Normalization for RNA-IP experiment

[jazz]

I have some Illumina sequencing reads from two RNA-IP experiments, one utilizing the test antibody and the other with a control antibody (IgG). I did a small scale sequencing and got 2 mill and 400,000 reads, respectively (after adapter clipping and QC). Now I am pondering on how to normalize the data. My naive thinking is that if I subsample 400,000 reads from my test antibody IP, then I can map those onto the genome along with the - control and that should be appropriate. Is there a flaw in my thinking here?
My colleague suggested normalizing using a negative control which is not supposed to bind my antibody.
Any suggestions as to what might be better suited in this case.

Thanks