Financial Mathematicg CT 1

Lecture Notes
on
Actuarial Mathematics
Jerry Alan Veeh
May 1, 2006
Copyright 2006 Jerry Alan Veeh. All rights reserved.
0. Introduction
The objective of these notes is to present the basic aspects of the theory of
insurance, concentrating on the part of this theory related to life insurance. An
understanding of the basic principles underlying this part of the subject will form a
solid foundation for further study of the theory in a more general setting.
Throughout these notes are various exercises and problems. The reader should
attempt to work all of these.
Also, problem sets consisting of multiple choice problems similar to those found
on examinations given by the Society of Actuaries are provided. The reader should
work these problem sets in the suggested time allocation after the material has been
mastered. The Tables for Exam M provided by the Society of Actuaries can be used
as an aid in solving any of the problems given here. The Illustrative Life Table
included here is a copy of the life table portion of these tables. The full set of tables
can be downloaded from the Societys web site. Familiarity with these tables is an
essential part of preparation for the examination.
Readers using these notes as preparation for the Society of Actuaries examination should master the material to the extent of being able to deliver a course on this
subject matter.
A calculator, such as the one allowed on the Society of Actuaries examinations,
will be useful in solving many of the problems here. Familiarity with this calculator
and its capabilities is an essential part of preparation for the examination.
1. Overview
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?
Such a random payment is called a contingent payment.
The theory of insurance can be viewed as the theory of contingent payments.
The insurance company makes payments to its insureds contingent upon the occurrence of some event, such as the death of the insured, an auto accident by an
insured, and so on. The insured makes premium payments to the insurance company
contingent upon being alive, having sufficient funds, and so on. A natural way to
model these contingencies mathematically is to use probability theory. Probabilistic
considerations will, therefore, play an important role in the discussion that follows.
The other central consideration in the theory of insurance is the time value of
money. Both claims and premium payments occur at various, possibly random,
points of time in the future. Since the value of a sum of money depends on the
point in time at which the funds are available, a method of comparing the value
of sums of money which become available at different points of time is needed.
This methodology is provided by the theory of interest. The theory of interest will
be studied first in a non-random setting in which all payments are assumed to be
sure to be made. Then the theory will be developed in a random environment, and
will be seen to provide a complete framework for the understanding of contingent
payments.
2. Elements of the Theory of Interest

A typical part of most insurance contracts is that the insured pays the insurer
a fixed premium on a periodic (usually annual or semiannual) basis. Money has
time value, that is, $1 in hand today is more valuable than $1 to be received one year
hence. A careful analysis of insurance problems must take this effect into account.
The purpose of this section is to examine the basic aspects of the theory of interest.
A thorough understanding of the concepts discussed here is essential.
To begin, remember the way in which compound interest works. Suppose an
amount A is invested at interest rate i per year and this interest is compounded
annually. After 1 year, the amount in the account will be A + iA = A(1 + i), and this
total amount will earn interest the second year. Thus, after n years the amount will
be A(1 + i)n . The factor (1 + i)n is sometimes called the accumulation factor. If
i 365n
) .
interest is compounded daily after the same n years the amount will be A(1 + 365
In this last context the interest rate i is called the nominal annual rate of interest.
The effective annual rate of interest is the amount of money that one unit invested
at the beginning of the year will earn during the year, when the amount earned is paid
at the end of the year. In the daily compounding example the effective annual rate
i 365
) 1. This is the rate of interest which compounded annually
of interest is (1 + 365
would provide the same return. When the time period is not specified, both nominal
and effective interest rates are assumed to be annual rates. Also, the terminology
convertible daily is sometimes used instead of compounded daily. This serves as
a reminder that at the end of a conversion period (compounding period) the interest
that has just been earned is treated as principal for the subsequent period.
Exercise 21. What is the effective rate of interest corresponding to an interest rate
of 5% compounded quarterly?
Two different investment schemes with two different nominal annual rates of
interest may in fact be equivalent, that is, may have equal dollar value at any fixed
date in the future. This possibility is illustrated by means of an example.
Example 21. Suppose I have the opportunity to invest $1 in Bank A which pays
5% interest compounded monthly. What interest rate does Bank B have to pay,
compounded daily, to provide an equivalent investment? At any time t in years

12t
365t
i
and 1 + 365
respectively.
the amount in the two banks is given by 1 + 0.05
12
Finding the nominal interest rate i which makes these two functions equal is now
an easy exercise.
Exercise 22. Find the interest rate i. What is the effective rate of interest?
Situations in which interest is compounded more often than annually will arise
2: Elements of the Theory of Interest
frequently. Some notation is needed to discuss these situations conveniently. Denote

by i(m) the nominal annual interest rate compounded m times per year which is
equivalent to the interest rate i compounded annually. This means that
i(m)
1+
m
!m
= 1 + i.
Exercise 23. Compute 0.05(12) .

An important abstraction of the idea of compound interest is the idea of continuous compounding.
If interest
is compounded n times per year the amount after

nt
i
t years is given by 1 + n . Letting n in this expression produces eit , and
this corresponds to the notion of instantaneous compounding of interest. In this
context denote by the rate of instantaneous compounding which is equivalent
to interest rate i. Here is called the force of interest. The force of interest is
extremely important from a theoretical standpoint and also provides some useful
quick approximations.
Exercise 24. Show that = ln(1 + i).
Exercise 25. Find the force of interest which is equivalent to 5% compounded
daily.
The converse of the problem of finding the amount after n years at compound
interest is as follows. Suppose the objective is to have an amount A n years hence.
If money can be invested at interest rate i, how much should be deposited today in
order to achieve this objective? The amount required is A(1 + i)n . This quantity is
called the present value of A. The factor (1 + i)1 is often called the discount factor
and is denoted by v. The notation vi is used if the value of i needs to be specified.
Example 22. Suppose the annual interest rate is 5%. What is the present value of
a payment of $2000 payable 10 years from now? The present value is $2000(1 +
0.05)10 = $1227.83.
The notion of present value is used to move payments of money through time
in order to simplify the analysis of a complex sequence of payments. In the simple
case of the last example the important idea is this. Suppose you were given the
following choice. You may either receive $1227.83 today or you may receive $2000
10 years from now. If you can earn 5% on your money (compounded annually)
you should be indifferent between these two choices. Under the assumption of an
interest rate of 5%, the payment of $2000 in 10 years can be replaced by a payment
of $1227.83 today. Thus the payment of $2000 can be moved through time using the
idea of present value. A visual aid that is often used is that of a time diagram which
shows the time and amounts that are paid. Under the assumption of an interest rate
of 5%, the following two diagrams are equivalent.
Two Equivalent Cash Flows
$2000
.......................................................................................................................................................
time (years)
10
$1227.83
.......................................................................................................................................................
time (years)
10
The advantage of moving amounts of money through time is that once all
amounts are paid at the same point in time, the most favorable option is readily
apparent.
Exercise 26. What happens in comparing these cash flows if the interest rate is
6% rather than 5%?
Notice too that a payment amount can be easily moved either forward or backward in time. A positive power of v is used to move an amount backward in time; a
negative power of v is used to move an amount forward in time.
In an interest payment setting, the payment of interest of i at the end of the
period is equivalent to the payment of d at the beginning of the period. Such a
payment at the beginning of a period is called a discount. Formally, the effective
annual rate of discount is the amount of discount paid at the beginning of a year
when the amount invested at the end of the year is a unit amount. What relationship
between i and d must hold for a discount payment to be equivalent to the interest
payment? The time diagram is as follows.
Equivalence of Interest and Discount
i
.......................................................................................................................................................
d
.......................................................................................................................................................
The relationship is d = iv follows by moving the interest payment back in time

to the equivalent payment of iv at time 0.
Exercise 27. Denote by d(m) the rate of discount payable m times per year that is
equivalent to a nominal annual rate of interest i. What is the relationship between
d(m) and i? Between d(m) and i(m) ? Hint: Draw the time diagram illustrating the two
payments made at time 0 and 1/ m.
Exercise 28. Treasury bills (United States debt obligations) pay discount rather
than interest. At a recent sale the discount rate for a 3 month bill was 5%. What is
the equivalent rate of interest?
The notation and the relationships thus far are summarized in the string of
equalities
!m
!m
i(m)
d(m)
1+i= 1+
= 1
= v1 = e .
m
m
Another notion that is sometimes used is that of simple interest. If an amount
A is deposited at interest rate i per period for t time units and earns simple interst,
the amount at the end of the period is A(1 + it). Simple interest is often used over
short time intervals, since the computations are easier than with compound interest.
The most important facts are these.
(1) Once an interest rate is specified, a dollar amount payable at one time can
be exchanged for an equivalent dollar amount payable at another time by
multiplying the original dollar amount by an appropriate power of v.
(2) The five sided equality above allows interest rates to be expressed relative
to a convenient time scale for computation.
These two ideas will be used repeatedly in what follows.
Problems
Problem 21. Show that if i > 0 then
d < d(2) < d(3) < < < < i(3) < i(2) < i.
Problem 22. Show that limm d(m) = limm i(m) = .
Problem 23. Calculate the nominal rate of interest convertible once every 4 years
that is equivalent to a nominal rate of discount convertible quarterly.
Problem 24. Interest rates are not always the same throughout time. In theoretical
studies such scenarios are usually modelled by allowing the force of interest to
depend on time. Consider the situation in which $1 is invested at time 0 in an
account which pays interest at a constant force of interest . What is the amount
A(t) in the account at time t? What is the relationship between A (t) and A(t)? More
generally, suppose the force of interest at time t is (t). Argue that A (t) = (t)A(t),
and solve this equation to find an explicit formula for A(t) in terms of (t) alone.
Problem 25. Suppose that a fund initially containing $1000 accumulates with a
force of interest (t) = 1/ (1 + t), for t > 0. What is the value of the fund after 5
years?
Problem 26. Suppose a fund accumulates at an annual rate of simple interest of i.
What force of interest (t) provides an equivalent return?
Problem 27. Show that d = 1 v. Is there a similar equation involving d(m) ?
Problem 28. Show that d = iv. Is there a similar equation involving d(m) and i(m) ?
Problem 29. Show that if interest is paid at rate i, the amount at time t under
simple interest is more than the amount at time t under compound interest provided
t < 1. Show that the reverse inequality holds if t > 1.
Problem 210. Compute the derivatives
d
d
d and .
di
dv

Solutions to Problems
Problem 21. An analytic argument is possible directly from the formulas.
For example, (1 + i(m) / m)m = 1 + i = e so i(m) = m(e / m 1). Consider m as a
continuous variable and show that the right hand side is a decreasing function
of m for fixed i. Can you give a purely verbal argument? Hint: How does
an investment with nominal rate i(2) compounded annually compare with an
investment at nominal rate i(2) compounded twice a year?
Problem 22. Since i(m) = m((1 + i)1/ m 1) the limit can be evaluated directly
using LHopitals rule, Maclaurin expansions, or the definition of derivative.
Problem 23. The relevant equation is 1 + 4i(1/ 4)
1/ 4
4
= 1 d(4) / 4 .
Problem 24. In the constant force setting A(t) = e t and A (t) = A(t).
The equationR A (t) = (t)A(t) can be solved by separation of variables to give
t
A(t) = A(0)e
(s) ds
.
R5
Problem 25.
The amount in the fund after 5 years is 1000e
1000eln(6)ln(1) = 6000.
(t) dt
Rt
(s) ds
0
Problem
26.
The
force
of
interest
must
satisfy
1
+
it
=
e
for all t > 0.
Z t
Thus
(s) ds = ln(1 + it), and differentiation using the Fundamental Theorem
0
of Calculus shows that this implies (t) = i/ (1 + it), for t > 0.

Problem 27. 1 d(m) / m = v1/ m .
Problem 28. d(m) / m = v1/ m i(m) / m.
Problem 29. The problem is to show that 1 + it > (1 + i)t if t < 1, with the
reverse inequality for t > 1. The function 1 + it is a linear function of t taking
the value 1 when t = 0 and the value 1 + i when t = 1. The function (1 + i)t is a
convex function which takes the value 1 when t = 0 and 1 + i when t = 1.
Problem 210.
d
d
d
d
d = (11/ (1+i)) = (1+i)2 , and = ( ln(v)) = v1 .
di
di
dv
dv

Solutions to Exercises
Exercise 21. The equation to be solved is (1 + 0.05/ 4)4 = 1 + i, where i is the
effective rate of interest.
Exercise 22. Taking tth roots of both sides of the equation shows that t plays
no role in determining i and leads to the equation i = 365((1+0.05/ 12)12/ 365 1) =
0.04989.
Exercise 23. 0.05(12) = 12((1 + 0.05)1/ 12 1) = 0.04888.
Exercise 24. The requirement for equivalence is that e = 1 + i.
Exercise 25. Here e = (1 + 0.05/ 365)365 , so that = 0.4999. So as a rough
approximation when compounding daily the force of interest is the same as the
nominal interest rate.
Exercise 26. The present value in this case is $2000(1 + 0.06)10 = $1116.79.
Exercise 27. A payment of d(m) / m made at time 0 is required to be equivalent
to a payment of i(m) / m made at time 1/ m. Hence d(m) / m = v1/ m i(m) / m. Since
v1/ m = (1 + i)1/ m = 1 + i(m) / m this gives d(m) / m = 1 v1/ m or 1 + i = (1 d(m) / m)m .
Another relation is that d(m) / m i(m) / m = (d(m) / m)(i(m) / m).
Exercise 28. The given information is d(4) = 0.05, from which i can be
obtained using the formula of the previous exercise as i = (1 0.05/ 4)4 1 =
0.0516.
10
3. Cash Flow Valuation

Most of the remainder of these notes will consist of analyzing situations similar
to the following. Cash payments of amounts C0 , C1 , . . . , Cn are to be received at
times 0, 1, . . . , n. A cash flow diagram is as follows.
A General Cash Flow
C0
Cn
C1
........................................................................................................................................................................................................................................................................................................................................................................................
...
n
0
1
The payment amounts may be either postive or negative. A positive amount denotes
a cash inflow; a negative amount denotes a cash outflow.
There are 3 types of questions about this general setting.
(1) If the cash amounts and interest rate are given, what is the value of the cash
flow at a given time point?
(2) If the interest rate and all but one of the cash amounts are given, what should
the remaining amount be in order to make the value of the cash flow equal
to a given value?
(3) What interest rate makes the value of the cash flow equal to a given value?
Here are a few simple examples.
Example 31. What is the value of this stream of payments at a given time t? The
payment Cj made at time j is equivalent to a payment of Cj vjt at time t. So the value
of the cash flow stream at time t is
n
X
Cj vjt .
j=0
Example 32. Instead of making payments of 300, 400, and 700 at the end of
years 1, 2, and 3, the borrower prefers to make a single payment of 1400. At what
time should this payment be made if the interest rate is 6% compounded annually?
Computing all of the present values at time 0 shows that the required time t satisfies
the equation of value 300(1.06)1 + 400(1.06)2 + 700(1.06)3 = 1400(1.06)t , and
the exact solution is t = 2.267.
Example 33. A borrower is repaying a loan by making payments of 1000 at the
end of each of the next 3 years. The interest rate on the loan is 5% compounded
annually. What payment could the borrower make at the end of the first year in
order to extinguish the loan? If the unknown payment amount at the end of the year
is P, the equation of value obtained by computing the present value of all payments
at the end of this year is P = 1000 + 1000v + 1000v2 , where v = 1/ 1.05. Computation
gives P = 2859.41 as the payment amount. Notice that the same solution is obtained
using any time point for comparison. The choice of time point as the end of the first
year was made to reduce the amount of computation.
3: Cash Flow Valuation
12
Problems
Problem 31. What rate of interest compounded quarterly is required for a deposit
of 5000 today to accumulate to 10,000 after 10 years?
Problem 32. An investor purchases an investment which will pay 2000 at the end
of one year and 5000 at the end of four years. The investor pays 1000 now and
agrees to pay X at the end of the third year. If the investor uses an interest rate of
7% compounded annually, what is X?
Problem 33. A loan requires the borrower to repay 1000 after 1 year, 2000 after 2
years, 3000 after 3 years, and 4000 after 4 years. At what time could the borrower
make a single payment of 10000 to repay the loan? Assume the interest rate is 4%
effective.
Problem 34. A note that pays 10,000 3 months from now is purchased by an
investor for 9500. What is the effective annual rate of interest earned by the
investor?
3: Cash Flow Valuation

Problem 31. The equation of value is 5000(1 + i/ 4)40 = 10000, from which
i = 0.0699.
Problem 32. The equation of value today is 2000v + 5000v4 = 1000 + Xv3
where v = 1/ 1.07. Thus X = 5773.16.
Problem 33. The exact time t is the solution of 1000v + 2000v2 + 3000v3 +
4000v4 = 10000vt , where v = 1/ 1.04. Thus t = 2.98 years.
Problem 34. The equation involving the annual rate of interest i is 9500(1 +
i)1/ 4 = 10000, from which i = 0.2277.
13
4. Sample Question Set 1

Solve the following 6 problems in no more than 30 minutes.
0.05
Question 41 . Fund A accumulates at a force of interest
at time t (t 0).
1 + 0.05t
Fund B accumulates at a force of interest 0.05. You are given that the amount in
Fund A at time zero is 1,000, the amount in Fund B at time zero is 500, and that the
amount in Fund C at any time t is equal to the sum of the amount in Fund A and
Fund B. Fund C accumulates at force of interest t . Find 2 .
31
A.
660
B.
21
440
D.
2 + e0.1
44 + 20e0.1
C.
1 + e0.1
22 + 20e0.1
E.
2 + e0.1
22 + 20e0.1
Question 42 . Gertrude deposits 10,000 in a bank. During the first year the bank
credits an annual effective rate of interest i. During the second year the bank credits
an annual effective rate of interest (i5%). At the end of two years she has 12,093.75
in the bank. What would Gertrude have in the bank at the end of three years if the
annual effective rate of interest were (i + 9%) for each of the three years?
A. 16,851
B.
17,196
D.
17,936
C.
17,499
E.
18,113
Question 43 . Fund X starts with 1,000 and accumulates with a force of interest
1
t =
for 0 t < 15. Fund Y starts with 1,000 and accumulates with an
15 t
interest rate of 8% per annum compounded semi-annually for the first three years
and an effective interest rate of i per annum thereafter. Fund X equals Fund Y at the
end of four years. Calculate i.
A. 0.0750
B.
0.0775
D.
0.0825
C.
0.0800
E.
0.0850
4: Sample Question Set 1
15
Question 44 . Jeff puts 100 into a fund that pays an effective annual rate of discount
2t
of 20% for the first two years and a force of interest of rate t = 2
, 2 t 4,
t +8
for the next two years. At the end of four years, the amount in Jeffs account is the
same as what it would have been if he had put 100 into an account paying interest at
the nominal rate of i per annum compounded quarterly for four years. Calculate i.
A. 0.200
B.
0.219
D.
0.285
C.
0.240
E.
0.295
Question 45 . On January 1, 1980, Jack deposited 1,000 into Bank X to earn

interest at the rate of j per annum compounded semi-annually. On January 1, 1985,
he transferred his account to Bank Y to earn interest at the rate of k per annum
compounded quarterly. On January 1, 1988, the balance at Bank Y is 1,990.76. If
Jack could have earned interest at the rate of k per annum compounded quarterly from
January 1, 1980 through January 1, 1988, his balance would have been 2,203.76.
Calculate the ratio k/ j.
A. 1.25
B.
1.30
D.
1.40
C.
1.35
E.
1.45
Question 46 . You are given two loans, with each loan to be repaid by a single
payment in the future. Each payment includes both principal and interest. The
first loan is repaid by a 3,000 payment at the end of four years. The interest is
accrued at 10% per annum compounded semi-annually. The second loan is repaid
by a 4,000 payment at the end of five years. The interest is accrued at 8% per
annum compounded semi-annually. These two loans are to be consolidated. The
consolidated loan is to be repaid by two equal installments of X, with interest at
12% per annum compounded semi-annually. The first payment is due immediately
and the second payment is due one year from now. Calculate X.
A. 2,459
B.
2,485
D.
2,521
C.
2,504
E.
2,537
16
Answers to Sample Questions

Rt
0.05
Question 41 . The amount in fund A at time t is A(t) = 1000e 0 1+0.05s ds = 1000+50t,

the amount in fund B at time t is B(t) = 500e0.05t and the amount in fund C at time
t is C(t) = A(t) + B(t). So 2 = C (2)/ C(2) = (A (2) + B (2))/ (A(2) + B(2)) =
(50 + 25e0.1 )/ (1100 + 500e0.1 ) = (2 + e0.1 )/ (44 + 20e0.1 ). D.
Question 42 . From the information given, 10000(1 + i)(1 + i 0.05) = 12093.75,
from which i = 0.125. Thus 10000(1 + i + 0.09)3 = 17, 936.13. D.
Question 43 . After 4 years the amount in fund X is 15000/ (15 4) = 15000/ 11
and the amount in fund Y is 1000(1.04)6 (1 + i). Equating these two gives i = 0.0777.
B.
2
R4
Question 44 . The amount Jeff actually has is 100(1 0.20) e 2 t dt = 312.50,

while what he would have under the other option is 100(1 + i/ 4)16 . Equating and
solving gives i = 0.2952.E.
Question 45 . The given information gives two equations. First, 1000(1+j/ 2)10 (1+
k/ 4)12 = 1990.76 and second 1000(1 + k/ 4)32 = 2203.76. The second gives k = 0.10,
and using this in the first gives j = 0.0799. Thus k/ j = 1.25. A .
Question 46 . From the information given, 3000(1.05)8 + 4000(1.04)10 = X +
X(1.06)2 , from which X = 2504.12. C .
5. Annuities, Amortization, and Sinking Funds

Many different types of financial transactions involve the payment of a fixed
amount of money at regularly spaced intervals of time for a predetermined period.
Such a sequence of payments is called an annuity certain or, more simply, an
annuity. A common example is loan payments. The concept of present value is
easily used to evaluate the worth of such a cash stream at any point in time. Here is
an example.
Example 51. Suppose you have the opportunity to buy an annuity, that is, for
a certain amount A > 0 paid by you today you will receive monthly payments of
$400, say, for the next 20 years. How much is this annuity worth to you? Suppose
that the payments are to begin one month from today. Such an annuity is called an
annuity immediate (a truly unfortunate choice of terminology). The cash stream
represented by the annuity can be visualized on a time diagram.
An Annuity Immediate
$400
...
$400
...
$400
$400
0
1
2
239
240
Clearly you would be willing to pay today no more than the present value of the
total payments made by the annuity. Assume that you are able to earn 5% interest
(nominal annual rate) compounded monthly. The present value of the payments is
240
X
(1 +
j=1
.05 j
) 400.
12
This sum is simply the sum of the present value of each of the payments using the
indicated interest rate. This sum is easily found since it involves a very simple
geometric series.
Exercise 51. Evaluate the sum.
Since expressions of this sort occur rather often, actuaries have developed some
special notation for this sum. Write an for the present value of an annuity which
pays $1 at the end of each period for n periods.
The Standard Annuity Immediate
0
........................................................................................................................................................................................................................................................................................................................................................................................
n1
...
Then
an =
n
X
j=1
vj =
1 vn
i
n+1
5: Annuities, Amortization, and Sinking Funds
18
where the last equality follows from the summation formula for a geometric series.
The interest rate per period is usually not included in this notation, but when such
information is necessary the notation is an i . The present value of the annuity in the
previous example may thus be expressed as 400a240 .05/ 12 .
A slightly different annuity is the annuity due which is an annuity in which the
payments are made starting immediately. The notation a n denotes the present value
of an annuity which pays $1 at the beginning of each period for n periods.
The Standard Annuity Due
1
........................................................................................................................................................................................................................................................................................................................................................................................
n1
...
n+1
Clearly
a n =
n1
X
j=0
vj =
1 vn
d
where again the last equality follows by summing the geometric series. Note that n
still refers to the number of payments. If the present time is denoted by time 0, then
for an annuity immediate the last payment is made at time n, while for an annuity
due the last payment is made at time n 1, that is, the beginning of the nth period.
Evidently, an = v a n , and there are many other similar relationships.
Exercise 52. Show that an = v a n .
The connection between an annuity due and an annuity immediate can be viewed
in the following way. In an annuity due the payment for the period is made at the
beginning of the period, whereas for an annuity immediate the payment for the
period is made at the end of the period. Clearly a payment of 1 at the end of the
period is equivalent to the payment of v = 1/ (1 + i) at the beginning of the period.
This gives an intuitive description of the equality of the previous exercise.
Annuity payments need not all be equal. Here are a couple of important special
modifications.
Example 52. An increasing annuity immediate with a term of n periods pays
1 at the end of the first period, 2 at the end of the second period, 3 at the end of
the third period, . . . , n at the end of the nth period. What is (Ia)n , the present
P
value of such an annuity? From the definition, (Ia)n = nj=1 jvj . Although this
is not a geometric series, the same technique can be used. This procedure gives
(Ia)n v(Ia)n = v + v2 + . . . + vn nvn+1 which gives (Ia)n = (an nvn+1 )/ (1 v) =
(an nvn )/ i.
Exercise 53. A decreasing annuity immediate with a term of n periods pays n
at the end of the first period, n 1 at the end of the second period, n 2 at the end
19
of the third period, . . . , 1 at the end of the nth period. Find (Da)n , the present value
of such an annuity.
Exercise 54. An annuity immediate with 2n 1 payments pays 1 at the end of the
first period, 2 at the end of the second, . . . , n at the end of the nth period, n 1 at the
end of the n + 1st period, . . . , 1 at the end of the 2n 1st period. What is the present
value of this annuity?
Example 53. A deferred annuity is an annuity in which the payments start at
some future time. A standard deferred annuity immediate in which payments are
deferred for k periods, has the first payment of 1 made at time k + 1, that is, at the end
of year k + 1. Notice that from the perspective of a person standing at time k, this
deferred annuity immediate looks like a standard n period annuity immediate. The
present value of a k year deferred, n year annuity immediate is denoted k| an . The
present value of the deferred annuity at time k is an . Bringing this to an equivalent
value at time 0 gives k| an = vk an . A time diagram shows that the deferred payments
can be obtained by paying back payments that are received during the first k periods.
Thus k| an = an+k ak .
Exercise 55. What is k| a n ?
Theoretically, an annuity could be paid continuously, that is, the annuitant
receives money at a constant rate of 1 dollar per unit time. The present value of such
an annuity that pays 1 per unit time for n time periods is denoted by an . Now the
value at time 0 of such a continuously paid annuity can be computed as follows. The
value of the dt
in the time interval from t to t + dt is vt dt = e t dt.
Z ndollars that arrive
n
1
v
e t dt =
.
Hence an =
0
Annuity payments can be made either more or less often than interest is compounded. In such cases, the equivalent rate of interest can be used to most easily
compute the value of the annuity.
Example 54. The symbol a(m)
n i denotes the present value of an annuity immediate
that pays 1/ m at the end of each mth part of a period for n periods under the
assumption that the effective interest rate is i per period. For example, if m = 12
and the period is a year, payments of 1/12 are made at the end of each month. What
is a formula for a(m)
n assuming the effective rate of interest is i per period? Notice
here that the payments are made more frequently than interest is compounded.
Using the equivalent rate i(m) makes the computations easy. Using geometric series,
1 Pnm
(m)
j
n
(m)
a(m)
= ian / i(m) .
n = m
j=1 (1 + i / m) = (1 v )/ i
Exercise 56. The symbol a (m)
n i denotes the present value of an annuity due that pays
1/ m at the beginning of each mth part of a period for n periods when the effective
20
periodic interest rate is i. Find a formula for a (m)

n assuming the effective periodic
rate of interest is i.
Exercise 57. The symbol (Ia)(m)
n is the present value of an annuity that pays 1/ m
at the end of each mth part of the first period, 2/ m at the end of each mth part of
the second period, . . . , n/ m at the end of each mth part of the nth period when the
effective annual interest rate is i. Find a computational formula for (Ia)(m)
n .
Thus far the value of an annuity has been computed at time 0. Another common
time point at which the value of an annuity consisting of n payments of 1 is computed
is time n. Denote by sn the value of an annuity immediate at time n, that is,
immediately after the nth payment. Then sn = (1 + i)n an from the time diagram.
The value sn is called the accumulated value of the annuity immediate. Similarly
sn is the accumulated value of an annuity due at time n and sn = (1 + i)n a n .
(m)
Exercise 58. Similarly, s(m)
n , and sn are the values of the corresponding annun ,s
ities just after time n. Find a formula for each of these in terms of sn .
Exercise 59. What do the symbols (Is)n and (Is)n represent?
Now a common use of annuities will be examined.
Example 55. You are going to buy a house for which the purchase price is $100,000
and the downpayment is $20,000. You will finance the $80,000 by borrowing this
amount from a bank at 10% interest with a 30 year term. What is your monthly
payment? Typically such a loan is amortized, that is, you will make equal monthly
payments for the life of the loan and each payment consists partially of interest and
partially of principal. From the banks point of view this transaction represents the
purchase by the bank of an annuity immediate. The monthly payment, p, is thus
the solution of the equation 80000 = pa360 0.10/ 12 . In this setting the quoted interest
rate on the loan is assumed to be compounded at the same frequency as the payment
period unless stated otherwise.
Exercise 510. Find the monthly payment. What is the total amount of the payments made?
An amortization table is a table which lists the principal and interest portions
of each payment for a loan which is being amortized. An amortization table can be
constructed from first principles. Denote by bk the loan balance immediately after
the kth payment and write b0 for the original loan amount. Then the interest part
of the kth payment is ibk1 and the principal amount of the kth payment is P ibk1
where P is the periodic payment amount. Notice too that bk+1 = (1 + i)bk P. These
relations allow the rows of the amortization table to be constructed sequentially.
21
Taking a more sophisticated viewpoint will exhibit a method of constructing

any single row of the amortization table that is desired, without constructing the
whole table. In the prospective method the loan balance at any point in time
is seen to be the present value of the remaining loan payments. The prospective
method gives the loan balance immediately after the kth payment as bk = Pank .
In the retrospective method the loan balance at any point in time is seen to be
the accumulated original loan amount less the accumulated value of the past loan
payments. The retrospective method gives the loan balance immediately after the
kth payment as bk = b0 (1 + i)k Psk . Either method can be used to find the loan
balance at an arbitrary time point. With this information, any given row of the
amortization table can be constructed.
Exercise 511. Show that for the retrospective method, b0 (1 + i)k Psk = b0 + (ib0
P)sk .
Exercise 512. Show that the prospective and retrospective methods give the same
value.
A further bit of insight is obtained by examining the case in which the loan
amount is an , so that each loan payment is 1. In this case the interest part of the kth
payment is iank+1 = 1 vnk+1 and the principal part of the kth payment is vnk+1 .
This shows that the principal payments form a geometric series.
Finally observe that the prospective and retrospective method apply to any series
of loan payments. The formulas obtained when the payments are not all equal will
just be messier.
A second way of paying off a loan is by means of a sinking fund.
Example 56. As in the previous example, $80,000 is borrowed at 10% annual
interest. But this time, only the interest is required to be paid each month. The
principal amount is to be repaid in full at the end of 30 years. Of course, the borrower
wants to accumulate a separate fund, called a sinking fund, which will accumulate to
$80,000 in 30 years. The borrower can only earn 5% interest compounded monthly.
In this scenario, the monthly interest payment is 80000(0.10/ 12) = 666.67. The
contribution c each month into the sinking fund must satisfy cs360 0.05/ 12 = 80000,
from which c = 96.12. As expected, the combined payment is higher, since the
interest rate earned on the sinking fund is lower than 10%.
Here is a summary of the most useful formulas thus far.
an
1 vn
i
sn
vn 1
i
vn an

vn 1
d
1 vn
d
(Ia)n =
a n nvn
i
(Is)n =
sn n
= vn (Ia)n
i
n an
i
(Ds)n =
nvn sn
= vn (Da)n
i
a n
(Da)n =
van
sn
22
vn a n
an
The reader should have firmly in mind the time diagram for each of the basic
annuities, as well as these computational formulas.
23
Problems
Problem 51. Show that an < an < a n . Hint: This should be obvious from the
picture.
Problem 52. True or False: For any two interest rates i and i , (1 + i)n (1 + i sn i ) =
1 + (i i)an i .
Problem 53. True or False:
1
1
=
+ i.
an sn
Problem 54. True or False:
a (m)
n
i
i(m)
i
+
an .
m
Problem 55. Show that a2n = an (1 + vn ).

Problem 56. Show that a3n = an + vn a2n = a2n + v2n an .
Problem 57. Suppose an annuity immediate pays p at the end of the first period,
pr at the end of the second period, pr2 at the end of the third period, and so on, until
a final payment of prn1 is made at the end of the nth period. What is the present
value of this annuity?
Problem 58. John borrows $1,000 from Jane at an annual effective rate of interest
i. He agrees to pay back $1,000 after six years and $1,366.87 after another 6 years.
Three years after his first payment, John repays the outstanding balance. What is
the amount of Johns second payment?
Problem 59. A loan of 10,000 carries an interest rate of 9% compounded quarterly.
Equal loan payments are to be made monthly for 36 months. What is the size of
each payment?
Problem 510. An annuity immediate pays an initial benefit of one per year, increasing by 10.25% every four years. The annuity is payable for 40 years. If the
effective interest rate is 5% find an expression for the present value of this annuity.
Problem 511. Humphrey purchases a home with a $100,000 mortgage. Mortgage
payments are to be made monthly for 30 years, with the first payment to be made one
month from now. The rate of interest is 10%. After 10 years, Humphrey increases
the amount of each monthly payment by $325 in order to repay the mortgage more
quickly. What amount of interest is paid over the life of the loan?
Problem 512. On January 1, an insurance company has $100,000 which is due to
Linden as a life insurance death benefit. He chooses to receive the benefit annually
over a period of 15 years, with the first payment made immediately. The benefit
24
he receives is based on an effective interest rate of 4% per annum. The insurance

company earns interest at an effective rate of 5% per annum. Every July 1 the
company pays $100 in expenses and taxes to maintain the policy. How much money
does the company have remaining after 9 years?
Problem 513. A loan of 10,000 is to be repaid with equal monthly payments of p.
The interest rate for the first year is 1.9%, while the interest rate for the remaining
2 years is 10.9%. What is p? What is the balance after the 6th payment? After the
15th payment? What are the principal and interest components of the 7th payment?
Of the 16th payment?
Problem 514. A loan of 10,000 is to be repaid as follows. Payments of p are to
be made at the end of each month for 36 months and a balloon payment of 2500 is
to be made at the end of the 36th month as well. If the interest rate is 5%, what is
p? What is the loan balance at the end of the 12th month? What part of the 13th
payment is interest? Principal?
Problem 515. A loan is being amortized with a series of 20 annual payments. If
the amount of principal in the third payment is 200, what is the amount of principal
in the last 3 payments? The interest rate is 4%.
Problem 516. A loan is amortized with 10 annual installments. The principal part
of the fifth payment is 20 and the interest part is 5. What is the rate of interest on
the loan?
Problem 517. A loan is amortized with payments of 1 at the end of each year for
20 years. Along with the fifth payment the borrower sends the amount of principal
which would have been paid with the sixth payment. At the time of the sixth
payment, the borrower resumes payment of 1 until the loan is repaid. How much
interest is saved by this single modified payment?
Problem 518. A loan of 1000 is being repaid by equal annual installments of 100
together with a smaller final payment at the end of 10 years. If the interest rate is
4%, show that the balance immediately after the fifth payment is 1000 60s5 .04 .
Problem 519. A loan of 1200 is to be repaid over 20 years. The borrower is to
make annual payments of 100 at the end of each year. The lender receives 5% on
the loan for the first 10 years and 6% on the loan balance for the remaining years.
After accounting for the interest to be paid, the remainder of the payment of 100 is
deposited in a sinking fund earning 3%. What is the loan balance still due at the
end of 20 years?
Problem 520. A loan is being repaid with 10 payments. The first payment is 10,
the second payment is 9, and so on. What is the amount of interest in the fourth
25
payment?
Problem 521. A standard perpetuity immediate is an annuity which pays 1 at the
end of each period forever. What is a , the present value of a standard perpetuity
immediate? A standard perpetuity due pays 1 at the beginning of each period
forever. What is a , the present value of a standard perpetuity due?
Problem 522. A standard perpetuity due has a present value of 20, and will be
exchanged for a perpetuity immediate which pays R per period. What is the value
of R that makes these two perpetuities of equal value?
Problem 523. You are given an annuity immediate paying $10 for 10 years, then
decreasing by $1 per year for nine years and paying $1 per year thereafter, forever.
If the annual effective rate of interest is 5%, find the present value of this annuity.
Problem 524. A loan is being repaid with a payment of 200 at the end of the first
year, 190 at the end of the second year, and so on, until the final payment of 110
at the end of the tenth year. If the interest rate is 6%, what was the original loan
amount?
Problem 525. A loan is being repaid with a payment of 200 at the end of the first
year, 190 at the end of the second year, and so on, until the final payment of 110 at
the end of the tenth year. The borrower pays interest on the original loan amount
at a rate of 7%, and contributes the balance of each payment to a sinking fund that
earns 4%. If the amount in the sinking fund at the end of the tenth year is equal to
the orginal loan amount, what was the original loan amount?
Problem 526. A loan of 1 was to be repaid with 25 equal payments at the end of
the year. An extra payment of K was made in addition to the sixth through tenth
payments, and these extra payments enabled the loan to be repaid five years early.
Show that K = (a20 a15 )/ a25 a5 .
Problem 527. A loan is being repaid with quarterly payments of 500 at the end of
each quarter for seven years at an interest rate of 8% compounded quarterly. What
is the amount of principal in the fifth payment?
Problem 528. A borrower is repaying a loan with 20 annual payments of 500
made at the end of each year. Half of the loan is repaid by the amortization method
at 6% effective. The other half of the loan is repaid by the sinking fund method
in which the interest rate is 6% effective and the sinking fund accumulates at 5%
effective. What is the amount of the loan?
Problem 529. A loan of 18000 is made for 12 years in which the lender receives 6%
compounded semiannually for the first six years and 4% compounded semiannually
26
for the last six years. The borrower makes seminannual payments of 1000, and
the balance after paying interest is deposited into a sinking fund which pays 3%
compounded semiannually. What is the net amount remaining on the loan after 12
years?
Problem 530. The interest on an inheritance invested at 4% effective would have
been just sufficient to pay 16000 at the end of each year for 15 years. Payments
were made as planned for the first five years, even though the actual interest earned
on the inheritance for years 3 through 5 was 6% effective instead of 4% effective.
How much excess interest had accumulated at the end of the fifth year?

Problem 51. Isnt the chain of inequalities simply expressing the fact that
getting a given amount of money sooner makes it worth more? An analytic proof
should be easy to give too.
Problem 52. True, since vn (1 + i sn i ) = vn + i an i = 1 ian i + i an i .
Problem 53. True. Write 1/ (1 vn ) = vn / (vn 1) = (vn 1 + 1)/ (vn 1) =
1 + 1/ (vn 1), multiply by i and use the definitions. This also has a verbal
explanation. 1/ an is the periodic payment to amortize a loan of 1 with n
payments. The loan can also be paid off by paying i per period and contributing
(m)
1/ sn to a sinking fund. Similar reasoning shows that 1/ a(m)
+ 1/ s(m)
n and
n = i
(m)
(m)
(m)
1/ a n = d + 1/ sn .
1/ m (m)
an = (1 + i(m) / m)(i/ i(m) )an .
Problem 54. True, since a (m)
n = (1 + i)
Problem 55. Breaking the period of length 2n into two periods of length n
gives a2n = an + vn an , and the result follows.
Problem 56. Just break the period of length 3n into two pieces, one of length
n and the other of length 2n.
n
X
Problem 57.
Direct computation gives the present value as
prj1 vj =
j=1
(pv pvn+1 rn )/ (1 vr) = p(1 vn rn )/ (1 + i r), provided vr 1.

Problem 58. From Janes point of view the equation 1000 = 1000(1 + i)6 +
1366.87(1 + i)12 must hold. The outstanding balance at the indicated time is
1366.87(1 + i)3 , which is the amount of the second payment.
Problem 59. An interest rate of 9% compounded quarterly is equivalent to an
interest rate of 8.933% compounded monthly. The monthly payment is therefore
10000/ a36 .0893/ 12 = 10000/ 31.47 = 317.69.
Problem 510. Each 4 year chunk is a simple annuity immediate. Taking the
present value of these chunks forms an annuity due with payments every 4 years
that are increasing.
Problem 511. The initial monthly payment P is the solution of 100, 000 =
Pa360 . The balance after 10 years is Pa240 so the interest paid in the first 10
years is 120P (100, 000 Pa240 ). To determine the number of new monthly
payments required to repay the loan the equation Pa240 = (P + 325)ax should be
solved for x. Since after x payments the loan balance is 0 the amount of interest
paid in the second stage can then be easily determined.
Problem 512. Since the effective rate of interest for the insurance company
is 5%, the factor (1.05)1/ 2 should be used to move the insurance companys
expenses from July 1 to January 1.
Problem 513.
The present value of the monthly payments must equal the
27

original loan amount. Thus 10000 = pa12 0.019/ 12 + (1 + 0.019/ 12)12 pa24 0.109/ 12 ,
from which p = 303.49. The balance after the sixth payment is most easily
found by the retrospective method. The balance is 10000(1 + 0.019/ 12)6
303.49s6 0.019/ 12 = 8267.24. The balance after the 15th payment is most easily
found by the prospective method. The balance is 303.49a21 0.109/ 12 = 5778.44.
The interest portion of the 7th payment is (0.0109/ 12)(8267.24) = 13.09 and
the principal portion is 303.49 13.09 = 290.40. Similar computations give the
interest portion of the 16th payment as 52.49 and the principal portion as 251.00.
Problem 514. The payment p = 235.20 since 10000 = pa36 .05/ 12 + (1 +
.05/ 12)36 2500. Using the retrospective method, the loan balance at the end of
the 12th month is 10000(1 + .05/ 12)12 ps12 .05/ 12 = 7623.65. The interest part
of the 13th payment is 31.77.
Problem 515. If p is the annual payment amount, the principal part of the
third payment is pv18 = 200. Thus p = 405.16, and the loan balance after the
seventeenth payment is pa3 0.04 = 1124.36.
Problem 516. Let p denote the annual payment amount. Then pv6 = 20 and
p(1 v6 ) = 5. Adding these two equations gives p = 25, so v6 = 0.8, from which
i = 0.0379.
Problem 517. The borrower saves the interest on the sixth payment, which is
1 v206+1 = 1 v15 .
Problem 518. The retrospective method gives the balance as 1000v5
100s5 .04 , which re-arranges to the stated quantity using the identity vn = 1 + isn .
Problem 519. The amount in the sinking fund at the end of 20 years is
40s10 (1.03)10 + 28s10 = 937.25, so the loan balance is 1200 937.25 = 262.75.
Problem 520. The loan balance immediately after the third payment is
(Da)7 = (7 a7 )/ i, so the interest paid with the fourth payment is 7 a7 .
Problem 521. Summing the geometric series gives a = 1/ i and a = 1/ d.
These results can also be obtained by letting n in the formulas for an and
a n .
Problem 522. From the information given, 1/ d = 20 (and so d = 1/ 20) and
Rv/ d = 20. Since v = 1 d = 19/ 20, R = 20/ 19.
Problem 523. What is the present value of an annuity immediate paying
$1 per year forever? What is the present value of such an annuity that begins
payments k years from now? The annuity described here is the difference of a
few of these.
Problem 524. The payments consist of a level payment of 100 together with
a decreasing annuity. The present value of these payments at the date of issue of
the loan is the loan amount, which is therefore 100a10 .06 +10(Da)10 .06 = 1175.99.
Problem 525. Let A be the original loan amount. The contribution to the
sinking fund at the end of year j is then 100 + 10(11 j) .07A, and these
28

contributions must accumulate to A. Thus 100s10 .04 + 10(Ds)10 .04 .07As10 .04 =
A, from which A = 1104.24.
Problem 526. Denote the original payment amount by p. Then p = 1/ a25 .
The fact that the additional payments extinguish the loan after 20 years means
1 = pa20 + v5 Ka5 . Thus K = (a25 a20 )/ v5 a5 a25 . The result follows since
v5 (a25 a20 ) = a20 a15 .
Problem 527. Using i = 0.02, the loan balance immediately after the fourth
payment is 500a24 , so the principal part of the fifth payment is 500 10a24 =
310.86.
Problem 528. Denote by L the loan amount. From the information given,
(500 L/ (2a20 0.06 ) 0.03L)s20 0.05 = L/ 2, so that L = 500/ (1/ 2s20 0.05 + 0.03 +
1/ 2a20 0.06 ) = 5636.12.
Problem 529. The semiannual contributions to the sinking fund are 460 for
the first six years and 640 for the last six years. The sinking fund balance at the
end of 12 years is (1.015)12 460s12 0.015 + 640s12 0.015 = 15518.83. So the net loan
balance is 18000 15518.83 = 2481.17.
Problem 530. The accumulated interest is (24000 16000)s3 .06 = 25468.80.
29

Exercise
51. The sum is the sum of the terms of a geometric series. So
P240
.05 j
1
241
)/ (1(1+0.05/ 12)1 ) =
j=1 (1+ 12 ) 400 = 400((1+0.05/ 12) (1+0.05/ 12)
60, 610.12.
Exercise 52. This follows from the formulas for the present value of the two
annuities and the fact that d = iv.
Exercise 53. The method of the example could be used again, but the formula
can also be obtained from (Ia)n + (Da)n = (n + 1)an , which gives (Da)n =
(n an )/ i.
Exercise 54. The present value is (Ia)n + vn (Da)n1 = (1 + an1 vn
vn an1 )/ i = (1 vn )(1 + an1 )/ i = a n an .
Exercise 55. This is the present value of an annuity with payments of 1 which
start at the beginning of period k + 1 (that is, at time k) and continue for a total
of n payments. Thus k| a n = vk a n = a k+n a k .
n
(m)
= da n / d(m) .
Exercise 56. As in the previous example, a (m)
n = (1 v )/ d
Exercise 57. Proceeding as in the derivation of the formula for (Ia)n gives
a n nvn
(Ia)(m)
.
n =
i(m)
Exercise 58. Direct computation using the parallel facts for the values at time
(m) (m)
0 give s(m)
n = isn / d(m) , and sn = isn / .
n = isn / i , s
Exercise 59. The symbol (Is)n is the value of an increasing annuity immediate
computed at time n; (Is)n is the value of an increasing annuity due at time n.
Exercise 510. Using the earlier formula gives a360 0.10/ 12 = 113.95 from
which p = 702.06 and the total amount of the payments is 360p = 252740.60.
Exercise 511. Simply use the identity vn = 1 + isn .
Exercise 512. Direct computation using the formulas gives ank = vk (an ak )
and P = b0 / an gives Pank = b0 vk (an ak )/ an = b0 (1 + i)k Psk .
30

Question 61 . The present value of a series of payments of 2 at the end of every
eight years, forever, is equal to 5. Calculate the effective rate of interest.
A. 0.023
B.
0.033
D.
0.043
C.
0.040
E.
0.052
Question 62 . An annuity immediate pays an initial benefit of one per year,

increasing by 10.25% every four years. The annuity is payable for 40 years. Using
an annual effective interest rate of 5%, determine an expression for the present value
of this annuity.
A. (1 + v2 )a20
a20
D.
2
s2
B. (1 + v )a20
C.
2 a20
Question 63 . Determine an expression for
E.
a40
a2
a5
.
a6
A.
a2 + a3
2 a3
B.
a2 + s3
1 + a3 + s2
D.
1 + a2 + s2
a3 + s3
C.
a2 + s3
a3 + s3
E.
1 + a2 + s2
1 + a3 + s2
32
Question 64 . Which of the following are true?

I. (an d )(1 + i) = an1
II. The present value of a 10 year temporary annuity immediate paying 10 per month
for the first eight months of each year is 120a10 a(12)
.
8/ 12
III. The present value of a perpetuity paying one at the end of each year, except
s
paying nothing every fourth year, is 3 .
i s4
A.
I and II only
B.
I and III only
D.
C.
II and III only
I, II, and III
E. The correct answer is not given

by A, B, C, or D
Question 65 . Warren has a loan with an effective interest rate of 5% per annum.
He makes payments at the end of each year for 10 years. The first payment is
200, and each subsequent payment increases by 10 per year. Calculate the interest
portion in the fifth payment.
A. 58
B.
60
D.
65
C.
62
E.
67
Question 66 . An investment fund accrues interest with force of interest t =

K
for 0 t 1. At time zero, there is 100,000 in the fund. At time one
1 + (1 t)K
there is 110,000 in the fund. The only two transactions during the year are a deposit
of 15,000 at time 0.25 and a withdrawal of 20,000 at time 0.75. Calculate K.
A. 0.047
B.
0.051
D.
0.150
C.
0.141
E.
0.154
33
Question 67 . You are given an annuity immediate with 11 annual payments of

100 and a final balloon payment at the end of 12 years. At an annual effective
interest rate of 3.5%, the present value at time 0 of all the payments is 1,000. Using
an annual effective interest rate of 1%, calculate the present value at the beginning
of the ninth year of all remaining payments.
A. 412
B.
419
D.
439
C.
432
E.
446
Question 68 . Using an annual effective interest rate j > 0, you are given
(1) The present value of 2 at the end of each year for 2n years, plus an additional
1 at the end of each of the first n years, is 36
(2) The present value of an n-year deferred annuity immediate paying 2 per year
for n years is 6
Calculate j.
A. 0.03
B.
0.04
D.
0.06
C.
0.05
E.
0.07
Question 69 . An 11 year annuity has a series of payments 1, 2, 3, 4, 5, 6, 5, 4, 3,

2, 1, with the first payment made at the end of the second year. The present value
of this annuity is 25 at interest rate i. A 12 year annuity has a series of payments 1,
2, 3, 4, 5, 6, 6, 5, 4, 3, 2, 1, with the first payment made at the end of the first year.
Calculate the present value of the 12 year annuity at interest rate i.
A. 29.5
B.
30.0
D.
31.0
C.
30.5
E.
31.5
34
Question 610 . Joan has won a lottery that pays 1,000 per month in the first year,
1,100 per month in the second year, 1,200 per month in the third year, etc. Payments
are made at the end of each month for 10 years. Using an effective interest rate of
3% per annum, calculate the present value of this prize.
A. 107,000
B.
114,000
D.
135,000
C.
123,000
E.
148,000
Question 611 . A 5% 10 year loan of 10,000 is to be repaid by the sinking fund

method, with interest and sinking fund payments made at the end of each year. The
effective rate of interest earned in the sinking fund is 3% per annum. Immediately
before the fifth years payment would have fallen due, the lender requests that the
outstanding principal be repaid in one lump sum. Calculate the amount that must
be paid, including interest, to extinguish the debt.
A. 6,350
B.
6,460
D.
6,850
C.
6,740
E.
7,000
Question 612 . A company agrees to repay a loan over five years. Interest
payments are made annually and a sinking fund is built up with five equal annual
payments made at the end of each year. Interest on the sinking fund is compounded
annually. You are given
(1) The amount in the sinking fund immediately after the first payment is X
(2) The amount in the sinking fund immediately after the second payment is Y
(3) Y/ X = 2.09
(4) The net amount of the loan immediately after the fourth payment is 3,007.87
Calculate the amount of the sinking fund payment.
A. 1,931
B.
2,031
D.
2,231
C.
2,131
E.
2,431
35

P
Question 61 . The information given is that 2 j=1 v8j = 5. Using geometric series
gives this equation as v8 / (1 v8 ) = 5/ 2, from which v = (5/ 7)1/ 8 and i = 0.0429. D .
Question 62 . Note that (1.05)2 = 1.1025, so that the quadrennial increase is v2 .
Writing out directly the present value of the payments gives (v + v2 + v3 + v4 ) + (v3 +
v4 + v5 + v6 ) + (v5 + v6 + v7 + v8 ) + . . . + (v19 + v20 + v21 + v22 ) which re-arranges to
a20 + v2 a20 , or B.
Question 63 . Write a5 = v + v2 + v3 + v4 + v5 , and to likewise for a6 . Multiply top
and bottom by v3 to get answer C.
Question 64 . That I is true follows from (an d/ )(1 + i) = (1 d en )e / =
(ve e(n1) )/ = an1 . II is false, since the middle factor should be a 10 . III is also
false. The correct value is s3 / ds4 . E.
Question 65 . The loan balance at the beginning of the fifth year is 240v + 250v2 +
. . . + 290v6 = 1337.84, where v = 1/ (1.05). The interest for the fifth year is therefore
0.05(1337.84) = 66.89. E.
R1
R1
R1
Question 66 . The equation of value is 100e 0 t dt + 15e 1/ 4 t dt 20e 3/ 4 t dt = 110.

