Okay, to answer my own question:
It works as intended. I was using dipole swivel antennas for my XBee's which were position close to each other, thus giving a strong signal link. After closer inspection I realized that such strong link resulted in PWM with 100% duty cycle, which appeared as a logic high. By covering one of the antennas with aluminium foil or just disconnecting it entirely I was finally able to see/measure the PWM.
I have, however, decided to use remote at command request frame instead as it's much simpler to implement for the time being.