Altruistic equity allocation

Char­i­ties im­plic­itly al­lo­cate credit be­tween their em­ploy­ees, founders, and fun­ders. For ex­am­ple, a char­ity might de­scribe it­self as “fund­ing con­strained” or “re­cruit­ing con­strained,” let­ting donors or hires know that their con­tri­bu­tions are par­tic­u­larly valuable right now.

I think there is room to do bet­ter, so I’m in­ter­ested in ex­per­i­ments with more ex­plicit mechanisms. In this post I’ll sug­gest one al­ter­na­tive that mir­rors for-profit equity al­lo­ca­tion. This mechanism is similar to and com­ple­men­tary with cer­tifi­cates of im­pact, but they can be adopted in­de­pen­dently.

Sev­eral peo­ple have con­sid­ered adapt­ing this kind of mechanism to non-profit credit al­lo­ca­tion, and I’m not try­ing to claim pri­or­ity here (just try­ing to start a more con­crete dis­cus­sion). In par­tic­u­lar, thanks to Carl Shul­man, Ja­cob Trefethen, and es­pe­cially Holden Karnofsky for dis­cussing similar ideas and crit­i­ciz­ing past ver­sions (not to say any­one en­dorses this ver­sion).

Is­su­ing al­tru­is­tic equity

Sup­pose that I’m a char­ity look­ing to raise funds for 2020. I’d like to raise these funds by is­su­ing new “shares” in my im­pact, ex­plic­itly dilut­ing all of my other sup­port­ers and em­ploy­ees. How would that work?

Imag­ine as­sign­ing the to­tal im­pact my char­ity will ever have to a col­lec­tion of shares; let’s ar­bi­trar­ily say 100 shares. A bunch of peo­ple have made this im­pact pos­si­ble—em­ploy­ees, founders, fun­ders, and other sup­port­ers. We could imag­ine do­ing a re­ally ex­ten­sive credit as­sign­ment ex­er­cise where we as­sign shares to each per­son based on the im­por­tance of their con­tri­bu­tion. I’m not go­ing to ad­vo­cate ac­tu­ally figur­ing out who owns how many shares, but I’ll talk about all of these peo­ple as the “share­hold­ers.”

If I is­sue a new share, then the amount of im­pact as­signed to each ex­ist­ing share is re­duced: it’s now spread across 101 shares in­stead of 100 shares. If I give the new share to a donor in ex­change for some money, then hope­fully the money in­creases my im­pact by more than 1%, so that ex­ist­ing share­hold­ers end up with more to­tal im­pact de­spite the dilu­tion.

The pro­posal is: when I raise money for 2020 I also an­nounce a (pre-fund­ing) val­u­a­tion, the sales price I’m will­ing to offer per share times the num­ber of out­stand­ing shares. Then I give donors an ap­pro­pri­ate num­ber of new shares, and dilute ex­ist­ing share­hold­ers. Char­i­ties choose these val­u­a­tions to max­i­mize the af­ter-dilu­tion im­pact of ex­ist­ing share­hold­ers, while donors de­cide whether to donate based on whether they are OK with the to­tal im­pact they’d be as­signed.

For ex­am­ple, if my val­u­a­tion is $100M and a donor donates $500k, then they re­ceive 0.5/​100.5 of the to­tal im­pact (0.5/​100.5), and ex­ist­ing share­hold­ers are diluted 0.5%. (None of these num­bers de­pend on the ab­solute num­ber of ex­ist­ing shares, 100 re­ally was ar­bi­trary and has no con­se­quences.)

No one has an ex­ter­nal in­cen­tive to be­have hon­estly. The sys­tem is in­tended as an aid for gen­er­ally-al­igned donors and em­ploy­ees to ac­tu­ally un­der­stand how much they are con­tribut­ing. That said, if the dilu­tion num­bers are pub­lished then at least “be co­op­er­a­tive by com­mu­ni­cat­ing hon­estly” is al­igned with “make my to­tal con­tri­bu­tion look as large as pos­si­ble.” And to the ex­tent peo­ple are good at max­i­miz­ing num­bers, it seems good to tie the num­bers to a plau­si­ble es­ti­mate for their real im­pact.

How is val­u­a­tion de­ter­mined?

If I raise at a low val­u­a­tion, then I’m tel­ling donors they get a large share of the credit for my fu­ture ac­tivi­ties, which should make it eas­ier to raise money but will dilute cur­rent share­hold­ers. If I raise at a high val­u­a­tion, then I’m tel­ling donors they get a small share of the credit, so it’s harder to raise money but ex­ist­ing share­hold­ers re­tain their im­pact. My choice is a bal­ance be­tween these con­sid­er­a­tions.