Simplifying gives (425/ 4)K + 95 = 110 from which K = 12/ 85 = 0.1411. C.
Question 67 . From the given information, 1000 = 100a11 .035 +(1.035)12 B, where
B is the amount of the balloon payment. Thus B = 150.87. The present value at the
beginning of the ninth year is 100a3 .01 + (1.01)4 B = 439.08. D.
Question 68 . Looking at the first annuity as paying 3 for 2n years and taking
back 1 in each of the last n years shows that 36 = 3a2n 6/ 2, from which a2n = 13.
Similarly, 6 = 2a2n 2an , so an = 10. But 6 = 2vn an also, so that vn = 6/ 20 Finally,
10 = an = (1 vn )/ i = (1 (6/ 20))/ i, from which i = 7/ 100. E.
Question 69 . Write T for the present value of the twelve year annuity and
E for the present value of the eleven year annuity. Looking at a time diagram
shows that T E = a6 , and direct computation gives E = va6 a6 = (a6 )2 . Thus
T = E + a6 = 25 + 5 = 30. B.
Question 610 . By breaking off the increasing part from a constant payment of
1000 per month for the life of the payments, the present value is 1000s12 (.03)(12) a10 .03 +
(1.03)1 100s12 (.03)(12) (Ia)9 = 147928.85. E.
Question 611 . The annual sinking fund contribution is 10000/ s10 .03 = 872.31.
The amount in the sinking fund just before the fifth payment is 872.31s4 .03 (1.03) =
36
3758.88. The amount due is 10000 3758.88 + 500 = 6741.12. C.

Question 612 . The sinking fund payment is X. Let the loan amount be A. Then
X = A/ s5 . Also Y = X(1 + i) + X = (2 + i)X = 2.09X. Thus i = .09. Finally
A Xs4 = 3007.87. So X = 3007.87/ (s5 s4 ) = 2130.85. C.
7. Brief Review of Probability Theory

Another aspect of insurance is that money is paid by the company only if some
event, which may be considered random, occurs within a specific time frame. For
example, an automobile insurance policy will experience a claim only if there is an
accident involving the insured auto. In this section a brief outline of the essential
material from the theory of probability is given. Almost all of the material presented
here should be familiar to the reader. The concepts presented here will play a crucial
role in the rest of these notes.
The underlying object in probability theory is a sample space S, which is simply
a set. This set is sometimes thought of as the collection of all possible outcomes
of a random experiment. Certain subsets of the sample space, called events, are
assigned probabilities by the probability measure (or probability set function)
which is usually denoted by P. This function has a few defining properties.
(1) For any event E S, 0 P[E] 1.
(2) P[] = 0 and P[S] = 1.
S
P
(3) If E1 , E2 , . . . are events and Ei Ej = for i j then P[ i=1 Ei ] = i=1 P[Ei ].
From these basic facts one can deduce all manner of useful computational formulas.
Exercise 71. Show that if A B are events, then P[A] P[B].
Another of the basic concepts is that of a random variable. A random variable
is a function whose domain is the sample space of a random experiment and whose
range is the real numbers.
In practice, the sample space of the experiment fades into the background and
one simply identifies the random variables of interest. Once a random variable has
been identified, one may ask about its values and their associated probabilities. All
of the interesting probability information is bound up in the distribution function of
the random variable. The distribution function of the random variable X, denoted
FX (t), is defined by the formula FX (t) = P[X t].
Two types of random variables are quite common. A random variable X with
distribution function FX is discrete if FX is constant except at at most countably
many jumps. A random
variable X with distribution function FX is absolutely
Z t
d
continuous if FX (t) =
FX (s) ds holds for all real numbers t.
ds
If X is a discrete random variable, the density of X, denoted fX (t) is defined by
the formula fX (t) = P[X = t]. There are only countably many values of t for which
the density of a discrete random variable is not 0. If X is an absolutely continuous
d
random variable, the density of X is defined by the formula fX (t) = FX (t).
dt
7: Brief Review of Probability Theory
38
Example 71. A Bernoulli random variable is a random variable which takes on

exactly two values, 0 and 1. Such random variables commonly arise to indicate the
success or failure of some operation. A Bernoulli random variable is discrete.
Exercise 72. Sketch the distribution function of a Bernoulli random variable with
P[X = 1] = 1/ 3.
Example 72. An exponentially distributed random variable Y with parameter
> 0 is a non-negative random variable for which P[Y t] = e t for t 0. Such a
random variable is often used to model the waiting time until a certain event occurs.
An exponential random variable is absolutely continuous.
Exercise 73. Sketch the distribution function of an exponential random variable
with parameter = 1. Sketch its density function also.
Exercise 74. A random variable X is uniformly distributed on an interval (a, b) if
X represents the result of selecting a number at random from (a, b). Find the density
and distribution function of a random variable which is uniformly distributed on the
interval (0, 1).
Exercise 75. Draw a picture of a distribution function of a random variable which
is neither discrete nor absolutely continuous.
Another useful tool is the indicator function. Suppose A is a set. The indicator
function of the set A, denoted 1A (t), is defined by the equation

1A (t) =
1 if t A
0 if t A.
Exercise 76. Graph the function 1[0,1) (t).

Exercise 77. Verify that the density of a random variable which is exponential
with parameter may be written e x 1(0,) (x).
Example 73. Random variables which are neither of the discrete nor absolutely
continuous type will arise frequently. As an example, suppose that a person has a fire
insurance policy on a house. The amount of insurance is $50,000 and there is a $250
deductible. Suppose that if there is a fire the amount of damage may be represented
by a random variable D which has the uniform distribution on the interval (0, 70000).
(This assumption means that the person is underinsured.) Suppose further that in the
time period under consideration there is a probability p = 0.001 that a fire will occur.
Let F denote the random variable which is 1 if a fire occurs and is 0 otherwise. The
size X of the claim to the insurer in this setting is given by
X = F (D 250)1[250,50000] (D) + (50000 250)1(50000,) (D) .
39
This random variable X is neither discrete nor absolutely continuous.

Exercise 78. Verify the correctness of the formula for X. Find the distribution
function of the random variable X.
Often only the average value of a random variable and the spread of the
values around this average are all that are needed. The X
expectation (or mean)
of a discrete random variable X is defined by E[X] =
t fX (t), while the ext
pectation
of an absolutely continuous random variable X is defined by E[X] =
Z
t fX (t) dt. Notice that in both cases the sum (or integral) involves terms of the
form (possible value of X) (probability X takes on that value). When X is neither

discrete nor absolutely continuous, the expectation is defined by the Riemann
R
Stieltjes integral E[X] =
t dFX (t), which again has the same form.
Exercise 79. Find the mean of a Bernoulli random variable Z with P[Z = 1] = 1/ 3.
Exercise 710. Find the mean of an exponential random variable with parameter
= 3.
Exercise 711. Find the mean and variance of the random variable in the fire
insurance example given above. (The variance of a random variable X is defined
by Var(X) = E[(X E[X])2 ] and is often computed using the alternate formula
Var(X) = E[X 2 ] (E[X])2 .)
The percentiles of a distribution (or random variable) are also sometimes used.
A pth percentile p of a random variable X is a number for which P[X p ] 1 p
and P[X p ] p. When X has an absolutely continuous distribution, these two
conditions can be replaced by the single condition P[X p ] = p. When p = 1/ 2, a
pth percentile is called a median of X.
Example 74. The median of the uniform distribution on the interval (0, 1) is 1/ 2.
The median of an exponential distribution with parameter is ln(1/ 2)/ . Notice
that this differs from the mean of the exponential distribution.
An important computational fact is that E[X + Y] = E[X] + E[Y] for any random
variables X and Y for which the expectations exist.
Computation of conditional probabilities will play an important role. If A and B
are events, the conditional probability of A given B, denoted P[A| B], is defined by
P[A| B] = P[A B]/ P[B] as long as P[B] 0. The intuition captured by the formula
is that the conditional probability of A given B is the probability of A recomputed
assuming that B has occured. This intuitive understanding is often the means by
which conditional probabilities are computed. The formal defintion is only used to
derive ways of manipulating these conditional probabilities.
40
Example 75. Two cards are drawn from a deck of 52 cards without replacement.
Suppose A is the event that the second card drawn is red and B is the event that the
first card drawn is red. The (unconditional) probability of A is 26/ 52, since half of
the cards are red. The conditional probability P[A|B] = 25/ 51, since knowing that
the first card is red leaves a deck of 51 cards of which 25 are red.
The events A and B are independent if P[A|B] = P[A]. The intuition underlying
the notion of independent events is that the occurance of one of the events does not
alter the probability that the other event occurs.
Similar definitions can be given in the case of random variables. Intuitively,
the random variables X and Y are independent if knowledge of the value of one of
them does not effect the probabilities of events involving the other. Independence
of random variables is usually assumed based on this intuition. One important fact
is that if X and Y are independent random variables then E[XY] = E[X] E[Y]. A
second important computational fact is that for independent random variables X and
Y, Var(X + Y) = Var(X) + Var(Y).
The conditional expectation of X given Y, denoted E[X |Y], is the random
variable which intuitively represents the expectation of X recomputed assuming the
value of Y is known. For independent random variables E[X |Y] = E[X]. A general
fact, known as the theorem of total expectation, is that for any random variables
X and Y, E[E[X |Y | ]] = E[X]. This fact is often used to simplify the computation
of expectations. A guiding principal in the application of this formula is that if in
attempting to compute E[X] the computation would be easy if the value of another
random variable Y were known, then compute E[X | Y] first and use the theorem of
total expectation.
Example 76. One iteration of an experiment is conducted as follows. A single six
sided die is rolled and the number D of spots up is noted. Then D coins are tossed
and the number H of heads observed is noted. In this case, E[H | D] = D/ 2, and
E[H] = E[D]/ 2 = 3.5/ 2. Direct computation would be quite involved.
When two or more random variables are studied at the same time, the probabilistic behavior of all the random variables as a group is usually of interest.
The joint distribution function of the random variables X and Y is defined by
FX,Y (s, t) = P[X s, Y t]. A similar definition is made for the joint distribution
function of more than two random variables. The joint distribution is discrete if
FX,Y is constant except for countably many jumps; the joint distribution is absolutely
Z s Z t
2
continuous if FX,Y (s, t) =
FX,Y (u, v) du dv for all s and t. If X and Y
u v
are jointly discrete, the joint density of X and Y is fX,Y (s, t) = P[X = s, Y = t];
if X and Y are jointly absolutely continuous the joint density of X and Y is
2
fX,Y (s, t) = s t FX,Y (s, t). If X and Y are independent, fX,Y (s, t) = fX (s)fY (t).
41
Problems
Problem 71. Suppose X has the uniform distribution on the interval (0, a) where
a > 0 is given. What is the mean and variance of X?
Problem 72. The moment generating function of a random variable X, denoted
MX (t), is defined by the formula MX (t) = E[etX ]. What is the relationship between
MX (0) and E[X]? Find a formula for Var(X) in terms of the moment generating
function of X and its derivatives at t = 0.
Problem 73. Express the Maclaurin expansion of MX (t) in terms of the moments
E[X], E[X 2 ], E[X 3 ],. . . of X. Hint: What is the Maclaurin expansion of ex ?
Problem 74. Find the moment generating function of a Bernoulli random variable
Y for which P[Y = 1] = 1/ 4.
Problem 75. A random variable X has the binomial distribution with parameters
n and p if X counts the number of successes in n independent Bernoulli trials
each with success probability p. Use the fact that X =
n
X
Bj where B1 , . . . , Bn are
j=1
independent Bernoulli random variables to compute the mean and variance of X.

What is the moment generating function of X?
Problem 76. A random variable G has the geometric distribution with parameter
p if G is the trial number of the first success in an infinite sequence of independent
Bernoulli trials each with success probability p. Let B denote the Bernoulli random
variable which is 1 if the first trial is a success and zero otherwise. Argue that
E[G|B] = B + (1 B)E[1 + G], and use this to compute E[G]. Also find the moment
generating function of G and Var(G)
Problem 77. A random variable N has the negative binomial distribution with
parameters r and p if N is the trial number of the first success in an infinite sequence
of independent Bernoulli trials each with success probability p. Find the mean,
variance, and moment generating function of N. Hint: Isnt N the sum of r
independent geometric random variables?

d2
d

Problem 78. Show that ln MX (t) = E[X] and 2 ln MX (t) = Var(X). This

dt
dt
t=0
t=0
is useful when the moment generating function has a certain form.
Problem 79. Find the moment generating function of a random variable Z which
has the exponential distribution with parameter . Use the moment generating
function to find the mean and variance of Z.
42
Problem 710. The probability generating function of a random variable X,

denoted PX (z), is defined by PX (z) = E[zX ], for z > 0. Probability generating
functions are often used instead of moment generating functions when the random
variable X is discrete. What is the relationship between the probability generating
function and the moment generating function?
Problem 711. A double indemnity life insurance policy has been issued to a
person aged 30. This policy pays $100,000 in the event of non-accidental death
and $200,000 in the event of accidental death. The probability of death during the
next year is 0.002, and if death occurs there is a 70% chance that it was due to an
accident. Write a random variable X which represents the size of the claim filed in
the next year. Find the distribution function, mean, and variance of X.
Problem 712. In the preceding problem suppose that if death occurs the day of
the year on which it occurs is uniformly distributed. Assume also that the claim
will be paid immediately at death and the interest rate is 5%. What is the expected
present value of the size of the claim during the next year?
Problem 713. Suppose X and Y are independent random variables with X having the exponential distribution with parameter and Y having the exponential distribution with parameter . Find the distribution of the random variable
X Y = min{X, Y}. Hint: Compute P[X Y > t].
Problem 714. Property insurance typically only pays the amount by which the
loss exceeds a deductible amount. Suppose the random variable L denotes the size
of the loss and d > 0 is the deductible. The insurance payment would then be
L d if L > d and zero otherwise. For convenience, the function x+ = x if x > 0
and 0 otherwise. Compute E[(L d)+ ] when L has an exponential distribution with
parameter .
Problem 715. Property insurance typically also has a limit on the maximum
amount that will be paid. If L is the random amount of the loss, and the maximum
paid by the insurance is a > 0, then the amount paid by the insurance company is
L a. Compute E[L a] when L has an exponential distribution with parameter .
Problem 716. True or False: For any number a and any number x, (xa)+ +(xa) =
x.
Problem 717. Show that for a random
variable Y which is discrete and takes non
negative integer values, E[Y] =
X
i=1
E[Y 2 ].
P[Y i]. Find a similar alternate expression for
43
Problem 718. Suppose

is a non-negative, absolutely continuous random variZ Y
able. Show that E[Y] =

P[Y > t] dt.
0

Problem 71. E[X] = a/ 2 and Var(X) = a2 / 12.
Problem 72. MX (0) = E[X].
x
Problem 73. Since e =
xk / k!,
k=0
MX (t) = E[etX ]
X
= E[ (tX)k / k!]
k=0
E[X k ]tk / k!.
k=0
Thus the coefficient of tk in the Maclaurin expansion of MX (t) is E[X k ]/ k!.

Problem 74. MY (t) = 3/ 4 + et / 4.
Since E[Bj ] = p, E[X] = np. P
Since Var(Bj ) = p(1 p),
n
t
B
tX
Var(X) = np(1 p). Now MX (t) = E[e ] = E[e j=1 j ] = E[etB1 . . . etBn ] =
tB1
tBn
t n
E[e ] . . . E[e ] = (1 p + pe ) .
Problem 75.
Problem 76. If the first trial is a success, then G = 1, while if the first
trial is a failure, the average waiting time until the first success will be 1 plus
E[G], since the trials are independent. The theorem of total expectation gives
E[G] = 1 + (1 p)E[G], from which E[G] = 1/ p. A similar argument shows that
E[etG |B] = et B + (1 B)MG+1 (t), from which MG (t) = pet / (1 (1 p)et ). From
this, Var(G) = (1 p)/ p2 .
Problem 77. Since N is the sum of r independent geometric random variables,
E[N] = r/ p, Var(N) = r(1 p)/ p2 , and MN (t) = (pet / (1 (1 p)et ))r .
Problem 79. MZ (t) = / ( t) for 0 t < .
Problem 710.
MX (t) = PX (et ).
From the definition, PX (z) = MX (ln(z)), or equivalently,
Problem 711. Let D be a random variable which is 1 if the insured dies in the
next year and 0 otherwise. Let A be a random variable which is 2 if death is due
to an accident and 1 otherwise. Then X = 100000AD.
Problem 712. If U is uniformly distributed on the integers from 1 to 365 then
E[100000ADvU ] is the desired expectation. Here v = 1/ (1 + 0.05(365) / 365).
Problem 713. Since X and Y are independent, P[X Y > t] = P[X > t] P[Y >
t] = e( +)t for t > 0. Thus X Y is exponential with parameter + .
Problem 714. Here E[(L d)+ ] = e d / .
44

Problem 715. Here E[L a] = (1 e a )/
Problem 716.
problems?
True. What does this have to do with the preceding two
Problem
717. Hint: In the usual formula for the expectation of Y write
P
i = ij=1 1 and then interchange the order of summation.
Problem 718. Use a trick like that of the previous problem. Double integrals
anyone?
45

Exercise 71. Using the fact that B = A (B \ A) and property (3) gives
P[B] = P[A] + P[B \ A]. By property (1), P[B \ A] 0, so the inequality
P[B] P[A] follows.
Exercise 72. The distribution function F(t) takes the value 0 for t < 0, the
value 2/ 3 for 0 t < 1 and the value 1 for t 1.
Exercise 73. The distribution function F(t) is 0 if t < 0 and 1 et for t 0.
The density function is 0 for t < 0 and et for t 0.
Exercise 74. The distribution function F(t) takes the value 0 for t < 0, the
value t for 0 t 1 and the value 1 for t > 1. The density function takes the
value 1 for 0 < t < 1 and 0 otherwise.
Exercise 75. The picture should have a jump and also a smoothly increasing
portion.
Exercise 76.
otherwise.
This function takes the value 1 for 0 t < 1 and the value 0
Exercise 78. The distribution function F(t) takes the value 0 if t < 0, the value
(10.001)+0.001(250/ 70000) for t = 0 (because X = 0 if either there is no fire or
the loss caused by a fire is less than 250), the value0.999+0.001(t +250)/ 70000
for 0 t < 50000 250 and the value 1 for t 49750.
Exercise 79. E[Z] = 0 (2/ 3) + 1 (1/ 3) = 1/ 3.
Exercise 710. The expectation is
integration by parts.
R
0
tf (t) dt =
R
0
t3e3t dt = 1/ 3 using
Exercise 711. Notice that the loss random variable X is neither discrete
nor absolutely continuous. The distribution function of X has two jumps:
one at t = 0 of size 0.999 + 0.001 250/ 70000 and another at 49750 of size
0.001 50000/ 70000. So E[X] = 0 (0.999 + 0.001 250/ 70000) +
R0.001
49750
t0.001/ 70000 dt + 49750 (0.001 0.001 50000/ 70000). The quantity
0
E[X 2 ] can be computed similarly.
46
8. Survival Distributions
An insurance policy can embody two different types of risk. For some types
of insurance (such as life insurance) the variability in the claim is only the time at
which the claim is made, since the amount of the claim is specified by the policy.
In other types of insurance (such as auto or casualty) there is variability in both the
time and amount of the claim. The problems associated with life insurance will be
studied first, since this is both an important type of insurance and also relatively
simple in some of its aspects.
The central difficulty in issuing life insurance is that of determining the length
of the future life of the insured. Denote by X the random variable which represents
the future lifetime of a newborn. For mathematical simplicity, assume that the
distribution function of X is absolutely continuous. The survival function of X,
denoted by s(x) is defined by the formula
s(x) = P[X > x] = P[X x]
where the last equality follows from the continuity assumption. The assumption
that s(0) = 1 will always be made.
Example 81. In the past there has been some interest in modelling survival functions in an analytic way. The simplest model is that due to Abraham DeMoivre. He
x
assumed that s(x) = 1
for 0 < x < where is the limiting age by which
all have died. The DeMoivre law is simply the assertion that X has the uniform
distribution on the interval (0, ).
Life insurance is usually issued on a person who has already attained a certain
age x. For notational convenience denote such a life aged x by (x), and denote the
future lifetime of a life aged x by T(x). What is the survival function for (x)? From
the discussion above, the survival function for (x) is P[T(x) > t]. Some standard
notation is now introduced. Set
t px
= P[T(x) > t]
t qx
= P[T(x) t].
and
When t = 1 the prefix is ommitted and one just writes px and qx respectively.
Generally speaking, having observed (x) some additional information about the
survival of (x) can be inferred. For example, (x) may have just passed a physical
exam given as a requirement for obtaining life insurance. For now this type of
possibility is disregarded. Operating under this assumption
t px
= P[T(x) > t] = P[X > x + t| X > x] =
s(x + t)
.
s(x)
8: Survival Distributions
48
Exercise 81. Write a similar expression for t qx .

Exercise 82. Show that for t s, t px = ts px+s s px .
There is one more special symbol. Set
t|u qx
= P[t < T(x) t + u]
which represents the probability that (x) survives at least t and no more than t + u
years. Again, if u = 1 one writes t| qx . The relations t| u qx = t+u qx t qx = t px t+u px
follow immediately from the definition.
Exercise 83. Prove these two equalitites. Show that t| u qx = t px u qx+t .
Exercise 84. Compute t px for the DeMoivre law of mortality. Conclude that under
the DeMoivre law T(x) has the uniform distribution on the interval (0, x).
Under the assumption that X is absolutely continuous the random variable T(x)
will be absolutely continuous as well. Indeed
P[T(x) t] = P[x X x + t| X > x] = 1
s(x + t)
s(x)
so the density of T(x) is given by

fT(x) (t) =
s (x + t)
fX (x + t)
=
.
s(x)
1 FX (x)
Intuitively this density represents the rate of death of (x) at time t.

Z
Exercise 85. Use integration by parts to show that E[T(x)] =

0
t px
dt. This
expectation is called the

Z complete expectation of life and is denoted by e x . Show
2
also that E[T(x) ] = 2
t t px dt.
0
Exercise 86. If X follows DeMoivres law, what is e x ?

The quantity
fX (x)
s (x)
=
1 FX (x)
s(x)
represents the death rate per unit age per unit survivor for those attaining age x, and is
called the force of mortality. Intuitively the force of mortality is the instantaneous
probability that someone exactly age x dies at age x. (In component reliability
theory this function is often referred to as the hazard rate.) Integrating both sides
of this equality gives the useful relation
x =
Z
s(x) = exp
t dt .
49
Exercise 87. Derive this last expression.

Exercise 88. Show that t px = e
R x+t
x
s ds
Exercise 89. What properties must a force of mortality have?

Exercise 810. Show that the density of T(x) can be written fT(x) (t) = t px x+t .
If the force of mortality is constant the life random variable X has an exponential distribution. This is directly in accord with the memoryless property of
exponential random variables. This memoryless property also has the interpretation
that a used article is as good as a new one. For human lives (and most manufactured
components) this is a fairly poor assumption, at least over the long term. The force
of mortality usually is increasing, although this is not always so.
Exercise 811. Find the force of mortality for DeMoivres law.
The curtate future lifetime of (x), denoted by K(x), is defined by the relation
K(x) = [T(x)]. Here [t] is the greatest integer function. Note that K(x) is a discrete
random variable with density P[K(x) = k] = P[k T(x) < k + 1]. The curtate
lifetime, K(x), represents the number of complete future years lived by (x).
Exercise 812. Show that P[K(x) = k] = k px qx+k .
Exercise 813. Show that the curtate expectation of life ex = E[K(x)] is given by
P
P
the formula ex = i=0 i+1 px . Hint: E[Y] = i=1 P[Y i].
50
Problems
Problem 81. Suppose x+t = t for t 0. Calculate t px x+t and e x .
Problem 82. Calculate
d
e x .
t px and
x
dx
Problem 83. A life aged (40) is subject to an extra risk for the next year only.
Suppose the normal probability of death is 0.004, and that the extra risk may be
expressed by adding the function 0.03(1 t) to the normal force of mortality for this
year. What is the probability of survival to age 41?
Problem 84. Suppose qx is computed using force of mortality x , and that qx is
computed using force of mortality 2x . What is the relationship between qx and qx ?
Problem 85. Show that the conditional distribution of K(x) given that K(x) k is
the same as the unconditional distribution of K(x + k) + k.
Problem 86. Show that the conditional distribution of T(x) given that T(x) t is
the same as the unconditional distribution of T(x + t) + t.
Problem 87. The Gompertz law of mortality is defined by the requirement that
t = Act for some constants A and c. What restrictions are there on A and c for this
to be a force of mortality? Write an expression for t px under Gompertz law.
Problem 88. Makehams law of mortality is defined by the requirement that
t = A + Bct for some constants A, B, and c. What restrictions are there on A, B and
c for this to be a force of mortality? Write an expression for t px under Makehams
law.
Problem 81. Here t px = e
Rt
0
x+s ds
= et
/2
and e x =
t px
dt = 2 / 2.
Problem 82.
d
e x =
t px = t px (x x+t ) and
x
dx
Z
0
x 1.
t px dt = x e
x
Problem 83. If t is the usual force of mortality then p40 = e
R1
0
40+s +0.03(1s) ds
Problem 84. The relation px = (px )2 holds, which gives a relation for the
death probability.
Problem 85. P[K(x) k + l|K(x) k] = P[k K(x) k + l]/ P[K(x) k] =
= P[K(x + k) + k l + k].
l qx+k
Problem 86. Proceed as in the previous problem.

Problem 87. A force of mortality must always be non-negative and have
infinite integral. Thus A > 0 and c 1 are required here. Direct substitution
gives t px = exp{A(cx cx+t )/ ln c}.
Problem 88. If c < 1 then A > 0 is required, while if c 1 then A + B > 0 is
required. Substitution gives t px = exp{At B(cx cx+t )/ ln c}.
51
Exercise 81. t qx = P[T(x) t] = P[X x + t|X > x] = P[x < X x + t]/ P[X >
x] = (s(x) s(x + t))/ s(x).
Exercise 82. t px = s(x + t)/ s(x) = s(x + s + (t s))/ s(x) = (s(x + s + (t s))/ s(x +
s))(s(x + s)/ s(x)) = ts px+ss px . What does this mean in words?
Exercise 83. For the first one, t|u qx = P[t < T(x) t + u] = P[x + t < X
t+u+x|X > x] = (s(x+t)s(t+u+x))/ s(x) = (s(x+t)s(x)+s(x)s(t+u+x))/ s(x) =
t+u qx t qx . The second identity follows from the fourth term by simplifying
(s(x + t) s(t + u + x))/ s(x) = t px t+u px . For the last one, t|u qx = P[t < T(x)
t + u] = P[x + t < X t + u + x| X > x] = (s(x + t) s(t + u + x))/ s(x) =
(s(x + t)/ s(x))(s(t + x) s(t + u + x))/ s(x + t) = t px u qx+t .
Exercise 84. Under the DeMoivre law, s(x) = ( x)/ so that t px =
( x t)/ ( x) for 0 < t < x. Thus the distribution function of T(x) is
t/ ( x) for 0 < t < x, which is the distribution function of a uniformly
distributed random variable.
R
R
R
Exercise 85.
R E[T(x)] = 0 tfT(x) (t) dt = 0 ts (x + t)/ s(x) dt = 0 s(x +
t)/ s(x) dt = 0 t px dt. The fact the limt s(x + t) = 0 is assumed, since everyone
eventually dies. The other expectation is computed similarly.
Exercise 86. Under DeMoivres law, e x = ( x)/ 2, since T(x) is uniform on
the interval (0, x).
Rx
Rx
Exercise 87.
0 t dt = 0 s (t)/ s(t) dt = ln(s(x)) + ln(s(0)) = ln(s(x)),
since s(0) = 1. Exponentiating both sides to solve for s(x) gives the result.
R x+t
ds
Exercise 88. From the previous exercise, s(x + t) = e 0 s . Using this fact,
the previous exercise, and the fact that t px = s(x + t)/ s(x) gives the formula.
Exercise
89. The force of mortality must satisfy x 0 for all x, and
Z
x dx = . This last condition is needed to make lim s(x) = 0.
x
Exercise 810. Since fT(x) (t) = s (x + t)/ s(x) and s (x + t) = s(x + t)x+t by
the previous exercise, the result follows.
Exercise 811. From the earlier expression for the survival function under
DeMoivres law s(x) = ( x)/ , so that x = s (x)/ s(x) = 1/ ( x), for
0 < x < .
Exercise 812. P[K(x) = k] = P[k T(x) < k+1] = (s(x+k)s(x+k+1))/ s(x) =
((s(x + k) s(x + k + 1))/ s(x + k))(s(x + k)/ s(x)) = k px qx+k .
Exercise
813. E[K(x)] =
P
p
.
j+1
x
j=0
P
i=1
P[K(x) i] =
P
i=1
P[T(x) i] =
i=1 i px
52
9. Life Tables
In practice the survival distribution is estimated by compiling mortality data in
the form of a life table. An example of a life table appears later in these notes.
The conceptual model behind the entries in a life table is this. Imagine that
at time 0 there are l0 newborns. Here l0 is called the radix of the life table and is
usually taken to be some large number such as 10,000,000. These newborns are
observed and lx is the number of the original newborns who are still alive at age x.
Similarly n dx denotes the number of the group of newborns alive at age x who die
before reaching age x + n. As usual, when n = 1 it is supressed in the notation.
Exercise 91. Show that n dx = lx lx+n .
The ratio
lx
is an estimate of s(x) based on the collected data. Assume that in
l0
lx
for non-negative integer values of x. Since earlier the assumption was
l0
made that the life random variable X is absolutely continuous, the question arises
as to how the values of the survival function will be computed at non-integer values
of x.
fact s(x) =
Under the assumption of the uniform distribution of deaths in the year of

death, denoted UDD, the number alive at age x + t, where x is an integer and
0 < t < 1, is given by
lx+t = lx tdx .
The UDD assumption means that the age at death of those who will die at curtate
age x is uniformly distributed between the ages x and x + 1. In terms of the survival
function the UDD assumption means
s(x + t) = (1 t)s(x) + ts(x + 1)
where x is an integer and 0 t 1. The UDD assumption is the one most commonly
made.
Direct application of the UDD assumption and the earlier definitions shows that
under UDD
t qx = tqx ,
t px
= 1 tqx ,
and
x+t =
whenever x is an integer and 0 < t < 1.
qx
1 tqx
9: Life Tables
54
The assumption of a constant force of mortality in each year of age means

that x+t = x for each integer age x and 0 < t < 1. This is equivalent to the formula
t
t
where = ln px . The constant force
t px = (px ) and also to s(x + t) = s(x) e
assumption is less widely used than UDD.
Having observed (x) may mean more than simply having seen a person aged x.
For example, (x) may have just passed a physical exam in preparation for buying
a life insurance policy. One would expect that the survival distribution of such a
person could be different from s(x). If this is believed to be the case the survival
function is actually dependent on two variables: the age at selection (application
for insurance) and the amount of time passed after the time of selection. A life table
which takes this effect into account is called a select table. A family of survival
functions indexed by both the age at selection and time are then required and notation
such as q[x]+i denotes the probability that a person dies between years x+i and x+i+1
given that selection ocurred at age x. As one might expect, after a certain period
of time the effect of selection on mortality is negligable. The length of time until
the selection effect becomes negligable is called the select period. The Society of
Actuaries (based in Illinois) uses a 15 year select period in its mortality tables. The
Institute of Actuaries in Britain uses a 2 year select period. The implication of the
select period of 15 years in computations is that for each j 0, l[x]+15+j = lx+15+j .
A life table in which the survival functions are tabulated for attained ages only
is called an aggregrate table. Generally, a select life table contains a final column
which constitutes an aggregate table. The whole table is then referred to as a select
and ultimate table and the last column is usually called an ultimate table. With
these observations in mind the select life tables can be used in computations.
Exercise 92. You are given the following extract from a 3 year select and ultimate
mortality table.
x
70
71
72
l[x]
l[x]+1
l[x]+2
7984
8016
7592
lx+3
7600
x+3
73
74
75
Assume that the ultimate table follows DeMoivres law and that d[x] = d[x]+1 =
d[x]+2 for all x. Find 1000(2|2 q[71] ).
9: Life Tables
55
Problems
Problem 91. Use the life table to compute
for fractional years.
1/ 2 p20
under each of the assumptions
Problem 92. Show that under the assumption of uniform distribution of deaths in
the year of death that K(x) and T(x) K(x) are independent and that T(x) K(x) has
the uniform distribution on the interval (0, 1).
Problem 93. Show that under UDD e x = ex + 12 .
9: Life Tables
Problem 91. Under UDD, t px = (1 t) + tpx so 1/ 2 p20 = 1/ 2 + 1/ 2p20 =
1
1 9,607,896
t
2 + 2 9,617,802 . Under constant force, t px = (px ) so 1/ 2 p20 = p20 .
Problem 92. For 0 t < 1, P[K(x) = k, T(x) K(x) t] = P[k T(x)
k + t] = k px t qx+k = k px (t tpx+k ) = tP[K(x) = k].
Problem 93. Use the previous problem to see that e x = E[T(x)] = E[K(x) +
(T(x) K(x))] = ex + E[T(x) K(x)] = ex + 1/ 2, since T(x) K(x) has the uniform
distribution on the unit interval.
56
9: Life Tables
Exercise 91. Since n dx is the number alive at age x who die by age x + n, this
is simply the number alive at age x, which is lx , minus the number alive at age
x + n, which is lx+n .
Exercise 92. The objective is to compute 10002|2 q[71] = 1000(2 p[71] 4 p[71] ) =
1000(l[71]+2 l[71]+4 )/ l[71] = 1000(l[71]+2 l75 )/ l[71] , where the effect of the selection period has been used. To find the required entries in the table proceed as
follows. Since 8016 7592 = 424 and using the assumption about the number of
deaths, l[72]+1 = 8016212 = 7804 and l72+3 = 7592212 = 7380. Since the ultimate table follows DeMoivres Law, l71+3 = (7600 + 7380)/ 2 = 7490. Again using the assumption about the number of deaths, l[71]+2 = (7984 + 7490)/ 2 = 7737
and l[71] = 7984 + 247 = 8231. So 10002|2 q[71] = 1000(7737 7380)/ 8231 =
43.37.
57

Solve the following 6 problems in no more than 30 minutes. You should
have the Tables for Exam M available for these problems.
1
Question 101 . You are given s(x) =
. Determine the median future lifetime
1+x
of (y).
A. y + 1
1
D.
y
B. y
C.
E.
1
1+y
Question 102 . You are given x = 0.1 for all ages x > 0. The probability that
(30) and (50) will die within 10 years of each other is p. Calculate p.
A. 0.1e1
B.
0.5e1
D.
0.5(1 e1 )
C.
e1
E.
1 e1
Question 103 . Given e 0 = 25, lx = x, 0 x , and T(x) is the future lifetime

random variable, calculate Var(T(10)).
A. 65
B.
93
D.
178
C.
133
E.
333
Question 104 . For a certain mortality table you are given (80.5) = 0.0202,
(81.5) = 0.0408, (82.5) = 0.0619, and that deaths are uniformly distributed
between integral ages. Calculate the probability that a person age 80.5 will die
within two years.
A. 0.0782
B.
0.0785
D.
0.0796
C.
0.0790
E.
0.0800
59
Question 105 . The future lifetimes of a certain population can be modeled as

follows. Each individuals future lifetime is exponentially distributed with constant
hazard rate . Over the population, is uniformly distributed over (1, 11). Calculate
the probability of surviving to time 0.5 for an individual randomly selected at time
0.
A. 0.05
B.
0.06
D.
0.11
C.
0.09
E.
0.12
Question 106 . An insurance agent will receive a bonus if his loss ratio is less
than 70%. You are given that his loss ratio is calculated as incurred losses divided
by earned premium on his block of business. The agent will receive a percentage of
earned premium equal to 1/3 of the difference between 70% and his loss ratio. The
agent receives no bonus if his loss ratio is greater than 70%. His earned premium
is 500,000. His incurred losses are distributed according to the Pareto distribution

600, 000 3
F(x) = 1
, x > 0. Calculate the expected value of his bonus.
x + 600, 000
A. 16,700
B.
31,500
D.
50,000
C.
48,300
E.
56,600
60

Question 101 . From the information given, P[T(y) > t] = s(y + t)/ s(y) = (1 +
y)/ (1 + y + t), and this is equal to 1/ 2 when t = 1 + y. A.
Question 102 . Since the force of mortality is constant, both lives have the same
exponential distribution. Z Thus p = P[T(30) < T(50) < T(30) + 10] + P[T(50) <
T(30) < T(50) + 10] = 2

(ex e(x+10) )ex dx = 1 e10 = 1 e1 . E.
0
Question 103 . From the form of lx , mortality follows DeMoivres law. Thus
= 50 since e 0 = 25. The random variable T(10) is therefore uniformly distributed
on the interval (0, 40) and Var(T(10)) = 133.33. C.
Question 104 . Under UDD, x+t = qx / (1 tqx ) so that qx = x+0.5 / (1 + 0.5x+0.5 ).
p80
Now 2 p80.5 = 0.5 p80.5 p81 0.5 p82 = 10.5q
(1 q81 )(1 0.5q82 ). The expression for
80
p
comes
from
the
fact
that
p
=
p
.5 80.5
80
0.5 80 0.5 p80.5 . Using the information gives the
survival probability as 0.9218 and the death probability as 0.0781. A.
Question 105 . The desired probability is E[e0.5 ] =
11
e0.5t dt/ 10 = 0.120. E.
Question 106 . If L is the incurred loss, the expected bonus is (500000/ 3)E[(.7
L/ 500000)+ ], using the given information. Now (ax)+ +(x a) = a, so the expected
bonus can be written as (1/ 3)(350000E[L350000]) = (1/ 3)(350000300000(1
(600000/ 950000)2 )) = 56, 555 using the formula on the supplied tables. E.
11. 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
x : n is a status which dies at the time of death of (x) if the death of (x) occurs 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.
11: Status
62
Problems
Problem 111. Use the life table to compute p40:50 and p40:50 assuming (40) and
(50) are independent lives.
1
Problem 112. Find a formula for the survival function of x : n in terms of the
survival function of (x).
1
Problem 113. If the UDD assumption is valid for (x), does UDD hold for x : n ?
1
Problem 114. Find a formula for the survival function of x : n .
1
Problem 117. If the UDD assumption is valid for each of (x) and (y) and if (x) and
(y) are independent lives, does UDD hold for x : y?
11: Status
Problem 111. From independence, p40:50 = p40 p50 =
p40:50 = 1 q40 q50 , by independence.
9,287,264 8,897,913
9,313,166 8,950,901 .
Also
1
1
Problem 112. P[T(x : n ) t] = t px for 0 t < n and P[T(x : n ) t] = n px
for t n.
1
1
Problem 113. The UDD assuption holds for x : n if and only if P[T(x : n )
1
1
k + t] = (1 t)P[T(x : n ) k] + tP[T(x : n ) k + 1] for all integers k and all
0 t 1. Now use the formula for the survival function found in the previous
problem to see that UDD does hold for the joint status.
1
Problem 114. Here P[T(x : n ) t] = 1 if t < n and is equal to n qx if t n.
Problem 115. Using the previous formula for the survival function shows that
UDD fails to hold on the interval (n 1, n).
Problem 116. The survival function for x : n is S(t) = t px for t < n and zero
for t n. So again the UDD condition will fail to hold on the interval (n 1, n).
Problem 117. The survival function for the joint status is S(t) = t px t py , and
in general UDD will not hold for the joint status.
63
11: Status
Exercise 111. The joint life status survives t time units if and only if both (x)
and (y) survive t time units. Using the independence gives s(t) = t px t py .
Exercise 112. Since a constant random variable is independent of any other
random variable, s(t) = t px t pn = t px if t n and 0 if t > n, by using the previous
exercise.
1
Exercise 113. The status x : n dies at time n if (x) is still alive at time n,
otherwise this status never dies.
64
12. 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.
13. 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
n-year term
A1x:n = E[vT 1[0,n] (T)]
whole life
n-year endowment
Ax = E[vT ]
Ax:n = E[vTn ]
m-year deferred n-year term
T
m| n Ax = E[v 1(m,n+m] (T)]
T
(IA)(m)
x = E[v [Tm + 1]/ m]
(IA)1x:n = E[vT [T + 1]1[0,n) (T)]
(DA)1x:n = E[vT (n [T])1[0,n) (T)]
whole life increasing mthly

n-year term increasing annually
n-year term decreasing annually
1
x:n
= n Ex = E[vn 1(n,) (T)]
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 .
13: Life Insurance
67
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 A1x:n . This
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 status x : 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 = A1x:n + vn n px Ax+n by conditioning on the event
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 = A1x:n + A 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
13: Life Insurance
68
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.
Insurances Payable the End of the Year of Death
Type
Net Single Premium
n-year term
A1x:n = E[vK+1 1[0,n) (K)]
whole life
Ax = E[vK+1 ]
n-year endowment
Ax:n = E[v(K+1)n ]
m-year deferred n-year term
m| n Ax
whole life increasing annually

n-year term increasing annually
(IA)x = E[vK+1 (K + 1)]

(IA)1x:n = E[vK+1 (K + 1)1[0,n) (K)]
n-year term decreasing annually
= E[vK+1 1[m,n+m) (K)]
(DA)1x:n = E[vK+1 (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 =
X
k=0
vk+1 k px qx+k =
X
k=0
vk+1
dx+k
.
lx
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 T(x) ]
= E[e (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).
13: Life Insurance
69
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 A(m)
x .
(m) (m)
Show that under UDD i 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.
13: Life Insurance
70
Problems
Problem 131. Write expressions for all of the net single premiums in terms of
either integrals or sums. Hint: Recall the form of the density of T(x) and K(x).
Problem 132. Show that A1x:n = iA1x:n , but that Ax:n iAx:n , in general.
Problem 133. Use the life table and UDD assumption (if necessary) to compute
A21 , A21:5 , and A1 .
21:5
Problem 134. Show that

dAx
= v(IA)x .
di
Problem 135. Assume that DeMoivres law holds with = 100 and i = 0.10.
Find A30 and A30 . Which is larger? Why?
Problem 136. Suppose x+t = and i = 0.10. Compute Ax and A1x:n . Do your
answers depend on x? Why?
Problem 137. Suppose Ax = 0.25, Ax+20 = 0.40, and Ax:20 = 0.55. Compute A
1
x:20
and A1x:20 .
Problem 138. Show that
(IA)x = vqx + v[Ax+1 + (IA)x+1 ]px .
What assumptions (if any) did you make?
Problem 139. What change in Ax results if for some fixed n the quantity qx+n is
replaced with qx+n + c?
13: Life Insurance

Problem 131. The densities required are fT(x) (t) = t px x+t and fK(x) (k) =
k px qx+k respectively.
Problem 132. A1 = (Ax e n n px Ax+n ) = iAx ivn n px Ax+n = iA1 .
x:n
x:n
Problem 133. Use A21 = iA21 , A21:5 = A1
21:5
+ 5 E21 and the previous problem.
Problem 134. Just differentiate under the expectation in the definition of Ax .

Problem 135. Clearly A30 > A30 since the insurance is paid sooner in the
continuous case. Under DeMoivres law the UDD assumption is automatic and
R 70 t
1
A30 = 70
dt.
0 e
Z
Problem 136. Using the form of gives Ax =
e t et dt = / ( + ).
Similarly, A1 = (1 en(+ ) )/ ( + ). The answers do not depend on x since

x:n
the lifetime is exponential and therefore ageless.
Problem 137. The two relations Ax = A1 + v20 n px Ax+20 and Ax:20 = A1 +
x:20
x:20
vn n px along with the fact that A 1 = vn n px give two equations in the two sought
after unknowns.
x:20
Problem 138. Either the person dies in the first year, or doesnt. If she doesnt
buy an increasing annually policy for (x + 1) and a whole life policy to make up
for the increasing part the original policy would provide.
Problem 139. The new benefit is the old benefit plus a pure endowment
benefit of cv at time n.
71
13: Life Insurance

Exercise 131. Since if the benefit is paid, the benefit payment occurs at time
n, n Ex = E[vn 1[n,) (T(x))] = vn P[T(x) n] = vn n px .
Exercise 132.R Under the assumptions given the net single premium is
E[1000vT(x) ] = 0 1000e0.05t (1/ 50)et/ 50 dt = 285.71.

Exercise 133. For the conditioning argument, break the expectation into two
pieces by writing Ax = E[vT ] = E[vT 1[0,n] (T)] + E[vT 1(n,) (T)]. The first expectation is exactly A1 . For the second expectation, using the Theorem of Total Exx:n
pectation gives E[vT 1(n,) (T)] = E[E[vT 1(n,) (T)| T n]]. Now the conditional
distribution of T given that T n is the same as the unconditional distribution of
T(x+n)+n. Using this fact gives the conditional expectation as E[vT 1(n,) (T)| T
n] = E[vT(x+n)+n 1(n,) (T(x + n) + n)]1(n,) (T) = vn Ax+n 1(n,) (T). Taking expectations gives the result. To use the time diagram, imagine that instead of buying
a whole life policy now, the insured pledges to buy an n year term policy now,
and if alive after n years, to buy a whole life policy at time n (at age x + n). This
will produce the same result. The premium for the term policy paid now is A1
x:n
and the premium for the whole life policy at time n is Ax+n . This latter premium
is only paid if the insured survives, so the present value of this premium is the
second term in the solution.
Exercise 134. Using the definition and properties of expectation gives Ax:n =
E[vT 1[0,n] (T) + vn 1(n,) (T)] = E[vT 1[0,n] (T)] + E[vn 1(n,) (T)] = A1 + A 1 .
x:n
x:n
Exercise 135. The net single premium for the pure endowment policy is
v5 5 p30 = 5 E30 == 0.74091.
Exercise 136.
vn lx+n / lx .
1
x:n
= E[vn 1[n,) (T)] = E[vn 1[n,) (K)] = A
1
x:n
= vn n px =
Exercise 137. Since term policies can be expressed as a difference of premiums for whole life policies, the answer is yes.
Exercise 138. The net single premium for a pure endowment policy is
5 E30 = 0.74091. For the endowment policy, the net single premium for a 5
year term policy must be added to this amount. From the relation given earlier,
A1 = A30 5 E30 A35 . The relationship between insurances payable at the time
30:5
of death and insurances payable at the end of the year of death is used to complete
the calculation. This gives A30:5 = (i/ )(0.10248) (i/ )(0.74091)(0.12872) +
0.74091.
Exercise 139. Notice that [mT(x)] is the number of full mths of a year
that (x) lives before dying. (Here [a] is the greatest integer function.) So the
number of mths of a year that pass until the benefit for the insurance is paid
is [mT(x)] + 1, that is, the benefit is paid at time ([mT(x)] + 1)/ m. From here
the derivation proceeds as above. A(m)
= E[v([mT]+1)/ m ] = E[v([m(TK+K)]+1)/ m ] =
x
K
([m(TK)]+1)/ m
E[v ]E[v
]. Now T K has the uniform distribution on the interval
(0, 1) under UDD, so [m(T K)] has
distribution over the integers
P the uniform
j/ m
0,. . . , m 1. So E[v([m(TK)]/ m ] = m1
v
(1/
m) = (1/ m)(1 v)/ (1 v1/ m )
j=0
72
13: Life Insurance

from the geometric series formula. Substituting this in the earlier expression
1 1/ m
gives A(m)
(1/ m)(1 v)/ (1 v1/ m ) = Ax / i(m) since i(m) = m(v1/ m 1).
x = Ax v v
73
14. 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 nonrandom 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
an =
vj =
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 =
n1
X
j=0
1 vn 1 vn
v =
=
.
1v
d
j
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[aK(x)+1 ] where the left member is
a notational convention. This formula gives
a x = E[aK(x)+1 ] = E[
1 vK(x)+1
1 Ax
]=
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)
x . Show that
(m)
(m) (m)
Ax + d a x = 1.
14: Life Annuities
75
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
90
1 A60
1 0.36913
= 90
= 1003.08.
1v
1 1/ 1.06
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
P
vj ]. A similar argument applies
total of K(x) n payments, and ax:n = E[ K(x)n
j=1
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)(n1)
vj ] = E[ 1v 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
14: Life Annuities
76
the obvious notation gives

1 A(m)
x
d(m)
i
1 i(m)
Ax
=
d(m)
i
(1 da x )
1 i(m)
=
d(m)
id
i(m) i
= (m) (m) a x + (m) (m)
i d
i d
a (m)
x =
where at the second equality the UDD assumption was used. Since this relationship
id
i i(m)
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
ax (m)
Tables for Exam M. The above relationship is then written as a (m)
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
Z n
1 e n
e t dt =
.
an =
0
A life annuity which is payable continuously will thus have actuarial present value
ax = E[aT(x) ] = E[
1 e T(x)
].
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
14: Life Annuities
77
due for a period of n years the accumulated value of the annuity at time n, denoted
n
by sn , is given by sn = (1 + i)n a n = (1+i)d 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
x:n
n Ex s
= 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 sx:n represents the amount of pure
endowment payable at time n which is actuarially equivalent to the annuity.
14: Life Annuities
78
Problems
Problem 141. Show that under UDD
(3)
x.
ax < a(2)
(2)
(3)
x < ax < < ax < < a
x < a
x < a
Give an example to show that without the UDD assumption some of the inequalities
may fail.
1
Problem 142. True or false: A1x:n = 1 d a 1x:n . Hint: When does x : n die?
Problem 143. True or false: sx:n sn .
Problem 144. Use the life table to calculate the actuarial present value of $1000
due in 30 years if (40) survives.
Problem 145. Use the life table to compute a21 .
Problem 146. Find a general formula for
table to compute 5|10 a 20 .
x
m| n a
and use it together with the life
Problem 147. Prove ax:n = 1 Ex a x+1:n .

