Exam LTAM

Updated 9/10/2018
Adapt to Your Exam

SURVIVAL DISTRIBUTIONS
SURVIVAL DISTRIBUTIONS
SURVIVAL DISTRIBUTIONS Express 𝑝𝑝’s or 𝑞𝑞’s in terms of 𝜇𝜇
Express 𝑝𝑝’s or 𝑞𝑞’s in terms of 𝜇𝜇 𝑙𝑙# 𝑙𝑙−
# − 𝑙𝑙#K6
𝑙𝑙#K6 𝑡𝑡 𝑡𝑡
jj 6 𝑞𝑞#
6 𝑞𝑞=
# = ==
𝑙𝑙# 𝑙𝑙# 𝜔𝜔 𝜔𝜔
−− 𝑥𝑥 𝑥𝑥
Probability Functions
Probability Functions 6 𝑝𝑝#
6 𝑝𝑝= [ [g.𝑝𝑝g#.𝑝𝑝⋅#𝜇𝜇⋅#Kg
# = 𝜇𝜇#Kg
d𝑠𝑠 d𝑠𝑠

𝑙𝑙#K:
𝑙𝑙#K: −− 𝑙𝑙#K:K6
𝑙𝑙#K:K6 𝑡𝑡 𝑡𝑡
Survival Function
Survival Function 6 6
6 6 𝑞𝑞#𝑞𝑞=
:|6:|6 # = ==
𝑇𝑇#𝑇𝑇
: future lifetime or time-to-death of (𝑥𝑥) 𝑙𝑙# 𝑙𝑙# 𝜔𝜔 𝜔𝜔
−− 𝑥𝑥 𝑥𝑥
# : future lifetime or time-to-death of (𝑥𝑥) 6 𝑞𝑞=
6 𝑞𝑞# # =
[[ . .
g 𝑝𝑝g#𝑝𝑝⋅#𝜇𝜇⋅#Kg
𝜇𝜇#Kg
d𝑠𝑠 d𝑠𝑠
∘ ∘ 𝜔𝜔 𝜔𝜔 −− 𝑥𝑥 𝑥𝑥
𝑆𝑆#𝑆𝑆(𝑡𝑡)
# (𝑡𝑡)
== Probability that (𝑥𝑥) survives 𝑡𝑡 years
Probability that (𝑥𝑥) survives 𝑡𝑡 years / / 𝑒𝑒#𝑒𝑒=
# =
:K6 :K6 22
== Pr[𝑇𝑇
Pr[𝑇𝑇# ># >𝑡𝑡] 𝑡𝑡] ∘ ∘ 𝑛𝑛 𝑛𝑛
𝑞𝑞#𝑞𝑞=
:|6:|6 [ [ g.𝑝𝑝g#.𝑝𝑝⋅#𝜇𝜇⋅#Kg
# = 𝜇𝜇#Kg
d𝑠𝑠 d𝑠𝑠
𝑒𝑒#:W|
𝑒𝑒#:W| = =W𝑝𝑝W#𝑝𝑝(𝑛𝑛)
# (𝑛𝑛)
+ +W𝑞𝑞W#𝑞𝑞q # q r r
== Pr[𝑇𝑇
Pr[𝑇𝑇/ >/ >𝑥𝑥 +𝑥𝑥 +𝑡𝑡|𝑇𝑇𝑡𝑡|𝑇𝑇
/ >/ >
𝑥𝑥] 𝑥𝑥] : : 22
𝑆𝑆/𝑆𝑆(𝑥𝑥
/ (𝑥𝑥
++ 𝑡𝑡) 𝑡𝑡)

== Moments
Moments Beta Distribution
Beta Distribution
(𝑥𝑥)
𝑆𝑆/𝑆𝑆 /(𝑥𝑥) 𝑙𝑙# 𝑙𝑙= } }
Complete Expectation
Complete Expectation # = 𝑘𝑘(𝜔𝜔
𝑘𝑘(𝜔𝜔 −− 𝑥𝑥)𝑥𝑥)
𝑆𝑆#𝑆𝑆(𝑡𝑡) must satisfy:
# (𝑡𝑡) must satisfy: • •First Moment
First Moment 𝛼𝛼 𝛼𝛼 𝛼𝛼 𝛼𝛼
• •𝑆𝑆#𝑆𝑆(0) 𝜇𝜇#𝜇𝜇= # = ⇒
⇒
𝜇𝜇 #K6
𝜇𝜇#K6
==
# (0)== 1 1 ∘ ∘
jj jj
𝜔𝜔 𝜔𝜔−− 𝑥𝑥 𝑥𝑥 𝜔𝜔 𝜔𝜔
−−(𝑥𝑥(𝑥𝑥
++ 𝑡𝑡) 𝑡𝑡)
• •𝑆𝑆#𝑆𝑆(∞)# (∞) == 0 0 𝑒𝑒#𝑒𝑒=
# =
𝐸𝐸[𝑇𝑇# ] #=
𝐸𝐸[𝑇𝑇 ]=[ [𝑡𝑡 ⋅𝑡𝑡 6⋅𝑝𝑝#6𝑝𝑝𝜇𝜇##K6
𝜇𝜇#K6
d𝑡𝑡d𝑡𝑡
== [ [ 6𝑝𝑝#6𝑝𝑝d𝑡𝑡
# d𝑡𝑡 𝑙𝑙#K6 𝜔𝜔 𝜔𝜔
𝑙𝑙#K6 −− ++ (𝑥𝑥(𝑥𝑥 } }
𝑡𝑡) 𝑡𝑡)
/ / / /
• •𝑆𝑆#𝑆𝑆(𝑡𝑡) is a non-increasing function of t
# (𝑡𝑡) is a non-increasing function of t 6 𝑝𝑝=
6 𝑝𝑝# # = ==   Ä Ä
• •Second Moment
Second Moment 𝑙𝑙# 𝑙𝑙# 𝜔𝜔 𝜔𝜔 −− 𝑥𝑥 𝑥𝑥
Actuarial Notations
Actuarial Notations jj jj ∘ ∘ 𝜔𝜔 𝜔𝜔
− −
𝑥𝑥 𝑥𝑥
𝐸𝐸[𝑇𝑇 l l]
# ]#=
𝐸𝐸[𝑇𝑇 [ [𝑡𝑡 l𝑡𝑡⋅l 6⋅𝑝𝑝#6𝑝𝑝𝜇𝜇##K6
= 𝜇𝜇#K6
d𝑡𝑡d𝑡𝑡
== [ [2𝑡𝑡2𝑡𝑡
⋅ 6⋅𝑝𝑝#6𝑝𝑝d𝑡𝑡
# d𝑡𝑡 𝑒𝑒#𝑒𝑒=# =
6 𝑝𝑝#
6 𝑝𝑝=
# =
Probability that (𝑥𝑥) survives 𝑡𝑡 years
Probability that (𝑥𝑥) survives 𝑡𝑡 years / / / /
𝛼𝛼 +𝛼𝛼 + 11

== Pr(𝑇𝑇
Pr(𝑇𝑇 # > # > 𝑡𝑡) 𝑡𝑡) • •Variance
Variance Gompertz’s Law
Gompertz’s Law
𝑆𝑆#𝑆𝑆(𝑡𝑡) ∘ ∘l l
== # (𝑡𝑡) l l] 𝜇𝜇#𝜇𝜇=# =
# #
𝐵𝐵𝑐𝑐𝐵𝐵𝑐𝑐
𝑐𝑐
𝑐𝑐
>> 1, 1,
𝐵𝐵 𝐵𝐵
>> 0 0
# ] #=
𝑉𝑉𝑉𝑉𝑉𝑉[𝑇𝑇
𝑉𝑉𝑉𝑉𝑉𝑉[𝑇𝑇 ]= 𝐸𝐸[𝑇𝑇
𝐸𝐸[𝑇𝑇# ]#− −
q𝑒𝑒q𝑒𝑒
# r# r

# #
6 𝑞𝑞=
6 𝑞𝑞# # =
Probability that (𝑥𝑥) dies within 𝑡𝑡 years
Probability that (𝑥𝑥) dies within 𝑡𝑡 years 𝐵𝐵𝑐𝑐𝐵𝐵𝑐𝑐 6 6
==Pr(𝑇𝑇
Pr(𝑇𝑇 6 𝑝𝑝=
6 𝑝𝑝# # =exp exp Ñ−Ñ− (𝑐𝑐(𝑐𝑐− − 1)Ö
1)Ö
# ≤ # ≤ 𝑡𝑡) 𝑡𝑡) Curtate Expectation
Curtate Expectation ln ln𝑐𝑐 𝑐𝑐
==𝐹𝐹# (𝑡𝑡)
𝐹𝐹# (𝑡𝑡) • •First Moment
First Moment

jj jj Makeham’s Law
Makeham’s Law
6 𝑝𝑝+
6 𝑝𝑝# # +6 𝑞𝑞#
6 𝑞𝑞=
# = 1 1 # #
𝑒𝑒#𝑒𝑒=
# =
𝐸𝐸[𝐾𝐾# ] #=
𝐸𝐸[𝐾𝐾 ]=ss
𝑘𝑘 ⋅𝑘𝑘Q|⋅𝑞𝑞Q|#𝑞𝑞=
# =
ssQ 𝑝𝑝Q#𝑝𝑝 # 𝜇𝜇#𝜇𝜇=# = 𝐴𝐴 +
𝐴𝐴 +
𝐵𝐵𝑐𝑐𝐵𝐵𝑐𝑐
𝑐𝑐
𝑐𝑐
>> 1, 1,
𝐵𝐵 𝐵𝐵
>> 0, 0,
𝐴𝐴 𝐴𝐴
≥≥ −𝐵𝐵
−𝐵𝐵
𝑞𝑞#𝑞𝑞=
:|6:|6 # =Probability that (𝑥𝑥) survives 𝑢𝑢 years
Probability that (𝑥𝑥) survives 𝑢𝑢 years # #
𝐵𝐵𝑐𝑐𝐵𝐵𝑐𝑐
Qt/
Qt/ Qtu
Qtu Ñ−Ñ− (𝑐𝑐(𝑐𝑐− − 6 6
and dies within the following 𝑡𝑡 years
and dies within the following 𝑡𝑡 years 6 𝑝𝑝=
6 𝑝𝑝# # =expexp 1)Ö1)Ö ⋅ exp(−𝐴𝐴𝐴𝐴)
⋅ exp(−𝐴𝐴𝐴𝐴)
• •Second Moment
Second Moment ln ln𝑐𝑐 𝑐𝑐
= =
: 𝑝𝑝:#𝑝𝑝⋅ #6 𝑞𝑞
⋅ #K:
6 𝑞𝑞#K: jj jj

= = Fractional Ages
Fractional Ages
: 𝑝𝑝:#𝑝𝑝−
# − 𝑝𝑝#𝑝𝑝 # l l]
:K6 :K6 𝐸𝐸[𝐾𝐾
𝐸𝐸[𝐾𝐾# ]#= =
ss𝑘𝑘 l𝑘𝑘⋅lQ|⋅𝑞𝑞Q|#𝑞𝑞=
# =
s(2𝑘𝑘
s(2𝑘𝑘
−−1)1)
Q 𝑝𝑝Q#𝑝𝑝
#
= = 𝑞𝑞#𝑞𝑞− UDD (0
UDD (0 ≤≤ 𝑡𝑡 < 𝑡𝑡 <1) 1)
:K6 :K6 # −
: 𝑞𝑞:#𝑞𝑞
# Qt/
Qt/ Qtu
Qtu
Use linear interpolation:
Use linear interpolation:

• •Variance
Variance
Curtate Future Lifetime
Curtate Future Lifetime l l] (𝑒𝑒(𝑒𝑒 𝑙𝑙#K6𝑙𝑙#K6 ==(1(1 −− 𝑡𝑡) 𝑡𝑡)
⋅ 𝑙𝑙#⋅ 𝑙𝑙+
# +
𝑡𝑡 ⋅𝑡𝑡𝑙𝑙#Ku
⋅ 𝑙𝑙#Ku

