Edith Cowan University Edith Cowan University

Research Online Research Online

Theses : Honours Theses

2006

Age-phasing and the use of life-cycle funds Age-phasing and the use of life-cycle funds

Shelley Farr Edith Cowan University

Follow this and additional works at: https://ro.ecu.edu.au/theses_hons

Part of the Family, Life Course, and Society Commons

Recommended Citation Recommended Citation Farr, S. (2006). Age-phasing and the use of life-cycle funds. https://ro.ecu.edu.au/theses_hons/1067

This Thesis is posted at Research Online. https://ro.ecu.edu.au/theses_hons/1067

Edith Cowan University  

 

Copyright Warning      

You may print or download ONE copy of this document for the purpose 

of your own research or study.  

The University does not authorize you to copy, communicate or 

otherwise make available electronically to any other person any 

copyright material contained on this site.  

You are reminded of the following:  

Copyright owners are entitled to take legal action against persons who infringe their copyright. 

 

A reproduction of material that is protected by copyright may be a 

copyright infringement. Where the reproduction of such material is 

done without attribution of authorship, with false attribution of 

authorship or the authorship is treated in a derogatory manner, 

this may be a breach of the author’s moral rights contained in Part 

IX of the Copyright Act 1968 (Cth). 

 

Courts have the power to impose a wide range of civil and criminal 

sanctions for infringement of copyright, infringement of moral 

rights and other offences under the Copyright Act 1968 (Cth). 

Higher penalties may apply, and higher damages may be awarded, 

for offences and infringements involving the conversion of material 

into digital or electronic form.

AGE-PHASING AND THE USE OF LIFE-CYCLE FUNDS

Shelley Farr Bachelor of Business (Finance & Accounting)

This thesis is presented in fulfilment of the requirements for the degree of Bachelor of Business (Honours)

Faculty of Business & Law Edith Cowan University

February,2006

USE OF THESIS

The Use of Thesis statement is not included in this version of the thesis.

ABSTRACT

The Superannuation Guarantee legislation has made many Australian employees

compulsory investors. The reality that many Australian employees are failing to save

adequate retirement benefits highlights the importance of selecting an appropriate

superannuation investment strategy. With a majority of members having their employer­

sponsored contributions in Defined Contribution Funds, it is ultimately members who

are responsible for making investment decisions. Given that Australian employees are

faced with myriad investment options, it is opportune to examine how members are

exercising investment choice. A key factor for a member to consider is whether their

investment strategy should be influenced by their age.

Using the member asset allocation data from four of Australia's larger superannuation

funds (HESTA, STA, GESB, and UniSuper), this thesis provides an empirical

investigation of a sample of members' investment strategy decisions and key

demographics, notably age. This thesis also presents an overview of life-cycle funds, an

emerging product in the superannuation industry. General comparisons between existing

life-cycle funds in terms of asset allocations, cost structures, and investment strategy,

allow for an understanding of the range of funds available.

Several key findings emerge. There is a considerable amount of variation among

the life-cycle products being offered in the United States and Australia. A comparison

of the asset allocation weights between the Australian life-cycle products and the default

investment options of Australian superannuation funds reveals that life-cycle products

decrease individuals' exposure to growth assets very early in the life-cycle. Age has a

significant relationship with the portfolio decisions, equity participation and equity

allocation. When additional member characteristics are taken into account including

gender, marital status, and risk profile, the relative importance of the age factor

diminishes significantly.

iii

DECLARATION

I certify that this thesis does not, to the best of my knowledge and belief:

(i) incorporate without acknowledgment any material previously submitted for a

degree or diploma in any institution of higher education.

(ii) contain any material previously published or written by another person except

where due reference is made in the text; or

(iii) contain any defamatory material.

I also grant permission for the Library at Edith Cowan University to make duplicate

copies of my thesis as required.

Signature: ..

Date: .....

iv

ACKNOWLEDGEMENTS

Thank you to my Honours Supervisor, Paul Gerrans, for his guidance, time, and

assistance throughout the Honours Program. I gratefully acknowledge Jacqui Whale for

her assistance with organising the superannuation data. I thank the Lecturers and Tutors

I had over the course of my studies for imparting their knowledge to me, with special

mention to Mahendra Chandra. I thank my family for their patience and encouragement.

Successful completion of this thesis was possible due to the support and guidance from

the abovementioned people.

v

TABLE OF CONTENTS

USE OF THESIS ABSTRACT DECLARATION ACKNOWLEDGEMENTS 1. INTRODUCTION

ii iii iv v 1

1.1. Retirement Investment Strategy and Age ......................................................... 3 1.2. Research Contribution and Questions .............................................................. 4 1.3. Findings Overview ........................................................................................... 7 1.4. Thesis Structure ................................................................................................ 7

2. LITERATURE REVIEW 8 2.1. Theoretical Literature ....................................................................................... 8

2.1.1. Modem Portfolio Theory ................................................................... 8 2.1.2. Optimal Lifetime Asset Allocation .................................................... 9 2.1.3. Asset Allocation and Annuitisation ................................................. 12

2.2. Empirical Literature ....................................................................................... 12 2.2.1. Age-Specific Patterns in Asset Allocation ....................................... 12 2.2.2. Labour Supply Flexibility and Asset Allocation .............................. 15 2.2.3. Australian Evidence on Age-Specific Patterns in Asset Allocation 15

3. SURVEY OF LIFE-CYCLE FUNDS 16 3.1. The U.S. Experience ....................................................................................... 16 3.2. Retirement Saving Behavioural Problems ..................................................... 19 3.3. A Comparison of Target-Date Funds ............................................................. 21

3.3.1. Asset Allocation ............................................................................... 21 3.3.2. Asset Holdings ................................................................................. 22 3.3.3. U.S. Expense Ratios ......................................................................... 23

3 .4. The Australian Experience ............................................................................. 23 3.5. Life-Cycle Funds as a Default Option ............................................................ 25 3.6. Limitations ofthe Life-Cycle Investment Strategy ........................................ 28 3.7. Summary ........................................................................................................ 30

4. DATA AND RESEARCH METHODS 31 4.1. Data Description and Definition of Variables ................................................ 31 4.2. Description of Funds and Investment Strategy Choices ................................ 34 4.3. Research Methods .......................................................................................... 35

4.3.1. Preliminary Analysis ........................................................................ 35 4.3.2. Regression Analysis ......................................................................... 35

5. FINDINGS AND DISCUSSION 38 5 .1. Preliminary Results ........................................................................................ 3 8

5.1.1. Equity Allocations ............................................................................ 38 5.2. Regression Results ......................................................................................... 45

5.2.1. Effect of Age on Equity Participation and Equity Allocation .......... 45 5.2.2. Effect of Age Relative to other Variables on Equity Participation and Equity Allocation ........................................................................................... 52

6. CONCLUSION 62 7. REFERENCES 65 8. APPENDICES 69

vi

1. INTRODUCTION

Superannuation fund assets now total $945.6 billion (Australian Prudential

Regulation Authority, 2006). Superannuation has played an increasingly important role

in the managed funds industry in the past two decades, due to a significant restructuring

of Australia's retirement income policy in the 1980's and 1990's (AXISS Australia,

2004).

Prior to the introduction of a number of compulsory superannuation initiatives in

the late 1980's and early 1990's, a majority of fund members were in a Defined Benefit

Fund (DBF) (Drew & Stanford, 2003). In a DBF, the employer bears the risk of

investments and promises to fund a defined benefit upon retirement, usually in terms of

an employee's final salary and years of employment. That is, the employer increases

contributions when investment returns are low, and decreases contributions when

investment returns are high (AXISS Australia, 2004). In contrast, a member in a

Defined Contribution Fund (DCF) has an individual account where their balance varies

with the investment choice and movements in the market.

The beginnings of compulsory superannuation were initially introduced in

Australia in 1986 through 'Award Super,' which required many employees to accept a

minimum three percent superannuation contribution in exchange for foregoing a wage

increase (AXISS Australia, 2004). The introduction of the Superannuation Guarantee

(Administration) Act 199i made many more Australian employees compulsory

investors and was largely responsible for mass compulsory superannuation coverage,

with coverage increasing from approximately 40 percent in the 1980's to over 90

percent in recent years (Drew & Stanford, 2003).

Under the Superannuation Guarantee legislation, it is mandatory for employers

to make contributions of a specified proportion of wage and salaries to a complying

superannuation fund, on behalf of employees2• Employees who are under the age of 70

and are earning more than $450 per calendar month are eligible for superannuation

1 Generally referred to as the Superannuation Guarantee (SG) legislation. 2 Payments must be made at least quarterly.

1

guarantee contributions3. The minimum contribution levy was originally set at three

percent in 1992, and has progressively increased to the current rate of nine percent.

Australia's current superannuation system has shifted to one that is characterised

by compulsion and DCF structures. DCF's became increasingly attractive to both

employers and employees. For employers, DCF's are preferable as DBF contribution

requirements may exceed the minimum contribution level required by SG legislation to

maintain benefit levels. Such schemes offer practicality to employees, especially those

who regularly transfer between jobs (AXISS Australia, 2004). As a result of this trend

towards DCF 's, investment risk is now being borne by members to a large extent. The

operation of DCF's is similar to a bank account, whereby employees are paid at least

nine percent of their annual salary into a superannuation account. The retirement benefit

consists of the accumulated values of these contributions, taking into consideration any

profits/losses determined by market movements. Thus, the key difference between

DBF's and DCF's is who bears the investment risk. At present, 62 percent of

superannuation fund assets are in DCF's, and approximately three percent of

superannuation fund assets are in DBF's, with the balance in hybrid schemes

(Australian Prudential Regulation Authority, 2006, p. 16).

A recent change to Australia's current superannuation system was the

Commonwealth choice of fund legislation4, which took effect from 1 July 2005. Under

this legislation, employer contributions are directed to a complying superannuation fund

of the employee's choice. From a potential 9.5 million employees, approximately 5.2

million additional employees gained the right to choose their superannuation fund upon

the introduction of the choice of fund legislation (Clare, 2005).

Another choice being offered to members is within-fund investment choice. The

Australian superannuation industry has followed a global trend of investment choice

expansion. For many, this has resulted in greater choice and control over the allocation

of pension assets (Gallery, Gallery, & Brown, 2004), catering for a wider range of

individual risk-return preferences.

The Superannuation Guarantee legislation has made Australian employees

compulsory investors. With a majority of members having their employer-sponsored

3 The SG obligation does not apply to (1) self-employed persons; (2) employees earning less than $450 per calendar month; and (3) employees aged 70 or over. 4 Superannuation Legislation Amendment (Choice of Superannuation Fund) Act 2004

2

contributions in DCF's, it is ultimately members who are responsible for making

investment decisions. Given that Australian employees are faced with myriad

investment options, it is opportune to examine how members are exercising investment

choice.

1.1. Retirement Investment Strategy and Age

A key factor for a member to consider is whether their investment strategy

should be influenced by their age. A popular rule of thumb with respect to retirement

investing is that the proportion of an individual's portfolio invested in equities should be

equal to 100 minus the investor's age (Bodie & Crane, 1997). Thus, a person who is 40

years old should invest 60% in equities, and a person who is 60 years old should invest

40% in equities. This strategy can be referred to as age-phasing, whereby individuals

reduce their exposure to risky assets as they become older.

In response to the perceived desire of investors to reduce their equity exposure

within their investment portfolio as they age, and to address the problems of inertia in

retirement asset allocation portfolio decisions, "life-cycle" funds (also known as "target­

date" funds) have been created (Poterba, Rauh, Venti, & Wise, 2005, p. 2). Such funds

"are one of the most rapidly growing financial products of the last decade," (Poterba,

Rauh, Venti, & Wise, 2005, p. 2).

Consistent with the concept of age-phasing, the main idea of a life-cycle fund is

to reallocate investments over time to be more conservative. Life-cycle funds are

managed funds that automatically reduce the exposure to more volatile asset classes,

primarily equity, as an investor ages. In choosing a life-cycle fund an investor selects an

appropriate target-date which refers to an individual's expected year of retirement. The

exposure to growth assets will gradually decrease until this particular date when asset

allocation will become most conservative.

Life-cycle funds first emerged in the U.S. investment environment during the

mid-1990s, and have experienced a significant amount of growth in recent years. At the

end of 2002, the U.S. target-date fund population had a total $15 billion in assets under

management. Over the three years to 2005, this grew to a total $70 billion in assets

under management. In addition to the significant growth of life-cycle funds in the

United States over the last decade, Ameriks & Zeldes (2004) note that the proliferation

3

of age-phasing advice has stimulated interest among academics in at least two

questions: (1) should individuals follow this advice and reduce their exposure to risk as

they age, and (2) do individuals actually follow this advice. This paper is concerned

with the second question, which will be investigated in the context of member choices

of four large Australian superannuation funds.

1.2. Research Contribution and Questions

The key contribution of this thesis is the survey of life-cycle funds which is an

emerging product in the superannuation industry. Being an emerging product, very little

information about life-cycle funds is available in Australia. Some recent U.S. based

studies have briefly referred to life-cycle products. A recent paper published by the

Boston-based Financial Research Corporation (FRC)5 is the only known study that

provides a comprehensive treatment of these products and is solely U.S. based. General

comparisons between existing life-cycle funds in terms of asset allocations, cost

structures, and investment strategy, will allow for an understanding of the range of

funds available, and to determine whether there is consistency among these products.

This thesis provides a comparison of a sample of individual superannuation

member asset allocation decisions at different ages with existing life-cycle products.

This will provide information on asset allocation patterns of members over the life­

cycle, allowing for some evidence as to the potential suitability of the application of

life-cycle funds in Australia.

Second, this thesis contributes to the existing empirical literature concernmg

life-cycle asset allocation. A study of this nature is important to Australia as most of the

empirical literature is limited to the U.S. Even in the U.S., few empirical studies

examine life-cycle patterns in portfolio choice until recently, partly due to a lack of

detailed data (Ameriks & Zeldes, 2004). The asset allocation decision is one that is

significant to all investors who must choose how to apportion total portfolio wealth to

various asset classes. Superannuation members face a number of risks, which include

(Rice Walker Actuaries, 2006, p. 2):

5 Lifecycle Fund Economics: Evaluating Next Generation Competitive Dynamics (FRC, 2006). This study provides discussion relating to the growth oflife-cycle funds, life-cycle fund differentiators (i.e. key investment-related and marketing features), the distribution landscape for life-cycle funds, life-cycle product economies, life-cycle portfolio growth in alternative product strategies (e.g. variable annuity products and 529 college savings plans), and success factors in the life-cycle fund market.

4

" Employer-contributions will be insufficient to fund an adequate retirement

benefit;

111 Retirement savings will not earn sufficient returns to generate a reasonable

retirement benefit;

111 Disposable income will not increase in the immediate period before retirement,

resulting in an inability to top-up retirement savings; and

1111 Taking an early retirement and thus losing a few years of retirement savings

accumulation and having a few extra years of consumption.

The above risks, in addition to the reality that approximately one fifth of

Australia's semi and fully retired population aged over 55 have failed to save for

retirement (Egan, 2006), highlights the fact that selecting an appropriate superannuation

investment strategy is one of the most important financial decisions members can make.

In terms of research methodology, a majority of studies have not distinguished

between the two portfolio decisions of equity participation and equity allocation.

Following Ameriks & Zeldes (2004), these two portfolio decisions were modelled

separately, accounting for the reality that many superannuation members are not

exposed to any level of equity at all. Also, some studies [see for example, Bodie &

Crane (1997); Schooley & Worden (1999)] have employed the Ordinary Least Squares

(OLS) procedure to estimate the general relationship between age and equity allocation.

