Insurance: Mathematics and Economics 54 (2014) 8492

Contents lists available at ScienceDirect

Insurance: Mathematics and Economics

journal homepage: www.elsevier.com/locate/ime

Asset allocation for a DC pension fund with stochastic income and

mortality risk: A multi-period meanvariance framework
Haixiang Yao a , Yongzeng Lai b, , Qinghua Ma a , Minjie Jian c
a

School of Informatics, Guangdong University of Foreign Studies, Guangzhou 510006, China

Department of Mathematics, Wilfrid Laurier University, Waterloo, Ontario, Canada, N2L 3C5

Department of Applied Mathematics, College of Science, South China Agricultural University, Guangzhou 510642, China

highlights

A multi-period meanvariance asset allocation for DC pension funds is studied.

Stochastic income and mortality risk are considered in the model.
Lagrange multiplier method and dynamic programming approach are used.
Explicit expressions of the efficient strategy and efficient frontier are derived.
Some special cases are discussed and some numerical analyses are presented.

article

info

Article history:
Received June 2013
Received in revised form
August 2013
Accepted 29 October 2013
Keywords:
Asset allocation
Defined contribution pension fund
Multi-period meanvariance
Stochastic income
Mortality risk

abstract
This paper investigates an asset allocation problem for defined contribution pension funds with stochastic
income and mortality risk under a multi-period meanvariance framework. Different from most studies
in the literature where the expected utility is maximized or the risk measured by the quadratic mean
deviation is minimized, we consider synthetically both to enhance the return and to control the risk
by the meanvariance criterion. First, we obtain the analytical expressions for the efficient investment
strategy and the efficient frontier by adopting the Lagrange dual theory, the state variable transformation
technique and the stochastic optimal control method. Then, we discuss some special cases under our
model. Finally, a numerical example is presented to illustrate the results obtained in this paper.
2013 Elsevier B.V. All rights reserved.

1. Introduction
According to the fund procurement and operation pattern, pension funds can mainly be divided into two types: one is defined
benefit (DB) pension funds, the other is defined contribution (DC)
pension funds. A DB pension fund is a pension fund where the benefits are set in advance by the sponsor. In such a fund, contributions are set and constantly adjusted so as to ensure that the fund

This research is supported by grants from the National Natural Science Foundation of China (No. 71271061, 61104138), Natural Science Foundation of Guangdong
Province (No. S2011010005503), Scientific and Technological Innovation Foundation of Guangdong Colleges and Universities (No. 2012KJCX0050), Science and Technology Planning Project of Guangdong Province (No. 2012B040305009), National
Social Science Foundation of China (No. 11CGL051), the Business Intelligence Key
Team of Guangdong University of Foreign Studies (No. TD1202), and National Statistical Science Research Projects (No. 2013LY101).
Corresponding author. Tel.: +1 519 884 0710x2107; fax: +1 519 884 9738.
E-mail addresses: yaohaixiang@gdufs.edu.cn (H. Yao), ylai@wlu.ca (Y. Lai),
mqh@mail.gdufs.edu.cn (Q. Ma), iamjianminjie@outlook.com (M. Jian).

0167-6687/$ see front matter 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.insmatheco.2013.10.016

remains in balance, and the risk is borne by the organizer of the

fund. A DC pension fund is a pension fund where contributions are
set and benefits therefore depend solely on the accumulation scale
and the return of the investment, thus the financial risk is borne by
the members.
Historically, DB pension funds have been the more popular
and preferred by workers since the management is easy and
the risk is borne by the sponsors of the fund. However, due to
the demographic evolution and the development of the capital
market, especially due to the population aging problem and the
longevity risk, DC has become more and more popular in the global
pension market in recent years, and more and more countries
have shifted completely or partially from a DB pension scheme to
a DC pension scheme. As a result, the study of DC pension fund
investment management has become a research hot topic in the
actuarial and the financial literature over the past decade. Due to
space limitations, only some important research results are briefly
introduced below.
By minimizing the risk measured by the quadratic mean
deviation (also called the target-based criterion), Vigna and

H. Yao et al. / Insurance: Mathematics and Economics 54 (2014) 8492

Haberman (2001) and Haberman and Vigna (2002) investigate

multi-period DC pension fund investment managements during
accumulation phase, and obtain the optimal investment strategies
by using the dynamic programming method. Gerrard et al. (2004)
extend the work of Vigna and Haberman (2001) to continuoustime setting and post-retirement case. Based on continuous-time
expected utility maximization: Deelstra et al. (2003) and Giacinto
et al. (2011) investigate optimal asset allocation for DC pension
funds with stochastic interest rates and a minimum guarantee
protection; Gao (2009) investigate the optimal portfolios for DC
pension funds under a constant elasticity of variance (CEV) by
applying the Legendre transform and dual theory; Han and Hung
(2012) consider a portfolio selection problem for DC pension
fund under inflation by using the stochastic control approach.
Using quadratic risk minimization criterion, Gerrard et al. (2012)
formulate and solve a stochastic control and optimal stopping
problem of finding the optimal time of annuitization for a DC
pension scheme in the de-cumulation phase. Based on targetdriven framework and prospect theory in behavioral finance, Blake
et al. (2013) study the optimal asset allocation problem for DC
pension funds.
As the contribution period in a pension fund is very long,
generally from 20 to 40 years due to promotion prospect and
changes in economic environment, it is crucial to allow a stochastic
term structure for the income (salary). As the contribution is often
a fixed percentage of salary, the contribution is also stochastic.
Therefore, it is crucial to take into account the income risk for a
DC pension fund management. Recently, by using continuous-time
utility maximization model, Cairns et al. (2006), Zhang and Ewald
(2010) and Ma (2011) investigate the optimal asset allocation
with stochastic income for DC pension funds in different market
settings, where the income process is described by a geometric
Brownian motion. Under the DC pension framework, Emms (2012)
study the lifetime investment and consumption problem with
stochastic income based on CARA utility maximization model.
A pension fund member may die before his retirement. In
such a case, his pension plan has to be terminated due to
mortality risks, such as traffic accident, fire hazard and serious
illnesses, etc. Although there are many studies in the literatures
on ordinary portfolio selection and life insurance problems with
uncertain time horizon, see for example, Martellini and Urosevic
(2006), Pliska and Ye (2007) and Christophette et al. (2008),
there are very limited studies aim at pension fund investment
managements. By maximizing the expected utility, Charupat and
Milevsky (2002) consider the optimal asset allocation in a variable
annuity contract that may have some relation to DC pension fund;
Hainaut and Devolder (2007) consider the optimal dividend policy
and the asset allocation of a DB pension fund under the mortality
risk. By minimizing the risk which is a quadratic target-based
cost function, Hainaut and Deelstra (2011) investigate optimal
contribution rate of a DB pension fund with a stochastic mortality
which is modeled by a jump process. But all these studies focus
on annuity contract and DB pension fund management problems.
Research on DC pension fund portfolio selection problems with
mortality risks is very limited.
Furthermore, researchers of the above mentioned studies either
maximize the expected utility or minimize the risk which is
measured by the quadratic mean deviation, but they do not
apply the mean-risk bi-objective criteria to study the pension
fund investment problems. It is well known that the mean-risk
framework has become one of the most basic frameworks in
the modern portfolio selection theory since Markowitz (1952)
published his seminal work on the meanvariance portfolio
selection. The meanvariance criterion is a mean-risk bi-objective
criterion that strikes a balance between enhancing the return and
controlling the risk, where the risk is measured by the variance of

