Case study: designing a new organisation that might be more effective than GiveWell’s top recommendation

Introduction

Several months ago, we wrote about an apparently easy way to create a charity that’s more effective than GiveWell’s top recommendations. It’s a simple idea: create an organisation that does nothing except fundraise for GiveWell’s top recommendations. It seems relatively easy to raise more than $1 for every $1 invested in fundraising, so it seems relatively easy to act as a multiplier on donations to other charities, and thus create an organisation with a cost-effectiveness ratio that’s higher than the charities themselves.

The idea has merit from other angles too, for instance, there’s reason to expect that charities don’t fundraise as much as would be optimal, and it’s a valuable service to them to take fundraising off their books. It’s also promising because the success of fundraising is relatively easy to quantify compared to many other meta-charity activities (like doing research or advocacy). This means that you can learn quickly whether you’re succeeding, which means you can improve quickly.

We were thrilled, therefore, to find that two 80k members, Joey and Xio, are planning to start an organisation that does exactly this. It’s called Effective Fundraising. Their plan is to start by writing grants for Against Malaria Foundation (AMF) and The Humane League for six months. If it goes well, they could hire more people. Longer term, they could expand into others kinds of fundraising. They chose grant writing because (i) surveys of average fundraising ratios found that grant writing earns an average return of around $8 for every $1 invested, which is higher than most other forms of fundraising (ii) the money can be raised within 6-18 months, unlike ‘chugging’ (asking for donations on the street) or door-to-door which takes several years to pay off.

Joey and Xio sought out a huge amount of advice on their strategy. They’re some of the most altruistic people we know, and recently put college on hold in Canada in order to work on effective altruist start-ups. One of the sources they consulted was 80,000 Hours, and we were very happy to help. Rob Wiblin and I spent several hours thinking about their strategy and several hours discussing with Joey and Xio. (They have also sought advice from the other employees of 80,000 Hours and our President, Will MacAskill).

Joey and Xio were mainly seeking advice on how best to execute the project. In the rest of this post, we outline the main recommendations we made during the course of the discussion. Although our discussions didn’t lead to an immediate change in their plans, we hope this contributes to the discussion that Joey and Xio have been leading on how best to fundraise for top GiveWell recommendations, and inspires others to consider similar ways to make a difference. In summary:

Summary

We think:

  1. Working on Effective Fundraising looks like a very strong option for building career capital.

  2. They could consider running more experiments before committing to grant-writing for 6 months.

  3. They may be having less impact than they could because they may not be supporting the most high priority causes.

  4. They should strongly consider hiring someone else to work for Effective Fundraising as a grant writer.

(1) We agree that working on Effective Fundraising looks like a very strong option for building career capital

By founding their own organisation, we expect them to learn a variety of very useful entrepreneurial skills very rapidly. They should also improve their ability to fundraise, which is a highly useful skill for someone looking to work in the non-profit sector, but also involves sales skills that are useful almost everywhere.

This seems better than the skill development they could have working at another effective altruist organisation, where they would probably be given less responsibility, particularly if they make sure they prioritise learning while running Effective Fundraising (which they are).

In addition, Effective Fundraising offers more long-term upside. If it works, it could grow into large, new organisation within effective altruism.

(2) They could consider running more experiments before committing to grant-writing for 6 months.

We both agree that it’s very important to test this project before scaling it up. We place higher weight on this factor than Joey and Xio, because because we think:

  1. There’s a large amount of uncertainty about much money they might expect to raise with different methods

  2. It seems relatively easy to learn more about how much money they might expect to raise by trying out different methods for a couple of months each.

  3. This information would be very valuable, because it could significantly improve how they spend a year of effort, and the information would be valuable to other effective altruists considering the same thing.

Moreover, we think testing grant writing has lower value of information than testing other fundraising, because the feedback is weak.

First, grant writing often has low success ratios. If your base rate of success is only 5%, then you will need to write a lot of grants to measure your ratio. We did a quick analysis, assuming that you have a 50% chance of being as good as professional grant writers (who we modelled as having three levels of ability, based on Institute of Fundraising (IoF) data and assuming a base rate of success of 5%), and a 50% chance of not being able to raise money via grant-writing. We then asked: how many attempts would you have to make before you can be confident that grant writing is better than a guaranteed 3 times pay-off? Assuming a binomial distribution on successes, and performing a bayesian update, we found:

  • It would take 47 failures in a row to become 90% confident that grant writing is worse

  • It would take 13 failures in a row for the expected value of grant writing to drop below a guaranteed 3 times return.

