The value of life, which we need to know if we are to correctly prioritize between interventions, is poorly captured by value of statistical life (VSL) estimates, and instead is better captured by community perspective evaluations.
Value of Statistical Life Estimates
VSL estimates may involve stated preferences of survey respondents (i.e. from asking them their willingness to pay to avert some risk) or revealed preference (e.g. looking at how much people are willing to pay for airbags in cars, or at how much they need to be compensated to work in a riskier job). Regardless, the general idea behind the VSL approach is that we have the following implicit equation:
Compensation needed = Probability of death * Badness of death
Hence, if we elicit the compensation needed, we can then just divide by the probability of death to get the badness of death. There are numerous flaws in this approach, however:
Problems leading to the misestimation of the compensation needed
- Humans are not hyperrational, and it's not as if survey respondents, or people taking up risky jobs, will bother to literally think about how much they subjectively value their own life in monetary terms, and then use the additional probability of death to calculate the amount of compensation needed to get them to undertake a risky course of action. Whatever number they throw out in response to surveys, or whatever risk premium they implicitly accept in taking up jobs, therefore, will just not be a good guide to the actual compensation needed, and cannot be used to estimate the badness of death.
- People exhibit insensitivity with respect to changes in small risks, reporting same or similar compensation amounts for risks that differ even in magnitudes. Hence, even if people were calculating the sums needed to compensate them for a risk, they may not be doing so accurately.
- Estimates may be made under duress (e.g. workers may fear quitting a risky job due to a lack of alternatives). The upshot of this is that the true equation people are implicitly solving is "Value from avoiding unemployment >= Probability of death * Badness of death", in which case the actual risk premium does not necessarily equal the compensation needed, since even if the former were below the latter people would still choose the risk so long as the value of avoiding unemployment were sufficiently high. This potentially leads to the systematic understatement of the compensation needed and hence the badness of death, if the value from avoiding unemployment > compensation needed > actual risk premium.
- People may substitute in the judgement of others when estimating the compensation needed, rather than using their own, true subjective monetary value of life (e.g. workers leaving it up to their boss to decide what risks are reasonable, even if the bosses do not value their lives as much as they themselves would).
- Estimates may be confounded by moral considerations (e.g. workers like firefighters may take up jobs not because they are appropriately compensated for it from a purely self-interested perspective, but because it benefits others sufficiently that they see it as worth doing, the risk to themselves notwithstanding). And so, parallel to the duress case, the true equation people are implicitly solving is "Social benefits >= Probability of death * Badness of death", and there is systematic understatement of the compensation needed and hence the badness of death, if the social benefits > compensation needed > actual risk premium.
- Estimates may be confounded by social considerations (e.g. workers valuing the social conditions in which the risk occurs, with them finding risks more acceptable when they trust their managers or have more say over the risk-taking). Therefore, the true equation people are implicitly solving is "Financial compensation + Social compensation >= Probability of death * Badness of death", in which case the risk premium (i.e. the financial compensation) systematically understates the total compensation amount needed, and cannot be used to calculate value of life in conditions where the social elements (e.g. having control over the risk) are eliminated.
- People have different preferences – some care a lot about not dying, and others less so. When measuring the value of life through compensating differentials between dangerous and less dangerous jobs, what we end up measuring is the value of life for those more risk loving and less concerned about dying (i.e. those systematically more likely to take up dangerous jobs), and we end up with an underestimate of the average person's value of life.
Problems leading to the misestimation of the probability of death
- Straightforwardly, people may not know the true risks when buying a product or taking up a job or the like.
Community Perspective Evaluations
An alternative approach to estimating the value of life, which the GiveWell-commissioned IDinsight survey pioneers, is the community perspective evaluation - asking survey respondents to take the perspective of a decision-maker for their community and choose between saving more lives or doubling income. And while there are worries about social desirability bias (i.e. people overstating how much they value life relative to income, so as to not appear cold and amoral), this can potentially be corrected for by applying a discount derived from past studies of social desirability bias in value self-reporting. This allows us to arrive at potentially more accurate moral weights, and this is the approach CEARCH itself takes. (Note: CEARCH is the cause prioritization research organization I work at; we came out from the recent Charity Entrepreneurship incubation round).
That said, there are also concerns over the SDB-corrected community perspective approach:
- The underlying community perspective estimate arbitrarily caps out at 10,000 cash transfers for doubling income for one year relative to saving a life, which means we may be systematically undervaluing life relative to income.
- There are uncertainties over the robustness of the specific error correction value used to offset the social desirability bias in the survey. It would be preferable to conduct additional surveys using the community perspective, but whose research design eliminates social desirability bias (e.g. through list experiments).
Overall, the SDB-corrected community perspective is still preferable to the VSL approach, because on balance, the flaws in both approaches point towards a likely undervaluation of life relative to income. To the extent that the SBD-corrected community perspective yields higher values than the VSL approach, the former probably gets us closer than the latter to the true, average value of life.
I'm a very big fan of the GiveWell/IDinsight's moral weights research in Kenya and Ghana, and think that this is a very promising area of meta-research, which can help us better estimate the value of life and hence better direct scarce resources (e.g. I have always thought that we undervalued life-saving charities like AMF relative to income-raising ones like SCI).
CEARCH is considering funding more research on this, and would be keen to work with any other organizations interested in this matter. All feedback (positive or negative) on the value of such additional moral weights research is, of course, welcome!
Hi Joel, thanks for this write-up and for the work you're doing on this. For some context, I'm the Chief Economist at IDinsight and worked on the GW-funded study you mentioned.
A few comments:
Some thoughts on the points you raised:
(1) On whether social desirability bias is an issue for VSL. My understanding is that the economics literature isn't concerned about this (nor was IDinsight, per the report) - which makes sense to me, because when people are asked to pay to avert small risks, they consider it a pragmatic decision rather than an explicitly moral one where they have to decide whether or not to let someone die for more money. The issue is hence less salient from a moral point of view, and less likely to trigger worries about how one appears to others (kind and compassionate, or cold and selfish). Just think of how much less salient refusing to pay to install a lifebuoy next to a pond is, vs refusing to jump in to save a drowning child right now, even if statistically the former nets out in expected value to the latter.
(2) If we think that both VSL and the community perspective are flawed attempts at getting the true value, and that both are downward biased (per the reasons discussed), then the higher SBD-corrected community perspective is probably a lower bound. In fact, my main worry is that there is significant downward bias - given the strong and very cogent moral reasoning expressed by respondents in the qualitative side of the survey ("life is priceless", "children have economic potential"), you could easily go up by one magnitude in trade-offs (i.e. 1 life to 100,000 cash transfers to double income) and still get a significant number of people going for that.
(3) & (4) Will definitely be interested in talking to your colleague at the Dignity Initiative, and will drop you an email to discuss your potential work in replicating your preference work! Am excited about your ideas, and would be happy to contribute any way I can.
Thanks for the clear explanation of the downside of VSL- I learned a lot!
Glad it was useful!