1.

Status A life insurance policy is sometimes issued which pays a benefit at a time which depends on the survival characteristics of two or more people. A status is an artificially constructed life form for which the notion of life and death can be well defined. Example 111. A common artificial life form is the status which is denoted n . This is the life form which survives for exactly n time units and then dies. Example 112. Another common status is the joint life status which is constructed as follows. Given two life forms (x) and (y) the joint life status, denoted x : y, dies exactly at the time of death of the first to die of (x) and (y). Exercise 111. If (x) and (y) are independent lives, what is the survival function of the status x : y? Exercise 112. What is survival function of x : n ? Occasionally, even the order in which death occurs is important. The status 1 : n is a status which dies at the time of death of (x) if the death of (x) occurs x before time n. Otherwise, this status never dies. 1 Exercise 113. Under what circumstances does x : n die? A final status that is commonly used is the last survivor status, x : y, which is alive as long as either (x) or (y) is alive.

2. Valuing Contingent Payments The central theme of these notes is embodied in the question, What is the value today of a random sum of money which will be paid at a random time in the future? This question can now be answered. Suppose the random amount A of money is to be paid at the random time T . The value of this payment today is computed in two steps. First, the present value of this payment AvT is computed. Then the expectation of this present value random variable is computed. This expectation, E[AvT ], is the value today of the random future payment. The interpretation of this amount, E[AvT ], is as the average present value of the actual payment. Averages are reasonable in the insurance context since, from the companys point of view, there are many probabilistically similar policies for which the company is obliged to pay benefits. The average cost (and income) per policy is therefore a reasonable starting point from which to determine the premium. The expected present value is usually referred to as the actuarial present value in the insurance context. In the next few sections the actuarial present value of certain standard parts of insurance contracts are computed. Insurance contracts generally consist of two parts: benefit payments and premium payments. A life insurance policy paying benefit bt if death occurs at time t has actuarial present value E[bT vT ]. In this context the actuarial present value is also called the net single premium. The net single premium is the average value of the benefit payments, in todays dollars. The net single premium is the amount of the single premium payment which would be required on the policy issue date by an insurer with no expenses (and no profit requirement). The actuarial present value of the premium payments must be at least equal to the actuarial present value of the benefit payments, or the company would go bankrupt. In the next section methods are developed for computing the net single premium for commonly issued life insurances. The following section develops methods for computing the actuarial present value of the premium payments. These two sets of methods are then combined to enable the computation of insurance premiums.

3. Life Insurance In the case of life insurance, the amount that is paid at the time of death is usually fixed by the policy and is non-random. Assume that the force of interest is constant and known to be equal to . Also simply write T = T (x) whenever clarity does not demand the full notation. The net single premium for an insurance which pays 1 at the time of death is then E[vT ] by the principle above. The net single premium would be the idealized amount an insured would pay as a lump sum (single premium) at the time that the policy is issued. The case of periodic premium payments will be discussed later. A catalog of the various standard types of life insurance policies and the standard notation for the associated net single premium follows. In most cases the benefit amount is assumed to be $1, and in all cases the benefit is assumed to be paid at the time of death. Keep in mind that a fixed constant force of interest is also assumed and that v = 1 (1 + i) = e . Insurances Payable at the Time of Death Type Net Single Premium n-year pure endowment A 1 = E = E[vn 1 n x (n, ) (T )]
x:n

n-year term whole life n-year endowment m-year deferred n-year term whole life increasing mthly n-year term increasing annually n-year term decreasing annually

A1 = E[vT 1[0,n] (T )] x:n Ax = E[vT ] Ax:n = E[vT n ]

)]

The bar is indicative of an insurance paid at the time of death, while the subscripts denote the status whose death causes the insurance to be paid. These insurances are now reviewed on a case-by-case basis. The first type of insurance is n-year pure endowment insurance which pays the full benefit amount at the end of the nth year if the insured survives at least n years. The notation for the net single premium for a benefit amount of 1 is A 1 (or x:n occasionally in this context n Ex ). The net single premium for a pure endowment is just the actuarial present value of a lump sum payment made at a future date. This differs from the ordinary present value simply because it also takes into account the mortality characteristics of the recipient. Exercise 131. Show that n Ex = vn n px .

