Enron email

Algorithmic Indexing

13 Nov 2000, 16:46 UTC

Re: VAR for Niamh portfolio "Financial Products"

From
Tanya Tamarchenko tanya.tamarchenko@enron.com
To
Niamh Clarke
Cc
Vince J Kaminski
Bcc
Vince J Kaminski
Date
Mon, 13 Nov 2000 08:46:00 -0800 (PST)
Niamh, 
I examined how VAR of "Financial Products" portfolio depends on 
term-structure of correlations between BRENT and 61NY. Here are the results 
(for positions on 11/6/00):

1). If we assume correlations of 1 across all forward contacts for Brent and 
61NY
as well as between Brent and 61NY for all maturities,

VAR=$424,199

2). If we calculate correlations based on first 18 contracts prices for both 
Brent and 61NY
together and assume longer maturity contracts are all perfectly correlated, 
then

VAR=$1,868,949

3). Same as (2) but correlations are based on 12 first contracts, everything 
beyond that is perfectly correlated, then:

VAR=$1,106,773.

Conclusion 1:  when we use higher correlations, VAR goes down (as we 
expected).
Conclusion 2: correlations depend on how many forward contracts we use to 
calculate them. We should use all liquidly traded contracts, because prices 
for those contain statistically reliable information.

I chose 18 initially, what is your suggestion?

Regards,

Tanya.