This approach was not used in this thesis. Instead, the Probit and Tobit procedures were

employed in working with limited dependent variables to compensate for the boundaries

at zero and one for equity participation and zero and 100 for the share of equity held in a

portfolio.

Finally, this thesis offers some conclusions about the relative importance of the

age variable in explaining equity ownership and equity allocation when additional

member characteristics are taken into account. This thesis is of relevance to fund

trustees, fund members, regulators, and employers.

First, under the Superannuation Industry (Supervision) Act 1993, fund trustees

are solely responsible and directly accountable for the prudential management of the

investment of an entity's assets. One of the key duties of fund trustees is to ensure that

their powers are exercised in the best interests of the members. This includes the

5

optimal construction and management of investment menus with members' interests in

mind. The discussion regarding life-cycle funds in this thesis will provide a helpful

information source for fund trustees who may be considering offering life-cycle

products.

Second, superannuation fund members themselves who are interested in

assistance beyond their default investment option will find this thesis useful. It will

allow them to compare their own asset allocation behaviour with that of other

Australian superannuation members, and expose them to (or increase their awareness

of) life-cycle funds.

Third, this thesis will assist funds in the creation of new products to cater for

specific investor needs, for example life-cycle investing. The summary of

superannuation member asset allocations and estimated sensitivities will be particularly

helpful to funds, in gauging the potential demand for products that support an age-based

investment strategy.

Finally, this thesis will provide information to employers in assessing whether or

not to offer life-cycle products to their employees.

This thesis addresses the following questions:

1. What is the status of the life-cycle funds markets in the United States and

Australia?

2. What does the investment literature and financial planning industry suggest

investors should do with their portfolio allocation as they age?

3. Using a sample of member choices from four large Australian superannuation

funds, how different are actual asset allocations to the life-cycle products

available?

4. How does age affect the likelihood of stock ownership in super accounts and

conditional on stock ownership, how does age affect the allocation of stock

within superannuation accounts?

5. Using additional demographic data, how important is age relative to other

member characteristics in explaining equity ownership and equity allocation in

member investment choices?

6

1.3. Findings Overview

Four key findings emerge from this thesis. First, there is a considerable amount

of variation among the life-cycle products being offered in the United States and

Australia, especially in terms of asset allocation structure, expense structure, and

investment strategy. Second, a comparison of the asset allocation weights between the

Australian life-cycle products and the default investment options of Australian

superannuation funds reveals that life-cycle products decrease individuals' exposure to

growth assets very early in the life-cycle. The ultimate risk of an under-allocation to

growth assets is an insufficient superannuation benefit upon retirement. Adopting life­

cycle funds as a default option could exacerbate this risk. Third, the age variable does

have a significant relationship with equity participation and equity allocation. The two

key trends exhibited are hump-shaped and negative linear age-ownership and age­

allocation profiles. Fourth, when additional member characteristics are taken into

account including gender, marital status, and risk profile, the relative importance of the

age factor diminishes significantly.

1.4. Thesis Structure

This thesis is organised as follows. Chapter 2 provides a review of the existing

theoretical and empirical literature concerning life-cycle asset allocation (age-phasing).

Next, a survey oflife-cycle funds is presented. Chapter 4 describes the data and research

methods employed. Chapter 5 presents and discusses the findings. Chapter 6 provides

concluding comments with suggestions for relevant parties.

7

2. LITERATURE REVIEW

The notion of "age-phasing" refers to the reduction in risky assets over the life­

cycle. There is much debate surrounding age-phasing. On one hand, financial

practitioners frequently advise their clients to shift their investments away from stocks

(towards bonds or other fixed interest securities) as they age. Alternatively, investments

theory has argued that individuals should invest a constant proportion of their portfolios

in equities regardless of age. More recently, there has been increasing interest in this

topic, as the existing theoretical and empirical literature provides conflicting views.

2.1. Theoretical Literature

2.1.1. Modern Portfolio Theory

Markowitz (1952) introduces the basic concepts of what today is referred to as

modem portfolio theory. His paper is the first to offer a mathematical formalisation of

the idea of diversification of investments: the mathematical version of "the whole is

greater than the sum of its parts," (Rubinstein, 2002, p. 1042).

Characterised by rational investment behaviour and uncertainty, Markowitz

(1952, p. 77) expresses that in making portfolio decisions, investors are concerned with

both risk and return, and the discounting of "anticipated" or "expected" returns.

Markowitz (1952, p. 79) introduces the "expected returns-variance of returns" (mean­

variance) rule, postulating that an investor should maximise expected portfolio return

while minimising portfolio variance of return (or alternatively, minimise variance for a

given level of expected return). Portfolios fulfilling these criteria are described as being

"efficient" (Markowitz, 1952, p. 82).

Markowitz distinguishes between the attainable set of portfolios and the efficient

set of portfolios. The efficient frontier is defined as the set of Pareto optimal expected

return, variance of return combinations (Markowitz, 1991, p. 470). The attainable set

(opportunity set) of portfolios is the entire set of portfolios that could be found from a

group of n securities. However, as the law of large numbers cannot be applied to a

portfolio of securities, and because returns from securities are too intercorrelated,

diversification cannot eliminate all risk.

8

As stated by Rubinstein, the most significant feature of Markowitz's work was

to "show that it is not a security's own risk that is important to an investor, but rather

the contribution the security makes to the variance of his entire portfolio" (2002, p.

1042). Thus, the decision to include a particular security in one's portfolio should not be

made by comparing the security's expected return and variance to other securities.

Rather, one should consider the expected return and variance of the whole portfolio of

securities which includes the security in question.

The above basic elements set the foundation for modem portfolio theory. The

next section deals with extensions of Markowitz's early work on portfolio selection.

2.1.2. Optimal Lifetime Asset Allocation

The existing literature contains myriad theoretical justifications both for and

against this concept, commonly referred to as "age-phasing."

Samuelson (1969), Merton (1969) and Mossin (1968) demonstrate that subject

to a set of assumptions individuals should invest a constant proportion of their portfolios

in equities regardless of age. The key assumptions of their models include: (1) time

additive and separable utility; (2) identically and independently distributed returns (that

is asset returns follow a random-walk); (3) frictionless and complete markets; (4)

investors have constant relative risk aversion; and (5) investors have no labour income

or non-tradeable assets.

The models specified by Samuelson (1969), Merton (1969) and Mossin (1968)

are multi-period models. According to Mossin, such models follow the structure that the

investor still has the objective of maximising expected utility of wealth, with the added

assumption that "the time between the present and his horizon can be subdivided into n

periods (not necessarily the same length), at the end of each of which return on the

portfolio held during the period materialises and he can make a new decision on the

composition of the portfolio to be held during the next period," (1968, p. 220).

Samuelson (1969) develops a generalised discrete-time model to examine the

problem of optimal portfolio selection and consumption rules for an individual. He

shows that the result of investing the same fraction in equities at all ages holds, despite

the young person's increased opportunity to recoup any losses suffered earlier in the

life-cycle. Also, Samuelson's analysis allows him to dispel the application of the law of

large numbers to the portfolio allocation problem (1969, p. 245). Similarly, Merton

9

(1969) and Mossin (1968) explore the same problem in the context of a continuous-time

model and arrive at the same conclusion as Samuelson (1969).

Compared to the Markowitz static one-period framework, the multi-period

models specified by Samuelson (1969), Merton (1969) and Mossin (1968) are more

general and assume that individuals make choices dynamically over time (as opposed to

making decisions "myopically") (Bodie & Crane, 1997, p. 14).

Campbell, Cocco, Gomes, & Maenhout (1999, p. 3) point out that the

assumptions held by Samuelson (1969), Merton (1969) and Mossin (1968) pose

significant restrictions on the results derived. Much of the more recent empirical

literature related to life-cycle asset allocation relaxes one or more of these assumptions.

In contrast to the arguments against the reduction of risky assets over the life­

cycle, the following discussion identifies some authors' cases offered in support of age­

phasing. First, Samuelson's "escrow" argument suggests that altering the utility

function to include a minimum subsistence level of wealth (while retaining the

assumption of independent and identical probability distributions and ignoring human­

capital complications), provides an explanation for age-phased risk reduction (1989, p.

9048).

The reasoning behind the inclusion of this minimum subsistence level is that it

reflects the reality that people save to ensure themselves (and their families) sufficient

funds to satisfy minimum consumption requirements upon retirement (Samuelson,

1989, p. 9048). To do this, an individual places a proportion of his wealth in a safe cash

fund and the remaining is allocated in constant proportions between "risky" and "safe"

assets. As time passes, "safe" assets (which includes the value of the escrow) become a

larger proportion of aggregate assets ("risky" and "safe") (McNaughton, Piggot, &

Purcal, 1999, p. 4).

McNaughton, Piggot, & Purcal (1999) oppose Samuelson's (1989) "escrow"

argument. In fact, they propose that escrow is likely to lead to an investment strategy

whereby the proportion in risky assets actually increases over the life cycle.

Alternatively, McNaughton, Piggot, & Purcal (1999, p. 3) demonstrate that for those

people who accumulate their assets through the working phase of life, financial age­

phasing may be the result of a "saving rule".

10

To explain this further, this argument imposes the assumption of a firm saving

habit, which may be something as simple as an individual saving $k per annum until

retirement. At any point in time, the value of this stream of savings must be considered

in calculating the fraction of the portfolio in hand to be invested in safe assets

(McNaughton, Piggot, & Purcal, 1999, p. 5). As stated by McNaughton, Piggot, &

Purcal (1999, p. 5), "if the Merton constant proportion rule is adopted, then age-phasing

of the portfolio in hand will result, because the realised portfolio will exclude the value

ofthe safe saving stream to come."

As another case for age-phased risk reduction, Bodie, Merton, & Samuelson

(1992) present the labour supply flexibility argument which incorporates continuous­

time portfolio theory, as popularised by Merton (1969). Their model considers two key

components- financial wealth and human capital. Following Merton's continuous-time

framework the market values of both wealth components (financial and human capital)

change continuously and stochastically (Bodie, 2002, p. 4). Additionally, the wage rate .. (the return on human capital) exhibits perfect positive correlation with the market return

on traded assets (Bodie, 2002, p. 4).

At each point in time, individuals decide: (1) the amount of their consumption,

(2) the proportion of their financial wealth to invest in risky assets (versus the safe

asset), and (3) the portion of their maximum possible labour income that they will

"spend" on leisure so as to maximise their discounted lifetime expected utility (Bodie,

2002, p. 4). In other words, young individuals have more flexibility regarding their

labour/leisure decisions (i.e. to increase labour and reduce leisure and consumption) to

offset losses resulting from holding risky assets. Older individuals experience

diminishing labour supply flexibility, thus they tend to decrease the risk in their

investment portfolios as they near retirement.

Finally, Samuelson (1991) and Kritzman (1994) suggest another argument,

which is based on how individuals view movements in equity prices. Kritzman (1994)

refers to this as the "time diversification" argument. In this case for age-phased

reduction in equity, the assumption of identically and independently distributed returns

is abandoned (i.e. asset returns do not follow a random-walk). Samuelson (1991) and

Kritzman (1994) propose that asset returns exhibit mean reversion, which leads to

increased risk-taking by those individuals who have long time horizons. The reasoning

behind this argument is that, if asset returns are mean reverting, then above-average

11

returns tend to cancel out below-average returns over long horizons (Kritzman, 1994, p.

14). That is, the younger a person is, the greater the individual's risk tolerance as a

result of the belief in the chances of prices bouncing back from a low level

(McNaughton, Piggot, & Purcal, 1999, p. 2).

2.1.3. Asset Allocation and Annuitisation

Investors facing longevity risk may prefer to annuitise wealth at retirement if

annuity markets are sufficiently complete. However, this exposes investors to annuity

risk That is, the utility derived from annuitised wealth may disappoint if market

conditions turn out to be unfavourable at retirement (Koijen, Nijman, & Werker, 2006).

Although the annuitisation of retirement savings is not as much of an issue for

Australian Superannuation, as most members can take their retirement savings as a

lump-sum, it is important to acknowledge the possibility that age-phasing patterns in

asset allocation may be partly explained by the decision to convert retirement savings to

an annuity.

2.2. Empirical Literature

2.2.1. Age-Specific Patterns in Asset Allocation

The results of the study conducted by Schooley & Worden (1999, p. 41)

provides general support of the notion that individuals base their portfolio decisions on

their time horizons and risk tolerance. They identify a curvilinear relationship between

age and the percentage of equity in a financial portfolio. That is, as an investor's age

increases, so does the percentage of financial assets held in equities. This increase

occurs at a decreasing rate, until at some point the percentage of assets invested in

equities begins to decline.

Yoo (1994), Poterba & Samwick (1997), and Agnew, Balduzzi, & Sunden

(2003) identify similar patterns. Yoo (1994) uses cross-sectional data from the 1962

Survey of Financial Characteristics of Consumers and the 1983 and 1986 Surveys of

Consumer Finances (SCFs). Investigating the relation between age and risk exposure of

individual portfolios, Yoo (1994) provides empirical results inconsistent with the

behaviour prescribed by economic theory and financial practitioners. He identifies a

"hump-shaped" pattern, noting that exposure to risky assets increases over the working

life and decreases after retirement. Yoo's (1994) multiple regression results identify a

12

significant relationship between age and portfolio composition, supporting the non­

linear age patterns identified.

It is important to note that in examining the relationship between age and equity

allocations, each of the above studies does not specify regressions that are conditional

on stock ownership. Ameriks & Zeldes (2004) note that because a significant portion of

U.S. households do not hold any wealth in the stock market, there is a need to consider

two choices: first, the decision of whether or not to own stock, and second, the decision

of how much stock to hold conditional on ownership.

In researching how portfolio allocation to equity varies with investor age,

Ameriks & Zeldes (2004) acknowledge the existence of an identification problem with

regard to age, time, and cohort effects. This inherent limitation relates to the reality that

there is no way to separately identify the three effects without imposing further

assumptions, as the age, time, and cohort variables do not change independently

(Ameriks & Zeldes, 2004). Ameriks & Zeldes (2004) demonstrate that different

interpretations based on a dataset are possible, depending on assumptions about age,

time, or cohort effects.

Poterba & Samwick (1997) use pooled cross-sectional data from the 1983, 1989

and 1992 SCFs, to distinguish between age and cohort effects on the fraction of net

worth allocated to various asset categories. They note a general increase in ownership

and allocation of "all taxable equity" (i.e. directly held stock and stock mutual funds and

brokerage accounts) until around age 43. After this point, the age profiles are relatively

flat. Additionally, Poterba & Samwick (1997) find that there are practically no

differences in the age profiles of ownership and allocation when cohort effects are

included in the model. They conclude that, " ... households do not necessarily follow the

popular finance advice to switch from stocks to bonds as they approach retirement"

(Poterba & Samwick, 1997, p. 18).

Agnew, Balduzzi, & Sunden (2003) model jointly the decision to hold equities

and the decision of how much to hold, including age and time effects and excluding

cohort effects in their regression specification. Their preliminary inspection of data

reveals a hump-shaped pattern in equity allocations over the life cycle. Results from

their initial censored multiple-regression analysis include that age has an overall

negative effect on the share held in equities. In fact, they find that an additional year

translates into a lower allocation to stocks by 93 basis points. Agnew, Balduzzi, &

13

Sunden (2003) also estimate a specification including both age and age squared, to

capture possible non-linearity in the relation between equity allocation and age. The

results identify that equity allocation peaks very early, at 32.5 years of age.

