Embed Size (px)
Citation preview
Risk Analysis for Mergers, Acquisitions, and Asset Valuation
Steve R. Dean, ASA, PEDAI Management Consultants, Inc.
2
Using Market Data in Risk Analysis
AgendaData AvailabilityThe Value of Data-Informed Risk AnalysisCapital Expenditure and Regulatory Risk
Analysis Applications Using Market Data
3
Using Market Data in Risk Analysis
Data Availability
Why Market Data?
The Rise of Data
Market Centers
Important Characteristics and Limitations
The Value of Data-Informed Risk Analysis
Capital Expenditure and Regulatory Risk Analysis Applications Using Market Data
4
Why Market Data?
Practical Reasons
Market data provide the clearest picture of what actual firms are paying for actual commodities
No subjectivity
We can shift from what should this asset be worth? to what is this asset worth?
Equilibrium in electricity markets can be fragile and elusive
Strategic Reasons
Prices as information (volume and liquidity)
Market data tell us how markets evolve
Bottom Line
Using actual market data produces more accurate values and improves management and investor decision making
5
The Rise of Data
From System
to Energy Clearing Prices (ECP)
Volatility is a consequence of healthy, competitive markets
Most Important Data for Risk Analysis
Primary Markets
Energy
Capacity
Ancillary Services
Secondary Markets
Emissions Credits
Fuel Prices
$0
$50
$100
$150
$200
$250
Jan-93
Jan-95
Jan-97
Jan-99
Jan-01
Jan-03
Peak LambdaPeak Load PriceOff Peak Price
NEPOOL Data: Introduction of ECP
6
Data AvailabilityWEAK
STRONG
AVERAGESTRONG
7
Energy Products
Four Key Features Distinguishing Types of Prices
Zonal vs Nodal
Zonal reduces market power potential; nodal is theoretically more efficient
Spot vs Long-Term Bilateral
Spot markets promote competition, but bilateral contracting moderates volatility
Portfolio vs Unit-Specific Scheduling and Bidding
Portfolio scheduling allows more flexibility, but makes it harder to manage congestion
FTR (Financial Transmission Rights) vs TCR (Transmission Congestion Rights)
Both rights are financial, not physical
8
Energy Products
Because of the physical characteristics of electricity, the price to generate electricity must be considered together with the price to transport it
Locational Marginal Prices
LMP = Energy + Congestion + Losses
Energy: marginal costs of generation
Congestion: compensation for out-of- merit-order dispatch due to transmission constraints
Losses: cost of line losses
The congestion premiums may be substantial during peak load periods for some regions (compared to equilibrium forecasts of marginal fuel- based energy prices)
www.nyiso.com
9
Contrasting Energy Markets
ERCOT
Zonal: 4 zones
Bilateral market with real- time balancing
Portfolio scheduling and bidding
TCRs
PJM
Real-time and day-ahead markets
Hourly LMPs
Nodal:
12 transmission zones
4 interfaces
4 hubs
1 overall load-weighted average
FTRs
10
Capacity Products
Used to ensure that sufficient resources exist and are accessible to system to satisfy peak load and protect reliability
Installed Capacity (PJM 1/1/1999 – 5/31/1999)
Total electricity resources required to meet peak system load over the planning period in accordance with reliability standards
Unforced Capacity (PJM since 6/1/1999)
Effective available capacity of a resource when forced outages (planned and scheduled maintenance) are taken into account
UCAP = ICAP x Average Unforced Outage Rate
Daily, Monthly, Multi-Month contract terms available for unforced capacity clearing market
11
Capacity PricesCapacity Credit Market - PJM East
$0
$50
$100
$150
$200
$250
$300
$350
$400
Jan-99 Dec-99 Dec-00 Dec-01 Dec-02
$/M
W-d
ay
0
500
1,000
1,500
2,000
2,500
3,000
3,500
4,000
4,500
Tota
l Cap
acity
(MW
)
Capacity PriceCapacity Cleared
12
Developing Markets
Ancillary Services
Real-power Balancing (frequency stability)
Voltage Stability (for customers)
Transmission Security
Economic Dispatch
Financial Trade Enforcement
Black Start Capacity/Regulation
Where will we see markets form?
Great interest in reducing the “socialized” nature of these services
13
Ancillary Service Markets (?)
