Another factor might be the time and effort to create documentation. The loss of service manuals has occurred simultaneous to the loss of good user manuals.
Back in 2009, I was hired by a small software company to do the English translation of their product. That involved several thousand text strings in the program (i.e. any text the program itself displayed, like the names of menu items, checkboxes, and other widgets, as well as error messages and the like), over a thousand context-sensitive help files. Then there was the 300-page manual, hundreds of FAQs, etc.
Writing a manual like that is a ton of work, even just in one language. Really good manuals actually fill two roles: how-to guides for less-experienced users, and reference guides for experienced ones. That’s a big challenge in itself. Companies have been less and less willing to do that, instead putting out short “quick start guides” and then writing shitty FAQs to add information that should have been in the manual to begin with, and leaving the rest up to community self-help by offering user forums. I think it’s shameful.
All of that is just for user documentation, which is something that a good technical writer can write mostly on their own. Public service documentation requires much more direct involvement of the engineering staff who actually designed the product. This makes that documentation even more expensive. And to boot, it often requires review by the legal department to ensure that no trade secrets are being divulged.
In the end, most service documentation nowadays is produced for internal use only, often written by the engineers themselves. This avoids the need to have it reviewed by legal and rewritten/edited by the technical communication department.