The Use of Decision Tools in Biotechnology
Project and Portfolio Decision Making
Svetlana Sigalova
Vertex Pharmaceuticals Inc
Corporate Finance
March 2010
Table of Contents
Background
Introducing decision tools & analytics to Vertex’s corporate strategy process
Deep dive: building portfolio tool to support the portfolio planning process & LRP
Expanding the tools: wins in one area cause a “viral” spread to other groups
© 2010 Vertex Pharmaceuticals Incorporated
2
My Background
•
Raised in Moscow, Russia
•
Moved to US to go to college
•
Changed 3 industries prior to joining the world of biotech
•
No math / science / excel modeling background
•
Internship with Vertex during second year of MBA
•
Currently work in the strategy & analytics group (corporate finance)
© 2010 Vertex Pharmaceuticals Incorporated
3
Introducing Decision Tools into Corporate Strategic Planning
2005
• Basic LRP Model
2006 - 2007
• Scenario based planning with expected values / probabilization
2008 - 2009
• Introduce decision tools concepts & outputs to existing scenario based planning
2010
• Utilize decision tool & process to better test key drivers & assumptions
• Use tools more broadly
© 2010 Vertex Pharmaceuticals Incorporated
4
Applying Decision Tools within Corporate Strategy & Analytics
Used for
•
•
Long-range planning
Portfolio optimization process
How
•
•
•
Revenue & growth projections
Budgeting
Resource Allocation
Why
•
•
•
Prepare for the “what if” outcomes
Provide key stakeholders with a range of outcomes
Align corporate goals with portfolio strategy
Fun Facts
•
•
•
Long-range planning takes place once a year
Portfolio process takes place twice a year
All functions of the company get involved
© 2010 Vertex Pharmaceuticals Incorporated
5
Building a Portfolio Tool for Vertex
What
•
•
Dynamic revenue and expense forecasting methodology
Monte-Carlo simulation
Why
•
•
•
•
Show boundaries
Vertex portfolio was expanding - binary scenarios becoming too numerous
Provide better “risk-adjustment” process
Take into consideration key stakeholders’ risk tolerance
Fun Facts
•
•
•
No excel modeling experience prior to this project
Academic knowledge of finance (not applicable to this project)
Lots of internal and external help whenever possible
© 2010 Vertex Pharmaceuticals Incorporated
6
Building a Portfolio Tool for Vertex
How does it
work?
•
•
•
•
Key assumptions from the long-range plan
Simulate pass or fail by stage, for each program
Some simulation of commercial outcomes
Certain programs are correlated
Outputs
•
•
•
Program contribution (revenue, development costs, EBIT)
Both strategic outputs & operational metrics
10th, 90th & 50th percentiles
Fun Facts
•
•
•
•
3mo to build & 1.5yrs of continuous revision
Original model was 25MB & took 3+ hrs to run
Runs one scenario at a time (26 assets x 6 outcomes for each asset)
Need approx. 23,000 iterations
© 2010 Vertex Pharmaceuticals Incorporated
7
Communicating Results to Non - Statisticians
How
•
•
•
Know your audience
Simple and easy to read outputs – no statistics background necessary
Added simulation results to an already established process
Why was it
successful?
•
•
•
•
Simplicity
Identified and focused on what’s important
Introduced new concepts at the right time
Implemented tools one process at a time
•
Explained the difference between Monte-Carlo and Latin Hypercube sampling to a
senior executive
Once the tool was complete, the whole group took Palisade training online
Fun Facts
•
© 2010 Vertex Pharmaceuticals Incorporated
8
Corporate Strategy: Sample Outputs
WW Revenue ($B)
10&90th Percentiles
Sc.1
Sc.2
Sc.3
How likely is particular scenario to occur?
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
EBIT ($B)
Corporate Goal
Conservative, Median
Aggressive, Median
90th Percentile
Conservative approach is unlikely to meet
corporate goals, given current portfolio
assumptions
2011 2012 2013 2014 2015 2016 2017 2018 2019 2020
© 2010 Vertex Pharmaceuticals Incorporated
9
Prob. Density
Portfolio Tool: Sample Outputs
Millions
Std Dev
Millions
Key takeaways
Without a crisp analysis summary any tool could become obsolete.
