero-inflated negative-binomial regression models. From two of the four models, we find that occupational groups were as likely to hire tobacco lobbyists as non-membership interests. Only in the second and fourth models, which were calculated using only interest groups that hired at least one multi-client lobbyist, were occupational groups less likely than institutions to hire tobacco lobbyists. In both models, these groups hired about 18 percent fewer tobacco lobbyists. The difference in results between the four models suggests that the hiring of tobacco lobbyists between institutional and occupational interests is correlated with the hiring of multi-client lobbyists and that numerous occupational groups hire only single-client lobbyists. ⁶ Results are quite consistent across all four models, however, for advocacy groups. These groups hired between 39.7 and 47.3 fewer tobacco lobbyists on average, depending on model specification. Unsurprisingly, as the number of multi-client lobbyists hired by any interest increased in number, the number of tobacco lobbyists hired increased as well.

Table 2. Tobacco lobbyists hired by healthcare interests

Note: State and period effects were included in all models but not reported. Errors are clustered by state. *p < 0.1; **p < 0.05; ***p < 0.01, two-tailed tests.

In the results not presented, the differences by group type in numbers of tobacco lobbyists remain largely the same whenever we exclude explicitly anti-drug and anti-cancer groups, and lung associations, from our sample. However, these groups were not immune from hiring tobacco lobbyists. Anti-drug groups were no less likely to hire tobacco lobbyists than institutions. Three of these organizations in our dataset hired at least one tobacco lobbyist. The Kansas Alcohol and Drug Addiction Counselors Association, the Arizona chapter of the Partnership for a Drug-Free America, and the Florida Alcohol and Drug Abuse Association all hired tobacco lobbyists in 1989 or 2009. Some anti-cancer groups, such as the Cancer Treatment Centers of America (an institution), two state chapters of the American Cancer Society, and the Haitian-American Association Against Cancer, hired tobacco lobbyists. Three state chapters of the American Association for Respiratory Care, a professional association for respiratory therapists, hired tobacco lobbyists.

With regard to intergroup competition for access, two of the four regression analyses support our third hypothesis. From the two models in which competition is correlated with tobacco lobbyists hired (at the p = 0.05 threshold), every additional group per legislator was correlated with all groups hiring between 9.7 and 12.3 percent additional tobacco lobbyists, on average. The results, although not entirely consistent, provide evidence that groups accepted a greater risk of conflicts in political environments in which more groups were competing for the attention of lawmakers (as in Kattelman 2015). Given that the numbers of lawmakers did not change in most states between the two waves of observations, much of this effect is likely driven by increases in state interest groups.

The results provide strong and consistent support for our fourth hypothesis: the implementation of anti-conflict laws appears to have dampened the numbers of tobacco lobbyists that healthcare interests hired. Between the two periods we studied, four states implemented anti-conflict laws: New Jersey, Pennsylvania, Texas, and Utah. Our results suggest that the implementation of these laws, as well as the existing laws in Idaho and Washington, led healthcare groups to hire between 72.2 and 77.5 percent fewer tobacco lobbyists on average. This provides evidence that lobbyists obeyed these laws due to either actual or potential conflicts and that interest groups accordingly dissolved relationships with conflicted lobbyists; but our results do not prove that perfect compliance occurred. In accordance with our fifth hypothesis, the implementation of additional lobbyist reporting standards (as measured by Newmark 2005) was correlated with healthcare groups hiring more tobacco lobbyists. This trend confirms our expectation and is consistent across all four models: with every additional reporting requirement a state adopted, healthcare interests hired between 18.4 and 20.4 percent more tobacco lobbyists. In the results not presented, anti-conflict laws and reporting requirements do not moderate or enhance the effects of intergroup competition on the hiring of tobacco lobbyists.

Some results presented in Table 2 are consistent with those presented by Goldstein and Bearman (1996). Although those authors did not propose a theory regarding which healthcare organizations hire tobacco lobbyists more often, they did examine what kinds of groups (broadly construed) hired such lobbyists. According to them (1140), “[f]ive major categories of health organizations [stood] out as employing [tobacco] lobbyists: physician professional associations and societies; hospitals, hospital associations, and health care associations; pharmaceutical organizations; optometry and chiropractic associations; and medical [or] healthcare corporations.” We note that none of these organizations are advocacy groups that consist of members not united by an economic or professional consideration.

Lobbyist compensation

Although we found consistent evidence for our hypotheses by examining the hiring of tobacco lobbyists across thousands of healthcare interests in all states, we conducted a second set of analyses to determine if any of the trends are due to differences in the compensation that lobbyists receive. Recall that tobacco lobbyists are compensated well in comparison to other lobbyists (Givel and Glantz 2001). It may be the case that tobacco lobbyists did not represent health occupational or advocacy groups due to lower pay rates. Indeed, advocacy groups in particular have been found to pay their lobbyists less compensation in comparison to other interest groups (Berry 1977; Strickland 2020). With a second set of analyses, we seek to determine if the trends we found earlier persist even when we control for the effects of compensation paid to lobbyists. Also, the second set of analyses allows us to determine if healthcare interests hire tobacco lobbyists more or less often than all other kinds of interests. The set also allows us to determine if health interests hire tobacco lobbyists less often than non-health interests within each organization type.

