I doubt length-of-stay is normally distributed. I would wager that the distribution is skewed heavily to the left. This would result in a that the median is much lower, with a mean that it is biased to the right due to a small number of unusually long leases.

tl;dr: the mean is almost certainly the wrong statistic.

How skewed do you think it would have to be for the mean stay to be “short” eg 2-3 years, given the upper bound of human lifetime?

This is one of those things that sounds smart but fails the common sense test.

That's fair. I suppose I'll have to think about this some more (maybe with a pencil and paper).

Perhaps you're right that the gap between median and mean doesn't change the final conclusion. I can't really say at this point.

Anecdotally, my closest acquaintance in NYC public housing lived there because she is disabled, unable to work, and has no one to take care of her due to age. She has lived in the same public apartment for decades because she is dependent on the state. Her son fled the apartment after living there his whole life, because he cooperated with NYPD on a murder investigation and was threatened by the local criminal gang. Her only roommate now is her orphaned teenage grandson who is vulnerable to the gangs and has ambitions beyond his immediate circumstances.