My donation budget and fallback donation allocation

In this (fairly long!) post, I describe how I'm currently thinking of my donation budget, and how I distinguish between opportunistic, high-marginal-value donations and fallback donations. If you're viewing this on the EA Forum, please use the table of contents (that you should see to the left of the post) to navigate the post and understand its overall structure.

Some of this goes over ground I previously tread on in my previous posts on my Q4 2021 donation and my Q1 2022 donation. However, this post is fully focused on my overall donation strategy rather than simply describing it in the context of specific donations.

I've written a few other blog posts in the past few years. Both the act of writing the posts and the comments that some of the posts have received have informed my thinking. These posts are:

• Donor strategies for separating "how much" from "where" to donate
• Levels of donations
• What regrets have you had regarding past donations?

How I see the value of various things I do

In general, I see the main vehicles through which I create value as follows:

1. Living my "regular" life, including doing my day job, interacting with family and friends, etc.
2. Using my time and money for personal projects with some altruistic bent, such as the donations list website or timelines wiki
3. Using my money for donations.

I expect that my value-add is roughly (1) > (2) > (3), though I expect that the value-add-per-unit-time may be higher for (2) and (3) than (1).

Given this overall view, I want to set rules on my donations in a way that they don't undercut (1) and (2). This informs the budget that I set for donations. I'll talk about the budget in much more detail later in this post.

Other related things I've written:

• In the "Sources of value" section of my post on handling stress, I talked about sources of purpose and pleasure. That is a bit more detailed and more about my moment-to-moment experience, but it has some overlap with how I see the overall value of things I do.

• In a post I wrote on three-tiered ethics on my old blog, I separated out three tiers of ethics: negative rights ethics, contract/responsibility ethics, and excellence ethics. Earlier tiers of ethics take precedence over later tiers of ethics; for instance, if my contract requires me to kill somebody, then I would rather break the contract than kill. In my list above, (1) falls under contract/responsibility ethics, whereas (2) and (3) fall mostly under excellence ethics (though to some extent under contract/responsibility ethics to the extent that I've made commitments).

Savings thresholds and donation budgets

The extreme view of "save till you no longer need to save, then donate all net savings above the threshold"

One view, that I'm fairly sympathetic to, is the view where you basically switch gears in life between saving and donation. To begin with, your goal is to save enough money to build a nest egg. Once you have achieved that, you switch to donating all your surplus income (i.e., income - consumption).

I feel like this view is, at least in the abstract, more logically sound than splitting your money between savings and donation. It's more risk-averse, in the sense that if financial challenges prevent you from hitting your savings goals, you conservatively don't donate and save whatever you can. But it's also generous beyond the point of hitting your savings goals.

I don't ultimately go with this extreme view, for various reasons I articulate, but I think it should be something for people to consider.

The other extreme view of treating donations as a fixed proportion of income or savings

In this view, you have a fixed rule around what proportion of your income or savings you donate periodically (for instance, every year). I talked about some of the predetermined rules people use for such things in my post Donor strategies for separating "how much" from "where" to donate.

I think this is a reasonably good strategy with its own advantages, but it's already relatively well-known so I won't say much more.

Finding a middle ground of sorts

Overall, I've picked a middle ground that combines some features from both extremes. I have in mind a few different savings thresholds:

1. Emergency savings threshold: This corresponds to something like 6-12 months of expenses. Before hitting this savings threshold, I would generally not donate (though, historically, I did donate while below the threshold, I think this may have been the wrong choice as I did face a shortfall of cash that was partly caused by the donations).

2. Money-won't-be-a-bottleneck-to-life-changes threshold: Basically this is a threshold of saving at which I can make changes to my life without being constrained on cash. For instance, I can finance a move to a new city, or make a downpayment on a house if I determine that buying is better in the long run than renting, or quit my job and try something out for a year or two. I don't have an exact threshold here but something like 2-5 years of savings seems about right for this.

3. I-have-enough-money-for-life threshold: This is a threshold where I basically have enough money to finance the rest of my foreseeable life, after making appropriate assumptions about interest rates. This is the most uncertain, particularly because it requires making predictions about the nature of the medium-term future. But if we don't believe in short timelines for very disruptive changes to the world, I think something like 10-30 times current annual spend is probably enough for this. The presence of various safety nets and cost-cutting measures can affect the exact calculation here.

My current situation is that I'm clearly past 1 and moderately past 2, but still well short of 3. When I was below 2, making donations traded off against having the flexibility to make life changes in a relatively tangible way, and I was therefore extremely selective about making donations.

Now that I'm past 2, making donations (that are small enough that I stay past 2) doesn't really trade off against life changes.

