SMA4043 Survival Models and Ruin Theory | Yumn S.

Yusoff

Chapter 1: Survival Models.

1.4 Actuarial notations
 The actuarial notations for survival and mortality probabilities are:

𝑡𝑝𝑥 = Pr(𝑇𝑥 > 𝑡) = 𝑆𝑥 (𝑡)

𝑡𝑞𝑥 = Pr(𝑇𝑥 ≤ 𝑡) = 1 − 𝑆𝑥 (𝑡) = 𝐹𝑥 (𝑡)

𝑢|𝑡 𝑞𝑥 = Pr(𝑢 < 𝑇𝑥 ≤ 𝑢 + 𝑡) = 𝑆𝑥 (𝑢) − 𝑆𝑥 (𝑢 + 𝑡)

Where
 𝑡𝑝𝑥 is the probability that (𝑥) survives to at least age 𝑥 + 𝑡.
 𝑡𝑞𝑥 is the probability that (𝑥) survives to at least age 𝑥 + 𝑡.
 𝑢|𝑡𝑞𝑥 is the probability that (𝑥) survives 𝑢 years, and then dies
in the subsequent 𝑡 years, that is between ages 𝑥 + 𝑢 and 𝑥 +
𝑢 + 𝑡. This is called a deferred mortality probability, because it
is the probability that death occurs in some interval following a
deferred period.

 The relationships below follow immediately from the definitions above:

𝑡𝑝𝑥 + 𝑡𝑞𝑥 =1

𝑢|𝑡 𝑞𝑥 = 𝑢𝑝𝑥 − 𝑢+𝑡𝑝𝑥

𝑢+𝑡𝑝𝑥 = 𝑡𝑝𝑥 𝑢𝑝𝑥+𝑡

1 𝑑
𝜇𝑥 = − 𝑥𝑝0
𝑝
𝑡 0
𝑑𝑥

Similarly,
1 𝑑 𝑑
𝜇𝑥+𝑡 = − 𝑝
𝑡 𝑥 → 𝑡𝑝𝑥 = − 𝑡𝑝𝑥 𝜇𝑥+𝑡
𝑡 𝑝𝑥 𝑑𝑡 𝑑𝑡
 SMA4043 Survival Models and Ruin Theory | Yumn S. Yusoff

𝑡
𝑡𝑝𝑥 = 𝑒𝑥𝑝 {− ∫ 𝜇𝑥+𝑠 𝑑𝑠}
0

𝑡
𝑡𝑞𝑥 =∫ 𝑠𝑝𝑥 𝜇𝑥+𝑠 𝑑𝑠
0

When 𝑞𝑥 is small, it follows that 𝑝𝑥 is close to 1, and hence 𝑠𝑝𝑥 is close to 1

for 0 ≤ 𝑠 < 1. Thus
1
𝑞𝑥 ≈ ∫ 𝜇𝑥+𝑠 𝑑𝑠 ≈ 𝜇𝑥+1⁄
2
0

Example 4.
Using the 𝐹0 (𝑡) given in Example 1, calculate both 𝑞𝑥 and 𝜇𝑥+1⁄ for 𝑥 = 20 and
2
for 𝑥 = 110.

1.5 Mean and Standard Deviation of 𝑇𝑥 .

 We consider the expected future lifetime of (𝑥), 𝐸[𝑇𝑥 ], is denoted in
actuarial notation by 𝑒𝑥0 .
 We also call this the complete expectation of life.
 SMA4043 Survival Models and Ruin Theory | Yumn S. Yusoff

∞
𝑒𝑥0 = ∫ 𝑡𝑓𝑥 (𝑡) 𝑑𝑡
0

∞
= ∫ 𝑡 𝑡𝑝𝑥 𝜇𝑥+𝑡 𝑑𝑡
0

∞
𝑑
= ∫ 𝑡 ( 𝑝 ) 𝑑𝑡
0 𝑑𝑡 𝑡 𝑥

∞
∞
= − (𝑡 𝑡𝑝𝑥 |0 −∫ 𝑡𝑝𝑥 𝑑𝑡)
0

∞
𝑒𝑥0 = ∫ 𝑡𝑝𝑥 𝑑𝑡
0

 For 𝐸[𝑇𝑥2 ],
∞
𝐸[𝑇𝑥2 ] = ∫ 𝑡 2 𝑡𝑝𝑥 𝜇𝑥+𝑡 𝑑𝑡
0

∞
𝑑
= ∫ 𝑡2 ( 𝑝 ) 𝑑𝑡
0 𝑑𝑡 𝑡 𝑥

∞
2 ∞
= − (𝑡 𝑡𝑝𝑥 |0 −∫ 𝑡𝑝𝑥 2𝑡 𝑑𝑡)
0

∞
= 2 ∫ 𝑡 𝑡𝑝𝑥 𝑑𝑡
0

 Then, the variance of 𝑇𝑥 can be calculated as

𝑉[𝑇𝑥 ] = 𝐸[𝑇𝑥2 ] − (𝑒𝑥0 )2

 SMA4043 Survival Models and Ruin Theory | Yumn S. Yusoff

Example 5.
Refer to Example 1, calculate 𝑒𝑥0 and 𝑉[𝑇𝑥 ] for 𝑥 = 30 and for 𝑥 = 80.

