That's not quite true. Even without changing the bitrate, you may change the codec being used, or the codec profile, or the minor version number, or the motion search quality, or the lowpass/highpass used, and since the data won't be the same as the input wav, you're going to have differences in the input signal analysis that become further differences in the output. Marginal errors accumulate across lossy reencodes.

The only possible justification for pulling out to wav is if it's already got a recompress in its future. If you're just trying to copy the audio stream, then regardless of any other arguments, you're wasting time and degrading signal by decompressing and recompressing it.