Giving now vs. later
by paulfchristiano
It has been observed that money spent helping the poor compounds over time in the same way that profit-oriented investments do (see Holden Karnofsky, or Giving What We Can). When I support the world’s poorest people I’m not just alleviating their suffering, I’m increasing the productivity of their lives. The recipients of aid go on to contribute to the world, and their contributions compound in turn.
Moreover, one can argue that the returns to aid are much higher than achievable financial returns. Equities make around 5% in real terms, while the world’s poorest earn returns upwards of 20%. And that’s not surprising—more people are interested in making money than are interested in improving the world, so we should expect it to be harder to find opportunities to make money than opportunities to improve the world.
So, the argument goes, altruists ought to donate as quickly as possible, doing good works that will compound out there in the world much faster than they will compound in their bank accounts. Make sense?
I think this argument is mistaken.
Here’s the problem: if you give your money well, it might compound much faster than it would have in your bank account—but only for a while. Over time the positive effects will spread out more and more, across a broader and broader group of beneficiaries, until eventually you are just contributing to a representative basket of all human activities. Eventually, my investment will compound at the rate of world economic growth, rather than at the particularly promising interest rate I was able to originally identify.
When I leave money in my bank account, it compounds slightly (1-2%, conservatively) faster than world economic growth. It does this for years, until I decide to spend it, for example on developing world aid. At that point it will earn anomalously high returns for a while, before being spread out throughout the world and then compounding in line with world economic growth. If I donate a year later, I earn 1 extra year of market returns, and 1 less year of compounding in line with economic growth. That’s a good deal—an extra 1-2%.
What I really care about is the extra multiplier I get right after I donate my money, before it resumes compounding at the usual rate. E.g., if my investment compounds 5% faster than economic growth for 15 years, my money is being effectively doubled, when compared to an investment that compounded with growth all along. That multiplier isn’t changing a lot over time (it might increase or decrease as good charitable opportunities arise or dry up, but there’s no good a priori reason to expect it to decrease steadily with time).
So if I’m considering donating to a cause like developing world aid, I shouldn’t donate sooner rather than later just because poor folks can earn higher returns than I can. I should instead look at the total multiplier I’m getting on my money, and donate iff I expect that multiplier is declining faster than [(market rates of return) – (economic growth)]. I don’t think we’ve found many causes for which this is plausibly the case, and I think that aid is definitely not one of them.
The basic issue is that philanthropists aren’t much less selfish than normal people, instrumentally. Even if you care about other people, you would much prefer that you make $1 than that someone else make $2, since when you make $1 you can use it to create more than $2 of value. (Of course this wouldn’t be true in a world with many altruists, in which the returns to altruistic endeavors were less fantastic.)
In the very long run I expect interest rates to fall / growth rates to rise until the two are equal, and I expect the multiplier on altruistic investments to decrease until it’s 1. But those events look to be a long way off.
Is giving later really an option?
There are at least four other reasons to give now rather than later:
- Giving can send a social signal, which is useful for encouraging more giving, building communities, demonstrating our generosity, and coordinating with charities. (Who are more likely to take a community of donors seriously than a community of people saying “maybe we’ll donate in the future.)
- Our priorities may change over time, and we would prefer that our future selves not have the option to keep money for themselves.
- Eventually we will die, and it is very hard to build institutions which will carry out our mission after our death.
- Taxes favor giving as you earn. (You can only deduct donations against this year’s income, and in the US at least you can only deduct 50% of your income in this way, so someone looking to donate a large fraction of their income will be deduction-opportunity-limited, and will want to use up all of the available deductions each year.)
I believe that these are serious considerations. However, “give now” is just one possible solution to these challenges, and if “give later” would otherwise be a much better strategy, it is worth searching for alternative solutions. Moreover, I expect that I’ll have a much better idea of how to spend charitable dollars in the future. So I would like to hold on to my money even if it weren’t earning interest, and I would need to confront these problems anyway.
