㬗

EF4327 Fixed Income Securities

Session 9: One-Step Binomial Trees
Dr. DU Jintao
City University of Hong Kong

Semester B, 2023/2024
Admin

• Presentation sequence: check the file “EF4327 presentation topics.xlsx”.

• All members must present, unless with a legitimate reason.

• https://www.cityu.edu.hk/arro/asmt/mitg_main.htm

• 15-20 min presentation +5 min Q&A.

• Public holiday in Week 10; no make-up session.

 ˊ
The Big Picture

• Basics: intro, discount factors vs. interest rates, yield curve, yield vs. return
• Bond pricing: ZCB, fixed/ floating rate coupon bond
• Interest rate risk management
• How to measure and manage the risk for your current fixed-income portfolio

E
• Duration, convexity, factor duration
• Immunization, asset liability management
• How to price and manage the risk for your future fixed-income portfolio
• FRA, bond futures, IRS, forward swap contract, and their hedging portfolios

ˊ
• Modeling the evolution of interest rates: binomial trees
• One step binomial trees
• Replicating Portfolio
• Market Price of Risk
• Risk-Neutral Probability
• Multi-step binomial trees
Outline

• Replicating Portfolio

• Market Price of Risk

• Risk-Neutral Probability
A One-Step Interest Rate Binomial Tree

• An interest-rate tree starts with the specification of the dynamics of the

short-term interest rate consistent with the information we have from
the yield curve:

Fixed

点
1
Reflects how we think
(Table 9.1)
the short rate will change
over the next 6 months.

• These dynamics reflect our views about future interest rates.

• We take the 6-month interest rate as exogenous.
A One-Step Interest Rate Binomial Tree
A One-Step Interest Rate Binomial Tree
ni
t6month wwu up
go
wceettestmtnpnu
50 up

Taieo
zeifeyypnnmu
or
in it down
50

Assumption nationis same

• All rates here are continuously compounded.
• Period vs. Time This doesn’t mean hj
• The Notation 𝑟 , the comp. freq.
T 0 1 2 3

J
a
点
nrrlf
, ,
• The tree implies E𝑟 = = 2.17% > 1.74%
• The tree imposes restrictions across interest rate instruments.
d 蕊 器
LJ6month
interestrate in6mouth
Binomial Tree for a 2-Period ZCB

o
• Pi,j(k) = the price, in period i at node j, of the bond maturing in period k.

Read from
Table 9.1 f ercer 100𝑒 . × .
goupandgoupagain

Ǘ
 executed or not
Pricing Method 1: Replication right.t I
zedo.ew.se
f sweaty.io e. o
o

iu OI
• Example: Consider the following option:
(Payoff at i =1) = 100 × max(2% – r1,0)
O
• So, on the tree we have hgmjn.texexcuate yrawe.co
(Payoff at i =1 if r1 = 3.39%) = 0
(Payoff at i =1 if r1 = 0.95%) =
• Consider the following portfolio:
Short 0.8554 units of the 1-period ZCB 囖
2 3.39 C0

_T.es zt.co95 a0
suau

a
Buy 0.8700 units of the 2-period ZCB
• In period 1, the value of the portfolio is given by
98319 了 aint IT
r1 = r1,u ⟹ 0.8700 × P1,u(2) – 0.8554 × 100 =
r1 = r1,d ⟹ 0.8700 × P1,d(2) – 0.8554 × 100 = 1.05
same as those of the option.

不
not abrtye
 3
Pricing Method 1: Replication

• Example: Consider the following option:

(Payoff at i =1) = 100 × max(2% – r1,0)
• So, on the tree we have
(Payoff at i =1 if r1 = 3.39%) = $0.00
(Payoff at i =1 if r1 = 0.95%) = $1.05
• Consider the following portfolio:
Short 0.8554 units of the 1-period ZCB
Buy 0.8700 units of the 2-period ZCB
• In period 1, the value of the portfolio is given by
r1 = r1,u ⟹ 0.8700 × P1,u(2) – 0.8554 × 100 = $0.00
r1 = r1,d ⟹ 0.8700 × P1,d(2) – 0.8554 × 100 = $1.05
same as those of the option.
Pricing Method 1: Replication

i
• So, for there to be no arbitrage, it must be that
The price of the option = the value of the portfolio in period 0
= 0.8700699 × 97.8925 − 0.8554466 × 99.1338
= $0.3697.

yield
www iexeented

• Again, all that matters is each instrument’s payoffs.

