Casino tourism, social cost and tax effects

Abstract

Local economic growth can be spurred by casino tourism, yet this may take place at the expense of the external regions where tourists live. We show that Pigouvian taxes should be imposed on gambling activity to attenuate its external cost. The tax may boost social welfare in the local and external communities even though casinos and tourists incur certain private losses due to their tax burdens. The tax can also mitigate gaming-biased unbalanced growth via resource reallocation, and improve the terms of trade for local welfare enhancement through tourism as an exporting industry. If taxes are collected from tourists but not fully refunded, local tax policy then plays the dual roles for both social cost reduction and public revenue generation. Our empirical study suggests the importance of casino taxation for preventing the ‘exported’ social cost from coming back to hurt the local economy, for too much such exportation may trigger visa policy restrictions by tourists' home communities.

Keywords:

• casino tourism
• social cost
• gambling taxation

Acknowledgements

This article is based on a research project funded by the University of Macau (under grant MYRG081 < Y2-L2>-FBA11-GXH). We also thank the IGS Editor and two anonymous referees for their valuable comments and suggestions, and Sara X. Chang for her excellent research assistance. The points made in this paper do not necessarily reflect the views of the University of Macau, and all errors, if any, are, of course, ours.

Notes

 1. The Dutch disease refers to a situation where strong growth in one dominant sector may have adverse impacts on other sectors in an economy. In the 1960s, large deposits of natural gas were discovered in the Netherlands, but the gas industry's expansion later led to the demise of a significant portion of the manufacturing sector due to the resulting surge in wages, energy costs and currency value. The Dutch disease is a widespread economic phenomenon: for example, as happened in Colombia during 1910–55 under the coffee boom that slowed down banana exports, and in Bangalore (India) recently because of the dominance of IT skills-based industries that damage other sectors. As pointed out by Nobel Economics Laureate Sir James Mirrlees (Duffin, 2007), the Dutch disease also exists in Macao as a gaming-biased, tourism-led economy, and has become increasingly serious; the rising ratio of gaming revenue to GDP from 41.7% in 2002 to 92.1% in 2011 clearly signals the rapid shrinking of all other sectors in Macao.

 2. A distinction between the private and social costs is needed for clarity in our analysis of casino taxation (Gu, 2007). As the sum of the private and external costs, social cost is widely misused in lieu of external cost when referring to a negative externality in the gaming literature. We will follow this tradition wherever no confusion is caused. A distinction between the social and private welfare of economic activity in one industry should be made both for that industry in a partial equilibrium analysis and for all industries in a general equilibrium study. Note that externalities generated from one industry may spill over into other industries in an economy. The pervasive negative impact of casino tourism is often overlooked in the literature on gaming-led economic development (Andersen, 1996).

 3. In the early 1900s, an economist named Arthur Cecil Pigou provided a simple but powerful idea: if an activity creates an external cost that is ignored by the activity performer, the government can induce him/her to act as if he/she bears all of the activity's social costs by imposing a tax (or fee) equal to the size of the marginal external cost at the efficient outcome. In recognition of Pigou's contribution, the use of taxes (or fees) to remedy negative externalities is referred to as Pigouvian taxation.

 4. All amounts of money in this paper are in US dollars unless otherwise stated.

 5. Unlike American players who have changed their attitudes away from table games toward more user-friendly, high-tech and low-pressure slot games, Chinese players stick to table games and traditional entertainment with little interest in computerized forms of gambling. In Macao, much of CGR is generated from VIP ‘high rollers’ who always focus their playing efforts on high-stake table games. CGR in the mass market also comes largely from table, not slot, games. There are certain factors that discourage slot play in Macao (Liu & Wan, 2011).

 6. There is a similar interpretation of estimate signs in Levitzky, Assane, and Robinson (2000), which are opposite to the usual expectation.

 7. Concretely, the saved external cost A (a gain) from the reduced gambling activity, including aa'cc' and ab'cd', is equal to aa'bb'cc'dd'a+abcd. The resultant combined loss B in the consumer and producer surpluses is ab'c” + ac”c' or ab'c'. Then, the net benefit (i.e. the gain minus the loss) from attenuating the external cost via casino taxation is A − B = aa'b+add' + abcd = a'cd'.

 8. This can be proved quickly by means of math. Suppose an IC is defined as U(λX, Y) = u _o , where 0 < λ ≤ 1 represents the adverse impact of external costs on the economy-wide utility. Private utility follows from λ = 1, while social welfare corresponds to λ < 1 with smaller λ indicating greater harm done to social utility by negative externalities. Clearly, |MRS _S | = λU _X /U _Y  < U _X /U _Y  = |MRS _P |, where marginal utility (MU) is positive and should be discounted by external costs in considering social welfare.

 9. The government gains revenue from a tax at the expense of consumers and producers, yet their combined losses exceed the tax revenue, so that there must be a pure loss that is not captured by anybody, from society's viewpoint. Such a loss is referred to as the deadweight loss of the tax or the excess burden of the tax.

10. The casino tax rate is high at 40% in the UK and Macao, 61% in Spain, 80% in Austria and France and 92% in Germany, but relatively low in other places, such as Australia (30%), Malaysia (25%), South Africa (21%), Singapore (15%) and South Korea (10%). In particular, the tax rate is very different between American states, e.g. 34% in Pennsylvania, 40% in Indiana and 50% in Illinois, but 6.75% in Nevada and 8% in New Jersey and Mississippi. These taxes are usually applied to gaming revenue, but player winnings may be subject to income taxes in some locations.

11. Three seasonal dummies (D ₁, D ₂, D ₃) included in the variance equation of the TGARCH(0, 1)-M model are not reported in Table 2. The estimated coefficient is significantly negative for D ₁ and D ₂ but insignificantly positive for D ₃, and all effects are very small in size. The lagged values of the regressand, (CGR( − 4), CGR( − 6)), along with the dummies, (D ₁, D ₂, D ₃), enter the mean equation of the GARCH(1, 1)-M model for estimation, but their results are excluded from Table 2. CGR( − 4) has a significantly positive effect on current CGR but the effect of CGR( − 6) is insignificantly negative. The estimated coefficients on the dummies, all negative, are significant for D ₂ and D ₃ but insignificant for D ₁. The probability for both the Obs-R ² and F-statistics has a value around 0.95 for one lag in the two GARCH models, showing that the null hypothesis of homoskedasticity cannot be rejected so that there is no longer ARCH(1) effect when estimation is conducted.