The demand for alcoholic beverages in British Columbia : an imperical investigation
Date
1986
Authors
Enemark, Philip Gordon
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
In comparison with Canada as a whole, British Columbians are average consumers of beer per adult, but are above average consumers of the more potent wines and spirits. In terms of pure alcohol equivalency per adult therefore, B.C. often tops the remaining provinces. There has been little investigation of the determinants of alcohol consumption in B.C. However, some studies do exist for both Canada and abroad. This thesis extends the methodology of these studies to an analysis of alcohol consumption in B.C. The B.C. results are then compared with those for other jurisdictions, with extra attention being paid to the results obtained by the Canadian studies.
Three demand equations are specified, one to explain the per adult consumption of each of beer, wine and spirits. The model is multiplicative and Ordinary Least Squares regression analysis is applied throughout. The analysis gives special attention to the derivation of price and income elasticities. Due to the importance of habit persistence the multiplicative model utilized is a variation of the stock adjustment model. Specifically, alcohol consumption is assumed to be continually adjusting to some long run equilibrium value with habit persistence impeding this ongoing process.
Multicollinearity is found to be a major problem and therefore ridge regression is used. Variables whose coefficient estimates carry high standard errors and/or unexpected signs and are not subject to considerable multicollinearity are then excluded from the three equations. Final specifications are decided upon, each of which is different for the three beverages analyzed. In all three equations, robust parameter estimates with the expected signs are obtained for own-price, income, the beer strike variable, and lagged consumption. The immigrants variable is also an apparently important influence on consumption in the negative direction. The legal drinking age dummy carries the expected positive sign and is statistically discernible in the wine and spirits equations. The only apparently significant cross-price effect in the expected direction is that for wine price in the spirits equation. The time trend variable remains only in the wine model, and possesses a positive sign. The inflation and unemployment variables only have statistically identifiable effects on spirits consumption, the former being positive and the latter negative.
The respective estimated short run own-price elasticities are -0.43, -0.75, and -0.26 for beer, wine, and spirits. Income elasticities are 0.12, 0.64, and 0.18. The spirits own-price estimate is very low in comparison with most of those calculated elsewhere. The other elasticity estimates above and all of the remaining results are in general accordance with other studies.