I think there's a likelihood that someone will lie about being raped, and a likelihood of them identifying a named individual (two numbers that should be considered seperately, since young girls who fall pregnant may lie to avoid blame, while not trying to blame anyone in particular).
I think that similarly, there's a percentage of retractions that will also be false, for various reasons.
This is similar to the statistics question about whether someone has cancer when a test comes back positive. It depends on the accuracy of the test and the base rate.
I could easily believe that there's more false retractions than false accusations (both as a percentage and an absolute number) so I'm not sure a retraction actually increases my belief in their not being a rape, particularly if a specific person is named, without taking into account other specific evidence.