FYI

Vince
---------------------- Forwarded by Vince J Kaminski/HOU/ECT on 02/12/2001 
11:22 AM ---------------------------


Eugenio Perez
02/08/2001 01:13 PM
To: Vince J Kaminski/HOU/ECT@ECT, Stinson Gibner/HOU/ECT@ECT, Rudi 
Zipter/HOU/ECT@ECT
cc:  
Subject: New Merchant Asset VaR Model

I was recently informed of a risk-reducing strategy that we have on the 
Merchant Asset portfolio.  It seems that we have a vertical bear spread 
hedging our holdings of Hanover Compression.

Since these holdings represent about 10% of the porfolio, I felt that 
parametric VaR would not be accurate enough.  As a result, I have changed the 
model to a Monte Carlo simulation with the following details:

Following my conversation with Vince, I changed the method of parameter 
estimation to RiskMetrics standard (assume a zero mean and estimate daily 
volatilities and correlations using exponential weighting with a 0.94 lambda).

I use the Cholesky decomposition to produce correlated random normals, which 
I then use to simulate Geometric Brownian Motion prices.

I treat the Hanover Compression stock holding and bear spread together as a 
synthetic short put at 34 and long call at 92.  I do full revaluation of the 
options with the Black-Scholes.  I keep volatility and the interest rate 
constant for the sake of computational speed.

I write my Monte Carlos through VBA and formulas in Excel, so this model does 
not need Crystal Ball.

Regards,



Eugenio