Know your objective & goal upfront. It will save you a lot of “downtime”.
© 2010 Vertex Pharmaceuticals Incorporated
10
Adjusting for Risk using Monte-Carlo Simulation
Traditional eNPV Calculation Method
Traditional P&L
2010
($M)
2011
Product Revenue
Gross Margin
-
Total Costs
Operating Profit
Tax
Net Profit
NPV ($M)
443
Stage
Preclin
Phase1
Phase2a
Phase2b
Phase3
USFiling
Cumulative
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
64
54
432
358
506
416
818
661
677
549
571
465
478
391
404
334
353
293
321
268
91
77
44
38
15
12
14
12
14
12
26
366
26
308
26
267
26
242
3
75
1
36
12
12
132
234
111
197
96
171
87
155
27
48
13
23
4
8
4
8
-
-
-
-
-
23
(23)
19
(19)
42
(42)
40
(40)
95
(95)
81
(81)
49
5
40
318
32
384
26
636
26
523
26
439
(23)
(19)
15
(57)
15
(55)
34
(130)
29
(111)
19
(15)
114
203
138
246
229
407
188
335
158
281
2010 2011
100% 100%
2012
56%
12
4
8
Portfolio Tool Risk-adjustment Method
Start
POS Cume
2007 100% 100%
2007.75 100% 100%
2010.25 56% 56%
2012.5 60% 34%
2014.25 75% 25%
2016 90% 23%
23%
Probability of starting Ph 3: probability of Ph 2a x probability of Ph 2b
or 56% x 60% = 34%
Probability of incurring costs in 3Q 2012: 34%
Probability of incurring revenues (launching the drug): 23%
Problem:
Traditional risk-adjustment method shows only a portion of costs/revenue associated with the drug. When failing a stage, company
incurs all of the development costs associated with that stage and not just “62%” of the cost.
When drug is approved, company incurs all of the revenue and not just “23%” of it.
Risk – adjusted P&L
2016
25%
2017
23%
2018
23%
2019
23%
2020
23%
2021
23%
2022
23%
2023
23%
2024
23%
2025
23%
2026
23%
2027
23%
Product Revenue
Gross Margin
-
-
-
-
-
-
16
14
99
82
116
96
188
152
156
126
131
107
110
90
93
77
81
67
74
62
21
18
10
9
Total Costs
Operating Profit
23
(23)
19
(19)
23
(23)
23
(23)
32
(32)
28
(28)
12
1
9
73
7
88
6
146
6
120
6
101
6
84
6
71
6
62
6
56
1
17
0
8
Tax
Net Profit
(23)
(19)
8
(32)
8
(31)
12
(44)
10
(38)
5
(4)
26
47
32
56
53
94
43
77
36
65
30
54
26
45
22
39
20
36
6
11
3
5
eNPV ($M)
61
Probability
-
2013
56%
2014
34%
2015
34%
© 2010 Vertex Pharmaceuticals Incorporated
2028
23%
2029
23%
3
3
-
3
3
-
2030
23%
3
3
Portfolio Tool Approach
Iteration
-
3
3
3
1
2
1
2
1
2
1
2
3
4
5
Outcome
Fail in Ph 2a
Fail in Ph 2b
Fail in Ph 3
Fail in Regulatory
Launch the drug
2010
(6)
(23)
(15)
(15)
(23)
2011
(19)
(19)
(19)
(19)
2012
(16)
(57)
(57)
(57)
2013
(55)
(55)
(55)
2014
(42)
(130)
(130)
2015
(111)
(111)
2016
(15)
2017
203
2018
246
2019
407
2020
335
8
Solution:
Simulate pass / fail by stage. Incur all of costs/revenue for that time period if the drug “passed” the stage.
Estimate “true average” eNPV of a project
By asset, eNPV & 50th percentile line should converge. Then overlay scenarios to relate work back to senior management.
