Lecture 4

Finance of Innovation
 Binomial Model. Creating a replicating
portfolio
The objective in creating a replicating portfolio is to
use a combination of risk free borrowing/lending
and the underlying asset to create the same cash
flows as the option being valued.
• Call = Borrowing + Buying Δ of the Underlying Stock
• Put = Selling Short Δ on Underlying Asset + Lending
• The number of shares bought or sold is called the
option delta.
The principles of no arbitrage then apply, and the
value of the option has to be equal to the value of
the replicating portfolio.
2
 Binomial Model. No Flexibility

DCF – use average CF, discounted at r.

 Rate r_project - use CAPM (compare with market).
 Compare with similar asset (with absolutely
correlated return) :
𝑟_𝑎𝑠𝑠𝑒𝑡 = (𝑆𝑜𝑃𝑜 + 𝑆𝑝𝑃𝑝)/𝑆 − 1

Correlation(r_project, r_asset)=1

Po Vo Po So
E(V) S
Vp Sp
Pp Pp

3
 Binomial Model. No Flexibility

Replicating portfolio includes 𝑚 units of a

correlated asset and В bonds, risk free rate is
𝑟0, :
𝑚𝑆𝑜 + 𝐵(1 + 𝑟0) = 𝑉𝑜 ,
𝑚𝑆𝑝 + 𝐵(1 + 𝑟0) = 𝑉𝑝
Find ms and Вs. Then PV = msS+Bs(1+ r0)
 Risk-neutral probabilities approach:
• PV accounts for all risks.
• Estimate risk-neutral probabilities .
• Discount at risk-free rate (r0).
4
 Binomial Model. Flexibility

Replicating portfolio with Flexibility.

• 𝑚𝑆𝑜 + 𝐵(1 + 𝑟0) = 𝑉𝑜 ,
• 𝑚𝑆𝑝 + 𝐵(1 + 𝑟0) = max(𝑉𝑑𝑝, 𝑉𝑝) = 𝑉𝑑𝑝, for Vdp>Vp
𝑉𝑜 − 𝑉𝑑𝑝
Hence: 𝑚𝑠 = ,
𝑆𝑜− 𝑆𝑝
𝐵𝑠 = (𝑉𝑜𝑆𝑝 – 𝑉𝑑𝑝𝑆𝑜)/(𝑆𝑝 − 𝑆𝑜)/(1 + 𝑟0)
Then:
𝑃𝑉𝑅𝑂𝐴 = 𝑚𝑠𝑆 + 𝐵𝑠 = (1/(1 + 𝑟0)) (((1 + 𝑟0) +
𝑆 − 𝑆𝑝)𝑉𝑜 + 𝑉𝑑𝑝(𝑆𝑜 − 𝑆 − (1 + 𝑟0))/(𝑆𝑜 − 𝑆𝑝)
Price of option: С=PVROA- PV.

5
 Binomial Model. Replicating portfolio
• Knowing PVROA calculate risk-adjusted return:
𝑟𝑅𝐴 = (𝑉𝑜𝑃𝑜 + 𝑉𝑝𝑃𝑝)/𝑃𝑉𝑅𝑂𝐴 − 1
• Discount CF at rRA is equivalent to replicating portfolio
approach.
Replicating portfolio for option
• Replace V0 with managerial flexibility 𝐶0 = 𝑉0 − 𝑉0 = 0.
• Replace V0 with managerial flexibility
𝐶𝑝 = max(𝑉𝑑𝑝, 𝑉𝑝) − 𝑉𝑝 = 𝑉𝑑𝑝 – 𝑉𝑝 > 0
Use replicating portfolio
• 𝑚𝑐𝑠 = (С𝑜 − С𝑝)/(𝑆𝑜 − 𝑆𝑝) is Δ of option since
• 𝐶0 − 𝑚𝑐𝑠 𝑆𝑜 = 𝐶𝑝 − 𝑚𝑐𝑠 𝑆𝑝 = 𝐵(1 + 𝑟0) is risk free CF.
6
 Binomial Model. Risk-neutral probabilities

• 𝐶0 − 𝑚𝑐𝑠 𝑆𝑜 = 𝐶𝑝 − 𝑚𝑐𝑠 𝑆𝑝 is risk free CF (equal).

