Cross-border casino competition, Externalities and Optimal Tax Policy: A Unified Theory with Quantitative Analysis

Highlights

• Develop a framework on cross-border casino competition with welfare analysis.

• Cross-border gambling makes aggregate casino demand more elastic.

• Cross-border competition in Detroit-Windsor induces both to lower casino taxes.

• The optimal tax mix features a shift from casino revenue to gambler tax in Detroit.

• The optimal tax mix in Windsor features a reversed shift.

Abstract

We develop a framework on cross-border competition in markets for goods with negative externalities and provide evidence for optimal fiscal policy with a special focus on taxation. We build the case of two bordering casinos with city governments setting taxes to maximize social welfare. Analytically, we show that cross-border casino gambling makes aggregate casino demand more elastic. By calibrating the model to fit the Detroit-Windsor market, our welfare analysis shows that cross-border competition induces both cities to lower casino taxes, while the optimal tax mix features a shift from the casino revenue tax to the good and service surcharge on gambling in Detroit but a reversed shift in Windsor. We also find a casino buy-out deal to not be credible because Windsor's willingness to pay Detroit to ban Michigan casinos is far below Detroit's willingness to accept giving up its casinos.

Introduction

Cross-border competition has become increasingly fierce for capturing external sources of revenues. It is particularly prominent in the casino industry where sizable externalities prevail. Across Europe, casino gambling revenues have increased from €8.03 billion in 2016 to €8.55 billion in 2017 with France, Great Britain, Germany and Switzerland being the largest markets. In North America, casino gambling is even larger with a sharp increase over the past two decades: revenues increased by 233% from $11.2 billion in 1993 to $37.3 billion in 2012 in the U.S. and in Canada from $6.4 billion in 1995 to $15.1 billion in 2010–2011. Both in Europe and in North America, casinos are often either government run or heavily taxed and thus play an important role for public households, providing, for example $8.6 billion in the U.S. and $13.5 in Canada.

Today there are 2021 casinos in Europe, 508 in the U.S. and 71 in Canada. Interestingly, many casinos were built along various borders across states and countries. Such cross-border settings are especially relevant in Europe due to the many borders and large number of casinos. Prominent examples include the setting in Lugano, Mendrisio, and Campione near the Switzerland-Italy border; Basel near the Swiss-French-German border; Baden-Baden near the France-Germany border; Rozvadov near the Germany-Czech border and the planned three new casinos in Liechtenstein. The likely best known example in the cross-border setting is in Detroit, U.S., and Windsor, Canada. While there have been some case studies on cross border casino-competition, a systematic study – especially including welfare implications and optimal taxation for casino regulatory policies – remains completely unexplored. We deem such a study not only an important contribution for gambling regulators but also a perfect example for cross-border competition with externalities in general.

City or national boundaries are locations of economic opportunity, especially if the existence of the border is itself the source of a monopoly situation that favors one side over the other (Krakover, 1997). Indeed, the border is a favorite site for the development of casinos, if an untapped, large market exists on the other side.¹ The monopoly situation, however, can turn into a highly competitive one, when casinos are positioned for new competition from the other side of the border. Once the border turns into a relentlessly competitive battleground, not only is the cake of the casino market redistributed, but each side of the border has to deal with the negative externalities generated by gambling casinos on both sides.² While bordering casinos generate demand from the other side of the border and create local jobs and other businesses, they also represent the import of tax income and the re-exportation of negative externalities that accompany the gamblers as they return to their home city (i.e., the export of external disorder costs). Such undesirable consequences have led many governments to use various taxes and/or regulations as a social guardian to control the social cost of gambling despite the revenue generating power of casinos. Casino externalities are an important factor that is often neglected in economic analyses of the casino market, despite the important role that it plays in institutionalizing relevant regulations.

Just how would the relentless competition in this growing industry affect recreational (regular) and problem (addicted) gamblers both on the intensive margin and on the extensive margin via cross-border gambling? What are the underlying driving forces influencing the intensity of cross-border gambling and the bordering casinos' pricing, possibly preferences for gambling, casino taxes, the population of gamblers of different types, and commuting cost, among others? How would the bordering governments’ casino tax policies depend on the extent of cross-border gambling and the associated negative (social disorder) and positive (income creation) externalities? Would it be better to impose a tax on casino revenue (i.e., a wagering tax) or to impose a good and service tax (GST) surcharge on gambling? How would the fiscal competition outcome in turn affect cross-border casino competition? We address these interesting questions with a systematic study of cross-border casino competition.