𝑉𝑉𝑉𝑉𝑉𝑉[𝐾𝐾# ] #=
𝑉𝑉𝑉𝑉𝑉𝑉[𝐾𝐾 ]= # ]#−
𝐸𝐸[𝐾𝐾
𝐸𝐸[𝐾𝐾 − # )#l )l
𝐾𝐾#𝐾𝐾: number of completed future years by (𝑥𝑥) prior
# : number of completed future years by (𝑥𝑥) prior
6 𝑞𝑞#6 𝑞𝑞=
# =𝑡𝑡 ⋅𝑡𝑡𝑞𝑞⋅#𝑞𝑞 #
to death
to death Temporary Expectation
Temporary Expectation 𝑞𝑞#𝑞𝑞#
W W W W 𝜇𝜇#K6𝜇𝜇#K6 ==
𝐾𝐾#𝐾𝐾= # = ⌊𝑇𝑇#⌊𝑇𝑇
⌋ # ⌋ ∘ ∘ 1− 1− 𝑡𝑡 ⋅𝑡𝑡𝑞𝑞⋅#𝑞𝑞#
𝑒𝑒#:W|
𝑒𝑒#:W|
== [ [𝑡𝑡 ⋅𝑡𝑡 6⋅𝑝𝑝#6𝑝𝑝𝜇𝜇##K6
𝜇𝜇#K6
d𝑡𝑡d𝑡𝑡
++ 𝑛𝑛 W𝑛𝑛𝑝𝑝W#𝑝𝑝=
# =
[ [6𝑝𝑝#6𝑝𝑝 d𝑡𝑡
# d𝑡𝑡
Pr[𝐾𝐾
Pr[𝐾𝐾 # =# = 𝑘𝑘]𝑘𝑘]= = Q 𝑝𝑝Q#𝑝𝑝∙#𝑞𝑞∙#KQ
𝑞𝑞#KQ == Q|𝑞𝑞Q|
#𝑞𝑞
# / / / / 𝑓𝑓# (𝑡𝑡)
𝑓𝑓# (𝑡𝑡)
= =6𝑝𝑝#6𝑝𝑝∙#𝜇𝜇∙#K6 𝜇𝜇#K6 == 𝑞𝑞#𝑞𝑞 #
Wdu
Wdu W W
Life Table
Life Table Constant Force of Mortality (0
Constant Force of Mortality (0 ≤≤𝑡𝑡 <
𝑡𝑡 <
1) 1)
𝑙𝑙#K6𝑙𝑙#K6 𝑒𝑒#:W|
𝑒𝑒#:W|
== ss
𝑘𝑘 ⋅𝑘𝑘Q|⋅𝑞𝑞Q|#𝑞𝑞+
# +
𝑛𝑛 ⋅𝑛𝑛 W⋅𝑝𝑝W#𝑝𝑝=
# =
ssQ 𝑝𝑝Q#𝑝𝑝 #
Use exponential interpolation:
Use exponential interpolation:
6 𝑝𝑝#
6 𝑝𝑝=
# = Qt/
Qt/ Qtu
Qtu ud6ud6 6 6
𝑙𝑙# 𝑙𝑙# 𝑙𝑙#K6𝑙𝑙#K6 ==(𝑙𝑙#(𝑙𝑙)# ) ⋅ (𝑙𝑙⋅ #Ku
(𝑙𝑙#Ku
) )
∘ ∘
(𝑝𝑝(𝑝𝑝 6 6
6 𝑑𝑑6#𝑑𝑑 # 𝑙𝑙# 𝑙𝑙− # − 𝑙𝑙#K6
𝑙𝑙#K6 Relationship between 𝑒𝑒
Relationship between 𝑒𝑒
# and 𝑒𝑒
# and 𝑒𝑒
# # 6 𝑝𝑝#6 𝑝𝑝=
# = # )# )
6 𝑞𝑞=
6 𝑞𝑞# # = == ∘ ∘
𝜇𝜇#K6𝜇𝜇#K6 == −− ln(𝑝𝑝
ln(𝑝𝑝
# ) # )
𝑙𝑙# 𝑙𝑙# 𝑙𝑙# 𝑙𝑙# 𝑒𝑒#𝑒𝑒≈
# ≈
𝑒𝑒#𝑒𝑒+
# +
0.5
0.5

𝑓𝑓# (𝑡𝑡)
𝑓𝑓# (𝑡𝑡)
= =6𝑝𝑝#6𝑝𝑝∙#𝜇𝜇∙#K6
𝜇𝜇#K6 == 𝑒𝑒 d{⋅6
𝑒𝑒 d{⋅6
⋅ 𝜇𝜇 ⋅ 𝜇𝜇
6 𝑑𝑑6#K:
𝑑𝑑#K: 𝑙𝑙#K: 𝑙𝑙#K:−− 𝑙𝑙#K:K6
𝑙𝑙#K:K6 Recursive Formulas
Recursive Formulas
𝑞𝑞#𝑞𝑞=
:|6:|6 # = ==
𝑙𝑙# 𝑙𝑙# 𝑙𝑙# 𝑙𝑙# ∘ ∘
𝑒𝑒#𝑒𝑒=
∘ ∘ ∘ ∘
Select & Ultimate Mortality
Select & Ultimate Mortality
# =
𝑒𝑒#:W|
𝑒𝑒#:W|
+ +W𝑝𝑝W#𝑝𝑝⋅#𝑒𝑒⋅#KW
𝑒𝑒#KW


The age at which a person is selected is denoted as
The age at which a person is selected is denoted as
Force of Mortality
Force of Mortality 𝑒𝑒#𝑒𝑒=
# =
𝑒𝑒#:W|
𝑒𝑒#:W|
+ +W𝑝𝑝W#𝑝𝑝⋅#𝑒𝑒⋅#KW
𝑒𝑒#KW
[𝑥𝑥].
[𝑥𝑥].
𝑑𝑑 𝑑𝑑 𝑑𝑑 𝑑𝑑 ∘ ∘ ∘ ∘ ∘ ∘
𝑓𝑓# (𝑡𝑡) d𝑡𝑡d𝑡𝑡6𝑝𝑝#6𝑝𝑝# d𝑡𝑡d𝑡𝑡
𝑓𝑓# (𝑡𝑡) 𝑙𝑙#K6𝑙𝑙#K6 𝑒𝑒#:zKW|
𝑒𝑒#:zKW| == 𝑒𝑒#:z|
𝑒𝑒#:z|
++ z𝑝𝑝z#𝑝𝑝⋅#𝑒𝑒⋅#Kz:W|
𝑒𝑒#Kz:W|

𝜇𝜇#K6
𝜇𝜇#K6 == == −− == −− Select mortality is written as 𝑞𝑞
Select mortality is written as 𝑞𝑞
[#]K6 where 𝑥𝑥 is the
[#]K6 where 𝑥𝑥 is the
(𝑡𝑡)
𝑆𝑆#𝑆𝑆# (𝑡𝑡) 6 𝑝𝑝#
6 𝑝𝑝# 𝑙𝑙#K6𝑙𝑙#K6 𝑒𝑒#:zKW|
𝑒𝑒#:zKW|
== 𝑒𝑒#:z|
𝑒𝑒#:z|
++ z𝑝𝑝z#𝑝𝑝⋅#𝑒𝑒⋅#Kz:W|
𝑒𝑒#Kz:W|
selected age and 𝑡𝑡 is the number of years after
selected age and 𝑡𝑡 is the number of years after
𝑓𝑓# (𝑡𝑡)
𝑓𝑓# (𝑡𝑡)
== 𝑆𝑆#𝑆𝑆(𝑡𝑡)
# (𝑡𝑡)
⋅ 𝜇𝜇⋅#K6
𝜇𝜇#K6
== 6 𝑝𝑝#6𝑝𝑝⋅#𝜇𝜇⋅#K6
𝜇𝜇#K6 𝑒𝑒#𝑒𝑒= 𝑝𝑝#𝑝𝑝(1
# = # (1
++ ) )
𝑒𝑒#Ku
𝑒𝑒#Ku selection.
selection.

Finding
Finding Special Mortality Laws
Special Mortality Laws The mortality after the select period is called the
The mortality after the select period is called the
W𝑝𝑝W#𝑝𝑝
Using Force of Mortality
# Using Force of Mortality
W W Constant Force of Mortality
Constant Force of Mortality ultimate mortality, where:
ultimate mortality, where:
W𝑝𝑝W#𝑝𝑝=
# =
exp
exp
Z−Z−
[ [𝜇𝜇#K6
𝜇𝜇#K6
d𝑡𝑡 d𝑡𝑡
\ \ 𝜇𝜇#𝜇𝜇=# =𝜇𝜇 𝜇𝜇 𝑞𝑞[#]K6 ==
𝑞𝑞[#]K6 𝑞𝑞#K6
𝑞𝑞#K6

/ /
#KW#KW 𝑝𝑝
6 # 6 # =
𝑝𝑝= 𝑒𝑒 d{6
𝑒𝑒 d{6

Common Approach
Common Approach
== exp
exp Z−Z−
[ [ 𝜇𝜇]𝜇𝜇 d𝑦𝑦 ∘ ∘ 1 1
] d𝑦𝑦
\ \
𝑒𝑒#𝑒𝑒= Read from the left to the right and then
Read from the left to the right and then
# # # =

𝜇𝜇 𝜇𝜇 continue downwards.
continue downwards.
Properties of Force of Mortality
Properties of Force of Mortality ∘ ∘ 11
𝑒𝑒#:W|
𝑒𝑒#:W|= = (1(1 −− 𝑒𝑒 d{⋅W
𝑒𝑒 d{⋅W
) )
• • 𝜇𝜇#K6
𝜇𝜇#K6
≥≥ 0 0 𝜇𝜇 𝜇𝜇 𝑥𝑥 𝑥𝑥 𝑞𝑞[#]
𝑞𝑞[#]
𝑞𝑞[#]Ku
𝑞𝑞[#]Ku
𝑞𝑞[#]Kl
𝑞𝑞[#]Kl
𝑞𝑞#Kà
𝑞𝑞#Kà

∞∞
• •∫/ ∫/𝜇𝜇#K6
𝜇𝜇#K6
d𝑡𝑡 d𝑡𝑡
== ∞ ∞ Uniform Distribution
Uniform Distribution 30 30

Adding/Multiplying a Constant
Adding/Multiplying a Constant 𝑙𝑙# 𝑙𝑙=
# = 𝑘𝑘(𝜔𝜔
𝑘𝑘(𝜔𝜔−− 𝑥𝑥)
𝑥𝑥) 31 31
∗ ∗ 11 11
• • 𝜇𝜇#K6
𝜇𝜇#K6
==𝜇𝜇#K6
𝜇𝜇#K6
++ 𝑘𝑘 ⇒ ∗
𝑘𝑘 ⇒W 𝑝𝑝W#∗𝑝𝑝=
# =
dQWdQW
W𝑝𝑝W#𝑝𝑝∙#𝑒𝑒∙ 𝑒𝑒 𝜇𝜇#𝜇𝜇= # = ⇒
⇒
𝜇𝜇 #K6
𝜇𝜇#K6
== 32 32
∗ ∗ Q Q 𝜔𝜔 𝜔𝜔
−− 𝑥𝑥 𝑥𝑥 𝜔𝜔 𝜔𝜔 (𝑥𝑥(𝑥𝑥
−− ++ 𝑡𝑡) 𝑡𝑡)
• •𝜇𝜇#K6
𝜇𝜇#K6
==𝑘𝑘 ∙𝑘𝑘𝜇𝜇∙#K6
𝜇𝜇#K6 ∗ e e f f
⇒⇒W 𝑝𝑝W#∗𝑝𝑝=
# = W𝑝𝑝W#𝑝𝑝# 𝑙𝑙#K6
𝑙𝑙#K6 𝜔𝜔 𝜔𝜔 −− (𝑥𝑥(𝑥𝑥
++ 𝑡𝑡) 𝑡𝑡) 33 33
6 𝑝𝑝=
6 𝑝𝑝# # = ==
𝑙𝑙# 𝑙𝑙# 𝜔𝜔 𝜔𝜔 −− 𝑥𝑥 𝑥𝑥

www.coachingactuaries.com Copyright © 2018 Coaching Actuaries. All Rights Reserved. 1