Using cross-sectional data from a 1996 survey, Bodie & Crane (1997) examine

the asset allocation behaviour of a group of TIAA-CREF participants. They provide

evidence of a significant negative relationship between age and equity exposure in self­

directed retirement accounts. Bodie & Crane (1997) conclude that an additional year

translates into a lower allocation to stocks by 0.6 percentage points (i.e. consistent with

recommendations of financial practitioners).

Brown, da Silva Rosa & McNaughton (2006) use a unique time-series data set

(1974- 2005) consisting of trade-by-trade information on managed funds, in a study

concerning the behaviour of Australian managed fund investors. Consistent with the

lifecycle theory of investing, they find that older investors are more risk-averse than

younger investors. To examine the effect of age on portfolio allocation, Brown, da Silva

Rosa & McNaughton (2006) compare the split of investor-fund-days ("growth" and

"conservative") between six age groups. They report that those individuals aged

between 20 and 30 have 60.5% of their investor-fund-days in growth funds and 12.7%

in "conservative." Those investors over 60 have only 46.2% of their investor-fund-days

in "growth" and 24.8% in conservative.

In examining the empirical relationship between age and portfolio choice,

Ameriks & Zeldes (2004) employ pooled cross-sectional data from five waves of the

SCFs between 1983 and 1998, and a panel data of TIAA-CREF accounts covering the

period 1987-1999. They model two portfolio choices: (1) the decision to hold stock or

not (using a Probit procedure) and (2) the choice of how much equity to hold,

conditional on stock ownership (using simple Ordinary Least Squares). Ameriks &

Zeldes (2004) consider three separate exclusion restrictions by estimating a model with

age and time effects and a model with age and cohort effects.

With respect to equity ownership probabilities, their regression results with no

cohort effects reveal a significant hump shape in the age-ownership profile. The

regression with no time effects shows a monotonic increase until around 50 years of

age, and a levelling off thereafter. With respect to the amount of stock held conditional

on stock ownership, Ameriks & Zeldes (2004) find that when cohort effects are

excluded from the specification, the estimated age-ownership profile is extremely flat.

14

The regression with no time effects shows a steady increase in the age-ownership

profile. However, Ameriks & Zeldes (2004) conclude that after taking into account

several complexities, including key features of individual portfolio behaviour, a

fundamental identification issue, and lack of perfect data, the evidence indicates that

individuals do not gradually decrease equity shares as they age.

2.2.2. Labour Supply Flexibility and Asset Allocation

Using panel data from the U.S. Health and Retirement Study (HRS), Benitez­

Silva (2002) conducts an empirical analysis of the relationship between measures of

labour supply flexibility and portfolio choice decisions by utility-maximising

individuals. After controlling for unobserved heterogeneity, Benitez-Silva (2002) finds

that individuals with more labour supply flexibility, on average, hold between 12% and

14% more wealth in stocks in their portfolios.

Similarly, Cocco, Gomes, & Maenhout (2005) provide evidence that the share

invested in equities decreases with age. They note that this is driven by the fact that the

labour income profile itself is downward sloping, that is labour supply flexibility

decreases as one becomes older.

2.2.3. Australian Evidence on Age-Specific Patterns in Asset Allocation

In a study of Australian share ownership, the Australian Stock Exchange (2004)

reports an increase in equity participation (direct and indirect share ownership) from

40% to 55% over the period 1998-2004. This study also indicates that the percentage of

individuals owning equity steadily increases with age.

Using data from the 2002 HILDA6 survey, the Reserve Bank of Australia (2004)

reports that the percentage of Australian households owning equity investments follows

a hump-shaped pattem: equity participation increases from 29% for the 16-34 age range

to 53% for the 55-64 age range, and then decreases to 34% for those aged 75 or over. In

addition, the median value of equity holdings for Australian households follows a

similar trend. Equity allocation increases from $7,000 for the 16-34 age range to

$48,000 for the 65-74 age range, and then declines to $30,000 for those aged 75 or over.

6 Household, Income, and Labour Dynamics in Australia

15

3. SURVEY OF LIFE-CYCLE FUNDS

One of the mam reasons for the popularity of life-cycle funds is that they

simplify asset allocation and fund selection for investors of different ages and risk­

preference groups, who are often unprepared and unwilling to make these decisions

themselves (Morentsov, 2006).

Behavioural finance studies [for example, Madrian & Shea (2001), Michell,

Mottola, Utkus, & Yamaguchi (2006), Agnew, Balduzzi, & Sunden (2003), Iyengar,

Jiang, & Huberman (2004), and Huberman & Jiang (2006)] indicate that some

individuals suffer from psychological biases with respect to retirement savings

decisions, such as inertia, procrastination, and choice overload. The existence of such

behavioural biases hinders the ability to make rational decisions. Thus life-cycle funds

offer an approach to retirement investing that does not require individuals to actively

make decisions but to put in place a semi-active investment strategy with one decision.

3.1. The U.S. Experience

Financial practitioners are frequently advising their clients to reduce their

exposure to risky assets as they become older: that is, to shift their investments away

from stocks and towards bonds or other fixed interest securities. The following

statement from a retirement strategies booklet from TIAA-CREF, a U.S. based

retirement fund with assets over US$3 80 billion, is typical:

If you retire early, you should receive a larger percentage of income from equity­based accounts, since your income will need to grow with inflation, perhaps for 30 years or more. As you age, you might consider switching to more predictable income sources, such as fixed-income and real estate account (TIAA-CREF, 2005, p. 32).

Similar views can be found in finance theory such as that by Malkiel (1996, pp.

404-405) who claims that "the longer the time period over which you can hold your

investments, the greater should be the share of common stocks in your portfolio."

The operation of life-cycle funds reflects the above advice. Life-cycle funds are

managed funds that automatically reduce the proportion of equities to total assets as an

16

investor ages. In choosing a life-cycle fund an investor selects a fund with a "target

date" closest to his expected retirement. Existing life-cycle funds generally offer target

dates ranging from 2010 to 2040, with five or ten-year increments. For example, the

U.S. fund family, Vanguard, offers a series of life-cycle funds (Vanguard Target

Retirement) with target dates 2010 to 2050, in five-year increments. A 55-year old in

2006 who is planning to retire at age 64 (nine years to retirement) could choose the

Vanguard Target Retirement 2015 fund. Upon the investor's retirement date, funds are

generally rolled into the lowest-risk fund of a target-date series typically referred to as

an "income" fund. Essentially, selecting an appropriate target date is the only decision

that investors are faced with when investing in a life-cycle fund.

First introduced to the U.S. investment environment during the mid-1990s, life­

cycle funds have experienced a significant amount of growth in recent years. At the end

of 2005, the U.S. target-date fund population had a total $70 billion in assets under

management, of which Fidelity Investments managed $50 billion. To put these figures

into context, in 2005, mutual funds managed a total $3.4 trillion in retirement account

assets. U.S. households owned a further $780 billion in mutual fund assets through

variable products at life insurance companies held outside of retirement accounts

(Investment Company Institute, 2006, p. 2). Figure 3.1 shows the growth of life-cycle

funds in the U.S. From modest beginnings in the mid-nineties, the amount of money

held in life-cycle funds gradually increased until 2001/2002, after which these funds

experienced rapid growth. As at June 2006, fund tracker Morningstar held 155 target­

date funds in its database (Carlson, 2006). To address the phenomenon of target-date

funds in the United States, Morningstar introduced three new mutual fund categories in

March 2006: "target-date 2000-2014" "target-date 2015-2029" and "target-date 2030+."

The five major providers of target-date funds in the U.S. are Fidelity Investments,

Vanguard, T.Rowe Price, Principal Investors, and Barclays.

17

Figure 3.1 Growth of Life-Cycle Funds Billions of dollars, year-end, 1996-2005

0 ~layer-Sponsored Defined Contribution Plans Ill Individual Retirerrent .A:x:.ounts rn Other Investors

~ IU

8 0 ~ ~ ffi

80.0.-~----~~----~--------~------------------------~-,

70.0

60.0

50.0

40.0

30.0

20.0

10.0

0.0+-~~~--~-=~-r~~,-==~~==~~~~-=~-r~~,-~~

1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

Years

Source: Investment Company Institute (2005, Figure 11)

Figure 3.1 also provides a breakdown of life-cycle fund assets between

Employer-Sponsored Defined Contribution (DC) Plans, Individual Retirement Accounts

(IRA's), and "Other" Investors. The main distribution channels for life-cycle funds are

through Employer-Sponsored Plans and IRA's. Combined, these two distribution

channels accounted for almost 90 percent of life-cycle fund assets as at the end of 2005

(Investment Company Institute, 2006, p. 10). Marquez (2005) reports that 38 percent of

401(k) plans offer life-cycle funds and 37 percent of plan participants use these funds. A

study completed by the Financial Research Corporation (2006) reports that the

attractiveness of life-cycle funds among investors and retirement plan sponsors is

expected to continue. It is anticipated that small DC plans will experience the largest

sales growth, as plan providers increasingly focus on the small-plan DC market and as

target-date funds prove to be the most cost-efficient asset allocation tool for this

particular segment of the market. Also, IRA's are expected to experience gradually

increasing sales as a growing number of job changers choose target-date funds for IRA

rollovers. On the other hand, little growth in target-date funds is expected in retail

accounts due to "inherent conflicts of interest in selling asset allocation products

through commission or fee-based accounts" (Financial Research Corporation, 2006, p.

14).

18

Based on the concept of life-cycle investing, target-date funds are marketed

mainly for retirement purposes. In particular, these funds have been designed to target

those investors who demand a simple investment vehicle that does not require asset

allocation rebalancing and frequent monitoring on their part. Target-date funds are

likely to be suitable for investors who want to adopt an age-based investment strategy,

are unlikely to seek financial advice, do not have the time and/or knowledge to actively

manage their retirement portfolio, or suffer from psychological barriers which hinder

the ability to make rational decisions.

3.2. Retirement Saving Behavioural Problems

The existence of psychological barriers among individuals is well-documented

in the behavioural finance literature. Empirical research suggests that when it comes to

saving for retirement, some individuals suffer from certain behavioural biases such as

procrastination, inertia, and choice overload. In an analysis of the 40l(k) savings

behaviour of the participants of a large U.S. company plan, Madrian & Shea (2001) find

that there is a significant increase in plan participation when automatic enrollment is in

place, compared to the low participation rates when employees must make an active

decision to participate. Specifically, they report an increase in participation rates from

49 percent to 86 percent, suggesting procrastination of the decision to participate in the

401(k) plan prior to automatic enrollment as a potential key explanation for the change

in participation rates (Madrian & Shea, 2001 ). Madrian & Shea (200 1) also document

"default" behaviour, whereby a large percentage of employees who became plan

participants via automatic enrollment did not change their default contribution rates and

default fund allocations over time. In other words, these employees suffered from inertia

in relation to the investment reallocation decision.

Using a dataset of more than 1500 retirement plans, Mitchell, Mottola, Utkus, &

Yamaguchi (2006) find that most participants display profound inertia. Over a two-year

period, a majority of participants (80 percent) initiate no trades, and an additional 11

percent makes only a single trade. Agnew, Balduzzi, & Sunden (2003) also provide

evidence of inertia in asset allocations. They find that the participants of a large

corporate 401(k) plan, who entered the plan before April1994 with a default allocation

of 100 percent to the risk-free asset, have average equity allocations that are

considerably lower than those of later entries, who were required to make an explicit

19

asset allocation choice (Agnew, Balduzzi, & Sunden, 2003). In other words, there is a

tendency for participants to maintain their initial default allocations rather than revise

their asset allocations.

The Australian superannuation industry has followed a worldwide trend of

investment choice expansion. In the current retirement savings environment, members

are faced with a vast menu of investment choices. The Australian Prudential Regulatory

Authority (2006, p. 31) reports that the average number of investment choices offered in

corporate, industry, public sector, and retail funds in 2005 was 4, 9, 6, and 61

respectively. For some individuals, the plethora of investment choice available can be

overwhelming, and may result in "choice overload".

A number of studies have investigated the effect of the number of investment

choices (size of the investment menu) on individuals' savings behaviour. Using a

sample of 647 401(k) plans, Iyengar, Jiang, & Huberman (2004) test the hypothesis that

employee participation rates decline as the number of fund options increase. They find

that the more funds a plan offers, the less likely an employee is to participate. Iyengar,

Jiang, & Huberman (2004) show that participation rates peaked at 75 percent when only

two funds were offered. Participation steadily decreased to a low of 60 percent when 59

funds were offered. In a study of 638 401(k) plans, Huberman & Jiang (2006) find that

participants generally use a small number of funds (no more than three or four)

regardless of the number of funds offered.

Overall, such findings suggest that the availability of too many investment options

can lead to individuals experiencing feelings of information overload, resulting in

increased use of the default option and declines in participation rates. The evidence is

similar in Australia. A significant proportion of superannuation members are passive

decision-makers, with 55.7% of superannuation assets being held in the default

investment strategy for the year ending June 2005 (Australian Prudential Regulation

Authority, 2006t

Target-date funds offer a potential solution for those individuals who require

additional assistance in managing their portfolios. The age-based automatic rebalancing

feature can be useful in overcoming psychological biases or any such problems

7 Based on the total assets of those superannuation entities with more than four members.

20

preventing individuals from taking a more active role in the management of their

retirement savings.

3.3. A Comparison of Target-Date Funds

A survey of current target-date funds suggests that while they are based on the

concept of age-phasing, they do vary considerably in relation to a number of key factors

including asset allocation structure, expense ratios, and investment objectives and

strategies employed. This is illustrated in Appendix A Table 1 which reports some key

investment details for those mutual funds tracked by Morningstar (US) which are

categorised as either "target-date 2000-20 14" "target-date 20 15-2029" or "target-date

2030+."

The following information is provided for each of the target-date funds in a

provider's fund series: Total Assets (total assets under management), Net Asset Value

(the fund's per-share price calculated as total assets of the fund less fees and expenses

divided by number of shares outstanding), Expenses (expense ratio, front load fee, and

deferred load fee), and Asset Allocation (cash, stocks, bonds, and "other").

The following observations have been made with respect to the information

based on target-date funds tracked by Morningstar (US) only. It is important to note that

some (if not all) target-date series offered by each ofthe fund families offer a number of

share classes for each fund in their series, which will therefore have different costs

attached.·

3.3.1. Asset Allocation

Overall, equity holdings range from nil to 98.8 percent. For example, none of

SunAmerica's High Watermark target-date portfolios have an allocation to equity. They

invest in cash and bonds only. On the other hand, Seligman's TargETFund portfolios

maintain a large allocation to equity, with the decrease in equity over the life-cycle

being replaced by allocation to bonds. Using the target-date of 2010 to compare the

target-date fund series, the largest stock holding is 65.3 percent (AllianceBernstein) and

the lowest stock holding is nil (SunAmerica).

21

The rate of decrease in equity allocation vanes between the target-date

portfolios. For example, the largest change in equitl is a movement of 44.8 percentage

points from 92.1 percent to 47.2 percent (Barclays' LifePath series). The smallest

change in equity is a movement of 20.6 percentage points from 81 percent to 60.4

percent (Schwab's Target series).

In terms of cash holdings, there is a general tendency to increase allocation to

cash as the retirement date draws closer. Using the target-date of 2010 to compare the

target-date portfolios, the largest cash holding is 26.6 percent (SunAmerica's High

Watermark C 2010) and the smallest cash holding is 0.5 percent (DWS-Scudder's

Target 2010).

