Risk Management at Enron

Presented by
Tanya Tamarchenko
Enron Research Group

23 February 2001
 Overview

 Enron today

 Value-At-Risk: Introduction

 Enron’s Value-At-Risk Model

 Extensions of Value-at-Risk

 Conclusions

Value @ Risk 2
 Enron Today

 Combination of physical presence in

the energy markets with expertise in
financial products

 A market-maker in energy derivatives

 Three core businesses:

- Cash and physical business
- Risk management
- Finance

Value @ Risk 3
 Top North American Gas & Power Marketers
(First Quarter 2000)

Natural Gas Electricity

(BCFD) 102.9 (BKWH)
20.6
87.9

12.0 55.0 54.4

10.8 10.7 50.4
9.9 9.6 45.6
8.5
7.7
28.2 26.0
&E
n

y
l

nt
la

a
ra
k
ro

eg

pr
ui

lia
Du

PG
Co
En

m
Aq

Re

Dy

Se

Value @ Risk 4
 Enron Today: International Portfolio
In Operation In Construction In Development
Power Plant Power Plant Power Plant
Pipeline Pipeline Pipeline
Other Other Other
E&P LPG Facility E&P
LPG Facility LNG Facility LPG Facility
LNG Facility Teesside
LNG Facility 1,875 MW
United Kingdom
London
Water Offices LNG Vessel
Sutton Bridge Poland
Gas Office Barge 790 MW Elektrocieplownia
Nowa Sarzyna
Distribution Locations Germany 116 MW
Wessex Bitterfeld
125 MW Jertovec
Electricity 240 MW
Italy
Distribution Milan Croatia
China
Zagreb
Turkey Beijing
Trakya Western-Eastern China South Korea
478 MW Natural Gas Pipeline Seoul

Italy Palestine Wuhan

Gaza SK-Enron
Sarlux EOG Loop Gas Distribution
Dominican 551 MW Power Project Sichuan-Wuhan
Republic Puerto Rico LNG Vessel Sichuan Natural Gas Pipeline LPG Import
Santo Domingo San Juan Dabhol- EOG - India
United Arab Emirates 135 Mc m Talasari 3,200 Bbls/d
3 E&P
EcoEléctrica Wuhan Qingshan
Puerto Plata Dubai Pipeline 38 MMcf/d 443 MW
Guatemala 507 MW Chengdu
Guatemala City 185 MW Logun Wuhan Wuchang Guam
LNG Facility
Puerto Quetzal Abu Dhabi 284 MW 189 MW Anigua
San Juan Gas Philippines
110 MW LDC 500,000 Tons Oman India
124 MW Jamaica LNG Export 1.6 MM Tons Delhi, Manila Piti
ProCaribe Trinidad and Hainan Island
Kingston Mumbai Thailand Subic Bay 80 MW
Tobago LNG Export Dabhol 160 MW 116 MW
EOG LNG Bangkok
IGL Progasco Vietnam Batangas
Nicaragua 115 MMcfd; Import
Venezuela 3,500 Bbls/d Nigeria Hanoi 110 MW
Corinto
70MW Caracas Lagos State Malaysia First Gas Power Co
Ventane Power Project India 2.6 MM Tons Fuel Supply
Dabhol LNG Export 30-35,000 Bbls/d
Panama Colombia Citadel Venezolana 90 MW Barge 2,184 MW
Bahía Las Minas Bogotá Benin 560 MW CCGT Phase I: 740 MW
Bachaquero III Singapore
355 MW Promigas Integrated Gas Phase II: 1,444 MW
Phase I: 245 MW 1,062 Miles Accro III & IV Gaspart Northeast & Power Project
Phase II: 110 MW Centragas CALIFE 80 MW
LDC Pantanal Energía 5 LDC
357 Miles 480 MW 20 Miles Indonesia
Phase I: 150 MW Jakarta
Phase II & III: 330 MW
Bolivia Maputo Iron &
Santa Cruz Brazil Steel Project
Cuiabá Pipeline Salvador, Sáo Paulo, Rio de Janeiro Integrated Steel Slab
TRANSREDES Manufacturing Facility 3.5 MTPY
3,093 Miles 385 Miles Riogás/CEG
LDC
Bolivia to Brazil Pande Gas Project
Pipeline Gaspart South 378 Miles
1,864 Miles Elektro 2 LDC
Eletricidade e South Africa Mozambique
Serviços Johannesburg Maputo
Obras Sanitarias Australia
de Mendoza Sydney
Argentina
Buenos Aires

TGS
4,104 Miles

Enron projects completed or underway

in more than 30 countries

Value @ Risk 5
 Value-at-Risk for Financial Instruments:
Definitions

 Value-at-Risk measures exposure of a portfolio of financial

instruments to potential losses resulting from:
- fluctuations of market prices, interest rates, exchange rates, etc.
- non-performance by counter parties

 Value-at-Risk technology has been developed for trading

portfolios. Use of Value-at-Risk for asset/liability management by
non-financial companies is a subject of debate.

