Any way you write it, it's more useful to understand it as a value variance. After all, it is the difference between two values--earned and planned--denominated in dollars; neither are a cost--both are a value measurement.
The nice thing about thinking about this as a value variance rather than a schedule variance is that the thought is consistent at all the boundary conditions and points. One boundary point is where EV reaches PV--all the planned value is earned. The EV - SV calculation is now $0. As a value variance: no conceptual problem: all the value has been earned.
As a schedule variance, we have to think like a pretzel to explain how the variance can be zero under two of three boundary conditions: the project finishes early, the project finishes on-time, the project finishes late. In all three cases, the variance can be zero (if all value is earned) but the three conditions are really quite different vis a vis schedule variance. Zero schedule variance doesn't compute; they share only one attribute: the value variance is zero.
Are you on LinkedIn? Share this article with your network by clicking on the link.