Not sure if the 'cost' is a 'slightly reduced sensitivity'. Please see the image below that shows two actual images acquired with identical optical parameters with and without Bayer CFA with the same (type of) sensor. Please notice the means on the right side.
Your sample clearly shows the increased sensitivity derived from filter elimination, but reminds me of other benefits of keeping the filter array: increased sensor progressiveness, i.e. dynamic range.
I criticised the Leica Monochrome for paradoxically having a more 'digital' response than RGB cameras, since RAW channel individual exposures act as a dynamic range enhancer on RGB sensors.
Normally the B/R channel clip around 1,5 stops later than the G channel. While in a Leica Monochrome you have to be very careful about clipping the highlights, on RGB sensors you have around 1,5 extra stops of highlight headroom since B&W (actually also colour information) can be reconstructed from a single RAW channel.
The sun in this scene:
Has the following RAW data in EV along the red line:
A monochrome highlight reconstruction strategy (dcraw -H 2):
Inpaint colour algorithm (RawTherapee's colour propagation):
Playing ETTR on a monochrome sensor without accurate RAW histograms is more dangerous than doing it on RGB sensors. Moreover, saving highlights information on a monochrome sensor yields to overall severe underexposure while a RGB sensor still provides a good SNR on the G channel to save the shadow areas. RGB colour filter array acts as an HDR-like sensor: B/R save the highlights, G saves the shadows.
Regards.