What is a salary formula?

A salary formula is a set formula that calculates a salary for an employee using a variety of inputs such as:

• A candidate’s previous experience,

• an employee’s tenure at your organization,

• the title level of the role,

• the location of the role or cost of living in that location,

• the employee’s competence at their role,

• the market rates for certain kinds of roles,

or other factors that might be relevant to compensating them.

A salary formula can calculate a specific salary or provide a range for negotiation between the organization and the candidate.

Salary formulas provide a fairer method for setting employee compensation. Instead of negotiating salaries based on market rates or an employee’s ability to negotiate, organizations apply a formula to determine the exact salary offered to each employee. This approach aims to eliminate salary disparities that may arise from negotiations or other subjective influences.

Salary formulas also help control costs at an organization. If you know exactly what the salary will be for a role, you can better budget your costs into the future. They also reduce bias: they reduce the risk of paying different employees different amounts, including based on their gender or other factors.

The primary downside of using salary formulas is that they might not always produce competitive salaries compared to a potential employee’s counterfactual. When designing a formula, you have to make sure that its outputs will produce salaries that lead to the candidates you want to hire saying yes to your job offers.

Some large institutions, such as universities and governments, use salary tables instead of formulas. For example, the General Schedule (GS) system outlines the salary levels applied to federal government employees. Jobs are assigned specific GS levels, and as employees move up the system in steps, they receive raises over time. Similarly, universities might have salary tables or salary bands for different roles. A band allows negotiation but within a constrained range.

Downsides of using salary formulas

There are some specific downsides to using a salary formula:

• They force you to generate a specific salary, which, if you mess up your formula, might not be competitive for the roles you’re hiring for.

• Alternatively, a poorly designed formula might pay more than you wanted to pay for some roles, which you might end up regretting (because you later face budget constraints, etc.).

• If you deviate from the formula, it might exacerbate a feeling of unfairness, because everyone else is being paid by a fair system while the person you deviated for is not. If you’re going to deviate from the formula and break a promise of fairness, morale might be better at the organization if you never had a formula at all.

• There might be some level of lock-in. For example, you might feel trapped using the formula and unable to go back because it feels very fair to staff, even if you later realize that it isn’t working for your situation.

• Salary formulas tend to produce salaries within a pretty restricted range. This makes sense for a lot of nonprofits, which don’t tend to pay their highest-paid employees many times more than their lowest-paid employees.

  • For some organizations this restriction might be a problem, especially those hiring highly technical staff. The market rates for their roles might be hard to achieve using your formula without heavily distorting it or using a large market rate adjustment. But if you use a really large market rate adjustment, it maybe means your formula just isn’t doing most of the work to create the salary — the market rate adjustment is doing it instead.

Designing a salary formula

To make a salary formula, you first have to decide what you actually care about. You can then use these features as inputs into the formula. The inputs you choose also will be incentives for your staff. If you reward promotions, for example, with more pay, then staff will probably want promotions more than they naturally would. Or, if you reward tenure with increasing amounts of pay, then staff might be incentivized to stay at the organization longer.

Inputs can be combined at different weights into an overall formula. For example, you might care significantly more about an employee’s skill than their previous experience, but you do want to reward both. If so, you would weight skill at a much higher level than experience.

Potential inputs to a formula

• Base salary

  • Often formulas will have a base salary that is modified to achieve the output. Depending on how your formula is designed, you can also increase the base over time with inflation to account for cost of living increases. This isn’t strictly necessary, but a base salary can help make it easier to put together the overall formula later.

  • If your formula does use a base salary, you could imagine it as the lowest salary your organization could possibly pay, such as the salary of the lowest-paid junior employee. When designing, you can always use a placeholder number for the base salary and tune it later after seeing the outputs from the formula.

• Experience

  • Many employers care about an employee’s previous experience, so you might use it as an input into your formula.

  • You may want to define how you measure previous experience. For example, does someone’s time in a Ph.D. count as experience? Does time in an unrelated field count?

  • Since what previous experience means can be a bit fuzzy, depending on how much you care about your formula being objective and consistently applicable — such that no matter who calculated it, it produces the same result — you might want to define this more or less specifically.

• Tenure

  • On top of increases that account for changes in cost of living or inflation, employers might want to reward employees for staying at their company longer, and thus could include tenure at the organization in the overall formula.

• Skill

  • You might want to reward an employee for their skill in their job. For example, you might have two employees who are in the same role, but one is a higher performer than the other. This employee might get a higher skill rating.

  • Of course, this might reduce the objectivity of the formula. Managers might rate “skill” in a biased way.

• Title level

  • Many organizations reward staff for advancing within their organizations—that is, they pay staff more as they are promoted. Your formula might account for this by, say, paying staff with junior titles less than staff with senior titles.

• Cost of living

  • If you have employees in multiple locations, the cost of living might vary between them, and so it might be prudent to pay employees different amounts to account for that.

  • You could adjust your salaries somewhat or fully relative to the cost of living.

  • For example, if City A has half the cost of living of City B, you might reduce the salary by half, or you might just reduce it a little bit, depending on how much you care about incentivizing employees to live in cheaper or more expensive places.

  • You might also increase the salary above the cost of living adjustment for employees who live in a specific location. For example, if you specifically care about employees living near your offices but also allow remote work, you might want to pay employees more if they live in a specific city, above and beyond the cost of living adjustment.

• Cost of labor

  • Different locations have different costs of labor. For example, London and Washington, DC have similar costs of living, but typical wages are very different. If you have remote employees in both of those cities, offering a salary that is similar based just on cost of living might not produce a competitive wage compared to their counterfactual.

• Market rates of roles

  • Different roles might pay very different market rates. For example, software engineers might make much more than teachers. If you’re hiring both types of roles and want to hire competitive talent for software engineers but also minimize the amount of money you’re spending on salaries, you might want to incorporate market rates for roles into your formula.

  • You might want to only adjust the formula for certain roles that are hard to hire for. It might not make sense to do this for all roles.

  • For example, if you typically hire candidates from other nonprofits, formulas can usually produce fairly competitive offers without a market rate adjustment, because salaries tend to fall within a relatively restricted range. But if you also need to hire software engineers who come from for-profits, then a market rate adjustment for software engineers specifically might make sense.

• Counterfactual earning potential

  • Maybe you want to adjust salaries based on what your staff would earn elsewhere.

  • For example, if you hire an ex-teacher and an ex-management consultant into the same role, their alternative earning potential might look pretty different.

  • You could adjust the salary based on a rough sense of this by looking at their resume, and considering their alternative options.

  • In practice, this might be pretty hard to do though, and probably could produce some strong feelings of unfairness (given that it might result in very different pay for the same work).

• Currency

  • If you have employees internationally, you might be paying staff in different currencies. Therefore, you might need to convert your salary into the employee’s local currency.

Considerations around combining inputs

Setting weights for inputs

After you decide on what your inputs are, you’re going to need to combine them. The first thing you need to do is decide how important each factor should be to the overall salary.

To start, I recommend setting a relative weight to each one. An easy way to do this is to answer the question “if the employee advanced on this measure by 1, what percentage raise should they get?” If they have 1 more year of experience, or had a title level 1 higher, how big should their raise be for that change? If the answer is 5%, for the time being, set the weight at 5%. You can always adjust later after seeing the salaries your formula outputs.

Some inputs might not need a weight but should instead affect the entire salary. Maybe you want to adjust the entire salary by cost of living such that salaries exactly match the differences in cost of living between two places. If City A has 50% of the cost of living of City B and an employee’s salary in City A is $50,000, then in City B it should be $100,000. If you choose to do this, you might apply the cost of living adjustment to the entire output of the salary formula, as opposed to setting a weight for it.

So after you decide on your inputs, set a weight for each one based on how important you think it is, or decide if it needs to impact the salary formula in a different way. Don’t worry too much if this feels a bit uncertain right now. You can always produce some dummy salaries in a spreadsheet using your weights when you’ve set up your formula, and decide if they make sense after seeing the outputs it produces and comparing them against, say, market rates.

So, let’s imagine a salary formula with the following inputs and weights:

• base_salary: $50,000

• title_level (we have 5—titles 1 through 5): 20%

• tenure: 5%

• col_adjustment: The entire salary is adjusted by relative local cost of living, with Philadelphia, Pennsylvania set to 1.

Now, how do we combine these into a salary?

Combining salaries multiplicatively or additively

Imagine that your base salary is $50,000, and your tenure adjustment is 5%.

You can combine salary inputs in two ways:

• Additively, where each year of tenure adds 5% of the base (e.g., $2,500 in our example). This will result in linear salary growth when the input is increased.

• Multiplicatively, where each additional year of tenure increases the entire salary by 5%. This will result in compounding salary growth when the input is increased.

The advantage of the additive approach is that, over time, the salaries of the highest-paid people will increase more slowly than the salaries of the lowest-paid people, meaning that your salary hierarchy will stay relatively flat. If your organization cares a lot about equity, this might be a valuable feature. However, it also might mean you can’t offer competitive salaries to your most senior employees because their salaries might not keep up with outside options.

The multiplicative approach increases salaries more quickly. For instance, if an employee earning $100,000 receives a 5% raise, their salary increases by $5,000, compared to an employee earning $50,000, whose salary only increases by $2,500. As a result, this method creates a wider gap between the highest- and lowest-paid employees over time.

Here is a simple formula for both approaches, and an example calculation for an employee who has a title level of 3, 5 years of tenure, and lives in a city that costs 85% of Philadelphia’s cost of living:

Additive Approach: salary = (base_salary + base_salary * title_level * title_level_weight + base_salary * tenure * tenure_weight) * col_adjustment

salary = (50,000 + 50,000 * 3 * 0.20 + 50,000 * 5 * 0.05) * 0.85 = $78,625

Since the weight of title level is 20% of the base salary, if their title level went to 4, they would get a (20% × $50,000 =) $10,000 raise.

Multiplicative Approach: salary = (base_salary * (1 + title_level_weight)^title_level * (1 + tenure_weight)^tenure) * col_adjustment

Or for our same employee:

salary = (50,000 * (1 + 0.20)^3 * (1 + 0.05)^5) * 0.85 = $93,730.12

So, because each additional title level is increasing the entire salary by 20% (not just the base), and each year of tenure is increasing the entire salary by 5%, the final salary is higher. A promotion to title level 4 would increase their salary by a further (93,730.12 × 0.2 =) $18,746.

To illustrate this, here is this employee’s salary under both formulas over time, at title level 3, as their tenure increases year-over-year. The additive salary increases linearly over time, adding the same 5% of $50,000 each year, while the multiplicative salary compounds, increasing by 5% of the total each year.

Mixing additive and multiplicative approaches, or other options

Of course, you might think that some inputs should be added and others multiplied in. To accomplish this, you add the additive inputs first, then multiply in the multiplicative ones. Say we wanted to add in our title level adjustment but multiply in our tenure one (i.e., every title level increase gives you the same dollar amount raise, but every year of tenure gives you a percentage bump):

salary = ((base_salary + base_salary * title_level * title_level_weight) * (1 + tenure_weight)^tenure) * col_adjustment

Or for our same employee:

salary = ((50,000 + 50,000 * 0.20 * 3) * (1 + 0.05)^5) * 0.85 = $86,787.15

Or, you might want some inputs to scale on a different trajectory, such as diminishing their effect over time. For example, you might care about previous experience, but past a certain point, you don’t think marginal years matter much. You could use a sigmoid to do this, like Wild Animal Initiative does:

experience_adjustment = 40,000 / (1 + exp(-0.25 * experience)) - 20,000

Here the first year of experience adds around $2,500, but the 5th year only adds a marginal $1,850, and the 20th year adds only about $75.

