An econometric study of alcohol consumption and associated risky and antisocial behaviour

This thesis, using two national individual-level datasets, undertakes an investigation of Australian consumers’ irresponsible behaviour with alcohol abuse and Australian households’ alcohol purchasing behaviour, by means of microeconometric modelling and demand system modelling techniques. The four self-contained but related studies carried out in this thesis provide valuable empirical evidence to inform alcohol tax policy formulation and other educational and regulatory measures. This thesis also involves a methodological extending to the literature in the area of demand system estimation. The first chapter presents individual level evidence from Australia to examine the factors associated with individuals’ participation in binge drinking and several alcohol-related antisocial and unlawful behaviours. In particular, to quantify the negative externalities of excessive alcohol consumption by product forms, the primary focus lies in the link with eleven types of alcoholic beverages. The role of binge drinking in increasing the likelihood for engaging in these drinking related negative behaviours is also examined. In this chapter, individual level data from a national representative survey are used and a system econometric model that allows unobservable factors for all negative behaviours to be correlated is employed. Potential mis-reporting in the self-reported data is accounted for. Results provide valuable evidence for more effective alcohol taxation as a tool for correcting differentiated negative externalities by beverage types. In the second chapter, a multinomial Logit modelling approach is employed to study Australian households' major alcoholic beverage purchasing patterns, in order to identify their relationship with certain households’ socioeconomic and demographic characteristics. In particular, eight alcohol purchasing patterns are identified from the data. The individual-level sample data used in the chapter are drawn from the latest two waves of the Australian Household Expenditure Survey (HES). The results achieved in this chapter are interesting, in the sense that they provide valuable information for the formulation of targeted educational campaigns or government economic policies. The third chapter in this thesis provides the much needed information in regard to alcohol demand elasticities, to inform tax policy. Specifically, own and cross-price elasticities and expenditure elasticities of demand for beer, wine and spirits are estimated. In addition, this chapter has also examined the substitutability of alcoholic drinks with non-alcoholic drinks by jointly estimating the demand elasticities for non-alcoholic drinks. As opposed to some other previous studies in the literature which use aggregated national accounting data, using micro level survey data, this chapter provides crucial information on how households’ certain demographic and socioeconomic characteristics relate to their alcohol purchasing behaviour. To deal with the difficulty of excessive zero expenditure observations in the individual level data, a segmented demand system estimation strategy is proposed. In particular, a demand system model is respectively estimated for each of the sixteen purchasing pattern groups identified from the data. Following the recent literature on the statistical approach to modelling the zero expenditure generating process, a demographically extended demand system approach within a multivariate sample selection framework is also specified and estimated. For comparison, a naïve demand system approach is estimated as well, which results in biased estimates. In this last chapter, a class of demand systems based on simple parametric specification of the indirect utility functions, but allowing for the parsimonious imposition of global regularity, is proposed. Demand systems in this class are completely flexible in rank, i.e., can be potentially specified to acquire as large a rank as required in empirical work. They also exhibit a clear and reasonable homothetic asymptotic behaviour. In an empirical application using Australian data, several examples from this class are estimated and compared with some popular alternatives in the literature.