If I have a fixed tar­get amount to raise for 2020, and a max­i­mum amount of time I’m will­ing to spend fundrais­ing, then there is a max­i­mum price I can charge while meet­ing my fundrais­ing goal, and so that’s where I want to end up:

If I’m in­ter­act­ing with large donors who all have a fixed thresh­old for “cost effec­tive enough,” then the donor de­mand curve is nearly ver­ti­cal. So I want to set my val­u­a­tion to be ex­actly at their thresh­old (though if the num­ber of char­i­ties and donors is small we have a bar­gain­ing prob­lem). In prac­tice most donors won’t have a to­tally ver­ti­cal de­mand curve, though for very large and in­ter­nally co­her­ent donors, it might be nearly ver­ti­cal. If there is a dis­tri­bu­tion of donors with differ­ent bars and differ­ent as­sess­ments, then I’ll end up with a smooth de­mand curve.

I can’t re­ally pre­dict this price in ad­vance, and so it will be de­ter­mined by some com­bi­na­tion of talk­ing with donors, re­vis­ing prices over the course of a fundraiser, and ad­just­ing across fundraisers to main­tain a de­sired level of run­way. For ex­am­ple, be­fore the fundraiser I may talk to large donors about what val­u­a­tions they think are ap­pro­pri­ate and try to find a price that makes sense. If I raise more money than I’m able to eas­ily spend, then I’ll in­crease my val­u­a­tion for fundrais­ing dur­ing the com­ing year.

In re­al­ity the amount of money I need to raise for 2020 is not fixed in ad­vance. I can main­tain a larger run­way, offer bet­ter salaries, hire more peo­ple, and so on. How much I should raise de­pends on the op­por­tu­nity cost of funds: is it worth spend­ing an ex­tra $1M to in­crease my to­tal im­pact by 1%? This in­for­ma­tion is com­mu­ni­cated to me via the price that donors are will­ing to pay. If I have a bet­ter val­u­a­tion, rais­ing $1 causes less dilu­tion and so I raise more money.

In this case, dis­cus­sion and grad­ual iter­a­tion are even more im­por­tant to dis­cov­er­ing an ap­pro­pri­ate price—and figur­ing out how ag­gres­sively I should spend money to im­prove my pro­duc­tivity or ex­pand ex­ist­ing pro­grams.

Where does this mat­ter?

I think be­ing more ex­plicit about al­tru­is­tic equity al­lo­ca­tion could po­ten­tially im­prove de­ci­sions, and maybe even de­crease to­tal time and angst. This sec­tion de­scribes some of the ar­eas I think could be im­proved. It’s not at all ex­haus­tive.

(I do think that this ex­per­i­ment, like most ex­per­i­ments, would prob­a­bly be a failure. So you should think of this more as a list of pos­si­ble ad­van­tages rather than strong pre­dic­tions, with the main up­side of an ex­per­i­ment be­ing value of in­for­ma­tion.)

Donor co­or­di­na­tion. There are of­ten mul­ti­ple donors in­ter­ested in fund­ing the same char­i­ties. I think the cur­rent de facto ap­proach, a mix of “split­ting” or “fung­ing,” is pretty un­satis­fy­ing. In­stead, we can ad­just the val­u­a­tion un­til sup­ply equals de­mand and each donor gives as much as they think is rea­son­able at the given val­u­a­tion (with­out hav­ing to think about other donors).

This re­sults in the most ex­cited donors sup­port­ing each char­ity rather than ev­ery­one chip­ping in. I think this is lo­cally effi­cient, but one con­cern is that more con­cen­trated donor bases are less ro­bust. A char­ity can try to miti­gate this risk by un­der­stand­ing how much they would need to re­duce their val­u­a­tion to get ad­di­tional donors in­ter­ested.

Spend­ing de­ci­sions. Char­i­ties have a lot of in­for­ma­tion about how they can spend money to im­prove their im­pact, the rel­a­tive promise of new pro­gram ar­eas, how use­ful they think run­way would be, and so on. Right now de­ci­sions are es­sen­tially made jointly by char­i­ties and donors, based on a dis­cus­sion that is rel­a­tively low band­width and not in­cen­tive al­igned. I’d pre­fer the situ­a­tion where donors com­mu­ni­cate the op­por­tu­nity cost of funds, and then char­i­ties make de­ci­sions about how to spend the money given that cost. I don’t think we could elimi­nate the need for donors to un­der­stand marginal de­ci­sions, but I think we could make things a lot more mod­u­lar than the sta­tus quo.

Com­mu­ni­cat­ing about RFMF. EA dona­tion de­ci­sions of­ten de­pend on as­sess­ing “room for more fund­ing,” while the effects of fund­ing are of­ten to dis­place other donors and/​or re­duce the at­ten­tion an or­ga­ni­za­tion has to spend fundrais­ing. I think we are pretty bad at think­ing and talk­ing about this, and that the equity model pro­duces ba­si­cally rea­son­able de­ci­sions at equil­ibrium. Donors can eval­u­ate RFMF by look­ing at val­u­a­tions rather than guessti­mat­ing, and char­i­ties can make a lo­cal greedy de­ci­sion about whether to spend more time fundrais­ing based on how much it will al­low them to re­duce their val­u­a­tion.