Problem 148. Show that (Ia)T + TvT = aT .
Problem 149. Use the previous problem to show that (Ia)x + (IA)x = ax . Here
(Ia)x is the actuarial present value of an annuity in which payments are made at rate
t at time t. Is there a similar formula in discrete time?
Problem 1410. Show that a x = 1 + ax and that
1
m
1 n
+ a(m)
(m)
x:n = a
x:n + m v n px .
Problem 1411. Show that a x:n = a x vn n px a x+n and use this to compute a 21:5 .
14: Life Annuities

Problem 141. Intuitively, as the type of annuity varies from left to right,
the annuitant receives funds sooner and thus the present value is higher. A
(m)
(m)
(m)
direct argument begins with a (m)
increases with m,
x = (1 Ax )/ d . Now d
(m)
(m)
while under UDD Ax = (i/ i )Ax also increases with m. So the numerator of the
expression for a (m)
x decreases with m and the denominator increases with m, so the
(m)
(m)
value decreases with m. A similar argument works for a(m)
x = (1(i/ d )Ax )/ i .
(2)
1/ 3
(3)
If (x) must die 5 months through the year, a x = 1/ 2 < 1/ 3 + v (1/ 3) = a x , at
least for some values of v.
Problem 142. The status dies only if (x) dies before time n. The result is true.
Problem 143. False, since sx:n = a x:n / n Ex = vn / n px + vn+1 px / n px + . . . +
v1 n1 px / n px vn + vn+1 + . . . + v1 = sn .
Problem 144. The value is 100030 E40 = 1000v30 6,616,155
9,313,166 .
Problem 145. Here a21 = ()a21 (), under UDD.
Problem 146.
x
m|n a
= m Ex a x+m m+n Ex a x+m+n .
Problem 147. The payment made at the end of the first year is only made
if (x) survives the year. From the point of view of the end of the first year, the
annuity immediate looks like an annuity due for (x + 1).
Problem 148.
Z T
t e t dt.
Use integration by parts starting with the formula (Ia)T =
Problem 1410. The annuity due pays 1 now for sure, and the remaining
payments look like an annuity immediate. For the temporary annuity due, the
payment of 1/ m now is certain, and the last payment of the annuity immediate
is only made upon survival. The rest of the payments are the same.
Problem 1411. To buy a life annuity due, buy an n year temporary annuity
due today, and if you are still alive n years from now buy a life annuity due.
79
14: Life Annuities

K(x)
Exercise 141. In this case, E[aK(x) ] = E[ 1vi
]=
1v1 Ax
.
i
Exercise 142. Here there

the geometric series
P are [mT] +j/ m1 payments, so using
([mT]+1)/ m
formula gives a (m)
= E[ [mT]
)/ (1 v1/ m )].
x
j=0 (1/ m)v ] = E[(1/ m)(1 v
1/ m
(m)
Now m(1 v ) = d , which gives the result.
Exercise 143. To get your moneys worth, you must live long enough so that
the present value of the annual dues you would pay if you were not a life member
will exceed $675. This gives a condition that K(60) must satisfy if you are to
get your moneys worth.
(K+1)n
(K+1)n
Exercise 144. For the first one a x:n = E[ 1v d ] = E[ 1v d 1[0,n1] (K)] +

(K+1)n
(K+1)
n
E[ 1v d 1[n,) (K)] = E[ 1vd 1[0,n1] (K)] + E[ 1v
d 1[n,) (K)] = (1/ d)(1 Ax:n ).
A similar argument shows that ax:n = (1/ i)(A1 + n px vn+1 ).
x:n
Exercise 145. The argument proceeds in a similar way, beginning with the
1v1/ m A(m)
x
relation a(m)
.
x =
i(m)
Exercise 146. Direct reasoning gives m|n a x = a x:n+m a x:m .
Exercise 147. The first relationship follows directly from the given equation
and the fact that Ax = E[e T(x) ]. Since T(x : n ) = T(x) n a similar argument
gives ax:n = (1/ )(1 Ax:n ).
80

Question 151 . You are given q60 = 0.020, q61 = 0.022, and the fact that deaths
are uniformly distributed over each year of age. Calculate e 60:1.5 .
A. 1.447
B.
1.457
D.
1.477
C.
1.467
E.
1.487
Question 152 . Simplify
(IA)x A1x:1
(IA)x+1 + Ax+1
A.
Ax
B.
Ax+1
D.
A1x:1
C.
Ax:1
E.
1
x:1
Question 153 . Z1 is the present value random variable for an n-year continuous
endowment insurance of 1 issued to (x). Z2 is the present value random variable for
an n-year continuous term insurance of 1 issued to (x). Calculate Var(Z1 ), given that
Var(Z2 ) = 0.01, vn = 0.30, n px = 0.80, and E[Z2 ] = 0.04.
A. 0.0036
B.
0.0052
D.
0.0144
C.
0.0098
E.
0.0148
Question 154 . You are given = 0,

0
t t px dt = g, and Var(aT ) = h where T is
the future lifetime random variable for (x). Express e x in terms of g and h.
A. h g
B.
C.
hg
gh
D.
E.
2g h
2h g
82
Question 155 . An insurance company has agreed to make payments to a worker

age x who was injured at work. The payments are 150,000 per year, paid annually,
starting immediately and continuing for the remainder of the workers life. After
the first 500,000 is paid by the insurance company, the remainder will be paid by a
reinsurance company. The interest rate i = 0.05 and t px = (0.7)t , for 0 t 5.5 and
t px = 0 for t > 5.5. Calculate the actuarial present value of the payments to be made
by the reinsurer.
A. Less than 50,000
B. At least 50,000 but less than
100,000
C. At least 100,000 but less than
150,000
D. At least 150,000 but less than

200,000
E.
At least 200,000
Question 156 . For a two year term insurance on a randomly chosen member
of a population you are given that 1/3 of the population are smokers and 2/3 are
nonsmokers. The future lifetimes follow a Weibull distribution with = 2 and
= 1.5 for smokers, and = 2 and = 2.0 for nonsmokers. The death benefit is
100,000 payable at the end of the year of death, and i = 0.05. Calculate the actuarial
present value of this insurance.
A. 64,100
B.
64,300
D.
64,900
C.
64,600
E.
65,100
83

Question 151 . Here
e 60:1.5 = E[T(60) 1.5]
Z
=
0
1.5
t t p60 60+t dt + 1.5P[T(60) > 1.5]
= q60 / 2 + p60 q61
Z
1
1.5
t dt + 1.5p601/ 2 p61
= 1.477.
D.
Question 152 . The numerator is the actuarial present value of an insurance which
pays nothing for death in the first year, 2 for death in the second year, and so on.
Thus (IA)x A1x:1 = vpx (Ax+1 + (IA)x+1 ) and the ratio is vpx = A 1 . E.
x:1
Question 153 . Here Z2 = vT(x) 1[0,n] (T(x)) and Z1 = vn 1[n,) (T(x)) + Z2 . Since the
indicators multiply to zero, E[Z12 ] = v2n n px + E[Z22 ] = (.30)2 (.80) + .01 + (.04)2 =
0.0836, and by additivity of expectation, E[Z1 ] = vn n px + E[Z2 ] = (.30)(.80) + .04 =
.28. Thus Var(Z2 ) = .0836 (.28)2 = 0.0052. B.
Question
154 . Since = 0, aT = T and thus h = Var(T). Also since E[T 2 ] =
Z
p
2
x = E[T] = 2g h. D.
t px dt, 2g = E[T ]. Hence e
0
Question 155 . The actuarial present value is

approximately. B.
5.5
10/ 3
150000e t t px dt = 55978.85,
Question 156 . The actuarial present value is 100000(1/ 3(vFS (1) + v2 (FS (2)
2
FS (1))) + 2/ 3(vFN (1) + v2 (FN (2) FN (1))) where FS (t) = 1 e(t/ 1.5) is the lifetime
2
distribution for smokers and FN (t) = 1 e(t/ 2) is the lifetime distribution for
nonsmokers. Plugging in gives the value as 64,558.99. C.
16. 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[aK(x)+1 ]
or
Ax Pax = 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 premium
for an n year term policy with premiums payable monthly is denoted P(12)
1 .
x:n
Exercise 162. Use the life table to find P1
30:10
. What is P(12)
?
1
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,
16: Net Premiums
85
the net annual premium P is the solution of the equation

Pa(12)
= 100000 A1x:65x + 75000 65x| 10 Ax + 50000 75x| 5 Ax + 25000 80x| 5 Ax
x:85x
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) (A1x:n ). The notation for other semi-continuous policies is similar.
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 PaK(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)?
16: Net Premiums
86
Problems
Problem 161. Show that if = 0 then P(Ax ) = 1/ e x .
Problem 162. Arrange in increasing order of magnitude: P(2) (A40:25 ), P(A40:25 ),
P(A40:25 ).
Problem 163. If 15 P45 = 0.038, P45:15 = 0.056 and A60 = 0.625 find P1
45:15
Problem 164. Use the equivalence principle to find the net annual premium for
a fully discrete 10 year term policy with benefit equal to $10,000 plus the return,
with interest, of the premiums paid. Assume that the interest rate earned on the
premiums is the same as the interest rate used in determining the premium. Use the
life table to compute the premium for this policy for (21). How does this premium
compare with 10000P1 ?
21:10
Problem 165. A level premium whole life insurance of 1, payable at the end of
the year of death, is issued to (x). A premium of G is due at the beginning of each
year provided (x) survives. Suppose L denotes the insurers loss when G = Px , L
denotes the insurers loss when G is chosen so that E[L ] = 0.20, and Var(L) = 0.30.
Compute Var(L ).
Problem 166. A policy issued to (x) has the following features.
(1) Premiums are payable annually.
(2) The first premium is twice the renewal premium.
(3) Term insurance coverage for $100,000 plus the difference between the first
and second premium is provided for 10 years.
(4) An endowment equal to the first year premium is paid at the end of 10 years.
(5) Death claims are paid at the moment of death.
Use the equivalence principle to find an expression for the renewal net annual
premium.
Problem 167. A $1000 whole life policy is issued to (50). The premiums are
payable twice a year. The benefit is payable at the moment of death. Calculate the
semi-annual net premium given that A50 = 0.3 and i = 0.06.
Problem 168. Polly, aged 25, wishes to provide cash for her son Tad, currently
aged 5, to go to college. Polly buys a policy which will provide a benefit in the
form of a temporary life annuity due (contingent on Tads survival) in the amount of
$25,000 per year for 4 years commencing on Tads 18th birthday. Polly will make 10
equal annual premium payments beginning today. The 10 premium payments take
the form of a temporary life annuity due (contingent on Pollys survival). According
16: Net Premiums
87
to the equivalence principle, what is the amount of each premium payment? Use
the life table and UDD assumption (if necessary).
Problem 169. Snow White, presently aged 21, wishes to provide for the welfare
of the 7 dwarfs in the event of her premature demise. She buys a whole life policy
which will pay $7,000,000 at the moment of her death. The premium payments for
the first 5 years will be $5,000 per year. According to the equivalence principle,
what should her net level annual premium payment be thereafter? Use the life table
and UDD assumption (if necessary).
Problem 1610. The Ponce de Leon Insurance Company computes premiums for
its policies under the assumptions that i = 0.05 and x = 0.01 for all x > 0. What
is the net annual premium for a whole life policy for (21) which pays a benefit of
$100,000 at the moment of death and has level premiums payable annually?
16: Net Premiums

Z
Problem 161. If = 0, Ax =
t px x+t dt = 1 and ax =
0
t px
dt = e x .
Problem 162. This is really a question about the present value of annuities,
since the insurance is the same in all cases. The ordering follows from a40:25 <
a (2)
< a 40:25 .
40:25
+ 15 E45 / a 45:15 and
Problem 163. Use the two equations P45:15 = P1
45:15
A45 15 E45 A60
= 15 P45 15 E45 A60 / a 45:15 with the given information.
=
P1
a 45:15
45:15
Problem 164. The present value of the benefit is 10000vK+1 1[0,10) (K) +
pvK+1 sK+1 1[0,10) (K) where p is the premium. The actuarial present value of the
benefit is 10000A1 + pax:10 p 10 Ex a 10 . The equivalence principle gives the
x:10
premium as 10000A1 / a 10 .
x:10
Problem 165. The loss random variable is (1 + G/ d)vK+1 G/ d from which

the mean and variance in the two cases can be computed and compared. In
particular Var(L) = (1 + Px / d)2 Var(vK(x)+1 ) and Var(L ) = (1 + G/ d)2 Var(vK(x)+1 ).
Also (1 + Px / d)E[vK+1 ] Px / d = 0, from which 1 + Px / d = 1/ (Ax 1), while
(1 + G/ d)E[vK+1 ] G/ d = 0.20, from which 1 + G/ d = 1.2/ (Ax 1). This gives
(1 + G/ d)/ (1 + Px / d) = 1.2 and Var(L ) = (1.2)2 (0.30).
Problem 166. If P is the renewal premium then (100000 + P)Ax:10 + 2P10 Ex =
P + Pax:10 .
Problem 167. The annual premium p satisfies pa(2)
50 = 1000A50 . UDD gives
a50 (2), and
A50 = ( / i)A50 , from which a 50 is obtained. Finally, a (2)
50 = (2)
the values of and can be obtained from the interest rate function table.
Problem 168. The premium p satisfies pa25:10 = 2500013|4 a 5 . Also a 25:10 =
a 25 10 E25 a 35 and 13|4 a 5 = a 5 17 E4 a 22 .
Problem 169. The premium p satisfies 7000000A21 = 5000a21:5 + p5 E21 a 26 .
Problem 1610. Here 100000A21 = Pa21 . Now since the force of mortality is
constant,
Z the future lifetime random variable is exponentially distributed so that
A21 =
0
e t 0.01e0.01t dt = 0.01/ ( + 0.01). Under the UDD approximation,
a 21 = (1 A21 )/ d = (1 ( / i)A21 )/ d.
88
16: Net Premiums

Exercise 161. From the table, P30 = A30 / a 30 = 0.10248/ 15.8561.
Exercise 162. Now P1
= A1 / a 30:10 . Also A1
= A30 10 E30 A40 .
30:10
30:10
30:10
Similarly, a 30:10 = a 30 10 E30 a 40 0. The other premium differs only in the
denominator, since P(12)
= A1 / a (12)
. Now a (12)
= a (12)
(12)
1
30 10 E30 a
40 . Since
30:10
30:10
30:10
30:10
a (12)
a30 (12) and a similar expression holds for a (12)
30 = (12)
40 , the value of the
annuity can be computed from the life table using the interest rate function table
as well. Note that the UDD assumption has been used here.
Exercise 163.
h Px
= Ax / a x:h .
Exercise 164. The net monthly premium is P/ 12 where P = (100000A1

75000v44 44 p21 A1
65:10
+ 50000v54 54 p21 A1
75:5
25000v59 59 p21 A1 )/ a (12)

21:64
80:5
21:44
. These
values can be computed from the life table using the techniques of an earlier
exercise.
Exercise 165. Using a n = (1 vn )/ d and a little algebra gives L = (1 +
P/ d)vK(x)+1 P/ d, so that Var(L) = (1 + P/ d)2 Var(vK(x)+1 ).
89
17. Insurance Models Including Expenses

A more realistic view of the insurance business includes provisions for expenses.
The profit for the company can also be included here as an expense.
The common method used for the determination of the expense loaded premium (or the gross premium) is a modification of the equivalence principle. According to the modified equivalence principle the gross premium G is set so that
on the policy issue date the actuarial present value of the benefit plus expenses is
equal to the actuarial present value of the premium income. The premium is usually
assumed to be constant. Under these assumptions a formula to determine G can be
easily written. Assume that the expenses in policy year k are ek1 and are paid at
time k 1, that is, at the beginning of the year. The actuarial present value of the
expenses is then given by
K(x)
X
E[
v ek ] =
k=0
vk ek k px .
k=0
Typically expenses are dependent on the premium. Also the sales commission is
usually dependent on the policy size.
Example 171. Suppose that the first year expenses for a $100,000 semi-continuous
whole life policy are 20% of premiums plus a sales commission equal to 0.5% of
the policy amount, and that the expenses for subsequent years are 10% of premium
plus $5. The gross premium G for such a policy satisfies
100000Ax + (0.20G + 500) + (0.10G + 5)ax = Gax .
An important, and realistic, feature of the above example is the large amount of
first year expense. Expenses are now examined in greater detail.
Example 172. Lets look at the previous example in the case of a policy for a
person aged 21. Assume that the interest rate is 6% and that the life table applies.
Then
100, 000A21 + 495 + 5a21
G=
= $516.76.
0.9a21 0.1
From this gross premium the company must pay $500 in fixed expenses plus 20%
of the gross premium in expenses ($120.85), plus provide term insurance coverage
for the first year, for which the net single premium is 100, 000A1 = $102.97. Thus
21:1
there is a severe expected cash flow strain in the first policy year! The interested
reader may wish to examine the article Surplus Loophole in Forbes, September
4, 1989, pages 44-48.
17: Insurance Models Including Expenses
91
Expenses typically consist of two parts. The first part of the expenses can be
expressed as a fraction of gross premium. These are expenses which depend on
policy amount, such as sales commission, taxes, licenses, and fees. The other part
of expenses consist of those items which are independent of policy amount such
as data processing fees, printing of actual policy documents, clerical salaries, and
mailing expenses.
Studying the gross premium as a function of the benefit provided can be useful.
Denote by G(b) the gross premium for a policy with benefit amount b. The value
G(0) represents the overhead involved in providing the policy and is called the
policy fee. Typically the policy fee is not zero. The ratio R(b) = G(b)/ b is called the
premium rate for a policy of benefit b and reflects (approximately) the premium
change per dollar of benefit change when the benefit amount is b.
Exercise 171. In the example above find R(b), the premium rate for a policy of
benefit b.
92
Problems
Problem 171. The expense loaded annual premium for an 35 year endowment
policy of $10,000 issued to (30) is computed under the assumptions that
(1) sales commission is 40% of the gross premium in the first year
(2) renewal commissions are 5% of the gross premium in year 2 through 10
(3) taxes are 2% of the gross premium each year
(4) per policy expenses are $12.50 per 1000 in the first year and $2.50 per 1000
thereafter
(5) i = 0.06
Find the gross premium using the life table.
Problem 172. A semi-continuous whole life policy issued to (21) has the following
expense structure. The first year expense is 0.4% of the policy amount plus $50. The
expenses in years 2 through 10 are 0.2% of the policy amount plus $25. Expenses in
the remaining years are $25, and at the time of death there is an additional expense
of $100. Find a formula for G(b). Compute G(1) and compare it to A21 .
Problem 173. Your company sells supplemental retirement annuity plans. The
benefit under such a plan takes the form of an annuity immediate, payable monthly,
beginning on the annuitants 65th birthday. Let the amount of the monthly benefit
payment be b. The premiums for this annuity are collected via payroll deduction
at the end of each month during the annuitants working life. Set up expenses for
such a plan are $100. Subsequent expenses are $5 each month during the premium
collection period, $100 at the time of the first annuity payment, and $5 per month
thereafter. Find G(b) for a person buying the plan at age x. What is R(b)?
Problem 174. A single premium life insurance policy with benefits payable at the
end of the year of death is issued to (x). Suppose that
(1) Ax = 0.25
(2) d = 0.05
(3) Sales commission is 18% of gross premium
(4) Taxes are 2% of gross premium
(5) per policy expenses are $40 the first year and $5 per year thereafter
Calculate the policy fee that should be charged.

Problem 171. 10000A30:35 + 0.35G + 0.05Ga30:10 + (0.02G + 25)a30:35 + 100 =
Ga30:35 .
Problem 172. bA21 + 0.002b + 25 + 0.002ba21:10 + 25a21 + 100A21 = G(b)a21 .
(12)
Problem 173. 12b65x| a(12)
a(12)
x +512
x +95+10065+ 1 x Ex = G(b)12ax:65x .
12
Problem 174. G(b) = bAx + 0.18G(b) + 0.02G(b) + 35 + 5ax .
93

Exercise 171. Since the premium is bR(b) when the benefit is b, the modified
equivalence principle gives bAx + (0.20bR(b) + 0.005b) + (0.20bR(b) + 5)ax =
bR(b)ax from which R(b) = (bAx + 0.005b + 5ax )/ b(0.9ax 0.20).
94

Question 181 . Simplify
A.
B.
P1
C.
A40:20
(s40:20 a 40:20 )20 E40

d.

a 40:20 2
D.
a 40:20 A
E.
P40:20
20 E40
1
40:20
40:20
Question 182 . A fully discrete 20 year endowment insurance of 1 is issued to (40).

The insurance also provides for the refund of all net premiums paid accumulated at
interest rate i if death occurs within 10 years of issue. Present values are calculated
at the same interest rate i. Using the equivalence principle, the net annual premium
A
payable for 20 years for this policy is 40:20 . Determine k.
k
A. 10 p40 a 10
B.
a 40:20 (IA)1
D.
a 40:20 a 40:10 + 10 E40 a 10
C.
a 40:10 + s10
E.
a50:10
10 E40 (
40:10
10 E40
+ s10 )
Question 183 . A 20 payment whole life insurance with annual premiums has the
following expenses:
Per Policy
Percent of Premium
First
Year
50
110%
Years
2-10
20
10%
Years
11 and after
20
5%
You are given a x = 16.25, a x:10 = 8.00, and a x:20 = 12.00. Gross premiums are
equal to the expense loaded premium and are expressed as f g + h where f is the rate
per $1 of face amount, g is the face amount, and h is the policy fee. Calculate h.
A. 27.00
B.
29.58
D.
35.50
C.
33.25
E.
39.44
96
Question 184 . For a continuous whole life annuity of 1 on (x), T(x), the future
lifetime of (x), follows a constant force of mortality of 0.06. The force of interest is
0.04. Calculate Pr(aT(x) > ax ).
A. 0.40
B.
0.44
D.
0.48
C.
0.46
E.
0.50
Question 185 . The distribution of Jacks future lifetime is a two point mixture.
With probability 0.60, Jacks future lifetime follows the Illustrative Life Table, with
deaths uniformly distributed over each year of age. With probability 0.40, Jacks
future lifetime follows a constant force of mortality = 0.02. A fully continuous
whole life insurance of 1000 is issued on Jack at age 62. Calculate the benefit
premium for this insurance at i = 0.06.
A. 31
B.
32
D.
34
C.
33
E.
35
Question 186 . For a whole life annuity due of 1 on (x), payable annually qx = 0.01,
qx+1 = 0.05, i = 0.05, and a x+1 = 6.951. Calculate the change in the actuarial present
value of this annuity due if px+1 is increased by 0.03.
A. 0.16
B.
0.17
D.
0.19
C.
0.18
E.
0.20
97
Question 187 . Company ABC issued a fully discrete three year term insurance of
1000 on Pat whose stated age at issue was 30. You are given q30 = 0.01, q31 = 0.02,
q32 = 0.03, q33 = 0.04, and i = 0.04. Premiums are determined using the equivalence
principle. During year 3 Company ABC discovers that Pat was really age 31 when
the insurance was issued. Using the equivalence principle, Company ABC adjusts
the death benefit to the level death benefit it should have been at issue, given the
premium charged. Calculate the adjusted death benefit.
A. 646
B.
664
D.
750
C.
712
E.
963
Question 188 . The pricing actuary at Company XYZ sets the premium for a fully
continuous whole life insurance of 1000 on (80) using the equivalence principle
and the assumptions that the force of mortality is 0.15 and i = 0.06. The pricing
actuarys supervisor believes that the Illustrative Life Table with deaths uniformly
distributed over each year of age is a better mortality assumption. Calculate the
insurers expected loss at issue if the premium is not changed and the supervisor is
right.
A. 124
B.
26
D.
37
C.
E.
220
98

Question 181 . Using 20 E40 s40:20 = a 40:20 and da 40:20 = 1 A40:20 gives
(s40:20 a 40:20 )20 E40
d = (1 20 E40 )/ a 40:20 d

a 40:20 2
= (1 20 E40 da 40:20 )/ a 40:20
= (A40:20 20 E40 )/ a 40:20
= A1
/ a 40:20
= P1
40:20
40:20
B.
Question 182 . The equivalence principle gives
Pa40:20 = A40:20 + E[PsK(40)+1 vK(40)+1 1[0,9] (K(40))].
Now the expectation is equal to P(a40:10 a 10 10 p40 . Thus
k = a 40:20 a 40:10 + a 10 10 p40
= v10 10 p40 a 50:10 + a 10 10 p40
= 10 E40 (a50:10 + s10 ).
E.
Question 183 . The equation for the gross premium P is gAx +302 0ax +0.05Pax:20 +
0.05ax:10 +P = Pax:20 , from which h = (30+20ax )/ (ax:20 10.05ax:10 0.05ax:20 ) =
35.5. D.
Question 184 . Here
Z
aT =
Z
and ax =
e0.04t dt = (1 e0.04T )/ 0.04,
e0.04t e0.06t dt = 10. Thus P[aT > 10] = P[T > ln(0.6)/ 0.04] =
e.06 ln(.6)/ .04 = 0.46. C.

Question 185 . Here the premium P satisfies
Z
0.6(1000)(i/ )A62 + 0.4(1000)

0.02e t0.02t dt
= P 0.6(1 (i/ )A62 )/ + 0.4

from which P = 31. A.
( +0.02)t
e
0
dt ,
99
Question 186 . Since a x = 1 + vpx a x+1 = 1 + vpx + v2 px px+1 a x+2 and a x+1 = 1 +
vpx+1 a x+2 , the given information shows a x+2 = 6.577. The original value is a x = 7.553
while the modified value is 7.730. The difference is 0.177. C.
Question 187 . Using the information and direct computation gives the original
premium as 18.88. The adjusted benefit is therefore 664, again by direct computation. B.
Question 188 . For the original calculations, a = 1/ ( + ) and A = 1 a, so
the premium charged is 150. Using the Illustrative Life Table gives the loss at issue
with this premium as 1000A 150a = 124. A.
19. Multiple Lives

The study of the basic aspects of life insurance is now complete. Two different
but similar directions will now be followed in the ensuing sections. On the one
hand, types of insurance in which the benefit is paid contingent on the death or
survival of more than one life will be examined. On the other hand, the effects of
competing risks on the cost of insurance will be studied.
The first area of study will be insurance in which the time of the benefit payment
depends on more than one life. Recall that a status is an artificially constructed
life form for which there is a definition of survival and death (or decrement). The
simplest type of status is the single life status. The single life status (x) dies exactly
when (x) does. Another simple status is the certain status n . This status dies at
the end of n years. The joint life status for the n lives (x1 ), . . . (xn ) is the status
which survives until the first member of the group dies. This status is denoted by
(x1 x2 . . . xn ). The last survivor status, denoted by (x1 x2 . . . xn ) is the status which
survives until the last member of the group dies.
When discussing a given status the question naturally arises as to how one would
issue insurance to such a status. If the constituents of the status are assumed to die
independently this problem can be easily solved in terms of what is already known.
Example 191. Consider a fully discrete whole life policy issued to the joint status
(xy). The net annual premium to be paid for such a policy is computed as follows.
Using the obvious notation, the premium, P, must satisfy
Axy = Paxy .
Using the definition of the joint life status gives
Axy = E[vK(x)K(y)+1 ]
and
a xy =
1 Axy
d
which are obtained as previously.

Exercise 191. Obtain an expression for Axy in terms that can be computed from
the life table.
Exercise 192. Is Axy + Axy = Ax + Ay ?
A useful technique for writing computational formulas directly is to ask the
question Under what conditions is a payment made at time t? The answer will
usually provide a computational formula.
19: Multiple Lives
101
Example 192. What is a xy ? This annuity makes a payment of 1 at time k if and

only if both (x) and (y) are alive at time k, and the probability of this is k pxy = k px k py .
Thus a xy =
vk k px k py .
k=0
If one is willing to assume an analytical law of mortality computations involving

joint lives can be simplified. Recall that two of the common analytical laws of
mortality are the Gompertz and Makeham laws. The Gompertz law has force of
mortality x = Bcx . The joint survival of two independent lives (x) and (y) is
probabilistically identical with the survival of a single life (w) if and only if
(xy)+s = x+s + y+s = w+s .
When (x) and (y) have mortality which follows Gompertz Law this relation holds if
w satisfies cx + cy = cw . A similar observation applies to Makehams law for which
the force of mortality is x = A + Bcx . In this case, however, mimicing the joint life
(xy) requires the use of a joint life (ww) at equal ages. Here w is the solution of
2cw = cx + cy .
Exercise 193. Verify these assertions.
A status can also be determined by the order in which death occurs. The idea
1
here is similar to that used for term insurance earlier in which the status x : n fails
at the time of death of (x) provided (x) dies first. As a more complicated example
2
the status (x : y) dies at the time of death of (y) provided (y) is the second to die.
Hence this status lives forever if (y) dies before (x). An insurance for such a status
is a simple case of what is known as a contingent insurance. Again, if the lives are
assumed to fail independently the computations can be reduced to those involving
cases already considered.
Exercise 194. Show that if X and Y are independent random variables and one
of them is absolutely continuous then P[X = Y] = 0. Hence under the standard
assumptions of this section no two people can die simultaneously.
One model for joint lives which allows for simultaneous death is the common
shock model. The intuition is that the two lives behave almost independently except
for the possibility of death by a common cause. The model is as follows. Let T (x),
T (y), and Z be independent random variables. Assume that T (x) and T (y) have
the distribution of the remaining lifetimes of (x) and (y) as given by the life table.
The random variable Z represents the time of occurrence of the common catastrophe
which will kill any survivors. The common shock model is that the true remaining
lifetimes of (x) and (y) are given as T(x) = min{T (x), Z} and T(y) = min{T (y), Z}
respectively. The key computational fact is that under the common shock model
t pxy = t px t py P[Z t].
19: Multiple Lives
102
Exercise 195. What is the probability that (x) and (y) die simultaneously in this
model?
For the special case in which T (x), T (y), and Z have exponential distributions
with parameters x , y , and z respectively, computations for the common shock
model are relatively easy, and will be explored in the problems.
19: Multiple Lives
103
Problems
Problem 191. Show
t pxy
= t pxy + t px (1 t py ) + t py (1 t px ).
Problem 192. Suppose x = 1/ (110 x) for 0 x < 110. Find

and e 20:30 .
10 p20:30 , 10 p20:30 ,
Problem 193. Find an expression for the actuarial present value of a deferred
annuity of $1 payable at the end of any year as long as either (20) or (25) is living
after age 50.
Problem 194. Find the actuarial present value of a 20 year annuity due which
provides annual payments of $50,000 while both (x) and (y) survive, reducing by
25,000 on the death of (x) and by 12,500 on the death of (y).
Problem 195. Show that n q1xy = n qxy2 + n qx n py .
Problem 196. Show that A1xy Axy2 = Axy Ay .
Problem 197. If x = 1/ (100 x) for 0 x < 100, calculate 25 q
2 .
25:50
Problem 198. If the probability that (35) will survive for 10 years is a and the
probability that (35) will die before (45) is b, what is the probability that (35) will
die within 10 years after the death of (45)? Assume the lives are independent.
Problem 199. Suppose that in the common shock model T (x), T (y), and Z have
exponential distributions with parameters x , y , and z respectively. Find the net
single premium for a continuous whole life policy of 1 on the joint life (xy). Assume
the force of interest is > 0.
Problem 1910. Find an expression for the net single premium for a continuous
1
whole life policy of 1 issued to (xy), a status which fails when (x) dies if T(x) < T(y).
Problem 1911. Find an expression for the net single premium for a continuous
2
whole life policy issued to (xy), where the benefit is paid on the death of (y) if
T(x) < T(y).
19: Multiple Lives

Problem 191. t pxy = P[[T(x) t] [T(y) t]] = P[T(x) t, T(y)
t] + P[T(x) t, T(y) t] + P[T(x) t, T(y) t].
Problem 192. From the form of the force of mortality, DeMoivres Law holds,
so 10 p20:30 = 10 p20 10 p30 = (80/ 90)(70/ 80). Also, 10 p20:30 = 1 10 q20 10 q30 , and
Z
Z 80
90 t 80 t
e 20:30 =
p
dt
=
dt.
t 20:30
90
80
0
0
Problem 193.
30| a20
+ 25| a25 30| a20:25 .
Problem 194. The annuity pays 12,500 for 20 years no matter what so the
actuarial present value consists of 3 layers giving 25, 000ax:20 + 12, 5000ay:20 +
12, 500a20 .
Problem 195. The event that (x) dies first and within n years occurs if (y)
dies second within n years (so that both die) or (x) dies within n years and (y)
survives n years.
Problem 196. If (x) dies before (y), the insurance on the left side pays 1 at the
death of (x) and takes back 1 at the death of (y); the insurance on the right side
does the same. If (y) dies before (x) the insurance on the left side pays nothing,
and neither does the insurance on the right side.
Problem 197. DeMoivres Law holds and a picture shows that the probability
is the area of a triangle, which is (1/ 2)252 / (50)(75) = 1/ 12.
R
198. P[T(35) > T(45) + 10] = 0 P[T(35) > t +R10]t p45 45+t dt =
RProblem
10]P[T(35) 10]t p45 45+t dt = a 0 P[T(45)+10 >

0 P[T(35) > t+10| T(35)
R
R
t + 10]t p45 45+t dt = a 0 (t p45 )2 45+t dt = 0 t p45 dtd t p45 dt = a/ 2. Thus the
desired probability is 1 a/ 2 b.
Problem 199. In this case T(xy) has an exponential distribution with parameter
x + y + z , so the net single premium is (x + y + z )/ (x + y + z + ).
Z
Problem 1910. The net single premium is
vt t px x+tt py dt.
Z
Problem 1911. The premium is
0
vt t qxt py y+t dt.
104
19: Multiple Lives

Exercise 191.
P Using the independence
P gives t pxy = t pxt py , so that Axy =
E[vK(xy)+1 ] = k=0 vk+1 (k pxy k+1 pxy ) = k=0 vk+1 (k pxk py k+1 pxk+1 py ).
Exercise 192. Intuitively, either (x) dies first or (y) dies first, so the equation
is true. This can be verified by writing the expectations in terms of indicators.
Exercise 193. Under Makeham the requirement is that (A+Bcx+s )+(A+Bcy+s ) =
(A + Bcw+s ) + (A + Bcw+s ) for all s, and this holds if cx + cy = 2cw .
Z
Exercise 194.
If X is absolutely continuous, P[X = Y] =
P[Y =
x] fX (x) dx. This probability is zero since P[Y = x] is non zero for at most
countably many values of x.
Exercise 195. In the common shock model, (x) and (y) die simultaneously if
T (x) > Z and T (y) >Z Z. By conditioning on the value of Z the probability of
simultaneous death is
P[T (x) > z]P[T (y) > z]fZ (z) dz when Z is absolutely
continuous. Here the independence of T (x) and T (y) was used.
105
20. Multiple Decrement Models

In contrast to the case in which a status is defined in terms of multiple lives, the
way in which a single life fails can be studied. This point of view is particularly
important in the context of the analysis of pension plans. In such a setting a person
may withdraw from the workforce (a death) due to accident, death, or retirement.
Different benefits may be payable in each of these cases. A common type of
insurance in which a multiple decrement model is appropriate is in the double
indemnity life policy. Here the benefit is twice the face amount of the policy if
the death is accidental. In actuarial parlance the termination of a status is called a
decrement and multiple decrement models will now be developed. These models
also go by the name of competing risk models in other contexts.
To analyze the new situation, introduce the random variable J = J(x) which is a
discrete random variable denoting the cause of decrement of the status (x). Assume
that J(x) has as possible values the integers 1, . . . , m. All of the information of
interest is contained in the joint distribution of the random variables T(x) and J(x).
Note that this joint distribution is of mixed type since (as always) T(x) is assumed
to be absolutely continuous while J(x) is discrete. The earlier notation is modified
in a fairly obvious way to take into account the new model. For example,
(j)
t qx
= P[0 < T(x) t, J(x) = j].
and
(j)
t px
= P[T(x) > t, J(x) = j].
Here q(j)
x gives the marginal density of J(x). To discuss the probability of death due
to all causes the superscript () is used. For example,
()
t qx
= P[T(x) t] =
m
X
(j)
t qx
j=1
()
and a similar expression for the survival probability holds. Although t q()
x + t px = 1
a similar equation for the individual causes of death fails unless m = 1. For the
force of mortality from all causes
()
x+t
=
fT(x) (t)
,
P[T(x) > t]
as before, and the force of mortality due to cause j is

(j)
x+t
=
fT(x) J(x) (t, j)

.
P[T(x) > t]
The force of mortality due to cause j represents the instantaneous death rate in an
imaginary world in which cause j is the only possible cause of death. For this reason,
and also directly from the defining formula, x() =
m
X
j=1
x(j) .
20: Multiple Decrement Models
107
Care must be exercised in the use of these formulas. In particular, while

Z
()
t px = exp{
the inequality
Z
(j)
t px exp{
()
ds}
x+s
(j)
x+s
ds}
is generally true. This latter integral does have an important use which is explored
below.
An important practical problem is that of constructing a multiple decrement life
table. To see how such a problem arises consider the case of a double indemnity
whole life policy. Assume that the policy will pay an amount $1 at the end of the
year of death if death occurs due to non-accidental causes and an amount of $2 if
the death is accidental. Denote the type of decrement as 1 and 2 respectively. The
present value of the benefit is then
vK(x)+1 1{1} (J(x)) + 2vK(x)+1 1{2} (J(x)) = J(x)vK(x)+1 .
To compute the net premium the expectation of this quantity must be computed.
This computation can only be completed if p(j)
x is known. How are these probabilities
calculated?
There are two basic methodologies used. If a large group of people for which
extensive records are maintained is available the actual survival data with the deaths
in each year of age broken down by cause would also be known. The multiple
decrement table can then be easily constructed. This is seldom the case.
Example 201. An insurance company has a thriving business issuing life insurance
to coal miners. There are three causes of decrement (death): mining accidents, lung
disease, and other causes. From the companys vast experience with coal miners a
decrement (life) table for these three causes of decrement is available. The company
now wants to enter the life insurance business for salt miners. Here the two causes of
decrement (death) are mining accidents and other. How can the information about
mining accidents for coal miners be used to get useful information about mining
accidents for salt miners?
A simple-minded answer to the question raised in the example would be to
simply lift the appropriate column from the coal miners life table and use it for the
salt miners. Such an approach fails, because it does not take into account the fact
that there are competing risks, that is, the accident rate for coal miners is affected by
the fact that some miners die from lung disease and thus are denied the opportunity
to die from an accident. The death rate for each cause in the absence of competing
risk is needed.
108
To see how to proceed the multiple decrement process is examined in a bit more
detail. As mentioned earlier, x(j) , the instantaneous death rate due to cause j, is the
death rate in an imaginary world in which cause j is the only cause of decrement.
Intuitively, this is because a person can not simultaneously die of 2 causes. Thus on
an instantaneous basis, there is no competing risk. This fact leads to the introduction
of the quantities
Z
t
(j) = exp{
t px
and
(j)
ds}
x+s
(j) = 1 t px(j) .
t qx
The probability t qx(j) is called the net probability of decrement (or absolute rate
of decrement). The absolute rate of decrement t qx(j) is the probability of decrement
for (x) within t years in the imaginary world in which cause j is the only cause of
decrement.
These probabilities may be used to obtain the desired entries in a multiple
decrement table as follows. First
()
t px =
m
Y
(j)
t px
j=1
since, as remarked earlier, x() =
m
X
x(j) . This shows how one can pass from the
j=1
absolute rate of decrement to total survival probabilities. Note that this relationship
implies that the rates are generally larger than the total survival probability.
Connecting the rates of decrement to the entries in the multiple decrement
table can be accomplished under several differnet types of assumptions. As a first
illustration, suppose that the force of mortality for each decrement over each year
(j)
of age in the multiple decrement table is constant. This means that x+t
= x(j) for
()
0 < t < 1 and 1 j m. Consequently, x+t
= x() too. So
Z
qx(j) =
=
0
()
s px
(j)
x+s
ds
()
s px
x(j) ds
x(j) Z 1 () ()
= ()
s px x ds
x 0
x(j) ()
= ()
qx
x
ln px(j) ()
=
qx
ln p()
x

or briefly,
px = p()
x
(j)
109
()
q(j)
x / qx
From this relation the rates of decrement can be found if the multiple decrement
table is given. Conversely, if all of the rates of decrement are known, the multiple
decrement table can be constructed by using this relation and the fact that 1 q()
x =
px()
m
Y
px(j) .
j=1
This solves the problem of computing the entries in a multiple decrement table
under the stated assumption about the structure of the causes of decrement in that
table.
Exercise 201. What happens if p()
x = 1?
Exercise 202. Show that the same formula results if one assumes instead that the
time of decrement due to each cause of decrement in the multiple decrement table
has the uniform distribution over the year of decrement. (This assumption means
(j)
that t q(j)
x = t qx .)
Exercise 203. Assume that two thirds of all deaths at any age are due to accident.
What is the net single premium for (30) for a double indemnity whole life policy?
How does this premium compare with that of a conventional whole life policy?
The previous computations were based on assumptions about the causes of
decrement within the multiple decrement table. In some contexts it is more sensible
to make assumptions about the structure of the individual causes of decrement as
if each were acting independently, that is, to make assumptions about the absolute
rate of decrement in the single decrement tables.
Example 202. Suppose we are designing a pension plan and that there are two
causes of decrement: death and retirement. In many contexts (such as teaching)
retirements might be assumed to all occur at the end of a year, while deaths can occur
at any time. How could we construct a multiple decrement table which reflects this
assumption?
The key observation here is that the force of mortality due to cause j can
be computed in two ways: within the multiple decrement table, and within the
imaginary world in which cause j is the only cause of decrement. The results of
these two computations must be equal, since on an instantaneous basis there are no
competing risks. Thus
d (j)
d (j)
s qx
s q
(j)
ds
ds x .
=
=
x+s
()
(j)
s px
s px
110
This key relationship, together with assumptions about mortality within the single
decrement table, is sufficient to find the connection between the rates of decrement
and the multiple decrement table.
One common assumption about a single decrement is the assumption of uniform
distribution of deaths in the year of death. In the multiple decrement context this
translates into the statement that for 0 t 1
(j) = t qx(j) .
t qx
Using this assumption together with the key relationship above permits the following
type of computation. The computations are illustrated for the case of 2 causes of
decrement. In this setting
Z
q(1)
x =
() (1)
s px x+s
=
1
=
0
(1)
ds
(1) s px(2) x+s
s px
ds
d (1)
s qx
(1)
(2) ds
s px s px
(1) ds
s px
= qx(1)
= qx(1)
(2) ds
s px
0
1
(1 sqx(2) ) ds
1
= qx(1) (1 qx(2) )
2
with a similar formula for q(2)
x . This procedure could be modified for different
assumptions about the decrement in each single decrement table.
Exercise 204. Construct a multiple decrement table in which the first cause of
decrement is uniformly distributed and the second cause has all decrements occur
at the end of the year. The pension plan described in the example above illustrates
the utility of this technique.
Another approximation which is used to connect single and multiple decrement
tables makes use of the life table functions
Z
Lx =
mx =
lx+t dt
lx lx+1
Lx
which are sometimes used in the single decrement case. Intuitively, Lx is the number
of person years lived by those dying with age between x and x + 1. Hence Lx is a
111
weighted average of lx and lx+1 with the weights determined by the pattern of death
in that year of age. So mx = dx / Lx is perhaps a more reasonable estimate of qx than
some other measures which have been used. The function mx is called the central
death rate at age x.
In the context of a multiple decrement table the central rate of death is used in
a special technique, called the central rate bridge. This technique is now briefly
described. Define
Z 1
() ()
t px x+t dt
q()
x
0
m()
=
=
Z
Z
x
1
1
()
()
t px dt
t px dt
0
and
m(j)
x
() (j)
t px x+t
and
mx =
(j)
dt
q(j)
x
=Z
()
t px
dt
0
(j)
dt
(j) x+t
t px
Z
0
t px
(j)
=Z
dt
0
()
t px
dt
qx(j)
t px
(j)
.
dt
The central rate bridge is based on the following approximation. First, under the
UDD assumption in each single decrement table
mx(j) =
qx(j)
.
1 12 qx(j)
Second, under the UDD assumption in the multiple decrement table

m(j)
x =
q(j)
x
.
1 12 q()
x
Thirdly, under the constant force assumption in the multiple decrement table
(j)
(j)
m(j)
x = x = mx .
Now assume that all of these equalities are good approximations in any case. This
assumption provides a way of connecting the single and multiple decrement tables.
There is no guarantee of the internal consistency of the quantities computed in this
way, since, in general, the three assumptions made are not consistent with each
other. The advantage of this method is that the computations are usually simpler
than for any of the exact methods.
Exercise 205. Show that each of the above equalities hold under the stated assumptions.
112
Problems
Problem 201. Assume that each decrement has a uniform distribution over each
year of age in the multiple decrement table to construct a multiple decrement table
from the following data.
qx(1)
0.020
0.022
0.028
Age
62
63
64
qx(2)
0.030
0.034
0.040
qx(3)
0.200
0.100
0.120
Problem 202. Rework the preceding exercise using the central rate bridge. How
different is the multiple decrement table?
Problem 203. In a double decrement table where cause 1 is death and cause 2 is
withdrawal it is assumed that deaths are uniformly distributed over each year of age
while withdrawals between ages h and h + 1 occur immediately after attainment of
()
(1)
(2)
age h. In this table one sees that l50
= 1000, q(2)
50 = 0.24, and d50 = 0.06d50 . What
(1)
is q50 ? How does your answer change if all withdrawals occur at midyear? At the
end of the year?
Problem 204. The following data was collected from independent samples at
various stages of the development of the apple maggot.
Development
Stage x
0 (egg)
1 (early larvae)
2 (late larvae)
3 (early pupae)
4 (late pupae)
Number
Observed
977
963
153
435
351
1 (predator)
0
0
65
88
78
Number of Deaths by Cause

2 (parasite)
3 (disease)
0
224
12
143
45
687
126
0
10
19
4 (other)
14
87
0
0
54
Use the given data to construct a multiple decrement life table using l0 = 10, 000.
Problem 205. Refer to the preceding problem and find the absolute rate of decrement for each stage of development and each cause of decrement.
Problem 206. Refer to the previous problem and construct the multiple decrement
table that would hold if death by disease were completely eliminated.
Problem 207. How would you construct a multiple decrement table if you were
given qx(1) , qx(2) , and q(3)
x ? What assumptions would you make, and what formulas
(3)
would you use? What if you were given qx(1) , q(2)
x , and qx ?