• If the option price > $0.3697, what can an arbitrageur do?
buglow an option.pre long upli.depatfhi
sell high
Replication: General Description

• Consider a portfolio with N1 units of the ZCB with maturity i = 1 and N2

f
units of the ZCB with maturity i = 2. At time i = 0 its value is
П0 = N1 × P0(1) + N2 × P0(2)
• At time i = 1,

tudo am
he in
П1,u = N1 × 100 + N2 × P1,u(2)
П1,d = N1 × 100 + N2 × zerrhduiginpēd2
P1,d(2) interesthate
andduringperiod
in
• If we want to replicate an instrument with values V1,u and V1,d,
П1,u = N1 × 100 + N2 × P1,u(2) = V1,u
П1,d = N1 × 100 + N2 × P1,d(2) = V1,d
which is a system of 2 equations in 2 unknowns (N1 and N2).
Ei
0 1.05
N2 器 98 3193 99 5261

Nz o 870069605
 NOM t lo87006905x98.3193 0
Replication: General Description M 0.855445799

• We can rewrite
П1,u = N1 × 100 + N2 × P1,u(2) = V1,u
П1,d = N1 × 100 + N2 × P1,d(2) = V1,d s
as
100 𝑃 , (2) 𝑁 𝑉,
=
100 𝑃 , (2) 𝑁 𝑉,
𝑃 (1) 𝑃 (2)
𝑁 100 𝑃 , (2) 𝑉,
• =
𝑁 100 𝑃 , (2) 𝑉,
Pricing by Replication

• Let’s revisit the option with payoff 100 × max(2% – r1,0).

– V1,u = $0 and V1,d = $1.05

𝑁 100 𝑃 , (2) 𝑉,
• =
𝑁 100 𝑃 , (2) 𝑉,
100 98.3193 0
= ,
100 99.5261 1.05

99.5261 −98.3193 0
= ,
( . . ) −100 100 1.05
0.824711 −0.814711 0
= ,
−0.828638 0.828638 1.05
−0.8554466

Ei
= .
0.8700699
Pricing by Replication: Another Example

tee
• Consider a swap with payoff $100 × (r1−2%)/2.
– V1,u = $100 × (3.39%−2%)/2 = $0.695

8
– V1,d = $100 × (0.95%−2%)/2 = −$0.525

f
 Lf

Pricing by Replication: Another Example

in
• Consider a swap with payoff $100 × (r1−2%)/2.
– V1,u = $100 × (3.39%−2%)/2 = $0.695

吉
– V1,d = $100 × (0.95%−2%)/2 = −$0.525
不
𝑁 100 𝑃 , (2) 𝑉,
•
𝑁
=
100 𝑃 , (2) 𝑉, _NX100 Nzx Ra以 Ud
0.824711 −0.814711 0.695 Nz P hu Ud
= , uh Pnu
−0.828638 0.828638 −0.525
=
1.001 sm
−1.011
. at
that is, the replicating portfolio is such that you go long 1.001 units
of the 1-period ZCB and go short 1.011 units of the 2-period ZCB.
2 eee
• The value of the swap is 1.001×99.1138−1.011×97.8925 = $0.259.
Pricing by Replication: Summary

• Given an interest rate instrument,

1. Compute its payoffs/values V1,u and V1,d.

2. Compute the weights N1 and N2 of the replicating portfolio

(consisting of two三ZCBs) using
𝑁 100 𝑃 , (2) 𝑉,
= .
𝑁 100 𝑃 , (2) 𝑉,

3. Then V0 = N1P0(1) + N2P0(2).

Itm iieuuf
set
Where Is the Probability p?

• The probability p of an increase in the short rate has been missing,

completely, from our discussion so far.
• But how is it possible that the price of an instrument that pays only when
the short rate is low is independent of the probability?
• It’s not. The probability is incorporated in the price of the 2-period ZCB:
ˊ_

• Other things equal, if p increases, the price of the ZCB will decrease
• and the price of a replicable instrument will also change accordingly.
Pricing Method 2: Adjusting for the Market Price of Risk

if
• Suppose you must price a claim to an uncertain payoff stream. What’d be
your first guess?

