Association for Public Policy Analysis & Management

Lotteries and casinos are often adopted as a means of generating revenues for state and local governments. While much research has focused on these forms of gambling, the use of “non-casino” gambling machines—known as video gaming terminals (VGTs)—in the US context has garnered little attention. To date, seven states have legalized the use of these devices, the most recent case being Illinois, where VGTs were legalized statewide in 2009 and first came into operation in 2012. Since legalization, Illinois has become the national leader in both the number of VGTs in operation and the number of establishments hosting them (2016 State of the States: The AGA Survey of the Casino Industry). As of February 2017, there were 25,363 VGTs operating across 5,824 establishments in Illinois (Illinois Gaming Board).

Despite their expansion across the country, substantive analyses of VGTs at the municipal level have not been attempted in any of the seven states where non-casino devices are legal. Consequently, little is known about 1) the motivation behind VGT adoption and 2) the determinants of VGT revenue-generating capacity at the local level. This analysis seeks to fill this gap in the research and, in so doing, to inform the decisions of state and local policymakers who may be considering VGT legalization and adoption.

Towards that end, we construct a rich longitudinal dataset spanning four years using data from Illinois. As local governments are fully empowered to allow or prohibit VGTs in their jurisdictions, this setting provides substantial variation in both VGT adoption and local revenue across both jurisdictions and time. Using a two-part model and a rich set of controls, our identification strategy first identifies the predictors of VGT adoption among local jurisdictions and then, conditional on adoption, estimates the determinants of VGT revenues. To address possible sample selection bias and to check the robustness of our results, we alternatively consider Tobit and Heckman Selection models in our analysis. It is our hope that this study spurs more interest in this topic, for which there exists many avenues for further research.