Financial Mathematics

ACTL2111 Financial Mathematics for Actuaries

Eric Cheung

UNSW Sydney
Risk and Actuarial Studies, UNSW Business School

Module 5: Stochastic Returns

1/16
 Financial Mathematics

Plan

Module 5: Stochastic Returns

Introduction
Accumulated Value of One Dollar
The Log-Normal Model

2/16
 Financial Mathematics
Module 5: Stochastic Returns
Introduction

Plan

Module 5: Stochastic Returns

Introduction
Accumulated Value of One Dollar
The Log-Normal Model

3/16
 Financial Mathematics
Module 5: Stochastic Returns
Introduction

Deterministic and Stochastic Returns

I So far we have assumed future interest rates are known

(deterministic).
I In practice, returns on investments and interest rates respond
to news and are constantly changing in a random fashion.
I For example, the returns on bonds will fluctuate randomly.
I Stochastic Models explicitly allow for this randomness by
introducing probability into models.
I Do not mix up the variability of spot rates w.r.t. maturity and
the randomness of interest rates!

3/16
 Financial Mathematics
Module 5: Stochastic Returns
Introduction

Probability Distribution

I An assumption is required for the probability distribution of

returns based on the historical data and its main features.
I The model chosen should also be easy to apply (and program)
I Interest rate (and asset price) models could have discrete or
continuous distributions.
I A commonly used continuous distribution is the Log-normal.

4/16
 Financial Mathematics
Module 5: Stochastic Returns
Accumulated Value of One Dollar

Plan

Module 5: Stochastic Returns

Introduction
Accumulated Value of One Dollar
The Log-Normal Model

5/16
 Financial Mathematics
Module 5: Stochastic Returns
Accumulated Value of One Dollar

Distribution of Accumulated Values

The accumulated value of $1 after n periods can be represented as

Sn = (1 + y1 ) (1 + y2 ) · · · (1 + yn )
Yn
= (1 + yt )
t=1

where yt is the effective interest rate (or return) in the t-th period.
Concerning the distribution of Sn :
I it can be derived under simple assumptions
I otherwise, we need simulation techniques
I we can determine the moments of Sn from
 !k 
h i Yn
E Snk = E  (1 + yt ) 
t=1
5/16
 Financial Mathematics
Module 5: Stochastic Returns
Accumulated Value of One Dollar

Independent, Identically Distributed Returns

Assume all yt ’s:

I are independent of each other; and
I are identically distributed
This is the simplest model we can assume.

Furthermore, denote

E [yt ] = j,
Var [yt ] = s 2 .

6/16
 Financial Mathematics
Module 5: Stochastic Returns
Accumulated Value of One Dollar

Moments of Sn

Consider
 !k 
n n
" #
h i Y Y
E Snk = E  (1 + yt ) =E (1 + yt )k
t=1 t=1

Since the yt are i.i.d. we have

 !k 
Y n n
Y h i
E (1 + yt )  = E (1 + yt )k
t=1 t=1

7/16
 Financial Mathematics
Module 5: Stochastic Returns
Accumulated Value of One Dollar

Expected Value

In particular
n
Y
E [Sn ] = E [(1 + yt )]
t=1
Yn
= (1 + E [yt ])
t=1
= (1 + j)n

8/16
 Financial Mathematics
Module 5: Stochastic Returns
Accumulated Value of One Dollar

Variance
We know that

Var (Sn ) = E Sn2 − (E [Sn ])2

 

so we need
n
Y h i
E Sn2 = E (1 + yt )2
 

t=1
Yn
E 1 + 2yt + yt2
 
=
t=1
Yn
1 + 2j + s 2 + j 2


=
t=1

n
= 1 + 2j + s 2 + j 2 .
9/16
 Financial Mathematics
Module 5: Stochastic Returns
Accumulated Value of One Dollar

Numerical Example
Determine the mean, variance of the accumulation of $1000 after
3 years. Returns are assumed to be random and i.i.d., with mean
0.11 and variance 0.0003.

10/16
 Financial Mathematics
Module 5: Stochastic Returns
The Log-Normal Model

Plan

Module 5: Stochastic Returns

Introduction
Accumulated Value of One Dollar
The Log-Normal Model

11/16
 Financial Mathematics
Module 5: Stochastic Returns
The Log-Normal Model

Log-Normal Model
Specifications
I the force of interest rt for the t-th period is assumed to be
normally distributed

rt ∼ N µ, σ 2 (iid)

I since
rt = ln (1 + yt )
this means that (1 + yt ) is log-normal distributed.
I note
I we have (1 + yt ) > 0 for all t, which is useful for modelling
asset prices (which cannot have a negative value)
I easy to apply in practice

11/16
 Financial Mathematics
Module 5: Stochastic Returns
The Log-Normal Model

Accumulated Value under i.i.d. Lognormal

Recall that the accumulated value of $1 is
n
Y
Sn = (1 + yt )
t=1

Notice that
n
!
Y
ln (Sn ) = ln (1 + yt )
t=1
= ln ((1 + y1 ) (1 + y2 ) .. (1 + yn ))
= ln (1 + y1 ) + ln (1 + y2 ) + .. + ln (1 + yn )

and recall that

ln (1 + yt ) ∼ N µ, σ 2

12/16
 Financial Mathematics
Module 5: Stochastic Returns
The Log-Normal Model

Accumulated Value under i.i.d. Lognormal

Thus ln (Sn ) is a sum of normal distributed random variables,

which is again normally distributed. Since they are assumed to be
i.i.d. we have:

E [ln (Sn )] = E [ln (1 + y1 ) + ln (1 + y2 ) + .. + ln (1 + yn )]

= nE [ln (1 + y1 )]

and

Var [ln (Sn )] = Var [ln (1 + y1 ) + .. + ln (1 + yn )]

= nVar [ln (1 + y1 )]

13/16
 Financial Mathematics
Module 5: Stochastic Returns
The Log-Normal Model

Lognormal Distribution

If X is lognormal with parameters µ, σ 2 then for positive k we have

 
h i
k 1 2 2
E X = exp kµ + k σ
2

Outline of proof:
If X is lognormal then Z = ln X is normal. Thus
Z
h Xi = e h i
E X k = E e kZ

and notice that the RHS of the last equation is the moment
generating function (mgf) of a normal distribution.

14/16
 Financial Mathematics
Module 5: Stochastic Returns
The Log-Normal Model

Lognormal Distribution
Useful result: X is lognormal with parameters µ, σ 2 then
 
1 2
E [X ] = exp µ + σ
2

and
    2
1 2 2 1 2
Var [X ] = exp 2µ + 2 σ − exp µ + σ
2 2
exp 2µ + 2σ 2 − exp 2µ + σ 2
   
=
exp 2µ + σ 2 exp σ 2 − 1
    
=
{E [X ]}2 exp σ 2 − 1
  
=

15/16
 Financial Mathematics
Module 5: Stochastic Returns
The Log-Normal Model

Numerical Example
Assume that returns yt are random and i.i.d., with mean 0.11 and
variance 0.0003. Furthermore assume that (1 + yt ) is log normally
distributed with parameters µ, σ 2 . Find the probability that $1000
accumulates to at least $1200 after 3 years.

16/16