()
()
First, p()
(j) =
62 = (.98)(.97)(.80) and q62 = 1 p62 . Also p62
ln p (j)
1q62
(j) . From the relation q(j)62 = 62() q()
62 the first row of the multiple decrement
ln p62
table can be found.
Problem 201.
(2)
(1)
Problem 203. From the information d50
= 240 and d50
= 14. Since withdrawals occur at the beginning of the year there are 1000 240 = 760 people
under observation of whom 14 die. So q50(1) = 14/ 760. If withdrawals occur at
year end all 1000 had a chance to die so q50
(1) = 14/ 1000.
Problem 204.
follows.
x
0 (egg)
1 (early larvae)
2 (late larvae)
3 (early pupae)
4 (late pupae)
5 (adult)
The life table constructed from the given information is as
lx
10000.00
2824.97
1543.03
766.47
341.83
150.95
dx
7175.03
1281.95
776.56
424.64
190.88
dx(1)
0.00
0.00
655.54
155.06
75.96
dx(2)
0.00
657.11
121.02
251.97
43.82
dx(3)
7031.73
369.62
0.00
17.62
18.50
dx(4)
143.30
255.22
0.00
0.00
52.59
q(2)
x
0.0000
0.2326
0.0784
0.3287
0.1282
q(3)
x
0.7032
0.1308
0.0000
0.0230
0.0541
q(4)
x
0.0143
0.0903
0.0000
0.0000
0.1538
This can also be given as follows.

x
0 (egg)
1 (early larvae)
2 (late larvae)
3 (early pupae)
4 (late pupae)
px
0.2825
0.5462
0.4967
0.4460
0.4416
qx
0.7175
0.4538
0.5033
0.5540
0.5584
q(1)
x
0.0000
0.0000
0.4248
0.2023
0.2222
Problem 205. Under the UDD assumption in the multiple decrement table,
the absolute rates of decrement are as follows.
x
0 (egg)
1 (early larvae)
2 (late larvae)
3 (early pupae)
4 (late pupae)
px
0.2825
0.5462
0.4967
0.4460
0.4416
qx
0.7175
0.4538
0.5033
0.5540
0.5584
q x
0.0000
0.0000
0.4460
0.2554
0.2777
(1)
q x
0.0000
0.2665
0.1033
0.3807
0.1711
(2)
q x
0.7103
0.1600
0.0000
0.0330
0.0762
(3)
q x
0.0249
0.1134
0.0000
0.0000
0.2016
(4)
Problem 206. Under UDD in the multiple decrement table, after setting the
absolute rate of decrement for disease to zero, the multiple decrement table
would be as follows.
x
0 (egg)
1 (early larvae)
2 (late larvae)
3 (early pupae)
4 (late pupae)
px
0.9751
0.6503
0.4967
0.4612
0.4780
qx
0.0249
0.3497
0.5033
0.5388
0.5220
q(1)
x
0.0000
0.0000
0.4248
0.2053
0.2300
q(2)
x
0.0000
0.2519
0.0784
0.3336
0.1327
q(3)
x
0.0000
0.0000
0.0000
0.0000
0.0000
q(4)
x
0.0249
0.0978
0.0000
0.0000
0.1593
Problem 207. The central rate bridge could be used. Is there an exact method
available?
113

Exercise 201. What would this mean for x() and the derivation?
(j)
()
Exercise 202. The assumption is that t q(j)
x = tqx for all j. Hence t px =
(j)
(j)
(j)
1 tq()
= d q / p() = qx / s p()
x and
x . Substitution and integration gives
R 1 (j) x+s ds s x s x
(j) ()
x+s ds
(j)
()
q
/
q
()
= (1 qx ) x x . Since p()
px = e 0
x = 1 qx , the result follows by
substitution.
Exercise 203. The actuarial present value of the benefit is (1/ 3) 1 Ax +

(2/ 3) 2 Ax = (5/ 3)Ax , from which the premium is easily calculated.
R
(2)
(1) 1
Exercise 204. Since cause 1 obeys UDD, q(1)
x = qx
0 s px ds as in the
(2)
(1)
(1)
derivation above. For cause 2, s px = 1 for s < 1, so qx = qx . For cause 2
R 1 (1) d (2)
proceed as in the derivation above to get q(2)
ds. Now s qx (2) is
x = 0 s px ds s qx
(2)
(2)
(2) (1)
0 except for a jump of size qx at s = 1. Hence qx = qx px = qx (2) (1 qx (1) ).
Notice that the integral has been interpreted as a Stieltjes integral in order to take
the jump into account.
Exercise 205. Under UDD in the single decrement table t px(j) = 1 tqx(j) and
R 1 (j)
R1
(j)
(j)
(j)
(j)
(j) (j)
1 (j)
t px x+t = qx so mx = 0 qx dt/ 0 (1 tqx ) dt = qx / (1 2 qx ). Under
(j)
()
UDD in the multiple decrement table x+s
= q(j)
x / s px so that substitution gives
the result. Under the constant force assumption in the multiple decrement table,
(j)
(j)
(j)
x+s
= x(j) for all j and m(i)
x = x = mx by substitution.
114

Question 211 . A life insurance on John and Paul pays
(1) 1 at the death of John if Paul is alive
(2) 2 at the death of Paul if John is alive
(3) 3 at the death of John if Paul is dead
(4) 4 at the death of Paul if John is dead.
John and Paul are independent risks, both age x. Calculate the actuarial present
value of the insurance provided.
A. 7Ax 2Axx
B.
7Ax 4Axx
D.
10Ax 4Axx
C.
10Ax 2Axx
E.
10Ax 5Axx
Question 212 . You are given lxy = lx ly and dxy = lxy lx+1:y+1 . Which of the
following is equivalent to dxy dx dy ?
"
#
px qy + qx py
A. lx+1 ly+1
px py
"
B.
lx+1 ly+1
"
C.
lx ly
px qy + qx py
qx qy
px qy + qx py
px py
#
"
px qy + qx py
qx qy
D.
lx ly
E.
lx ly px qx + qy py
116
Question 213 . In a double decrement table you are given

x
25
26
q(1)
x
0.01
0.01
q(2)
x
0.15
0.10
lx()
8,400
(1)
Calculate the effect on d26
if q(2)
25 changes from 0.15 to 0.25.
A. decrease by 10
B.
decrease by 5
D.
increase by 10
C.
increase by 5
E.
increase by 15
Question 214 . Given the following data from a double decrement table, calculate
(2)
d65
.
()
(1) l63
= 500
(1)
(2) q63 = 0.050
(3) q(2)
63 = 0.500
(4) 1| q(1)
63 = 0.070
(5) 2| q(1)
63 = 0.042
(2)
(6) 2 q63 = 0.600
()
(7) l66
=0
A.
100
B.
105
D.
114
C.
109
E.
119
Question 215 . A multiple decrement table has 2 decrements, death (d) and
withdrawal (w). Withdrawals occur once a year three-fourths of the way through
the year of age. Deaths in the associated single decrement table are uniformly
distributed over each year of age. You are given lx() = 1000, qx(w) = 0.2, and
()
lx+1
= 720. Calculate dx(d) .
A. 80
B.
83
D.
93
C.
90
E.
95
117
Question 216 . In a triple decrement table, lives are subject to decrements of death
(d), disability (i), and withdrawal (w). The total decrement is uniformly distributed
()
(w)
= 23, 000, m(d)
over each year of age, lx() = 25, 000, lx+1
x = 0.02, and mx = 0.05.
Calculate q(i)
x , the probability of decrement by disability at age x.
A. 0.0104
B.
0.0112
D.
0.0128
C.
0.0120
E.
0.0136
()
Question 217 . In a double decrement table, l30
= 1000, q30(1) = 0.100, q30(2) =
(1)
()
(2)
0.300, 1| q30 = 0.075, and l32 = 472. Calculate q31 .
A. 0.11
B.
0.13
D.
0.15
C.
0.14
E.
0.17
Question 218 . For a last survivor insurance of 10,000 on independent lives (70)
and (80) you are given that the benefit, payable at the end of the year of death, is
paid only if the second death occurs during year 5. Mortality follows the Illustrative
Life Table, and i = 0.03. Calculate the actuarial present value of this insurance.
A. 235
B.
245
D.
265
C.
255
E.
275
Question 219 . For a last survivor whole life insurance of 1000 on (x) and (y)
the death benefit is payable at the moment of the second death. The independent
random variables T (x), T (y), and Z are the components of a common shock model.
T (x) has an exponential distribution with xT (x) (t) = 0.03 for t > 0. T (y) has an
exponential distribution with yT (y) (t) = 0.05 for t > 0. Z the common shock random
variable, has an exponential distribution with Z (t) = 0.02 for t > 0. The force of
interest is = 0.06. Calculate the actuarial present value of this insurance.
A. 0.216
B.
0.271
D.
0.368
C.
0.326
E.
0.423
118

Question 211 . The present value of benefit random variable is vJ 1J<P + 2vP 1P<J +
3vJ 1P<J + 4vP 1J<P = 3vJ + 4vP 2vJ 1J<P 2vP 1P<J . The expectation is thus 7Ax 2Axx .
A.
Question 212 . From the given information, dxy dx dy = lx ly lx+1 ly+1 dx dy =
(lx+1 +dx )(ly+1 +dy )lx+1 ly+1 dx dy = lx+1 dy +ly+1 dx = lx+1 ly qy +ly+1 lx qx = lx+1 ly+1 (qy / py +
qx / px ).A.
(1)
()
Question 213 . With the current table d26
= 84, and l25
(1 .01 .15) = 8400 so
()
()
l25 = 10000. With the change, l26 = 10000(1 .01 .25) = 7400 giving the new
(1)
value of d26
= 74. A.
()
Question 214 . The first 3 facts give l64
= 500 25 250 = 225. The fourth
() (1)
(1)
() (2)
fact gives 0.07 = p63 q64 , so d64 = 35. The sixth fact gives 0.60 = q()
63 + p63 q64
(2)
and using what has been determined, d64 = 50 so that l65 = 140. The fifth fact
() ()
(1)
gives 0.042 = q(1)
65 p64 p63 , from which d65 = 21, from which the last fact gives
(2)
d65
= 140 21 = 119. E.
(d) (w)
(d)
Question 215 . Since p()
= .9. Thus
x = px px , the information given yields px
under the UDD assumption, there are 75 deaths in the first three-fourths of the year,
leaving 925 alive of which 20% withdraw. Thus there are 740 left alive after the
withdrawals, of which 20 must die since lx+1 = 720. Thus there were 95 deaths. E.
()
()
(j)
Question 216 . Under the UDD assumption, m(j)
x = qx / (1 0.5qx ). Since qx =
(w)
(d)
()
2000/ 25000 = 0.08, q(i)
x = qx qx qx = 0.08 0.96(0.02) 0.96(0.05) = 0.0128.
D.
()
Question 217 . From the information, p()
30 = (.9)(.7) = .63 so that l31 = 630.
(1)
() (1)
(1)
(1)
(2)
Now 1| q30 = p30 q31 so that q31 = .075/ .63. Thus d31 = 75 and d31 = 83. Hence
q(2)
31 = 83/ 630 = 0.131. B.
Question 218 . The benefit is paid if (70) dies in year 5 and (80) dies in year 4 or
before, or if (80) dies in year 5 and (70) dies in year 5 or before. This probability is
5664051 47.31
2660734
2660734 113.69
5396001
(1
)+
(1
) = 0.27223.
6616155 1000
3914365
3914365 1000
6616155
Multiplication of the probability by 10, 000v5 gives the value as 234.83. A.
Question 219 . The value of a last survivor policy is the sum of the individual
policies minus the value of a joint survivor policy. Now T(x) has an exponential
distribution with parameter .05; T(y) is exponential with parameter 0.07; and the
joint life is exponential with parameter 0.10. The desired value is thus .05/ .11 +
.07/ .13 .1/ .16 = 0.368. D.
22. Insurance Company Operations

The discussion thus far has been about individual policies. In the next few
sections the operations of the company as a whole are examined. This examination
begins with an overview of the accounting practices of an insurance company. This
is followed by a study of the behavior of the loss characteristics of groups of similar
policies. This last study leads to another method of setting premiums for a policy.
23. Net Premium Reserves

A realistic model for both insurance policies and the method and amount of
premium payment is now in hand. The next question is how accounting principles
are applied to the financial operations of insurance companies.
A basic review of accounting principles is given first. There are three broad
categories of items for accounting purposes: assets, liabilities, and equity. Assets
include everything which is owned by the business. Liabilities include everything
which is owed by the business. Equity consists of the difference in the value of the
assets and liabilities. Equity could be negative. In the insurance context liabilities
are referred to as reserve and equity as surplus. When an insurance policy is
issued the insurance company is accepting certain financial obligations in return for
the premium income. The basic question is how this information is reflected in
the accounting statements of the company. Keep in mind that this discussion only
concerns how the insurance company prepares accounting statements reflecting
transactions which have occurred. The method by which gross (or net) premiums
are calculated is not being changed!
Example 231. Suppose the following data for an insurance company is given.
Income for Year Ending December 31, 1990
Premiums
341,000
Investment Income
108,000
Expenses
112,000
Claims and Maturities
93,000
Increases in Reserves
Net Income
Assets
Reserves
Surplus
Balance Sheet
December 31, 1989
1,725,000
500,000
December 31, 1990
1,433,000
The missing entries in the tables can be filled in as follows (amounts in thousands). Total income is 341 + 108 = 449 while total expenses are 112 + 93 = 205,
so net income (before reserve contributions) is 449 205 = 244. Now the reserves
at the end of 1989 are 1, 725 500 = 1, 225, so the increase in reserves must be
1, 433 1, 225 = 208. The net income is 244 208 = 36. Hence the 1990 surplus
is 536 and the 1990 assets are 1,969.
The central question in insurance accounting is How are liabilities measured?
23: Net Premium Reserves
121
The answer to this question has some very important consequences for the operation
of the company, as well as for the financial soundness of the company. The general
equation is
Reserve at time t = Actuarial Present Value at time t of future benefits
Actuarial Present Value at time t of future premiums.
The only accounting assumption required is one regarding the premium to be used
in this formula. Is it the net premium, gross premium, or ???
The only point of view adopted here is that liabilities are measured as the
net level premium reserves. This is the reserve computed under the accounting
assumption that the premium charged for the policy is the net level premium. To
see that this might be a reasonable approach, recall that the equivalence principle
sets the premium so that the actuarial present value of the benefit is equal to the
actuarial present value of the premiums collected. However, after the policy is
issued the present value of the benefits and of the un-collected premiums will no
longer be equal, but will diverge in time. This is because the present value of the
unpaid benefits will be increasing in time and the present value of the uncollected
premiums will decrease in time. The discrepency between these two amounts at any
time represents an unrealized liability to the company. To avoid a negative surplus
(technical bankruptcy), this liability must be offset in the accounting statements of
the company by a corresponding asset. Assume (for simplicity) that this asset takes
the form of cash on hand of the insurance company at that time. How does one
compute the amount of the reserve at any time t under this accounting assumption?
This computation is illustrated in the context of an example.
Example 232. Consider a fully discrete whole life policy issued to (x) in which
the premium is payable annually and is equal to the net premium. What is the
reserve at time k, where k is an integer? To compute the reserve simply note that
if (x) has survived until time k then the (curtate) remaining life of x has the same
distribution as K(x + k). The outstanding benefit has present value vK(x+k)+1 while
the present value of the remaining premium income is a K(x+k)+1 times the annual
premium payment. Denote by k L the random variable which denotes the size of the
future loss at time k. Then
kL
= vK(x+k)+1 Px a K(x+k)+1 .
The reserve, denoted in this case by k Vx , is the expectation of this loss variable.
Hence
x+k .
k Vx = E[k L] = Ax+k Px a
This is called the prospective reserve formula, since it is based on a look at the
future performance of the insurance portfolio. The prospective reserve formula is
122
the statement that the reserve at the end of policy year k is the expected future loss
on the policy.
A word about notation. In the example above the reserve has been computed for
a discrete whole life policy. The notation for the reserves for other types of policies
parallel the notation for the premiums for the policy. Thus k V1x:n is the reserve at
time k for a fully discrete n year term policy. When discussing general principles
the notation k V is used to denote the reserve at time k for a general policy.
Exercise 231. What types of policies have reserves t V(A1x:n ), k V(A1x:n ), and k V(ax )?
Certain timing assumptions regarding disbursements and receipts have been
made in the previous computation. Such assumptions are always necessary, so they
are now made explicit. Assume that a premium payment which is due at time t is
paid at time t+; an endowment benefit due at time t is paid at time t+; a death benefit
payment due at time t is assumed to be paid at time t, that is, just before time t.
Interest earned for the period is received at time t. Thus t Vx includes any interest
earned and also the effects of any non-endowment benefit payments but excludes
any premium income payable at time t and any endowment payments to be made at
time t. Also assume that the premium charged is the net level premium. Therefore
the full technical description of what has been computed is the net level premium
terminal reserve. One can also compute the net level premium initial reserve
which is the reserve computed right at time t. This initial reserve differs from the
terminal reserve by the amount of premium received at time t and the amount of the
endowment benefit paid at time t. Ordinarily one is interested only in the terminal
reserve.
In the remainder of this section methods of computing the net level premium
terminal reserve are discussed. For succintness, the term reserve is always taken
to mean the net level premium terminal reserve unless there is an explicit statement
to the contrary.
Exercise 232. Show that k Vx = 1
reasonable?
a x+k
. From this lim k Vx = 1. Why is this
k
a x
Exercise 233. Use the Life Table to compute the reserve for the first five years
after policy issue for a fully discrete whole life policy to (20). Assume the policy
amount is equal to $100,000 and the premium is the net premium.
The reserve can be viewed in a different way. Typically an insurance company
has many identical policies in force. One may benefit by studying the cash flow
associated with this group of policies on the average. Here is an example.
Example 233. Let us examine the expected cash flow associated with a whole life
123
policy issued to (x). Assume the premium is the net level premium and that the
policy is fully discrete. In policy year k + 1 (that is in the time interval [k, k + 1))
there are the following expected cash flows.
Time
k
k
k + 1
k + 1
Income
(benefits just paid, interest just received)
Px
qx+k
i(k Vx + Px )
Cash on Hand
k Vx
k Vx + Px
k Vx + Px qx+k
(1 + i)(k Vx + Px ) qx+k
This final cash on hand at time k +1 must be equal to the reserve for the policies
of the survivors. Thus
px+k k+1 Vx = (1 + i)(k Vx + Px ) qx+k .
This provides an important formula connecting successive reserves.
Exercise 234. Show that 1 Ex+k k+1 Vx = k Vx + Px vqx+k .
The analysis of the previous example illustrates a general argument connecting
the reserves at successive time points.
kV
= Actuarial Present Value at time k of benefits payable in [k, k + 1)

Actuarial Present Value at time k of premiums payable in [k, k + 1)
+ Actuarial Present Value at time k of the reserves at time k + 1.
Such recursive formulas for reserves are especially useful for computational purposes.
Example 234. As a more concrete illustration of the cash flow analysis, consider
a fully discrete 1,000,000 5 year term insurance issued to (21). Using the Illustrative
Life Table, this policy has net annual premium 1069.72. The expected cash flow
analysis is as follows.
Year k
1
2
3
4
5
Expected
Premium
Income
1069.72
1068.59
1067.42
1066.20
1064.95
Expected
Benefit
Payments
1061.73
1095.56
1132.51
1173.10
1217.54
Expected
Reserve
Contributions
72.18
41.48
5.77
-35.75
-83.67
Expected
Total
Reserves
72.18
113.65
119.43
83.67
0.00
Expected
Interest
Income
64.18
68.45
70.86
71.14
68.92
Expected
Net
Cash Flow
0.00
0.00
0.00
0.00
0.00
Notice that the reserves build up in the earlier, low mortality years in order to
compensate for the higher mortality rate in the later years.
Example 235. In contrast to the previous example, consider the same policy with
expenses of 400 at policy inception, sales commission of 20% of gross premium for
124
each of the first 2 years and 10% of the premium for each year thereafter, and an
annual profit of 35% of the premium. The gross premium is 2289.64, and the cash
flow analysis is as follows.
Expected Expected
Expected
Expected Expected Expected
Expected
Expected Expected
Premium Benefit
Reserve
Total
Interest
Policy
Commission Policy
Net
Year k Income Payments Contributions Reserves Income Expenses Payments
Profits Cash Flow
1
2289.64 1061.73
72.18
72.18
37.82
400.00
457.93
801.37
-465.75
2
2287.21 1095.56
41.48
113.65
38.14
0.00
457.44
800.52
-69.65
3
2284.70 1132.51
5.77
119.43
50.09
0.00
228.47
799.65
168.40
4
2282.11 1173.10
-35.75
83.67
60.46
0.00
228.21
798.74
178.27
5
2279.42 1217.54
-83.67
0.00
68.92
0.00
227.94
797.80
188.73
Because of the expenses, the cash flow in the early years is negative.
Exercise 235. What is the expected total cash flow over the life of the policy?
There are other ways to compute the reserve. First the reserve may be viewed as
maintaining the balance between income and expenses. Since at time 0 the reserve is
0 (because of the equivalence principle) the reserve can also be viewed as balancing
past income and expenses. This leads to the retrospective reserve formula for a
fully discrete whole life policy as
k Ex k Vx
= Px a x:k A1x:k .
This formula is derived as follows. Recall that

Ax = A1x:k + vk k px Ax+k
and
a x = a x:k + vk k px a x+k .
Since the reserve at time 0 is zero,

0 = Ax Px a x = A1x:k + vk k px Ax+k Px a x:k + vk k px a x+k
where k is an arbitrary positive integer. Rearranging terms and using the prospective
formula for the reserve given above produces the retrospective reserve formula.
Exercise 236. Sometimes the retrospective reserve formula is written as
h Vx
= Px a x:h / h Ex h kx = Px sx:h h kx
where h kx is called the accumulated cost of insurance. Find an expression for h kx .

How would t kx be computed?
Example 236. The retrospective formula can also be written entirely in terms of
premiums. In the case of the whole life policy, k Vx = (Px P1x:k )/ P 1 .
x:k
125
Exercise 237. Verify the formula given in the example.

A retrospective reserve formula can be written for any type of policy.
Obtaining expressions for the reserve for any of the many possible types of
insurance policy is now relatively straightforward. Doing this is left as an exercise
for the reader. One should keep in mind that the important point here is to be able
to (ultimately) write a formula for the reserve which one can compute with the data
available in the life table. Hence continuous and/or mthly payment schemes need
to be reduced to their equivalent annual forms. Recursive formulas are also often
used.
126
Problems
Problem 231. True or False: For 0 k < n, k Vx:n = 1
k = n?
a x+k:nk
. What happens at
a x:n
Problem 232. Find a formula for the reserve at the end of 5 years for a 10 year
term policy with benefit $1 issued to (30) on a net single premium basis.
Problem 233. Show that for 0 t n

t V(Ax:n )
= P(Ax+t:nt ) P(Ax:n ) ax+t:nt .
This is called the premium difference formula for reserves. Find similar formulas
for the other types of insurance.
Problem 234. Show that for 0 t n
P(Ax:n )
t V(Ax:n ) = 1
P(Ax+t:nt )
Ax+t:nt .
This is called the paid up insurance formula for reserves. Find similar formulas
for the other types of insurance.
Problem 235. Find P1x:n if n Vx = 0.080, Px = 0.024 and P
1
x:n
= 0.2.
Problem 236. Given that 10 V35 = 0.150 and that 20 V35 = 0.354 find 10 V45 .
Problem 237. Write prospective and retrospective formulas for 40
20 V(A20 ), the reserve at time 20 for a semi-continuous 40 payment whole life policy issued to
(20).
Problem 238. For a general fully discrete insurance let us suppose that the benefit
payable if death occurs in the time interval (h 1, h] is bh and that this benefit is
paid at time h, that is, at the end of the year of death. Suppose also that the premium
paid for this policy at time h is h . Show that for 0 t 1
t px+k k+t V
+ v1t t qx+k bk+1 = (1 + i)t (k V + k ).
This gives a correct way to interpolate reserves at fractional years.

Problem 239. In the notation of the preceding problem show that for 0 t 1
k+t V
= v1t (1t qx+k+t bk+1 + 1t px+k+t k+1 V) .
Problem 2310. Show that under UDD, hk V(Ax:n ) = (i/ )hk V1x:n + hk V 1 .
x:n
127
(m)
. This gives
Problem 2311. Show that under UDD, hk Vx:n
= hk Vx:n + (m)h P(m)
x:n k V1
x:n
the reserves for a policy with mthly premium payments in terms of the reserves for
a policy with annual premium payments.
Problem 2312. Show that hk V (m) (Ax:n ) = hk V(Ax:n ) + (m)h P(m) (Ax:n )k V1x:n under
UDD.
Problem 2313. The amount at risk in year k for a discrete insurance is the
difference between the benefit payment payable at the end of year k and k V. Find
the mean and variance of the amount at risk in year 3 of a 5 year term policy issued
to (30) which pays a benefit of 1 at the end of the year of death and has net level
premiums.
Problem 2314. Suppose that 1000 t V(Ax ) = 100, 1000P(Ax ) = 10.50, and =
0.03. Find ax+t .
Problem 2315. Calculate 20 V45 given that P45 = 0.014, P
0.030.
1
45:20
= 0.022, and P45:20 =
Problem 2316. A fully discrete life insurance issued to (35) has a death benefit of
$2500 in year 10. Reserves are calculated at i = 0.10 and the net annual premium
P. Calculate q44 given that 9 V + P = 10 V = 500.

Problem 231. Use the prospective formula and Ax:n + da x:n = 1 to see the
formula is true. When k = n the reserve is 1 by the timing assumptions.
Problem 232. Prospectively the reserve is A1
35:5
Problem 233. Use the prospective formula and the premium definitions.
Problem 235. From the retrospective formula n Exn Vx = Px a x:n A1 . Now
x:n
divide by a x:n .
Problem 236. Use the prospective formula and the relation Ax + da x = 1 to
obtain k Vx = 1 a x+k / a x .
Problem 237. The prospective and retrospective formulas are
a20:20 A1 )/ 20 E20 .
A40 Pa40:20 and 40
20 V(A20 ) = (P
40
20 V(A20 )
20:20
Problem 238. The value of the reserve, given survival, plus the present value
of the benefit, given death, must equal the accumulated value of the prior reserve
and premium.
Problem 2313. The amount at risk random variable is 1{2} (K(30)) 3 V1
30:5
Problem 2314.
Ax + ax = 1.
Use the prospective reserve formula and the relationship
Problem 2315. Use the retrospective formula.

Problem 2316. By the general recursion formula 1 E4410 V = 9 V +P2500vq44 .
128

Exercise 231. t V(A1 ) is the reserve at time t for a fully continuous n year
x:n
term insurance policy, k V(A1 ) is the reserve at time k for a semi-continuous n
x:n
year term policy, and k V(ax ) is the reserve at time k for a life annuity.
Exercise 232. Since Ax + da x = 1, k Vx = Ax+k (Ax / a x )ax+k = 1 da x+k (1
da x )ax+k / a x = 1 a x+k / a x .
Exercise 233. The reserve amounts are easily computed using the previous exercise as 1000001 V20 = 100000(1 16.4611/ 16.5133) = 316.11, 1000002 V20 =
649.17, 1000003 V20 = 985.59, 1000004 V20 = 1365.57, and 1000005 V20 =
1750.71.
Exercise 234. Just multiply the previous equation by v = (1 + i)1 .
Exercise 235. The expected total cash flow over the life of the policy is zero,
since that is the requirement used in determining the gross premium.
Exercise 236. h kx = A1 / k Ex , which can be easily computed from the life
x:k
table using previous identities.
Exercise 237. Divide the retrospective formula by k Ex to isolate the reserve
term. Then use the fact that k Ex / a x:k = P 1 .
x:k
129
24. Asset Shares
An alternate methodology used to study an insurance portfolio is to examine the

progress of the asset share, or contribution to the company assets, associated with
a given group of identical policies. Formally, the asset share for a group of policies
at time k is the ratio of total company assets generated by these policies at time k to
the number of such policies in force at time k.
Asset share computations can be most easily visualized in a situation in which
the company sold, at time 0, l0 policies of the given type issued to a group of insureds
aged x and to assume that the results in each period will be the probabilistically
expected amounts. Let k AS denote the asset share at time k (on a terminal basis),
()
lx+k
denote the number of original policy holders surviving to age x + k, G denote
the gross premium, ck denote the fraction of the gross premium paid at time k for
(1)
expenses, ek denote the per policy expenses at time k, dx+k
denote the number of
policy holders dying at age (x + k), bk+1 denote the benefit paid at time k + 1 for a
(2)
death at age x + k, and dx+k
the number of policy holders withdrawing at age (x + k).
Standard cash flow analysis gives the fundamental relationship
()
k+1 AS lx+k+1
()
(1)
(2)
= (k AS + G ck G ek ) (1 + i)lx+k
bk+1 dx+k
k+1 CV dx+k
where k+1 CV is the cash value, or withdrawal benefit, paid to those insureds who
cancel their policy at age x + k. Typically, k+1 CV = k+1 V, the reserve for the policy
at time k + 1. The insurance is assumed to be on a fully discrete basis and the
cash value is paid at the end of the year of withdrawal. Dividing both sides of this
()
equation by lx+k
produces a second useful recursion formula connecting successive
asset shares
()
k+1 AS px+k
(2)
= (k AS + G(1 ck ) ek ) (1 + i) bk+1 q(1)
x+k k+1 CV qx+k .
Some of the uses of this idea are now illustrated.

One natural use of the idea of asset shares is to determine the gross premium
G required in order to achieve a certain asset goal at a future time. To do this, note
that multiplying the first equation above by vk+1 gives
k+1 ()
lx+k+1
k+1 AS v
()
()
k AS vk lx+k
= (G(1 ck ) ek ) vk lx+k
(1)
(2) k+1
(bk+1 dx+k
+ k+1 CV dx+k
)v .
24: Asset Shares
131
Using the fact that 0 AS = 0 and summing this telescoping series gives
n ()
n AS v lx+n
n1
X
k+1 ()
lx+k+1
k+1 AS v
k=0
n1
X
=G
()
(1 ck )vk lx+k
k=0
n1
X
()
k AS vk lx+k
()
ek vk lx+k
k=0
n1
X
(1)
(2)
bk+1 dx+k
+ k+1 CV dx+k
vk+1 .
k=0
One can now easily solve for the gross premium G.

A technique similar to the asset share computation can be used for a different
purpose. Suppose the asset requirement for the kth year of policy life, k F, is set so
that with this amount on hand there is a high probability of meeting all expenses.
Then as before
k+1 F
(1)
(2)
p()
x+k = (k F + G ck G ek ) (1 + i) bk+1 qx+k k+1 CV qx+k .
If these computations were done under conservative assumptions the company may
wish to return part of the excess to the insured in the form of dividends. Let k D
denote the amount available for dividends at the end of the kth period. Also denote
with a hat the values of the respective quantities which were observed in practice.
Then
(k+1 F + k+1 D) p ()
k G e k ) (1 + i) bk+1 q (1)
(2)
x+k = (k F + G c
x+k k+1 CV q
x+k .
Subtracting the first equation from the second gives
()
k+1 D p
x+k
=(k F + G)(i i)
+ [(Gck + ek )(1 + i) (Gck + e k )(1 + i)]
+ (1 k+1 F)bk+1 (q(1)
(1)
x+k q
x+k )
+ (k+1 CV k+1 F)(q(2)
(2)
x+k q
x+k ).
Exercise 241. Write a formula for k+1 D under the additional assumptions that
k+1 CV = k+1 F and that dividends are paid to insureds who die or withdraw.
Computations similar to the above can be used to compare the predicted asset
share with that obtained in experience.
24: Asset Shares
132
Problems
Problem 241. If 10 AS1 is the asset share at the end of 10 years based on premium
G1 and 10 AS2 is the asset share at the end of 10 years based on premium G2 , find a
formula for 10 AS1 10 AS2 .
Problem 242. A policy providing death benefit of $1000 at the end of the year
of death has a gross premium of $25. Suppose we are given i = 0.05, 10 AS = 160,
(2)
c10 = 0.1, e10 = 2.50, q(1)
x+10 = 0.003 (decrement 1 is death), qx+10 = 0.1 (decrement 2
is withdrawal), and 11 CV = 170. What is 11 AS?
24: Asset Shares

Problem 241.
10 AS1
10 AS2 = (G1 G2 )
Problem 242. Use the relation

(2)
bk+1 q(1)
x+k + k+1 CV qx+k .
P9
()
k+1 AS px+k
k=0 (1
() ()
ck )vk9 lx+k
/ lx+10 .
= (k AS + G(1 ck ) ek ) (1 + i)
133

Question 251 . You are given the following values calculated at = 0.08 for two
fully continuous whole life policies issued to (x):
Policy #1
Policy #2
Death
Benefit
4
6
Premium
0.18
0.22
Variance
of Loss
3.25
Calculate the variance of the loss for policy #2.

A. 4.33
B.
5.62
D.
6.83
C.
6.37
E.
9.74
Question 252 . Which of the following are correct expressions for ht V(Ax:n ) for
t h h?
i
I. h P(Ax+t:nt ) h P(Ax:n ) ax+t:ht
"
#
h P(Ax:n )
II. 1
Ax+t:nt
h P(Ax+t:nt )
III. h P(Ax:n )sx:t t kx
A.
I and II
B.
I and III
D.
All
C.
II and III
E.
None of A, B, C, or D
135
Question 253 . A fully discrete whole life insurance provides for payment of its
net level reserve in addition to the face amount of 1. t V x is the reserve for this
insurance, and x V x = 0. Which of the following are expressions for the net annual
premium?
Px1 t+1
v qx+t
I. t=0
a x
II. 2P
x
Px1 t+1
v t| qx (1 + t+1 V x )
III. t=0
a x
A.
I and II
B.
I and III
D.
All
C.
II and III
E.
Question 254 . A fully discrete whole life insurance is issued to (x). You are given
4
that Px = 11
, t Vx = 0.5, and a x+t = 1.1. Calculate i.
A. 0
B.
0.04
D.
0.10
C.
0.05
E.
0.25
Question 255 . For a fully discrete two year term insurance of 400 on (x), i = 0.1,
400P1x:2 = 74.33, 4001 V1x:2 = 16.58, and the contract premium equals the benefit
premium. Calculate the variance of loss at issue.
A. 21,615
B.
23,125
D.
31,175
C.
27,450
E.
34,150
136
Question 256 . For a 10 year deferred whole life annuity of 1 on (35) payable
continuously mortality follows DeMoivres law with = 85. The interest rate i = 0
and level benefit premiums are payable continuously for 10 years. Calculate the
benefit reserve at the end of five years.
A. 9.38
B.
9.67
D.
10.36
C.
10.00
E.
10.54
Question 257 . For a fully discrete whole life insurance with non-level benefits
on (70) the level benefit premium for this insurance is equal to P50 . Also, q70+k =
q50+k + 0.01 for k = 0, 1, . . . , 19, q60 = 0.01368, k V = k V50 for k = 0, 1, . . . , 19, and
11 V50 = 0.16637. Calculate b11 , the death benefit in year 11.
A. 0.482
B.
0.624
D.
0.648
C.
0.636
E.
0.834
Question 258 . For a fully discrete 3 year endowment insurance of 1000 on (x),
qx = qx+1 = 0.20, i = 0.06, and 1000Px:3 = 373.63. Calculate 1000(2 Vx:3 1 Vx:3 ).
A. 320
B.
325
D.
335
C.
330
E.
340
137

Question 251 . If L1 is the loss for Policy #1 then L1 = 4vT 0.18aT = (4 +
.18/ .08)vT 0.18/ .08 and similarly L2 = (6 + 0.22/ 0.08)vT .22/ .08. Thus Var(L2 ) =
(8.75)2 Var(vT ) = (8.75/ 6.25)2 Var(L1 ) = 6.37. C.
Question 252 . The prospective formula would give I except that the first term
should be for an h t payment premium. So I is incorrect. Similarly in II the
denominator factor should have a left subscript of h t. Only III is correct. E.
Question 253 . Direct application of the equivalence principle shows that III is
P + vpx+kk+1 V.
correct. The general reserve formula gives k V = vqx+k (1 + k+1 V)

P
vk+1 qx+k / a 10x , and I holds.
Multiply this by vk and sum to see that P = x1
k=0
Considering the special case in which x = 1 shows that II fails. B.
Question 254 . Here 0.5 = t Vx = Ax+t Px a x+t = 1 (d + Px )ax+t , from which
d = 1/ 11 and i = 1/ 10. D.
Question 255 . The information about the reserve and premium give qx+1 = 0.25.
Using this and the premium information gives qx = 0.17. The loss at issue random
variable L = (400v P)1{0} (K(x)) + (400v2 Pv P)1{1} (K(x)) (Pv + P)1[2,) (K(x)),
and these terms are disjoint. Since E[L] = 0, Var(L) = E[L2 ] = (400v P)2 qx +
(400v2 Pv P)2 px qx+1 + (Pv + P)2 2 px = 34150.56. E.
Here t p35 = 1 t/ 50 for 0 < t < 50. The premium P satisfies
Question 256 .
Z
50
10
Z
P
0
t p35
dt = P
5
t p40
10
t p35
dt giving P = 16/ 9. The reserve at time 5 is then

5
45
t p40
dt
dt = 9.38. A.
Question 257 . Using the formula connecting successive reserves gives p6011 V50 =
(10 V50 +P50 )(1+i)q60 and also p8011 V = (10 V +P50 )(1+i)q80 b11 , since the premiums
are the same. The first term on the right side of each equation is the same, so using
the first equation the value of this common term is q60 + p6011 V50 = 0.1777. Using
this in the second equation gives b11 = 0.6479. D.
Question 258 . Since this is an endowment, (10002 V + 1000P)(1 + i) = 1000,
from which 10002 V = 569.76. Using the retrospective reserve formula, px 10001 V =
1000P(1 + i) 1000qx , giving 10001 V = 245.06. Thus 1000(2 V 1 V) = 324.70. B.
26. The Individual Risk Model

An insurance company has a large number of policies in force at any given
time. This creates a financial risk for the company. There are two aspects to the
problem of analyzing this risk. First, one must be able to estimate the amount of
risk. Secondly, one must be able to model the times at which claims will occur in
order to avoid cash flow difficulties. The problem of modeling the amount of risk
will be studied first.
In the individual risk model the insurers total risk S is assumed to be expressable
in the form S = X1 + . . . + Xn where X1 , . . . , Xn are independent random variables
with Xi representing the loss to the insurer on insured unit i. Here Xi may be quite
different than the actual damages suffered by insured unit i. In the closed model
the number of insured units n is assumed to be known and fixed. A model in which
migration in and out of the insurance system is allowed is called an open model.
The individual risk model is appropriate when the analysis does not require the
effect of time to be taken into account.
The first difficulty is to find at least a reasonable approximation to the probabilistic properties of the loss random variables Xi . This can often be done using data
from the past experience of the company.
Example 261. For short term disability insurance the amount paid by the insurance
company can often be modeled as X = cY where c is a constant representing the
daily rate of disability payments and Y is the number of days a person is disabled.
One then is simply interested in modelling the random variable Y. Historical data
can be used to estimate P[Y > y]. In this context P[Y > y], which was previously
called the survival function, is referred to as the continuance function. The same
notion can be used for the daily costs of a hospitalization policy.
The second difficulty is to uncover the probabilistic properties of the random
variable S. In theoretical discussions the idea of conditioning can be used to find
an explicit formula for the distribution function of a sum of independent random
varibles.
Example 262. Suppose X and Y are independent random variables each having
the exponential distribution with parameter 1. By conditioning
Z
P[X + Y t] =
Z t
=
0
P[X + Y t|Y = y] fY (y) dy
P[X t y] fY (y) dy
Z t
=
0
= 1 et tet
1 e(ty) fY (y) dy
26: The Individual Risk Model
139
for t 0.
This argument has actually shown that if X and Y are independent and absolutely
continuous then
Z
FX+Y (t) =
FX (t y) fY (y) dy.
This last integral is called the convolution of the two distribution functions and is
often denoted FX FY .
Exercise 261. If X and Y are absolutely continuous random variables show that
X + Y is also absolutely continuous and find a formula for the density function of
X + Y.
Exercise 262. Find a similar formula if X and Y are both discrete. Use this formula
to find the density of X + Y if X and Y are independent Bernoulli random variables
with the same success probability.
An approach which requires less detailed computation is to appeal to the Central
Limit Theorem.
Central Limit Theorem. If X1 , . . . , Xn are independent random variables then the
distribution of
n
X
Xi is approximately the normal distribution with mean
i=1
and variance
n
X
n
X
E[Xi ]
i=1
Var(Xi ).
i=1
The importance of this theorem lies in the fact that the approximating normal
distribution does not depend on the detailed nature of the original distribution but
only on the first two moments.
Example 263. You are a claims adjuster for the Good Driver Insurance Company
of Auburn. Based on past experience the chance of one of your 1000 insureds being
involved in an accident on any given day is 0.001. Your typical claim is $500. What
is the probability that there are no claims made today? If you have $1000 cash
on hand with which to pay claims, what is the probability you will be able to pay
all of todays claims? How much cash should you have on hand in order to have a
99% chance of being able to pay all of todays claims? What assumptions have you
made? How reasonable are they? What does this say about the solvency of your
company?
Using the Central Limit Theorem, if an insurance company sold insurance at
the pure premium not only would the company only break even (in the long run)
but due to random fluctuations of the amount of claims the company would likely
go bankrupt. Thus insurance companies charge an amount greater than the pure
premium. A common methodology is for the company to charge (1 + ) times
140
the pure premium. When this scheme is followed is called the relative security
loading and the amount (pure premium) is called the security loading. This
is a reasonable procedure since the insureds with larger expected claims pay a
proportionate share of the loading. The relative loading is usually adjusted to
achieve a certain measure of protection for the company.
Example 264. Suppose that a company is going to issue 1,000 fire insurance
policies each having a $250 deductible, and a policy limit of $50,000. Denote by
Fi the Bernoulli random variable which is 1 if the ith insured suffers a loss, and
by Di the amount of damage to the ith insureds property. Suppose Fi has success
probability 0.001 and that the actual damage Di is uniformly distributed on the
interval (0,70000)). What is the relative loading so that the premium income will be
95% certain to cover the claims made? Using the obvious notation, the total amount
of claims made is given by the formula
S=
1000
X
Fi (Di 250)1[250,50000] (Di ) + 500001(50000,) (Di )
i=1
where the Fs and the Ds are independent (why?) and for each i the conditional
distribution of Di given Fi = 1 is uniform on the interval (0,70000). The relative
security loading is determined so that
P[S (1 + ) E[S]] = 0.95.
This is easily accomplished by using the Central Limit Theorem.
Exercise 263. Compute E[S] and Var(S) and then use the Central Limit Theorem
to find . What is the probability of bankruptcy when = 0?
Another illustration is in connection with reinsurance. Good practice dictates
that an insurance company should not have all of its policy holders homogeneous,
such as all located in one geographical area, or all of the same physical type. A
moments reflection on the effect of a hurricane on an insurance company with all
of its property insurance business located in one geographic area makes this point
clear. An insurance company may diversify its portfolio of policies (or just protect
itself from such a concentration of business) by buying or selling reinsurance. The
company seeking reinsurance (the ceding company) buys an insurance policy from
the reinsurer which will reimburse the company for claims above the retention
limit. For stop loss reinsurance, the retention limit applies on a policy-by-policy
basis to those policies covered by the reinsurance. The retention limit plays the
same role here as a deductible limit in a stop loss policy. Usually there is one
reinsurance policy which covers an entire package of original policies. For excess
of loss reinsurance, the retention limit is applied to the total amount of claims for the
package of policies covered by the insurance, not the claims of individual policies.
141
Example 265. You operate a life insurance company which has insured 2,000 30
year olds. These policies are issued in varying amounts: 1,000 people with $100,000
policies, 500 people with $500,000 policies, and 500 people with $1,000,000 policies. The probability that any one of the policy holders will die in the next year is
0.001. Stop loss reinsurance may be purchased at the rate of 0.0015 per dollar of
coverage. How should the retention limit be set in order to minimize the probability that the total expenses (claims plus reinsurance expense) exceed $1,000,000 is
minimized? Let X, Y, and Z denote the number of policy holders in the 3 catagories
dying in the next year. Then X has the binomial distribution based on 1000 trials
each with success probability 0.001, Y has the binomial distribution based on 500
trials each with success probability 0.001, and Z has the binomial distribution based
on 500 trials each with success probability 0.001. If the retention limit is set at r
then the cost C of claims and reinsurance is given by
C = (100000 r)X + (500000 r)Y + (1000000 r)Z

+ 0.0015 1000(100000 r)+ + 500(500000 r)+ + 500(1000000 r)+ .
Straightforward, computations using the central limit theorem provides an estimate
of P[C 1, 000, 000].
Exercise 264. Verify the validity of the above formula. Use the central limit
theorem to estimate P[C 1, 000, 000] as a function of r. Find the value(s) of r
which minimize this probability.
142
Problems
Problem 261. The probability of an automobile accident in a given time period is
0.001. If an accident occurs the amount of damage is uniformly distibuted on the
interval (0,15000). Find the expectation and variance of the amount of damage.
Problem 262. Find the distribution and density for the sum of three independent
random variables each uniformly distributed on the interval (0,1). Compare the
exact value of the distribution function at a few selected points (say 0.25, 1, 2.25)
with the approximation obtained from the central limit theorem.
Problem 263. Repeat the previous problem for 3 independent exponential random
variables each having mean 1. It may help to recall the gamma distribution here.
Problem 264. A company insures 1000 essentially identical cars. The probability
that any one car is in an accident in any given year is 0.001. The damage to a car
that is involved in an accident is uniformly distributed on the interval (0,15000).
What relative security loading should be used if the company wishes to be 99%
sure that it does not lose money?

Problem 261. The amount of damage is BU where B is a Bernoulli variable
with success probability 0.001 and U has the uniform distribution.
1000
X
Problem 264. The loss random variable is of the form
Bi Ui .
i=1
143

Exercise 261. Differentiation
of the general distribution function formula
R
above gives fX+Y (t) = fX (t y)fY (y) dy.
Exercise 262. In the discrete case the same line of reasoning
P
Pmgivesn fX+Yty(t) =
Y (y). Applying this in the Bernoulli case, fX+Y (t) =
y fX (ty)f
y=0 ty p (1

m

P
n
n+m t
p)nt+y my py (1 p)my = pt (1 p)n+mt m
=
p
(1
p)m+nt .
y=0 ty y
t
Exercise 263. The loading is chosen so that E[S]/ Var(S) = 1.645, from
the normal table. When = 0 the bankruptcy probability is about 1/ 2.
Exercise 264. Direct computation using properties of the binomial distribution
gives E[C]
= (100000 r) 1 + (500000 r) (1/ 2) + (1000000 r) (1/ 2) +
0.0015 1000(100000 r)+ + 500(500000 r)+ + 500(1000000 r)+ and also
Var(C) = (100000r)2 0.999+(500000r)2 0.999/ 2+(1000000r)2 0.999/ 2.
The probability can now be investigated numerically using the Central Limit
Theorem approximation.
144
27. The Collective Risk Model and Ruin Probabilities

Some of the consequences of the collective risk model will now be examined.
In the collective risk model the time at which claims are made is taken to account.
PN(t)
Here the aggregate claims up to time t is assumed to be given by k=1
Xk where
X1 , X2 , . . . are independent identically distributed random variables representing the
sizes of the respective claims, N(t) is a stochastic process representing the number
of claims up to time t, and N and the Xs are independent. The object of interest is
the insurers surplus at time t, denoted by U(t), which is assumed to be of the form
U(t) = u + ct
N(t)
X
Xk
k=1
where u is the surplus at time t = 0, and c represents the rate of premium income.
Special attention will be given to the problem of estimating the probability that the
insurance company has negative surplus at some time t since this would mean that
the company is ruined.
To gain familiarity with some of the ideas involved, the simpler classical gamblers ruin problem will be studied first.
28. Stopping Times and Martingales

A discrete time version of the collective risk model will be studied and some
important new concepts will be introduced.
Suppose that a gambler enters a casino with z dollars and plays a game of chance
in which the gambler wins $1 with probability p and loses $1 with probability
q = 1 p. Suppose also that the gambler will quit playing if his fortune ever reaches
a > z and will be forced to quit by being ruined if his fortune reaches 0. The main
interest is in finding the probability that the gambler is ultimately ruined and the
expected number of the plays in the game.
In order to keep details to a minimum, the case in which p = q = 1/ 2 will
be examined first. Denote by Xj the amount won or lost on the jth play of the
game. These random variables are all independent and have the same underlying
distribution function. Absent any restrictions about having to quit the game, the
fortune of the gambler after k plays of the game is
z+
k
X
Xj .
j=1
Now in the actual game being played the gambler either reaches his goal or is ruined.
Introduce a random variable, T, which marks the play of the game on which this
occurs. Technically
P
T = inf{k : z + kj=1 Xj = 0 or a}.
Such a random variable is called a random time. Observe that for this specific
random variable the event [T k] depends only on the random variables X1 , . . . , Xk .
That is, in order to decide at time k whether or not the game has ended it is not
necessary to look into the future. Such special random times are called stopping
times. The precise definition is as follows. If X1 , X2 , . . . are random variables and
T is a nonnegative integer valued random variable with the property that for each
integer k the event [T k] depends only on X1 , . . . , Xk then T is said to be a stopping
time (relative to the sequence X1 , X2 , . . .).
P
The random variable z + Tj=1 Xj is the gamblers fortune when he leaves the
casino, which is either a or 0. Denote by (z) the probability that the gambler
P
leaves the casino with 0. Then by direct computation E[z + Tj=1 Xj ] = a(1 (z)).
A formula for the ruin probability (z) will be obtained by computing this same
expectation in a second way.
Each of the random variables Xj takes values 1 and 1 with equal probability,
P
so E[Xj ] = 0. Hence for any integer k, E[ kj=1 Xj ] = 0 too. So it is at least plausible
P
P
that E[ Tj=1 Xj ] = 0 as well. Using this fact, E[z + Tj=1 Xj ] = z, and equating this
28: Stopping Times and Martingales
147
with the expression above gives z = a(1 (z)). Thus (z) = 1 z/ a for 0 z a
are the ruin probabilities.
There are two important technical ingredients behind this computation. The first
is the fact that T is a stopping time. The second is the fact that the gambling game
with p = q = 1/ 2 is a fair game. The notion of a fair game motivates the definition
of a martingale. Suppose M0 , M1 , M2 , . . . are random variables. The sequence is a
martingale if E[Mk | Mk1 , . . . , M0 ] = Mk1 for all k 1. In the gambling context, if
Mk is the gamblers fortune after k plays of a fair game then given Mk1 the expected
fortune after one more play is still Mk1 .
Exercise 281. Show that Mk = z +
Pk
j=1
Xj (with M0 = z) is a martingale.

2
Example 281. The sequence M0 = z2 and Mk = z + kj=1 Xj k for k 1 is also

a martingale. This follows from the fact that knowing M0 , . . . , Mk1 is the same as
knowing X1 , . . . , Xk1 and the fact that the Xs are independent.
Exercise 282. Fill in the details behind this example.
The important computational fact is the Optional Stopping Theorem which
states that if {Mk } is a martingale and T is a stopping time then E[MT ] = E[M0 ]. In
the gambling context this says that no gambling strategy T can make a fair game
biased.

2
Example 282. Using the martingale M0 = z2 and Mk = z + kj=1 Xj k for

k 1 along with the same stopping time T as before can provide information
about the duration of the gamblers stay in the casino. The random variable MT =
2

P
z + Tj=1 Xj T has an expectation which is easily computed directly to be E[MT ] =
a2 (1 (z)) E[T]. By the optional stopping theorem, E[MT ] = E[M0 ] = z2 .
Comparing these two expressions gives E[T] = a2 (1 (z)) z2 = az z2 as the
expected duration of the game.
The preceding example illustrates the general method. To analyze a particular
problem identify a martingale Mk and stopping time T. Then compute E[MT ] in two
ways, directly from the definition and by using the optional stopping theorem. The
resulting equation will often reveal useful information.
Uncovering the appropriate martingale is often the most difficult part of the
process. One standard method is the following. If X1 , X2 , . . . are independent and
identically distributed random variables define
Wk =
et
Pk
j=1
E[e
Xj
Pk
j=1
Xj
.
]
148
Notice that the denominator is nothing more than the moment generating function
of the sum evaluated at t. For each fixed t the sequence Wk is a martingale (here
W0 = 1). This follows easily from the fact that if X and Y are independent then
E[et(X+Y) ] = E[etX ] E[etY ]. This martingale is called Walds martingale (or the
exponential martingale) for the X sequence.
Exercise 283. Show that {Wk : k 0} is a martingale no matter what the fixed
value of t is.
In many important cases a non-zero value of t can be found so that the denominator part of the Wald martingale is 1. Using this particular value of t then makes
application of the optional stopping theorem neat and easy.
To illustrate the technique consider the following situation which is closer to
that of the collective risk model. Suppose the insurer has initial reserve z and
that premium income is collected at the rate of c per unit time. Also, Xk denotes
the claims that are payable at time k, and the Xs are independent and identically
P
distributed random variables. The insurers reserve at time k is then z + ck kj=1 Xj =
P
z + kj=1 (c Xj ). Denote by T the time of ruin, so that
T = min{k : z + ck
k
X
Xj 0}.
j=1
The objective is to study the probability (z) that ruin occurs in this setting.
As a first step, notice that if E[c Xj ] 0, ruin is guaranteed since premium
income in each period is not adequate to balance the average amount of claims in
the period. So to continue, assume that E[c Xj ] > 0.
Under this assumption, suppose there is a number so that E[e(cXj ) ] = 1.
This choice of
in Walds martingale makes the denominator 1, and shows that
P
k
Mk = e(z+ck j=1 Xj ) is a martingale. Computing the expectation of MT using the

Optional StoppingP
Theorem gives E[MT ] = E[M0 ] = ez . Computing directly gives
E[MT ] = E[e(z+cT
T
j=1
Xj )
|T < ] (z). Hence
(z) = ez / E[e(z+cT
PT
j=1
Xj )
|T < ].
A problem below will show that < 0, so the denominator of this fraction is larger
than 1. Hence (z) ez . The ruin probability decays exponentially as the initial
reserve increases.
The usual terminology defines the adjustment coefficient R = . Thus
(z) eRz . A large adjustment coefficient implies that the ruin probability declines
rapidly as the initial reserve increases.
149
Problems
Problem 281. By conditioning on the outcome of the first play of the game show
that in the gamblers ruin problem (z) = p(z + 1) + q(z 1). Show that if p = q
there is a solution of this equation of the form (z) = C1 + C2 z and find C1 and C2
by using the natural definitions (0) = 1 and (a) = 0. Show that if p q there is a
solution of the form (z) = C1 + C2 (q/ p)z and find the two constants. This provides
a solution to the gamblers ruin problem by using difference equations instead of
probabilistic reasoning.
Problem 282. In the gamblers ruin problem, show that if p q the choice t =
ln(q/ p) makes the denominator of Walds martingale 1. Use this choice of t and the
optional stopping theorem to find the ruin probability in this case.
Problem 283. Suppose p q in the gamblers ruin problem. Define M0 = z and
P
Mk = z + kj=1 Xj k(p q) for k 1. Show that the sequence Mk is a martingale and
use it to compute E[T] in this case.
Problem 284. Suppose that c > 0 is a number and X is a random variable which
takes on only non-negative values. Suppose also that E[c X] > 0. Show that if
c X takes on positive and negative values then there is a number < 0 so that
E[e(cX) ] = 1.

