Ineffective Altruism

Texas Lt. Gov. Dan Patrick hung up a morality piñata when he had the audacity to state, on record, that he’d be willing to take risks with his life to prevent an economic meltdown:

No one reached out to me and said, ‘As a senior citizen, are you willing to take a chance on your survival in exchange for keeping the America that all America loves for your children and grandchildren?’ And if that is the exchange, I’m all in

Like many other moral-minded folk, Cuomo took a free swing at the piñata a few hours later, and snapped back with this zinger:

While this tweet is genetically engineered to be re-twatted, it is a ridiculous statement.  People put prices on human life all the time. Insurance companies price human lives. Legislators do it all the time when enacting regulations meant to protect human life, at a cost. 

On the flip side, it’s common to price out the cost of saving a life as well.  Effective Altruism is a niche but deeply principled movement which goes through great lengths to, with exacting rigor, price out the most effective way to save lives (and then, generally, donate 80% of their salary, time, and organs to those causes).

GiveWell is one of them.  They annually put together a version of this spreadsheet which calculates which charities are able to do the most work with the fewest dollars. 

It’s worth checking out. The math is more complicated than a naive observer would expect. It turns out that the shittiest things nature can do to a person often doesn’t kill them — Hookworms reduce educational attainment, blind children (!), and reduce life earnings, but rarely if ever… kill anyone. But because being blind and poor really sucks, many of GiveWell’s “most effective” charities attempt to eliminate hookworms and similar parasites.  

The way to massage this math onto a linear scale is to compute dollars per QALY saved, where QALY stands for Quality Adjusted Life Year — the equivalent of one year in perfect health.  By this measure, saving the life of an infant who would otherwise die in childbirth may save 75 QALYs, while saving the life of a senile, bedbound 80 year old may save 15 Quality Adjusted Life Hours.

This is a reasonable and principled method of making financial investments.  If you put stock in the Efficient Politics Hypothesis of government, stop here and feel good about the choices we’ve made.

2020

We have decided to, as a nation, spend an indefinite amount of time-money at home eating ice cream and mastrubating to Netflix, in a crusade to stop* the spread of COVID-19.  

*by some difficult to define metric

How much time-money?  $2.2 trillion is the down-payment stimulus package.  It’s a very conservative lower-bound estimate of the cost of recovery (it’s not expected that this will fix the economy by any means), so we can run with it.

How many lives are we saving?  The high upper bound is 2.2 million (assuming a 100% infection rate (not generally considered realistic), with a fatality rate of .66% (best estimate)).

This works out to a conveniently round, very lower bound, $1,000,000 per life saved (in the outcome that the US quarantine does prevent the majority of those deaths).  What about those QALYs? I’m not going to try sum it out, but we can look at a few suggestive stats:

  • The average age at death (due to COVID-19) in Italy is 79.5.  Italy’s average life expectancy is, for reference, 82.5*
  • 99% of Italian COVID-19 victims had pre-existing conditions (same source).

*I understand that the average life expectancy of a living 79 year old is higher than 82, which is why I’m not doing the math, so please shut up so we can move on.

We can speculate that no, we will not be saving very many QALYs.  But we can use the crude ‘lives saved’ metric instead to generously lower-bound our math, and run with 2.2 million raw.

Effectiver Altruism

My question: how does this calculus compare to effective altruism?  I was genuinely curious, because $1,000,000 per life saved is somewhat disproportionate to the charity pitches you see on television:

COMMERCIAL CUT. Enter UNICEF USA. 

“Save a Child for Only $39 per day*”

Exit right.

*assuming 70 year life expectancy @ $1,000,000 per life

I tried to find a toplist of “problems we can solve for X dollars to save Y lives per year”.  I did not find one. GiveWell (entirely reasonably) calculates the payout of donating to a specific charity, not of speculatively eliminating entire facets of human suffering.

So I put together a list.  These numbers aren’t precise.  They are very speculative.  My goal was to understand the orders of magnitude involved.

My focus was on problems we could solve that don’t involve serious tradeoffs, and don’t require hard political choices.  Trying to solve “war”, “suicide”, or “alcoholism” don’t cost money per se, they require societal committment we can’t put a price tag on.  For the most part, this leaves diseases.

I started with the highest-preventable-death diseases in the developing world, and ended up with 7 “campaigns” where we could non-controversially plug in money on one end, and pull extant wretched masses out of the other.   When considering the payout in lives saved from eradicating a disease, I used 30 years, because using “forever” is unfair (I’m sure there’s a time-decay value on life an actuary would prefer, but this was simple, and it doesn’t really change the conclusion).