For each of their target-date funds, DWS-Scudder and Wells Fargo allocate very

little to cash (no more than two percent), with allocations being concentrated on equity

and bonds. SunAmerica actually decreases its allocation to cash in favour of bonds as

the retirement date draws closer.

In terms of bond holdings, there is a general tendency to increase the allocation

to bonds as the retirement date draws closer. Increases in bond allocation vary between

the target-date fund series. For example, the largest change in bonds9 is a movement of

44.3 percentage points from 10.2 percent to 54.5 percent (Wells Fargo's Advantage DJ

Target). The smallest change in bonds is a movement of 15.6 percentage points, from

11.9 percent to 27.5 percent (Schwab's Target).

Overall, inspection of the asset allocation figures for the target-date funds in

Appendix A Table 1 suggests that they become conservative very quickly. For example,

DWS-Scudder's Target 2014 portfolio only allocates 26.4 percent to equity. An

individual investing in this portfolio in 2006 would still have approximately seven years

until retirement plus life expectancy period.

3.3.2. Asset Holdings

Target-date funds often employ a fund-of-funds approach. That is, the

investments within the life-cycle funds are other funds offered by the sponsoring fund

family. They invest in their own family mutual funds. For example, Fidelity's Freedom

8 Measured as the percentage point difference in equity between those portfolios with a 2040 target-date and those portfolios with a 2010 target-date. 9 Measured as the percentage point difference in bonds between those portfolios with a 2040 target-date and those portfolios with a 2010 target-date.

22

target-date funds are well-diversified among its own actively managed funds. The

Fidelity Freedom 2040 holds five international funds, including Fidelity Japan, Fidelity

Southeast Asia, and Fidelity Europe. Some of its other fund holdings include Fidelity

Mid-Cap Stock, Fidelity Blue Chip Growth, and Fidelity Investment Grade Bond.

On the other hand, Wells Fargo's Outlook target-date funds take a different

approach, as they invest directly in shares and bonds rather than mutual funds. Top

holdings in the Wells Fargo Outlook 2040 include Microsoft, Pfizer, Citigroup, and

General Electric.

3.3.3. U.S. Expense Ratios

The fees attached to life-cycle funds generally tend to decrease as the retirement

date draws closer. For example, Russell's LifePoints 2040 portfolio has an expense ratio

of 1.51 percent, which gradually decreases to 1.35 percent for the LifePoints 2010

portfolio. Of the 21 fund families who offer target-date funds as categorised by

Morningstar, Vanguard offers the lowest-cost series ofthese funds and does not attach

either a Front or Deferred load.

The difference between the highest and lowest expense ratios within each of the

fund families' target date series', range from nil to 0.84 percent. Different expense ratio

patterns are evident among the funds. For example, Barclays, Principal Investors, and

Wells Fargo maintain the same expense ratios for all funds in their target-date series;

DWS-Scudder, JPMorgan, MassMutual, MFS, Russell, Schwab, Seligman,

VantagePoint Funds, and T.Rowe Price progressively decrease their expense ratios as

the target-date draws closer; Hartford Mutual Funds decrease their expense ratios

equally from the latest to the earliest occurring target-date. StateFarm on the other hand

actually slightly increase their expense ratios as the target-date draws closer.

3.4. The Australian Experience

Life-cycle funds are now beginning to emerge in Australia. In 2005, Russell

Investment Group introduced Australia's first series oflife-cycle superannuation funds­

Russell LifePoints target portfolios. Russell (Australia) adopts a manager-of-managers

approach, whereby each of its life-cycle portfolios invests in a mix of growth and

income investments via the Russell Multi-Manager funds (Russell Investment Group,

2005). According to Dunstan (2006), Russell's LifePoints portfolios are starting to

23

attract interest from superannuation funds in addition to members of Russell's master

trust.

Aside from Russell Investment Group, the only other known providers of life­

cycle funds at the time of writing are AMP and Virgin Money10• With AMP,

superannuation contributions are invested in the portfolio structured for the relevant age

group. However, investors have the option of switching one-up or one-down from the

portfolio structure of their age group in order to take a more conservative or more

aggressive approach. The core investment of the Life Stages series is the AMP Balanced

Direct fund. AMP also adopts a multi-manager approach through investing in its Future

Directions investment options.

In contrast to the life-cycle portfolios offered by Russell and AMP, Virgin offers

two different choices, balanced and aggressive. Each option contains four funds which,

like AMP's LifeStages series, are based on investors' ages. Virgin invests in funds

managed by Macquarie Funds Management. Both Virgin and AMP stress that their life­

cycle funds are intended to be a 'whole of working life' strategy. That is, they are not

intended to be combined with any other investment option. Combining the use of a life­

cycle fund with other investment options negates the automatic age-based rebalancing

feature of the funds. Life-cycle funds are based on the principle of age-based investing.

In order for results to be realised, a full commitment of retirement savings is required. If

investors are prepared to manage and rebalance their own portfolios, then life-cycle

funds do not offer any advantages.

There are two key ways in which life-cycle funds operate in terms of automatic

rebalancing: progressive operation and step-like operation. With a progressive

operation, the fund's asset allocation changes frequently. For example, a fund's

investment mix may be moved towards a more conservative position on a continuous or

annual basis. A step-like operation involves changes in the fund's asset allocation when

an investor reaches a threshold age. This type of operation is employed by the Virgin

Money LifeStage Tracker and AMP LifeStages series'. As an example, the asset

allocation for an individual who invests in Virgin's "under 40's mix" fund will be

automatically adjusted once the investor reaches the entry age for the next portfolio, the

"over 40's mix". With Russell LifePoints, asset allocation is re-balanced automatically

every year, progressively reducing the allocation to growth-type assets.

10 The LifeStages and Life Stage Tracker series respectively.

24

3.5. Life-Cycle Funds as a Default Option

The default investment options within U.S. 401(k) plans are typically money

market funds. Approximately sixty-seven percent of plans have money market or stable

value funds as their default option (Marquez, 2005). Research from the Vanguard

Center for Retirement and Research notes that plan sponsors are increasingly focusing

on the need to adopt more balanced default investment options in recognition of the

long-term nature of retirement savings plans (Utkus, 2004). One potential solution to

this more "balanced" approach is the adoption of life-cycle funds as a default option,

offering a series of funds containing a mixture of asset classes (both equity and income

securities) based on the investor's expected time to retirement, which is automatically

rebalanced as the investor ages (Utkus, 2004). Life-cycle funds are offered by more than

one third of 401(k) plans in the United States (Marquez, 2005), where they increasingly

serve as a default option.

How life-cycle products would serve as a default option to members within the

Australian superannuation environment is unclear, in part because these products are

very new to Australia. Life-cycle funds also challenge the status quo of existing

superannuation fund default options. Unlike the typical money market funds employed

as default options in a majority of 401(k) plans in the U.S., current default options

offered within Australian superannuation funds are generally positioned more towards a

balanced or diversified investment strategy. An analysis concerning investment

strategies in Australian superannuation funds, conducted by Rice Walker Actuaries,

provides some insight into this issue by warning that members should avoid using life­

cycle products that shift away from growth assets at too young an age, as "they are too

conservative and they cost members money" (Rice Walker Actuaries, 2006, p. 8). They

argue that adopting life-cycle funds as a default option exposes funds to the risk of

members returning in the future and claiming that they lost their money due to the

trustee strategy.

The analysis states that the default options of Australian superannuation funds

typically invest 70 to 80 percent in growth assets, "reflecting the view that super is a

long term investment, and that growth assets offer the highest returns" (Rice Walker

Actuaries, 2006, p. 4). Providing the average asset allocations for Australian default

investment strategies (broken down into corporate, industry, public sector, and retail

25

superannuation funds), Table 3.1 confirms that Australian superannuation fund default

options are generally geared more towards growth assets. The table indicates that there

is a large degree of consistency among the key superannuation sectors in terms of

average asset allocations in the default investment options. The last column reports that

on average, the total proportion of superannuation assets allocated to "growth-type"

assets (shares and property) was 64 percent in 2005, with the majority of default

strategy assets being held in equities.

Table 3.1 Asset Allocation of Default Investment Strategy- Proportion of Assets (%) Entities with more than four Members

Aust Shares Int'l Shares Property -Listed -Unlisted Aust Fix Interest Int'l Fix Interest Cash Other

Corporate Industry 36 35 25 23

4 3 14 5 4 9

3 7 10 5 5 13

Total 100% 100% Source: APRA (2006, Table 16)

Public Sector 32 29

4 5 11 7 8 4

100%

Retail 30 17

4 2 19 4 8 15

100%

Total 33 23

4 4 13 5 7 10

100%

To compare the asset allocation between the Australian life-cycle products and

the default investment options of Australian superannuation funds, Tables 3.2 to 3.4

provide the asset allocation weights for the Australian superannuation funds data that

have been employed in this thesis (HESTA, STA, GESB, and UniSuper), the existing

Australian life-cycle funds, and a fund series that is representative ofthe U.S. life-cycle

funds population.

The funds examined in this thesis are four of Australia's larger not-for-profit

superannuation funds. HESTA, STA, and UniSuper are industry funds, whereas GESB

is a public-sector fund. Further information regarding these funds is provided in Chapter

4.

Looking first at the default option asset allocations of the four superannuation

funds, the allocation to risky assets generally falls within the 60 percent to 75 percent

range. Although all four life-cycle funds begin with significantly higher allocations to

growth assets (85 percent to 100 percent), it is clear that the funds shift away from

growth assets early in the life-cycle. For those individuals who are approximately 13

26

years (3 years) from their expected retirement date, the corresponding fund invests as

little as 46 percent (20 percent) in growth assets.

Table 3.2 Industry Superannuation Fund Default Asset Allocation by Asset Class This table shows the asset class exposures for each fund's default investment option for the 2005/2006 financial year.

Date Cash Fixed Property Int Aust Int11 Shares Shares

HESTA "Core Pool" 1/7/2005 4.0 12.0 12.0 23.0 30.0 STA "Balanced Plan" 1/7/2005 5.8 9.5 8.7 26.7 36.9 GESB "Conservative Plan" 1/7/2005 2.0 68.0 5.0 15.0 10.0 UniSuper "Balanced" 1/7/2005 0.0 30.0 10.0 25.0 27.5

Infras'ture

9.0

7.3

Priv Abs Other Equity Rtn

Str

4.0 3.0 3.0

3.0 2.3

7.5 Source: HESTA, STA, GESB, and UniSuper 2005/2006 annual reports

Table 3.3 Australian Life-CJ!.cle Fund Asset Allocation bJ!. Asset Class

Cash Aust Int'l Alt's12 Aust Int'l Prop Bonds Bonds Shares Shares

Russell Investment Graue. "LifePoints" 2040 0.0 0.0 0.0 0.0 50.0 40.0 10.0 2030 1.0 11.0 5.0 3.0 37.0 35.0 8.0 2020 8.5 18.5 11.5 1.5 26.0 25.0 9.0 2010 22.5 23.5 14.0 0.0 17.5 15.0 7.5 AMP "LifeStaz.es" 2040 1.1 4.7 1.9 2.0 44.6 43.1 2.8 2030 3.3 18.8 9.7 6.5 28.7 24.0 9.1 2020 12.5 24.0 17.0 5.0 19.0 15.5 7.0 2010 32.0 21.7 16.1 4.0 11.7 8.9 5.6 Virz.in Money "Life Stage Tracker" (Balanced choice) 2040 15.0 47.0 32.0 6.0 2030 30.0 38.0 24.0 8.0 2020 50.0 25.0 15.0 10.0 2010 80.0 10.0 5.0 5.0 Fidelity Investments "Freedom JJ

2040 2.6 12.3 84.2 2030 2.8 14.3 82.1 2020 4.0 25.4 69.8 2010 11.1 38.8 49.0

11 For the GESB "Conservative Plan", the Fixed Interest asset class includes inflation-linked bonds and global bonds. 12 The "Alternatives" asset class refers to an income-type asset class for Russell LifePoints, where as it refers to a growth-type asset class for AMP Life Stages.

27

Source: Russell Investment Group PDS (Issued: 30 September 2005), AMP PDS (Issued: 3 September 2005), Virgin Money PDS (Issued: 7 November 2005), and Fidelity Freedom prospectus (Issued: 30 May 2006)

Table 3.4 US. Representative Life-Cycle Fund Asset Allocation by Class The Fidelity Freedom target-date funds are used as a representative for the U.S. target­date fund population as they have the largest amount of total net assets.

Cash Bonds Shares Other Fidelity Investments "Freedom " 2040 2.6 12.3 84.2 0.9 2030 2.8 14.3 82.1 0.8 2020 4.0 25.4 69.8 0.7 2010 11.1 38.8 49.0 1.1

3.6. Limitations of the Life-Cycle Investment Strategy

As suggested by Rice Walker Actuaries (2006), a risk of an under-allocation to

equity during the life-cycle is underperformance of the retirement investment and

ultimately, an insufficient superannuation balance upon retirement. Additionally, with

current mortality statistics suggesting an extended life expectancy13, members are faced

with increased longevity risk. That is, the risk of living longer than expected and

outliving savings. Thus, the need to maintain a long-term focus and hold a sufficient

level of growth assets in the retirement portfolio at all times is of increasing importance.

On the other hand, an over-allocation to growth assets at certain points in an

individual's life-cycle can also pose risks. For example, holding a large proportion of

growth assets in the period leading up to retirement exposes retirement savings to short­

term market fluctuations, potentially resulting in losses and a lower retirement benefit.

Whether or not life-cycle funds are an appropriate retirement investment vehicle

essentially depends on the individual investor. Life-cycle funds offer simplification of

the investment process. Individuals can be passive-active. That is they elect to not do

anything but allow a manager to adjust their portfolio automatically. In the current

superannuation environment which is characterised by choice, life-cycle funds may be

an attractive option for members who feel overwhelmed by the plethora of investment

choice available. Similarly, such funds may be appropriate for members who do not

want to actively manage their own retirement portfolio due to psychological biases like

13 According to ABS Life Tables, the average worker retiring in 2003 (at age 65) can expect to 17 years if male, and 21 years iffemale (Rice Walker Actuaries, 2006).

28

procrastination and inertia, or a lack of time and/or knowledge. However, whilst life­

cycle funds offer an age-based, "set and forget" type investment strategy over the static

investment strategy employed by default options, such products are not without their

shortcomings. Some key potential disadvantages associated with the use of life-cycle

funds are identified next.

First, life-cycle funds assume limited differences among investors, leading to

oversimplification of investment decisions. For example, individuals of the same age

may differ in terms of expected retirement date, risk tolerance, needs at retirement,

existing portfolio allocation, and wealth. Each life-cycle fund is typically geared to

investors solely on age. That is, those in their 20's/30's are placed in the most

aggressive fund, and those who are older are invested in one of the more conservative

funds. While age may be one of the factors that determines an investor's appetite for

equity, it may not be the most important one.

Second, life-cycle funds shift away from growth assets early in the life-cycle. As

previously pointed out, an under-allocation to growth assets can lead to

underperformance of the retirement investment resulting in an inadequate retirement

benefit.

Third, life-cycle funds are designed to be a "one-stop-shop." That is, investors

should commit 100 percent of their retirement savings and future contributions to the

one fund. Investing in several life-cycle funds or combining a life-cycle fund with other

investments defeats the purpose of the "whole of working life" strategy espoused by

life-cycle funds, and works against the automatic rebalancing feature of the funds

(Marquez, 2005). Combining the use of a life-cycle fund with other funds is also more

costly.