Depending on the specific factor, you can combine these approaches (though at the cost of making your formula more complex).

Finally, you might also add a flat amount to each salary. For example, maybe you think every staff member should have an extra $5,000 on top of their salary to pay for well-being expenses that can’t be covered by the employer directly. You might handle it by increasing salaries by $5,000. This amount isn’t affected by any of the multipliers or adjustments year-over-year; it’s just an additional $5,000, added at the end of the calculation.

Updating the formula over time

Year-over-year, you might need to update your formula. Obviously, if you include tenure as a component, this will increase automatically for staff and they’ll presumably receive raises. You might also want to make sure that new offers to candidates are competitive, so you may want to increase the base salary over time, potentially with inflation or cost of labor increases, or other factors.

This is fairly straightforward to do, putting aside some currency-specific considerations, especially using a multiplicative approach as outlined above. With a multiplicative approach, increasing the base salary with inflation will increase all salaries by the same degree. But, it’s worth bearing in mind that if you have both a tenure adjustment and an inflation adjustment to the base salary, then salaries will increase above inflation over time. This is because staff will receive both their tenure adjustment and the inflation adjustment. If your tenure adjustment is 5% and inflation is 3%, a staff member might get an 8% raise. This is also true in the additive approach, but less severe due to the scaling differences outlined above.

You might also want to increase your salaries over time below inflation. For example, if you increase your base salary by, say, inflation minus 1% and have a tenure adjustment, your current staff will still get above-inflation raises year-over-year, but new offers will not increase at the rate of inflation. This might be because you’re trying to stay competitive with local salaries that are also not increasing at the rate of inflation. Similarly, you could increase your base salary by the change in the cost of labor, rather than inflation.

Special considerations

Cost of living and cost of labor

Including cost of living or cost of labor in your formula seems straightforward, but can introduce some tricky situations. Because you’ll likely need to set a benchmark that your cost of living adjustment is measured against, the relative differences in cost of living can shift in unexpected ways due to variance in change over time across cities.

For example, if your baseline is set to Philadelphia and it has a cost of living factor of 1, but Philadelphia sees a massive increase in cost of living due to a surge in popularity one year, while Washington, DC does not see a similar increase, then using Philadelphia as your benchmark would cause Washington, DC salaries to go down. In practice, you probably don’t want salaries to decrease year-over-year, especially for staff actively at the organization. So it might be better to set your cost of living benchmark to a national index or something less prone to year-over-year variance. Alternatively, you could make cost of living a smaller or less weighty factor in your overall formula, so these changes are offset by other increases (like inflation adjustments or tenure increases).

My personal view is that many remote organizations shouldn’t include cost of living considerations in their formula. They might actually be neutral about where staff live. If the organization is neutral about where staff live, why should it pay different amounts based on where staff choose to live?

Of course, not adjusting for cost of living in some sense incentivizes staff to move to cheaper places. But, my personal view is that it is fine to leave these things at the discretion of staff, and not be an organizational matter, since the organization doesn’t really care where the staff member lives.

I do think organizations should adjust salaries above the cost of living when they care about people living in specific places. For example, if you have an office and allow remote staff, you might want to adjust above the cost of living for staff who live in the city where your office is. That way you’re actively incentivizing staff to move to that location, which you might prefer to them working remotely.

I also think organizations hiring internationally should probably adjust for cost of living or cost of labor across countries, where the variance in costs can be extremely high (relative to within a single country). But, I think this variation is best tackled by setting different base salaries in each currency that your organization pays in, and not adjusting the formula further. That way, you also don’t get extreme versions of the problems I outlined above (which can occur because inflation can be very different between different countries and currencies).

Currencies

If you pay staff in different countries, there are two ways you might handle currencies in your formula:

1. Your formula calculates salaries in one currency and then converts them to the local currency. You then update conversion rates every year.

2. You have different base salaries in each of the currencies you pay staff in.

Option one might be appealing because it’s simpler in some sense. You only have one base salary and you just track currency rates over time; however, it introduces an unfortunate downside. If the local currency strengthens or weakens against your base currency, you might be forced to give non-inflation related raises or decrease staff salaries if you’re strictly following your formula.

For example, if a staff member’s salary converts from 50,000 in the base currency to 65,000 in the local currency this year, but the exchange rate shifts such that — even with a raise in base currency terms— the formula suggests paying them 63,000 in local currency next year, it would feel like a salary decrease to the employee.

Option two usually works better. To use this approach, you set base salaries in each currency you pay in. You can do this by spot-converting between currencies, or by using a cost of labor or cost of living adjustment when converting — for example, using purchasing power parity rates (see the previous section for more reasoning here) or using cost of living / cost of labor databases like Numbeo or Expatisan.

Then you treat each base salary as independent, never converting it back to the original currency. For example, if you pay people in both dollars and pounds and after a year the dollar inflation rate is 2% and the pound inflation rate is 2.5%, you would raise the dollar base salary by 2% and the pound base salary by 2.5%, independent of what the exchange rates imply (though of course these inflation rates might correlate heavily with what the currency exchange rates imply). This ensures that local inflation is accounted for in each currency, which helps if inflation is highly variable between countries, and ensures that your formula won’t accidentally produce a salary decrease for anyone year-over-year. If you need to pay a staff member in a new currency, you simply follow the same procedure to add the currency at the time they start.

The downside of this approach is that your organization is bearing the costs of foreign exchange — if your base currency weakens, you’ll pay more to hire staff in other countries. But, this seems sensible because employees don’t normally expect their salaries to decrease because a currency they aren’t paid in weakened.

Bands instead of specific outputs

Maybe you don’t want to go all the way to using a salary formula and still want salary bands that you can negotiate within. Bands are appealing because they constrain the negotiation to a specific range. You know the highest you’re willing to pay for a role, and maybe you know the lowest you’re willing to pay.

The easiest way to create bands from the outputs of a salary formula is simply to have a percentage variance that you’ll allow away from that salary. For example, if your formula produces a salary of $100,000 and you allow 10% variance, you can make offers for that role within $90,000 to $110,000. This ensures that your salaries are roughly close to equitable (since there are constraints on negotiation) but at the same time gives you some ability to negotiate or offer people a bit more should you need to.

I’m not sure this makes a ton of sense to do in most cases. Generally, I think that either your bands will be really large to account for the large differences in pay that people might expect, in which case what’s the point of doing the formula anyway? Or the bands will be relatively small, in which case they might not actually provide that much utility. I do think it’s an option for people who like to have a little wiggle room in their salary formula.

You might also use a strict formula, and apply bands to limit the outputs of the formula (preventing edge cases where your formula produces a really low or really high salary).

How do you know if the salaries it outputs are competitive?

It’s usually worth stress-testing your formula against benchmarks. There are various public or private salary benchmarking datasets available, but you can also do a rough version of this yourself by:

1. Going through recent job advertisements from other organizations that your applicants might apply for (e.g., on 80,000 Hours, Probably Good, or Animal Advocacy Careers).

2. Pulling salary ranges for comparable roles from these job boards.

3. Seeing if the salaries your formula produces seem competitive with the other roles candidates might be applying for.

This is imperfect in a variety of ways, especially because job advertisements don’t always reflect the actual salaries organizations are paying, but it is a signal of what your peer organizations think is a competitive salary on the market. If you think they are getting strong applicants, it might be a good indicator of whether or not you will too.

If you do decide to use benchmarked salary data, it’s probably worth making sure that it specifically applies to the region you’re hiring in (salaries can vary wildly, even city-to-city within a country), and is actually for the right industry. For example, if you look at salary data for all nonprofits, you’ll see salaries across a wide range of roles and organizations — small local nonprofits with a few staff usually pay very differently than multi-billion dollar foundations.

Finally, it’s worth considering the salary expectations of your candidates. For example, if you’re hiring a software engineer, they might be looking at industry roles that pay much more than nonprofits. This might mean that benchmarking data that isn’t role-specific could be misleading.

My default formula recommendation

My personal view is that it’s easiest to keep salary formulas fairly simple. If I launched a remote organization today, I’d likely use a multiplicative formula that includes:

• Base salary

  • I would set a base salary equal to a very entry-level role in the field, and increase it at a sub-inflation rate annually (e.g., inflation minus 1% or pegged to cost of labor). I’d use a sub-inflation rate because other adjustments will mean current staff receive above-inflation raises annually, but new salaries will be kept in line with normal, sub-inflation wage growth.

• Tenure (at ~2%–5% increase per year)

  • I think that providing above-inflation increases annually is important for ensuring good staff retention. The downside is that your costs can rapidly increase relative to inflation. Adjusting for tenure on top of a sub-inflation base salary increase allows you to keep wage growth relative to inflation in control while still giving staff reasonable annual raises.

• Title level (at ~5%–15% increase per level)

  • On some level, I don’t really like that more senior people in organizations are paid more. But I also think staff expectations generally mean that this is required to retain high-performing people who are rising within your organization.

  • Title level adjustments here basically mean “a staff member gets a salary increase when promoted.”

• Skill (at ~2%–5% increase per level)

  • I think that rewarding highly skilled employees is much more important than rewarding title level increases. For most organizations, I think impact is driven by relatively few of the staff. Mechanisms for retaining those people are important, and skill adjustments seem like the easiest way to do so.

  • I also think that these introduce some level of subjectivity to the formula — if you have managers set skill ratings, bias could be introduced. This means you need good, easy guidance for setting these ratings and training for managers.

• Maybe market rate adjustments for specific, hard-to-hire roles with high counterfactual earnings (e.g., lawyers)

  • I think that formulaic approaches can break when hiring specific, hard-to-hire roles. I’d rather not break my formula by negotiating over salary. So I prefer to use a market rate adjustment for specific roles that have high counterfactual earnings, and thus might be hard to hire for, such as software engineers or lawyers.

So, a formula might look like:

salary = base_salary * (1 + tenure_rate)^years * (1 + title_rate)^level * (1 + skill_rate)^level * market_rate_adjustment

I personally am inclined against doing cost of labor or cost of living adjustments based on employees’ locations, unless I specifically want someone to be in a specific location, or talent is unusually constrained. I’m somewhat inclined to think that if the formula is producing roughly competitive salaries—say, for people anywhere in the US except the most expensive cities—and I have strong talent pools for my roles, then adjusting for cost of living mildly incentivizes the organization to hire people in cheaper places (a distortion to hiring on skills you care about) and can get complicated culturally (people feel bad about their salary being lowered if they move).

However, if I ran an organization with an office, or there was specifically an organizational need for people to be in a specific place, I might be inclined to include an adjustment for people to live in that specific place—I want to actively incentivize people to move there.

I also would not include previous experience by default. I find it pretty hard to calculate (e.g., what experience counts as relevant?) and it can lead to fuzziness and negotiation over what counts as relevant.

I personally think multiplicative salaries better capture how salaries change over time at other organizations, but the result of using this approach will be that the gap between the highest- and lowest-paid employees grows over time.

