Volatility in commodity prices and financial markets, compounded by technical uncertainty, make it difficult for mining professionals to assess risk exposures and identify factors influencing their Strategic Capital Management (“SCM”) decisions. Integrated Valuation and Risk Modeling (“IVRM”) methods are a toolkit comprising advanced finance theory, risk management concepts, decision analytics, and numerical methods that can be used to analyze a wide range of SCM problems. Types of applications include competing project development strategies, innovative financing structures, mergers and acquisitions, and corporate portfolio risk.
06_Joeri Van Speybroek_Dell_MeetupDora&Cybersecurity.pdf
Improving Strategic Capital Management with Integraated Valuation and Risk Modelling methods
1. Improving Strategic Capital Management
with Integrated Valuation and Risk
Modelling methods
Michael Samis, SCM Decisions
David Laughton, DL Consulting / University of Alberta
Originally presented March 3, 2017 at PDAC, Toronto, Canada
Current revision date: October 4, 2018
Photo: Jason Benz Bennee / Shutterstock
2. Page 2
► The course presenters acknowledge with gratitude the following professionals for
providing background material and comments to various sections of this course:
Jim Whyte, Ontario Securities Commission
Lawrence Devon Smith, LD Smith Associates
Charles Beaudry, International Explorers and Prospectors Inc.
Rosemary Niechcial, WayPoinT Infrastructure
Graham Davis, Colorado School of Mines
Andrew Tuck, EY
Lisa Lin, EY
John Steen, University of Queensland
Peter Monkhouse, Monkhouse Consulting
John Seddon, Control Risks
► Any errors or omissions are the responsibility of the course presenters.
Acknowledgements —
3. Page 3
Agenda
Integrated Valuation and Risk Modelling
Modelling commodity price uncertainty
Information gathering – exploration / pilot studies
Management flexibility – staged development
Alternative finance – streaming
Corporate portfolios – country risk effects
Organizational issues
Building IVRM into Enterprise Risk Management
“Of all those expensive and uncertain projects,
… there is none perhaps more perfectly ruinous
than the search after new silver and gold
mines.”
Adam Smith (1776), The Wealth of Nations, Book IV,
Chapter VII, page 610.
“As miners and explorers, we need to consider
that extreme volatility is the new normal. We
need to do things differently if we are to
effectively manage volatility.”
Paraphrasing a Canadian mining CEO (January, 2017).
Photo: photomatz / Shutterstock
4. Page 4
► The mining industry has been criticized for wasting billions of dollars during
the last boom. Much of this is attributed to over-priced acquisitions BUT
static valuation methods may bear some of the blame due to inherent biases.
► In particular, static DCF valuation methods may:
► Incorrectly estimate expected cash flows due to the “Flaw of Averages”1. The
presence of option-like financing / tax structures and management flexibility may
cause static investment models to generate biased cash flow estimates.
► Excessively discount future cash flows which may lead to frontloading capital
through building larger and more expensive projects than is prudent – especially
when staged development is possible.
► Ignore the effects of operating leverage even while adjusting for financial leverage.
The high PE attributed to royalty streams is due to low operating leverage without
adequately recognizing the possibility of cash flows being lower than expected.
► Together, these issues can cause problems when analysing mining
investment opportunities (both project and financial).
Industry background —
Static cash flow models are widely used in the mining industry
1. Savage, S (2002). The Flaw of Averages. Harvard Business Review, November.
5. Page 5
► Dynamic cash flow modelling has been considered as an alternative to static
DCF methods since at least the mid-1960’s for non-financial investment
decision-making.
► Significant advances have been made in the academic literature on
valuation-related issues over the last 25 years:
► Schwartz E.S. (Journal of Finance, 1997) on realistic uncertainty models for
commodity forecasts.
► Longstaff F.A and E.S. Schwartz (The Review of Financial Studies, 2001) on
Monte-Carlo methods for valuing American options.
► Further, recent advances in computing power allows the analysis of
moderately complex flexible project designs and financing arrangements in a
reasonable timeframe.
► Dynamic models will always be a somewhat crude representation of an actual
investment opportunity. There are still important insights to be gained by
considering an investment over a range of possible future business environments.
Industry background —
Dynamic cash flow modelling as an alternative
6. Page 6
► Mining investments are designed, developed, and operated while exposed to
high levels of geological, technical, market, political, and social uncertainty.
► Our cash flow models should explicitly recognize these risk exposures while
balancing the need to remain a practical tool.
► a practical analytical tool.
Industry background —
What should we be looking for in a cash flow modelling approach?
Cash flow model
Desirable traits for a cash flow modelling method Static Dynamic
Skills sets needed to perform analysis are widely available √ X
Compatible with current investment decision and risk management
processes √ X
Recognizes the cash flow, value, and risk distortions created by the
interaction of uncertainty and contingent finance / tax / flexibility X √
Risk exposure can be modelled, measured, and communicated with a range
of methods and formats across an organization X √
Provides a framework in which to describe forecast uncertainty and how an
investment adapts to forecast changes X √
7. Page 7
► Corporate acceptance of dynamic cash flow models has been slow even with
its benefits. The slow uptake is the result of:
Poor communication on the part of practitioners about why moving from a
static to a dynamic cash flow analysis is a good idea, and…
The costs of integrating information from a dynamic cash flow analysis
into design and investment decision processes that are built on a static
future view.
► The objective of today`s course is to advocate for the greater use of dynamic
cash flow models by:
► Recognizing the differences between static and dynamic cash flow models.
► Illustrating the range of dynamic cash flow applications in natural resource
industries.
► Making quantitative risk analysis (especially capital investment risk) an integral part
of the analysis supporting an investment decision.
► Discussing the organizational issues linked to dynamic cash flow modeling.
Industry background —
Slow acceptance of dynamic cash flow modelling by industry
8. Page 8
► Questions to ask when considering a move to dynamic cash flow analysis:
► Should we restrict our investment analysis to using a single forecast of future
conditions or expand our analysis to include how conditions may change?
► Can we realistically describe how our forecasts change over time and the
relationships between these forecasts?
► Do we have the ability to respond to different business conditions in the future by
adapting investment / operating policy or using contingent financing / taxation
structures?
► Is there a particular investment problem type that we should analyse with dynamic
cash flow models? Is there a primary investment strategy for balancing risk and
reward in an uncertain world?
► Should dynamic models be used only in situations in which it will provide a different
conclusion to static analysis or is the expanded perspective of dynamic modelling
enough to justify its use?
Industry background —
Questions to ask before moving to dynamic cash flow analysis
10. Page 10
The SCM challenge
1. Are we missing relevant insights by relying on static cash flow models?
2. Can we better understand the risk + reward trade-offs of capital management
decisions with dynamic cash flow models?
Strategic capital management (SCM) —
Managing capital in support of business objectives
Protecting the balance sheet:
How to maintain company liquidity?
Responsive operations
Improved risk monitoring
Adaptable capital structure
Raising capital:
Is capital structure aligned
with strategy?
Divestiture readiness
Innovative finance
Optimizing the corporate portfolio:
How is portfolio performance maximized?
Focused performance metrics
Capture synergies
Systematic portfolio reviews
Investing capital:
Which assets support strategy?
Acquisition readiness
Structure creatively
Leading design / analytical practice
Strategic
capital
management
(SCM)
Two questions for SCM professionals
Graphic adapted from EY Capital Agenda
11. Page 11
0.00
1.00
2.00
3.00
4.00
5.00
6.00
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
Copperprice
(real,December31,2016;US$/lb)
Time (date)
Consensus forecast at past forecast date (narrow solid lines)Historic spot price Long-term forecast at past forecast date
► SCM analysis often relies on static forecasts that are updated for changes in
business outlook – actual descriptions of forecast uncertainty are overlooked.
► Commodity price forecasts may be generated with industry marginal cost analysis,
supply-demand studies, consensus forecasts and financial market information.
► Effectively describing uncertainty in corporate forecasts requires asking:
► How do spot prices and other variables move around our forecasts?
► How are corporate forecasts revised / updated as business conditions change?
Strategic capital management —
Limitations of static cash flow models
Long-term copper forecasts from consensus forecasts
Source: Consensus Economiccs; Reuters; EY analysis
12. Page 12
► Static SCM analysis also ignores our ability to manage uncertainty through
investment / operational flexibility and contingent finance.
► Modelling our ability to manage uncertainty requires thinking about:
► Can we approach capital investment and operations such that we reduce the risks
of sunk capital and efficiently adapt operations when the outlook changes?
► Are there contingent finance possibilities that will improve capital investment
efficiency and provide resilient financing structures?
Strategic capital management —
Limitations of static cash flow models
80ktpd / 28.8mtpa capacity
0
4
20 30
Project year
Legend
Decision point:
50ktpd / 18.0mtpa capacity
140ktpd / 50.4mtpa capacity
40
Yes
No
No
Yes
No
Yes
Expand capacity to
110ktpd / 39.6mtpa in Year 7?
7
Expand capacity to
140ktpd / 50.4mtpa in Year 10?
10
Expand capacity to
80ktpd / 28.8mtpa in Year 4?
110ktpd / 39.6mtpa capacity
53 year horizon
36 year horizon
29 year horizon
25 year horizon
Financing terms adapting for outlookInvestment / operations adapting for outlook
13. Page 13
Annual updating of investment
models and forecast. Simple
discounting adjustments.
Simple scenario-based risk
analysis.
Benefit: May be acceptable
for some go / no-go decisions.
Analytical skills commonly
available. Aligns with current
corporate decision processes.
Cost: Static investment
description. Results can be
biased or misleading.
Dynamic modelling of primary
risk exposures and the ability
to manage uncertainty.
Adaptive risk adjustments.
Quantitative risk analysis.
Benefit: Detailed modelling of
investment problem. Granular
distinctions between
investments. Provides risk
exposure information.
Cost: Requires improved
numerical skills. Corporate
decision processes need to be
adapted.
Strategic capital management —
What can we do to include uncertainty in our SCM analysis?
Passive uncertainty
recognition
Active uncertainty
recognition
Focus on improving our ability
to forecast the future
investment environment.
Develop ability to time metal
price cycles.
Benefit: Little or no risk in
cash flow projection if
successful.
Cost: Unrealistic and insanely
expensive.
Attempt to reduce
forecast error
Three approaches to recognizing uncertainty
14. Page 14
► Moving from using static cash flow models to the select use of dynamic cash
flow models takes effort and time. It is important to be clear on whether there
are benefits to offset the costs.
Building a dynamic cash flow model has the
following three benefits…
Reduced cash flow estimation errors associated
with flexibility and contingent finance / taxation –
avoids the “flaw-of-averages”A risk dimension that recognizes cash
flow variability and allows us to better
measure and communicate risk. Improved representation of the investment
decision by explicitly recognizing a
greater range of investment features
… which leads to better information quality
supporting the investment decision.
Strategic capital management —
Should we move from static to dynamic cash flow models?
15. Page 15
Integrated valuation and risk modelling —
Creating a risk dimension for infrastructure SCM analysis
Risk-based
SCM modelling
Models of
uncertainty
Numerical
methods
Finance
theory
Risk
measures
Statistical
analysis
Tools for
communication
Measuring
value, return,
capital eff.
