If we want to donate money, should we give it away now or invest it to give away later? The answer depends on many considerations, including our expected rate of return, the chance of our personal values changing, and the question of whether we live at the “Hinge of History” — a time with high-impact opportunities that will soon vanish.
In this talk, Phil Trammell of the Global Priorities Institute discusses how we can evaluate these considerations and change our giving strategy in response to what we learn.
Below is a transcript of the talk, which we’ve lightly edited for clarity. You can also watch it on YouTube or discuss it on the EA Forum.
Today, I'll be sharing some thoughts on how to think about the timing of philanthropic efforts. There are a lot of considerations that go into this question.
I don't have time to cover most of them, but here are four that I'll discuss, along with a simple model for weighing them against one another:
These are the considerations we'll weigh against one another. What we'll find is that in ordinary times, if you just assume that hingeyness is constant, you don't need to think about it; philanthropists should probably invest almost all of their resources.
The caveat I'll make is that I'm using a rather broad definition of “investments.” An investment is anything that allows you to spend money now and results in more money going toward your moral goals in the near future. So fundraising and movement building count. Building schools doesn't count, because even though it increases the earnings of the people who go to the school [and receive an education], the money's out of your hands once the school's built.
But introducing the hingeyness consideration makes things ambiguous. And if the hingeyness is extreme enough, then you should spend a lot.
I'll just say a few words about the model and some of its implications.
First, what is the interest rate?
It's quite high. The U.S. stock market has averaged an interest rate return of over 7% per year for the last century. At that rate, your money doubles roughly every decade, or multiplies by a factor of two to the tenth power every 100 years, which is 1,024. That's probably not the relevant interest rate for our purposes, for various reasons. Let's say it’s 4.4%.
If we are patient philanthropists — that is, we want to do good but don't care about when we do it — we should consider whether this interest rate [4.4%] is high or low. Is it an appealing opportunity for investing, spending, or even borrowing?
There's a very simple observation that should lead us to think that [an average 4.4% interest rate] is, from a patient philanthropist’s perspective, probably high most of the time. This is what's known as the Ramsey formula, which is the backbone of the economics around discounting. Let's say there's consumption growth of 2% per year. And there's a parameter, eta (𝜂), which captures the extent to which marginal utility for people diminishes in consumption, such that it gets more expensive to help someone at a rate of 𝜂 multiplied by the rate at which they’re getting richer per year. So if people are 2% richer next year, it's 2.4% more expensive to buy them a util [a unit of human welfare].
The reason why people [might choose not to] invest $1.044 next year instead of $1.00 now is they discount their own welfare at some positive rate — let’s say 2%. That's what delta (𝛿) denotes. An individual’s own mortality risk could be part of that. But if you're a patient philanthropist, you don't care whether you help someone now or help their descendants, so you should remove delta from the equation. If you wait, you can do 2% more good than you'd be doing by just giving this year, if your goal is to help people simply by increasing their consumption. Impatience is incorporated into market interest rates; in general, if you're patient, it’s a good idea to invest.
There are some obvious caveats to this.
You don't want to completely remove the [so-called] pure time discount rate, [since it actually incorporates risks] you do also care about — risks like extinction, expropriation, or value drift. And you won't actually do 2% more good each year than if you wait to give, because if [the amount of money you set aside gets] really big, then further growth to your fund will have diminishing returns. But if your goal is just increased human consumption, the [risk] of getting so big that you'd actually push down people's marginal utility probably [is remote]. That wouldn’t happen for a very long time.
The main takeaway, therefore, is that in normal times, you probably want to take advantage of this [opportunity and give later].
Upon first encountering this simple observation, people often [voice] a litany of objections. Unfortunately, I won't have time to go through all of them. I’ve just listed them [on the slide below] with one-liner responses.
For example, people sometimes say, "Well, what if the poor are getting richer so much faster than the interest rate that it's actually getting more expensive to help them each year?" This can only happen for a temporary period of time. Eventually the good that you'll do by investing, if nothing structurally changes, will outstrip the good that you would have done by giving to the poor now — even if the factor is 250 to 1. In that case, you'll have done more good in 275 years or so by waiting, if this delta equals 2% forever.
If your goal is just to help people, and you think that people are imperfect at helping themselves — in particular by being impatient — then the paternalism objection doesn't really hit home.
I want to share some intuitions beyond the somewhat glib one I just gave for thinking that patience might be a super power that, if we make use of it, allows us to have a really big impact on the future.
Suppose you're at an auction house and some baseball cards are up for auction. You're a baseball fan; you and the friends you went to the auction with are obsessed with baseball cards. [If no one else at the auction is a baseball fan], you should be able to outbid the other people at the auction house for the baseball cards.
That's similar to the position we're in [as philanthropists who care about the future]. The future, in some sense, is this large, four-dimensional object, and [people in the effective altruism movement] think about it a lot — and most people don't. There should be a way for [patient philanthropists] to buy [the future] from its current owners. There'd be a weird market failure if that weren't possible.
