Portfolio Risk Analysis
Kimber Hardy
November 2012
2
The Flaw of Averages
Sam L. SavageJohn Wiley & Sons, Inc., 2009
3
Pipeline values and risk…
… single, average values can be misleading
…. “plans based on average assumptions are wrong on average”
4
Role of Portfolio Management
Portfolio Management
• Align portfolio with Strategy• Maintain portfolio balance• Maximize portfolio value
Portfolio Management
• Align portfolio with Strategy• Maintain portfolio balance• Maximize portfolio value
Company StrategyR&D investment allocation
Company StrategyR&D investment allocation
Resource Management Resource Management
TA1TA1 TA2TA2 TA3TA3 NMEDev
NMEDev LCMLCM ResRes
Total R&D InvestmentTotal R&D Investment
Company StrategyCompany Strategy
Resource supply and demand, by project and Department
Total R&D InvestmentTotal R&D Investment
Project Prioritization & Optimization
Pipeline-derived salesValue and Risk
Portfolio Value Change
Project Strategic Commercial Operational CostsProject score
Rank
A 90 70 50 60 68 1
B 65 65 50 75 64 2
C 85 85 30 50 63 3
D 75 80 30 65 63 4E 55 45 55 90 61 5
-500
0
500
1'000
1'500
2'000
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Launch probabilityN
PV
€ m
illio
ns
Pipeline value change: + 1100 m € (+24%)
Pip
elin
e va
lue
(€m
illio
ns)
-800400
1’500 5’600
4’500
20112010
0
20
40
60
80
100
120
2012 2013 2014 2015
Res
ou
rce
use
(%
)
Dept A
Dept B
Dept C
Dept D
0
500
1'000
1'500
2'000
2'500
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
€ m
illi
on
s
50%
25%
10%
75%
90%1'000
1'500
2'000
2'500
3'000
5
eNPV: often used for (late-stage) project valuation
Success
Failure
50%
50%
eNPV*
0 4 7 10
–50 90 100 100 100
–50 0 0 0 –50
Probability of Success
Cash Flows by Year
NPV*
25
eNPV: to account for timing and technical risks, NPV and probabilities are combined
*at 10% discount rate
6
-250 -50 0 +200 +400
project value (€m)
p
rob
ab
ilit
y
0.01
0.02
0.03
0.04
0.05
Fails in Preclinical. Project stopped
Successful Development &
Registration. Generates sales
-250 -50 0 +200 +400
Mean value €80m
project value (€m)
p
rob
ab
ilit
y
0.01
0.02
0.03
0.04
0.05Fails in Clinical. Project stopped
Project values: range of possible outcomes
Mean value by itself doesn’t capture
enough about the value and risk of the
project
7
Many ways to present project values and risks e.g.
Phase III Filing TPC outcome Market environment Outcome Probability
ME positive40%
Optimum TPC10%
60%ME negative
ME positive40%
Succeed Base TPC90% 80%
60%ME negative
ME positive40%
Minimum TPC10%
60%ME negative
Succeed45% 10%
Fail
55%Fail
Scenario 6 Launch 2.4%
Fail 4.5%
Fail 55.0%
Scenario 4 Launch 19.4%
Scenario 5 Launch 1.6%
Scenario 2 Launch 2.4%
Scenario 3 Launch 13.0%
Scenario 1 Launch 1.6%
-250 -50 0 +200 +400
project value (€m)
pro
ba
bil
ity
0.01
0.02
0.03
0.04
0.05
NPV given Technical Success, Rebif in Colorectal Cancer
$5,800
Low
2001
$5MM
$5,800
0%
7
20%
20%
$7,100
High
2003
$28MM
$7,100
-3%
40
40%
40%
TPP
0 50 100 150 200 250 300 350 400 450 500
Product Profile
EU Market Share
US Final Market Share
EU Annual Launch Costs
EU Price Growth Rate
EU Initial Price
US Annual Launch Costs
EU Launch Year
Material cost per mcg
US Initial Price
Base Value: $170 million
MAP
30%
30%
25
-2%
$6,500/yr
$20MM
2002
Base
$6,500/yr
BaseCase
Sensitivity analysis Decision trees and scenarios
Monte Carlo simulation Value change over time
8
How to analyse and present pipeline values, ranges and risks?
9
Typical portfolio analyses
Project Strategic Commercial Operational CostsProject score
Rank
A 90 70 50 60 68 1
B 65 65 50 75 64 2
C 85 85 30 50 63 3
D 75 80 30 65 63 4E 55 45 55 90 61 5
-500
0
500
1'000
1'500
2'000
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Launch probability
NP
V
€ m
illi
on
s
Pipeline value change: + 1100 m € (+24%)
Pip
elin
e va
lue
(€m
illio
ns)
-800400
1’500 5’600
4’500
20112010
0
500
1'000
1'500
2'000
2'500
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
€ m
illio
ns
50%
25%
10%
75%
90%1'000
1'500
2'000
2'500
3'000
Pipeline-derived sales
Project prioritization
Project value and risk
Pipeline value change
10
It can be very useful to show an analysis of potential sales with ranges …
0
500
1'000
1'500
2'000
2'500
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
€ m
illio
ns
50%
25%
10%
75%
90%1'000
1'500
2'000
2'500
3'000Pipeline-derived sales *
*indicates that there is a 10% probability that sales will be greater than the value shown and 90% probability that it be less
11
Pipeline value change: + €1’100m (+24%)
Pip
elin
e va
lue
(€ m
illi
on
s)
-800400
1’500 5’600
4’500
20112010
… or an analysis of portfolio values with an indication of confidence limits
12
An analogy with Financial Portfolios Probability analyses increasingly used: e.g. assessing downside risk in pension portfoliosPension PlanningSimulation
Initial Amount ($000) 1'000 24% chance of depletion.Withdrawal/yr. ($000) 75
Portfolio of Funds*Gold 10%Mixed Fund 25%Large Cap 25%Cash 40%
10%90% $0
$500
$1'000
$1'500
$2'000
0 5 10 15 20Years
Fu
nd
s R
emai
nin
g
Median 10% 90%*Illustrative only! Adapted from www.ProbabilityManagement.org
13
13
Pipeline Value as the sum of project mean eNPVs
505
599
625
286
820
250
450
488
835
163
580
0
1'000
2'000
3'000
4'000
5'000
6'000
Proj AProj BProj CProj DProj E
Proj Z
Total pipeline value = €5’600m
14
14
Pipeline Value as the sum of project eNPVs
This is a mean value
How to show more information about the portfolio: values and risks?