Donor advised funds, or more generally donating to 501(c)3’s which invest, may resolve some of these problems. Giving to a DAF serves as a binding commitment to give, and therefore overcomes problem 2 and some of problem 1. Contributions to DAFs have the same tax status as donations, so overcome problem 4. Potentially the existence of such funds could also facilitate community-building and coordination with charities, perhaps even more effectively than an established record of giving.
At the moment donor advised funds typically charge management fees of about 0.6% / year on top of normal management fees, and tend to place more restrictions on grants than are legally mandated. In principle, it seems that US charities are free to invest arbitrarily (and subsume arbitrary for-profit enterprises), and use investment income to finance their activities or make grants.
It is tempting to imagine a charitable fund designed to accommodate very long-term giving, organized as a 501(c)3 and not charging significant annual management fees. Such a fund might seek out philosophically aligned donors and adopt a governance structure which codifies their shared values (representing a more credible commitment to to the charities that might want to coordinate with those donors). It might go further and actively manage its funds, hiring philosophically aligned traders and analysts, who would be happier to see the surplus they create split between themselves and philanthropists with similar values. I don’t know whether this is possible, but it seems worth thinking about.
Problem 3 appears to be significantly more difficult, but might be addressed in part by the same mechanisms. Donor advised funds provide a legal framework in which donated funds can be preserved without suffering from estate taxes or the mandated 5% payout of foundations. If there were enough like-minded philanthropists, it may be possible to allow each philanthropist to pass of advisory control of their funds to a like-minded successor before they became unable to oversee them themselves. Making such a fund stable enough to resist gradual value drift or pillaging would be a significant challenge.
The historical record is conspicuously devoid of successful enterprises of this kind, which suggests that future outings in this genre are doomed to fail. But I don’t think this is such strong evidence:
- I know of few serious attempts to create institutions which would build up wealth while resisting value drift or collapse. Today may be special just because today we think of different plans than our ancestors thought of.
- We have access to more effective communication/coordination mechanisms than were available historically. It may now be possible to reliably identify a like-minded and similarly able successor who shares your values even if the social prevalence of those values is one in a million or lower.
- I think there is a reasonable probability that we are in an unprecedentedly stable legal and political regime which will last longer than any preceding regime. Moreover, we now have a global political infrastructure which lets us build financial entities that can survive changes or collapse in one country by moving to another.
So if “give much later” looks like a good strategy, I think we should seriously consider trying to make it work, rather than throwing up our hands and saying “at least we can get some value by giving now.”
Can good works grow faster than the world around them?
The argument above is based on an empirical claim: good deeds tend to grow at the same speed as the world at large, rather than growing much faster. Equivalently: measured as a fraction of the world, the positive changes we effect don’t continue growing indefinitely. (While money in your bank account steadily increases as a fraction of the world’s wealth.) For some interventions, like existential risk reduction or undirected economic boosts, this is true almost definitionally (and I consider risk reduction an unusually forward-looking strategy, rather than unusually short-sighted).
Something that compounds faster than economic growth necessarily represents an increasing fraction of the world, and if it continues compounding faster than economic growth for long enough it will eventually make up a large fraction of the world and single-handedly increase the growth rates. This is rarely realistic for the kinds of investments we consider.
Instead, what happens is that the returns from an investment are typically not that concentrated. If I improve the educational system, this may make the world significantly better, but very little of that improvement will occur within the educational system itself. Instead, that improvement will be scattered out across many different endeavors (which will in turn spread their outputs even more widely). Even if the returns to education are very large, they will eventually spread out through society and compound at the usual rate unless the product [(returns to education) times (fraction of returns that are reinvested in education)] exceeds the economic growth rate.
In general, for a class X of investments to compound faster than economic growth rate, it would need to be the case that [(returns to X) times (fraction of returns reinvested in X)] exceeds the growth rate. If X is a small or very specialized class of investments (like education), it is hard for “fraction of returns reinvested in X” to be large. If X is a large natural class of investments (like the economy of the US), it is hard for “returns to X” to be far above market returns.