• Difference from replicating portfolio:
𝑚𝑐𝑠 is obtained from CF of hedging portfolio for any scenario
(portfolio behave like bond).
• Martingale condition:
(𝐶0 − 𝑚𝑐𝑠 𝑆𝑜)/(1 + 𝑟0) = 𝐵 = (𝐶 − 𝑚𝑐𝑠 𝑆)
• Recalling 𝑚𝑐𝑠 = (С𝑜 − С𝑝)/(𝑆𝑜 − 𝑆𝑝)
• Obtain
С = 𝐵 + 𝑚𝑐𝑠 𝑆 = (𝑃’𝑜С𝑜 + 𝑃’𝑝 С𝑝)/ (1 + 𝑟0), where
1 + 𝑟0 +𝑆 − 𝑆𝑝
𝑃’𝑜 =
𝑆𝑜− 𝑆𝑝
𝑆𝑜− 𝑆 − 1 + 𝑟0
𝑃’𝑝 = are risk-neutral probabilities and P’o+ P’p=1.
𝑆𝑜− 𝑆𝑝

7
 Binomial Model. Assumptions

1. The higher PV, the more valuable option.

2. Important assumption – no arbitrage
opportunities, i.e. assets with equal CF have
equal prices (law of one price)
3. Another assumption: existence of correlated
tradable asset i.e. complete market.

8
 Binomial Model. Choice of Correlated Asset

Use price of commodity in mining and oil and

gas extracting companies.
Sometimes there is no such asset.
Then use asset itself - Market Asset Disclaimer.
In our case:
𝑚𝑉𝑜 + 𝐵(1 + 𝑟0) = 𝑉𝑜 ,
𝑚𝑉𝑝 + 𝐵(1 + 𝑟0) = max(𝑉𝑑𝑝, 𝑉𝑝).

9
 Binomial Model

Po u 2V
Use uV
Po
Pp
udV
V С Po
or 𝑢 = 𝑆𝑜/𝑆,
Pp dV
d 2V
𝑑 = 𝑆𝑝 /𝑆, Pp
Risk neutral probabilities:
𝑃’𝑜 = ((1 + 𝑟0) − 𝑑)/(𝑢 − 𝑑)
𝑃’𝑝 = (𝑢 − (1 + 𝑟0))/(𝑢 − 𝑑)
Start from the last period:
𝑃𝑉’ = (1/ (1 + 𝑟0)^2)((𝑃’𝑜)2𝑢2 + 2𝑃’𝑜𝑃’𝑝𝑢𝑑 + (𝑃’𝑝)2𝑑2)𝑉

