A quick follow-up to this morning's blog. Just thinking about it, it occured to me that it's not enough for the benefit you get from the other party to be greater than the benefit to person with next best fit. Given that this is a negotiation situation and that participation or non-participation is usually voluntary, if the other party knows what the value to you is, they may raise their price to equal the benefit anyway.
So T could set the price to A not equal to G(T,B) but equal to G(T,A) if T knows what G(T,A) is. In part, of course, this is solved by the mutual nature of the relationship - the synergy created is too important to both parties to try for such pricing (or in other words both parties think they're getting a good deal and don't bother). But it also makes a case for not letting the other party know exactly what their value to you is. Basically, A should show T that T's value to A is G(T,B) + d, where d (delta) < (G(T,A) - G (T,B)). That way even if T prices at benefit to A, not at next best alternative value (so C(T,A) = (G(T,B) +d)), A still stands to gain a net benefit equal to
G (T,A) - G(T,B) - d. In plain English, of course, this is called playing hard to get.