Thanks for making the ex ante versus ex post distinction. But it makes me confused about the penultimate paragraph; if I am offered an job at an org and am comparing to earning to give, shouldn’t I be using the (currently unpublicised) ex ante numbers, not these ex post numbers?
The risks of bad personal fit, costs of senior staff time, costs of fast hiring in general, and time taken to build trust are all still in the future at the point, and don’t apply to the earning to give alternative. As far as I can tell, the only cost which has already been sunk at that point is the cost of evaluating me as a candidate. In my experience of working on the recruiting side of a non-EA org, this is far smaller than the other costs outlined, in particular the costs of training and building trust. I’m curious if the EA orgs feel differently.
In general though, I don’t think attempting to save these numbers by pointing out how hiring is subtly much more expensive than you would think interacts much with my objection to these numbers, since each additional reason you give me for why hiring is wildly expensive for EA orgs is yet another reason to prefer earning to give, precisely because it does not impose those costs! All these reasons simply mean the numbers should be lower than what you get from an ex post phrasing of the question, at least insofar as they are being used for what appears to be their intended purpose, namely comparison of options.
I wrote ‘most relevant’ to contrast that group with people who know they’ll never be able to do direct work at the organisations surveyed, for whom those figures are largely irrelevant. As you say, they’re still not quite right for that group because you’re only part way through the process (I think by the time you’re offered a job, about half the costs have been paid, though that depends on whether you had a trial period).
They would be even more relevant to someone considering quitting—but as they can talk to their colleagues, they probably wouldn’t be using this survey.
The costs of hiring makes applying to join a project look worse, but it also makes earning to give to fund the salaries of new staff look worse too. If you’re expecting to donate hoping the money will be used to fund new hires, it seems like it should cancel out.
The three ways around bearing those costs that I can see at first glance are i) earning to give to prevent a group from laying off staff for lack of funding, ii) earning to give to buy capital goods rather than hire people, iii) staying in your current direct work job rather than switching.
Since most of the EA orgs in question are heavily constrained in hiring by whatever level of growth they can manage or feel comfortable with (that’s kinda the whole point of the OP, right?), it would not generally be my assumption that additional funds would be used for extra hiring compared to the counterfactual. I grant that if that is the assumption, these effects seem to cancel.
Other ways not listed in your last paragraph.
-earning to give to allow orgs to raise salaries
-earning to give to fund regranting
-earning to give to fund things like targeted advertising (you may have intended to cover this category in ‘capital goods’, I’m not sure)
These things are much closer to my model of where extra funding to at least CEA and 80k in at least the past 18 months has gone, not into additional hiring.
A quick note on another aspect that I think Rob might have underemphasised:
Hiring could generate very high returns to the organisations but often not as high as other marginal activities.
Here’s a simple model: If you think that many EA orgs spend about $1m per year and generate $10m per year of “value” (as rough orders of magnitude), then even if a marginal hire generates $1m “excess” value to the org, that’s only a 10% increase in impact for that year. This is good but there could easily be even more important things for senior management to focus on.
This would explain why the orgs don’t hire many people.
However, it would still be consistent with the idea that marginal hires are valuable and can have more impact by working at the org than by earning to give, since each generates $1m.
It would also mean conditional on an org running a hiring process, it’s better for that process to go as well as possible and get the best possible applicants.
In that simple model, it appears to me that marginal hires are worth $1m minus the counterfactual use of senior staff time (hypothetically, and relevantly, suppose all the possible hires for a position decided to earn to give instead overnight. It would not be the case that the world was $1m worse off, because now senior staff are freed up to do other things). If there are in fact even more important things for senior management to focus on, this would be a negative number.
More realistically, we could assume orgs are prioritising marginal hiring correctly relative to their other activities (not obvious to me, but a reasonable outside view without delving into the org particulars I think), in which case the value of a marginal additional hire would simply be ~0.
So again, I appreciate the attempt to boil down the area of disagreement to the essentials, and even very largely agree with the essential models and descriptions you and Rob are using as they happen to match my experience working and recruiting for a different talent-constrained organisation, but feel like this kind of response is missing the point I’m trying to make, namely that these ex post numbers, even if accurate on their own terms, are not particularly relevant for comparison of options.
I agree with the thrust of your point: if hiring is very costly, then that’s reason against working at EA orgs vs etg.
It still seems like there could be a world in which hiring is high return but often (but not always) lower return than some other marginal activities. This will mean both that (i) recent hires are valuable (ii) the orgs don’t hire many people (iii) hiring is sometimes worthwhile. I was just trying to show how these could be consistent.
Going back to the overall issue of why the dollar figures are high, I don’t think the “hiring is costly” point is the main thing going on, and think this post might have emphasised that too much. The post was more trying to answer the question of why the orgs don’t hire more often despite the high dollar figures. The broader question of why the figures are high and what to do based on them is discussed briefly in the original article about the survey, but hasn’t yet been tackled in depth.