85

the portfolios return. Along with the break through of solving of

the dynamic meanvariance model (see Zhou and Li (2000) and
Li and Ng (2000) for continuous-time case and multi-period case,
respectively), the dynamic meanvariance model also become one
of the most important frameworks to study asset allocation and
assetliability management problems, see, for example, Fu et al.
(2010), Costa and Oliveira (2012), Chen et al. (2008), Wu and Li
(2011), Chiu and Wong (2013) and Yao et al. (2013a,b). But, to
our knowledge, only very few studies in the literature apply the
dynamic meanvariance models to study pension fund investment
problems. By using the continuous-time meanvariance model,
Delong et al. (2008) and Josa-Fombellida and Rincn-Zapatero
(2008) study the optimal investment and contribution strategies
for DB pension funds; Vigna (2012) investigates the portfolio
selection problem for DC pension funds. But they only focus on the
continuous-time case without touching the multi-period case.
To the best of our knowledge, no research work in the literature
investigates the multi-period version of DC pension fund asset
allocation problems by using the meanvariance framework.
This paper attempts to explore this topic. In this paper, we
consider a DC pension fund investment management problem
by using the multi-period meanvariance model. In addition, we
incorporate stochastic income and mortality risk into our model.
Mathematically, the inclusion of stochastic income and mortality
risk increases the difficulty in solving the Bellman equation which
comes from the dynamic programming approach. Specifically,
the inclusion of the stochastic income adds a new state variable
to the model, namely, there are two state variables for our
problem: the wealth and the wage income. Furthermore, after
incorporating the mortality risk, our asset allocation problem for
a DC pension fund becomes more complicated. Therefore, adding
the stochastic income and the mortality risk into the multi-period
meanvariance model drastically increase the computational
complexity in obtaining the closed form solutions. These solutions
cannot be obtained easily by using the embedding technique
used in Li and Ng (2000) since the calculation procedure of the
embedding technique is quite troublesome. Different from most
of the researches in the literature on multi-period meanvariance
models which adopt the embedding technique, we use the state
variable transformation technique and the Lagrange dual theory
to solve the model synthetically. Compared to the embedding
technique, our approach is relatively simple in procedure settings
and calculations.
The remainder of the paper is organized as follows. The market
setting is described in Section 2. The multi-period meanvariance
asset allocation model for DC pension funds with stochastic income
and mortality risk are set up in the same section. In Section 3,
the original model is transformed into a standard multi-period
stochastic control problem by using the Lagrange multiplier, then
the corresponding analytical solution is derived by using the state
variable transformation technique and the dynamic programming
approach. Closed form expressions for the efficient investment
strategy and the efficient frontier are obtained in Section 4 by
using the Lagrange dual theory. Some special cases are discussed
in Section 5. In Section 6, some numerical analyses are presented
to illustrate our results. The paper is concluded in Section 7.
2. Model formulation
Consider a financial market consisting of n + 1 assets that
may include a risk free asset or be all risky. Denote by ek =
(e0k , e1k , e2k , . . . , enk ) the random returns of these n + 1 assets over
period k (i.e., the time interval from time k to time k + 1), k =
0, 1, . . . , T 1, where represents the transpose of a matrix
or a vector. This paper considers a multi-period asset allocation
problem for a DC pension fund. Suppose that a representative

86

H. Yao et al. / Insurance: Mathematics and Economics 54 (2014) 8492

pension fund member enters a pension plan at time 0, and plans

to retire at time T . Before his retirement, he needs to contribute
certain amount of money at the beginning of every period in a
predefined way. Upon his retirement, he can convert the wealth
of his pension fund into annuity so that he can receive a scheduled
pension at every period after his retirement. If the member dies
before his retirement, his heir can withdraw all the money of the
pension fund. Denote by x0 and y0 (>0) the initial fund paid and
the initial income, respectively. Let yk be the wage income received
at time k. It is assumed in this paper that the wage income is
stochastic and satisfies the following dynamics
yk+1 = qk yk ,

k = 0, 1, . . . , T 1,

(1)

where qk is an exogenous random variable representing the

stochastic growth rate of the wage income over period k. It is also
assumed that the wage income cannot be negative. So we suppose
that qk > 0 almost surely for all k = 0, 1, . . . , T 1. Suppose
ck yk is the amount that the member contributes at time k, where
ck is a deterministic variable depending on k only. Let xk and zk
denote the wealth of the pension fund just before and just after his
contribution at time k, respectively. Then we have zk = xk + ck yk .

time T of the pension fund plan is k; if he dies after T 1, the

actual terminated time T = T . I.e.,

T =

k,
T,

k 1 < k and 1 k T 1,
> T 1.

(5)

We now discuss how to find the probability (mass) function for T .

Let S (t ) (t > 0) be the survival probability of the pension fund
member, i.e.,
S (t ) = Pr( t | > 0),

(6)

where, Pr() is the probability measure. It is well known that

(see Charupat and Milevsky (2002) and Pliska and Ye (2007)) the
survival probability S (t ) can be expressed as the following form
S ( t ) = e

t
0

(s)ds

(7)

where (s) is the instantaneous hazard rate (mortality intensity).

