The case for investing to give later

Edit 07/​07/​20 at 1.30 pm BST: Thank you for your com­ments so far! In par­tic­u­lar, there were a few use­ful com­ments on some of the ‘con­ser­va­tive’ es­ti­mates be­ing too op­ti­mistic. As a re­sult, I’ve up­dated some of those and re­placed the sheet model by this Guessti­mate model.

This model more ac­cu­rately rep­re­sents the un­cer­tainty I have about these es­ti­mates and doesn’t re­quire the use of vague terms such as “con­ser­va­tive”. Fur­ther­more, it in­cludes the “pa­ram­e­ter un­cer­tainty” fac­tor, which is found by com­par­ing the es­ti­mated yearly im­pact mul­ti­plier with the 10-yearly one: it seems to be an im­por­tant fac­tor that shouldn’t be left out of the anal­y­sis.

For clar­ity, I’d also like to fur­ther em­pha­size that the pri­mary pur­pose of this post is not to make a case for my cur­rent es­ti­mates, but to in­vite in­put on (1) the over­all model, (2) miss­ing fac­tors and (3) which fac­tors to pri­ori­tize for re­search. In­deed, my cur­rent es­ti­mates should be taken with a huge grain of salt, as I have hardly spent any time on most of them so far: re­fin­ing these es­ti­mates is the pur­pose of the re­main­der of this re­search pro­ject. How­ever, as men­tioned above, com­ments on where you think my cur­rent es­ti­mates are wrong are ob­vi­ously wel­come as well!

Introduction

After hav­ing to pause the pro­ject for a while, we have re­cently re­sumed work on the idea of a long-term in­vest­ment fund at Founders Pledge. The next step is a re­search pro­ject on the im­pact of ‘in­vest­ing to give later’ as a philan­thropic strat­egy more gen­er­ally. This will help us de­cide whether to launch the fund and to what ex­tent to pri­ori­tise it.

In this post, I out­line the key fac­tors that bear on this ques­tion as I cur­rently see them, af­ter pre­limi­nary re­search, and draw a ten­ta­tive con­clu­sion. Please note that these rep­re­sent my per­sonal views, and not cur­rently those of Founders Pledge.

I would ap­pre­ci­ate any thoughts on (1) im­por­tant fac­tors that are miss­ing, (2) faults in my rea­son­ing and method­ol­ogy, (3) which fac­tors to pri­ori­tize deep­en­ing out fur­ther, and (4) re­sources or con­nec­tions that would be helpful in do­ing that. Please leave these in the com­ments on this post or reach out at sjir@founder­spledge.com.

Be­fore launch­ing into the con­tent, I’d like to par­tic­u­larly thank Phil Tram­mell for the sub­stan­tial con­tri­bu­tion of his work on pa­tient philan­thropy (see also this 80,000 Hours pod­cast) to this pro­ject so far and for dis­cus­sion and com­ments. I’m also grate­ful to Michael Dick­ens, Sasha Cooper, and my col­leagues Ai­dan Goth and John Halstead for com­ments on a draft of this post. Fi­nally, as will be clear from the linked sources be­low, this work leans heav­ily on ear­lier work by other peo­ple in the effec­tive al­tru­ism com­mu­nity.

Edit 03/​07/​20 at 7 pm BST: It hap­pens Michael Dick­ens was work­ing on a very similar topic in par­allel, and just pub­lished his work as well. We both wrote our post be­fore see­ing each other’s, so any over­lap in con­tent is co­in­ci­den­tal.

Summary

I go into 7 key fac­tors that de­ter­mine how in­vest­ing to give later com­pares to giv­ing now for an in­di­vi­d­ual in­vestor-philan­thropist in the cur­rent situ­a­tion. To­gether, these paint a rel­a­tively strong pic­ture in favour of in­vest­ment: it seems plau­si­ble that one can have pos­i­tive real, net fi­nan­cial re­turns and a grow­ing share of the econ­omy in ex­pec­ta­tion—tak­ing into ac­count value drift and ex­pro­pri­a­tion risks—and that there are sig­nifi­cant benefits to in­vest­ment in terms of learn­ing. Pa­ram­e­ter un­cer­tainty likely fur­ther strength­ens the case for in­vest­ment.