Em­ploy­ees and donors. The EA com­mu­nity has some­times been in­con­sis­tent in how it talks and thinks about the rel­a­tive im­pact of dona­tions and di­rect work. Is­su­ing equity helps quan­tify the im­pact of marginal dona­tions, which I think would im­prove think­ing and dis­cus­sion on this topic. You’d get more mileage if you also quan­tified how much al­tru­is­tic equity you were giv­ing to em­ploy­ees as part of their com­pen­sa­tion, but just see­ing val­u­a­tions is already a lot bet­ter than the sta­tus quo.

Think­ing about lev­er­age. Some spec­u­la­tive pro­jects may mostly be prov­ing an idea or fa­cil­i­tat­ing a big scale-up in the fu­ture, which would re­quire rais­ing more money and hiring more peo­ple. I think that the equity model is a rea­son­able way to think about this kind of im­pact—once the idea is proven, later donors in­vest at a much higher val­u­a­tion, and early donors re­tain a rea­son­able share of the sub­se­quent im­pact. If you can’t raise at that higher val­u­a­tion, then the story about lev­er­age may be over­count­ing. (In some cases you would need to ex­plic­itly sep­a­rate “tranches” of im­pact in or­der to re­ally make the lev­er­age case work, if differ­ent donors care about differ­ent parts of the im­pact, which would be more of a pain.)

Equity in what?

In ad­di­tion to choos­ing what frac­tion of equity they sell to donors, a char­ity needs to choose ex­actly what im­pact they are sel­l­ing—if a donor is buy­ing 1% of my im­pact for $1M, what ex­actly are they buy­ing 1% of?

I think the most nat­u­ral op­tions are “all of this char­ity’s im­pact” or “all of this char­ity’s fu­ture im­pact.” Each of these have is­sues:

  • Sel­ling fu­ture im­pact is tricky, be­cause of­ten past ac­tivi­ties fa­cil­i­tate fu­ture ac­tivi­ties and draw­ing the line is re­ally un­clear (i.e. it de­pends on its own credit al­lo­ca­tion prob­lem).

  • Sel­ling past im­pact can be tricky, be­cause some­times past ac­tivi­ties are very differ­ent from cur­rent ac­tivi­ties, the peo­ple who sup­ported past ac­tivi­ties might be less ex­cited about fu­ture ac­tivi­ties, or the lines might be­come blurry be­tween in­side and out­side the char­ity.

In some cases there will be an in­ter­me­di­ate that makes sense. For ex­am­ple, I might sell the im­pact of “all of my re­search, all of my fu­ture ac­tivi­ties, and all of the past work that fa­cil­i­tated those fu­ture ac­tivi­ties,” if it’s eas­ier to draw lines around that then around ei­ther “all my fu­ture im­pact” or “all my im­pact.”

The most im­por­tant thing is for char­i­ties and donors to be on the same page about what is be­ing in­cluded.

What could this ac­tu­ally look like?

Here’s how I’d imag­ine an ex­per­i­ment right now:

  • A non-profit de­cides they want to try this kind of ex­per­i­ment in a fu­ture fundraiser.

  • They have pre­limi­nary dis­cus­sions with past and pre­sent sup­port­ers, es­pe­cially cur­rent em­ploy­ees and big donors, try­ing to find a val­u­a­tion that could make both donors and past donors/​em­ploy­ees happy. If they can’t find such a val­u­a­tion, then it’s an in­ter­est­ing op­por­tu­nity for some re­flec­tion.

  • They de­cide on a fundraiser to roll it out and a pre­limi­nary val­u­a­tion to offer; they have some more dis­cus­sions with big fu­ture donors (who might be giv­ing out­side of the fundraiser proper) con­firm­ing they feel OK about it.

  • As part of the fundraiser, they pub­li­cly state a pre-fund­ing val­u­a­tion. This need not be em­pha­sized or in-your-face, but it should be clear to any­one who is look­ing for it.

  • Pub­lish the to­tal dilu­tion in a pub­lic log. E.g. if they raise $2M at a $50M val­u­a­tion, an­nounce that pre­vi­ous stake­hold­ers were diluted by ~4%.

To make this use­ful, it would be good for a first ex­per­i­ment to also in­clude some ex­po­si­tion. For ex­am­ple, it would be great for an early im­ple­menter to:

  • Ex­plain roughly how they ar­rived at their val­u­a­tion and how they feel about it.

  • Doc­u­ment any les­sons from the pro­cess and over­all how smoothly it went.

  • Get some­one ex­ter­nal to en­gage se­ri­ously with the pro­posed val­u­a­tion and pro­cess, and think about ways in which it makes sense or doesn’t make sense.

I think an ini­tial ex­per­i­ment could give in­for­ma­tion on how peo­ple feel about this kind of ar­range­ment, iron out de­tails, provide pub­lic baselines so that the next per­son doesn’t feel so in the dark, and ver­ify­ing that noth­ing ter­rible hap­pens.