We develop a theoretical model of cross-border casino competition highlighting the following salient features that are important but largely ignored in the existing literature. First, we model separately the behavior of recreational and problem gamblers and analyze their differential decisions on cross-border gambling. Second, we, on the one hand, allow the bordering casinos to compete with each other for the source of demand from both sides of the border. On the other hand, we permit the two competing cities’ governments to be active, where they can set their optimal tax policy (the casino revenue tax on casino operators and the GST surcharge on gambling) to achieve the highest local welfare. In other words, we analyze cross-border casino competition for both casino gambling revenues and tax revenues, as observed in the real world. Third, for the normative analysis, we consider that “travel to use” casino services may generate local externalities, possibly negative (social disorders) or positive (income creation). Thus, by engaging in tax competition, both governments take into account the “import” of tax revenues and the “export” of external disorder costs. Finally, we provide the first attempt to quantitatively conduct positive and normative analyses by calibrating the theoretical model to fit the Detroit-Windsor data.

The study of Detroit-Windsor casino competition is an example that is well-suited to our analysis for several important reasons (to be fully elaborated in Section 4 below). Briefly put, such competition reflects the historically relentless rivalry of two cities over one of the busiest commercial borders between the U.S. and Canada.³ In addition, it is of a large scale, with participation rates at 71% and 66% in Michigan and Ontario, respectively, and with large cross-border gambling consisting of 80% of customers in Windsor originating in Detroit. Moreover, such competition is accompanied by an active city government competition policy in the form of various tax incentives. Furthermore, the data are very rich, featuring significant cross-city heterogeneities with regard to population size and preferences towards gambling, thus permitting various policy analyses, including city-specific optimal casino taxation that is valuable to policymakers. In addition, two sharp events – the increased commuting costs due to 911 and the decreased population size in Detroit due to the declining automobile industry – have caused external shocks to the setting and allow for interesting analyses.

We solve the equilibrium backward. We first solve the optimization problem of individual gamblers of each type (recreational and problem), obtaining individual demand as well as cross-border gambling decisions. We then determine the (Bertrand) price competition of the two bordering casinos. We further pin down the optimal tax policy imposed by the two competing cities’ governments. The competition between the two bordering casinos is subsequently affected by the tax policy, in addition to the commuting cost of border crossing, the heterogeneous preferences for casino gambling and the differential population size of the two cities. That is, a full equilibrium of casino gambling involves both cross-border casino competition and cross-border casino tax competition. In particular, upon fully calibrating the border casino competition between Detroit and Windsor, we quantitatively assess the casino tax effects and determine the optimal casino tax policy for each city in the presence of cross-border casino competition.

Among many theoretical results, we choose to highlight three sets of findings that are all related to cross-border gambling. First, we show that under a reasonable assumption the demand elasticity for casino gambling is greater than one for recreational gamblers but less than one for problem gamblers, both rising with the preset payout ratio. While a higher commuting cost discourages cross-border gambling, the overall cross-border gambling intensity (for both problem and recreational patrons) is increasing in the own city's casino price and GST surcharge on gambling. Second, the presence of cross-border casino gambling provides an outside option to gamblers, thus leading to an elastic aggregate demand for casino services despite the addictive nature of gambling. Interestingly, in a city whose residents have stronger preferences for casino gambling, the net flows of cross-border gambling are more pronounced. As a consequence, a lower commuting cost that encourages agents to cross the border to gamble would reduce this city's casino monopoly power, thereby making its aggregate demand for casinos more elastic. On the contrary, a lower commuting cost makes the price elasticity of casino demand in a city inhabited by people with weaker preferences for casino gambling less elastic. In short, a lower commuting cost favors cross-border casino business in a city with a weaker taste for gambling. Third, a larger population size in the rival city makes the local city's price elasticity of casino demand less elastic. This may raise local casino prices and induces cross-border gambling. While an increase in the local population may make the local city's price elasticity of casino demand more elastic and local casino prices lower, the resulting negative effect on cross-border gambling is offset by the positive population scale effect, thereby leading to an ambiguous outcome.