Global Hunger

Hunger is the stereotypical “big bad problem”, and it wasn’t hard to find data about deaths:

Around 9 million people die of hunger and hunger-related diseases every year, more than the lives taken by AIDS, malaria and tuberculosis combined.

(for the record, this gave me some good leads on other problems).  How much would it cost to actually fix hunger?

Estimates of how much money it would take to end world hunger range from $7 billion to $265 billion per year.  

Pulling the high estimate, we get… 

Price Tag$265 billion
Lives Saved9,000,000
Cost Per Life$29,444 / life 

Malaria

Malaria sucks, and a lot of smart people want to spend money to get rid of it.  How many people does Malaria kill?

In 2017, it was estimated that 435,000 deaths due to malaria had occurred globally

What would it take to actually eliminate Malaria?

Eradicating malaria by 2040 would cost between $90 billion and $120 billion, according to the Gates Foundation

We can highball this estimate to get …. 

Price Tag$120 billion
Lives Saved13,050,000
Cost Per Life$9,195 / life

Tuberculosis

Tuberculosis is still a huge killer in the developing world, but it’s a killer we can put rough numbers on:

The Lancet study said reducing tuberculosis deaths to less than 200,000 a year would cost around $10 billion annually… a chronic lung disease which is preventable and largely treatable if caught in time, tuberculosis is the top infectious killer of our time, causing over 1.6 million deaths each year.

Price Tag$10 billion
Lives Saved42,000,000
Cost Per Life$7,143 / life

The math here is fuzzier than I’m comfortable with, but works out in the same ballpark as Malaria, so I feel OK about the result.

AIDS

Again, this wasn’t the cleanest math puzzle, but this report pegs the cost of ending AIDs at $26.2 billion/year for 16 years.  At 770,000 deaths per year from AIDs, we can (again, more mathematically than I like, ballpark the bill and lives saved over 30 years:

Price Tag$366.8 billion
Lives Saved10,780,000
Cost Per Life$33,963 / life

Maternal mortality

Like Tuberculosis, it ends up in the same ballpark as Malaria, so I’m inclined to believe it’s not more than half-asinine.

Dying in childbirth is bad, and kills people.  How much public health spending would it take to eliminate it?  Again, it was really hard to find good estimates, but we find that 

Researchers at UNFPA and Johns Hopkins University calculated that the annual cost of direct services, such as paying for medical staff, drugs and supplies when a woman is giving birth, will reach $7.8bn (£6.2bn) by 2030, up from an estimated $1.4bn last year.

To save how many lives?  

About 295 000 women died during and following pregnancy and childbirth in 2017

I’m honestly not sure how to compare these numbers, but if we ballpark that the $7.8 billion saves at least that number (?) each year, we work out to 

Price Tag$7.8 billion
Lives Saved295,000
Cost Per Life$26,440 / life

If you don’t like the fuzziness of the math, feel free to ignore it, or multiply it by 10.  Or whatever.

Measles

Measles is bad.  To convince you to vaccinate your children, I will attach a picture of a child with measles.

In the US, measles doesn’t flat-out kill many people, but in the developing world, it does:

Worldwide more than 140,000 people died from measles in 2018

What would it cost to actually eradicate measles?

Eradicating measles by 2020 is projected to cost an additional discounted $7.8 billion and avert a discounted 346 million DALYs between 2010 and 2050

Using the 30 year window I’ve been using, we end up with:

Price Tag$7.8 billion
Lives Saved140,000
Cost Per Life$1,857 / life

This is a shockingly low number, and can only conclude either that (1) I messed something up, or (2)  that we are a terrible, horrible, species for not having eradicated this decades ago.

Global Warming

Stepping outside of diseases, what about something big?  Global warming is big.

How many deaths might be attributed to climate change in the next century?  Obviously this is a make-believe number, but the number is definitely at least ‘several’:  

A report on the global human impact of climate change published by the Global Humanitarian Forum in 2009, estimated more than 300,000 deaths… each year

This is the lowest bound I can find short of flat-out AGW denialism.  It’s easy to find genocide-level projections, assuming crop failures in the developing world several orders of magnitude higher.  I won’t use them.

What’s the cost of fixing global warming?  Long-term, there’s no good answer yet, because the technology doesn’t exist.  But there are (speculative) ways we can slow it down for reasonable sums via carbon sequestration: 

Returning that land to pasture, food crops or trees would convert enough carbon into biomass to stabilize emissions of CO2, the biggest greenhouse gas, for 15-20 years… With political will and investment of about $300 billion, it is doable

We can use these numbers to price tag the cost/payoff of delaying global warming:

Price Tag$300 billion
Lives Saved6,000,000
Cost Per Life$50,000 / life

This is the most speculative guesstimate of all, so if you want to ignore it too, feel free.