Contingent
finance
Contingent
taxation
Risk
assessment
Flexible
project
design
Contingent
corporate
strategy
Dynamic
portfolio
optimization
Balance
sheet risk
ERM
IVRM building
blocks
Project
analysis
Corporate
portfolio
► Integrated Valuation and Risk Modelling (IVRM) is a dynamic cash flow
modelling framework that creates a risk dimension for SCM analysis with:
► Finance theory ► Risk management concepts ► Numerical methods
► Decision analytics ► Statistical analysis ► Communication tools
16. Page 16
► An IVRM analysis is built on:
► A description of the SCM decision – sources of uncertainty, investment structure
and portfolio effects.
► Application of the IVRM analytical toolkit – method of cash flow estimation,
measurement of investment benefit, risk assessment and portfolio analysis.
Integrated valuation and risk modelling —
Making static cash flow models dynamic
Optimization
Single/multiple objectives
Efficient frontier
Qualitative
Scenario/sensitivity
Simulation-based
DCF / RO NPV
Return
Capital efficiency
Static
Simulation / lattice
Decision tree
Alternatives
Interdependence
Constraints
Cash flow
Owner options
Finance / tax
Type / source
Resolution
Correlation
Uncertainty
Investment
structure
Portfolio
effects
Cash flow
estimation
Value and
return
Risk
analysis
Portfolio
analysis
IVRM
analytical
framework
Description of a SCM decision IVRM analytical toolkit
17. Page 17
0
500
1000
1500
2000
2500
0 5 10 15 20
Goldprice($/oz)
Project time (year)
Year 0, 5, 10 forecast from specific price 10%/90% forecast confidence bdy
Simulated price from forecast date
0
500
1000
1500
2000
2500
0 5 10 15 20
Goldprice($/oz)
Project time (year)
10%/90% forecast confidence bdyYear 0, 5, 10 forecast from specific price
Simulated price from forecast date
► Stochastic processes are used to
describe commodity price and
long-term forecast behaviour in
financial markets.
► A stochastic process describes the
possible changes of a variable
through time – a set of uncertainty
distributions indexed by time.
► A key feature is updating future
distributions (mean / associated
variance) for recent price moves.
► Graphs on the right compare non-
updating vs updating price models.
► There is no forecast updating in
the upper graph.
► Which price path better reflects
price moves in financial markets?
Some key features of IVRM —
Commodity price uncertainty described by stochastic processes
Price movements without updating
Price movements when there is updating
18. Page 18
► Future operating and investment decisions will be made across a range of
possible future business and financial environments.
► Cash flow and value calculations need to recognize how cash flow structure
and investment / operating policy adapts to business conditions.
► Simulation and optimization (eg lattice dynamic programming) numerical methods
will likely be needed to avoid the “Flaw of Averages”.
Some key features of IVRM —
Contingent cash flow and value calculations
Contingent cash flow calculation Contingent value calculation
A mature mine with high costs is renegotiating an
existing 6% royalty. A sliding-scale royalty is
proposed that reduces royalty rates at current
price levels in exchange for higher rates when
prices are higher.
A mine design team is assessing a future open pit
/ underground development decision in a range of
commodity price environments. Management
wants to understand the prices at which the pit
should be deepened or an UG mine developed.
Royalty
10% RoyRate if AuPrice > $1,800
8% RoyRate if AuPrice: $1,600 $1,800
UOz AuPrice 6% RoyRate if AuPrice: $1,400 $1,600
4% RoyRate if AuPrice: $1,200 $1,400
2% RoyRate if AuPrice $1,200
A
Time t
PV develop underground mine ,
PV develop open pit pushback ,
Value Maximum of
PV continue mining existing pit ,
Close mine
t
t
t
t
1. Savage, S (2002). The Flaw of Averages. Harvard Business Review, November.
19. Page 19
► DCF and Real Option (“RO”) risk adjustments are used for NPV calculations.
► Mining professionals often talk about RO value as being distinct from NPV. It is not.
► RO and DCF cash flow calculations are structurally very similar. RO and DCF
risk adjustment differences can be important when calculating value.
Some key features of IVRM —
DCF and option-based risk adjustment methods
DCF present value calculation RO present value calculation
MetalPrice RDF RA Price
Metal amount
OpCost
CAPEX
Time & residual risk adj.
RA revenue
RA operating profit
RA net cash flow
Present Value of net cash flow
Price Metal amount Revenue
OpCost
CAPEX
Time & risk adj.
Operating profit
Net cash flow
Present Value of net cash flow
Abbreviations
RDFMetal: Risk discount factor for a specific metal.
RA: Risk-adjusted
20. Page 20
► Mining investment risk is often
assessed with scenario analysis.
► Scenarios are selected in a
qualitative manner.
► IVRM uses numerical methods
(e.g. simulation) to generate a
very large number of scenarios
for specific uncertainties such as
price that are consistent with
assumed behaviour (e.g. price
movements in markets).
► This information can be used to
gain insights about cash flow in
various business environments.
Some key features of IVRM —
Ability to consider a much larger number of cash flow scenarios
Cash flow database from simulation
Cash flow
calculation
dimension
Cash flow scenario analysis
High price scenario
Time 0 1 2 … T
Price …
Metal amount …
Revenue …
Op cost …
EBIT …
Tax …
CAPEX …
Net cash flow …
Discount factor …
PV net cash flow …
NPV
Base case scenario
Time 0 1 2 … T
Price …
Metal amount …
Revenue …
Op cost …
EBIT …
Tax …
CAPEX …
Net cash flow …
Discount factor …
PV net cash flow …
NPV
Low price scenario
Time 0 1 2 … T
Price …
Metal amount …
Revenue …
Op cost …
EBIT …
Tax …
CAPEX …
Net cash flow …
Discount factor …
PV net cash flow …
NPV
21. Page 21
► Dynamic cash flow modelling reduces the biases of static cash flow models
and expands our ability to measure risk and communicate risk exposure.
Some key features of IVRM —
Expanded ability to communicate investment benefits and risk
Static cash flow model IVRM with dynamic cash flow
Investment benefits summarized by… Investment benefits summarized by…
Net present value Profitability index Net present value Profitability index
IRR Payback period Modified IRR Payback period
Risk exposure assessed by… Risk exposure assessed by…
Sensitivity analysis Sensitivity analysis Event probabilities
Conditional expectations Uncertainty measures
Loss thresholds
Analysis communicated with… Analysis communicated with…
Summary statistics Spider diagrams Summary statistics Spider diagrams
Expected CF graphs Expected CF graphs Confidence bdys
Decision trees Decision boundaries
Histograms
22. Page 22
Better
understanding
of your
investment,
more informed
SCM decision
making.
► Supports understanding of key project, company, and market
factors that influence value and risk which are not visible with
static SCM analysis.
► Provides an excellent means of communicating investment
uncertainty characteristics and their impact on value and
corporate risk exposure.
► Promotes SCM conversations that you may not have had
before.
Integrated valuation and risk modelling —
Communication is the key IVRM value proposition for SCM
► IVRM helps generate and communicate SCM insights and provides support
for decision-making. It is not:
► A ploy to calculate a higher investment NPV for a favoured but challenged project.
► A substitute for extensive industry experience.
24. Page 24
► Long-range metal price forecasts and the uncertainty around those forecasts
are a key input into the analysis supporting natural resource SCM decisions.
► Forecasts influence corporate strategy, project design, financing, taxation,
community relations and government policy, among other things.
► Price forecasts are generated with a range of techniques, incorporating insights and
information from market participants and market analysts.
► Unfortunately, with static cash flow models and annual planning cycles, we
often ignore how our SCM decisions are impacted by updates to our long-
range forecasts over the planning cycle.
Modelling commodity price uncertainty —
The importance of long-range forecasts
25. Page 25
► The natural resource industries often recognize long-range forecast price
uncertainty with scenario analysis (price decks).
► Long-range forecast scenarios are sometimes probability weighted to include the
effects of price uncertainty in decision making and valuation. This approach to
uncertainty modelling ignores long-term forecast updating.
Modelling commodity price uncertainty —
Scenario analysis and long-range forecasts
Price deck
Scenario Au price
Blue sky $1,500
Higher $1,400
High $1,300
Forecast $1,200
Low $1,100
Lower $1,000
Lights out $ 900
Price deck
Scenario Au price Probability
Blue sky $1,500 5%
Higher $1,400 10%
High $1,300 20%
Forecast $1,200 30%
Low $1,100 20%
Lower $1,000 10%
Lights out $ 900 5%
Expected price $1,200
The uncertainty
around the forecast
may be taken into
account by assigning
probability weights to
each scenario
26. Page 26
Modelling commodity price uncertainty —
Three components of an uncertainty model
Price
forecast
Forecast
updating
Three features of a complete
model of price uncertainty
Price
variability
Price variability
describing uncertainty
around a forecast
► The model we use
generates a price
distribution at each
future time point.
Forecast updating
allowing for dynamic
expectations
► Future expectations
change as future prices
change.
Forecasts generated by:
► Supply / demand
projections.
► Cost curve models.
► Consensus forecasts.
► Financial market
information.
► However, price decks and their associated probability-weights are an
incomplete model of price uncertainty – we still need to recognize forecast
updating over time.
27. Page 27
Modelling commodity price uncertainty —
Model selection and parameter estimation
Hypothesis testing assesses whether an
assumed model of a commodity price
movements is consistent with historical
price behaviour.
Model selection Parameter selection
The econometric modelling techniques to
estimate inputs for price uncertainty
models from historical market information.
.
Mathematical
description of
commodity price
behavior
Statistical tests performed to identify:
► Behaviour characteristics.
► Whether model choice is reasonable.
Form Parameters
Price model types and inputs:
► A wide range of inputs from historic spot
prices, forward curves, and consensus
forecasts.
► Models can display a range of behavior
where spot prices and forecasts move in
sync, spot prices move around forecasts,
and the possibility of jumps.
28. Page 28
► Non-reverting models are used to describe the price movements of financial
stocks, precious metals, FX and possibly a few base and minor metals.
► Long-term forecasts move in lockstep with spot price movements. A 2% rise in the
spot price results in a 2% increase in the long-term forecast price.
► Uncertainty increases with term (time from today).
► Limitation: Applies only to financial stocks, precious metals and FX rates.
0
500
1,000
1,500
2,000
2,500
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
Goldprice
(real;December31,2016;US$/loz)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
Types of commodity price uncertainty models —
Single factor non-reverting models
29. Page 29
0
50
100
150
200
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
WTIOilprice
(real;December31,2016;US$/bbl)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
► …
Types of commodity price uncertainty models —
Single factor reverting models
► Reverting models describe base metal and energy price movements.
► A constant real or nominal long-term forecast. Spot price varies around and reverts
to the long-term forecast price.
► Uncertainty saturates with term, reducing long-life project cash flow discounting.
► Need to update the long-term forecast for market regime changes.
► Limitation: A single long-term forecast that does not change over time.
30. Page 30
0
50
100
150
200
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
WTIOilprice
(real;December31,2016;US$/bbl)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
► …
Types of commodity price uncertainty models —
Two-factor models
► Two-factor models better reflect base metal and energy price movements.
► Both spot price and long-term forecast price are uncertain.
► Uncertainty increases with term. Variability in the long-term forecast can generate
option value for long-life base metal and energy projects.
► Limitation: Parameterization using historical prices results in uncertainty
levels (indicated by confidence intervals) that are unreasonably high.