Fourth, as life-cycle funds are a very new product, they have not established

much of a track record, posing performance measurement difficulties. Further, the

significant differences between available life-cycle funds, in terms of asset allocation

structure, expense structure, investment strategies and objectives, raises questions as to

the best way to perform peer group performance comparisons when the funds are so

different.

Fifth, life-cycle funds generally carry higher fees and expenses than typical

"balanced" or "diversified" funds.

29

3.7. Summary

Life-cycle funds challenge the status quo of existing "balanced" funds typically

offered as default investment options within Australian superannuation funds. The

advantages of implementing life-cycle products as a default option within

superannuation funds may not be as extensive in the Australian context compared to the

United States. There a majority of retirement plans currently employ money market

funds as default options, which is inconsistent with the long-term nature of retirement

savings. The adoption of life-cycle products as a default option could also potentially

threaten the role of financial advisers in deciding asset mixes (Dunstan, 2006), leading

to a possibly reduced financial adviser role within the superannuation industry though

investors likely to choose such products may have been less likely to use advisers in the

first place. Life-cycle funds can be helpful for investors who want to adopt an age-based

investment strategy, are unlikely to seek financial advice, do not have the time and/or

knowledge to actively manage their retirement portfolio, or suffer from psychological

barriers such as procrastination, inertia, and choice overload. Through their automatic

rebalancing feature, such funds offer simplification of the retirement investment

process.

Life-cycle funds are not without their limitations. Some of the limitations of life­

cycle funds include, the assumption of limited differences among investors, a shift away

from growth assets early in the life-cycle, combining a life-cycle fund with other

investments works against the automatic rebalancing feature of the funds and is also

more costly, and life-cycle funds have not established a track record which poses

performance measurement difficulties.

30

4. DATA AND RESEARCH METHODS

This chapter first provides a description of the data used in this thesis and a brief

comparison of the two key superannuation fund data sets in terms of summary statistics.

Also in this chapter is an explanation of the research methods employed, including a

preliminary analysis which involved sorting member asset allocations by age, gender,

and marital status, and comparing actual member allocations with existing life-cycle

fund asset allocations, and a regression framework involving the Probit and Tobit

methods.

4.1. Data Description and Definition of Variables

The primary data employed in this thesis has been provided by four Australian

superannuation funds: the Health Employees Superannuation Trust (HESTA), the

Superannuation Trust of Australia (STA), the Government Employees Superannuation

Board (GESB), and UniSuper. Combined, these four superannuation funds had $57

billion in assets under management and over 2.3 million members as at 30 June 200614.

The fund data employed in this study is divided into two key data sets:

1. Full sample: collective information containing membership data obtained from

the funds for all member investment strategy changes since the respective fund

first introduced investment choice (93042 observations); and

2. Restricted sample: combined data containing the base membership data and

additional survey information for those members who have exercised choice

since the respective fund fi~st introduced investment choice and responded to a

survey which examined the intentions of the members to change their

investment strategy and make extra contributions (2315 observations )15•

Appendix B Table 1 contains definitions of the variables used in the regression

analysis. Details of this analysis are explained in the following subsection. Summary

14 HESTA, STA, UniSuper, and GESB 2005/2006 Annual Report. 15 This data was collected by a PhD student working on a sponsored project associated with this thesis.

31

statistics16 for the main variables used in this study are reported below in Tables 4.1 and

4.2.

Table 4.1 Summary Statistics of Fund Member Characteristics: Full Sample This table presents summary statistics relating to the superannuation characteristics of HESTA, STA, and GESB members in the full sample, including equity participation, age, gender, member contributions, fund, and year of choice.

Variable Obs Mean St Dev Min Max Equity Participation:

Yes 81826 No 11213 Unknown 3

Total 93042 Eguity Allocation (%) 93039 55.27 28.42 0.00 100.00 Age at Choice 86154 38.72 11.73 11.00 77.00 Gender: Male 38297 Female 47848 Not Specified 6897

Total 93042 Total contributions in 2004 ($) 65826 3,694.61 5,549.13 0.12 219,165.00 Fund: HESTA (1) 47843 STA (2) 25698 GESB (3) 19501

Total 93042 Calendar Year of Choice:

1995 42 1996 129 1997 928 1998 1185 1999 2111 2000 4899 2001 19238 2002 15980 2003 25540 2004 22987 Unknown 3

Total 93042

16 Observations with missing values are not included in calculations of summary statistics.

32

Table 4.2 Summary Statistics of Fund Member Characteristics: Restricted Sample This table presents summary statistics relating to the superannuation characteristics of HESTA, STA, GESB, and UniSuper members in the restricted sample including equity participation, age, gender, member contributions, fund, year of choice, and net wealth.

Variable Obs Mean St Dev Min Max Equity Participation:

Yes No

Total Equity Allocation (%) Age Gender: Male Female Not Specified

Total Marital Status: Married/I)e-facto Separated/I)ivorce/Widow Single Unknown

Total Total contributions in 2004 ($) Fund:

HESTA (1) STA (2) GESB (3) UniSuper (4)

Total Calendar year of choice:

1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Total Net Wealth:

$0-$100000 $100001-$200000 $200001-$300000 $300001-500000 $500001-$700000 Above$700000 Unknown

Total

2009 306

2315 2315 2315

1001 1306

8 2315

1732 255 288 40

2315 2163

755 178 561 821

2315

3 3

21 57 139 147 540 313 610 482 2315

238 227 291 479 368 591 121

2315

57.23 30.56 0.00 100.00 44.96 10.70 15.43 104.82

7,425.76 9,462.04 0.00 114,274.50

33

In both data sets, the majority of members have equity in their investment

strategy. The average member asset allocations to equity are comparable between both

data sets, with 55.27 percent for the full sample and 57.23 percent for the restricted

sample. The average member age is slightly higher for the restricted sample ( 45

compared to 39). Also, the majority of individuals are female (51 percent for the full

sample and 56 percent for the restricted sample). The average member contribution in

2004 was higher for the restricted sample ($7,426 compared to $3,694), and most asset

allocation changes took place between 2001 and 2004.

Data regarding U.S. and Australian life-cycle funds and information on asset

allocations and cost structures in particular have been sourced from the Morningstar

(U.S.) website and product disclosure statements, and have been discussed in Chapter

Three. Strategic asset allocations for the "default" investment options offered by

HESTA, STA, GESB, and UniSuper have been obtained from their 2005/2006 annual

reports. General Australian superannuation default strategy statistics including

percentage of total superannuation assets held in the default investment strategy and

average asset allocation of the default strategy were obtained from the Australian

Prudential Regulatory Authority (APRA) published data.

4.2. Description of Funds and Investment Strategy Choices

As at June 2006, HESTA had 550,000 members (primarily from the health and

community services industry) and 45,000 employers were registered with the fund

(HESTA, 2006). HESTA first offered choice to its members in 1995. STA initially

offered choice to its members in July 1997. On 1 July 2006, STA merged with the

Australian Retirement Fund (ARF) and Finsuper to form AustralianSuper.

AustralianSuper represents more than 1.2 million members and over $20 billion in

funds under management. As at June 2006, GESB had over 270,000 members

(primarily from the health and community services industry), 45,000 participating

employers, and GESB managed $6.85 billion in funds for its members (GESB, 2006).

GESB first offered choice to its members in April 2001. UniSuper had 364,948

members as at 30 June 2006 and total assets under management of $19.3 billion

(UniSuper, 2006). UniSuper first offered choice to its members in 1998.

34

HESTA, STA, and GESB offer members the choice ofreadymade options (a set

of specified investment strategies set by the fund) or a do-it-yourself strategy, whereby

members actively choose their own investment strategy. Members may also blend the

readymade options and the do-it-yourself strategy. UniSuper offers members a set of

readymade investment options. The investment offerings of the four funds are

summarised in Appendix B Table 2.

4.3. Research Methods

4.3.1. Preliminary Analysis

Before employing a regression framework to examine equity participation and

equity allocation within member superannuation accounts, a summary of actual age­

based member asset allocations is presented. This involves arranging members who

made investment choice changes to their asset allocation into age quintiles. For each of

the major asset classes, average allocations are then calculated. Member asset

allocations are also sorted by gender (broken further into age quintiles) and marital

status. This analysis complements the regression analysis.

4.3.2. Regression Analysis

Regression analysis was used to individually investigate the two key portfolio

decisions: equity participation and equity allocation. To examine the effect of age on the

likelihood of stock ownership, a Probit regression was employed. As equity

participation is a discrete variable (that is, member chooses equity or member does not

choose equity), it is inappropriate to use the simple Ordinary Least Squares method.

Also, the use of a more sophisticated binary response model like Probit allows certain

limitations of the linear probability model to be overcome (Wooldridge, 2006)

including: (1) fitted probabilities can be negative or greater than one; and (2) the partial

effect of any independent variable (appearing in level form) is constant. The Probit

model is estimated using Maximum Likelihood estimation, whereby the most likely

values of regression parameters are estimated given the specific data set.

Equation 1 describes the relationship between the likelihood of equity

participation and age:

(1)

35

where

<I> is the standard cumulative normal distribution;

SHARE_ ALLOCi,r denotes whether a member has an equity allocation (1) or not (0);

AGEi, is the age of the member (in years) when decision made;

x;8 is a vector of explanatory variables which are described further in Appendix B

Table 1; and

u i 1 is a normally distributed error term.

Two versions of the above model are estimated in the next chapter. In the first

version, equity participation was regressed on age, gender, and member contributions

only. This is the basic member information made available by the funds and available to

the funds. In the second version, equity participation was regressed on a number of

additional variables including marital status, risk profile, education, whether a member

invests in shares or managed investment funds external to their superannuation account,

and net wealth (excluding superannuation) to explore conditioning of age effects on the

likelihood of equity participation. This data was sourced through a survey conducted on

participants in 2005 as part of a larger study into the four funds.

To investigate the independent effect of age on the level of equity allocation, and

the significance of age relative to other member characteristics in explaining equity

allocation, a Tobit regression (censored regression) was used. Using Ordinary Least

Squares is inappropriate in this case as it ignores the fact that some of the observations

in the sample are censored, that is members can allocate between zero and 100 percent

only with no short-selling permitted in any asset class. Ordinary Least Squares would

lead to biased estimates of equity allocation. The Tobit model, which is estimated using

the Maximum Likelihood technique, is well-suited to the dependent variable employed

in this study as equity allocation is equal to zero (non-stock ownership) for a significant

fraction of the sample, and the positive values for equity allocation are approximately

continuously distributed (Wooldridge, 2006, p. 595). Only those members who hold

equity in their superannuation portfolios were included, since equity participation and

equity allocation conditional on ownership were being analysed independently.

Equation 2 describes the relationship between equity allocation and age.

36

SHARESi,r = {3 0 + f31AGEi,r + x;o + ui,r, if 0 < SHARESi,t < 1;

(2)

where

SHARESi,t denotes a member's percentage allocation to equity in their superannuation

account conditional on them having equity and is bounded at 1 00%;

AGEi 1 is the age ofthe member (in years) when decision made;

x;o is a vector of explanatory variables which are described further in Appendix B

Table 1; and

ui 1 is a normally distributed error term.

Similar to the estimations of the Probit model in Equation 1 specified

previously, estimation of the Tobit model involved regressing equity allocation on: (1)

age, gender, and member contributions only; and (2) additional explanatory variables,

including marital status, risk profile, education, whether a member invests in shares or

managed investment funds, and net wealth in order to test the relative importance of age

in determining equity allocation conditioned on these variables.

Each of the regressions explained above were performed for each individual

superannuation Fund 17• All regressions are controlled for time effects or "trends" in

equity ownership and allocation as described by Ameriks & Zeldes (2004), by including

year dummy variables. That is, the regressions control for any potential effects related

to the date of observation (the year a particular investment choice is made).

Further, rather than include age as a continuous variable in the regressions,

which assumes a linear relationship between age and equity participation/equity

allocation, members were classified into age quintiles to capture possible non-linear age

effects. Analytical procedures were performed using Eviews (version 5).

17 HESTA, STA, and GESB (for the full sample) and HESTA, STA, GESB, and UniSuper (for the restricted sample).

37

5. FINDINGS AND DISCUSSION

This chapter summarises superannuation member asset allocations sorted by age,

gender (broken further into age quintiles) and marital status, and presents a comparison

of actual age-based member allocations with that suggested by available life-cycle

products. It also presents and discusses the results of an empirical investigation of a

sample of members' investment strategy decisions and key demographics, notably age.

A set of Probit and Tobit regressions was performed on the full sample and restricted

sample of superannuation fund data. The full sample contains 93042 observations

including membership data collected by the four superannuation funds (HESTA, STA,

GESB, and UniSuper), for all member investment strategy changes since investment

choice was introduced by the respective funds. The restricted sample contains 2315

observations including the combined membership and additional survey information for

members who have exercised choice since the respective fund first introduced

investment choice, and responded to a survey which examined the intentions of the

members to change their investment strategy and make extra contributions. All

regressiOns were performed for each individual fund resulting in a total of 14

regressiOns.

5.1. Preliminary Results

5.1.1. Equity Allocations

Tables 5.1 and 5.2 summarise member asset allocations, showing the average

percentage of the superannuation portfolio held in cash, fixed interest, property, shares,

and an asset class termed 'other'. The observations in the two key datasets are sorted by

age, gender (further sorted by age), and marital status18. Appendix C Tables 1 and 2

report the average relative asset allocations for fund members. This analysis excludes

observations where members adopt only one readymade option19 as opposed to a do-it­

yourself strategy, or a mixture of both the readymade option and the do-it-yourself

strategy. The analysis is therefore based on members who can be regarded as actively

18 Information regarding members' marital status is available for the restricted data set only. 19 A specified investment strategy which is set by the fund.

38

choosing an asset class allocation level by choosing a given percentage or have chosen

an allocation by blending readymade options.

Overall, the results suggest that age has a discernible effect on the asset

allocation. There are some marked patterns in superannuation member asset allocation

which both support and run counter to the age-phasing advice frequently offered. In

Table 5.1, the overall average allocation to shares is 55.41 percent. When sorted by age

group, a hump-shaped pattern is evident in the allocation to shares; on average,

members under 36 allocate 56.99 percent of their superannuation account to shares,

members in the 36-42 age groups allocate 57.25 percent to shares, and members in the

43-48 age group allocate 60.01 percent to shares. For the 49-54 age group the average

allocation decreases to 52.99 percent, and falls further to 41.93 percent for those

members over 55. The average allocations to "safer" assets (cash and fixed interest)

exhibit an inverse pattern to this, decreasing during a key portion of an individual's

working life, and increasing significantly in the immediate period before retirement.

This hump-shaped pattern in equity allocation is generally consistent with existing

empirical findings [for example, Yoo (1994), Poterba & Samwick (1997), Agnew,

Balduzzi, & Sunden (2003), and Ameriks & Zeldes (2004)].

Using pooled cross-sectional data from five waves of the Survey of Consumer

Finances between 1983 and 1998, and TIAA-CREF panel data covering the 1987 to

1999 period, Ameriks & Zeldes (2004) examine the relationship between equity

allocation and age. To address the inherent identification problem that time, age, and

cohort effects do not change independently, Ameriks & Zeldes (2004) estimate a model

with age and time effects only, and a model with age and cohort effects only. Based on

a regression specification with age and time effects (excluding cohort effects), they find

that the share of equity in financial assets has a hump-shape pattern with age.

