Modeling the response to smoking bans in gambling venues

Presenter: Joseph Hirschberg, University of Melbourne

Abstract

The Australian state of Victoria has over 520 clubs and hotels (eating and drinking establishments) that contain more than a total of 25,000 electronic gaming machines (i.e. pokies, slots). In September 2002 a ban on smoking in the area where the machines are located was instituted. In this study we establish which characteristics of these venues led to changes in the revenue from these machines and thus the subsequent differences in the resulting changes in the gaming patterns in those areas where the smoking ban came into effect.

The relationship between habitual gambling and smoking has been established in a number of earlier studies. The link between changes in gambling revenue and the institution of venue smoking bans has also been reported in earlier studies both of this smoking ban (see Lal and Siahpush 2008) and of other bans instituted in other locations (see Glantz and Smith 1994). In this study we examine how the measured shifts in gambling revenues can be used to establish which characteristics of the gambling venue and the customers are most important in determining the magnitude of the shift. Using spatial relationships we construct a unique set of regional demographic characteristics based on the proximity of each gaming venue to surrounding communities. By using a panel data series of annual gaming expenditures from 1993 to 2007 for 65 municipalities we estimate a model designed to incorporate the spatial, time-series and random coefficients elements of this analysis. From these results we are able to attribute how different components of the customer base reacted to the smoking bans.

The results of this study can be used to better determine how this type of policy change can influence both gambling and smoking behavior. This is done by establishing if we can determine who changed their behavior and what the ramifications are for future policy shifts in this area.

To date the basic data set has been constructed and preliminary results have shown that the age and income distribution of the local area influenced the degree to which the shift has taken place. In addition, we found that the use of the spatially defined “market” area for the venues yielded much better results than the use of the aggregate municipal census data that does not account for the density of the venue locations in the Melbourne metropolitan area as opposed to the more rural shires. In addition, the regional data also allows us to control for the proximity of some venues to the neighboring state of New South Wales where smoking bans in gaming venues were not put into effect until after our study period.