Risks of as­set loss, risks of value drift, and un­cer­tainty about the availa­bil­ity of high-im­pact giv­ing op­por­tu­ni­ties over time weak­ens our con­fi­dence in this con­clu­sion, but not to the ex­tent that it changes it. There are how­ever some ways of ‘in­vest­ment-like’ giv­ing, such as en­courag­ing oth­ers to in­vest, which might com­pete with di­rect fi­nan­cial in­vest­ment.

Ex­clud­ing those ‘in­vest­ment-like’ giv­ing op­por­tu­ni­ties, my cur­rent best guess es­ti­mate is that an in­vestor-philan­thropist will on av­er­age be able to mul­ti­ply the to­tal im­pact of their funds by ~1.01 in one year and by >>10[1] in ten years. This is if they in­tend to spend the funds on longter­mist ob­jec­tives; I in­tend to add es­ti­mates from a short-ter­mist per­spec­tive at a later point.

Defin­ing ‘in­vest­ing to give later’

Here we con­sider ‘in­vest­ing to give later’ to be a philan­thropic strat­egy in which one en­gages in for-profit in­vest­ing with the in­ten­tion to donate the prin­ci­pal and ex­pected prof­its at a later time point.

In par­tic­u­lar, we’ll con­sider the case from the point of view of an al­tru­is­tic and strate­gic in­di­vi­d­ual in­vestor-philan­thropist with limited re­sources (<$100 mil­lion) and in the im­me­di­ate situ­a­tion, i.e. whether this in­di­vi­d­ual should in­vest the com­ing few years rather than spend this year, as­sum­ing he/​she can­not co­or­di­nate with oth­ers. This is dis­tinct from the ques­tion of what is the op­ti­mal spend­ing rate in the effec­tive al­tru­ism com­mu­nity as a whole, which Phil Tram­mell ex­plores in his pa­per, or the ques­tion of at which point in time and to what ex­tent an in­vestor-philan­thropist should change strat­egy. If we con­clude here that in­vest­ment is the op­ti­mal choice right now this doesn’t im­ply (1) that this will hold if a lot of value-al­igned oth­ers also start do­ing it or (2) that this will hold at a later time point.

Fi­nan­cial re­turns on investment

Ar­guably the largest ad­van­tage of in­vest­ing is that it can ex­po­nen­tially grow fi­nan­cial re­sources, which can be used for good at a later point. The S&P 500 has had an in­fla­tion-ad­justed an­nu­al­ized re­turn of ~7% since its in­cep­tion in 1926. We need to ad­just this for se­lec­tion bias, as there have been mul­ti­ple mar­kets in other coun­tries that have done a lot worse, or have even ceased to ex­ist (e.g. the Rio de Janeiro Stock Ex­change). A re­cent Credit Suisse re­port at­tempts this for global equity re­turns and finds an an­nu­al­ized real re­turn of ~5% from 1900 to 2019.

There is con­tro­versy about whether in­fla­tion-ad­justed prices us­ing the Con­sumer Price In­dex are ac­cu­rate: it’s likely the CPI is bi­ased up­wards, and hence that av­er­age re­cent global re­turns have been (much) higher than 5%.

For the pur­pose of this dis­cus­sion, we’ll stick with 5% as a con­ser­va­tive es­ti­mate for real ex­pected re­turns on in­dex fund in­vest­ing. This, in turn, is a lower-bound on ex­pected real re­turns from in­vest­ing more gen­er­ally: higher re­turns seem pos­si­ble with other types of in­vest­ments, e.g. lev­er­aged or ven­ture cap­i­tal in­vest­ment, if one is able to ex­ploit risk, in­for­ma­tion and/​or mar­ket ac­cess pre­miums. We hence con­ser­va­tively as­sume that a skil­led in­vestor can achieve 7% ex­pected real re­turns. Note that this will come at the ex­pense of higher risk, but that this is much less im­por­tant for al­tru­is­ti­cally-minded in­vestors, though it needs to be taken into ac­count to some ex­tent.

