The VSL is important for policy, which is why I devote a serious amount of time to it in my ECONS102 class. However, it's not the sort of concept where I would have expected to see a book targeted at general readers. So, Kip Viscusi's Pricing Lives was an interesting surprise to me. However, if anyone were going to tackle the challenge of writing a book on VSL for a general readership, it would have been Viscusi. He didn't invent the VSL (that honour goes to Nobel Prize winner Thomas Schelling), but has arguably had the greatest influence on its use and development, and has made countless research contributions in this area.
The book is relatively easy to read, and especially so for someone with a basic grounding in economics. However, I found it somewhat repetitive (even to the extent where some of the endnotes also appeared in the text), and it could easily have been shortened substantially without any loss to the narrative. Viscusi starts by defending the concept of monetising the VSL, rather than leaving lives saved as some un-measured benefit of policy:
In an earlier study of the use of benefit-cost analysis to value water-resource projects, I found that government agencies placed almost exclusive emphasis on the monetized effects... In that era, environmental consequences received qualitative discussion but were largely set aside in favor of emphasis on the series of tangible economic benefits that were monetized.In other words, the only things that count in decision-making, are unfortunately only those things that can be counted (and measured). Which is why the VSL has been such an important policy tool. However, Viscusi sees the VSL as being underutilised, and the main aim of the book is to open the readers' eyes to the potential:
...to greatly broaden the menu of uses of the VSL, which to date has largely been confined to benefit assessments in the analysis of proposed major regulations.Specifically, Viscusi presents three main opportunities for the VSL: (1) in the realm of business decision-making, such as in deciding which product safety improvements are worthwhile (and which are too costly to implement); (2) in the courts, such as in deciding on appropriate damages awards that would adequately incentivise safety improvements; and (3) a broader use in government agencies, such as in the size of regulatory sanctions.
I found the discussion of product safety to be especially good, especially the unexpected incentive effects caused by initial corporate use of benefit-cost analysis:
In the 1970s, 1980s, and 1990s... all the major US automobile companies undertook detailed economic analyses of the cost and risk implications of safety-related product characteristics. However, frank assessments of the risks and costs of different design possibilities led these companies to be vilified in the press and penalized by juries for undertaking such safety studies, not simply for specific alleged deficiencies in the analyses... These adverse experiences no doubt have contributed to the corporate abandonment of systematic assessments of safety decisions.The courts used previous safety assessments against the companies, so the companies responded by ceasing to conduct rigorous safety assessments. Viscusi presents an (obvious) solution to this dilemma, which is to make the assessments inadmissible in court, provided the company used an appropriate measure of the VSL in assessing the benefits of the safety-related product characteristics.
The book does a great job of outlining some of the other controversies related to the VSL, such as whether the VSL should vary by age (should lives saved of younger people be worth more than lives saved of older people) or income (should lives saved of wealthier people be worth more than lives saved of less wealthy people). While acknowledging that context matters, Viscusi appears to lean towards 'no' in terms of age, and 'maybe yes' in terms of income (but only when the associated policy costs would not spill over onto other groups). The book also tackles the difference between statistical lives and identified lives, which is particularly challenging (and not easy to explain, so I will refrain from doing so here).
Overall, this is a broad book on a seemingly narrow topic. Definitely, this is recommended reading for those with an interest in the subject of policy analysis, or law and economics, and especially for those wanting to understand the VSL better. This is an important policy concept, and destined to remain so.
Interesting. I concluded, perhaps 20 years ago, that "the system" here in Australia valued a life at very approximately $1m. [Say, 20 x our per-capita GDP]*. (Had never seen it admitted to or discussed in any public forum - before or since). Subsequent examples persuade me that it remains a reasonable estimate.
ReplyDeleteI can't imagine a way of estimating and standardising across different countries/economies/cultures. But, were it possible to do so, it would be fascinating to make comparisons [such as*].
Viscusi spends a whole chapter on that topic (it is well worth reading if you're interested). He essentially concludes that the best approach may be to transfer the VSL that is estimated using US labour data (which is roughly US$10 million) to other countries, because that is the highest quality estimate (meaning the estimated with the least uncertainty, and based on revealed rather than stated preferences). To transfer to other countries, Viscusi recommends an income elasticity of VSL of 1.
DeleteSo, given the ratio of GNI per capita between Australia and the US (51360/59160) that gives a VSL of US$8.7 million for Australia (or about AU$12.4 million).
A similar calculation for New Zealand gives a VSL of US$6.6 million (based on a GNI ratio of 38780/59160), or NZ$9.9 million. That's far higher than the official VSL here, which is around $4.2 million.
Thank you for additional details.
ReplyDeleteTo start with a US figure and transfer/translate (exchange rates, ppp etc.) seems to me (retired CA) to be ridiculous. Rather, if each country used its own per capita GDP (**expressed in its own currency**) and an agreed multiplier related to working years lost upon death of an average citizen, it would give a starting-point/reference figure that would make some intuitive sense. Adjust that - stating reasons - to deal with particular case under consideration.
I often think that the level of public debate would benefit from a wider understanding of some elementary economics concepts : sunk cost, marginal cost, opportunity cost, that money is a fungible : come to mind.
Economists ensconced in academe need to do some more work on VSL before it joins the above list. :)
The problem with using per capita GDP (or income) as a basis for a value of statistical life is that it assumes that a person's only worth is the income that they generate over their working life. Viscusi calls this the 'cost of death' approach. In my ECONS101 class, we call it the 'human capital' approach. Either way, it radically underestimates the value that people place on lives saved (or on risk reductions) based on actual revealed preferences (such as, how much of a wage premium is attached to riskier jobs).
DeleteHaving said that, I do agree that local data is probably better, if good data are available, because risk preferences differ between countries in ways that may not be adequately captured by differences in incomes.
However, there is a tradeoff here. A VSL based on revealed preferences data is more defendable that one based on stated preferences. But you need a lot of data to get reasonable revealed preference estimates. Which is why countries like NZ (and even the UK) use stated preference surveys to estimate the VSL.
I still think the concept needs to be grounded at some point to an amount which has some intuitive appeal as a 'real' value. Such as a standardised 'cost of death' baseline (which would not vary greatly around,say $1.3m, for each of the three countries discussed).
DeleteUsers of the VSL then need to explain how "in this particular instance" they justify using, say, $10m, rather than the standard $1.3m. If the explanations rely either on "revealed" or "stated"(surveyed) preferences ....................... expect a cynical look from this jaded, old accountant.