By (5) and (6), the probability mass function of T is given by
pk := Pr(T = k) =

S (k 1) S (k), k = 1, . . . , T 1,
S (T 1), k = T .

(8)

According to (7) and (8), we have

k1

T 1

(s)ds

e
(s)ds
> 0,

(s)ds

k = 1, . . . , T 1,

Remark 1. In our model, ck can depend on k. Compared with the

case where ck is a constant c irrelevant to k, our model is more
adaptable and flexible. We point out that it meets the requirements
of the actual needs in most cases. For example, generally speaking,
the pension fund member contributes every month, but one day is
a period for the pension fund investment. Obviously, in this case,
the pension fund member needs not to contribute for every period.
If over period k, he needs not to contribute, we can set ck = 0;
otherwise, we can set ck > 0.

pk =

Remark 2. In order to make our model to be more general, we do

not assume that ck 0. For example, when ck < 0, ck yk can be
interpreted as the consumption of the member or the distribution
of the pension fund over period k, k = 1, 2, . . . , T . Therefore,
our model can also be used to study the DC pension fund asset
allocation problem in the de-cumulation phase.

Assumption 1. E[ek ek ] > 0, i.e., E[ek ek ] is a positive definite

matrix for k = 0, 1, . . . , T 1.

Suppose that a pension fund can be invested in n + 1 assets in

the market. Denote by uik the amount invested in the ith asset over
period k, i = 1, 2, . . . , n and k = 1, 2, . . . , T . Then incorporation
of the contribution ck yk at the beginning of period k, the
amount
n
i
invested in the 0th asset over period k is (xk + ck yk )
i=1 uk ,
k = 1, 2, . . . , T . Therefore, xk follows the dynamics

xk+1 = e0k

(xk + ck yk )

uik

i=1

= ( xk +

ck yk e0k

eik uik

i=1

+ Pk uk ,

(2)

where Pk = (
,
,...,
(u1k , u2k , . . . , unk ) .
From (1) and (2), it follows that
e1k

xk+1 + ck+1 yk+1 = (xk +

e0k

e2k

ck yk e0k

e0k

enk

) and uk

e0k

+ ck+1 qk yk + Pk uk .

(3)

Notice that zk = xk + ck yk , then by (3), the dynamics of zk satisfies

zk+1 = zk e0k + ck+1 qk yk + Pk uk .

(4)

Though the pension fund member plans to retires at time T .

However, in reality, he may die before T and in this case his pension
fund plan has to be terminated before the retired time T due to
mortality risks, such as traffic accident and serious illness. Suppose
the pension fund member is alive at time t = 0 and denote by his
lifetime (the time of death), where is a positive random variable.
If he dies during the (k 1)th time period, the actual terminated

k = T.

(9)

Let k be the -field representing the information available

till time k. We call the investment strategy u = {uk ; k =
0, 1, . . . , T 1} admissible if uk is measurable with respect to k .
Let k denote the collection of all admissible strategies starting
at time k. Let E[] and Var[] denote the expectation operator and
variance operator, respectively. Throughout this paper, we make
the following assumptions similar to most literatures.

Assumption 2. The random series k = (Pk , qk ) for k

0, 1, . . . , T 1 are statistically independent.

Assumption 3. is statistically independent of k for k

0, 1, . . . , T 1.

where 0 is the n dimension zero vector

Assumption 4. E[Pk ] = 0,
for k = 0, 1, . . . , T 1.
Remark 3. Assumption 1 is used in Li and Ng (2000) and it
means that the financial assets in the market are not redundant.
Assumption 2 shows that the returns of financial assets and
wage income are statistically independent among different time
periods. Similar assumptions are also used in some literatures. For
example, Li and Ng (2000) assume that the returns of the financial
assets are statistically independent among different time periods.
Assumptions 3 and 4 are used in Wu and Li (2011) and Yao et al.
(2013b).
The optimal asset allocation problem for a DC pension fund under
the multi-period meanvariance framework refers to the problem
of finding the optimal admissible investment strategy such that the
variance of the terminal wealth is minimized for a given expected
terminal wealth level d, i.e.,

min Var[xT ] := E[x2T ] d2 ,

u0

s.t. E[xT ] = d, (1)(2).

(10)

The solution u = {uk ; k = 0, 1, . . . , T 1} of Problem (10)

is called an efficient investment strategy. The point (Var[xT ], d)
corresponding to an efficient investment strategy on the variancemean space is called an efficient point. The set of all the efficient
points forms the efficient frontier in the variance-mean space.

H. Yao et al. / Insurance: Mathematics and Economics 54 (2014) 8492

87

3. Solution scheme

Then by the dynamic programming principle, we obtain the

Bellman equation for Problem (16) as follows

For convenience, we define p0 = 0. By the law of total

probability, under Assumption 3 we have

fk (zk , yk ) = min E pk (zk ck yk )2

uk

+ 2pk (zk ck yk ) + fk+1 (zk+1 , yk+1 )]

= pk zk2 + pkck2 y2k 2pk ck zk yk + 2pk zk 2pk c

k yk
0

+
min
E
f
z
e
+
c
q
y
+
P
u
,
q
y
,

k
+
1
k
k
+
1
k
k
k
k
k
k
k

uk

2pT (zT cT yT )

fT (zT , yT 2) = pT (zT cT yT ) +
= pT zT 2pT cT zT yT + pT cT2 y2T + 2pT zT 2pT cT yT .

T
T

E[x ] =
E[xT |T = s] Pr(T = s) = E
ps xs ,

T
s=0
s=0

T
T

2
2

E[xT |T = s] Pr(T = s) = E
ps xs .
E[xT ] =
s=0

(11)

s=0

Thus, Model (10) is equivalent to

2
2

min E
ps xs d ,

u0
s=0

s.t. E
ps xs = d, (1)(2).

(12)

s =0

The equality constraint E

s=0

p s xs

= d in Model (12) can

be eliminated by the Lagrange method. We first fix a Lagrange

multiplier 2 and turn to solve the following optimization problem

T
T

min E
2
2
p x d + 2 E
p x d
,
s s

u0

s s

s=0

s=0

(13)

s.t. (1)(2).

To this end, we first focus on solving Problem (13). Since

d2 + 2 E

ps x2s

s=0

=E

p s xs

s=0

ps x2s + 2ps xs

d2 2d,

(14)

s=0

following optimization problem in the sense that they share the

same optimal solution

min E
0

ps x2s

+ 2ps xs

s.t. (1)(2).

