Short-sighted alcohol tax effects

CNN reports on a new study [pdf] in the American Journal of Public Health that notes that higher alcohol prices via taxes leads to the saving of lives. That spin of the results makes all of the difference-- the researchers correctly note that the increase in alcohol taxes leads to "immediate and sustained reductions alcohol-related disease mortality." [emphasis added] The real issue here is that taxation causes substitution away from the taxed good (beer, wine and spirits) and for the answer that everyone cares about-- what's the overall effect on mortality-- to become apparent, we'd need to look at the spillover effects of this tax into related activities. Are all of the high-value drinkers substituting into lower quality alcohol? Maybe they're drinking less and eating more potato chips on the couch watching football. The direct effect is clear, but the truth comes in looking at the secondary outcomes. The authors don't extend their research beyond alcohol, nor beyond drect alcohol related deaths by disease (i.e. drunk driving), so they aren't trying to overstep their bounds. CNN gave it a bit too much range, however.

For what it's worth, Figure 2 doesn't provide a compelling story. Personally, if I don't see it in the simplest of terms, I'm not going to buy any level of statistical fudgery. And at first glance, under the right circumstances, an increased propensity for alcohol-related death over the second span could generate the downward dichotomy at 1983 that the authors attribute to the tax shock. Also, the variance of the error terms doesn't seem to be constant across the span either. And are they drawing the last line in the thrid section with 8 data points over 2 years?

Again, I believe the paper doesn't try to say too much, but the policy implications of papers like this can't be underestimated if they get into the wrong, MADD-inspired hands.