With your first two paragraphs, I just want to step back and point out that things get pretty confusing when you include all opportunity costs. When you do that, the return of every action except the single best action is zero or negative. Being close to zero is actually good. It’s probably less confusing to think in terms of a ranked list of actions that senior staff could take.
I also expect the orgs partially take account of the opportunity costs of staff time when reporting the dollar value figures, though it’s hard to be sure. This is why next year we’ll focus on in-depth interviews to better understand the figures rather than repeating the survey.
With your first two paragraphs, I just want to step back and point out that things get pretty confusing when you include all opportunity costs. When you do that, the return of every action except the single best action is zero or negative. Being close to zero is actually good. It’s probably less confusing to think in terms of a ranked list of actions that senior staff could take.
True, but I haven’t accounted for all the opportunity costs, just one of them, namely the ‘senior staff time’ opportunity cost. If you are in fact close to 0 after that cost alone (i.e. being in a situation where a new hire would use x time and generate $1m, but an alternative org-improvement action that could be taken internally would generate $950k), that isn’t ‘good’, it’s awful, because one of those actions incurs opportunity costs on the applicant side (namely, and at the risk of beating a dead horse, the cost of not earning to give), but the other does not.
So we could look at this as a ranked list of potential senior staff actions, but to do so usefully the ranking needs to be determined by numbers that account for all the costs and benefits to the wider world and only exclude senior staff time (i.e. use $1m minus opportunity cost to applicant minus salary minus financial cost of hiring process per successful hire etc.), not this gross $1m number.
Similarly, potential applicants to EA orgs making a ranked list of their options should include all costs and benefits that aren’t tied to them, i.e. they should subtract senior staff time from the $1m number, if that hasn’t been done for them already. Which is what I’ve in fact been recommending people do. But my experience is that people who haven’t been directly involved in hiring during their career radically underestimate the cost of hiring, and many applicants fall into that camp, so for them to take account of this is not trivial. I mean, it’s not trivial for the orgs either, but I do think it is relatively easier.
I also expect the orgs partially take account of the opportunity costs of staff time when reporting the dollar value figures, though it’s hard to be sure. This is why next year we’ll focus on in-depth interviews to better understand the figures rather than repeating the survey.
Given this conversation, I’m pretty skeptical of that? My experience with talking to EA org leaders is that if I beat this horse pretty hard they back down on inflated numbers or add caveats, but otherwise they say things that sound very like ‘However, it would still be consistent with the idea that marginal hires are valuable and can have more impact by working at the org than by earning to give, since each generates $1m.’, a statement it sounds like we now both agree is false or at least apt to mislead people who could earn to give for slightly less than $1m into thinking they should switch when they shouldn’t.
For the benefit of third parties reading this thread, I have had conversations in this area with Ben and other org leaders in the past, and I actually think Ben thinks about these issues more clearly than almost anybody else I’ve spoken to. So the above paragraph should not be read as a criticism of him personally, rather a statement that ‘if you can slip up and get this wrong, everybody else is definitely getting this wrong, and I speculate that you might be projecting a bit when you state they are getting it right’. The only thing Ben personally has done here is been kind enough to put the model in writing where I can more-easily poke holes in it.
I’m feeling a bit set upon here. I was talking about a different topic (why people hire slowly) and it seems like we got our wires crossed and are now debating the value of a marginal hire. Looking back, I see how my earlier comment led us off in a confusing direction.
If we’re discussing the value of a marginal hire, I totally agree that the survey figures DON’T include the costs of hiring someone in the first place. That’s why I brought up the ex-ante ex-post distinction in the first place.
This means that someone considering working at an EA org should use a lower figure for the estimate of their value-add (we agree). In particular, they should subtract the opportunity costs of the time spent hiring them. (This varies a lot on the situation, but one month of senior staff time seems like a reasonable ballpark to me.)
However, just to be clear, I don’t think they should subtract the opportunity costs of senior staff time spent on their on-going management, since I think the natural interpretation of the survey question includes these. (If a new hire could raise $1m of donations, but would take up management time that could have raised $800k otherwise and has a salary of $100k, it would be odd for the org say that they’d need to be compensated with $1m if the hire disappeared. Rather, the answer should be $100k. I expect most orgs were aiming to include these costs, though of course they might not have made a good estimate. This is what I thought you were talking about when I said I think orgs partially take opportunity costs into account.)
Where does this leave the survey figures? We could super roughly estimate the value of a month of senior staff time at an org is 5x the value of a month of junior staff time, so this would super roughly reduce the value of junior hires over three years by 5⁄36 = 14%.
Re. your first paragraph, I don’t know why you chose to reply to my comment specifically, since as far as I can tell I’ve never been asking ‘why do people hire slowly’.