In fact, we can literally do this. If you buy land and then sell a 100-year lease on the land, you're buying the right to say what happens to the land from 100 years out until the end of time. And a 100-year lease on the land usually costs about 90% as much as the land. Basically, for a 90% discount, you're getting the land, or almost all of it [if your ownership persists for hundreds or thousands of years after the lease is up].
That’s what investment is. It’s saying, "Hey, impatient person, you can have some resources just for a while, and then give more back to me later."
People sometimes speculate that we're living at a very special time with respect to giving — that our donations can have a giant impact on the future. But what this amounts to is the observation that we're also living at a special time with respect to the value of investing. If the dynamics of pure time preference play out, eventually patient people will own more and more of the world. Interest rates will no longer incorporate pure time preference [to the same extent, because most assets will be held by people who don’t have such preferences]. In a sense, the future will have been bought. So this, too, is a fleeting opportunity.
The idea [in the slide above] is to suppose that you face diminishing returns parameterized by eta (𝜂). Your impact at a given time is some constant, multiplied by the amount you’re spending, transformed by these diminishing returns. And suppose that a community of patient philanthropists got together and acted as a single philanthropic agent. What if the community had no income stream, but began with a fixed pool of assets?
Let’s assume that's the situation we are in. How much should we spend? It turns out that it's the number represented by the formula [in the slide above]. And if eta equals one, which is the case of logarithmic impact, then it's really simple: It's just your philanthropic discount rate.
If there's a 0.5% chance per year that either we go extinct or your money is stolen, then that also turns out to be the amount that you should spend every year. And you should invest [the other] 99.5% of your resources every year.
What if, on the other hand, you start out at the other extreme? You have no resources, and the community is a group of young people with nothing in their pockets, but they have an income stream. Then what fraction of that income should be put to philanthropic purposes each year, and what portion should be invested?
I think most small donors think of themselves as being in this position, but they give everything that they plan to give each year instead of investing it in a fund to give later. But it turns out that you should spend this ratio: the philanthropic discount rate divided by the market impatience rate — perhaps about 25%.
Now if this played out forever, then you’d grow indefinitely a bit faster than the economic growth rate.
In 400 years [a community investment project] might [yield something] like 10% of the global wealth. In response to that, people will say, "If this is so easy, then why hasn’t it been done before?" That would be a whole separate talk. But people have tried to do it in various ways. I don't think they've done a great job. It probably wouldn't work anyway [even if it were done well], but that's already accounted for in ẟ<sub>p</sub>. A 0.5% philanthropic expropriation rate would mean that the half-life of an effort to buy the future would be 200 years [by which time, there’s a 50% chance that your investments disappear because of some unforeseen event]. After 400 years, there would be only a quarter-chance that [your investments] would remain — but the fact that [this long-term strategy] hasn't worked isn't strong evidence that it is totally impossible.
Weighing against this model of patience is the possibility that we really are living in a special time. This possibility was articulated by the late great philosopher Derek Parfit. He said:
He said that in 2011. And he was talking about the next few centuries, so the hinge isn't very narrow. It's not like we have to give money this year because we're living at the hinge of history. Maybe the “hinge-iest” year will be in 400 years. It's worth distinguishing what we mean when we talk about this hinge.
You might also look back over the past few hundred years and ask which moments have been the most pivotal, leading up to the way that we’re currently thinking about what is best for civilization going forward. Maybe our values would have been very different if World War II or the American Revolution had gone differently.
John Adams said something similar [when he wrote] the U.S. Constitution; he foresaw that what he was doing would last for thousands of years.
Below is the same formula from before, but now you have this h(t) factor, which scales how much impact you have as a function of how much you’re spending at a given time.
I have no idea what the actual “hingeyness” schedule [of American history] has been, but maybe it looks something like this [see slide below]. I'm just illustrating the concept — it's a continuous version of the idea of being at the hinge of history.
We'll let h follow an arbitrary, finite Markov process, which just means that whatever state you're in, there's some chance that you could move to any of the other states.
Each has its own interest rate. Each has its own discount rate. And the optimal spending policy turns out to be the one that satisfies this formula:
To shed just a little bit of light on how the “hingeyness” consideration weighs against the [approach of waiting to let interest accumulate], let's work through the numbers to determine how quickly you should spend if you're at a moderately hingey time.
Trivially, if you just ratchet up the h high enough, then you should spend a lot. But let's look at an intermediate case. Let's say the philanthropic discount rate is, to my mind, relatively high: 1% per year. Let's say eta equals one. (It turns out that the interest rate actually doesn't matter in this case.) And let's say you have two states — one of which isn't so hingey (“state 1” or s<sub>1</sub>), and one of which is 10 times hingeyer (s<sub>2</sub>). Once you're in s<sub>1</sub>, you typically stay there; there's only a 10% chance that you'll be bumped into s<sub>2</sub>. And when you're in s<sub>2</sub>, there's an 80% chance you’ll revert to s<sub>1</sub>. So, if you find yourself in s<sub>2</sub>, you have a lot of motivation to spend.