31. Page 31
0
50
100
150
200
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
WTIOilprice
(real,December31,2016;US$/bbl)
Time (date)
Forward curve forecast at past forecast date (narrow solid lines)Historic spot price Long-term forecast at past forecast date
► Outlook for long-term prices can also change dramatically over short periods.
► The increase in oil demand from China in 2003 had an impact on prices that was
sudden, dramatic, and unexpected.
► The decline in oil prices as a result of increased Saudi production was sudden,
dramatic, and unexpected.
► These sudden price forecast changes may be the result of price jumps or
periods of high volatility. They are random and can happen at any time.
Types of commodity price uncertainty models —
Jumps / high vol periods creating sudden market outlook changes
Upward move from
Increased demand
Downward move from
Increased production
WTI oil spot price and quarterly forward-implied forecast from January 1, 2000
32. Page 32
0
50
100
150
200
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
WTIOilprice
(real;December31,2016;US$/bbl)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
► …
Types of commodity price uncertainty models —
Two-factor model with jumps or high volatility periods
► Two-factor reverting models extended to include a jump factor or high
volatility periods for unexpected large changes in long-term forecast.
► Jumps or high volatility periods absorb some of the long-term forecast volatility.
► Simulated price behavior may be closer to what we see in markets.
► Limitation: Increased complexity and simulation time.
33. Page 33
0
500
1,000
1,500
2,000
2,500
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20 1-Jan-25 1-Jan-30
Goldprice
(real,December31,2016;US$/oz)
Time (date)
Analyst forecast (narrow solid lines)Historic spot price
Consensus forecast Forward curve forecast
0
500
1,000
1,500
2,000
2,500
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20 1-Jan-25 1-Jan-30
Goldprice
(real,December31,2016;US$/oz)
Time (date)
Analyst forecast (narrow solid lines)Historic spot price
Consensus forecast Forward curve forecast
► Gold is primarily held as an
investment asset with few industrial
uses.
► Range of business outlooks at both
dates. Analyst price forecasts much
more divergent 5 years ago after a
large run up in prices.
► Forward market long-term forecast
had a greater change over 5 years
than consensus long-term forecast.
Gold price uncertainty model —
Analyst / consensus / forward long-term price forecasts
Analyst forecasts – December 31, 2011
Analyst forecasts – December 31, 2016
Source: Consensus Economics; Reuters; M Samis analysis
Source: Consensus Economics; Reuters; M Samis analysis
Forecast
date
Long-term forecast price ($/oz; 30/12/16)
Analyst Calculated / market
Low High Consensus Forward
31-Dec-11 837 2,117 1,230 1,670
31-Dec-16 915 1,576 1,222 1,143
34. Page 34
0
500
1,000
1,500
2,000
2,500
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
Goldprice
(real,December31,2016;US$/oz)
Time (date)
Consensus curve forecast at past forecast date (narrow solid lines)
Historic spot price Long-term forecast at past forecast date
0
500
1,000
1,500
2,000
2,500
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
Goldprice
(real,December31,2016;US$/oz)
Time (date)
Forward curve forecast at past forecast date (narrow solid lines)
Historic spot price Long-term forecast at past forecast date
Gold price uncertainty model —
Price behavior and forecast updating
Consensus forecast
Forward-implied forecast
Source: Consensus Economics; Reuters; M Samis analysis
Source: Reuters; M Samis analysis
► Consensus forecasts display
anchoring – forecast updates are
less reactive to spot market
movements than forward-implied
forecast.
► Forward-implied forecasts respond
quickly to market movements as
long-term forecasts move upwards
and downwards in a parallel
fashion.
► Analyst forecasts rely on information
from non-market participants and
may have limitations compared with
the actual financial trades
embedded in forward contracts.
35. Page 35
0
500
1,000
1,500
2,000
2,500
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
Goldprice
(real;December31,2016;US$/loz)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
► Gold prices are modelled as a non-reverting process around an updating
long-term forecast. Consistent with gold being a store of perceived value.
► Volatility is estimated using either historical prices or implied from option prices.
► No statistical evidence of reversion in spot prices and forward prices.
► The stochastic model here assumes a flat forecast in real dollars at each date. The
model can have upward or downward trending forecasts at each date.
Gold price uncertainty model —
Simulated prices with one factor NREV uncertainty model
36. Page 36
0
10
20
30
40
50
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20 1-Jan-25 1-Jan-30
Silverprice
(real,December31,2016;US$/oz)
Time (date)
Analyst forecast (narrow solid lines)Historic spot price
Consensus forecast Forward curve forecast
0
10
20
30
40
50
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20 1-Jan-25 1-Jan-30
Silverprice
(real,December31,2016;US$/oz)
Time (date)
Analyst forecast (narrow solid lines)Historic spot price
Consensus forecast Forward curve forecast
► Silver has mainly industrial uses with
some investment interest. It is mainly
produced as a by-product and as a
result it is less responsive to supply /
demand signals.
► A range of analyst long-term price
forecasts at both dates suggesting
divergent business outlooks.
► Analysts were more in agreement in
2016 than in 2011 (less spread).
Silver price uncertainty model —
Analyst / consensus / forward long-term price forecasts
Sources: Reuters and Consensus Economics
Analyst forecasts – December 31, 2011
Analyst forecasts – December 31, 2016
Source: Consensus Economics; Reuters; M Samis analysis
Source: Consensus Economics; Reuters; M Samis analysis
Forecast
date
Long-term forecast price ($/oz; 30/12/16)
Analyst Calculated / market
Low High Consensus Forward
31-Dec-11 16.50 31.52 23.75 27.02
31-Dec-16 13.30 23.71 18.60 14.07
37. Page 37
0
10
20
30
40
50
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
Silverprice
(real,December31,2016;US$/oz)
Time (date)
Consensus curve forecast at past forecast date (narrow solid lines)
Historic spot price Long-term forecast at past forecast date
0
10
20
30
40
50
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
Silverprice
(real,December31,2016;US$/oz)
Time (date)
Forward curve forecast at past forecast date (narrow solid lines)
Historic spot price Long-term forecast at past forecast date
► << uses >>
► << consensus comment >>.
► << forward forecast comment >>
► …
Silver price uncertainty model —
Price behavior and forecast updating
Consensus forecast
Forward-implied forecast
Source: Consensus Economics; Reuters; M Samis analysis
Source: Reuters; M Samis analysis
► Consensus forecasts again display
anchoring – forecast updates are
less reactive to spot market
movements than forward-implied
forecast.
► Forward-implied forecasts move
with spot prices with some
backwardation in high price
environments.
► Forward markets may be revealing
either non-reverting prices or weak
long-term reversion to a price of
approximately $20/oz.
38. Page 38
0
10
20
30
40
50
60
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
Silverprice
(real;December31,2016;US$/loz)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
► Silver prices are modelled as a non-reverting process around an updating
long-term forecast. This is reflective of by-product production as some
production is less responsive to price signals.
► Volatility is estimated from historic price data.
► The stochastic model here assumes a flat forecast in real dollars at each date.
► Past econometric analysis could support weak reversion.
Silver price uncertainty model —
Simulated prices with one factor NREV uncertainty model
39. Page 39
0.00
1.00
2.00
3.00
4.00
5.00
6.00
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20 1-Jan-25 1-Jan-30
Copperprice
(real,December31,2016;US$/lb)
Time (date)
Analyst forecast (narrow solid lines)Historic spot price
Consensus forecast Forward curve forecast
0.00
1.00
2.00
3.00
4.00
5.00
6.00
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20 1-Jan-25 1-Jan-30
Copperprice
(real,December31,2016;US$/lb)
Time (date)
Analyst forecast (narrow solid lines)Historic spot price
Consensus forecast Forward curve forecast
► Copper spot price trend influenced
by supply and demand adjustments
over time. These adjustments create
a long-term price within a narrow
band.
► Long-term analyst forecasts appear
to show reversion while forward-
implied forecasts are responsive to
current price levels.
Copper price uncertainty model —
Analyst / consensus / forward long-term price forecasts
Analyst forecasts – December 31, 2011
Analyst forecasts – December 31, 2016
Source: Consensus Economics; Reuters; M Samis analysis
Source: Consensus Economics; Reuters; M Samis analysis
Forecast
date
Long-term forecast price ($/lb)
Analyst Calculated / market
Low High Consensus Forward
31-Dec-11 1.91 3.20 2.62 3.23
31-Dec-16 1.93 3.01 2.56 2.30
40. Page 40
0.00
1.00
2.00
3.00
4.00
5.00
6.00
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
Copperprice
(real,December31,2016;US$/lb)
Time (date)
Forward curve forecast at past forecast date (narrow solid lines)
Historic spot price Long-term forecast at past forecast date
0.00
1.00
2.00
3.00
4.00
5.00
6.00
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
Copperprice
(real,December31,2016;US$/lb)
Time (date)
Consensus curve forecast at past forecast date (narrow solid lines)
Historic spot price Long-term forecast at past forecast date
► Reversion exhibited by both
consensus and forward-implied
forecasts.
► Note the regime change (jump / high
volatility period) in 2005, where the
long-term forecast changed in both
consensus and forward-implied
forecasts.
► As of Jan 2016, analysts were more
optimistic than forward-implied
forecasts.
► Difference may reflect copper
market risk premium embedded in
analyst forecasts.
Copper price uncertainty model —
Price behavior and forecast updating
Consensus forecast
Forward-implied forecast
Source: Consensus Economics; Reuters; M Samis analysis
Source: Reuters; M Samis analysis
41. Page 41
0.00
1.00
2.00
3.00
4.00
5.00
6.00
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
Copperprice
(real;December31,2016;US$/lb)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
Copper price uncertainty model —
Simulated price scenario with two factor + jump uncertainty model
► Copper prices modelled with a two factor + jump process to describe forecast
uncertainty and forecast shocks.
► Short-term price volatility and long-term forecast volatility is estimated from historic
spot price and forward price information.
► Model jumps interpreted to reflect demand shocks such as increased demand from
developing countries (2005) and GFC (2008).
42. Page 42
0
50
100
150
200
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20 1-Jan-25 1-Jan-30
WTIOilprice
(real,December31,2016;US$/oz)
Time (date)
Analyst forecast (narrow solid lines)Historic spot price
Consensus forecast Forward curve forecast
0
50
100
150
200
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20 1-Jan-25 1-Jan-30
WTIOilprice
(real,December31,2016;US$/bbl)
Time (date)
Analyst forecast (narrow solid lines)Historic spot price
Consensus forecast Forward curve forecast
► Oil / diesel is often an important cost
component of mining operations
which creates uncertainty about
future operating costs levels.
► Analysts forecasts are less scattered
than in 2011.
► Energy costs have fallen along with
metal prices.
► Costs and revenues tend to move in
tandem.
