The rub there is that you'll have varying styles in which error handling statements appear in your code. With simplistic instances appearing one way and less simplistic ones appearing similarly or differently depending on where each lands on the complexity spectrum. The idiomatic approach is for all instances to be written in the same way regardless of technical nuances.

All of that aside, I've come to learn that passing errors up the call stack without any wrapping or handling is a code smell. It is less than useless for me to attempt setting the value of cell A1 in an Excel sheet to "Foo" and then receive an out-of-range error because the developer made no attempt to even inform me of the context around the error. Let alone handling the error and attempting to correct state before giving up.

In my Excel example, the cause of the error was a data validation problem a couple columns over (F or so). The error was legitimately useless in troubleshooting.