Yes. To expand on what Bob said, you have probably seen a range of recommended headroom because there is no magic number and it depends on all these things:
-the nature of material itself
-how loud it is pushed
-what limiter is used and if it has oversampling or intersample peak detection
-the target encoded format
-how much headroom you leave been the limiter output and 0dBFS
It’s best to use something like the Encoder Checker that comes with WaveLab, or Sonnox ProCodec is a very useful app if you’re serious about this. I like the Sonnox ProCodec because you can perform offline encodes with reproducible results to verify. With most (or maybe all?) encoder simulators, you will see more randomized results with the peak levels when you have the encoder running live because the encoder bitstream is not perfectly in sync with the start of the file you’ll eventually encode or send in for digital distribution.
Lower bitrate codecs will produce and higher (and more frequent) peak levels when encoded from WAV than higher bitrate codecs will. So if you want your material to be safe all the way down to 128kpbs, you may have to lower the output ceiling to -1dBFS or more. If you only care about being safe down to 320kbps, you might be able to get away with -0.5dBFS.
However, this depends on what the source material is, how high the average and peak levels are mastered, if the limiter used is doing some oversampling or intersample peak detection, and as I said, the specific codec you plan to encode to (128, 256, 320, AAC, Ogg Vorbis, etc.)
So for me, I make an educated guess based on the material and the limiter I’m using, render the WAVs and then do an accurate offline test in Sonnox ProCodec to be sure clipping doesn’t occur at whatever the target codec for distribution. If everything checks out, I’m done. If ProCodec reports clipping, I go back to WaveLab, adjust, and test again.