Hi again,

 As, I stated is a prior emails, the contact details for the RiskCalc folks 
are as follows:

1.  Marc Brammer
Director of Strategic Sales
tel:  800.523.2627, x5706
mbrammer@MoodysRMS.com

2. Neal Clark
Vice President
Tel:  800.523.2627, x5701
nclark@MoodysRMS.com

 Neal said that he would be willing to set up a meeting between us and their 
quants to better understand their model, the various inputs and whether or 
not the model can be tweaked for less than the 17 specified inputs.

 Let me know if you guys wish to have Neal set up a conference between 
Moody's quants and Enron folks in Houston and London.


Ciao,
Iris


 -----Original Message-----
From:  Salmon, Scott  
Sent: Thursday, April 19, 2001 3:00 PM
To: Mack, Iris; Dhar, Amitava
Cc: Chaney, Craig; Mumford, Mike; Kirkpatrick, Eric; Brent, Richard; Parsons, 
Ben; Albanis, George
Subject: RE: D&B Contact Names

Iris/Amitava,

I may have already said this, but we've had so many e-mails and telephone 
calls flying back and forth that I wanted to make sure I at least got this 
much out to you.  Clearly, I would like you to set up the meeting with the 
RiskCalc gurus to dig deeper into their methodology (as much as they'll tell) 
and the impact of missing variables and any other idiosyncracies the model 
may have.  Also try to get a better feel for how well they think the model 
extends to other countries.  I know from a previous conversation with Lea 
Carty and reading about the model that it was calibrated off North American 
names but they suspect it might extend to other countries.  Conversely, I 
know they are working efforts within Moody's Risk Management Services to 
develop private (and probably public) firm models specific to regions.  One 
they mentioned was Australia and also within Western Europe.  Data, as 
always, was the issue.  See if the quants have any perspective on their 
success/plan in that area.

Lastly, we truly appreciate the aggressive efforts you both have made toward 
pushing off this private model initiative and I'm very confident we'll be 
able to nail a firm plan down upon receipt of either D&B and/or Experian data.

Cheers,
Scott
---------------------- Forwarded by Scott Salmon/EU/Enron on 19/04/2001 20:51 
---------------------------


Mike Mumford@ECT
19/04/2001 17:55
To: Scott Salmon/EU/Enron@Enron
cc:  

Subject: RE: D&B Contact Names


---------------------- Forwarded by Mike Mumford/LON/ECT on 19/04/2001 17:59 
---------------------------
From: Iris Mack/ENRON@enronXgate on 19/04/2001 09:23 CDT
To: Mike Mumford/LON/ECT@ECT
cc: Amitava Dhar/Corp/Enron@ENRON 

Subject: RE: D&B Contact Names

Hi,

 Two RiskCalc sales people made a presentation this week about their product.

 They stated that they could set up a meeting with their quants and our 
people to discuss the model input and what happens if we have less than 17 
inputs.

 
Regards,
Iris

 -----Original Message-----
From:  Mumford, Mike  
Sent: Thursday, April 19, 2001 1:37 AM
To: Mack, Iris
Subject: RE: D&B Contact Names

Ooops... part II.

I just threw out 12 as a number for something we might develop... it's really 
not based on anything.  We know 17 will work for pre-packaged programs.  We 
also know there will be a great number of names with less than the full 17... 
I assume we would pursue another model (internal or external) to provide 
probably less accurate numbers but at least available for names with fewer 
inputs.  

Mike





From: Iris Mack/ENRON@enronXgate on 18/04/2001 15:02 CDT
To: Mike Mumford/LON/ECT@ECT
cc: Amitava Dhar/Corp/Enron@ENRON, Scott Salmon/EU/Enron@Enron, Eric 
Kirkpatrick/EU/Enron@Enron 

Subject: RE: D&B Contact Names

Hi Mike,

 Thanks for your email.

 What do you mean by "buying blind  all major data associated with these 
specific DUNS numbers"?

 I am puzzled about why you would be looking at 17 vs 12 fields.  I know 
RiskCalc requires 17 inputs and seem to apply that the model does not work 
for less than 17.  Yesterday the RiskCalc sales people could not guarantte 
that their model would work for less than the required 17 data inputs.  They 
suggested that this would be a question to direct towards their quants who 
developed the model.

Regards,
Iris

 -----Original Message-----
From:  Mumford, Mike  
Sent: Wednesday, April 18, 2001 1:20 PM
To: Mack, Iris
Cc: Dhar, Amitava; Salmon, Scott; Kirkpatrick, Eric
Subject: RE: D&B Contact Names

Iris,

Thanks.

After our meeting we stuck around to discuss things a little further.

Specifically with respect to global counterparty names... we have DUNS 
cross-references on about 70 of active names (13k out of 18.5k).  We could 
greatly accellerate purchase of some useful data by buying blind all major 
data associated with these specific DUNS numbers.

Pro  -  Data gets in on names we know we want to review ASAP, but without 
resolution on which specific fields are most useful.  We also get to test for 
completeness of data (how many have all 17 fields... how many have only 12 of 
which fields, etc.).

Con  -  Costs could be significantly higher... buying info we don't know we 
need (converse of buying only the 17 fields doesn't seem best answer).  We 
would end up buying additional info later... overhead costs and piecemeal 
buying will increase costs.  IT development time would be greatly 
increased... multiple file formats... developing to unknown maximum size, 
etc.... redoing when more data is purchased.


Any thoughts.



Mike