Professional Documents
Culture Documents
Present Value of Annuities Under Random Rates of Interest by Abraham Zaks Technion I.I.T. Haifa ISRAEL and University of Haifa, Haifa ISRAEL
Present Value of Annuities Under Random Rates of Interest by Abraham Zaks Technion I.I.T. Haifa ISRAEL and University of Haifa, Haifa ISRAEL
By
Abraham Zaks
and
Abstract
Some attempts were made to evaluate the future value (FV) of the expected
value and the variance for some classical cash flows (CF). The motivation stemmed
from some recursive formulas. Similar recursive formulas do not hold for the case of
the present value (PV) due to the lack of independence of some random variables .
One can get some estimates for the PV using the results about the FV. We study the
PV and we overcome the difficulty of independence by reversing the order of the CF.
It turns out that we get similar recursive formulas for the PV as for the FV in the
classical cash flows. It turns out to be similar to that used to evaluate the FV.
1. Introduction
The annuity is due if the payments are made in the beginning of each year , and
if the payments are made at the end of each year we term the annuity as in arrear .
Let the interest in the year ti be ji , and assume that these ji for i = 1, . . .
2
E(1+ ji ) = 1+j and Var(1+ ji ) = s for all i , i = 1 , . . . , n . ( 1.1)
We investigate the expected value and the variance of the present value (PV).
Dufresne (1989) and Bedard and Dufresne (2001) study similar PV under a different
set of assumptions. The future value (FV) is discussed in McCutcheon and Scott
(1986), by Zaks (2001) and by Burnecki , Marciniuk and Weron (2003). For a series
of k yearly payments PV(k) denotes the present value at the beginning of the first
year of the payments , and FV(k) denotes the future value at the end of the kth year .
2. Future Value
For the FV of the annuities due , let Sk denote the random value of the FV of an
annuity due of k payments evaluated at the end of k years ,then S = C1(1+ji ) .
1
The random variables ( Sk + Ck ) and (1+ jk+1) remain independent for all k. Let :
Scott (1986) where specific formulas for E( Sk )= µ k , and for E( Sn )= µ n are given.
In case (1d) it is well known that for all k , k = 1 , . . . , n the following hold :
mk+1 - µk+1
2
=( mk +2µ k Ck + Ck2 )[( 1+ j )2 +s 2 ]-[( µ k + Ck )( 1+ j )] 2
mk+1 -µk+1
2
=[( 1+ j )2 + s 2 ](mk -µk2 )+µk+1
2
s 2 /( 1+ j )2 (2.7)
Set S0 = 0 ,and Var( S0 ) = 0 ,and add (2.8) for k = 1 , . . . , n to get:
errors were noticed and corrected by Burnecki , Marciniuk and Weron (2003)).
3. Interlude.
Our next aim is to investigate the PV case . The PV(k) of an annuity is the value
A direct approach similar to the one we used for the FV case , seems impossible
since in the recursive formulas for the PV(k) the random variables that arise are not
annuities. The reader may compare it to that used by Dufresne (1989)). We proceed to
explore the PV in a way similar to the one used for the FV.
At this point we poimt out that estimates for the value of the PV may be easily
and these values may be used to get estimates for the PV(k) , using the FV(k) .
It is important to observe that the above relations apply to the case of annuities
that consists of a series of k annual payments made during the first k years , and
where FV(k) is their value at the end of the kth year and PV(k) is their value at the
4. Present Value
-1
It follows tha: 1- di =( 1+ jn-i+1 ) . Let us denote :
2
E( 1- di )=1- d and Var( 1- di )= z for all i , i = 1 , . . . , n . ( 4.1)
-1
In general the relation 1- d = ( 1+ j ) does not hold .
5
The problem that arises , when trying to follow a line of thought similar to the
one used for the FV in the second section , is that the random variables involved are
no longer independent .
consider all the above cases with Ck , for k = 1 , . . . , n, and set Dk = Cn-k+1 in each
In particular , the PV(k) of that annuity is the value in the beginning of the (n-k+1) th
The formulas for the values of PV(k) as annuities certain at a fixed yearly rate of
discount ○ are discussed in McCutcheon and Scott (1986). The classical cases are:
(2a) PV(k) = ak
= D1 , and
of k yearly payments evaluated at the beginning of the first year ,then A 1
= A
A (1- d )+ D (4.2)
k+1 k k+1 k+1 , for k = 1 , . . . , n - 1
) = θ
E(A 2 ) = t
and E(A for k = 1 , . . . , n (4.3)
k k k k
) = tk - θ 2
θ k = FV(k) as given above and , Var( A (4.4)
k k
The detailed evaluation of PV(k) is known e.g. McCutcheon abd Scott (1986),
) = θ .
) = θ k , and in particular for E( A
so there are known formulas for E( A k n n
) , and Var( A
that lead to the values of E( A ) , via successive iterations.
n n
To verify this point one observes that in the case (1d) it is well known that :
and ,in the other cases One can derive the recursive formulas , for k = 1 , . . . , n – 1:
2
tk+1 - θ k+1 = [(1- d)2 + z 2 ](tk - θ k2 ) + θ k2 z 2 (4.7)
)/[(1- d)2 + z 2 ] k ,and θ (k)= θ 2 /[(1- d)2 + z 2 ] k
Set M(k) = Var(A then :
k k
may be achieved along the details as introduced by Zaks (2001) and Burnecki ,
A similar approach for the PV of annuities due can be used for that of annuities
in arrear and leads to similar results . Starting by replacing (4.1) with the equalities :
We observe that Ak and (1- d k+1 ) are independent random variables and Dk+1 is
References
Bedard D., Dufresne D. (2001) Pension Funding with Moving Average Rates of
Burnecki K., Marciniuk A., Weron A. (2003) Annuities Under Random Rates of
Dufresne D. (1989) Stability of pension systems when rates of return are random ,
IME 8 , 71-76
Address
e-mail : azaks@techunix.technion..ac.il
Acknowledgement
This research was supported by the Fund for the Promotion of Research at the
Techhnion
Footnote