[Epistemic Status: Fairly confident under the assumptions provided below]
TL;DR: Many individual EA donors should not donate throughout the year (e.g., at each paycheck). Instead, they should probably regularly invest those funds (e.g., in an index fund) and donate the money therein to a matched Facebook fundraiser on Giving Tuesday. The expected value of a counterfactually matched donation outweighs the discounting of future donations, even if saved for a whole year. This implies many EAs are making a mistake by donating throughout the year instead of saving all donations for Giving Tuesday.
How Much to Discount?
I assume that the proper discounting rate for present-vs.-future donations is 16%/yr., as that was the median response in the latest 80,000 Hours survey. In reality, this is probably too high. If so, my conclusions are even stronger.
EAs can hedge against this by investing donations in an index fund.* I use and recommend Vanguard as a brokerage firm because they have no transaction fees for buying and selling nearly all U.S. exchange-traded funds. I also recommend using Vanguard ETFs and mutual funds because they essentially operate at cost. For the sake of this analysis, I assume you invest money you would otherwise presently donate in their international small-cap index ETF, VSS.† VSS has historically earned annual returns of 9.86%.
Of course, if you think that giving later is generally better, then this whole post is moot. See this post for sources on this question.
Probability of Matching on Giving Tuesday '19
For my analysis, I assumed (conservatively, I think) that a donor has a 30% chance of getting her donation matched by Facebook on 2019. This year, 65% of donations were matched. However, my subjective probability of getting a donation matched on GT 2019 is substantially lower due to:
1. Uncertainty about whether Facebook will run another Giving Tuesday;
2. If so, what the details will be (e.g., individual matching limitations);
3. Increased competition for matched donations;
4. Regression to the mean.
Note that my conclusion holds even if this probability is as low as 11%.
EAs Should Invest-to-Give, Then Donate on Giving Tuesday
Under the above assumptions about discounting and matching, according to my calculations, EAs should:
1. Invest their donation money all year in an index fund;
2. Donate all that money in matched EA fundraiser on Giving Tuesday. Join this Facebook group to coordinate for this.
This holds even for money a year in advance, which, using the above numbers, has an expected value of 118% its nominal value using this strategy.
Obviously this only applies to donors who can donate through the Facebook Giving Tuesday platform. Join the aforementioned Facebook group for more discussion of this.
If you disagree with my numerical assumptions—or have special considerations not captured above—I encourage you to download my spreadsheet and adjust them on your own.
*If you are risk-averse in your donations, then you should choose a lower-risk, lower-yield option for your savings. Brendon Wong lists recommendations for such savings amounts here, and reasons one might want to be somewhat risk-averse in donations here. Even if you assume 2% annual yields, it makes sense to save all year then give on GT.
† I originally recommended a large-cap ETF. However, small cap ETFs generally get better returns on investment (with higher associated risk). So, in the interest of promoting and endorsing risk neutrality, I'm recommending VSS.
I am currently more than 30% confident that I will get at least one donation matched by Facebook on Giving Tuesday 2019. This is not conditioned on Facebook doing another match this year or on anything else.
I in fact got at least one donation matched by Facebook on Giving Tuesday 2019. (And 2020.)
To automatically invest money each paycheck so that you don't have to remember, mutual funds are another option besides ETFs (it's not always possible to automatically invest in an ETF). Mutual funds have higher account minimum balances, though.
Notes on this, from someone who was fairly involved in the GT process:
It seems valuable for someone to write up a more detailed document on timing considerations: "give now or give later" is a popular question, but often implies giving many years later; "when to give in the next 12 months" is very different.
One more thing which seems important: There are other ways to optimize a donation besides timing! Once you know how much you'll give, and where, you have many options for how to share that information; you can write about it, post on social media, set up your own "match" for friends (make it truly counterfactual, and try to discourage EA people from using up matching funds that might instead attract non-EA people)
Great points. I agree re Double Up Drive that it was worthy of much deeper investigation to figure out the counterfactual nature of it. Perhaps there was even a way donors could have *made it* more counterfactual via doing a great job using their donation as an opportunity to signal and influence others' behaviors. I briefly considered donating to it just so that I could write a message to the donors who were offering the match encouraging them to donate the full amount regardless of whether or not the match was reached and have my message possibly be heard.
