Sure, I agree that "can't afford" is doing a lot of work there, and most people considering nonprofit roles are coming from positions of relative privilege (especially globally). I should have clarified that I'm benchmarking large nonprofits, large philanthropies, and for-profits versus other nonprofits. 

What I meant is more about opportunity costs and life constraints rather than subsistence: someone with student loans, family to support, or who's trying to afford living in/near a high-cost city might find the financial gap genuinely prohibitive, even if they're still well-off in absolute terms.  Also, smaller nonprofits typically offer minimal or nonexistent benefits: limited healthcare, no retirement matching, sparse parental leave, while also having high education requirements and geographic constraints (often expensive cities). 

Thanks for the post. I think the golden handcuffs concern is worth flagging, but it's likely outweighed by the bigger problem: talented people who never enter the social sector because they can't afford the pay cut.

There's also a structural issue here. Nonprofits can underpay because they're extracting a "mission commitment subsidy" from employees. And this gets reinforced by funders/donors who often treat salaries as wasteful overhead rather than core inputs to doing good work. That creates perverse incentives where orgs compete to look "efficient" by under-investing in people.

Somewhat obvious note: These organizations can pay more because they're exceptionally well-funded. The real issue isn't that some orgs pay well, it's the unevenness. If compensation were uniformly reasonable across the sector, people could move based on impact and fit rather than financial need. 

That all said, publishing compensation philosophies would be broadly beneficial (and more fully in line with transparency). 

(FWIW, these points are also broadly true in the public sector.)

A note regarding this:

In Swapcard, the app used for EAG(x) events, it is possible to schedule meeting times with more than one person (up to eight, using the same method as scheduling a 1-on-1).  I would really like to see this feature used more regularly, but I suspect few know it exists.