Problem 282. (z) =
(q/ p)a (q/ p)z

(q/ p)a 1 .
Problem 283. E[T] =
z
qp
a 1(q/ p)z
qp 1(q/ p)a .
Problem 284. Define a function f (v) = E[ev(cX) ]. Then f (v) = E[(c

X)ev(cX) ] and f (v) = E[(c X)2 ev(cX) ] > 0. Thus f is a convex function and
the graph of f is concave up. Now f (0) = 1 and f (0) = E[(c X)] > 0. Thus
the graph of f is above 1 to the right of 0, and below 1 (initially) to the left of
0. Since c X takes on negative values, limv f (v) = , so there is a negative
value of v at which f (v) = 1, by continuity.
150

Exercise 281. Knowing M0 , . . . , Mk1 is the same
as knowing X1 , . . . , Xk1 . So
Pk1
E[Mk |M0 , . . . , Mk1 ] = E[Mk | X0 , . . . , Xk1 ] = z + j=1
Xj + E[Xk | X0 , . . . , Xk1 ] =
Mk1 since the last expectation is 0 by independence.
P
2
P
2
k1
Exercise 282. First write Mk = z + k1
X
+
X
k
=
z
+
X
+
j
k
j
j=1
j=1
Pk1
2
2Xk (z + j=1 Xj ) + Xk k. Take conditional expectations using the fact that Xk is
independent of the other Xs and E[Xk ] = 0 and E[Xk2 ] = 1 to obtain the result.
Pk
Pk1
t
X
t
X
Exercise 283. Independence gives E[e j=1 j ] = E[e j=1 j ]E[etXk ]. Direct
computation of the conditional expectation gives the result.
151
29. The Collective Risk Model Revisited

The ideas developed in connection with the gamblers ruin problem will now
be used to compute the ruin probability in the collective risk model. Since the
processes are now operating in continuous time the details are more complicated
and not every step of the arguments will be fully justified.
P
In this setting the claims process is N(t)

k=1 Xk where X1 , X2 , . . . are independent
identically distributed random variables representing the sizes of the respective
claims, N(t) is a stochastic process representing the number of claims up to time t,
and N and the Xs are assumed to be independent. The insurers surplus is given by
P
U(t) = u + ct N(t)
k=1 Xk , where u > 0 is the surplus at time t = 0 and c > 0 is the rate
at which premium income arrives per unit time. The probability of ruin with initial
surplus u will be denoted by (u).
As in the discrete time setting, the Wald martingale will be used together with the
Optional Stopping Theorem in order to obtain information about the ruin probability.
Here the denominator P
of the Wald martingale is E[eU(t) ], and the first step is to find
N(t)
a 0 so that E[e(ct k=1 Xk ) ] = 1 no matter the value of t.
P
N(t)
Xk . Now for each
The new element in this analysis is the random sum k=1
fixed t, N(t) is a random variable which is independent of the Xs. The moment
generating function of this sum can be easily computed by conditioning on the value
of the discrete random variable N(t).
E[e
PN(t)
k=1
Xk
] = E[E[e
=
=
=
=
X
j=0
E[e
E[e
j=0

X
j=0
PN(t)
k=1
Xk
PN(t)
k=1
Pj
k=1
ej ln(E[e
Xk
|N(t) = j] P[N(t) = j]
Xk
] P[N(t) = j]
j
E[eX ]
X ])
| N(t)]]
P[N(t) = j]
P[N(t) = j]
j=0
= MN(t) (ln(MX ())).

PN(t)
Hence there is a 0 so that E[e(ct k=1 Xk ) ] = 1 if and only if ect MN(t) (ln(MX ())) =
1 for all t. Suppose for now that there is a number R > 0 so that
eRct MN(t) (ln(MX (R))) = 1
for all t. This number R is called the adjustment coefficient. The existence of an
adjustment coefficient will be investigated a bit later. Using R as the value of in
29: The Collective Risk Model Revisited
153
Walds martingale shows that

PN(t)
Wt = eR(u+ct
k=1
Xk )
is a martingale.
P
Define a stopping time Ta by Ta = inf{s : u + cs N(s)

k=1 Xk 0 or a} where
a is an arbitrary but fixed positive number. Intuitively, Ta is a stopping time in an
appropriate sense in the new continuous time setting. Now by the Optional Stopping
Theorem, E[WTa ] = eRu . Direct computation gives
E[WTa ] = E[e
R(u+cTa
PN(Ta )
k=1
+ E[eR(u+ct
Xk )
PN(t)
k=1
| u + cTa
Xk )
N(T
Xa )
Xk
k=1
N(T
Xa )
|u + cTa
0] P[u + cTa
N(T
Xa )
Xk
k=1
N(T
Xa )
Xk a] P[u + cTa
0]
Xk a].
k=1
k=1
Since this equation is valid for any fixed positive a, and since R > 0, limits can be
P a)
Ra
taken as a . Since lima P[u+cTa N(T
=0
k=1 Xk ) 0] = (u) and lima e
the following result is obtained.
Theorem. Suppose that in the collective risk model the adjustment coefficient R > 0
P
satisfies eRct MN(t) (ln(MX (R))) = 1 for all t. Let T = inf{s : u + cs N(s)
k=1 Xk 0}
be the random time at which ruin occurs. Then
eRu
(u) =
E[e
R(u+cT
PN(T)
k=1
Xk )
| T < ]
eRu .
Exercise 291. Why is the last inequality true?

As in the discrete time model, the existence of an adjustment coefficient guarantees that the ruin probability decreases exponentially as the initial surplus u
increases.
In general there is no guarantee that an adjustment coefficient will exist. For
certain particular types of models the adjustment coefficient can explicitly be found.
Moreover, a more detailed analysis of the claims process can be made in these
special cases.
The more restrictive discussion begins by examining the nature of the process
N(t), the total number of claims up to time t. A common assumption is that this
process is a Poisson process with constant intensity > 0. What this assumption
means is the following. Suppose W1 , W2 , . . . are independent identically distributed
exponential random variables with mean 1/ and common density e x 1(0,) (x).
154
The Ws are the waiting times between claims. The Poisson process can then
be viewed as the number of claims that arrive up to time t. This means that
P
N(t) = inf{k : k+1
j=1 Wj > t}. It can be shown that for any fixed t the random variable
N(t) has the Poisson distribution with parameter t and that the stochastic process
{N(t) : t 0} has independent increments, that is, whenever t1 < t2 < . . . < tn are
fixed real numbers then the random variables N(t2 ) N(t1 ), . . . , N(tn ) N(tn1 ) are
independent. Using this, direct computation gives
E[eN(t) ] =
ej P[N(t) = j]
j=0
ej e t ( t)j / j!
j=0
= e t
(e t)j / j!
j=0
t(e 1)
=e
This simple formula for the moment generating function of N(t) leads to a simple
formula for the adjustment coefficient in this case. The general equation for the
adjustment coefficient was earlier found to be eRct MN(t) (ln(MX (R))) = 1. Taking
logarithms and using the form of the moment generating function of N(t) shows that
the adjustment coefficient is the positive solution of the equation
+ cR = MX (R).
An argument similar to that given in the discrete time case can be used to show that
there is a unique adjustment coefficient in this setting.
Exercise 292. Verify that the adjustment coefficient, if it exists, must satisfy this
equation.
Example 291. Suppose all claims are for a unit amount. Then MX () = e so the
adjustment coefficient is the positive solution of + cR = eR . Note that there is
no solution if c . But in this case the ruin probability is clearly 1.
Exercise 293. Show that if c E[X] the ruin probability is 1. Show that if
c > E[X] the adjustment coefficient always exists and hence the ruin probability
is less than 1.
The previous exercises suggest that only the case in which c > E[X] is of
interest. Henceforth write c = (1 + ) E[X] for some > 0. Here is the relative
security loading.
Even more detailed information can be obtained when N(t) is a Poisson process.
To do this define a stopping time Tu = inf{s : U(s) < u} to be the first time that
155
the surplus falls below its initial level and denote by L1 = u U(Tu ) the amount by
which the surplus falls below its initial level. Then
Z
1
P[Tu < , L1 y] =
(1 FX (x)) dx.
(1 + )E[X] y
The proof of this fact is rather technical.

proof :
Let h > 0 be small. Then P[N(h) = 0] = e h 1, P[N(h) = 1] = he h h and

P[N(h) 2] 0. Denote by R(u, y) the probability that with an initial surplus of u the first
time the surplus drops below 0, the surplus actually drops below y. Conditioning on the
value of N(h) gives
Z
R(u, y) (1 h)R(u + ch, y) + h
R(u x, y)fX (x) dx +
fX (x) dx .
u+ch+y
Re-arranging gives
Z
R(u, y) R(u + ch, y)

= R(u + ch, y) +
ch
c
c
fX (x) dx.
u+ch+y
Now take limits as h 0 to obtain

Z
Z
u


fX (x) dx.
R (u, y) = R(u, y) +
c
c 0
c u+y
Since R(u, y) (u) eRu , both sides can be integrated with respect to u from 0 to .
Doing this gives
R(0, y) =
R(u, y) du +
0
Z
0
R(u x, y)fX (x) dx du +
fX (x) dx du.
0
u+y
Interchanging the order

Z of integration in the double integrals shows that the first double
integral is equal to
R(u, y) du, while the second double integral is equal to
0
xy
fX (x) du dx =
y
=
y
Z
=
(x y)fX (x) dx
xfX (x) dx yP[X y]
(1 FX (x)) dx
after integration by parts. Substitution now completes the proof after using c = (1+) E[X].
This formula has two useful consequences. First, by taking y = 0, the probability that the surplus ever drops below its initial level is 1/ (1 + ). Second, an
explicit formula for the size of the drop below the initial level is obtained as
1 Zy
P[L1 y| Tu < ] =
(1 FX (x)) dx.
E[X] 0
156
This expression can be evaluated in certain cases.

Exercise 294. Derive this expression for P[L1 y| Tu < ].
Exercise 295. What is the conditional distribution of L1 given Tu < if the claim
size has an exponential distribution with mean 1/ ?
Exercise 296. Show that the conditional moment generating function of L1 given
Tu < is (MX (t) 1)/ (tE[X]).
This information can also be used to study the random variable L which repreP
sents the maximum aggregrate loss and is defined by L = maxt0 { N(t)
k=1 Xk ct}.
Note that P[L u] = 1 (u) from which the distribution of L has a discontinuity at the origin of size 1 (0) = / (1 + ), and is continuous otherwise. In
fact a reasonably explicit formula for the moment generating function of L can be
obtained.
Theorem. If N(t) is a Poisson process and L = maxt0 {
ML () =
proof :
PN(t)
k=1
Xk ct} then
E[X]
.
1 + (1 + )E[X] MX ()
Note from above that the size of each new deficit does not depend on the initial starting
point of the surplus process. Thus
D
X
L=
Aj
j=1
where A1 , A2 , . . . are independent identically distributed random variables each having the
same distribution as the conditional distribution of L1 given Tu < , and D is a random
variable independent of the As which counts the number of times a new deficit level is
reached. From here the moment generating function of L can be computed using the
same methodology as earlier to obtain ML (t) = MD (ln MA (t)). Since D is geometric with
success probability 1/ (1 + ) and A has the same distribution as L1 , the computations can
be completed by substitution and simplification.
Exercise 297. Complete the details of the proof.

This formula for the moment generating function of L can sometimes be used
to find an explicit formula for the distribution function of L, and hence (u) =
1 P[L u].
Example 292. Suppose X is exponential with mean 1 and that = 1. Then
1
substitution gives ML (t) = 12 + 12 12t
, after using long division (or partial fractions).
The second term is half of the moment generating function of an exponential random
variable with mean 1/ 2, while the first term is half the moment generating function
157
of a random variable that is degenerate at zero. The ruin probability is therefore

(u) = P[L > u] = e2u / 2 for u > 0. Notice that (0) = 1/ 2.
The analysis of random sums of the form
N
X
Xj in which the Xs are inde-
j=1
pendent and identically distributed has played a key role in the preceding analysis. The distribution of such a sum is called a compound distribution, with
N as the compounding variable and X as the compounded variable. The conditioning method used earlier shows that the moment generating function of such
a sum is MN (ln MX (t)). The first two moments are E[
N
X
Xj ] = E[N]E[X] and
j=1
Var( Nj=1 Xj ) = E[N] Var(X) + (E[X])2 Var(N). These formulas are very useful computationally.
The case in which the compounding variable is Poisson is especially interesting. A random variable S has the compound Poisson distribution with Poisson
parameter and mixing distribution F(x), denoted CP( , F), if S has the same disP
tribution as Nj=1 Xj where X1 , X2 , . . . are independent identically distributed random
variables with common distribution function F and N is a random variable which is
independent of the Xs and has a Poisson distribution with parameter .
Example 293. For each fixed t, the aggregate claims process CP( t, FX ).
Example 294. If S has the CP( , F) distribution then the moment generating
function of S is
Z
MS () = exp{
(eu 1) dF(u)}.
This follows from the earlier general derivation of the moment generating function
of a random sum.
Exercise 298. Suppose that S has the CP( , F) distribution and T has the CP(, G)
distribution and that S and T are independent. Show that S + T has the CP( +
, + F + + G) distribution.
This last property is very useful in the insurance context. Because of this
property the results of the analysis of different policy types can be easily combined
into one grand analysis of the companys prospects as a whole. A compound Poisson
distribution can also be decomposed.
Example 295. Suppose each claim is either for $1 or $2, each event having
probability 0.5. If the number of claims is Poisson with parameter then the
amount of total claims, S, is compound Poisson distributed with moment generating
function
MS () = exp{0.5 (e 1) + 0.5 (e2 1)}.
158
Hence S has the same distribution as Y1 + 2Y2 where Y1 and Y2 are independent
Poisson random variables with mean / 2. Thus the number of claims of each size
are independent!
Example 296. The collective risk model can be used as an approximation to
the individual risk model. In the individual risk model the claim amount is often
represented by a product Bj Xj in which B is a Bernoulli random variable which
represents whether a claim is paid or not and X is the amount of the claim. Then
BX =
B
X
j=1
Xj
N
X
Xj
j=1
where N has a Poisson distribution with parameter P[B = 1] and X1 , X2 , . . . are

independent random variables each having the same distribution as X. Thus the
distribution of BX may be approximated by the CP(P[B = 1], FX ) distribution.
The analysis of the aggregate loss random variable L introduced the idea of
mixing. A mixture of distributions often arises when the outcome of an experiment
is the result of a two step process.
Example 297. A student is selected at random from the student population and
given an examination. The score S of the student could be modeled as the result of a
two stage process. The selection of the student from the student population could be
considered as the selection of an observation from a normal population with mean 70
and variance 100. This is a statement about the distribution of student abilities. Such
a model is stating that the average student score is 70. Let the random variable M
denote the result of this selection. The selected student then takes a particular exam,
and the score on the exam depends on the ability M and the variability caused by the
examination itself. So the final score might have the normal distribution with mean
M and variance 16. Conditioning on M shows that the moment generating function
2
2
of the score S on the examination is MS (t) = E[E[etS | M]] = E[etM+8t ] = e70t+58t ,
making use of the form of the moment generating function of the normal distribution.
So the score S is normally distributed with mean 70 and variance 116.
Example 298. In an earlier example the distribution of L was computed. The
distribution could be interpreted as the result of a two stage experiment. In the first
stage, the surplus either drops below its initial level or it doesnt. In the second
stage, the amount by which the surplus drops below the initial level is determined.
In the particluar case consider, with = 1, there is a 50% chance the surplus doesnt
drop below its initial level. So L = 0 with probability 1/ 2. If the surplus does drop
below its initial level, the size of the drop has an exponential distribution with mean
1/ 2.
Some additional examples of mixing are given in the problems.
159
Problems
Problem 291. If N has a Poisson distribution with parameter express P[N = k]
in terms of P[N = k 1]. This gives a recursive method of computing Poisson
probabilities.
Problem 292. Show that if X takes positive integer values and S has the CP( , FX )
P
distribution then x P[S = x] = k=1 kP[X = k]P[S = x k] for x > 0. This
is called Panjers recursion formula. Hint: First show, using symmetry, that
E[Xj |S = x, N = n] = x/ n for 1 j n and then write out what this means.
Problem 293. Suppose in the previous problem that = 3 and that X takes on the
values 1, 2, 3, and 4 with probabilities 0.3, 0.2, 0.1, and 0.4 respectively. Calculate
P[S = k] for 0 k 40.
Problem 294. Suppose S1 has a compound Poisson distribution with = 2 and
that the compounded variable takes on the values 1, 2, or 3 with probabilities 0.2,
0.6, and 0.2 respectively. Suppose S2 has a compound Poisson distribution with
parameter = 6 and the compounded variable takes on the values 3 or 4 with
probabilities 1/2 each. If S1 and S2 are independent, what is the distribution of
S1 + S2 ?
Problem 295. The compound Poisson distribution is not symmetric about its
mean, as the normal distribution is. One might therefore consider approximation of
the compound Poisson distribution by some other skewed distribution. A random
variable G is said to have the Gamma distribution with parameters and if G has
density function
1 x
fG (x) =
x e 1(0,) (x).
()
It is useful to recall the definition and basic properties of the Gamma function in this
connection. One easily computes the moments of such a random variable. In fact
the moment generating function is MG () = (/ ) . The case in which = 1/ 2
and 2 is a positive integer corresponds to the chisquare distribution with 2
degrees of freedom. Also the distribution of the sum of n independent exponential
random variables with mean 1/ is a gamma distribution with parameters n and .
For approximation purposes the shifted gamma distribution is used to approximate
the compound Poisson distribution. This means that an , , and x is found so that
x + G has approximately the same distribution as the compound Poisson variate.
The quantities x, , and are found by using the method of moments. The first
three central moments of both random variables are equated, and the equations are
then solved. Show that when approximating the distribution of a compound Poisson
160
random variable S the method of moments leads to

4[Var(S)]
=
[E[(S E[S])3 ]]2
3
4 E[X 2 ]
2
E[X 3 ]
2Var(S)
2E[X 2 ]
=
=
E[(S E[S])3 ]
E[X 3 ]
2
2 E[X 2 ]
2[Var(S)]2
= E[X]
.
x = E[S]
E[(S E[S])3 ]
E[X 3 ]
Problem 296. A random variable X is a mixture of exponential random variables
if the value of X is determined in the following way. Fix a number 0 < p < 1.
Perform a two stage experiment. In the first stage, select a number U at random in
the interval (0, 1). For the second stage, proceed as follows. If U < p select the
value of X to be the value of an exponential random variable with parameter 1 . If
U > p select the value of X to be the value of an exponential random variable with
parameter 2 . Show that the density of X is fX (x) = p1 e1 x + (1 p)2 e2 x for
x 0. Show that the moment generating function of X is E[etX ] = 1pt + 1p
.
2 t
Problem 297. What is the density of a random variable X with moment generating
function E[etX ] = (30 9t)/ 2(5 t)(3 t) for 0 < t < 3?
Problem 298. In the continuous time model, if the individual claims X have
density fX (x) = (3e3x + 7e7x )/ 2 for x > 0 and = 1, find the adjustment coefficient
and (u).
Problem 299. In the continuous time model, if the individual claims X are discrete
with possible values 1 or 2 with probabilities 1/ 4 and 3/ 4 respectively, and if the
adjustment coefficient is ln(2), find the relative security loading.
Problem 2910. Use integration by parts to show that the
adjustment coefficient in
Z
the continuous time model is the solution of the equation
erx (1FX (x)) dx = c/ .
0
Problem 2911. In the continuous time model, use integration by parts to find
ML1 (t). Find expressions for E[L1 ], E[L12 ] and Var(L1 ). Here L1 is the random
variable which is the amount by which the surplus first falls below its initial level,
given that this occurs.
Problem 2912. Find the moment generating function of the maximum aggregate
loss random variable in the case in which all claims are of size 5. What is E[L]?
Hint: Use the Maclaurin expansion of MX (t) to find the Maclaurin expansion of
ML (t).
161
Problem 2913. If (u) = 0.3e2u + 0.2e4u + 0.1e7u , what is the relative security
loading?
Problem 2914. If L is the maximum aggregate loss random variable, find expressions for E[L], E[L2 ], and Var(L) in terms of moments of X.
Problem 2915. In the compound Poisson continous time model suppose that
= 3, c = 1, and X has density fX (x) = (e3x + 16e6x )/ 3 for x > 0. Find the relative
security loading, the adjustment coefficient, and an explicit formula for the ruin
probability.
Problem 2916. In the compound Poisson continous time model suppose that
9x
= 3, c = 1, and X has density fX (x) = e3x/ 5 for x > 0. Find the relative security
25
loading, the adjustment coefficient, and an explicit formula for the ruin probability.
What happens if c = 20?
Problem 2917. The claim number random variable is sometimes assumed to have
the negative binomial distribution. A random variable N is said to have the negative
binomial distribution with parameters p and r if N counts the number of failures

before the rth success in a sequence of independent Bernoulli trials, each having
success probability p. Find the density and moment generating function of a random
variable N with the negative binomial distribution. Define the compound negative
binomial distribution and find the moment generating function, mean, and variance
of a random variable with the compound negative binomial distribution.
Problem 2918. In the case of fire insurance the amount of damage may be quite
large. Three common assumptions are made about the nature of the loss variables in
this case. One is that X has a lognormal distribution. This means that X = eZ where
Z is N(, 2 ). A second possible assumption is that X has a Pareto distribution.
This means that X has a density of the form x0 / x+1 1[x0 ,) (x) for some > 0. Note
that a Pareto distribution has very heavy tails, and the mean and/or variance may
not exist. A final assumption which is sometimes made is that the density of X is a
mixture of exponentials, that is,
fX (t) = (0.7)1 e1 t + (0.3)2 e2 t
for example. After an assumption is made about the nature of the underlying
distribution one may use actual data to estimate the unknown parameters. For each
of the three models find the maximum likelihood estimators and the method of
moments estimators of the unknown parameters.
Problem 2919. For automobile physical damage a gamma distribution is often
postulated. Find the maximum likelihood and method of moments estimators of the
unknown parameters in this case.
162
Problem 2920. One may also examine the benefits, in terms of risk reduction,
of using reinsurance. Begin by noting the possible types of reinsurance available.
First there is proportional reinsurance. Here the reinsurer agrees to pay a fraction
, 0 1, of each individual claim amount. Secondly, there is stoploss
reinsurance, in which the reinsurer pays the amount of the individual claim in
excess of the deductible amount. Finally, there is excess of loss reinsurance in
which the reinsurer pays the amount by which the claims of a portfolio of policies
exceeds the deductible amount. As an example, the effect of stoploss reinsurance
with deductible d on an insurers risk will be analyzed. The amount of insurers
risk will be measured by the ruin probability. In fact, since the ruin probability is
so difficult to compute, the effect of reinsurance on the adjustment coefficient will
be measured. Recall that the larger the adjustment coefficient, the smaller the ruin
probability. Initially (before the purchase of reinsurance) the insurers surplus at
time t is
U(t) = u + ct
N(t)
X
Xj
j=1
where c = (1 + ) E[X] and N(t) is a Poisson process with intensity . The

adjustment coefficient before the purchase of reinsurance is the positive solution of
+ cr = MX (r).
After the purchase of stop loss reinsurance with deductible d the insurers surplus is
U (t) = u + c t
N(t)
X
(Xj d)
j=1
where c = c reinsurance premium. Note that this process has the same structure
as the original one. The new adjustment coefficient is therefore the solution of
+ c r = MXd (r).
By examining the reinsurance procedure from the reinsurers standpoint the reinsurers premium is given by
(1 + ) E[(X d)1[d,) (X)]
where is the reinsurers relative security loading. With this information the new
adjustment coefficient can be computed. Carry out these computations when = 2,
= 0.50, = 0.25, d = 750, and X has a exponential distribution with mean 500.
Problem 2921. Repeat the previous problem for the case of proportional reinsurance.
Problem 2922. The case of excess of loss reinsurance leads to a discrete time
model, since the reinsurance is applied to a portfolio of policies and the reinsurance
163
is paid annually (say). The details here a similar to those in the discussion of the
discrete time gamblers ruin problem. Analyze the situation described in the previous
problem if the deductible for an excess of loss policy is 1500 and the rest of the
assumptions are the same. Which type of reinsurance is better?
Problem 2923. In the discrete time model, suppose the Xs have the N(10, 4)
distribution and the relative security loading is 25%. A reinsurer will reinsure a
fraction f of the total portfolio on a proportional basis for a premium which is 140%
of the expected claim amount. Find the insurers adjustment coefficient as a function
of f . What value of f maximizes the security of the ceding company?

Problem 292. By symmetry, E[X1 | P
S = x, N = n] = x/ n while direct com
putation gives E[X1 |S = x, N = n] =
k P[X1 = k| S = x, N = n]. Now
k=1P
P[X1 = k, S = x, N = n] = P[X1 = k, nj=1 Xj = x, N = n] = P[X1 =
P
P
k] P[ nj=2 Xj = x k] P[N = n] = P[X1 = k] P[ nj=2 Xj = x k] P[N = n
1] / n = P[X1 = k] P[S
P = x k, N = n 1] / n. Making this substitution gives
xP[S = x, N = n] = k=1 kP[X1 = k] P[S = x k, N = n 1]. Summing both
sides on n from 0 to gives the result.
Problem 294. The sum has a compound Poisson distribution with = 8.
Problem 296. Compute the distribution function of X by conditioning on
U to obtain FX (x) = P[X1 x] p + P[X2 x] (1 p) for x 0 where X1 and
X2 are exponentially distributed random variables with parameters 1 and 2
respectively.
Problem 297. Use partial fractions and the previous problem to see that X
is a mixture of two exponentially distributed random variables with parameters
1 = 3 and 2 = 5 and p = 1/ 4.
Problem 298. Here MX (t) = (5t 21)/ (t 3)(t 7) and E[X] = 5/ 21. This
leads to R = 1.69. Also ML (t) = 1/ 2 0.769/ (t 1.69) 0.280/ (t 6.20) using
partial fractions. Hence the density of L is fL (t) = 0.769e1.69t + 0.280e6.20t
together with a jump of size 1/ 2 at t = 0. (Recall that L has both a discrete and
absolutely continuous part.) Thus (u) = 0.454e1.69u + 0.045e6.20u .
Problem 299. Here = 10/ 7 ln(2) 1 = 1.0609.
Problem 2911. The density of L1 is fL1 (t) = (1 FX (t))/ E[X] for t 0.
Integration by parts then gives ML1 (t) = (MX (t) 1)/ tE[X]. Using the Maclaurin
E[X 2 ]
t+
expansion of MX (t) = 1 + tE[X] + t2 E[X 2 ]/ 2 + . . . then gives ML1 (t) = 1 +
2E[X]
E[X 3 ] 2
t + . . ., from which the first two moments of L1 can be read off.
6E[X]
2
E[X ]
Problem 2912. ML (t) = 5t/ (1 + 5(1 + )t e5t ) = 1 + t 2E[X]
+ . . ..
Problem 2913. Here = 2/ 3 since (0) = 1/ (1 + ).

Problem 2914. Substitute the Maclaurin expansion of MX (t) into the expression for moment generating function of L in order to get the Macluarin expansion
of ML (t).
1
1
Problem 2915. Here = 4/ 5 and R = 2. Also ML (t) = 4/ 9+(8/ 9) 2t
+(4/ 9) 4t
2u
4u
so that (u) = (4/ 9)e + (1/ 9)e .
9
(3/ 5 t)2 so that E[X] = 10/ 3 and = 9/ 10
Problem 2916. Here MX (t) = 25
when c = 1. Since < 0, there is no adjustment coefficient and the ruin
1
probability is 1. When c = 20, = 1 and R = 0.215. Also ML (t) = 0.119/ (t
2
164

0.215) + 0.044/ (t 0.834) by partial fractions. The density of the absolutely
continuous part of L is fL (t) = 0.119e.0.215t 0.044e0.834t , and the distribution
of L has a jump of size 1/ 2 at the origin. So (u) = 0.553e0.215u 0.053e0.834u .
Problem 2917. The compound negative binomial distribution is the distribuN
X
tion of the random sum
Xi where N and the Xs are independent, N has the
i=1
negative binomialdistribution, and the Xs all have the same distribution. Now
r
k
r
t r
P[N = k] = k+r1
r1 p (1 p) for k 0 and MN (t) = p (1 (1 p)e ) . Now use
the general result about the moment generating function of a random sum.
Problem 2920. Here MX (t) = (1 500t)1 for t < 1/ 500. The adjustment
coefficient before reinsurance is then R = 1/ 1500. The reinsurance premium is
(1 + 0.25)2E[(X 750)1(750,) (X)] = 278.91 and the insurers new adjustment
coefficient is the solution of 2 + (1500 278.91)R = MX750 (R) which gives
R = 0.00143.
Problem 2921. As in the preceding problem, R = 1/ 1500 before reinsurance.
Suppose the insurer retains 100(1)% of the liability. The reinsurance premium
is then (1 + ) E[X] = 1250. The adjustment coefficient after reinsurance
is then the solution of 2 + (1500 1250)R = 2M(1)X (R) = 2MX ((1 )R). So
R = (2)/ 500(5 2 11 +6) which is always at least 1/ 1500. Notice that since
< here, the insurer should pass off all of the risk to the reinsurer. By using
= 1 the insurer collects the difference between the original and reinsurance
premiums, and has no risk of paying a claim.
Problem 2922. The computational details here are quite
PN complicated. In
a time interval of unit length the total claims are C =
j=1 Xj where N is a
Poisson random variable with parameter . Now recall that in the discrete time
setting the adjustment coefficient is the solution of the equation E[eR(cC) ] = 1.
As before c = 1500. Also MC (t) = e (MX (t)1) . So the adjustment coefficient
before reinsurance is 1/ 1500. The reinsurance premium with deductible 1500
is (1 + )E[(C 1500)1(1500,) (C)] = 568.12. This is obtained numerically
by conditioning on the value of N and using the fact that conditional on N = k,
C has a gamma distribution with parameters = k and = 1/ 500. The new
adjustment coefficient solves E[eR(1500568.12C1500) ] = 1.
2
Problem 2923. Here MX (t) = e10t+2t , the premium income is 12.5 for each
time period, and the adjustment coefficient is the solution of e12.5t MX (t) = 1
which gives R = 1.25. The reinsurance premium is 14f so that after reinsurance
the adjustment coefficient satisfies e(12.514f )t MX ((1 f )t) = 1, which gives
R = (5 8f )/ 4(1 2f + f 2 ). The value f = 1/ 4 produces the maximum value of
R, namely 4/ 3.
165

Exercise 291. Since R > 0 and u + cT
denominator expectation is at least 1.
PN(T)
k=1
Xk 0 when T < the
Exercise 292. MU(t)u (R) = 1 holds if and only if ctR t(MX (R) 1) = 0,
which translates into the given condition.
Exercise 293. If c E[X] premium income is less than or equal to the
average rate of the claim process. So eventually the company will be ruined
by a run of above average size claims. By Maclaurin expansion, MX (R) =
1 + E[X]R + E[X 2 ]R2 / 2 + . . . and all of the coefficients are positive since X is a
positive random variable. So + cR MX (R) = (c E[X])R E[X 2 ]R2 / 2 . . .
is a function which is positive for R near 0 and negative for large values of R.
Thus there is some positive value of R for which this function is zero.
Exercise 294. From the definition of conditional probability, P[L1 y| Tu <
] = P[L1 y, Tu < ]/ P[Tu < ] and the result follows from the previous
formula and the fact that P[Tu < ] = 1/ (1 + ).
Exercise 295. Since in this case FX (t) = 1 e t for t > 0, direct substitution
gives P[L1 y| Tu < ] = 1 e y for y > 0.
Exercise 296. Given Tu < the density of L1 is (1FX (y))/ E[X] for y > 0. Using integration
by parts then gives the conditional moment
function of
Rgenerating
R
L1 as 0 ety (1FX (y))/ E[X] dy = ety (1 FX (y))/ tE[X]0 + 0 ety fX (y)/ tE[X] dy =
(MX (t) 1)/ tE[X]. Notice that the unconditional distribution of L1 has a jump of
size / (1 + ) at the origin. The unconditional moment generating function of
L1 is / (1 + ) + (MX (t) 1)/ (1 + )tE[X].
Since P[D = k]P
= (/ (1 + ))(1/ (1 + ))k for k = 0, 1, 2, . . .,
D
P
A
t
k
conditioning gives ML (t) = E[e j=1 j ] = E[MA (t)D ] =
k=0 MA (t) (/ (1 +
k
))(1/ (1 + )) = (/ (1 + ))/ (1 MA (t)/ (1 + )) = / (1 + MA (t)) and this
simplifies to the desired result using the formula of the previous exercise.
Exercise 297.
Exercise 298. Using the independence, MS+T () = MS ()MT () and the result
follows by substituion and algebraic rearrangement.
166

Question 301 . Lucky Tom finds coins on his way to work at a Poisson rate of 0.5
coins/minute. The denominations are randomly distributed with 60% of the coins
worth 1, 20% of the coins worth 5, and 20% of the coins worth 10. Calculate the
conditional expected value of the coins Tom found during his one-hour walk today,
given that among the coins he found exactly ten were worth 5 each.
A. 108
B.
115
D.
165
C.
128
E.
180
Question 302 . You are given that the claim count N has a Poisson distribution
with mean , and that has a gamma distribution with mean 1 and variance 2.
Calculate the probability that N = 1.
A. 0.19
B.
0.24
D.
0.34
C.
0.31
E.
0.37
Question 303 . A special purpose insurance company is set up to insure one single
life. The risk consists of a single possible claim. The claim is 100 with probability
0.60, and 200 with probability 0.40. The probability that the claim does not occur
by time t is 1/ (1 + t). The insurers surplus at time t is U(t) = 60 + 20t S(t), where
S(t) is the aggregate claim amount paid by time t. The claim is payable immediately.
Calculate the probability of ruin.
A. 4/ 7
B.
3/ 5
D.
3/ 4
C.
2/ 3
E.
7/ 8
168
Question 304 . Taxicabs leave a hotel with a group of passengers at a Poisson rate
of = 10 per hour. The number of people in each group taking a cab is independent
and is 1 with probability 0.60, 2 with probability 0.30, and 3 with probability 0.10.
Using the normal approximation, calculate the probability that at least 1050 people
leave the hotel in a cab during a 72 hour period.
A. 0.60
B.
0.65
D.
0.75
C.
0.70
E.
0.80
Question 305 . A company provides insurance to a concert hall for losses due
to power failure. You are given that the number of power failures in a year has a
Poisson distribution with mean 1, the ground up loss due to a single power failure
is 10 with probability 0.3, 20 with probability 0.3, and 50 with probability 0.4.
The number of power failures and the amounts of losses are independent. There is
an annual deductible of 30. Calculate the expected amount of claims paid by the
insurer in one year.
A. 5
B.
D.
12
C.
10
E.
14
Question 306 . An investment fund is established to provide benefits on 400

independent lives age x. On January 1, 2001, each life is issued a 10 year deferred
whole life insurance of 1000 payable at the moment of death. Each life is subject
to a constant force of mortality of 0.05. The force of interest is 0.07. Calculate the
amount needed in the investment fund on January 1, 2001, so that the probability,
as determined by the normal approximation, is 0.95 that the fund will be sufficient
to provide these benefits.
A. 55,300
B.
56,400
D.
59,300
C.
58,500
E.
60,100
169
Question 307 . You are given that the number of claims has mean 8 and standard
deviation 3, while the individual losses have a mean of 10,000 and a standard
deviation of 3,937. Using the normal approximation, determine the probability that
the aggregate loss will exceed 150% of the expected loss.
A. (1.25)
B.
(1.5)
D.
1 (1.5)
C.
1 (1.25)
E.
1.5(1)
Question 308 . An insurance company sold 300 fire insurance policies. One
hundred of the policies had a policy maximum of 400 and probability of claim per
policy of 0.05. Two hundred of the policies had a policy maximum of 300 and a
probability of claim per policy of 0.06. You are given that the claim amount for each
policy is uniformly distributed between 0 and the policy maximum, the probability
of more than one claim per policy is 0, and that claim occurrences are independent.
Calculate the variance of the aggregate claims.
A. 150,000
B.
300,000
D.
600,000
C.
450,000
E.
750,000
Question 309 . A risky investment with a constant rate of default will pay principal
and accumulated interest at 16% compounded annually at the end of 20 years if it
does not default, and zero if it defaults. A risk free investment will pay principal
and accumulated interest at 10% compounded annually at the end of 20 years. The
principal amounts of the two investments are equal. The actuarial present values of
the two investments are equal at time zero. Calculate the median time until default
or maturity of the risky investment.
A. 9
B.
10
D.
12
C.
11
E.
13
170
Question 3010 . For an insurer with initial surplus of 2 the annual aggregate claim
amount is 0 with probability 0.6, 3 with probability 0.3, and 8 with probability 0.1.
Claims are paid at the end of the year. A total premium of 2 is collected at the
beginning of each year. The interest rate is i = 0.08. Calculate the probability that
the insurer is surviving at the end of year 3.
A. 0.74
B.
0.77
D.
0.85
C.
0.80
E.
0.86
Question 3011 . X is a random variable for a loss. Losses in the year 2000 have
a distribution such that E[X d] = 0.025d2 + 1.475d 2.25 for d = 10, 11, . . . , 26.
Losses are uniformly 10% higher in 2001. An insurance policy reimburses 100% of
losses subject ot a deductible of 11 up to a maximum reimbursement of 11. Calculate
the ratio of expected reimbursements in 2001 over expected reimbursements in the
year 2000.
A. 110.0%
B.
110.5%
D.
111.5%
C.
111.0%
E.
112.0%
Question 3012 . Insurance for a citys snow removal costs covers four winter
months. There is a deductible of 10,000 per month. The insurer assumes that the
citys monthly costs are independent and normally distributed with mean 15,000
and standard deviation 2,000. To simulate four months of claim costs, the insurer
uses the Inverse Transform Method where small random numbers correspond to low
costs. The four numbers drawn from the uniform distribution on [0, 1] are 0.5398,
0.1151, 0.0013, and 0.7881. Calculate the insurers simulated claim cost.
A. 13,400
B.
14,400
D.
20,000
C.
17,800
E.
26,600
171
Question 3013 . A new insurance salesperson has 10 friends, each of whom is

considering buying a policy. Each policy is a whole life insurance of 1000 payable
at the end of the year of death. The friends are all age 22 and make their purchase
decisions independently. Each friend has a probability of 0.10 of buying a policy.
The 10 future lifetimes are independent. S is the random variable for the present
value at issue of the total payments to those who purchase the insurance. Mortality
follows the Illustrative Life Table and i = 0.06. Calculate the variance of S.
A. 9,200
B.
10,800
D.
13,800
C.
12,300
E.
15,400
172

Question 301 . If A, B, and C are the number of coins of the respective denominations found by Tom, then A, B, and C are independent Poisson random variables with
parameters 18, 6, and 6 (per hour). Thus E[A + 5B + 10C|B] = 18 + 50 + 60 = 128.
C.
Question 302 . Here P[N = 1|] = e , so P[N = 1] = E[e ]. Since
M (t) = E[et ] = (0.5/ 0.5t)0.5 from the given information, the desired probability is
the derivative of the moment generating function at t = 1. Thus P[N = 1] = 0.1924.
A.
Question 303 . The company is ruined if there is a claim of 100 before time 2
or a claim of 200 before time 7. So ruin occurs if there is any claim before time
2, or a claim of 200 between times 2 and 7. The ruin probability is therefore
2/ 3 + (7/ 8 2/ 3)(0.40) = 3/ 4. D.
Question 304 . The number N of cabs leaving the hotel in a 72 hour period
is Poisson with parameter 720. If A, B, and C are the number of 1, 2, and 3
person cabs then these random variables are independent Poisson random variables
with parameters 432, 216, and 72. The mean number of people leaving is then
432 + 2(216) + 3(72) = 1080 and the variance is 432 + 4(216) + 9(72) = 1944, and
the approximate probability is P[Z >= (1050 1080)/ 1944] = P[Z >= 0.680] =
0.751.D.
Question 305 . The expected amount is E[(L 30)+ ] = E[L 30] E[(L
30)1[0,30) (L)], where L is the loss due to power failure. Now P[L = 0] = e1 ,
P[L = 10] = 0.3e1 and P[L = 20] = 0.3e1 + (0.3)2 e1 / 2. So the last expectation is 30e1 (20)0.3e1 10(0.3e1 + (0.3)2 e1 / 2) = 14.51. Using this gives
E[(L 30)+ ] = 29 30 + 14.51 = 13.51. E.
Question 306 . The loss random variable for the ith policy is
Li = 1000e Ti 1[10,) (Ti ),
where Ti is the future lifetime of the ith policy holder. Thus
Z
E[Li ] = 1000
and
2
e t .05e.05t dt = 125.49
10
E[Li2 ]
= 1000
e2 t .05e.05t dt = 39, 300,
10
from which Var(Li ) = 23, 610. For the total loss S, E[S] = 50, 199 and Var(S) =
9440000. The amount required is 50199 + 1.645 9440000 = 55, 253. A.
173
Question 307 . The total loss T has E[T] = 8(10, 000) = 80, 000 and Var(T) =
8(3937)2 + 32 (10, 000)2 = 1023999752. Thus
q
P[T 120000] P[Z 40000/ Var(T)] = P[Z 1.25].

C.
Question 308 . Each of the one hundred policies has expected loss 200(0.05) = 10
and mean square loss (400)2 (.05)/ 3, giving the per policy variance as 2566.66. For
each of the two hundred policies the mean loss is 150(0.06) = 9 and the mean square
loss is (300)2 (0.06)/ 3, giving the per policy variance as 1719. The total variance is
therefore 100(2566.66) + 200(1719) = 600, 466. D.
Question 309 . The time T until default has an exponential distribution with
parameter . Hence (1.16)20 e20 = (1.10)20 and = 0.0531. The median of the
distribution of T is ln(1/ 2)/ = 13.05. (If this had turned out to be larger than 20,
the median of T 20, which is what is sought, would be 20.)E.
Question 3010 . Making a tree diagram with i = 0 shows that the survival probability is 0.738. This follows because the initial cash on hand is 2 + 2 = 4, so the
possible cash at the end of the first year is either 4, 1, or 4. The first two of these
cases lead to cash at the end of year 2 of 6, 3, or 2, or 3, 0, and 5. From here, the
probability of ruin in year 3 is easily determined. The only change using i = 0.08 is
to make the reserve strictly positive at a node where the reserve is 0 using i = 0. A.
Question 3011 . First note that (xa)+ a = x(2a)xa. For 2000, the expectation
is E[(X 11)+ 11] = E[X 22] E[X 11] = 7.15. For 2001 the expectation is
E[(1.1X 11)+ 11] = 1.1E[(X 10)+ 10] = 1.1(E[X 20] E[X 10]) = 7.975.
The ratio is 7.975/ 7.15 = 1.115. D.
Question 3012 . The corresponding z values are 0.10, 1.2, 3.0, and 0.80, from
the normal table. The corresponding costs are 15200, 12600, 9000, and 16600, and
the costs after the deductible are 5200, 2600, 0, and 6600, giving a total cost of
14,400. B.
Question 3013 . The mean of each policy is 0.1(1000)A22 and the second moment
is 0.1(1000)22 A22 from which the variance of each policy is 1584.30, using the
values from the table. The variance for all of the policies is therefore 15843. E.
31. Related Probability Models

In the next section a probability model is discussed which can be used for
transactions other than life insurance.
Discrete time Markov chains are often used as models for a sequence of random
variables which are dependent. One application of such stochastic processes is as a
model for the length of stay of a patient in a nursing home.
32. Discrete Time Markov Chains

In many situations the random variables which serve naturally as a model are
not independent. The simplest kind of dependence allows future behavior to depend
on the present situation.
Example 321. Patients in a nursing home fall into 3 categories, and each category
of patient has a differing expense level. Patients who can care for themselves with
minimal assistance are in the lowest expense category. Other patients require some
skilled nursing assistance on a regular basis and are in the next higher expense
category. Finally, some patients require continuous skilled nursing assistance and
are in the highest expense category. One way of modeling the level of care a
particular patient requires on a given day is as follows. Denote by Xi the level of
care this patient requires on day i. Here the value of Xi would be either 1, 2, or
3 depending on which of the 3 expense categories is appropriate for day i. The
random variables {Xi } are not independent.
Possibly the simplest type of dependence structure for a sequence of random
variables is that in which the future probabilistic behavior of the sequence depends
only on the present value of the sequence and not on the entire history of the
sequence. A sequence of random variables {Xn : n = 0, 1, . . .} is a Markov chain if
(1) P[Xn {0, 1, 2, 3, . . .}] = 1 for all n and
(2) for any real numbers a < b and any finite sequence of non-negative integers
t1 < t2 < < tn < tn+1 ,
P[a < Xtn+1 b| Xt1 , . . . , Xtn ] = P[a < Xtn+1 b|Xtn ].
The second requirement is referred to as the Markov property. Intuitively, the
Markov property means that the future behavior of the chain depends only on the
present and not on the more distant past.
The possible values of the chain are called states.
Exercise 321. Show that any sequence of independent discrete random variables
is a Markov chain.
Because of the simple dependence structure a vital role is played by the transition probabilities P[Xn+1 = j|Xn = i]. In principle, this probability depends not
only on the two states i and j, but also on n. A Markov chain is said to have
stationary transition probabilities if the transition probabilities P[Xn+1 = j|Xn = i]
do not depend on n. In the examples here, the transition probabilities will always
be assumed to be stationary, and the notation Pi,j = P[Xn+1 = j| Xn = i] will be used.
32: Discrete Time Markov Chains
176
would be
If the chain has non-stationary transition probabilities the notation P(i,j)
n
needed.
The transition probabilities are collected into the transition matrix of the chain.
When the transition probabilities are stationary, the transition matrix is P = [Pi,j ].
When the transition probabilities are not stationary the matrix Pn = [Pn(i,j) ] holds the
transition probabilities from the state occupied at time n.
Exercise 322. Show that if P is a transition matrix then
P
j
Pi,j = 1 for each i.
The transition probabilities together with the distribution of X0 determine completely the probabilistic behavior of the Markov chain.
Example 322.
nursing home example, suppose the transition
In the previous
0.9 0.05 0.05
0.1
matrix is P = 0.1 0.8
. The probability that a patient who enters at
0 0.05 0.95
time 0 in state 1 is in state 1 at time 1 and state 2 at time 2 is then 0.9 0.05, using
the first two entries in the first row of the transition matrix.
Example 323. In many cases, costs are associated with each state. In the nursing
home example, suppose the cost of being in state i for one day is i, and that this
expense must be paid at the end of the day. The expected present value of the costs
for the first two days of care for a patient entering in state 1 at time 0 would be
1v + (0.9 1 + 0.05 2 + 0.05 3)v2 . Here v is based on the daily interest rate. This
computation again uses the first row of the transition matrix.
Example 324. In more complicated models, costs may be associated with the
transitions between states. The typical convention is that the transitions occur
at the end of each time period and the transition costs are incurred at that time
point. The transition costs are typically collected into a matrix, with the (i, j) entry
of the matrix n C being the cost of a transition from state i to statej at time n.
0 1 1
Suppose in the nursing home example the transition costs are n C =

1 0 1
1 1 0
for n 1. Such a matrix indicates a cost of 1 for any day the patient changes care
level. Suppose the costs of being in a particular state are as before. The actuarial
present value of the expenses for the first 2 days for a patient entering in state 1 is
(1+0.11)v+(0.901+0.052+0.053+0.900.11+0.050.21+0.050.051)v2 .
The last terms in each case represent the expenses for state transitions at the end of
the day.
Example 325. The gamblers ruin problem illustrates many of the features of a
Markov chain. A gambler enters a casino with $z available for wagering and sits
177
down at her favorite game. On each play of the game, the gambler wins $1 with
probability p and loses $1 with probability q = 1 p. She will happily leave the
casino if her fortune reaches $a > 0, and will definitely leave, rather unhappily,
if her fortune reaches $0. Denote by Xn the gamblers fortune after the nth play.
Clearly {Xn } is a Markov chain with P[X0 = z] = 1. The natural state space here is
{0, 1, . . . , a}.
Exercise 323. Find the (a + 1) (a + 1) transition matrix.
Even with the simplifying assumption of stationary transition probabilities the
formula for the joint distribution of the values of the chain is unwieldy, especially
since in most cases the long term behavior of the chain is the item of interest. Fortunately, relatively simple answers can be given to the following central questions.
(1) If {Xn } is a Markov chain with stationary transition probabilities, what is the
limiting distribution of Xn ?
(2) If s is a state of a Markov chain with stationary transition probabilities how
often is the process in state s?
As a warm up exercise for studying these questions the n step transition
probabilities defined by Pni,j = P[Xn+m = j| Xm = i] and the corresponding n step
transition probability matrix P(n) will now be computed.
Exercise 324. Show that P[Xn+m = j| Xm = i] does not depend on m.
Theorem. The n step transition probability matrix is given by P(n) = Pn where P is
the transition probability matrix.
proof :
The case n = 1 being clear, the induction step is supplied.

Pni,j = P[Xn+m = j| Xm = i]
= P[[Xn+m = j]
!
[Xn+m1 = k] ]| Xm = i]
k=0
P[[Xn+m = j, Xn+m1 = k]| Xm = i]

k=0
P[Xn+m = j| Xn+m1 = k, Xm = i] P[Xn+m1 = k| Xm = i]

k=0
P[Xn+m = j| Xn+m1 = k] P[Xn+m1 = k|Xm = i]

k=0
Pk,j P(n1)
i,k .
k=0
The induction hypothesis together with the formula for the multiplication of matrices
conclude the proof.
178
When the transition probabilities are not stationary the matrix of probabilities
= P[Xn+k = j|Xn = i] is obtained by k Pn = [k P(i,j)
n ] = Pn Pn+1 . . . Pn+k1 .
(i,j)
k Pn
Using this lemma gives the following formula for the density of Xn in terms of
the density of X0 .
( P[Xn = 0] P[Xn = 1] . . . ) = ( P[X0 = 0] P[X0 = 1] . . . )Pn .
If the transition probabilites are not stationary, the matrix Pn must be replaced by
matrix product P0 P1 . . . Pn1 .
Exercise 325. Verify that this formula is correct.
Consequently, if Xn converges in distribution to Y as n then
( P[Y = 0] P[Y = 1] . . . ) = lim ( P[Xn = 0] P[Xn = 1] . . . )
n
= lim ( P[X0 = 0] P[X0 = 1] . . . )Pn
n
= n
lim ( P[X0 = 0] P[X0 = 1] . . . )Pn+1
= ( P[Y = 0] P[Y = 1] . . . )P
which gives a necessary condition for Y to be a distributional limit for the chain,
namely, the density of Y must be a left eigenvector of P corresponding to the
eigenvalue 1.
Example 326. For the nursing home chain given earlier there is a unique left
eigenvector of P corresponding to the eigenvalue 1, after normalizing so that the
sum of the coordinates is 1. That eigenvector is (0.1202, 0.1202, 0.7595). Thus a
patient will, in the long run, spend about 12% of the time in each of categories 1
and 2 and about 76% of the time in category 3.
Exercise 326. Find the left eigenvectors corresponding to the eigenvalue 1 of the
transition matrix for the gamblers ruin chain.

