I've had the good fortune to work for several good managers, either as direct supervisors or as senior managers, up to the Corporate VP level. That includes people in small companies, in Fortune 500 companies, and even active duty Army officers.
What I've observed is that the top levels of management DO NOT want to listen to what the good engineering managers try to tell them, about topics like staff training and retention, schedules or resources (e.g. hardware/capital expenditures.) Instead, the CxO level people promote those who tell them what they want to hear. It's not universal, but many of the good managers I've had are products of deliberate leadership/management training, rather than being promoted from 'nerd' to 'boss' and left to figure it out on their own. Part of that training is how to talk to the CxO level and how to make arguments in terms of corporate business case, objectives, etc.
The only good news is that at least in this millennium, the number of top managers/CxOs who actually know something about software, has increased. They're still a minority, but you may well find a VP who understands that software isn't "that crappy stuff that always makes our systems late, so we'll 'fix' it by throwing more cheap bodies at it." (I got really tired of the engineering VPs whose experience was in hardware, and whose ideas of software systems engineering was framed by "that FORTRAN course I took in college...")
One interesting model that was popular in the early '90s may deserve another look. Some research labs* split managerial duties, separating technical leadership from administration. Where some organizations got into trouble with that model was not treating both classes of managers as equals. The technical leaders too often got marginalized, because the administrators were the ones that talked about the kinds of stuff CEO/CFO wanted to discuss. It takes a tremendous investment at the CxO level to institute a program that recognizes and grows technical leadership as distinct from, frankly, beancounting.
* It runs in my mind that DEC's Western Research Labs was one of the organizations that implemented this approach successfully.