However, I'm still well short of 3. If I were above 3, I would likely adopt the extreme position of "donate all net savings" at least in principle (though various practical considerations may still require me to hold money while I figure out good donation targets). But I'm not above 3, which means that I'm still trying to grow my savings. The question then is: what sort of rule do I use to determine the amount of money I can donate?

Haste considerations

The main reason I want to donate now rather than wait till I'm at or above 3 is haste considerations: I think there are a lot of compelling opportunities that exist right now. Crudely, at least for small amounts of donations, my sense is that the social rate of return to improving the world right now is greater than the private rate of return (in terms of interest accrued, as well as money-as-cushion) of keeping it.

Picking a donation budget

As I documented in my Q4 2021 donation post, I set a donation target of $1,000 per month starting July 2021. The budget is cumulative, and I don't have to spend it, so whatever I don't spend accrues, allowing me to save within the budget for bigger donations.

As of October 2022, I've been in this regime for 16 months, accumulating a budget of $16,000. I've spent $4,000 so far, so I have $12,000 to donate.

How did I pick this donation budget? Essentially, I wanted to pick a value that was both big enough to be worth the overhead of thinking about it, and small enough that it didn't interfere with my regular life or personal projects, which I believe to ultimately be more valuable.

The "not too big" angle

• What kind of variation in net savings would I consider small enough to not worry about? If my donation budget is within that range of variation, it's probably okay.

• What kind of variation in after-tax income would likely not be decision-relevant to choosing to join or stay in a job? If my donation budget is within that range of variation, it's probably okay.

The "not too small" angle

I wanted a donation budget that feels substantive and gives me "room to play" with making serious donations. I thought about this from a few angles:

• How does it compare with my other budget line items? At $1,000 per month, or $12,000 per year, my donation budget is only exceeded by two line items: rent and contract work. My donation budget is in excess of my other nontrivial line items such as food, health insurance, web hosting, and health care.

• What level of donation does it place me at? Assuming I donate about once a year, it puts me at level 2, which is a high enough level that I can do some meaningful investigation for each donation, and where there starts being a meaningful probability of meeting highly specific funding gaps.

How should the donation budget evolve over time?

If my savings continue to increase, slowly and steadily, should I increase the proportion of money I allocate to my donation budget, thereby slowing down the rate at which I accumulate money? I'm fairly unsure about the right approach here. However, for reasons of simplicity, I've decided against making continuous changes to the rate at which I add to the donation budget. In other words, I'll continue to add $1,000/month to the donation budget for a while, until I revisit things (either due to life changes or due to a reassessment of where I am with respect to savings).

Windfall profits and the donation budget

My income and spending rate define the backdrop of my donation budget decision. What if I have a windfall profit? This is not merely a theoretical question -- I've had significant fluctuations in income due to cryptocurrency price changes, and I expect to make a sudden amount of money as a result of exercising and selling stock options at the tech company I recently left.

My current thinking is to not let windfall profits affect the donation budget, at least for profits that comprise less than one additional year of net saving.

Borrowing against future donation budget

What if I find a truly compelling and time-sensitive opportunity, and I want to donate more to it than my donation budget allows? In extreme cases, I'm open to the idea of borrowing against my future donation budget. I consider this fairly unlikely in the near term, partly because I've already accumulated a fair amount in the donation budget.

Minimum donation targets and fallback donations

When I initially picked a donation budget, it only set an upper bound, with no expectation that I'd actually make donations at any specific rate. At the time, the concept of a donation budget was relatively new, and I wasn't sure what sort of donation opportunities I would be interested in.

However, now (October 2022) I want to also set some rule to make sure I do spend at least some of my donation budget, while still allowing for accumulation to finance potentially larger donations in the future.

The original concept of fallback donation targets

In my Q4 2021 donation post, I talked about fallback donation targets. I wrote:

There are a number of "fallback" donation targets including the EA Funds and the GiveWell Maximum Impact Fund. If I had enough savings that I was actively looking to give it away (and here I am thinking of something in the range of millions of dollars, which would be enough to last me a lifetime and then some), my strategy would be something like this:

1. Proactively investigate donation opportunities that are likely to beat the fallback donation targets.
2. After exhausting those (either not finding any within a predetermined time limit, or meeting their funding gaps), donate the rest of the money to the fallback donation targets.

I am currently not at the level of savings where I'm actively looking to give it away. So, in particular, I won't do step 2; I will limit my donation efforts to step 1. And even within step 1, my investigation's "proactive" nature will be limited, because giving money away is not a priority. Moreover, I will give more priority to relatively time-sensitive donation opportunities, and not focus on other donation opportunities that I can revisit later when I have more money and more information. So it really boils down to just:

Investigate time-sensitive donation opportunities that are likely to beat the fallback donation targets, and make donations to them if I am moderately convinced of them.

