Exam MLC Cheat Sheet(s)

James Nylen

November 2, 2009

1 Basics

än = (1 + i) an → s̈n = (1 + i) sn
δ ∗ = 2δ → ν ∗ = ν 2 ; i∗ = 2i + i2 ; d∗ = 2d − d2
Z x+t
t px = exp (− µs ds)
x

1 1
Under DML: µx+t = ; t px µx+t =
ω−x−t ω−x
∞
X px
ex = k px ; under CF =
k=1
qx

Under UDD (and therefore DML): e̊x = ex + .5

Under DML: e̊x:n = (n) n px + n2 n qx

1
2 Insurances
1
Ax:n = Ax:n + n Ex
1 1
1 Ax − Ax:n Ax − Ax:n
Ax = Ax:n + n Ex Ax+n → n Ex = ; Ax+n =
Ax+n n Ex

(IA)x = Ax + ν px (IA)x+1
1 1
(DA)x:n = n ν qx + ν px (DA)x+1:n−1

Ax = ν qx + ν px Ax+1 = ν qx + ν 2 1| qx + ν 2 2 px Ax+2
2
Ax = ν 2 qx + ν 2 px 2 Ax+1
i 1 δ 1
Āx:n = Ax:n + n Ex ; Ax:n = Āx:n + n Ex
δ i
n| Āx = n Ex Āx+n

2.1 Under CF
µ
(I¯Ā)x = Āx āx =
(µ + δ)2
Āx:n = Āx (1 − n Ex ) + n Ex
1 q
Ax:n = Ax (1 − n Ex ) = (1 − n Ex )
q+i
1
Āx:n = Āx (1 − n Ex )

2.2 Under DML

āω−x
Āx =
ω−x
a
Ax = ω−x
ω−x
1 an
Ax:n =
ω−x

2
3 Annuities
1 − Ax:n
äx:n =
d
Ax:n = 1 − d äx:n
n
X
äx:n = ν k k px
k=0

äx = 1 + ν px äx+1
äx:n + n Ex = ax:n + 1
äx = äx:n + n| äx

n| äx = n Ex äx+n = äx − äx:n

äx:n = än + n| äx = äx + än − äx:n
2
Āx:n − (Āx:n )2
V ar(Y = āT (x)∧n ) = (analogous to āx:n )
δ2

3.1 Under CF

āx = āx:n + n| āx = āx (1 − n Ex ) + n Ex āx+n

1+i
äx:n = (1 − n Ex )
q+i

4 Miscellaneous

4.1 Payable mth-ly

m m
d(m) d(m)
 
1− =1−d → d=1− 1−
m m
āx < ä(m)
x < äx

3
A(m)
x = 1 − d(m) ä(m)
x

m−1
ä(m)
x ≈ äx −
2m
(m) 1
ä∞ =
d(m)

4.1.1 Under UDD

(m)
äx:n = α(m) äx:n − β(m) (1 − n Ex ) ($1 when pmts start - $1 when pmts end)
i
A(m)
x = Ax
i(m)

4.2 Statistics and Percentiles

p
Minimum sufficient amount = n F E[Z] + φ(P r) n F 2 V ar(Z)
X X
For mixtures of RVs: E[Y ] = wi E[Yi ] ; E[Y 2 ] = wi E[Yi2 ] ; ...
i i

100p’th percentile of the PV RV = ν h for insurances, āh for annuities.

For insurances: d px − h px = 1 − p (where d = deferral period)
For annuities: d px − h px = p
 µ/δ
T T µ
Under CF, P r(āT > āx ) = P r(ν < Āx ) = P r(ν < µ āT ) =
µ+δ

5 Premiums

E[L] = Ax − P äx
1 d Ax
Under EP: Px = −d=
äx 1 − Ax

4
 Ax
For n-pay whole life: n Px =
äx:n
1 1
Under CF: Px = P x:n = νq ; P̄x = P̄ x:n =µ
2 2
Āx − (Āx )2

P̄
V ar(L) = 1 + (2 Āx − (Āx )2 ) ; under EP =
δ (1 − Āx )2
For FCWL under EP, CF: V ar(L) = 2 Āx

6 Reserves

6.1 Standard Formulas

1 1
Px:n = − d → äx:n =
äx:n d + Px:n
d Ax:n Px:n
Px:n = → Ax:n =
1 − Ax:n d + Px:n
Prospective formula: t Vx:n = PVFB − PVFP = Ax+t:n−t − Px:n äx+t:n−t
The following three formulas hold under EP:
äx+t:n−t
• Annuity-ratio: t Vx:n = 1 −
äx:n
Ax+t:n−t − Ax:n
• Insurance-ratio: t Vx:n =
1 − Ax:n
Px+t:n−t − Px:n
• Premium-ratio: t Vx:n =
Px+t:n−t + d
Retrospective formula: t Vx:n = AVPP − AVPB = Px:n s̈x:t − t kx
A1
where t kx = accumulated cost of insurance = x:t
t Ex

