Journalistic accounts and my own elite interviews have highlighted the importance of mobilizing employees in the growing political sway of the marijuana industry, especially in states that have adopted adult-use legalization . A second reason members might support industries with a presence in their districts draws on the logic of structural power . Because their re-election prospects depend in part on economic performance , members of Congress have an incentive to support policies that benefit business interests central to economies in the places they represent—even in the absence of active corporate political mobilization. Governors also, for the same reason, might use their sway with members of Congress to advocate for federal laws that align with state policies and programs . While marijuana industry’s economic contribution remains small relative to major industries like healthcare and energy, it is highly labor intensive and, in many states, growing rapidly . Moreover, high taxes on marijuana are often used to fund state programs in areas like education and criminal justice, and also to bolster general fund revenues . The importance of industry tax revenue for budgets and programs in the states they represent thus gives members of Congress representing legalizing states another reason to support pro-marijuana federal laws. In addition to conferring structural power and the ability to mobilize employees or members, state policies that benefit particular organized interests also might provide those interests with a greater capacity to deploy financial resources: lobbying and campaign contributions. Though money is generally ineffective at buying roll-call votes in Congress,rolling tables research suggests it can shape how members allocate their time and attention.
Hall and Wayman’s seminal study, for instance, found that, while PAC contributions from organized interests had no effect on roll call voting, contributions did affect the time that members spent working on issues promoted by contributors—a result that has been corroborated experimentally . Of course, firms and unions can also contribute to the campaigns of members representing districts where they do not have an economic presence. In the marijuana case, as I will show, the industry has mostly targeted members representing legalizing states, but also contributed to campaigns elsewhere. The focus on members representing legalizing states may reflect a strategy of seeking to increase the time and attention that members already inclined to support marijuana reform—perhaps because of other mechanisms associated with industry growth in their districts—spend on the issue. The mechanisms discussed above focus on organized interests, but public policies also affect individual-level mobilization and attitudes . Citizens living in states that adopt reforms—after experiencing them “on the ground”—may become more comfortable with their national adoption. In cases like marijuana where new policies establish new markets and products, consumers can also be a powerful coalition—especially when organized by the firms that sell to them . In considering member responsiveness to shifts in individual-level political behavior, there is ample evidence that members’ roll call votes are correlated with the preferences of constituents in their districts , and that policy is broadly dynamically responsive to shifts in public attitudes . At the same time, more recent findings suggest that member responsiveness might be decreasing. As the major parties have polarized, a greater share of variation in member behavior is explained by partisanship, so a competitive district might be represented very differently depending on the outcome of a close election .
More broadly, studies show that even if member behavior is correlated with voter preferences, there remains a large overall gap between public preferences and public policy . One reason is that politicians misperceive the preferences of their constituents. Comparing surveys of state legislators to Cooperative Congressional Election Study data, Broockman and Skovron find that lawmakers consistently believe the preferences of their constituents are more conservative than they actually are. These biases likely extend to the U.S. Congress. Conducting surveys of senior congressional staffers, Hertel-Fernandez, Mildenberger, and Stokes find that staffers have skewed perceptions of public attitudes driven, they argue, by a reliance on conservative and business interests for policy information. This brings us to a final potential mechanism: state policy as signal of constituent preferences. When states adopt particular policies, it provides information to members of Congress representing those states about the preferences of their constituents. This is especially true in cases where state policy is enacted via initiative, which has been a crucial element of liberalizing state marijuana laws. In addition to providing information, these votes can provide political ammunition, giving election opponents the chance to highlight cases where members of Congress are “out-of-step” with their constituents’ expressed preferences. Each of these mechanisms—shifts to interest group resources, shifts in individual-level mobilization, and signals of constituent preferences—are likely to produce positive feedback from state policy to congressional representation. I would therefore expect members to respond to the adoption of policies in the states they represent to be more likely to support aligned policies at the federal level. In the next section, I introduce the empirical context and design for examining this relationship. Marijuana politics has several features that make it particularly suitable for investigating the causal effect of state policy on representation in Congress. First, as I discuss below, the key role of the ballot initiative in state legalization of marijuana provides exogenous variation in likelihood of legalization that can be leveraged for causal inference.