More generally one thing I have updated a lot on after this past giving season is that I now believe that for small donors the signaling value of their donations matter a lot more. For example, a GWWC member earning $50k/year and donating $5k/year has a certain amount of credibility that conceivably could be used to help influence much larger donors to donate more and more effectively. How one's donation and one's communications around one's donation is going to be perceived by much larger donations may in fact count for more than the value of one's donation itself.
Minor point: I don't think that regression to the mean is a reason to expect the probability of an EA's donation being matched in 2019 to be less than the probability of it getting matched in 2018.
Just to flag that the case for this is much much weaker outside the USA.
The matching limits for donations outside the US is much lower and you may also lose your tax benefits of donating.
I recommend modifying this:
To something like this:
Done, thanks :-)
I think there are two big benefits of donating monthly and automatically as opposed to on Giving Tuesday:
1) you can't forget to do it.
2) you don't see a big amount pile up during the year and then on Giving Tuesday think "am I really going to give that much away??"
Of course if you know you won't fall prey to either of these scenarios then it makes more sense to do as you describe.
Small-cap stocks and emerging market stocks generally perform better than large-cap ones (with higher variance). In the interest of promoting risk-neutrality regarding donations, I'm updating this post and the associated calculations to recommend either VSS (international small-cap index) or VWO (emerging markets).
Thanks for writing this up. This is what I did in 2018 and I have no regrets. It clearly seemed to me to be the right thing to do after I saw my Giving Tuesday 2017 donations get matched. I considered writing a similar post to this one after Giving Tuesday 2017 to encourage others to hold off on donating until Giving Tuesday 2018, but I don't think I ever did due to (if I recall correctly):
(1) a worry that I would dissuade people from donating and worried that some of those people would suffer from value drift and no longer want to donate at all come Giving Tuesday, and
(2) I wasn't sure if I had sufficient evidence to give others confidence that come Giving Tuesday 2018 they'd be able to get their donations matched.
This year for a US donor who is not concerned about value drift being a significant risk I agree with Cullen that this is probably a great bet again.
And note that you can adjust for the probability of value drift in the spreadsheet :-)
FB had a limit of $20k/donor this year, and I think that's much more likely to go down than up. So depending how much you're donating there's not much reason to save more than than for Giving Tuesday.
There's also the 1% PayPal match (plus 2% cash back) that's been in December each year. At a 16%/year discount rate it's worth waiting a couple months for that 3% but not all year.
Thanks Jeff. I agree.
Note that, strictly speaking, the formula is the value of a given dollar if invested until GT. So, if we're referring to the 20,001st dollar, then we should probably adjust the probability of it getting matched down, way below my 30%. (Due to the time it takes to make several donations on Giving Tuesday morning, that probability should curve down before that anyway). But if you still assign it a probability of 12%, it's worth doing now.
My personal credence that donations above $20K would be matched is probably much lower than this for the reasons you identify, though.
It's also notable that PayPal announced that they were doing a dollar-for-dollar match of up to $500k in donations through the PayPal Giving Fund on Giving Tuesday, separately from the Facebook match.
That post was made on Giving Tuesday itself and the details there are very limited, but I found this post by an organization with a date of Nov 23, along with the match start and end time. I'm going to dig into this some more. Maybe we can ask some EA-aligned organizations signed up for the PayPal Giving Fund if they can keep an eye out for this and give us an early tipoff.
There's also other counterfactual matching opportunities that tend to arise around the same time though.
How did Facebook's $7 million match in 2019 compare with the probabilities modeled?
I see two reasons to adjust the benefit of this practice downward:
1) Some of Facebook's potential match would otherwise have nonzero benefit, and some (small) fraction would go to EA causes.
2) (More importantly) You can't donate stock through the Facebook Fundraiser platform. Donating stock (or index funds) allows you to not incur capital gains taxes, which for most people would be about 15% LTCG tax * 10% return = 1.5% per year.
What was the largest amount that any individual got matched on GT? Given that this year there were only 15 seconds of matching funds, can one person get through enough forms in time to give a lot?
A number of people got $20k matched out of $20k donations. This required 8 donations of $2500 each.
I got the whole $20k: https://www.jefftk.com/p/facebook-donation-match
Do people who try to give huge amounts run the risk of the transaction being rejected by their bank (as fraudulent), and then not giving in time to be matched?
Yes, this was a problem in both 2017 and 2018, and our 2018 match percent would have been higher than 65% without these problems. I think Cullen's estimates account for this though.
In 2018 we conducted a survey try to to understand this better. We should be able to use some of the results to make better recommendations in 2019.