Earning to Give vs. Altruistic Career Choice Revisited

A com­monly voiced sen­ti­ment in the effec­tive al­tru­ist com­mu­nity is that the best way to do the most good is gen­er­ally to make as much money as pos­si­ble, with a view to­ward donat­ing to the most cost-effec­tive char­i­ties. This is of­ten referred to as “earn­ing to give.” In the ar­ti­cle To save the world, don’t get a job at a char­ity; go work on Wall Street William MacAskill wrote:

Top un­der­grad­u­ates who want to “make a differ­ence” are en­couraged to forgo the al­lure of Wall Street and work in the char­ity sec­tor … while re­search­ing eth­i­cal ca­reer choice, I con­cluded that it’s in fact bet­ter to earn a lot of money and donate a good chunk of it to the most cost-effec­tive char­i­ties, a path that I call “earn­ing to give.” In gen­eral, the char­i­ta­ble sec­tor is peo­ple-rich but money-poor. Ad­ding an­other per­son to the la­bor pool just isn’t as valuable as pro­vid­ing more money, so that more work­ers can be hired.

In pri­vate cor­re­spon­dence, MacAskill clar­ified that he wasn’t ar­gu­ing that “earn­ing to give” is the best way to do good, only that it’s of­ten bet­ter than work­ing at a given non­profit. In a re­cent com­ment MacAskill wrote

I think there’s too much em­pha­sis on “earn­ing to give” as the *best* op­tion rather than as the *baseline* op­tion

and raises a num­ber of counter-con­sid­er­a­tions against “earn­ing to give. De­spite this, the idea that “earn­ing to give” is op­ti­mal has caught on in the effec­tive al­tru­ist com­mu­nity, and so it’s im­por­tant to dis­cuss it.

Over the past three years, I my­self have shifted from the po­si­tion that earn­ing to give is philan­throp­i­cally op­ti­mal, to the po­si­tion that it’s gen­er­ally the case that one can do more good by choos­ing a ca­reer with high di­rect so­cial value than by choos­ing a lu­cra­tive ca­reer with a view to­ward donat­ing as much as pos­si­ble.

In this post I’ll out­line some ar­gu­ments in fa­vor of this view.

Re­sponses to MacAskill’s Considerations

In the ar­ti­cle To save the world, don’t get a job at a char­ity; go work on Wall Street, MacAskill gives three con­sid­er­a­tions in fa­vor of “earn­ing to give.” I re­spond to these con­sid­er­a­tions be­low. What I write should be read as a re­sponse to the ar­ti­cle, rather than to MacAskill’s views.

Var­i­ance in cost-effec­tive­ness of charities

MacAskill wrote

… char­i­ties vary tremen­dously in the amount of good they do with the money they re­ceive. For ex­am­ple, it costs about $40,000 to train and provide a guide dog for one per­son, but it costs less than $25 to cure one per­son of sight-de­stroy­ing tra­choma. For the cost of im­prov­ing the life of one per­son with blind­ness, you can cure 1,000 peo­ple of it…it’s un­likely that you can work for only the very best char­i­ties. In con­trast, if you earn to give, you can donate any­where, prefer­ably to the most cost-effec­tive char­i­ties, and change your dona­tions as of­ten as you like.

GiveWell has spent about five years look­ing for the best giv­ing op­por­tu­ni­ties in global health, and its cur­rent #1 ranked char­ity is Against Malaria Foun­da­tion (AMF). GiveWell es­ti­mates that AMF saves an in­fant’s life for ~ $2,300, not count­ing other benefits. Th­ese other benefits not with­stand­ing, AMF’s cost per DALY saved is much higher than the im­plied cost per DALY saved as­so­ci­ated with the figure cited for cur­ing sight-de­stroy­ing tra­choma.

GiveWell may have missed giv­ing op­por­tu­ni­ties in global health that are much more cost-effec­tive than AMF is, but given the amount of time, en­ergy and at­ten­tion that GiveWell spent on its search, one should have a strong prior against the pos­si­bil­ity that one can eas­ily find a bet­ter giv­ing op­por­tu­nity in global health. So a plau­si­ble es­ti­mate of the cost-effec­tive­ness of donat­ing to the best char­ity that de­liv­ers di­rect global health in­ter­ven­tions is much lower than the above quo­ta­tion sug­gests.

