Challenges in Capital Adequacy UH-GEMI 3rd Annual Energy

1
Challenges in Capital AdequacyUH-GEMI 3rd Annual
Energy Trading Marketing Conference Rebuilding
the BusinessHouston, TexasJanuary 20, 2005

• Laurie Brooks
• VP Risk Management and Chief Risk Officer
• Public Service Enterprise Group

UNIVERSITY of HOUSTON Global Energy Management
Institute
2
Capital Adequacy and Capital AllocationConnected?

• Capital Adequacy
• How much capital is required to achieve the
  companys stated goals and objectives?
• Capital Allocation
• How should corporations allocate capital between
  competing demands?

3
Capital Adequacy for Energy Transactors
  1.  Capital for what?         Business models
regulated utilities, merchant generators,  marketi
ng and trading entities         Economic capital
vs liquidity adequacy                 Banking
models                 SP liquidity survey
Measures - EaR vs CFaR, role of stress
testing, market risk vs credit risk trade-offs,
role of ECE and PFE 2.    Why energy is
different - impact of following on margin/cash
requirements                  volatilities       
           sector ratings      
storability                  regulatory
intervention                  age and depth of
markets                  contract terms
                 risk  mgt tool
availability  3.    Capital how?         Access
to capital markets         Diversification of
cash flows         Credit mitigations             
    role of netting and clearing                
stair stepped margining agts.  
4
Capital Use by Activity
5
Market Risk Trading vs. Non-Trading Activities
Non-Trading
Trading

• Positions to facilitate marketing
• Proprietary trading positions

• Positions generated by asset/customer business
• Strategic buy and hold hedges

Purpose

• Illiquid or buy and hold positions
• Holding period measured in months/years

• Liquid, actively funded positions across many
  markets
• Holding period measured in days/weeks

Liquidity

• Asset/customer-driven embedded options
• Long holding period makes non-linearity material

• Price-driven exchange traded or OTC options
• Short holding period allows linear approximations

Optionality

• Long-term volatilities and correlation
• Mean reversion, seasonality simulation, Earnings
  at Risk

• Short-term volatilities and correlation
• Jump diffusion, intra-day VaR analytical,
  simulation

Valuation
Risk Management/ Intervention

• Structured solutions, contract renegotiations,
  asset sales and purchases
• Management of regulatory process

• VaR limit reduction, stop loss limits, hedging
  with traded instruments

6
Key Concepts of Capital Adequacy Three Risk Types
The framework for determining capital adequacy
for economic value requires an estimation of
economic capital and thus quantifying the
following significant risks

• Market Risk - Variation of portfolio market value
  due to a change in a market price or rate, as
  well as a change in energy demand
• Credit Risk - Variation of portfolio market value
  due to default or a credit downgrade of an issuer
  or counterparty
• Operative Risk (term to address Operations and
  Operational risk collectively)
• Operations - The risk associated with delivering
  or producing physical energy
• Operational - The risk of direct or indirect loss
  resulting from inadequate or failed internal
  processes, people, and systems or from external
  events

7
Key Concepts of Economic Capital Adequacy Market
Risk
Modeling Approaches
Price Behavior Process
Market Exposures
Pros/Cons
Comments
Analytical
Closed-form approach for modeling price movements
Works well for linear type exposures

• Pros
• Simple and fast
• Easy to change as assumptions change
• Cons
• Does not capture optionality well
• Minimal ability to model complexities over a
  longer period of time

• Works well for determining shorter-term price
  moves for a trading portfolio
• Can be used as a quick metric to help manage
  portfolio positions

Simulation
Robust methodology for mean reversion, jumps,
linking, spot, and forward prices
Full revaluation at each price iteration better
approximates nonlinearity of asset/option
positions

• Pros
• Robust
• Captures optionality
• Provides a full distribution of outcomes
• Cons
• Complex to construct the simulation model
• Only as good as model input parameters
• For historical simulation, values are constrained
  to conform to history which may be irrelevant due
  to market, economic, or regulatory changes

• As the time horizon is extended and the need to
  model certain energy price characteristics
  increases, simulation becomes a more suitable
  solution. Meanwhile, the technical difficulties
  increase and the model needs to be modified to
  fit the long-term simulation purpose.