Real-power Balancing (frequency stability)
The most “well-developed” ancillary service market; “spinning” reserves
Voltage Stability (for customers)
No possibility for bilateral market-making because of the externalities (how A and B trade with each other can affect C)
Centralized exchange in parallel with real-time market is possible, but difficult due to the difficulty in transmitting reactive power (highly localized; losses are about 10x greater than real power)
Transmission Security
Service must be provided by the ISO
ISO can sell transmission rights, but physical coordination in real-time is required
The “security-constrained unit commitment problem” is very difficult
Economic Dispatch
Centralized day-ahead power exchanges (like PJM)
Financial Trade Enforcement
ISO must provide (much like CBOT’s OCC)
Black Start Capacity/Regulation
Centralized market is possible
14
Apples and Oranges: Market Comparability
Zonal vs Nodal Pricing
On- vs Off-Peak Pricing
Firm vs Non-Firm Pricing
“Into” Pricing/Congestion
Liquidity and the informational content of prices
“Representative,” “Indicative,” and one-sided prices
Past as prologue?
What doesn’t looking at the past tell us?
15
Using Market Data in Risk Analysis
Data Availability
The Value of Data-Informed Risk Analysis
Differences between typical assumptions and actual data
The valuation impact of using actual market data
Capital Expenditure and Regulatory Risk Analysis Applications Using Market Data
16
Standard Modeling Assumptions
Prices are lognormal
Volatility is constant
Are these valid?
How big are the differences?
17
Hypothetical vs Actual DataHenry Hub Gas Prices
0%5%
10%15%20%25%30%35%
$1.40 $2.20 $3.00 $3.80 $4.60 $5.40
$/MMbtu
Freq
uenc
y
ActualLognormal
Error Between Hypothetical and Actual
0%
5%
10%
15%
20%
$1.40 $2.40 $3.40 $4.40 $5.40
$/MMbtu
Pro
babi
lity
Entergy Electricity Prices
0%5%
10%15%20%25%30%
$10 $20 $30 $40 $50
$/MWh
Freq
uenc
y
ActualLognormal
Error Between Hypothetical and Actual
0%
5%
10%
15%
20%
$10.00 $22.50 $35.00 $47.50
$/MWh
Pro
babi
lity
18
Hypothetical vs Actual DataTime-Dependent Volatility
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Jun-99
Dec-99
Jun-00
Dec-00
Jun-01
Dec-01
Jun-02
Dec-02
Two-
Mon
th R
ollin
g C
oeffi
cien
t of V
aria
tion
Clustered Volatility
0%
200%
400%
600%
800%
1000%
1200%
1400%
1600%
May-99
Nov-99
May-00
Nov-00
May-01
Nov-01
May-02
Nov-02
Squ
ared
Pric
e C
hang
e
19
More on Volatility…
20
The Valuation Impact of Hypothetical vs Actual Data
“…but does it matter?”
Estimate the spark spread-determined capacity factor of a GTCC unit buying at Henry Hub and selling into Entergy
8500 heat rate, daily discretion
Use actual 2002 data vs traditional assumptions
Test against actual 2003 data (January-May)
21
Actual 2002 Results
Average Std. DeviationEntergy 28.50$ 5.33$ Henry Hub 3.36$ 0.72$
52.70% (whole year)56.60% (January - May only)
Capacity FactorCapacity Factor
…but what about 2003?
22
Comparative 2003 Forecasts
Conventional “Normal” Assumption
Prices follow “conventional” distribution, based on 2002 mean and standard deviation
Capacity Factor 49.1%
Simulation Approach with Historical Data
Prices incorporate all aspects of historical data (mean-reversion, stochastic volatility, etc.), as well as more realistic distributional forms
Capacity Factor 18.5% ± 0.6%Simulation also provides more data about plant operation
Actual January-May 2003 Plant Performance
Capacity Factor 17.9%
($5.0)
($2.5)
$0.0
$2.5
$5.0
$7.5
$10.0
Jan-03 Feb-03 Mar-03 Apr-03 May-03$/
MW
h
Actual Margin Average Spark Spread5th Percentile 95th Percentile
Average spark spread rising as summer months approach
23
Using Market Data in Risk Analysis
Data Availability
The Value of Data-Informed Risk Analysis
Capital Expenditure and Regulatory Risk Analysis Applications Using Market Data
Economic Analysis of Emissions-Related Capital Expenditures
Regulatory Risk Analysis of Fleet Configuration Decision-Making
24
Analysis Overview
Analysis uses classical economic marginal cost techniques coupled with actual market and operational data
Only install new equipment if the average total cost (including capital) is less the than marginal cost of production with the existing configuration (assumes capital is sunk for existing assets).
