---------------------- Forwarded by Steven J Kean/HOU/EES on 04/20/2000 01:44 
PM ---------------------------
From: Vladimir Gorny@ECT on 04/20/2000 01:27 PM
To: James D Steffes/HOU/EES@EES, Karen L Barbour/HOU/ECT@ECT, Scott 
Stoness/HOU/EES@EES, Scott Gahn/HOU/EES@EES, Harry Kingerski/HOU/EES@EES
cc: Ted Murphy/HOU/ECT@ECT, David Gorte/HOU/ECT@ECT, Steven J 
Kean/HOU/EES@EES 
Subject: Curve Validation

Outlined below are some details of the existing curve validation process and 
its application to the proposed validation of tariff curves.

Wholesale Curve Validation

The existing process incorporates validation of all wholesale curves and 
includes monthly validations and random curve reviews. Every month, results 
of the validation are summarized in a memo and distributed to the Senior 
Management. Results of the random curve reviews are brought up to the 
attention of Desk Heads responsible for the respective curves and Senior 
Management as necessary.

Curve Assessment
 Green: 
  - curve is reasonable
  - small P&L impact from curve shift
 Yellow:
  - curve is illiquid
  - medium-to-large position and P&L impact from curve shift
  - some ability to validate prices from external sources
 Red:
   - curve mark is inaccurate (based on external data obtained)
  -  curve is illiquid: no ability to validate externally
  - significant positions and P&L impact from curve shift

Curves that are inaccurately marked based on data obtained from external 
sources are urged by the Senior Management to be corrected. In the cases 
where external data is not available or clear, the Senior Management urges 
the respective Desk Head not to increase an already large illiquid position 
or to built in a risk premium by widening the respective bid/offer spread and 
not to take on positions inside of that spread.

Selection Criteria
The criteria for monthly curve selection are the following:
 P&L sensitivity
 Largest positions
 Staleness (curves have not been marked for a period of time)
 Completeness (curves that have not been validated in a long time)

The main criteria for random curve validation is new deals.

Curve Validation Tools
Broker Quotes/Other external data
Graph Analysis (curve slope vs. a peer group, curve humps/sculpting vs. a 
peer group, curve shift vs. position shift, etc.)
New Deal Analysis (comparison of curve marks and executed/quoted prices)
Boundary Analysis/V@R Simulations (comparison of curve to simulated curve 
boundaries using Enron's V@R Engine)
SAVA Model Comparison (comparison of curves to model generated results)


Tariff Curve Validation

Scope/Timing:
ASAP - 20 curves with the largest positions related to deals pending review: 
(Frito-Lay, Quaker Oats, Starwood) - to ensure timely and smooth approval of 
these deals.
Next 30-days - 20 curves with the largest positions/P&L impact
Monthly/Quarterly - 20 curves selected based on criteria described above
Randomly - curves with significant positions/P&L impact, resulting from new 
deals

Data Gathering:

This Month:  Listing of 20 curves (related to deals in progress) and the 
related support requested by EGA (to be provided by Scott Stoness)
On-going Basis:  The process of obtaining data should be streamlined and the 
curve validation process should be relatively painless:
    - RAC should have access to the curve data (Minal Dalia is fully devoted 
to EES and is currently residing on the EES floor)
    - Curve assumptions should be documented (to be provided by Scott Stoness)

Execution/Reporting:

RAC, EGA and EES Risk Management Group work together on curve validation 
utilizing available tools and work on developing new tools 
Utilize the Wholesale Curve Assessment methodology (Green, Yellow, Red) 
Generate a curve validation report (see template attached)
Provide feedback to the Desk Heads and distribute the Curve Validation Report 
to Enron and EES Senior Management 



Other Curves

Other curves related to EES business should be gradually incorporated into 
the aforesaid process (ancillary services, congestion, DSM and labor related 
curves) - let us set a timetable.

In conclusion, curve validation results should be evaluated in conjunction 
with other risk analytics: stress testing of prices, volatilities and 
correlations, Value-at-Risk/Capital-at-Risk analysis, performance measures, 
capital allocation framework, etc.

Vlady.