The second type of insurance is n-year term insurance. The net single premium for a benefit of 1 payable at the time of death for an insured (x) is denoted A1 . This x:n type insurance provides for a benefit payment only if the insured dies within n years 1 of policy inception, that is, at the time of death of the statusx : n . The third type of insurance is whole life in which the full benefit is paid no matter when the insured dies in the future. The whole life benefit can be obtained by taking the limit as n in the n-year term insurance setting. The notation for the net single premium for a benefit of 1 is Ax . Exercise 132. Suppose that T (x) has an exponential distribution with mean 50. If the force of interest is 5%, find the net single premium for a whole life policy for (x), if the benefit of $1000 is payable at the moment of death. Exercise 133. Show that Ax = Ax:n + vn n px Ax+n by conditioning on the event 1 T (x) n and also by direct reasoning from a time diagram by looking at the difference of two policies. The fourth type of insurance, n-year endowment insurance, provides for the payment of the full benefit at the time of death of the insured if this occurs before time n and for the payment of the full benefit at time n otherwise. The net single premium for a benefit of 1 is denoted Ax:n . Exercise 134. Show that Ax:n = Ax:n + A 1 . 1
x:n

Exercise 135. Use the life table to find the net single premium for a 5 year pure endowment policy for (30) assuming an interest rate of 6%. The m-year deferred n-year term insurance policy provides provides the same benefits as n year term insurance between times m and m + n provided the insured lives m years. All of the insurances discussed thus far have a fixed constant benefit. Increasing whole life insurance provides a benefit which increase linearly in time. Similarly, increasing and decreasing n-year term insurance provides for linearly increasing (decreasing) benefit over the term of the insurance. Direct computation of the net single premium for an insurance payable at the time of death is impossible using only the life table. For example, Ax = Z vt t px x+t dt. As will be seen below, under the UDD assumption, all of these net 0 single premiums can be easily related to the net single premium for an insurance

that is payable at the end of the year of death. The definition and notation for these net single premiums will now be introduced. The only difference between

these insurances and those already described is that these insurances depend on the distribution of the curtate life variable K = K(x) instead of T . The following table introduces the notation. Type Insurances Payable the End of the Year of Death Net Single Premium A1 = E[vK+1 1[0,n) (K)] x:n Ax = E[vK+1 ] Ax:n = E[v(K+1) n ]
m n Ax

n-year term whole life n-year endowment m-year deferred n-year term whole life increasing annually n-year term increasing annually n-year term decreasing annually

= E[vK+1 1[m,n+m) (K)]

(IA)x = E[vK+1 (K + 1)] (IA)1 = E[vK+1 (K + 1)1[0,n) (K)] x:n (DA)1 = E[vK+1 (n x:n K)1[0,n) (K)]

These policies have net single premiums which can be easily computed from the information in the life table. To illustrate the ease of computation when using a life table observe that from the definition Ax = vk+1
X

vk+1 k px qx+k =

dx+k lx

k=0

k=0

In practice, of course, the sum is finite. Similar computational formulas are readily obtained in the other cases. Exercise 136. Show that A 1 = A 1 and interpret the result verbally. How would x:n x:n you compute A 1 using the life table?
x:n

Under the UDD assumption formulas relating the net single premium for insurance payable at the time of death to the corresponding net single premium for insurance payable at the end of the year of death can be easily found. For example, in the case of a whole life policy Ax = E[e = E[e
T (x)

] ]

(T (x) K(x)+K(x))

= E[e (T (x) K(x)) ] E[e K(x) ] 1 = (1 e )e E[e (K(x)+1) ] i = Ax where the third equality springs from the independence of K(x) and T (x) K(x) under UDD, and the fourth equality comes from the fact that under UDD the

random variable T (x) K(x) has the uniform distribution on the interval (0,1).

Exercise 137. Can similar relationships be established for term and endowment policies? Exercise 138. Use the life table to find the net single premium for a 5 year endowment policy for (30), with death benefit paid at the moment of death, assuming an interest rate of 6%. Exercise 139. An insurance which pays a benefit amount of 1 at the end of the mth part of the year in which death occurs has net single premium denoted by Ax(m) . (m) Show that under UDD i(m) Ax = Ax . One consequence of the exercise above is that only the net single premiums for insurances payable at the end of the year of death need to be tabulated, if the UDD assumption is made. This leads to a certain amount of computational simplicity.

