Re-examining the economic determinants of alcohol consumption in Canada: controlling for the presence of common correlated effects

ABSTRACT

Alcoholic beverages represent both an important source of revenue and a driver of expenditure for provincial governments in Canada. As a result, the pricing of alcoholic beverages has substantial public policy implications. In this context, we re-examine existing work estimating the demand for three classes of alcoholic beverage (beer, wine, and spirits) by controlling for common correlated effects. The results reveal that any conclusions regarding the government’s ability to influence alcohol consumption is sensitive to the assumption that the relationship between the demand for alcoholic beverages and economic variables is identical across provinces.

KEYWORDS:

• Panel data
• common correlated effects
• alcoholic beverages
• price elasticity

JEL CLASSIFICATION:

• C13
• C23
• D12
• I18

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

¹ When considering the policy implications of alcohol sales, it is important to note that through the provincial liquor boards, provincial governments’ markup and set the wholesale price of alcoholic beverages in Canada.

² As with any good, depending on the price elasticity of demand, an increase in the price of any class of alcoholic beverages may either increase or decrease the revenue collected.

³ All significant at the 5% level.

⁴ Consumption data from 1982 to 1988 are drawn from The Control and Sale of Alcoholic Beverages (63-202-X). Data from 1989 to 2013 are taken from Cansim Table 183-0015. It should be noted that change in reporting standards makes data from 2014 incompatible with previous data for our purposes. Price Data is drawn from the CPI – Cansim Table 326-0020. Income and unemployment are drawn from Cansim Tables 384-0037 and 282-0001, respectively.

⁵ This test statistic is based on the standard normal distribution.

⁶ It is important to note that seemingly unrelated regressions may also correct for the presence of CCE. However, this approach does not produce results, which are directly comparable to Ogwang and Cho (2009).