5
6.2 Other Formulas
i
Under UDD: n V (Āx ) = Āx+n − P (Āx ) äx+n = n Vx
δ
For WL or term under CF: k V = 0
(k V + πk ) (1 + ik+1 ) = qx+k bk+1 + px+k k+1 V = k+1 V + qx+k (bk+1 − k+1 V )
Initial benefit reserve for year k = k−1 V + πk−1
Net amount at risk for year k = bk − k V
n−1
X
nV = P s̈n − ν qx+k (bk+1 − k+1 V ) (1 + i)n−k
k=0

Hattendorf: V ar(k L)
= ν 2 (bk+1 − k+1 V )2 px+k qx+k
+ ν 4 (bk+2 − k+2 V )2 px+k+1 qx+k+1 px+k
+ ν 6 (bk+3 − k+3 V )2 px+k+2 qx+k+2 2 px+k
+ ...

P −P
6.3 P Problems

• Accumulated difference of premiums = difference in reserves

Px s̈x:n = n Vx + n kx = n V x + n kx (1)
Px:n s̈x:n = n Vx:n + n kx = 1 + n kx (2)
1 1
P x:n s̈x:n = n V x:n + n kx = 0 + n kx (3)
n
n Px s̈x:n = n Vx + n kx = Ax+n + n kx (4)
1
(4) − (3) : (n Px − P x:n ) s̈x:n = nn Vx − n V x:n
1
= Ax+n
(2) − (4) : (Px:n − n Px ) s̈x:n = n Vx:n − nn Vx = 1 − Ax+n
(2) − (1) : (Px:n − Px ) s̈x:n = n Vx:n − n Vx = 1 − n Vx
1 1
(1) − (3) : (Px − P x:n ) s̈x:n = n Vx − n V x:n = n Vx
n Ex 1
• P x:n1 s̈x:n = s̈x:n = 1 → Replace s̈x:n with above.
äx:n P x:n1
(nk Vx = kth reserve for n-pay WL = Ax+k − n Px äx+k:n−k )

6
7 Multiple Lives
Z n Z n
n qxy = n px n py µx+t µy+t dt → n qxx = (n px )2 2 µx+t dt
0 0

n|m qxy = n+m qxy − n qxy = n+m qx n+m qy − n qx n qy

Z ∞
e̊xy = t px t py dt
0

e̊x + e̊y = e̊xy + e̊xy

Axy with common shock = Ax[µ=µx +µcs ] + Ay[µ=µy +µcs ] − Axy[µ=µx +µy +µcs ]
CoV (T (xy), T (xy)) = E[T (xy) T (xy)] − e̊xy e̊xy = (e̊x − e̊xy ) (e̊y − e̊xy )

7.1 Contingent Functions

Z n
1
n q xy = P r(T (x) < T (y) ∩ T (x) < n) = t pxy µx+t dt
0
Z n
2
n q xy = P r(T (y) < T (x) < n) = t px µx+t t qy dt
0

7.1.1 Identities

1
n q xy + n q xy1 = n qxy
2
n q xy + n q xy2 = n qxy = n qx n qy
1 2
n q xy + n q xy = n qx

7.1.2 Under DML

1
n q xy = n qx n/2 py (x dies)(y still alive when x expected to die)
2 1
n q xy = n qx n qy 2
(both die)(right order)

7
7.1.3 Under CF

1 µx
∞ q xy = µx e̊xy =
µx + µy
1 µx 1
n q xy = µx e̊xy:n = µx (e̊xy (1 − n pxy )) = n qxy = ∞ q xy n qxy
µx + µy

7.2 Contingent Insurances

Z ∞
1
Āxy = ν t t pxy µx+t dt
0
Z ∞ Z ∞
2 t
Āxy = ν t px µx+t t qy dt = ν t t pxy µy+t Āx+t dt
0 0

7.2.1 Identities

1 2
Āx = Āxy + Āxy
1
Āxy = Āxy + Āxy1
2
Āxy = Āxy + Āxy2

7.2.2 Under DML

1
Āxy = qx āy

7.2.3 Under CF

1 µx
Āxy =
µx + µy + δ
Z ∞
Āxy2 = ν t t pxy µx+t Āy+t dt = Āxy
1
Āy (Āy when x dies, if x dies first)
0

8
7.3 Reversionary Annuities

āx|y (āfailing|surviving ) = āy − āxy (y gets paid after x dies)