Fur­ther­more, the phe­nomenon of the op­ti­mizer’s curse sug­gests that all char­i­ties with ro­bust case for fairly high cost-effec­tive­ness are closer in cost-effec­tive­ness to AMF than ex­plicit cost-effec­tive­ness calcu­la­tions in­di­cate. This nar­rows the var­i­ance in cost-effec­tive­ness amongst char­i­ties.

So the ad­van­tage of be­ing able to choose a char­ity to sup­port and change at any time is smaller than the above quo­ta­tion sug­gests.

Discrep­ancy in earnings

MacAskill wrote:

An­nual salaries in bank­ing or in­vest­ment start at $80,000 and grow to over $500,000 if you do well. A life­time salary of over $10 mil­lion is typ­i­cal. Ca­reers in non­prof­its start at about $40,000, and don’t typ­i­cally ex­ceed $100,000, even for ex­ec­u­tive di­rec­tors By en­ter­ing fi­nance and donat­ing 50% of your life­time earn­ings, you could pay for two non­profit work­ers in your place—while still liv­ing on dou­ble what you would have if you’d cho­sen that route.

The as­sump­tion “if you do well” is a very strong one. Only about 1% of Amer­i­cans make ~$500k/​year. There are some peo­ple who have a strong com­par­a­tive ad­van­tage in fi­nance, for whom “earn­ing to give” to give may be es­pe­cially com­pel­ling. But peo­ple who are able to make ~$500k/​year in fi­nance who don’t have a large com­par­a­tive ad­van­tage in fi­nance have very strong trans­fer­able skills. Such peo­ple are sig­nifi­cantly more ca­pa­ble than the av­er­age non-profit worker, and can plau­si­bly have a big­ger im­pact than 2 or 3 such work­ers by work­ing di­rectly on some­thing with high so­cial value.


MacAskill wrote:

…“mak­ing a differ­ence” re­quires do­ing some­thing that wouldn’t have hap­pened any­way…The com­pe­ti­tion for not-for-profit jobs is fierce, and if some­one else takes the job in­stead of you, he or she likely won’t be much worse at it than you would have been. So the differ­ence you make by tak­ing the job is only the differ­ence be­tween the good you would do, and the good that the other per­son would have done.

I would guess that there are some highly cost-effec­tive hu­man­i­tar­ian in­ter­ven­tions that are suffi­ciently easy to im­ple­ment that the im­ple­menters are eas­ily re­place­able. I could eas­ily imag­ine that this is the case for vac­ci­na­tion efforts.

But fund­ing op­por­tu­ni­ties for these in­ter­ven­tions can be thought of as “low hang­ing fruit.” Broad mar­ket effi­ciency sug­gests that such in­ter­ven­tions will be funded. And in­deed, GiveWell has found that straight­for­ward im­mu­niza­tion efforts are already largely funded, to the point that GiveWell has been un­able to find giv­ing op­por­tu­ni­ties for in­di­vi­d­ual donors in this area.

This sug­gests that at the mar­gin, very high value hu­man­i­tar­ian efforts re­quire highly skil­led and highly mo­ti­vated la­bor­ers.

High skil­led la­bor­ers are a rel­a­tively small sub­set of la­bor­ers, so there are fewer peo­ple available to do these sorts of jobs than other jobs. Do­ing a hard, non-rou­tine job well re­quires high mo­ti­va­tion. The col­lec­tion of peo­ple who are suffi­ciently highly mo­ti­vated to do a hard job with high so­cial value that doesn’t pay well, and who could oth­er­wise be mak­ing much more money, largely con­sists of peo­ple who are try­ing to have a sig­nifi­cant pos­i­tive so­cial im­pact.

So sup­pose that you’re a highly skil­led la­borer de­cid­ing whether to “earn to give” or take a job with high so­cial value that re­quires high skills and mo­ti­va­tion. If you don’t take the job with high so­cial value, your coun­ter­fac­tual re­place­ment is likely be one of the fol­low­ing:

1. Sub­stan­tially less ca­pa­ble than you on ac­count of hav­ing low skills, or low al­tru­is­tic mo­ti­va­tion.

2. A highly skil­led per­son with high mo­ti­va­tion, who would be do­ing some­thing else with high so­cial value if you had taken the job, and who can’t do this be­cause they have to do the job that you would have done.