It turns out that in s<sub>1</sub>, a community of patient philanthropists should spend 0.3% of its resources. And in s<sub>2</sub>, it should only spend 2.4%. Even in moderately hingey times, the patient philanthropist spends little.
How revisionary is this, really?
When we’re referring to a community of patient philanthropists, it’s a nebulous crowd. But let's think about the effective altruism (EA) community. There’s roughly $15 billion in EA dollars [available to be spent] if you count the Open Philanthropy Project (and others). You also have to count resources that aren't part of actual assets. Let's say there are 10,000 people involved in effective altruism, earning an average of $100,000 per year each. And let's say 10% of that, on average, is committed to EA causes, rather than being consumed privately. That would be $100 million a year, and EA wages would grow at the same rate as everyone else's. However, we would discount that amount each year to arrive at the present value of a given year’s future wages, because we're only considering the net present value. In other words, if we could borrow against all of those wages, how much would we have?
There is a geometric series, with $100 million divided by r minus g, yielding $5 billion of net-present value of EA wages. [When you add this to the $15 billion in currently available funding], that’s perhaps $20 billion. Therefore, if you think that we're living in a moderately hingey time right now — but not a time during which an event like World War III will determine everything that happens afterward — then we should spend something like $480 million a year.
The Open Philanthropy Project’s grants for the first third of 2019 totaled $125 million. Multiply that by three and it yields $375 million — and then there's everyone else (excluding the Open Philanthropy Project).
Perhaps we're already spending roughly what we would be if we [had determined that we’re living] in a moderately hingey time. But I don't think that’s why [we’re spending $480 million]. I think there are a lot of small donors who are donating 100% of the wages that they plan to give to charity as they earn those wages. Meanwhile, the Open Philanthropy Project is spending more slowly than it plans to in the coming decade or two because it's capacity-constrained with respect to research analysts. That's my understanding. Therefore, I think we might be spending about the right amount, but for the wrong reasons.
One justification for this could be that we really are living in a very hingey time, so we should be ratcheting up our spending. Another could be that much of the expenditure [qualifies as] what I'm calling investment [in movement-building]; it's a form of fundraising, like giving to 80,000 Hours, to [recruit] more people who will give later. But that doesn't seem to be the case if you [look at] what EAs spend their money on.
Another justification might be that in the future there'll be so many people who are sympathetic to EA exogenously, or who are part of a long-term trend of social values moving toward EA’s current values, that we should spend a lot now. In a sense, there's a lot of low-hanging fruit now that will soon be picked.
But if you don't think that — if you think that we're currently a bit inclined to overspend — where should we go from here?
One thing we might reconsider is earning to give. I know it was popular in EA before this influx of money came in — particularly from the Open Philanthropy Project — and now many people have deprioritized it. But maybe that was too strong of a conclusion to draw. We might also reconsider the importance of frugality. Even if there aren't a lot of seemingly great giving opportunities right now, you can always just invest. And actually, there's quite a lot of value in doing that.
As an individual donor, I think it might be good to consider [investing in] a donor-advised fund in order to reduce the rate of value drift. Donor-advised funds are tax-exempt funds that you can set up for yourself such that you give to the fund, let your money grow tax-free, and then eventually donate it to a charitable cause. You can't take the money out for yourself.
I think the final implication is that if we really are planning on these century-long timescales, we should think hard about how best to design long-lasting philanthropic institutions and coordinate [how different institutions plan their giving] as time proceeds.
Moderator: Thank you. Does your conclusion mean that EAs should consider finding good investment opportunities or [are you recommending that they] set up a [new] fund?
Phil: Those both seem like worthwhile things to do. I think something that should happen eventually — and that a lot of people have expressed interest in — is [to set up a] fund that explicitly meets the giving desires of people who think that we should save for a very long time.
But I'm not really recommending that. I'm just saying that if you conceive of all EA activity as this fund over time, maybe it should consider allocating more to efforts that involve saving for a long time. Does that make sense?
Moderator: Okay. To follow up, how do you take into account the fact that [if someone invests for the long term], other people or agents will be the ones actually spending [their] money?
Phil: Yeah, that [creates risk from] institutional value drift. If we think hard about it, we can lower the risk of that. We’ll never [reduce the risk] to zero. But one thing you might do is that, instead of entrusting this giant fund to one person and having them hand it down through the generations, you could set up a committee of people. They might vote on whom to swap out every 10 or 20 years.
You could also look at other institutions — for example, religious institutions that have been faithful to their values over time — and see what might be learned from them. It is a big empirical question. I haven't thought about it as much as I would like to.
Moderator: Thank you. There is a clarifying question from the audience: Can you explain what you said about buying land and renting it?
Phil: You can buy a patch of land and then sell a 100-year lease on it. You give someone else the right to do what they like with it for the first 100 years, and you'll be paid about 90% of what you paid to purchase the land. In a sense, you just paid 10% of the price of the land for the right to say what happens to it [after 100 years have passed].
Moderator: All right. Thank you so much, Phil.