So far, I've mostly followed the above, and the only two donations I've made since adopting a donation budget have been to things that I deemed time-sensitive and that, at least at the margin at which I donated, and based on my values and the information available to me at the time of donation, seemed in my mind to beat fallback donation targets.

Downward update on existence of compelling and time-sensitive opportunities and my time availability to investigate them

One update over the last year has been in terms of the existence of compelling and time-sensitive donation opportunities, as well as my ability to investigate them to a level of depth where I feel comfortable donating to them. Specifically, my rough impression now is something like this:

• There aren't that many compelling and time-sensitive opportunities that I am well-positioned to fund. Existing funding infrastructure from other donors, who have both more funds and more investigative resources, seems to cover many of these opportunities. And even if there's an opportunity that I initially find compelling, the fact that these other donors are passing on the opportunity would likely make me want to investigate it more.

• My time availability to investigate opportunities is not reliably high (and in particular it may not be high during the windows when the time-sensitivity exists). My current estimate is that I most likely don't expect to have enough confidence to allocate more than $5,000 per year to compelling and time-sensitive opportunities, and I might very well allocate much less.

Resultant desire to "get some money out the door" by donating to fallback donation targets

Given the above factors, I've been leaning more toward making donations to fallback donation targets (such as the EA Funds or the GiveWell Maximum Impact Fund). My listed fallback donation targets all have several (related) advantages:

• They are regrantors rather than grant endpoints, and they have a reasonably thoughtful and rigorous approach to identifying endpoint donees. This derisks things quite a bit; in general, I expect them to have pretty large room for more funding and to make reasonably good decisions.

• They are more legible and therefore give me a stronger sense of having "done good" with my donation, as well as communicate to others more clearly that I'm doing good. They offer a credible signal of my "effective altruist" sympathies.

• My incremental investigative effort is limited to choosing between very high-level things, rather than getting into specific operational questions. To wit, it's something like, what matters more at the margin: directing funding to animal welfare, directing funding to the most cost-effective programs in global health, or directing funding to AI safety / biorisk / longtermist areas? While this is not a super-easy choice (I think all of these have merits) it is something where I can give my current best guess based on fairly broad considerations. It therefore needs less investigative effort. And even if I choose something and change my mind later, the "damage done" seems relatively small.

A more general point is that, based on responses to my regrets post, it seems that people (in the reference class that I belong to) don't generally have significant regrets about having donated too much or left too little money for other donations or non-charitable uses. Obviously, I want to build in the right set of safeguards for my own use case, but I should probably expect less regret from my future self than my current self feels in anticipation.

Trading off between having money available for compelling and time-sensitive opportunities versus getting money out the door

The trade-off feels real to me: do I keep more money (within my donation budget) for potential compelling opportunities, or do I get money out the door?

Overall, I think that if I were able to find compelling and time-sensitive opportunities that use up my donation budget, I'm happy to spend on those and not donate to fallback donation targets at all (even though this does have some psychic costs that you can infer by reversing the benefits of fallback donation targets listed above). This is primarily an intuition that thoughtful iconoclasm is more valuable than thoughtlessly following the herd, even though iconoclasm is more high-variance.

However, I don't expect to be able to find enough compelling and time-sensitive opportunities that use up my donation budget, and I want to get money out the door soon. So how do I square this?

My solution for now: minimum fallback donation target assuming enough donation budget

The idea I had is to do a periodic (e.g., end-of-year) evaluation as follows. If by the end of the year, I have $10,000 or more unallocated in my donation budget, I will allocate $5,000 to a single fallback donation target. The target can vary by year, but in any given year it should represent my best judgment among the pool of options (the EA Funds and the GiveWell Maximum Impact Fund).

Here's how I justify this solution.

"Small enough" for the donation target

All the listed fallback donation targets have annual spends ranging from the hundreds of thousands to hundreds of millions. Based on their processes, it seems that none of them would see any meaningful changes in their marginal value function with a $5,000 donation. So, I don't have to worry too much about donation dynamics and room for more funding.

"Big enough" for the donation target

$5,000 is big enough for the donation target that I don't feel like I'm wasting their time. It falls at level 2 in my levels of donation post, which is enough for the donee to collaborate with me on administrative matters, and for the donation to very clearly be worth more than any administrative overhead incurred by either me or the donee.