I think I’ve already explained why I don’t agree with your later paragraphs and see little value in repeating myself, so we should probably just leave it there.
Thanks for making the ex ante versus ex post distinction. But it makes me confused about the penultimate paragraph; if I am offered an job at an org and am comparing to earning to give, shouldn’t I be using the (currently unpublicised) ex ante numbers, not these ex post numbers?
The risks of bad personal fit, costs of senior staff time, costs of fast hiring in general, and time taken to build trust are all still in the future at the point, and don’t apply to the earning to give alternative. As far as I can tell, the only cost which has already been sunk at that point is the cost of evaluating me as a candidate. In my experience of working on the recruiting side of a non-EA org, this is far smaller than the other costs outlined, in particular the costs of training and building trust. I’m curious if the EA orgs feel differently.
In general though, I don’t think attempting to save these numbers by pointing out how hiring is subtly much more expensive than you would think interacts much with my objection to these numbers, since each additional reason you give me for why hiring is wildly expensive for EA orgs is yet another reason to prefer earning to give, precisely because it does not impose those costs! All these reasons simply mean the numbers should be lower than what you get from an ex post phrasing of the question, at least insofar as they are being used for what appears to be their intended purpose, namely comparison of options.
I wrote ‘most relevant’ to contrast that group with people who know they’ll never be able to do direct work at the organisations surveyed, for whom those figures are largely irrelevant. As you say, they’re still not quite right for that group because you’re only part way through the process (I think by the time you’re offered a job, about half the costs have been paid, though that depends on whether you had a trial period).
They would be even more relevant to someone considering quitting—but as they can talk to their colleagues, they probably wouldn’t be using this survey.
The costs of hiring makes applying to join a project look worse, but it also makes earning to give to fund the salaries of new staff look worse too. If you’re expecting to donate hoping the money will be used to fund new hires, it seems like it should cancel out.
The three ways around bearing those costs that I can see at first glance are i) earning to give to prevent a group from laying off staff for lack of funding, ii) earning to give to buy capital goods rather than hire people, iii) staying in your current direct work job rather than switching.
Since most of the EA orgs in question are heavily constrained in hiring by whatever level of growth they can manage or feel comfortable with (that’s kinda the whole point of the OP, right?), it would not generally be my assumption that additional funds would be used for extra hiring compared to the counterfactual. I grant that if that is the assumption, these effects seem to cancel.
Other ways not listed in your last paragraph.
-earning to give to allow orgs to raise salaries
-earning to give to fund regranting
-earning to give to fund things like targeted advertising (you may have intended to cover this category in ‘capital goods’, I’m not sure)
These things are much closer to my model of where extra funding to at least CEA and 80k in at least the past 18 months has gone, not into additional hiring.
Hey Alex,
A quick note on another aspect that I think Rob might have underemphasised:
Hiring could generate very high returns to the organisations but often not as high as other marginal activities.
Here’s a simple model: If you think that many EA orgs spend about $1m per year and generate $10m per year of “value” (as rough orders of magnitude), then even if a marginal hire generates $1m “excess” value to the org, that’s only a 10% increase in impact for that year. This is good but there could easily be even more important things for senior management to focus on.
This would explain why the orgs don’t hire many people.
However, it would still be consistent with the idea that marginal hires are valuable and can have more impact by working at the org than by earning to give, since each generates $1m.
It would also mean conditional on an org running a hiring process, it’s better for that process to go as well as possible and get the best possible applicants.
In that simple model, it appears to me that marginal hires are worth $1m minus the counterfactual use of senior staff time (hypothetically, and relevantly, suppose all the possible hires for a position decided to earn to give instead overnight. It would not be the case that the world was $1m worse off, because now senior staff are freed up to do other things). If there are in fact even more important things for senior management to focus on, this would be a negative number.
More realistically, we could assume orgs are prioritising marginal hiring correctly relative to their other activities (not obvious to me, but a reasonable outside view without delving into the org particulars I think), in which case the value of a marginal additional hire would simply be ~0.
So again, I appreciate the attempt to boil down the area of disagreement to the essentials, and even very largely agree with the essential models and descriptions you and Rob are using as they happen to match my experience working and recruiting for a different talent-constrained organisation, but feel like this kind of response is missing the point I’m trying to make, namely that these ex post numbers, even if accurate on their own terms, are not particularly relevant for comparison of options.
Hey Alex,
I agree with the thrust of your point: if hiring is very costly, then that’s reason against working at EA orgs vs etg.
It still seems like there could be a world in which hiring is high return but often (but not always) lower return than some other marginal activities. This will mean both that (i) recent hires are valuable (ii) the orgs don’t hire many people (iii) hiring is sometimes worthwhile. I was just trying to show how these could be consistent.