1.6 Curtate Future Lifetime.

 In many insurance applications we are interested not only in the future
lifetime of an individual, but also in what is known as the individual’s
curtate future lifetime.
 The curtate future lifetime random variable is defined as the integer part
of future lifetime, and is denoted by 𝐾𝑥 for a life aged 𝑥.
 We can find the probability of 𝐾𝑥 by noting that for 𝑘 = 0, 1, 2, … , 𝐾𝑥 = k
if and only if (𝑥) dies between the ages of 𝑥 + 𝑘 and 𝑥 + 𝑘 + 1. Thus, for
𝑘 = 0, 1, 2, …
𝑃𝑟[𝐾𝑥 = 𝑘] = Pr[𝑘 ≤ 𝑇𝑥 < 𝑘 + 1]

= 𝑘| 𝑞𝑥

= 𝑘𝑝𝑥 − 𝑘+1𝑝𝑥
 SMA4043 Survival Models and Ruin Theory | Yumn S. Yusoff

= 𝑘𝑝𝑥 − 𝑘 𝑝𝑥 𝑝𝑥+𝑘

= 𝑘𝑝𝑥 𝑞𝑥+𝑘

 The expected value of 𝐾𝑥 is denoted by 𝑒𝑥 , so that 𝑒𝑥 = 𝐸[𝐾𝑥 ], and is

referred to as the curtate expectation of life. So

𝐸[𝐾𝑥 ] = 𝑒𝑥

= ∑ 𝑘. 𝑃𝑟[𝐾𝑥 = 𝑘]
𝑘=0

= ∑ 𝑘( 𝑘𝑝𝑥 − 𝑘+1𝑝𝑥 )
𝑘=0

= ( 1𝑝𝑥 − 2𝑝𝑥 ) + 2( 2𝑝𝑥 − 3𝑝𝑥 ) + 3( 3𝑝𝑥 − 4𝑝𝑥 )+⋯

=∑ 𝑘 𝑝𝑥
𝑘=1

 Similarly,
∞

𝐸[𝐾𝑥2 ] = ∑ 𝑘 2 ( 𝑘𝑝𝑥 − 𝑘+1𝑝𝑥 )

𝑘=0

= ( 1𝑝𝑥 − 2𝑝𝑥 ) + 4( 2𝑝𝑥 − 3𝑝𝑥 ) + 9( 3𝑝𝑥 − 4𝑝𝑥 )+⋯

∞ ∞

= 2 ∑ 𝑘 𝑘𝑝𝑥 − ∑ 𝑘 𝑝𝑥
𝑘=1 𝑘=1
 SMA4043 Survival Models and Ruin Theory | Yumn S. Yusoff

= 2 ∑ 𝑘 𝑘𝑝𝑥 − 𝑒𝑥
𝑘=1

 An approximation to a simple relationship between 𝑒𝑥0 and 𝑒𝑥 is

1
𝑒𝑥0 ≈ 𝑒𝑥 +
2

1.7 Parametric Survival Models.

 In this section we explore some continuous probability distribution
functions that at times serves as survival models.
 The uniform distribution is defined as
1
𝑓(𝑡) = , 0≤𝑡 ≤𝜔
𝜔

 The exponential distribution is defined as

𝑓(𝑡) = 𝜃𝑒 −𝜃𝑡 , 𝜃 ≥ 0, 𝑡 ≥0

 The Gompertz distribution is defined by its hazard rate as

𝜆(𝑥) = 𝐵𝑐 𝑥 , 𝑥 ≥ 0, 𝐵 > 0, 𝑐>1

 The Makeham distribution is obtain by modifying the Gompertz

distribution and it is defined as

𝜆(𝑥) = 𝐴 + 𝐵𝑐 𝑥 , 𝑥 ≥ 0, 𝐵 ≥ 0, 𝑐 > 1, 𝐴 > −𝐵

 SMA4043 Survival Models and Ruin Theory | Yumn S. Yusoff

 The Weibul distribution is defined as

𝜆(𝑥) = 𝑘. 𝑥 𝑛 , 𝑥 ≥ 0, 𝑘 > 0, 𝑛 > −1

 The Gamma distribution is defined as

1 𝑥−ξ
𝛼−1 (− 𝛽 )
𝑓(𝑥) = (𝑥 − ξ) 𝑒 , 𝑥 > ξ, 𝛼 > 0, 𝛽>0
𝛽 𝛼 Г(𝛼)

Example 6.
Let 𝜇𝑥 = 𝐵𝑐 𝑥 , x > 0 where B and c are constants such that 0 < B < 1 and c
> 1. This model is called Gompertz’ law of mortality. Derive an expression
for Sx(t).