4. Life Annuities The premium payment part of the insurance contract is examined by developing techniques for understanding what happens when premiums are paid monthly or annually instead of just when the insurance is issued. In the non random setting a sequence of equal payments made at equal intervals in time was referred to as an annuity. Here interest centers on annuities in which the payments are made (or received) only as long as the insured survives. An annuity in which the payments are made for a nonrandom period of time is called an annuity certain. From the earlier discussion, the present value of an annuity immediate (payments begin one period in the future) with a payment of 1 in each period is n X 1 vn j an = v = i j=1 while the present value of an annuity due (payments begin immediately) with a payment of 1 in each period is a n =
n X1 n n vj = 1 v = 1 v . d 1 v j=0

These formulas will now be adapted to the case of contingent annuities in which payments are made for a random time interval. Suppose that (x) wishes to buy a life insurance policy. Then (x) will pay a premium at the beginning of each year until (x) dies. Thus the premium payments represent a life annuity due for (x). Consider the case in which the payment amount is 1. Since the premiums are only paid annually the term of this life annuity depends only on the curtate life of (x). There will be a total of K(x) + 1 payments, so the actuarial present value of the payments is a x = E[a K(x)+1 ] where the left member is a notational convention. This formula gives a x = E[a K(x)+1 ] = E[ 1 Ax 1 vK(x)+1 ]= d d

as the relationship between this life annuity due and the net single premium for a whole life policy. A similar analysis holds for life annuities immediate. Exercise 141. Compute the actuarial present value of a life annuity immediate. What is the connection with a whole life policy? Exercise 142. A life annuity due in which payments are made m times per year and each payment is 1 m has actuarial present value denoted by a (m) . Show x that

(m) Ax + d .

(m)

a (m) = 1. x

Example 141. The Mathematical Association of America offers the following alternative to members aged 60. You can pay the annual dues and subscription rate of $90, or you can become a life member for a single fee of $675. Life members are entitled to all the benefits of ordinary members, including subscriptions. Should one become a life member? To answer this question, assume that the interest rate is 6% so that the Life Table at the end of the notes can be used. The actuarial present value of a life annuity due of $90 per year is 1 A60 1 0.36913 = 90 = 1003.08. 1 v 1 1 1.06

90

Thus one should definitely consider becoming a life member. Exercise 143. What is the probability that you will get at least your moneys worth if you become a life member? What assumptions have you made? Pension benefits often take the form of a life annuity immediate. Sometimes one has the option of receiving a higher benefit, but only for a fixed number of years or until death occurs, whichever comes first. Such an annuity is called a temporary life annuity. Example 142. Suppose a life annuity immediate pays a benefit of 1 each year for n years or until (x) dies, whichever comes first. The symbol for the actuarial present value of such a policy is ax:n . How does one compute the actuarial present value of such a policy? Remember that for a life annuity immediate, payments are made at the end of each year, provided the annuitant is alive. So there will be a PK(x) n total of K(x) n payments, and ax:n = E[ j=1 vj ]. A similar argument applies in the case of an n year temporary life annuity due. In this case, payments are made at the beginning of each of n years, provided the annuitant is alive. In this P (K(x)+1) n case a x:n = E[ K(x) (n 1) vj ] = E[ 1 v d ] where the left member of this equality j=0 introduces the notation. Exercise 144. Show that Ax:n = 1 d a x:n . Find a similar relationship for ax:n . Especially in the case of pension benefits assuming that the payments are made monthly is more realistic. Suppose payments are made m times per year. In this case each payment is 1 m. One could begin from first principles (this makes a good exercise), but instead the previously established facts for

insurances together with the relationships between insurances and annuities given above will be used. Using

the obvious notation gives

1 A(m) x d (m) i 1 i(m) Ax = d(m) i 1 i(m) (1 da x = ) d(m) i(m) i id = (m) (m) a x (m) (m) i d i d +

where at the second equality the UDD assumption was used. Since this relationship i i(m) id is very useful, the actuarial symbols (m) = (m) (m) and (m) = (m) (m) are i d i d introduced. The value of these functions for selected values of m are included in the Tables for Exam M. The above relationship is then written as a (m) = (m)a x x (m) using these symbols. Exercise 145. Find a similar relationship for an annuity immediate which pays 1 m m times per year. Exercise 146. An m year deferred n year temporary life annuity due pays 1 at the beginning of each year for n years starting m years from now, provided (x) is alive at the time the payment is to be made. Find a formula for m n a x , the present value of this annuity. (When n = , the present value is denoted m a x .) A useful idealization of annuities payable at discrete times is an annuity payable continuously. Such an annuity does not exist in the real world, but serves as a useful connecting bridge between certain types of discrete annuities. Suppose that the rate at which the benefit is paid is constant and is 1 per unit time. Then during the time interval (t, t + dt) the amount paid is dt and the present value of this amount is e t dt. Thus the present value of such a continuously paid annuity over a period of n years is Zn 1 e n t an = e dt = 0 . A life annuity which is payable continuously will thus have actuarial present value 1 e T (x) ax = E[aT (x) ] = E[ ]. Exercise 147. Show that Ax = 1 ax . Find a similar relationship for ax:n .

