Empirical Evidence for the Value of Version Control?

We received this by email:

I use version control for my software, and I encourage others to do so, but I have no experimental evidence to base that decision. I pulled out my old copy of Code Complete (it's a first edition), and the only reference it makes is to "Moore 1992", which is a private communication that says that Microsoft considers their internal use of version control to be a competitive advantage.

The common practices I know of are:

1. no version control
2. every once in a while make a backup, either as a tar/zip file or copy everything into a new directory
3. use filesystem versioning, like what was on a VAX, or Time Machine on a Mac, or Dropbox for a distributed multi-version file system
4. use a version control system; though this in turn can vary from SCCS and RCS to Fossil and Veracity

In addition, there's a difference between the needs of a single developer vs. a small team, vs. a large, distributed team.

Is there published experimental evidence showing that a version control system is more useful than, say, developing using Dropbox? I tried looking for the relevant papers but I don't know how to search that field and I couldn't find anything.

It's a good question—does anyone have an answer?