ATT 268452.txt

                                                                  EXHIBIT 99.279

I think this misses the point, and there is an important policy issue that
should not be overlooked.

The PJM data shows that a constraint can be violated under different
contingencies. The different contingencies can mean that the PTDFs change
depending on the contingency, resulting in a different set of flows across that
constraint (and others). This means that to remain fully hedged, a trader may
have to acquire a different mix of FGRs to match the changed flows resulting
from each contingency. If the contingency changes, the mix of required FGRs may
change again.

Listing these different contingency/constraint pairs shows that it is not
enough to predict which constraints will be binding, as some advocates have
claimed; you also need to know why the constraint is binding and what that does
to the relevant PTDFs, because that will define the mix of FGRs you need. Each
contingency could have a different set of PTDFs for the same transaction, so
you need a different mix of FGRs to maintain the same hedge. However, you will
not know which mix you need until you know which of many possible contingencies
actually occurred (or did not occur). This can lead to the ISO having to
publish a "library" of PTDF tables, one for each possible contingency, with
traders having to consult the appropriate table and readjust their FGR
portfolio each time a different contingency arises that affects any of the
identified constraints. I do not know whether the market will find it easy,
difficult or impossible to do this, but the requirement to do it has to be
considered when addressing the merits of any flowgate rights approach.

This problem had an effect on the hybrid proposal developed by the MISO working
group, which chose to "freeze" the grid assumptions on a month-ahead basis for
purposes of defining the values of FGRs. If the RTO tries to "simplify" this
problem by freezing the grid configuration assumptions and corresponding PTDFs
for any period, you are then confronted with tough dilemmas. During the freeze
period, the RTO must socialize the difference between the actual congestion
costs based on actual grid conditions (and the actual contingency that made the
constraints binding) and the FGR congestion credits based on the frozen grid
conditions/PTDFs/contingencies. If you try to limit the socialization risks by
shortening the freeze period -- e.g., from a year down to a month or week or
less -- then you have precluded the market from acquiring any longer-term price
certainty. So while the market gets subsidized rights during the freeze period,
they lose longer-term hedging/price certainty. Conversely, a long-term freeze
to improve price certainty can increase the degree of cost socialization. These
are not easy trade offs either way. The data in Andy's paper helps us to see
this problem.

Andy's paper is a warning that the problems are multi-faceted, and that we
can't assume the numbers of constraints or contingencies will be either small
or predictable. We have to design transmission rights to deal effectively and
efficiently with these uncertainties, rather than assume they will not arise
and expect someone else to bear the risks of our being wrong.

Regards,



John Chandley


                                    Page 1