Obesity costs

Another day, another cost study. This time we're told that obesity costs New Zealand between $722,000,000 and $849,000,000.

Here's a thought experiment. Imagine that we wave a costless magic wand and every obese person suddenly has a BMI right in the middle of the healthy range. Would long term health costs increase or decrease? Well, we know that we'd save a bunch of current costs of treating diabetes. But everyone dies of something, eventually. And many of those somethings affect the Ministry of Health's budget. If the diseases from which the obese would otherwise die are less expensive than the diseases that do affect the obese, then obesity has net cost on the health system: the cost difference and time discounting. But we really should net out the costs that the obese would have imposed had they not imposed obesity-related costs.

Total health care costs were estimated at $623.9 million. The additional costs were costs of productivity losses associated with obesity: premature mortality and excess sick leave. Here's the two approaches they used.

The "Human Capital Approach" takes the present discounted value of all future earnings up to the age of retirement for those who died prematurely due to obesity. The "Frictional Cost Approach" takes three months' base salary assuming average salaries and workforce participation rates, then adding an additional training and recruitment cost of 97% of the FCA premature death cost. They added to both of these costs of short term absenteeism: the overweight and obese take an extra 0.4 days' sick leave on average in Australia; this was multiplied by obesity prevalence rates and average population daily wage rates. Depending on whether the FCA or HCA method were used, that gives us the additional $98-225 million.

But there are a few problems here too. Obesity is pretty concentrated in lower income strata. Some of this will be obesity issues resulting in lower salaries, but it seems awfully likely that the kinds of things that lead to obesity are also the kinds of things that lead to lower income. Using average wage rates and average workforce participation rates then is a pretty poor proxy for wage rates among those whose obesity issues were so severe that they died early as consequence. You could make a case for it where obesity caused the lower wages; when it's more a consequence of other factors that drive income, as well as having its own negative effect on income, not so much.

The study's authors note this:

With regards to the estimates of the lost productivity costs, it could be argued that obesity and overweight is more prevalent among low SES groups and therefore using an average wage to calculate the productivity cost estimates is not a true reflection of the costs. Our analysis was based on the availability of average wage and participation rates data in NZ.

But there's an awful lot of data out there that would let you scale wage rates among the obese. For example, the 2008/2009 NZ Adult Nutrition Survey. They sort people by NZ Deprivation Index 2006 quintiles, where 1 is least deprived and 5 is most deprived. 41% of Decile 5 females class as obese, with only 26.8% normal; 48.5% of Decile 1 females class as normal and 16.5% as obese. The male profile is flatter in income. Here's the table.

All you'd need to do is weight wages by relative quintile populations and quintile wage rates. You could at least get an average fractional multiplier to apply to the average age-wage-gender profile.

I wonder how many multiples of the total annual health care budget you'd get if you added up all the different "Costs to the health system" figures.

In related news, the Science Media Centre asked me for comment on a new metastudy on the effects of fat taxes. Here's what I provided; bits of this may be showing up in a few places.

"Taxes and subsidies to encourage healthy eating are notoriously difficult to administer in the real world. They're the kind of thing that sounds simple, but wind up being a bit of a compliance nightmare. Why subsidise fresh vegetables while ignoring frozen or canned vegetables, which are just as nutritious and favoured by poorer cohorts? Where then do we draw the line between a frozen unprocessed product and one that's too processed to count any longer? If minted peas count, where do we draw the line between a frozen vegetable and a processed food product? I'm reminded of a recent Australian case where a judge had to decide whether a mini-ciabatta counted as bread (GST-exempt) or as a cracker (subject to GST). The food importer flew in a bread expert from Italy as a witness in the case. These kinds of systems lead companies to find strong tax advantage in having their products being declared compliant. The accountants, lawyers and bread-deciders can do well out of it, but that's not necessarily to the national advantage. Denmark recently abandoned its one-year-old fat tax because of the administrative burden it placed on Danish firms."

"The paper finds a ten percent increase in the price of soft drinks via a soda tax would reduce soda consumption by 9.3 percent but would only reduce overall energy consumption by 0.2%. People who want to consume sweet things shift away from taxed sweet things to less-taxed sweet things."

"The paper focuses on the effects of tax and subsidies intended to affect consumer behaviour. Another way of tackling the problem is to look at how consumer behaviour changes with other changes in the relative prices of healthy and less healthy foods. Jonah Gelbach, Jon Klick and Tom Stratmann used American data from 1982 to 1996 to see whether changes in the relative prices of healthy and unhealthy foods over time across U.S. regions affected obesity rates. Because overall food prices vary by a lot more than any real world fat tax, their study gives us a way of examining what can happen with very large price increases. They found that the relative price of healthy food really does little to determine overall obesity rates. While we should not expect large health benefits from fat or sugar taxes, the administrative costs are real and substantial."

See also this Mercatus primer.