Actual historical data can be incorporated for:
Emissions prices
Fuel prices
Operational data on emissions rates
Questions
What equipment should be installed?
What market events would cause managers to change their decisions?
How do emissions-related capital expenditures affect project values?
How should managers allocate a limited capital budget across an entire fleet, given that future regulatory events are uncertain?
25
Single-Configuration, Single-Plant Example: NOX
$53.38
$58.38
$0
$100
$0 $1,000 $2,000 $3,000 $4,000NOx Allowance Price $/tonScenario-specific
Value
NOx Emissions Costs
Slope is based on NOx emissions rate
Annualized Capital, Fuel, and O&M Costs
26
Analysis of Multiple Possible Configurations for a Particular Plant
$63
$65
$68
$71
$73
$2,000 $2,250 $2,500 $2,750 $3,000
as isscrubcomb opt (CO)lo nox burnerofascrco + scrco + lnb/ofasbaciaci + bagdry + sbdry + sb + acicofire biomass
Optimal cost
NOx Allowance Price
Configurations
Configuration crossover points
Most expensiveconfiguration
$/MWh Spread betweenbest and worst
27
Fleet-Level Regulatory Risk Analysis
Plants with few optimal configurations are easier to plan future operations
State 1 State 2 State 3 … State n # ConfigsPlant 1 SB Dry + SB SB … SB 2Plant 2 OFA SB OFA … OFA 2Plant 3 SCR Biomass OFA … OFA 3Plant 4 SB Biomass SB … SB 1Plant 5 As Is As Is As Is … As Is 1Plant 6 SCR Biomass SCR … SCR 3Plant 7 As Is Biomass As Is … As Is 3Plant 8 As Is As Is As Is … As Is 1Plant 9 As Is As Is As Is … As Is 1Plant 10 As Is As Is As Is … As Is 1
Optimal Configuration in Each Scenario
28
Dominant Strategy Regret Table
Uncertainty over scenario probabilities can be addressed through minimizing the expected future regret given that
another scenario were chosen. The State 2 scenario carries with it the potential for substantial regret (and is therefore
deserving of considerable further analysis).
State 1 State 2 State 3 … State n DominantPlant 1 -$ (0.61)$ -$ … -$ SBPlant 2 -$ (39.09)$ -$ … -$ SCRPlant 3 -$ (0.95)$ -$ … -$ SBPlant 4 -$ (47.37)$ -$ … -$ As IsPlant 5 -$ (4.05)$ -$ … -$ OFAPlant 6 -$ -$ -$ … -$ As IsPlant 7 -$ -$ -$ … -$ As IsPlant 8 -$ -$ -$ … -$ As IsPlant 9 -$ -$ -$ … -$ As IsPlant 10 (0.29)$ (15.70)$ -$ … -$ OFAYearly Costs (199,661)$ (41,557,347)$ -$ … -$
$/MWh Cost of Using Dominant Configuration in Each Scenario
29
Summary
Most regions now have a wide variety of data available
New players in energy markets are used to using historical data in their analyses – generators will be judged increasingly by data-intensive metrics, rather than by strategic value
As markets continue to evolve, new sources of data will become available, providing investors with new ways of analyzing investment values and risks
Careful use of historical data can significantly improve decision-making accuracy
Data availability facilitates quantitative risk analysis and simulation-based valuation techniques; these are the new standards
30
Appraisal &Valuation
EngineeringAsset
Management
DecisionAnalysis
31
Core Practice Areas
Power Energy Industrial Facilities
Electric Generation Coal Chemical
Transmission Alternative Fuels Waste Processing
Distribution Natural Gas Steel
DAI Management Consultants 1370 Washington Pike Bridgeville, PA 15017
(412) 220-8920 voice(412) 220-8925 [email protected]
