Thoughts on patient philanthropy

Summary

• Risk-free interest rates are currently very low. Therefore, patient philanthropy can only work with risky assets, such as stocks.
• For risky assets, the attractiveness of such investments depends on one’s degree of risk aversion. Risk-averse investors should consider the certainty equivalent (i.e., the guaranteed return that’s equally desirable as the risky return) of investing rather than expected return. 
• There are some reasons why patient philanthropy is differentially good for those who primarily wish to reduce suffering, compared to those with other longtermist goals.
• Overall, I believe we should pursue both financial and non-financial investments (such as movement-building and cause prioritisation research). This is in part because there are not that many highly urgent interventions to directly reduce s-risks right now, and there might be more in the future. 

Introduction

Patient philanthropy, or patient longtermism, is the idea that we can have more impact by not spending resources (in particular, money) now, but instead investing and growing them to spend more in the future. For an introduction to the idea, see here; for more in-depth discussion, see:

• 80,000 Hours’ podcast with Philip Trammell
• Philip Trammell’s paper 
• Some discussion by Robin Hanson: 1, 2, 3
• Relevant EA Forum posts: 1, 2, 3, 4

In this post, I outline some additional thoughts on the degree to which suffering reducers should invest rather than spend now. 

The term “invest” or “patient philanthropy” also includes non-financial “investments” like movement building and cause prioritisation research. Depending on how high the “rate of return” of such non-financial investments is (see here for more details), those forms of patient philanthropy may be better than financial investments. However, in this post, I will focus on the latter, as financial investments are easier to analyse and less dependent on specific activities.

(Disclaimer: This post is mostly a write-up of notes on specific questions I’m interested in, and therefore not suitable as an introduction to or as an overview of the topic of patient philanthropy. It may also be hard to follow if you are not yet familiar with the ideas.)

How large are investment returns?

One possible implementation of patient philanthropy is to invest in risk-free assets, earning the risk-free interest rate. However, in the current (August 2020) market environment, the risk-free interest rate is very low (see here), and is arguably below the expected growth rate or inflation¹. 

This calls into question a common story behind patient philanthropy: namely, that most people are impatient, whereas longtermists are not, resulting in high interest rates from which the latter can profit. While that sounds plausible, the current low-interest environment contradicts that story; apparently, some sort of “patience” seems not unique to longtermists. This is rather puzzling, and it’s not clear what is going on. I suspect that one factor is that many wealthy people don’t quite know what to do with their money – that is, immediate consumption seems pointless – so they just save and invest it, resulting in a form of patience and thereby low interest rates. (Also, interest rates are primarily set by central banks rather than reflecting people’s pure time preferences.) 

Thinking in terms of centuries or millennia rather than decades still seems unique to longtermists. But if investing is good over a 200-year timespan, then it must also be good over 20-year timespans, since the (expected) return over 200 years is the product of (expected) returns over many 20-year periods. So perhaps ultra-patient investors can’t easily profit from the fact that others are merely patient on the scale of a few decades, because the latter is enough to drive interest rates down.²

To achieve higher investment returns, one needs to invest in risky assets, in particular, in the stock market (possibly using leverage).³ My estimate of the expected (nominal) return of a global stock portfolio is ~5% per year⁴, which does significantly exceed the growth rate (or inflation). The gains from this kind of investment are perhaps more due to the arguably greater risk tolerance of altruists compared to the average investor, rather than patience – but of course, either can support the idea that we should invest rather than spend now.⁵ 

However, when using risky assets, expected returns are not the right yardstick (unless one is entirely risk-neutral), as there is also some loss of expected utility due to the risk involved. It seems better to consider the certainty equivalent, which is always lower than expected returns (unless one is entirely risk-neutral). For lognormal market returns with certain parameters, and a given level of risk aversion, the certainty equivalent can be calculated using the formula on slide 12 here. 

For instance, suppose we invest with 2x leverage, the risk aversion parameter (using isoelastic utility / CRRA) is 0.5, and we model yearly returns of the leveraged portfolio with a lognormal distribution with parameters mu = 0.02, sigma = 0.4. The expected return is 10.5% per year, but the certainty equivalent is a yearly gain of 6.1%.⁶ This is much less, but still exceeds the growth rate plus the philanthropic discount rate by a significant margin. Investing therefore seems quite attractive given those assumptions. (By contrast, for an investor with logarithmic utility and 1x leverage, the certainty equivalent is 3%⁷, which is much closer to growth or inflation rates.)

Patient philanthropy seems particularly attractive from a suffering-focused perspective

We can, in general, do more to prevent a risk in the years (or perhaps decades) before it materialises, whereas (direct) work done a very long time in advance is much less effective.⁸ Since we are arguably⁹ far away from serious s-risks (because they require technological maturity, e.g. due to transformative AI), patient philanthropy appears particularly appealing from a suffering-focused perspective. Rather than work (directly) on speculative distant risks now, it makes sense to focus on building up resources (through financial and non-financial investments) until we are closer to s-risks and can see more clearly how s-risks occur and how we can prevent them.¹⁰

By contrast, longtermists that are worried about existential risks have more reason to spend now, if they think such risks will plausibly materialise in the nearer future.¹¹ Put differently, hingeyness is value-relative, and the distribution is arguably more skewed towards later times if one is primarily concerned about s-risks. (In particular, most s-risks materialise later than x-risks.)

