Data/Method: We focus on the impact of the total number of cigarettes smoked per day on individuals’ body mass index and on the likelihood of being overweight and obese. A primary challenge in estimating the relationship between smoking and body weight is to account for the unobserved factors that may simultaneously affect smoking and body weight. Unobserved personality traits (such as addictive propensities) or institutional and environmental factors (such as a local environment that fosters a sedentary life style, low food prices, high concentration of fast food restaurants, low cigarette prices, availability of cigarettes) can be correlated with both smoking and weight outcomes. Further more, it is also possible that there exists a reverse causality or simultaneity bias in the estimates, i.e. obesity may induce cigarette smoking as a mechanism for weight control or vice versa. In other words, smoking measure could be endogenous.Instrumental variables estimation with province level fixed effects is used to correct for the endogeneity of cigarette smoking.
Results: We find a moderate negative relationship between cigarette smoking and body mass index. Smoking is also negatively related to being overweight and obese, but the marginal effects are small and statistically insignificant for being obese. Quantile regression analyses reveal that the association between smoking and body mass index (BMI) is quite weak among subjects whose BMIs are at the high end of the distribution but are considerably stronger among subjects in the healthy weight range. These results thus reconcile an inverse average effect of smoking on body weight with the absence of any significant effect on obesity. From a policy perspective these findings suggest that, while smoking cessation may lead to moderate weight gain among subjects of healthy weight, the effects on obese subjects are minimal and should not be expected to increase obesity prevalence rates.