INSURANCE INSURANCE ANNUITIES ANNUITIES

Type of Constant Force Uniform Distribution Type of

𝐄𝐄𝐄𝐄𝐄𝐄 𝜇𝜇 𝐄𝐄𝐄𝐄𝐄𝐄
Insurance 𝑎𝑎êîd#| Annuities
𝐴𝐴̅# = 𝐴𝐴̅# =
Discrete 𝜇𝜇 + 𝛿𝛿 𝜔𝜔 − 𝑥𝑥 Due; Discrete
j j
QKu 𝜇𝜇 𝑎𝑎êW| 𝑎𝑎̈ # = s 𝑣𝑣 Q ⋅ Q 𝑝𝑝#
𝐴𝐴# = s 𝑣𝑣 ⋅ Q|𝑞𝑞# ̅
𝐴𝐴 u = e1 − W𝐸𝐸# f 𝐴𝐴̅ u =
#:W| 𝜇𝜇 + 𝛿𝛿 #:W|
Qt/ 𝜔𝜔 − 𝑥𝑥 Qt/
Whole Life
Continuous 𝜔𝜔 − (𝑥𝑥 + 𝑛𝑛) Immediate; Discrete
Whole Life
j W𝐸𝐸# = 𝑒𝑒 d({Kï)W W𝐸𝐸# = 𝑣𝑣 W ⋅ 𝑎𝑎# = 𝑎𝑎̈ # − 1
𝐴𝐴̅# = [ 𝑣𝑣 6 ⋅ 6𝑝𝑝# 𝜇𝜇#K6 d𝑡𝑡 𝜔𝜔 − 𝑥𝑥
/

Continuous
Calculate l𝐴𝐴 and l𝐴𝐴̅ similarly to 𝐴𝐴 and 𝐴𝐴̅, but with j
Discrete
double the force of interest, 𝛿𝛿. Equivalently, 𝑎𝑎ê# = [ 𝑣𝑣 6 ⋅ 6𝑝𝑝# d𝑡𝑡
𝐴𝐴u#:W| = 𝐴𝐴# − W𝐸𝐸# ⋅ 𝐴𝐴#KW replace 𝑣𝑣 with 𝑣𝑣 l , or replace 𝑖𝑖 with 2𝑖𝑖 + 𝑖𝑖 l . /

Term Life Due; Discrete

Continuous Variances 𝑎𝑎̈ #:W|
êêê = 𝑎𝑎̈ # − W𝐸𝐸# ⋅ 𝑎𝑎̈ #KW

̅
𝐴𝐴 u = 𝐴𝐴̅# − W𝐸𝐸# ⋅ 𝐴𝐴̅ #KW
#∶W|
Discrete 𝑉𝑉𝑉𝑉𝑉𝑉[𝑍𝑍] Temporary Immediate; Discrete
Discrete Life 𝑎𝑎#:W|
êêê = 𝑎𝑎̈ #:W|
êêê − 1 + W 𝐸𝐸#
l
u Whole Life 𝐴𝐴# − (𝐴𝐴# )l
Deferred W|𝐴𝐴# = 𝐴𝐴# − 𝐴𝐴#:W| = W𝐸𝐸# ⋅ 𝐴𝐴#KW Continuous
Whole Life Continuous Endowment Insurance l l 𝑎𝑎ê#:W|
êêê = 𝑎𝑎
ê# − W𝐸𝐸# ⋅ 𝑎𝑎ê#KW
êêê − e𝐴𝐴#:W|
𝐴𝐴#:W| êêêf
W|𝐴𝐴 ̅ # = 𝐴𝐴 ̅ #
̅
− 𝐴𝐴 u = W𝐸𝐸# ⋅ 𝐴𝐴̅ #KW
Due; Discrete
#∶W|
Discrete Replace A with 𝐴𝐴̅ for continuous cases. Deferred W|𝑎𝑎̈ # = 𝑎𝑎̈ # − 𝑎𝑎̈ #:W|
êêê = W𝐸𝐸# ⋅ 𝑎𝑎̈ #KW

𝐴𝐴 u = W𝐸𝐸# = 𝑣𝑣 W W𝑝𝑝# Recursive Formulas Whole Life Continuous

Pure #:W|
𝐴𝐴# = 𝑣𝑣𝑞𝑞# + 𝑣𝑣𝑝𝑝# ⋅ 𝐴𝐴#Ku
Endowment W|𝑎𝑎
ê# = 𝑎𝑎ê# − 𝑎𝑎ê#:W|
êêê = W𝐸𝐸# ⋅ 𝑎𝑎
ê#KW
Continuous 𝐴𝐴u#:W| u
êêê = 𝑣𝑣𝑞𝑞# + 𝑣𝑣𝑝𝑝# ⋅ 𝐴𝐴#Ku:Wdu|
êêêêêêê
N/A Certain-
(𝐼𝐼𝐼𝐼)# = 𝑣𝑣𝑞𝑞# + 𝑣𝑣𝑝𝑝# (𝐴𝐴#Ku + (𝐼𝐼𝐼𝐼)#Ku ) 𝑎𝑎êêêêêêê
êêê = 𝑎𝑎
#:W| êêêê
W| + W|𝑎𝑎
ê#
and-Life
Discrete
j
1/mthly Insurance
𝐴𝐴
#:W|
= 𝐴𝐴u#:W| + W𝐸𝐸# (z) 1 (𝐼𝐼𝑎𝑎̈ ) # = s(𝑘𝑘 + 1)𝑣𝑣 Q ⋅ Q 𝑝𝑝#
Endowment 𝐾𝐾# = ⌊𝑚𝑚𝑇𝑇# ⌋
Insurance Continuous 𝑚𝑚 Qt/
(z) j
Prô𝐾𝐾# = 𝑟𝑟ö = õ𝑝𝑝# ⋅ u 𝑞𝑞#Kõ = u 𝑞𝑞# ̅ ê) #
̅
𝐴𝐴 ̅
= 𝐴𝐴 u + W𝐸𝐸# z
õ |
z
(𝐼𝐼 𝑎𝑎 = [ 𝑡𝑡𝑡𝑡 6 ⋅ 6𝑝𝑝# d𝑡𝑡
#:W| #:W| j /

𝐴𝐴#
(z)
= s 𝑣𝑣 (QKu)/z ⋅ Q (𝐼𝐼𝐼𝐼) #:W|
êêê + (𝐷𝐷𝐷𝐷)#:W|
êêê
j u 𝑞𝑞# Varying
|
z z
(𝐼𝐼𝐼𝐼)# = s(𝑘𝑘 + 1)𝑣𝑣 QKu ⋅ Q|𝑞𝑞#
Qt/
Annuities = (𝑛𝑛 + 1)𝑎𝑎#:W|
êêê
(𝒎𝒎) W
Qt/ í 𝒙𝒙
Relationship between 𝑨𝑨𝒙𝒙 , 𝑨𝑨𝒙𝒙 , and 𝑨𝑨 ̅ ê) #:W|
(𝐼𝐼 𝑎𝑎 6
j êêê = [ 𝑡𝑡𝑡𝑡 ⋅ 6 𝑝𝑝# d𝑡𝑡
̅ ̅)# = [ 𝑡𝑡𝑡𝑡 6 ⋅ 6𝑝𝑝# 𝜇𝜇#K6 d𝑡𝑡
(𝐼𝐼 𝐴𝐴 (Under UDD Assumption) /
W
/ 𝑖𝑖
W
𝐴𝐴̅# = 𝐴𝐴# í 𝑎𝑎ê) #:W|
(𝐷𝐷 êêê = [ (𝑛𝑛 − 𝑡𝑡)𝑣𝑣 6 6𝑝𝑝# d𝑡𝑡
̅ ̅) u
(𝐼𝐼 𝐴𝐴 6
êêê = [ 𝑡𝑡𝑡𝑡 ⋅ 6 𝑝𝑝# 𝜇𝜇#K6 d𝑡𝑡
#:W| 𝛿𝛿 /
Varying / 𝑖𝑖 u ̅ ê) #:W| í 𝑎𝑎ê) #:W|
𝐴𝐴̅ u êêê = 𝐴𝐴 #:W| êêê
(𝐼𝐼 𝑎𝑎 êêê + (𝐷𝐷 êêê = 𝑛𝑛𝑎𝑎
ê#:W|
êêê
Insurance í 𝐴𝐴̅) #:W|
(𝐷𝐷 u
êêê #:W|
𝛿𝛿
W 𝑖𝑖 Uniform
= [ (𝑛𝑛 − 𝑡𝑡)𝑣𝑣 6 ⋅ 6𝑝𝑝# 𝜇𝜇#K6 d𝑡𝑡 W|𝐴𝐴# =
̅ 𝐴𝐴 Constant Force
𝛿𝛿 W| # Distribution
/ 𝑖𝑖 u 1
(𝐼𝐼𝐼𝐼) u u
êêê + (𝐷𝐷𝐷𝐷) #:W|
êêê = (𝑛𝑛 𝐴𝐴̅#:W|
êêê = 𝐴𝐴 #:W|
u
êêê + 𝐴𝐴 #:W|
êêê
Integrate directly,
#:W|
𝛿𝛿 𝑎𝑎ê# =
+ 1)𝐴𝐴u#:W| 𝜇𝜇 + 𝛿𝛿 or use 𝐴𝐴̅ # = 1 − 𝛿𝛿𝑎𝑎ê#
(z) 𝑖𝑖
̅ ̅) u í 𝐴𝐴̅) u 𝐴𝐴# = (z) 𝐴𝐴#
êêê = 𝑛𝑛 ⋅ 𝐴𝐴̅ #:W| 1
u Integrate directly, or
(𝐼𝐼 𝐴𝐴 êêê + (𝐷𝐷 êêê 𝑖𝑖
#:W| #:W| 𝑎𝑎ê#:W|
êêê = e1 − W𝐸𝐸# f use 𝐴𝐴
̅ = 1 − 𝛿𝛿𝑎𝑎ê#:W|
êêê

Percentiles 𝜇𝜇 + 𝛿𝛿 #:W|

The 100𝑝𝑝th percentile of Z is the value 𝑧𝑧° such that: Variances
Prô𝑍𝑍 ≤ 𝑧𝑧° ö = 𝑝𝑝

To calculate 𝑧𝑧° : Discrete 𝑉𝑉𝑉𝑉𝑉𝑉[𝑌𝑌]

1. Draw a graph with Z on y-axis and 𝑇𝑇# on x-axis. l
𝐴𝐴# − (𝐴𝐴# )l
2. Identify the parts of the curve where 𝑍𝑍 ≤ 𝑧𝑧° . Whole Life
𝑑𝑑l
Determine the value of 𝑇𝑇# that corresponds to l
l
those parts. êêê − e𝐴𝐴#:W|
𝐴𝐴#:W| êêê f
Temporary Life
3. Use the value of 𝑇𝑇# from Step 2 to calculate 𝑧𝑧° . 𝑑𝑑l

Replace A with 𝐴𝐴̅ and 𝑑𝑑 with 𝛿𝛿 for continuous

cases.

www.coachingactuaries.com Copyright © 2018 Coaching Actuaries. All Rights Reserved. 2

Recursive Formula PREMIUMS PREMIUMS RESERVES RESERVES

𝑎𝑎̈ # = 1 + 𝑣𝑣𝑝𝑝# ⋅ 𝑎𝑎̈ #Ku
Net Future Loss Net Premium Reserve
Relationship between Insurances and Annuities /𝐿𝐿 = 𝑃𝑃𝑃𝑃(f. benefits) − 𝑃𝑃𝑃𝑃(f. premiums) Prospective Method

Discrete 6 𝑉𝑉 = 𝐸𝐸𝐸𝐸𝑉𝑉6 (f. benefits) − 𝐸𝐸𝐸𝐸𝑉𝑉6 (f. premiums)
Fully Discrete Fully Continuous

