3 Methods To Differentiate Term Insurance
3 Methods To Differentiate Term Insurance
3 Methods To Differentiate Term Insurance
Method 2: Applying the Leibniz rule (AM p. 701 first result) to the formula
n t
=
Ax1:n ∫0 v t px µ x +t dt
and noting the two limits of the integral are constant, we have
∂ 1 n ∂
= Ax:n ∫ vt ( t px µ x +t )dt .
∂x 0 ∂x
By the product rule,
∂ ∂ ∂ ∂
( t px µ x +t )= ( t px )µ x +t + t px µ x +t = t px (µ x − µ x +t )µ x +t + t px µ x +t .
∂x ∂x ∂x ∂t
∂ ∂ ∂ ∂
(That µ x +t = µ x +t follows from g ( x + t )= g ′( x + t )= g ( x + t ) which is true
∂x ∂t ∂x ∂t
because of the chain rule.) Hence, we have
∂ 1 n n ∂
Ax:n = µ x Ax1:n − ∫ vt t px (µ x +t ) 2dt + ∫ vt t px µ x +t dt .
∂x 0 0 ∂t
The last integral “invites” you to integrate by parts,
n ∂ n
∫0 t Ex ∂t µ x+t d=t ∫0 t Ex dµ x+t
t =n
( t Exµ x+t ) t =0 − ∫0 µ x +t d ( t E x )
n
=
( n Exµ x+n − µ x ) − ∫0 ( )
n
= µ x + t v t d t p x + t p x dv t .
t t
Method 3: Because vt t px = e ∫0 e ∫0
− ( δ+µ x + s )ds n − ( δ+µ x + s )ds
,=
we have Ax1:n ∫0 µ x + t dt .
By changing the time variables t and s to age variables
y = x+t and z = x+s, we have
z= y
e ∫z = x
y= x + n − ( δ+µ z )dz
=Ax1:n ∫y = x µ y dy .
By the Leibniz rule,
∂ − ∫z = x ( δ+µ z )dz
z= y
∂ 1 y= x + n
= A
∂x x:n ∫y = x
∂x
e µ y dy
=z x= +n
∂ ∂
z x
+ e ∫ ( x + n) − e ∫
− ( δ+µ z )dz − ( δ+µ z )dz
z=x z=x
µ x+n µx x.
∂x ∂x
Now, by the chain rule and Leibniz rule,
( )
=z y=z y
∂ − ∫ z x= ∂
e ∫z x
( δ+µ z )dz − ( δ+µ z )dz z= y
=
e= × − ∫ (δ + µ z )dz
∂x ∂x z=x
z= y z= y
∂
= e ∫z = x × − 0 + 0 − ( δ + µ x ) x = ( δ + µ x ) e ∫z = x
− ( δ+µ z )dz − ( δ+µ z )dz
.
∂x
∂ 1
Hence, Ax:n = (δ + µ x ) Ax1:n + n Ex µ x + n − µ x .
∂x
∂
Question: A =?
∂x x:n
Solution (By using the answer above): Because A=
x:n
Ax1:n + n Ex and because
( )
z=
x+n z=
x+n
∂= ∂ − ∫z x= ∂
= e ∫z x
( δ+µ z )dz − ( δ+µ z )dz z= x+ n
n=
Ex e × −∫ (δ + µ z )d
=z n E x ( −µ x + n + µ x )
∂x ∂x ∂x z =x
(the last formula is #10(i) in Midterm Exam 1; it can also be obtained by using
∂
n=px n px (µ x − µ x + n ) ), we have
∂x
∂
Ax:n = (δ + µ x ) Ax1:n − µ x + n Ex µ x = δAx1:n − µ x (1 − Ax:n ) .
∂x