© 2010 Vertex Pharmaceuticals Incorporated
© 2010 Vertex Pharmaceuticals Incorporated
18
11
Table of Contents
Background
Introducing decision tools & analytics to Vertex’s corporate strategy process
Deep dive: building portfolio tool to support the portfolio planning process & LRP
Expanding the tools: wins in one area cause a “viral” spread to other groups
© 2010 Vertex Pharmaceuticals Incorporated
12
Using Palisade Suite Outside of Corporate Finance
Who
•
•
•
Business development team
Commercial group
Scientists & program leaders
What
•
•
•
Bidding & term sheet process
Sales forecasting (share, price, competition)
Sensitivity analysis
How
•
•
•
@ Risk
Top Rank
Precision Tree
Why
•
•
•
•
Ability to show multiple outcomes with assigned probabilities
Generate average of outcomes
Identify key drivers and pressure-test assumptions
Educate non-finance audience about the effect of assumptions
© 2010 Vertex Pharmaceuticals Incorporated
13
Appendix
Corporate Finance
March 2010
Corporate Budgeting: Sample Output
Budget Proxy
90th Percentile (All Succeed)
Mean (Risk-Adj)
Accounting for risk, the portfolio investment plan for . . .
• Year 1: Falls right within the budget
• Year 2: Leaves small budget cushion
Millions
• Year 3: Opportunity to invest aggressively
Year 1
Year 2
Year 3
Decision tools help to improve the corporate budgeting processes
© 2010 Vertex Pharmaceuticals Incorporated
15
Informing Business Development Process using @Risk
Monte Carlo Analysis Output
Define Variables for simulation
eNPV (after-tax)
Base
Range
Discount Rate
X%
7% - 10%
POS (Cumulative)
22.5%
15% - 30%
Price
$xK
-5K / +10K
Market Share
X%
25% - 45%
Likelihood of Early Launch *
x%
0% - 2x%
1
X% VRTX Win
$XXXM
X% VRTX Lose
0.8
Probability of VRTX Win
Base Assumptions
0.6
0.4
0.2
Bid (Millions)
Simulation can help to inform the bid strategy
© 2010 Vertex Pharmaceuticals Incorporated
16
Sensitivity Analysis using Top Rank
eNPV Sensitivity Analysis
Impact by Input
Efficacy (Penetration)
Pricing (X, XX, 2.5X)
Launch Timing (+/- 1y r)
Ph2 PoS (+/- 10%)
300%
250%
200%
150%
100%
50%
0%
-50%
-100%
-150%
Total Dev l Costs (+/- 50%)
Percent Change from Base
Help the team focus on key drivers of the program value.
© 2010 Vertex Pharmaceuticals Incorporated
17
Creating “weighted” Commercial Forecast with @Risk
Base Case
• Price = branded products
• Novel oral peak share
< XXTrial Version
@RISK
• VRTX share of novel
oral < XXPurposes Only
For Evaluation
Lower Bound Sensitivity Range
Upper Bound Sensitivity Range
• Price < 30% vs. branded products
• Price = branded products
• Novel oral peak share < X
• Novel oral peak share ~3X
• VRTX share of novel oral < X
• VRTX share of novel oral > X
Product Peak Sales (Millions)
Sensitivity ranges reflect the identified risk and opportunities in the commercial profile of the program
© 2010 Vertex Pharmaceuticals Incorporated
18
Program Valuation using @Risk
Program Value
90.0%
5.0%
Indication “A”
5.0%
Program Value
Regression - Mapped Values
Net Price
Product share
Indication “A”
has Trial
5x greater
@RISK
Version impact on the variability of
For
Evaluation
Purposes
Only
the program value than
any other indication
Other indications
@RISK
Trial Version
Mean pre-taxFor
eNPV
$X
Evaluation Purposes Only
90% Confidence Range $.8X – $1.6X
Therapy peak share
Program eNPV (Millions)
Longer right tale indicates the upside potential
outweighs the downside risks