WTI oil price uncertainty model —
Analyst / consensus / forward long-term price forecasts
Analyst forecasts – December 31, 2011
Analyst forecasts – December 31, 2016
Source: Consensus Economics; Reuters; M Samis analysis
Source: Consensus Economics; Reuters; M Samis analysis
Forecast
date
Long-term forecast price ($/lb; 31/12/16)
Analyst Calculated / market
Low High Consensus Forward
31-Dec-11 81.74 127.05 104.08 88.81
31-Dec-16 46.52 68.41 58.43 51.11
43. Page 43
0
50
100
150
200
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
WTIOilprice
(real,December31,2016;US$/bbl)
Time (date)
Forward curve forecast at past forecast date (narrow solid lines)
Historic spot price Long-term forecast at past forecast date
0
50
100
150
200
1-Jan-00 31-Dec-04 31-Dec-09 1-Jan-15 1-Jan-20
WTIOilprice
(real,December31,2016;US$/bbl)
Time (date)
Consensus curve forecast at past forecast date (narrow solid lines)
Historic spot price Long-term forecast at past forecast date
► Long-term forecasts affected by oil
price rise in 2008. Even after the
2008 financial crisis, long-term
forecasts reverted to a higher oil
price level.
► Reversion exhibited by both
consensus and forward-implied
forecasts after 2008.
► Consensus long-term forecasts and
forward-implied forecasts are in
broad agreement.
WTI oil price uncertainty model —
Price behavior and forecast updating
Consensus forecast
Forward-implied forecast
Source: Consensus Economics; Reuters; M Samis analysis
Source: Reuters; M Samis analysis
44. Page 44
0
50
100
150
200
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
WTIOilprice
(real;December31,2016;US$/bbl)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
WTI oil price uncertainty model —
Simulated price scenario with two factor + jump uncertainty model
► WTI oil prices modelled with a two factor + jump process to describe forecast
uncertainty and forecast shocks.
► Short-term price volatility and long-term forecast volatility is estimated from historic
spot price and forward price information.
► Model jumps interpreted to reflect supply and demand shocks such as shale oil
technology (2007), Saudi production ramp up (2014) and OPEC supply cuts (2017).
45. Page 45
-10%
0%
10%
20%
30%
40%
50%
0 1 2 3 4 5
Meanpercentageerror
Forecast term (years)
Spot forecast Forward-implied forecast Consensus forecast
-10%
0%
10%
20%
30%
40%
50%
0 1 2 3 4 5
Meanpercentageerror
Forecast term (years)
Spot forecast Forward-implied forecast Consensus forecast
-10%
0%
10%
20%
30%
40%
50%
0 1 2 3 4 5
Meanpercentageerror
Forecast term (years)
Spot forecast Forward-implied forecast Consensus forecast
-10%
0%
10%
20%
30%
40%
50%
0 1 2 3 4 5
Meanpercentageerror
Forecast term (years)
Spot forecast Forward-implied forecast Consensus forecast
► The mining industry is often skeptical of using forward curves to infer price
forecasts. Mean Percentile Error (MPE) of quarterly “naïve” spot price,
forward-implied and consensus forecasts from January 1, 2000 suggests:
► Consensus tends to have largest long-term forecast MPE for each commodity.
► Gold, silver, WTI oil have lowest MPE with spot and forward-implied forecasts.
Comparing forecasting methods —
Which did better - consensus or forward-implied forecasts?
Gold forecast MPE Copper forecast MPE
Silver forecast MPE WTI oil forecast MPE
46. Page 46
► Single factor and more complex stochastic uncertainty models can be used to
describe price forecast uncertainty when reasonably parameterized. Relying
on scenario decks to describe future price uncertainty:
Over simplifies how price forecasts may change in the future, which leads to…
An inadequate description of future price behaviour, which generates…
Lower quality value and risk information on which to base a SCM decision.
► These descriptions of commodity price behaviour can be incorporated into
cash flow models to better understand how our investments will perform and
can be best managed across a range of business environments.
Modelling commodity price uncertainty —
Some final thoughts…
47. Page 47
Value of information – exploration / pilot studies
Photo: Adwo / Shutterstock
48. Page 48
► Value-of-Information problems assess the benefit of learning about a project
prior to making a large investment decision.
► The mining project life-cycle is a sequence of Value-Of-Information problems from
early exploration through to mine closure.
► However, industry understanding of how Value-Of-Information projects create
value is sporadic.
Value-of-Information in the mining industry
A critical component of the industry value creation process
Mining
project
cycle
Step
5
Step
3
Step
6
Step
2
Step
4
Step
1
Operating mine during which additional
investments are made to improve efficiency,
reduce costs, and develop brownfields.
Development stage during
which large amounts of capital
are invested to build the mine
Closure at which additional costs may
be incurred to ensure site is shut down
with little or no future impact
Early exploration in which a series of
small investments are used to identify
promising targets.
Additional capital is required during later-stage
exploration in order to learn about the reserve /
resource potential of a deposit.
Design stage during which different mine
and mill designs are considered against a
range of economic, technical and social
measures.
49. Page 49
► Exploration is a classic value-of-information activity in the mining industry.
► Targets are advanced from the identification stage to a reserve statement
supporting mine development by gathering information through a series of
escalating investments.
► Targets are either advanced or rejected / sidelined using exploration information to
rank targets / resources on a combination of geological and economic measures.
Value-of-Information in the mining industry
Managing geological uncertainty of an exploration pipeline
Target identification
Geological
anomalies
Exploration
targets
Project feasibility
Resource / reserve definition
Operating mine /
Project development
Follow-up targets
Advanced targets
Inferred resource
Measured / indicted
resources
Reserves
Reject
Reject
Reject
50. Page 50
► Mine and mill designs for a project at the pre-development stage may have
performance uncertainty linked to dilution / grades / capacity / recovery which
require testing to gain information about the commercial scale performance.
► Pilot plants and other tests appear expensive when their value creation role is
not understood.
► How much would you pay to know your mill operates on the blue curve and so
avoid investing capital in a mine and mill that operates on the purple curve?
Value-of-Information in the mining industry
Managing performance uncertainty of a new mill / mine design
0%
20%
40%
60%
80%
100%
120%
0 12 24 36
% of Design Capacity
Months After Start‐Up
McNulty Ramp‐Up Curves
Type 1
Type 2
Type 3
Type 4
Plant ramp-up curves:
Type 1: Mature technology; extensive pilot testing for risky
unit operations.
Type 2: Some prototype technology; severe conditions;
incomplete pilot testing.
Type 3: Same as 2 but with very limited pilot testing;
incomplete knowledge of feed characteristics.
Type 4: Same as 3 but complex flowsheet and
misunderstood chemistry; pilot testing for product
quality but not process parameters.
51. Page 51
► Innovation is currently a much
discussed industry topic focusing on:
► Automation ►Machine learning
► Energy storage ►Digital information
► Insights from the high-tech start-up
industry (eg Lean Startups) are
complementing existing processes to
improve investment efficiency.
► Test critical assumptions and disprove
/ adapt an idea before spending large
amounts of capital.
► However, despite the tub-thumping,
the benefits from innovation are
(highly) uncertain so that some form
of economic analysis is required that
is reflective of the problem.
Value-of-Information in the mining industry
Managing the uncertainty of an innovation program
Build − Assess − Learn feedback loop
Adapted from:
Ries, E. (2011). The Lean Startup. Crown Publishing. New York. 320p.
Build
Lear
n
Assess
Ideas
Proto-
type
Data
52. Page 52
Example: Valuing a new mill technology
Background
► A global mining company is considering
whether to invest in a new mill
technology across its operations to
improve recovery and reduce costs.
► The new technology requires investment
of US$586 million over three years:
Year 1: US$ 19 million
Year 2 & 3: US$567 million
► Once installed, the present value of cash
flows over ten years from additional
metal recovery and reduced costs is
expected to be US$1.2 billion.
► However, there is no guarantee that the
new technology will actually work (ie
there is uncertainty).
53. Page 53
► Capital of US$586 million is invested over three years with two outcomes:
1. Technology works and has a present value of benefit of US$1.2 billion with a
probability of X%.
2. Technology does not work − there is ZERO benefit with a probability of 1 − X%.
► Net Present Value (“NPV”) is calculated by probability weighting the two
outcomes.
0
200
400
600
800
0%20%40%60%80%100%
ProjectNPV(US$million)
Probability of technology success (%)
All-in refernce project
Example: Valuing a new mill technology
“All-in” reference investment case
0 1 2 3
Spend US$586 million to
roll out new technology
across the company
Technology success
Career Making Value Creation
(“CMVC”) with value benefit of
US$1.2 billion.
Outcome NPV = US$686m.
Technology fail
Career Ending Value Loss
(“CEVL”) with a ZERO value
benefit.
Outcome NPV = −US$483m
NPV of All-in investment case across
a range of success probabilities
Decision tree for All-in investment case
Tech project does
not proceed when
the probability of
success is less
than 41%.
54. Page 54
Example: Valuing a new mill technology
Alternative investment case – test the new technology (pilot)
► Testing a technology with a pilot
program can create additional
value if it allows you to avoid
investing in a flawed technology.
► Particularly if it signals technology
failure prior to making a large
capital investment.
► Information quality is an important
determinant of whether a pilot
program will create value.
► False Positive information signals
success when the technology is a
failure – wastes capital.
► False Negative information signals
failure when the technology is a
success – missed value creation.
Modified decision tree for the pilot investment case
3
Successful
pilot
test
0 1 2
Spend US$567m
to continue
installing new
technology?
Yes; spend
$567m on new
technology
Tech success
CMVC with benefit
of US$1.2 billion.
NPV = US$686m −
PV of pilot cost
Tech fail
CEVL with a ZERO
value benefit.
NPV = −US$483m
− PV of pilot cost
Pilot fail
Cut Losses (“CL”) by cancelling
technology program. Avoid
spending capital of US$567m.
NPV = −US$17m − PV of pilot
cost
Failed
pilot test
Interpret
pilot test
results
Spend
US$19m
on new
tech and
US$Xm
on pilot
program
This outcome is
never achieved.
55. Page 55
0
200
400
600
800
0%10%20%30%40%50%60%70%80%90%100%
ProjectNPV(US$million)
Probability of technology success (%)
All-in project Pilot cost = $0m Pilot cost = $50m Pilot cost = $100m Pilot cost = $150m
Example: Valuing a new mill technology
Value benefit of “100% reliable” pilot project information
► GREEN graph area represents the value of a pilot program generating “100%
reliable” information for a range of technology success probabilities and
program costs.
► 100% reliable information refers to information from a pilot program that never
generates FALSE NEGATIVES or FALSE POSITIVES. This is an unrealistic
assumption and is used to create a reference case.
► Program costs reduces both value and the success probability range over which a
pilot program creates value.
Maximum pilot program value is
US$273m when the probability of
success is 41%.
High success probability
case Maximum pilot program
value is US$140m when the
probability of success is 70%.
Low success probability case
Maximum pilot program value is
US$194m when the probability
of success is 30%.
56. Page 56
0%
20%
40%
60%
80%
100%
0%20%40%60%80%100%
Outcomeprobability(%)
Information reliability given tech fail (%)
Pilot+Tech success prob Pilot success+Tech fail prob Pilot fail prob
0%
20%
40%
60%
80%
100%
0%20%40%60%80%100%
Outcomeprobability(%)
Information reliability given tech success (%)
Pilot+Tech success prob Pilot success+Tech fail prob Pilot fail prob
Example: Valuing a new mill technology
Why unreliable information impacts pilot project value
► The possibility of unreliable information shifts probability between the
different outcomes of the pilot program decision tree.
► FALSE NEGATIVE information given that the technology is a success reduces the
probability of a CMVC outcome while increasing the probability of CL outcome (ie
increasing probability of missing a successful tech innovation).