DC .tn
• Expected Value of the Present Value of the Payoffs.
• Well, this doesn’t work.
– St. Petersburg Paradox rii 点
悲 notgooddeal youwin
立 六 uitaxitsi getoeuanatuan
• Because not only does the probability distribution matter in valuing the
claim, but also people’s attitude toward risk.
is risk and
every one

• How can we measure the risk premium?

uawe ofinstrument
• (Dollar) Risk Premium _expected rypne
∆
E𝑒 𝑃 (2) − 𝑃 2
– For our example, 98.0658−97.8925 = $0.1733.
• So what? 1 Lhnsu
pupmui p.ua
cyieucmei
 Pricing Method 2: Adjusting for the Market Price of Risk

T.ee
• It turns out the risk premium per unit risk (the market price of risk) is the
same across all traded interest rate instruments:

E𝑒 ∆
𝑉 −𝑉 E𝑒
price ∆
𝑃 (2) − 𝑃 (2) Risk Premium
i
itpenodeg_fn cpg.tt
𝑉, −𝑉,
=
𝑃 , 2 − 𝑃 , (2)
=
Risk
=: 𝜆
instrument
anddifferentpossible esapǐu an
– Recall that V’s denote the values of an arbitrary instrument.
– According to the definition above, 𝜆 < 0, indicating that the bond
price decreases when the short rate increases.
Pricing Method 2: Adjusting for the Market Price of Risk

ˊ
• It turns out the risk premium per unit risk (the market price of risk) is the
same across all traded interest rate instruments:

E 𝑒 ∆𝑉 − 𝑉 E 𝑒 ∆ 𝑃 (2) − 𝑃 (2) Risk Premium

= = =: 𝜆
𝑉, −𝑉, 𝑃 , 2 − 𝑃 , (2) Risk

– Recall that V’s denote the values of an arbitrary instrument.

– According to the definition above, 𝜆 < 0, indicating that the bond
price decreases when the short rate increases.

∆𝑉
• 𝑉 =E𝑒 − 𝜆 (𝑉 , − 𝑉 , )
Pricing Based on the Market Price of Risk: Examples

Ü Efti6.p.ua _Po
Run
∆ . × . . .
( ) ( )
• 𝜆 = =
, , ( ) . .
nu
= −0.1436.
P. e
z_Pri.eof.pt
• Option
– E𝑒

0e
∆𝑉
o O.in O_o
= 𝑒 . × . 0 + 1.05 = $0.5205
– Risk Adjustment = −0.1436 × 0 − 1.05 = $0.1508
– 𝑉 = 0.5205 − 0.1508 = $0.3697.
zoz.cn 15week9
• Swap
5
ie𝑉
∆
tomaturity
. × .
u uwt.p
– E𝑒 = 𝑒 0.695 − 0.525 = $0.084
– Risk Adjustment = −0.1436 × 0.695 + 0.525 = −$0.175
– 𝑉 = 0.084 + 0.175 = $0.259.
e
What If We Do Not Know p?

__
• What if we miscalculate p and use a wrong number?
What If We Do Not Know p? 2
• What if we miscalculate p and use a wrong number?
• It turns out it doesn’t matter as far as asset pricing is concerned:
λ0, too, depends on p, so if we miscalculate p, we will miscalculate
the risk adjustment as well (see next slide). to be
notwine
• One error exactly counterbalances the other. P is correct
• For the risk adjustment method to work, p doesn’t have to be correct. It
just must be used consistently within the model.
What If We Do Not Know p?

ˊ
t s t
be

loilrrotseet

find wn
 What If We Do Not Know p?

For p = 0.6448, risk premium = 0, i.e., it is as if market participants are risk neutral.

Ǜzeno adjustment
yfEsgcnmostuse o
Pricing Method 3:
s
Risk-Neutral Probability

• Since it doesn’t matter which probability p we use to compute the value

of an instrument, why don’t we use the one, denoted by p*, that leads to
the simplification λ0 = 0?
– Since λ0 is common to all securities, so should p*.

__
• With the risk-neutral probability p*, we can simply say “the value of an
arbitrary instrument is its expected present value (under the risk-neutral
probability measure).”
• P0(2) = E*[e−r0×∆P1] LHS
品品time
=e −r0×∆ × [p* × P1,u(2) + (1 – p*n_n) × P1,d(2)]
or
楍 和
e P0 2   P1,d 2 
r 0
zpěnd2
在

G
p 
*
g
P1,u 2   P1,d 2 
CRHs
74 x 5
e X97.8925 99.5261
P 98.3193 995261
a6448
 1 0 3552

pffrf
Risk-Neutral Pricing

• This leads to a third recipe for pricing interest rate derivatives:

1. Compute the risk neutral probability p*.
2. Compute the price of the instrument using
o
V0 = E*[e−r0×∆V1].

• It is important to note that there’s no underlying assumption that market

participants are risk-neutral.
– They are in fact risk-averse, and there is a positive risk premium.