Value @ Risk 6
 Value-at-Risk: Definition

 Value-at-Risk - dollar loss that may be

experienced in the value of a portfolio of
financial instruments:
– over a defined time period
– with a given probability
– due to market fluctuations

 Typically calculated for one day or two calendar

weeks (10 trading days)

 The time period may be selected to reflect

position liquidation time requirements

Value @ Risk 7
 Value-at-Risk: Illustration

Value @ Risk 8
 Value-at-Risk: Importance

 A requirement for any trading organization or a

major user of derivatives

 Internal factors:
– A summary of risks for senior management
and the Board of Directors.
– Tool for setting trading limits and for
performance measurement.

 External factors:
– Regulatory agencies, credit rating
agencies, creditors.

Value @ Risk 9
Extension of VAR to the Energy Industry

 Value-at-Risk developed originally for trading

portfolios of financial instruments
 Extension to portfolios of liquid, energy related
financial instruments is relatively straightforward
 Complications arise from inclusion of physical
transactions and/or extension to physical
operations of a plant or a pipeline
 Complications also arise from differences in the way
energy prices behave versus the way equity prices
or interest rates behave
 Value-at-Risk for equity stakes in private
companies is also not well defined
 Applicability of Value-at-Risk technology
to non-trading activities is a subject of a debate
Value @ Risk 10
 Backtesting

 Losses should exceed Value-at-Risk once

in 20 days on average at the 5% level

 Backtesting is hard for illiquid commodities

Value @ Risk 11
 0
100

20
40
60
80

(60)
(40)
(20)
1/ 5/ 1998
1/5/1998
2/ 5/ 1998

Value @ Risk
3/ 5/ 1998

4/ 5/ 1998

5/ 5/ 1998

6/ 5/ 1998

7/5/1998
7/ 5/ 1998

8/ 5/ 1998
Summer 1998
Power Price Spike

9/ 5/ 1998

10/ 5/ 1998
Curve Shift P&L

11/ 5/ 1998
Mitch
Hurricane

12/ 5/ 1998

1/ 5/ 1999

12
1/5/1999
Value at Risk

2/ 5/ 1999

3/ 5/ 1999

4/ 5/ 1999

5/ 5/ 1999
V@R Limit

6/ 5/ 1999

7/ 5/ 1999
(1/5/98 to 12/30/99)

7/5/1999
8/ 5/ 1999

9/ 5/ 1999
Backtesting ENRON V@R

10/ 5/ 1999

11/ 5/ 1999

12/ 5/ 1999
12/5/1999
 Energy Value-at-Risk
Numerical Techniques

 Analytical: Mean/Variance Technique

 Simulation
- Historical
- Monte Carlo

 Stress tests (predefined scenarios)

 Enron’s recommendation:
- Monte Carlo simulations
- supported by statistical analysis of historical prices;
- full re-pricing of entire portfolio in spite of computational
burden;

Value @ Risk 13
 Enron’s Value-at-Risk Model

• Inputs:
– Positions for each commodity (gamma – delta approach currently);
– Forward price curve for each commodity (provided by traders);
– Volatility curve for each commodity to be used in simulations;
– Correlations across commodities as well as across forward contracts
of different maturity;
• Outputs: distribution of returns for each portfolio in the portfolio hierarchy

ENRON

GAS POWER LIQUIDS

Value @ Risk 14
 VAR: Simulation Mechanism

 Portfolio holdings (positions) are represented through

Deltas (i) and Gammas (i), i=1,2, … N.

 Obtain new price, Fi,sim, from Monte Carlo simulations

 Change in position value for the contract i is:

i*(Fi,sim - Fi) + 0.5*i*(Fi,sim - Fi)2

 A number of simulations of Portfolio Value Change are

calculated

 The 5th percentile is used to obtain VAR.