10
 Option to abandon. Replicating portfolio
If we have option to abandon (PUT) with strike Х such that
𝑢𝑑𝑉 > 𝑋 > 𝑑2𝑉
Time to expiration 2 years
Replicating portfolio for node С: 𝑚𝑑𝑉 + 𝐵 and
𝑚(𝑑𝑢𝑉) + (1 + 𝑟0)𝐵 = 𝑢𝑑𝑉
𝑚(𝑑2𝑉) + (1 + 𝑟0)𝐵 = 𝑋.
Find m* and B*.
∗ ∗
Exercise in 1 year if 𝑚 𝑑𝑉 + 𝐵 > Х. Option doesn’t expire
at C.
Find replicating portfolio for 0 and find 𝑃𝑉𝑓 with flexibility.
Then 𝑃𝑉𝑓 – 𝑃𝑉 is the price of flexibility
11
 Option to abandon. Risk adjusted return and
risk-neutral probabilities
• Similar replicating portfolio for the
option.
• Risk adjusted rate of return (RAR) at node
C:
𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑃𝑉
𝑅𝐴𝑅 𝐶 = −1
𝑉𝑎𝑙𝑢𝑒 𝑎𝑡 𝐶
= (𝑃𝑜𝑢𝑑𝑉 + 𝑃𝑝𝑋)/(𝑚 ∗ 𝑑𝑉 + 𝐵 ∗) − 1
• In risk neutral probabilities:
Put= (P’o Put o + P’p Put p)/ (1+r0)
12
 How are Real Options Assessed and Calculated?
Six steps for using the binomial model
1. Select the number of intervals within the period for which you
wish to carry out the binomial calculation and calculate n
(number of intervals divided by t)
2. Use S and n to calculate the implied upward (u) and
downward (d) movements in the value of the underlying asset
for the selected number of intervals.
3. Using the implied mean (current value of the underlying asset)
and the upward (u) and downward (d) movements calculated in
2 above, construct a roll-out (event tree) of the underlying asset
based on the assumption that it can only move upwards by u or
downwards by d for each interval. Use a geometric assumption
for the roll-out, multiplying the underlying asset by u or d for the
next part of the roll-out. When you reach the end of the roll-out,
you should have all possible values for the underlying asset.
13
 How are Real Options Assessed and Calculated?
Six steps for using the binomial model
4. Establish the decision rule. In this case (a European call
option), the decision rule is that if the value is below the exercise
price, you should not exercise the option, which is worth zero. If
the value of the underlying asset is above the exercise price, the
option will be exercised and will be worth the value of the asset
less the exercise price.
5. Use risk free discount rate, and the risk neutral probabilities of
upward and downward movements (p) and (1-p).
6. Fold back the values using the probabilities and risk free
discount rate calculated in 5 above and bring the future values
back to the previous interval node. Continue this process until
you reach the present to obtain the current value of the option.

14
 Binomial Model. ROA

Calculate PV of CFs, built event tree

Apply decision rules

Calculate PV of CFs, with flexibility at the last nodes

Apply replicating portfolio

Calculate PV of CFs, with flexibility at the previous

nodes

Apply replicating portfolio

Calculate Real option price at the begining

15
 Example. Portfolio of options
Biofuel production. NPV= $50 mln. Offer is $52 mln.
Real options:
• Building can be sold for $25 mln.
• Equipment allows to increase production by 20%
with additional costs of $10 mln.
• Equipment allows to decrease production by 50%
and sell part for $15 mln
Binomial model (PV is GBM in discrete time)
Annual volatility is 20%.
Time period is 4 years and risk free is 8%.
All above costs and sale prices are stable over time. 16
 Portfolio of options

Начало
Start 11 year
год 2 year
2 год 3 год
3 year 4 год
4 year
A B C D E
$111,28 (E1)
$91,11 (D1)
$74,59 (C1) $74,59 (E2)
$61,07 (B1) $61,07 (D2)
$50 $50,00 (C2) $50,00 (E3)
$40,94 (B2) $40,94 (D3)
$33,52 (C3) $33,52 (E4)
$27,44 (D4)
$22,47 (E5)

u= exp(0.20*1), d=1/u
Event tree with static CF with no flexibility.
17
 Portfolio of options
Опцион
Option toликвидации
abandon max(K A ; S )
Опцион
Option toрасширения
extend max(1.2S  K E ; S )
Опцион
Option toсокращения
contract max(0.5S  KC ; S )

where:
KA – exercise price of option to abandon,
КЕ – exercise price of option to expand,
КС – exercise price of option to contract.
All 3 options are American.

общее
Generalправило
decision принятия
rule решений:
max(S; K A ;1.2S  K E ; 0.5S  KC )

18
 Portfolio of options. Year 4
E
max(111.28; 25;1.2 *111.28  10; 0.5 *111.28  15)  123.53
Исполнить опцион
Best choice:
1 расширения
Option to extend
E
max(74.59; 25;1.2 * 74.59  10; 0.5 * 74.59  15)  79.51
Исполнить опцион
Best choice:
2 расширения
Option to extend
E
max(50; 25;1.2 * 50  10; 0.5 * 50  15)  50
Не исполнять
Best choice:
3 опционы
do nothing
E
max(33.52; 25;1.2 * 33.52  10; 0.5 * 33.52  15)  33.52
Не исполнять
Best choice:
4 опционы
do nothing
E
max(22.47; 25;1.2 * 22.47  10; 0.5 * 22.47  15)  26.24
Исполнить
Best опцион
choice:
5 сокращения
Option to contract

Apply replicating portfolio in those previous nodes

where we can exercise options in future.

