The individual pixels ARE smaller on a smaller sensor! The micro 4/3 sensor is half the size of a full frame and the megapixels are about the same, so there's 1/4 the light absorption for each pixel at the same aperture and focal length.
Yes, but this is not always the case, and you've got to keep all other things equal if you want to talk about this stuff in a meaningful way. Talking about ISO performance varying according to pixel size, sensor size, etc. is just confusing the issue. The exposure is calculated based on the amount of light hitting a given area of the sensor, period... no matter whether there's a lot of pixels in that area or not, or whether there's more pixels outside that area or not.
The exposure numbers have been the same since the early days of FILM photography, when pixels didn't enter the equation at all! You're too stuck in the current iteration of photographic technology to think about the problem in the right way. You've got to abstract things a bit, wipe away the technicalities of pixels and sensors and just think about light coming through the lens and hitting an object that is sensitive to light.
I have a 5D, and I had a 20D. They used almost the same sensor technology, the 20D had 8.2MP while the 5D had 12MP. Their pixel pitch wasn't exactly the same, but it was relatively close. I think 6.4µ for the 20D and 8.2µ on the 5D. You could stick the same lenses on both, set the ISO's and shutter speeds and apertures to the same values, and you'd get approximately the same image from both (one just cropped to the middle).
I could take a Sigma 30mm f/1.4 and it would work great with the 20D. Put it on the 5D, and the corners would be quite a bit darker, but the middle part of the image would be exactly the same as it was on the 20D.
Or take the opposite situation, take a full-frame lens like a Canon 70-200, put it on the 5D and you get the roughly the same photo as you do with the lens mounted on the 20D with the same exposure settings; but the 20D is cropped. The cropping throws away a good amount of the light that was transmitted through the lens (and that landed outside of the smaller sensor area of the 20D) but the exposure -- for the pixels that were exposed -- is the same.
ISO is ISO, aperture is aperture, and from the photographer's point of view it doesn't change at all from full-frame, to crop, to m4/3, to P&S (outside of DOF, of course). Yes the manufacturers might "juice" the numbers somehow, but overall they are trying to match the final result with the same film standards that photographers have used for 100 years. There is too much established already for a company to mess with anything too much.
There are plenty of photogs who use light meters, which tell you what your exposure needs to be based on the light available. These meters are the exact same as they were back in the film days. If a camera's exposure didn't match the numbers from the light meter, it would be found out pretty quickly. Granted I have seen some cases where one camera or lens might vary from another by 1/2 stop or so, but those are the rare cases.