In a study of7,000 retirement accounts in a large 401(k) plan, Agnew, Balduzzi,

& Sunden (2003) reveal a hump-shaped age-equity allocation profile when observations

are sorted by age, with mean equity allocations being 37.50 percent, 42.50 percent, and

44.01 for the 35, 35 to 44, and 45 to 54 age groups. For the 55 to 64 age group the mean

allocation decreases to 37.65 percent, and falls further to 4.75 percent for those

participants over 65. Although their initial regression results suggest that age has a

negative effect on the equity allocation (more specifically, each additional year

translates into a lower allocation to stocks by 93 basis points), a specification including

39

both age and age squared to capture possible non-linearity in the relationship between

equity allocation and age reveals that equity allocation peaks very early, at 32.5 years of

age.

The results for the smaller dataset of members replying to the survey presented

in Table 5 .2, suggest a negative linear age-equity ownership profile; the average equity

allocations are 63.53 percent, 60.87 percent, 58.85 percent, 55.35 percent, and 46.26

percent for the under 36, 36-42, 43-48, 49-54, and over 55 age groups respectively.

When those observations containing only one readymade option are excluded

from the two datasets, the asset allocation percentages remain quite similar to those in

Table 5.1 and Table 5.2 however there are some notable differences. As seen in

Appendix C Table 1, although the average allocation to equity does primarily increase

with age and then decrease (from the 43-48 age group), the initial increase is relatively

flatter than previously; there is not a great deal of change in equity allocation within the

first few age groups. In Appendix C Table 2, the average percentage of the

superannuation portfolio held in equity falls with age with the exception of the 43-48

age group, in which the equity percentage rises just over one percentage point. An

additional observation is that when sorted by age only, the decrease in equity allocation

from the 49-54 age group to the over 55 age group, is considerable. This is maintained

across all four breakdowns discussed above.

When observations are sorted by gender the results show that on average, male

members allocate a larger percentage of their superannuation accounts to equity than

their female counterparts. In the full sample presented in Table 5.1, males allocate 55.71

percent to equity compared to 50.28 percent for females. In the restricted sample

presented in Table 5.2, males allocate 59.21 percent to equity compared to 55.26

percent for females. Appendix C Table 1 and Table 2 show that the exclusion of those

observations with only one readyrnade option does not affect this observation. This is in

line with the view of previous studies that men are more overconfident relative to

women or report more confidence in areas such as finance (Barber & Odean, 2001).

Prince (cited in Barber & Odean, 2001, p. 265) asserts that men tend to feel more

competent than women do in financial matters. As a result, it is expected that men trade

more and hold more risky portfolios than women. Barber & Odean, (2001) estimate the

portfolio risk characteristics in terms of gender. The results from their time-series

40

regression analysis reveals that women are inclined to hold less risky portfolio positions

than their male counterparts.

When the observations for both datasets are sorted by both gender and age, the

age-equity ownership patterns evident are similar to those exhibited when the

observations are sorted by age only. In the full sample presented in Table 5.1, equity

allocation increases with age and then decreases. For males, the average equity

allocations are 57.29 percent and 58.45 percent for the under 36 and 36-42 age groups.

The average allocation then decreases to 56.36 percent for the 43-48 age group, and

further decreases to 51.28 percent and 39.76 percent for the 49-54 and over 55 age

groups respectively. For females, those members under the age of 36 allocate 56.76

percent of their superannuation portfolio to shares. This increases and peaks at 59.71

percent for the 36-42 age group. The average allocation then decreases to 58.65 percent

for the 43-48 age group, and further decreases to 54.29 percent and 44.26 percent for the

49-54 and over 55 age groups. One key difference between the mean allocations sorted

by age and the mean allocations sorted by gender and age, is that those allocations

sorted by gender and age that exhibit a hump-shaped pattern appear to peak earlier in

the life-cycle (36-42 age group) rather than later ( 43-48 age group).

A general negative linear age-equity ownership profile is evident in Table 5.2 as

average equity allocations decrease as a function of age, with the exception of the 49-54

age group for males, where the average allocation actually increases by just over two

percentage points. For males, the average equity allocations are 66.19 percent, 61.13

percent, 60.45 percent, 62.63 percent, and 44.51 percent for the under 36, 36-42, 43-48,

49-54, and over 55 age groups respectively. For females, average allocations are 61.66

percent, 60.66 percent, 57.46 percent, 50.50 percent, and 47.48 percent for the under 36,

36-42, 43-48, 49-54, and over 55 age groups respectively. When those observations

containing only one readymade option are excluded from the two datasets, the asset

allocation percentages are relatively similar to those in Table 5.1 and Table 5.2. For the

full sample presented in Table 5.1, a hump-shaped pattern prevails peaking at the 36-42

age group for both males and females. For the restricted sample presented in Table 5.2,

the average allocation to shares decreases monotonically for females, but is less stable

for males.

Another important observation is related to the marital status of individuals.

Married members tend to hold a higher allocation to equity in their superannuation

41

accounts. In the full sample presented in Table 5.2, the average equity allocation for

married members is equal to 58.32 percent, while unmarried members allocate an

average of 54.33 percent. In Appendix C Table 2, the average allocation for married

members is equal to 43.91 percent, while the average for unmarried members is 39.97

percent. This is consistent with the finding of Agnew, Balduzzi, & Sunden (2003), who

report that the average allocation to equity for married participants is 42.88 percent

compared to 36.52 for single participants. They (Agnew, Balduzzi, & Sunden, 2003)

offer two potential explanations for this pattern: (1) idiosyncratic labour-income shocks,

and (2) a stronger bequest motive. First, Agnew, Balduzzi, & Sunden (2003) explain

that couples with dual income earners experience diversification of idiosyncratic labour­

income shocks which exposes their non-financial income to less risk (Agnew, Balduzzi,

& Sunden, 2003, p. 198). As a result, these individuals are encouraged to undertake

more aggressive asset allocations than single investors. Second, they state that the

bequest motive is stronger for married couples. The bequest motive extends an

investor's horizon beyond his life span, and models of optimal portfolio choice predict

higher allocation to equities the longer the time horizon (Agnew, Balduzzi, & Sunden,

2003, pp. 198-199). Following this argument, married couples have a longer time

horizon than single participants, as a result of their stronger bequest motive and

therefore have a higher allocation to equities.

42

Table 5.1 Average Asset Allocations by Group- Full Sample This table presents statistics for average member asset allocations, showing the average percentage of the superannuation held in cash, fixed interest, property, shares, and 'other'. Observations are sorted by age only, and gender and a~e.

Obs Cash F.Interest Pro~erty Shares Other ALL 93042 9.74 17.09 12.49 55.41 5.28 AGE: <36 35745 9.49 16.00 12.12 56.99 5.40 36-42 18189 8.15 17.22 12.27 57.25 5.11 43-48 12362 6.60 14.82 13.89 60.01 4.68 49-54 11029 9.69 19.48 12.81 52.99 5.02 55+ 8829 16.94 25.55 11.04 41.93 4.54 Unknown 6888 11.70 11.71 13.82 55.23 7.54 GENDER: Male 38297 9.22 17.97 11.64 55.71 5.47

<36 15646 8.58 17.01 11.60 57.29 5.51 36-42 6380 7.52 17.19 12.02 58.45 4.82 43-48 6916 7.48 19.00 12.55 56.36 4.61 49-54 4788 10.09 22.22 11.90 51.28 4.51 55+ 4564 17.24 28.91 9.96 39.76 4.13 Unknown 3 4.38 3.47 11.82 71.12 9.22

Female 47848 13.55 18.52 12.83 50.28 4.82 <36 20090 10.20 15.22 12.52 56.76 5.31 36-42 7874 7.55 14.42 13.24 59.71 5.09 43-48 9378 7.54 15.12 13.55 58.65 5.12 49-54 6240 9.38 17.39 13.51 54.29 5.42 55+ 4265 16.61 21.95 12.19 44.26 4.98

Unknown 1 30.00 27.00 12.00 28.00 3.00

Gender unknown 6897 11.68 11.71 13.83 55.24 7.55

43

Table 5.2 Average Asset Allocations by Group Restricted Dataset This table presents statistics for average member asset allocations, showing the average percentage of the superannuation held in cash, fixed interest, property, shares, and 'other'. Observations are sorted by age only, gender and age, and marital status.

Obs Cash F.Interest Pro,eert~ Shares Other ALL 2315 10.35 17.35 10.36 57.23 6.21 AGE: <36 476 5.51 16.34 9.93 63.53 6.99 36-42 414 8.57 15.47 11.18 60.87 5.51 43-48 558 9.54 16.89 10.31 58.85 5.97 49-54 452 10.65 18.20 10.82 55.35 6.33 55+ 415 18.45 20.05 9.62 46.26 6.25 GENDER: Male 1001 10.08 16.48 9.41 59.21 6.01 <36 196 4.47 14.27 9.56 66.19 7.11 36-42 183 8.94 14.64 10.15 61.13 6.44 43-48 260 9.83 15.88 9.96 60.45 5.18 49-54 181 5.92 17.94 8.59 62.63 6.06 55+ 181 21.84 20.14 8.53 44.51 5.52

Female 1306 10.60 18.08 11.02 55.76 6.29 <36 280 6.23 17.79 10.18 61.66 6.88 36-42 231 8.27 16.13 12.00 60.66 4.59 43-48 298 9.29 17.77 10.62 57.46 6.67 49-54 271 13.81 18.38 12.31 50.50 6.52 55+ 226 16.28 20.47 10.04 47.48 6.41

Gender unknown 8 3.00 6.25 22.38 51.25 17.13 MARITAL STATUS: Married/De-facto 1732 10.58 16.76 9.86 58.32 5.81 Unmarried 543 9.22 18.62 12.12 54.33 7.57 Unknown 40 15.56 25.43 8.48 49.62 4.09

The actual member asset allocations described above are not completely consistent with

the operation of life-cycle funds. Whereas life-cycle funds follow a general decreasing

trend (either through a progressive or step-like operation), the above trends suggest that

members of the superannuation funds either follow a decreasing pattern or hump-shaped

pattern in their asset allocations over the life-cycle.

44

5.2. Regression Results

5.2.1. Effect of Age on Equity Participation and Equity Allocation

Tables 5.3 and 5.4 summarise the estimation results20 from the "restricted"

regressions, whereby equity participation and equity allocation are regressed on age,

gender, member contributions, and a set ofyear dummies only.

The results confirm that there is a significant relationship between member age

and (1) equity participation, and (2) equity allocation. As seen in Table 5.3, the

estimated coefficients on all age group variables are statistically significant for STA,

GESB, and two of the four categories for HESTA at the 99% level, and at 95% for one

of the four categories for REST A. The results suggest that only for HESTA members is

the relationship between age and equity participation non-linear or supportive of a

hump-shaped pattern. Relative to the base age group (under 36), those members in the

36 to 42 age group are more likely to hold shares in their superannuation account. From

this point, members are increasingly less likely to hold shares than members under 36.

For STA and GESB, the probability of equity participation decreases monotonically as

members become older.

The results for STA suggest that male members are more likely to hold shares

than female members and the amount of members' total contributions to their

superannuation account has a positive effect on the likelihood of equity participation

(that is, the higher the contribution amount the more likely a member is to hold equity),

however this estimated effect is modest. All time dummies are statistically significant at

the 99% level.

The Tobit regression results (in the form of estimated coefficients) in Table 5.4

also exhibit a hump-shaped pattern across all funds; members initially increase their

equity allocations up to a certain age, after which point the amount held in equity

decreases. Relative to the base age group, those members in the 36 to 42 age group hold

more shares in their superannuation account. The results for HESTA suggest that

members in the 43 to 48 age group also hold more shares than the base group (but less

than the 36 to 42 age group). From this point, members hold increasingly less shares

than members under 36.

20 The heteroskedasticity- robust standard errors reported in parentheses have been estimated using the quasi-maximum likelihood (Huber/White) method. Note that in the case of binary dependent variable

45

Other general findings include: male members hold more equity in their super

accounts than their female counterparts; the higher the contribution amount, the larger

the portion of the superannuation account that is held in equity (again, this effect is

modest); and, all time dummies are statistically significant21 at the 99% level.

It should be noted however, that although the estimated Probit and Tobit

regression results allow for determining the statistical significance of the variables of

interest, and whether a particular variable has a positive or negative effect on the

dependent variable, they do not provide the magnitude of the effect an explanatory

variable has on the dependent variable. The estimated coefficients from a limited

dependent variable model cannot be interpreted as the marginal effect on the dependent

variable. In the case of binary response models, unlike the linear probability model

which assumes constant marginal effects, the Probit model implies diminishing

magnitudes of the partial effects (Wooldridge, 2006, p. 592).

models, these standard errors are not robust to heteroskedasticity; they are robust to certain misspecifications of the underlying distribution of y (QMS, 2004). 21 With the exception of the 2003 time dummy variable for STA.

46

Table 5.3 Regression Results for Pro bit Estimation of Equity Participation: Full Sample

P(SH _ALLOCi,t = 1)= <I>(/3 0 + /31AGEi,t + x;o + ui,t)

This table presents the estimation results from the "restricted" Probit regressions, whereby equity participation was regressed on age, gender, member contributions, and a set of year dummies only. The following are coefficient estimates, not marginal effects.

Dependent Variable: Equity Participation (SH ALLOC) Estimate

Variable HESTA STA GESB Constant 1.615933 0.860023 1.427831

(0.025661) (0.035686) (0.062956) Age Categories:

36-42 0.032026 -0.188799*** -0.276802*** (0.028431) (0.037657) (0.061543)

43-48 -0.052326** -0.367312*** -0.321091 *** (0.025781) (0.035813) (0.058015)

49-54 -0.151289*** -0.604358*** -0.680473*** (0.029232) (0.038224) (0.060921)

55+ -0.302194*** -0.983167*** -1.135710*** (0.0311202 (0.0380952 (0.065020)

Member Characteristics:

Male -0.033416 0.036005 -0.031697 (0.021694) (0.029923) (0.041442)

Er cants 1.09E-05*** 1.40E-06 1.87E-05** ~2.42E-062 ~1.55E-06) (8.85E-06)

Calendar Years: 1997 1.695431 ***

(0.280952) 1998 0.934702***

(0.122807) 1999 1.899083***

(0.134524) 2000 1.809132***

(0.099094) 2001 0.392054 *** 0.783246*** 0.873412***

(0.058650) (0.055548) (0.053773) 2003 -0.598120*** -0.136007*** -0.249957***

(0.024430) (0.029622) (0.058080) 2004 -0.237352*** 0.682776*** 0.564739***

(0.028367) (0.037414) (0.079646) Observations 35318 17609 12831 Obsw/Dep= 0 3720 3241 781 Obsw/Dep= 1 31598 14368 12050

Pseudo R 2 0.047335 0.177322 0.186300

LR Stat (9 df) 1125.566 2981.857 (13 df) 1096.482

Note: robust standard errors are in parentheses. Estimated using the quasi-maximum likelihood (Huber/White) method. In the case of binary dependent variable models, these standard errors are not robust to heteroskedasticity; they are robust to certain misspecifications of the underlying distribution of y (QMS, 2004).

***, **, * denotes significance at 1%, 5%, 10% confidence level, respectively.

47

Table 5.4 Regression Results for Tobit Estimation of Equity Allocation: Full Sample

SHARESi,t = {3 0 + {31AGEi,t + x;o + ui,t, if 0 < SHARESi,t < 1

This table presents the estimation results from the "restricted" Tobit regressions, whereby equity allocation was regressed on age, gender, member contributions, and a set of year dummies only. The following are coefficient estimates, not marginal effects.

