We’re only scratching the surface in the design space of “how can people with money (funders) support provide who need money (grantees)”.
- Restricted vs Unrestricted
- Restricted tends to be the default — “I’ll give you money for this project”. But this is quite annoying for nonprofits; reduces their flexibility and ability to react
- Lump sum vs Over time
- Over time would be less expensive (wrt cost of capital to funder); keep incentives aligned. Could have checkpoints to see regular reports of more work.
- Is it good for funders to have more touchpoints with their grantees?
- Failure mode: funders telling grantees how to do their work, when grantees are the actual expert
- To individuals vs to orgs
- Microgrants to individuals has become very popular: Emergent Ventures, Fast Grants, ACX Grants, etc
- Grants to individuals may make them less likely to work within an org
On the surface, it might seem that grants are the most desirable. Free money! Some problems with grants:
- Implicit restriction on what you can spend it on
- Grantees have to try and predict “does this make the funder happy”
- Unlike in for-profit companies, where the objectives are clearer
- Grants move “PR risk budget” onto the funder
- Cf “Every grant is also a bounty”
- Necessitates more screening by the funder, more cautiousness. If you make one bad grant, everyone comes at you
- But more screening also weeds out the weird but promising ideas
- Less mechanical alignment between grantee and funder.
- The incentive of a grantor is to pick grants that look good instead of grants that will pay out well in the future
- A loan or an equity investment requires a story of some kind, that
- Most expensive for the grantor
- When you give a loan, there’s some chance of default, but in expectation you’ll be able to make back your capital
- Grantors have to ask to raise more money after, and would have to optimize for “good-lookingness” of grants
- Free money might attract scammers?
- Free money might encourage free-spending (or appear to)
- Loans to individuals
- E.g. student loans? Common for expensive long programs like med school
- Loans to nonprofit orgs
- E.g. contingent on future fundraising
- Would larger grantmakers be unwilling to pay future loans, if what that does is just subsidize earlier grantmakers?
- If a grantmaker has promised funding but is slow about delivering it (cough SFF), then a loan can help smoothen operations
- Providing access to finance in Africa was a popular thing. why didn’t this end up working out/seeming cost effective, esp in comparison to GiveDirectly?
- Unsecured vs secured
- Generally hard to securitize the kinds of loans given in EA…
- $300k loan: securitized against Oli’s personal assets
Why get loans in EA, compared to from a commercial source?
- Lower interest rates, because:
- More trust within EA community. An EA funder has much more context about the work of the grantee, compared to a for-profit bank.
- The grantee and funder operate in the same community, so the grantee is more socially incentivized not to default
- Funders and grantees share incentives. If you’re going to conduct positive-sum trade, you’d rather it go to someone who shares your values. Grantee prefers profits go to funder; funder prefers grantees get cheaper money.
- In the case of a default, more willing to write off as a grant
- Can move faster - this doesn’t currently happen at all
- Extremely unclear that commercial sources would provide loans to upskill in safety
Note: Loans don’t have to be net profitable to collect meaningful returns
- Can have zero interest rate
- Or negative interest rate! Kind of a hybrid, putting you on a scale closer to a grant
- “I’ll lend you $12k now; you pay back $1k a month for the next 6 months”
- For a total interest rate of -50%.
- Income share agreements
- An economists’s dream. Funders do better as the grantee makes more.
- Never really took off at Upstart or Lambda School but still seems kind of reasonable, especially for…
- Upskilling to work at a top TAIS lab (eg OpenAI, Anthropic, Deepmind) where salaries are $250-500k
- Example: could sell 5% of future salary earned by TAIS
- What about work that is valuable but not highly compensated? Happens a lot in nonprofit orgs, e.g. if OP would prefer 1 good hire over $1.5m more funding, but pays $150k in salary to that good hire.
- Ideally OP pays some of that surplus to the seed funder…?
- Literal equity
- Great for for-profits (C-corps, LLCs)
- Not allowable for nonprofits
- Impact certificates
- Kind of like the above, though currently contingent on prizes
- Perhaps: nonprofits should view funds raised as revenue income stream
Is more “financialization” within EA good?
- Financialization makes explicit what used to be implicit. Is this good?
- Classic study: when a daycare started charging fees for late pickups, parents felt empowered to just come in late
- Austin thinks the daycare should have just charged more (until you hit the market clearing rate/happy price)
- Explicit metrics give very concrete feedback
- Yes yes, Goodheart, but Goodheart doesn’t mean “let’s stop counting how many users we’re serving”
- Money is a great proxy for impact, one of the best that we have.
- EA was founded on something like “financializing altruism”
- There is something warm & fuzzy about “free” though, which may encourage better community norms?
- Too much financialization reduces the necessity for trust
- Once upon a time, to get to the airport for a late night flight, my dad would call up his brother; with the understanding that when my brother needed the same service, he would help out. Now… we just call Ubers.
- Vulnerability & dependence are what forms strong relationships
- Early Christians: no charging within each other?