āx|y:n = āy:n − āxy:n
ān |x = n| āx = n Ex āx+n (n = failing status)

8 Multiple Decrements
0
Y X
px(j) = p(τ) (τ)
x = 1 − qx = 1 − qx(j) (even given no assumptions)
j j
Z 1
0 (j) 0 (j) 0 (j) (j)
qx = 1 − px ; px = exp (− µx+t dt)
0
∞
X X
(j)
`(j)
x = dk → `(τ)
x = `(j)
x
k=x j

n−1 (j) Z n
(j)
X d k (τ) (j)
n qx = (τ)
= t px µx+t dt (affected by changes in any decrement)
k=0 `x 0

(j)
If the forces of each decrement are proportional, then the qx ’s have the same
(τ)
proportions relative to qx .

8.1 Density Functions

(j)
fT,J (t, j) = t p(τ)
x µx+t

(j) Z ∞
`x (τ) (j)
fJ (j) = P r(J = j) = (τ)
= t px µx+t dt = ∞ qx(j)
`x 0

(τ)
fT (t) = t p(τ)
x µx+t

(j)
µx+t
fJ|T (j|t) = P r(J = j | T = t) = (τ)
µx+t

9
8.2 Under UDDMDT
0 (j) (τ)
px(j) = (p(τ)
x )
qx /qx

8.3 Under UDDASDT

0 (j) 0
t qx = (t) qx(j) , 0 ≤ t ≤ 1 , for all j
0 0
For m = 2: qx(1) = qx(1) (1 − 12 qx(2) )
0 0 0 0 0
For m = 3: qx(1) = qx(1) (1 − 21 qx(2) − 21 qx(3) + 13 qx(2) qx(3) )

9 Expenses

% of premium, % of benefit amount, and fixed expenses are incurred

at the beginning of each policy year (% of premium expenses are only
incurred while premium is still being paid.)
Settlement charges are incurred when the death benefit is paid.
Extended EP: E[0 Le ] = 0 → PVFB + PVFE = PVF(ELP)
where ELP = expense loaded premium = P + expense loading.
Expense-augmented reserve: k V E = E[k Le ] = PVFB + PVFE − PVF(ELP)
Expense reserve = PVFE − PVF(L = level expense loading to Px )

9.1 Asset Shares

(d) (w)
Two decrements: = death; = withdrawal
(d) (w) (τ)
(k AS + G (1 − ck ) − ek ) (1 + i) = bk+1 qx+k + k+1 CV qx+k + k+1 AS px+k
n−1 h
X i (1 + i)n−k
(d) (w)
n AS = G(1 − ck ) − ek − bk+1 ν qx+k − k+1 CV ν qx+k (τ)
k=0 n−k px+k

10
10 Poisson Processes
λk
P r(Poisson(λ) = k) = e−λ
k!
Z t
m(t) = λ(x) dx ; effective λ between a and b = m(b) − m(a)
0

Thinned Poisson “sub-processes” are independent.

10.1 Compound Poisson Processes

Let S(t) = X1 + X2 + ... + XN (t) be a process such that frequency follows

a Poisson process N (t), and severity is represented by Xi (IID per event).
Then:
• E[S] = λN E[Xi ]
• V ar(S) = λN E[Xi2 ] = λN (V ar(Xi ) + (E[Xi ])2 )

10.2 Mixed Poisson Processes and Negative Binomial

Suppose N (t) is a Poisson process with a rate that is Gamma-distributed

with parameters (α, θ). Then:
N (t) ∼ NB(r = α, β = t θ)
Recall:
• X ∼ Γ(α, θ) → E[X] = αθ ; V ar(X) = αθ2
• X ∼ NB(r, β) → E[X] = rβ ; V ar(X) = rβ(β + 1)
Suppose X ∼ NB(r, β). Let t = r, b = β, x = k. Then:
  t  x
t+x−1 1 b
P r(X = k = x) =
x 1+b 1+b

11
11 Markov Chains

Q = transition matrix for 1 period (homogeneous)

kQ = transition matrix for k periods = Qk (homogeneous)

Qn = transition matrix from time n to time n + 1 (non-homogeneous)
Q(i,j) = P r(in state j next period | in state i now) (Q(row,column) )
(i,j)
kQ = P r(in state j k periods from now | in state i now)
P (i) = Q(i,i) = P r(stay in state i for 1 period)
(i)
kP = (P (i) )k = P r(stay in state i for k periods) (6= k Q(i,i) )
For Markov-modeled insurance, keep in mind that you can only die once (so
the probability that the insured was already dead must be subtracted out
from each probability vector).

12