I don't know what NIH says about the problem you described, but I have
seen applications in which one of our investigator describes a research
project which cannot be executed at our school.  In the good ones, the PI
has found the missing resources at another school and planned on a formal
collaboration involving subcontracts; in the weak ones, the application
attempts to hide our insufficiency by being vague on the details of how
the project will be executed.  The sponsored programs office usually
receives the weak applications too late to try to put formal
collaborations in place, and so we try to convince the institutional
signatory not to submit the application.
 We currently hold two large grants (>$500,000/year) for which we
were not equipped to deliver what we promised and for which the PI did
not plan adequate collaboration beforehand.  This sounds like the
situation you describe--for surely your PI knew, or should have known, of
this problem when he prepared the application.   We've simply required
the PI to make formal contractual arrangements.  Except it hasn't always
been simple because of factors like unplanned indirect costs for the
collaborating institution, complex regulatory transactions (the IRB and
IACUC), and the PIs concern about somehow losing control of the project
by contracting part of it.
 I think the answer is for the school to spot such difficulties
before the application goes out and fix them beforehand with careful
planning.  Hopefully the spotting would be done by the PIs department
chair, for when the sponosored programs office makes such a call, it's
accused of being intrusive in matters properly left to faculty judgment;
but if the chair won't do it, somebody should.  Once such an award is
made, you have to make the best of a difficult situation, presumably by
teaching the PI about formal collaboration.  Some of the important
principles are that grants are awarded to institutions, not individuals,
and that "employment" is a true condition that must be considered, even
in academic settings.   I certainly would resist the idea of simply
transferring a grant awarded to my institution to another one; and I
should think the target institution would have trouble accepting the
arrangment when the PI is employed by another organization.

Good luck!

Peter J. Dolce
Director
Office of Sponsored Programs
Meharry Medical College
Nashville, TN  37208
P 615 327 6703
F 615 327 6738