(15)

s=0

For convenience, we make a state variable transformation for

Problem (15). Problem (15) can be transformed into

min E

u0

Setting t = 0, we find that the optimal values to Problem (13) and

(16) are f0 (z0 , y0 ) and (f0 (z0 , y0 ) d2 2d), respectively, where
z0 = x0 + cy0 .
To find the explicit expression of fk (zk , yk ), we construct the
series wk , hk , k , k , k and gk satisfying the recurrence relations
and boundary conditions as follows.

wk = pk + wk+1 Ak ,
hk = pk + hk+1 Jk ,

(19)
(20)

h2k+1

Dk ,
T = 0 ,
wk+1
k = k+1 Ck pk ck + wk+1 ck+1 Ck ,
T = cT pT ,

2
2
k = k+1 E[qk ] + pk ck + (wk+1 ck+1 + 2k+1 ) ck+1 Bk

2
k+1 E[q2k ] Bk ,

w
k +1

T = cT2 pT ,

g = gk
+1 E[qk ] pk ck + hk+1

k
k+1
ck+1 Mk
(E[qk ] Mk ) ,

wk+1

gT = cT pT ,

k = k+1

ps (zs cs ys ) + 2ps (zs cs ys ) ,

2

s =0

(16)

Ak = E[(e0k )2 ] E[e0k Pk ]E1 [Pk Pk ]E[e0k Pk ],

B = E[q2k ] E[qk Pk ]E1 [PkPk ]E[qk Pk ],

k
Ck = E[e0k qk ] E[e0k Pk ]E1 Pk Pk E[qk Pk ],
Dk = E[Pk ]E1 [Pk Pk ]E[Pk ],

0
0
1

Jk = E[ek ] [ek Pk ]E [P1k Pk ]E[Pk ],

Mk = E[qk ] E[qk Pk ]E [Pk Pk ]E[Pk ],

(21)
(22)

(23)

(24)

and

k1

fk (zk , yk ) = min E
ps (zs cs ys )2

uk

s=k

+ 2ps (zs cs ys )) (zk , yk )

s.t. (1) and (4).

i =k

() = 1. We give a useful lemma first.

Lemma 1. Suppose that the series {lk } satisfies the recursion formula
lk = lk+1 tk + sk , k = 0, 1, . . . , T 1 and lT is given. Then
T 1

lk = lT
since xs = zs cys . In the following, we adopt the dynamic
programming approach to solve Problem (16).
Let fk (zk , yk ) be the optimal value function of Problem (16)
starting from time k with initial states zk and yk , that is

(17)

(25)

and E1 [Pk Pk ] is the inverse matrix of E[Pk Pk ], i.e., E1 [Pk Pk ] =

(E[Pk Pk ])1 .
We now further derive the computational formulas for the
k1
series wk , hk , k , k , k and gk . For convenience, define i=k () = 0

s.t. (1) and (4)

wT = pT ,
hT = p T ,

where

and d2 2d is fixed, Problem (13) is equivalent to the

(18)

ti +

i=k

T 1
i 1

si

i=k

tj .

(26)

j=k

This lemma can be proved easily by repeated backward

iterations.
By Lemma 1, (19) and (20), we have

T
1
T 1
i1
T
i 1

w
=
p
A
+
p
A
=
p
Aj ,

k
T
i
i
j
i

i=k

i=k

j =k

i =k

j =k

T
1
T 1
i 1
T
i1

h
=
p
J
+
p
J
=
p
Jj ,

k
T
i
i
j
i

i=k

i =k

j =k

i=k

j =k

(27)

88

H. Yao et al. / Insurance: Mathematics and Economics 54 (2014) 8492

for k = 0, 1, . . . , T . After obtaining the expressions for wk and hk ,

by Lemma 1, (21) and (22), we can further express k and k in
terms of wk and hk as follows

i1
T
1
T 1

Cj

c
p
C
+
c
C

c
p
)
(w

k
T
T
i
i
+
1
i
+
1
i
i
i

j=k
i=k
i =k

T 1
i
T
i1

=
wi+1 ci+1
Cj
ci pi
Cj ,

i=k
j=k
i=k
j=k

T 1

h2i+1

k =

Di , k = 0 , 1 , . . . , T .
wi+1
i=k

Then for k, by Bellman equation (18), we obtain

fk (z , y) = pk z 2 + pk ck2 y2 2pk ck zy + 2pk z 2pk ck y

+ min E fk+1 ze0k + ck+1 qk y + Pk uk , qk y

uk

= pk z + pk ck2 y2 2pk ck zy + 2pk z 2pk ck y

2
+ min E wk+1 ze0k + ck+1 qk y + Pk uk + k+1 q2k y2
uk

+ 2k+1 ze0k + ck+1 qk y + Pk uk qk y + k+1 2

+2hk+1 ze0k + ck+1 qk y + Pk uk + 2gk+1 qk y

2

(28)

= pk z 2 + pk ck2 y2 2pk ck zy + 2pk z 2pk ck y

For k = 0, 1, . . . , T 1, let

k = ck2 pk + (wk+1 ck+1 + 2k+1 ) ck+1 Bk

k+1 E[q2k ] Bk ,
wk+1

k+1

k = ck pk + hk+1 ck+1 Mk
(E[qk ] Mk ) .
wk+1

+ wk+1 z 2 E[(e0k )2 ] + wk+1 ck2+1 y2 E[q2k ]

+ 2wk+1 ck+1 yz E[e0k qk ] + k+1 y2 E[q2k ]
(29)

+ 2k+1 yz E[qk e0k ] + 2k+1 ck+1 y2 E[q2k ]

(30)

+ 2hk+1 z E[e0k ] + 2hk+1 ck+1 yE[qk ]

+ 2gk+1 E[qk ]y + k+1 2 + min wk+1 uk

uk

E Pk Pk uk + 2 wk+1 z E[e0k Pk ] + (wk+1 ck+1

+ k+1 ) yE[qk Pk ] + hk+1 E[Pk ] uk .

Then, (23) and (24) can be simplified as follows

k = k+1 E[q2k ] + k ,
gk = gk+1 E[qk ] + k ,

T = cT2 pT ,
gT = cT pT .