Dependent Variable: Equity Allocation (SHARES) Estimate

Variable HESTA STA GESB Constant 64.65755 56.92620 61.16975

(0.344541) (0.535129) (0.574870) Age Categories:

36-42 1.896169*** 0.822122* 0.586314* (0.380522) (0.436620) (0.326814)

43-48 1.331016*** -0.221786 -1. 711664*** (0.362292) (0.451044) (0.335684)

49-54 -0.482485 -3.171667*** -3.397259*** (0.415857) (0.580716) (0.501597)

55+ -4.149244*** -8.592687*** -7.382350*** (0.500664) ~0.7250982 (0.810674)

Member Characteristics:

Male 2.763531 *** 0.798730** 2.342647*** (0.308650) (0.398863) (0.293967)

Er cants 1.26E-06 0.000158*** 1.55E-05 (2.44E-052 (2.43E-05) (5.80E-05)

Calendar Years: 1997 10.71085***

(0.854515) 1998 -4.963126*** 9.680279***

(0.498334) (0.817416) 1999 -2.397903*** -7.654025***

(0.484552) (0.682267) 2000 6.863969*** 4.661718***

(0.405979) (0.512912) 2001 7.706509*** 7.880753*** 5.251008***

(0.460982) (0.595672) (0.544927) 2003 -6.798784*** -0.157651 -3.927489***

(0.378365) (0.558409) (0.786086) 2004 -1.826013*** 7.371387*** 2.361901 ***

(0.412079) (0.559904) (0.764905) Observations 30960 14368 12050

Pseudo R 2 0.047362 0.066733 0.065397

Log Likelihood -140682.4 -62785.20 -49380.87

Note: robust standard errors are in parentheses. Estimated using the quasi-maximum likelihood (Huber/White) method. ***,**,*denotes significance at 1%, 5%, 10% confidence level, respectively.

48

To summarise the effect of the explanatory variables on the probability of equity

participation (SH_ALLOC) and the amount of equity held (SHARES), the standard

method of computing the marginal effects at the mean of the explanatory variables (that

is, replacing each explanatory variable with its sample mean) is employed (Dougherty,

2002, p. 291) and requires the calculation of scale factors22. Tables 5.5 and 5.6 report

the estimated 200423 marginal effects for the age variables.

Table 5.5 Estimated Probit Age Marginal Effects (2004) This table presents the estimated 2004 age marginal effects from the "restricted" Probit regressions, whereby equity participation was regressed on age, gender, member contributions, and a set of year dummies only.

HESTA STA GESB 36-42 43-48 49-54 55+

Table 5.6

0.004724 -0.008663 -0.028420 -0.067545

Estimated Tobit Age Marginal Effects (2004)

-0.028392 -0.023746 -0.069760 -0.029740 -0.148968 -0.109198 -0.327121 -0.303731

This table presents the estimated 2004 age marginal effects from the "restricted" Tobit regressions, whereby equity allocation was regressed on age, gender, member contributions, and a set of year dummies only.

36-42 43-48 49-54 55+

HESTA STA GESB 0.009319 0.007039 -0.003215 -0.043270

0.000752 -0.000245 -0.005879 -0.038745

0.000007 -0.000041 -0.000136 -0.000919

With respect to the likelihood of equity participation, the marginal effects for

HESTA members reveal a marginal hump shape in the age-participation profile, with an

initial increase early in the life-cycle, and decreasing from the 43 to 48 age group.

Ameriks & Zeldes (2004) also find a hump-shaped age-ownership profile when they

22 As all explanatory variables (except ER _ CONTS) included in the regression models are binary variables, the marginal effects are calculated as follows: the binary explanatory variables are set to one

~

and the sole continuous variable is set to its average in order to generate a value for the index, x' f3 . The

standard normal cumulative distribution function is applied to this value to compute a scale factor, then multiplied by the respective AGE coefficient. Age marginal effects are calculated for each age dummy variable, for each calendar year included in a particular regression. As the education and wealth explanatory variables are categorical, it does not make sense to set all variables to one when calculating the marginal effects. The following variables are set to one to calculate the index value; POST GRAD DEG and HIGH WEALTH. 23 The-age ma~ginal effects reported in the above tables are those for the year 2004, to allow for a meaningful comparison between the funds.

49

estimate a model with age and time effects, excluding cohort effects from the

specification24. For STA and GESB members, the marginal effects exhibit a general

negative relationship between age and the probability of equity participation. For all

funds, the probability of owning equity in the superannuation account decreases

considerably between the 49 to 54 and over 55 age groups.

In relation to the amount of equity held (conditional on equity participation), the

marginal effects suggest that for members of HESTA and STA, the amount of equity

held in their superannuation accounts follows a very slight hump-shaped pattern25.

However this pattern is less evident for GESB members, with the estimated age-equity

allocation profile being extremely flat.

In terms of time effects, the estimated age marginal effects across the different

years do not reveal any consistent trends in equity participation and equity allocations

related to the date of observation. However, there is some marked variation in equity

participation and equity allocations for some years which are not reported in the tables

but are discussed below.

Firstly considering the funds on an individual basis the age-participation profile

for HESTA members was relatively flat during 2001, but much more pronounced in

2003 and 2004 exhibiting a hump-shaped profile. The marginal effects for the first three

age groups are similar with the only noteworthy difference being that during 2004, there

was an increased probability for members to hold equity in their superannuation

accounts. In terms of equity allocations, the 2000 and 2001 age-equity allocation

profiles are flat. The remaining years are all quite similar to each other, with the notable

difference being that for the first two age groups, members held increasingly slightly

more equity in their accounts in 1998 and 2003 compared to 1999 and 2004. This

pattern is reversed for the two oldest age groups in 1998 and 2003, whereby members

allocated increasingly less equity than in 1999 and 2004.

A potential reason for these "trends" in equity participation is that high stock

returns during these periods may have generated increased public attention, resulting in

some members learning about and participating in the stock market Ameriks & Zeldes

24 However, their regression results indicate a significant hump-shaped pattern. The regression specifications estimated by Ameriks & Zeldes (2004) differ to those estimated in this thesis, as they estimate fixed-effects models. 25 The amount of equity held peaks between the age of 43 to 48 for HESTA members, and 36 to 42 for STA members.

50

(2004). The increase in the size of investment menus in recent years may also be a

possible reason for such "trends".

For STA members, the 1997, 1999, and 2000 age-ownership profiles are all flat,

whereas the remaining years exhibit an overall negative age effect on the likelihood of

holding equity. This effect is most evident in the 2003 marginal effects, where the

progressive reduction in the likelihood of equity participation is higher across all age

groups compared to the other years. In relation to equity allocation, for all years (except

1999) the age-ownership profiles are flat for the first three age groups. The 1999 profile

stands out as the decrease in the amount held in shares is larger for the two oldest age

groups than in any other year. Perhaps this suggests that the circumstances of the

members in the older age groups change significantly at these age groups and as a result

there is a need to revise the superannuation portfolio to one that is more appropriate for

their position in the life-cycle.

Overall for GESB members the marginal effects for all years are similar, with

the exception of the 49 to 54 and over 55 age groups, where the decrease in the

likelihood of equity participation is 9 percent (13 percent) smaller than 2001 (2003) for

the 49 to 54 age group, and 13 percent (15 percent) smaller than 2001 (2003) for the

over 55 age group. In terms of equity allocations, the 2001 and 2004 age-equity

allocation profiles are flat. The major difference is visible from the 2003 marginal

effects where the decrease in equity for the over 55 age group is significantly larger than

in 2001 and 2004.

Comparing the 2001, 2003 and 200426 marginal effects across the three funds, it

IS evident that in terms of equity participation the age-ownership profile is flat for

HESTA, and exhibits a decreasing trend for STA and GESB across all three years. In

terms of equity allocation the age marginal effects for GESB produce an extremely flat

age-equity allocation profile, whereas those for HESTA and STA present a general

negative trend which is more visible for STA during 2001 and 2002.

Overall, the marginal effects suggest that age is an important variable in

determining equity participation and equity allocation; this is similar to the existing

literature though the results are mixed. Two conflicting patterns are found with respect

to the age-equity participation and age-equity allocation profiles, hump-shaped and

26 These are the three years common to all three funds.

51

decreasing. The hump-shaped pattern is consistent with the findings of Ameriks &

Zeldes (2004) and Y oo ( 1994 ), whilst the decreasing pattern is consistent with the

findings ofBodie & Crane (1997) and Schooley & Worden (1999).

5.2.2. Effect of Age Relative to other Variables on Equity Participation and

Equity Allocation

Appendix C Tables 5.7 and 5.8 summarise the results from estimating the

regressions whereby equity participation and equity allocation are regressed on age,

gender, member contributions, time dummies, and a set of additional explanatory

variables as described in Section 4.

Overall, when additional variables are introduced, the age-equity

participation/allocation profiles exhibited are extremely varied across the four

superannuation funds. In the case of equity participation, HESTA members aged 36 to

42 are less likely to hold shares than members under 36; members in the 43 to 48 and 49

to 54 groups are increasingly more likely to hold shares than the base age group, whilst

members over the age of 55 are less likely to hold shares than the base age group. For

the STA and GESB funds, members between 36 and 42 are more likely to hold shares

than the under 36 group; members in the 43 to 48, 49 to 54, and 55 plus age groups are

increasingly less likely to hold shares than the base age group. For UniSuper, members

of all age groups are less likely to hold shares than the base age group, with the

coefficients exhibiting no apparent trend.

Other general findings include: male members are more likely to hold equity in

their superannuation accounts than their female counterparts; the higher the contribution

amount, the larger the portion of the superannuation account that is held in equity,

however this effect is exceedingly small; those individuals who are married or in a de­

facto relationship are more likely to hold shares than unmarried members27; individuals

who are willing to take substantial risk in expectation of earning substantial returns are

more likely to hold shares than those individuals who are prepared to assume low to

average financial risks; investing in shares or managed fund accounts outside of their

superannuation account increases the probability of a member owning equity; and,

individuals who have an average or high level of net wealth are less likely to hold shares

than individuals with lower net wealth.

27 This is the case for the REST A and GESB funds only.

52

The effects of the education variables on equity participation are less clear, with

the regressions producing mixed results across the four funds. For HESTA, members

who have completed a certificate, trade or diploma level of education, a bachelor

degree, or a post graduate degree are more likely to hold shares relative to individuals

who have completed some level of secondary education. For STA, members who have

completed a higher level of education are less likely to hold shares in their

superannuation account than members limited to some level of secondary education. For

GESB, members who have completed a certificate, trade or diploma level of education

or a post graduate degree, are less likely to hold shares than the base group; members

with a bachelor degree are more likely to hold shares than the base group. For

UniSuper, members who have completed a certificate, trade or diploma level of

education are less likely to hold shares than the base group; members with a bachelor

degree or post graduate degree are more likely to hold shares than the base group.

As displayed by the Probit results, the estimated Tobit coefficients on the age

variables also suggest mixed patterns in equity allocation. For HESTA, members in the

first three age groups hold more of their super account in shares than members under 36,

with the amount allocated to shares decreasing as members age. For UniSuper, members

of all age groups hold less equity in their super accounts than members under 36. A

decreasing trend is evident up until the 49 to 54 age range, after which point the amount

allocated to equity actually increases. There is no clear trend evident for STA and

GESB.

Other general findings include: male members hold more equity in their super

accounts than their female counterparts; the higher the contribution amount, the smaller

the portion of the superannuation account that is held in equity (although this effect is

modest); married members hold more shares than unmarried members; individuals who

are willing to take substantial risk hold more shares than those individuals who are

prepared to assume low to average financial risks; individuals who have completed a

higher level of education hold a larger portion of their superannuation accounts in

equity; members who invest in shares or managed fund accounts outside of their super

accounts hold more shares; and average to high net-worth individuals hold more shares

in their super accounts.

53

Table 5.7 Regression Results for Pro bit Estimation of Equity Participation: Restricted Sample

P(SH _ALLOCi,t =l)=ct>(fio + fi,AGE;, 1 +x;o +u;,1 )

This table presents the results from estimating the Probit regressions, whereby equity participation was regressed on age, gender, member contributions, marital status, risk profile, education, whether a member invests in shares or managed investment funds external to their superannuation account, net wealth (excluding superannuation), and a set of year dummies. The following are coefficient estimates, not marginal effects.

De,eendent Variable: Eguitl: Partici,eation (SH ALLOC) Estimate

Variable HESTA STA GESB UniSu,eer Constant 1.043422 1.097737 1.155851 0.287083

(0.316705) (0.716706) (0.420719) (0.325699) Age Categories:

36-42 -0.098723 0.975765* 0.083097 -0.081512 \

(0.251382) (0.528095) (0.351762) (0.218209) 43-48 0.105601 -0.308671 -0.085926 -0.035294

(0.270428). (0.393744) (0.330771) (0.229523) 49-54 0.133688 -0.360600 -0.362308 -0.430496*

(0.252897) (0.598528) (0.381396) (0.233492) 55+ -0.232546 -1.274025** -0.550659 -0.336276

(0.262625) (0.526203) (0.4590472 (0.236523) Member Characteristics:

Male 0.025793 0.300144 0.068587 0.239354* (0.158501) (0.272893) (0.269501) (0.129651)

Er cants 5.86E-06 5.94E-05** 6.84E-05 6.34E-06 (7.96E-06) (2.61E-05) (6.00E-05) (5.72E-06)

Married/Defacto 0.329486* -0.025712 0.119201 -0.221350 (0.168660) (0.328198) (0.228780) (0.174020)

High Risk 0.363422** 0.996958** 0.818604*** 0.318120** (0.152860) (0.388720) (0.292190) (0.130183)

Cert/Trade/Dip 0.087502 -0.244605 -0.318299 -0.105883 (0.204440) (0.514765) (0.326411) (0.265790)

Bachelor Deg 0.196306 -1.480967** 0.242741 0.148271 (0.227277) (0.619509) (0.380039) (0.256544)

Post Grad Deg 0.047769 -0.313423 -0.197643 0.022828 (0.239870) (0.662422) (0.355565) (0.239354)

Shares/MIF 0.151936 0.322818 0.727179*** 0.240282* (0.153941) (0.290089) (0.269087) (0.135229)

Avg Wealth -0.118298 -0.685833* -0.626624** -0.014606 (0.181502) (0.394670) (0.308552) (0.174915)

High Wealth -0.054248 -0.073895 -0.770781 ** -0.282091 (0.198692) (0.458153) (0.378963) (0.1977812

Calendar Years: 1997

1998 0.683397* (0.363815)

1999 0.965457*** (0.216587)

2000 1.694088*** (0.439681)

2001 0.572937 0.309637*** 0.974118*** (0.437353) (2.626398) (0.267770)

54

2003 -1.024579*** -0.743937** -0.990281 *** 0.519798*** (0.188328) (0.325857) (0.317939) (0.151786)

2004 -0.044242 1.082482** 1.422983*** (0.208866) (0.463334) (0.240112)

Observations 598 154 456 673 Obsw/Dep= 0 75 21 27 119 Obsw/Dep= 1 523 133 429 554

Pseudo R 2 0.158758 0.372111 0.359655 0.158251

LR Stat (17 df) 71.69320 45.65007 (16 73.73222 (16 99.37536 (20 df) df) df)

Note: robust standard errors are in parentheses. Estimated using the quasi-maximum likelihood (Huber/White) method. In the case of binary dependent variable models, these standard errors are not robust to heteroskedasticity; they are robust to certain misspecifications of the underlying distribution of y (QMS, 2004). * * *, * *, * denotes significance at 1%, 5%, 10% confidence level, respectively.

