Tanya,
              I went through the comparisons for the Liquids curves and the 
appearance of clear parallel shifts, etc, do begin to emerge when fewer 
forward prices are used.  It looks sensible.  I have passed the graphs over 
to the liquids people, and I have asked them to identify rough term structure 
months when illiquidity begins for these curves.  It might coincide with your 
assumptions.  I am surprised by Brent and Dubai, which should be WTI-clones.

Naveen  




Tanya Tamarchenko@ECT
10/04/2000 04:35 PM
To: Naveen Andrews/Corp/Enron@ENRON, Vladimir Gorny/HOU/ECT@ECT
cc: Vince J Kaminski/HOU/ECT@ECT, Kirstee Hewitt/LON/ECT@ECT 

Subject: Re: factor loadings for primary curves  

Naveen & Vlady,
Jin Yu finished debugging the vatrfacs code and now it calculates factor 
loadings for every "primary" curve (except power curves).
I am sending you the calculated factors:

Most of them don't look good. 60 forward prices were used in calculations for 
each commodity.
I reran the code using fewer forward prices depending on the commodity
(12 prices for C3GC,MTBE,NC4,SO2,
17 prices for NXHO, 18 - for SA,
24 for C2GC, LAX_JFK,,
30 -for Condensate, Dubaicrude, Brent,,
48 for NSW, Semichem-Risi)
These results are in 

Most of them look much better.
Please, review.
We will have to add a column in rms_main_curve_list to specify how many 
forward prices we want to use for each commodity,
and then use the new factors in the VAR model.

Tanya.