Salary formula examples

Some organizations publish their salary formulas and methodologies. If designing a new one, it might be helpful to look at these for inspiration:

• Wild Animal Initiative

• Giving What We Can

• Animal Charity Evaluators

Charitable endowments, in the classical sense, are funds that charities hold and invest, spending only the interest they make on the investments. Endowments as a source of revenue for nonprofits are distinct as the revenue is not contingent on recurring engagement with a donor — the organization just needs to prudently invest and spend its funds, and doesn’t need to apply for grants for that funding.

Endowments are common among universities in the US, and also used by many (especially older) nonprofits to provide a source of revenue independent from their donors. But these mechanisms are basically not used in the high-impact charity sector, except by donors themselves (who might give out the returns from their endowments as grants).

I argue three points that are somewhat in tension with each other:

• Endowments often waste a lot of charitable resources that could be better used otherwise.

• We should be equally skeptical of donor endowments as we are of charity endowments.

• There are certain circumstances where having endowed funds is extremely useful, and would be beneficial for an organization’s impact (vs. spending the funds directly).

  • And correspondingly, donors should be somewhat open to granting funds that could be put into quasi and term endowments (but not permanent endowments, see below) to organizations in certain circumstances.

What are endowments?

In a US 501(c)3 context, there are several forms of endowments.

• Perpetual/permanent endowments: A donor designates that their funds be held in the endowment forever, and the organization only spends the gains or interest.

• Term endowments: A donor designates that their funds be held over a certain time period (e.g., many years), with the organization only spending the returns until that term ends.

• Quasi-endowments: An organization chooses to invest its own funds, and spend the returns in an endowment-like manner.

Endowments can also be “restricted” where their gains are spent on specific programs, or unrestricted, where the organizations choose how they are used. These endowment types can also be combined (by an individual donor, or across institutional funds). Often, when you hear “this university has a $10 billion dollar endowment,” it means something like “this organization has a giant basket of various gifts from the last few hundred years, with different sets of rules governing specific portions of these funds in some complicated manner.”

Endowments often come up in the context of universities. For example, a common critique of elite educational institutions from the left is “why don’t they spend their endowments down and give their very good education to more people?” But, for many large universities, this is somewhat difficult. Endowments are typically governed in the US via a uniform law (e.g., law adopted identically by every state) called The Uniform Prudent Management of Institutional Funds Act (UPMIFA). Under UPMIFA, funds that a donor designates to a perpetual endowment are basically required to be held and invested (and not directly spent) unless various conditions are met which might look like:

• The donor releases the restriction of the funds to the endowment (which they can only do if they are living — a problem for older universities)

• The organization petitions a state attorney general and gets approval to spend the funds because the restriction is now illegal or impossible to achieve.

• If the fund is less than $100,000 and over 20 years old, and the restriction is unlawful or impractical.

For universities, these conditions usually aren’t met. For example, Harvard’s endowment is around $52 billion USD, and produces around 37% of Harvard’s annual revenue. Of that $52B, around 80% is legally restricted and cannot easily be spent directly. This means that even if Harvard shifted strategies and wanted to massively increase endowment spending, it could only do so with around 20% of its assets. And only 5% of those assets are fully unrestricted — much of the endowment is restricted to specific programs, professorships, or other purposes.

Endowments therefore can be an immense source of income for institutions that have them ($2.5B in the last year for Harvard), but also are restraining — Harvard has access to this level of income basically forever (if they manage the funds well), but practically can’t do anything else with these funds. The donors who funded this endowment will directly impact how Harvard spends its assets basically indefinitely into the future (or until the law is changed).

Endowments waste a lot of charitable resources

Say I have $100 million USD (lucky me). I decide to donate this to my preferred charity. But my preferred charity doesn’t seem able to spend $100M easily. Maybe it seems like they can usefully spend only $10M each year, and will continue to be able to spend that much impactfully into the future.

But I’d like them to be less stressed by fundraising, so maybe I give them the entire $100M. They carefully and prudently invest it, and are able to get a consistent 6% return above inflation. I’ve given them the ability to basically get $6M of their funding, indefinitely into the future. This is probably a huge relief to the charity — $6M would be a lot of money to raise year after year. And, theoretically, in the long run (say after 17 years), I’ve given them way more than $100M.

But of course, if I hadn’t held the funds, I would have done something else with it. What might I have done?

• Given it to other charities

  • I’ll assume for the purposes of this exercise that I think the charity spending $6M per year over whatever amount of time into the future will continue to be impactful. So this option seems not much better from an impact perspective.

  • The main upside of giving it away now to multiple charities (if I’m really itching to get rid of the money) is that I have more confidence in how it will be used — even if I think my preferred charity is the most effective, there isn’t a guarantee it will continue to be effective. Maybe there is a chance my preferred charity will decline in effectiveness, such that a gift with a more certain but lower impact now looks better than an uncertain impact in the future.

  • Of course, giving it all away now to many charities who can spend it immediately, on net, results in less money going to charity in the long run — unless I think charitable opportunities will get much worse over time, I might be causing less money overall to be spent charitably, so I also might need to prefer my present opportunities, or the value created by current interventions, to justify this (see intersections with patient philanthropy, below).

• Invested it and given it away over time, like the charity.

  • Of course, I can also invest the funds, just like the charity. Assuming both the charity and I are equally good at managing funds (or hiring money managers), then this results in the same amount going to charity over time. But now I retain control instead of the charity retaining control.

  • If I just give the $6M I generate to the charity every year, they now have to worry about fundraising a bit more, so they are a little worse off.

  • But the advantage is I retain option value — I can choose to stop giving to my preferred charity (say, because they became less effective), and give to something else.

• Invested it and make more money to give away

  • I somehow got $100M in the first place! That might mean I’m pretty good at making money.

  • Depending on my circumstances, there is some chance I can make a lot more money than I would just investing it in the market — maybe I can use it to start a business, or invest in some startups, or take other opportunities that aren’t available to the charity (or at least, opportunities the charity would never take).

  • While foundations and donor advised funds are subject to responsible investment requirements, individuals might not be, which means they can take greater risks with the funds (and thus see higher upside) than charities are legally able to, at least in certain circumstances.

  • If I am successful, I might generate a lot more money to give to charity in the long run, and this could offset the cost of not giving it to charity immediately.

    • Of course, I might be wrong and lose it! That would be worse, but I should presumably only take this path if it seems valuable in expectation.

    • And this is further subject to some kind of optimal stopping problem. Presumably at some point I should cash out and donate the money (again, see patient philanthropy discussion below).

Overall, these are all pretty plausible scenarios — even though it would be great for this specific charity to get an endowment, there are lots of totally reasonable worlds where I actually shouldn’t give it to them (at the cost of them losing some independence from funders) — it’s therefore not surprising that many high-impact charities don’t have huge endowments — their funders might have been making mistakes if they’d funded an endowment instead of pursuing other strategies.

But there are a few other reasons to think endowments might not be a great choice for funders.

Most (all?) things don’t need to exist or be funded forever

The best case against endowments is that they lock charitable assets into inflexible arrangements. In some cases, this effectively means that the assets are permanently locked into being invested and having their returns generate income for the organization holding them.

While this model might make sense for universities and hospitals, many charitable projects just don’t need to exist forever — their impact might be expected to be had only over a limited period of time. This might be because the project effectively resolves the issue they work on, or there are diminishing returns to continuing to work on it.

There might be some high-impact institutions that would want to exist indefinitely, such as research organizations, where an endowment might provide some benefit. But it’s unclear to me that research institutions can demonstrate that they can be highly effective or innovative forever. Of course, funding might be a barrier to continued impact, but institutions can change, atrophy, face bureaucratic drift, or otherwise stop being effective stewards of funds.

And many organizations just have inherently time-limited missions. I hope that we can one day end malaria — at that point, we might not need organizations that try to stop its harms. Giving these organizations perpetual endowments is not going to be beneficial in the long run.

I think that these points make a strong case against donors making gifts required to be endowed perpetually. At a minimum, it seems likely that effective endowments will be either limited in term, or made at the charity’s discretion (e.g., the quasi-endowment, described above).

Endowments often create worse incentives for the charities

Ideally, donors are able to provide some level of check on charity activities — they regularly review the activities and impact of the charity, and their funding is conditional (again, in an ideal world) on the charity continuing to use funds effectively.

Of course, in reality, charitable resources aren’t distributed ideally by impact: even highly effective donors might rely on heuristics like trust in the specific charity, the charity’s self-reported plans in a grant application, community and confidant impressions of their work, external research, etc. All of these heuristics and tools can cause mistakes.

But despite this, donors are providing a regular check on charity activities through their due diligence and decisions to continue or stop giving — if the charity stops being as impactful, impact-focused donors might stop giving, and direct their resources to more useful projects. Because of this, charities have strong incentives to align with donor interests, and if those donors are interested in impact, charities have stronger incentives to focus on impactful projects beyond their own intrinsic motivation alone.

However, endowments break this pattern — because endowments allow charities to receive some of their income independent of their activities, they remove these types of incentives. Charities can pursue their own goals, independent of donor interest.

This could be good and bad. Donors can make mistakes, or make strategic pivots that make them less impactful for external reasons (like their values changing or social pressure). If they change their priorities, they force charities they fund to change priorities, find new funders, or cut back on their work. These could all be bad outcomes for effective charities.

But having a regular check on a charity’s activities could also stop the charity from doing less impactful work. I can think of many examples of charities that closed their doors or reduced programs because they lost funding — charitable dollars would have been wasted if they had endowments to keep these less effective programs going. And because the cultural costs of change inside an organization are large, charity incentives probably point slightly away from staying impactful even on their own — an endowment might make this worse.

Many critiques here also apply to donor “endowments”

Of course, charities aren’t the only kinds of entities that can have “endowments.” Donors also have “endowments,” and often operate as such — many wealthy people choose to give away a portion of invested assets, or might have funds in foundations or donor advised funds that literally operate as an endowment.

These structures are subject to the exact same critiques:

• Their funding vehicle probably shouldn’t be funded forever.

  • The same logic that applies to charities applies to donor vehicles. A donor’s foundation or DAF doesn’t need to exist in perpetuity — the problems and opportunities it was created to address will change, the founder’s values might become outdated or wrong, and the people managing it will turn over.

  • The Ford Foundation’s well-documented value drift from Henry Ford’s original intentions is a classic example: what was once a family foundation focused on Michigan shifted to a global focus very different from its original intentions (for better or worse).

• They don’t have good incentives to stay impactful.

  • Private foundations and DAFs lack the external accountability that charities (theoretically) face from donors. Who checks whether a foundation is doing effective grantmaking? For operating charities, donors can provide this function. For foundations or individual donors, the answer is usually that no one has real leverage — board members at foundations are often family or close associates, and the IRS and legal authorities care about legal compliance rather than impact.

  • Foundation staff develop their own institutional interests. Research on nonprofit mission drift suggests organizations routinely shift away from founding purposes as internal stakeholders pursue their own goals. Foundations have even less external pressure to resist this than operating charities do.

  • DAFs have been critiqued (I think unfairly) as “warehousing” charitable assets — donors get immediate tax benefits while funds sit uninvested in charitable work — and this critique applies some of the same logic I’ve applied to charity endowments. The median DAF payout at community foundations is around 9%, meaning most assets are being held rather than deployed.