55

Table 5.8 Regression Results for Tobit Estimation of Equity Allocation: Restricted Sample

SHARESu = /3 0 + f31AGEu + x;o + u; 1 , if 0 <SHARES; 1 < 1 ' ' ' '

This table presents the results from estimating the Tobit regressions, whereby equity allocation was regressed on age, gender, member contributions, marital status, risk profile, education, whether a member invests in shares or managed investment funds external to their superannuation account, net wealth (excluding superannuation), and a set of year dummies. The following are coefficient estimates, not marginal effects.

De,eendent Variable: Eguity Allocation (SHARES) Estimate

Variable HESTA STA GESB UniSu,eer Constant 54.50655 47.43495 57.95451 51.74440

(4.521426) (6.661787) (5.180169) (4.953385) Age Categories:

36-42 1.860445 -0.371872 0.050850 -2.895242 (3.664709) (3.897533) (1.750327) (3.273255)

43-48 0.443235 2.992684 -2.599022 -7.117905** (3.335332) (4.403460) (1.773382) (3.275731)

49-54 0.026307 -3.142878 -1.690093 -10.14016*** (3.238227) (5.556036) (2.225338) (3.491886)

55+ -4.446415 -2.336623 -6.099996* -6.226883 (3.666758) (4.992463) (3.446762) (3.801800)

Member Characteristics:

Male -0.592003 -0.673406 3.618318*** 4.780589** (2.194862) (2.978350) (1.334755) (2.168621)

Er cants 1.96E-05 -0.000241 * -0.000542 -3.50E-05 (0.000108) (0.000140) (0.000370) (8.72E-05)

Married/Defacto 2.555992 3.931112 -1.888969 2.703624 (2.277771) (4.745221) (1.677344) (2.754752)

High Risk 2.710976 19.94000*** 2.509112** 9.998554*** (2.173330) (3.060298) (1.260120) (2.327598)

Cert/Trade/Dip 5.259207* 0.908023 2.912247 2.719121 (2.800515) (3.563060) (2.011205) (4.258665)

Bachelor Deg 5.774478* 3.811540 0.827926 4.553709 (3.092143) (4.041567) (1.832049) (3.848205)

Post Grad Deg 8.224229** 1.806737 -0.721052 2.536051 (3.350874) (4.655946) (2.218134) (3.620095)

Shares/MIF 2.483863 3.863539 2.136913* 0.385847 (1.964550) (2.580811) (1.284442) (2.212474)

Avg Wealth 2.037760 -2.031410 2.972276* -0.950896 (2.466793) (3.243280) (1.542904) (2.818605)

High Wealth 2.757656 2.954130 4.727437** 1.653663 (2.633597) (3.605282) ~2.2062421 (3.313820)

Calendar Years: 1997 -0.769267

(7.277825) 1998 -6.512154* -9.625099 24.79187***

(3.467844) (8.955236) (5.808822) 1999 -3.474323 -6.898588 23.38157***

(4.892123) (6.473921) (3.516936) 2000 5.942643* 0.666110 13.63047***

(3.509581) (5.820776) (4.171905) 2001 5.205543 1.192199 6.855095 23.76737***

(3.870962) (6.309185) (4.221869) (3.562738)

56

2003 -15.00762*** -9.304140 2.205452 7.000829** (3.738709) (6.711229) (4.901957) (3.103769)

2004 -0.449095 6.392597 1.406957 18.19898*** (3.2755332 (5.737482) (4.817612) (3.459001)

Observations 523 133 429 554

Pseudo R 2 0.137085 0.377726 0.102955 0.199131

Log Likelihood -2340.919 -546.6047 -1704.106 -2530.974 Note: robust standard errors are in parentheses. Estimated using the quasi-maximum likelihood (Huber/White) method.

***, **, * denotes significance at 1%, 5%, 10% confidence level, respectively.

57

Once again, in isolation the age-equity ownership/allocation patterns exhibited

by the regression results can be somewhat misleading. To consider the magnitudes of

the age effects requires calculating the marginal effects. Tables 5.9 and 5.10 report the

estimated 2004 marginal effects for the age variables.

Table 5.9 Estimated Probit Age Marginal Effects (2004) This table presents the estimated 2004 age marginal effects from the Probit regressions, whereby equity participation was regressed on age, gender, member contributions, marital status, risk profile, education, whether a member invests in shares or managed investment funds external to their superannuation account, net wealth (excluding superannuation), and a set ofyear dummies.

36-42 43-48 49-54 55+

Table 5.10

HESTA STA -0.007730 0.005600 0.006698 -0.022975

0.000004 -0.000264 -0.000369 -0.020163

Estimated Tobit Age Marginal Effects (2004)

GESB UniSuper -0.004326 -0.001705 -0.043329 -0.028818

This table presents the estimated 2004 age marginal effects from the Tobit regressions, whereby equity allocation was regressed on age, gender, member contributions, marital status, risk profile, education, whether a member invests in shares or managed investment funds external to their superannuation account, net wealth (excluding superannuation), and a set of year dummies.

36-42 43-48 49-54 55+

HESTA STA 0.001686 0.000000 0.000506 0.000000 0.000032 0.000000 -0.010854 0.000000

GESB 0.000000 -0.000003 -0.000001 -0.000027

UniSuper -0.000831 -0.004000 -0.009035 -0.003044

In relation to the Probit and Tobit marginal effects, similar patterns exhibited by

the corresponding age coefficients prevail for the funds. It is important to note though,

that the marginal effects show that the magnitudes of the effect of the age variables on

the likelihood of equity participation and equity allocation are much smaller than the

reported coefficients.

There is a great deal of consistency in the Probit and Tobit marginal effects

across all years, with no significant time effects present. Across all age groups and

funds, a majority of the marginal effects are less than two percent with all of the STA

and GESB Tobit marginal effects being equal to zero.

58

One salient feature of the above tables is that all marginal effects are

significantly smaller than the corresponding marginal effects from the restricted

regressions. This is especially evident from the Tobit marginal effects which take on

either zero or trivial values when additional member characteristics are taken into

consideration. Clearly, inclusion of the additional variables to the regression

specifications results in the age factor becoming less significant in determining: (1) the

likelihood of equity participation in superannuation accounts, and (2) the amount of

equity held in superannuation accounts. This is also supported by the fact that across all

funds, there is a large increase in the pseudo-R 2 values for the Probit and Tobit

regressions that include the additional member characteristics. For example, the model

including the additional variables for HESTA shows a pseudo-R 2 of 15.88 percent

compared to 4. 73 percent for the restricted model.

The preliminary and regression results discussed above provide mixed evidence

as to the relationship between age and equity participation/equity allocation. Most

importantly though, the regression results and estimated marginal effects generally

suggest that age plays a significant role in determining (1) the probability of a member

holding equity in their superannuation account, and (2) the amount of equity held in the

superannuation account. One of the key patterns exhibited by the results in this study, a

hump shape, agree with those of a number of previous studies including Schooley &

Worden (1999), Yoo (1994), Poterba & Samwick (1997), Agnew, Balduzzi, & Sunden

(2003), and Ameriks & Zeldes (2004). However, similar to the finding of Bodie &

Crane (1997), some of the results suggest that there is a negative linear relationship

between age and equity participation/equity allocation.

Some of the relationships between equity participation/equity allocation and the

member characteristics are generally consistent with previous findings, whilst others do

not and are unclear. For example, the estimated coefficients for the regressions suggest

that the following groups of individuals are more likely to hold equity, and conditional

on equity participation, own more equity: (1) males, (2) individuals with higher

contribution amounts, and (3) individuals who are married or in a de-facto relationship.

Ameriks & Zeldes (2004) report similar findings, noting that male 401(k) participants

invest 19.3 percent more in equities than their female counterparts and married

participants invest 14.4 percent more in equities than their single counterparts.

59

In the two regresswns including the HIGH_RISK variable, the estimated

coefficients suggest that individuals who are willing to take substantial risk are more

likely to hold shares (and hold more shares, conditional on equity participation) than

those individuals who are only prepared to assume low to average financial risks. This

is consistent with the general risk-return theory that investors are willing to bear higher

risk for a higher potential return.

The unclear results relate to the estimated coefficients on the education and

wealth variables. First, the regression results produce mixed results for the education

variables across the four funds. For STA, members who have completed a higher level

of education are less likely to hold shares in their superannuation account than members

limited to some level of secondary education. For GESB, members who have completed

a certificate, trade or diploma level of education or a post graduate degree, are less

likely to hold shares than the base group, while members with a bachelor degree are

more likely to hold shares than the base group. The results from the HESTA fund

reflects more closely the relationship that would generally be expected between

education level and equity participation, where individuals with a higher level of

education are more likely to hold shares in their superannuation account.

The estimated coefficients on the wealth variables propose that individuals who

have an average or high level of net wealth are less likely to hold shares than individuals

with lower net wealth. One would expect the opposite relationship to prevail, as

members who are better-off financially will have a higher level of awareness of and

play a more active role in their financial affairs.

The findings ofthis thesis have several key implications. First, the results of the

preliminary and regression analyses suggest two key patterns with respect to age and

equity participation/allocation, hump-shaped and linear negative. Results also suggest

that while age is an important factor in determining equity participation and equity

allocation, when additional factors are taken into account, the importance of the age

variable diminishes significantly. Thus, there is a need for superannuation funds to

consider the effect of member characteristics in addition to the age variable in the

design and structure of funds.

Also, it is important for members to understand the consequences of an under­

allocation to growth assets in their superannuation accounts. A risk of an under­

allocation to equity during the life-cycle is underperformance of the retirement

60

investment and ultimately, an insufficient superannuation balance upon retirement.

Equally important is the need to understand the consequences of an over-allocation to

growth assets. Holding a large proportion of growth assets in the period leading up to

retirement exposes retirement savings to short-term market fluctuations, potentially

resulting in losses and a lower retirement benefit. The key goal at this stage of the life­

cycle is capital preservation.

Third, fund trustees need to approach the issue of offering life-cycle products as

a superannuation default option with care. They must assess whether or not life-cycle

funds should be implemented in the best interests of members.

61

6. CONCLUSION

There is much debate surrounding the age-phasing strategy, whereby investors

reduce their exposure to growth assets in favour of "safe" assets as they become older.

On one hand, financial practitioners frequently advise their clients to shift their

investments away from stocks (towards bonds or other fixed interest securities) as they

age. Alternatively, investment theory has argued that individuals should invest a

constant proportion of their portfolios in equities regardless of age. The existing

theoretical and empirical literature provides conflicting views regarding age-phasing.

Life-cycle funds (also known as target-date funds) have been created in response

to the perceived desire of investors to reduce their equity exposure within their

investment portfolio as they age, and to address psychological biases such as inertia in

retirement asset allocation portfolio decisions (Poterba, Rauh, Venti, & Wise, 2005).

Such funds were first introduced to the U.S. investment environment during the mid­

nineties and have experienced a significant amount of growth in recent years. At the end

of 2005, the U.S. target-date fund population had a total $70 billion in assets under

management. Life-cycle funds are now beginning to emerge in Australia as a

superannuation investment strategy.

This thesis examined several questions. First, it provided an overview of the life­

cycle fund markets in the United States and Australia. Second, it summarised the

theoretical and empirical literature relating to life-cycle asset allocation. Third, using a

sample of member choices from four large Australian superannuation funds, this thesis

investigated the difference between actual asset allocations and the asset allocations

suggested by life-cycle products. This thesis employed a regression framework to

examine the effect of age on (1) the likelihood of stock ownership and (2) the allocation

of stock within superannuation accounts (conditional on stock ownership). Further,

using additional demographic data, this thesis investigated the relative importance of

age in explaining equity ownership and equity allocation in member investment choices.

Four key findings emerged from this thesis. Existing life-cycle funds being

offered in the United States and Australia vary considerably in terms of asset allocation

structure, expense structure, and investment strategy. A comparison of the asset

allocations between the Australian life-cycle products and the Australian

62

superannuation fund default investment options revealed that life-cycle products

decrease individuals' exposure to growth assets very early in the life-cycle. Third, the

age variable has a significant relationship with equity participation and equity

allocation. The two key trends exhibited are hump-shaped and negative linear age­

ownership and age-allocation profiles. Finally, when additional member demographic

variables like gender, marital status, risk profile, education, and wealth are taken into

account there is a decline in the relative importance of age in explaining equity

ownership and equity allocation

A number of implications result from this thesis. First, the preliminary and

regression results suggest two key patterns with respect to age and equity

participation/allocation - hump-shaped and decreasing. The importance of the age

variable diminishes significantly when additional factors are taken into account. Thus,

superannuation funds need to consider the effect of member characteristics in addition

to age in the design and structure of funds. Second, members need to be aware of the

consequences of an under-allocation and over-allocation to growth assets in their

superannuation accounts. Third, in choosing an appropriate asset allocation,

superannuation members need to take all individual-specific characteristics into account

when selecting a retirement investment strategy. Finally, fund trustees should approach

the issue of offering life-cycle products as a superannuation default option with care, as

they are required to exercise their responsibilities in the best interests of members.

A potential limitation of this thesis is the nature of the sample studied. The

sample includes four of Australia's larger industry superannuation funds (HESTA, STA,

GESB & UniSuper). There are a total of 92 industry superannuation funds in Australia

(Australian Prudential Regulation Authority, 2006), so the sample is not completely

representative of the industry superannuation funds sector and superannuation funds in

general. Thus, observations and results are not necessarily reflective of all

superannuation members. Also, as this thesis is concerned with the portfolio decisions

(equity participation and equity allocation) of superannuation members in an Australian

context only, then the findings may not be applicable to retirement participants in other

countries.

In terms of suggestions for further research in the area of age-based asset

allocation, there are a number of potential areas to be explored. It would be interesting

to extend this thesis to consider the expenses associated with life-cycle funds and of the

63

services provided by these funds. Another possible study concerning life-cycle funds

would entail a comparison of life-cycle funds and other investment vehicles in terms of

the differential fees across these investment options.

Applying a similar regression analysis to the member asset allocation data for a

larger number of Australian superannuation funds would provide findings that more

accurately reflect the retirement savings behaviour of Australian superannuation fund

members in general.

64

7. REFERENCES

Agnew, J., Balduzzi, P., & Sunden, A. (2003). Portfolio Choice and Trading in a Large 401(k) Plan. The American Economic Review, 93(1), 193-215.

Ameriks, J., & Zeldes, S. P. (2004). How Do Household Portfolio Shares Vary With Age: National Bureau of Economic Research.

Australian Prudential Regulation Authority. (2006). Annual Superannuation Bulletin. Sydney: Australian Prudential Regulation Authority.

Australian Stock Exchange. (2004). Australian Share Ownership Study. Sydney, New South Wales: Australian Stock Exchange Ltd.

AXISS Australia. (2004). Managed Funds in Australia. Sydney: AXISS Australia.

Barber, B. M., & Odean, T. (2001). Boys will be Boys: Gender, Overconfidence, and Common Stock Investment. Quarterly Journal of Economics, 116(1 ), 261-292.

Benitez-Silva, H. (2002). Labor Supply Flexibility and Portfolio Selection: An Empirical Analysis: University of Michigan, Michigan Retirement Research Centre.

Bodie, Z. (2002). Life-Cycle Finance in Theory and in Practice. Boston: Boston University School ofManagement.

Bodie, