Willis Eschenbach had a post recently at Watts Up With That on Parasitic Losses or Parasitic Loads depending on your point of view. Parasitic Loads are things that don't directly contribute to the desired product but make the production more enjoyable or tolerable, like A/C in your car or power brakes/steering which require energy from your engine and rob some fuel mileage where if you had large rock hard tires, no brakes and never turned, you would get great mileage once or twice but not many would enjoy the ride.
I included "rock hard tires" because soft tires with good road grip and a nice cushy suspension also reduce fuel mileage a touch and are more like like parasitic losses than loads. Energy is knowingly diverted to produce the A/C, power brakes and steering but not the suspension or the aerodynamics of the vehicle which are designed to reduce frictional loss while acknowledging creature comfort, manufacturing cost and sex (sales) appeal. This is one of the problems with the 100 mile per gallon car, people wouldn't buy one. There is a trade off between efficiency and sex appeal.
Willis described latent and sensible cooling as "parasitic" losses in a typical one dimensional evaluation of a complex climate system evaluated through a radiant energy only perspective. Possibly a better way would be a radiant efficiency perspective. At whatever is the actual top of the atmosphere, the climate system has close to 100% radiant efficiency and in the depths of the oceans nearly 0% radiant efficiency. If you consider the ocean surface to be "the" surface then with a temperature of 20C, input energy of 200 Wm-2, latent/sensible combined cooling of 100 Wm-2 you have a couple of choices for defining your radiant efficiency. At first blush, 100 Wm-2 of "parasitic" loss with 200 Wm-2 input would indicate 50% radiant efficiency. However, 20C would have a radiant energy of 418 Wm-2, which with the 200 input and 100 output leaves 118 Wm-2 of interacting energy. That energy can be stored or returned by the atmosphere. Since the 100 Wm-2 latent/sensible cooling decreases the actual "surface" temperature, the effective temperature of the "surface" is 518 Wm-2, so you could consider the radiant efficiency to be 418/518 or 81% efficient in only the simplistic up/down perspective. You could also consider the input 200 Wm-2 against the effective temperature and get 200/518 or 38.6% efficient which is more reasonable with respect to Carnot efficiency. This may seem confusing, but if the temperature is stable, then the output has to equal the input meaning the selected "surface" is less efficient at losing energy which is the whole premise of the Greenhouse effect (GHE).
Then if you consider the "imbalance" of about 0.5 Wm-2, the GHE imbalance is 0.5/518 or 0.9% which is pretty much exactly where we stand. There is a small imbalance due to the changing composition of the atmosphere and albedo changes at the "surface" or stratosphere or anywhere you would like to look for small changes. Pick a different "surface" you get different results. That is pretty much the GHE game, find the "surface" with the most sex appeal for your side of the argument.