Second, legalization has produced clear, sizable, and fast changes to states’ policy and interest group landscapes: between 2010 and 2020 industry revenue increased by nearly tenfold . Beyond being a suitable empirical case to examine policy feedback dynamics in Congress, the politics of marijuana is important to understand because of the policy implications. Marijuana prohibition has direct and sizable consequences for people’s lives. In 2018, with marijuana already having been legalized for adult-use in 10 states, 40 percent of total drug arrests in the U.S. were for marijuana-related offenses—with a full 92 percent of those arrests just for possession . Convictions for marijuana possession can produce life-altering costs, affecting eligibility for public housing and student financial aid, employment opportunities, child custody determinations, and immigration status among other things . For these and other reasons, many advocates see marijuana policy as a crucial piece of broader criminal justice reform . Marijuana policy also has important economic implications. As legalization has advanced, industry revenue has grown steadily from a total of $3.5 billion in legal sales in 2014 to over $13.5 billion in legal sales in 2019 and marijuana industry is now one of the fastest areas of job growth in the U.S. . Though marijuana remains a Schedule 1 drug at the federal level,cannabis grow supplies in the past 25 years state actions have spearheaded a steady liberalization of marijuana policy. California’s Proposition 215 of 1996, which permitted the use of marijuana for medical purposes, initiated a wave of state medical marijuana laws. By the end of 2020, the use of marijuana for medical purposes was legal in 33 states, with another 14 states permitting marijuana with limited THC content for medical use. More recent years have seen the expansion of adult-use marijuana legalization at the state level; between 2012 and the end of 2020, 15 states legalized marijuana for adult use. The state-level liberalization of marijuana laws has been driven by a combination of increasing public favorability and well-funded advocacy organizations working across the country. Support for marijuana legalization increased from 31 percent of the public in 2000 to 68 percent in 2020 . Advocates have taken advantage of favorable public opinion by relying heavily on citizen initiatives, and organizations like Marijuana Policy Project have developed expertise in running initiative campaigns. Even as the legal landscape of marijuana has shifted dramatically at the state level, federal law has remained largely stagnant. Lack of progress at the federal level has led to growing conflict between state and federal laws, leaving the burgeoning industry in a highly fragmented legal environment. But in addition, federal prohibition limits industry access to banking and other financial services and limits small businesses’ access to tax deductions. Warming public attitudes, industry growth, and growing costs from state-federal policy conflicts have produced momentum for reform in Congress.
Several pro-marijuana bills were introduced in the 115th Congress, but Republican majorities kept them from being brought to floor votes. With various forms of legalization continuing to spread across the states, and Democrats taking control of the House in 2018 elections, advocates and industry interests saw the 116th Congress as a crucial opportunity to advance reform at the federal level. As one journalist wrote: “This is the first Congress in history where, going into it, it seems that broad marijuana reforms are actually achievable” . Efforts from advocates and industry coalesced around three broad goals—each with a related proposed bill—in the 116th Congress. First, and narrowest in scope of the three, was providing the marijuana industry with greater access to banking services. The proposed SAFE Banking Act would “create protections for depository institutions that provide financial services to cannabis-related legitimate businesses and service providers for such businesses” . The second major goal was broadly protecting industry and consumers in states that have legalized marijuana from federal interference or prosecution. The STATES Act would exempt individuals and corporations operating legally according to state law from federal enforcement. The third and broadest goal was amending the Controlled Substances Act to end federal prohibition on marijuana. The MORE Act would both end federal prohibition as well as expunge prior convictions. Notably, support for the MORE Act comes to a greater extent from advocates than from industry interests, which have focused on narrower bills.26Even using state policy variation for leverage, causally estimating the feedback effects of prior policies on the actions of lawmakers poses inferential problems. Since policy adoption is nonrandom, any observed relationship between sub-national policy and member behavior might be driven simply by a correlation in the preferences of sub-national policymakers and members of Congress—not by the theoretical mechanisms discussed above. In this case, the types of states that legalize marijuana are also likely to be the types of states that elect representatives that are more progressive on marijuana policy, making it difficult to estimate the effect of legalization on representation in Congress. This paper relies primarily on an instrumental variables design for causal inference. The IV design draws specifically on the fact that citizen initiatives have been a fundamental tool for legalization advocates. The first 8 states to legalize marijuana for adult-use did so via citizen initiative. For pro-marijuana organizations, whether states allowed initiatives has been a major factor in determining where to allocate time and resources. The importance of the initiative, according to one advocate, stems from the fact that the public generally holds more liberal views on marijuana than representatives in state legislatures.As of the end of 2020, whether a state allowed citizen initiatives was highly correlated with whether it permitted marijuana for adult use and whether it allowed medical marijuana —the first requirement for a valid instrument. To serve as a valid instrument, initiative status must also, conditional on observables, only be associated with member behavior through the mechanism of legalization . There is good reason to think this is the case. Initiative processes were generally put into place around the turn of the 20th century in response to pressure from elements of the Progressive movement— long before marijuana policy was a salient issue. If initiative status were related to congressional representation on marijuana issues through mechanisms other than legalization, then we would expect these rules to also be related with factors generally associated with the behavior of members of Congress. But, as indicated by Figure 1, whether a state allows citizen initiatives is unrelated to the factors political scientists generally believe to drive congressional behavior. First, initiative status is uncorrelated with measures of congressional ideology in the 116th Congress. In addition, it is slightly negatively correlated with 2016 Democratic presidential vote share, which should bias results downwards to the degree it is not accounted for in analysis. Finally, it is neither correlated with state-level measures of attitudes towards marijuana legalization nor state-level measures of social liberalism in the mass public from 2000 to 2010 .