I was about to compose a post saying essentially the same things as in Clio's post above, but he saved me the trouble :-)
Ralph has particularly emphasized the part about "not all components are designed to this standard," meaning that it is common for balanced interfaces to have inadequately low output impedance, or to be otherwise poorly designed. That would seem likely to be the major reason for differences being reported between balanced cables.
Kijanki 1-19-11: You can spend arm and a leg for a balanced IC as well. They're also made of fancy materials - both metals and insulation. There is no reason why it should be different IMHO. Purity of metal still applies since impurity reside between crystals while insulation material directly affects capacitance and dielectric absorption.
This assumes that differences in impurities between crystals and differences in dielectric absorption are sonically significant, neither assumption having been proven based on widely recognized science, as far as I am aware. Although neither assumption has been disproven either, as far as I am aware.
Low capacitance, of course, is easily obtainable in low priced as well as high priced cables, and is only significant if output impedance is high and run length is long.
Regards,
-- Al