𝐴𝐴# = 1 − 𝑑𝑑𝑎𝑎̈ # Special Formulas

™´ Ku Æ´
Whole Life l /𝐿𝐿 = 𝑣𝑣 − 𝑃𝑃𝑎𝑎̈ êêêêêêêêê
™´ Ku| /𝐿𝐿 = 𝑣𝑣 − 𝑃𝑃𝑎𝑎êêêêê
Æ´ | For a fully discrete whole life insurance policy:
𝐴𝐴 # = 1 − (2𝑑𝑑 − 𝑑𝑑l ) l𝑎𝑎̈ #
𝑃𝑃 ™ Ku 𝑃𝑃 𝑃𝑃 Æ 𝑃𝑃 𝑎𝑎̈ #K6
𝐴𝐴#:W| = ¨1 + ≠ 𝑣𝑣 ´ − = ¨1 + ≠ 𝑣𝑣 ´ − 6 𝑉𝑉 = 1 −
Temporary êêê = 1 − 𝑑𝑑𝑎𝑎̈ #:W|
êêê 𝑑𝑑 𝑑𝑑 𝛿𝛿 𝛿𝛿 𝑎𝑎̈ #
l l l 𝐴𝐴#K6 − 𝐴𝐴#
Life êêê = 1 − (2𝑑𝑑 − 𝑑𝑑 ) 𝑎𝑎
𝐴𝐴 #:W| ê#:W|
êêê

Fully Discrete 6 𝑉𝑉 =
1 − 𝐴𝐴#
Continuous 𝐸𝐸ô /𝐿𝐿 ö = 𝐴𝐴# − 𝑃𝑃𝑎𝑎̈ #

Gross Premium Reserve

𝐴𝐴#̅ = 1 − 𝛿𝛿𝑎𝑎ê# 𝑃𝑃 l Prospective Method
Whole Life l 𝑉𝑉𝑉𝑉𝑉𝑉ô /𝐿𝐿 ö = ¨1 + ≠ ô l𝐴𝐴# − (𝐴𝐴# )l ö
𝐴𝐴 # = 1 − (2𝛿𝛿) l𝑎𝑎ê# ≤
𝑑𝑑 6 𝑉𝑉 = 𝐸𝐸𝐸𝐸𝑉𝑉6 (f. ben.) + 𝐸𝐸𝐸𝐸𝑉𝑉6 (f. exp.) − 𝐸𝐸𝐸𝐸𝑉𝑉6 (f. pre.)
𝐴𝐴̅ #:W|

Temporary êêê = 1 − 𝛿𝛿𝑎𝑎
ê#:W|
êêê Fully Continuous
Expense Reserve
Life l
𝐴𝐴̅ #:W|
êêê = 1 − (2𝛿𝛿) 𝑎𝑎
l
ê#:W|
êêê 𝐸𝐸ô /𝐿𝐿 ö = 𝐴𝐴̅# − 𝑃𝑃𝑎𝑎ê# Expense Premium = Gross Premium − Net Premium
º ≤

𝑃𝑃 l 6 𝑉𝑉 = 6𝑉𝑉 − 6𝑉𝑉 W
Annuities with mthly Payments 𝑉𝑉𝑉𝑉𝑉𝑉ô /𝐿𝐿 ö = ¨1 + ≠ ô l𝐴𝐴#̅ − (𝐴𝐴̅# )l ö º
(z) (z) 𝛿𝛿 6 𝑉𝑉 = 𝐸𝐸𝐸𝐸𝑉𝑉6 (f. exp.) − 𝐸𝐸𝐸𝐸𝑉𝑉6 (f. exp. premium)
𝐴𝐴# = 1 − 𝑑𝑑(z) 𝑎𝑎̈ #
UDD Assumption Equivalence Principle Recursive Formula
(z)
𝑎𝑎̈ # = 𝛼𝛼(𝑚𝑚) ⋅ 𝑎𝑎̈ # − 𝛽𝛽(𝑚𝑚) • 𝐸𝐸ô /𝐿𝐿 ö = 0 • Net premium reserve
(z) ⇒ 𝐸𝐸𝐸𝐸𝐸𝐸(f. premiums) = 𝐸𝐸𝐸𝐸𝐸𝐸(f. benefits) e 6𝑉𝑉 + 𝑃𝑃f(1 + 𝑖𝑖) = 𝑞𝑞#K6 ⋅ 𝑏𝑏 + 𝑝𝑝#K6 ⋅ 6Ku𝑉𝑉
êêê = 𝛼𝛼(𝑚𝑚) ⋅ 𝑎𝑎̈ #:W|
𝑎𝑎̈ #:W| êêê − 𝛽𝛽(𝑚𝑚)(1 − W 𝐸𝐸# )
𝐸𝐸𝐸𝐸𝐸𝐸(benefit) • Gross premium reserve
(z)
= 𝛼𝛼(𝑚𝑚) ⋅ W|𝑎𝑎̈ # − 𝛽𝛽(𝑚𝑚) ⋅ W𝐸𝐸# ⇒ Net Premium =
W|𝑎𝑎̈ # 𝐸𝐸𝐸𝐸𝐸𝐸(annuity) e 6𝑉𝑉 ≤ + 𝐺𝐺 − 𝑒𝑒f(1 + 𝑖𝑖) = 𝑞𝑞#K6 ⋅ (𝑏𝑏 + 𝐸𝐸)
𝑖𝑖𝑖𝑖 • Variance ≤
+ 𝑝𝑝#K6 ⋅ 6Ku𝑉𝑉
𝛼𝛼(𝑚𝑚) = (z) (z)
𝑖𝑖 𝑑𝑑
l
𝐴𝐴# − (𝐴𝐴# )l Thiele’s Differential Equation
𝑖𝑖 − 𝑖𝑖 (z)
𝛽𝛽(𝑚𝑚) = (z) (z) Discrete l
𝑉𝑉𝑉𝑉𝑉𝑉ô /𝐿𝐿 ö = 𝑏𝑏 ⋅ d
𝑖𝑖 𝑑𝑑 (1 − 𝐴𝐴# )l 6 𝑉𝑉 = 𝛿𝛿6 ⋅ 6 𝑉𝑉 + 𝐺𝐺6 − 𝑒𝑒6 − e𝑏𝑏6 + 𝐸𝐸6 − 6 𝑉𝑉 f𝜇𝜇#K6
z dz d𝑡𝑡
𝑖𝑖 (z) 𝑑𝑑 (z)
Z1 + \ = Z1 − \ = (1 + 𝑖𝑖) = (1 − 𝑑𝑑)du l
𝐴𝐴̅ # − (𝐴𝐴̅# )l Euler’s Method
𝑚𝑚 𝑚𝑚 Continuous 𝑉𝑉𝑉𝑉𝑉𝑉ô /𝐿𝐿 ö = 𝑏𝑏 l ⋅
(1 − 𝐴𝐴̅# )l • Forward Euler Approximation:
Note:
d 6KΩ𝑉𝑉 − 6 𝑉𝑉
• The formulas for 𝛼𝛼(𝑚𝑚) and 𝛽𝛽(𝑚𝑚) are provided on í to 𝐴𝐴′𝑠𝑠 and 𝑎𝑎′𝑠𝑠 for endowment insurance. 𝑉𝑉 =
Add 𝑛𝑛| d𝑡𝑡 6 ℎ
the LTAM Table.
• Backward Euler Approximation:
Gross Premium
• The values for 𝛼𝛼(𝑚𝑚) and 𝛽𝛽(𝑚𝑚) when 𝑖𝑖 = 0.05 are ≤ d 6 𝑉𝑉 − 6dΩ𝑉𝑉
/𝐿𝐿 = 𝑃𝑃𝑃𝑃(f. benefits) + 𝑃𝑃𝑃𝑃(f. expenses) 6 𝑉𝑉 =
also given in the LTAM Table. d𝑡𝑡 ℎ

− 𝑃𝑃𝑃𝑃(f. gross premiums)

Woolhouse’s Formula (3 terms) Equivalence Principle For net premium reserve, drop expense-related
𝑚𝑚 − 1 𝑚𝑚 l − 1 • 𝐸𝐸ô /𝐿𝐿≤ ö = 0 terms and replace 𝐺𝐺6 with net premium.
(z)
𝑎𝑎̈ # ≈ 𝑎𝑎̈ # − − (𝜇𝜇# + 𝛿𝛿) ⇒ 𝐸𝐸𝐸𝐸𝐸𝐸(f. gross premiums)

2𝑚𝑚 12𝑚𝑚 l Interim Reserves (𝟎𝟎 ≤ 𝒔𝒔 < 𝟏𝟏)

If the question asks to use the Woolhouse’s = 𝐸𝐸𝐸𝐸𝐸𝐸(f. benefits) + 𝐸𝐸𝐸𝐸𝐸𝐸(f. expenses)
• Exact value:
formula with two terms, just drop the last term. Variance e 6𝑉𝑉 + 𝑃𝑃f(1 + 𝑖𝑖) g − g𝑞𝑞#K6 ⋅ 𝑏𝑏 ⋅ 𝑣𝑣 udg
For a fully discrete whole life policy: 6Kg𝑉𝑉 =
If 𝜇𝜇# is not available, approximate 𝜇𝜇# as: g 𝑝𝑝#K6
1 𝐺𝐺 − 𝑒𝑒õ ™ Ku 𝐺𝐺 − 𝑒𝑒õ
≤
/𝐿𝐿 = ¨𝑏𝑏 + 𝐸𝐸 + ≠ 𝑣𝑣 ´ − • Linear approximation:
𝜇𝜇# ≈ − (ln 𝑝𝑝#du + ln 𝑝𝑝# ) 𝑑𝑑 𝑑𝑑
2 6Kg𝑉𝑉 = e 6𝑉𝑉 + 𝑃𝑃f(1 − 𝑠𝑠) + 𝑠𝑠 ⋅ 6Ku𝑉𝑉
𝑚𝑚 − 1 (𝑒𝑒
+ ¥ − 𝑒𝑒õ )
(z)
êêê ≈ 𝑎𝑎̈ #:W|
𝑎𝑎̈ #:W| êêê − e1 − W𝐸𝐸# f 𝐺𝐺 − 𝑒𝑒õ l l Modified Reserve
2𝑚𝑚 ≤ö
𝑉𝑉𝑉𝑉𝑉𝑉ô /𝐿𝐿 = ¨𝑏𝑏 + 𝐸𝐸 + ≠ ô 𝐴𝐴# − (𝐴𝐴# )l ö
𝑚𝑚 l − 1 𝑑𝑑 Full preliminary term (FPT): The policy is treated
− ô𝜇𝜇 + 𝛿𝛿 − W𝐸𝐸# (𝜇𝜇#KW + 𝛿𝛿)ö 1. Replace 𝐴𝐴 and d with their continuous as if it were issued one year later, with the first
12𝑚𝑚 l # counterparts for fully continuous policies. year of the policy being treated as if it were a one-
If the interest rate is 0: í to 𝐴𝐴′𝑠𝑠 for endowment insurance. year term insurance.
2. Add 𝑛𝑛|
∘ 1 1
𝑒𝑒# ≈ 𝑒𝑒# + − 𝜇𝜇

• FPT net premium
2 12 # Portfolio Percentile Premium
1st year modified net premium = 𝐴𝐴u#:u|í
𝑆𝑆 = 𝐿𝐿u + 𝐿𝐿l + ⋯ + 𝐿𝐿∂
Percentiles 𝐴𝐴#Ku
𝐸𝐸[𝑆𝑆] = 𝑁𝑁 ⋅ 𝐸𝐸[𝐿𝐿] Renewal modified net premium =
The 100pth percentile of Y is the value 𝑦𝑦° such as 𝑉𝑉𝑉𝑉𝑉𝑉[𝑆𝑆] = 𝑁𝑁 ⋅ 𝑉𝑉𝑉𝑉𝑉𝑉[𝐿𝐿] 𝑎𝑎̈ #Ku
that:
• FPT reserve
Using the portfolio percentile premium principle, ¬√Æ
Prô𝑌𝑌 ≤ 𝑦𝑦° ö = 𝑝𝑝 u𝑉𝑉# = 0
the premium is set such that there is a specified ¬√Æ
To calculate 𝑧𝑧° : 𝑉𝑉
6 # = 6du𝑉𝑉#Ku
probability (x%) that the total loss is negative:
1. Draw a graph with Y on y-axis and 𝑇𝑇# on x-axis.
Pr[𝑆𝑆 < 0] = 𝑥𝑥%
2. Identify the parts of the curve where 𝑌𝑌 ≤ 𝑦𝑦° .
Determine the value of 𝑇𝑇# that corresponds to Percentile of 𝟎𝟎𝑳𝑳
those parts. The 100pth percentile of /𝐿𝐿 is the value 𝜋𝜋° such
3. Use the value of 𝑇𝑇# from Step 2 to calculate 𝑦𝑦° . that Prô /𝐿𝐿 ≤ 𝜋𝜋° ö = 𝑝𝑝. To determine 𝜋𝜋° :
1. Graph /𝐿𝐿 on y-axis and 𝑇𝑇# on x-axis.
2. Identify the parts of the curve where /𝐿𝐿 ≤ 𝜋𝜋° .
Determine the value of 𝑇𝑇# that corresponds to
those parts.
3. Use the value of 𝑇𝑇# from Step 2 to calculate 𝜋𝜋° .