► FALSE POSITIVE information given that the technology is a failure reduces the
probability of a CL outcome while increasing the probability of CEVL outcome (ie
increasing probability of sinking capital into a dud technology).
FALSE NEGATIVE
30% chance of tech success / zero pilot cost
FALSE POSITIVE
30% chance of tech success / zero pilot cost
57. Page 57
0
50
100
150
200
0%20%40%60%80%100%
Valueofpilotprogram
(US$million)
Information reliability given tech success (%)
Pilot cost = $100m Pilot cost = $50m Pilot cost = $0m
0
50
100
150
200
0%20%40%60%80%100%
Valueofpilotprogram
(US$million)
Information reliability given tech success (%)
Cost = $150m Cost = $100m Cost = $50m Cost = $0m
Example: Valuing a new mill technology
Value impact of FALSE NEGATIVE information
► FALSE NEGATIVE information shifts the underlying probability of a CMVC
outcome (tech success) to an outcome in which the pilot program wrongly
flags a flawed technology (CL).
► Reducing FALSE NEGATIVE information is important when the new technology is
likely a success as missing out on value creation has a cost. A 1% improvement in
information quality increases pilot value by US$4.9m.
► It is less important to reduce FALSE NEGATIVE information when new technology
is a long-shot as technology success is unlikely. Improving information quality by
1% increases value by US$1.4m.
70% probability of tech success 30% probability of tech success
58. Page 58
► FALSE POSITIVE information shifts the underlying probability of the CL
outcome (pilot program flagging a dud technology) to an outcome where
capital is wrongly invested in a flawed technology (CEVL).
► When the new technology is likely successful, reducing FALSE POSITIVE
information is still desirable as avoiding wasted capital is a good outcome.
Improving information quality by 1% increases pilot value by US$2.1m.
► Reducing FALSE POSTIVE information is more important when the new
technology is a long-shot as investing in dud technology destroys value. A 1%
improvement in information quality increases pilot value by US$3.3m.
0
50
100
150
200
0%20%40%60%80%100%
Valueofpilotprogram
(US$million)
Information reliability given tech fail (%)
Cost = $150m Cost = $100m Cost = $50m Cost = $0m
0
50
100
150
200
0%20%40%60%80%100%
Valueofpilotprogram
(US$million)
Information reliability given tech fail (%)
Pilot cost = $100m Pilot cost = $50m Pilot cost = $0m
Example: Valuing a new mill technology
Value impact of FALSE POSTIVE information
70% probability of tech success 30% probability of tech success
59. Page 59
► Pilot programs for new technology also have an important capital risk
management benefit as they provide an opportunity to avoid investing large
capital amounts in a technology failure.
► Projects with high success probabilities benefit on a relative basis almost as much
as a project with a low probability of success.
► Pilot programs reduce capital risk exposure by approximately 50% or better when
information quality is reasonable in both instances.
0
100
200
300
400
0%20%40%60%80%100%
Expectedcapitalloss
(US$million)
Information reliability given tech fail (%)
All-in case Cost=$150m Cost=$100m Cost=$50m Cost=$0m
0
100
200
300
400
0%20%40%60%80%100%
Expectedcapitalloss
(US$million)
Information reliability given tech fail (%)
All-in case Cost=$100m Cost=$50m Cost=$0m
Example: Valuing a new mill technology
Capital risk management benefits of a pilot program
70% probability of tech success 30% probability of tech success
60. Page 60
Some concluding thoughts for this example…
► Value-of-Information projects create value through learning about a project
prior to making a large investment decision - high-risk projects such as
exploration programs are not viable otherwise.
► Learning creates value by providing an option to avoid making a large investment in
a new but flawed technology. There are important capital-risk management
benefits as well.
► It is important to assess the interplay between the cost of gathering information, the
reliability of the information generated, and the scheduling of capital investments.
► The mining industry as a rule has processes in place to manage these projects.
However, we are not particularly adept at valuing these projects or understanding
how they create value.
► The primary benefit of being able to value a Value-Of-Information project is in
the understanding of how these projects create value.
► This understanding allows us to improve the efficiency of investment schedules and
the learning objectives of these projects.
► The actual value itself is less important given the subjectivity of the inputs.
61. Page 61
► In 2005, a petroleum company (“O&GCo”) is developing a new off-shore
natural gas field that is expected to emit 1 million tonnes of CO2 per year from
production start in 2009 to expected closure in 2055.
► Global warming will likely result in CO2 emissions incurring a carbon tax (possibly
significant) at some time during the field operation.
► CO2 emissions at the field can be mitigated by building a Carbon Capture and
Sequestration (“CCS”) facility to inject carbon in a nearby depleted reservoir.
► Senior project management has asked for a detailed analysis of the policy
and investment implications of building and operating a CCS facility so they
can make recommendations to the Board’s Investment Committee.
A CCS facility is being considered to prevent
CO2 emissions from an undeveloped NatGas
field. Is the CCS facility’s ability to manage
gas and carbon price uncertainty valuable?
Assess value and build / operate policy by
considering market and technical risk exposures
combined with ability to decide when and how to
build and operate the CCS plant.
Example: CCS and market/technical uncertainty
Background
Issue: Solution:
62. Page 62
► The CCS facility requires a year to build at a cost of $120m.
► Once built, the facility costs $1m to operate and consumes an expected 0.7
Gcf of NatGas field production to compress/transport/inject carbon into the
storage reservoir.
► The CCS facility can be temporarily closed at an annual $0.5m maintenance cost.
► Two competing CCS facility designs:
► Basic design: Capital and operating costs as stated above.
► Enhanced design: Increase capital costs by $5m and reduce NatGas consumption
by 10% to expected 0.63 Gcf.
► The CCS facility may create value for O&GCo by avoiding the cost of emitting
CO2 into the atmosphere. The cash flow calculation when operating is:
Example: CCS and geological uncertainty
Design, development and operation of the CCS facility
2Net cash flow CO emitted x
NatGas consumed x
Fixed operating cost
not 2CO Price
NatGas price
63. Page 63
► There is uncertainty about the storage reservoir porosity that impacts the
amount of NatGas consumed to inject CO2.
► Prior knowledge about the depleted reservoir and similar geological
formations elsewhere suggest the following annual NatGas consumption
outcomes once the CCS facility is running:
High consumption: 25% probability that annual consumption is 0.875 Gcf.
Base consumption: 50% probability that consumption is 0.7 Gcf.
Low consumption: 25% probability of 0.525 Gcf of NatGas consumed.
► Management can wait to resolve this uncertainty or they resolve this
uncertainty now before building the CCS facility with a 1-Year reservoir study
costing $5m.
► Performing the reservoir study also has a design benefit in that improving the
understanding of reservoir porosity reduces NatGas consumption by 10%
regardless of consumption outcome.
Example: CCS and geological uncertainty
Geological uncertainty linked to storage reservoir porosity
64. Page 64
0.00
2.50
5.00
7.50
10.00
12.50
1/01/90 1/01/00 1/01/10 1/02/20
NatGasprice
(real;September30,2005;US$/mmcf)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
► From the perspective of 2005, NatGas prices are expected to revert to a
$7.50/mmcf expected long-term forecast tied to LNG supply price.
► Single factor reverting model with a volatility of 20% and uncertain restoring
force with a 3 year time scale.
► A two factor model may be a better representation of market behaviour. A one-
factor model provides simplicity and we compensate by using a higher volatility.
Example: CCS and geological uncertainty
NatGas price uncertainty
65. Page 65
► A one factor model reflecting rising carbon price to a median of $30/tonne in
2030 before flattening from backstop technology.
► Parameters determined based on expert panel analysis of interplay of physical
system uncertainty, technology, economics and politics
► Longer reversion time scale (4 years), greater short-term uncertainty
► Maximum short-term uncertainty in 2025 at 32.5%
Example: CCS and geological uncertainty
Carbon pricing uncertainty
66. Page 66
Example: CCS and geological uncertainty
Risk adjustments
► Risk discounting accomplished by using distributions centred around forward
prices (red lines in graph below for gas and previous page for CO2).
► Risk discounting in gas forward prices 3% per year for each 10% annual
uncertainty, 4% for CO2
67. Page 67
► Carbon and NatGas price movements are not perfectly correlated.
► Initially, carbon and Nat gas price correlated as increasing carbon prices drive
greater fuel switching from coal / oil to NatGas.
► After 2045, negative correlation as tighter constraints and higher carbon prices lead
to new technology that is cleaner than NatGas. Increases in carbon prices lead to
reduced demand for NatGas.
Example: CCS and geological uncertainty
Relationship between carbon and natural gas prices
Correlationbetweencarbon
andNatGasprices
68. Page 68
► Construction of the NatGas field platform contains an initial development
decision point for the CCS facility.
► Project managers have an option to spend $10m to extend the platform
during construction to provide space for the CCS facility if it is built later.
► It will cost $30m to retrofit the platform for the CCS facility if an extension is not
initially built for CCS.
Example: CCS and geological uncertainty
Initial NatGas field development alternatives
CCS facility / reservoir
appraisal decision
space
Pay $10m now to extend platform for CCS
facility in order to avoid a $30m retrofit if
CCS built later?
Yes
CCS facility / reservoir
appraisal decision
space
No
2005 2006 2007 2008 2009 2010
to 2055
Construct platform with CCS extension
Construct platform without CCS extension
69. Page 69
► From 2010, project managers have 4 options for building the CCS facility.
1) Don’t build the CCS facility; release CO2 into the atmosphere and pay carbon tax.
2) Build a Basic CCS facility for $120m ($90m with pre-built platform extension).
3) Build an Enhanced CCS facility to reduce NatGas consumption by 10% for $125m
($95m with pre-built platform extension).
4) Perform 1-Year reservoir study for $5m to resolve porosity uncertainty. Then make
the CCS facility development decision (Don’t build / Basic / Enhanced).
Example: CCS and geological uncertainty
CCS plant development decisions during field operation
CCS facility / reservoir
decision space
CCS facility
decision
space
Run reservoir study for
$5m and decrease
NatGas consumption by 10%
CCS facility /
reservoir decision
space
Time t-1 year Time t Time t+1 year
to 2055
Build enhanced CCS for $125m ($95m)
Build basic CCS
for $120m ($90m)
Wait
Wait
Build basic CCS
for $120m ($90m)
Time t+2 year
Build basic CCS
for $125m ($95m)
Don’t build CCS
70. Page 70
► The DCF and RO NPVs for the CCS facility and initial development policy:
► Both DCF and RO NPV calculations show that the CCS facility adds value.
► However, the two NPV calculations have different initial investment policies.
► The DCF calculation recommends not extending the NatGas field platform for CCS
(ie retrofit later at a higher cost) while the RO calculation recommends that the
option to extend the platform for the CCS facility should be exercised.
► RO recommendation is driven by a larger value of future capital cost savings:
1) occurs earlier,
2) discounted at a lower rate.
Example: CCS and geological uncertainty
Initial development recommendation – DCF vs Real Options
Risk adjustment approach
CCS facility NPV
($ million) Investment policy
DCF (10% discount rate) 10.0 Don’t extend platform for CCS
RO (forward curve + RF rate) 13.6 Extend platform for CCS
71. Page 71
► Simulation used to estimate value and delineate operating policy.
