How Much Money Saves A Life?
It costs more than you'd think. (Introductory, unlikely to be very interesting to experienced EAs.)
Sometimes people wonder how much they would have to donate to save a person's life. Let's say you gave money to a charity that provides health care to people in the developing world: maybe it provides antiretroviral treatments to people living with HIV, or gives insecticide-treated nets to people in regions where malaria is a problem, or gives children Vitamin A supplements to prevent deficiency. How much money would you have to donate before you know that you saved a single person's life?
A study suggests that the average person thinks that it costs $488 to save a life by donating to the average developing-world charity. That is, the average person thinks it is very cheap to save a life in the developing world. However, researchers have discovered, saving a life in the developing world costs nearly ten times what people think it does.
Charities tend to go "it costs three dollars to buy a bednet and save a person's life!" You know that charities exaggerate, so you go "maybe it actually costs three hundred dollars?" But this is still an underestimate.
The charity evaluator GiveWell is the most rigorous charity evaluation organization in the world. They pore through thousands of different charities and select only a small number which meet their strict standards for quality. They have found four programs which they believe are most effective at saving lives: medicine to prevent malaria; insecticide-treated bednets, which also prevent malaria*; supplements which prevent vitamin A deficiency; and cash incentives to get people to bring in their children for vaccines. If you donate to any of these charities, it will cost between $3000 and $5000 per life saved equivalent.
"Per life saved equivalent" is hiding a little bit of statistical wizardry. All of these charities have multiple effects. For example, malaria and vitamin A deficiency make it hard for kids to learn in school and tend to make kids less healthy adults, so as adults children who had malaria or were deficient in vitamin A are poorer. It wouldn't make sense to talk about how good malaria prevention is without mentioning that kids who don't have malaria are much richer as adults!
[The below two paragraphs are inaccurate for GiveWell’s current version of the cost-effectiveness spreadsheets—see Erin’s excellent comment.]
So what happens is that GiveWell asks their staff how much increased wealth is equivalent to saving one child's life, then converts the economic benefits to the number of lives they're "worth." (There are also some benefits to malaria prevention-- like people not suffering through the experience of getting sick-- that are so small that GiveWell doesn't include them in their spreadsheets.) So when we say "per life saved equivalent," what we mean is that donating $3000-$5000 has a positive effect on the world that is the same as the positive effect of saving a person's life.
GiveWell’s staff have thought a lot about global health, but they don’t necessarily have any special expertise on how much money is equivalent to saving one life. A lot of people assume that the average of GiveWell’s staff estimates is probably close enough to what they’d think if they thought about it, but you don’t have to. You can copy GiveWell's cost-effectiveness analysis and put in your own moral weights, if you are unsatisfied with GiveWell's, but it's not very user-friendly.
Why is the cost per life saved equivalent so high? The cost of any individual item is very cheap: it costs only $5 for a bednet and $1 for a vitamin A supplement. And bednets save people's lives from malaria, right? So why isn't it $5 per life saved?
Well, not everyone who gets vitamin A supplementation or a free bednet would have gotten sick without them. Of the people who got sick, many of them wouldn't have died. So you have to distribute a lot of vitamin A and bednets before you can be certain you prevented a death.
A lot of treatments don't always work. Sometimes people get vaccines or medicine that prevents malaria and still get sick, because our treatments don't work perfectly. Some programs have to be used correctly to have an effect: for example, some people might not hang up a bednet, or hang it up improperly.
Some treatments have negative effects. For example, medicine that prevents malaria can make children sick or contribute to the malaria parasite developing resistance to the drugs we use to treat malaria. These negative side effects are incorporated into GiveWell's models.
Finally, sometimes if one charity didn't distribute vitamin A, another charity would. There are a lot of charities in the world, and a lot of them are doing the same programs. So if you distribute Vitamin A, some other charity might say "well, we were going to distribute vitamin A there, but that charity seems to have it covered." Presumably in that case the charity would spend the money on some other less cost-effective treatment. So in that case the true effect of your donation is not to cause vitamin A to be distributed-- it's to free up the resources of a different charity so that they can do whatever they would otherwise do with that money. Of course, sometimes no one else would distribute the vitamin A if you didn't. It can be hard to know which one is true. GiveWell tries to estimate this as best they can and incorporates it into their models.
In conclusion: a lot of people think that it only costs a few hundred dollars to save a life, and all developing-world charities are equally good. In fact, the best charities are many thousands of times better than other charities, and when you fully consider every factor involved it costs about three to five thousand dollars to save a life.
*Tangent for certain readers: you may have heard that insecticide-treated bednets are ineffective because people use them for fishing instead of putting them over their beds. The Against Malaria Foundation, which GiveWell recommends donating to, checks whether people put the nets over their beds, and finds that 75-90% of recipients use their bednets. There is no evidence that fishing with insecticide-treated bednets harms fish stocks. GiveWell's cost-benefit analysis already accounts for the fact that some people don't use their bednets.
Hi Ozy,
I think GiveWell's $3000-$5000 figure is intended to be, and can reasonably be read as, the actual cost per life saved, not just life saved equivalent. Here https://www.givewell.org/charities/top-charities/citations#Why_is_the_cost_to_save_a_life_so_much_higher_than_the_cost_per_nettreatment is one place I would really expect to see some nod to the concept of life-saved-equivalent if it was being used, and I don't see it. Looking at the 2021 CEA, the "cost per life saved" boxes really are what they say on the tin; e.g., B168, estimating that AMF's program in the DRC saves one life for $3,713, refers only to the amount spent, the total number of actual lives saved, and a downside adjustment for things like wastage. Income increases for children covered by bednets are also in the spreadsheet, but they seem to only be used for "units of value generated" estimates, not the cost per life saved.