By Lemma 1 and (30), for k = 0, 1, . . . , T , we obtain the explicit

expressions for k and gk

i1
T
1
T 1

2
2

E[q2j ],

=
c
E
[
q

p
]
+

k
i
T
T
i

i =k
T 1

i =k
T 1

j =k
i1

(31)

E[qj ].
i
E[qi ] +

gk = cT pT
i=k

i=k

It is known from Proposition 1 that wk+1 > 0.On the other hand,
it is known from Li and Ng (2000) that E Pk Pk is positive definite
under Assumption 1. Therefore, the first order necessary condition
(which is also sufficient) about uk gives the optimal strategy
uk = E1 Pk Pk

It is well known that Assumption 1 implies that

see Li and Ng (2000). By (27), we have the following proposition.

Proposition 1. For k = 0, 1, . . . , T , we have wk > 0.

Theorem 1. Let z = zk and y = yk for simplicity. Then the

solution to Bellman equation (18), namely the optimal value function
of Problem (16), is given by
fk (z , y) = wk z 2 + 2k zy + k y2 + 2hk z + 2gk y + k 2 ,

(32)

where wk , hk , k , k , k and gk are defined by (19)(24) for k =

0, 1, . . . , T .
Proof. We prove this theorem by mathematical induction on k.
For k = T , by the boundary conditions of (19)(24), we have

wT z + 2T zy + T y + 2hT z + 2gT y + T
= pT z 2 2pT cT zy + pT cT2 y2 + 2pT z 2pT cT y.

fT (z , y) = pT z 2pT cT zy +

hk+1

wk+1

k+1
wk+1

pT cT2 y2

E [P k ] .

(34)

fk (z , y) = pk z 2 + pk ck2 y2 2pk ck zy + 2pk z 2pk ck y

+ wk+1 z 2 E[(e0k )2 ] + wk+1 ck2+1 y2 E[q2k ] + 2wk+1 ck+1 yz E[e0k qk ]

+ 2hk+1 z E[e0k ] + 2hk+1 ck+1 yE[qk ] + 2gk+1 E[qk ]y

+ k+1 2 wk+1 z E[e0k Pk ] + (wk+1 ck+1 + k+1 ) yE[qk Pk ]

1
k+1

0
+ hk+1 E[Pk ] E
Pk Pk z E[ek Pk ] + y ck+1 +
wk+1

hk+1
E[qk Pk ] +
E[Pk ] .
wk+1
Simplifying the above formula and by (25), we have
fk (z , y) = (pk + wk+1 Ak ) z 2 + 2 [pk ck + (wk+1 ck+1

+ k+1 ) Ck ] zy + 2 (pk + hk+1 Jk ) z + k+1

On the other hand, according to the boundary condition of Bellman

equation (18), it follows that
2

z E[e0k Pk ] + y ck+1 +

+ k+1 y2 E[q2k ] + 2k+1 yz E[qk e0k ] + 2k+1 ck+1 y2 E[q2k ]

We are now ready to state and to prove the following theorem.

Substituting (34) into (33) yields

Ak = E[(e0k )2 ] [e0k Pk ]E1 [Pk Pk ]E[e0k Pk ] > 0,

E[qk Pk ] +

j =k

(33)

+ 2pT z 2pT cT y.

Therefore, for k = T , (32) holds.

Suppose that for k + 1, (32) holds, i.e.,
fk+1 (z , y) = wk+1 z 2 + k+1 y2 + 2k+1 zy + 2hk+1 z
+ 2gk+1 y + k+1 2 .

h2k+1

wk+1

+ pk ck2 + k+1 E[q2k ] + wk+1 ck2+1 + 2k+1 ck+1 Bk

2
2k+1 2

E[qk ] Bk y
wk+1

+ 2 pk ck + gk+1 E[qk ] + hk+1 ck+1 Mk

k+1

(E[qk ] Mk ) y.
wk+1

Dk

H. Yao et al. / Insurance: Mathematics and Economics 54 (2014) 8492

Then, by (19)(24), it follows that

fk (z , y) = wk z + 2k czy + k c y + 2hk z + 2gk cy + k .
2

2 2

This means that (31) holds for k. By the Principle of Mathematical

Induction (PMI), (32) holds true for k = 0, 1, . . . , T , and the
theorem is proved. 

Again substituting (38) into (37) and noticing that z0 = x0 + c0 y0 ,

the optimal value of the meanvariance model (10), namely, the
minimum variance is obtained as follows
Var [xT ] =

1 + 0

1
1 + 0

(35)

where z0 = x0 + cy0 . By Theorem 1 and (35), it follows that

+ 2h0 z0 + 2g0 y0 + 0 2 d2 2d
= 0 2 + 2 (h0 z0 + g0 y0 d) + w0 z02

(37)

Proposition 2. k < 0 for k = 0, 1, . . . , T 1.

Proof. It is known from Li and Ng (2000) that under Assumption 1,
E[Pk Pk ] is positive definite. Then so is E1 [Pk Pk ]. By Assumption 4,

then Dk = E[Pk ]E1 [Pk Pk ]E[Pk ] > 0. By (9), pT > 0,

E[Pk ] = 0,
and by Proposition 1, wk+1 > 0 for k = 0, 1, . . . , T 1. Therefore,
by (27) and (28), we have

Proposition 2 shows that 0 < 0. Therefore, by (36), the optimal

solution of optimization problem (37) exists. By the first-order
condition, we obtain the optimal solution as follows

(xk + ck yk ) E[e0k Pk ]

k+1
+ yk ck+1 +
E[qk Pk ]
wk+1
1

uk = E

Pk Pk

(h0 (x0 + c0 y0 ) + g0 y0 d) hk+1

E[Pk ] .
0 wk+1

1
1 + 0

(h0 (x0 + c0 y0 ) + g0 y0 )2 .

(41)

We summarize the above results in the following theorem.

The model in the previous section is referred to as the General

case. In this section, we discuss some special cases of our model.
Special case 1: The case of no pension contribution. In this case,
we need only to set ck = 0 for k = 0, 1, . . . , T . Our model
degenerates to an ordinary multi-period meanvariance portfolio
selection model. It is known from (28) that k = 0 for k =
0, 1, . . . , T . Then by (29), k = k = 0 for k = 0, 1, . . . , T 1.
Hence, by (31), it follows that k = gk = 0 for k = 0, 1, . . . , T .
Therefore, in this case, by (39) and (40), the efficient investment
strategy and the efficient frontier can be simplified as
uk = E1 Pk Pk

xk E[e0k Pk ]

1 + 0

+ w0

(h0 x0 d) hk+1
E[Pk ] ,
0 wk+1

(42)

h0 x 0

1 + 0

h20

1 + 0

x20 ,

(43)

respectively, where wk , hk and k are also given by (27) and (28).