  • On the other hand, 2-3 successes out of the first 5 or so would push grant writing strongly ahead. (You can see the spreadsheet here)

Second, you have to wait 6-12 months to hear the results.

Note that both of these concerns can be mitigated to some extent by starting with easy grants that have close deadlines.

Joey and Xio, however, think that the reasons behind grant writing are sufficiently strong that they should start immediately and do it for six months as a test. In particular they respond:

  • The uncertainty is less than made out, because there is some data provided by the IoF on the returns of different methods. In particular, the average grant-writer makes 8 times their salary, and the bottom 15% make 5 times their salary. So, even if they perform at the bottom of grant writers, it will be better to do grant-writing than most other forms of fundraising.

  • It’s harder to experiment than it looks. “You need licenses to do door-to-door fundraising and street fundraising, and it takes years for you to determine if the recurring donations have a good ratio. Legacies take several years, and sometimes decades to see returns. Both corporate and high net worth donors take considerable time to learn the field and become good at. In 6 months it might be possible to sufficiently test one of these. We think the data, while not as strong as we would like, is enough to show that grant writing is promising, and that it is worth a six to twelve month test.”

We note some countervailing considerations:

  • It’s unclear how good the Institute of Fundraising data is, because they don’t publish their methods. For instance, they may not correct for survivorship bias—maybe fundraisers who don’t succeed don’t survive till the end of the year, so don’t appear in the data as collected. We’d also expect the figures to be systematically overstated, because charities have an incentive to keep fundraising costs low, so will try to put fundraising costs under different budget lines (moreover, this seems easier to do with grant writing or high net-worth fundraising than with chugging, because much grant writing is done by managers who have other responsibilities).

  • The IoF data seems to be the returns earned by people who currently work as grant-writers. Grant-writers who earn less than 5 times their salary lose their jobs very quickly, so the data obviously shows that the ratios are higher than 5 times. This information doesn’t tell us much about how much money Joey and Xio might expect to earn by grant-writing.

  • There are reasons to expect Joey and Xio will earn less. The average grant-writer is paid over £30,000 per year, which is a professional salary and probably equivalent to something like earning £60,000 in the corporate sector. This suggests that it’s a competitive field. On the other hand, this might be balanced out by the fact that (i) Against Malaria Foundation should be relatively easy to fundraise for (ii) Joey and Xio will prioritise learning and feedback so that they improve at grant writing as quickly as possible.

  • Even if the licenses for other types of fundraising take several days of full-time work to get, it will still be worth it. Moreover, there seem to be ways to do tests without a license, for instance, you don’t collect payment details and instead just collect pledges for donations and email addresses.

In the end, we didn’t reach agreement. Joey and Xio still intend to commit to grant writing, while we suspect it may be better to prioritise experimentation slightly higher.

(3) They may be having less impact than they could because they may not be supporting the most high priority causes.

We agree that supporting GiveWell recommendations is highly valuable and that the project has an attractive flexibility, since (to some extent) it will be possible to fundraise for different organisations as the evidence changes.

We think, however, that AMF and The Humane League might not be the most promising uses of money currently available, so Joey and Xio might be able to increase expected returns by funding someone else.

Joey and Xio’s response is that: They are a fundraising organization, not a charity evaluation organization. They take top recommendations from Givewell and Effective Animal Activism, two of the most recognized charity evaluators in their respective fields.

We think that going with GiveWell’s recommendations is a reasonable move as it’s plausibly the best charity evaluator in the world, but we also think it’s worth putting more effort into the decision if your aim is really to maximise expected returns.

First, many leading experts both within and outside of effective altruism don’t think that GiveWell’s recommendations clearly offer the best returns. As three examples within global poverty, IPA, AidGrade and Giving What We Can are experts on evidence-based aid, and they don’t fully agree with GiveWell although there is overlap (see here, here and here). Disagreements are even larger if you also consider charities outside the global poverty cause.

More importantly, even GiveWell don’t claim that giving to AMF is generically the use of money with the highest expected return. Rather they claim that it’s the best donation opportunity to an existing organisation that’s proven, scalable, has room for more funding and is transparent enough that uninvolved third parties can see that it is. This leaves open the possibility that there are organisations that might be more effective to fund, work for or set up:

  1. Unproven organisations—these could worth investing in (a) to gain information about what works or (b) because they offer high expected return despite low certainty.

  2. Organisations that Joey and Xio have special insight into not available to GiveWell.

  3. Organisations that Joey and Xio could support to gain valuable information about what works that wouldn’t give GiveWell the same level of information.

