How expensive is a can of Coke if social cost is priced in? | Hacker News Reader

How expensive is a can of Coke if social cost is priced in?

kurtsp.com

34 pointsarikr14 comments

Multiply the cost of one case of diabetes ($85,200) by the marginal chance of developing the disease (18%), and divide by the number of 12oz cans consumed over that decade (2555) gives $6.00.

The marginal increase is not 18%, it's 18% of the base rate.

Agreed.

The base rate of getting type 2 diabetes is 11%. So if you increase your consumption of sugar sweetened beverages to one serving per day above the average, your chance of getting diabetes is now ~13%.

You increase your absolute risk by 2%. So the average marginal healthcare cost is $85,200*.02 = ~$1700.

I'm not sure how he goes from 1 serving per day to 2555 per decade. I would assume it would be ~3650 servings.

So for every soda you drink, it would add $0.47 in additional medical costs.

But the study also controls for adiposity since people with higher body fat are more likely to get diabetes and are also more likely to drink sugar sweetened beverages.

If you control for that then each drink adds $0.21 in additional medical costs.

> I'm not sure how he goes from 1 serving per day to 2555 per decade. I would assume it would be ~3650 servings.

3650/30*21 = 2555 - 1 serving per work day?

A serving is ~3/4 of a can.

Yeah, that makes more sense!

I tried to read the research article, but I'm not sure about the notation. One can of coke increase the risk

a) from x% to x+18%?

b) from x% to x*(1.18)%?

The paper's conclusion that a can of coke per day is associated with a 'relative risk' of 1.18. That means X% -> x*(1.18)%, where X is whatever the baseline risk is. The blog introduces the result as increasing 'the likelihood of developing diabetes by 18%', which means the same thing.

It then incorrectly translates this to a 'marginal chance of 18%' -- which means X% -> X+18%. And the math uses that.

X%->X+18% is 'a marginal risk of 18%', or 'an increase of 18 percentage points' or 'a 0.18 increase in absolute risk'.

Neither seems reasonable. The a is definitely wrong, or 6 cans would give you diabetes guarantee, which is wrong. The b grows slower, but is still uncapped and speeds up as it grows.

The 18% transformation is associated with the change in risk due to a serving of coke every day for a decade. Option b) is correct.

I forgot to add "per day" in my comment. (I was aware of it, I just forget to write it.) So option b) is almost correct.

I read this article expecting to hear about negative externalities, but it seems the social cost described here is entirely in "private costs", at least under the assumption of no government-sponsored healthcare.

Would be interesting analysis if done correctly. Unfortunately author mistakes absolute for relative risk. To be fair, this isn't an infrequent thing.

What if you live in an economy with an NHS, and the direct and indirect costs of diabetes mediation are less? (not that it's good, but it becomes an argument in the application of indirect cost into pricing, which breaks the macdonalds burger index apart into shards of social policy costs)

that's okay because they're going to pass those costs onto the community.

Reply on news.ycombinator.com