Special case 2: The terminal time is deterministic. We need only
to set the mortality intensity (s) = 0 on [0, T ]. Then by (9), it
follows that
i = 1, 2, . . . , T 1; pT = 1.

In this case, wk , hk , k , k , k and gk can be simplified as

T
1
T
1
T 1
T
1 J 2

Di ,
wk =
Ai , hk =
Ji , k =

i=k
i =k
i =k
j=i+1 j

T
1
T 1
i

c
C
+
w
c
Cj ,
k
T
i
i+1 i+1

i=k

(39)

and

pi = 0 ,
(38)

Substituting (38) into (34) and noticing that z = zk , y = yk and

zk = xk +ck yk , we obtain the optimal strategy of the meanvariance
model (10), namely the efficient investment strategy as follows

+ 0 y20

Var [xT ] =

T 1

h2i+1
h2
Di T DT 1 = pT DT 1 < 0,
wi+1
wT
i=k

h0 z0 + g0 y0 d

Varmin [xT ] := w0 (x0 + c0 y0 )2 + 20 (x0 + c0 y0 ) y0

for k = 0, 1, . . . , T 1. This completes the proof of the proposition.



we can obtain the global

5. Some special cases

In order to show the existence of the solution to for Problem (37),

we give the following proposition first.

k =

h0 (x0 +c0 y0 )+g0 y0

,
1+0

(40)

(36)

According to the Lagrange dual theory (see Luenberger (1968)),

the optimal value of Problem (10) (which is equivalent to Problem
(12)) can be obtained by maximizing H (z0 , y0 , ) over , i.e.,
Var [xT ] = max H (z0 , y0 , ).

1 + 0

Theorem 3. For a given expected terminal wealth E[xT ] = d

(d dmin ), the efficient investment strategy and the efficient frontier
of the multi-period meanvariance DC pension funds investment
management problem (10) with stochastic income and mortality risk
are given by (39) and (40), respectively.

H (z0 , y0 ) = w0 z02 + 20 z0 y0 + 0 y20

+ 20 z0 y0 + 0 y20 d2 .

(h0 (x0 + c0 y0 ) + g0 y0 )2 .

Setting d = dmin :=
minimum variance

By the previous analysis in Section 3, we know that the optimal

value of Problem (13) is

h0 (x0 + c0 y0 ) + g0 y0

+ w0 (x0 + c0 y0 ) + 20 (x0 + c0 y0 ) y0 + 0 y20

Theorem 2. The optimal strategy for Problem (16) is given by (34).

H (z0 , y0 , ) = f0 (z0 , y0 ) d 2d,

By the proof of Theorem 1, we have the following result.

4. Efficient investment strategy and efficient frontier

89

i =k

j =k

T
1
T 1
i1

k = cT2
E[q2i ] +
i
E[q2j ],

i
=
k
i
=
k
j
=
k

T
1
T 1
i 1

gk = cT
E[qi ] +
i
E[qj ],
i=k

i=k

j =k

(44)

90

H. Yao et al. / Insurance: Mathematics and Economics 54 (2014) 8492

where

Substituting the above data into (27), (28) and (31) gives

k = (wk+1 ck+1 + 2k+1 ) ck+1 Bk k+1 E[q2k ] Bk ,

wk+1

k+1

k = hk+1 ck+1 Mk
(E[qk ] Mk ) .
wk+1

= 0.0127,
0 = 2.7557,
g0 = 0.8813,
0

=
(
2
.
5164
,
2
.
0926
,
1
.
6610
,
1
.
3097
,

1.0243, 0.7932, 0.6065),

= (1.9533, 1.2562, 0.7611, 0.4054,

0.1560, 0.0128, 0.1213),

h = (1.1807, 1.1405, 1.0097, 0.8922,

0.7866, 0.6917, 0.6065),

(45)

The efficient investment strategy and the efficient frontier are also
given by (39) and (40), respectively.
Special case 3: The case with a risk-free asset in the market.
Suppose that the 0th asset is the risk-free asset. Then e0k is a nonstochastic constant, k = 0, 1, . . . , T 1. By (25), we have
Ak = (e0k )2 (1 Dk ) ,

Ck = e0k (E[qk ] Dk ) ,

Jk = e0k (1 Dk ) .

(46)

where w
= (w0 , w1 , . . . , w6 ), = (0 , 1 , . . . , 6 ), h =
(h0 , h1 , . . . , h6 ). Then, substituting (48) into (39), we obtain the
efficient investment strategy as

uk =

According to (27), (28) and (46), the expressions of wk , hk and k

can be rewritten as

T
i 1

wk =
pi
(e0j )2 1 Dj ,

i =k
j =k

i1
T

e0j 1 Dj ,
p
h
=

i
k

i=k

j =k

(47)

j =k

In this case, k , k and gk are again given by (28) and (31). In

addition, the efficient investment strategy and the efficient frontier
are still given by (39) and (40), respectively.

In this section, we provide some numerical analyses to illustrate

the results obtained in this paper.
Consider a pension fund member who enters a DC pension fund
scheme at time 0 with an initial fund paid x0 = 3 and plans to
retire at time T = 6. Suppose that his initial income is y0 = 1 and
he needs to contribute 20% of his income at the beginning of every
period, i.e., ck = 0.2 for k = 0, 1, . . . , T . Due to the mortality risk,
the actual terminating time of his pension fund scheme is T . The
probability distribution of T is defined by (5)(9). For simplicity,
we assume that the mortality intensity (s) is independent of time
s, and always equal to 0.1, i.e., (s) = 0.1 for s [0, T ]. Then by
(9), The probability distribution of T is given by
p1 = 0.0952,

p2 = 0.0861,

p3 = 0.0779,

p4 = 0.0705,

p5 = 0.0638,

p6 = 0.6065.

