Altruistic equity allocation

post by Paul_Christiano · 2019-10-16T05:54:49.426Z · score: 79 (30 votes) · EA · GW · 5 comments

Contents

  Issuing altruistic equity
  How is valuation determined?
  Where does this matter?
  Equity in what?
  What could this actually look like?
None
5 comments

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 [EA · GW], 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:

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:

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:

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.

5 comments

Comments sorted by top scores.

comment by cwgoes · 2019-10-17T10:26:34.249Z · score: 3 (2 votes) · EA(p) · GW(p)

I find the broad proposal of quantifying impact in an equity-like form and providing that "impact equity" to contributing stakeholders proportionally to their contribution to the organization's impact in an attempt to improve the efficiency of the matching market between organizations and donors compelling. I also think it may have several second-order effects, of which some may be positive and some may be negative, which merit consideration.

More efficient matching markets between organizations and donors

It seems plausible that quantizable impact equity would substantially simplify the donor coordination problem and reduce the communication complexity involved in charity spending decisions. As I understand your proposal, that simplification will result from the distillation of the charity's program options into a single (or perhaps several, since there may be threshold effects in implementing particular programs) impact equity offers, which can then be evaluated by donors on a comparative basis to other impact equity offers they have from other charities. Charities then need not spend time proposing particular programs to individual donors, and donors needn't understand the details of the particular programs the charity might pursue.

Less clear to me is how comparable impact equity of different charities would be - would there be a specific metric (e.g. estimated QALYs saved) or would donors construct individual conversion rates (at least implicitly) based on their evaluations of how effective charities are likely to be over their lifetimes? The former further simplifies the evaluation complexity for donors but might be hard to evaluate for many kinds of programs and could lead to problematic incentives (if donor decisions are being made somewhat automatically, charities which overestimate their QALYs saved will have an advantage), while the latter might be harder for a donor to understand or estimate than the impact of a specific program (although this may vary widely based on the charity). This is different than venture capital investing, where the input investment into and eventual expected returns of all potential companies are in the same currency.

Alterations of donor incentives

The reification of contributions into measurable impact equity could lead to a more competitive game between donors since this impact equity confers a measurable kind of status which can easily be publicly displayed. This might have strong positive effects - since the status gained in return is more valuable, donors might donate more - concretely, imagine if everyone's "impact score" were displayed beside their forum profile or Twitter account. Measurable impact equity might also lead to perverse incentives to overestimate impact, depending on how impact equity from different organisations is compared (see above paragraph).

If the impact return curve of charities is some kind of power law, impact equity might cause donors to contribute smaller amounts to many newer organisations in the expectation that most will fail to have much impact but one or two will succeed and "return the fund". I don't know enough about the current funding landscape to judge whether newer organisations are currently over-funded or under-funded.

Alterations of employee incentives

Impact equity could be given to donors only, or given to donors and organisation employees. The latter is the closest to the for-profit equity model, where the founders start with all of the equity and then parcel it out to employees and investors in recompense for work or capital. Giving impact equity to employees might have positive effects - as for donors, it is tangible status and might encourage more employees to join the organisation or encourage existing employees to work harder to increase the future impact of the organisation (and thus the value of their impact equity). However, it might also lead to discontent if employees don't consider the impact equity allocations to be fair (whether between different employees, between employees and founders, or between employees and investors). One other advantage of not quantizing the individual contributions of employees is that they can sum up to more than 100% - all twenty employees of an organisation may each believe that they are responsible for at least 10% of its success, which is mathematically inconsistent but may be a useful fiction (and in some sense it could be true - there may be threshold effects such that if any individual employee left the impact of the organisation would actually be 10% worse) - if impact equity is explicitly parceled out, everyone's fractions will sum to 1.

comment by Paul_Christiano · 2019-10-17T15:28:33.017Z · score: 4 (3 votes) · EA(p) · GW(p)
would there be a specific metric (e.g. estimated QALYs saved) or would donors construct individual conversion rates (at least implicitly) based on their evaluations of how effective charities are likely to be over their lifetimes?

It would come down to donor predictions, and different donors will generally have quite different predictions (similar to for-profit investing). I agree there is a further difference where donors will also value different outputs differently.

One other advantage of not quantizing the individual contributions of employees is that they can sum up to more than 100% - all twenty employees of an organisation may each believe that they are responsible for at least 10% of its success, which is mathematically inconsistent but may be a useful fiction (and in some sense it could be true - there may be threshold effects such that if any individual employee left the impact of the organisation would actually be 10% worse) - if impact equity is explicitly parceled out, everyone's fractions will sum to 1.

I mostly consider this an advantage of quantifying :)

(I also think that impacts should sum to 1, not >1---in the sense that a project is worthwhile iff there is a way of allocating its impact that makes everyone happy, modulo the issue where you may need to separate impact into tranches for unaligned employees who value different parts of that impact.)

However, it might also lead to discontent if employees don't consider the impact equity allocations to be fair (whether between different employees, between employees and founders, or between employees and investors).

This seems like a real downside.

comment by Mati_Roy · 2019-11-19T22:56:33.385Z · score: 2 (2 votes) · EA(p) · GW(p)

Added here: https://causeprioritization.org/Moral_economics

comment by cole_haus · 2019-10-17T20:17:39.224Z · score: 2 (2 votes) · EA(p) · GW(p)

A charity can try to mitigate this risk by understanding how much they would need to reduce their valuation to get additional donors interested.

A Vickrey auction should be a nice way of addressing this problem. The charity ends up with a complete, honest listing of all bidder's valuations. Google's IPO was structured along these lines.

How is valuation determined?

I think it's also nice to think about this from an "inside view". Typically, an asset is priced such that the price equals the net present value of the expected stream of rents the asset generates. The ability to calculate a valuation from (slightly) more tangible inputs means that the charity doesn't have to pick a totally arbitrary number when they're starting their price-finding/fundraising process. This definition also helps guide the "Equity in what?" discussion.

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.

I know you have this and other caveats throughout, but I definitely worry that it'll be hard to ignore the numbers when they create perverse incentives (Goodhart's law, Campbell's law, etc.). Market mechanisms have lots of problems and I wouldn't want to import those into the charitable sector unnecessarily (which is not to say that markets don't have good features or that the charitable sector doesn't have structural issues). I'll have to think about this more carefully, but it seems like it would be nice to design a mechanism which is tailored for the unique features of the charitable sector (e.g. incentives ought to be more aligned and less zero sum, any potential market is probably fairly illiquid).

Another thing to think about: Equity typically confers voting rights. Is that appropriate here? Why or why not? The why not argument that comes to mind immediately is: In typical for-profit companies, voting rights are useful as a way of disciplining managers that have their own private incentives. Hopefully, the incentives of donors and charity management are already fairly aligned.

comment by ishi · 2019-10-18T17:57:34.605Z · score: 0 (5 votes) · EA(p) · GW(p)

I only skimmed this and am not familiar much with investing, but the issues seem to me whether people are investing to make money, or investing to make 'social impact' (for the greater good) on goals they support. and also whether the investment actually will 'pay off' in terms of either making money, having social impact, or both. I think forecasting 'impact' is the hardest one. Some venture capital firms succeed , others fail.

My view is if impact predictions are correct then any 'dilution' effects on shareholder value will be temporary. In a sense current shareholders are loaning some of their shares to others to invest in the project in the hope its a good investment. (I think in a way what is called MMT--popular in some economics circles---is a version of this idea).