Gambling taxes are highly regressive—it targets the poor and uneducated and violates the principles of vertical equity.

This study concludes unambiguously that the tax imposed on gambling activities in Switzerland is regressive. Lower income individuals contribute proportionally more to the total state revenues than do higher income groups. Indeed, the Suits index shows a negative value of –0.184, indicating a clear regressive tax pattern. The regression analysis corroborates this first result. The income elasticity of 0.199 is clearly and statistically lower than one, providing further proof of a regressive tax pattern. Decreasing income by 10% results in a decrease of only 1.99% in the gambling budget. Therefore, the economic incidence clearly shows that the final distribution of the tax burden is proportionally larger for individuals with lower incomes. Thus, this tax violates one of the main qualities of a good tax in Switzerland, namely vertical equity. Vertical equity suggests that individuals with higher income should have a larger tax burden. However, this is obviously not the case in the present study.

Another type of inequality is related to educational level. In the regression analysis, this coefficient is found to be negative. This pattern shows that the more educated an individual is, the less money he/she will spend on gambling. According to our analysis, state gambling revenues are expected to rise if the population that has access to gambling is poorly educated. These two results lead to the surprising conclusion that the revenues to the state from games of chance will be higher if the population is poor and not well educated. This finding is in line with the findings of Abt, Smith, and Christiansen (1985).

Why this subpopulation may be encouraged to spend money on gambling is addressed by Wisman (2006). He highlighted three reasons why poor people may be more “vulnerable to the promise of getting rich by ‘investing’ in gambling” (Wisman, 2006). The first reason is related to human capital. Because they have less access to high-quality schools, poor people have difficulty developing human capital, finding well-paid jobs or accessing credit to start businesses. Second, due to their poor human capital, low-income populations have a higher discount rate, and so they are more oriented toward the present. Poor people generally have a shorter life expectancy and live in a culture that is more preoccupied with the present because of uncertainties about the future. The third reason is that because poor people possess little control over their everyday lives, work or financial means, choosing lucky numbers provides them with a sense of control and participation. The first reason highlighted by Wisman may not be applicable in Switzerland because high-quality schools are accessible to all socio-economic classes. However, the overrepresentation of lower income groups in gambling expenditures may be a combination of a lower discount rate and the quest for the artificial feeling of control provided by gambling.

Several limitations to this study must be considered when interpreting the results. First, we focus our analysis on the tax incidence for gambling in Switzerland. We do not account for the budgetary incidence of this activity. This incidence is difficult to compute for Switzerland because many expenses derived from gambling tax are not earmarked. While part of the tax imposed on casinos is transferred to the OAI, the other part is not earmarked and is given to the cantons, which are free to use this money as they wish. Due to this specific repartitioning, considering the benefits of casinos may decrease the regressivity of the present study. However, this observation does not hold for the tax lottery redistribution. Because an important part of the revenues collected by this tax are redistributed for cultural or leisure activities, the regressivity may increase. Therefore, if we properly account for the benefits of this tax, the budgetary incidence may differ for casinos and lotteries. Another limitation of this study is related to how individuals reported their gambling expenditures. Rather than a continuous variable, the answers were designed to be grouped into a categorical variable with six categories, which reduces the precision of the answers. Moreover, the question asked how much the respondent spent on gambling. As shown by Blaszczynski et al. (2006), this question is biased and is subject to ambiguity because the question “how much money do you spend gambling?” may be interpreted in different ways and may lead to misinterpretation. Generally, this question is asked to determine the difference between the amount of money a person has at the beginning of a gambling session and the amount of money he or she has at the end. In other words, the information of interest is the net value of the money spent. Blaszczynski et al. (1997) concluded that in response to this question, two-thirds of the respondents interpret this as their net expenditure. The rest of the respondents relate their answer to turnover or other alternative interpretations. In this study, we assume that the answer to the question asked in the SHS reflected the net expenditure. Finally, we are unable to distinguish between casino and lottery expenditures. Because the incomes of these two categories of gamblers may be different, grouping these two types of expenditures may have an impact on the final results.

This study is in line with the international literature that finds that the gambling tax on casinos and lotteries exhibits regressive patterns (Mason et al., 1989). This conclusion was expected because several international studies have shown that poor and uneducated individuals tend to gamble more and to experience gambling problems more often. Thus, it is not unexpected to find that these individuals spend a proportionally larger part of their revenue on games of chance in Switzerland and that this tax violates vertical equity.