Of course, ultimately, the money that is given to the charity is owned by the donor, and given at their discretion, so it seems odd to frame conversations around their “endowment” in this light. But from the lens of “there are X dollars that could be given, and Y charities that could receive it. How do we optimally distribute funds?” these questions still arise.

Intersection with patient philanthropy

Patient philanthropy argues that donors should often invest and give later rather than giving now, because:

• Investment returns allow more giving over time

• Future opportunities might be more impactful

• Our understanding of what’s effective will improve

This might seem to support charity endowments — if holding assets is good for donors, why not for charities?

But patient philanthropy arguments are also about who should hold assets during the waiting period, not just whether to wait. The arguments above suggest donors should usually retain control during the waiting period, because:

• Donors can redirect funds if the intended recipient becomes less effective

• Donors may have access to higher-return investment opportunities

• Donors can respond to unexpected giving opportunities

Charity endowments lock in both the decision to wait and the recipient.

The exception is when the charity is better positioned than the donor to identify the right moment to deploy funds. For some organizations with deep domain expertise working on uncertain timelines — maybe some AI safety organizations, or groups preparing for specific future scenarios — this could be true. But it’s a higher bar.

But endowments can be useful

Despite endowments basically looking ineffective on many fronts, they have one incredibly useful function: they create institutions that are independent from their donors.

There are several places where having an institution that can basically operate without continuous approval of its donors might be useful:

• Organizations that provide oversight for a community, especially a community those donors are engaged in

  • Sometimes, donors are deeply embedded in the communities they fund. These communities might create oversight bodies or ombuds bodies that serve a protective role, such as mediating conflicts, handling interpersonal issues, and similar.

  • Because donors wield a huge amount of power in nonprofit spaces, oversight bodies that depend on those donors for annual funding face obvious conflicts of interest. An ombudsperson who could lose their job if a major funder is unhappy with their findings cannot be fully independent.

  • Endowing such organizations creates genuine independence. The oversight body can investigate complaints about major donors, make uncomfortable recommendations, and publish critical findings without worrying about next year’s grant. This is probably one of the strongest cases for endowments in high-impact philanthropy.

• Some research organizations

  • Research often benefits from long time horizons and freedom from short-term pressures. A researcher pursuing a decade-long project shouldn’t need to repackage their work annually to appeal to shifting funder priorities.

  • But the case here is weaker than for oversight bodies. Research quality can be evaluated (imperfectly) by outside observers, so donor oversight might actually improve outcomes by cutting funding to unproductive projects. The risk is that donors optimize for legible outputs rather than important ones — but an endowed research organization can drift toward unimportant work too, without the check of needing to justify its existence.

  • I suspect that it is worth endowing some aspects of research with very long feedback loops, to insulate it from year-to-year shifts in donor priorities.

• Organizations with very long-term theories of change

  • Some organizations work on problems where impact is expected to continue to exist or only materialize over decades or centuries.

  • For charities with similarly long time horizons, dependence on annual funding creates pressure to demonstrate near-term results that may not align with their actual theory of change.

  • But this case also has limits. Very few organizations can credibly claim their theory of change operates on a multi-decade timescale in ways that can’t be decomposed into nearer-term milestones. And organizations claiming long time horizons are especially vulnerable to mission drift because feedback loops are so long.

Independence from donors doesn’t necessarily mean worse impact

I’ve argued above that donor oversight provides a check on charity effectiveness, and that endowments remove this check. But the relationship between donor oversight and impact isn’t straightforward.

Donor oversight optimizes for what donors can observe and evaluate. This works well when donor priorities align with actual impact and when impact is measurable. It works poorly when:

• Donor priorities diverge from impact (e.g., donors might prefer visible programs over less visible, more effective ones, prefer direct work over research, etc.)

• Impact is hard to measure or has long feedback loops

• The most effective work is uncomfortable or unpopular with donors

• Donor attention is fickle or subject to trends

For organizations in these categories, independence from donors might actually improve impact. An endowed organization can pursue unpopular but effective strategies, invest in long-term capacity building, and resist pressure to change in response to outside forces.

The key question is whether the organization has good internal incentives and governance to stay impactful without external pressure. This probably depends on:

• Strong founding values and culture around impact

• Board composition that prioritizes mission over organizational perpetuation

• Some external feedback mechanism, even if not donor-based (e.g., peer review, community approval, regular hiring)

• Built-in time limits or sunset provisions that force periodic reconsideration of how funds are spent.

So when should donors or ecosystems consider endowments as part of their giving?

Here are my very tentative best guesses:

Quasi-endowments should almost always be acceptable

Quasi-endowments — where an organization chooses to invest unrestricted funds rather than spend them immediately — should rarely be controversial. The organization retains full discretion to change course if circumstances change. This is just prudent financial management for any organization that expects to operate beyond the next year.

Donors should probably be comfortable with organizations building modest reserves and investing them. The question is one of degree: an organization sitting on 10 years of operating expenses while claiming funding constraints is different from one maintaining 1-2 years of runway. And an organization with more funds in reserve will inherently make a worse argument to donors that it has a lot of room for more funding.

Term endowments make sense for specific use cases

If a donor-restricted endowment seems appropriate, it should almost always have an end date. A 10-20 year term endowment, for example, might capture most of the benefits of independence while forcing periodic reconsideration of whether the arrangement still makes sense.

The strongest cases for term endowments are:

• Oversight bodies where independence is structurally necessary

• Organizations with genuinely long theories of change that are hard to evaluate from short-term outcomes.

• Situations where the organization needs to be insulated from short-term donor volatility but not forever

Endowments probably shouldn’t cover 100% of operating costs

Even for organizations that benefit from endowment income, maintaining some dependence on ongoing fundraising preserves useful accountability. An organization funded 50% by endowment returns and 50% by ongoing grants has meaningful independence while still facing external pressure to demonstrate value.

The exception might be oversight bodies, where even partial funding dependence could compromise independence.

Endowments via bequests have different considerations

Bequests are a natural fit for endowment-style giving because the donor won’t be around to provide ongoing oversight anyway. A donor who plans to give $10M at death faces a choice between directing it to immediate spending, giving it to an existing endowment-like vehicle (like a community foundation), or creating a new term-limited fund.

The case for bequest-funded endowments is stronger than for living donors precisely because the alternative isn’t “donor retains option value” but “money goes somewhere without donor input either way.”

Of course, there are ways to avoid this — a donor might designate someone they trust to manage their assets after their death. But for many individuals, the largest giving opportunity in their life will come when they pass away, and the timing of their death might not align with the optimal time to give. Endowments might help manage this.

This is absolutely not legal advice, and all people considering launching organizations should consult with lawyers.

You can launch a 501(c)3 really quickly if you want. If you have a funder lined up, you can go from an idea to having a nonprofit with 501(c)3 status and with staff on payroll within around 40 days. I think this is around the fastest you can consistently do this, though if lucky, you can make it happen a bit faster (I just think you’re significantly less likely to succeed at getting 501(c)3 status in that window).

This outcome isn’t guaranteed, and steps in this list can go wrong. In particular, expediting your 501(c)3 application, discussed below, has, in my experience, about a 60% success rate for working within your requested timeline (but has a 95%+ success rate for expediting *to some extent*). So that means that doing this, there is something like a 40% chance it’ll take longer than you expect, because you’ll have to wait longer to get 501(c)3 status.

Note that having 501(c)3 status doesn’t hold you up from receiving grants or setting up payroll. The IRS treats organizations that intend to file for 501(c)3 status as 501(c)3s retroactively to their date of incorporation as long as they get the status within 27 months. And, donor gifts in that period are retroactively tax deductible (once you receive the status). You can open bank accounts, run payroll, get insurance, etc. without 501(c)3 status. Public charities can mostly make grants to you. However, donor advised funds and private foundations might have a harder time getting funding to you without the status.

These comments are based on my own experience starting several 501(c)3s, and seeing several dozen go through the expedite process. Everything in this post is very much not legal advice, and you should always seek legal advice when forming a nonprofit.

What is a 501(c)3?

501(c)3 is a special tax status for US corporations. Most corporations in the US are formed at the state-level (e.g. in the State of Delaware). 501(c)3 is a tax status granted by the federal government that makes your corporation exempt from federal income tax, and makes donations to your organization tax deductible.

After you form, you have 27 months to secure 501(c)3 status. During this period, the federal government treats you as a 501(c)3. And, while you’re pre-status, you can still open a bank account, have staff, etc. So, many of the steps described below can happen in parallel, even if you’re waiting for your status.

The maximum speed path

1. Get a donor who only wants to give to you if you’re a 501(c)3 (∞ days)

The only way to get 501(c)3 status remotely quickly is to have a donor lined up who will only give you funds if you’re a 501(c)3 and who will remove the offer of funding if you don’t get it by a certain date, such that you can credibly claim you need the status quickly to get the funds.

If you don’t have this, getting 501(c)3 takes 8-10 months (and it’s unclear if this timeline has changed (for better or worse) in the current era of IRS staffing cuts).

Ideally, this donor is giving somewhere between 10% and 50% of your first year budget. I’ve heard people say it can be higher than 50%, but I’ve only seen successful 501(c)3 expedite requests in that range.

It helps if the donor is a private foundation or a donor advised fund, as these organizations can credibly claim that they need you to be a 501(c)3 (since they’d otherwise have to do expenditure responsibility, and many DAFs are unwilling to). But, this isn’t strictly necessary.

It can help if the donor is willing to write a letter that says “I’ll give you the money if you get 501(c)3 status but this date, otherwise you lose it” but this isn’t strictly necessary.

You’ll need to write a letter yourself to include in your application. Aaron Hamlin has a good template for this.

Finally, I’ve heard lawyers advise that the IRS doesn’t like it if the donor is a board member or other person affiliated with the organization. However, I’ve also seen the IRS expedite applications where the donor is a board member, so I’m not sure how true this is. But it probably doesn’t hurt for the donor to be unrelated.

Ideal case

• The donor is a DAF or private foundation

• The donor is giving 10%-50% of your first year of operating budget.

• The donor writes a letter awarding you the grant, with the exploding deadline stated clearly.

• The donor isn’t a board member or person related to the organization.

2. Incorporate in Delaware or another speedy state (2 days)

Next, you need to form a corporation in a specific state. Work with a registered agent service to do this, which will provide an address in the state you’re incorporating in. Note that this address is not a mailing address — it’s exclusively used for communication with the state about corporate activity. Don’t use it elsewhere.

I usually recommend Northwest Registered Agent. It’s part of a massive, shadowy registered agent conglomerate, but it’s the segment of that conglomerate that has the best customer service.

You can choose what state to incorporate in. Often people incorporate in the state that they live in (because you’ll probably need to register there anyway), or Delaware. But registering as a foreign corporation is easy, and registering in Delaware is very fast, so I prefer to register in Delaware, then do a foreign qualification in the state I live in. Definitely seek legal advice on this / this is not legal advice.

You can expedite Delaware registration for a few hundred dollars and incorporate in under 24 hours, but it’s generally pretty fast even if you don’t do this.