In ad­di­tion to im­perfect in­for­ma­tion and ac­cess, the ex­is­tence of this op­por­tu­nity can largely be ex­plained by the pure time prefer­ence of most mar­ket ac­tors and their risk aver­sion. An al­tru­is­tic and strate­gic philan­thropist is much less risk-averse (the ex­tent to which may de­pend on the cause area) and doesn’t have a (strong) pure time prefer­ence: even if he/​she cares more about the cur­rent gen­er­a­tion, this is likely for per­son-af­fect­ing rea­sons, and those con­sider per­son­hood rather than time.

Risks of loss or value drift

The gained fi­nan­cial value will not be fully con­verted into philan­thropic value if it is lost be­fore it can be spent (for other rea­sons than in­vest­ment losses) or if it’s spent on less valuable ac­tivi­ties. This could hap­pen in a va­ri­ety of ways, e.g. via le­gal challenges, theft, gov­ern­ment tax­a­tion, ex­is­ten­tial catas­tro­phes, or value drift of an in­vest­ment ve­hi­cle’s man­age­ment.

Loss

Ex­is­ten­tial risk es­ti­mates can be used as a lower bound for the loss rate. This is es­pe­cially true from a longter­mist per­spec­tive, as al­most by defi­ni­tion your money will be worth a lot less af­ter an ex­is­ten­tial catas­tro­phe has hap­pened. It also seems rea­son­able from a short-ter­mist/​per­son-af­fect­ing per­spec­tive, as many plau­si­ble ex­is­ten­tial risks will di­rectly lead to a loss of as­sets. Toby Ord’s best guess es­ti­mate in the Precipice is a risk of one out of six in the com­ing cen­tury, which con­verts into a yearly rate of ~0.2%. From Ord’s ar­gu­men­ta­tion it seems like he thinks the risk is in­creas­ing (most of the risk comes from fu­ture tech­nolo­gies), so we should per­haps re­vise this lower bound to 0.1% for in­vest­ment in the cur­rent mo­ment. On the other hand, many global catas­trophic risks could be suffi­cient for as­set loss, which our best guess es­ti­mate should take into ac­count.

As an up­per bound for the loss rate, we can look at the refer­ence class of ex­it­ing non­prof­its: a strate­gic in­vestor-philan­thropist should eas­ily be able to out­live the av­er­age non­profit, which has to deal with a lot more pres­sures (in­clud­ing fundrais­ing) and for which as­set loss is only one of the po­ten­tial rea­sons to cease to ex­ist. In the US, the yearly exit rate of non­prof­its is ~4% (be­tween 3% and 5%).

There are rea­sons to es­ti­mate the rele­vant loss rate much closer to the lower than the up­per bound. Tak­ing an in­side view per­spec­tive, the short-term risk of loss is ar­guably low: be­yond global catas­trophic risks and war it’s hard to imag­ine many sce­nar­ios (other than in­vest­ment risks) in which an in­vestor-philan­thropist would lose their as­sets in the cur­rent en­vi­ron­ment of (seem­ingly) sta­ble prop­erty rights.[2] Fur­ther­more, for a pro­por­tion of the sce­nar­ios one could imag­ine there would be warn­ing signs (e.g. of an im­pend­ing war), which would al­low the in­vestor-philan­thropist to change their strat­egy in time.[3] Note also that sce­nar­ios in which the in­vestor-philan­thropist is forced to spend their money should not be counted as loss risks.[4]

Value drift

The risk of value drift is even harder to es­ti­mate, but an im­por­tant fac­tor. For in­stance, these three sources (1,2,3) col­lec­tively sug­gest a yearly value drift rate of ~10% for in­di­vi­d­u­als within the effec­tive al­tru­ism com­mu­nity.

How­ever, the short-term value drift rate also seems much eas­ier to in­fluence pos­i­tively, most eas­ily via a proper de­sign[5] of a le­gal ve­hi­cle used by the in­vestor-philan­thropist. One can, for in­stance, com­mit one’s funds to be given to char­i­ta­ble en­tities by in­vest­ing from a donor-ad­vised fund, and ap­point a com­mit­tee of trustees to spread the risk of value drift.

