It seems that, ideally, yes.
But “Note that this criterion only requires the value to be uniformly distributed , not random in any sense. A good randomizing function is (barring computational efficiency concerns) generally a good choice as a hash function, but the converse need not be true.”