In Delaware, 501(c)3s are usually incorporated as an “exempt” corporation. This corporation type only requires 1 board member (the IRS requires 3 when you register as a 501(c)3 but you can incorporate with 1). The IRS cares about two things in your incorporation documents in particular: that it states you are formed exclusively to do 501(c)3 stuff, and that it states upon dissolution, you’ll distribute assets to other 501(c)3s. Template articles for exempt corporations from registered agents usually. include this, but it’s worth double-checking them before they are filed.

Registered agent template articles of incorporation will usually have charitable purposes listed as “any charitable purpose” or similar. You can make this a more specific purpose. Sometimes, this might matter a ton so if you’re planning on becoming a leading AI lab, you should probably have a lawyer look at them.

DO NOT pay the registered agent to get your employer identification number (EIN). They’ll often add it as a service by default for free, so you have to actively uncheck a box requesting it. This is tempting, but the IRS takes weeks to respond to their applications, so it’s a mistake if you’re trying to move quickly. See section 4 below for information on getting your EIN yourself.

3. Get an address (1 day)

You’ll need an address on file with the IRS. You should choose a stable address, that you can reliably receive mail at. Don’t use your registered agent’s address in the state of incorporation — registered agents will usually only open mail from the secretary of state for you.

You can use an office, a home (though if you’re on a lease, make sure it is okay), or a mailbox for this. Note that this address will be public. Changing your address with the IRS is annoying to do, and the IRS will occasionally make large, annoying mistakes if you start using a different address (like assigning the new address a different employer identification number). So, make sure you can reliably access this address into the future.

Many virtual office or mailbox providers will require you to sign and notarize a post office document to receive your mail — you can do this online, and it takes just a few minutes, but you’ll need an US ID typically.

What state should the address be in?

For many functions, like insurance, you’ll need to choose a “headquarters” in a specific state. If you have a main office, this location might be obvious. Often, people will choose the state where their executive director is located.

The state you choose has at least one extremely important impact on your organization: how expensive your health insurance quotes will be.

Health insurance companies in the US set rates partially based on the state of their customers, and they rely heavily on your mailing address to make this determination. Some states have significantly higher health insurance costs. For example, having an address in New York State for health insurance will probably double the quotes you’ll receive compared to pricing you’ll get if you are located in Pennsylvania.

If you use my preferred payroll provider, Justworks, the state you’re in will also impact if you get UnitedHealthcare (bad insurance (in my opinion) and slow to quote) or Aetna (good insurance and fast to quote) insurance plans offered to you. I believe this is also true for other PEOs, so it is worth asking the PEO how your address impacts health insurance plans you’re offered.

How to find a virtual mailbox provider

Virtual mailbox providers use a gig-economy-esque business where they provide a software tool, and various businesses receive and scan your mail. When you sign up with a service like Anytime Mailbox or PostScan Mail, both of which are fine options, they provide software to a local business, who receives and scans your mail.

So, when choosing an address, search the address in Google Maps before committing, and see what kind of business is getting your mail (you can look at it in Streetview if nothing comes up). It might be a grocery store or a liquor store or a mechanic. I prefer to find a mail service provider, who is in the business of processing mail, as they are typically more reliable and faster. You can also look up the business up and see if they have positive reviews online before committing!

Once you find a good provider, commit to your address!

4. DO NOT pay a third party service to get your EIN. Get it yourself (5 minutes)

You’ll need an employment identification number (US tax ID). Registered agents can apply for these for you, but they usually fax or mail in the application (I’m not sure if they are required to but they always do). This takes weeks for the IRS to respond to!

If you have a US social security number, you don’t have to mail it in! You can apply online. It takes 5 minutes, and the form basically requires just your address (from above), a bit of information about your business, plus the reason you need it (usually to hire employees and to file tax returns). You’ll get it issued instantaneously. Save the PDF that the IRS website generates at the end of the application, as it is extremely annoying to get a copy of it if you don’t.

If you aren’t a US person, you’ll need to mail/fax in the application anyway, so have your registered agent do it to save you the trouble, but you’ll need to add two weeks to the timeline outlined here, as the 1023 and bank won’t be able to move forward without an EIN.

5. Adopt policies (0 days)

Before you apply for 501(c)3 status, your board should adopt bylaws (articles governing the corporation) and a board conflict of interest policy. When you incorporate, you usually need to file bylaws with the state, and your registered agent will often file some boilerplate ones. You can work with a lawyer to update those, or adapt one of many templates online. Similarly, there are many templates for board conflict of interest policies online that should meet IRS requirements.

6. File Form 1023 with an expedite request (5 days)

Form 1023 is your application for 501(c)3 status. If you’re doing anything remotely weird, complicated, or not obviously charitable, you should get a lawyer to help you complete it! You probably should get a lawyer regardless, and the below isn’t legal advice, and you shouldn’t rely on it as such.

You can file Form 1023 yourself online. It costs $600, and takes around 20 minutes to complete, besides writing out your narrative description of your activities (which is important to get right). In Form 1023, you’ll describe your organization, list your board members, describe various policies you have, how you’ll use the intellectual property you’ll create, etc.

This is important to get right! This is the IRS’s understanding of what you do as a charity, and your 501(c)3 status is conditioned on doing the activities you describe in your Form 1023.

In the 1023, you’ll write a narrative description of your activities. Lawyers will write you a treatise with lots of reference to charity law. The IRS will also accept three sentence descriptions of what you plan to do that are very high level. When I’ve dug into this with lawyers, I’ve not gotten a good understanding of why they take a different approach (usually, they are under the impression that the IRS will reject very short applications, but that isn’t true). But, the IRS does evaluate your charitable status on the basis of this description, so it should describe your work accurately!

Besides a narrative, you’ll also need a 3 year budget for your organization. This can be a projection based on your current plans.

You’ll need to attach a variety of documents to your 1023, including your articles of incorporation, your bylaws, and if you plan to expedite your application, a letter from your organization requesting expediting (template here). You can also include a letter from your funder as additional evidence of the expedite request, but it isn’t required.

Then you just submit, and sit back and wait! (or do all the steps below, which you don’t need 501(c)3 status to advance — you can complete them while you wait for your status).

7. Expedite your 1023 properly

Your expedite request is not guaranteed. In my experience, something like 60% of expedite requests are met in the window originally requested. However, 100% of expedite requests I’ve seen submitted are approved in a shorter period than the typical review period (6-8 months) — it might just take a few months instead of 30 days.

If you work with a lawyer on your 1023, they very likely will not do the following steps, but the following steps will significantly increase the chances of your expedite request working, so it’s worth asking them about it.

Expediting your 501(c)3 status doesn’t stop when you submit it! My rough understanding of what happens when you click the “expedite” box in the application is that your application goes into the 1023 review queue. Someone has the job of sorting through that queue, and the review backlog is months long. Your expedited application request is mixed in with every other request. It doesn’t go into an “expedite” track. So, you need to get your expedite request out of that queue, and assigned to a review officer!

The best thing you can do to achieve this is to fax your expedite request to the IRS. This is because when the fax machine goes off, someone notices it, sees that you requested an expedited review, and pulls your application out of the queue and assigns it to someone. The IRS’s exempt fax number is here. Online fax services like Dropbox Fax work great, and cost around $1.50. You just need to send your expedite request letter, and a cover sheet explaining what application it is in relation to.

Then, wait around 1-2 weeks, and call the IRS tax exempt toll-free number. You’ll have to wait on hold for 1-2 hours typically, but you get to listen to the dulcet tones of Fresh Optimism…

…so that’s something. When you get connected, you can ask about the status of your application, and specifically ask to be connected to the review officer. You can politely emphasize why you need it expedited, and thank them profusely. Be genuinely nice to them and they are helping you out! I don’t recommend calling in again.

8. Get 501(c)3 status (30 days)

At best, it’ll take around 30 days to get an expedited application approved. Sometimes, you’ll get a letter from the IRS asking for clarification about something in your application. This letter will indicate the review officer (if you didn’t do the above and want to get in touch). My sense is these letters are usually a positive sign (e.g. the IRS’s review is almost complete), and after responding to it, you’ll get a response fairly quickly.

But, while you wait for your 501(c)3 status to be approved, you can still proceed with opening up other operational necessities!

9. Open a bank account (2 days if American)

You’ll need a bank account in your business’s name. My typical recommendation is that businesses open an account with Mercury first if trying to move quickly, then an account with Chase after they get 501(c)3 status.

I recommend Mercury for a handful of reasons:

• They are fast: they will approve an application from an American in around 24 hours, and from a non-American in around 3 days.

• They are profitable, (which other fintech startup banks that are also fast are not) which suggests to me that they are slightly less likely to fail randomly (which would be annoying, since they hold your money!)

• They have a good user interface, good customer service, and are easy to work with.

I recommend opening an account with Chase after becoming a 501(c)3 because I think a good financial risk mitigation strategy is having a systemically important bank account and Chase is the least bad options among major US ones. But, it will be a significantly worse experience than opening the Mercury account, and it won’t be fast.

10. Setup payroll (2 weeks)

US payroll takes around 2 weeks to set up, but can be faster if needed. My default recommendation is Justworks, a PEO). I recommend them because their customer service is typically very good, and PEOs seem better for basically any organization that doesn’t have enough staff to require a full-time person working on US payroll compliance. You pay slightly more per month, but don’t have to deal with workers compensation audits, you pay lower health insurance and workers compensation insurance rates, and don’t have to deal with state-level registrations for employment tax accounts in most states.

Health insurance takes around 30 days to set up in the US, and 401(k)s take around 45 days, so if you plan to offer these benefits, you’ll need more time.

11. Set up accounting (1 day)

Finally, you’ll need to set up accounting for tracking financial transactions. Often, organizations will partner with a bookkeeping firm (I’m happy to provide recommendations).

If you choose to do bookkeeping yourself, I recommend Quickbooks Online (which is terrible, but so are all its competitors) because it is the most widely used, and has the best integrations with other software. But, unless you plan to have very high volumes of transactions immediately, wait until you get your 501(c)3 status to buy it. If you wait, you can purchase it through TechSoup, a service that provides discounted software for nonprofits. You can save around $1,000 a year buying it via TechSoup (or more for fancier versions), but it is painful to transfer from a normal account to a TechSoup account, so just buy it later, and catch up on your bookkeeping then.

If you are going to work with an outside firm, and they plan to charge you for the Quickbooks subscription they’ll use, have them wait too — spreadsheets are usually fine for tracking things in the 30 days until you get your 501(c)3 status.

Things to do next

After you complete these steps, you basically are good to go and be a nonprofit! Things that you probably should be thinking about early in the course of your operations include:

• Getting insurance (1 month)

  • Many organizations secure both Directors and Officers and General Liability insurance. If you have employees, you’ll also need workers compensation insurance (though if you work with Justworks or another PEO, it will be provided via the PEO).

• Doing board things (1 month)

  • You likely should have a board meeting early on to set your strategy and plans!

For impact-minded donors, it’s natural to focus on doing the most cost-effective thing. Suppose you’re genuinely neutral on what you do, as long as it maximizes the good. If you’re donating money, you want to look for the most cost-effective opportunity (on the margin) and donate to it.

But many organizations and individuals who care about cost-effectiveness try to influence the giving of others. This includes:

• Research organizations that try to influence the allocation or use of charitable funds.

• Donor advisors who work with donors to find promising opportunities.

• People arguing to community members on venues like the EA Forum.

• Charity recommenders like GiveWell and Animal Charity Evaluators.