Given the availa­bil­ity of these strate­gies, my best-guess es­ti­mate for the short-term[6] value drift rate is cur­rently 2% for a strongly com­mit­ted and strate­gic in­vestor-philan­thropist. How­ever, I have a lot of un­cer­tainty about this[7] and this es­ti­mate de­pends a lot on the the hy­po­thet­i­cal in­vestor-philan­thropist in ques­tion, so I in­vite the reader to make their own es­ti­mates based on the case they are con­sid­er­ing.[8]

Availa­bil­ity of opportunities

In­de­pen­dent of whether we are able to de­tect them, the value of the best available fund­ing op­por­tu­ni­ties changes over time and is de­pen­dent on the amount of cap­i­tal one has.

Diminish­ing returns

First, at any point in time, marginal so­cial re­turns to philan­thropic cap­i­tal are plau­si­bly diminish­ing with re­spect to the amount of cap­i­tal one spends. There might be ex­cep­tions to this, how­ever: cer­tain pro­jects might have in­creas­ing marginal re­turns and some might even re­quire a min­i­mum amount of cap­i­tal to have any chance of suc­cess at all. Creat­ing a new global in­sti­tu­tion could be an ex­am­ple.

More im­por­tantly, when one is try­ing to im­prove the world rather than one’s own life, one should con­sider one’s spent philan­thropic cap­i­tal as a marginal con­tri­bu­tion to all of the cap­i­tal that is spent in a value-al­igned way. This means that diminish­ing re­turns will prob­a­bly only play a sig­nifi­cant role at very large amounts of spent per­sonal philan­thropic cap­i­tal (>$100m). Hints of this can be ob­served em­piri­cally by look­ing at the scale of room for fund­ing of GiveWell’s recom­men­da­tions, or con­sid­er­ing that Open Philan­thropy is now able to spend more than $200m on a yearly ba­sis on fund­ing op­por­tu­ni­ties that meet their bar.

We should dis­t­in­guish the above diminish­ing marginal re­turns to per­sonal philan­thropic spend­ing from the diminish­ing marginal re­turns that could re­sult from in­creased con­cur­rent spend­ing by other al­igned philan­thropists, or by ac­tors that have at least similar in­stru­men­tal goals. Here, too, there could be ex­cep­tions due to some ma­jor philan­thropic pro­jects only be­ing fea­si­ble with a min­i­mum amount of to­tal cap­i­tal ded­i­cated to them. Fur­ther­more, to­tal al­igned spend­ing is not guaran­teed to in­crease over time. In the case of the effec­tive al­tru­ism com­mu­nity, an in­crease seems likely for at least the fore­see­able fu­ture, es­pe­cially given the plans of Open Philan­thropy/​Good Ven­tures to in­crease spend­ing over time and the re­cent ar­rival of Ben Delo as an­other UHNW effec­tive al­tru­ist donor. This is less clear for a more gen­eral cat­e­gory of in­stru­men­tally al­igned spend­ing.

A chang­ing world

Se­condly and prob­a­bly most im­por­tantly, there might be more or less im­pact­ful op­por­tu­ni­ties available due to a world that is ex­oge­nously in flux.

From a longter­mist per­spec­tive, this con­sid­er­a­tion is strongly re­lated to the de­bate on ‘hingey­ness’, though here we ex­clude the ex­oge­nous learn­ing fac­tor dis­cussed be­low. One in­tu­ition says that the ear­lier in time, the more of the fu­ture you still have to in­fluence, so the bet­ter the op­por­tu­ni­ties that will be available to you. On the other hand, this might be a very small effect, given the po­ten­tial length of time still left to us. More­over, it seems most likely to be dom­i­nated by other (more lo­cal) fac­tors, such as whether hu­man­ity has just de­vel­oped the power to de­stroy it­self, or is go­ing through a pe­riod of ex­cep­tional eco­nomic growth. I won’t ad­dress the full de­bate here in de­tail: it seems to be far from a done dis­cus­sion, and it cur­rently seems wise to ex­plore in­ter­ven­tions that are op­ti­mal across the spec­trum of views. More im­por­tantly, most ar­gu­ments about this be­ing a spe­cial time (e.g. those from ex­is­ten­tial risk) con­cern the whole cen­tury or the next few cen­turies we are liv­ing in rather than the spe­cific year, so it seems rea­son­able to as­sume the next few years will be quite similar to this year (in ex­pec­ta­tion) in terms of their hingey­ness.

