risk management & real options v. designing a system means sculpting its value shape stefan...

Risk Management & Real Options

V. Designing a system means sculpting its value shape

Stefan ScholtesJudge Institute of Management

University of Cambridge

MPhil Course 2004-05

2 September 2004 © Scholtes 2004

Where are we?

I. IntroductionII. The forecast is always wrong

I. The industry valuation standard: Net Present Value

II. Sensitivity analysisIII. The system value is a shape

I. Value profiles and value-at-risk charts

II. SKILL: Using a shape calculator

III. CASE: Overbooking at EasyBedsIV. Developing valuation models

I. Easybeds revisitedV. Designing a system means sculpting its value shape


Project design

Value shape / risk profile helps us analyse strengths and weaknesses of the project and optimise it

Focus on TAILS of the distribution• What causes the left tail (losses / threats) and what can we do to

avoid it? • What causes the right tail (profits / opportunities) and what can we

do to amplify it?

Aim: Re-design the project so that its value shape moves further to the right (towards the higher values)

KEY MESSAGE OF THIS SESSION:

DESIGNING A PROJECT MEANS SCULPTING ITS RISK SHAPE


Project design

Designing a project means sculpting its risk profile

Histogram

0.0%

2.0%

4.0%

6.0%

8.0%

10.0%

12.0%

14.0%

-£22

0,00

0,00

0

-£19

0,00

0,00

0

-£16

0,00

0,00

0

-£13

0,00

0,00

0

-£10

0,00

0,00

0

-£70

,000

,000

-£40

,000

,000

-£10

,000

,000

£10,

000,

000

£40,

000,

000

£70,

000,

000

£100

,000

,000

£130

,000

,000

£160

,000

,000

£190

,000

,000

£220

,000

,000

£250

,000

,000

£280

,000

,000

£310

,000

,000

£330

,000

,000

£360

,000

,000

£390

,000

,000

£420

,000

,000

£450

,000

,000

£480

,000

,000

Values

Fre

qu

ency

Design 1

Design 2

Value-at-risk chart

0.0%

10.0%

20.0%

30.0%

40.0%

50.0%

60.0%

70.0%

80.0%

90.0%

100.0%

-200,000,000 -100,000,000 0 100,000,000 200,000,000 300,000,000 400,000,000

Target value

Pro

bab

ilit

y th

at r

eali

sed

val

ue

is l

ess

than

tar

get

val

ue

Design 1

Design 2


Design decisions

Where?

When?

How big?

With whom?

Etc.


Group work

Parking garage case I


The flaw of averages

Tacit assumption: “The system value calculated on the basis of of average conditions is the average system value”

THIS IS WRONG!

Parking garage valuation based on static NPV is far from the average NPV of the Monte Carlo simulations, although the demand data for the static NPV is the correct average demand


Arguing on the basis of averages

Jane: “James, Paul, thanks for coming. You know we need to decide on the launch of the new computer games that we have developed over the last months. James, you’ve promised to produce cost, sales volume, and margin projections.”

James: “Well, boss, I’ve checked our historic data. We’ve sold 75 games of this sort over the past 4 years. Our average sales volume is 130,000 units and our average margin is £11. The average launch cost is about £800K”

Jane: “So our revenue projection is 130,000*£11=£1,43 M against an investment of £800K. That’s a juicy profit of £630K. Let’s go for it.”


Thinking in scenarios

Paul: “Well, Jane, you know that we have been less than impressed with the financial performance of our games over the past years. I have looked at our data again. The launch cost estimate of £800,000 is pretty reliable but when it comes to sales, the data varies a lot. Our marketing people tell us that a game becomes a success only if it reaches a critical sales volume of around 100,000 units. Then it sells itself without much advertising. If a game doesn’t reach the critical volume, we keep it in the shops and sell it as a niche product; there are always freaks who love our games. Over the past 4 years about 50% of our launched games became mainstream games, selling an average of 250,000 copies each at an average unit margin of £4. The niche games sold about 10,000 copies on average but we could command a high unit margin of £18 because this type of market is fairly price inelastic and our competitors don’t develop competing products. It would be great if we could predict in advance whether or not a game becomes mainstream but that is notoriously difficult. We have to bite the bullet and invest the launch cost before we know the sales success.”

Jane: “Okay, okay, Paul. I am well aware that there is a risk that the new game will not cover the launch cost. But we are selling many of these games. So we are well diversified and can therefore safely base our analysis on averages.”

IS JANE RIGHT?



Business projections of performance measures based on average conditions are typically NOT averages of the performance measures!

Formally:

])[()]([ XEfXfE

E: ExpectedE: Expectedvalue value

f(): Systemf(): Systemperformanceperformancemeasuremeasure

X: UncertaintiesX: Uncertaintiesthat determinethat determinesystem performance system performance



])[()]([ XEfXfE

E = ExpectedE = Expectedvalue value

f() = Systemf() = Systemperformanceperformancemeasuremeasure

X = UncertaintiesX = Uncertaintiesthat determinethat determinesystem performance system performance

This is what we are This is what we are interested in and what theinterested in and what theMonte Carlo model calculatesMonte Carlo model calculates

This is what a projection-This is what a projection-based model calculatesbased model calculates


System constraints and uncertainty

The mean is whereThe mean is wherethe histogram “balances”the histogram “balances”

NPVNPV

ProbabilityProbability““mass”mass”

Histogram of NPVHistogram of NPV




NPVNPV

Constraint cuts off some Constraint cuts off some good NPV scenariosgood NPV scenarios





NPVNPV





NPVNPV

The probability mass is now out of balance!The probability mass is now out of balance!




The mean is whereThe mean is wherethe histogram“balances”the histogram“balances”

NPVNPV

New balance pointNew balance pointto the left: averageto the left: average

is reducedis reduced




Constraints are often invisible in number-based models• Implicitly incorporated in the projections

Constraints will often cut off some good NPV scenarios• Bad results are not balanced out by good results

O̵ Expectation is reduced

The larger the uncertainty… • the more mass is assigned to both good and bad results• The more NPV-mass is cut off by constraint• The further to the left the balance point moves

… the larger the negative effect of constraints

Can we see the effect of uncertainty in the presence of system constraints in a numbers model?

• Compare static NPV of parking garage


Conclusion: Flaw of averages

The flaw of averages is one of the most prevalent “valuation traps”

Bear in mind: System performance based on average conditions is NOT the same as average system performance

Have seen that system constraints induce the flaw

Monte Carlo simulation avoids the flaw of averages

Sam Savage has coined the term “flaw of averages” and has collected a host of stories around the flaw

• google Sam to find out more

risk management & real options v. designing a system means sculpting its value shape stefan...

Documents