© 2010 Vertex Pharmaceuticals Incorporated
eNPV Sensitivity (Millions)
No commercial forecast assumptions outside of
indication “A”, can move valuation by greater than 10M
19
Traditional eNPV Calculation Method
Traditional P&L
2010
($M)
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
2025
2026
2027
2028
2029
2030
14
12
14
12
Product Revenue
Gross Margin
-
-
-
-
-
-
64
54
432
358
506
416
818
661
677
549
571
465
478
391
404
334
353
293
321
268
91
77
44
38
15
12
Total Costs
Operating Profit
23
(23)
19
(19)
42
(42)
40
(40)
95
(95)
81
(81)
49
5
40
318
32
384
26
636
26
523
26
439
26
366
26
308
26
267
26
242
3
75
1
36
12
Tax
Net Profit
(23)
(19)
15
(57)
15
(55)
34
(130)
29
(111)
19
(15)
114
203
138
246
229
407
188
335
158
281
132
234
111
197
96
171
87
155
27
48
13
23
4
8
NPV ($M)
443
Stage
Preclin
Phase1
Phase2a
Phase2b
Phase3
USFiling
Cumulative
Start
POS Cume
2007 100% 100%
2007.75 100% 100%
2010.25 56% 56%
2012.5 60% 34%
2014.25 75% 25%
2016 90% 23%
23%
-
12
12
4
8
4
8
Probability of starting Ph 3: probability of Ph 2a x probability of Ph 2b
or 56% x 60% = 34%
Probability of incurring costs in 3Q 2012: 34%
Probability of incurring revenues (launching the drug): 23%
Risk – adjusted P&L
Probability
2010 2011
100% 100%
2012
56%
2013
56%
2014
34%
2015
34%
2016
25%
2017
23%
2018
23%
2019
23%
2020
23%
2021
23%
2022
23%
2023
23%
2024
23%
2025
23%
2026
23%
2027
23%
Product Revenue
Gross Margin
-
-
-
-
-
-
16
14
99
82
116
96
188
152
156
126
131
107
110
90
93
77
81
67
74
62
21
18
10
9
Total Costs
Operating Profit
23
(23)
19
(19)
23
(23)
23
(23)
32
(32)
28
(28)
12
1
9
73
7
88
6
146
6
120
6
101
6
84
6
71
6
62
6
56
1
17
0
8
Tax
Net Profit
(23)
(19)
8
(32)
8
(31)
12
(44)
10
(38)
5
(4)
26
47
32
56
53
94
43
77
36
65
30
54
26
45
22
39
20
36
6
11
3
5
eNPV ($M)
61
© 2010 Vertex Pharmaceuticals Incorporated
2028
23%
2029
23%
3
3
-
2030
23%
3
3
-
3
3
-
3
3
3
1
2
1
2
1
2
20
Portfolio Tool Risk-adjustment Method
Problem:
Traditional risk-adjustment method shows only a portion of costs/revenue associated with the drug. When failing a stage, company
incurs all of the development costs associated with that stage and not just “62%” of the cost.
When drug is approved, company incurs all of the revenue and not just “23%” of it.
Portfolio Tool Approach
Iteration
1
2
3
4
5
Outcome
Fail in Ph 2a
Fail in Ph 2b
Fail in Ph 3
Fail in Regulatory
Launch the drug
2010
(6)
(23)
(15)
(15)
(23)
2011
(19)
(19)
(19)
(19)
2012
(16)
(57)
(57)
(57)
2013
(55)
(55)
(55)
2014
(42)
(130)
(130)
2015
(111)
(111)
2016
(15)
2017
203
2018
246
2019
407
2020
335
Solution:
Simulate pass / fail by stage. Incur all of costs/revenue for that time period if the drug “passed” the stage.
Estimate “true average” eNPV of a project
© 2010 Vertex Pharmaceuticals Incorporated
21
Vertex Pharmaceuticals Inc.
History
• Founded in 1989, public in 1991 (ticker: VRTX)
• Developed 2 HIV drugs to date, commercialized by GSK
Pipeline
• 2 drugs in Ph.3 trials: HCV & Cystic Fibrosis
• 4 drugs in Ph.2 trials
• 2009 R&D expense - $551M
© 2010 Vertex Pharmaceuticals Incorporated
$45
8.5
6.8
Price Per Share
Market Capitalization, Billions
Financials
• 1 profitable quarter
• Over $3B in cumulative losses to date
• Raised over $1B in the last 2 years (43% stock dilution)
• NASDAQ 100 index best performing stock in 2008
Vertex Historical Performance
6.2
4.7
4.9
2Q 08
3Q 08
4.6
4.7
4Q 08
1Q 09
3.4
$0
1Q 08
Share Price
2Q 09
3Q 09
4Q 09
Market Cap
22