From a short-ter­mist/​per­son-af­fect­ing and hu­man-cen­tric per­spec­tive, a rele­vant ques­tion is how fast dis­eases and poverty might be erad­i­cated ex­oge­nously, as this might in­fluence how much good you can do. A rough but use­ful proxy for how fast this is hap­pen­ing is the num­ber of peo­ple liv­ing in ex­treme poverty. This was 730 mil­lion in 2015 and was pro­jected by the World Bank to re­duce to 480 mil­lion by 2030, which im­plies a yearly rate of re­duc­tion of 2.7%. Another data point is the global growth rate of ~2%. Both are, how­ever, no di­rect prox­ies for the availa­bil­ity of cost-effec­tive op­por­tu­ni­ties to help peo­ple. A more di­rect but also more noisy way of look­ing at this ques­tion is by con­sid­er­ing GiveWell’s cost-effec­tive­ness es­ti­mates over time. In 2012, they es­ti­mated the cheap­est way to save a life[9] to be $2300 for bed­nets, and in 2019 this was still $2300 (also for bed­nets). Look­ing at their cost-per-life-saved-equiv­a­lent num­bers, Michael Dick­ens has even es­ti­mated a strong de­crease from $2066 to $443 in the same time pe­riod, though this is ar­guably more strongly in­fluenced by learn­ing than by new or bet­ter giv­ing op­por­tu­ni­ties be­ing available. Taken to­gether, my best guess is that there cur­rently is a rel­a­tively slowly in­creas­ing cost to helping peo­ple, if at all.[10]

Ex­oge­nous learning

Another im­por­tant ad­van­tage of in­vest­ing over giv­ing now is that it will al­low an in­vestor-philan­thropist to learn about bet­ter giv­ing op­por­tu­ni­ties over time. Here we are talk­ing about the abil­ity of the in­vestor-philan­thropist to iden­tify the best op­por­tu­ni­ties that are available, rather than whether those op­por­tu­ni­ties are available (see above).

We should dis­t­in­guish be­tween two forms of learn­ing: en­doge­nous and ex­oge­nous. En­doge­nous learn­ing is the learn­ing that the in­vestor-philan­thropist brings about them­selves, e.g. by fund­ing re­search or try­ing things out. Op­por­tu­ni­ties for en­doge­nous learn­ing can be a rea­son to give now rather than to in­vest (see the sec­tion on com­pound­ing re­turns on giv­ing be­low).

Ex­oge­nous learn­ing in­cludes ad­vances in the sci­en­tific com­mu­nity, new philan­thropic in­ter­ven­tions be­ing in­vented and/​or tried out, moral progress, and more. It also cap­tures the time needed for rele­vant knowl­edge to be­come available, e.g. an ex­per­i­ment might take time, re­search might need to be done in a cer­tain or­der, or there might be a tal­ent con­straint in a re­search area that takes time to be re­solved. When learn­ing is done ex­oge­nously, there are ad­van­tages to wait­ing and hence in­vest­ing.

90% of all sci­en­tists who ever lived are al­ive to­day, and we should ex­pect big gains in knowl­edge across the board, though maybe not as much as that num­ber would sug­gest on first sight. More im­por­tantly though, effec­tive al­tru­ism is still a very young en­deav­our, so an in­vestor-philan­thropist should ex­pect there to be a high rate of ex­oge­nous learn­ing in the effec­tive al­tru­ism com­mu­nity in the short to medium term.

