Wages, Alcohol Use, and Smoking: Simultaneous Estimates

This paper estimates a simultaneous model of moderate and problem drinking, smoking, and wages using a random sample of employed Canadian men. The results indicate that sample selection into alcohol and tobacco use is not negligible. With all else in the system held constant, moderate and heavy drinking are both associated with considerably higher wages than abstention from drinking, whereas smoking is associated with lower wages. Allowing for feedback from wages to substance use is important: if wages are excluded from the substance use equations, the premium to heavy drinking disappears, the premium for moderate drinking rises, and the penalty to smoking is diminished.