The effects of cigarette costs on BMI and obesity
About 30% of Americans are currently obese, which is roughly a 100% increase from 25 years ago. Public health officials have consequently become alarmed because recent research indicates that societal costs of obesity now exceed those of cigarette smoking and alcoholism. Cigarette taxes may have exacerbated the prevalence of obesity. In 1964, the US Surgeon General issued its first report relating smoking and health, and since that time, federal and state governments have increased cigarette taxes in a successful effort to reduce cigarette smoking. However, because cigarette smoking and obesity seem inversely related, cigarette taxes may have simultaneously increased obesity.
This paper examines the effects of cigarette costs on BMI and obesity and finds that they have significant positive effects. This paper attempts to reconcile conflicting evidence in the literature by controlling more carefully for correlation with state-specific time trends using panel data.
Results indicate that the net benefit to society of increasing cigarette taxes may not be as large as previously thought, though this research in no way concludes that they should be decreased to prompt weight loss.
That’s some caveat (the italics are the author’s). I’d like to write a paper requiring that sort of defence mechanism, one day.
USA Today’s coverage of the Mississippi Dumb Obesity Bill even had a map:
That’s the geo-location of obesity in the US, according to the CDC. What isn’t already phenomenally expensive for the State is mopped up by you – Americans are spending some USD30bn (USD41bn in today’s dollars) on weight-loss wastes of time.
So to Charles Baum and his paper (BYO Orchestra – sorry). His paper argues that taxation on cigarettes contains a heretofore hidden opportunity cost/unintended consequence: cigarettes, as diet-suppressants, are a pathway to non-obesity (for want of a better description). There is a social gradient to cigarette-smoking (poor people do it more) and to obesity (poor people are it more, for a variety of income-related reasons). Increasing cigarette prices (Healthy People 2010, for example, wants taxation to be USD2 per pack – a nearly 100% increase) will therefore have some effect on decreasing the quantity of cigarettes demanded – it will also (according to Baum) have a subsequent effect on obesity.
Controlling for a bundle of covariates (not reported: click for a larger image):
We see our outcome. The models, spread across several tables, are, however, quite unstable (in that the results are very sensitive to specification), so there is a significant residual question over the value of the information we’re receiving here. Frequentistedly (I made that up) we observe statistical significance, but my Bayesian switches all flipped with these results. I’m not convinced even by the descriptive statistics that suggest heterogeneity amongst obesity and cigarettes prices/taxes: there are just too many state and environmental variables (I think – although he does control for these in the regression analysis).
One of the things that I didn’t see mentioned in this paper was the effect on smoking intensity. The impact of cigarette taxation not only on consumption but on intensity is known – it is the intensity, for example, that contributes a great deal to the carcinogenic effect of smoking.
I would expect, then, that the compensating increase in smoking intensity would also have an affect on BMI and obesity according to the Baum models, but such a confounder was not mentioned. It seems that, instead, a linear relationship is assumed between the consumption of cigarettes and the intensity with which they are smoked, rather than the more likely concave function.
Amongst other things this, if significant, negates the entire argument that Baum is trying to make. The kindness of tax increases is questioned precisely because the increase in intensity is more harmful: would the same increase in intensity not also leave BMI either unchanged, or at least changed in an unkown direction? Getting back to the sensitivity of the results to model specification – this would explain a lot of that. The actual intake of nicotine, etc. per smoker is simply not being measured, but we know that it is not constant; nor is it linearly-determined by either cigarette consumption or cigarette prices.
It’s an interesting idea. I think the paper is missing some critical pieces, but it’s certainly an interesting policy angle to examine.