From a short-ter­mist/​per­son-af­fect­ing per­spec­tive, as an illus­tra­tion, con­sider that GiveWell has only been around for 13 years, An­i­mal Char­ity Eval­u­a­tors only for 8 years and Founders Pledge re­search only for 3 years, with new high-im­pact giv­ing op­por­tu­ni­ties be­ing dis­cov­ered by these or­gani­sa­tions on a reg­u­lar ba­sis. On the other hand, look­ing back at the ex­am­ple above, GiveWell has had broadly similar top giv­ing recom­men­da­tions since 2012, though they have only re­cently started con­sid­er­ing policy in­ter­ven­tions.

From a longter­mist per­spec­tive, ex­oge­nous learn­ing is an (even) more im­por­tant fac­tor. Firstly, longter­mism as a more for­mal idea has only very re­cently been de­vel­oped, though in­sti­tu­tional work has been done on it at least since the found­ing of the Fu­ture of Hu­man­ity In­sti­tute in Oxford in 2005. Se­condly, the num­ber of peo­ple work­ing full time on what is the best thing to do from a longter­mist per­spec­tive is prob­a­bly less than 200[11]. Thirdly, fund­ing op­por­tu­nity re­search speci­fi­cally is even younger than for short-ter­mism, with the first in­sti­tu­tional re­search prob­a­bly be­ing car­ried out by Open Philan­thropy around 2014, and ex­is­ten­tial risk cur­rently be­ing the only well-es­tab­lished in­ter­ven­tion area.

In ad­di­tion to un­cer­tainty about the best fund­ing op­por­tu­ni­ties to achieve some defined goal, there is a lot of un­cer­tainty and de­bate even about what the goals should be: should we be short-ter­mist/​per­son-af­fect­ing or longter­mist; to what ex­tent should we in­clude an­i­mals in our moral con­cern; should we ul­ti­mately mostly/​only aim for some (broad) mea­sure of sub­jec­tive well-be­ing or are there other things that are im­por­tant to con­sider as well? Given how re­cent it has been for many of these ideas to gain se­ri­ous trac­tion (e.g. Peter Singer’s An­i­mal Liber­a­tion was only pub­lished in 1975, though the idea had ob­vi­ously been around for a lot longer), it’s likely we have a lot to learn still and could hope to learn more rel­a­tively soon.

Lastly, we might learn more about in­vest­ing to give later as a philan­thropic strat­egy it­self. To my knowl­edge, Phil Tram­mell’s pa­per is the first for­mal doc­u­ment to out­line the case for this, and in this re­search pro­ject I will likely only be able to scratch the sur­face. From the fac­tors pre­sented here, the differ­ence in im­pact po­ten­tial be­tween in­vest­ing and giv­ing now could plau­si­bly be very large, cer­tainly over longer time scales. And there is a rele­vant asym­me­try: un­less one has good rea­son to be­lieve there is an ex­traor­di­nary and timely giv­ing op­por­tu­nity available right now, it seems the ex­pected cost of in­vest­ing and wait­ing on more in­for­ma­tion (for a limited time) if giv­ing now turns out to be bet­ter is lower than the ex­pected cost of giv­ing now and fore­go­ing the op­por­tu­nity to ever in­vest if in­vest­ing turns out to be bet­ter.

Pa­ram­e­ter uncertainty

There is a lot of un­cer­tainty about many if not all of the pa­ram­e­ters dis­cussed above, e.g. the ex­pected fi­nan­cial re­turns, the ex­pro­pri­a­tion and value drift rates, and the learn­ing rate. If these pa­ram­e­ters com­bined are likely to cause com­pound­ing pos­i­tive effects, then this un­cer­tainty it­self can fur­ther in­crease the ex­pected value of in­vest­ing over time.

In math­e­mat­i­cal terms: for any yearly rate of so­cial re­turn r > 0, and any con­stant q > 0, the world in which r = q de­liv­ers a smaller multi-year so­cial re­turn than the world in which r is dis­tributed as a non-de­gen­er­ate ran­dom vari­able R > 0 where E[R] = q.

My model sug­gests this to be an im­por­tant fac­tor: there is a large differ­ence be­tween the one-year and ten-year im­pact mul­ti­plier for in­vest­ing to give later.