Investment seems to have anomalously good internal returns. I think this is possible because someone is actively applying significant effort to ensure that the returns from one opportunity are reinvested exclusively in the most promising opportunities (and a lot of effort has gone into building the legal and economic infrastructure that can support this). The same thing applies within a firm, although usually firms eventually encounter diminishing returns and their growth slows or stops.
Even so, the above argument implies that it is impossible for an investor to earn market returns for too long (or else they would eventually have all of the money). In practice, investors are able to earn market returns while they invest, but eventually they take their money out of the market or die, and they do not form lasting lineages of investors.
There may be other opportunities that have the same important properties as investment. For example, I might be able to efficiently support individuals or organizations that share my values, and therefore might be able to pursue other high-return investments after they have used up their own capacity for growth. It is plausible to me that supporting a growing community of “effective altruists” might have this effect.
Catch-up growth in developing countries is a plausible source of temporary higher multipliers for foreign aid. I predict that growth in China will much exceed world growth for a good while, and that this too shall pass as it approaches the technological frontier.
Really appreciate the post Paul. There is not enough written on this.
As Carl says though, growth in incomes for poor folks of a rapidly developing country could exceed average investment returns for the next few decades at least. If you were only expecting to wait that long that consideration could dominate.
For poverty-focussed donors, there is also a real chance that the best giving opportunities a few decades from now will be less effective than those today, if we continue to see rapid improvements in health, and reductions in poverty. On the other hand, schemes to generate leverage may get more effective in part due to the ideas of the EA movement.
But, like you, my guess is that speeding growth in the EA movement, if you think it is effective and that rapid growth is more useful than it is risky, dominates these issues.
>As Carl says though, growth in incomes for poor folks of a rapidly developing country could exceed average investment returns for the next few decades at least. If you were only expecting to wait that long that consideration could dominate.
It seems to me that there are basically 2 relevant rates of return, the interest rate and the growth rate, plus a constant one-time multiplier you obtain when you give your money (before it spreads out throughout the world). I agree that upcoming growth in the developing world increases the one-time multiplier for giving in the developing world. But if you wait to give, there will be different one-time multipliers in the future. To first order, what you are trading is “Investment returns now, giving multiplier later” vs. “giving multiplier now, growth later.” What you need to compare is “investment vs. growth” not “investment vs. giving.”
As you point out, if there are good opportunities now but won’t be in the future (perhaps because extreme inequality is a transient historical phenomenon) then we must also trade off against the vanishing opportunities. But that requires a very separate argument. For example, it would be compelling if you showed that a philanthropist in the 1800’s would have faced much better giving opportunities, and one in the 1600’s would have faced better opportunities still, since that would support a historical record of opportunities shrinking and we could start to talk about how fast that shrinking is occurring.
(Also, I don’t think the multiplier for developing world aid is the biggest available. I would guess that tech development is higher.)
I think that growth of the EA movement is good and may well be the best thing to work on right now, but I also caution that in this case as well it is possible to be misled by looking at the short-term returns. I pointed this out in my post on time-discounting. I think we should instead consider the fixed multiplier you obtain by spending 1 year to cause an N-person, mature EA community to arrive (1/K) years sooner.
Notionally, K is the current size of the EA community, and so you get a multiplier on your actions which depends on the ratio between the current size of the community and the long-term size. But as we scale up the targeted size, we (1) run into doubts about the extent to which the members of that large community are similar to us and (2) growth of the community at that scale becomes increasingly entangled with exogenous factors.
Hey Paul,
Great article, I really enjoyed discussing this with you. At the beginning of the article, you write that you get 1-2% more return in your bank account than the world economy gets overall. According to the CIA world factbook, world economic growth has been 3-5% for the past few years: https://www.cia.gov/library/publications/the-world-factbook/geos/xx.html
What bank do you use? Because I’m switching!!
-Eric
Wikipedia gives 4% growth worldwide over the last 110 years. I think a reasonable basket of equities worldwide earned something like 5% over the same period (while US index funds earned 6.5%). I discuss this a bit in my post on time discounting, though I’m not super confident of the numbers.