8
Key Concepts of Economic Capital Adequacy Credit
Risk

• Expected Loss
• Represents the average loss that a company could
  expect to incur over a given horizon
• Unexpected Loss
• Measures the uncertainty of losses around the
  expected loss

9
Key Concepts of Economic Capital Adequacy
Operative Risk Scorecard
CA Framework Key Concepts

• Scorecard Approach
• Can be used for operations and operational risk
  to identify risks, determine frequency and range
  of costs, and assesses the effectiveness of
  controls and mitigation techniques in place. It
  is subjective, but now that the SEC has mandated
  the COSO framework for Sarbanes Oxley 404
  compliance, standards will be set. In particular,
  the Capability Maturity Model can be adapted to
  set standards for a scorecard approach and is
  already used by many audit firms. Additionally, a
  company may want to use CCRO Best Practices from
  earlier white papers as a qualitative assessment
  of where companies stand with regard to CCRO
  recommendations.
• Regardless of the scorecard criteria used, a
  scorecard approach can form the basis for
  continuous improvement processes for internal
  controls to mitigate operative risk. It can also
  reflect improvement in the risk-control
  environment in reducing the severity and
  frequency of future losses.

10
Key Concepts of Economic Capital
AdequacyOperative Risk Risk Taxonomy
CA Framework Key Concepts

• The risk taxonomy is a system for organizing
  types of operative risks by serving as a family
  tree, aggregating risks by various
  characteristics. The level of aggregation at
  which each characteristic presents itself may be
  determined individually.
• There is no standardized risk taxonomy, but
  certain characteristics should be used to create
  the groupings
• Risk classes (people, processes, systems, asset
  damages) the broadest classes of risks
• Subcategories could include whether the risk is
  internal or external, a type of fraud, or a
  natural disaster
• Risk activity examples specific activities or
  events that could cause a loss, such as rogue
  trading, hurricane, model risk, or pipeline
  rupture.

11
Key Concepts of Liquidity Adequacy

• Fixed Payments - This would include, but is not
  limited to fixed charges such as debt service,
  dividends, debt/equity retirement and current
  portion of committed, maintenance and
  non-discretionary capital expenditures.
• Contingent Liquidity Contingent liquidity is
  synonymous with unexpected change or variation in
  liquidity. While economic capital protects
  against losses in the companys economic value,
  contingent liquidity is held to support the risk
  of unexpected reduction in cash. Includes
• Cash Flow at Risk
• Trigger events
• Downgrade event
• Loss of threshold
• Adequate assurance
• Debt/equity trigger
• Contingency events
• Operational/Operations Risk
• Credit/counterparty termination default

12
Key Concepts Combined Capital
CA Framework Key Concepts
Methodology
Description
Advantages
Disadvantages
Assumption
Simple Sum
Derive economic capital for credit, market, and
operative risk, then sum them

• Easy to implement
• Most conservative view of risk

• Overestimates risk
• Results in the lowest level of capital adequacy

Correlation assumed to be perfect among risk
components
Modern Portfolio Theory
From historical data, determine an explicit
correlation among credit, market, and operative
risk economic capital
Attempts to represent the actual correlation
among risks, rather than a conservative assumption
Requires a time series of credit, market, and
operative risk economic capital that is
reasonably robust
Assumes that some risks are uncorrelated,
allowing for lower risk and improved capital
adequacy
Monte Carlo Simulation
Using consistent parameters, simulate risk
factors to produce a joint distribution of
outcomes
The most robust perspective of risks and their
interaction if modeled correctly

• Requires a large amount of research, analytical,
  and technical resources
• Ensuring assumptions are correct is critical

Material risk inputs can be parameterized
accurately
13
Key Concepts Correlation Math Refresher
CA Framework Key Concepts

• In a two asset portfolio with equal investment in
  assets A and B, the VaR of the portfolio (at 95
  confidence) VaRAB 1.65 ?AB where ?AB is the
  standard deviation of returns of the portfolio
• where ?AB is the correlation between AB
  (do the returns move together?)
• Remember (ab)2 a22abb2 and
• Then if ?AB 1
• So Portfolio VaR VaRA VaRB!
• If ?AB0, (Square root sum of squares)
• The truth 0 lt ?AB lt 1 lies somewhere in between
  and
• lt ?AB lt ?A?B
• Square root sum of squares Simple Sum

14
The Risk Management team at PSEG demonstrated the
CCROs framework using a sample asset portfolio.

Example

• This example illustrates how the CCRO framework
  can be used in practice
• We will walk you through the following
  implementation steps
• Portfolio setup
• Methodology
• Pre-simulation
• Simulation
• Results
• We will also discuss some of the firm and systems
  resources required