ㄨ
• Examples 咬 6448Xot ret.nu
–znoggé174
Option: 0.9913 × (0.6448 × 0 + 0.3552 × 1.05) = 0.3697
– Swap: 0.9913 × [0.6448 × 0.695 + 0.3552 × (−0.525)] = 0.259
Risk Neutral Expectation of Future Interest Rates

i
• The expected future interest rate under the risk neutral probability is
given by
• E*[r1] = p* × r1,u + (1 – p*) × r1,d = 0.6448 × 3.39% + 0.3552 × 0.95% =
2.5234%
• This number is higher than the true expected interest rate computed,
which was equal to E[r1] = 2.17%
– Risk neutral pricing includes the risk premium in the probability of an up
move (from p to p*) and thus increases the predicted future interest rate
– However, this does not mean that the market participants expect the interest
rate in six months to be 2.5234% instead of 2.17%
• How close is the risk neutral expectation of interest rates to the forward
rate?
Risk Neutral Expectation of Future Interest Rates

• The forward rate is given by: zxhl991338197.8925 252

f(0,1,2) = -ln(F(0,1,2))/0.5=2 × ln(P0(1) / P0(2)) = 2.52%
• This has two implications:
– Forward rates are not equal to the market expectation of future interest rates
• If we observe high forward rates, we should think about two possibilities:
1. Market participants expect higher future interest rates
2. They are strongly averse to risk, and thus the price of long-term bonds is low today
– The forward rate is not equal to the risk neutral future interest rate either,
although they are quite close
• Recall that risk neutral pricing is based on the notion of dynamic replication, which
o involves trading in securities
• Interest rates are related to securities prices through a convex relation
• Thus, the divergence between rates, is because there is a convexity adjustment
missing to make both interest rates the same

Ywtweek5 6 1_
f
Consider the Interest Rate Tree in the Table
sdfexěef
i=0 i=1

t=0 t=0.5
r1,u=4% With prob. p=1/2
r0=2%
r1,d=1% With prob. 1-p=1/2

a) Compute the expected 6-month interest rate E[r1]

b) The 1-year treasury bill is traded at P0(2)=97.4845. What is the
continuously compounded forward rate for the periods i=1 to i=2? How
does it compare with the expected rate computed in (a)?

f
c) Compute the market price of risk λ.
d) Compute the risk neutral probability p . *
rrutpzz
下

e) Price the option whose payoff at i=1 is:

100 × max(r1-2%, 0)
ssutriskuthnlpriws
f) Price the bond option whose payoff at i=1 is:
V0
max(P1(2)-98.5, 0)

uploaded Exelv andsdudm

 ˊ
of ER

s ref.lisk

neutuml Pwbnhtg de
we
another

portfolio
pniywdeh
eplzhn

End twee
 ocalc
sub c 不在

final

excn co

contact
tnnaud Gwnp
F10 1 27
t 100
o i 1.5 t F Coin
ZC0 1 100 100ECF
1015 1.5
Z 0 1.5 e
F 6 1 1.5
ZG i
érco.Dxl
Ch 08111.5 1.5
o 81
e
aq
of
Flo 1 2
2Co
zc
2 éēhanf
ércoDxi
51 2
éc 0.8
09
loo_Go EG Co 5 t ZCo Dt loozco
C 2么 05
0 05 05
eh
z Go g e e

zco.y.ie e
Ehulof 1
of
a
iselfee
Elr ÌX4 ÍX1 0.025 2.5

b P 2 9.7.4845
forwand rate In F 10.5
1 0 2

2 In Po i 1 Po 2

to
2 In e 0.974845
5
2x 1h ē 0.974845

0.030954
0980199
Pn 2

P 2 o 974gsy
0.995012
P d 2

2 05
70 e ooi199to9g 12 0.974854

0.980199 0.995012

0.198041 i
EIE P 2 P 2

Pu 2 P d 2
d
ě P
pt 2 Pnd 2

Pi u 2 Pnd 2
e2 05x 0 974845 0995012
0.70031
6.980199 0.995012
1.375121969
e prze at 1t 2
v.in
_ˊVud 0
2 05 0
e 0700131x it6.29969
option price 1.386131

f
0 Viu

1.0012 Vi d
2
Prlie e X0.5x
0.70031X0
0.29969x 1.0012

0.297355