By calibrating the model to fit the casino competition between Detroit and Windsor, we obtain additional findings from positive analysis. First, if a larger city whose residents have stronger preferences for casino gambling (Detroit) raises its casino tax (either a casino revenue tax or a GST surcharge on gambling), the cross-border consumption of Detroit would exhibit an extensive margin response in the sense that the proportion of the cross-border gamblers would increase, although the cross-border casino consumption per gambler would decrease. While the competition brought about by Detroit's casinos hurts neighboring gambling revenues, such a loss in Windsor is less than Detroit's gain, leading to an expansion in the overall Detroit-Windsor casino market. Second, a higher wagering tax is more effective in reducing the casino disorder cost than a GST surcharge on gambling. It exhibits nonequivalence in the tax burden between the casino revenue tax and the GST surcharge on gambling. Third, in contrast to the responses to tax shifts, a higher commuting cost leads to an intensive margin response whereby the cross-border gamblers from Detroit to Windsor decrease, but each gambler consumes more. When a rising commuting cost discourages cross-border gambling, the city with stronger preferences and a larger population absorbs greater demand and tax revenue, which are accompanied by higher disorder costs. Fourth, the drop in Detroit's overall gambling population hurts its neighboring casino, Windsor, more severely. By contrast, Detroit's casino demand and revenue, tax revenue, and disorder costs are more responsive to its proportion of problem gamblers. Fifth, while individual demand for gambling is more responsive to the wagering tax than the GST surcharge on gambling, the government's tax competition tends to make the cross-border casino competition more intense regardless of the instruments of the casino tax.

Our quantitative welfare analysis also leads to several interesting findings regarding the optimal casino tax policy. First, we establish the optimal policy based on a single casino tax instrument of a city, given the alternative tax instrument and its rival's tax policy at the benchmark values. We find that cross-border competition induces both city governments to lower each tax compared to the pre-existing rate. The relentless competition pushes cities to tolerate problem gamblers to visit their casinos; it is more pronounced in Windsor as it relies more on cross-border gamblers and has lower disorder costs. Second, we conduct a tax incidence exercise, solving the optimal tax mix in a city given its rival's tax policy at the benchmark values. We find that while Detroit has a favorable tax mix away from the casino revenue/wagering tax, Windsor has one away from the GST surcharge on gambling to attract cross-border gamblers. Third, we perform a welfare-based pairwise casino competition in one tax instrument, fixing the other tax policy at the benchmark values. We find that, in order to better compete with the neighboring casino, it is optimal for both cities to lower casino revenue tax rates and GST surcharges on gambling compared to those where the rival's tax policy is given (i.e., the optimal policy of a single casino tax or a tax mix obtained above). More interestingly, it is optimal for Detroit to set relatively high casino taxes to control the social cost of gambling by preventing Windsor's problem gamblers from crossing the border. By contrast, it is optimal for Windsor to aggressively set lower casino taxes to enhance casino and tax revenues and income creation by pulling in the cross-border visitor, including some problem gamblers from Detroit. Finally, we find that Windsor's willingness to pay Detroit in order for Detroit to ban its casinos is far below Detroit's willingness to accept giving up its casinos. Thus, such a casino buy-out deal in this cross-border rival scenario is not credible.

This paper is a first attempt to develop a model that is rich enough to capture the central features of cross-border casino and tax competition, but simply tractable to yield valuable insights toward addressing the interesting questions mentioned above. In both our analytical and quantitative methods, we have developed a richer framework of fiscal competition with “travel to use” services generating local externalities that could be negative or positive, or, from a different angle, a richer framework of spatial competition with competing governments that are active in setting their optimal tax policy. There is a lack of a comprehensive analysis of casinos in both positive and normative aspects because of the paucity of theoretical modeling in this field. Among rare exceptions, Sauer (2001) develops a political competition model to study how gambling restrictions lower the level of gambling, whereas Chang et al. (2010) study the entry and tax regulation of oligopolistically competitive privately-run casinos and government-run casinos in the presence of casino gambling externalities. Neither examine cross-border casino competition, which is the primary focus of our paper. Quantitatively, our welfare analysis provides policy implications to the casino policymakers of the bordering cities and offers new insights into the sparse economic literature on casino gambling. To date, efforts to consider optimal gambling taxation have been limited primarily to lottery games. In viewing a casino tax as a Pigouvian tax to correct for externalities, more research is needed to assess the size of the externalities involved and design optimal corrective taxes.

Our model-based quantitative approach offers such assessment in a systematic manner, which is valuable as well for other broader studies with limited data. For example, our paper may be applied to the literature on cross border shopping for goods by taking advantage of price differences between the bordering countries, such as the prices of tobacco and alcohol, due to different sales or sin/excise taxes (such as Delaware versus Pennsylvania or New Jersey). With different population sizes and preferences, we can differentiate between regular and heavy addicts, analyzing their consumption behaviors and their differential external impacts on welfare. Moreover, our framework may also be applied to the tourism literature, especially because the cross-border tourism competition is usually associated with sophisticated negative congestion or environmental externalities accompanied by positive income creation (such as the rivalry between Hong Kong and Macau).