To conduct the second set of analyses, we expand a dataset compiled by Strickland (2020) consisting of all the interest groups that registered to lobby in five states where exact compensation statistics are available for every lobbyist–client pairing. Unfortunately, although more than 20 states collect compensation totals from lobbyists, only 5 provide exact compensation broken down by client in formats that are readily accessible. (Such granular information is necessary for understanding how well clients pay different sets of lobbyists.) For example, lobbyists in Alaska often report hourly rates but do not indicate how many hours they spent representing each client. In Indiana, clients often report compensation totals for entire lobby firms and not for individual lobbyists. Compensation figures for lobbyists in Texas are reported in increments, so the exact figures are not available. Rather, in Kentucky, Maine, Maryland, Mississippi, and South Carolina, lobbyists are required to report the total pay received from each client every few months or years. This second dataset consists of more than 3,500 interest groups that registered to lobby in these states in 2018. ⁷ We added to the original dataset the numbers of tobacco lobbyists hired by each interest, and the amounts of compensation paid to all lobbyists from each group. As we did for healthcare interests in all states, we also coded all interests (i.e., not only health interests) in the five states according to their sources of support. As before, interests were classified as institutions with no members, economic (occupational or labor-related) membership groups, and non-economic groups.

Table 3 reports the total number of interest groups active in each of the three categories separated by healthcare status. From the second column, we find that roughly 17 percent of all the groups in our five-state sample are healthcare interests. All but one of the different kinds of interests in our sample hired an average of two or more lobbyists in 2018: the exception is non-healthcare occupational interests. Numbers of tobacco lobbyists hired vary more across groups with only around 2 percent of healthcare advocacy groups hiring such lobbyists. (Only one such group hired more than one tobacco lobbyist.) There is also variance in numbers of multi-client lobbyists hired: advocacy groups hired these agents less often than did other groups. With regard to compensation, advocacy groups also appear to pay their sets of lobbyists less.

Table 3. Lobbyist compensation in five states, 2018

Note: Compensation presented in 2018 U.S. dollars.

In a second series of regression analyses, we estimate the number of tobacco lobbyists hired by each group while also controlling for additional group-level variables: numbers of multi-client lobbyists hired and total compensation paid to all lobbyists. Among the 3,537 observations, the numbers of tobacco lobbyists hired per group are overdispersed. To model such overdispersion, we estimate a series of negative-binomial regression models. Moreover, since all of the observations are from a single year and only five states, we do not estimate coefficients for various state-level variables such as competition for access, reporting requirements, or total tobacco groups. Rather, we include state effects in our models to capture differences in state context.

We first begin by examining the numbers of tobacco lobbyists hired only by the 593 healthcare groups active in the states. Our results are presented in Table 4. In the first model, we estimate a model to determine if the group-level differences we found in our first set of analyses also persist in the smaller sample. As before, healthcare interests were coded according to their types: institutions, economic-based membership groups, and other membership groups. We also control the number of multi-client lobbyists hired by each group. From the first model, we find that trends first presented in Table 2 are also present in the smaller sample from only five states: healthcare advocacy groups hired around 75 percent fewer tobacco lobbyists than healthcare institutions. As before, there is no statistically discernible difference between the numbers of tobacco lobbyists hired by economic membership groups and those hired by institutions. Importantly, these results are also found in the second model where we hold constant total pay amounts. The substantive difference in hiring between healthcare advocacy groups and institutions is practically unchanged.

Table 4. Tobacco lobbyists hired in five states, 2018

Note: All models use the negative-binomial variance function. State effects are included in all models but not reported.

Robust standard errors are reported. *p < 0.1; ** < 0.05; ***p < 0.01, two-tailed tests.

In the third and fourth models, we examine hiring among all interests active in the five states. While also controlling for numbers of multi-client lobbyists and total pay amounts, we examine whether healthcare interests generally hired fewer tobacco lobbyists than all other interests, and also if the different kinds of healthcare interests hired fewer tobacco lobbyists than non-healthcare interests. From the third model, we find that healthcare interests generally were no more or less likely to hire tobacco lobbyists than the other interests in our sample. The results from the fourth regression explain why: the effect is confined to healthcare advocacy groups, who hired about 76 percent fewer tobacco lobbyists than all other interests. Recall that only 80 of the 593 healthcare groups in our sample are advocacy groups, so trends in hiring among those groups were not sufficiently prevalent to produce an overall effect that could be detected for all healthcare interests.