"Small enough" for me

There are two aspects of this:

• Since this allocation is done only if I have at least $10,000 accrued in my donation budget, it pretty much never causes me to run out of money to donate for compelling and time-sensitive opportunities. In the worst case, I'll end up having $5,000 left over in the budget after this donation. Given the rate at which I've been finding compelling and time-sensitive opportunities so far, this should be more than enough.

• It's pretty much guaranteed that the fallback donation targets won't account for more than 5/12 of my total donation budget, which seems aligned with my overall intuition around how to drive the most value. And as I already articulated, my donation budget as a whole is small enough as to not interfere with the other things I want to do in life.

"Big enough" for me

$5,000 per year is comparable to my annual expense on food. It's comparable to my annual expense on health insurance, and exceeds my annual expense in most categories (rent and contract work are the only categories where I spend more). As such, it's a sizable amount that reflects the importance of donations as a way to do good.

Donation bunching

I might want to engage in donation bunching for tax considerations. For this reason, if I haven't made any other donations at all in a given tax year, I might choose to defer the fallback donation to the beginning of the next tax year, so there's more potential for bunching. In my current jurisdiction, the tax year matches the calendar year, so basically this would mean donating in January of the next year.

When should I start applying fallback donation targets?

I'm currently undecided between whether to start applying fallback donation targets starting with end-of-2022 or wait till end-of-2023.

The argument for waiting till 2023

I had originally planned to spend a significant amount of time in 2022 on the donations list website as well as other projects that would give me more clarity on what's going on in the philanthropy space. I expected this to inform my worldview, both in terms of finding compelling and time-sensitive donation opportunities and in terms of choosing more wisely between fallback donation targets.

However, 2022 was a year of several other changes for me, including a job change and moving house. In addition to job transition, I'm also spending time this year on some long-overdue devops work related to the servers hosting my websites, and my current plan is to focus on this devops work for the rest of 2022. So I don't expect to spend time on the donations list website until 2023, or to spend significant time thinking about where to donate (beyond the time spent writing this blog post and possibly a successor blog post).

Given these considerations, if I wait till the end of 2023, I'll have more information and I'll have had more time to explore. I'll also have more money to play with, and various parts of the ecosystem would have matured more (for instance, there will be more clarity on the role that the FTX Future Fund will play). All of these make the case for waiting till the end of 2023.

The argument for starting in 2022

There are a few reasons I want to start in 2022:

• In general, the argument for doing this is also an argument for doing this now. Much of the argument is not sensitive to the exact start date, and it makes sense to start sooner rather than later.

• I already made a donation in January of 2022 that just about gets me past the standard deduction, so I get donation bunching benefits by starting in 2022.

• I expect to make a windfall of money related to exercising and selling stock options at the tech company I left. While this isn't significant enough to change my donation budget (for reasons discussed earlier) it does seem like it'll address any liquid cash considerations that arise from starting in 2022. It's well in excess (even after taxes) of the $5,000 that I might spend this year if I started this year.

Overall leaning

Overall, I'm leaning toward starting in 2022, but I plan to continue to think about this through October and November to see if other considerations come up that I hadn't thought of before.

Selecting between fallback donation targets

I am not making a decision right now, but when I do, I will likely pick between one of the four EA Funds:

• Global Health and Development Fund
• Animal Welfare Fund
• Long-Term Future Fund
• EA Infrastructure Fund

While I was initially thinking of the GiveWell Maximum Impact Fund / GiveWell Top Charities Fund, it seems like the Global Health and Development Fund is strictly better, to the extent that it rolls money to GiveWell top charities if it doesn't find better opportunities.

I would have liked to have the donations list website up-to-date before this analysis!

It would have been good if the donations list website included up-to-date information on the EA Funds before I did this analysis, so I could review considerations more comprehensively. For now, I'm relying on the fund pages on the EA Funds site (as linked above).

Current preference order

My current weakly held preference order is something like this:

Animal Welfare Fund > Long-Term Future Fund > EA Infrastructure Fund >> Global Health and Development Fund

Why I rank global health and development last: not neglected!

The Global Health and Development Fund seems to be fairly well-funded; for instance, in 2022 the fund has already allocated a little over $10 million.

Moreover, it funges with GiveWell's funds, that raise tens of millions of dollars a year. The purpose of the Global Health and Development Fund is to allocate money to high-risk, speculative things in global health and development first, and then allocate the remaining money to GiveWell top charities. But it looks like the 95%+ of the 2022 allocation so far has been to GiveWell top charities. The situation was a bit different in 2021, but overall it looks like, at least based on the most recent information, the fund doesn't have a consistent and compelling stream of opportunities different from GiveWell top charities.

