Analysis using citizen initiative rules as an instrument, though, as discussed above, can provide causal leverage. The key treatment is a measure of state marijuana legalization status at the end of the 116th Congress.In the main analysis, I code the treatment variable as 0 for states with neither medical nor adult-use, 1 for states with medical marijuana, and 2 for states with adult-use legalization.For outcomes, first, I record binary measures of whether members sponsored or co-sponsored each of the SAFE Banking Act, STATES Act, and MORE Act. I also estimate a broader marijuana bill sponsorship score by computing the proportion of the 14 priority pieces of legislation promoted by the industry group National Cannabis Industry Association sponsored or co-sponsored by each member. Two-stage least squares regression is used to estimate the effect of liberalization of state level marijuana law on these outcomes. The first stage predicts adult-use marijuana legalization from the ballot initiatives variable. First stage results presented in the appendix demonstrate that citizen initiative rules are a strong instrument for legalization. The second stage estimates the relationship between predicted legalization and bill sponsorship.I estimate models both with and without state- and member-level covariates: party-identification , ideology ; and state-level covariates: 2016 Democratic presidential vote share; and social liberalism of the mass public. Estimates are presented in Figure 3. For the SAFE Banking Act and the STATES Act, I estimate that state-level marijuana legalization increased the likelihood that members sponsored liberal marijuana bills.
The non-covariate adjusted coefficient of for the SAFE Banking Act, for instance,cannabis grow supplies indicates a 1-point shift in legalization status is associated with an increase of 24 percentage points in likelihood that members sponsored the Act. I estimate effects of similar magnitude for the STATES Act. I do not estimate a statistically significant effect for the MORE Act, which may be driven by the fact that sponsorship of this bill was more partisan than the others . Turning to members’ broader bill sponsorship scores, I find evidence of a causal relationship. The coefficient of .06 suggests that a 1-point shift in legalization status is associated with an increase of .06 , which corresponds to additional NCIA supported bills sponsored on average.In developing the paper’s theoretical argument, I proposed that state policy might affect representation in Congress by, first, shaping the political economy in ways that influence the pressures faced by reelection-seeking members; and second, by sending signals that provide information about constituent preferences. While providing a precise decomposition of the role of these potential mechanisms in the empirical case is not possible with the available data, in this section I bring together quantitative and qualitative evidence to provide some insight as to how these mechanisms have contributed to the overall effect observed. As part of gathering qualitative data, I conducted 8 semi-structured elite interviews with individuals working in marijuana politics and policy. In considering how state legalization affected the pressures faced by members of Congress, it is worth considering, first, the extent to which legalization has actually affected organized economic interests. The answer is: quite a lot, especially in states adopting adult-use legalization. According to NCIA, as of 2018 the average state with medical marijuana featured sales of $21 per capita , while the average state with adult use legalization featured sales of $130 per capita .Revenue growth in legalizing states has increased the capacity of industry interests to engage politically at the national level.
To examine exercise of instrumental power, I draw on lobbying and campaign contributions data collected by the Center for Responsive Politics. The data reveal a sharp increase in lobbying from marijuana industry coinciding with recent state adoption of adult-use legalization. Annual federal lobbying from the marijuana industry has grown from just 45,000 dollars in 2012—the year that Colorado and Washington voters legalized marijuana for recreational use by ballot initiative—to nearly 6 million dollars in 2019 . Campaign contributions data also suggest that legalization has affected the political presence of marijuana industry. Firms in the industry did not donate to congressional campaigns prior to the 2018 cycle. In the 2018 cycle, however, marijuana industry interests contributed in 19 percent of House races in states with adult-use marijuana, and just 2 percent of House races elsewhere. Discrepancies for the Senate were less stark, with contributions in 7 percent of races with adult-use marijuana, and 5 percent of races elsewhere. IV analysis again using citizen initiative rules to instrument for legalization suggests this relationship is causal . In addition to leveraging its growing resources for lobbying and campaign contributions, the marijuana industry has leveraged its economic growth to engage politically by mobilizing consumers and employees. For instance, in Colorado, Governor Jared Polis collaborated with industry interests to turn out marijuana consumers and industry employees in his 2018 reelection. As part of this effort, the campaign matched the state’s database of marijuana employees to the voter file to identify potential supporters, and then sent them targeted text messages and mailers .
