The Value of Life is Too High
Small changes add up to absurd conclusions
Somebody considering transportation improvements, medical care, or pollution mitigation will face tradeoffs between saving lives, and increasing consumption. Staying alive is not the only thing which we care about – we would not want to give up travel, social interaction, or our homes in order to eke out a year more of living, on average. We must be able to say that some interventions to save lives fail a cost-benefit test, and implicitly, we say that a human life has a monetary value.
The standard way to figure out what this value is to infer it from observed behavior. Suppose, for example, that there are two jobs which require identical skills but which have different levels of mortality risk. In equilibrium, the jobs should pay different amounts, with the gap in pay representing the level of pay which makes people indifferent to the risk. Alternatively, we can directly measure how people alter their behavior in order to estimate the value. No one has done this, but I think one can use the data in Currier, Glaeser, and Kreindler (2023) to estimate the value of a life from how much Uber drivers are willing to slow down by. Other studies infer the value placed on life from which cars people buy, or whether they are willing
There are some problems with this approach though. It is obviously extremely difficult to find jobs which really do differ only in their mortality risk – as I wrote about yesterday, non-wage compensation is largely impossible to measure. It will be an underestimate, if we assume that people have positive spillovers, and do not fully take these into account in their decision to live. The approach can only find the value of a human life for small changes in the risk of mortality, and cannot be scaled up to larger changes in risk. More importantly, people behave in fundamentally incoherent ways regarding their own mortality. It is considerably more deadly, for example, for people to drive than to take an airplane. Yet, many people, due to their misplaced fear of airplanes, drive and die at a far higher rate in car crashes. The four-fold pattern of risk aversion will show up in what people are willing to do to preserve their lives too.
That we can only infer the marginal changes, and that these marginal changes are likely exaggeratedly high, is actually incredibly important. There was a recent study on the effects of Medicaid expansion from Wyse and Meyer which found that expanding Medicaid was indeed effective at saving lives. The paper itself is interesting and well-worth reading, but I wanted to focus on one particular claim. The cost to save one life, on average, was $5.4 million, or $179,000 per year. Is this something we should be spending on?
If you believe the literature from revealed preference, it is well-below the value we’d be willing to spend on life preservation, and we should do it. Yet, this is just total nonsense! $179,000 considerably exceeds US GDP per capita. Interpreting this figure as saying that we should simply convert our entire government into Medicaid would be misguided. Instead, we have to recognize that our estimates of the value of a human life are universally too high. They represent what we are willing to pay to mitigate miniscule risks, and not the actual value we place on a human life.
I don’t know what the value of a human life should be. All I do know is that they are too high. The value of a human life should try to take into account the expected spillovers onto other people, as well as how much they will personally consume. It may be socially corrosive, but our value of a human life should vary with class or intelligence, because richer and smarter people will benefit others more. The value of lives should be higher in richer countries, again simply because they will experience more. A better method might be to take how much we would be willing to spend on healthcare expenditures as a share of lifetime income, and then finding the value of a life as a function of the average income of a country. Perhaps we could infer the value of life from what people pay for life insurance.
In any case, when we use the implied value of a human life from marginal changes in traffic safety as a guide to healthcare policy, we are making a serious error. We should stop doing this.
I am puzzled by your conclusion and the idea that you claim to support - "I don’t know what the value of a human life should be. All I do know is that they are too high". I can normally follow your argumentation, but there is too much elided here for me to really grok why one should arrive at something akin to "society should spend less money on saving lives, particularly poor lives" based on evidence/logic. When you notice something like "small changes = absurd conclusions", my first instinct is to question the validity of the absurd conclusions and look for the signals that aren't being accounted for, rather than barrel towards extrapolation station.
Your last paragraph could reveal a lot about your implicit philosophies - but you don't dive into them; perhaps you haven't evaluated why you hold these views yourself? Particularly: "It may be socially corrosive, but our value of a human life should vary with class or intelligence, because richer and smarter people will benefit others more. The value of lives should be higher in richer countries, again simply because they will experience more." It might be wise to include a contrasting statement that reflects that you (if you were in charge) would not be tempted to implement some kind of death panels that deny care for poor sick people who are "too socially expensive" to save. Otherwise, I think it is reasonable to be unsure what your perspective is. Do you actually believe that people who are rich or smart deserve more life than those who are dumb or poor? If you do, how do you measure rich and poor, smart and dumb? (to me, this is where things get more interesting, and we can learn more about how economics reflects our philosophy).
It is useful and important to discuss the economics of human lives! That said, economics divorced from any underlying philosophical argumentation feels hollow and incomplete to me. I believe you are a conscientious person, but the analysis you presented here lacks humanity to me.
Presumably as you increase Medicaid spending the marginal lives saved return on the dollar decreases considerably from $179k or whatever it is currently. So "this implies we should spend our entire GDP on Medicaid, which is nonsense" does not follow because approving the marginal dollar spend at ~$179k (or whatever it is currently) does not imply continuing to spend after you've already increased spend and thereby reduced your additional marginal return.