There is one other idea of importance. In the annuity certain setting one may be interested in the accumulated value of the annuity at a certain time. For an annuity

due for a period of n years the accumulated value of the annuity at time n, denoted n by sn , is given by s n = (1 + i)n a n d = (1+i) 1 . The present value of sn is the same as the present value of the annuity. Thus the cash stream represented by the annuity is equivalent to the single payment of the amount sn at time n. This last notion has an analog in the case of life annuities. In the life annuity context
n Ex

s x:n = a
x:n

where n Ex = vn n px is the actuarial present value (net single premium) of a pure endowment of $1 payable at time n. Thus s x:n represents the amount of pure endowment payable at time n which is actuarially equivalent to the annuity.

5. Net Premiums The techniques developed for analyzing the value of benefit payments and premium payments are now combined to compute the size of the premium payment needed to pay for the benefit. To develop the ideas consider the case of an insurer who wishes to sell a fully discrete whole life policy which will be paid for by equal annual premium payments during the life of the insured. The terminology fully discrete refers to the fact that the benefit is to be paid at the end of the year of death and the premiums are to paid on a discrete basis as well. How should the insurer set the premium? A first approximation yields the net premium, which is found by using the equivalence principle: the premium should be set so that actuarial present value of the benefits paid is equal to the actuarial present value of the premiums received. Using the equivalence principle the net premium P for a fully discrete whole life policy should satisfy E[vK(x)+1 ] = PE[a K(x)+1 ] or Ax Pa x = 0. From here the net premium, which in this case is denoted Px , is easily determined. Exercise 161. Use the life table to find the net premium, P30 , for (30) if i = 0.06. The notation for other net premiums for fully discrete insurances parallel the notation for the insurance policies themselves. For example, the net annual (1 premium for an n year term policy with premiums payable monthly is1denoted P 2) .
x:n

Exercise 162. Use the life table to find P1

30:10 .

What is P 1

(12) 30:10

Exercise 163. An h payment whole life policy is one in which the premiums are paid for h years, beginnning immediately. Find a formula for h Px , the net annual premium for an h payment whole life policy. Example 161. As a more complicated example consider a recent insurance advertisement which I received. For a fixed monthly premium payment (which is constant over time) one may receive a death benefit defined as follows: 100000 1[0,65) (K(x)) + 75000 1[65,75) (K(x)) + 50000 1[75,80) (K(x)) + 25000 1[80,85) (K(x)). What is the net premium for such a policy? Assume that the interest rate is 5% so

that the life table can be used for computations. Using the equivalence principle,
.

the net annual premium P is the solution of the equation Pax:85

= 100000 Ax:65 1

+ 75000 65

x 10 Ax

+ 50000 75

x 5 Ax

+ 25000 80

x 5 Ax

in terms of certain term and deferred term insurances. Exercise 164. Compute the actual net monthly premium for (21). The methodology for finding the net premium for other types of insurance is exactly the same. The notation in the other cases is now briefly discussed. The most common type of insurance policy is one issued on a semi-continuous basis. Here the benefit is paid at the time of death, but the premiums are paid on a discrete basis. The notation for the net annual premium in the case of a whole life policy is P(Ax ). The net annual premium for a semi-continuous term policy with premiums payable mthly is P(m) (A1 ). The notation for other semi-continuous policies is similar. x:n Policies issued on a fully continuous basis pay the benefit amount at the time of death and collect premiums in the form of a continuous annuity. Obviously, such policies are of theoretical interest only. The notation here is similar to that of the semi-continuous case, with a bar placed over the P. Thus P(Ax ) is the premium rate for a fully continuous whole life policy. The equivalence principle can also be viewed in a slightly different way that is more useful for a probabilistic analysis of the insurance process. For concreteness consider the case of a fully discrete whole life policy with benefit 1 and annual premiums. The prospective loss on such a policy when the annual premium is P is the loss random variable L = vK(x)+1 Pa K(x)+1 . Notice that L is nothing more than the present value, at the time the policy is issued, of the difference between the policy benefit expense and the premium income. The equivalence principle sets the premium P so that E[L] = 0, that is, the expected loss on the policy is zero. A more detailed probabilistic analysis of the policy can be made by studying how the random variable L deviates from its mean. This will be discussed in more detail later. Exercise 165. For the fully discrete whole life policy with premium P, what is Var(L)?