Please refer to pages 61-67 of the white paper
for a full description of the example.
15
We chose to model the asset-level impacts over a
year of different risks on a company over time.
Example Setup

• We modeled market, credit and operative risks
  jointly in one simulation versus separately
• Felt there was better intuition and that we could
  better justify a choice of the assumptions
• Calculation process seemed clear based on this
  approach
• Used a 1-year holding period and ran 5,000 trials
  with a 95 CI
• We modeled a five-year time horizon, with price
  changes modeled as follows
• Year 1 spot
• Year 2-5 forward prices
• We chose a variety of assets and parameters.
• Three different generating assets and fuel types
• Assets are in three different pools

Generating Plant
Power Pool
Capacity
VOM
Heat Rate
Fuel Type
Book Value
Gas-fired combined cycle
ECAR
850
3.98
7.25
Natural Gas
510,448,931
Coal-fired, base load
NEPool
375
2.51
10.3
Coal
49,720,351
Jet kero-fired peaking
PJM
500
34.48
15.7
Jet Kero
11,094,684
16
Market Risk Calculations
Example Setup

• Unhedged market risk
• Minimum (realized generation over 12 months)
  (Expected generation value of the remaining
  term) (Initial expected value of the
  generation)
• Hedged market risk
• (Unhedged market risk) (Realized and unrealized
  trading profit or loss)

17
Credit Risk Calculations
Example Setup
Counterparty
Rating
1-Year Probability of Default
Commodity
Counterparty A
CCC
27.87
Fuel coal, natural gas, jet kero
Counterparty B
BBB
0.34
Power NEPool, PJM, Cinergy
Counterparty C
BB
1.16
Fuel and power

• Calculated as the sum of credit loss across the
  twelve months of simulations, as a function of
  counterparty risk and power pool risk
• The company has three counterparties
• Counterparty A is used for fuel procurement
• Counterparty B is used for power sales
• Counterparty C is used for speculative trading.
• The recovery rate is assumed to be 10.
• Each power pool has collateral requirements that
  are a function of the companys credit rating,
  tangible net worth and activity in the pool
• Value is calculated under two potential ratings,
  BBB (credit limit 80,000,000) and BB (credit
  limit 4,000,000)

18
Operative Risk Calculations
Example Setup

• Operations loss
• Sum of lost profit from plants not running at
  full capacity
• Operational loss (if applicable)
• Hidden trade on the books whose value is set to
  the largest negative value of all the trading
  positions on the book.

19
Liquidity calculations
Example Setup
Liquidity risk is defined as the minimum cash
flow point in a simulation.

• Prior month realized P/L (retained earnings)
• Current month generation P/L
• Collateral posted
• Accounts receivable
• Accounts payable
• Full margin on mark-to-market
• Credit loss
• Operations loss
• Operational loss

Monthly cash flow
20
Hedging affects liquidity in offsetting ways.
Example Setup

• Liquidity risk is increased by hedging in the
  following ways
• Creates cash outflows due to full margining on
  mark-to-market
• Creates the possibility of credit loss
• Liquidity risk is decreased by hedging in the
  following ways
• Decreases the amount of cash needed to be posted
  to power pools since that is determined by net
  activity.
• Decreases the distribution of realized P/L from
  generation

The net effect of hedging was a decrease in the
liquidity risk.
21
Three key methodology choices drive our model
Example Methodology
Method
Pros
Cons
Joint simulation of credit, market, and operative
risks (versus assumed correlations)

• Consistency
• More data available to check micro relationships
  rather than portfolio relationship
• Can change micro assumption and rerun
• Are not assuming answer

• Increases memory need and computer time
• Necessitates more simplifying assumptions,
  leading to less accurate estimates of component
  risks

Risk modeling
Correlated Brownian Motion for Energy Forward
Prices

• Most practical method with 3 power pools and 3
  types of fuel for 5 years
• Would be difficult to jointly calibrate more
  complex model for diversity and tenure of
  portfolio

• Easier to believe for forward prices rather than
  spot prices still oversimplifies reality
• Probably overstates volatility for longer-dated
  contracts

Energy forward prices
Daily power prices are normally distributed with
mean equal to forward price and standard
deviation equal to historical daily spot standard
deviation

• Allows for analytical determination of MWs of
  generation and generation value
• No need to do daily simulation