The sway of marijuana industry and marijuana voters in Polis’s 2018 bid was a major reason why former Senator Cory Gardner, who anticipated a tough re-election in 2020 , made marijuana such a priority in the 116th Congress.Outside of industry mobilization, another potential mechanism is that state legalization leads the broader public in legalizing states to be more liberal on marijuana, which then drives members of Congress to respond by supporting marijuana reform. To investigate the association between marijuana legalization and public attitudes, I use state-level estimates of support for marijuana legalization collected by Caughey and Warshaw .Using a difference-indifferences design, I compare changes to public opinion in legalizing states to changes in public opinion over the same time period in a set of similar non-legalizing states. More specifically, I leverage a method recently developed by Xu , which is particularly useful since it allows for implementation of synthetic controls in cases of multiple treated units and variable treatment periods. The method uses a linear interactive fixed effects model to impute counter factuals for each treated unit . I consider legalization of marijuana for medical use and for adult use separately.Figure 6 plots average support for marijuana legalization over time for both the treated states and the synthetic controls. In addition to estimates of state-level support for legalization,dry racks for weed the model also includes estimates of mass ideology . If legalization led to greater public support for marijuana, we would expect the solid series representing legalizing states to jump above the dashed series at treatment . The evidence, though, suggests no such effect. Indeed, public opinion, for both adult-use and medical, is slightly more favorable in the synthetic control group, though not statistically distinguishable from the treated group. This suggests that state-level legalization has not disproportionately improved public opinion in the states where it is adopted.However, even if it did not drive improved public opinion, it is likely that marijuana legalization—especially in the majority of cases when it was enacted via ballot initiative— provided a signal of public favorability to lawmakers in Congress. This is the mechanism that, along with growing influence of industry in legalizing states, interviewees working as lobbyists and policy advocates were most likely to raise. It is difficult to investigate quantitatively, though. One analysis that can provide insight into the importance of this mechanism is exploring the relationship between the length of time since legalization and representation. If signaling were driving the effect of legalization on representation, we would expect members representing states with more recent legalization votes to adopt more pro-marijuana positions. If industry influence were more fundamental, we would expect members representing states that legalized further in the past, and where industry had a greater amount of time to develop, to adopt more pro-marijuana positions. Figure 7 presents models with members’ bill sponsorship scores and roll calls on the votes for which a positive effect of legalization was estimated as outcomes, and the number of years of legalization for a members’ home state as the key independent variables.
The left panel presents results for adult-use states and the right panel presents results for medical-use states . Broadly speaking, results lend support for the industry influence mechanism—not signaling. For adult-use states, in both bivariate and covariate-adjusted models, years since legalization is positively and statistically significantly associated with both bill sponsorship and roll-call outcomes. I recover similar results in the bivariate models of medical use legalization, but they are not robust to covariate adjustment. This may reflect the adult-use legalization tends to have a much greater effect on industry growth than medical-use legalization. Using an IV design that leverages exogenous variation in marijuana legalization from longstanding differences in the availability of citizen initiatives, I have shown that the policy landscapes in the states they represent affect the behavior of members of Congress. I observed strongest effects for bill sponsorship, but also effects on certain lower-profile roll-call votes. The evidence suggests the strongest mechanism driving these effects is growing industry influence in legalizing states, though other mechanisms—particularly signaling of public preferences— cannot be ruled out. The set of analyses is not without its limitations. One limitation is a short temporal window. Since adult-use legalization and the emergence of the marijuana industry are relatively new phenomena, data is limited. For instance, only a few roll-call votes related to marijuana legislation have been taken since the state legalization wave began. As more data become available, researchers will be able to extend the analyses performed here. In addition, while this study indicates a causal link between state-level adult-use legalization and representation in Congress, there remains uncertainty with respect to the mechanisms. Finally, it should be noted that these analyses likely underestimate the full effects of state legalization on the politics in Congress. While this paper demonstrates the effect of legalization on members representing legalizing states relative to non-legalizing states, mechanisms like growing industry presence in Congress are likely to affect legislators across the country—they are simply more pronounced in legalizing states. Future empirical work can build on the theoretical framework developed here to explore the role of different mechanisms in different policy cases. For instance, the role of signaling is likely unusually important in this case due to the role of the ballot initiative compared to cases where state policy reform is achieved via legislation. The findings on the role of state policy in structuring state political economies and, as a result, the pressures faced by members is likely to be more broadly generalizable to other policy areas. Consider, for instance, climate change. Industries reliant on the burning of fossil fuels are enormously powerful in American politics, spending vast amounts of money on lobbying and campaign contributions in federal as well as state and local politics . Organized interests enriched by the extraction and burning of fossil fuels have also become a key organizing force within the Republican Party . The power of these groups in our politics is built atop a set of policies in place across the federal system that not only fail to adequately price the negative externalities of burning fossil fuels , but also subsidize the production of fossil fuels . But as states governments continue to adopt and strengthen policies driving the transition to renewable energy , this is likely to reshape the pressures faced by members of Congress, potentially opening new opportunities for federal policy. Or consider policing. The killing of George Floyd in May 2020 led to widespread protests calling for actions across levels of government to enact major policing reforms, including in Congress . Though these reforms are widely popular , a major impediment to their enactment is the power of police unions, which leverage financial resources and ability to mobilize members to prevent reforms . Analysis drawing on the framework proposed here might explore the degree to which the power of police unions to prevent reform in Congress is bolstered by pro-police state and local policies. One key scope condition for this mechanism is the degree to which policy areas feature strong vested interests dependent on material benefits from government policies . Considering the role of state and local policy in congressional representation, another key scope condition is the degree to which governance is shared between federal, state, and local levels—a core feature of American federalism . The scope of sub-national authority in American federalism and increasingly active role of state governments in American politics means it is crucial that we develop a better understanding of the implications of state policies for the broader polity.