Maybe you should have said “stock” instead of “bank account” – I think most people think of a checking or savings account when you say “bank account.”
Banks pay much lower interest in exchange for the ability to get your money back at will. You can get higher returns by buying GICs that lock up your money or bonds (and there does seem to be an equity premium).
Plus interest rates are unusually low right now, for several reasons.
[…] written about saving vs. giving before, focusing on the issue of interest rates vs. returns on good deeds. But for now, I think there is a […]
[…] Paul Christiano, Giving now vs. later […]
[…] written about saving vs. giving before, focusing on the issue of interest rates vs. returns on good deeds. But for now, I think there is a […]
For point 4 under “Is giving later really an option?” in the US you can carryover charitable deductions that exceed 50% of your income (for up to 5 years). See section 4.2 and the corresponding example here: http://en.wikipedia.org/wiki/Charitable_contribution_deductions_in_the_United_States
I think this goes backwards—if you want to donate before you have income you can do that and roll the donations forward, but if you want to donate after you have income you can’t.
[…] Giving can send a social signal, which is useful for encouraging more giving, building communities, demonstrating our generosity, and coordinating with charities. (more) […]
[…] we may eventually fall back to what we would have done anyway–my own view is that this wouldn’t be so bad–or we can reevaluate the situation and perhaps hold out for an opportunity even further in […]
Nice post, and an interesting topic!
I think it is important to distinguish between social and private rates of return though. As altruists, we should care more about the former than the latter.
Your argument is a lot more relevant for the option of donating to GiveDirectly, where you give cash to somebody and most of your impact will come through that person’s consumption and them investing it and getting a high (private) rate of return in the short run.
However, I think your argument is less relevant for donations to, for example, deworming organizations. Here, you are choosing what the money is being spent on, and it is an activity with a remarkably high (social) rate of return – both in terms of short term health (and perhaps long term if NTDs really can be eliminated/drastically reduced in the long run), and future incomes through improved education outcomes. Miguel and Kremer (2004) do a back of the envelope calculation: ‘Under these assumptions, deworming increases the net present value of wages by over $30 per treated child at a cost of only $0.49.’
I think you mention this, but I also think that if you are donating to global poverty causes we should expect the rate of return of our donations to decrease over time, due to diminishing marginal returns. (Hopefully) we are doing a decent job at finding the best interventions, and as these ‘fill up’ (like AMF), and less people live in poverty, we should expect the ‘low hanging fruit’ to be gone in 20-30 years.
I agree that social returns to e.g. deworming might be much higher than private investment returns to the poor; but it seems like the overall argument applies more-or-less unchanged. The social returns take the form of benefits to the poor, which then compound at the same rate as GiveDirectly (modulo adjustment for accruing to a different sort of person), and then the story is just the same. So we can think of this as a better bang-for-your-buck. If you believe Miguel and Kremer, an extra 60x, + whatever net benefits aren’t captured by wages, – a few times for growth discounting, though honestly my real estimates are quite a bit lower than this. Do you think a further adjustment is needed?
For the most part the interventions we are investing in (e.g. including deworming) are introduced by technological or social developments. So if we are thinking about using up the low hanging fruit, what we care about is whether technological progress is introducing new quality-adjusted opportunities faster or slower than they are being used up.
There are a few particular reasons to suspect “slower” that I have thought about, though I’m sure I’m missing many.
One reason is that as the world gets wiser (e.g. as philanthropists get better at picking winners, or society gets better at allocating resources to important projects more generally), people will be more quickly addressing whatever good opportunities exist. I think this is a really big consideration: are you getting wiser faster or slower than the world at large? For most EA’s, and certainly for myself, I think the answer is “faster.” So I think that this particular discount is more than offset by my own decisions getting better. That said, if you weren’t learning particularly quickly then this consideration would turn the other way (or if you are learning quickly but are even more optimistic about the world as a whole).
