How should health benefits be aggregated?

In my ECONS102 class, in the health economics topic, we cover the value of a statistical life (increasingly now referred to as the value of a preventable fatality). The value of a statistical life (VSL) is a statistical construct that can be used in cost-benefit evaluations as a measure of the benefits of saving lives. So, for example, if government is considering building road safety improvements on a stretch of highway, the VSL can be used as a measure of the benefits of the improvements, which can then be weighed up against the costs of those improvements.

The VSL comes in for criticism though, mainly on moral or ideological grounds (often along the lines of 'you cannot place value on a human life'). However, the alternatives of ignoring the benefits of lives saved, or assuming those lives saved have infinite value, clearly lead to implausible implications for decision-making. For example, if government truly believed that lives had infinite value, they should immediately ban all vehicular travel, because regardless of the costs that such a ban would impose on society, those costs would be outweighed by the infinite benefits from lives saved due to fewer fatal motor vehicle accidents.

A similar critique was raised in this article from The Conversation yesterday by Ilan Noy (Victoria University of Wellington), in the context of measuring the burden of disasters:

But “value of life” prices can vary a lot between and even within countries. There is also an understandable public distaste for putting a price tag on human life. Governments typically don’t openly discuss these calculations, making it difficult to assess their legitimacy.

Noy they suggests: 

An alternative is a “life years lost index”. It is based on the World Health Organization (WHO) measure of “disability-adjusted life years” (DALY), calculated for a long list of diseases and published in a yearly account of the associated human costs.

In conventional measurements of the impact of disaster risk, the unit used is dollars. For this alternative index, the unit of measurement is “lost life years” — the loss of the equivalent of one year of full health.

Both the VSL and the life years lost index can be used to measure the health burden of disasters. They are both ways that can be used to aggregate the health benefits of disaster prevention or mitigation, or the health costs of disasters.

You might believe that avoiding placing a value on lives lost, as the life years lost index does, is an improvement over the VSL. However, there is an implicit assumption that you are making when doing so, that you probably don't even realise that you are making, and which you would probably find just as distasteful as placing a value on human lives. The assumption is that every lost life year has the same value. That's a necessary assumption in order to add up the lost life years from different people.

That assumption doesn't seem so bad, right? But let's think about the implications of making that assumption. If a young person dies aged five years, when they had a life expectancy of 85 years, then that represents 80 lost life years. So, that young person's death would add 80 to the 'life years lost index'. If an older person dies aged 85 years, when they had a life expectancy of 90 years, then that represents 5 lost life years. So, that older person's death would add 5 to the 'life years lost index'. In other words, the young person's life is worth 16 times the older person's life, when you calculate the 'life years lost index'.

Moreover, if government was making decisions on the basis of the 'life years lost index', they would be well justified to devote excess resources to saving young people and protecting their health. After all, saving a young person leads to a significantly greater improvement in the index than does saving an older person. If you extend the index to considering disability, then any treatment that reduces disability or improves health among young people is similarly going to be preferred over the same treatment being offered to older people. Are you still feeling good about not valuing lives saved?

To be fair, the value of statistical life has exactly the opposite problem. It values all lives equally, so a life saved near the end of life is valued the same as a life saved near its beginning. That doesn't seem so bad on the surface, until you think about what that means for the value of a life year. Each year of additional life added for the older person I used in the example above would be valued at 16 times each year of additional life added for the young person. So, a government would be well justified to divert resources to treatments for older people that extend their lives, rather than the same treatments for younger people.

Aggregating health benefits is a difficult problem. These issues (and several related issues) are well covered in Kip Viscusi's excellent book Pricing Lives (which I reviewed here). There isn't a perfect solution to this dilemma, and economics cannot answer the question of which option - pricing lives saved (as the value of a statistical life does), or life years (as the life years lost index does) - is better. It is quite reasonable to hold an opinion either way, and quite unreasonable to impose your opinion on others. Critiques of VSL like the one that Noy uses do little to help people understand the issues, or to make an informed judgement for themselves.