For example, what’s the probability that the value is less than €2’000m? 505
599
625
286
820
250
450
488
835
163
580
0
1'000
2'000
3'000
4'000
5'000
6'000
Proj AProj BProj CProj DProj E
Proj Z
Total pipeline value = €5’600m
15
Defined scenarios for projects
Scenarios
High (x%) Base (y%) Low (y%)
Competition
Pricing & Reimbursement
Price
Patent status
Launch date
Year of peak sales
Marketing Strength
Etc.
16
Scenarios vs. Sensitivity analysisBoth useful, but prefer scenarios as more clearly defined
NPV given Technical Success, Rebif in Colorectal Cancer
$5,800
Low
2017
$5MM
$5,800
0%
7
20%
20%
$7,100
High
2015
$28MM
$7,100
-3%
40
40%
40%
TPP
0 50 100 150 200 250 300 350 400 450 500
Product Profile
EU Market Share
US Final Market Share
EU Annual Launch Costs
EU Price Growth Rate
EU Initial Price
US Annual Launch Costs
EU Launch Year
Material cost per mcg
US Initial Price
Base Value: $170 million
MAP
30%
30%
25
-2%
$6,500/yr
$20MM
2016
Base
$6,500/yr
BaseCase
Sensitivity analysis
17
Global peak sales(€ millions)
20%
60%
20%
= scenario probability
Sales forecasts & costs defined for each scenario
1036
819
519
High Base case
Low
e.g. for sales...
18
Project probability of success
CMR & KMR benchmarks
– Useful background information when assessing project probabilities
– However, benchmarks for certain therapeutic areas have insufficient data
– And certain types of project differ markedly from benchmark• E.g. phase III projects from small vs. large Pharma companies
Phase III Oncology trials: 2000 to 2009*
Company size Positive Phase III
<$300m 0/21 (0%)
>$1’000m 21/27 (78%)
*A. Feuerstein & M.J. Ratain. J. National Cancer Institute.103(20):1-2. Nov 2nd 2011
19
Commercial and Technical Risks: project data needed for Monte Carlo simulation of portfolio value
54%
46%
750
400
200
-240
-220
NPVPhase III Filing Scenario Outcome Probability
20%
Succeed
60%
Succeed
Fail
Fail
Scenario 1
Scenario 2
Scenario 3
Launch
Launch
Launch
Fail
Fail
11%
11%
32%
6%
40%
20%
60%
10%
90%
60%
40%
20
Portfolio value: Monte Carlo simulation
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Cumulative distribution
€ m
illi
on
s
Proj A
Proj B
Proj C
Proj A
Proj B
Proj C
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
21
Portfolio Value Ranges
« Value at Risk »22% of median value
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
Cumulative distribution
€ m
illi
on
s
0
1000
2000
3000
4000
5000
6000
7000
8000
9000
10000
22
Portfolio Value Ranges
The shape of the curve describes the value – risk profile of the portfolio
23
-1'000
0
1'000
2'000
3'000
4'000
5'000
6'000
7'000
8'000
9'000
0% 5% 10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%
100%
Cumulative distribution
Po
rtfo
lio
val
ue
(€ m
illi
on
s)
Portfolio values: “the flaws of averages”
Portfolio AMean value: €500m
Low Risk/ Low Reward
Portfolio BMean value: €500m
High Risk/ High Reward
24
Impact of correlation on portfolio risk/reward profile
Four phase II projects: 70% correlation vs. no correlation
-2'000
0
2'000
4'000
6'000
8'000
10'000
12'000
14'000
0% 5% 10%
15%
20%
25%
30%
35%
40%
45%
50%
55%
60%
65%
70%
75%
80%
85%
90%
95%10
0%
Cumulative distribution
Val
ue
(€m
)
4 projects. 70% correlation
4 projects. No correlation
70% correlation
no correlation
25
But a final word on keeping presentations to the Board simple. At most this…
0
500
1'000
1'500
2'000
2'500
2012 2013 2014 2015 2016 2017 2018 2019 2020 2021
€ m
illio
ns
50%
25%
10%
75%
90%1'000
1'500
2'000
2'500
3'000
26
Pipeline value change: + €1’100m (+24%)
Pip
elin
e va
lue
(€ m
illi
on
s)
-800400
1’500 5’600
4’500
20112010
… or this to show confidence limits
27
The Flaw of Averages
Sam L. SavageJohn Wiley & Sons, Inc., 2009