266 Multiple state models
For the EPVs of the lump sum payments we have the following formulae:
⎝ ∞
Āy = e−δt ( t pxy
00 01 02 23
µx+t:y+t + t pxy µy+t ) dt,
0
⎝ ∞
Āx = e−δt ( t pxy
00 02 01 13
µx+t:y+t + t pxy µx+t ) dt,
0
⎝ ∞
Āxy = e−δt t pxy
00
(µ01 02
x+t:y+t + µx+t:y+t ) dt,
0
⎝ ∞
Āxy = e−δt (t pxy
01 13 02 23
µx+t + t pxy µy+t ) dt,
0
⎝ ∞
Ā1xy = e−δt t pxy
00 02
µx+t:y+t dt,
0
⎝ n
Ā1xy:n = e−δt t pxy
00 02
µx+t:y+t dt.
0
From these formulae we can derive the important formula
Āxy = Āx + Āy − Āxy (8.27)
which can be explained in the same way as formulae (8.25) and (8.26) by
considering cash flows.
Note that the relationship between āx and Āx in equation (5.14) extends to
the joint life case, so that
1 − Āxy
āxy = . (8.28)
δ
The proof of this is left to Exercise 8.4.
The formulae for EPVs have been written in terms of probabilities derived
from our model. Since none of the states in the model can be re-entered once it
has been left, we have
ii ii
t pxy ≡ t pxy for i = 0, 1, 2, 3
8.9 Joint life and last survivor benefits 267
so that using formula (8.8)
⎞ ⎝ t ⎠
00 01 02 03
p
t xy = exp − (µ x+s:y+s + µ x+s:y+s + µ x+s:y+s ) ds , (8.29)
0
⎞ ⎝ t ⎠
11 13
t px = exp − µx+s ds ,
0
⎞ ⎝ t ⎠
22
t py = exp − µ23
y+s ds ,
0
and, for example,
⎝ t
01 00 01 11
t pxy = s pxy µx+s:y+s t−s px+s ds. (8.30)
0
Assuming as usual that we know the transition intensities, probabilities for
the model can be evaluated either by starting with Kolmogorov’s forward
equations, (8.14), and then using Euler’s, or some more sophisticated, method,
or, alternatively, by starting with formulae corresponding to (8.29) and (8.30)
and integrating, probably numerically.
Example 8.10 (a) Derive the following expression for the probability that the
husband has died before reaching age x + t
⎝ t ⎝ t⎝ s
00 02 00 01 11 13
s pxy µx+s:y+s ds + u pxy µx+u:y+u s−u px+u du µx+s ds.
0 0 0
(b) Now suppose that µ01 23 02 13
x+t:y+t = µx+t and µx+t:y+t = µy+t for all t ≥ 0.
Show that
(i) the probability that both husband and wife are alive at time t is
⎞ ⎝ t ⎠ ⎞ ⎝ t ⎠
exp − µ13
x+s ds exp − µ23
y+s ds , (8.31)
0 0
(ii) the probability that the husband is alive and the wife is dead at
time t is
⎞ ⎝ t ⎠⎫ ⎞ ⎝ t ⎠⎬
exp − µ13
x+s ds 1 − exp − µ23
y+s ds , (8.32)
0 0
