I recently came across the following question posted by Paul Buchheit (the founder of Gmail):
Assume that I’m going to get rid of $20,000 and my only concern is the “common good”. Which of these is the best use of the money: give it to the Gates foundation, buy a hybrid car, invest it in a promising startup, invest it in the S&P500, give it to the US government, give it to a school, other?
We provide advice on how to maximise the social impact of your career based on several years of research alongside academics at Oxford, and many of our users donate money as a way to do good with their careers. I liked this way of posing the question – it’s both broad and concrete. So I spent an hour writing out a rough answer, to serve as an introduction to some of our key ideas.
I’ll take each option in turn and eliminate the worst ones, then compare a shortlist at the end.
The main benefit of a hybrid car is reduced CO2 emissions. The average car in the US emits about 4 tons of CO2 per year. If you could cut this to zero (to get an upper bound on how good a more efficient car could be), then you’d reduce emissions by 4 tons of CO2 per year.
Economists have made a variety of estimates of the social cost of a ton of CO2. A recent survey of over 300 economists gave a median value of $32 per ton (with considerable uncertainty).1 So, avoiding 4 tons of emissions each year would produce about $130 of social value per year for the lifetime of the car. Supposing you keep the car for 10 years, that’s a maximum of $1300 of social value. And that doesn’t seem especially great for a $20,000 investment.
I expect this produces some good, though there is reason for skepticism: when subjected to randomised controlled trials, 50-75% of large social programs don’t work (though the remainder mostly work, so on average they’re probably positive).
It’s consensus in economics that there are diminishing returns to income – i.e. giving a dollar to a rich person produces less welfare than giving a dollar to a poor person. Economists most commonly model the relationship between income and welfare as logarithmic, and there’s considerable empirical evidence supporting this relationship (and some evidence that the returns to additional income are even weaker than logarithmic).
Just on the basis of diminishing returns to income, we’d expect giving cash to the poorest people in the world to produce welfare benefits around 100 times larger than giving to the average American. So I expect giving the money to the US government to be beaten by giving it to a poor Kenyan, which you can now do cheaply and reliably through Give Directly (though there are some tricky questions around the long-run effects of giving to the US vs. the developing world).
As a first approximation, the effects of giving to a school are similar to welfare spending – you’re transferring money to poor Americans (or perhaps rich Americans if you give to a rich school). So it also doesn’t seem promising compared to giving to the global poor.
Giving to a school, however, looks more promising because better education (probably) speeds up innovation, creates good citizens and has many other positive spillover effects. And you may be better at picking ways to spend the money than the US government. It would be especially promising if the school is committed to improving teaching practices and other innovation within education (e.g. KIPP schools).
However, there are also some serious reasons to doubt that education is the best cause to focus on if you want to maximise the common good. Here’s five reasons. In short, it’s really hard to know what works in education, and it’s already perhaps the most popular cause among donors, so there’s a lot of money already going into it.
Invest in the S&P
Investing the money (compared to, say, hiding it under your bed) increases the supply of capital, which should decrease the cost of capital, potentially allowing business to grow faster. However, there’s lots of other potential economic effects, so the overall impact is unclear. I asked one of the leading expert economists on interest rates what they thought the net social benefit of investing in the S&P would be, and they said it’s really uncertain either way, so we can probably take the expected value of this effect to be zero.
I think the main benefit of investing in the S&P is to compound the money so that you have more to give in the future.
The question of whether to give now or invest and give later is tricky, and depends on your situation and a couple of big judgement calls. Here’s an in-depth analysis. We summarised the findings with this flow chart:
Invest in a startup
This is better than investing in the S&P500.
The expected return is higher. It’s true the risk is also higher, but if your aim is to help the common good, there’s not much reason to be risk-averse (see an introduction to why here and more detail here. So if you’re investing to give later, invest in whatever has the highest expected return.
You increase the supply of capital just as much as with investing in the S&P500, so it’s equally good on that front.
Investing in the S&P500 doesn’t help to improve the allocation of capital, whereas if you’re picking startups then there’s additional scope to have a social impact by improving allocation. This is especially true if you have expertise in picking startups, and because startup investing is probably one of the least efficient areas of the market.
