How meta-charity can dominate direct charity

Solid scientific evidence has confirmed that deworming is a best buy in development aid. The direct charities that REG recommends thus either perform deworming interventions or interventions that are comparably cost-effective. Charity cost-effectiveness is measured by the extent of the life-saving/medical, educational, economic and social benefits they provide. An intervention whose cost-effectiveness has been shown to be comparable to deworming is anti-malarial bed-net distribution.

REG takes a strategic approach to charity: We strive to maximize the expected value we get out of the resources we invest in pursuit of charitable goals. This approach necessitates research into strategies that can “beat” the best existing direct charities (deworming, anti-malaria). One upshot of our research is “meta-charity”: By successfully fundraising for the best direct charities, it should be possible to achieve an even greater charitable impact per dollar invested.

The decisive questions, though, are the following:

1) Is meta-charity measurable?
2) If so, is it provably more cost-effective than direct deworming or anti-malaria interventions?

With regard to 1), we argue that if direct interventions such as deworming are measurable, then so is respective meta-charity: Scientific evaluators of direct charities (such as GiveWell.org) make a compelling case that direct charity is in fact measurable. The current best buys in development aid can be expected to save a life for an approximate marginal $3,400. (Considering them “best buys” necessitates belief in their measurability, needless to say.)

The fundraising effectiveness of projects that aim to raise money for these best buys is intrinsically easy to measure: We can just look at the fundraising expenses and compare them to the total amount of money raised. Take the example of REG: Last year, the organisation had overall expenses of about $50k and generated total donations of more than $500k – which equals a fund-ratio greater than 10 : 1. If we can measure the fundraising effectiveness of a meta-charity; and if the life-saving effectiveness of the recipient direct charity is measurable too; then it follows that the life-saving effectiveness of the meta-charity is indeed measurable.

With regard to 2), the question can be answered by checking whether the meta-charity’s fund-ratio is greater than 1 : 1, i.e. whether for every dollar invested into its fundraising, more than one dollar ends up at the recipient direct charities.

The story is more complicated than that, though, since the counterfactual behaviour of the donors reached by the meta-charity’s fundraising efforts must be taken into account. The decisive questions here are:

2.1) Would the donors reached by the meta-charity have donated their money anyway? What fraction of it?
2.2) If so, would the money have gone to direct charities as cost-effective as the ones the meta-charity is fundraising for?

As regards 2.1), might REG’s superficial fund-ratio of $500k : $50k = 10 : 1 be an illusion? How much of the $500k would have been donated anyway? The evidence suggests: Not much. The conversations with our members have shown that a clear majority of them were inspired to consider (unusually large) donations due to REG’s influence. In an extremely pessimistic case, REG’s actual fund-ratio may be 5 : 1. This estimate is much more conservative than what the evidence should lead us to believe in, which is a fund-ratio of at least 8 : 1.

As regards 2.2), how cost-effective are the charities that REG members would otherwise have donated to? The answer to this question is even clearer than the answer to 2.1). The approach of strategically maximizing the life-saving expected value of donations is novel. To the extent that REG donors would have donated anyway, their money would most likely have gone to rather randomly selected charities, i.e. to charities that are (expectedly) median in life-saving cost-effectiveness. As the research by charity evaluators such as GiveWell.org has clearly shown, the very top charities are 10x or even 100x as cost-effective median charities. (Medical interventions in the poorest countries tend to be about 100x as cost-effective as medical interventions in rich countries: Adding one quality-adjusted year of life costs about $30,000 in the UK, while the same can be achieved for less than $300 by deworming or anti-malarial interventions in very poor countries.)

Given the considerations pertaining to 2.2), REG would likely be multiplying the life-saving effect of the donations it receives even if its fund-ratio were just, for instance, 1.1 : 1.

It follows that an actual multiplier effect of REG’s meta-charity efforts of 10 : 1 is a somewhat conservative and thus very realistic estimate. The example of REG shows that meta-charities (for additional examples, see this post) can dominate the direct charities they are fundraising for.