Going back to the overall issue of why the dollar figures are high, I don’t think the “hiring is costly” point is the main thing going on, and think this post might have emphasised that too much. The post was more trying to answer the question of why the orgs don’t hire more often despite the high dollar figures. The broader question of why the figures are high and what to do based on them is discussed briefly in the original article about the survey, but hasn’t yet been tackled in depth.
With your first two paragraphs, I just want to step back and point out that things get pretty confusing when you include all opportunity costs. When you do that, the return of every action except the single best action is zero or negative. Being close to zero is actually good. It’s probably less confusing to think in terms of a ranked list of actions that senior staff could take.
I also expect the orgs partially take account of the opportunity costs of staff time when reporting the dollar value figures, though it’s hard to be sure. This is why next year we’ll focus on in-depth interviews to better understand the figures rather than repeating the survey.
True, but I haven’t accounted for all the opportunity costs, just one of them, namely the ‘senior staff time’ opportunity cost. If you are in fact close to 0 after that cost alone (i.e. being in a situation where a new hire would use x time and generate $1m, but an alternative org-improvement action that could be taken internally would generate $950k), that isn’t ‘good’, it’s awful, because one of those actions incurs opportunity costs on the applicant side (namely, and at the risk of beating a dead horse, the cost of not earning to give), but the other does not.
So we could look at this as a ranked list of potential senior staff actions, but to do so usefully the ranking needs to be determined by numbers that account for all the costs and benefits to the wider world and only exclude senior staff time (i.e. use $1m minus opportunity cost to applicant minus salary minus financial cost of hiring process per successful hire etc.), not this gross $1m number.
Similarly, potential applicants to EA orgs making a ranked list of their options should include all costs and benefits that aren’t tied to them, i.e. they should subtract senior staff time from the $1m number, if that hasn’t been done for them already. Which is what I’ve in fact been recommending people do. But my experience is that people who haven’t been directly involved in hiring during their career radically underestimate the cost of hiring, and many applicants fall into that camp, so for them to take account of this is not trivial. I mean, it’s not trivial for the orgs either, but I do think it is relatively easier.
Given this conversation, I’m pretty skeptical of that? My experience with talking to EA org leaders is that if I beat this horse pretty hard they back down on inflated numbers or add caveats, but otherwise they say things that sound very like ‘However, it would still be consistent with the idea that marginal hires are valuable and can have more impact by working at the org than by earning to give, since each generates $1m.’, a statement it sounds like we now both agree is false or at least apt to mislead people who could earn to give for slightly less than $1m into thinking they should switch when they shouldn’t.
For the benefit of third parties reading this thread, I have had conversations in this area with Ben and other org leaders in the past, and I actually think Ben thinks about these issues more clearly than almost anybody else I’ve spoken to. So the above paragraph should not be read as a criticism of him personally, rather a statement that ‘if you can slip up and get this wrong, everybody else is definitely getting this wrong, and I speculate that you might be projecting a bit when you state they are getting it right’. The only thing Ben personally has done here is been kind enough to put the model in writing where I can more-easily poke holes in it.
Hey Alex,
I’m feeling a bit set upon here. I was talking about a different topic (why people hire slowly) and it seems like we got our wires crossed and are now debating the value of a marginal hire. Looking back, I see how my earlier comment led us off in a confusing direction.
If we’re discussing the value of a marginal hire, I totally agree that the survey figures DON’T include the costs of hiring someone in the first place. That’s why I brought up the ex-ante ex-post distinction in the first place.
This means that someone considering working at an EA org should use a lower figure for the estimate of their value-add (we agree). In particular, they should subtract the opportunity costs of the time spent hiring them. (This varies a lot on the situation, but one month of senior staff time seems like a reasonable ballpark to me.)
However, just to be clear, I don’t think they should subtract the opportunity costs of senior staff time spent on their on-going management, since I think the natural interpretation of the survey question includes these. (If a new hire could raise $1m of donations, but would take up management time that could have raised $800k otherwise and has a salary of $100k, it would be odd for the org say that they’d need to be compensated with $1m if the hire disappeared. Rather, the answer should be $100k. I expect most orgs were aiming to include these costs, though of course they might not have made a good estimate. This is what I thought you were talking about when I said I think orgs partially take opportunity costs into account.)
Where does this leave the survey figures? We could super roughly estimate the value of a month of senior staff time at an org is 5x the value of a month of junior staff time, so this would super roughly reduce the value of junior hires over three years by 5⁄36 = 14%.
Re. your first paragraph, I don’t know why you chose to reply to my comment specifically, since as far as I can tell I’ve never been asking ‘why do people hire slowly’.
I think I’ve already explained why I don’t agree with your later paragraphs and see little value in repeating myself, so we should probably just leave it there.
Yes, I apologise for that, we were talking at cross purposes.