| Summary: | The main objective to this meta-analysis is to estimate the price elasticity of alcohol demand after correcting for publication bias and to identify the attributes that explain the data heterogeneity. By employing a funnel asymmetry test, this study confirms and corrects for the presence of the publication bias and suggests that price elasticites for wine and spirits have decreased. This finding is in line to the changing drinking patterns, where wine and spirits become more attractive beverages then beer. A stronger publication bias has been confirmed also under the assumption of nonlinearity, which is derived by applying the "stem-based" method. To account for model uncertainty we use Bayesian model averaging and derive that the length of the data, highly frequency data and estimates under the specification for unconditional demand function, contribute the most to the data heterogeneity. Heavy drinkers are not responsive to the price changes, exhibiting a totally inelastic alcohol demand. Country development is important and low and medium income countries may affect the estimated alcohol price elasticities.
|
|---|