19
 Portfolio of options
Начало
Start 11 year
год 2 year
2 год 3 год
3 year 4 year
4 год
A B C D E
$123,53 (E1)
$100,07 (D1)
$80,94
(C1) $79,51 (E2)
$65,46 (B1) $64,02 (D2)
$53,03 $51,78 (C2) $50,00 (E3)
$42,14 (B2) $40,94 (D3)
$33,93 (C3) $33,52 (E4)
$28,72 (D4)
$26,23 (E5)
• option to expand in nodes Е1 and Е2,
• option to abandon is worthless.
• option to contract can be exercised in year 3.
• After contraction option to expand and option to abandon
are worthless Then for D4, we us decision rule:

20
 Portfolio of options
Value of the project with options
$53,03 млн.> $52 млн.
Value of options is
$53,03 млн.- $50 млн. = $3,03 млн.
Each option value
Реальные опционы
Options: Стоимость
Value ($ млн.)
Опцион расширения
Option to extend 2.98
Опцион ликвидации
Option to abandon 0.03
Опцион сокращения 0.05
Option to contract
Сумма цен опционов 3.06
Sum of options

Real options in the project depend on each other.

21
 Compound options
Option on option in case of stages.
Example. Exploration. Stages:
 Preliminary estimation (1 year, I= 0.5 mln prob. of success is 0.3)
 Geological research (1 year, I= 1 mln prob. of success is 0.6)
 Advanced geological research(2 years, I= 3 mln prob. of success is 0.8)
After: invest 15 mln and obtain 25 mln.
RAR=10% for all stages
Сag= (0.8*10+0.2*0)/(1.1)^2=6.61 0.8 NPV=25-15=10
NPVag=6.61-3=3.31
Option on
Сg= (0.6*3.31+0.4*0)/1.1=1.805 0.6
AG
0
NPVg=1.805-1=0.805 0.3
Option on
GR
0.2

Сpe=0.232 0.4 0
PE
NPVpe=0.232-0.5<0
No value.
0.7 0
22
 Continuous Time, Black, Scholes, Merton Formula

As the time interval is shortened, the limiting distribution, as t ->

0, can take one of two forms.
• If as t -> 0, price changes become smaller, the limiting
distribution is the normal distribution and the price process
is a continuous one.
• If as t->0, price changes remain large, the limiting distribution
is the poisson distribution, i.e., a distribution that allows for
price jumps.
The Black-Scholes model applies when the limiting distribution is
the normal distribution , and explicitly assumes that the
priceprocess is continuous and that there are no jumps in
asset prices.

23
 Continuous Time, Black, Scholes, Merton Formula

Value of project, V, will evolve over time.

• μ = Expected return on V. This expected return will be
consistent with the project’s (nondiversifiable) risk.
• d = Payout rate on project. This is the rate of cash payout,
as fraction of V.
• So μ = 𝛿 + expected rate of capital gain.
First, suppose there is No Risk.
• Then rate of capital gain is: ∆𝑉/𝑉 = (𝜇 − 𝛿)∆𝑡
• and 𝜇 = 𝑟𝑓 , the risk-free interest rate.

24
 Continuous Time, Black, Scholes, Merton Formula

Now, suppose V is risky. Then:

∆𝑉
= 𝜇 − 𝛿 ∆𝑡 + σ𝜀𝑡
𝑉
where 𝜀𝑡 is random, zero-mean. So 𝑉 follows a random walk, like
the price of a stock.
• If all risk is diversifiable, 𝜇 = 𝑟𝑓 .
• If there is nondiversifiable risk, 𝜇 > 𝑟𝑓 .
Write process for V as:
𝑑𝑉
= 𝜇 − 𝛿 𝑑𝑡 + σ𝑑𝑧𝑡
𝑉
• where 𝑑𝑧𝑡 = 𝜀𝑡 dt is the increment of a Wiener process, and
𝜀𝑡 normally distributed
• So V follows a geometric Brownian motion (GBM). 25
 Continuous Time, Black, Scholes, Merton Formula

С(𝑇) = 𝑉(0)𝑁(𝑑1) − 𝐾 exp(−𝑟𝑓𝑇) 𝑁(𝑑2) ,

where
𝑁(𝑑1) = (pseudo)-probability call in the money
at expiry date
= option delta
1 /2 1 /2
𝑑1 = (ln(𝑉/𝐾) + 𝑟𝑓𝑇)/(𝜎𝑇 ) + 0.5(𝜎𝑇 );
1 /2
𝑑2 = 𝑑1 − 𝜎𝑇 .

