It's useful to warn engineers against taking decisions without any supporting facts and numbers, just taking for granted what salesmen say. When just connected to the grid, an induction motor will probably have more than necessary flux, because it will probably work below its nominal power. The assumption that one can exactly match the size of the motor to the application is unrealistic. The motor cannot be sized for exactly the required torque for many reasons, including the fact that sizes are at a fixed step, and grid voltage and load are not exactly constant. It can then happen that, even accounting for variable frequency drive (VFD) losses, the reduced flux level at lower-than-nominal torque can give VFD an advantage, if the drive control has a careful efficiency approach.
I tried to follow back the citations for the formulas used in losses calculation in the white paper, and they seem quite strange. In particular, it makes little sense to consider 3hp of losses in the VFD at no current, even for a >200hp nominal power.
Almost all applications are sized for the hp that will be needed when there is eventual wear and degradation in the mechanical system, not for the hp required to drive the load when everything is brand-spanking new. That means that there are service factors and margins built in to the sizing from the very beginning. Consider a fan system. Eventually, the system will develop leakage in the ducting, so we add enough fan and power to cover needs PLUS a 15% margin for leakage. Eventually the fan itself will wear and not produce design airflow without a little extra power, so that another 10% margin. Dampers or control vanes definitely wear and leak over time, so we add another 15% margin. Hence, when new, the fan and the drive power are stacked with capability to produce much more air than needed, so 100% rated FLOW, may only require 70% speed. Even with this full RATED FLOW application, the VFD would be a quick payback while running at 100% rated flow, 100% of the time. The other factor that is forgotten by many is that the VFD does correct the power factor of the motor, which may or may not be significant, depending upon electric power rate structure and the speed/design of the electric motor, but should not be neglected. There is a reason engineers are paid to understand and evaluate such things, it IS too easy to oversimplify!