► The following plots show DCF and RO investment / operating policies for a
large number of NatGas and carbon price simulation outcomes in 2010.
► DCF suggests wait - CCS value is too low.
► RO suggests 1) build Enhanced at high CO2 price, 2) appraise at middle prices
(gas savings are important), and 3) wait at low CO2 prices.
Example: CCS and geological uncertainty
Comparison of 2010 DCF / RO CCS development policies
RO
72. Page 72
► With DCF, higher CO2 required to build CCS facility.
► Enhanced CCS with reduced gas use is only supported at very high NatGas prices.
► RO suggests investing in CCS at lower CO2 prices as the present value
(“PV”) of future CO2 costs are greater.
► Present value of future gas costs are also greater with RO so more likely to
introduce the Enhanced CCS design. Small possibility of testing porosity first.
Example: CCS and geological uncertainty
Comparison of 2016 DCF / RO CCS development policies
RO
73. Page 73
► With both DCF and RO, Enhanced CCS facility becomes less likely as there
is less time to amortise the initial CCS plant cost. Gas savings are also
generated over a shorter period.
► RO more likely to suggest building a CCS facility during 2022 as PV of future
CO2 cost stream is greater.
Example: CCS and geological uncertainty
Comparison of 2022 DCF / RO CCS development policies
RO
74. Page 74
► Near end of potential operation, less value for CO2 under RO risk valuation.
► DCF risk premium is 7% per year.
► RO short-term CO2 risk premium: (0.4 * 0.25 per year = 10% per year
► Price of CO2 risk = 0.4 annualized
► Short term CO2 price uncertainty = 0.25 annualized
Example: CCS and geological uncertainty
Comparison of 2032 DCF / RO CCS development policies
RO
75. Page 75
► Probability of DCF CCS build recommendation is greatest after 15 years of
NatGas field life when carbon prices have increased.
► Probability of doing a porosity test is highest for RO early in NatGas field life
when the investment in cost certainty can be offset against a longer period of
reduced costs.
Example: CCS and geological uncertainty
Event probabilities for DCF / RO CCS development
DCF CCS build decision
RO porosity test for basic design
RO porosity test for enhanced design
Probability
76. Page 76
► This example has shown that complex sequential decisions can be analysed
quantitatively in order to gain useful insights about how a project could be
developed across a range of business environments.
► The approach to applying a risk adjustment also matters to the analysis.
► The use of an aggregate DCF discount rate can result in undervaluing the future
costs related to carbon capture and reduced gas consumption so that there is less
incentive to invest in cost saving technologies.
► RO picks up the effects of natural gas and CO2 price reversion on risk discounting
and places a higher value on future natural gas and CO2 costs. As a result, cost
saving technologies such as CCS and the enhanced design are more likely to be
developed.
► Modelling complex decision processes requires skill and industry experience.
Analysis of future actions is an approximation of the actual problem and will
require future revisions as new information becomes available.
Example: CCS and geological uncertainty
Some final thoughts…
78. Page 78
► Flexibility is a value benefit as it allows management to limit operating
losses during adverse situations and amplify the cash flow effects of
positive business conditions over the project’s lifetime.
► There are two factors affecting the value of flexibility. These are:
► Response to new information: Management can change development or
operating policy in response to new information.
► The characteristics of uncertainty: Flexibility can be very valuable when new
information equally affects the full term structure of a forecast. Flexibility is often
worth more at a gold mine due to gold’s non-reverting behavior than at copper
mines where the copper price may exhibit reversion.
► Some examples of flexibility include:
► Investment deferral ► Compound deposit development
► Early closure / temporary closure ► Expand / reduce capacity
► Raise / lower cutoff grade ► Change processing technology
► Information gathering (exploration)
Benefiting from flexibility —
Flexibility: The value of avoiding loss and extending gains
79. Page 79
► Consider a gold mine which produces 100k ozs per year for 10 years at a
cost of $1,000 per oz with a current gold forecast of $1,250 per oz.
► A simple estimate of cumulative expected net cash flow with a Static CF
model is $250 million. Here are a couple of questions:
1) What is the hidden assumption behind a Static CF model?
2) How does a mine shutdown option affect the cash flow estimate?
Benefiting from flexibility —
An example of responding to new information (price change)
-60
-40
-20
0
20
40
60
80
100
120
140
0 2 4 6 8 10
Operatingnetcashflow(US$million)
Project year
Static cash flow Simulated NOFLEX expected CF Simulated FLEX expected CF
90% CB (NOFLEX + FLEX) 10% CB (NOFLEX) 10% CB (FLEX)
Cash flows from static and
NOFLEX / FLEX simulated models Cash flows histograms from NOFLEX / FLEX models
80. Page 80
Deferral
period
Deferral
period
1.50
2.00
2.50
3.00
3.50
4.00
4.50
0 2 4 6 8 10 12 14 16
Copperprice($/lb)
Project year
Scenario 1
Scenario 2
Scenario 3
Scenario 4
Scenario 5
► Consider an undeveloped copper project that costs $275 million to develop.
Building now has a DCF NPV of ─$32.1 million based on a long-term copper
forecast of $3.00/lb. The final build decision can be deferred for 6 years at
which time one of five price forecasts may occur with equal probability.
► A couple of questions:
1) Is the deferral option valuable and why?
2) How does reverting / non-reverting price behaviour affect option value?
Benefiting from flexibility —
An example of reverting vs non-reverting price effects
1.50
2.00
2.50
3.00
3.50
4.00
4.50
0 2 4 6 8 10 12 14 16
Copperprice($/lb)
Project year
Scenario 1
Scenario 2
Scenario 3
Scenario 4
Scenario 5
NPV ($ million)
Copper price Year 6 Scenario No flexibility Build flexibility
scenario copper price probability Scenario Weighted Scenario Weighted
1 3.90 20% 20.2 4.0 20.2 4.0
2 3.45 20% -1.6 -0.3 0.0 0.0
3 3.00 20% -32.1 -6.4 0.0 0.0
4 2.55 20% -53.1 -10.6 0.0 0.0
5 2.10 20% -84.4 -16.9 0.0 0.0
Future expected value -30.2 4.0
Reverting
price forecast
Non-reverting
price forecast
NPV ($ million)
Copper price Year 6 Scenario No flexibility Build flexibility
scenario copper price probability Scenario Weighted Scenario Weighted
1 3.90 20% 479.9 96.0 479.9 96.0
2 3.45 20% 223.9 44.8 223.9 44.8
3 3.00 20% -32.1 -6.4 0.0 0.0
4 2.55 20% -288.1 -57.6 0.0 0.0
5 2.10 20% -544.1 -108.8 0.0 0.0
Future expected value -32.1 140.8
81. Page 81
► The current approach to mine design often ignores the flexibility of mining
projects. Design is treated as a set of mutually exclusive choices rather than
sequential interrelated development decisions.
► Consider the following example:
Management at an UG mine is considering how to extend mine life. There are three choices:
1) Develop Existing Reserves for $160m
2) Develop Existing Reserves and a recently discovered Middle Zone for $265m.
3) Develop Existing Reserves, the Middle Zone and a down-dip Deep Zone for $350m.
Existing reserves
Existing reserves ($160m)
Benefiting from flexibility —
Current approach to mine design ignores flexibility
Design choice with static models Design choice with IVRM decision tree
0 2 4 6 8 10 12 14
Existing reserves / Middle Zone ($265m)
Existing reserves / Middle Zone / Deep Zone ($350m)
2 4 6 8 10 12 14
Develop
Middle Zone?
No
0
Develop
Deep Zone?Yes
Yes
Existing reserves+
Middle Zone
No
Existing reserves+
Middle Zone+Deep Zone
82. Page 82
► A mining company (“MinCo”) is studying the development of a gold project
with a high-grade open pit (“HG Pit”), a low-grade pushback (“LG Pushback”
or “LGP”) and an underground extension (“UG Zone”).
► A combined resource of 112.4 million tonnes containing a payable 6.5 million ozs.
► Three design alternatives are being studied with a maximum mill capacity of
18,000 tpd. Each design has a unique capital investment pattern ranging
from frontloaded investment to a staged investment profile.
► Total lifetime capital expenditure is $1,225 million for all designs.
► There is no clear choice as the three designs have seemingly similar NPVs
with a long-term gold forecast of $1,200/oz.
A mining company is considering three
design alternatives for a gold project with
similar NPVs but different upfront CAPEX.
How do you choose between the designs?
Compare the three designs based on capital risk
exposure and development flexibility. Generate
risk and policy information by simulating metal
prices and linking results to design features.
Example: Managing capital investment risk
Background
Issue: Solution:
83. Page 83
► Standard investment analysis considers each design alternative separately.
► Frontloaded CPX: Develop HG Pit and UG Zone together for $1,125 million. ROM capacity is
18ktpd. Develop LG Pushback in Year 13 for $100 million. ROM capacity for LG Pushback is
18ktpd. Mine life is 21 years.
► Staged CPX (1): Develop HG Pit for $775 million. ROM capacity is 18ktpd. Combine LG
Pushback and UG Zone development in Year 10 for $450 million. ROM capacity for Combined
LG Pushback and UG Zone is 18ktpd. Mine life is 21 years.
► Staged CPX (2): Develop HG Pit for $775 million. ROM capacity is 18ktpd. Develop LG
Pushback in Year 10 for $100 million. ROM capacity is 18ktpd. UG Zone developed in Year 16
for $350 million. ROM capacity for LG Pushback is 7ktpd. Mine life is 25 years.
Example: Managing capital investment risk
Three development alternatives
Project time (year)
0 5 10 15 20 25
B3
Staged CPX (2): Sequential HG Pit + LG Pushback + UG Zone
B2
B1
B1Investment decision timing point Full project development branch
Staged CPX (1): HG Pit + Combine LG Pushback / UG Zone
Frontloaded CPX: Combine HG Pit / UG Zone + Late LG Pushback
D1
NF
84. Page 84
Frontloaded CPX Staged CPX (1) Staged CPX (2)
-600
-400
-200
0
200
400
600
800
0 5 10 15 20 25
Cashflowamount($million)
Project year
Capital expenditure Operating profit
-600
-400
-200
0
200
400
600
800
0 5 10 15 20 25
Cashflowamount($million)
Project year
Capital expenditure Operating profit
-600
-400
-200
0
200
400
600
800
0 5 10 15 20 25
Cashflowamount($million)
Project year
Capital expenditure Operating profit
0
100
200
300
400
500
600 800 1,000 1,200 1,400 1,600 1,800
Averageannualcashflow
($million)
Gold price ($/oz)
Overall HG LG UG
0
100
200
300
400
500
600 800 1,000 1,200 1,400 1,600 1,800
Averageannualcashflow
($million)
Gold price ($/oz)
Overall HG LG+UG
0
100
200
300
400
500
600 800 1,000 1,200 1,400 1,600 1,800
Averageannualcashflow
($million)
Gold price ($/oz)
Overall HG + UG LG
Example: Managing capital investment risk
Cash flow information for the three design alternatives
Netcashflowand
capitalprofileCashflowrisk
► The cash flow information generated by a static cash flow model is limited.
► Amount and timing of cash flow is provided but risk is communicated with simple
measures linked to sensitivity analysis.
► Risk measures difficult to generate with a static cash flow model.