Overall, I think GiveWell top charities present a moderately compelling case, but this cuts both ways: the case is compelling to a wider range of people. In general, helping "humans" "now" seems like an easier sell than other things, and GiveWell has created the (probably correct) perception of having identified rigorous programs that deliver on the promise of cost-effectively helping humans now. GiveWell has both historical success and the prospect of raising more money. Open Philanthropy alone has been upping its allocation to GiveWell top charities.

While I do think that what GiveWell is doing is very valuable, it isn't neglected relative to the other more niche cause areas covered by the other funds, and given my set of values, it is pretty unlikely to be close to the most cost-effective.

Comparing animal welfare, the long-term future, and EA infrastructure

I think animal welfare matters, and I think the long-term future matters. I think society as a whole, and probably even the effective altruist penumbra (people familiar with EA-ish stuff in general) probably undervalue both these areas relative to their importance. I think that the EA core has zeroed in effectively on these areas, since at least Luke's 2013 post that identified the four focus areas that would eventually each get their own fund. And through what might seem to be circular reasoning, EA's ability to identify animal welfare and the long-term future itself vindicates EA to some extent, making EA infrastructure valuable to invest in.

All three areas seem important and neglected. All of them have challenges with measuring tractability. My current thinking is as follows:

• The long-term future, viewed abstractly, is probably the most important. Specific things like AI safety and biorisk also seem fairly important, even if you only partly buy the cases made by people focusing on those areas.

• Animal welfare seems the most neglected, though it's hard to say (again, I was hoping to do more work on the donations list website in a way that would inform this).

  • The existence of players such as FTX Future Fund and the Survival and Flourishing Fund, that focus mostly on the long-term future and somewhat on EA infrastructure, but basically don't spend at all on animal welfare, partly informs this view.

  • Even among the EA Funds, the Animal Welfare Fund seems to have the smallest budget.

• Animal welfare seems to naively be the most tractable, but it's not very clear to me how tractable it is at the margin. Here, specifically, the activities of the Animal Welfare Fund are informative: it's funded several small animal welfare organizations around the world, some of which have very small budgets, and might be playing an important role in nurturing the growth of animal welfare movements around the world.

Given all these considerations, I'm leaning toward treating animal welfare as the most high-value area to donate to. However, in terms of personal focus on the non-donation front (e.g., in terms of what to talk to people about, or what to read/learn about), I think stuff related to the long-term future, global catastrophic risks, and in particular AI safety and biosecurity seem more important. I also think that if the FTX Future Fund and the Survival and Flourishing Fund didn't exist, I might well have gone with the Long-Term Future Fund as my top choice.

I have more thoughts on this, including thoughts on how to justify spending resources on anything other than AI safety if imminent AI timelines are plausible. But I want to mull these further and make a final decision as of the time I actually want to allocate the $5,000.

How I plan to learn and improve

Regardless of whether I start the process of making fallback donations in 2022 or wait till 2023 to do so, I'm hoping to be in a better position in the coming years. Some of the ways I plan to improve my epistemics:

• I'll spend substantial effort on the donations list website in 2023, to a point where it can usefully inform my donation decisions and my understanding of the overall landscape.

• I'll continue to be involved (through financial support and guidance) in a number of other projects such as Org Watch and timelines wiki that each, in different ways, helps illuminate the landscape.

• I'll be following what's going on in places like Metaculus, the EA Forum, various EA-related podcasts, as well as other high-quality sources of discussion, to continue to build and improve my models of the world.

While I hope that all these will help drive better donation decisions for my personal donations, I see them as relevant in a larger sense: they help improve my overall model of the world, and at least in some cases (such as the ones where I'm working on public websites or documenting my thoughts in blog posts), could help improve others' model of the world as well.

Seeking your thoughts

I wrote this post mostly to record my own thoughts rather than to make a particular point for other people. But if you were patient enough to read enough of the post, I appreciate it and hope you found something interesting enough to be worth your while!

If you have thoughts on any parts of the post, I would appreciate them. These parts in particular are ones where I'm most interested in hearing thoughts, but don't limit yourself to just these!

• What do you think about my solution in terms of conditionally allocating a portion of my donation budget to fallback donation targets? Would you recommend a different approach that has most of the advantages of my approach, but that is even better on at least some counts?

• Are there crucial considerations I seem to be missing regarding how to decide between the EA Funds? I say "seem to be" because there's probably stuff in my head I didn't write down -- so I may not actually be missing them -- but I would still love to hear what you think is missing in the articulated portion of my thinking.

• Any thoughts on how soon I should get started with making donations to fallback donation targets? Should I start in 2022, or wait till 2023 when I expect to have built a more substantive picture?