26
 Continuous Time, Black, Scholes, Merton Formula

 Exercise price K is usually invested capital.

 S(t) – PV discounted to t
 Т -time to solution.
 Risk free T-bills or T-bonds for US and OFZ for Russia
 Volatility:
• Volatility of own shares,
• Volatility of shares of similar company in the same country,
• Volatility of shares of similar company in the other country
(adjustment from Damodaran website).

27
 Black, Scholes, Merton Formula
and Binomial Method

Black-Scholes is a single equation which you can

program into your Excel worksheet and enter the 5
values to obtain the option value.
• It is elegant and useful for simple real options with a single
source of uncertainty
• There is one decision date, however, which makes it
inappropriate for American options and complex options
• If you can use it, it is the simplest solution but …
The binomial model is more involved to calculate
and…
• It is useful in a larger range of real options applications
• It retains the appearance of DCF
• It is transparent and highly visual
• It handles American and European options
28
 Continuous Time, Other Cases

Mean reverting process:

𝑑𝑉
= 𝜆 𝑉ത − 𝑉 𝑑𝑡 + σ𝑑𝑧𝑡
𝑉

Process with jumps (regulatory or government

decisions)
𝑑𝑉
= 𝜇 − 𝛿 𝑑𝑡 + σ𝑑𝑧𝑡 + 𝜗𝑑𝑞𝑡
𝑉

29
Using Monte Carlo to Construct an Event Tree

30
Using Monte Carlo to Construct an Event Tree

31
 Example Monte-Carlo Approach
Model prices according to stochastic process.
Example GBM:
𝑑𝑆(𝑡)/𝑆(𝑡) = 𝜇𝑑𝑡 + 𝜎𝑑𝑊(𝑡) = 𝑟0𝑡 + 𝜎𝑑𝑊’(𝑡),
where 𝑊’(𝑡) = 𝑊(𝑡) + ((𝜇 − 𝑟0)/ 𝜎)𝑡 is BM process ΔW’(t)~N(0, Δt)
w.r.t. risk neutral probabilities
𝑆(𝑡) = 𝑆(0)exp((𝜇 − 𝜎2/2)𝑡 + 𝜎𝑊(𝑡))
= 𝑆(0)exp((𝑟0 − 𝜎2/2)𝑡 + 𝜎𝑊’(𝑡))
1. Choose S(0).
2. Estimate μ and σ.
3. Generate u from standard normal 𝑁 (𝜇, 𝜎).
4. Then: Δ𝑆(𝑡) = 𝑆(𝑡 + Δ𝑡) − 𝑆(𝑡),
/
or : 𝑆(𝑡 + Δ𝑡) − 𝑆(𝑡) = 𝑆(𝑡)(𝑟0Δ𝑡 + 𝜀(𝑡)𝜎Δ𝑡1 2), where
𝜀(𝑡)~𝑁(0, 1).
5. Obtain trajectory S(t) for t from 0 to T.
6. Simulate n=1000 trajectories and obtain 𝑆𝑖(𝑇), where i=1,…,n
7. 𝐶 = 𝐸’(max(𝑆(𝑇) − 𝑋, 0)) = Σ(max(𝑆(𝑇) − 𝑋, 0))/𝑛
32
 Drawbacks of ROA

ROA for event trees use expert opinion for scenarios.

BS unjustified:
 Illiquidity of RO,
 Lack of correlated asset
 Time dependency in volatility
 Arbitrage opportunities
 Differences between real and financial options
Higher estimates with ROA.

33
 Thank you
for your attention!