These are endeavors where you’re specifically trying to influence the giving of others. And when you influence the giving of others, you don’t get full credit for their decisions! You should only get credit for how much better the thing you convinced them to do is compared to what they would otherwise do.

This is something that many people in EA and related communities take for granted and find obvious in the abstract. But I think the implications of this aren’t always fully digested by the community. In particular, often, when looking at an intervention, it being highly cost-effective is less important than who paid for it — if the donor would have otherwise funded something similarly cost-effective, the intervention isn’t actually making that much difference on the margin.

As a quick demonstration, say as a donor advisor you have two options:

• Option 1: You can influence Big EA Donor to move $1,000,000 from something generating 95 units of value per dollar to something generating 105 units per dollar.

  • You’ve created ($1,000,000 * 105 - $1,000,000 * 95) = 10,000,000 units of value.

  • This is often what I expect a typical EA Forum post arguing that X animal welfare campaign is better than Y one, or similar, would achieve. I think the amount of money moved by these arguments is usually small, and you’re mostly refining between very good options (though of course, moving money across causes might change this equation substantially).

• Option 2: You can influence Big Normie Foundation to move $1,000,000 from something generating 0 unit of value per dollar (because it is useless) to something generating 10 units of value per dollar.

  • You’ve created the same amount of counterfactual value as the Big EA Donor example. This is just as much impact! Big Normie Foundation is still doing very ineffective charity, but it’s still a huge amount of counterfactual impact.

But many in the high impact charity community might celebrate “I got The Humane League to run more effective campaigns” as a much bigger win than “I got a normie foundation to give to something mildly better than what they were doing before, but still significantly worse than GiveWell,” even though the impact of both these actions might be the same.

I think a lot of this is that the community often pays too little attention to who is paying for a specific intervention. And I think paying attention to this more could have fairly substantial implications:

Impact is largely a function of what the donor would have done otherwise.

Most high-impact charity is funded by relatively few donors. For example, charities I’ve worked at have often received 80% or more of their funding from a single funder.

I think these large funders are often very thoughtful. Imagine I start a new project, and if these donors didn’t fund my project, they might have spent their funds on something else that was still highly impactful. It might have been worse than my project, but still would probably be pretty impactful. By my project existing, some other charity doesn’t grow quite as much, or otherwise doesn’t get as much funding.

The impact that my project can claim isn’t what we achieve — it’s what we achieve compared to what would have happened if we didn’t run the project. And, what would have happened is our donors would have given to another project, hopefully a less effective one (since otherwise our donors would be making a mistake to give to us), and the overall impact for the world would have only been a bit worse.

Is improving the use of effective or ineffective charitable dollars easier?

I don’t have a good answer to this question. There are lots of reasons for thinking improving effective philanthropy is easier than improving ineffective philanthropy:

• Effective philanthropy is motivated in part by being effective, so might be more motivated to change in response to cost-effectiveness arguments.

• Relatedly, ineffective philanthropy might have reasons for pursuing their work that are unrelated to cost-effectiveness, and these could be more difficult to argue against.

• Advocates for effective philanthropy might be better messengers for people motivated by impact than “normie” nonprofits.

But fundamentally, it seems like there are far more opportunities to impact ineffective philanthropy than effective philanthropy.

• If we think that charity impact massively varies, with the best opportunities being many orders of magnitude more impactful than the average, then there are correspondingly far more opportunities at lower impact levels.

• At a minimum, it seems likely that organizations that can generate 1 unit of value per dollar can absorb way more funding than organizations that can generate 100 units of value per dollar.

• It seems likely that there are way more organizations at the 10 unit level than the 100 unit level.

• So it might be significantly easier in principle to convince a philanthropist to move from giving at the 1 unit to the 10 unit level, even if not arguing on the basis of cost-effectiveness: there are just more opportunities to move a donor from 1 to 10 units of value than from 101 to 110.

How do people respond to these “lower impact” interventions?

I think that impact-minded people by default have a strong tendency to look at overall impact. But that’s not what the actual impact of an intervention is: projects are only as good as their counterfactual.

As an example (where I have some complicated conflicts of interest): here is an EA Forum post where a program is announced. The program is funded by an extremely non-EA funder (a New York City community foundation). The program is potentially somewhat effective (working on wild animal welfare, lots of animals potentially impacted!). The comments immediately raise the question of whether this is an impactful use of funds, since wild animals live in far greater numbers outside of cities than in them.

But “is this impactful?” is not the right question. A better question is “what’s the best possible thing that a foundation focused on improving New York City might fund?” Under those lights, it’s pretty amazing anything EA-related got funded at all. The funder seems to mostly fund culture and art programs in New York. Getting funding for something that might actually be mildly impactful should be seen as a huge victory: it’s plausibly far more cost-effective than convincing EAs to switch their donations between two very good opportunities.

To be clear, I don’t think these commenters are off-base in their criticism. I’m not sure if I believe the best wild animal welfare research opportunities are in urban settings. But if you asked me “what’s the best possible thing we could convince a foundation that can only give money to stuff in New York City to do?” then funding urban wild animal welfare programs is probably very high on my list, especially when the counterfactual is funding community gardens. The value of this project is a lot more complicated than “is this the best possible thing to do for wild animal welfare?” It’s measured by a question more like “how much better is this than what otherwise would happen?”

What are the implications of paying a lot more attention to funding counterfactuals?

I think that carefully paying attention to the counterfactual use of funds might end up having fairly radical implications for how many organizations approach their work:

• Donor advisors should pay more attention to how their advisees might otherwise give.

  • Efforts to influence the donations of EA advisees, or Anthropic donors, or other people who might already buy into high impact charity frameworks might be inherently less effective than influencing people who don’t buy cost-effectiveness arguments or otherwise are much less likely to donate to high impact charities.

  • Of course, it could be so much easier to influence people motivated by cost-effectiveness that this isn’t true, but I haven’t seen any formal analysis of the relative difficulty, or organizations doing donation advising trying to measure the counterfactual, besides Giving What We Can (to their credit).

• Charity evaluators should think about their impact as partially just “moving money around,” not counterfactual donations.

  • Charity evaluators who have an engaged audience of donors might, via changing recommendations over multiple years, be mostly engaging in “moving money around” between similarly cost-effective opportunities.

  • Charity evaluators are partially in the business of keeping these donors engaged (which is still valuable), but most of their value probably comes from their ability to get new donors engaged (and giving to more effective things)—it wouldn’t surprise me if most of the impact of a charity evaluator comes from engaging new donors, not from their evaluations being perfectly accurate (at least, past a certain point).

• People should often be excited about projects that move ineffective philanthropy toward more effective outcomes, even if those outcomes aren’t that effective.

  • I think there are likely many philanthropic efforts that just don’t look particularly good from an EA lens, but are probably as cost-effective as EA ones, primarily because instead of moving EA dollars between interventions, they move non-EA dollars to something significantly more cost-effective.

Objections to this argument.

There are lots of ways that these arguments probably wouldn’t hold. But a few that I think are most likely:

• Even among the best opportunities, there are massive differences in cost-effectiveness.

  • Especially for hits-based philanthropy, you might expect the difference in value between high impact opportunities to be extremely high, such that moving money between two high impact opportunities is actually as valuable as moving money from a non-high impact to a higher impact opportunity.

  • Similarly, it seems likely that across causes, cost-effectiveness is massively different (I think it’s pretty likely that different EA causes, for example, are many orders of magnitude different in cost-effectiveness). So there might be equally valuable work convincing animal welfare donors to give to reducing existential risk, or vice versa, depending on your views.

• The counterfactual use of talent is more important than the counterfactual use of funds.

  • The use of funds isn’t the only counterfactual—the use of talent also matters! In the wild animal welfare example in New York City, one could argue talented people are working on projects that might not be the most impactful thing they can do. Their counterfactual for their time isn’t doing worse things—it’s doing more impactful things! And if parts of the movement or certain roles are more talent-constrained than funding-constrained, the cost here could be greater than the impact of changing how the funding is used.

  • I think that this objection is fairly strong. It seems plausible to me that programmatically-focused people shouldn’t work on less impactful things, and that these arguments mostly apply to donor advisors, charity evaluators, and similar.

• There is more philanthropic money than there are highly effective opportunities.

  • In some spaces, (for example, animal welfare), I suspect that there are not many highly impactful giving opportunities. Single large funders might struggle to find good opportunities to give. If this is the case, than their marginal donations might be extremely impactful — the counterfactual might be them not giving the funding at all, or giving to something much much worse.

• This is all pretty impractical because looking at counterfactuals is too hard.

  • I think that counterfactuals are pretty hard to identify, and that knowing exactly what the impact of, say, a specific piece of donation advise is might be hard to identify due to this.

  • But, this doesn’t diminish the fact that improving the impact of ineffective charity by a marginal amount is just as good for the world as improving impact of effective charity marginally.

See also Marcus Davis and Kieran Greig’s work here for another implementation of this idea for research projects specifically.

Over the last few years, I helped run several dozen hiring rounds for around 15 high-impact organizations. I’ve also spent the last few months talking with organizations about their recruitment. I’ve noticed three recurring themes:

Candidates generally have a terrible time
Work tests are often unpleasant (and the best candidates have to complete many of them), there are hundreds or thousands of candidates for each role, and generally, people can’t get the jobs they’ve been told are the best path to impact.

Organizations are often somewhat to moderately unhappy with their candidate pools
Organizations really struggle to find the talent they want, despite the number of candidates who apply.

Organizations can’t find or retain the recruiting talent they want
It’s extremely hard to find people to do recruitment in this space. Talented recruiters rarely want to stay in their roles.

I think the first two points need more discussion, but I haven’t seen much discussion about the last. I think this is a major issue: recruitment is probably the most important function for a growing organization, and a skilled recruiter has a fairly large counterfactual impact for the organization they support. So why is it so hard for organizations to hire them?

Recruitment is high leverage and high impact

The majority of an organization’s impact flows through its recruitment function. Since most organizations in the high-impact charity sector are research and advocacy organizations, most of their impact is a direct function of the strength of their talent. I think that choosing who to hire is basically the second most consequential decision an organization makes for its overall impact (after choosing what to work on), and importantly, it’s a decision organizations are often willing to have early-career or more junior staff be involved in.

Additionally, basically all other consequential factors in an organization’s impact are downstream of recruitment:

• Is your strategy strong? It probably depends on hiring people who think well about strategy.

• Is your culture good? It probably depends on hiring strong managers.

• Is your organizational structure effective? It probably depends on hiring the right people for the right roles.

If a recruiter is directly involved in the assessment of candidates, they directly generate value for the organization based on the strength of their choices and assessments. But even recruiters who don’t directly evaluate candidates generate large amounts of value. They might screen initial candidates or help design assessments while a hiring manager makes advancement decisions. That is still a major factor in shaping the outcome of a hiring process.

And recruiters also reduce the time cost of hiring. Hiring is incredibly expensive for organizations. I’ll typically spend 50-60 hours in a given hiring process for a direct report, and expect a similar amount in total is spent by other people involved. If I run an organization hiring even a few roles per year, it’s easy to see the costs of hiring adding up. Recruiters can directly help reduce this burden, both through their work and through process improvements like designing more effective evaluations.