0 1
Example 327. Consider the Markov chain with transition matrix P =
.
1 0
This chain will be called the oscillating chain. The left eigenvector of P corresponding to the eigenvalue 1 is ( 1/ 2 1/ 2 ). If the chain starts in one of the states
there is clearly no limiting distribution.
Exercise 327. Show that this last chain does not have a limiting distribution.
The oscillating chain example shows that a Markov chain need not have a
limiting distribution. Even so, this chain does spend half the time in each state,
so the entries in the left eigenvector do have an intuitive interpretation as long run
fraction of the time the chain spends in each state.
179
The nursing home chain is an example in which the limiting behavior of the
chain does not depend on initial state of the chain. For the gamblers ruin chain,
the limiting behavior does depend on the initial state of the chain, as is intuitively
reasonable. The distinction between these two types of behavior can be understood
with a bit of effort.
For each state i define the random variable Ni to be the total number of visits
of the Markov chain to the state i. Possibly, Ni = . A state i for which P[Ni =
|X0 = i] = 1 is a state which is sure to be revisited infinitely many times. Such
a state is said to be recurrent. A non-recurrent state, that is, a state i for which
P[Ni = |X0 = i] < 1 is said to be transient. Amazingly, for a transient state i,
P[Ni = |X0 = i] = 0. Thus for each state i the random variable Ni is either always
infinite or never infinite.
Exercise 328. Show that if i is a transient state then Ni is a geometric random
variable, given that the chain starts at i.
Checking each state to see whether that state is transient or recurrent is clearly
a difficult task with only the tools available now. Another useful notion can greatly
simplify the job. The state j is accessible from the state i if there is a positive
probability that the chain can start in state i and reach state j. Two states i and j are
said to communicate, denoted ij, if each is accessible from the other.
Example 328. Consider the coin tossing Markov chain X in which Xn denotes the
outcome of the nth toss of a fair coin in which 1 corresponds to a head and 0 to a
tail. Clearly 01.
Example 329. In the gamblers ruin problem intuition suggests that the states 0
and a are accessible from any other state but do not communicate with any state
except themselves. Such states are absorbing. The other states all communicate
with each other.
Exercise 329. Prove that the intuition of the preceding example is correct.
Example 3210. In the nursing home example, all states communicate with each
other.
Importantly, if ij then i is recurrent if and only if j is recurrent.
Exercise 3210. For the coin tossing chain, is the state 1 recurrent?
Exercise 3211. What are the recurrent states for the gamblers ruin chain?
The existence of transient states in the gamblers ruin chain forced the limiting
180
distribution to depend on the initial state. The fact that all states of the nursing home
chain communicate caused the limiting behavior to not depend on the initial state.
In order to described completely the behavior of the limiting relative frequency
of the occupation time of a given state, some additional notation is required. Denote
by fi,j the probability that the chain ever enters state j given that the chain is currently
in state i. Denote by i,i the expected number of time steps between visits to state
i. (For a transient state, i,i = and it is possible for i,i = even for a recurrent
state.) The central result in the theory of Markov chains says that for any two states
i and j, given that the chain begins at time zero in state i,
lim
total number of visits to state j by time n fi,j

=
.
n
j,j
This is the result that was anticipated based on previous examples. For the gamblers
ruin chain, the expectations j,j only when j is one of the absorbing states, and
the limiting relative frequencies always depend on the initial state i. For the nursing
home and oscillating chains the probabilities fi,j = 1 for all i and j since all states
communicate and are recurrent. The limiting relative frequency does not depend on
the initial state in such cases.
The distinction between the existence of this limiting relative frequency and a
limiting distribution depends on the notion of the period of a state, and will not be
explored here.
181
Problems
Problem 321. Suppose the chain has only finitely many states all of which communicate with each other. Are any of the states transient?
Problem 322. Suppose Ni,j is the total number of visits of the chain to state j given
that the chain begins in state i. Show that for i j, E[Ni,j ] =
E[Nk,j ] Pi,k . What
k=0
happens if i = j?
Problem 323. Suppose the chain has both transient and recurrent states. Relabel
the states so that
states are listed first. Partition the transition matrix in
the transient

PT Q
to blocks P =
. Explain why the lower left block is a zero matrix. Show
0 PR
that the T T matrix of expectations E[Ni,j ] as i and j range over the transient states
is (I PT )1 .
Problem 324. In addition to the 3 categories of expenses in the nursing home example, consider also the possibilities of withdrawal from the home and death. Sup
0.8 0.05 0.01 0.09 0.05
0.5 0.45 0.04 0.0 0.01
pose the corresponding transition matrix is P =

0.05 0.15 0.70 0.0 0.10
0.0
0.0
0.0
1.0
0.0
0.0
0.0
0.0
0.0
1.0
where the states are the 3 expense categories in order followed by withdrawal and
death. Find the limiting distribution(s) of the chain. Which states communicate,
which states are transient, and what are the absorption probabilities given the initial
state?
Problem 325. An auto insurance company classifies insureds in 2 classes: (1)
preferred, and (2) standard. Preferred customers have an expected loss of $400 in
any one year, while standard customers have an expected loss of $900 in any one
year. A driver who is classified as preferred this year has an 85% chance of being
classified as preferred next year; a driver classified as standard this year has a 40%
chance of being classified as standard next year. Losses are paid at the end of each
year and i = 5%. What is the net single premium for a 3 year term policy for an
entering standard driver?

Problem 321. No. Since all states communicate, either all are transient or
all are recurrent. Since there are only finitely many states they can not all be
transient. Hence all states are recurrent.
Problem 322. The formula follows by conditioning on the first step leaving
X
state i. When i = j the formula is E[Ni,i ] = 1 +
E[Nk,i ] Pi,k , by the same
argument.
k=0
Problem 323. Going from a recurrent state to a transient state is not possible.
Express the equations of the previous problem in matrix form and solve.
Problem 324. The 3 expense category states communicate with each other and
are transient. The other 2 states are recurrent and absorbing. The probabilities
fi,j satisfy f0,4 = 0.8f0,4 + 0.05f1,4 + 0.05f2,4 + 0.09 and 2 other similar equations,
from which f0,4 = 0.382, f1,4 = 0.409, and f2,4 = 0.601.
Problem
325.
From the given information the transition matrix is P =

.85 .15
. The two year transition probabilities for an entering standard
.6 .4
driver are found from the second row of P2 to be ( .75 .25 ). The premium is
900v + (.6 400 + .4 900)v2 + (.75 400 + .25 900)v3 = 1854.90.
182

Exercise 321. Because of the independence both of the conditional probabilities in the definition are equal to the unconditional probability P[a < Xtn+1 < b].
Exercise 322.
P
j
Pi,j =
P
j
P[X1 = j| X0 = i] = P[X1 R |X0 = i] = 1.
1 0 0 0 0
q 0 p 0 0
0 q 0 p 0
Exercise 323.
0 0 q 0 p
. . . . .
. . . . .
. . . . .
0 0 0 0 0
...
...
...
...
..
.
...
0
0
0 .
..
.
1
Exercise 324. Use induction on n. The case n = 1 is true from the definition
of stationarity. For the induction
step assume the result holds when
P
P n = k. Then
P[Xk+1+m = j| Xm = i] = b P[Xk+1+m
=
j,
X
=
b|
X
=
i]
=
k+m
m
b P[Xk+1+m =
P
j|Xk+m = b]P[Xk+m = b| Xm = i] = b P[Xk+1 = j| Xk = b]P[Xk = b| X0 = i] =
P[Xk+1 = j| X0 = i], as desired.
P
P
Exercise 325. P[Xn = k] = i P[Xn = k|X0 = i]P[X0 = i] = i Pni,k P[X0 = i]
which agrees with the matrix multiplication.
Exercise 326. Matrix multiplication shows that the left eigenvector condition
implies that the left eigenvector x = (x0 , . . . , xa ) has coordinates that satisfy
x0 + qx1 = x0 , qx2 = x1 , pxk1 + qxk+1 = xk for 2 k a 2, pxa2 = xa1 and
pxa1 + xa = xa . From these equations, only x0 and xa can be non-zero, and these
two values can be arbitrary. Hence all left eigenvectors corresponding to the
eigenvalue 1 are of the form (c, 0, 0, . . . , 0, 1 c) for some 0 c 1.
Exercise 327. P[Xn = 1] is 0 or 1 depending on whether n is odd or even, so
this probability has no limit.
Exercise 328. Let p be the probability that the chain ever returns to state
i given that the chain starts in state i. Since i is transient, p < 1. Then
P[Ni = k| X0 = i] = pk (1 p) for k = 0, 1, . . ., since once that chain returns to i it
forgets it ever left.
Exercise 329. If the current fortune is i, and i is not 0 or a, then the fortune j
can be obtained in | j i| plays of the game by having | j i| wins (or losses) in
a row.
Exercise 3210. Yes, 1 is recurrent since state 1 is sure to be visited infinitely
often.
Exercise 3211. The only recurrent states are 0 and a.
183

Question 331 . In the state of Elbonia all adults are drivers. It is illegal to drive
drunk. If you are caught, your drivers license is suspended for the following year.
Drivers licenses are suspended only for drunk driving. If you are caught driving
with a suspended license, your license is revoked and you are imprisoned for one
year. Licenses are reinstated upon release from prison. Every year, 5% of adults
with an active license have their license suspended for drunk driving. Every year,
40% of drivers with suspended licenses are caught driving. Assume that all changes
in driving status take place on January 1, all drivers act independently, and the adult
population does not change. Calculate the limiting probability of an Elbonian adult
having a suspended license.
A. 0.019
B.
0.020
D.
0.036
C.
0.028
E.
0.047
Question 332 . For the Shoestring Swim Club, with three possible financial states
at the end of each year, State 0 means cash of 1500. If in state 0, aggregate member
charges for the next year are set equal to operating expenses. State 1 means cash
of 500. If in state 1, aggregate member charges for the next year are set equal to
operating expenses plus 1000, hoping to return the club to state 0. State 2 means
cash less than 0. If in state 2, the club is bankrupt and remains in state 2. The club
is subject to four risks each year. These risks are independent. Each of the four
risks occurs at most once per year, but may recur in a subsequent year. Three of the
four risks each have a cost of 1000 and a probability of occurrence 0.25 per year.
The fourth risk has a cost of 2000 and a probability of occurrence 0.10 per year.
Aggregate member charges are received at the beginning of the year, and i = 0.
Calculate the probability that the club is in state 2 at the end of three years, given
that it is in state 0 at time 0.
A. 0.24
B.
0.27
D.
0.37
C.
0.30
E.
0.56
185
Question 333 . Rain is modeled as a Markov process with two states. If it rains
today, the probability that it rains tomorrow is 0.50. If it does not rain today, the
probability that it rains tomorrow is 0.30. Calculate the limiting probability that it
rains on two consecutive days.
A. 0.12
B.
0.14
D.
0.19
C.
0.16
E.
0.22
186

Question 331
. If the states
are D(riving), S(uspended), and J(ailed) the transition
.95 .05 0
0 .4
matrix is
. The last column shows that 0.4 S = J while the
.6
1
0
0
second column shows that 0.05 D = S . Combining these with the requirement that
D + S + J = 1 gives S = 1/ 21.4 = 0.0467. E.
3
Question
! 332 . The transition probabilities are P00 = 0.9(0.75) = 0.3796, P01 =
3
0.9
(0.25)(0.75)3 = 0.3796, P02 = 0.2406, P10 = P00 , P11 = P01 , and P12 = P02 .
1
The desired probability is P02 +P00 P02 +P00 P00 P02 +P00 P01 P12 +P01 P12 +P01 P11 P12 +
P01 P10 P02 = 0.562. E.
Question 333 . Using the four states RR, RS, SR, and
SS for the occurrence
of
0.5 0.5 0
0
0
0 0.3 0.7
R(ain) or S(un) on pairs of days gives the transition matrix

.
0.5 0.5
0
0
0
0 0.3 0.7
The first two stationary equations show that RR = RS = SR while the last two
equations show that 0.7 SR = 0.3 SS . Thus RR = 3/ 16 = 0.1875. D.
34. Life Table at 6% Interest

x
0
5
10
15
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
lx
10,000,000
9,749,503
9,705,588
9,663,731
9,617,802
9,607,896
9,597,695
9,587,169
9,576,288
9,565,017
9,553,319
9,541,153
9,528,475
9,515,235
9,501,381
9,486,854
9,471,591
9,455,522
9,438,571
9,420,657
9,401,688
9,381,566
9,360,184
9,337,427
9,313,166
9,287,264
9,259,571
9,229,925
9,198,149
9,164,051
9,127,426
9,088,049
9,045,679
9,000,057
8,950,901
8,897,913
8,840,770
8,779,128
8,712,621
8,640,861
8,563,435
8,479,908
8,389,826
8,292,713
8,188,074
8,075,403
7,954,179
7,823,879
7,683,979
7,533,964
1000qx
20.42
0.98
0.85
0.91
1.03
1.06
1.10
1.13
1.18
1.22
1.27
1.33
1.39
1.46
1.53
1.61
1.70
1.79
1.90
2.01
2.14
2.28
2.43
2.60
2.78
2.98
3.20
3.44
3.71
4.00
4.31
4.66
5.04
5.46
5.92
6.42
6.97
7.58
8.24
8.96
9.75
10.62
11.58
12.62
13.76
15.01
16.38
17.88
19.52
21.32
a x
16.8010
17.0379
16.9119
16.7384
16.5133
16.4611
16.4061
16.3484
16.2878
16.2242
16.1574
16.0873
16.0139
15.9368
15.8561
15.7716
15.6831
15.5906
15.4938
15.3926
15.2870
15.1767
15.0616
14.9416
14.8166
14.6864
14.5510
14.4102
14.2639
14.1121
13.9546
13.7914
13.6224
13.4475
13.2668
13.0803
12.8879
12.6896
12.4856
12.2758
12.0604
11.8395
11.6133
11.3818
11.1454
10.9041
10.6584
10.4084
10.1544
9.8969
1000Ax
49.00
35.59
42.72
52.55
65.28
68.24
71.35
74.62
78.05
81.65
85.43
89.40
93.56
97.92
102.48
107.27
112.28
117.51
122.99
128.72
134.70
140.94
147.46
154.25
161.32
168.69
176.36
184.33
192.61
201.20
210.12
219.36
228.92
238.82
249.05
259.61
270.50
281.72
293.27
305.14
317.33
329.84
342.65
355.75
369.13
382.79
396.70
410.85
425.22
439.80
10002 Ax
25.92
8.45
9.37
11.33
14.30
15.06
15.87
16.76
17.71
18.75
19.87
21.07
22.38
23.79
25.31
26.95
28.72
30.63
32.68
34.88
37.26
39.81
42.55
45.48
48.63
52.01
55.62
59.48
63.61
68.02
72.72
77.73
83.06
88.73
94.76
101.15
107.92
115.09
122.67
130.67
139.11
147.99
157.33
167.13
177.41
188.17
199.41
211.13
223.34
236.03
10005 Ex
728.54
743.89
744.04
743.71
743.16
743.01
742.86
742.68
742.49
742.29
742.06
741.81
741.54
741.24
740.91
740.55
740.16
739.72
739.25
738.73
738.16
737.54
736.86
736.11
735.29
734.40
733.42
732.34
731.17
729.88
728.47
726.93
725.24
723.39
721.37
719.17
716.76
714.12
711.24
708.10
704.67
700.93
696.85
692.41
687.56
682.29
676.56
670.33
663.56
656.23
100010 Ex
541.95
553.48
553.34
552.69
551.64
551.36
551.06
550.73
550.36
549.97
549.53
549.05
548.53
547.96
547.33
546.65
545.90
545.07
544.17
543.18
542.11
540.92
539.63
538.22
536.67
534.99
533.14
531.12
528.92
526.52
523.89
521.03
517.91
514.51
510.81
506.78
502.40
497.64
492.47
486.86
480.79
474.22
467.12
459.46
451.20
442.31
432.77
422.54
411.61
399.94
100020 Ex
299.89
305.90
305.24
303.96
301.93
301.40
300.82
300.19
299.49
298.73
297.90
297.00
296.01
294.92
293.74
292.45
291.04
289.50
287.82
286.00
284.00
281.84
279.48
276.92
274.14
271.12
267.85
264.31
260.48
256.34
251.88
247.08
241.93
236.39
230.47
224.15
217.42
210.27
202.70
194.72
186.32
177.53
168.37
158.87
149.06
139.00
128.75
118.38
107.97
97.60
34: Life Table at 6% Interest

x
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
lx
7,373,338
7,201,635
7,018,432
6,823,367
6,616,155
6,396,609
6,164,663
5,920,394
5,664,051
5,396,081
5,117,152
4,828,182
4,530,360
4,225,163
3,914,365
3,600,038
3,284,542
2,970,496
2,660,734
2,358,246
2,066,090
1,787,299
1,524,758
1,281,083
1,058,491
858,676
682,707
530,959
403,072
297,981
213,977
148,832
99,965
64,617
40,049
23,705
13,339
7,101
3,558
1,668
727
292
108
36
11
1000qx
23.29
25.44
27.79
30.37
33.18
36.26
39.62
43.30
47.31
51.69
56.47
61.68
67.37
73.56
80.30
87.64
95.61
104.28
113.69
123.89
134.94
146.89
159.81
173.75
188.77
204.93
222.27
240.86
260.73
281.91
304.45
328.34
353.60
380.21
408.10
437.29
467.65
498.94
531.20
564.15
598.35
630.14
666.67
694.44
731.87
a x
9.6362
9.3726
9.1066
8.8387
8.5693
8.2988
8.0278
7.7568
7.4864
7.2170
6.9493
6.6836
6.4207
6.1610
5.9050
5.6533
5.4063
5.1645
4.9282
4.6980
4.4742
4.2571
4.0470
3.8442
3.6488
3.4611
3.2812
3.1091
2.9450
2.7888
2.6406
2.5002
2.3676
2.2427
2.1253
2.0152
1.9123
1.8166
1.7275
1.6450
1.5686
1.5005
1.4343
1.3812
1.3223
1000Ax
454.56
469.47
484.53
499.70
514.95
530.26
545.60
560.93
576.24
591.49
606.65
621.68
636.56
651.26
665.75
680.00
693.98
707.67
721.04
734.07
746.74
759.03
770.92
782.41
793.46
804.09
814.27
824.01
833.30
842.14
850.53
858.48
865.99
873.06
879.70
885.94
891.77
897.19
902.26
906.98
911.43
915.64
920.47
927.08
943.40
10002 Ax
249.20
262.83
276.92
291.46
306.42
321.78
337.54
353.64
370.08
386.81
403.80
421.02
438.42
455.95
473.59
491.27
508.96
526.60
544.15
561.57
578.80
595.79
612.51
628.92
644.96
660.61
675.83
690.59
704.86
718.61
731.83
744.50
756.61
768.13
779.08
789.44
799.21
808.39
817.00
825.03
832.53
839.25
845.83
851.12
857.04
10005 Ex
648.27
639.66
630.35
620.30
609.46
597.79
585.25
571.81
557.43
542.07
525.71
508.35
489.97
470.57
450.19
428.86
406.62
383.57
359.79
335.40
310.56
285.44
260.21
235.11
210.36
186.21
162.90
140.69
119.79
100.43
82.78
66.97
53.08
41.15
31.12
22.92
16.36
11.37
7.56
4.93
2.99
1.76
0.98
0.52
0.26
100010 Ex
387.53
374.36
360.44
345.77
330.37
314.27
297.51
280.17
262.31
244.03
225.46
206.71
187.94
169.31
151.00
133.19
116.06
99.81
84.59
70.56
57.83
46.50
36.61
28.17
21.13
15.42
10.91
7.47
4.93
3.13
1.90
1.10
0.60
0.31
0.15
0.07
0.03
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
188
100020 Ex
87.37
77.38
67.74
58.54
49.88
41.86
34.53
27.96
22.19
17.22
13.04
9.61
6.88
4.77
3.19
2.05
1.27
0.75
0.42
0.22
0.11
0.05
0.02
0.01
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00

x
0
5
10
15
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
a xx
16.1345
16.6432
16.4660
16.2187
15.9005
15.8272
15.7502
15.6696
15.5851
15.4967
15.4041
15.3073
15.2062
15.1005
14.9901
14.8750
14.7549
14.6298
14.4995
14.3640
14.2230
14.0766
13.9246
13.7670
13.6036
13.4344
13.2594
13.0786
12.8919
12.6994
12.5011
12.2971
12.0873
11.8720
11.6513
11.4252
11.1941
10.9580
10.7172
10.4720
10.2227
9.9696
9.7131
9.4535
9.1911
8.9266
8.6602
8.3926
8.1241
7.8552
1000Axx
86.73
57.93
67.96
81.96
99.97
104.12
108.48
113.04
117.82
122.83
128.07
133.55
139.27
145.26
151.50
158.02
164.82
171.90
179.27
186.94
194.92
203.21
211.81
220.74
229.99
239.56
249.47
259.70
270.27
281.16
292.39
303.94
315.81
328.00
340.49
353.29
366.37
379.74
393.37
407.24
421.35
435.68
450.20
464.90
479.75
494.72
509.80
524.95
540.15
555.36
10002 Axx
50.89
16.51
18.13
21.67
27.00
28.33
29.77
31.33
33.01
34.82
36.77
38.87
41.12
43.55
46.16
48.96
51.96
55.18
58.63
62.32
66.26
70.48
74.98
79.77
84.89
90.32
96.11
102.25
108.76
115.65
122.95
130.67
138.80
147.38
156.41
165.90
175.85
186.28
197.18
208.57
220.44
232.79
245.62
258.93
272.69
286.91
301.56
316.62
332.09
347.92
a x:x+10
16.2844
16.4093
16.1541
15.8187
15.3934
15.2962
15.1945
15.0883
14.9774
14.8617
14.7411
14.6154
14.4845
14.3484
14.2068
14.0598
13.9071
13.7488
13.5848
13.4150
13.2393
13.0579
12.8705
12.6774
12.4784
12.2737
12.0633
11.8474
11.6260
11.3994
11.1677
10.9311
10.6898
10.4441
10.1944
9.9409
9.6840
9.4240
9.1614
8.8966
8.6301
8.3623
8.0938
7.8249
7.5563
7.2885
7.0221
6.7574
6.4952
6.2360
1000Ax:x+10
78.24
71.17
85.62
104.60
128.67
134.18
139.94
145.95
152.22
158.77
165.60
172.71
180.12
187.83
195.84
204.16
212.80
221.76
231.05
240.66
250.60
260.88
271.48
282.41
293.68
305.26
317.17
329.39
341.92
354.75
367.87
381.26
394.92
408.82
422.96
437.31
451.85
466.57
481.43
496.42
511.50
526.66
541.86
557.08
572.28
587.44
602.53
617.50
632.34
647.02
189
10002 Ax:x+10
34.71
19.17
22.70
28.49
37.00
39.11
41.39
43.83
46.46
49.28
52.31
55.56
59.03
62.75
66.72
70.97
75.50
80.34
85.48
90.96
96.78
102.96
109.52
116.46
123.80
131.56
139.75
148.38
157.46
166.99
177.00
187.48
198.44
209.88
221.81
234.22
247.10
260.46
274.27
288.54
303.24
318.35
333.85
349.73
365.94
382.46
399.26
416.30
433.53
450.93

x
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
a xx
7.5866
7.3187
7.0520
6.7872
6.5247
6.2650
6.0088
5.7565
5.5086
5.2655
5.0278
4.7959
4.5700
4.3507
4.1381
3.9326
3.7344
3.5438
3.3607
3.1855
3.0181
2.8587
2.7071
2.5633
2.4274
2.2991
2.1784
2.0651
1.9590
1.8600
1.7678
1.6823
1.6032
1.5304
1.4635
1.4022
1.3466
1.2963
1.2510
1.2104
1.1741
1.1439
1.1147
1.0944
1.0715
1000Axx
570.57
585.74
600.83
615.82
630.68
645.37
659.88
674.16
688.19
701.95
715.41
728.54
741.32
753.74
765.77
777.40
788.62
799.41
809.77
819.69
829.16
838.19
846.77
854.91
862.60
869.86
876.70
883.11
889.11
894.72
899.93
904.77
909.25
913.38
917.16
920.63
923.78
926.63
929.19
931.49
933.54
935.25
936.90
938.05
939.35
10002 Axx
364.09
380.58
397.35
414.36
431.58
448.96
466.46
484.03
501.64
519.23
536.75
554.16
571.41
588.45
605.25
621.75
637.91
653.70
669.08
684.02
698.48
712.45
725.89
738.79
751.14
762.91
774.11
784.73
794.77
804.22
813.09
821.39
829.12
836.29
842.92
849.02
854.60
859.66
864.25
868.37
872.07
875.16
878.15
880.24
882.60
a x:x+10
5.9802
5.7283
5.4809
5.2385
5.0014
4.7701
4.5450
4.3263
4.1146
3.9099
3.7125
3.5227
3.3406
3.1663
2.9998
2.8412
2.6905
2.5476
2.4125
2.2851
2.1652
2.0527
1.9475
1.8493
1.7579
1.6731
1.5947
1.5226
1.4564
1.3958
1.3405
1.2920
1.2458
1.2091
1.1706
1.1395
1.1124
1.0892
1.0695
1.0531
1.0397
1.0289
1.0205
1.0141
1.0093
1000Ax:x+10
661.50
675.76
689.76
703.48
716.90
730.00
742.74
755.11
767.10
778.69
789.86
800.60
810.91
820.78
830.20
839.18
847.71
855.80
863.44
870.66
877.44
883.81
889.77
895.33
900.50
905.30
909.73
913.81
917.56
920.99
924.12
926.87
929.48
931.56
933.74
935.50
937.03
938.35
939.46
940.39
941.15
941.76
942.24
942.60
942.87
190
10002 Ax:x+10
468.44
486.02
503.62
521.21
538.72
556.11
573.34
590.36
607.12
623.59
639.71
655.46
670.79
685.67
700.08
713.99
727.37
740.21
752.49
764.20
775.34
785.89
795.86
805.25
814.05
822.29
829.96
837.06
843.63
849.67
855.21
860.10
864.78
868.49
872.43
875.61
878.39
880.78
882.81
884.50
885.89
887.00
887.87
888.54
889.03
35. Interest Rate Functions at 6%

m
1
2
4
12
i(m)
0.06000
0.05913
0.05870
0.05841
0.05827
d(m)
0.05660
0.05743
0.05785
0.05813
0.05827
i/ i(m)
1.00000
1.01478
1.02223
1.02721
1.02971
d/ d(m)
1.00000
0.98564
0.97852
0.97378
0.97142
(m)
1.00000
1.00021
1.00027
1.00028
1.00028
(m)
0.00000
0.25739
0.38424
0.46812
0.50985
36. Practice Examinations

The remaining sections contain practice examinations. These were constructed
from old Course 150 examinations of the Society of Actuaries and also some of the
relevant parts of Course 3 examinations. Each practice examination consists of 8
questions and should be completed in no more than 40 minutes.
37. Practice Examination 1

1. For an insurance company you are given
(1) Surplus at time t 1 is 1,220,000
(2) Reserves at time t are 4,805,000
(3) Premiums for year t are 850,000
(4) Expenses for year t are 350,000
(5) Investment Income for year t is 512,000
(6) Claims for year t are 335,000
(7) Net Income for year t is 552,000
Premiums and expenses are paid at the beginning of the year. Claims and investment
income are paid at the end of the year. Calculate the interest rate earned on assets
for year t.
A.
0.07
B.
0.08
C.
0.09
D.
0.10
E.
0.11
37: Practice Examination 1
194
2. Upon payment of a death benefit a beneficiary age 40 is given the following

options:
(1) a lump sum payment of 10,000
or
(2) an annual payment of K at the beginning of each year guaranteed for 10
years and continuing as long as the beneficiary is alive.
The two options are actuarially equivalent. You are given i = 0.04, A40 = 0.30,
A50 = 0.35, and A1
= 0.09. Calculate the value of K.
40:10
A.
538
B.
543
C.
545
D.
548
E.
549
3. A continuous whole life insurance is issued to (50). Z is the present value random
variable for this insurance. You are given
(1) Mortality follows DeMoivres Law with = 100
(2) Simple interest with i = 0.01
(3) bt = 1000 0.1t2
Calculate E[Z]
A.
250
B.
375
C.
500
D.
625
E.
750
195
4. A 25 year mortgage of 100,000 issued to (40) is to be repaid with equal annual

payments at the end of each year. A 25 year term insurance has a death benefit which
will pay off the mortgage at the end of the year of death including the payment then
due. You are given i = 0.05, a 40:25 = 14, and 25 q40 = 0.2. Calculate the net annual
premium for this term insurance.
A.
405
B.
414
C.
435
D.
528
E.
694
5. Which of the following are true?

I. t+u qx u qx+t for t 0 and u 0
II. u qx+t t|u qx for t 0 and u 0
III. If mortality follows DeMoivres Law, the median future lifetime of (x) equals
the mean future lifetime of (x).
A.
I and II
B.
I and III
C.
II and III
D.
All
E.
196
6. A fully continuous insurance policy is issued to (x) and (y). A death benefit of
10,000 is payable upon the second death. The premium is payable continuously
until the last death. The rate of annual premium is K while (x) is alive and reduces
to 0.5K upon the death of (x) if (x) dies before (y). Calculate K given that = 0.05,
ax = 12, ay = 15, and axy = 10.
A.
79.61
B.
86.19
C.
88.24
D.
93.75
E.
103.45
7. You are given

(1) v2 = 0.75
(2) qx+t = 0.2
(3) Ax+t+1 = 0.5
(4) 2 Ax+t+1 = 0.3
(5) k L is the random variable representing the prospective loss at the end of k
years for a fully discrete whole life insurance issued to (x)
Calculate
Var(t L)
.
Var(t+1 L)
A.
0.9
B.
1.0
C.
1.1
D.
1.2
E.
1.3
197
8. You are given that mortality follows DeMoivres Law with = 100 and that
x and y are independent lives both aged 90. Calculate the probability that the last
survivor of x and y will die between ages 95 and 96.
A.
0.05
B.
0.06
C.
0.10
D.
0.11
E.
0.20
198
Solutions to Practice Examination 1

1. Since Net Income is Premiums plus Investment Income minus Expenses, Claims,
and change in Reserves, the change in reserves is 125,000. So the reserves at time
t 1 are 4,680,000. Thus the assets at time t 1 are 5,900,000. The return on assets
is 512/ (5900 + 850 350) = 0.08. B.
2. The equation of value is 10000 = K(a10 + v10 10 p40 a 50 ). Since A50 = 1 da 50 ,
a 50 = 16.9. Since A40 = A1
+ v10 10 p40 A50 , v10 10 p40 = 0.60. Finally, a 10 = 8.433.
40:10
Using these values gives K = 538.3484. A.
3. The discount factor for an amount paid at time t under simple interest is 1/ (1 + it).
Z
Thus E[Z] = (1/ 50)

E.
50
(100 0.1t)/ (1 + 0.01t) dt = (1/ 50)
50
1000 10t dt = 750.
4. The present value of the benefit is Pa25K(40) v1 vK(40)+1 1[0,25) (K(40)), where
P = 100000/ a25 0.05 = 7095.25 is the annual mortgage payment. The expected
v26 25 q40 ). Now A1
= A40:25 v25 25 p40 =
value of this benefit is (P/ d)(A1
40:25
40:25
1 da 40:25 v25 25 p40 = 0.0971. Plugging in and dividing the expected value of the
benefit by a 40:25 gives the annual premium as 434.68. C.
5. I is true since t+u qx = t qx + t pxu qx+t t qxu qx+t + t pxu qx+t = u qx+t . II is true since
t| u qx = t pxu qx+t . III is true since under DeMoivres law, T(x) is uniformly distributed.
D.
6. Here Axy = Ax + Ay Axy = 1 ax + 1 ay (1 axy ) = 0.15. The annuity
for the premium has present value (K/ 2)axy + (K/ 2)ax . Thus 10000(0.15) = (29/ 2)K
and K = 103.448. E.
7. Here t L = vK(x+t)+1 P((1 vK(x+t)+1 )/ d) = (1 + P/ d)vK(x+t)+1 P/ d. Similarly,
K(x+t+1)+1
P/ d. The ratio of the variances is therefore (2 Ax+t
t+1 L = (1 + P/ d)v
A2x+t )/ (2 Ax+t+1 A2x+t+1 ). Now Ax+t = qx+t v + px+t vAx+t+1 = 0.5196, and 2 Ax+t = qx+t v2 +
px+t v22 Ax+t+1 = 0.33. Hence the variance ratio is 1.2. D.
8. The joint distribution of T(x) and T(y) is uniform over a square of side length 10.
The event in question only occurs if the pair (T(x), T(y)) lies in an L shaped strip
with vertices at (0, 5), (5, 5), (5, 0), (6, 0), (6, 6), and (0, 6). Since the area of this
strip is 11, the probability is 11/ 100. D.

1. From a life table with a one year select period you are given
x
85
86
l[x]
1000
850
d[x]
100
100
e [x]
5.556
Assume that deaths are uniformly distributed over each year of age. Calculate
e [86] .
A.
5.04
B.
5.13
C.
5.22
D.
5.30
E.
5.39
2. L is the loss random variable for a fully continuous whole life insurance of 1
issued to (x). You are given that the premium has been determined by the equivalence
principle, that Var(vT ) = 0.0344, and that E[vT ] = 0.166. Calculate Var(L).
A.
0.0239
B.
0.0495
C.
0.4896
D.
0.8020
E.
1.2470
200
3. For a double decrement table you are given

(1) Each decrement is uniformly distributed within each year of age in the
associated single decrement table.
()
(2) l60
= 1000
(2)
(3) q60
= 0.20
(1)
(4) d60
= 81
(1)
Calculate 60.5
.
A.
0.082
B.
0.086
C.
0.090
D.
0.094
E.
0.098
4. For a continuous whole life insurance (Z = vT , T 0), we have E[Z] = 0.25.

Assume the forces of mortality and interest are each constant. Calculate Var(Z).
A.
0.04
B.
0.08
C.
0.11
D.
0.12
E.
0.19
201
5. For a fully continuous whole life insurance of 1 issued to (x) the expense augmented loss variable is given as
Le = L + X
where
(1) L = vT P(Ax )aT
(2) X = I + (g e)aT
(3) I is the initial expenses
(4) g is the annual rate of continuous maintenance expense
(5) e is the annual expense loading in the premium
(6) = 0.05
(7) ax = 12
(8) Var(vT ) = 0.1
(9) g = 0.0010
(10) e = 0.0033
Net and expense loaded premiums are calculated according to the equivalence
principle. Calculate Var(Le ).
A.
0.252
B.
0.263
C.
0.278
D.
0.293
E.
0.300
202
6. An insurance company decides to waive all future premiums on a fully continuous

whole life insurance policy of 1000 issued to (x). The variance of the loss random
variable after the change is 81% of the variance of the loss random variable before
the change. The force of interest is 0.03. Calculate 1000P(Ax ).
A.
3.00
B.
3.33
C.
3.67
D.
3.90
E.
4.20
7. (I n a)x is equal to E[Y] where

(
Y=
if 0 T < n
(Ia)T
(Ia)n + n(n| aTn ) if T n.
You are given that x = 0.04 for all x and that = 0.06. Calculate
A.
ne0.1n
B.
10e0.1n
C.
e0.1n
D.
e0.1n
E.
10
8. Calculate
15 k45 = 0.15.
A.
0.55
B.
0.60
C.
0.65
D.
0.70
E.
0.75
15 V45:20
given that P45:20 = 0.03, A1
45:15
d
(I n a)x .
dn
= 0.06, d = 0.054, and
203

1. From the given table, l86 = 900 and l87 = 750. Also, e [85] = (1/ 2) + e[85] =
(1/ 2)+p[85] (1+e86 ), e [86] = (1/ 2)+e[86] = (1/ 2)+p[86] (1+e87 ) and e86 = p86 (1+e87 ). The
first equation gives e86 = 4.6178, using this value in the third gives (1+e87 ) = 5.5413,
and using this in the second gives e [86] = 5.3894. E.
2. Here L = (1 + P/ )vT P/ . Since E[L] = 0 by the equivalence principle,
P/ = E[vT ]/ (1 E[vT ]) = 0.1990. So Var(L) = (1 + P/ )2 Var(vT ) = 0.0495. B.
3. From the UDD assumption in the single decrement table, q(1) = q (1 0.5q ),
(1)
(1)
(1)
(1)
so q = 0.09 from the given information. Again by UDD, x+t
= q x / (1 tq x )
(1)
giving 60.5
= 0.0942. D.
(1)
4. Using the given information, E[Z] =

Z
2
(2)
e t et dt = / ( + ) = 1/ 4. Thus
0
2 t
3 = . Also E[Z ] =
e e
0
1/ 7 1/ 16 = 9/ 112 = 0.0804. B.
dt = / ( + 2 ) = 1/ 7. So Var(Z) =
5. Here Le = vT P(Ax )aT + I + (g e)aT = vT (1 + P(Ax )/ + (e g)/ ) + constants.

So Var(Le ) = Var(vT )(1 + P(Ax )/ + (e g)/ )2 = 0.2933. D.
6. Here writing P for 1000P(Ax ), Loriginal = (1000+P/ )vT P/ and Laf ter = 1000vT .
So the ratio of the variances is 10002 / (1000 + P/ )2 = 0.81. Thus P = 3.33. B.
7. The random variable Y is the expected present value of a cash stream which is
t if t nZ and n if t n.
Z Computing directly using the given information yields
n
E[Y] =
0
10e0.1n . B.
te0.1t dt + n
e0.1t dt = 100 100e0.1n . The derivative is therefore
8. Here 15 k45 = A1 / 15 E45 , so 15 E45 = 0.40 and A45:15 = 0.46. Thus a 45:15 = 10.
45:15
The retrospective method gives 15 V45:20 = (P45:20 a 45:15 A1 )/ 15 E45 = 0.60. B.
45:15

1. You are given that
(1) dx = k for x = 0, 1, 2, . . . 1
(2) e 20:20 = 18
(3) Deaths are uniformly distributed over each year of age.
Calculate 30|10 q30 .
A.
0.111
B.
0.125
C.
0.143
D.
0.167
E.
0.200
2. A whole life insurance pays a death benefit of 1 upon the second death of (x)
and (y). In addition, if (x) dies first a payment of 0.5 is payable at the time of his
death. Mortality follows Gompertz law. Calculate the net single premium for this
insurance.
cx
)
2cw
where cw = cx+y
A.
Aw (1 +
B.
2(Ax + Ay ) Aw (2 +
C.
Aw (1 +
D.
2(Ax + Ay ) Aw (2 +
E.
Ax + Ay Aw (1
cx
)
2cw
cx
)
2cw
where cw = cx+y
where cw = cx + cy
cx
)
2cw
cx
)
2cw
where cw = cx + cy
where cw = cx + cy
205
3. PAx is the net annual premium for a fully discrete whole life insurance of 1
calculated using mortality table A and interest rate i. PBx is the net annual premium
for a fully discrete whole life insurance of 1 calculated using mortality table B and
interest rate i. For all ages the probability of survival from age x to age x + 1 has the
relationship pAx = (1 + c)pBx , where the superscript identifies the table. Determine an
expression for PAx PBx in terms of functions based on table B.
A.
Ax (at interest rate i) Ax (at interest rate
B.
Ax (at interest rate
C.
1
1
(at interest rate i) (at interest rate
a x
a x
D.
1
(at interest rate
a x
E.
Px (at interest rate i) Px (at interest rate
ic
)
1+c
ic
)
1+c
ic
)
1+c
Ax (at interest rate i)
ic
)
1+c
1
(at interest rate i)
a x
ic
)
1+c
4. For a multiple decrement table with 3 decrements you are given that each decrement is uniformly distributed within each year of age in the associated single decre1
1
1
, qx(2) =
, and qx(3) =
. Calculate
ment table. Also you are given qx(1) =
20
10
19
(2)
qx .
A.
0.081
B.
0.091
C.
0.093
D.
0.095
E.
0.100
206
5. A whole life insurance has annual premiums payable at the beginning of the year
and death benefits payable at the moment of death. The following expenses are
allocated to this policy at the beginning of each year:
% of Premium
30%
10%
First Year
Renewal
Per 1000 of Insurance

3.00
0.00
Per Policy
150
50
You are given that Ax = 0.247 and that a x = 13. A level policy fee is used to
recognize per policy expenses in the expense loaded premium formula. Calculate
the minimum face amount such that the policy fee does not exceed 50% of the
expense loaded premium.
A.
2,650
B.
3,000
C.
3,450
D.
5,300
E.
6,000
6. For a select and ultimate mortality table with a one year select period q[x] = 0.5qx .
Determine Ax A[x] .
(1 A[x] )
A.
2A1
B.
A1
(1 Ax+1 )
C.
A1
(1 A[x+1] )
D.
0.5 A1
(1 Ax+1 )
E.
0.5 A1
(1 Ax )
[x]:1
[x]:1
[x]:1
[x]:1
[x]:1
207
7. A multiple decrement table has two causes of decrement: (1) accident; (2) other
than accident. You are given
(1) y(1) = A for some A > 0
(2) y(2) = Bcy for some B > 0, c > 1
What is the probability that (x) dies due to accident?
A.
A
e x
B.
e x
A
C.
Aex
D.
1
Aex
E.
A
e x
8. You are given that 10 E30 = 0.35, a30:9 = 5.6, and i = 0.10. Calculate A1
30:10
A.
0.05
B.
0.10
C.
0.15
D.
0.20
E.
0.25
208

1. The first and third assumptions imply that lifetimes obey DeMoivres law. Since
30| 10 q30 = P[30 T(30) 40], evidently > 60 since none of the answers are zero.
Now FT(20:20 ) (t) = t q20:20 = t/ ( 20) for 0 t < 20 and is 1 for larger t. Thus the
Z
20
t dt + 20(1
distribution has a jump at t = 20, so that 18 = e 20:20 = (1/ ( 20))
0
20/ ( 20)) = 20 200/ ( 20). Thus = 120 and the desired probability is
10/ ( 30) = 1/ 9 = 0.111. A.
2. Choose
w so that cw = cx + cy . Then t pxy = t pw under Gompertz law, and
Z
A1xy =
e t t pxy x+t dt = Aw (cx / cw ). The premium is Axy + (1/ 2)A1xy = Ax + Ay
Axy + (1/ 2)Aw (cx / cw ) = Ax + Ay Aw + (1/ 2)Aw (cx / cw ) = Ax + Ay Aw (1 cx / cw ). E.

3. Here PA PB = (1da Ax )/ a Ax (1 da Bx )/ a Bx = 1/ a Ax 1/ a Bx . Now a Ax =
X
k=0
vk k pAx =
k=0
vk (1 + c)k k pBx = a Bx at interest rate i where 1/ (1 + i ) = (1 + c)/ (1 + i), that is
i = (i c)/ (1 + c). D.
Z
4. Here q =
dt =
dt = q
(1 t/ 20)(1
tp
t p t p t p
0
0
0
t/ 19) dt = 0.09478, upon computing the integral by multiplying out the integrand.
D.
(2)
()
(2)
x+t
(1)
(3)
(2)
(2)
x+t
(2)
5. B.
6. Here Ax = vqx + vpx Ax+1 and A[x] = vq[x] + vp[x] Ax+1 since the select period is 1
year. Thus Ax A[x] = (1/ 2)vqx (1 Ax+1 ) = vq[x] (1 Ax+1 ) = A1 (1 Ax+1 ). B.
[x]:1
7. The probability is
0
() (1)
t px+t x+t
dt = Aex . C.
8. Here a 30:10 = 1+a30:9 = 6.6. So A1

= A30:10 10 E30 = 1da 30:10 0.35 = 0.05.
30:10
A.

1. You are given the following extract from a 3 year select and ultimate mortality
table:
x
70
71
72
l[x]
l[x]+1
l[x]+2
7984
8016
7592
lx+3
7600
x+3
73
74
75
Assume that the ultimate table follows DeMoivres Law and that d[x] = d[x]+1 =
d[x]+2 for x = 70, 71, 72. Calculate 1000(2| 2 q[71] ).
A.
26.73
B.
32.43
C.
43.37
D.
47.83
E.
48.99
2. You are given 20 P25 = 0.046, P25:20 = 0.064, and A45 = 0.640. Calculate P1
25:20
A.
0.008
B.
0.014
C.
0.023
D.
0.033
E.
0.039
210
3. You are given that L is the loss random variable for a fully continuous whole life
insurance issued to (25), and that Var(L) = 0.2, A45 = 0.7, and 2 A25 = 0.3. Calculate
20 V(A25 ).
A.
0.3
B.
0.4
C.
0.5
D.
0.6
E.
0.7
4. (x) and (y) purchase a joint-and-survivor annuity due with an initial monthly
benefit amount equal to 500. You are given
(1) If (x) predeceases (y) the benefit amount changes to 300 per month
(2) If (y) predeceases (x) the benefit changes to B per month
(3) The annuity is actuarially equivalent to a single life annuity due on (x) with
a monthly benefit amount equal to B
(4) a (12)
= 10
x
= 14
(5) a (12)
y
(12)
(6) a xy = 8
Calculate B
A.
520
B.
680
C.
725
D.
800
E.
1025
211
(1)
2
5. You are given the following for a double decrement table: x+0.5
= 199
, q(2)
x = 0.01,
and each decrement is uniformly distributed over each year of age in its associated
single decrement table. Calculate 1000q(1)
x .
A.
9.95
B.
10.00
C.
10.05
D.
10.10
E.
10.15
6. For a single premium, continuous whole life insurance issued to (x) with face
amount f you are given
(1) Ax = 0.2
(2) Percent of premium expenses are 8% of the expense loaded premium
(3) Per policy expenses are 75 at the beginning of the first year and 25 at the
beginning of each subsequent year
(4) Claim expenses are 15 at the moment of death
(5) i = 5%
(6) Deaths are uniformly distributed over each year of age
(7) The expense loaded premium is expressed as gf + h
Calculate h.
A.
505
B.
508
C.
511
D.
514
E.
517
212
7. An individual age 30 purchases a fully continuous 50,000 20 year endowment

policy. At the end of 10 years she surrenders the policy in return for reduced
paid up insurance. You are given A40:10 = 0.538, a45:5 = 4.18, P(A30:20 ) = 0.027,
= 0.06, and cash values are equal to the net level premium reserves. Calculate
the reserve on this reduced amount of insurance five years after the origianal policy
was surrendered.
A.
17,600
B.
19,500
C.
23,000
D.
24,100
E.
26,200
8. You are given that mortality follows DeMoivres Law and that Var(T(50)) = 192.
Calculate .
A.
98
B.
100
C.
107
D.
110
E.
114
213

1. The given facts imply in order l75 = 7380, l74 = 7490, l[71]+2 = 7737, and
l[71] = 8231. The desired probability is (l[71]+2 l75 )/ l[71] = 43.37/ 1000. C.
2. Here A25 = 20 P25 a 25:20 and A25:20 = P25:20 a 25:20 . Also A25 = A1
+ 20 E25 A45 .
25:20
Equating this to the first expression for A25 gives 20 P25 = P1
+ 20 E25 A45 / a 25:20 .
25:20
Since A25:20 = A1 + 20 E25 , P25:20 = P1 + 20 E25 / a 25:20 . Using these two equations
25:20
25:20
involving P1
to eliminate the term involving E gives P1
= 0.0140. B.
25:20
25:20
3. Here L = (1 + P/ )vT P/ , so Var(L) = (1 + P/ )2 Var(vT ). From E[L] = 0,

A25 = (P/ )/ (1 + P/ ). Using these two facts gives a quadratic equation in P/ ,
from which P/ = 1. Finally, 20 V(A25 ) = A45 (P/ )(1 A45 ) = 2A45 1 = 0.40. B.
4. The equation of value is 12Ba(12)
= 3600a(12)
a(12)
a(12)
x
y + 12B
x + (2400 12B)
xy , from
which the given information yields B = 725. C.
(1)
5. Here x+0.5
= q x / (1 0.5q x ), from which q x = 0.01. Also q(2) = q (1
(1)
(2)
(1)
(2)
0.5q ) from which q = 2/ 199 and q(1) = q (1 0.5q ) = 198/ (199)(100) =
9.949/ 1000.A.
(1)
(1)
(1)
(2)
6. The equation of value is f Ax + 0.08P + 50 + 25ax + 15Ax = P, from which

h = (50 + 25ax + 15Ax )/ 0.92. Since 1 da x = Ax = ( / i)Ax , computation gives
h = 516.89. E.
7. C.
8. Since T(50) is uniform on the interval (0, 50), Var(T(50)) = ( 50)2 / 12,
Thus = 98. A.