www.coachingactuaries.com Copyright © 2018 Coaching Actuaries. All Rights Reserved. 3

MARKOV CHAINS
MARKOV CHAINS MULTIPLE DECREMENT MODELS
MULTIPLE DECREMENT MODELS Fractional Ages (0 ≤ 𝑠𝑠 < 1)

¥ƒ

• UDD in the multiple decrement table:
6 𝑝𝑝# : probability someone in state 𝑖𝑖 at age 𝑥𝑥 is in Multiple Decrement Tables (MDT) (ƒ) (ƒ)
state 𝑗𝑗 (where 𝑗𝑗 may equal 𝑖𝑖) at age 𝑥𝑥 + 𝑡𝑡 Decrements are dependent on each other. g 𝑞𝑞# = 𝑠𝑠 ⋅ 𝑞𝑞#
(Œ) (Œ)
¥¥ Discrete Probabilities g 𝑞𝑞# = 𝑠𝑠 ⋅ 𝑞𝑞#
6 𝑝𝑝# : probability someone in state 𝑖𝑖 at age 𝑥𝑥
remains in state 𝑖𝑖 until age 𝑥𝑥 + 𝑡𝑡 6du 6du • Constant forces of decrement:
(ƒ) (ƒ) (ƒ)
(ƒ) (Œ) (ƒ) (ƒ) 𝜇𝜇# 𝑞𝑞# g 𝑞𝑞#
6 𝑞𝑞# = s Q 𝑝𝑝# 𝑞𝑞#KQ = s Q|𝑞𝑞#
Continuous Probabilities (Œ) = (Œ) = (Œ)
• Direct Approach
Qt/ Qt/ 𝜇𝜇# 𝑞𝑞# g 𝑞𝑞#
z (ƒ)
(Œ) (ƒ) (ƒ) 𝑞𝑞# (Œ) g
6 𝑞𝑞# = s 6 𝑞𝑞#
6
¥¥ ¥ƒ g 𝑞𝑞# = (Œ) “1 − e𝑝𝑝# f ”
6 𝑝𝑝# = exp ∆− [ s 𝜇𝜇#Kg d𝑠𝑠 » ƒtu 𝑞𝑞#
/
ƒ«¥ (Œ) (Œ)
6 6 𝑝𝑝# + 6𝑞𝑞# = 1 Associated Single Decrement Tables (ASDT)
¥∙
= exp Z− [ 𝜇𝜇#Kg d𝑠𝑠 \ 6K:du The associated single decrements are independent.
(ƒ) (Œ) (ƒ) (Œ) (ƒ) 6
6|: 𝑞𝑞# = s Q 𝑝𝑝# 𝑞𝑞#KQ = 6𝑝𝑝# :𝑞𝑞#K6
/ ‘(ƒ) (ƒ)
For permanent disability model: 6 𝑝𝑝# = exp Z− [ 𝜇𝜇#Kg d𝑠𝑠\
Qt6 /
6
¥ƒ ¥¥ ¥ƒ ƒƒ d
6 𝑝𝑝# =[ g 𝑝𝑝# ⋅ 𝜇𝜇#Kg ⋅ 6dg𝑝𝑝#Kg d𝑠𝑠 Life Table (ƒ)
⇒ 𝜇𝜇#K6 = −
‘(ƒ)
lne 6 𝑝𝑝# f
/
(ƒ) d𝑡𝑡
6 𝑑𝑑 #

(ƒ) d ‘(ƒ)
• Approximation 6 𝑞𝑞# = (Œ)
6
6 𝑞𝑞#
𝑙𝑙# ‘(ƒ) ‘(ƒ) (ƒ) (ƒ) d𝑡𝑡
Kolmogorov’s Forward Equations: 6 𝑞𝑞# =[ g 𝑝𝑝# 𝜇𝜇#Kg d𝑠𝑠 ⇒ 𝜇𝜇#K6 = ‘(ƒ)

W (Œ) (Œ) (Œ) / 6 𝑝𝑝#
d ¥ƒ ¥Q Qƒ ¥ƒ ƒQ (Œ) 6 𝑑𝑑 # 𝑙𝑙# − 𝑙𝑙#K6 ‘(ƒ) ‘(ƒ)
𝑝𝑝 = se 6 𝑝𝑝# ⋅ 𝜇𝜇#K6 − 6 𝑝𝑝# ⋅ 𝜇𝜇#K6 f 6 𝑞𝑞# = (Œ) = 6 𝑝𝑝# + 6𝑞𝑞# = 1
d𝑡𝑡 6 # 𝑙𝑙#
(Œ)
𝑙𝑙# z
Qt/
Q«ƒ (Œ) (Œ) ‘(ƒ)
(Œ) 𝑙𝑙#K6 6 𝑝𝑝# = ’ 6𝑝𝑝#
= Pr(Start in 𝑖𝑖, move into 𝑗𝑗) 6 𝑝𝑝# = (Œ) ƒtu
−Pr(Start in 𝑖𝑖, move out of 𝑗𝑗) 𝑙𝑙#
(ƒ) ‘(ƒ) (ƒ)
Key Relationships between 6 𝑞𝑞# and 6 𝑞𝑞#

Euler’s Method: (ƒ) :𝑑𝑑 #K6

¥ƒ ¥ƒ Qƒ 6|: 𝑞𝑞# = (Œ) No Assumption
1. 6KΩ𝑝𝑝# ≈ 6𝑝𝑝# + ℎ ∑WQt/e 6𝑝𝑝#¥Q ⋅ 𝜇𝜇#K6 − 𝑙𝑙#
Q«ƒ (Œ) ‘(ƒ)
¥ƒ ƒQ Continuous Probabilities 6 𝑝𝑝# = ∏z
ƒtu 6 𝑝𝑝#
6 𝑝𝑝# ⋅ 𝜇𝜇#K6 f (Œ) (ƒ)
¥ƒ d (ƒ) • ÷ 6 𝑝𝑝# = 1 − ∑z
ƒtu 6 𝑞𝑞#

6 𝑞𝑞#
ℎ𝜇𝜇# 𝑖𝑖 ≠ 𝑗𝑗 6
2.
¥ƒ
Ω 𝑝𝑝# ≈ à (ƒ) (Œ) (ƒ) (ƒ) d𝑡𝑡
6 𝑞𝑞# =[ g 𝑝𝑝# 𝜇𝜇#Kg d𝑠𝑠 ⇒ 𝜇𝜇#K6 = (Œ)
(Œ) (Œ)
1 − ℎ𝜇𝜇#¥∙ 𝑖𝑖 = 𝑗𝑗 6 𝑝𝑝# + 6 𝑞𝑞# = 1

/ 6 𝑝𝑝#
z d (fi)
Premiums (Œ) (ƒ) ⎧ 𝜇𝜇(ƒ) = d‹ ‹›´
For an insurance on (𝑥𝑥) currently in state i that 𝜇𝜇#K6 = s 𝜇𝜇#K6 ⎪ #K6 (fl)
‹ °´
pays $1 immediately upon every transition to state ƒtu • d ‡(fi)
6 ⎨ (ƒ) ›
d‹ ‹ ´
j: (Œ) (Œ) ⎪𝜇𝜇#K6 = °‡(fi)
j 6 𝑝𝑝# = exp Z− [ 𝜇𝜇#Kg d𝑠𝑠\ ⎩ ‹ ´
¥ƒ Qƒ /
𝐴𝐴#̅ = [ 𝑒𝑒 dï6 s 6𝑝𝑝#¥Q ⋅ 𝜇𝜇#K6 d𝑡𝑡 6 UDD in Multiple-Decrement Tables (UDDMDT)
(Œ) (Œ) (Œ)
6 𝑞𝑞# = [ g 𝑝𝑝# 𝜇𝜇#Kg d𝑠𝑠
/
j
Q«ƒ (0 ≤ 𝑠𝑠 < 1)
/ (fi)
= [ 𝑒𝑒 dï6 ⋅ Pr[start in 𝑖𝑖, move into 𝑗𝑗]d𝑡𝑡 d (Œ) d (Œ) ›´
6 𝑞𝑞# − 6 𝑝𝑝# ‘(ƒ) (Œ) (fl)
/
⇒
(Œ)
𝜇𝜇#K6 = d𝑡𝑡 (Œ) = d𝑡𝑡 (Œ) g 𝑝𝑝# = e g 𝑝𝑝# f›´
For an annuity on (𝑥𝑥) currently in state i that pays
6 𝑝𝑝# 6 𝑝𝑝#
$1 per year while the person is in state j: 6K: UDD in Associated Single Decrement Tables
j (ƒ) (Œ) (ƒ)
¥ƒ ¥ƒ 6|: 𝑞𝑞# =[ g 𝑝𝑝# 𝜇𝜇#Kg d𝑠𝑠
(UDDASDT)(0 ≤ 𝑠𝑠 < 1)
6 𝑝𝑝# d𝑡𝑡
dï6
𝑎𝑎ê# = [ 𝑒𝑒 6 For 2 decrements:
/
j ‘(l)
Insurance Applications (u) ‘(u) 𝑠𝑠 l𝑞𝑞#
g 𝑞𝑞# = 𝑞𝑞# Z𝑠𝑠 − \
¥ƒ ¥ƒ
𝑎𝑎̈ # = s 𝑣𝑣 Q Q 𝑝𝑝# Consider a whole life policy: 2

Qt/
For 3 decrements:

Discrete
Reserves (u) ‘(u)
‘(l) ‘(à) ‘(l) ‘(à)
𝑠𝑠 l e𝑞𝑞# + 𝑞𝑞# f 𝑠𝑠 à𝑞𝑞# ∙ 𝑞𝑞#
z j
• Direct Approach g 𝑞𝑞# = 𝑞𝑞# ·𝑠𝑠 − + „
𝐸𝐸𝐸𝐸𝐸𝐸[benefits] = s s 𝑣𝑣 QKu𝑏𝑏
(ƒ) (Œ) (ƒ) 2 3
Use prospective method. Q 𝑝𝑝# 𝑞𝑞#KQ
CF in MDT or ASDT (0 ≤ 𝑠𝑠 < 1)
• Approximation ƒtu Qt/
j (fi)
›´
Thiele’s Differential Equation: (Œ)
𝐸𝐸𝐸𝐸𝐸𝐸[annuity] = s 𝑣𝑣 Q Q 𝑝𝑝# =e
‘(ƒ) (Œ) › (fl)
g 𝑝𝑝# g 𝑝𝑝# f
d ´
(¥)
𝑉𝑉 (¥) = 𝛿𝛿6 6𝑉𝑉 (¥) − 𝐵𝐵6 Qt/
d𝑡𝑡 6
W Continuous
¥ƒ ¥ƒ
− s 𝜇𝜇#K6 e𝑏𝑏6 + 6𝑉𝑉 (ƒ) − 6𝑉𝑉 (¥)f j MULTIPLE LIVES
MULTIPLE LIVES
(Œ) (ƒ)
ƒt/ 𝐸𝐸𝐸𝐸𝐸𝐸[benefits] = [ 𝑣𝑣 6 6𝑝𝑝# 𝜇𝜇#K6 d𝑡𝑡

