I think many of those points have force, which is part of why I generally favor only partial COL adjustment. I tentatively agree with you that orgs should generally consider the actual cost of employing each individual when making a hiring decision, such that the candidate in a lower-COL location will usually have an advantage.[1]
Working at EA organizations which offer cost of living adjustments becomes more attractive to people who need them, and less attractive to nomads or internationals
The implied model is my head is that the prospective employee will accept a certain amount of sacrifice, but not more than that. A fully unadjusted salary expects candidates with sticky ties to a higher-COL location to sacrifice more, either by breaking those ties or by accepting a much deeper drop in standard of living / consumption than candidates in other locations. It will lead to many of those candidates opting out.
Given the current distribution of EAs, unadjusted salaries would take many candidates off the table, and in many cases that would be pretty problematic. I’m thinking of an analogy to non-EA nonprofit salaries. As people have pointed out elsewhere, often these are set at a level where only those who rely on a wealthier partner or parent to provide the financial support are able to stay in those positions long-term. That’s a shrewd strategy for non-EA nonprofits if their applicant pool is deep, and having the third-best rather than the best candidate in a position doesn’t make a huge difference. I’m not convinced that is presently the situation in most EA orgs.
The org could, of course, pay salaries high enough to keep SF-based candidates in the pool no matter where the employee was located. But its ROI for doing so does not seem high.
By taking into account previous history considerations, you are incentivizing people to create those considerations so that they will be taken into account in the future.
I think many of those considerations are created by an individual’s pre/non-EA life. My sense of where feels like “home” is influenced by where I grew up. If one of my goals is for my son to have a deep relationship with his grandparents, the grandparents are located where they are and the costs of semi-frequent travel are what they are. These considerations are not “create[d]” by the candidate in a way that is responsive to incentives. Moreover, unless someone immediately lands an EA job, their location incentives are more likely shaped by the general employment market in their field—not by EA employers.
I do have some concerns that this could create bidding dynamics that could be unhealthy. Does the org go back to the # 1 candidate and say that the calculator gives $100K for the low-COL candidate, we like you 10% better, can you work for $109K even though the calculator says $130K for your location? Would it then go to the # 2 candidate and give them a chance to undercut the # 1 candidate by working for less?
FWIW: The CoL adjustment is the thing with the widest spread of views within the pay survey that we did. I think that we’ve arrived at near the least bad option for our team at the moment. An employee location does factor into some roles more than others, for example if someone is likely to be donor facing it is much more valuable to the organisation (and worth paying more) for them to be located in places like SF, NYC, London, Sydney etc where we have a high number of (potentially larger) donors.
That’s a great point. I would think that a position that required the employee work in a certain location should have a 100 percent locality adjustment, as the employee’s location is determined by employer needs rather than employee choice.
It gets more interesting where there is some benefit to the employer, but not enough to strongly prefer or require a location or locations. E.g., does an employee being located in the Bay serve an AI safety org’s interest to some extent, due to easier ability to collaborate with others doing similar work in and out of work hours?
Yep. It’s complicated and needs to be tailored to the organisation and it’s needs at the time. So far all our roles have been remote but we have seen an employees location as something that might have some cost/benefit to consider but not a requirement.
I think many of those points have force, which is part of why I generally favor only partial COL adjustment. I tentatively agree with you that orgs should generally consider the actual cost of employing each individual when making a hiring decision, such that the candidate in a lower-COL location will usually have an advantage.[1]
The implied model is my head is that the prospective employee will accept a certain amount of sacrifice, but not more than that. A fully unadjusted salary expects candidates with sticky ties to a higher-COL location to sacrifice more, either by breaking those ties or by accepting a much deeper drop in standard of living / consumption than candidates in other locations. It will lead to many of those candidates opting out.
Given the current distribution of EAs, unadjusted salaries would take many candidates off the table, and in many cases that would be pretty problematic. I’m thinking of an analogy to non-EA nonprofit salaries. As people have pointed out elsewhere, often these are set at a level where only those who rely on a wealthier partner or parent to provide the financial support are able to stay in those positions long-term. That’s a shrewd strategy for non-EA nonprofits if their applicant pool is deep, and having the third-best rather than the best candidate in a position doesn’t make a huge difference. I’m not convinced that is presently the situation in most EA orgs.
The org could, of course, pay salaries high enough to keep SF-based candidates in the pool no matter where the employee was located. But its ROI for doing so does not seem high.
I think many of those considerations are created by an individual’s pre/non-EA life. My sense of where feels like “home” is influenced by where I grew up. If one of my goals is for my son to have a deep relationship with his grandparents, the grandparents are located where they are and the costs of semi-frequent travel are what they are. These considerations are not “create[d]” by the candidate in a way that is responsive to incentives. Moreover, unless someone immediately lands an EA job, their location incentives are more likely shaped by the general employment market in their field—not by EA employers.
I do have some concerns that this could create bidding dynamics that could be unhealthy. Does the org go back to the # 1 candidate and say that the calculator gives $100K for the low-COL candidate, we like you 10% better, can you work for $109K even though the calculator says $130K for your location? Would it then go to the # 2 candidate and give them a chance to undercut the # 1 candidate by working for less?
FWIW: The CoL adjustment is the thing with the widest spread of views within the pay survey that we did. I think that we’ve arrived at near the least bad option for our team at the moment. An employee location does factor into some roles more than others, for example if someone is likely to be donor facing it is much more valuable to the organisation (and worth paying more) for them to be located in places like SF, NYC, London, Sydney etc where we have a high number of (potentially larger) donors.
That’s a great point. I would think that a position that required the employee work in a certain location should have a 100 percent locality adjustment, as the employee’s location is determined by employer needs rather than employee choice.
It gets more interesting where there is some benefit to the employer, but not enough to strongly prefer or require a location or locations. E.g., does an employee being located in the Bay serve an AI safety org’s interest to some extent, due to easier ability to collaborate with others doing similar work in and out of work hours?
Yep. It’s complicated and needs to be tailored to the organisation and it’s needs at the time. So far all our roles have been remote but we have seen an employees location as something that might have some cost/benefit to consider but not a requirement.