Taxing tobacco

Recent UM Ph.D. Adam Cole points me to this interesting study on tobacco taxes produced by the US Treasury Department.

For me, the most interesting bit is in the appendix. The study authors would like to estimate the volume of cigarettes consumed in the US on which tax has not been paid. One reasonable way to do this starts with using administrative data to determine the number of cigarettes on which federal taxes are paid. Administrative data works well here because federal taxes are paid at the production site and careful records are kept. The number of cigarettes consumed, in contrast, can be estimated using data from a nationally representative survey, in this case the National Health Interview Survey (NHIS), a data set widely used in research in health economics.

The problem is that when you do this, you discover that the number of cigarettes consumed lies well below the number on which taxes are paid - see Table 4. Given that cigarettes degrade in quality when stored reasonably rapidly, which indicates that the excess cigarettes are not being stored by consumers, this finding suggests that the NHIS measure of cigarette consumption has a downward bias, presumably due to "non-classical" measurement error in smoking incidence and/or (probably and) in number of cigarettes smoked conditional on reported incidence. The study does about the only thing one can do in this situation, which is to present estimates - see Table 5 - of the lost tax revenue as a function of assumptions about the downward bias in the NHIS consumption measure. One obvious recommendation here is to improve our knowledge of the degree of downward bias in smoking consumption measures by doing some sort of validation study. More generally, this study provides a nice example of how research on measurement, which might seem rather arcane, actually has a real world payoff.

The other bit of the study that stands out to me is this:

Recommendation 3: Allow enforcement officials to pay investigative expenses with proceeds gained through undercover operations.

Problem: Currently, tobacco tax enforcement programs are funded principally through agency appropriations or from forfeiture proceeds arising from asset forfeitures in concluded criminal cases. Additional funding through the use of proceeds gained through undercover investigations would expand investigative resources without the use of additional appropriated funds.

Recommendation: Existing law should be amended to authorize TTB to use proceeds gained from undercover tobacco tax enforcement operations to fund its investigations.

I disagree with this bit for the same reason I dislike asset forfeiture laws. I do not think that law enforcement should be a profit center for the government. Making it such strongly enhances the incentives for misbehavior by law enforcement officials.