Micro four thirds - does the crop factor apply to MFT lenses as well?

...of missing the forest for the trees.  We don't compare one pixel from a 16 MP photo to one pixel from a 24 MP photo, but 2 pixels from a 16 MP photo against 3 pixels from a 24 MP sensor, since 2 pixels of a 16 MP photo cover the same portion of the scene as 3 pixels from a 24 MP photo.

What we will find is that the same total amount of light falls on the 2 pixels from the 16 MP photo as falls on 3 pixels from the 24 MP photo, since the same total amount of light falls on the sensor for a given scene, perspective, DOF, and shutter speed, and, as the EM5 and D600 sensors have the same QE (Quantum Efficiency -- the proportion of light falling on the sensor that is recorded).  This means that the photon noise (the noise from the light itself) will be the same.

Next, we have read noise (the amount of noise added by the sensor and supporting hardware).  As it turns out, this is a function of the ISO setting for many sensors, including the EM5 and D600.  At ISO 200 on the EM5 and ISO 800 on the D600, there is actually less read noise for 3 pixels on the D600 than 2 pixels on the EM5 (however, at higher ISOs, the situation reverses).

It's worth noting that the read noise is only significant in the portions of the photo that receive very little light -- the bulk of the photo is dominated by read noise (unless we're talking about very low light photography, or pushing up the shadows a massive amount).

Thus, the total noise for 3 pixels from the D600 at f/4 1/200 ISO 800 will have a tad less noise than 2 pixels from the EM5 at f/2 1/200 ISO 200, and thus for the photo as a whole, as well.  However, at higher ISOs, the situation will reverse, and the EM5 will have a slight noise advantage for equivalent photos over the D600.