Of course, this is not an open-and-shut case. In particular, if one believes that radical change will happen soon, e.g. due to transformative AI, then we are now at a unique window of opportunity for s-risk reduction as well as other longtermist goals. In this case, hingeyness is extremely high for many value systems, and investing arguably does not make much sense.¹²

Even without AI or similar transformative events, one may think that a harmful political or social dynamic will be locked in much earlier (say, over the coming decades) and will ultimately cause the s-risk later on. Or perhaps events that happen soon influence later s-risks through some path dependency, so that spending now is most effective. However, while this is surely possible, it seems that we are (for now) unable to identify clear pathways of this sort, or promising levers to influence them, given vast uncertainty. (Candidates include reducing malevolence, moral circle expansion, or improving the political system.) 

Perhaps the strongest argument for spending now is that work on s-risks gets very little overall funding right now (unlike e.g. work on biorisk or nuclear risk). It seems plausible that it will never be that neglected in the future, especially if movement building efforts are successful. Assuming diminishing marginal returns, this can be a strong reason to spend sooner. 

—

Appendix: Investment returns can be correlated with hingeyness

For instance, if there is a major catastrophe or a third world war, then risky assets will likely plummet in value and patient philanthropy will not work well.¹³ If such a scenario offers much better opportunities for impact than normal times (which is unclear), then that is a potentially strong reason to be more risk-averse (in the hope that we have money left to spend for large impact in those scenarios). Higher risk-aversion is indirectly a reason against patient philanthropy because investing in stocks is not that good if you are risk averse (and less risky assets do not offer high returns).

Conversely, in the case of continuing peace and prosperity, returns will likely be high but perhaps charities will not be very funding-constrained (and such a scenario may contain only small amounts of suffering anyway). 

On the other hand, one could also argue that both investment returns and hingeyness will be high in the case of a growth mode change (i.e. much faster economic growth, resulting in a GDP doubling in less than a year), e.g. due to transformative AI. Overall, it seems unclear what the upshot of such considerations is.

1. Discussions of patient philanthropy (e.g. Philip Trammell’s paper) often compare interest rates to the growth rate, but it is not clear to me whether the growth rate is the right yardstick, rather than inflation or something else. This depends on what exactly one wishes to do with the money. For instance, in the special case of cash transfers now vs. later, using the growth rate seems to be right; whereas inflation seems to be a better yardstick when using money to help wild animals (because the cost scales with inflation but not growth).
2. This “argument from modularity” may not hold for all investments: for instance, 100-year leases (as mentioned in the podcast with Philip Trammell) may still allow deals of the form “the impatient receive the near-term in exchange for the patient inheriting the long term”, because the timescale is too long for most people. But it does hold for the most common investments, such as stocks, bonds, or real estate.
3. Note that expected stock returns are, at least theoretically, tied to interest rates. That is, the expected return of stocks is the risk-free interest rate plus the risk premium. Therefore, the low interest rate is also a reason to expect lower stock returns than we’ve observed historically.
4. Historical stock returns are even higher – up to 10% per year – but I think this needs to be adjusted downwards to account for survivorship bias (i.e. we are observing only those stock markets that were successful, not those that collapsed) and the current low-interest environment (cf. previous footnote). Also, current (Sep 2020) valuations of (US) stocks in P/E are fairly high, which perhaps suggests lower future returns.
5. If we could get stock returns (plus risk) without a time delay, we’d perhaps do that and spend now; but it’s just not possible to get this gain from lower risk aversion without a time delay.
6. Calculated using the above formula on slide 12 here.
7. I’m modeling yearly returns as lognormal with mu = 0.03, sigma = 0.2. Expected returns are 5.1%.
8. There are also arguments to the contrary. For instance, earlier work may be more effective due to field-building effects. See here for a review.
9. A possible exception could be scenarios where transformative AI happens soon, as discussed below.
10. Another possible argument for this view is that work on suffering-focused cause prioritisation has (allegedly) failed to identify any interventions or opportunities that are clearly highly effective in terms of reducing s-risks.
11. I think the probability of extinction soon is very low, but that discussion is beyond the scope of this post.
12. This also depends on whether civilisation reaches a steady state quickly after transformative AI, or whether and how money can still translate to impact in a “post-TAI” world. Another possibility is that a compromise or world government would allocate power partially based on wealth (see e.g. here).
13. Also, perhaps it is quite hard to preserve wealth in such scenarios. You might just be expropriated. For instance, in WW 1 and 2 states needed vast amounts of funding for the war, so they usually ended up expropriating the owners of capital in one way or another, e.g. through taxation, printing money, or forced bonds that weren’t repaid. But perhaps smart investors would be able to bypass those constraints.