ƒ«¥ / Joint Life

(ƒ)
𝜇𝜇# 𝑇𝑇#] = mine𝑇𝑇# , 𝑇𝑇] f
Euler’s Method: 𝐸𝐸𝐸𝐸𝐸𝐸[benefits] = (Œ) (Under CF)
d 6 𝑉𝑉
(¥)
− 6dΩ𝑉𝑉
(¥) 𝛿𝛿 + 𝜇𝜇 6 𝑝𝑝#] + 6 𝑞𝑞#] = 1
(¥)
6 𝑉𝑉 = j
d𝑡𝑡 ℎ 𝐸𝐸𝐸𝐸𝐸𝐸[annuity] = [ 𝑣𝑣 6 (Œ)
6 𝑝𝑝# d𝑡𝑡 :|6 𝑞𝑞#] = :𝑝𝑝#] ⋅ 6 𝑞𝑞#K::]K:
/
= :𝑝𝑝#] − :K6𝑝𝑝#]

= :K6𝑞𝑞#] − : 𝑞𝑞#]
Independent Lives
6 𝑝𝑝#] = 6 𝑝𝑝# ⋅ 6 𝑝𝑝]

6 𝑞𝑞#] = 6 𝑞𝑞# + 6 𝑞𝑞] − 6 𝑞𝑞# ⋅ 6 𝑞𝑞]

𝜇𝜇#K6:]K6 = 𝜇𝜇#K6 + 𝜇𝜇]K6

www.coachingactuaries.com Copyright © 2018 Coaching Actuaries. All Rights Reserved. 4

Moments Contingent Probabilities Reserve Recursion for Policies with Multiple
∘
j 6 States
𝑒𝑒#] = [ 6 𝑝𝑝#] d𝑡𝑡 u
6 𝑞𝑞#] =[ g 𝑝𝑝#] ∙ 𝜇𝜇#Kg 𝑑𝑑𝑑𝑑 Assuming there are 𝑚𝑚 + 1 states and cash flows
/ /
j 6 are made every h years:
l (ƒ)
𝐸𝐸 “e𝑇𝑇#] f ” = 2 [ 𝑡𝑡 ⋅ 6𝑝𝑝#] d𝑡𝑡 u
6 𝑞𝑞#] =[ g 𝑝𝑝#] ∙ 𝜇𝜇]Kg 𝑑𝑑𝑑𝑑 e 6𝑉𝑉 (ƒ) + ℎ𝑃𝑃6 f(1 + 𝑖𝑖) Ω
/ / z
j 6 ƒQ (Q) ƒQ
l = s Ω 𝑝𝑝#K6 eℎ𝐵𝐵6KΩ + 𝑏𝑏6KΩ + 6KΩ𝑉𝑉 (Q) f
𝑒𝑒#] = s Q 𝑝𝑝#] 6 𝑞𝑞#] =[ g 𝑝𝑝# ∙ 𝜇𝜇#Kg ⋅ e1 − g 𝑝𝑝] f𝑑𝑑𝑑𝑑
/ Qt/
Qtu
6

l If lump sum benefit is assumed to be paid in the
Last Survivor 6 𝑞𝑞#] =[ g 𝑝𝑝] ∙ 𝜇𝜇]Kg ⋅ e1 − g𝑝𝑝# f𝑑𝑑𝑑𝑑
/ middle of an interval:

êêêê = maxe𝑇𝑇# , 𝑇𝑇] f
𝑇𝑇#] Relationships (ƒ)
e 6𝑉𝑉 (ƒ) + ℎ𝑃𝑃6 f(1 + 𝑖𝑖) Ω
6 𝑝𝑝#] + 6𝑞𝑞#] = 1 6 𝑞𝑞#] + 6 𝑞𝑞#] = 6 𝑞𝑞#]
u u z
ƒQ (Q) ƒQ
u l
= 6𝑞𝑞# = s Ω 𝑝𝑝#K6 eℎ𝐵𝐵6KΩ + 𝑏𝑏6KΩ (1 + 𝑖𝑖)Ω/l + 6KΩ𝑉𝑉 (Q) f
:|6 𝑞𝑞#] = : 𝑝𝑝#] − :K6𝑝𝑝#] 6 𝑞𝑞#] + 6 𝑞𝑞#]
u l Qt/
= :K6𝑞𝑞#] − : 𝑞𝑞#] 6 𝑞𝑞#] = 6 𝑞𝑞#] + 6 𝑞𝑞# ⋅ 6𝑝𝑝]

l l Activities of Daily Living (ADLs):

𝑞𝑞
6 #] + 6 𝑞𝑞#] = 6 𝑞𝑞#]
Independent Lives • Bathing
𝑞𝑞u
∞ #]
u
+ ∞ 𝑞𝑞#] = 1 • Dressing
6 𝑞𝑞#] = 6 𝑞𝑞# ⋅ 6 𝑞𝑞]
𝑞𝑞 l l
+ ∞ 𝑞𝑞#] = 1 • Eating
6 𝑝𝑝#] = 6 𝑝𝑝# + 6 𝑝𝑝] − 6 𝑝𝑝# ⋅ 6𝑝𝑝] ∞ #]
u l • Toileting
6 𝑝𝑝# 𝜇𝜇#K6 ⋅ 6𝑞𝑞] + 6𝑝𝑝] 𝜇𝜇]K6 ⋅ 6 𝑞𝑞# ∞𝑞𝑞#] = ∞𝑞𝑞#]
𝜇𝜇#] (𝑡𝑡) = • Continence
6 𝑝𝑝# + 6𝑝𝑝] − 6𝑝𝑝# ⋅ 6 𝑝𝑝] Contingent Insurance

• Transferring
j
Moments 𝐴𝐴u̅#] = [ 𝑣𝑣 6 ⋅ 6𝑝𝑝#] ⋅ 𝜇𝜇#K6 d𝑡𝑡
∘
j
/
𝑒𝑒#] = [ 6 𝑝𝑝#] d𝑡𝑡 j PROFIT TESTS PROFIT TESTS
/ 𝐴𝐴l̅#] = [ 𝑣𝑣 6 ⋅ 6𝑝𝑝# ⋅ 𝜇𝜇#K6 ⋅ 6𝑞𝑞] d𝑡𝑡
j
/ Profits for Traditional Products

𝑒𝑒#] = s Q 𝑝𝑝#] The profit per policy in force at time t is
Relationships
Qtu
𝐴𝐴u̅#] + 𝐴𝐴̅ u ̅
#] = 𝐴𝐴 #]
Pr6 = e 6du𝑉𝑉 + 𝑃𝑃6 − 𝐸𝐸6 f(1 + 𝑖𝑖) − 𝑞𝑞#K6du𝐷𝐷𝐵𝐵6
Relationships between (𝒙𝒙𝒙𝒙) Status and 𝐴𝐴l̅#] + 𝐴𝐴̅ l = 𝐴𝐴 ̅ #] − 𝑝𝑝#K6du 6𝑉𝑉
#]
(𝒙𝒙𝒙𝒙
êêêê) Status
𝐴𝐴u̅#] + 𝐴𝐴l̅#] = 𝐴𝐴̅ # Change in Reserve
𝑇𝑇#] + 𝑇𝑇#] = 𝑇𝑇# + 𝑇𝑇]

𝑇𝑇#] ⋅ 𝑇𝑇#] = 𝑇𝑇# ⋅ 𝑇𝑇] Reversionary Annuities Δ 6𝑉𝑉 = 6du𝑉𝑉 (1 + 𝑖𝑖) − 𝑝𝑝#K6du 6𝑉𝑉

• Make payments to (y) after (x) has died: Profit Vector
6 𝑝𝑝#] + 6 𝑝𝑝#] = 6 𝑝𝑝# + 6 𝑝𝑝]
∘ ∘ ∘ ∘ 𝑎𝑎ê#|] = 𝑎𝑎ê] − 𝑎𝑎ê#] Pr = (Pr/ Pru Prl … PrW )
𝑒𝑒#] + 𝑒𝑒#] = 𝑒𝑒# + 𝑒𝑒]
∘ ∘ ∘ ∘ • Make payments only when exactly one life is

êêê + 𝑒𝑒#]:W|
𝑒𝑒#]:W| êêê = 𝑒𝑒#:W|
êêê + 𝑒𝑒]:W|
êêê Profit Signature
alive:
𝑒𝑒#] + 𝑒𝑒#] Profit per policy issued
êêêê = 𝑒𝑒# + 𝑒𝑒] 𝐸𝐸𝐸𝐸𝐸𝐸(annuities) = 𝑎𝑎ê#] − 𝑎𝑎ê#]
in force at time
𝐴𝐴̅#] + 𝐴𝐴̅#] = 𝐴𝐴̅ # + 𝐴𝐴̅] Π = Pr6 ⋅ Prob “ Î in force at time 0”
𝑡𝑡 − 1
𝑎𝑎ê#] + 𝑎𝑎ê#] = 𝑎𝑎ê# + 𝑎𝑎ê] Profit signature: (Π/ Πu Πl … ΠW )
LONG-TERM INSURANCE COVERAGE
LONG-TERM INSURANCE COVERAGE
W𝐸𝐸#] + W 𝐸𝐸#] = W 𝐸𝐸# + W𝐸𝐸]
where

Disability Income Insurance (DII) Π/ = Pr/
Covariance of 𝑇𝑇#] and 𝑇𝑇#] Continuous Sojourn Annuity
∘ ∘ ∘ ∘ Π6 = Pr6 ⋅ 6du𝑝𝑝# , 𝑡𝑡 = 1, 2, 3, … , 𝑛𝑛
Cove𝑇𝑇#] , 𝑇𝑇#] f = Cove𝑇𝑇# , 𝑇𝑇] f + e𝑒𝑒# − 𝑒𝑒#] fe𝑒𝑒] − 𝑒𝑒#] f The EPV of an n-year continuous sojourn annuity

Cove𝑇𝑇# , 𝑇𝑇] f = 0 if 𝑇𝑇# and 𝑇𝑇] are independent on (x) in state i that pays $1 per year continuously Profit Measures
while the life remains in state i is: NPV

Exactly One Life Survives W
∞
¥¥ ƒ
Pr(exactly one life survivies 𝑡𝑡 years) 𝑎𝑎ê#:W|
êêê = [
¥¥ dï6
6 𝑝𝑝# 𝑒𝑒 d𝑡𝑡 NPV = s Πƒ ⋅ 𝑣𝑣õ
/
= 6 𝑝𝑝#] − 6𝑝𝑝#] ƒt/
EPV of benefit of an n-year DII: where 𝑟𝑟 = risk discount or hurdle rate
= 6 𝑝𝑝# + 6 𝑝𝑝] − 2 ⋅ 6𝑝𝑝#]

W Partial NPV
/u uu
Relationships between Insurance Policies, 𝑎𝑎ê#:W|
êêê = [
// /u
6 𝑝𝑝# 𝜇𝜇#K6 𝑎𝑎
ê#K6:Wd6|
êêêêêêê 𝑒𝑒
dï6
d𝑡𝑡 Q
/ ƒ
Annuities, and Premiums NPV(𝑘𝑘) = s Πƒ ⋅ 𝑣𝑣õ
With waiting period of w years, the EPV is: ƒt/
𝐴𝐴#] = 1 − 𝑑𝑑𝑎𝑎̈ #]
IRR
1 − 𝐴𝐴̅ #] êêêê
WdÁ
uu uu ∞
𝑎𝑎ê#]
êêêê = [ // /u
6 𝑝𝑝# 𝜇𝜇#K6 q𝑎𝑎
ê#K6:Wd6|
êêêêêêê − 𝑎𝑎ê#K6:Á|
êêêê r 𝑒𝑒
dï6
d𝑡𝑡
𝛿𝛿 / NPV = s Πƒ ⋅ 𝑣𝑣õ = 0
ƒ

𝑎𝑎̈ #]:W|êêê = 𝑎𝑎̈ #] − W𝐸𝐸#] ∙ 𝑎𝑎̈ #KW:]KW

ƒt/
With waiting period of w years and benefit term of
1 DPP
êêêê =
𝑃𝑃#] − 𝑑𝑑 m years, the EPV is:
𝑎𝑎̈ #]
êêêê

