By definition, power dissipation in an object that has no internal source of power must be zero or positive.
LTspice has some bizarre behavior that seems to be related to two things, (1) was a "save" executed, or not, between sims, and 2) is the component (resistor seems to be vulnerable) isolated from a direct "hard" power source... don't know how transistors might behave, but Larkin claimed his negative power was in a transistor, with a value way to large to be rounding error.
I see no point in further wasting my time on a subject that no one else actually has a scientific interest in pursuing... just casting stones at me because I questioned the sacred cow ;-) ...Jim Thompson