Let us try to unpack some of the issues raised by this interesting question.
The first thing to bear in mind is that the units for masking ABR stimuli have to be in SPL – this is because a white noise masker is required to effectively mask the relatively wide bandwidth ABR stimuli (e.g. click sounds are broadband, as are CE-Chirp. Other signals like toneburst that are more frequency specific exhibit spectral splatter due to their short duration, and the bandwidth of narrow-band CE-Chirps cover several hundred Hz). Since there is no nHL (or even HL) correction for white noise the unit of level is SPL.
The level of masking noise required depends not on comparative loudness in relation to the stimulus (which is presented to the test ear), but on the level of masker in the non-test ear required to mask out any cross-heard sound in the non-test ear. Based on the study by Lightfoot et al (2010), we know what these levels are.
References and caveats
Lightfoot, G., Cairns, A., and Stevens, J. (2010). Noise levels required to mask stimuli used in auditory brainstem response testing. International Journal of Audiology, 49:, pp794-798.