1. Y is the present value random variable for a 30 year temporary life annuity of 1
payable at the beginning of each year while (x) survives. You are given i = 0.05,
2
= 0.0694, and A1x:30 = 0.1443. Calculate E[Y 2 ].
30 px = 0.7, A1
x:30
A.
35.6
B.
47.1
C.
206.4
D.
218
E.
233.6
2. You are given 10 V25 = 0.1 and 10 V35 = 0.2. Calculate 20 V25 .
A.
0.22
B.
0.24
C.
0.26
D.
0.28
E.
0.30
3. A multiple decrement table has two causes of decrement: (1) death by accident
and (2) death other than by accident. You are given that a fully continuous whole
life insurance issued to (x) pays 1 on a non-accidental death and 2 if death results by
(1)
accident and that x+t
= , the force of interest. Calculate the net single premium
for this insurance.
A.
(1 + )Ax
B.
(2 + )Ax
C.
Ax + 1
D.
2 Ax
E.
215
4. The expense loaded premium, G, for a fully discrete 3 year endowment insurance
of 1000 issued to (x) is calculated using the equivalence principle. Expenses are
paid at the beginning of each year. You are given
(1) 1000Px:3 = 323.12
(2) G = 402.32
(3) qx = 81
(4) qx+1 = 17
(5) i = 0.10
(6)
Expenses
Percentage of Premium
Per Policy
First Year
30%
8
Renewal
10%
4
Calculate the expense reserve at the end of the first year.
A.
40
B.
54
C.
62
D.
65
E.
71
5. You are given that A1x:n = 0.4275, = 0.055, and x+t = 0.045 for all t. Calculate
Ax:n .
A.
0.4600
B.
0.4775
C.
0.4950
D.
0.5245
E.
0.5725
216
6. L is the loss random variable for a fully discrete 2 year term insurance of 1 issued
to (x). The net level annual premium is calculated using the equivalence principle.
You are given qx = 0.1, qx+1 = 0.2, and v = 0.9. Calculate Var(L).
A.
0.119
B.
0.143
C.
0.160
D.
0.187
E.
0.202
7. You are given that male mortality is based on a constant force of mortality
with = 0.04, and that female mortality follows DeMoivres Law with = 100.
Calculate the probability that a male age 50 dies after a female age 50.
A.
0.5(1 3e2 )
B.
0.5(1 e2 )
C.
0.5
D.
0.5(1 + e2 )
E.
0.5(1 + 3e2 )
217
8. For a fully discrete participating whole life insurance of 1000 issued to (35) you
are given
(1) The fund share is equal to the cash value
(2) Dividends are paid at the end of each year, up to and including the year of
death or withdrawal
(3) The cash value at the end of 20 years is 304
(4) i = 0.04
(d)
(5) q(d)
54
= 0.005
54 = 0.010; q
(6) e19 = 4.00; e 19 = 3.00
(7) G = 15.00
(8) 20 D = 15.02
Calculate i.
A.
7.5%
B.
7.7%
C.
7.9%
D.
8.1%
E.
8.3%
218

1. Here Y = (1/ d)(1 vK+1 1[0,30) (K) v30 1[30,) (K)), so that d2 E[Y 2 ] = 1 2(A1x:30 +
v30 30 px ) + 2 A1x:30 + v60 30 px , which gives E[Y 2 ] = 218.00. D.
2. Notice that P25 = A25 / a 25 = (1 da 25 )/ a 25 , so that P25 + d = 1/ a 25 . Thus
25 = 1 (P25 + d)a25 = 1 a 35 / a 25 . Similarly, 10 V35 = 1 a 45 / a 35
10 V25 = A35 P25 a
and 20 V25 = 1 a 45 / a 25 . Hence 20 V25 = 1 a 45 / a 25 = 1 (1 10 V35 )(1 10 V25 ) =
1 (.9)(.8) = .28. D.
Z
3. The net single premium is Ax +
(1)
e t t p()
x x+t dt = Ax + ax = 1. E.
4. C.
n
0.1n
5. On one hand Ax:n = A1x:n + v n px = A1x:n + e
. Also A1x:n =
e t et () dt =
(/ + )(1 e0.1n ). Solving this last equation gives e0.1n = 0.05, and using this in
the first equation gives Ax:n = 0.4775. B.
6. Here L = v1{0} (K) + v2 1{1} (K) P Pv1[1,) (K). Since E[L] = 0, P = 0.1303.
Then squaring gives Var(L) = E[L2 ] = v2 qx + v4 px qx+1 + P2 + P2 v2 px 2PVqx
2Pv3 px qx+1 + 2P2 vpx = 0.1603. C.
Z
7. Conditioning on the time of death of the female gives

e2 )/ 2. B.
8. A.
50
(1/ 50)e0.4t dt = (1

1. Assume mortality follows DeMoivres Law for 0 x < . Which of the following expression equal x ?
mx
I.
, for x 1
1 + 0.5mx
II. n| qx for 0 n x 1
1
III.
e x
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
2. For a double decrement table where cause 1 is death and cause 2 is withdrawal
you are given
(1) Deaths are uniformly distributed over each year of age in the associated
single decrement table
(2) Withdrawals occur at the beginning of each year
()
(3) l20
= 1000
(2)
(4) q20 = 0.25
(1)
(2)
(5) d20
= 0.04 d20
Calculate q20
(1) .
A.
0.0089
B.
0.0100
C.
0.0114
D.
0.0133
E.
0.0157
220
3. Z is the present value random variable for a special continuous whole life insurance issued to (x). You are given for all t that x+t = 0.01, t = 0.06, and bt = e0.05t .
Calculate Var(Z).
A.
0.033
B.
0.037
C.
0.057
D.
0.065
E.
0.083
4. An insurance benefit pays 1 at the later of n years or the failure of the status xy.
Which of the following correctly express the net single premium for this benefit?
I. vn n qxy + vn n pxy Ax+n:y+n
II. Axy Axy:n

III. vn n px Ax+n + n py Ay+n n pxy Ax+n:y+n + n qxy
A.
None
B.
I only
C.
II only
D.
III only
E.
5. You are given 5 p50 = 0.9, 5 p60 = 0.8, q55 = 0.03, and q65 = 0.05. Calculate
5| q50:60 .
A.
0.0011
B.
0.0094
C.
0.0105
D.
0.0565
E.
0.0769
221
6. A whole life insurance issued to (25) provides the following benefits.

(1) the death benefit, payable at the end of the year of death, is equal to 20,000
up to age 65 and 10,000 thereafter
(2) the net single premium is refunded at age 65 if the insured is still alive
You are given
(1) A25 = 0.1
(2) A65 = 0.2
(3) 40 p25 = 0.8
(4) v40 = 0.2
Calculate the net single premium for this insurance.
A.
2,000
B.
2,400
C.
3,000
D.
4,000
E.
4,800
7. L is the lossatissue random variable for a fully continuous whole life insurance
of 1 with premiums based on the equivalence principle. You are given
(1) E[v2T ] = 0.34
(2) E[vT ] = 0.40.
Calculate Var(L).
A.
0.080
B.
0.300
C.
0.475
D.
0.500
E.
1.125
222
8. You are given

(1) Z is the present value random variable for an insurance on the lives of (x)
and (y) where
(
vT(y) T(x) T(y)
Z=
0
T(x) > T(y)
(2) (x) is subject to a constant force of mortality 0.07
(3) (y) is subject to a constant force of mortality 0.09
(4) (x) and (y) are independent lives
(5) = 0.06.
Calculate E[Z].
A.
0.191
B.
0.318
C.
0.409
D.
0.600
E.
0.727
223

Z
1. Since mx = qx /
0
n| qx
dt = 2qx / (2 qx ), mx / (1 + 0.5mx ) = qx = x under UDD,
t px
and I is ok. Also,

= 1/ ( x) = qx = x , so II holds. III fails since e x = ( x)/ 2
under DeMoivre. A.
2. Here t p 20 = 750/ 1000 for t > 0 since withdrawals occur at the beginning of each
(2)
(1)
year and q(2)
20 = 0.25. Using this gives q20 =
0.75q
Since
= 250,
0.01/ 0.75 = 0.0133. D.
(1)
20 .
(2)
d20
(1)
d20
= 10 and
Z
3. Using the given information E[Z] =

Z
() (1)
t p20 20+t dt =
q(1)
20
tp
(1)
(1)
20+t
t p 20 dt =
(2)
= 0.01. Using this gives q 20 =

(1)
e0.07t+0.05t 0.01 dt = 1/ 2 and E[Z 2 ] =
e0.1t0.13t 0.01 dt = 1/ 3, so that Var(Z) = 1/ 12 = 0.0833. E.
4. I fails because n pxy should be n pxy . II fails since the formula does not account for
the case in which xy dies before n. III holds since it is equivalent to the corrected
form of I. D.
5. Direct reasoning gives P[5 T(50 : 60) 6] = P[T(50) 5]P[5 T(60)
6] + P[T(60) 5]P[5 T(50) 6] + P[5 T(50) 6]P[5 T(60) 6] =
(0.1)(0.8)(0.05) + (0.2)(0.9)(0.03) + (0.9)(0.03)(0.8)(0.05) = 0.01048. C.
6. The equation for the premium P is P = 20000A25 10000v40 40 p25 A65 + Pv40 40 p25 ,
from which P = 2000 using the given information. A.
7. Here L = vT PaT = (1 + P/ )vT P/ . Since E[L] = 0, P/ = A/ (1 A) = 2/ 3.
Thus Var(L) = (1 + P/ )2 Var(vT ) = (1 + 2/ 3)2 (0.34 (0.40)2 ) = 0.50. D.
8. The given information and definition of Z gives
Z
E[Z] =
0
e0.06y (0.07)e0.07x (0.09)e0.09y dy dx
= (0.07)(0.09)/ (0.15)(0.22)
= 0.1909.
A.

1. Which of the following functions can serve as a force of mortality?
(1) Bcx where B > 0, 0 < c < 1, x 0
(2) B(x + 1)0.5 where B > 0, x 0
(3) k(x + 1)n where k > 0, n > 0, x 0
A.
1 and 2 only
B.
1 and 3 only
C.
2 and 3 only
D.
1, 2, and 3
E.
The correct answer is not given by A, B, C, or D.
2. Z is the present value random variable for an nyear endowment insurance of 1

issued to (x). The death benefit is payable at the end of the year of death. Y is the
present value random variable for a special nyear temporary life annuity issued to
(x). A payment of 1 is made at the end of each year for n years if (x) is alive at the
beginning of that year. You are given
(1) Var(Z) = 0.02
(2) i = 0.05
Calculate Var(Y).
A.
8.0
B.
8.4
C.
8.8
D.
9.2
E.
9.6

3. You are given
(1) Px = 0.01212
(2) 20 Px = 0.01508
(3) P 1 = 0.06942
x:10
(4) 10 Vx = 0.11430
Calculate 20
10 Vx .
A.
0.04264
B.
0.11430
C.
0.15694
D.
0.20548
E.
0.31978

x
30
31
32
Calculate 3 q(1)
30 .
A.
0.0555
B.
0.0577
C.
0.0614
D.
0.0656
E.
0.0692
q(1)
x
q(2)
x
0.020
0.050
q()
x
0.075
lx()
dx(1)
1850
54
dx(2)
130
225
226
5. You are given

(1) 1000(IA)50 = 4996.75
(2) 1000A1 = 5.58
50:1
(3) 1000A51 = 249.05

(4) i = 0.06
Calculate 1000(IA)51 .
A.
5,042
B.
5,073
C.
5,270
D.
5,540
E.
5,571
6. A fully discrete whole life insurance of 1 with a level annual premium is issued
to (x). You are given
(1) L is the loss at issue random variable if the premium is determined in
accordance with the equivalence principle.
(2) Var(L) = 0.75
(3) L is the loss at issue random variable if the premium is determined such
that E[L ] = 0.5.
Calculate Var(L ).
A.
0.3333
B.
0.5625
C.
0.7500
D.
1.1250
E.
1.6875

7. You are given
(1) Mortality follows Makehams law.
(2) (ww) is the equivalent equal age status for the joint life status (xy).
(3) (x) and (y) are independent lives and x y.
Which of the following are true?
(1) t pw = (t px )(t py )
(2) t px + t py 2t pw
(3) axy aww
A.
None
B.
1 only
C.
2 only
D.
3 only
E.
8. You are given FX (x) = 1
1
for x 0. Which of the following are true?
x+1
1
x+1
(2) 49 = 0.02
(3) 10 p39 = 0.80
(1) x p0 =
A.
1 and 2 only
B.
1 and 3 only
C.
2 and 3 only
D.
1,2, and 3
E.
The correct answer is not given by A, B, C, or D
227
228

Z
1. Since 0 < c < 1,
Bcx dx < , so (1) does not work. (2) and (3) both give
non-negative functions with infinite integral, and so both work. C.

2. Here
Z = vK+1 1[0,n) (K) + vn 1[n,) (K)
and Y = aK+1 1[0,n) (K) + an 1[n,) (K) = (1/ i) (vK+1 / i)1[0,n) (K) (vn / i)1[n,) (K) =
(1 Z)/ i. Thus Var(Y) = Var(Z)/ i2 = 8. A.
x+10:10 and 10 Vx = Ax+10 Px a x+10 . Also, Ax = Px a x
3. Now 20
10 Vx = Ax+10 20 Px a
and Ax = 20 Px a x:20 , and equating these gives Px a x = 20 Px a x:10 . Using the relations
a x = a x:10 + v10 10 px a x+10 and a x:20 = a x:10 + v10 10 px a x+10:10 together with the fact
that P 1 = v10 10 px / a x:10 allows the previous equality to be rewritten as Px a x+10 =
x:10
x+10:10
20 Px a
20
10 Vx
+ (20 Px Px )/ P
(20 Px Px )/ P
1
x:10
1 .
x:10
Using this in the expression for
10 Vx
gives
10 Vx
from which 20
10 Vx = 0.156939. C.
()
(1)
4. Direct computation using the given table entries gives l30
= 2000, d30
= 20 and
(1)
()
()
(1)
(1)
(1)
q30 = 0.01. Also q31 = 0.07 so that l32 = 1720.5. Thus 3 q30 = q30 + p()
30 q31 +
() () (1)
p30 p31 q32 = 0.01 + (0.925)(0.02) + (0.925)(0.93)(54/ 1720.5) = 0.0555. A.
5. Now (IA)50 = A50 + vp50 (IA)51 and A50 = A1 + vp50 A51 = 0.23914. Since
50:1
vq50 = 5.58/ 1000, vp50 = (1.06)(0.9941). Plugging in gives (IA)51 = 5072.99. B.
6. Here L = (1 + P/ d)vK+1 P/ d and L = (1 + P / d)vK+1 P / d. The equivalence
principle gives E[L] = 0, from which P/ d = Ax / (1 Ax ) and 1 + P/ d = 1/ (1 Ax ).
Since E[L ] = 1/ 2, (1 + P / d) = (3/ 2)/ (1 Ax ). So Var(L )/ Var(L) = 9/ 4, giving
Var(L ) = 27/ 16 = 1.6875. E.
7. Here (t pw )2 = t px t py , so (1) is false. Using the fact that t pw = t px t py and squaring
the proposed inequality (2) shows that (2) holds if and only if (t px t py )2 0. So
(2) is false. Using the fact that axy = ax + ay axy , and a similar fact for aww shows
that (3) holds if and only if ax + ay 2aw , and this holds since t px + t py 2t pw , as
was shown in disproving (2). D.
8. (1) is true since x p0 = s(x) = 1 FX (x). (2) is true since 49 = s (49)/ s(49) =

(1/ (x + 1)2 )/ (1/ (x + 1))
= 1/ 50. (3) holds since 10 p39 = s(49)/ s(39) = 0.80. D.
x=49

1. For a fully discrete whole life insurance of 10,000 issued to (x) the asset share
goal at the end of 20 years is K. A trial gross premium, H, results in an asset share
at the end of 20 years equal to 20 AS1 . You are given
(1) H = 100
(2) K 20 AS1 = 500
(3) ck = 0.5 for k = 0 and ck = 0.1 for 1 k 19 where ck denotes the fraction
of the gross premium paid at time k for expenses.
(4) i = 0.05
(5) 20 p()
x = 0.5
P19 k ()
(6) k=0 v k px = 9.
G is the gross premium that produces an asset share at the end of 20 years equal to
K. Calculate G.
A.
110.96
B.
112.24
C.
124.47
D.
130.86
E.
132.47
2. You are given

(1) Ax and ax are based on force of interest and force of mortality x+t .
(2) Ax and ax are based on force of interest k + and force of mortality x+t .
(3) A
x is based on force of interest and force of mortality k + x+t .
Determine A
x Ax .
A.
kax
B.
Ax Ax
C.
Ax + kax
D.
(k )ax + ax
E.
(ax ax )
230
3. For a fully continuous continuously decreasing 25 year term insurance issued to

(40) you are given
(1) bt = 1000a25t for 0 t 25
(2) Fully continuous net annual premium is 200
(3) A50:15 = 0.6
(4) i = 0.05 and = 0.04879
Calculate the net premium reserve at the end of 10 years for this insurance.
A.
600
B.
650
C.
700
D.
750
E.
800
4. For a triple decrement table you are given

(3)
(1) q(1)
50 = q50
(1)
(2) q(2)
50 = 2q50
(1)
(3) 50+t
= log 2, 0 < t < 1
Assume a constant force of decrement for each decrement over each year of age.
Calculate 1000q50(2) .
A.
531
B.
630
C.
750
D.
766
E.
794
231
5. You are given

(1) Mortality follows DeMoivres Law.
(2) q 1 = 0.8
80:98
(3) (80) and (98) are independent lives.

Calculate .
A.
100
B.
102
C.
105
D.
107
E.
110
6. A 10 year deferred fully discrete whole life insurance is issued to (x). The death
benefit during the deferral period is the return of the net level annual premiums
accumulated with interest at the rate used to calculate the premium. The death
benefit after the deferral period is 10,000. Premiums are payable only during the
deferral period. You are given
(1) i = 0.03
(2) 10 px = 0.88
(3) a x+10 = 12.60
(4) a x = 16.70.
Calculate the net level annual premium.
A.
350
B.
433
C.
522
D.
536
E.
633
232
7. You are given

(1) Ax = 0.25
(2) Ax+20 = 0.40
(3) Ax:20 = 0.55
(4) i = 0.03 and = 0.02956.
Assume deaths are uniformly distributed over each year of age. Calculate 1000Ax:20 .
A.
550
B.
551
C.
552
D.
553
E.
554
8. For a fully discrete whole life insurance you are given

(1) Gross annual premium is 10.0
(2) Net annual premium is 9.0
(3) Expenses, incurred at the beginning of each year, are 0.5 in the first year and
increase at a compound rate of 10% each year.
(4) px = 0.9 for all x
(5) i = 0.06.
Calculate the expected surplus at the end of year 3 for each initial insured.
A.
0.4
B.
0.7
C.
1.0
D.
1.4
E.
1.9
233

1. B.
2. Here A
x =
Z
0
e t (x+t +k)e
Rt
0
k+x+s ds
dt = Ax +kax = 1( +k)ax +kax = 1 ax .
Thus A
x Ax = 1 ax (1 ax ) = (ax ax ). E.
Z
3. The prospective formula gives
10 V =
25
10
bt vt10 t10 p50 50+t10 dt 200a50:15 .
Writing bt = 1000(1 v25t )/ and breaking the integral as the sum of two terms
(1000/ )v15 15 q50 200a50:15 = (1000/ )(A50:15
shows that 10 V = (1000/ )A1
50:15
v15 ) 200a50:15 = 798.99. E.
4. Generally q(j) = ln(p )q() / ln p() . This and (2) gives 2 ln(p ) = ln(p ). Now
(1)
(2)
(2)
(3) gives p = 1/ 2. Thus p = 1/ 4 and q = 3/ 4. C.
(j)
(1)
(2)
5. The joint distribution of T(80) and T(98) is uniform over a rectangle with side
lengths 80 and 98. The region in which (98) dies first consists of a
triangle with both legs of length 98 and a rectangle with sides of length 98
and 80 ( 98) = 18.

The probability that (98) dies first is therefore
(1/ 2)( 98)2 + 18( 98) / ( 98)( 80) = (/ 2 31)/ ( 80). Setting
this equal to 0.8 and solving gives = 110. E.
6. The equation of value is Pax:10 = 10000v10 10 px Ax+10 + E[PsK+1 vK+1 1[0,10) (K)] =
10000v10 10 px Ax+10 + Pax:10 Pa10 10 px . Rearranging and solving gives P = 536.09.
D.
7. Ax:20 = Ax v20 20 px (Ax+20 1), which gives v20 20 px = 1/ 2. Similarly, Ax:20 =
Ax v20 20 px (Ax+20 1) = 550.74/ 1000, using the usual relationship between Ax and
Ax . B.
8. The net premium can not contribute to the surplus nor cover expenses. So the
extra 1 in gross premium over net premium must cover expenses and contribute to
the surplus. The accumulated expected surplus is therefore (1 .5)(1.06)3 + (0.9)(1
.5(1.1))(1.06)2 + (0.9)2 (1 .5(1.1)2 )(1.06) = 1.389. D.

(1) q(1)
71 = 0.02
(2) q(2)
71 = 0.06
(3) Each decrement is uniformly distributed over each year of age in the double
decrement table.
Calculate 1000q71(1) .
A.
20.57
B.
20.59
C.
20.61
D.
20.63
E.
20.65
2. For a fully continuous 20 year deferred life annuity of 1 issued to (35) you are
given
(1) Mortality follows DeMoivres law with = 75
(2) i = 0
(3) Premiums are payable continously for 20 years.
Calculate the net premium reserve at the end of 10 years for this annuity.
A.
1.667
B.
3.889
C.
6.333
D.
6.667
E.
7.222
235
3. You are given

(1) 10| a x = 4.0
(2) a x = 10.0
(3) sx:10 = 15.0
(4) v = 0.94.
Calculate A1x:10 .
A.
0.24
B.
0.34
C.
0.44
D.
0.54
E.
0.64
4. Z is the present value random variable for an n year term insurance payable at
the moment of death of (x) with bt = (1 + i)t . Determine Var(Z).
A.
B.
n qx
C.
Ax n qx
D.
n qx n px
E.
Ax (n qx )2
236
5. You are given

(1) (70) and (75) are independent lives.
(2) Mortality follows DeMoivres law with = 100.
(3) a75 = 8.655.
Calculate A1
70:75
A.
0.2473
B.
0.2885
C.
0.3462
D.
0.4167
E.
0.6606
6. Assume mortality follows DeMoivres law for 0 x < . The median future
lifetime of (x) is denoted by m(x). Which of the following are equal to x for
1 x 1?
qx1
I.
px1
1
II.
2m(x)
mx
III.
1 + 0.5mx
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
237
7. A special fully discrete 2 year endowment insurance with a maturity value of

1000 is issued to (x). The death benefit in each year is 1000 plus the net premium
reserve at the end of that year. You are given
(1) i = 0.10
(2) qx+k = (0.1)(1.1)k , for k = 0, 1.
Calculate the net level annual premium.
A.
508
B.
528
C.
548
D.
568
E.
588
8. A fully discrete three year endowment insurance of 1000 issued to (x) has a level
expense loaded premium, G, equal to the net level premium plus an expense loading
e. You are given
(1) Expenses incurred at the beginning of the year are 18% of G plus 13 in the
first year and 7% of G plus 5 in the renewal years.
(2) The expense reserve two years after issue equals 16.10.
(3) G = 342.86.
Calculate 1000Px:3 .
A.
252.05
B.
275.90
C.
297.76
D.
305.14
E.
329.96
238

1. Here q(1) = q() ln(p )/ ln p() , and q() = 0.08 from the given information. So

0.02/ 0.08
(1)
(1)
p = p()
= 0.97937, which gives 1000q = 20.63. D.
(1)
2. Here t p35:20 = 1 t/ 40 for 0 t 20 and is zero otherwise. Using i = 0 then

Z
gives a35:20 =
20
20
1 t/ 40 dt = 15. Similarly, a55 = 10, and from the equation
Pa35:20 = v 20 p35 a55 , P = 1/ 3. Using the retrospective reserve method, the reserve
after 10 years is (1/ 3)a35:10 / 10 p35 = 3.8889. B.
3. Since a x = a x:10 + v10 10 px a x+10 and 10| a x = v10 10 px a x , a x:10 = 6. Now 15 = sx:10 =
a x:10 / v10 10 px , so v10 10 px = 6/ 15. Finally, A1x:10 = Ax:10 v10 10 px = 1 da x:10 6/ 15 =
0.24. A.
4. Here Z is Bernoulli, with success corresponding to death before time n. Thus
Var(Z) = n qxn px . D.
Z
5. Here A1
70:75
=
0
Z
t
v t p70t p75 70+t dt = (1/ 30)
vt t p75 dt = a75 / 30 = 0.2885. B.
6. Here qx1 = 1/ ( x + 1), so qx1 / px1 = 1/ ( x), and I holds. II holds since
m(x) = ( x)/ 2. III holds since mx = 2qx / (2 qx ). D.
7. The equation of value for the premium P is 1000v2 px px+1 + (1000 + 2 V)v2 px qx+1 +
(1000 + 1 V)vqx = P(1 + px v). Now 2 V = 1000, and the general formula connecting
successive reserves gives px+12 V = (1 + i)(1 V + P) qx+1 (1000 + 2 V). Using these
gives 1009 = 1 V + P. Using these two facts and the given values in the equation of
value gives P = 528.01. B.
8. C.

1. You are given
(1) 15 P30 = 0.030
(2) P30:15 = 0.046
(3) P1
= 0.006
30:15
Calculate A45 .
A.
0.462
B.
0.600
C.
0.692
D.
0.785
E.
0.900
2. Z is the present value random variable for a special increasing whole life insurance
with benefits payable at the moment of death of (50). You are given
(1) bt = 1 + 0.1t
(2) vt = (1 + 0.1t)2
(3) t p50 50+t = 0.02 for 0 t < 50
(4) log 2 = 0.7, log 3 = 1.1
Calculate Var(Z).
A.
0.01
B.
0.02
C.
0.03
D.
0.04
E.
0.05

(1) qx(2) = 2qx(1)
(2) qx(1) + qx(2) = q()
x + 0.18
Calculate qx(2) .
A.
0.2
B.
0.3
C.
0.4
D.
0.6
E.
0.7
4. You are given

(2) i = 0.04 and = 0.0392
(3) n Ex = 0.600
(4) Ax:n = 0.804
Calculate 1000P(Ax:n ).
A.
153
B.
155
C.
157
D.
159
E.
161
240

5. You are given
(2)
x
lx
35
100
36
99
37
96
38
92
39
87
I. 1| 2 q36 = 0.091
II. m37 = 0.043
III. 0.33 q38.5 = 0.021
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
6. You are given

(1) t kx = 0.30
(2) t Ex = 0.45
(3) Ax+t = 0.52
Calculate t V(Ax ).
A.
0.22
B.
0.24
C.
0.30
D.
0.39
E.
0.49
241
242
7. In the SOA Companys rate manual, the expense loaded annual premium for a
fully discrete whole life insurance issued to (x) has a premium rate per 1000 of
insurance equal to 12. You are given
(1) Expenses incurred at the beginning of the year are 70% of the expense loaded
premium plus a per policy expense of 17 in the first year, 10% of the expense
loaded premium plus a per policy expense of 8 in the renewal years.
(2) Ax = 0.125
(3) d = 0.05
S is the assumed average policy size used by the SOA Company to derive the expense
loaded premium. Calculate S.
A.
410
B.
1791
C.
2287
D.
2355
E.
2623
8. You are given

100
9
(2) x+t = k for all t
(3) = 4k
(1) Var(aT ) =
Calculate k.
A.
0.005
B.
0.010
C.
0.015
D.
0.020
E.
0.025
243

+ A45 (15 E30 / a 30:15 ). Similarly,
1. Since A30 = A1
+ 15 E30 A45 , 15 P30 = P1
30:15
30:15
P30:15 = P1
+ 15 E30 / a 30:15 . Combining gives A45 = (15 P30 P1 )/ (P30:15
30:15
30:15
P1 ) = 0.60. B.
30:15
Z
1
2. Direct computation gives E[Z] =
(0.02) dt = 0.2 ln 6. Also E[Z 2 ] =
0 1 + 0.1t
Z
1
(0.02) dt = 0.2(5/ 6). Thus Var(Z) = 0.0371. D.
0 (1 + 0.1t)2
3. Now q() = 1 p() = 1 (1 q )(1 q ) = q + q q q . Substituting

(2)
this expression in (2) and using (1) gives q = 0.06. D.
(1)
(2)
(1)
(2)
(1)
(2)
4. Now 1000P(Ax:n ) = 804/ a x:n . Since A1x:n = 0.804 0.600 = 0.204, A1x:n =
( / i)(0.204), giving a x:n = 5.2021. Thus the premium is 154.55. B.
5. Since 1|2 q36 = (96 87)/ 99 = 0.0909, I holds. Since m37 = q37 / (1 q37 / 2) =
4/ (96 2) = 0.0426, II holds. Since 0.33 q38.5 = 5(0.33)/ (92 2.5) = 0.0186, III fails.
A.
6. Here P = Ax / ax = 1/ ax , so P/ = 1/ (1Ax ) = 0.5848, since Ax = A1x:t +t Ex Ax+t =
t Ex (t kx + Ax+t ) = 0.369. Hence the reserve is Ax+t Pax+t = (1 + P/ )Ax+t (P/ ) =
0.2393. B.
7. The given data yield a x = 17.5. The equation of value is SAx + 8ax + 9 + 0.1Gax +
0.6G = Gax , where G is the gross premium. Now G = (S/ 1000)12 from the rate
table information. Using this and solving for S gives S = (8ax + 9)((0.012)(0.9ax
0.6) Ax ) = 2623.23. E.
Z
8. Since aT = (1 vT )/ , Var(aT ) = Var(vT )/ 2 . Now E[vT ] =

Z
e t t px x+t dt =
e t ekt k dt = k/ ( + k) = 1/ 5. Similarly, E[(vT )2 ] = k/ (2 + k) = 1/ 9. So
Var(vT ) = 1/ 9 1/ 25 = 0.0711. Thus from the first equation here, k = 0.20. D.

1. Y is the present value random variable for an annual life annuity due of 1 issued
on the lives of (x) and (y). For the first 15 years a payment is made if at least one of
(x) and (y) is alive. Thereafter, a payment is made only if exactly one of (x) and (y)
is alive. You are given
(1) a xy = 7.6
(2) a x = 9.8
(3) a y = 11.6
(4) 15| a xy = 3.7
Calculate E[Y].
A.
9.7
B.
9.9
C.
10.1
D.
10.3
E.
10.5

(1) lx(1) = 105vx
(2) lx(2) = 105 x
Calculate x(2) for x = 1.
A.
1
104
B.
1
105
C.
1
106
D.
1
104a2 1
E.
1
105a2 1

3. You are given
(1) 1000Ax:n = 563
(2) 1000Ax = 129
(3) d = 0.057
(4) 1000n Ex = 543
Calculate n| ax .
A.
7.07
B.
7.34
C.
7.61
D.
7.78
E.
7.94
4. You are given

(2) 45.5 = 0.5
Calculate e 45:1 .
A.
0.4
B.
0.5
C.
0.6
D.
0.7
E.
0.8
245
246

x
35
36
37
lx()
1000
828
dx(1)
39
dx(2)
41
69
Assume each decrement is uniformly distributed over each year of age in the double
decrement table. Calculate the absolute rate of decrement due to cause 1 for age 36.
A.
0.026
B.
0.050
C.
0.064
D.
0.080
E.
0.100
6. You are given

(1) Ax+1 Ax = 0.015
(2) i = 0.06
(3) qx = 0.05
Calculate Ax + Ax+1 .
A.
0.60
B.
0.86
C.
1.18
D.
1.30
E.
1.56
247
7. You are given

(1) Fully continuous whole life insurances of 1 are issued to both Matthew and
Zachary.
(2) Fully continuous net level annual premiums for each insurance are determined in accordance with the equivalence principle.
(3) Matthew is subject to a constant force of mortality 1 .
(4) Because of his hang gliding hobby, Zachery is subject to a constant force of
mortality 1 + 2 .
(5) The force of interest is for both insurances.
Determine the excess of Zacharys premium over Matthews premium.
A.
B.
C.
1 2
D.
2
(1 + 2 + )(1 + )
E.
1 2 + 22 1
(1 + 2 )(1 + 2 + )
8. Z is the present value random variable for a discrete whole life insurance of 1
issued to (x) and (y) which pays 1 at the first death and 1 at the second death. You
are given
(1) ax = 9
(2) ay = 13
(3) i = 0.04
Calculate E[Z].
A.
0.08
B.
0.28
C.
0.69
D.
1.08
E.
1.15
248

1. Here E[Y] = a xy 15| a xy = a x + a y a xy 15| a xy = 10.1. C.
2. Here x(2) = dxd lx(2) / lx() = 1/ (lx(1) + lx(2) ). Now l1(1) + l1(2) = 105v + 104 = 105a2 1,
which gives the value of 1(2) = 1/ (105a2 1). E.
3. First note that n| ax = n Ex ax+n . Since Ax = A1x:n + n Ex Ax+n and Ax:n = A1x:n + n Ex ,
the given information yields Ax+n = 109/ 543. Thus 1 + ax+n = a x+n = (1 Ax+n )/ d =
14.022, and using this in the first equation gives n| ax = 7.071. A.
4. By UDD, 1/ 2 = 45.5 = q45 / (1 0.5q45 ), from which q45 = 0.40. Thus e 45:1 =
Z
t p45 dt =
1 t(0.40) dt = 0.80. E.
()
(1)
5. Using the table gives l36
= 920 and d36
= 23. Since q(1) = ln(p )q() / ln(p() ),
(1)
()
(1)
(1)
p = (p() )q / q = (828/ 920)23/ 92 = 0.9740, from which q = 0.0260. A.
(1)
6. Since Ax = vqx + vpx Ax+1 , the given information yields Ax = 0.5841, giving
Ax + Ax+1 = 1.1832. C.
Z
7. For Matthew, a =
0
e t e1 t dt = 1/ ( + 1 ), from which the premium is
A/ a = (1 a)/ a = 1/ a = 1 . Similarly, the premium for Zachery is 1 + 2 ,

giving the difference as 2 . B.
8. Here E[Z] = Ax + Ay = 1 da x + 1 da y = 2 d(2 + ax + ay ) = 1.0769. D.

1. You are given
k
1
2
3
4
a k
1.00
1.93
2.80
3.62
Calculate a x:4 .
A.
1.6
B.
1.8
C.
2.0
D.
2.2
E.
2.4
2. You are given

(1) Mortality follows de Moivres law
(2) Var(T(15)) = 675
Calculate e 25 .
A.
37.5
B.
40.0
C.
42.5
D.
45.0
E.
47.5
k1| qx
0.33
0.24
0.16
0.11
250
3. L1 is the loss at issue random variable for a fully continuous whole life insurance
of 1 on the life of (x) with a net level annual premium determined by the equivalence
principle. You are given
(1) ax = 5.0
(2) = 0.080
(3) Var(L1 ) = 0.5625
L2 is the loss at issue random variable for this insurance with a premium which is
4/3 times the net level annual premium. Calculate the sum of the expected value of
L2 and the standard deviation of L2 .
A.
0.3
B.
0.4
C.
0.6
D.
0.7
E.
0.9
4. A fully discrete last survivor insurance of 1 is issued on two independent lives

each age x. Net annual premiums are reduced by 25% after the first death. You are
given
(1) Ax = 0.4
(2) Axx = 0.55
(3) a x = 10.0
Calculate the initial net annual premium.
A.
0.019
B.
0.020
C.
0.022
D.
0.024
E.
0.025
251
5. G is the expense loaded level annual premium for a fully discrete 20 payment
whole life insurance of 1000 on the life of (x). You are given
(1) G = 21
(2) 1000Ax = 202
(3) a x:20 = 11.6
(4) d = 0.06
(5) Expenses are incurred at the beginning of the year
(6) Percent of premium expenses are 12% in the first year and 3% thereafter
(7) Per policy expenses are k in the first year and 2 in each year thereafter
Calculate k.
A.
5.8
B.
6.8
C.
7.8
D.
8.9
E.
11.2
6. For a multiple decrement table you are given

(1) t p()
x = 1 0.03t, 0 t 1
(1)
(2) x+t = 0.02t, 0 t 1
Calculate m(1)
x .
A.
0.00970
B.
0.00985
C.
0.00995
D.
0.01000
E.
0.01015
252
7. An increasing whole life insurance pays k + 1 at the end of year k + 1 if (80) dies
in year k + 1, k = 0, 1, 2, . . .. You are given
(1) v = 0.925
(2) The net single premium for this insurance is 4 if q80 = 0.1.
P is the net single premium for this insurance if q80 = 0.2 and qx is unchanged for
all other ages. Calculate P.
A.
3.40
B.
3.66
C.
3.75
D.
3.87
E.
3.94
8. You are given

(1) k < 0.5n, where k and n are integers
(2) k Vx:n = 0.2
(3) a x:n + a x+2k:n2k = 2ax+k:nk
Calculate k Vx+k:nk .
A.
0.20
B.
0.25
C.
0.30
D.
0.35
E.
0.40
253

1. Direct computation gives a x:4 = 1(0.33) + 1.93(0.24) + 2.80(0.16) + 3.62(0.11) +
3.62(1 0.33 0.24 0.16 0.11) = 2.218. D.
2. From the variance information ( 15)2 / 12 = 675, so = 105. Thus T(25) is
uniform on (0, 80) and e 25 = 40. B.
3. Here P1 = A/ a = (1 a)/ a = 0.12. Since Var(L1 ) = (1 + P1 / )2 Var(vT ),
Var(vT ) = 0.090. Now E[L2 ] = (P1 / 3)a = 0.20 by using the premium information.
Also, Var(L2 ) = (1+P2 / )2 Var(vT ) = 0.81. The required sum is 0.810.20 = 0.70.
D.
4. The equation of value is Axx = P((1/ 4)axx +(3/ 4)axx ). Now Axx = 2Ax Axx = 0.25.
Also a xx = 2ax a xx . From (1) and (3), d = 0.06, and from (2), a xx = 7.5. Plugging
in gives P = 0.0222. C.
5. The equation of value is 1000Ax + 0.03Gax:20 + 0.09G + 2ax + k 2 = Gax:20 . The
given information yields a x = 13.3, and plugging in then gives k = 7.802. C.
Z
6. Here m(1)
x =
Z
() (1)
t px x+t dt/
()
t px
dt = 0.009949. C.
7. For the original mortality, 4 = vq80 + vp80 S = v(0.1) + v(0.9)S, from which
S = 4.6937. Now P = v(0.2) + v(.8)S = 3.6583. B.
8. Here k Vx:n = Ax+k:nk Px:n a x+k:nk = 1 a x+k:nk / a x:n since Px:n = Ax:n / a x:n =
1/ a x:n d. Similarly, k Vx+k:nk = 1 a x+2k:n2k / a x+k:nk . From this and (3), (1
1
= a x:n / a x+k:nk = 2 (1 k Vx+k:nk ), and solving gives k Vx+k:nk = 1/ 4. B.
k Vx:n )

1. You are given
(1) (x) and (y) are independent lives
(2) (ww) is the equivalent joint equal age status for (xy) assuming Makehamss
law with c = 2
(3) (z) is the equivalent single life status for (xy) assuming Gompertzs law with
c=2
Calculate z w.
A.
B.
C.
D.
E.
2. T is the random variable for the future lifetime of (x). Determine Cov(aT , vT ).
A.
(A2x 2 Ax )/
B.
(A2x 2 Ax )
C.
D.
(2 Ax A2x )
E.
(2 Ax A2x )/
255
3. You are given

n
4
5
6
A1
65:n
0.106
0.133
0.164
(IA)1
65:n
0.250
0.385
0.571
Calculate (DA)1 .
65:5
A.
0.227
B.
0.369
C.
0.394
D.
0.413
E.
0.580
4. For a special fully discrete whole life insurance of 1000 issued on the life of (75)
increasing premiums k are payable at time k for k = 0, 1, 2, . . .. You are given
(1) k = 0 (1 + i)k
(2) Mortality follows de Moivres law with = 105
(3) i = 0.05
(4) Premiums are calculated in accordance with the equivalence principle.
Calculate 0 .
A.
33.1
B.
39.7
C.
44.3
D.
51.2
E.
56.4
256
5. For a fully discrete whole life insurance with level annual premiums on the life
of (x) you are given
(1) i = 0.05
(2) qx+h1 = 0.004
(3) The initial reserve for policy year h is 200
(4) The net amount at risk for policy year h is 1,295
(5) a x = 16.2
Calculate the terminal reserve for policy year h 1.
A.
179
B.
188
C.
192
D.
200
E.
205
6. For a double decrement table where cause 1 is death and cause 2 is disability you
assume
(1) Disabilities occur at the beginning of each year
(2) Deaths occur at the end of each year
(3) q64(1) = 0.010
(4) i = 0.04
H is the net single premium for a one year term insurance issued to (64) which
refunds the net single premium at the moment of disability and pays 1000 at the
moment of death if disability has not occurred. Calculate H.
A.
9.52
B.
9.62
C.
9.78
D.
10.00
E.
10.24
257
7. You are given

(1) q70 = 0.040
(2) q71 = 0.044
Calculate e 70:1.5 .
A.
1.435
B.
1.445
C.
1.455
D.
1.465
E.
1.475
8. Z is the present value random variable for an insurance on the lives of Bill and
John. This insurance provides the following benefits.
(1) 500 at the moment of Bills death if John is alive at that time, and
(2) 1000 at the moment of Johns death if Bill is dead at that time.
You are given
(1) Bills survival function follows de Moivres law with = 85
(2) Johns survival function follows de Moivres law with = 84
(3) Bill and John are both age 80
(4) Bill and John are independent lives
(5) i = 0
Calculate E[Z].
A.
600
B.
650
C.
700
D.
750
E.
800
258

1. From the given information 2x + 2y = 2 2w and 2x + 2y = 2z , so that 2zw = 2
from which z w = 1. D.
2. Since aT = (1 vT )/ , E[aT vT ] = (A 2 A)/ . Thus Cov(aT , vT ) = E[aT vT ]
E[aT ] E[vT ] = (A 2 A)/ A(1 A)/ = (A2 2 A)/ . A.
3. Here (DA)1
65:5
= 6A1
65:5
(IA)1
= 0.413. D.
65:5
4. Since T(75) is uniform on (0, 30), A75 =
29
X
vk+1 / 30 = a30 / 30 = 0.5124. The
k=0
actuarial present value of the premiums is
29
X
kk p75 vk = 0
k=0
29
X
(30 k)/ 30 = 0 (30
k=0
14.5). Thus 0 = 1000A75 / (30 14.5) = 33.05. A.
5. From (3), 200 = h1 V + and from (4), 1295 = b h V. The general reserve
relation ph V = (1+i)(h1 V + )bq then gives h V = 204.82, from which b = 1499.82.
Since bAx = a x , using the value of b and (5) gives = 21.16, which from (3) gives
h1 V = 178.84. A.
6. Here H = 1000q v = 9.615. B.
(1)
7. Here e 70:1.5 =
1.5
t p70
dt =
0
1 tq70 dt + p70
0.5
1 tq71 dt = 1.4547. C.
8. Here E[Z] = 500P[Bill dies before John] + 1000P[John dies after Bill]. Now the
joint distribution of John and Bills remaining lives is uniform over a rectangle of
area 20. The part of this region in which John dies after Bill is a triangle of area
(0.5)(4)(4) = 8, giving the probabilities as 0.4. Thus E[Z] = 1500(0.4) = 600. A.

1. For a special decreasing term insurance on the life of (x) you are given
(1) Z is the present value random variable and
Z=
vK+1
0
a K+1
s5
0K<5
K5
(2) i = 0.05
(3) Px:5 = 0.19
Calculate the net level annual premium for this insurance.
A.
0.010
B.
0.012
C.
0.014
D.
0.016
E.
0.018
260
2. Z1 is the present value random variable for an n year term insurance of 1 on

the life of (x). Z2 is the present value random variable for an n year endowment
insurance of 1 on the life of (x). You are given
(1) vn = 0.20
(2) n px = 0.50
(3) E[Z1 ] = 0.23
(4) Var(Z1 ) = 0.08
(5) Death benefits are payable at the moment of death
Calculate Var(Z2 ).
A.
0.034
B.
0.044
C.
0.054
D.
0.064
E.
0.074
3. For a fully discrete life insurance on the life of (x) you are given
(1) 1 AS = 39
(2) 1 CV = 0
(3) The probability of decrement by death, q(1)
x , equals 0
(4) The probability of decrement by withdrawal, q(2)
x , equals 0.1
(5) q (2)
x = 0.2; all other experience factors equal the assumptions
c
Calculate 1 AS.
A.
34.7
B.
35.1
C.
39.0
D.
42.9
E.
43.9
261
4. You are given sx:n = ax:n / n Ex . Which of the following are true?
I. n| Ax n+1| Ax = v qx+n n Ex
II. Ax:n Ax:n+1 = d n Ex
1
III. sx:n+1 sx:n =
n+1 Ex
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
5. You are given

(1) qx = 0.25
(2) ln(4/ 3) = 0.2877
A
Based on the constant force of mortality assumption, the force of mortality is x+s
,
0 < s < 1. Based on the uniform distribution of deaths assumption the force of
B
B
A
mortality is x+s
, 0 < s < 1. Calculate the smallest s such that x+s
x+s
.
A.
0.4523
B.
0.4758
C.
0.5001
D.
0.5242
E.
0.5477

6. Y is the present value random variable for a benefit based on (x) such that

Y=
an
aT
0 T(x) n
T(x) > n
Determine E[Y].
A.
ax:n
B.
ax:n + n| ax
C.
an + n| ax
D.
an + vn ax+n
E.
n qx
an + n| ax
7. You are given that qy = 0.25, q2xy = 0.12, and qxy = 0.14. Calculate qxy1.
A.
0.02
B.
0.04
C.
0.11
D.
0.16
E.
0.23
8. You are given that Ax:n = 0.20 and d = 0.08. Calculate n1 Vx:n .
A.
0.90
B.
0.92
C.
0.94
D.
0.96
E.
0.98
262
263

1. Simplification gives Z = (vK+1 v5 )/ (1 v5 ) for K < 5. Thus E[Z] = (A1x:5
v5 5 qx )/ (1 v5 ) = (Ax:5 v5 )/ (1 v5 ). From (3), a x:5 = 1/ (Px:5 + d) = 4.2084. The
desired premium solves E[Z] = Pax:5 , and plugging in gives P = 0.0176. E.
2. Here Z2 = Z1 + vn 1[n,) (T) and Z22 = Z12 + v2n 1[n,) (T). Using the given information
produces E[Z2 ] = 0.23 + (.2)(.5) = 0.33, and E[Z22 ] = 0.08 + (0.23)2 + (0.2)2 (0.5) =
.1529, giving Var(Z2 ) = 0.0440. B.
3. E.
4. I is true since n| Ax n+1| Ax = n Ex Ax+n n+1 Ex Ax+n+1 = n Ex (Ax+n vpx+n Ax+n+1 ) =
+ EA
= Ax:n = A1x:n + n Ex v = Ax:n + n Ex (v
n Ex vqx+n . II is true since Ax:n+1 = A1
x:n n x x+n:1
1). III fails since sx:n+1 sx:n = a x:n+1 / n+1 Ex ax:n / n Ex = (1 + ax:n )/ n+1 Ex ax:n / n Ex =
1/ n+1 Ex + ax:n (1/ n+1 Ex 1/ n Ex ), and this last term is not zero. A.
B
A
= 0.25/ (1 0.25s) = 1/ (4 s). Thus
= ln(.75) = ln(4/ 3) while x+s
5. Here x+s
A
B
x+s x+s if and only if 1/ (4 s) ln(4/ 3) or s 4 1/ ln(4/ 3) = 0.5242. D.
6. Here Y = an 1[0,n) (T) + aT 1[n,) (T) = an 1[0,n) (T) + (1 vT )1[n,) (T)/ . Thus E[Y] =
an n qx + (1/ )n px (1/ )n| Ax . Now n| Ax = n Ex Ax+n = n Ex (1 ax+n ) = n Ex n| ax .
Making this substitution and simplifying gives E[Y] = an + n| ax . C.
7. The desired probability is the probability that (y) dies during the next year while
(x) is still alive. This can occur if (y) dies and (x) doesnt, or if both die with (x)
dying after (y). The associated probabilities are qy qxy = 0.11 and q2xy = 0.12,
giving the total probability as 0.23. E.
8. Here n Vx:n = 1 because of the endowment part. So the standard recursion gives
p = (1 + i)(n1 Vx:n + ) q so that n1 Vx:n = v . The premium = Ax:n / a x:n =
dAx:n / (1 Ax:n ) = 0.02. Using this gives the reserve as 0.90.A.

1. Z is the present value random variable for a whole life insurance of 1 payable at
the moment of death of (x). You are given x+t = 0.05 for t 0 and that = 0.10.
d
I. Ax = 0
dx
II. E[Z] = 13
III. Var(Z) = 51
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
265
2. For a 30 year deferred, annual life annuity due of 1 on (35) you are given that
R is the net single premium for this annuity if the net single premium is refunded
at the end of the year of death for death during the deferral period, and N is the net
single premium for this annuity if the net single premium is not refunded. Which
of the following correctly expresses R N?.
A
35
30| a
I. 35:30
1 A1
35:30
A1
(A35:30 A35 )
II. 35:30
d(1 A1 )
35:30
(1 da 35:30 ) 30| a 35
III.
da 35:30
A.
None
B.
I only
C.
II only
D.
III only
E.

I. t+r px r px+t for t 0 and r 0
II. r qx+t t|r qx for t 0 and r 0
III. If s(x) follows DeMoivres Law, the median future lifetime of (x) equals the
mean future lifetime of (x)
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
266
4. An insurance company issues a fully discrete whole life insurance on (x). You
are given
(1) The level expense loading, c, is 5
(2) Expenses ek1 incurred at the beginning of policy year k are
Policy Year k
ek1
1
10
2
8
3
6
(3) i = 0.05
(4) k px = (1.08)k for k = 1, 2, . . .
(5) Net premium reserves are recognized as the measure of liabilities
For this policy, the company plans to establish initial surplus, u(0), such that expected
assets equal expected liabilities at the end of 3 years. Calculate u(0).
A.
5.79
B.
8.42
C.
9.36
D.
9.75
E.
12.28
267
5. You are given that T(x) and T(y) are independent, that the survival function for
(x) follows DeMoivres Law with = 95, and that the survival function for (y) is
based on a constant force of mortality y+t = for t 0. Assume that n < 95 x.
Determine the probability that (x) dies within n years and predeceases (y).
A.
en
95 x
B.
nen
95 x
C.
1 en
(95 x)
D.
1 en
95 x
E.
1 en +
en
95 x
6. You are given x = A + ex for x 0, and that 0.5 p0 = 0.50. Calculate A.

A.
0.26
B.
0.09
C.
0.00
D.
0.09
E.
0.26
268
7. You are given

(1) X is the present value random variable for a 25 year term insurance of 7 on
(35)
(2) Y is the present value random variable for a 25 year deferred, 10 year term
insurance of 4 on the same life
(3) E[X] = 2.80, E[Y] = 0.12
(4) Var(X) = 5.76, Var(Y) = 0.10
Calculate Var(X + Y).
A.
4.75
B.
5.19
C.
5.51
D.
5.86
E.
6.14
(1)
1
1
8. For a double decrement table you are given qx(2) = 18 , 1| q(1)
x = 4 , and qx+1 = 3 .
Calculate qx(1) .
A.
1/4
B.
1/5
C.
1/6
D.
1/7
E.
1/8
269

1. Because of the constant
force of mortality, Ax does not depend on x and I
Z
0.1t 0.05t
holds. Now E[Z] =
e e
0.05 dt = 0.05/ 0.15 = 1/ 3, and similarly E[Z 2 ] =
0
0.05/ 0.25 = 1/ 5. Thus II holds and III fails. A.

2. First 30| a 35 + RA1
= R, so R = 30| a 35 / (1 A1 ). Also N = 30| a 35 , so
35:30
35:30
R N = A1 30| a 35 / (1 A1 ), and I fails. Now A35 = A1
+ 30 E35 A65 and
35:30
35:30
35:30
65 ) = 30 E35 d30| a 35 . Making this substitution shows that II
30 E35 A65 = 30 E35 (1 d a
holds. III fails since III has endowment insurances where the correct formula has
term insurances. C.
3. I fails since t+r px = t pxr px+t . II holds since t| r qx = t pxt qx+r . III holds since T(x) has
a uniform distribution. C.
4. B.
Z
5. Conditioning on the time of death of (x) gives

0
x). C.
R 0.5
6. Here 0.5 = 1/ 2 p0 = e
A+ex dx
= eA/ 2(e
0.5 1)
1
ex dx = (1en )/ (95
95 x
. Solving gives A = 0.0889. D.
7. Here XY = 0 since their dependence on T is supported in disjoint intervals. Thus

E[(X + Y)2 ] = E[X 2 ] + E[Y 2 ] = Var(X) + (E[X])2 + Var(Y) + (E[Y])2 . The given
information and this formula gives Var(X + Y) = 5.188. B.
() (1)
()
8. Since 1| q(1)
x = px qx+1 , the given information yields px = 3/ 4. Since (1q )(1
(2)
(1)
q ) = p() , the given information gives q = 1/ 7. D.
(1)

I.
n Ex = n Ex (x x+n )
x
II.
n Ex = n Ex (x+n + )
n
1
t Ex
III.
=
for t n
n Ex
tn Ex+n
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
2. For a 10 year term insurance of 10,000 with death benefits payable at the end of
the year of death of (30) you are given
(1) A1
= 0.015
30:10
(2) a 30:10 = 8
(3) 10 E30 = 0.604
(4) i = 0.05
(6) Level true fractional premiums are determined in accordance with the equivalence principle
Calculate the additional annual premium for this insurance if premiums are paid in
monthly rather than semi-annual installments.
A.
0.05
B.
0.10
C.
0.15
D.
0.20
E.
0.25

3. For a 2 year select and ultimate mortality table you are given
(1) Ultimate mortality follows the Illustrative Life Table:
x
lx
dx
93
530,974
127,890
94
403,084
105,096
95
297,988
84,006
96
213,982
74,894
97
139,088
66,067
98
73,021
49,289
99
23,732
23,732
(2) q[x] = 0.5qx for all x
(3) q[x]+1 = 0.5qx+1 for all x
(4) l[96] = 10, 000
Calculate l[97] .
A.
4047
B.
4076
C.
4094
D.
4136
E.
4158
271
272
4. You are given

(1) Mortality follows Gompertz law for both (35) and (40)
(2) c10 = 4
(3) A35:40 = 0.6
(4) T(35) and T(40) are independent
Calculate A1
35:40
A.
0.20
B.
0.24
C.
0.28
D.
0.30
E.
0.32
5. A whole life insurance of 10,000 payable at the moment of death of (x) includes
a double indemnity provision. This provision pays an additional death benefit of
10,000 during the first 20 years if death is by accidental means. You are given
()
(1)
(1)
= 0.05, x+t
= 0.005 for t 0, and x+t
= 0.001 for t 0, where x+t
is the force
of decrement due to death by accidental means. Calculate the net single premium
for this insurance.
A.
910
B.
970
C.
1030
D.
1090
E.
1150
273
6. A 10 year deferred life annuity due on (x) includes a refund feature during the
deferral period providing for return of the net single premium with interest at the
end of the year of death. Interest is credited at rate i. You are given i = 0.05 and
the terminal reserve at the end of 9 years equals 15.238. Calculate the net single
premium for this annuity.
A.
9.355
B.
9.823
C.
14.512
D.
15.238
E.
16.000
7. You are given that mortality follows the Illustrative Life Table:
x
54
55
56
..
.
64
65
66
lx
8,712,711
8,640,918
8,563,495
..
.
7,684,067
7,534,074
7,373,448
1000qx
8.24
8.96
9.75
..
.
19.52
21.32
23.29
and that i = 0.05. Calculate 2 A55:10 .