Organizations struggle to hire recruiters

Among hiring rounds I’ve run for organizations, two types of rounds have consistently failed to find strong candidates: leadership roles and recruiting roles. Leadership roles are understandable, as these hires are obviously very consequential for organizations, and there might be very few people who are a good fit.

My impression is that there is a gap, both in candidate pools and in service providers, of highly aligned, competent recruiting talent. When I’ve run hiring rounds for recruitment roles, organizations have been unhappy with the talent pool and often failed to make successful hires.

Why do organizations feel so unhappy with their recruitment talent pools? I think a major reason is that doing recruiting well, and especially managing a recruiting function, requires three core competencies:

Project management
Recruiters have to manage thousands of applications, handle individual emails to hundreds of people, and generally execute big, complex project management tasks. This is genuinely difficult and doing it well requires being thoughtful about communications, highly organized, and able to process large amounts of information quickly.

Decision-making under uncertainty
Hiring is inherently extremely uncertain. Our best selection methods aren’t that good, and organizations that hire according to theoretically optimal selection methods still regularly face performance issues in their staff that they shouldn’t expect if hiring perfectly predicted skills. And hiring is rife with uncertainty beyond “are you choosing the right person?”:

• Did you scope the role well? Is the role you scoped actually going to work the way you expect?

• Was your candidate pool competitive? Maybe the salary was too low or maybe you didn’t reach the right audience. You don’t know who didn’t apply, so it’s hard to directly assess whether you’re scoping the role and process well.

• Is anyone in the world actually a good fit for this role? The shorter the list of people who are actually qualified, the worse you should expect your candidate pool to be.

Ultimately, doing recruitment well is about successful project management, but it’s also about building tools to make decisions under uncertainty, and then testing and improving these tools over time. Doing this requires research skills, data analysis skills, and other competencies that aren’t usually considered part of the recruiter profile. The hiring tools organizations have tried (interviews, work tests, etc.) are pretty limited in the scheme of things. There is a ton of opportunity to try new methods for evaluating candidates, and for testing and improving those methods over time.

Understanding of the ecosystem
For organizations in specific cause areas, recruiters who understand their ecosystem can make a major difference. It’s unclear to me if this is a good thing, but in reality, organizations rely heavily on referrals from other organizations and signals about candidates’ context and value alignment. Having recruiters who can see those signals and interpret them, or decide when they’re worth ignoring, helps improve the evaluations organizations make of candidates.

For organizations, it might really matter, for example, that a recruiter recognizes that a background working on responsible scaling policy at Anthropic is meaningfully different from a background working at Pause AI, despite both presenting as organizations working to ensure that AI is developed safely. Or it might make a major difference to an animal welfare organization that a candidate came out of Direct Action Everywhere vs. Shrimp Welfare Project.

To be clear, I’m not convinced that using context clues like this is particularly wise for organizations to do (or at least, to use as a major point of assessment). But it is part of how groups approach recruitment, and having recruiters they can rely on to do the same is part of what they’re looking for.

These three skills are very different: candidates need both the operational skills of doing project management well and the research and empirical skills of trying to make good decisions under uncertainty. Finding someone with both of these who also has context on the ecosystem is even harder. Organizations looking for all three routinely fail.

Many of the people applying to recruitment roles emphasize their experience in recruitment. This isn’t the background organizations need

When hiring recruiters, usually the hiring pool is made up of many people who have a high degree of project management and operational skills. Candidates tend to be operationally focused and emphasize this part of their experience and background.

But often candidates won’t have either the research/empirical skills to explore ideas to improve hiring or won’t be interested in this aspect of it. And often organizations are looking for candidates with significantly higher context on the ecosystem they’re in than the candidates have.

The differentiator for organizations in hiring recruiters seems to usually be someone having all of these skills, not just the operational ones. Finding people at that intersection is much harder.

Almost no one is appropriately obsessed with hiring

The hiring tools organizations have are really rudimentary. Basically, organizations have a brief scan of resumes, some answers to short-answer prompts, a few interviews, and a few hours sampling candidates’ work. Even organizations with great recruiting practices routinely struggle with performance issues internally. And organizations I’ve spoken with have struggled to automate (or don’t trust the automation of) large parts of their review, meaning hiring anyone requires dozens of hours of reviewing materials of candidates who won’t be advanced. All of this suggests that there are major gains to be had from trying to improve hiring at organizations.

But my impression is that most organizations aren’t focused on things like experimenting with their hiring, running tests to try to improve the impact of their approaches, evaluating candidates they rejected to see if they might have made mistakes, or otherwise doing the basic steps it might take to make hiring more effective.

Given the importance of hiring, this lack of obsession is surprising. Hiring successfully is a critical part of organizational impact, but barely any time is spent trying to deeply understand how to do it well.

High-impact organizations need people who are completely obsessed with hiring. By this, I mean people obsessed with answering a question like, “How do I attract the right pool of people to a role and identify the best candidates out of tens of thousands, in a way that is minimally onerous for both candidates and my organization?”

I think that there is a huge opportunity for people who are obsessed with these kinds of questions. Answering them well unlocks a huge amount of impact for organizations, and we genuinely don’t have answers for them.

The state of evidence on hiring practices is bad

A lot of the reason that organizations struggle with hiring is that the evidence base for how to assess people is terrible.

There isn’t a (cheap) way to run true RCTs on hiring and evaluating hiring rounds often faces a fundamental statistical issue: in progressive rounds, you eliminate the people who did poorly in your assessment. But doing well in a hiring round is a separate skill from doing well in a job, so you’re inherently filtering on a different skill set, to some extent, than the one you want to measure. So you will never get information on the relative performance of people you rejected and be able to make confident claims about the predictiveness of your hiring.

That being said, we can still collect some evidence from hiring rounds: industrial/organizational psychologists have tried to study what hiring methods are most predictive of later performance. While all these assessments suffer the issues outlined above, we can still glean something from them. But unfortunately, the evidence they produce is mixed at best:

• The best methods of assessing candidates are at best only moderately correlated with later job performance. While presumably you can improve this by assessing candidates using multiple methods, if you believe, like I do, that the distribution of talent is extremely right-tailed, then moderate correlations just aren’t going to be particularly useful for finding the most talented people. You can do a good job eliminating candidates who aren’t a great fit, but if talent follows a right-tailed distribution, confidently determining who the best candidate is is much more important and much harder.

• The field suffers from the statistical issues of all psychology. For example, for years, general mental aptitude (GMA) has been thought to be highly correlated with job performance. Then, in 2021, some researchers claimed that a major study underpinning this claim had (in their view) made some dubious statistical choices which, when corrected, led to GMA being a much worse indicator of job performance. Without weighing in on the debate, that it is a major debate indicates that these questions are not settled even among the researchers who have spent their lives studying them.

Separately from choosing the right evaluation method, organizations have to design the evaluation well. If I am hiring for a software engineer and have them swim a mile for their work test, I’m fundamentally not testing the right skills for the job. Recruiters have to answer difficult questions like:

• How do I even figure out what the right skills for a job are?

• What do I do when these skills are in conflict with other things I care about? Many of the highest-performing people I know work really long hours. Is “working really long hours” actually a skill I want to filter on? Or is it one of many potential proxies for something else that matters more?

Retaining strong recruiters is really hard

Many of the people who are best at recruiting and convincing talented people to join their organizations aren’t in recruiting roles. Instead, they are in leadership or research roles. And it often seems like convincing someone to stay in a recruiting role, especially if they have research skills, is difficult. But this seems like a mistake — finding recruiters with the data analysis skills, research literacy, and operational skills to run experiments, improve systems, project-manage hiring rounds, and overall make the recruitment experience positive for candidates is really difficult.

Organizations do sometimes find these people. However, many of the best recruiters don’t stay in these positions for long. Instead, they often move into management, leadership, or research roles. This exacerbates the issue: many of the people with the most experience in running effective recruitment don’t actually work in recruitment. The skills that make someone excellent at recruitment might also make them an excellent researcher or excellent project manager. And because recruiting isn’t seen as particularly high status, people doing incredibly impactful work often move out of it to other roles.

I’m not sure what to do about this, but currently, my best guess is that recruiting should be viewed as having significantly higher impact than most people think it has, so there is greater social reward for staying in these roles.

Why might this be less important than I think?

High-impact organizations are usually pretty good at hiring, so finding better recruiters doesn’t make that much of a difference

I think that most high-impact organizations are doing very reasonable things in their hiring: they’re running multiple kinds of assessments (e.g., work tests, structured and unstructured interviews, work trials, etc.). Relative to other nonprofits, most organizations I work with in EA, AI governance, etc., are doing a good job at hiring.

If organizations are pretty good at hiring well, maybe there isn’t that much to be gained from hiring a recruiter, or recruiting talent would have the greatest counterfactual impact by going to organizations with the worst recruiting practices.

I suspect this is wrong though. I have rarely seen organizations do things I think are fundamental to successfully evaluating and improving their recruitment, like:

• Conducting job analyses

• Evaluating the later performance of candidates they rejected

• Backtesting their recruitment practices to see if they predicted later performance

This is too hard

I don’t think that the issues with recruitment I describe here are limited to high-impact philanthropy. I suspect that major for-profits, who theoretically should have heavy incentives to find the best talent, also struggle with them.

This indicates to me that there are just hard problems in doing this well. We might only make marginal improvements from the current best approaches.

I find this pretty convincing and think it’s possible I’ll look back at this post in 10 years and think it was misguided.

But again, I’ve not seen anyone try to just take recruitment as seriously as I think it deserves to be taken. I think given the stakes, it deserves a shot.

I’m trying to find people interested in this kind of approach to hiring. If this is you, please reach out.

This post isn’t meant as a pitch, but I’m trying to find people who are interested in exploring roles in recruitment, who are obsessed with hiring, or are interested in taking a highly empirical/experimental approach to hiring. If this is you, get in touch! I’m both looking for candidates for my own work and am regularly asked for referrals for recruiting roles.

Nothing in this document is intended as legal advice. We are not attorneys, and you should always seek legal counsel on compliance decisions. This post is meant as a starting point for thinking about using EORs, not a comprehensive or exhaustive guide to working with or evaluating them.

What are employers of record?

Employing via an employer of record (EOR) is an arrangement where your company, based in Country A, contracts the EOR to hire staff in Country B, who are seconded to work on your projects. Functionally, you’re contracting the EOR to provide you with staff to support your projects, and you pick those staff, how much they are paid, and within certain boundaries, their benefits.

This model allows companies in one country to hire in another country without establishing a local entity, doing compliance registrations or filings, etc. It makes it relatively easy to have staff globally.

Employers of record are sometimes called PEOs (professional employment organizations), but it is worth noting that in the US, the phrase “PEO” refers to a separate model of co-employment where some compliance and administrative functions are outsourced, including the named employer on W-2s. This US PEO is very different, and the benefits and downsides outlined in this document do not apply to US PEOs — I am specifically using EOR to not cause this confusion. EORs are not a co-employment model — the EOR is the sole employer, and the company contracts the EOR for services.

Many employers of records also offer international contractor support. These services are usually less useful, but they might provide some functionality for dealing with foreign bank accounts (e.g. US nonprofits often have trouble sending funds to many Indian banks, but the EOR is typically a for-profit, so won’t have this issue), even if they are just passing through. This guide specifically looks at using EORs for employees, not independent contractors.