3. Nonex­is­tent.

So the re­place­abil­ity con­sid­er­a­tion car­ries less weight than it might seem.

Ad­mit­tedly there’s a coun­ter­con­sid­er­a­tion — broad mar­ket effi­ciency cuts both ways, and one could imag­ine that the low hang­ing fruit in work­ing di­rectly on pro­jects with high so­cial value is also plucked, and this counter-con­sid­er­a­tion pushes in fa­vor of “earn­ing to give.” I have a fairly strong in­tu­ition that “if you don’t fund it, some­body else will” is more true than “if you don’t do it, some­body else will” so that this counter-con­sid­er­a­tion is out­weighed. It’s im­por­tant to note that many pro­jects of high so­cial value are the first of their kind, and that find­ing some­body else to ex­e­cute such a pro­ject is highly non­triv­ial. I think that it’s also rele­vant that 114 billion­aires have signed the Giv­ing Pledge, com­mit­ting to giv­ing 50+% of their wealth away in their life­times.

In any case, there isn’t a clear-cut, un­con­di­tional ar­gu­ment that fa­vors “earn­ing to give”: whether “earn­ing to give” is the best op­tion very much de­pends on nu­anced em­piri­cal con­sid­er­a­tions rather than a gen­eral ab­stract ar­gu­ment.

Other im­por­tant con­sid­er­a­tions that fa­vor an al­tru­is­tic career

There are ad­di­tional im­por­tant con­sid­er­a­tions that fa­vor pur­su­ing a ca­reer with high so­cial value over “earn­ing to give”:

Asym­met­ric im­pli­ca­tions of the ex­is­tence of small prob­a­bil­ity failure modes

In Ro­bust­ness of Cost-Effec­tive­ness Es­ti­mates and Philan­thropy, I de­scribed how a large col­lec­tion of small prob­a­bil­ity failure modes con­spires to sub­stan­tially re­duce the ex­pected value of a fund­ing op­por­tu­nity. The same is­sue ap­plies to choos­ing a nar­row ca­reer goal with a view to­ward di­rectly hav­ing a high pos­i­tive so­cial im­pact. But a worker has more ca­pac­ity than a donor does to learn whether small prob­a­bil­ity failure modes pre­vail in prac­tice, and can switch to a differ­ent job if he or she finds that such a failure mode pre­vails.

Here’s an ex­am­ple. Sup­pose that you go to med­i­cal school with a view to­ward the pos­si­bil­ity of perform­ing cleft palate surg­eries in the de­vel­op­ing world. It’s prob­a­bly the case that the op­por­tu­nity isn’t as promis­ing as it seems. But if you try it, then you’ll be able to see how effec­tive the in­ter­ven­tion is first­hand. If it’s highly effec­tive, then you can keep do­ing it. If it’s not highly effec­tive, then you can ex­plore other pos­si­bil­ities, such as

  • Start­ing your own surgery or­ga­ni­za­tion.

  • Switch­ing to do­ing a differ­ent kind of surgery in the de­vel­op­ing world, such as cataract re­moval.

  • Work­ing in a poor com­mu­nity in the de­vel­oped world (which could have a big­ger im­pact than work­ing in the de­vel­op­ing world ow­ing to flow-through effects).

  • Work­ing for a biotech com­pany.

  • Get­ting in­volved in clini­cal med­i­cal re­search.

  • Other things that haven’t oc­curred to me.

By ex­per­i­ment­ing, one can hope to hone in on a job that has both high os­ten­si­ble cost-effec­tive­ness, and and a rel­a­tively small mass of small prob­a­bil­ity failure modes.

Altru­is­tic ca­reers ex­tend be­yond the non­profit world

Even on the as­sump­tion that “earn­ing to give” is bet­ter than work­ing at a non­profit, it doesn’t fol­low that “earn­ing to give” op­ti­mizes so­cial im­pact. There are ways to have a pos­i­tive so­cial im­pact in the for-profit world, in sci­en­tific re­search, and in the gov­ern­ment.