A.
0.329
B.
0.402
C.
0.476
D.
0.550
E.
0.631
1000Ax
349.09
361.29
373.74
..
.
481.32
495.53
509.86
1000(2 Ax )
157.36
166.63
176.33
269.94
283.63
297.73
274
8. G is the gross annual premium for a fully discrete whole life insurance. You are
given
(1) No deaths or withdrawals are expected during the first two policy years
(2) i = 0.1
(3) Expenses are incurred at the beginning of each policy year
(4) Percent of premium expenses are 7% of G each year
(5) Per policy expenses are 10 for year 1 and 2 for year 2
(6) The expected asset share at the end of year 2, 2 AS, equals 12.66
Calculate G.
A.
12.25
B.
12.35
C.
12.45
D.
12.55
E.
12.65
275

R x+n
1. Note that n Ex = e n e x s ds . The Fundamental Theorem of Calculus then gives
n Ex = n Ex (x x+n ) and I holds. Similarly,

n Ex = n Ex (x+n + ) and II fails.
x
n
The ratio in III simplifies to tn Ex+n and III fails. E.
2. The equations for the two annual premium rates are 10000A1
and 10000A1
30:10
(m)
10 E30 a
40 ,
= P(12) a (12)
.
30:10
30:10
= P(2) a (2)
30:10
= (m)ax (m), so from a (m)

Now a (m)
30 =
x
a (m)
+
making this substitution and solving gives a (m)
= (m)a30:10
30:10
30:10
(m)(10 E30 1). Now (2) = 1.000, (2) = 0.2561, (12) = 1.000 and (12) =
0.4665. Making these substitutions gives P(2) = 18.9879 and P(12) = 19.1893 for a
difference of 0.2013. D.
3. Since the ultimate table is given, l[96] = l98 / p[96] p[96]+1 = 116079 using the
given information. Similarly, l[97] = l99 / p[97] p[97]+1 = 46979. Since by (4) l[96] is
renormalized to 10000, the renormalized value of is l[97] = 10000(46979/ 116079) =
4047.2. A.
4. Because of Gompertz law, (35 : 40)t = 35+t + 40+t = Bc35+t (1 + c5 ) = 335+t
by (2). Thus A35:40 = 3A1 , and A1
= 0.60/ 3 = 0.20. A.
35:40
5. Here Ax =
Z
0
20
35:40
e0.05t e0.005t 0.005 dt = 0.005/ 0.055 = 1/ 11, and also A1x:20 =
e0.05t e0.005t 0.001 dt = (1 e0.055(20) )/ 55 = 0.0121. The sum of these two times
10,000 gives the premium as 1030.88. C.

6. The retrospective method gives 9 Ex 9 V = P P E[v(K+1) vK+1 1[0,9) (K)] since the
premium P is refunded with interest if death occurs in the deferral period. The
expectation is 9 qx , and making this substitution gives v9 9 V = P = 9.822, using the
given information. B.
7. Here 2 A55 = 2 A55:10 10 2 E55 + 10 2 E55 2 A65 , where 10 2 E55 = v20 l65 / l55 . Rearranging
and using the table gives 2 A55:10 = 0.4020. B.
8. D.

1. A fully discrete whole life insurance with annual premiums payable for 10 years
is issued on (30). You are given the death benefit is equal to 1000 plus the refund
of the net level annual premiums paid without interest, and that premiums are
calculated in accordance with the equivalence principle. Determine the net annual
premium for this insurance.
A.
1000A30
a 30:10 + 10 10| A30
B.
1000A30
a 30:10 10 10| A30
C.
1000A30
a 30:10 (IA)1
30:10
D.
E.
a 30:10
1000A30
(IA)1
+ 10 10| A30
a 30:10
1000A30
(IA)1
10 10| A30
30:10
30:10
2. A multiple decrement table has two causes of decrement: (1) accident and
(2) other than accident. You are given y(1) = 0.0010 and y(2) = 0.0005(100.05 )y .
Determine the probability of death by accident for (x) in terms of e x .
A.
0.0005 e x
B.
0.0010 e x
C.
0.0050 e x
D.
0.0100vex
E.
0.0500 e x
277
3. You are given that deaths are uniformly distributed over each year of age, i = 0.05,
q35 = 0.01, and that A36 = 0.185. Calculate A35 .
A.
0.1797
B.
0.1815
C.
0.1840
D.
0.1864
E.
0.1883
4. For a special fully discrete whole life insurance on (55) you are given
(1) Initial net annual premiums are level for 10 years. Thereafter, net annual
premiums equal one-half of initial net annual premiums.
(2) Death benefits equal 1000 during the first 10 years, and 500 thereafter
(3) A55 = 0.36129, A65 = 0.49553, a 55 = 13.413, a 65 = 10.594, l55 = 8, 640, 918,
l65 = 7, 534, 074
(4) i = 0.05, v10 = 0.613913
Calculate the initial net premium.
A.
8.54
B.
10.81
C.
17.08
D.
21.62
E.
34.16
278
5. Which of the following are correct expressions for t V(Ax )?

Ax+t Ax
I.
1 Ax+t
P(A1x:t ) ax:t
II. P(Ax )sx:t
t Ei
x
h
III. P(Ax+t ) P(Ax ) ax
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
6. You are given qx = 0.04, x+t = 0.04 + 0.001644t for 0 t 1, and y+t =
0.08 + 0.003288t for 0 t 1. Calculate qy .
A.
0.0784
B.
0.0792
C.
0.0800
D.
0.0808
E.
0.0816
279
7. A whole life insurance of 1 with benefits payable at the moment of death of

(x) includes a double indemnity provision. This provision pays an additional death
benefit of 1 for death by accidental means. S is the net single premium for this
insurance. A second whole life insurance of 1 with benefits payable at the moment
of death of (x) includes a triple indemnity provision. This provision pays an
additional death benefit of 2 for death by accidental means. T is the net single
premium for this insurance. You are given
(1) The force of decrement for death by accidental means is constant and equal
to
(2) The force of decrement for death by other means is constant and equal to 5
(3) There are no other decrements
Determine T S.
A.
S
12
B.
S
8
C.
S
7
D.
S
4
E.
S
2
0.024
for t 0, and = 0.03. Calculate the
ln(0.4)
will exceed 20.
8. For (x) you are given x+t =

probability that aT(x)
A.
0.45
B.
0.55
C.
0.67
D.
0.74
E.
0.82
280

1. Here 1000A30 + PE[vK+1 (K + 1)1[0,10) (K)] + 10PE[vK+1 1[10,) (K)] = Pa30:10 , from
which P = 1000A30 / (a30:10 (IA)1
1010| A30 ). E.
30:10
2. Here q(1)
x =
() (1)
t px x+t
dt = 0.001 e x . B.
3. Here A35 = A1 + vp35 A36 = (iv/ )q35 + vp35 A36 = 0.1842. Now A35 = ( / i)A35 =
35:1
0.1797. A.
4. Here 1000A55 50010 E55 A65 = P(a55 (1/ 2)10 E55 a 65 ). The given information
yields 10 E55 = v10 l65 / l55 = 0.5353, and using this and the other given values yields
P = 21.618. D.
5. The prospective formula gives t V(Ax ) = Ax+t P(Ax )ax+t = Ax+t (Ax / ax )ax+t =
(Ax+t Ax )/ (1 Ax ) using ax = (1 Ax )/ . This shows that I fails. The retrospective
formula gives II directly. III rearranges to be equal to I after inserting the definitions
of the premiums, so III fails too. E.
6. Since y+t = 2x+t , py = (px )2 , from which qy = 1 (1 qx )2 = 0.0784. A.
Z
7. Now x() = 6 so that S =
Z
()
vt t p()
x x+t dt +
vt t p()
x dt = 6/ ( + 6) +
/ ( + 6) = 7/ ( + 6). Similarly, direct computation gives T = 8/ ( + 6).

So T S = / ( + 6) = S/ 7. C.
8. Here P[aT > 20] = P[(1 vT )/ > 20] = P[T > ln(1 20 )/ ln(v)] =
exp( ln(1 20 )/ ln(v)) = 0.4493. A.

1. A company issues fully discrete level premium whole life insurances of 1000 on
each of 1000 independent lives age 90. You are given
(1) Premiums are determined by the equivalence principle
(2) There are no expenses or taxes
(3)
x
lx
a x
1000Ax
90
1,058,511
3.630
827.13
91
858,696
3.404
837.89
92
682,723
3.175
848.80
93
530,974
2.936
860.15
(4) i = 0.05
(5) Death is the only decrement
Calculate the expected total fund the company will have at the end of 3 years.
A.
95,800
B.
107,500
C.
113,300
D.
121,400
E.
135,200
282
2. L is the loss at issue random variable for a fully discrete whole life insurance of
1 on (x). The annual premium charged for this insurance is 0.044. You are given
Ax = 0.40, a x = 10, and Var(L) = 0.12. An insurer has a portfolio of 100 such
insurances on 100 independent lives. Eighty of these insurances have death benefits
of 4 and 20 have death benefits of 1. Assume that the total loss for this portfolio is
distributed normally. Calculate the probability that the present value of the gain for
this portfolio is greater than 22.
A.
0.01
B.
0.07
C.
0.10
D.
0.16
E.
0.25
3. Two independent lives, both age x, are subject to the same mortality table.
Calculate the maximum possible value of t pxx t px .
A.
1
16
B.
1
8
C.
1
4
D.
1
2
E.

4. For a two year select and ultimate mortality table you are given
[x]
l[x]
l[x]+1
lx+2
30
1000
998
995
31
996
994
988
32
994
990
982
33
987
983
970
283
x+2
32
33
34
35

I. 2 p[31] > 2 p[30]+1
II. 1| q[31] > 1| q[30]+1
III. 2 q[33] > 2 q[31]+2
A.
None
B.
I only
C.
II only
D.
III only
E.
5. You are given that 1000q60 = 13.76, A60 = 0.34487, and A61 = 0.35846. Calculate
i.
A.
0.050
B.
0.055
C.
0.060
D.
0.065
E.
0.070
284

(1) Each decrement has a constant force of decrement over each year of age
(2) The following table of values
j
x(j)
1
0.2
2
0.4
3
0.6
Calculate q(2)
x .
A.
0.20
B.
0.23
C.
0.26
D.
0.30
E.
0.33
7. Z is the present value random variable for an insurance on (x) defined by

Z=
(6 K)vK+1
0
K = 0, 1, 2, 3, 4, 5
K6
where vK+1 is calculated at force of interest . Which of the following are true?
I. Z is the present value random variable for a 5 year decreasing term insurance
payable at the end of the year of death of (x).
P
II. E[Z] = 5k=0 (6 k) k| A1x:1
III. E[Z 2 ] calculated at force of interest equals E[Z] calculated at force of interest
2
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
285
8. L is the loss at issue random variable for a fully continuous whole life insurance
of 2 on (x). This insurance has a total level annual premium rate of 0.09. You are
given x+t = 0.04 for t 0 and = 0.06. Calculate Var(L).
A.
0.02
B.
0.09
C.
0.32
D.
0.56
E.
1.10
286

1. Direct computation gives 3 V = 1000A93 Pa93 = 191.15, since P = 1000A90 / a 90 =
227.86. Now the reserve only exists for the survivors, so the amount the company
will have, on average, is 1000(l93 / l90 )191.15 = 95, 885. A.
2. Now E[L] = Ax 0.044ax = 0.04, and for the total portfolio the loss, T,
has E[T] = 80(4)(0.04) + 20(0.04) = 12.8 and Var(T) = (80)(16)Var(L) +
20Var(L) = 156. Using the normal approximation gives P[T < 22] = P[Z <
(22 (12.8))/ 156] = P[Z < 0.7376] = 1 0.77 = 0.23. E.
3. Since t pxx = 1 (1 t px )2 , the maximum value of the difference is the same as the
maximum value of the quadratic x x2 on the interval 0 x 1, which is 1/ 4. C.
4. Direct computation gives 2 p[31] = 988/ 996 = 0.9920 and 2 p[30]+1 = 988/ 998 =
0.9900 so I holds. Also 1| q[31] = (994/ 996)(994 988)/ 994 = 0.0060 and 1| q[30]+1 =
(995/ 998)(995 988)/ 995 = 0.0070, so II fails. Finally, 2 q[33] = 17/ 987 = 0.0172
while 2 q[31]+2 = 18/ 988 = 0.0182 and III fails. B.
5. Since A60 = vq60 + vp60 A61 , 1 + i = (q60 + p60 A61 )/A60 = 1.065. D.
6. Here p() = e1.2 and p
e1.2 )/ 1.2 = 0.2329. B.
(2)
()
()
= e0.4 , so that q(2)
x = q ln(q )/ ln(p ) = 0.4(1
(2)
7. I fails since the factor given is (6 K) and not (5 K). II holds by direct
computation, and III fails due to the presence of the (6 K) term. E.
8. Since L = 2vZT 0.09aT = (2 + 0.09/ )vT 0.09/ , Var(L) = (2 + 0.09/ )2 Var(vT ).
Now E[vT ] =
e t et dt = / ( + ) and E[v2T ] = / ( + 2 ) by similar
0
computations. So Var(vT ) = 0.09 and Var(L) = 1.102. E.

1. S is the actuarial present value of a continuous annuity of 1 per annum payable
while at least one of (30) and (45) is living, but not if (30) is alive and under age 40.
Which of the following is equal to S?
I. a45 + a30 a30:45:10
II. a45 + 10| a30 a30:45
III. a45 + 10| a30 10| a30:45
IV. a45 + 10| a30 a30:45:10
A.
None
B.
I only
C.
II only
D.
III only
E.
IV only
2. Assume mortality follows DeMoivres Law for 0 x . Which of the following expressions equal x ?
1
I.
2ex
II. n| qx , 0 n x 1
mx
III.
,x 1
1 + 0.5mx
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.

3. You are given
(1) A35 = 0.17092,A40 = 0.20799
(2) a 35 = 17.410616, a 40 = 16.632258
(3) i = 0.05
(5) (4) = 1.00019, (4) = 0.38272
(6) 10005 V35 = 44.71
(7) a (4)
35 = 17.031204
(4)
Calculate 1000(5 V35
5 V35 ).
A.
0.17
B.
0.45
C.
1.00
D.
3.72
E.
3.81
288
289
4. For a fully discrete level benefit whole life insurance you are given
(1) Expenses, incurred at the beginning of each year, are
Type of Expense
Expense
Fraction of premium
0.25
Per 1000 of Insurance
2.00
Per Policy
30.00
(2) The assumed average policy size is 20,000
(3) S is the expense loaded level annual premium for an insurance of 25,000 if
the approximate premium rate method is used for per policy expenses
(4) T is the expense loaded level annual premium for an insurance of 25,000 if
the policy fee method is used for per policy expenses
Calculate S T.
A.
0.00
B.
2.50
C.
5.00
D.
7.50
E.
10.00
5. Which of the following can serve as survival functions for x 0?

I. s(x) = exp(x 0.7(2x 1))
1
II. s(x) =
(1 + x)2
III. s(x) = exp(x2 )
A.
I and II only
B.
I and III only
C.
II and III only
D.
I, II, and III
E.
290
6. Z is the present value random variable for a special whole life insurance with
death benefits payable at the moment of death of (x). You are given (for t 0)
bt = e0.05t , t = 0.06, and x+t = 0.01. Calculate Var(Z).
A.
0.037
B.
0.057
C.
0.063
D.
0.083
E.
0.097
7. You are given qx = 0.5, qx+1 = 0.5, qx+2 = 1.0, and

A.
1
4
B.
1
3
C.
1
2
D.
2
3
E.
3
4
d
a x = (3.5)2 Ex . Calculate i.
di

(1) Each decrement is uniformly distributed over each year of age in its associated single decrement table
(2) qx(1) = 0.1000
(3) qx(2) = 0.0400
(4) qx(3) = 0.0625
Calculate 1000q(1)
x .
A.
94.00
B.
94.55
C.
94.96
D.
95.00
E.
100.50
291

1. Direct reasoning gives the value of the annuity as a30:45 a30:10 = a30 + a45
a30:45 a30:10 = 10| a30 + a45 a30:45 , and II holds. III and IV fails since the last term
in each expression is not equal to a30:45 . I fails since the difference of I and II is
easily seen to be non-zero. C.
2. Since x = 1/ ( x) and e x = ( x)/ 2, I holds. II also holds since n| qx =
(( x n)/ ( x)) (1/ ( n x)) = x . III also holds by algebraic manipulation
Z
from mx = qx /
(1 tqx ) dt) = 2qx / (2 qx ). D.
(4)
3. Since a (4)
a35 (4) = 17.0312 and similarly a (4)
40 = 16.2527, 1000P35 =
35 = (4)
(4)
(4) (4)
1000A35 / a (4)
40 =
35 = 10.0357. Thus the reserve is 10005 V35 = 1000A40 1000P35 a
44.8828. Subtracting the given value for 10005 V35 gives the difference as 0.1728.
A.
4. E.
5. A survival function must have s(0) = 1 and be non-increasing. I fails since
s (x) = s(x)(1 0.7(2x ln 2)) > 0 for x near 0. II and III both work. C.
Z
6. Direct computation gives E[Z] =

e0.05t e0.06t e0.01t 0.01 dt = 1/ 2 while E[Z 2 ] =
0
1/ 3 by similar direct computation. So Var(Z) = 1/ 3 1/ 4 = 1/ 12 = 0.0833. D.
7. Since qx+2 = 1, a x = 1 + vpx + v2 2 px = 1 + v/ 2 + v2 / 4 since px = 1/ 2 and 2 px = 1/ 4.
d
d
Since v =
(1 + i)1 = v2 , the given information becomes v2 / 2 v3 / 2 =
di
di
3.5v2 / 4, from which v = 2(3.5/ 4 1/ 2) = 3/ 4 and i = 1/ 3.B.
Z
(1)
8. From the given information q

tq ) dt = 0.09496. C.
(2)
=
0
Z
() (1)
t px x+t
dt =
0
q (1 tq )(1
(1)
(2)

1. You are given
(1) (x) is subject to a uniform distribution of deaths over each year of age
(2) (y) is subject to a constant force of mortality of 0.3
(3) q1xy = 0.045
(4) T(x) and T(y) are independent
Calculate qx .
A.
0.052
B.
0.065
C.
0.104
D.
0.214
E.
0.266
2. For a fully discrete 3-year endowment insurance of 1000 on (x) you are given
(1) k L is the prospective loss random variable at time k
(2) i = 0.10
(3) a x:3 = 2.70182
(4) Premiums are determined by the equivalence principle.
Calculate 1 L given that (x) dies in the second year from issue.
A.
540
B.
630
C.
655
D.
720
E.
910
293
3. For a double decrement model

(1)
(1) t p 40 = 1 t/ 60, for 0 t 60
(2)
(2) t p 40 = 1 t/ 40, for 0 t 40
()
Calculate 40
(20).
A.
0.025
B.
0.038
C.
0.050
D.
0.063
E.
0.075
4. For independent lives (35) and (45)

(1) 5 p35 = 0.90
(2) 5 p45 = 0.80
(3) q40 = 0.03
(4) q50 = 0.05
Calculate the probablity that the last death of (35) and (45) occurs in the 6th year.
A.
0.0095
B.
0.0105
C.
0.0115
D.
0.0125
E.
0.0135
294
5. For a fully discrete whole life insurance of 100,000 on (35) you are given
(1) Percent of premium expenses are 10% per year
(2) Per policy expenses are 25 per year
(3) Per thousand expenses are 2.50 per year
(4) All expenses are paid at the beginning of the year
(5) 1000P35 = 8.36
Calculate the level annual expense loaded premium using the equivalence principle.
A.
930
B.
1041
C.
1142
D.
1234
E.
1352
6. Kings of Fredonia drink glasses of wine at a Poisson rate of 2 glasses per day.
Assassins attempt to poison the kings wine glasses. There is a 0.01 probability that
any given glass is poisoned. Drinking poisoned wine is always fatal instantly and
is the only cause of death. The occurences of poison in the glasses and the number
of glasses drunk are independent events. Calculate the probability that the current
king survives at least 30 days.
A.
0.40
B.
0.45
C.
0.50
D.
0.55
E.
0.60
295
7. Insurance losses are a compound Poisson process where

(1) The approvals of insurance applications arise in accordance with a Poisson
process at a rate of 1000 per day.
(2) Each approved application has a 20% chance of being from a smoker and an
80% chance of being from a non-smoker.
(3) The insurances are priced so that the expected loss on each approval is 100.
(4) The variance of the loss amount is 5000 for a smoker and is 8000 for a
non-smoker.
Calculate the variance for the total losses on one days approvals.
A.
13,000,000
B.
14,100,000
C.
15,200,000
D.
16,300,000
E.
17,400,000
8. Z is the present value random variable for a whole life insurance of b payable at
the moment of death of (x). You are given
(1) = 0.04
(2) x (t) = 0.02 for t 0
(3) The single benefit premium for this insurance is equal to Var(Z).
Calculate b.
A.
2.75
B.
3.00
C.
3.25
D.
3.50
E.
3.75
296

Z
1. Direct computation gives q1xy =
1
t px x+tt py
qx (1 e0.3 )/ 0.3, from which qx = 0.0521. A.
dy = qx
1
t py
dt = qx
e0.3t dt =
2. The premium is 1000Ax:3 / a x:3 = 1000(1 da x:3 )/ a x:3 = 279.21. The loss is
1000v 279.21 = 629.88. B.
d (j)
(j)
()
(1)
t p 40 / t p 40 the given information yields 40 (20) = 40 (20) +
dt
(2)
40
(20) = 1/ 40 + 1/ 20 = 0.0750. E.
(j)
3. Since 40
(t) =
4. Direct reasoning gives the probability as (0.90)(0.03)(1 0.80) + (0.80)(0.05)(1

0.90) + (0.90)(0.80)(0.03)(0.05) = 0.0105. B.
5. The gross premium G satisfies 100, 000A35 + 0.10Ga35 + 25a35 + 250a35 = Ga35 ,
from which G = (100, 000P35 + 275)/ 0.90 = 1234.44. D.
6. Poisoned glasses appear at a Poisson rate of 0.02 per day or 0.60 per 30 days. So
the probability of survival is e0.60 = 0.5488. D.
7. The smokers and non-smokers arrive according to independent Poisson processes
with rates of 200 and 800 per day respectively. The total variance is thus 200Var(S)+
(E[S])2 200 + 800Var(N) + (E[N])2 800 where S and N are the loss variables for an
individual smoker and non-smoker. Substitution gives this value as 17, 400, 000. E.
Z
8. Here E[Z] =
0
be t e0.02t 0.02 dt = b/ 3, and similarly E[Z 2 ] = b2 / 5. Since the
premium is equal to the variance, b/ 3 = b2 / 5 b2 / 9 from which b = 3.75. E.

1. For a special 3 year term insurance on (30) you are given
(1) Premiums are payable semiannually.
(2) Premiums are payable only in the first year.
(3) Benefits, payable at the end of the year of death, are
k
0
1
2
bk+1
1000
500
250
(4) Mortality follows the Illustrative Life Table.

(5) Deaths are uniformly distributed within each year of age.
(6) i = 0.06.
Calculate the amount of each semiannual benefit premium for this insurance.
A.
1.3
B.
1.4
C.
1.5
D.
1.6
E.
1.7
2. A loss X follows a 2-parameter Pareto distribution with = 2 and unspecified

5
parameter . You are given E[X 100| X > 100] = E[X 50|X > 50]. Calculate
3
E[X 150| X > 150].
A.
150
B.
175
C.
200
D.
225
E.
250
298
3. The scores on the final exam in Ms. Bs Latin class have a normal distribution
with mean and standard deviation equal to 8. is a random variable with a
normal distribution with mean equal to 75 and standard deviation equal to 6. Each
year Ms. B chooses a student at random and pays the student 1 times the students
score. However, if the student fails the exam (score 65) then there is no payment.
Calculate the conditional probability that the payment is less than 90 given that there
is a payment.
A.
0.77
B.
0.85
C.
0.88
D.
0.92
E.
1.00
4. For a Markov model with three states, Healthy (0), Disabled (1), and Dead (2)
0.70 0.20 0.10
(1) The annual transition matrix is given by

0.10 0.65 0.25 with the
0
0
1
states listed vertically and horizontally in the order 0, 1, 2.
(2) There are 100 lives at the start, all Healthy. Their future states are independent.
Calculate the variance of the number of the original 100 lives who die within the
first two years.
A.
11
B.
14
C.
17
D.
20
E.
23
299
5. An insurance company issues a special 3 year insurance to a high risk individual.

You are given the following homogeneous Markov chain model.
(1)
State 1
State 2
State 3
State 4
active
disabled
withdrawn
dead
0.4 0.2 0.3 0.1

0.2 0.5
0 0.3
with the states listed
The transition probability matrix is

0
0
1
0
0
0
0
1
vertically and horizontally in the order 1, 2, 3, 4.
(2) Changes in state occur at the end of the year.
(3) The death benefit is 1000, payable at the end of the year of death.
(4) i = 0.05
(5) The insured is disabled at the end of year 1.
Calculate the actuarial present value of the prospective death benefits at the beginning
of year 2.
A.
440
B.
528
C.
634
D.
712
E.
803
300
6. For a fully discrete whole life insurance of b on (x) you are given
(1) qx+9 = 0.02904
(2) i = 0.03
(3) The initial benefit reserve for policy year 10 is 343.
(4) The net amount at risk for policy year 10 is 872.
(5) a x = 14.65976.
Calculate the terminal benefit reserve for policy year 9.
A.
280
B.
288
C.
296
D.
304
E.
312
7. For a special fully discrete 2 year endowment insurance of 1000 on (x) you are
given
(1) The first year benefit premium is 668.
(2) The second year benefit premium is 258.
(3) d = 0.06.
Calculate the level annual premium using the equivalence principle.
A.
469
B.
479
C.
489
D.
499
E.
509

8. For an increasing 10 year term insurance you are given
(1) bk+1 = 100, 000(1 + k) for k = 0, 1, . . . , 9
(2) Benefits are payable at the end of the year of death.
(4) i = 0.06
(5) The single benefit premium for this insurance on (41) is 16,736.
Calculate the single benefit premium for this insurance on (40).
A.
12,700
B.
13,600
C.
14,500
D.
15,500
E.
16,300
301
302

1. The semiannual premium P satisfies the equation 1000vq30 + 500v2 p30 q31 +
250v3 2 p30 q32 = P + Pv1/ 2 1/ 2 p30 . The life table gives P = 1.276. A.
2. Since X = (X x)+ +X x, E[X x| X > x]P[X > x] = E[(X x)+ ] = E[X]E[X x].
Using the formulas for the Pareto distribution found in the tables for exam M shows
that E[X x| X > x] = x + . The given information implies 100 + = (5/ 3)(50 + ),
from which = 25 and E[X 150|X > 150] = 150 + 25 = 175. B.
3. Since E[etS ] = E[E[etS | ]] = E[et+64t / 2 ] = e75t+100t / 2 , the student score S has
a normal distribution with mean 75 and variance 100. The desired probability is
P[S < 90| S > 65] = P[1 < N(0, 1) < 1.5]/ P[N(0, 1) > 1] = 0.9206. D.
2
4. The probability that a single individual goes from healthy to dead in 2 years or
less is 0.1 + (0.7)(0.10) + (0.2)(0.25) = 0.22. So the number dying within 2 years
is binomial with parameters 100 and 0.22, giving the variance as 100(0.22)(0.78) =
17.16. C.
5. To collect, the insured must die in either 1 or 2 years, giving the value as
1000(0.30v + v2 ((0.2)(0.1) + (0.5)(0.3))) = 439.90. A.
6. The premium P = bAx / a x = b(1 da x )/ a x = 0.0391b. Also 343 = 9 V + P,
872 = b 10 V, and (9 V + P)(1.03) = bqx+9 + px+910 V = qx+9 (b 10 V) + 10 V. Using the
earlier information in the last equation gives 10 V = 327.97, from which b = 1199.97,
P = 46.92 and 9 V = 296.08. C.
7. On the one hand 1000Ax:2 = 668 + 258vpx , while also 1000Ax:2 = 1000vqx +
1000v2 px , since this is endowment insurance. Equating these two expressions gives
px = 0.91 and Ax:2 = 0.888 from which the premium is 1000dAx:2 / (1 Ax:2 ) =
479.05.B.
8. Here 100, 000(IA)1
41:10
100, 000(IA)1
40:10
the relation A1
40:10
= 16, 736. Now by reasoning from the time line diagram,

= 100, 000A1
40:10
+100, 000vp40 (IA)1
41:10
10v10 9 p41 q50 . Using
= A40 10 E40 A50 and the life table gives the value as 15, 513. D.

1. For a fully discrete whole life insurance of 1000 on (x)
(1) Death is the only decrement.
(2) The annual benefit premium is 80.
(3) The annual contract premium is 100.
(4) Expenses in year 1, payable at the start of the year, are 40% of contract
premiums.
(5) i = 0.10
(6) 10001 Vx = 40
Calculate the asset share at the end of the first year.
A.
17
B.
18
C.
19
D.
20
E.
21
304
2. For a collective risk model the number of losses, N, has a Poisson distribution
with = 20. The common distribution of the individual losses has the following
characteristics.
(1) E[X] = 70
(2) E[X 30] = 25
(3) P[X > 30] = 0.75
(4) E[X 2 |X > 30] = 9000
An insurance covers aggregate losses subject to an ordinary deductible of 30 per
loss. Calculate the variance of the aggregate payments of the insurance.
A.
54,000
B.
67,500
C.
81,000
D.
94,500
E.
108,000
3. For a collective risk model

(1) The number of losses has a Poisson distribution with = 2.
(2) The common distribution of the individual losses is
x
1
2
fX (x)
0.6
0.4
An insurance covers aggregate losses subject to a deductible of 3. Calculate the

expected aggregate payments of the insurance.
A.
0.74
B.
0.79
C.
0.84
D.
0.89
E.
0.94
305
4. A discrete probability distribution has the following properties:

(1) pk = c(1 + 1k )pk1 for k = 1, 2, . . .
(2) p0 = 0.5
Calculate c.
A.
0.06
B.
0.13
C.
0.29
D.
0.35
E.
0.40
5. A fully discrete 3 year term insurance of 10,000 on (40) is based on a double

decrement model, death and withdrawal:
(1) Decrement 1 is death.
(1)
(2) 40
(t) = 0.02, t 0
(3) Decrement 2 is withdrawal, which occurs at the end of the year.
(2)
(4) q 40+k = 0.04, k = 0, 1, 2
(5) v = 0.95
Calculate the actuarial present value of the death benefits for this insurance.
A.
487
B.
497
C.
507
D.
517
E.
527
306
6. You are given

(1) e 30:40 = 27.692
x
(2) s(x) = 1 , 0 x
(3) T(x) is the future lifetime random variable for (x).

Calculate Var(T(30)).
A.
332
B.
352
C.
372
D.
392
E.
412
7. For a fully discrete 5 payment 10 year decreasing term insurance on (60) you are
given
(1) bk+1 = 1000(10 k) for k = 0, 1, . . . , 9
(2) Level benefit premiums are payable for five years and equal 218.15 each.
(3) q60+k = 0.02 + 0.001k, k = 0, 1, . . . , 9.
(4) i = 0.06
Calculate 2 V, the benefit reserve at the end of year 2.
A.
70
B.
72
C.
74
D.
76
E.
78
307
8. You are given

(1) T(x) and T(y) are not independent.
(2) qx+k = qy+k = 0.05, k = 0, 1, 2, . . .
(3) k pxy = 1.02k px k py , k = 1, 2, . . .
Into which of the following ranges does ex:y , the curtate expectation of life of the
last survivor status, fall?
A.
ex:y 25.7
B.
25.7 < ex:y 26.7
C.
26.7 < ex:y 27.7
D.
27.7 < ex:y 28.7
E.
28.7 < ex:y
308

1. In terms of the gross premium G, 1 AS px = (G expenses)(1 + i) 1000qx . Using
(6), px 1 Vx = 80(1 + i) 1000qx , from which qx = 48/ 960, and thus 1 AS = 16.8. A.
P
2. The problem is to compute Var( Nj=1 (Xj 30)+ ). By the usual conditioning
argument, this variance is E[N]Var((X 30)+ ) + (E[(X 30)+ ])2 Var(N). Using the
relation X = (X 30) + (X 30)+ together with (1) and (2) gives E[(X 30)+ ] =
70 25 = 45. Now E[(X 30)2+ ] = E[(X 30)2 | X > 30] P[X > 30]. Squaring
out and using (3) and (4) gives this value as 3,375. Thus the variance sought is
20(1, 350) + (2, 025)(20) = 67, 500. B.
P
3. The objective is to compute E[( Nj=1 Xj 3)+ ]. The relation W = W 3 + (W 3)+

P
allows the computation of E[ Nj=1 Xj 3] to be made instead. A table of values
yields this last expectation as 2.0636, and the original expectation as 2(1.4)2.0636 =
0.7364.A.
4. This is an (a, b, 0) distribution, from which the probabilities are negative binomial
probabilities. C.
0.02
(0.96) = 0.9410
5. Since withdrawal is always at the end of the year, p()
x = e
0.02
2 ()
for all x. The actuarial present value is 10, 000(v(1 e
) + v px (1 e0.02 ) +
0.02
3 () 2
) = 506.60. C.
v (px ) (1 e
6. Here T(30) is uniform on the interval (0, 30), so e 30:40 =

Z
40
40
t p30
dt =
1 t/ ( 30) dt = 40 800/ ( 30). Equating this to the given value yields
= 95, and Var(T(30)) = (65)2 / 12 = 352.08. B.

7. Since 0 V = 0 the recursive reserve formula gives (0 V +218.15)(1+i)10, 000qx =
px 1 V, from which 1 V = 31.8765. Using the recursion again gives (1 V + 218.15)(1 +
i) 9, 000qx+1 = px+1 2 V, from which 2 V = 77.659. E.
8. Since exy = ex + ey exy , and ex = ey =
X
k=1
X
k=1
k px =
(0.95)k = 19, and exy =
k=1
(1.02)(0.95)2k = (1.02)(0.95)2 / (1 (0.95)2 ) = 9.4415, exy = 28.5585. D.

1. Subway trains arrive at your station at a Poisson rate of 20 per hour. 25% of the
trains are express and 75% are local. The types and number of trains arriving are
independent. An express gets you to work in 16 minutes and a local gets you there
in 28 minutes. You always take the first train to arrive. Your co-worker always takes
the first express. You are both waiting at the same station. Calculate the conditional
probability that you arrive at work before your co-worker, given that a local arrives
first.
A.
37%
B.
40%
C.
43%
D.
46%
E.
49%
2. Beginning with the first full moon in October deer are hit by cars at a Poisson
rate of 20 per day. The time between when a deer is hit and when it is discovered by
highway maintenance has an exponential distribution with a mean of 7 days. The
number hit and the times until they are discovered are independent. Calculate the
expected number of deer that will be discovered in the first 10 days following the
first full moon in October.
A.
78
B.
82
C.
86
D.
90
E.
94
310
3. You are given

(1) x (t) = 0.03, t 0
(2) = 0.05
(3) T(x) is the future lifetime random variable.
(4) g is the standard deviation of aT(x) .
Calculate P[aT(x) > ax g].
A.
0.53
B.
0.56
C.
0.63
D.
0.68
E.
0.79
4. (50) is an employee of XYZ Corporation. Future employment with XYZ follows

a double decrement model.
(1) Decrement 1 is retirement.
(1)
(1)
(2) 50
(t) = 0 for 0 t < 5 and 50
(t) = 0.02 for t 5.
(3) Decrement 2 is leaving employment with XYZ for all other causes.
(2)
(2)
(4) 50
(t) = 0.05 for 0 t < 5 and 50
(t) = 0.03 for t 5.
(5) If (50) leaves employment with XYZ he will never rejoin XYZ.
Calculate the probability that (50) will retire from XYZ before age 60.
A.
0.069
B.
0.074
C.
0.079
D.
0.084
E.
0.089
311
5. For a life table with a one year select period, you are given
(1)
x
80
81
l[x]
1000
920
d[x]
90
90
lx+1
e [x]
8.5

Calculate e [81] .
A.
8.0
B.
8.1
C.
8.2
D.
8.3
E.
8.4
6. For a fully discrete 3 year endowment insurance of 1000 on (x), i = 0.05 and
px = px+1 = 0.7. Calculate the second year terminal benefit reserve.
A.
526
B.
632
C.
739
D.
845
E.
952
312
7. For a fully discrete whole life insurance of 1000 on (50) you are given
(1) The annual per policy expense is 1.
(2) There is an additional first year expense of 15.
(3) The claim settlement expense of 50 is payable when the claim is paid.
(4) All expenses, except the claim settlement expense, are paid at the beginning
of the year.
(5) Mortality follows DeMoivres law with = 100.
(6) i = 0.05
Calculate the level expense loaded premium using the equivalence principle.
A.
27
B.
28
C.
29
D.
30
E.
31
8. The repair costs for boats in a marina have the following characteristics.
Boat Type
Power boat
Sailboat
Luxury yachts
Number
100
300
50
Repair Probability
0.3
0.1
0.6
Mean Repair Cost

300
1000
5000
Repair Cost Variance

10,000
400,000
2,000,000
At most one repair is required per boat each year. The marina budgets an
amount Y equal to the aggregate mean repair costs plus the standard deviaton of the
aggregate repair costs. Calculate Y.
A.
200,000
B.
210,000
C.
220,000
D.
230,000
E.
240,000
313

1. If an express arrives within 28 16 = 12 minutes, the co-worker will arrive
first. Since 12 minutes is 1/ 5 hour, and since the times between express trains is
exponential with mean 1/ 5, the probability the co-worker does not arrive first is
e5(1/ 5) = 0.3679. A.
2. The probability that a deer hit at time t is discovered by time 10 is 1 e(10t)/ 7 .
Given the number of deer hit, the times at which they were hit are each uniform on
the interval from 0 to 10. So the probability
that a deer that is hit in the first 10 days
Z
10
is discovered in the first 10 days is (1/ 10)

1 e(10t)/ 7 dt = 0.4677. Since there
0
are 200 hits on the average in 10 days and each hit is discovered with probability
0.4677, the expected number of discoveries is 200(0.4677) = 93.55. E.
3. Under the exponential life model, Ax = / ( + ) = 3/ 8 and 2 Ax = / ( + 2 ) =

3/ 13, by direct computation. Thus ax = (1 Ax )/ = 12.5. Also, since aT(x) =
(1 vT(x) )/ , Var(aT(x) ) = (2 Ax A2x )/ 2 , so g = 6.0048. The desired probability is
P[aT(x) > 6.5] = P[vT < 0.6752] = P[T > 7.8537] = e7.85370.03 = 0.7901. E.
4. In order to retire at age 50 + t the employee
must survive all
causes to this age and
Z
Z
10
() (1)
then instantly retire. So the probability is
t p50 50+t dt =
5
0
Z 10
2 0.25 0.50
0.10.02t 0.10.03t
e
e
(0.02) dt = (e
e
) = 0.0689. A.
5
5
10
(1)
(2) (1)
t p 50 t p 50 50+t
dt =
5. Direct computation from the table gives p[80] = 910/ 1000, p[81] = 830/ 920,
and p81 = 830/ 910. By UDD, e [80] = 1/ 2 + e[80] = 1/ 2 + p[80] (1 + e81 ) = 1/ 2 +
p[80] (1 + p81 (1 + e82 )), from which 1 + e82 = 8.5421. Similarly, e [81] = 1/ 2 + e[82] =
1/ 2 + p[81] (1 + e82 ) = 8.2065. C.
6. Using the given information yields a x:3 = 1 + vpx + v2 px px+1 = 2.111. So using the
prospective method, 10002 V = 1000(Ax+2:1 Px:3 a x+2:1 ) = 1000(v(1da x:3 )/ a x:3 ) =
526.31. A.
7. The equivalence principle gives 1000A50 + 1a50 + 15 + 50A50 = Ga50 . Using
P
P
50k 1
k+1
k+1
= a50 / 50 = 0.3651. Then
DeMoivre, A50 = 49
= 49
k=0 k px qx+k v
k=0 50 50k v
a 50 = 13.3325 and G = 30.8799. E.
8. The repair cost for each category is of the form
B
X
Xj where B is a binomial
j=1
random variable counting the number of boats of that type which need repair
and Xj is the repair cost for the jth boat. The expected
cost is thus 30(300) +

30(1000) + 30(5000) = 189, 000, and the variance is (300)2 (21) + 30(10, 000) +

10002 (27) + 30(400, 000) + 50002 (12) + 30(2, 000, 000) = 401, 190, 000. Thus

Y = 189, 000 + 20, 029 = 209, 029. B.
314

1. For an insurance
(1) Losses can be 100, 200, or 300 with respective probabilities 0.2, 0.2, and
0.6.
(2) The insurance has an ordinary deductible of 150 per loss.
(3) Y P is the claim payment per payment random variable.
Calculate Var(Y P ).
A.
1500
B.
1875
C.
2250
D.
2625
E.
3000
2. You are given x = 0.05 for 50 x < 60 and x = 0.04 for 60 x < 70. Calculate
4| 14 q50 .
A.
0.38
B.
0.39
C.
0.41
D.
0.43
E.
0.44
316
3. The distribution of a loss X is a two point mixture. With probability 0.8, X has
a two parameter Pareto distribution with = 2 and = 100. With probability 0.2,
X has a two parameter Pareto distribution with = 4 and = 3000. Calculate
P[X 200].
A.
0.76
B.
0.79
C.
0.82
D.
0.85
E.
0.88
4. For a special fully discrete 5 year deferred whole life insurance of 100,000 on
(40) you are given
(1) The death benefit during the 5 year deferral period is return of benefit
premiums paid without interest.
(2) Annual benefit premiums are payable only during the deferral period.
(4) i = 0.06
(5) (IA)1 = 0.04042
40:5
Calculate the annual benefit premiums.

A.
3300
B.
3320
C.
3340
D.
3360
E.
3380
317
5. You are pricing a special 3 year annuity due on two idependent lives, both age
80. The annuity pays 30,000 if both persons are alive and 20,000 if only one person
is alive. You are given
k
1
2
3
k p80
0.91
0.82
0.72
and that i = 0.05. Calculate the actuarial present value of this annuity.
A.
78,300
B.
80,400
C.
82,500
D.
84,700
E.
86,800
6. Company ABC sets the contract premium for a continuous life annuity of 1
per year on (x) equal to the single benefit premium calculated using = 0.03 and
x (t) = 0.02, for t 0. However, a revised mortality assumption reflects future
mortality improvement and is given by x (t) = 0.02 for t 10 but x (t) = 0.01 for
t > 10. Calculate the expected loss at issue for ABC (using the revised mortality
assumption) as a percentage of contract premium.
A.
2%
B.
8%
C.
15%
D.
20%
E.
23%
318
7. A group of 1000 lives each age 30 sets up a fund to pay 1000 at the end of the
first year for each member who dies in the first year, and 500 at the end of the second
year for each member who dies in the second year. Each member pays into the fund
an amount equal to the single benefit premium for a special 2 year term insurance
with
(1) Benefits b1 = 1000 and b2 = 500
(3) i = 0.06
The actual experience of the fund is as follows.
k
0
1
Interest Rate Earned

0.070
0.069
Number of Deaths
1
1
Calculate the difference, at the end of the second year, between the expected
size of the fund as projected at time 0 and the actual fund.
A.
840
B.
870
C.
900
D.
930
E.
960
319
8. In a certain town the number of common colds an individual will get in a year
follows a Poisson distribution that depends on the individuals age and smoking
status. The distribution of population and the mean number of colds are as follows.
Children
Adult non-smokers
Adult smokers
Proportion of Population
0.30
0.60
0.10
Mean number of colds

3
1
4
Calculate the conditional probability that a person with exactly 3 common colds
in a year is an adult smoker.
A.
0.12
B.
0.16
C.
0.20
D.
0.24
E.
0.28
320

1. The definition of Y P means that it is the conditional distribution of the claim
size given that a claim is made. Since the deductible is 150, a loss of 100 will not
generate a claim. So the values of Y P are 50 with probability 0.2/ 0.8 = 0.25 and
150 with probability 0.6/ 0.8. The variance is directly computed to be 1875. B.
2. Using the given information gives 4 p50 = e(0.05)(4) and 14 p54 = e(0.05)6(0.04)8 =
e0.62 . So 4|14 q50 = 4 p50 (1 14 p50 ) = e0.20 (1 e0.62 ) = 0.3783. A.
3. Using the information in table A.2.4.1 of the tables for exam M gives the probability as 0.8(1 (100/ 300)2 ) + 0.2(1 (3000/ 3200)4 ) = 0.7566. A.
4. The premium G satisfies 100, 0005 E40 A45 + G(IA)1 = Ga40:5 . Since a 40:5 =
40:5
a 40 5 E40 a 45 , direct use of the tables gives G = 3362.51. D.
5. The probability that exactly one of the two is alive at time k is the complement
of the probability that either both are alive or both are dead, that is, 1 (k p80 )2 (1
2
2
k p80 ) . Using this and direct reasoning gives the value as 30, 000 + 30, 000v(0.91) +
2
2
2
2
2
2
20, 000v(1 (0.91) (1 0.91) ) + 30, 000v (0.82) + 20, 000v (1 (0.82) (1
0.82)2 ) = 80431.70. B.
Z
6. Originally, ax =
0.03t 0.20.01(t10)
10
0.03t 0.02t
dt = 20. The revised value is
10
e0.03t e0.02t dt +
dt = (1 e0.5 )/ 0.05 + e0.5 / 0.04 = 23.032. The loss is thus 3,
or 300/ 20 = 15% of the original premium. C.

7. The expected balance is 0, since the premium is the net premium. The net
premium is 1000vq30 + 500v2 p30 q31 = 2.1587. The fund starts with 2, 158.70, earns
interest at the given rates and pays one claim at the end of each year, leaving 900 at
the end of 2 years. C.
8. The probability of exactly 3 colds for a single person of each group is e3 e3 / 3! =
0.2240, e1 13 / 3! = 0.0613, and e4 43 / 3! = 0.1954 respectively. Bayes theorem
(or direct reasoning) gives the desired probability as 0.1(0.1954)/ ((0.1)(0.1954) +
(0.6)(0.0613) + (0.3)(0.2240)) = 0.1582. B.

Financial Mathematicg CT 1

Uploaded by

Copyright:

Available Formats

Financial Mathematicg CT 1

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Financial Mathematicg CT 1

Uploaded by

Copyright:

Available Formats

Lecture Notes

Copyright 2006 Jerry Alan Veeh. All rights reserved.

Copyright 2006 Jerry Alan Veeh. All rights reserved.

Copyright 2006 Jerry Alan Veeh. All rights reserved.

2. Elements of the Theory of Interest

2: Elements of the Theory of Interest

frequently. Some notation is needed to discuss these situations conveniently. Denote

Exercise 23. Compute 0.05(12) .

2: Elements of the Theory of Interest

The relationship is d = iv follows by moving the interest payment back in time

2: Elements of the Theory of Interest

2: Elements of the Theory of Interest

2: Elements of the Theory of Interest

of Calculus shows that this implies (t) = i/ (1 + it), for t > 0.

2: Elements of the Theory of Interest

3. Cash Flow Valuation

3: Cash Flow Valuation

3: Cash Flow Valuation

4. Sample Question Set 1

Copyright 2006 Jerry Alan Veeh. All rights reserved.

4: Sample Question Set 1

Question 45 . On January 1, 1980, Jack deposited 1,000 into Bank X to earn

4: Sample Question Set 1

Answers to Sample Questions

Question 41 . The amount in fund A at time t is A(t) = 1000e 0 1+0.05s ds = 1000+50t,

Question 44 . The amount Jeff actually has is 100(1 0.20) e 2 t dt = 312.50,

5. Annuities, Amortization, and Sinking Funds

Copyright 2006 Jerry Alan Veeh. All rights reserved.

5: Annuities, Amortization, and Sinking Funds

5: Annuities, Amortization, and Sinking Funds

5: Annuities, Amortization, and Sinking Funds

periodic interest rate is i. Find a formula for a (m)

5: Annuities, Amortization, and Sinking Funds

Taking a more sophisticated viewpoint will exhibit a method of constructing

5: Annuities, Amortization, and Sinking Funds

5: Annuities, Amortization, and Sinking Funds

Problem 54. True or False:

Problem 55. Show that a2n = an (1 + vn ).

5: Annuities, Amortization, and Sinking Funds

he receives is based on an effective interest rate of 4% per annum. The insurance

5: Annuities, Amortization, and Sinking Funds

5: Annuities, Amortization, and Sinking Funds

5: Annuities, Amortization, and Sinking Funds

Direct computation gives the present value as

(pv pvn+1 rn )/ (1 vr) = p(1 vn rn )/ (1 + i r), provided vr 1.

The present value of the monthly payments must equal the

5: Annuities, Amortization, and Sinking Funds

5: Annuities, Amortization, and Sinking Funds

5: Annuities, Amortization, and Sinking Funds

6. Sample Question Set 2

Question 62 . An annuity immediate pays an initial benefit of one per year,

Question 63 . Determine an expression for

Copyright 2006 Jerry Alan Veeh. All rights reserved.

6: Sample Question Set 2

Question 64 . Which of the following are true?

I and III only

II and III only

I, II, and III

E. The correct answer is not given