No. We're trying to compute the product VMON*IMON.
Imagine that the mosfet current indicator VMON is zero, but the drain voltage VMON varies. Both PNP bases are at 0 volts, and the current through R1 (which is symbolic of fet D-S voltage VMON) splits between the two PNPs. Both voltages Z1 and Z2 go up as VMON goes up, but there's no power dissipation in the fet, and we're trying to compute dissipation.
The *difference* between the collector currents represents mosfet power.
!!!
Sure. Lots of people don't like this circuit, especially returning the PNP bases and collectors to ground. Too bad, it works.