The Political Economy of Legalized Sports Gaming: Evidence from California’s Proposition 26

INTRODUCTION

Casinos on tribal land in the U.S. have been a significant part of the U.S. gaming industry until the U.S. Supreme Court’s 1978 decision in California v. Cabazon Band of Mission Indians. ¹ In 2018, the U.S. Supreme Court decision in Murphy v. National Collegiate Athletic Association created the opportunity for each state to regulate sports gaming. ² Since states received the authority to legalize sports gaming, it has expanded rapidly, with a majority of states allowing legalized sports betting either online or in-person. ³ Sports betting is still illegal in the two most populous states: Texas and California.

In this commentary, we analyze county-level voting on the 2022 California referendum Proposition 26—the Legalize Sports Betting on American Indian Lands Initiative—to better understand the political economy of sports gaming adoption. The initiative would have legalized in-person sports betting at American Indian gaming casinos and licensed racetracks in California. Only 33% of voters said yes on Proposition 26. While others have provided an excellent analysis of Proposition 26 from a political and legal perspective, we feel that an empirical analysis would complement existing research. ⁴ In this commentary, we use an empirical median voter model to determine what relationships existed between constituent demographics and geographic location and vote totals. Doing so can help to adjudicate between different arguments as to why Proposition 26 failed and also to help inform future referendums.

Before proceeding with our empirical analysis, it is important to note that our work cannot provide insight into the effect of Proposition 27 on Proposition 26. On the ballot at the same time as Proposition 26, Proposition 27 would have legalized online and mobile sports betting. It was soundly defeated. Over 50 American Indian tribes opposed Proposition 27 as it was viewed as leading to an end of their exclusivity with respect to gaming in California. ⁵ As a result, American Indian tribes devoted all their resources to defeating Proposition 27 instead of supporting Proposition 26. ⁶ It is important to keep this context in mind when evaluating our analysis.

EMPIRICAL APPROACH

To analyze the determinants of voting on the legalization of sports betting in California, we use an empirical median voter model. ⁷ The median voter model says that the preferences of the median voter in a district, along with special interest and institutional variables, can explain a wide variety of policy outcomes. While median voter models typically include similar control variables to represent the median voter’s preferences, such as education levels, gender, and income, we also look at the economics of gaming literature for additional variables.

Our literature review found three articles that use the median voter model to explain gaming outcomes. Bonneau and Hall (2022) ⁸ use it to understand why Maine voters rejected a referendum to construct a casino in the border county of Oxford, Maine. Areas with higher incomes and younger voters were less likely to support the referendum. Similarly, Calcagno et al. (2010) ⁹ look at state-level casino legislation and find that older voters were negatively related to commercial casino legislation. Neto et al. (2016) ¹⁰ employ a median voter model to explain the expansion of nontribal casinos in Kansas. They find that having an out-of-state casino nearby was associated with yes votes. These articles, as well as the median voter literature more generally, inform our choice of variables.

County-level referendum results were obtained from the California Secretary of State, and our demographic data were obtained from the US Census Bureau, American Community Survey 5-Year Estimates 2017–2023. (2024). From the American Community Survey, we obtained county-level data on median household income, the percentage of the population that is female, median age, the percentage of people with a college degree or higher, and the percentage of the population that is Native American. In addition, we create binary variables indicating whether the county has a casino and whether it borders a state where sports gaming is legal. Table 1 shows these variables and their summary statistics.

While many of the covariates, such as income, female, and age, are included in nearly every median voter analysis, including the ones on gaming, we would also note that a higher education measure (College or Above%) was not included in any of the previous gaming literature. Three other variables warrant additional discussion. First, our expectation is that Native% will be positively related to voting yes on Proposition 26, as it would have enhanced the value of tribal casinos. Second, we include a binary variable equaling one if there is a casino in the county to capture broader economic spillovers beyond just Native Americans. Finally, we include a binary variable equaling one if the county borders a state that already has legalized sports gaming. (All of California’s neighboring states allow online sports gaming, as does Mexico.) Our prior is that individuals in these counties are more likely to be familiar with sports gaming as it operates in legalized states and would be more likely to have a positive view of Proposition 26. There also may be concerns about wanting to capture revenue that may be lost to cross-border gaming.

Table 2 presents our empirical results using ordinary Least Squares. We estimate four different models. Model 1 only includes the four basic demographic variables. Model 2 adds Native%. Model 3 adds Casino and Model 4 adds Border.

Our primary findings are that age is negatively related to voting yes on Proposition 26, and College and Above% is positively related. This suggests that while younger voters were interested in the expansion of sports gaming, the fact that age is positively related to voter turnout meant fewer yes votes on sports gaming expansion. To our knowledge, the positive relationship between College and Above% is a first finding in the gaming literature. Counties with more citizens with higher education were more likely to vote yes on Proposition 26, ceteris paribus.

In Models 2 and 3 we see Native% is positive and statistically significant. While not surprising, given that Proposition 26 would enhance the status of Native American casinos, it is still important to recognize the power of the Native American vote on the issue. From a referendum passage standpoint, however, their lower population numbers make it important to have a much broader coalition to achieve passage. While Native% does become statistically insignificant in Model 4 once Border is included, this is likely due to the positive correlation between Native% and Border. We do not find support for hypotheses that the larger economic impact of casinos (Casino) or the percentage of the population that is female played an important role in the failure of Proposition 26.

CONCLUDING THOUGHTS

Our findings highlight the importance of three demographic categories that help explain the failure of Proposition 26. While college education and being of Native American ancestry were positively related to voting yes on Proposition 26, the magnitude of their voting block was insufficient to ensure passage. Without a larger, more broad-based coalition in favor of sports gaming only at casinos, similar referendums are unlikely to pass. This is especially true if there are additional issues, such as Proposition 27 diverting away political advertising.