Review of Fundraising Activities of EAF in 2018post by stefan.torges (storges) · 2019-06-04T17:34:52.644Z · score: 46 (19 votes) · EA · GW · 3 comments
Summary Review of 2018 Overview Counterfactual estimates Raising for Effective Giving Regranting Expense and opportunity cost estimates Donation distribution By cause area By charity Development over time Future plans None 3 comments
This is an update on all fundraising activities of the Effective Altruism Foundation designed to raise funds for other high-impact charities. This mainly includes Raising for Effective Giving and our tax-deductible regranting.
- In 2018, we raised $9,968,506 for high-impact charities. Conservatively, we estimate that $5,545,783 (56%) would not have been donated otherwise (Raising for Effective Giving: $5,160,173; regranting: $385,609).
- We estimate the combined expenses for these two projects in 2018 to have been about $171,263. That means for every $1 spent on these projects we raised another $32. Taking into account opportunity costs of about $375,000, the net impact of these two projects was $4,999,499 last year.
- The donations we raised were split between different cause areas as follows: global health & development (40%), long-term future (32%), animal welfare (23%), and other causes (6%).
- Last year’s efforts of exploring fundraising projects in the blockchain community and among German-speaking philanthropists did not result in any major donations and we have stopped them.
- Our fundraising projects have had continual growth over the past few years, with money raised increasing even faster than the associated expenses. From 2017 to 2018 donations that would not have been made otherwise increased by $977,824 (21.4%). Expenses even decreased from $289,618 to $171,263 because we invested considerably less effort into Raising for Effective Giving.
- Going forward, we intend to maintain our philanthropic activities at the current level.
Review of 2018
We pursue two projects that raise money for other high-impact charities: Raising for Effective Giving (REG) and a regranting service. REG is a fundraising project aimed at the professional poker community, high-stakes players in particular. With our regranting service, we enable donors to take advantage of tax deductions. Because we are a charitable organization in Germany, Switzerland, and the Netherlands, donors in these countries can make tax-deductible contributions to high-impact charities which do not have charitable status in the home countries of the donors. They make a tax-deductible donation to us, the Effective Altruism Foundation, and we, in turn, regrant their donations to highly effective charities. To this end, we processed 5,867 individual donations last year. These donations are separate from any donations we processed as part of our project Raising for Effective Giving.
In 2018, we registered a total of $9,968,506 in donations for other high-impact charities due to these two projects. Conservatively, we estimate that $5,545,762 (56%) of that would not have been donated otherwise. Our estimate only accounts for some effects and might be biased in various ways. This figure represents our best guess. Rethink Priorities has produced a more detailed model [EA · GW] for similar work of RC Forward and we might adopt some parts of their model for future estimates. We estimate the combined expenses for these two projects to be about $171,263. That means for every $1 spent on these projects we raised another $34. The net impact were donations worth $5,374,499 (= $5,545,762 – $171,263). Factoring in estimated staff opportunity costs of $375,000 (assuming some well-paid earning to give opportunity), the net impact goes down to about $4,999,499 (= $5,545,762 – $171,263 – $375,000).
Most of these funds was raised through our work with Raising for Effective Giving.
Last year, we also invested about 100 hours each into testing various fundraising efforts in the blockchain community and in the community of German (would-be) philanthropists. We ultimately decided against pursuing this further. The main reasons were:
- As an organization, we concluded that research into potential worst-case risks from AI is comparatively more neglected and that we have a comparative advantage at pursuing this type of research.
- S-risks from AI as a cause seem particularly ill-suited for fundraising outside of the EA community.
- Funding has become less of a bottleneck for the EA community as a whole, and for EAF in particular.
I want to thank Alfredo Parra, Kim Korte, and Thomas Moispointner for their help over the past year with all of these projects. Without their efforts, this would not have been possible.
Raising for Effective Giving
With Raising for Effective Giving, we accounted for counterfactual giving in the following way: We counted donations made through our main website as influenced by us. We sometimes also did so when donors gave to our recommended charities directly. In the latter case, we included the donation if the donor confirmed to us that our influence was essential. This metric leaves out any donations to our recommended charities from people who never get in touch with us (also see our FAQ on this question). However, this also assumes that the donors would not have learned about effective altruism or effective giving otherwise. Some likely would have. We have not attempted to account for either of these effects but given the dominant role REG has played in the charitable giving of the poker community, both strike us as fairly minor.
In 2018, most donations came from the Double Up Drive we helped run in December. We registered total donations of $5,437,174 from this campaign. In the end, however, we counted only $3,251,377 (60%) as having been influenced by us. As a first step, we counted all the matching funds that had been provided ($2,718,587) because the matchers learned about effective altruism through Raising for Effective Giving. However, the Double Up Drive had also been advertised by the charities themselves. So it’s very likely that some people contributed to the matching challenge who would have made a donation in any case. For all donors ≥$10,000, we individually estimated the share that they would have donated anyways by looking at their donation history as well as inquiring with the donor in question and the charity they gave to. We conservatively assumed that only 10% was due to the matching challenge if we had no information whatsoever. The resulting average was 15%. From this set of donors, we extrapolated to all the remaining ones (<$10,000) we had no prior information on by using this average.