DPP = min[𝑡𝑡: NPV(𝑡𝑡) > 0]
Wd(zKÁ)
𝛿𝛿𝐴𝐴̅ #] [ // /u
6 𝑝𝑝# 𝜇𝜇#K6
uu
q𝑎𝑎ê#K6:zKÁ|
êêêêêêêêê − 𝑎𝑎
uu
ê#K6:Á|
êêêê r 𝑒𝑒
dï6
d𝑡𝑡

𝑃𝑃#] = Profit Margin

1 − 𝐴𝐴#] ̅ /
WdÁ
uu uu
NPV
Note: The list above is not exhaustive; similar +[ // /u
6 𝑝𝑝# 𝜇𝜇#K6 q𝑎𝑎
ê#K6:Wd6|
êêêêêêê − 𝑎𝑎ê#K6:Á|
êêêê r 𝑒𝑒
dï6
d𝑡𝑡 Profit margin =
Wd(zKÁ)
𝐸𝐸𝐸𝐸𝐸𝐸(f. premiums)
relationships can be applied to other forms of

insurance/annuities with appropriate adjustments.

www.coachingactuaries.com Copyright © 2018 Coaching Actuaries. All Rights Reserved. 5

Zeroized Reserves Defined Contribution Pension Plans Normal Cost
1. Begin with the last year and work backwards 𝐴𝐴𝐴𝐴(pension fund) = 𝐸𝐸𝐸𝐸𝐸𝐸(pension benefits) 6 𝑉𝑉 + 𝐶𝐶6 = EPV(benefits for mid-year exits)+𝑣𝑣𝑝𝑝#// 6Ku𝑉𝑉

2. Set the profit for the year to zero then solve for Defined Benefit Pension Plans
If there are no mid-year exits:
the beginning-of-year reserve Annual Retirement Benefit = 𝑛𝑛 ⋅ 𝑆𝑆¯˘˙ ⋅ 𝛼𝛼
6 𝑉𝑉
3. If the reserve is negative, set to zero and repeat where 𝐶𝐶6 =
𝑡𝑡
this entire process again until time 0 𝑛𝑛 = total number of years of service

𝑆𝑆¯˘˙ = final average salary
Gain by Source
𝛼𝛼 = accrual rate SURVIVAL MODEL ESTIMATION
SURVIVAL MODEL ESTIMATION
Gain in the order of expenses, interest, and
• Two methods to calculate the amount of Kaplan-Meier and Nelson-Aalen Estimators
mortality ( ′ = actual):
retirement benefit: Empirical Distribution
Expense: e𝑒𝑒6 − 𝑒𝑒6‘ f(1 + 𝑖𝑖6 ) + e𝐸𝐸6 − 𝐸𝐸6‘ f𝑞𝑞#K6du

§ PUC: projects salary to retirement or exit # of data points = 𝑥𝑥

Interest: e𝑖𝑖6‘ − 𝑖𝑖6 fe 6du𝑉𝑉 + 𝐺𝐺6 − 𝑒𝑒6‘ f Pr(𝑋𝑋 = 𝑥𝑥) =
date 𝑛𝑛
# of data points ≤ 𝑥𝑥

Mortality: e𝑞𝑞#K6du − 𝑞𝑞#K6du

‘ fe𝐷𝐷𝐷𝐷6 + 𝐸𝐸6‘ − 6𝑉𝑉 f § TUC: calculates salary based on employee’s 𝐹𝐹W (𝑥𝑥) = Pr(𝑋𝑋 ≤ 𝑥𝑥) =
𝑛𝑛
Actual Profit current age
𝐹𝐹W (𝑥𝑥) ⋅ [1 − 𝐹𝐹W (𝑥𝑥)]
2 [𝑆𝑆W (𝑥𝑥)] = Var
Var 2 [𝐹𝐹W (𝑥𝑥)] =
Using the actual experience: • Early retirement 𝑛𝑛

Annual Retirement Benefit
Actual Profit = e 6du𝑉𝑉 + 𝐺𝐺6 − 𝑒𝑒6‘ f(1 + 𝑖𝑖6‘ ) Kaplan-Meier Estimator
‘
= 𝑛𝑛 ⋅ 𝑆𝑆¯˘˙ ⋅ 𝛼𝛼 ⋅ (1 − pension reduction factor) ƒ
−𝑞𝑞#K6du (𝐷𝐷𝐷𝐷6 + 𝐸𝐸6‘ ) − ‘
𝑝𝑝#K6du ⋅ 6𝑉𝑉 𝑠𝑠¥
• Withdrawal without COLA 𝑆𝑆3 e𝑦𝑦ƒ f = ’ ¨1 − ≠
Expected Profit 𝑟𝑟¥
Annual Retirement Benefit = 𝑛𝑛 ⋅ 𝑆𝑆¯˘˙ ⋅ 𝛼𝛼 ¥tu
Using the assumed experience:

• Withdrawal with COLA 𝑆𝑆3 e𝑦𝑦ƒ f

Expected Profit = e 6du𝑉𝑉 + 𝐺𝐺6 − 𝑒𝑒6 f(1 + 𝑖𝑖6 ) ℎ4 e𝑦𝑦ƒ f = 1 −
Annual Retirement Benefit 𝑆𝑆3 e𝑦𝑦ƒduf
−𝑞𝑞#K6du (𝐷𝐷𝐷𝐷6 + 𝐸𝐸6 ) − 𝑝𝑝#K6du ⋅ 6𝑉𝑉 = 𝑛𝑛 ⋅ 𝑆𝑆¯˘˙ ⋅ 𝛼𝛼 Tail Correction

Total Gain ⋅ (1 + COLA)˝˛ˇ˘˝˛!˛˙ˇ "#˛d$˘ˇ%&˝"$"' "#˛

• Efron’s tail correction:
𝑆𝑆3 (𝑡𝑡) = 0 for 𝑡𝑡 > 𝑢𝑢
Total Gain = Actual Profit − Expected Profit Funding the Benefits • Klein and Moeschberger's tail correction:
= Gain from Expenses Actuarial Liability 𝑆𝑆3 (𝑢𝑢), for 𝑢𝑢 < 𝑡𝑡 < 𝛾𝛾
+ Gain from Interest 𝑆𝑆3 (𝑡𝑡) = 5
6 𝑉𝑉 = 𝐸𝐸𝐸𝐸𝐸𝐸(all accrued benefits at time 𝑡𝑡) 0, for 𝑡𝑡 ≥ 𝛾𝛾
+ Gain from Mortalities
• Brown, Hollander, and Korwar’s tail correction:
Normal Contribution 𝑆𝑆3 (𝑡𝑡) = 𝑆𝑆3 (𝑢𝑢)6/: for 𝑡𝑡 > 𝑢𝑢

6 𝑉𝑉 + 𝐶𝐶6 = EPV(benefits for mid-year exits)+𝑣𝑣𝑝𝑝#// 6Ku𝑉𝑉

Nelson-Aalen Estimator
PENSION MATHEMATICS
PENSION MATHEMATICS

ƒ
• Two methods of funding benefits: PUC and TUC 𝑠𝑠
Valuation of Benefits 7e𝑦𝑦ƒ f = s ¥
𝐻𝐻
• If there are no mid-year exits: 𝑟𝑟¥
Motivations u ¥tu
1. Attract potential employees § PUC: 𝐶𝐶6 = ⋅ 6𝑉𝑉 7
6 𝑆𝑆3 e𝑦𝑦ƒ f = 𝑒𝑒 d8e]fif
2. Provide incentive for employees to stay § TUC: 𝐶𝐶6 = q
6Ku
⋅
)´
− 1r 6𝑉𝑉

3. Facilitate turnover of older employees 6 )´*+ Variance of Estimators

4. Provide tax-friendly compensation • Greenwood’s Approximation:
Retiree Health Benefits ƒ
5. Pressure from trade unions Benefit Premium Annuity for age x at time t 𝑠𝑠¥
6. Reward employees who have contributed to the 2 ô𝑆𝑆3 e𝑦𝑦ƒ fö = ô𝑆𝑆3 e𝑦𝑦ƒ föl s
Var
j
𝐵𝐵(𝑥𝑥 + 𝑘𝑘, 𝑡𝑡 + 𝑘𝑘) 𝑟𝑟¥ (𝑟𝑟¥ − 𝑠𝑠¥ )
company’s success 𝑎𝑎̈ , (𝑥𝑥, 𝑡𝑡) = s 𝑣𝑣 Q Q 𝑝𝑝# Z \
¥tu

𝐵𝐵(𝑥𝑥, 𝑡𝑡) Used for Kaplan-Meier
Replacement Ratio Qt/

• Klein’s Estimation:
pension income in the year after retirement Value of retiree health benefit at retirement for a ƒ
𝑅𝑅 = 𝑠𝑠¥ (𝑟𝑟¥ − 𝑠𝑠¥ )
salary in the year before retirement life retiring at age xr in t years: 𝐵𝐵(𝑥𝑥𝑥𝑥, 𝑡𝑡)𝑎𝑎̈ , (𝑥𝑥𝑥𝑥, 𝑡𝑡) 2 ô𝐻𝐻
Var 7e𝑦𝑦ƒ fö = s

𝑟𝑟¥à
¥tu
Salary Projection When healthcare premiums increase exponentially ƒ
𝑠𝑠¥ (𝑟𝑟¥ − 𝑠𝑠¥ )
S: Salary 𝑆𝑆̅: Rate of salary with age and at a constant annual inflation rate 2 ô𝑆𝑆3 e𝑦𝑦ƒ fö = ô𝑆𝑆3 e𝑦𝑦ƒ föl s
Var
where: 𝑟𝑟¥à
s: Salary scale 𝑠𝑠̅: Rate of salary function ¥tu
• 𝑐𝑐 = 𝐵𝐵(𝑥𝑥 + 1, 𝑡𝑡)⁄𝐵𝐵(𝑥𝑥, 𝑡𝑡) Used for Nelson-Aalen
• Constant percentage of increase

𝑆𝑆]K6 = 𝑆𝑆] (1 + 𝑥𝑥%)6

• 𝑗𝑗 = annual rate of inflation for healthcare costs
In general,
2 ô𝑆𝑆3 e𝑦𝑦ƒ fö = ô𝑆𝑆3 e𝑦𝑦ƒ föl ⋅ Var

• Salary Scale Q (1 Q Var 2 ô𝐻𝐻

7e𝑦𝑦ƒ fö

𝐵𝐵(𝑥𝑥𝑥𝑥 + 𝑘𝑘, 𝑡𝑡 + 𝑘𝑘) = 𝑐𝑐 + 𝑗𝑗) 𝐵𝐵(𝑥𝑥𝑥𝑥, 𝑡𝑡)
𝑠𝑠] 1 + 𝑖𝑖
𝑆𝑆] = 𝑆𝑆# ⋅ 𝑎𝑎̈ , (𝑥𝑥𝑥𝑥, 𝑡𝑡) = 𝑎𝑎̈ #õ|¥∗ where 𝑖𝑖 ∗ = − 1
𝑠𝑠#

𝑐𝑐(1 + 𝑗𝑗)
• Rate of salary

𝑠𝑠̅] Actuarial Value of Total Health Benefit (AVTHB)

𝑆𝑆]̅ = 𝑆𝑆#̅ ⋅ 𝐴𝐴𝐴𝐴𝐴𝐴𝐻𝐻𝐵𝐵6
𝑠𝑠̅#

/ˆd#
u 𝑆𝑆#̅ u
𝑟𝑟#KQ Q
̅ 𝑑𝑑𝑑𝑑 =
𝑆𝑆] = [ 𝑆𝑆]K6 ⋅ [ 𝑠𝑠̅ 𝑑𝑑𝑑𝑑 = s 𝑣𝑣 𝐵𝐵(𝑥𝑥 + 𝑘𝑘, 𝑡𝑡 + 𝑘𝑘)𝑎𝑎̈ , (𝑥𝑥 + 𝑘𝑘, 𝑡𝑡 + 𝑘𝑘)
𝑠𝑠̅# / ]K6

