Wednesday, 25 August 2010

And more alcohol costs

Australia's all a-buzz with its latest study of the cost of alcohol to society. They get a really big number - even bigger than Collins & Lapsley's prior really big number.

Let's take a look at just one part of that cost figure.

At page 133, they provide some estimates of the economic costs of intangible harms Australians suffer as a result of someone else's drinking. Their survey respondents answer questions that give them a health-related quality of life score. A one-point drop in that score, they say, costs the individual the equivalent of $50,000. Respondents are sorted between those who cannot identify a known drinker who has negatively affected them, those knowing a drinker whose drinking has affected them "a little", and those knowing a drinker whose drinking has affected them "a lot". They take the mean quality of life score for each group, multiply the differences from baseline by $50,000, extrapolate to population averages, and come up with about $6.4 billion in costs.

There are a couple of pretty obvious problems here, if I understand their method correctly. Which I may not, but I'd love to hear if I'm wrong on this count. It really looks like they're just comparing sample means. If so, and if there are any other differences - differences unrelated to alcohol - between folks who have a close relationship with a problem drinker and those who don't, then the mean comparison is hopelessly confounded. There are all kinds of bad social circumstances that are predictors of problem drinking. Those same circumstances would lead to lower quality of life scores regardless of whether there's any drinking. Regression analysis is needed, not comparison of means.

At the paper's outset, they explicitly reject mainstream economic analysis's insistence that external costs are the ones that matter:
A small tradition within economics, critical of the cost-of-illness tradition discussed below, has endeavoured to confine the estimated costs to strictly-defined “external costs”, that is, costs imposed on others by the alcohol consumption of the drinker (Heien and Pittman, 1992, Manning, et al., 1989). However, these studies have been quite rigid in excluding costs which they do not regard as true externalities. Thus Heien & Pittman (1992) exclude costs to others in the drinker’s family on the grounds that these “are basically internalized within the family”. They assume there is no external cost on the grounds that a family unit internalizes the alcohol harm that a member in the family causes.
However, this case is only true if the family jointly owns all the resources. [EC: Umm, no.] Each individual in the family is entitled to their own labour and time. When drinker A in the family harms family member B, resulting in a loss of output due to injuries or loss of time spent seeking services, those losses will only be internalized if A and B’s labour and time are jointly owned. In terms of property – be it money or belongings – even if it may be jointly owned no matter who bought it, there are civil and criminal laws that prohibit drinker A damaging it. In this example, if drinker A harms B, that harm is clearly not internalized.

In their rigid approach to what constitutes an “externality”, Heien and Pittman (1992) also exclude injuries to a passenger in a drink driver’s car on the grounds the passenger “has accepted the risk of riding with an abuser”.
Sorry. If you're in the car, you're party to an implicit contract with the driver. It's then internal, not external.

And how joint ownership is required for cost internalization is beyond me. You can do it with Coasean bargaining with complete private individual ownership. You don't need the passengers to be joint owners of the airline for the costs of a screaming baby on the plane to be internalized if the passengers are in a contractual nexus with the airline.

This doesn't bode well for the rest of the paper. But if anybody can provide further insights on whether I'm right that they're just comparing means on intangible costs, as noted at the outset, I'd like to hear it.

No comments:

Post a comment