I did not mention the sampling rate, I referred to the bit-rate, i.e. 24bits instead of 16bits
Thanks for the clarification. My confusion was because "rate" usually implies a time-dependent entity (sample rate), and the number of bits in a sample is usually referred to as "bit length".
In your original comment, you stated "high-bit rate mastering which they claim is more important than a high bit rate end-product", which is true if we are talking about or number of bits per sample and not rates. When recording and processing, it is useful to have more than 16 bits (on recording, 24 bits are normally used, in processing the intermediate calculations might be done with much higher number of bits, and often in floating point format). When recording, 24 bits give more headroom and margin in case levels aren't perfect, and in processing you minimize rounding errors.
When the processing is done, the material can be normalized to an optimal level, and no more than 16 bits are needed. I have checked out thousands of recordings, and I have not, so far, come across a single commercial recording that uses a range that exceeds 16 bits, so the least significant bits in a 24-bit recording, as distributed, is purely noise.