Why do drinkers earn more? Job characteristics as a possible link

After some initial controversy, an inverted U-shape relationship between the consumption of alcohol and earnings seems to be an established result, at least in North America. It has been dubbed a "drinking premium", at least in the lower portion of the consumption curve. It is still unclear, perhaps even counter-intuitive, why such a drinking premium exists and the literature suggests it is not causal but results rather from selection effects. We suggest here that part of the premium is linked to occupation: some occupations pay better, controlling for the usual human capital determinants, and also attract drinkers or induce workers to drink more. Using a sample of full-time employed or self-employed individuals aged 25-64 and not in poor health from the 2015-16 Canadian Community Health Survey (CCHS), we confirm the existence of a drinking premium and a positive return to the quantity or frequency of drinking up to high levels of consumption. Using information on jobs held by respondents, linked to a data set of job characteristics, we find that controlling for job characteristics reduces the premium or return to drinking by approximately 30% overall, and up to 50% for female workers.