/ 𝑙𝑙#
Qt/
• Relationships /ˆd#
𝑟𝑟#KQ Q
Rate of salary function to salary scale: = 𝐵𝐵(𝑥𝑥, 𝑡𝑡) s 𝑣𝑣 ∗ 𝑎𝑎̈ ∗
u 𝑙𝑙# ¥ #KQ|¥
Qt/
𝑠𝑠] = [ 𝑠𝑠̅]K6 d𝑡𝑡
/ Actuarial Liability at time t, tV
Salary scale to rate of salary function: 𝑠𝑠̅] ≈ 𝑠𝑠]d/.ˆ /ˆd#
𝑡𝑡 𝑟𝑟#KQ Q
s ∙ 𝑣𝑣 𝐵𝐵(𝑥𝑥 + 𝑘𝑘, 𝑡𝑡 + 𝑘𝑘)𝑎𝑎̈ , (𝑥𝑥 + 𝑘𝑘, 𝑡𝑡 + 𝑘𝑘)
𝑡𝑡 + 𝑘𝑘 𝑙𝑙#
Qt/

www.coachingactuaries.com Copyright © 2018 Coaching Actuaries. All Rights Reserved. 6

Confidence Interval MORTALITY IMPROVEMENT
MORTALITY IMPROVEMENT The Lee Carter Model

Linear Confidence Interval for 𝑆𝑆(𝑡𝑡): ln 𝑚𝑚(𝑥𝑥, 𝑡𝑡) = 𝑙𝑙𝑙𝑙(𝑥𝑥, 𝑡𝑡) = 𝛼𝛼# + 𝛽𝛽# 𝐾𝐾6 + 𝜖𝜖#,6
Single-Factor Mortality Improvement Scales
2 ô𝑆𝑆3 (𝑡𝑡)ö
𝑆𝑆3 (𝑡𝑡) ± 𝑧𝑧(uK°)/l :Var 𝑞𝑞(𝑥𝑥, 𝑡𝑡) = 𝑞𝑞(𝑥𝑥, 0)(1 − 𝜑𝜑# )6 where:

Two-Factor Mortality Improvement Scales • 𝐾𝐾6 = 𝐾𝐾6du + 𝑐𝑐 + 𝜎𝜎Q 𝑍𝑍6

Log-transformed Confidence Interval:
Cubic Spline • 𝜖𝜖#,6 is sufficiently small to be negligible
• For 𝑆𝑆(𝑡𝑡):
e𝑆𝑆3 (𝑡𝑡)u/;< , 𝑆𝑆3 (𝑡𝑡);< f where A cubic function that joins the short-term factors Constraints used to fix identifiability problem:
and long-term factors in a smooth fashion #Q 6S
2 ô𝑆𝑆3 (𝑡𝑡)ö
𝑧𝑧(uK°)/l:𝑉𝑉𝑉𝑉𝑉𝑉

Age-Based Cubic Spline s 𝛽𝛽# = 1.0 and s 𝐾𝐾6 = 0.0

𝑈𝑈) = exp ⎛ ⎞
𝑆𝑆3 (𝑡𝑡) ⋅ ln 𝑆𝑆3 (𝑡𝑡) 𝐶𝐶G (𝑥𝑥, 𝑡𝑡) = 𝑎𝑎𝑡𝑡 à + 𝑏𝑏𝑡𝑡 l + 𝑐𝑐𝑐𝑐 + 𝑑𝑑
#t#R 6t6R

⎝ ⎠ With these constraints:
Cohort-Based Cubic Spline 6S
• For 𝐻𝐻(𝑡𝑡): 𝐶𝐶H (𝑥𝑥, 𝑡𝑡) = 𝑎𝑎∗ 𝑡𝑡 à + 𝑏𝑏 ∗ 𝑡𝑡 l + 𝑐𝑐 ∗𝑡𝑡 + 𝑑𝑑∗ ∑6t6 𝑙𝑙𝑙𝑙(𝑥𝑥, 𝑡𝑡)
R
7(𝑡𝑡)
𝐻𝐻 𝛼𝛼# =
7(𝑡𝑡) ⋅ 𝑈𝑈8 \ where Solve for the constants in the age-based and 𝑡𝑡W − 𝑡𝑡/ + 1
Z , 𝐻𝐻
𝑈𝑈8 cohort-based cubic splines using these 4 Central death rate improvement factor:
2 ô𝐻𝐻
𝑧𝑧(uK°)/l:𝑉𝑉𝑉𝑉𝑉𝑉 7(𝑡𝑡)ö constraints: 𝑚𝑚(𝑥𝑥, 𝑡𝑡)
1. The starting value of the spline matches the 𝜑𝜑z (𝑥𝑥, 𝑡𝑡) = 1 −
𝑈𝑈8 = exp ⎛ ⎞ 𝑚𝑚(𝑥𝑥, 𝑡𝑡 − 1)
7(𝑡𝑡)
𝐻𝐻 improvement factor at 2007.

⎝ ⎠ 2. The starting derivative of the spline matches the In the Lee Carter model:

1 slope of the improvement function at 2007. 1 − 𝜑𝜑z (𝑥𝑥, 𝑡𝑡)~logN(𝜇𝜇 = 𝛽𝛽# 𝑐𝑐, 𝜎𝜎 = 𝛽𝛽# 𝜎𝜎Q )
𝑈𝑈) = 3. The ending value of the spline matches the
𝑈𝑈8 The Cairns-Blake-Dowd (CBD) Models
improvement factor at 2027.
Age-Based Estimators 4. The ending derivative of the spline matches the Logit Function
Individual Data slope of the improvement function at 2027. 𝑥𝑥
• Exact Exposure:
logit(𝑥𝑥) = ln
Improvement factor for age x in year t: 1 − 𝑥𝑥
𝑞𝑞Bƒ = 1 − 𝑒𝑒 dCfi/ºfi 𝑞𝑞(𝑥𝑥, 𝑡𝑡)
𝜑𝜑(𝑥𝑥, 2007 + 𝑡𝑡) = 0.5𝐶𝐶G (𝑥𝑥, 𝑡𝑡) + 0.5𝐶𝐶H (𝑥𝑥, 𝑡𝑡) 𝑙𝑙𝑙𝑙(𝑥𝑥, 𝑡𝑡) = logit[𝑞𝑞(𝑥𝑥, 𝑡𝑡)] = ln
• Actuarial Exposure: 1 − 𝑞𝑞(𝑥𝑥, 𝑡𝑡)
𝑑𝑑ƒ where t is the number of years measured from 𝑒𝑒 T›(#,6)
𝑞𝑞Bƒ = 2007 ⇒ 𝑞𝑞(𝑥𝑥, 𝑡𝑡) =
𝑒𝑒ƒ 1 + 𝑒𝑒 T›(#,6)

Note that exposures for deaths are counted until Mortality Rate with Mortality Improvement The Original CBD Model
6
the end of the age interval (u) (l)

𝑞𝑞(𝑥𝑥, 𝑡𝑡) = 𝑞𝑞(𝑥𝑥, 0) ’e1 − 𝜑𝜑(𝑥𝑥, 𝑘𝑘)f 𝑙𝑙𝑙𝑙(𝑥𝑥, 𝑡𝑡) = 𝐾𝐾6 + 𝐾𝐾6 (𝑥𝑥 − 𝑥𝑥)

Assumption Description Qtu where:

The Lee Carter Model • 𝑥𝑥 is the average age in the data set
Advance birthday to (u) (u) (u)
• 𝐾𝐾6 = 𝐾𝐾6du + 𝑐𝑐 (u) + 𝜎𝜎Q+ 𝑍𝑍6
Age Last Birthday Central Death Rate
policy date (l) (l) (l)
u 𝐾𝐾6 = 𝐾𝐾6du + 𝑐𝑐 (l) + 𝜎𝜎QN 𝑍𝑍6
𝑞𝑞# ∫ 6 𝑝𝑝# 𝜇𝜇#K6 d𝑡𝑡
𝑚𝑚# = u = / u (u) (l)
• 𝐸𝐸ô𝑍𝑍6 𝑍𝑍6 ö = 𝜌𝜌, − 1 ≤ 𝜌𝜌 ≤ 1
Advance/retreat ∫/ 6 𝑝𝑝# d𝑡𝑡 ∫/ 6𝑝𝑝# d𝑡𝑡 (¥) (ƒ)
Age Nearest Birthday birthday to nearest 𝐸𝐸ô𝑍𝑍6 𝑍𝑍: ö = 0 for 𝑡𝑡 ≠ 𝑢𝑢, 𝑖𝑖 = 1,2, 𝑗𝑗 = 1,2

policy date • Assuming constant force of mortality between
Advantages of the original CBD model over the Lee
integer ages:
Carter model:
𝑚𝑚# = 𝜇𝜇
Anniversary-to- Study starts and ends • Fewer parameters
𝑞𝑞# = 1 − 𝑒𝑒 dz´
Anniversary on policy anniversary • Less parameter uncertainty
• Assuming UDD between integer ages:

Disadvantage of the original CBD model over the

Variance of Estimators 𝑞𝑞# Lee Carter model:
• Exact Exposure: 𝑚𝑚# = • Fit to population is sometimes worse
1
𝑑𝑑ƒ 1 − 𝑞𝑞#
2 ô𝑞𝑞Bƒ ö = e1 − 𝑞𝑞Bƒ fl ⋅
Var 2 The CBD M7 Model
𝑚𝑚#
𝑒𝑒ƒl 𝑞𝑞# = (u) (l)
1 𝑙𝑙𝑙𝑙(𝑥𝑥, 𝑡𝑡) = 𝐾𝐾6 + 𝐾𝐾6 (𝑥𝑥 − 𝑥𝑥)
• Actuarial Exposure: 1 + 𝑚𝑚#
2 (à)
+𝐾𝐾6 ((𝑥𝑥 − 𝑥𝑥)l − 𝑠𝑠#l ) + 𝐺𝐺6d#
𝑞𝑞B e1 − 𝑞𝑞Bƒ f
2 ô𝑞𝑞Bƒ ö = ƒ

Var Normal and Lognormal Random Variables where:
𝑒𝑒ƒ 𝑋𝑋~𝑁𝑁(𝜇𝜇, 𝜎𝜎 l ) represents a normally distributed ´Q (#d#)N
∑´V´
Interval-based Data random variable, X, with mean 𝜇𝜇 and variance 𝜎𝜎 l . • 𝑠𝑠#l = R

#Q d#R Ku

Exposures without uniform assumption 𝑍𝑍~𝑁𝑁(0,1) represents a standard normal random • 𝐺𝐺6d# introduces a cohort time series
𝑒𝑒ƒ = 𝑃𝑃ƒ + 𝑛𝑛ƒ variable with a mean of 0 and a variance of 1.

Advantages of the CBD M7 model:

Exposures with uniform assumption

𝑌𝑌 = 𝑒𝑒 L ~logN(𝜇𝜇, 𝜎𝜎) is a lognormal random variable • Includes a cohort effect

• Exact Exposure:
with parameters 𝜇𝜇 and 𝜎𝜎. • Includes a quadratic age difference term
𝑒𝑒ƒ = 𝑃𝑃ƒ + 0.5e𝑛𝑛ƒ − 𝑤𝑤ƒ − 𝑑𝑑ƒ f MN
• Actuarial Exposure: • 𝐸𝐸[𝑌𝑌] = 𝑒𝑒 {K N
N
𝑒𝑒ƒ = 𝑃𝑃ƒ + 0.5e𝑛𝑛ƒ − 𝑤𝑤ƒ f • 𝑉𝑉𝑉𝑉𝑉𝑉[𝑌𝑌] = (𝐸𝐸[𝑌𝑌])le𝑒𝑒 O − 1f

Multiple State Estimation

Transition intensity:
𝑑𝑑¥ƒ
𝜇𝜇̂ ¥ƒ =
𝑇𝑇¥
𝑑𝑑
2 (𝜇𝜇̂ ¥ƒ ) = ¥ƒl
Var
𝑇𝑇¥

Copyright © 2018 Coaching Actuaries. All Rights Reserved. 7

www.coachingactuaries.com
www.coachingactuaries.com Personal copies
Copyright permitted.
© 2018 Resale
Coaching or distribution
Actuaries. is prohibited.
All Rights Reserved. 7