INSTITUT Veolia Environnement

  • Home
  • Report n°4: How Much to Spend for the Protection of Health and Environment

Report n°4: How Much to Spend for the Protection of Health and Environment

Mortality

For the valuation of accidents the relevant parameter is the so-called "value of statistical life" VSL. Typical values recommended for policy decisions in Europe and North America are in the range of 1 to 5 M€. Previous phases of ExternE [1998, 2000] had used values around 3 M€, chosen as average of the VSL studies that had been carried out in Europe. More recently ExternE [2004] carried out a new CV study and lowered the value to 1 M€. For cancers ExternE [2004] uses 2 M€, higher than VSL because many studies find that people are willing to pay a premium to avoid a cancer death.

But for the mortality impact of NOx, SO2, PM and O3, which is primarily cardio-pulmonary, VSL is not relevant because the loss of life expectancy (LE) per premature death is much shorter than for accidents. Furthermore, one can show [Rabl 2003] that the total number of premature deaths due to air pollution cannot even be determined. One of the reasons is that air pollution cannot be identified as cause of any individual death; it is only a contributing, not a primary cause of death. Epidemiological studies of total (as opposed to acute) air pollution mortality cannot distinguish whether the observed result is due to a few people suffering a large loss of LE or many suffering a small loss. It is quite plausible that everybody's life is shortened to some extent by pollution, in which case every death would be due to pollution. Number of deaths is therefore not a meaningful indicator of the total air pollution mortality (even though several authors who do not understand this point have published numbers). Rather one has to use loss of LE which is indeed a meaningful indicator.

For the valuation one needs the value of a life year (VOLY). By contrast to hundreds of VSL studies that have been carried out in many industrialized countries, VOLY has received little attention until recently. A significant step forward was taken by Krupnick et al [2002] who developed a questionnaire specifically for the CV of air pollution mortality which they have applied in several countries (Canada, Japan and USA). More recently this questionnaire has also been applied in France, Italy and the UK [ExternE 2004]. The application in France [Desaigues et al 2004] involved not only the original questionnaire of Krupnick et al but also the test of several variants, in particular variants that phrased the elicitation question directly in terms of life expectancy gain (rather than risk of dying as in the original version). A crucial point that needs to be explained very carefully in such a questionnaire is that air pollution mortality does not cut off a few months of misery at the end of life but causes "accelerated aging". Based on the results in France, Italy and the UK, ExtenE [2004] is now using a VOLY of 50,000 €.

However, the uncertainties are still very large, about a factor of two in either direction. Furthermore, little or no VOLY information is available for developing countries. The lack of VOLY data is especially troubling for countries like China and India which are increasingly concerned about their air pollution. Only an upper bound on VOLY for almost all countries has recently been derived by Heck [2004] who considers the correlation between life expectancy and GNP/capita (corrected for purchase power), see Fig.4. A public policy to reduce pollution is ultimately paid by a reduction of disposable income, which in turn entails an increase in mortality - the "poverty kills" phenomenon already analyzed by Keeney [1995], although in terms of VSL (and only for the USA) rather than VOLY (and for all countries) as does Heck. If a policy costs more than Heck's upper bound per life year saved, its net effect is a decrease in life expectancy. Typically Heck's upper bound on VOLY is 3 to 10 times GNP/capita (adjusted for purchase power parity), although the uncertainties are considerable.

figure 4

For another perspective on the monetary value of a VOLY, one can look at data on actual expenditures for risk reductions in various sectors (transport, public health, etc). A very comprehensive set of data for the USA has been assembled by Tengs et al [1995]. Analogous data for Sweden have been published by Ramsberg and Sjöberg [1997]. For medicine and fatal injury reduction the costs per life year saved are around $20,000 in Sweden and $40,000 in the USA, but for toxin control the median cost is two orders of magnitude higher in the USA.

In view of the lack of reliable and generally accepted guidelines for VOLY it is instructive to have another perspective, by asking how much a rational individual is willing to pay for an extra year of life, interpreting rationality in the sense of maximizing life time expected utility. That has been done in a recent paper by Rabl [2004] who considered a public policy whose cost is paid in equal annual installments by the general population, via a direct tax or an increase in the price of consumer goods. There is a tradeoff between the utility gain due to increased life expectancy and the utility loss due to decreased consumption. A rational policy corresponds to optimizing this tradeoff, i.e. maximizing the total life time utility. The rational VOLY is the ratio of the consumption loss and the life expectancy gain at the optimum. The resulting VOLY turns out to be equal to the GDP per capita, times a factor that is in the range of about 0.5 to 2, depending on discount rate and variation of quality of life with age. This approach is especially interesting for developing countries where no other information on VOLY is available.

The results of these various studies are summarized in Table 2. Since the VOLY used by ExternE has changed over the years, different publications present different results for the damage costs. The differences are even larger because at the same time there has been enormous progress in epidemiology, leading to improved DRFs. However, the resulting changes in €/kg of pollutant are well within the uncertainty intervals that have been estimated.