RPTP Is a Strong Reason to Consider Giving Later

Peo­ple hop­ing to do the most good with their giv­ing face a trade­off be­tween (a) giv­ing now and (b) in­vest­ing to give more later. If the giver doesn’t ex­pect to learn any­thing over time about the best places to give, the ques­tion of when to give boils down roughly to the ques­tion of whether the in­ter­est rate at which they could in­vest ex­ceeds the “char­i­ta­ble dis­count rate” at which do­ing a unit of good is grow­ing more costly.

For me, and I think for pretty much any (util­i­tar­ian-lean­ing) as­piring effec­tive al­tru­ist right now, the “learn­ing” con­sid­er­a­tion should swamp all else. We cur­rently lack, as far as I can tell, any good way to fore­cast the long-run im­pacts of our ac­tions, and in that lack I think all of our at­tempts at char­ity are about as likely to do harm as to do good. But more peo­ple are start­ing to think about the prob­lem se­ri­ously, and there is at least a sliver of hope that progress will be made over the com­ing years or decades. In the mean­time, there is noth­ing to do but in­vest and wait—or, per­haps, fund bet­ter pri­ori­ti­za­tion re­search.

But if your goal as a philan­thropist is to spend money at time t to in­crease welfare around time t, and more com­plex fund­ing op­por­tu­ni­ties are not un­der con­sid­er­a­tion, it seems to me that there’s a good a pri­ori rea­son to think it’s usu­ally bet­ter to give later (even if we ig­nore the clue­less­ness/​learn­ing is­sue) which I haven’t seen ex­pressed el­se­where. The rea­son is that mar­ket in­ter­est rates are set in part by peo­ple’s “rates of pure time prefer­ence” (RPTP). If in­vestors were perfectly pa­tient—i.e. if they sought to max­i­mize the sum of their own non-dis­counted welfare over the course of their lives—an equil­ibrium in­ter­est rate of 7% would mean that a unit of welfare next year was pro­jected to cost about 7% more, in dol­lar terms, than a unit of welfare this year. But in­vestors are not perfectly pa­tient; they dis­count their fu­ture welfare at some pos­i­tive rate. If that rate is, say, 2%, then the in­differ­ence point of 7% re­turns im­plies that the rate at which the cost of welfare is ris­ing (R) is only 5%. Philan­thropists with zero RPTP can there­fore do 2% more good for oth­ers by in­vest­ing at 7% and giv­ing next year.

This is not a small con­cern. Some re­cent liter­a­ture on dis­count­ing, for in­stance, has ob­served that the “near-zero so­cial dis­count rate” rea­son­ing usu­ally used to jus­tify ex­ten­sive ac­tion against long-term risks like cli­mate change also im­plies a need to pro­mote in­vest­ment in gen­eral, with op­ti­mal cap­i­tal gains sub­sidies of as much as 50% (fi­nanced by cor­re­spond­ingly high taxes on pre­sent con­sump­tion).

Per­haps the cost of welfare is grow­ing more quickly for some pop­u­la­tions than oth­ers, and per­haps some of those pop­u­la­tions are cur­rently top con­tenders for our char­ity. For in­stance, per­haps the cost of helping the world’s poor­est is ris­ing more quickly than 7% per year, as is some­times claimed, due to the par­tic­u­larly fast progress be­ing made in global de­vel­op­ment. (Scott Alexan­der re­ports Elie Hassen­feld ba­si­cally mak­ing this point a few years ago.) If this is true, then in­deed, we would do less good giv­ing next year than giv­ing this year.

But this one-year re­la­tion­ship must be tem­po­rary. Over the course of a long fu­ture, the rate of in­crease in the cost of pro­duc­ing a unit of welfare as effi­ciently as pos­si­ble can­not, on av­er­age, ex­ceed R. Other­wise, the most effi­cient way to good would even­tu­ally be more costly than one par­tic­u­lar way to good—just giv­ing money to or­di­nary in­vestors for their own con­sump­tion. And since the long-run av­er­age rate of in­crease in the cost of welfare is bounded above by R (“5%“), in­vest­ing at R + RPTP (“7%“) must even­tu­ally re­sult in an en­dow­ment able to buy more welfare than the en­dow­ment we started with.

At first glance, this leads to the para­dox­i­cal con­clu­sion that we should in­vest for­ever and never give. The re­s­olu­tion of this para­dox is that one way or an­other, the op­por­tu­nity to in­vest at a rate R + RPTP will even­tu­ally not be available. There are var­i­ous rea­sons an en­dow­ment could come to lose this op­por­tu­nity. It could face idiosyn­cratic con­straints (the money is go­ing to be seized in a few years). Or, with in­creas­ing wealth, in­vest­ment op­por­tu­ni­ties could dry up in gen­eral (de­pend­ing on how util­ity de­creases with con­sump­tion, peo­ple might grow so rich that the only pro­jects worth in­vest­ing in would be those that earned ex­tremely high re­turns, and even­tu­ally fewer such pro­jects might ex­ist than the size of the en­dow­ment). Or in­vest­ment op­por­tu­ni­ties could van­ish for other rea­sons, such as the im­pend­ing end of the world. In the last case, if the other con­straints don’t hold, the best thing to do is to in­vest for­ever, un­til one mas­sive act of char­ity on the last day.

But un­der or­di­nary cir­cum­stances, to a first ap­prox­i­ma­tion, if a philan­thropist’s plan is to spend his money at some time t to in­crease con­sump­tion-based welfare as effi­ciently as pos­si­ble at time t—and if these con­sid­er­a­tions are not swamped by oth­ers be­yond the scope of this post, such as the risk of value drift—then it seems the philan­thropist should wait.

[Edited 4 Nov. 2018 [1] to in­clude the link to Elie Hassen­feld mak­ing the point about van­ish­ing giv­ing op­por­tu­ni­ties in global poverty, and [2] to weaken the last sen­tence so that it em­pha­sizes the limited scope of this post. Edited 7 Nov. 2018 [3] to point out this limited scope ear­lier on, so that it’s clear that this ar­gu­ment doesn’t ap­ply to re­search fund­ing.]