Since these donations are not the result of a dedicated fundraising effort, it’s likely that a significant portion of the total would have been donated in any case, even if we had not provided this service. We used a simple model to account for this counterfactual giving: While donors would not have been able to deduct the donations from their taxable income (or take advantage from similar tax schemes), it’s unlikely that this would have caused them to donate nothing whatsoever. However, they would donate less without the deduction since their taxable income increases, which in turn increases their taxes. This effectively represents a price increase for the donation. This increase is determined by the effective marginal tax rate, which in turn depends on the income of the donor and the size of the donation. The relationship between price increase and the corresponding change in quantity is governed by price elasticity (PE). For simplicity, we’ll assume a price elasticity of –1, i.e. unitary elasticity. We’re using the formula for arc elasticity which takes the price elasticity at the midpoint between the two relevant values on the demand curve. We are not very confident that this is the right way to approach this effect but didn’t want to invest additional time into checking alternative methods in detail. If others in the community have a stronger grasp of these issues, we’d appreciate all feedback. The basic formula is as follows:
PE = ((Q2 – Q1) / ((Q1 + Q2) / 2)) / ((P2 – P1) / ((P1 + P2) / 2))
In our case, Q2 is the donation volume as we observe it right now. Q1 is the donation volume we would have observed if donations were not tax-deductible, i.e., the counterfactual donation volume. P1 is equal to Q1 as the cost of these donations are equal to their volume: making a $100 donation simply costs $100 when donations are not tax-deductible. P2 is the cost of the donations as we observe them right now, i.e., when they’re tax deductible. This is equal to the cost of making the donations minus any tax savings from deducting the donations from the taxable income. The tax savings are equal to the donation volume times the relevant marginal tax rate: Making a $100 donation reduces the taxable income by $100, which in turn saves taxes equal to $100 times the relevant marginal tax rate. So we arrive at:
–1 = ((A – C) / ((C + A) / 2)) / (((A – A * MTR) – C ) / ((C + (A – A * MTR) ) / 2)),
where C equals the counterfactual donation volume (= Q1 = P1), A equals the actual donation volume (= Q2), and MTR equals the marginal tax rate.
Solving for C, we arrive at:
C = A * sqrt(–MTR + 1)
Our impact (I) then is equal to the actual donation volume (A) minus the counterfactual donation volume (C):
I = A – A * sqrt(–MTR + 1) = A * (1 – sqrt(–MTR + 1))
The effective marginal tax rate differs from country to country, and from state to state in some cases. In Germany, the median income is about €34,000 (~$38,000) per year which would put the marginal tax rate at 33% (single household, no children). In Switzerland, the median income is about CHF 78,000 (~$78,000) per year. Unfortunately, income taxes are determined locally to a significant extent. A rough estimate gives a marginal tax rate of 17%. In the Netherlands, the median income is about €36,000 (~$40,000) per year which would put the marginal tax rate at 38.1% (single household, no children). We rounded down to 38%. We’ll assume 0% for any other countries since donations to EAF are not tax-deductible there.
Using these numbers, we arrive at ~$385,609 (15.5%) as having being donated due to us providing this service. This likely underestimates the impact: From what we can tell, the studies on the price elasticity of donations were mainly done on changes to the tax code, not changes to the charitable status of individual charities. In the latter case, we’d expect significant substitution effects because donors decide to support other charities. In this case, they might donate to considerably less effective charities which are, however, tax deductible in their respective countries. There are also some other effects that Rethink Priorities has looked into for their study on the impact of RC Forward [EA · GW].
Expense and opportunity cost estimates
We estimate the 2018 expenses for both projects to be $171,263. We can only estimate this number because the staff members involved in these projects also have other responsibilities within the Effective Altruism Foundation. We also stopped accounting for expenses on a project-level in 2018 since the Effective Altruism Foundation has become a lot more integrated. So expenses cannot be neatly attributed to one project or another. Our estimate is based on figures from previous years and estimates of the relevant staff members on how much time they spent on the projects in question.
Dividing these expenses between REG and our regranting service is also a challenge. We did so by using a weighted formula that takes into account the respective donation count and donation volume. This yielded a 40/60 split between REG and our regranting service: $68,505 and $102,758 respectively.
To account for opportunity costs, we took the estimated 1.5 full-time equivalents dedicated to these two projects in 2018 and multiplied them with $250,000, a crude estimate for a very well-paid earning to give opportunity. This results in an opportunity cost estimate of $375,000. Assuming the same split between projects (40/60), we get $150,000 for REG and $225,000 for our regranting service.
By cause area
Development over time
Our fundraising projects have had continual growth over the past few years, with money raised increasing even faster than the associated expenses. Last year we even saw a decrease in expenses because we invested considerably less staff time into Raising for Effective Giving. We didn’t try to find many new donors and focused on maintaining the most important existing relationships instead.
Some clarifying endnotes on this table.
Going forward, we intend to maintain our philanthropic activities at the current level. We will continue to provide a donation infrastructure for German, Swiss, and Dutch donors. We will maintain Raising for Effective Giving, focusing on the most important relationships and opportunities.
As we wrote on this forum in December [EA · GW], we will focus much more on research compared to previous years. For instance, we hosted a workshop on s-risks in Berlin in March and another one in the Bay Area in May.
Since we are not a fiduciary, we cannot legally commit to forwarding any donations to the desired charities. However, we very strongly take into account the wishes of the donors when regranting donations, and have always done so in the past. ↩︎
After briefly reviewing the evidence on this, this seems about right. The only meta-analysis we could find gives a mean elasticity of –1.44 (standard deviation of 1.21). When the meta-analysis excluded outliers, they found an elasticity of –1.11. However, this report seems to suggest that such a high elasticity likely overstates the real effects. ↩︎
This includes time spent exploring fundraising efforts in the blockchain community and in the community of German (would-be) philanthropists. ↩︎
The numbers for REG differ slightly from the ones in the respective annual reports for 2014 and 2015. That’s because we decided to retroactively exclude donations made to either REG or EAF during those years. We also decided not to include estimates for opportunity costs since for previous years, we don’t have reliable estimates for full-time equivalents dedicated to these projects. We estimated the counterfactuals for regranting donations by using the same marginal tax rates estimates as for 2018 and adjusting based on the donation volume per country for that year. ↩︎
Comments sorted by top scores.