A second reason is that you might expect further progress to be fundamentally yielding worse opportunities over time, because of diminishing returns. In 2000 we may be developing a cure to a painful and debilitating disease which a philanthropist can fund, while in 2100 we may be slightly increasing the quality of people’s leisure which a philanthropist could fund but to lesser effect. I don’t think this is obvious in general though. My own view is that the most important philanthropic goal is having an influence over the far future, and our successive technological developments has been more rather than less important on those grounds, as we have acquired the ability to permanently and irreversibly mess stuff up. Eventually I agree that these returns will also start diminishing.
A third reason that applies if you think that some current issue X is very important is that issue X will tend to become less important over time as the world changes. If you are interested in climate change or global poverty for example, those are likely to eventually become a less attractive thing to work on. But overall I don’t see great reasons to expect today’s pressing issues to be more important than tomorrow’s pressing issues, except because of the first two considerations I mentioned (the world becoming wiser, and diminishing returns to innovations), so I think this would be double-counting. On the other hand, you might have particular insight into a present-day issue, and that would be a consideration in favor of working on that issue before it disappears.
[Of course all of this is on top of growth discounting.]
Hi Paul – really sorry I didn’t respond sooner to this comment. I’ve just had reason to revisit this blog and realized I never replied!
Re. social returns – I think I agree that your model can accommodate most cases of social returns by simply having a larger initial return, and for a longer period, before the usual argument about these dissipating into general growth. But I think there may also be some investments with a particularly high rate of return which is sustained indefinitely, or certainly over the period of savings to give later that you are considering – essentially investments that have dynamics such that returns do not dissipate into general growth.
The best example I have in mind here is the deworming example. Imagine that you have the opportunity to eliminate schistosomiasis (or some other NTD) from a given geographic area (and for simplicity assume that you can be fairly sure this elimination is permanent). Then I don’t think it is right to think of this as a one off benefit to current residents that will then dissipate into general growth. Instead I think this should be thought of as a benefit to successive generations who live in an area without worms – hence the ‘initial return’ is repeated each time period (until the worms would otherwise have been eliminated – this counterfactual will determine when the general growth argument then kicks in). This type of argument doesn’t necessarily rely on a binary {eliminate, don’t eliminate} case, but it does require a certain type of dynamics – i.e. a permanent shift onto a different path of costs/benefits so benefits are sustained at a higher rate than growth over a longer period. I don’t know enough about the dynamics of epidemiology to know how plausible this is in most cases. In the worm case, reinfection is common, and worm loads go back up after 6-12 months, but my understanding is that with large enough sustained investment over a large enough geographic space, you can permanently reduce or eliminate worms.
Re. the argument for investments in far future – I agree with the logic, and am persuaded by the arguments in principal. This isn’t really a discussion for this comment thread, but my personal view is that the signalling aspect of donations in the next few years are particularly important in growing the EA movement, and that this dominates decision making for me (I funnel most of my giving through movement building now – running giving games etc). I am also yet to be convinced that we can be very confident at all about investments made in far future stuff, especially relatively small donations, though I am quite ignorant on this. Plus if larger, lumpy investments are more productive in this area, this is another reason to save to give later.
[…] analyse this problem, we use a simple model, which we were introduced to by Paul Christiano. Although simple, we think this model captures and explains the key considerations. However, it […]
[…] Once the benefits spread out even further, we can think of them as contributing to one big pool of world assets, growing at the world growth-rate. We can see that the growth of the benefits produced by the intervention must eventually correspond to the world growth-rate, otherwise they’ll occupy a larger and larger share of the world’s assets see more detail here. […]
>Moreover, one can argue that the returns to aid are much higher than achievable financial returns. Equities make around 5% in real terms, while the world’s poorest earn returns upwards of 20%. And that’s not surprising—more people are interested in making money than are interested in improving the world.
I think this _is_ surprising: the opportunity to earn such high returns is not only available to the donor, but to the recipient of the donation (at least for adults), in the form of credit. So the disparity between social and market returns can’t be explained by the comparatively low interest in philanthropy.