• Ignores operating constraints on plants
• Splitting monthly prices into two normal
  distributions (normal and extreme days) captures
  peaking value more accurately
• Does not allow for fuels to vary by day

Daily power prices
22
Pre-Simulation prior to running our simulations,
we calculated a number of initial values.
Example Pre-Simulation
Pre-Simulation Calculations

• Initial expected value of the assets
• Calculated based on the current forward prices
  for fuels and power
• Expected fuel purchases and expected output to be
  sold to counterparties
• Calculated based on current forward prices
• Randomly-generated positions in power and fuels
• Constrained to be a quarter of the size of
  outright positions
• Used to simulate a speculative trading operation

23
Simulation we generated the inputs to credit and
operational performance.
Example Simulation
Correlated
Market risk simulation
Generation model
forward prices
Market risk
- power
Marginal cost
of fuel (VOM
Correlated
heat rate)
forward
prices - fuel
MTM - A/R -
A/P on trading
Credit excess/loss
contracts
Credit risk simulation
Probability
of default
Operational profit/loss
Probability
of outage
Operative risk simulation
Probability of trader misconduct
60 product months x 6 products x 12 monthly
steps of random standard normal pulls 7 risks
x 12 monthly steps of uniform random variables
pulled
24
Results Unhedged vs. Hedged Assets
Example Results
By hedging assets, market risk is reduced by less
than the additional economic capital required for
credit risk, increasing economic capital adequacy.

Note the simulation was also run with all
counterparties set at BBB to reflect the average
rating of many portfolios. The credit risk
remained at zero with a 95 confidence level,
while market risk was reduced from 23 million to
6 million.
25
Results Portfolio Effect
Example Results
Illustration of the mathematical factEC? 0
(square root sum of squares) lt EC lt ? lt 1 (Monte
Carlo simulation) lt EC?1 (simple sum)
Available vs. Required Capital
Sq. Root
Monte Carlo
( millions)
Sum of Squares
Simulation
Simple Sum
By analyzing capital requirements for unhedged
assets as part of a portfolio vs. individually,
the example illustrates how diversification
reduces the economic capital required for market
and operative risks.
Net Assets - Debt
285.6
285.6
285.6
Required Economical Capital
Market Risk
22.5
22.5
22.5
Credit Risk
0.0
0.0
0.0
Operative Risk
23.2
23.2
23.2
Diversification Effect - Across Risks
-13.4
-11.8
0.0
Total Required Economic Capital
32.3
33.9
45.7
Economic Capital Adequacy
253.3
251.7
239.9
Diversified
Available vs. Required Capital
Combined-
Total Individual
Component
( millions)
Disclaimer the closeness of the Monte Carlo (MC)
and Square Root Sum of Squares is not
representative. In general, one shouldnt assume
that SRSS is a good proxy for MC.
Coal
Cycle
Peaking
Assets
Total Portfolio
Risk
Net Assets
49.7
510.4
11.1
571.3
571.3
Debt
24.9
255.2
5.5
285.6
285.6
Required Economical Capital
Market Risk
7.0
27.6
3.5
38.1
22.5
-15.7
Credit Risk
0.0
0.0
0.0
0.0
0.0
0.0
Operative Risk
22.3
3.4
2.3
27.9
23.2
-4.7
Diversification Effect - Across Risks
-11.1
-2.9
-1.6
-15.6
-11.8
3.8
Total Required Economic Capital
18.2
28.1
4.1
50.5
33.9
-16.5
Economic Capital Adequacy
6.6
227.1
1.4
235.2
251.7
26
Why Emerging Practices?
Example Results

• These are recommendations for internal use and
  experimentation for companies to better
  understand and quantify the capital and cash
  requirements of the merchant energy business
  these are not recommendations for external
  communication or new disclosure.
• No one is going to implement all of these
  recommendations over night.
• Most of us have some capability to begin looking
  at the components of Capital Adequacy and
  liquidity requirements through the use of tools
  that we already have in place but which require
  extension and modification to achieve the more
  sophisticated views that result from the white
  paper recommendations. This should be a
  controlled evolutionary process - in most cases,
  the less sophisticated tools that we already have
  in place generate more conservative answers than
  the sophisticated approaches do.
• Why we will implement these ideas over time
• Better than what we have now
• Emphasize need to look both long term and short
  and to look at cash flow as well as earnings and
  value
• Ideas and methodologies useful in decision making
•