Value @ Risk 15
Modules of Enron’s Value-at-Risk Model
 US Gas: forward contracts on NYMEX plus
hundreds of basis locations
– quite developed market;
– prices are liquid up to a few years forward;
– history of prices available for more than
5 years daily;
– highly seasonal prices;

 US power: forward contracts for peak and off- peak

power for different US regions
– developing market, deregulation started recently;
– prices behavior changes from one year to the next;
– highly seasonal prices;
– low liquidity beyond 12 months forward;

Value @ Risk 16
Modules of Enron’s Value-at-Risk Model
(continued)

 Other commodities: liquids contracts (Crude Oil,

Unleaded Gasoline, Propane, Butane, etc), Coal,
Pulp and Paper, Condensate;

 European markets: UK Gas, UK Power,

Continental Gas, Continental Power, Nordic
Power;

 MG metals positions;

 Intramonth gas positions;

 Intramonth power positions;

Value @ Risk 17
 Main Concepts of Enron’s VAR Model
(continued)

Simulation of forward prices term-structure:

 A portfolio of commodity contracts has exposure the
entire term structure of forward prices
 Reshaping of the forward price curve may significantly
affect the performance of a portfolio
 Modeling of the term structure requires capturing the
correlations between different prices along the forward
price curve

Value @ Risk 18
 HJM Model
(Example: PCA results for Crude Oil)

0.8

0.6

0.4

0.2

-0.2 1 4 7 10 13 16 19 22 25 28 31 34 37 40 43 46 49 52 55 58

-0.4
Shift factor
-0.6
Slope factor
-0.8 Reshaping factor
-1
Contract maturity

Value @ Risk 19
 Main Concepts of Enron’s VAR Model
(continued)

 Statistical analysis of historical forward prices for

selected subset of all traded curves:

- recent history of forward prices is used;

- parameters are estimated for selected price processes, the data
fit is examined;
- correlations are captured across term structure as well as
across different commodities;
- analysis is repeated regularly;

Value @ Risk 20
 Volatility: Implied versus Historical

 The critical input which is not directly observable

 Volatility can be inferred from the option prices or estimated from

historical prices

 Historical volatility: standard deviation of logarithmic price returns:

In (Pt /Pt-1), annualized through multiplication by a relevant factor

Value @ Risk 21
 NG Prompt Month Volatilities
(Historical and Implied) from 1/1/95 to 9/12/00)

Forward volatility

Historical forward forward volatility

1.6

1.4

1.2

0.8

0.6

0.4

0.2

0
1 101 201 301 401 501 601 701 801 901 1001 1101 1201 1301

Value @ Risk 22
 Extensions of Value-At-Risk Model

 Incorporating different price processes:

- Brownian motion (GBM)

- Extension of GBM for forward prices (HJM)
- Mean-Reversion models (floor reverting processes)
- Jump-Diffusion model
- Combination of Mean-Reversion and Jump-Diffusion
model;
- Volatility switching models;
- Capturing “fat tails”;

Value @ Risk 23
Value @ Risk
Price ($)
10

0
1
2
3
4
5
6
7
8
9

1/3/1995

7/3/1995

1/3/1996

7/3/1996

1/3/1997

7/3/1997

24
1/3/1998

Tim e
7/3/1998

1/3/1999

7/3/1999

1/3/2000
NG prompt contract prices (1/1/95-02/15/00)

7/3/2000

1/3/2001
Choosing Appropriate Price Process
 Choosing Appropriate Price Process

North Path 15 prompt contract prices

(12/98-02/15/00)

350

300

250

200

150

100

50

0
12/1/1998

2/1/1999

4/1/1999

6/1/1999

8/1/1999

10/1/1999

12/1/1999

2/1/2000

6/1/2000

10/1/2000

12/1/2000

2/1/2001
4/1/2000

8/1/2000
Time

Value @ Risk 25
 Risk Analytics: Component VaR

 VAR decomposition/dissection by
– time-buckets
– traders
– sub-portfolios
 Useful in identifying risk-contributors and
hedges in a complex portfolio.

Value @ Risk 26
 Component VaR- Desk
Component VaR by Desk

8/25/00
8/24/00
8/21/00
In 000s

Desk5 added short position on 8/25

Desk1 Desk2 Desk3 Desk4 Desk5 Desk6

Value @ Risk 27
 Conclusions:
Challenges in Implementing Effective Risk
Management System in an Energy Firm

• Calibration of the model: choosing appropriate price processes

and their parameters estimation, volatility specification;
• Capturing price patterns of energy prices: seasonality, gapping
behavior;
• Capturing extreme events, the tails of distributions, non-normal
returns;
• From “risk measurement” to “risk management” - making risk
modeling system an active tool in portfolio management process;

Value @ Risk 28