Com­pound­ing re­turns on giv­ing now

A fi­nal fac­tor to con­sider in favour of giv­ing now is whether there might be giv­ing op­por­tu­ni­ties that them­selves have larger com­pound (so­cial) re­turns than in­vest­ing does.

From a short-ter­mist/​per­son-af­fect­ing and hu­man-fo­cused per­spec­tive, an ar­gu­ment that is of­ten brought up is that di­rect global health and poverty in­ter­ven­tions may have com­pound gains for benefi­cia­ries that out­weigh com­pound in­vest­ment gains. Phil Tram­mell ex­plains in his pa­per (sec­tion 5.1.2 and 5.1.3) why this is al­most cer­tainly not the case, at least over longer timescales, for both the­o­ret­i­cal and em­piri­cal rea­sons. The main con­cep­tual point is that even though some­one in poverty might ob­tain gains above the world growth rate (~2%) for a few years, these gains are (1) likely to dis­perse rapidly and (2) par­tially used for con­sump­tion with­out com­pound­ing re­turns. If the benefi­ciary of a dona­tion is not them­selves repri­ori­tiz­ing the spend­ing of their gained re­sources (health, money, knowl­edge, etc.) on the best available giv­ing op­por­tu­ni­ties in the world (or in­vest­ing those re­sources), then the yearly gains of those re­sources will at some point be bounded from above by the growth rate.

If one is only con­cerned with effects in the very short term (e.g. 10 years) though, it could be that the short-term higher re­turns are a rea­son to give now rather than in­vest. How­ever, with­out such short limits on the time hori­zon of effects, this by it­self doesn’t af­fect the com­par­i­son: in­vest­ing even only for 1 year and then giv­ing would be bet­ter than giv­ing now, as the ex­tra early gains from in­vest­ing would out­strip the bounded fu­ture gains from giv­ing now.

There are, how­ever, some giv­ing op­por­tu­ni­ties which are ‘in­vest­ment-like’ and could in prin­ci­ple have higher com­pound­ing re­turns than in­vest­ment, even in the longer term, and both from a short-ter­mist/​per­son-af­fect­ing and longter­mist per­spec­tive.

One ob­vi­ous can­di­date is en­courag­ing other peo­ple with al­igned val­ues to in­vest: at a high enough suc­cess rate per dol­lar spent on this, it is clear that this would beat di­rect in­vest­ment, and the main un­cer­tainty is whether such a suc­cess rate can be achieved. More gen­eral effec­tive al­tru­ism move­ment-build­ing may also qual­ify, but this is only true with cer­tainty if it leads to a high enough rate of peo­ple join­ing and a large enough rate of those peo­ple choos­ing to in­vest.

Se­condly, there is ca­pac­ity build­ing, in­clud­ing en­doge­nous learn­ing: if cer­tain ac­tivi­ties (e.g. re­search) in­crease the im­pact one can have from that mo­ment on­wards un­til one’s last dol­lar is spent, those ac­tivi­ties have com­pound gains that when large enough could ex­ceed those of in­vest­ment. The case for those ac­tivi­ties be­comes even stronger if they also af­fect the abil­ity of oth­ers to have a pos­i­tive im­pact. How­ever, we should take into ac­count the coun­ter­fac­tual: if this ca­pac­ity build­ing will hap­pen any­way (i.e. can be achieved ex­oge­nously), and will be funded by some­one who wouldn’t oth­er­wise have in­vested, this would likely bal­ance the scales in favour of in­vest­ing.

Syn­the­sis of factors

In this Guessti­mate model, I have made pre­limi­nary es­ti­mates for the im­por­tance of each of the fac­tors from a longter­mist[12] per­spec­tive. I have com­bined these to come to a very ten­ta­tive es­ti­mate of the ex­pected im­pact mul­ti­plier for in­vest­ment rel­a­tive to giv­ing now, both over 1 year and over 10 years.