85. Page 85
Frontloaded CPX Design
Combined HG Pit and UG Zone
development
Staged CPX (1) Design
HG Pit and then Combined
LG Pushback / UG Zone
development
0
500
1,000
1,500
2,000
2,500
600 800 1,000 1,200 1,400 1,600 1,800
NPV ($ million)
Gold price ($/oz)
► Conventional cash flow analysis
suggests the Frontloaded CPX
design generates the most value.
► Capital investment efficiency of the
Staged CPX (1) design is slightly
higher (7.5%) reflecting delayed
capital expenditure
► Frontloaded CPX design is
preferred for the project when gold
prices are above $1,170/oz. The
Staged CPX (1) design is preferred
at prices below this point.
► All designs appear to have similar
sensitivity to changes in gold price.
Example: Managing capital investment risk
Standard investment analysis with static cash flow
D1 NF: Design choice and gold price sensitivity
Investment benefit
Design NPV(5%) Profitability
alternative ($ million) index
Frontloaded CPX 535 0.511
Staged CPX (1) 526 0.549
Staged CPX (2) 495 0.545
86. Page 86
0
500
1,000
1,500
2,000
2,500
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
Goldprice
(real;December31,2015;US$/loz)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
Example: Managing capital investment risk
Introducing gold price uncertainty
► Gold price uncertainty is modelled with a non-reverting distribution with an
initial long-term forecast of $1,200/oz.
► Key features include:
► Long-term forecasts move in lockstep with spot price movements. A 2% rise in the
spot price results in a 2% increase in the long-term forecast price.
► Uncertainty increases with term (time from today).
87. Page 87
Staged CPX (2)
0%
100%
200%
300%
400%
500%
0 5 10 15 20 25
CashflowCoV(%)
Project year
Annual cash flow CoV
-600
-400
-200
0
200
400
600
800
0 5 10 15 20 25
Cashflowamount($million)
Project year
Capital expenditure Expected operating profit
90% cash flow CB 10% cash flow CB
Staged CPX (1)Frontloaded CPX
Example: Managing capital investment risk
Cash flow information for the three design alternatives
Netcashflowand
capitalprofile
Cashflowrisk
► The introduction of a gold price uncertainty model provides a greater range of
cash flow information.
► Cash flow amounts are supplemented with a range of risk information such as cash
flow variability and level of uncertainty.
-600
-400
-200
0
200
400
600
800
0 5 10 15 20 25
Cashflowamount($million)
Project year
Capital expenditure Expected operating profit
90% cash flow CB 10% cash flow CB
-600
-400
-200
0
200
400
600
800
0 5 10 15 20 25
Cashflowamount($million)
Project year
Capital expenditure Expected operating profit
90% cash flow CB 10% cash flow CB
0%
100%
200%
300%
400%
500%
0 5 10 15 20 25
CashflowCoV(%)
Project year
Annual cash flow CoV
0%
100%
200%
300%
400%
500%
0 5 10 15 20 25
CashflowCoV(%)
Project year
Annual cash flow CoV
88. Page 88
► Investment benefits are unaffected by modelling gold price uncertainty.
► Other projects may have different static and dynamic NPVs from non-linearities.
► Risk information from simulation suggests project designs are risky.
► Conditional profitability index (PI) losses are high. Expect to lose $1.10 for every
$1.00 of capital invested if NPV negative.
► Conditional NPV loss for each design is also high at $1.1 billion if NPV is negative.
► Risk levels seem excessive at this point in our analysis.
-2,000 -1,500 -1,000 -500 0 500 1,000 1,500 2,000 2,500 3,000 3,500
Frontloaded
CPX
Staged
CPX (1)
Staged
CPX (2)
NPV outcomes ($ million)
Dynamiccashflow/
nodesignchoice
Expected gain
$2,084
Expected NPV
$495
Expected loss
-$1,062
Expected gain
$2,219
Expected NPV
$527
Expected loss
-$1,138
Expected gain
$2,200
Expected NPV
$535
Expected loss
-$1,170
Example: Managing capital investment risk
Investment benefits and risk exposure (no future design choice)
NPV / risk exposure map
Profitability PI risk exposure
index (PI) PI loss PI gain
0.51 -1.14 2.07
0.55 -1.18 2.29
0.55 -1.16 2.29
89. Page 89
► Future design flexibility can be reinterpreted as a decision tree which maps
decision timing (yellow boxes) and project closure (grey boxes).
► Multiple possible development paths are grouped into Frontloaded CPX and
Staged CPX (1) & (2) designs.
Example: Managing capital investment risk
Representing design flexibility with a decision tree
Frontloaded CPX design
Staged
CPX (1)
Staged CPX (2)
design
At D3, choose between:
1. Develop LG Pit+UG Zone for
$450 million,
2. Develop LG Pit for $100 million,
3. Exhaust HG Pit reserves.
Project time (year)
0 5 10 15 20 25
B3
Develop UG Zone for $350 million
or exhaust LG Pit reserves?
Develop LG Pit for $100 million
or exhaust HG Pit+UG Zone reserves?
Develop HG Pit+UG Zone for $1,125 million
or develop HG Pit for $775 million?
LG Pit
Combine HG Pit + UG Zone
HG Pit
Combine LG Pit + UG Zone
LG Pit
UG Zone
B2
X1
B1D1 Design decision point X1 Early closure point Full project development branch
D1
Flex
B1
X2 X3
D2
D3
D4
90. Page 90
Staged CPX (1) & (2) Designs
HG Pit and then choose
LG Pit / UG Zone development policy
Frontloaded
CPX design
0
500
1,000
1,500
2,000
2,500
600 800 1,000 1,200 1,400 1,600 1,800
NPV ($ million)
Gold price ($/oz)
Frontloaded CPX Design
Combined HG Pit and UG Zone
development
Staged CPX (1) Design
HG Pit and then Combined
LG Pushback / UG Zone
development
0
500
1,000
1,500
2,000
2,500
600 800 1,000 1,200 1,400 1,600 1,800
NPV ($ million)
Gold price ($/oz)
Example: Managing capital investment risk
Future design flexibility also impacts the initial investment decision
D1 Flex: Initial design choice with flexibility
► Recognizing future design flexibility
can alter your initial investment
decision.
► A static cash flow model suggests
the Frontloaded CPX design is
preferred when the Time 0 gold
price is above $1,170/oz.
► When future design flexibility is
recognized, the Frontloaded CPX
design is preferred only if the Time
0 gold price is above $1,525/oz.
► The presence of flexibility tends to
delay investment – the preference
here is to defer capital investment
until later unless gold prices are
high.
D1 NF: Initial design choice with static model
91. Page 91
Exhaust
HG Pit
LG Pushback
and then UG Zone by
investing $100 million
(and then $350 million)
Combine LG Pushback
and UG Zone by
investing $450 million.
Develop LG Pushback
by investing $100 million
Exhaust
Combined HG Pit
and UG Zone
0
500
1,000
1,500
2,000
2,500
600 800 1,000 1,200 1,400 1,600 1,800
NPV ($ million)
Gold price ($/oz)
0
500
1,000
1,500
2,000
2,500
600 800 1,000 1,200 1,400 1,600 1,800
NPV ($ million)
Gold price ($/oz)
Example: Managing capital investment risk
Design flexibility at future decision points
D3: Design flexibility in Year 10 of Staged CPX
► Design flexibility allows investment
risk to be managed.
► For Frontloaded CPX, the choice in
Year 13 is whether to invest $100
million or close the mine early.
► For Staged CPX (1) & (2), the choice
in Year 10 is invest $450 million or
$100 million or nothing (close early).
Gold price Development action
Above $1,350
Combine LG Pushback and
UG Zone
$900 - $1,350 LG Pushback then UG Zone
Below $900 Exhaust HG Pit
D2: Design flexibility in Year 13 of Frontloaded CPX
Gold price Development action
Above $1,030 Develop LG Pushback
Below $1,030 Exhaust HG Pit + UG Zone
92. Page 92
► Recognizing design flexibility provides the following analytic refinements:
► Value increases by 60% (≈$500m to $850m) and capital efficiency increases by
50% (≈$0.55 to $0.81). Preferred design is now staged development.
► Risk levels are much lower (about 50% lower) with staged development as capital
is only invested if business environment is favourable.
Example: Managing capital investment risk
Investment benefit and the risk levels of flexible development
-2,000 -1,500 -1,000 -500 0 500 1,000 1,500 2,000 2,500 3,000 3,500
Frontloaded
CPX
Staged
CPX (1)
Staged
CPX (2)
Frontloaded
CPX
Staged CPX
NPV outcomes ($ million)
Dynamiccash
flow+decisiontree
Dynamiccashflow/
nodesignchoice
Expected gain
$2,100
Expected NPV
$845
Expected loss
-$537
Expected gain
$2,153
Expected NPV
$639
Expected loss
-$982
Expected gain
$2,084
Expected NPV
$495
Expected loss
-$1,062
Expected gain
$2,219
Expected NPV
$527
Expected loss
-$1,138
Expected gain
$2,200
Expected NPV
$535
Expected loss
-$1,170
Profitability PI risk exposure
index (PI) PI loss PI gain
0.51 -1.10 2.07
0.55 -1.18 2.29
0.55 -1.16 2.29
0.60 -0.97 2.03
0.81 -0.76 2.21
NPV / risk exposure map
93. Page 93
► This IVRM case study highlights the importance of recognizing uncertainty
and its impact on design choices. In this instance, ignoring flexibility by using
a static cash flow model to support the investment decision:
Undervalues the ability to stage project development, which leads to…
Front-loading of capital investment at $1,200/oz gold, which creates…
Reduced investment efficiency and needless capital risk for your investors.
► There are a number of extensions to this example:
► Cost uncertainty ► Geological uncertainty
► Intermediate timing of the UG Zone ► Early closure
► Capacity increases ► Satellite deposits
► Exploration planning ► Project / corporate risk budgeting
Example: Managing capital investment risk
Some final thoughts…
95. Page 95
► Mining companies are increasingly using a mix of traditional and alternative
finance to manage industry volatility while building a foundation for growth
and the execution of corporate strategy. Governments are also participating
in mining projects through alternative non-tax arrangements.
► Examples of alternative finance / government participation include: .
► Commodity-linked debt ► Sliding-scale royalties
► Earn-outs ► Off-take agreements with quotation periods
► Streams ► Capital repayments from carried interests
► Windfall / profit-sharing taxes ► Finite tax-loss carry forward periods
► It is curious to note that many of the above financial assets are analysed with
dynamic cash flow models for financial reporting purposes but not in other instances.
► Dynamic cash flow analysis is required to better understand how the interaction
of equity, traditional + alternative finance, and government interests distributes
value and risk in a volatile industry.
► Static cash flow models may introduce misleading estimates of cash flow and risk
when financing and tax agreements contain contingent cash flow structures.
Financing in the mining industry
Cash flow distortions from changing business conditions
96. Page 96
0
50
100
150
200
250
300
350
400
700 800 900 1000 1100 1200 1300 1400 1500 1600
Contingentpaymentperozdelivered($)
Gold price ($/oz)
Contingent payment per AU oz delivered
0
5
10
15
20
25
30
0 1 2 3 4 5
BankCo.netcashflow
($million)
Year
Simulated cash flow Static cash flow model
► The high coupon rates for debt
issued by single-asset companies
building a mine may be reduced by
an option giving the debt holder
exposure to higher metal prices.
