The effect of drinking and smoking on the labour market outcomes of low-income young adults

Abstract

Among adults the causal ‘drinking bonus’ and ‘smoking penalty’ have been estimated to be as large as 12% and 24%, respectively. The magnitudes of these effects compare in size with many active labour market programs targeted at low-income young adults. This article extends the literature by examining these relationships in such a group. Somewhat surprisingly the data indicate that just as in the greater population young drinkers have more favourable labour market outcomes than nondrinkers. However, when a fixed-effects approach is used to identify causal impacts there is no evidence that drinking has a positive impact on labour market outcomes and some evidence for negative returns to drinking. The smoking penalty is estimated to be much smaller among this group and not statistically significant. Finally, estimates suggest that the observed correlations between consumption and labour market outcomes are biased by unobserved characteristics of the individual as well as unobservables that change over time that are likely causing the treatment decision.

Keywords:

• alcohol
• smoking
• labour market

JEL Classification:

• I19
• J20
• J30

Notes

¹ 25 July 2007 Gallop poll. http://www.gallup.com/poll/28213/Latest-Gallup-Update-Shows-Cigarette-Smoking-Near-Historical-Lows.aspx#1.

² US Department of Health and Human Services. The Health Consequences of Smoking: A Report of the Surgeon General. US Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2004.

³ ‘Happy hour for drinkers' wages’, CNN Money.com, 14 September 2006.

⁴ A government white paper, US Department of Labor (1995), summarized literature on the economic impacts of employment and training programs and found successful programs increase earnings between 5% and 15%. The returns to Job Training Partnership Act (JTPA) for disadvantaged adults have been estimated at 10% for men and 15% for women. Studies of long-term classroom training estimate returns between 6% and 12% for each year of post-secondary education. Job Corps has been estimated to increase weekly earnings for disadvantaged young adults, 10% to 15%.

⁵ Deaton (1974) and Deaton and Muellbauer (1980).

⁶  The Deformed Transformed.

⁷  Address to the Washington Temperance Society, Springfield, Illinois, 22 February 1842.

⁸ Not all studies have found a negative relationship between abuse and labour market outcomes. In a working paper by Chatterji and DeSimone (2006) find that 10th graders who binge drink experience better labour market outcomes 10 years later.

⁹ Bettendorf and Dijkgraaf (2005).

¹⁰ As early as 1930 Weber argued that the Protestant religion stimulated capitalism and resulted in higher incomes. The so-called ‘Protestant work ethic’ has not been the only hypothesis however. Greater belief in religion may increase the utility derived from ‘good works’, thus causing a substitution away from observable income towards unobservable ‘good works’.

¹¹ MacDonald and Shields (2001) assert that they expect nonacute illness like diabetes and asthma to not affect occupational attainment though both can be associated with work interruptions and frequent medical appointments likely to affect ones career progression.

¹² Fogarty (2004) and Huang et al. (2004).

¹³ In fact Auld's estimated coefficients on price in the first stage are not statistically significant.

¹⁴ Flores-Lagunes (2007).

¹⁵ Alternatively drinking could be modelled as a continuous or categorical variable to capture different returns to the quantity of alcohol consumed. When this is done with my data I cannot reject the hypothesis that the relationships between labour market outcomes and different levels of treatment are the same.

¹⁶ All observables (except drinking and status and Job Corps eligibility) are recorded as of quarter 0. This makes them pretreatment observables in relation to changes in drinking and smoking behaviour. Job Corps eligibility is 0 for all participants in quarter 0 and 0 or 1 depending on randomization status in later periods.

¹⁷ All estimates use robust SEs to account for heteroscedasticity, clustered at the participant level. The full coefficient results for this and later regressions are available upon request.

¹⁸ The model was relaxed to test if the effect of starting or stopping consumption was statistically different in absolute value terms. I found that it was not.

¹⁹ Results are not sensitive to defining treatment as consumption in both periods.

²⁰ Of course quitting smoking can still reduce current expenditures on an unhealthy habit and in turn reduce future medical related expenditures.