This ap­proach has im­por­tant limi­ta­tions, and this is only a first iter­a­tion, but I think it’s im­por­tant to have an ex­plicit model to (1) bring the dis­cus­sion out of a va­guer “there are ar­gu­ments on both sides” realm and come to de­ci­sion-rele­vant best guesses and (2) be bet­ter able to iden­tify which un­cer­tainty would be most valuable to re­solve and pri­ori­tize re­search efforts as a re­sult. I en­courage you to du­pli­cate the model and make your own es­ti­mates, and to sug­gest al­ter­na­tives and im­prove­ments to the (very rudi­men­tary) method­ol­ogy!

Ex­clud­ing those ‘in­vest­ment-like’ giv­ing op­por­tu­ni­ties, my cur­rent best guess es­ti­mate is that a longter­mist in­vestor-philan­thropist will on av­er­age be able to mul­ti­ply the to­tal im­pact of their funds by ~1.01 in one year and by >>10[1:1] in ten years.


  1. Guessti­mate’s Monte Carlo simu­la­tion sam­ple limit of 5000 means the ac­cu­racy of its ten-year es­ti­mate is very limited: re­run­ning 10-20 times yields es­ti­mates rang­ing from 13 to 13000. ↩︎ ↩︎

  2. I plan to look fur­ther into whether and to what ex­tent this is the case. ↩︎

  3. In­ci­den­tally, such a time would ar­guably (in ex­pec­ta­tion) have an above-av­er­age availa­bil­ity of high-im­pact giv­ing op­por­tu­ni­ties, cf. the cur­rent situ­a­tion as dis­cussed un­der ‘Availa­bil­ity of op­por­tu­ni­ties’. ↩︎

  4. There is an in­ter­est­ing open ques­tion of whether the loss rate is (pos­i­tively or nega­tively) cor­re­lated with the size of one’s in­vest­ments. There seem to be ar­gu­ments on both sides. ↩︎

  5. An im­por­tant con­sid­er­a­tion in the de­sign is the (seem­ing) trade-off be­tween pre­vent­ing value drift and al­low­ing for po­ten­tial value im­prove­ment over time. An ex­am­ple of a way to bal­ance those—in the case of a char­i­ta­ble in­vest­ment fund—would be to ap­point trustees that se­lect their own suc­ces­sors. ↩︎

  6. The short-term yearly value drift risk in­tu­itively seems a lot lower than the yearly value drift risk be­yond an in­vestor’s life­time. How­ever, if an in­vestor is able to ap­point a suc­ces­sor that he/​she deems to have at least as good val­ues as him-/​her­self, the short- and longer-term risks are ar­guably very similar. ↩︎

  7. To im­prove my es­ti­mates for the risk of loss and the risk of value drift and to learn more about what might help to min­i­mize these risks, I aim to do a deeper dive both into rele­vant refer­ence classes (e.g. UK char­i­ta­ble en­tities that in­vest a sig­nifi­cant pro­por­tion of their as­sets) and case stud­ies (e.g. waqfs). ↩︎

  8. Granted, there are ob­vi­ous difficul­ties in es­ti­mat­ing one’s own ex­pected value drift rate. ↩︎

  9. For com­pa­ra­bil­ity be­tween the es­ti­mates I’m con­sid­er­ing benefits from di­rect mor­tal­ity re­duc­tion. ↩︎

  10. As an aside, note that the cost-effec­tive­ness of the giv­ing op­por­tu­ni­ties available to us might be cor­re­lated with the fi­nan­cial re­turns we are able to make on in­vest­ment or the risk of ex­pro­pri­a­tion at that par­tic­u­lar time. This could both strengthen (it al­lows for mis­sion hedg­ing) or weaken (we might need the cap­i­tal ex­actly when re­turns are low) the case for in­vest­ment. How­ever, note that this cor­re­la­tion would likely be a lo­cal one in time, whereas if one in­vests over a longer timescale, most of the fi­nan­cial re­turns may ac­crue by or be­yond this point, mean­ing that this would only have a limited in­fluence on the to­tal philan­thropic value achieved. ↩︎

  11. This is a guess; I would wel­come any available data on this. ↩︎

  12. I haven’t yet made es­ti­mates from a short-ter­mist/​per­son-af­fect­ing per­spec­tive, but in­tend to do so at a later point. ↩︎