Compare & Contrast

My original goal was to build a snarky visualization game which invited users to bin-pack global problem solving which worked out to less than $2T.  I was foiled, because you could do literally everything on this list for less — by my (fuzzy) calculations, you could solve global hunger*, malaria, tuberculosis, delay global warming 20 years, cure AIDs, eliminate maternal mortality, and eliminate measles, for “only” $1.4T.

*to be fair, this one is annual, not a permanent elimination.

But I had already invested the time learning how to use Data Studio, so I made the chart anyway:

(you can play with it yourself here)

Conclusion

What I feel confident saying — even using wildly generous numbers, since I am:

  • using the absolute upper bound for US COVID-19 deaths,
  • using crude deaths for a disease which primarily affects the elderly, instead of QALYs when comparing to diseases which affect primarily the young,
  • using just one ($2.2T) of many recovery packages we’re going to pay for, and
  • generously upper/lower bounding all the numbers

is that 

The COVID-19 economic shutdown is 20x as expensive per life as any other public health intervention the US could fund.  The most expensive intervention on this list — “delaying global warming” — cost $50,000/head.  We’re paying $1,000,000/head for COVID-19.

Now, there is a range of valid value statements, depending on your priors and beliefs in how creative fiscal policy can be:

  • “We should do both”
  • “We don’t have money to do either”
  • “Maybe civilization was a bad idea”

I’m not claiming to be a superhero here.  I’m not an Effective Altruist, and probably don’t register as an altruist at all.  But cheap platitudes annoy me, especially when used to shut down arguments.

In the end, the most meaningful, easiest, way Cuomo could have qualified his Tweet would have been 

We will not put a dollar value on American life

It’s not a great look, or a great tweet.  But as far as I can tell, it’s the only way to make the numbers — maybe — add up.

4 thoughts on “Ineffective Altruism

  1. I agree with most of the rationalism on all this article. The only thing that do not match here is the fact that we are using data that is still being created.

    Thing is, currently even in Italy we are in the dawn of Heath system collapse. Which means, for now we are only starting to see people dying not because COVID-19 has a high mortality rate, Italian doctors currently are choosing who lives or who dies using exactly the QALY measure to choose whom to treat. But, as the virus spread more and more people needs hospital treatment, and the demographic distribution of cases in which people needs treatment as scattered in our whole age pyramid.

    My point on this is, even if we base our approach on QALY metrics, currently we only see a small parcel of potential deaths due to health system collapse. And in the lack of treatment even the young people would die.

    Given the data we currently have, we are creating a false conflict here between economy vs saving lives. Cause we are considering that only older people dies. We only see this demographic dying because we currently still can count on the health infrastructure to harbor the young.

    1. Those are reasonable points, and I don’t want to speak for choices in other countries, which may or may not align with these costs. My main objection is to the mindset that there is _no_ cost at which saving lives becomes the incorrect choice. I think it’s dangerous and biases us towards short-term, visible, first-world fixes, and can blind us longer-term, less visible alternatives.

      And honestly, my biggest takeway is “if you support the quarantine, there are a lot of other causes you should be actively supporting as well”.

  2. I agree with most of the rationalism on all this article. The only thing that do not match here is the fact that we are using data that is still being created.

    Thing is, currently even in Italy we are in the dawn of Heath system collapse.

    Most people die for Covid-19 not because it has a big death rate. But because those with harsh symptoms that need medical attention are not being able to get it.

    For now we are only starting to see people dying not because COVID-19 but for the system collapse.

    Italian doctors currently are choosing who lives or who dies using exactly the QALY measure to choose whom to treat. But, as the virus spread more and more people needs hospital treatment, and the demographic distribution of cases in which people needs treatment as scattered in our whole age pyramid.

    My point on this is, even if we base our approach on QALY metrics, currently we only see a small parcel of potential deaths due to health system failure. And without treatment even the young people would die.

    Given the data we currently have, we are creating a false conflict here between economy vs saving lives. Cause we are considering that only older people dies. We only see this demographic dying because we currently Use the QALY measure to choose who lives and dies.

    Thing is, with more and more younger people needing attention we will see the age of “person treated” starting to lower, at a point that the number and age of person dying would impact the economy as much as the quarantine.

    Of course I am not pledging to say that this is a simple problem. But that we need to understand this POV too.

    If we don’t flatten the curve, we are not talking about old people lives we are talking about everyone that needs treatment dying without it.

    Sorry, I messed up the last comment.

    1. No worries, and thanks. 100% agree that enforcing most “flatten the curve” rules — sanitization, distancing, masks — makes sense, and that a hospital breakdown will introduce problems of its own. We need to be scaling up medical supplies as quickly as humanly possible

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s