Monday 10 July 2017

Counting calories

Alcoholic beverages in New Zealand don't require nutritional labeling. It's an anomaly relative to other food and drink, but does it make sense to add labeling requirements to alcohol just to have consistency across products?

I'd missed NZIER's assessment when it came out but had it pointed to me last week.

NZIER carefully goes through the complicated chains necessary for labeling to have an effect, and why the likelihood of any effect at all is very small, and how the international studies struggle to find any effect. Because they look not to have been able to see any way of finding an effect, they instead asked a different question: How big would the reduction in obesity have to be for compulsory labeling to pass cost-benefit?

After answering that question, they warn that their numbers are likely to be lower bound figures: more people would likely need to shift weight as consequence of the rule than they're expecting. Why? Because the obesity cost estimates are an average over the category, and those at the thinner end of the category (and less expensive end) might be the ones more likely to shift. I'd also want to check whether the obesity cost figure is incremental to any overweight cost figure, or is the total cost of being in the category relative to someone of normal BMI - because otherwise benefits may again be overstated.

They benchmark cost-effectiveness not against any cost-benefit analysis, but against existing programmes. Health Star was justified on similar basis, they report, and it was estimated to be cost-effective if obesity and overweight dropped by 0.04%, or 1513 people per year. Even that isn't a clean comparison though because the Health Star figure looked at changes in obesity and overweight, not just obesity; this study just looks at obesity.

And their winding up is pretty clear on the caveats. They give zero assessment of whether the policy would be cost effective, just the number of people, as a minimum, who would have to stop being obese as consequence of the policy for the policy to be effective. They provide no indication of whether that number could be expected to be reached, and specifically note that they cannot estimate it because there is no basis on which they could do so. And while they include some high-benefit cases based on purported welfare effects of obesity rather than just fiscal cost, they also have a whole appendix section on why they think those costs are overestimated.

One caveat they didn't put up: you can get perverse effects if you mandate putting up calorie counts while banning advertising of health benefits of moderate consumption. Other foods that have positive health consequences for moderate consumption at least can let people know about it.

Consider the kind of person who has zero clue that alcohol contains calories. If we have low information consumers, why would we expect them to be poorly informed only on the calories margin? Suppose that person also does not know that moderate drinking reduces all-source mortality risk by 16% or thereabouts, but has heard the repeated warnings about heavy drinking. Our consumer, let's say, is consuming 3 standard drinks per week and, after hearing about the calories, drops to 2. Knowing about the health benefits of moderate drinking but not the calorie count might have had the person shift from 3 to 4; knowing about health benefits and calories might have had the person stand pat at 3.

If health-conscious but poorly informed people who read labels are the ones most likely to change behaviour, they're less likely to be heavy drinkers to start with (because they're health conscious). If that's the case, then labeling could have the perverse effect of most substantially reducing consumption among the cohort receiving net health benefits from consumption, and that effect would need to be weighed in (and is here ignored).

It would be mildly fun to do the same extreme bounds style analysis to show how many thin moderate drinkers would need to shift to abstinence for every obese person who became not-obese for the policy to wind up doing harm.

I also wonder about trade effects if imported products would be subject to compulsory labeling but are not currently labeled. We could then be back in the annoying case where importers have to stick stupid labels on everything, like they do for standard drink counts. I don't know whether places from which we currently import have labeling requirements that would conform to what NZ/AUS would be thinking about, or whether this would be additional. NZIER notes no trade hassles if the thing's non-discriminatory, but you could get effects on competition and reduced access to imported products, which would have harms for consumers.

In any case, it's tough to see any strong case for mandatory labeling in that report.

No comments:

Post a Comment