Example: MinCo issues a debt package that
also delivers a small amount of gold to a bank
in exchange for paying a much lower coupon
rate. The package includes a provision in
which the bank pays the mining company up to
$300/oz when gold prices are above $1,000/oz.
► Hidden issue: Cap on bank gold payment
restricts Mining Co’s exposure to high gold
prices on future gold delivery.
► Dynamic CF model is needed to see impact.
A static model estimates the finance cost is
11% and cash outflow to the bank is $95m.
Finance cost is 16% and outflow to the bank
is $105m with a dynamic model.
Financing mining projects —
Debt with embedded commodity price derivatives
Contingent bank payment on each
gold ounce delivered
Bank net cash flow on gold delivery
97. Page 97
0%
5%
10%
15%
20%
25%
900
950
1,000
1,050
1,100
1,150
1,200
1,250
1,300
1,350
1,400
1,450
1,500
1,550
1,600
1,650
1,700
1,750
1,800
NSRpayoutrate(%)
Gold price (US$/oz)
Base NSR Sliding scale NSR
0
10
20
30
40
50
60
0 2 4 6 8
Cashflow(real;US$million)
Time (year)
Expected RoyCo NSR cash flow 90% confidence bdy
10% confidence bdy Static RoyCo NSR cash flow
► …
Financing mining projects —
Royalties and streaming
► Mining companies can raise capital
by pre-selling production through
royalties and streams. Embedded
options may cause unanticipated
shifts in risk and return.
Example: MinCo negotiates with RoyCo a
lower base rate on an existing 6% royalty in
exchange for higher rates if gold prices
increase. Immediate impact is to lower mine
costs in a $1,200 gold price environment.
► Hidden issue: The possibility of much higher
royalty rates may result in a hidden value
transfer to RoyCo.
► A dynamic CF model may reveal a hidden
value transfer. A static model forecasts
average annual RoyCo cash flows of $6.4m
while a dynamic model forecasts RoyCo
cash flow of $13.4m with average 10%
confidence boundary of $43m.
Sliding royalty rate schedule
Expected royalty cash flows –
static vs dynamic models
98. Page 98
0
20
40
60
80
0 25 50 75 100 125 150 175
Totalearn-inpayment(US$million)
Cumulative after-tax cash flow (US$ million)
Total earn-in payment
Earn-in payment (static cumulative cash flow +
static contingent payment) = $67.5 million)
Cash flow range: $50m to $95m
100% of after-tax cash flows
Cash flow range: Below $50m,
after-tax cash flows to JUNIORCO
Cash flow range: $95m to $140m
50% of after-tax cash flows
to MINCO
Cash flow range:
Above $140m, earn-in
payments capped
at $67.5m
0%
20%
40%
60%
80%
100%
0
10
20
30
40
50
0 1 2 3 4 5 6 7
Cumulativeprobability
ofearlyclosure
Earn-incashflow($million)
Year
Earn-in cash flow - static Earn-in cash flow - simulated
90%/10% cash flow CB 90%/10% cash flow CB
Cum. early closure probability
► … Total earn-out payment
Earn-in cash flow + closure probability
Financing project acquisitions —
Reconciling buyer / seller value expectations with earn-outs
► Buyer / seller value differences in
M&A can sometimes be bridged by
linking part of the purchase price to
future performance. Dynamic
models are most likely required to
account for contingent payments.
Example: MinCo is selling a mature high cost
mine to JuniorCo. There is disagreement on
value so MinCo asks for an earn-out based on
cumulative cash flow to capture the possibility
of higher gold prices and mine cash flows.
► Hidden issue: A cap on earn-out payments
and the possibility of early closure may cause
lower earn-out cash flows than expected.
► A dynamic CF model shows that expected
earn-out cash flows are indeed lower than a
static forecast. Total earn-out CFs are
$67.5m with the static model while a dynamic
model forecasts earn-out CFs of $34m.
99. Page 99
► A mining company (“MinCo”) is developing a new gold project and needs to
improve its financial position due to a decline in metal prices. Reducing the
remaining project capital requirements is one possibility.
► The company has rejected raising equity or increasing debt but would consider a
streaming deal as it is considered non-dilutive.
► A streaming company (“StreamCo”) has offered staged upfront payments to
the project during construction in exchange for the future delivery of gold at
25% of the spot price.
► MinCo’s board views the stream proposal as generally favourable but would
like more clarity about the risk effects of the stream.
Owners of a developing gold project are
considering a stream agreement to improve
their financial position. How does the stream
impact project viability and risk exposure?
Test the stream’s impact on the project over a
range of gold prices with simulation to obtain
unbiased cash flows estimates and an indication
of risk exposure.
Example: Stream financing at a gold mine
Background
Issue: Solution:
100. Page 100
► The gold project is planned as a combined open pit / underground operation
with a reserve base of 3.9 million ozs. The project has a 14 year operating
life of which the last 2 years are processing a low-grade stockpile.
► Annual mill through-put is 7.7 million tonnes with annual payable gold
production of 310 thousand ozs (“koz”) during operation and 74 koz when
processing the stockpile.
► Average annual mining / milling / G&A / TCRC costs of $578/oz when operating
and $870/oz during the last two years of processing.
► Annual sustaining CAPEX is $103/oz for the first 11 years and $45/oz afterwards.
► Project is one year into a three year build program. Total CAPEX is $887
million (“m”) with the balance of $788m incurred over the next two years.
► The government participates in the project through corporate income tax
levied at a rate of 25% on both mine and stream taxable income.
► Development CAPEX is depreciated with a simple 10-year straight line schedule.
► Stream deliveries during operations are not tax-deductible for MinCo.
Example: Stream financing at a gold mine
Mine operations and government taxes
101. Page 101
0
10
20
30
40
50
900
950
1,000
1,050
1,100
1,150
1,200
1,250
1,300
1,350
1,400
1,450
1,500
1,550
1,600
1,650
1,700
1,750
1,800
Cashflow(US$million)
Gold price (US$/oz)
Total stream revenue Stream delivery payment
► StreamCo will make upfront payments totaling $160m — $95m in Year 1 and
$65m in Year 2.
► During mine operations, StreamCo will pay 25% of the spot gold price for
7.5% of gold production to a maximum gold delivery amount of 300 koz.
► Thereafter, StreamCo receives 3.75% of gold production at 25% of the gold price.
Cumulative stream gold deliveries Stream cash flow at
average annual gold production of 310 koz
Example: Stream financing at a gold mine
Stream terms
0.00
0.10
0.20
0.30
0.40
0.0
0.5
1.0
1.5
2.0
2.5
3.0
3.5
4.0
4.5
5.0
5.5
6.0
GolddeliveredtoStreamCo(millionoz)
Cumulative gold production (million oz)
7.75% production share 3.75% production share
102. Page 102
StreamEquity with streamEquity / no stream
Netcashflowand
capitalprofileCashflowrisk
Example: Stream financing at a gold mine
Cash flow information from a static model
► A limited amount of information is generated by a static cash flow model.
► Amount and timing of cash flow is provided but cash flow risk is communicated with
simple measures linked to sensitivity analysis.
103. Page 103
Static NPV gold price sensitivity
► Equity NPV has a small decline from
entering the streaming deal while
capital efficiency increases by 20%.
► Equity profitability index (PI)
increases from 0.52 to 0.63 from the
$160 million upfront payment.
► Equity NPV more sensitive to gold
price changes than Stream NPV.
► Equity and Stream NPVs are
negative when gold prices are below
$1050/oz.
Example: Stream financing at a gold mine
Standard investment analysis with static cash flow
Investment benefit
NPV Profitability
Stakeholder ($ million) index (PI)
Equity / no stream (DR=7%) 372 0.52
Equity with stream (DR=7%) 356 0.63
Stream (DR=5%) 18 0.12
104. Page 104
0
500
1,000
1,500
2,000
2,500
1/01/75 12/31/84 1/01/95 1/01/05 1/01/15 1/01/25 1/02/35
Goldprice
(real;December31,2015;US$/loz)
Time (date)
Historic spot price Year 0, 5, 10 forecast from a specific future spot price
Simulated spot price from forecast date 10% / 90% forecast confidence boundary
Example: Stream financing at a gold mine
Introducing gold price uncertainty
► Gold price uncertainty is modelled with a non-reverting distribution with an
initial long-term forecast of $1,250/oz.
► Key features include:
► Long-term forecasts move in lockstep with spot price movements. A 2% rise in the
spot price results in a 2% increase in the long-term forecast price.
► Uncertainty increases with term (time from today).
105. Page 105
StreamEquity with streamEquity / no stream
Netcashflowand
capitalprofile
CashflowriskExample: Stream financing at a gold mine
Cash flow information from a dynamic model with no early closure
► The introduction of a gold price uncertainty model and simulation provides a
greater range of cash flow information.
► Cash flow amounts are supplemented with a range of risk information such as cash
flow variability and the level of uncertainty.
106. Page 106
► Modelling gold price uncertainty reveals a small tax non-linearity.
► Dynamic Equity NPVs are 12% lower than static NPVs.
► Stream NPV less affected by non-linear tax effects.
► Risk information suggests the Stream absorbs risk from Equity.
► Lower 10% NPV confidence boundary moves from –$733m to –$676m. Equity
NPV conditional loss declines from –$490m to –$454m.
► Stream also reduces the benefits of higher gold price environments for Equity.
-800 -600 -400 -200 0 200 400 600 800 1,000 1,200 1,400
Equity / no
stream
Equity with
stream
Stream
NPV outcomes ($ million)
Dynamiccashflow/
noearlyclosure
Expected gain: $57
Expected NPV: $17
Expected loss
-$32
Expected gain
$833
Expected NPV
$311
Expected loss
-$454
Expected gain
$896
Expected NPV
$327
Expected loss
-$490
1,600
Example: Stream financing at a gold mine
Investment benefits and risk exposure (no early closure)
Profitability PI risk exposure
index (PI) PI loss PI gain
0.46 -0.69 1.26
0.55 -0.80 1.47
0.11 -0.22 0.38
NPV / risk exposure map
107. Page 107
► Early closure flexibility limits cash
flow losses due to low gold prices.
► Early closure price set to All-In-
Sustaining-Cost (“ASIC”) during
production.
► Closure price increases after Yr 11
due to low grades / stockpile ops.
► Cumulative probability of early
close is 29% to Yr 10 and 58% for
Life-of-Mine (“LOM”).
► Stream financing increases chance
of early closure.
► The early closure price increases
on average by $45/oz.
► Cumulative probability of early
close increases to 34% by Yr 10.
Early closure boundaries
Cumulative early closure probability
Example: Stream financing at a gold mine
Early closure flexibility
108. Page 108
► Early closure has minimal impact
on Equity cash flow uncertainty to
Year 10 and then generates a
greater reduction of uncertainty
during later years.
► Average annual cash flow CoV for
the first 10 years is 106% with no
close and 102% with early closure.
► Stream cash flows become more
uncertain when early closure is
recognized.
► Average annual cash flow CoV
increases from 49% to 90% after
recognizing early closure.
Equity with stream
Stream
Example: Stream financing at a gold mine
Impact of early closure flexibility on cash flow uncertainty