Why use an employer of record?

Using an EOR has a lot of advantages for an organization with a limited footprint in another country. By paying a relatively premium fee, you get several benefits.

• You don’t have to do local employment registrations

  • The employer of record handles all local employment registrations and filings on behalf of your staff. You don’t have to know anything about these, and don’t have to deal with them.

• You (theoretically) don’t have to deal with local employment compliance

  • EORs will often tout themselves as handling local compliance. This is true when things are going well, but because EORs pass all employment costs on to you, if the EOR makes a compliance mistake you’ll likely have to pay for it. See more in Risks below.

  • EORs can also handle visas in some countries, which can be a significant benefit when hiring.

  • Some countries make it pretty difficult for residents to be contractors. In these countries, often the only way to hire someone is via an EOR.

• You don’t have to shop for benefits in the new country

  • Since EORs usually offer relatively normal benefits in each country they support, you can give people a typical benefit package for the country they are in.

• You can give foreign staff a relatively “normal” employment experience

  • Since staff will have normal paychecks, tax withholding, and benefits, their employment experience under an EOR will often feel “normal” relative to being an independent contractor, where they’ll have to deal with their own tax withholding, etc.

• You get access to global talent

  • The biggest upside of EORs, especially because you can offer a normal employment experience, is that they let you hire people in more countries, which makes your talent pool wider.

What are the risks of using an employer of record?

Relative to hiring normally in the US, EORs seem to pose a ton of risk. They are expensive, and don’t actually protect your business that much from compliance risk in the countries they hire in.

• Country-specific risks

  • When you hire staff in a new country, the EOR will handle compliance filings and ensure there is a legal contract in place. But, it won’t protect you from employment compliance issues.

  • If you try to, for example, fire the staff member for reasons that the EOR perceives as illegal, it might stop you (and you’ll be on the hook for continuing to pay their salary).

  • The EOR usually won’t tell you much about local employment compliance, so you’ll still have to learn about it yourself to protect your business.

  • The employer tax rate might vary widely in other countries (e.g. in some it is as low as 5% of salary, in others as high as 40%). You’ll pay this rate.

  • You’ll be subject to other local risks, like permanent establishment risk, collective bargaining agreements, or risks when terminating staff. The EOR usually won’t inform you much about risk from these.

  • Different countries might have different benefit requirements than you’re used to. The EOR might not make you aware that a specific country requires offering a certain amount of parental leave, etc.

  • How easy it is to terminate employees varies widely by country. This is especially surprising to US employers, who might be used to at-will employment. In some European countries, for example, employers need to go through a court process to terminate a contract for performance-related reasons. In some countries, termination is often handled through negotiating with the employee a payment to quit. The EOR usually doesn’t discuss this risk with your company.

• Asymmetry in handling disputes

  • Despite saying that they handle compliance, typically EORs pass all employment costs on to you. If the EOR is fined for inappropriately terminating one of your staff, you’ll pay the fine. If the EOR fails to file something for the staff, you’ll pay the fine. If an employee sues the EOR, you’ll pay any costs.

  • Because the EOR does not pay these fines, it isn’t incentivized to minimize them. The EOR wants to end any disputes as quickly as possible, and pass the cost on to you. This means they’d usually rather pay more to exit the dispute quickly, rather than pay less.

  • The EOR’s lawyers are minimizing the EOR’s risk, not yours.

    • Because of this, if you ever have an employment dispute for an employee under an EOR, you’ll either have to accept paying a massive premium to work with the EOR, or hire your own local employment lawyer to assist and navigate the situation, and provide feedback to the EOR on how to proceed.

      • The EORs will usually listen to your outside counsel, but they don’t have to, and could choose to handle the situation in a different way. You’re exposed to the risk from their decision-making, and again, they don’t want to minimize costs but want to exit the dispute quickly.

  • EORs do not protect you from local employment risk – they make it more severe by putting a negotiation partner between you and the risk that isn’t incentivized to minimize your costs. They might facilitate you hiring in another country, but they aren’t meaningfully reducing risk usually because of this.

• Currencies

  • If you hire staff who are paid in a different currency, the EOR will usually want to charge you in your local currency (e.g. you hire staff in Europe, but are based in the US – you pay USD and they are paid EUR).

    • This means that without complex financial maneuvers, like regular rebalancing of currencies you hold, you can’t hedge against currency risk easily while using them.

    • EORs often give worse-than-market rates for currency conversions, so you may pay an extra premium beyond their fee to use them.

• Collective bargaining agreements

  • In many countries, most employees are under sectoral collective bargaining agreements. This is especially unusual for companies used to employment in the US/UK to encounter, so it bears worth mentioning.

  • The union the employee is part of might negotiate additional benefits or raises throughout the year that you will be subject to. For example, if the union negotiates a raise for all its staff, your costs might randomly increase, or you might have to break a pay system you established. Or, you might have to offer benefits in certain countries that you wouldn’t or can’t offer elsewhere.

    • You will never be party to these negotiations, so have no agency over how they go and the outcomes from them.

• Legal risk

  • EORs are legally grey in many countries! Pre-COVID, this was a relatively small industry. During the pandemic, especially via a few startups exploding in popularity, they’ve become much more widely used. But, many countries have not really worked out how they want to regulate EORs in the long run. This means that the regulatory frameworks governing EORs might change over the next decade, and you may lose the ability to hire someone through one in some countries.

  • Some EORs seem to take unusual risks with employment law — for example, they might offer visas or other benefits that it is ambiguous they are legally able to. Often, these companies are startups who are pushing the boundaries of compliance to make their products better. But, the nature of their business model is a lot of the downside risk (at least in terms of fines) is passed on to you!

• Permanent establishment risk

  • While EORs might help you handle employment filings, they don’t necessarily mean you escape the risk of a country’s government deciding you are “doing business” in that country.

  • You might face the risk of being seen as having an establishment in a foreign country (and thus being required to incorporate there and register with tax agencies).

  • This risk might be higher if you’re using the EOR to hire senior staff, if you generate a lot of revenue in the foreign country, or if you have offices or other indicators that you don’t just have a few remote employees there.

• Reliance on subcontractors

  • To offer employment in many countries, EORs either need a lot of subsidiaries and legal entities, or will use subcontractors in each country who then serve as an EOR for the EOR.

    • In countries where they are using a subcontractor, all of the risks here are magnified – there are multiple legal entities between you and your employees who do not have aligned incentives with you or each other when there is a dispute.

  • It seems like a consistent issue is that when EORs subcontract their contracts, the subcontractor provides worse service.

  • You can ask your EOR directly when you start working with them if they use subcontractors, or if your employees will be employed by an entity they control.

• Reliability

  • EORs are trying to run payroll in many countries (sometimes over 100). Many of the most popular ones are startups. This is not an easy task — some have to operate under over a hundred employment law frameworks, etc. Even the best ones make a lot of mistakes — far more than the average US payroll provider, for example. Expect that over time, employees might occasionally not be paid on time or have paperwork filed incorrectly, even if you’re working with the best-reviewed ones.

What should you look at when evaluating hiring in a new country?

EORs offer the ability to hire staff in most countries, but this doesn’t mean you should hire staff in all those countries. The more countries you are operating in, the more risk your organization is taking on. You’re working under more legal frameworks, potential employment issues, etc.

In general, it seems good to limit the number of countries you hire in to the countries where most of your talent pool is. If there are lots of talented people for a role in the US, you probably don’t need to hire in the UK. Often, it might be better from a risk perspective to identify the 2-3 countries that will give you access to 80% of the best people in a field, rather than hiring in 10 countries to get to 90%.

But, if you are planning on hiring staff in a new country, here are some things to consider:

• Can you afford the employer tax rate?

  • Employer tax rates vary wildly by country – some are as high as 40%. Before choosing to hire in a specific country, make sure you can afford this tax burden.

• Do termination laws align with your needs?

  • The ease with which staff can be terminated or laid off varies to a significant degree by country. Even with the EOR, you’ll be subjected to these rules. Some countries require layoffs for financial reasons to be done in order of seniority. Some require courts to be involved in routine terminations. Some require a high degree of transparency when terminating staff.

  • If you’re based in the US, you should generally assume the ease of firing people in the US is the exception, not the rule. In every other country, your employees will have contracts that govern their employment, with statutory notice periods for terminations, and usually much stronger employment protections.

  • Before hiring in any country, make sure you know what you’re getting into, and consider mitigation measures like probationary periods that provide you with more options early on during an employee’s tenure.

• Are you okay with a collective bargaining agreement (CBA)?

  • In some countries, all employees within an EOR might be under a collective bargaining agreement. Make sure you understand the terms of that CBA if you are subject to one, and understand historically how the CBA has changed costs for employers.

  • In some countries, CBAs will have negotiated specific benefits (e.g. more leave, workplace perks, etc.). Going into the employment engagement, you should know what these are, and feel able to pay them.

• Can you afford the statutory benefits?

  • The US has minimal required statutory benefits — some states require minor amounts of paid family leave or sick leave, but overall these benefits are slim. Other countries have significantly more.

  • In particular, many countries have mandatory vacation leave, parental leave, and other types of leave.

    • Some of this leave in many countries, such as parental leave, is covered by social insurance programs. But often, the employer is required to cover part or all of it.

    • This means that employing outside the US has hidden costs that need to be accounted for, and which won’t be realized until someone takes a leave.

• Can you hire someone just as good in a country you’re already working in?

  • Generally, the compliance uncertainty and risk of hiring outside your own country probably means that unless the hiring pool is significantly improved, it might be prudent to not hire internationally.

How to use an employer of record

If you are proceeding with using an employer of record, I suggest taking the following steps:

• Identify the countries that are most important to hire in for your roles, and make sure their employment framework is going to work for you.

  • I think it is highly risky or a major mistake to significantly increase the number of countries you’re open to hiring in — instead, a better approach might be to identify the short-list of countries that the majority of people you want to hire are located in, and evaluate each of them individually on criteria like:

    • Employer tax burden

    • Ease of terminations

    • CBAs that staff will be under within an EOR

  • Generally, you should make sure that you feel at least somewhat comfortable with the employment framework you’re entering into.

    • Do you understand how employment disputes and terminations are typically resolved? Do you feel comfortable navigating that if such a dispute arises for you?

    • Do you have a sense of customary norms for employment that might not be obvious and are easy to accidentally break (e.g. 13th/14th month salaries in Europe and South America, meal cards in Brazil, anti-humiliation laws in New Zealand)?

• For each country you’re considering hiring in, verify that your EOR is not using subcontractors.

  • It’s probably worth getting referrals from another organization that has used that EOR in that country, to see if there have been major issues.

• Prior to signing up with the EOR, see if your Master Services Agreement has an exclusivity clause, and make sure that you understand it.

  • Some EORs require your company to work exclusively with them, or might have clauses that prevent you from exiting the EOR relationship and hiring the staff directly. This is inherently limiting for your organization, especially if you later choose to open a local entity due to determining you have a permanent establishment. It might be worth verifying that your MSA does not have this clause.

• Make sure that the employee understands the EOR arrangement.

  • The employee is going to get a contract that doesn’t say “your company name” and instead says “EOR name.” They are also going to have to provide documents to the EOR, talk with the EOR’s HR staff, etc.

    • Make sure that they fully understand the arrangement as well, so none of this is surprising.