Suppose that the pension fund can be invested in four risky assets
(indexed as 0, 1, 2, 3). For convenience, suppose that the market
parameters are independent of time k and are listed as follows

E[Pk Pk ] =

0.2365
0.0719
0.1184

0.0719
0.3449
0.1378

0.1184
0.1378 ,
0.3262

E[
] = 1.1689,
E[( ) ] = 1.2468,
E[ ] = 1.0430,
E[qk ] = 1.0284,
E[q2k ] = 1.4257,

E[qk Pk ] = (0.0083, 0.0220, 0.0657) ,

E[Pk ] = (0.0255, 0.0015, 0.0004) ,
E[e0k Pk ] = (0.0827, 0.0924, 0.0446) .
e0k qk
e0k

(4.6595 d) hk+1
+
0.0127wk+1

0.1342
0.0150 ,
0.0437

Var [xT ] = 77.9389 (d 4.7192)2 + 19.0355.

In the following, we discuss some special cases. The related
parameters are the same as above unless stated otherwise.
Special case 1: The case without pension contribution. In this
case, ck = 0 for k = 0, 1, . . . , T . Then by (42) and (43), we obtain
the efficient investment strategy

uk =

0.3176
(3.5421 d) hk+1
0.2323 xk +
0.0127wk+1
0.0766

0.1342
0.0150 ,
0.0437

and the efficient frontier

6. Numerical illustration

0.1654

0.3176
k+1
0.0075
0.2323 (xk + 0.2yk ) + yk 0.2 +
wk+1
0.2646
0.0766

where the values of wk+1 , k+1 and hk+1 (k = 0, 1, . . . , 5) are

given by (48). Plugging (48) into (40), we obtain the efficient
frontier as

T 1
i

wi+1 ci+1
e0j E[qj ] Dj
k =

i =k
j =k

i 1
T

e0j E[qj ] Dj .
ci pi

i=k

(48)

e0k 2

Var [xT ] = 77.9389 (d 3.5875)2 + 9.9403,

where the values of wk+1 and hk+1 (k = 0, 1, . . . , 5) are also
given by (48). From Fig. 1, compared with the General case, we find
that the efficient frontier shifts to the lower left but the shape of
the efficient frontier is invariant. It is not difficult to understand
this. Because in this case there is no pension contribution, and so
the wealth accumulation for the pension fund is reduced. So, for a
given risk (measured by variance) level of its terminal wealth, the
pension fund has lower expected terminal wealth level. Therefore,
the efficient frontier moves down.
Special case 2: The terminal time of the pension fund is
deterministic. In this case (s) = 0 for all s [0, T ]. Then by
(9) pi = 0, i = 0, 1, . . . , 5, p6 = 1. Substituting the data into (44),
we have

0 = 0.0159,
0 = 4.1160,
g0 = 1.1921,

w
= (3.0239, 2.5147, 2.0911, 1.7389, 1.4461, 1.2025, 1),
(49)
= (2.7619, 1.8650, 1.1807, 0.6645, 0.2805, 0, 0.2),

h = (1.2310, 1.1891, 1.1486, 1.1095, 1.0717, 1.0352, 1).

Then by (49), (39) and (40), we obtain the efficient investment
strategy

uk =

0.1654

0.3176
k+1
0.0075
0.2323 (xk + 0.2yk ) + yk 0.2 +
wk+1
0.2646
0.0766

(5.1312 d) hk+1
+
0.0159wk+1

0.1342
0.0150 ,
0.0437

H. Yao et al. / Insurance: Mathematics and Economics 54 (2014) 8492

91

minimum variance are also strictly larger than 0. But compared

with the global minimum variance 19.0355 in the General case, the
global minimum variance 1.1508 in this case has been dramatically
reduced.
Fig. 1 indicates that when the expected terminal wealth is
larger than that corresponding to the intersection point of the two
efficient frontiers in the General case and Special case 3, for the
same expected terminal wealth level, the variance of the General
case is smaller than that of Special case 3. It is also not difficult to
understand this. That is because there is one more risky asset in the
General case, namely there is one more risk source in the General
case. Obviously, inclusion of one more risk source is beneficial to
hedge the integral risk of the pension fund. This explains the fact
that when the expected terminal wealth is big enough, the variance
of the terminal wealth is smaller than that of Special case 3.
7. Conclusion
Fig. 1. Efficient frontiers for the General case and Special cases 13.

and the efficient frontier

Var [xT ] = 61.9407 (d 5.2141)2 + 26.0029,
where the values of wk+1 , k+1 and hk+1 (k = 0, 1, . . . , 5) are
given by (49). Fig. 1 shows that the efficient frontier moves to
the upper right compared with the General case. The economic
implications are stated as follows. Since there is no mortality
risk in this case, the pension scheme would not be forced to
terminate before the retirement of the member. Then, compared
with the General case, it takes longer time for the pension fund
to accumulate its wealth. Therefore, for a given risk level of the
terminal wealth, the pension fund has higher expected terminal
wealth level. This explains the upward movement of the efficient
frontier in this case.
Special case 3: There is a risk-free asset in the market. Suppose
that the 0th asset is risk-free. Then, e0k is a constant. Suppose that
e0k = 0.015, k = 0, 1, . . . , 5, and the market parameters for the
other three assets are the same as above. In this case, the efficient
frontier is
Var [xT ] = 82.8673 (d 4.5540)2 + 1.1508,

uk =

0.0952
k+1
0.0412 (xk + 0.2yk ) yk 0.2 +
w
k+1
0.0610

(4.4997 d) hk+1 0.0924

0.0400 ,
+
0.0119wk+1
0.0592

0.1019
0.2394
0.3028

References

where the values of wk+1 , k+1 and hk+1 (k = 0, 1, . . . , 5) are

given as follows

w
= (1.3142, 1.2419, 1.0837, 0.9427,

0.8172, 0.7056, 0.6065),

= (1.0139, 0.7147, 0.4802, 0.2839,

0.1210, 0.0128, 0.1213),

= (1.1380, 1.1077, 0.9855, 0.8754,

h

0.7762, 0.6869, 0.6065).

Acknowledgments
The authors are grateful to the anonymous referee(s) for giving
them very useful suggestions and comments.

and the efficient investment strategy is

Starting from the actual requirements of asset allocation for

DC pension funds, we consider synthetically the two objectives
to enhance the yield and to control the risk for a problem of
DC pension fund investment management in this paper. More
specifically, we establish the multi-period meanvariance asset
allocation model for a DC pension fund with stochastic income and
mortality risk in this paper. By adopting the dynamic programming
method and the Lagrange dual theory, we obtain the closedform expressions for the efficient investment strategy and the
efficient frontier. We also discuss some special cases of our model.
Finally, we present a numerical example to illustrate the results
obtained in this paper. We find some interesting results via
numerical analyses, and also give the corresponding economic
interpretations. Our model can be extended in several ways. For
example, we can extend our model to the cases with more realistic
conditions, such as stochastic market environment, parameter
uncertainty, and minimum guarantee protection. Using the multiperiod meanvariance model to study DB pension fund investment
management problems is another research topic in the near future.

