That's quite possible. I've done a certain amount of it myself. A couple of programs that I wrote for the factory 15+ years ago are being used continually for critical adjustment and testing of subassemblies. All told it's a few thousand lines of Visual Basic. Not "clean code" but carefully documented with a complete theory of operation that could be used as a spec for a professionally written version.
My view is that it's not a failing, any more than "software development can't be estimated" is, but a fact of life. Every complex organization faces the dilemma of big versus little projects, and ends up having to draw the line somewhere. It makes the most sense for the business, and for developer careers, to focus on the biggest, most visible projects.
The little projects get conducted in shadow mode. Perhaps a benefit of Excel is a kind of social compromise, where it signals that you're not trying to do IT work, and IT accepts that it's not threatening.
There's a risk, but I think it's minimal. Risk is probability times impact, measured in dollars. The biggest risks come from the biggest projects, just because the potential impact is big. Virtually all of the project failures that threaten businesses come from big projects that are carried out by good engineers using all of the proper methods.