You may be able to target companies with especially high positive externalities or that create an especially large consumer surplus through positive innovation e.g. clean energy, meat substitutes, and services for the global poor. If you can do this with only a small tradeoff in financial returns, then that could increase your impact even further.
By doing in-depth research and focusing on public goods, moreover, you’d expect the foundation to be able to beat a simple cash transfers as pursued by Give Directly.
But before pursuing this option though, consider:
The foundation doesn’t actively solicit donations. In their guidelines for people who want to donate, they say: “From time to time, people generously offer to contribute money to the foundation. We prefer that people give directly to our grantee organizations rather than to the foundation if they want to help advance the causes we’re passionate about. We have the stable funds we need to help us fulfill our mission, but our grantees often do not.” It makes sense – the foundation already has over $42bn, and it’s going to be hard to manage that much money, even for Bill Gates.
How well do your values match Gates’? If there’s a serious mismatch here, then it may not turn out to be efficient to give through the foundation. For instance, Gates has especially focused on immediate needs in the past, such as global health. Global health is an extremely promising cause, but some people think the bigger story is technology and that you can have even more impact by thinking about setting up humanity for the long-term (e.g. Elon Musk and Peter Thiel). The foundation also focuses heavily on US education, which I have reservations about (as noted above).
Personally, among large foundations, I’d give to the Open Philanthropy over Gates at the margin. OpenPhil is a partnership between Good Ventures (a foundation established by Dustin Moskotivz, through which he intends to donate about $8bn) and GiveWell. Why would I give here?
They publish most of their research and reasoning. This helps to build a body of knowledge around how to most effectively donate to charity, improving the sector as a whole. It also makes it easier to trust their findings. In contrast, it’s harder to know what’s going on in the Gates Foundation.
Better alignment with my values. OpenPhil also has the explicit mission of maximising the common good, but they take an even more wide ranging approach than the Gates Foundation. While Gates has mainly stuck to US education and global health, Open Phil is exploring US Policy advocacy, biosecurity, biomedical research and many other areas. See this list of causes they have begun investigating.
The main reason against is that OpenPhil has less track record.
Another problem is that OpenPhil doesn’t currently accept small donations. However, you can donate $20,000 to fund GiveWell’s operations, which goes towards funding very similar research and saves OpenPhil money, meaning the effect is similar to giving to OpenPhil itself.
If you don’t have sufficient value alignment with either of these options, then it may be worth doing your own research, though you’ll have to trade against spending much more of your own time.
The top two options from the list are:
Give to the Gates Foundation; or perhaps even better, Open Philanthropy.
Give it to a promising startup with positive externalities, then donate later. (If you can’t invest in startups, or don’t have relevant expertise, invest in the next highest return assets available).
Choosing between these two is tough. The main uncertainty is how you fare on the now vs. later flow chart above.
It seems like you’re unsure about the best cause, so my overall guess is:
Give several percent of wealth fairly soon, with a focus on “giving to learn”. In particular, use your giving to build up relationships with other groups focused on maximising their impact, such as Open Philanthropy, and to do other experiments, so you can figure out how best to use money in the future.
Invest the remainder of your money to give later. Invest as much of this as you can in high potential startups, especially those you expect to produce large positive externalities. As your views about how to best use money for the common good settle down, start committing on a big scale.
If you’re reading this and would like more advice on how to maximise the social impact of your own donations, post a question in the comments, or drop me an email ([email protected]). We can go into a lot more depth on all of the issues above. Also see an intro to our career advice.
“The median of 311 published estimates of the social cost of carbon is $116/metric ton of carbon (Richard Tol, ‘An updated analysis of carbon dioxide emission abatement as a response to climate change’, 2012, http://www.copenhagenconsensus.com/sites/default/files/climateemissionsabatement.pdf). Note that this is measured in metric tons of carbon, and so to convert to metric tons of carbon dioxide we need to multiply this figure by the ratio of their relative molecular masses, 12/44, giving a social cost of carbon of $32 per metric ton of carbon dioxide. If we were to use the mean rather than the median estimate, we would get a social cost of carbon of $48 per metric ton of CO2.↩