In the fifth model, we test for whether health institutions hired fewer tobacco interests than non-health institutions while holding constant numbers of multi-client lobbyists and overall pay amounts. The sixth and seventh models apply similar tests to occupational and advocacy groups, respectively. Examples of non-health-related advocacy groups include environmental and good-government organizations such as the Sierra Club and Common Cause. In general, we find no evidence that health-related groups avoid tobacco lobbyists more often than other groups when holding constant the membership type of the organizations. This suggests that, although health membership groups hire fewer tobacco lobbyists than health institutions, similar differences (based on membership types) may be found among non-health interests. These findings do not contradict our expectations but rather reveal the universal avoidance of tobacco lobbyists by advocacy groups. ⁸

Discussion

In this study, we proposed that interest groups consider the potential for harm when choosing lobbyists. Following McMunigal (2001), we argued that risk and harm are not equivalent in that some lobbyists who represent seemingly adversarial interests may not cause harm to their clients’ interests. Rather, groups consider and balance the risks and benefits of hiring particular kinds of lobbyists. By turning to both scholarly and popular accounts, we identified several factors that may have revealed different levels of risk acceptance by interest groups. These factors were related to groups’ sources of support and their institutional and legal environments. Upon assembling a large dataset of healthcare interests and the numbers of tobacco lobbyists they each hired, we found evidence that non-economic membership groups were less accepting of risk than other groups, that inter-group competition for access spurred groups to accept more risk and that lobby laws such as anti-conflict laws and reporting requirements had countervailing effects. Although the implementation of anti-conflict laws resulted in healthcare interests hiring fewer tobacco lobbyists, additional reporting requirements resulted in more such lobbyists being hired. A second, smaller dataset revealed that differences in group support continued to explain differences in tobacco lobbyist hire rates even when compensation amounts were held constant. Our findings generally suggest that groups are choosy regarding the kinds of lobbyists they hire and consider the risk of harm to their interests.

Our findings contribute to scholars’ knowledge of how interest groups hire lobbyists. Although a substantial amount of research suggests that organized interests have limited means for preventing shirking or various risks of harm (see Drutman 2015), our findings suggest that groups preemptively hire lobbyists who they think provide a proper balance between risk and effectiveness. In other words, groups are not powerless to prevent shirking. Our findings join a growing list of studies that suggest that groups behave strategically when choosing which advocates to hire and that group resources and context matter for such hiring trends (see Strickland and Stauffer 2022; Strickland and Crosson 2023).

More important, however, are our findings’ implications for representation and policy. Although healthcare interests may accept risk and hire tobacco lobbyists who they think can deliver access and influence, there is no reliable means of measuring actual harm since lobbyists may misconstrue policy losses as victories (Drutman 2015; Holyoke 2016). There is no evidence that any of the groups mentioned by Gray and Lowery (1996a) and Kersh (2000) that were victims of predatory lobbyists ever knew or understood the true costs (including harms) of being represented by their chosen agents. This suggests that, although interest groups are truly unable to judge the benefits and costs of their agents but nevertheless try, the costs of (mis)representation are potentially great. The informational asymmetries that allow lobbyists to dissemble and misconstrue positions may cost groups dearly and give lawmakers inaccurate impressions of groups’ preferences.

Therefore, our findings’ implications for policy are meaningful for improving the representation of interest groups. If institutions or laws may be reformed to result in more lateral access and fewer risky hires, then lawmakers may gain confidence in the abilities of lobbyists to convey group preferences accurately. Although numbers of organized interests competing for access and influence have increased across the United States over time, numbers of legislators have not and access is now (at least theoretically) more difficult to achieve than before. This has only benefited long-time lobbyists with established relationships or reputations. Unfortunately still, recent reforms of assembly sizes have tended to favor shirking the institutions (see Strickland 2022 for example). Based on our findings, then, the enactment of anti-conflict laws holds the most promise for enhancing the quality of representation that lobbyists provide. ⁹ Such laws are now in effect in only seven states, but the ongoing growth of conflicts and multi-client advocacy in general suggest that these laws are needed more today than ever. Although advocacy groups are less likely to hire conflicted and multi-client agents than other groups, the potential for harm remains for all interests.

We hope that others may build on our findings. Although scholars have some understanding of the benefits and costs of joining coalitions (see Dwidar 2022; Holyoke 2009; Hula 1999; Junk 2019), scholars have little knowledge about the benefits and costs of hiring multi-client lobbyists outside of formal coalitions. Our findings align particularly well with those of a study on coalition lobbying: Beyers and De Bruycker (2018) show that the kinds of organizations that hire tobacco lobbyists in the United States (i.e., organizations without members) are also the most willing to partner with other organizations when lobbying in Europe, including those that make for “strange bedfellows.”

Separately, there remain numerous other opportunities for research on multi-client lobbying outside of formal coalitions. It remains unclear how exactly interest groups locate lobbyists for hire (i.e., word of mouth, advertising, association directories), particularly contractors who already represent tobacco interests, and how clients may discourage lobbyists from harming their interests. For example, lobbyists in several states are allowed to be paid on a contingent basis by clients (see Susman and Martin 2007). Although such contingent-fee clauses may theoretically align the interests of multi-client lobbyists with the interests of their clients, it remains unclear how often lobbyists agree to such arrangements. Also, in addition to the ostensible conflicts between health and tobacco interests, conflicts between other sets of clients may emerge, including between health and alcohol interests or dental and sugar interests. The emergence and influence of cannabis interests present new opportunities to study ostensible conflicts. Finally, it also remains unclear how often lobbyists actually abuse the trust of their clients. As shown earlier, numerous examples of harm are documented but there is little systematic examination of abuse among lobbyists.