Consumer behaviour in the scratch card market: a double-hurdle approach

Abstract

This paper investigates how various socio-demographic as well as psychological factors are related to the purchasing frequency of Le Millionaire scratch card in Mauritius. In order to obtain a refined description and understanding of consumer behaviour, special attempts have been made to control for the personality traits of an individual. The OCEAN model of personality is thereby considered. We employ cross-sectional micro-data acquired from a household drop-off survey with a sample size of 1135 observations. Within a double-hurdle framework, gender, residential location, educational level, household income, alcohol consumption, other forms of gambling activities, and personality traits are found to be determining factors affecting an individual's decision to gamble in the scratch card market. Our findings help us to draw attention to the consequences of risk-taking behaviours and how personality-specific policies can be designed to influence gambling behaviour.

Keywords::

• gambling
• household drop-off survey
• double-hurdle model
• personality traits
• scratch cards

Jel Classification::

• C23
• L83

Notes

Special thanks are due to Carlos O. Criado, Robert Lundmark, Thomas F. Rutherford, Kristoffer Bäckström, Alex Blaszczynski and three anonymous referees for their comments and suggestions. Any errors are the responsibility of the authors.

^1. According to Potenza, Fiellin, Heninger, and Mazure (2002, p. 721), ‘gambling can be defined as placing something of value at risk in the hopes of winning something of relatively much greater value’.

^2. See http://www.crown-lodge.net/english/private-lodge-champ-mars-mauritius.html

^3. The first lotto draw was held on 7 November 2009. See http://www.lotteryinsider.com/lottery/mauritiu.htm

^4. See http://www.wildjackcasino.com/history-scratch-cards.html

^5. Roberts (2009, p. 140) defines personality traits ‘as the relatively enduring patterns of thoughts, feelings and behaviours that reflect the tendency to respond in certain ways under certain circumstances’ (emphasis in the original paper).

^6. These studies relate to wage determination (Braakmann, 2009), female labour force participation (Wichert & Pohlmeier, 2010), worker self-selection (Dohmen & Falk, 2010), marginal utility of income (Boyce & Wood, 2011) and unemployment duration (Uysal & Pohlmeier, 2011).

^7. Impulse buying arises when a consumer has a sudden irresistible urge to purchase an item, while compulsive buying occurs when there is repetitive and excessive shopping which may be due to boredom, tension or anxiety (Solomon, 2002).

^8. See http://business.mega.mu/2011/05/16/nearly-rs500-million-were-spent-millionaire-cards/

^9. The questionnaire is available upon request.

^10. Some respondents reported figures such as Rs 20.00 or Rs 40.00 with respect to Le Millionaire expenditures. These values may have been confused with expenses on other types of instant scratch lottery tickets (e.g. La Faya scratch cards which costs Rs 40.00) or gambling activities (e.g. lotto playing where the minimum bet is Rs 20.00). Also, the minimum age to gamble in Mauritius is 18 and some respondents were found to be under this age. Any flawed or partial responses were excluded from the final dataset.

^11. A Tobit model which includes all the personality traits has also been run. This model relies greatly on the assumption of a normally distributed error term and the validity of this assumption can be tested. Vincent (2010) proposes a Box Cox Lagrange Multiplier (LM) test for the null hypotheses of the standard Tobit model against the alternative of a more general non-linear specification. Failure to accept the null hypothesis implies the rejection of the Tobit model. The critical values for the Box Cox LM test are obtained through 5000 bootstrap replications and the computed test statistics is 41.305. The null hypothesis is strongly rejected.

^12. The presence of outliers can lead to type I or type II errors. However, arbitrary omission of observations can lead to sample selection bias (Wen, Tsai, Wu, & Chen, 2013). Approximately 0.61% of the respondents report a weekly expenditure on scratch cards of Rs 1000.00 or more. To check the influences of outliers, the double-hurdle models were rerun without these observations. Results remain roughly the same.

^13. The coefficients of determination are quite low and this is not unusual in cross-sectional analysis. For instance, while employing 1912 observations to run an ordered probit model, Grytten, Carlsen, and Skau (2009) compute a McFadden R² of 0.17.

^14. To examine any quadratic effect as postulated by Hirschi and Gottfredson (1994), an age-squared variable is included in the model. No significant impact on gambling decisions is uncovered.

^15. The marginal effects denote how much the conditional probability of the outcome variable changes when there is a change in the value of a regressor, ceteris paribus. Assuming a probit model, these can be derived from the standardized normal distribution:.

^16. According to Cox et al. (2000), smoking in Mauritius has been decreasing across all age and ethnic groups and across different levels of income and education.

^17. Decision regarding other forms of gambling (e.g. football betting, horse racing, lotto playing, etc.) may be determined simultaneously with the scratch-off gambling decision due to the underlying characteristics influencing the decision to gamble in any form. For instance, individual A may be intrinsically less risk adverse than individual B and each individual is facing an underlying risk preference. The other gambling variable attempts to capture some, if not all, of those intrinsic influences. This may produce endogeneity bias and thus the coefficient on one variable (or all) could be overstated. The lognormal double-hurdle model is rerun by excluding the other gambling variable and no significant change in the other variables is detected. Any bias coming from the other gambling activities variable can be considered as minimal.

^18. Westernized countries generally have higher sensation-seeking scores than Asian ones (Schultz & Schultz, 2008).

^19. Humphreys et al. (2010) and Tan et al. (2010) investigate only the socio-demographic factors as determinants of gambling in their study.

Additional information

Notes on contributors

Vishal Chandr Jaunky

Vishal C. Jaunky received his PhD in Economics from the University of Mauritius in September 2009. He worked as a lecturer in Economics at the University of Mauritius from September 2008 to December 2010. He was a postdoctoral fellow in Energy Economics at ETH Zürich from January 2011 to December 2012. He is currently undertaking further postdoctoral research in Forest Economics at the Luleå University of Technology. His research revolves around natural resource, energy and environmental economics and more recently behavioural economics.

Bhooshan Ramchurn

Bhooshan Ramchurn received his BSc in Economics and Finance from the University of Mauritius in October 2012. He is currently employed at the Mauritius Union Assurance Co. Ltd as Fund Administrator in the Investment Department. In his latest research, he focuses on consumer behaviour in the food and gambling markets.