His­tor­i­cal Precedent

For the most part, the peo­ple who have had the biggest pos­i­tive im­pact on the world haven’t had their im­pact by “earn­ing to give.”

There are a few pos­si­ble ex­cep­tions, such as Bill Gates and War­ren Buffett, whose philan­thropic ac­tivi­ties could be hav­ing a huge im­pact (though it’s hard to tell from the out­side) and could well out­strip the value that they con­tributed through their la­bor. But they ap­pear to have an un­usu­ally high ra­tio of wealth to di­rect pos­i­tive im­pact of their work, and so ap­pear to be un­rep­re­sen­ta­tive.

Steve Jobs’ high­est net worth was on the or­der of $10 billion, whereas Bill Gates’ high­est net worth was on the or­der of $100 billion. I don’t think that Bill Gates con­tributed 10x as much as Steve Jobs to tech­nol­ogy, and I don’t think that Jobs could have had a big­ger so­cial im­pact by donat­ing than through his work (which had mas­sive pos­i­tive flow-through effects). I ac­knowl­edge that Jobs is a cherry picked ex­am­ple, but I think that the gen­eral prin­ci­ple still holds.

Main­stream consensus

Few peo­ple think that “earn­ing to give” is the best way to make the world a bet­ter place. This could be at­tributable to ir­ra­tional­ity or to low al­tru­ism, but my ex­pe­rience is that there are many peo­ple who care about global welfare, or just welfare within a spe­cific cause, and many peo­ple who are highly in­tel­li­gent. In light of the ex­is­tence of illu­sory su­pe­ri­or­ity, one should be wary of hold­ing an im­plicit view that one knows more about how to make the world a bet­ter place than the vast ma­jor­ity of the pop­u­la­tion.

Steel­man­ning wealth maximization

It’s worth high­light­ing some fac­tors that fa­vor choos­ing a ca­reer with a view to­ward max­i­miz­ing wealth in some situ­a­tions:

  • Com­par­a­tive ad­van­tage — Some peo­ple are un­usu­ally good at mak­ing money rel­a­tive to do­ing other things. Such peo­ple may do bet­ter to “earn to give” than to try to choose a job that has a di­rect pos­i­tive im­pact (which they’re rel­a­tively bad at).

  • The mar­ket mechanism — In the for-profit world, max­i­miz­ing wealth is of­ten cor­re­lated with max­i­miz­ing pos­i­tive so­cial im­pact, and so can be used as a proxy goal for max­i­miz­ing pos­i­tive so­cial im­pact.

  • Con­nec­tions and per­sonal growth — Peo­ple with high earn­ings are gen­er­ally more ca­pa­ble and more knowl­edge­able than peo­ple in other con­texts, and tend to be well con­nected, so po­si­tion­ing one­self among such peo­ple can in­crease one’s prospects of soar­ing to greater heights. Jeff Be­zos started his ca­reer in fi­nance, and later cre­ated Ama­zon, which has had mas­sive pos­i­tive so­cial im­pact (both di­rect, and via flow-through effects).

  • Unusual val­ues — If one cares about causes that very few peo­ple care about, then it could be difficult to find fund­ing for work on them, so “earn­ing to give” could be nec­es­sary. I don’t be­lieve this to be the case, but it’s a con­sid­er­a­tion that’s been raised by oth­ers, and so is worth men­tion­ing.

Clos­ing sum­mary

There are many ar­gu­ments against the claim that “earn­ing to give” is gen­er­ally the best way to max­i­mize one’s pos­i­tive so­cial im­pact, and I be­lieve that choos­ing a job where one can do as much good as pos­si­ble through one’s work is gen­er­ally the best way to max­i­mize one’s pos­i­tive so­cial im­pact. How­ever, for some peo­ple in un­usual situ­a­tions, “earn­ing to give” may be the best way to have a pos­i­tive so­cial im­pact.

Note: I formerly worked as a re­search an­a­lyst at GiveWell. All views ex­pressed here are my own.

Ac­knowl­edge­ments: I thank Nick Beck­stead, Mo­dusPonies and Will Crouch for helpful feed­back on an ear­lier ver­sion of this ar­ti­cle.