(50)

We point out that although there is a risk-free asset in the market

in this case, the global minimum variance Varmin = 1.1508 is
strictly larger than zero. The reason is as follows: not only the
risky asset can cause risk, but also the two factors, stochastic
income and mortality risk, can generate the integral risk during
the accumulation process of the pension fund. So, even if all the
wealth is invested in the risk-free asset in this case, the global

Blake, D., Wright, D., Zhang, Y., 2013. Target-driven investing: optimal investment
strategies in defined contribution pension plans under loss aversion. J. Econom.
Dynam. Control 37, 195209.
Cairns, A.J.G., Blake, D., Dowd, K., 2006. Stochastic lifestyling: optimal dynamic asset
allocation for defined contribution pension plans. J. Econom. Dynam. Control
30, 843877.
Charupat, N., Milevsky, M.A., 2002. Optimal asset allocation in life annuities: a note.
Insurance Math. Econom. 30, 199209.
Chen, P., Yang, H.L., Yin, G., 2008. Markowitzs meanvariance assetliability
management with regime switching: a continuous-time model. Insurance
Math. Econom. 43, 456465.
Chiu, M.C., Wong, H.Y., 2013. Meanvariance principle of managing cointegrated
risky assets and random liabilities. Oper. Res. Lett. 41, 98106.
Christophette, B.S., Nicole, E.K., Monique, J., Lionel, M., 2008. Optimal investment
ecisions when time-horizon is uncertain. J. Math. Econom. 44, 1001113.
Costa, O.L.V., Oliveira, A.D., 2012. Optimal meanvariance control for discrete- time
linear systems with Markovian jumps and multiplicative noises. Automatica 48,
304315.
Deelstra, G., Grasselli, M., Koehl, P.F., 2003. Optimal investment strategies in the
presence of a minimum guarantee. Insurance Math. Econom. 33, 189207.
Delong, , Gerrard, R., Haberman, S., 2008. Meanvariance optimization problems
for an accumulation phase in a defined benefit plan. Insurance Math. Econom.
42, 107118.
Emms, P., 2012. Lifetime investment and consumption using a defined-contribution
pension scheme. J. Econom. Dynam. Control 36, 13031321.
Fu, C.P., Ali, L.L., Li, X., 2010. Dynamic meanvariance portfolio selection with
borrowing constraint. European J. Oper. Res. 200, 313319.
Gao, J.W., 2009. Optimal portfolios for DC pension plans under a CEV model.
Insurance Math. Econom. 44, 479490.

92

H. Yao et al. / Insurance: Mathematics and Economics 54 (2014) 8492

Gerrard, R., Haberman, S., Vigna, E., 2004. Optimal investment choices postretirement in a defined contribution pension scheme. Insurance Math. Econom.
35, 321342.
Gerrard, R., Hjgaard, B., Vigna, E., 2012. Choosing the optimal annuitization time
post retirement. Quant. Finance 12 (7), 11431159.
Giacinto, D.M., Federico, S., Gozzi, F., 2011. Pension funds with a minimum
guarantee: a stochastic control approach. Finance Stoch. 15, 297342.
Haberman, S., Vigna, E., 2002. Optimal investment strategies and risk measures in
defined contribution pension schemes. Insurance Math. Econom. 31, 3569.
Hainaut, D., Deelstra, G., 2011. Optimal funding of defined benefit pension plans. J.
Pension Econ. Finance 10 (1), 3152.
Hainaut, D., Devolder, P., 2007. Management of a pension fund under mortality and
financial risks. Insurance Math. Econom. 41, 134155.
Han, N.W., Hung, M.W., 2012. Optimal asset allocation for DC pension plans under
inflation. Insurance Math. Econom. 51, 172181.
Josa-Fombellida, R., Rincn-Zapatero, J.P., 2008. Meanvariance portfolio and
contribution selection in stochastic pension funding. European J. Oper. Res. 187,
120137.
Li, D., Ng, W.L., 2000. Optimal dynamic portfolio selection: multiperiod
meanvariance formulation. Math. Finance 10, 387406.
Luenberger, D.G., 1968. Optimization by Vector Space Methods. Wiley, New York.
Ma, Q.P., 2011. On optimal pension management in a stochastic framework with
exponential utility. Insurance Math. Econom. 49, 6169.

Markowitz, H., 1952. Portfolio selection. J. Finance 7 (1), 7791.

Martellini, L., Urosevic, B., 2006. Static meanvariance analysis with uncertain time
horizon. Manag. Sci. 52, 955964.
Pliska, S.R., Ye, J.C., 2007. Optimal life insurance purchase and consumption/investment under uncertain lifetime. J. Bank. Finance 31, 13071319.
Vigna, E., 2012. On efficiency of meanvariance based portfolio selection in defined
contribution pension schemes. Quant. Finance 122. iFirst.
Vigna, E., Haberman, S., 2001. Optimal investment strategy for defined contribution
pension schemes. Insurance Math. Econom. 28, 233262.
Wu, H.L., Li, Z.F., 2011. Multi-period meanvariance-portfolio selection with
Markov regime switching and uncertain time-horizon. J. Syst. Sci. Complexity
24, 140155.
Yao, H.X., Lai, Y.Z., Li, Y., 2013a. Continuous-time meanvariance assetliability
management with endogenous liabilities. Insurance Math. Econom. 52,
617.
Yao, H.X., Zeng, Y., Chen, S.M., 2013b. Multi-period meanvariance assetliability
management with uncontrolled cash flow and uncertain time-horizon. Econ.
Model. 30, 492500.
Zhang, A., Ewald, C.O., 2010. Optimal investment for a pension fund under inflation
risk. Math. Methods Oper. Res. 71, 353369.
Zhou, X.Y., Li, D., 2000. Continuous-Time meanvariance portfolio selection: a
stochastic LQ framework. Appl. Math. Optim. 42, 1933.