As always Igor, a very well written explanation of the use of change of measure techniques for multi-curve dynamics in the evaluation of interest rate derivatives. However I am still confused about some of the details in the section Libor adjustment in multi-curve framework. In the paragraph just above the definition of cDist and cdrift , you have Subscript[E, OIS][Fe] = Subscript[E, e][Fd Bj]. In the first place Fe is defined relative to the Libor estimation curve and the measure Subscript[Q, e], so the LHS of the equation should be Subscript[E, e][Fe] which then leads to a contradiction. I think that you meant to use Fd which is measured wrt the OIS curve giving Subscript[E, OIS][Fd] = Subscript[E, e][Fd Bj] so that according to Girsanov's theorem we have the Radon-Nikodym derivative Bj = d Subscript[Q, d]/d Subscript[Q, e]. Now we can substitute for Fd with the equation Fd == Fe Aj, so that the RHS becomes Subscript[E, e][Aj Fe Bj] == Aj Subscript[E, e][Fe Bj], where Aj is the forward basis/adjustment. This is now consistent with Aj *Subscript[E, e][Fe Bj] == cdfift and the definition of Aj.
I most probably have got something mixed up, but would appreciate it if you could better explain how Subscript[E, OIS][Fe] = Subscript[E, e][Fd Bj] is consistent with the change of measure and the definitions for cdrift and Aj.
Regards
Michael
