I am interested to know where arXiv stands against the journals. Not being peer-reviewed, and being so huge, certainly could contribute to arXiv having a very low impact factor, but since ultra-high-quality work also gets published there, it might still have a higher impact factor than some journals that only contain rather poor quality work that couldn't get published anywhere else.
Of course the "whole of arXiv" could have a different impact factor than specific arXiv channels, such as "astro-ph" or "quant-ph".