  4. Small organisations (if room for more funding is less than several hundred thousand dollars, GiveWell won’t normally investigate them, though there are reasons to think many of these opportunities are taken, at least outside of meta-charity style projects).

  5. Organisations that don’t exist yet, but could be created if a donor expressed interest in them or a credible entrepreneur started them. GiveWell has suggested that the approach of strategically picking a cause and then being an active donor could yield better results, at least in some cases.

  6. Organisations that GiveWell hasn’t had time to investigate yet.

  7. Organisations that don’t share GiveWell’s values (e.g. GiveWell doesn’t currently factor animal welfare into their evaluation). Going with GiveWell recommendations means ignoring opportunities to support organisations that fall into these categories, which plausibly contain organisations that are higher impact.

Note that after this post was written, Holden replied (see the comments below), stating that GiveWell does believe that giving to AMF is currently the charitable donation opportunity with the highest expected return, in contraction to the statement above

One response could be that making calls like this is difficult, so it’s better just to stick with GiveWell. But Joey and Xio have already decided to funnel 50% of the funds raised to Effective Animal Activism’s top recommendation, which has to date invested far less research time than GiveWell in charity evaluation. It seems inconsistent to overrule GiveWell in this case but not in the other cases.

As an aside, note that if you think that the best opportunities will typically be in these categories, then the idea of Effective Fundraising is slightly less attractive, because opportunities like these are typically difficult to fundraise for as an outsider (and seem easier to support through direct work or earning to give). We think it’s plausible that the best opportunities in these categories will often be more attractive than proven, scalable charities, because (i) value of information is very important because we’re highly uncertain what’s best and opportunities like this offer more information about what works (ii) being an active donor or entrepreneur opens up the space of projects a great deal, which increases the chance finding a good one provided you have a search strategy that’s somewhat effective (iii) broad market efficiency—if it has already been recognized by GiveWell, then it willl be attracting the attention of some major funders.

(4) They should strongly consider hiring someone else to work for Effective Fundraising as a grant writer.

Employing someone to write grants for AMF seems like a good use of money, but it’s something that you don’t need an effective altruist to do. Rather, it might be better to find a grant writer with a track record of several years and ask them to write the grants. This seems more proven and likely to result in higher returns more quickly. The fundraiser could also be paid partially on commission, reducing the downside to Effective Fundraising. This would free Joey and Xio up to work on projects that are less easy to outsource. Or, at least it could help them learn to grant write quicker. I offered to help them raise money to pay for such a person.

Joey and Xio raised some important additional points:

  1. It would be difficult to raise the money, especially if they don’t want it to take funds away from effective charities.

  2. It’s too risky—they prefer to trial the project themselves before scaling up.

  3. They’re worried an external fundraiser might not be effectiveness oriented, which would make the organisation less flexible.

  4. They did not feel an external fundraiser, costing three times their combined salary, would likely raise 6 times more per hour than they would.

We agree (1) and (3) are concerns, though don’t seem to sink the idea. Rather, it would be better to make some tests to see if (1) and (3) are true.

(4) is not the relevant comparison. First, what matters is marginal effectiveness not average effectiveness. Even if the extra fundraiser has a lower fundraising ratio, it could still be worth hiring them because it increases the total impact. Second, the ratio of money raised to salary is not what’s relevant from the perspective of maximising impact. Joey and Xio are happy to work on impressively low salaries, which boosts the headline cost-effectiveness of Effective Fundraising, but their opportunity cost also matters. Joey and Xio may have a higher opportunity cost than a fundraiser, because they may have the option to work on another highly effective project.

(2) also doesn’t seem right. If Joey and Xio try to write grants for AMF and fail, then we haven’t learned that Effective Fundraising is a bad idea. Rather, we’ve just learned that Joey and Xio shouldn’t write grants. In this case, it seems like you’d still want to find someone else to try to write grants for AMF. If Joey and Xio try to write grants for AMF and succeed, then they’ll want to hire someone else to scale up. Either way, they’ll want to someone else to be employed as a grant writer, and delaying that process is just slowing down moving money to AMF. Moreover, it seems that it’s less risky to hire a professional with a track record than to do grant writing themselves.

We didn’t reach agreement and Joey and Xio intend to delay hiring another staff member for at least 6 months.

Further reading

If you would like to get involved, contact Joey and Xio.

If you want to read more about Effective Fundraising, see this discussion.

You can also check out an interview between our trustee Nick Beckstead, Joey and a donor here.

If you’re interested in doing something similar yourself, we’d be interested in doing a case study on your choice.

Crossposted from the 80,000 Hours blog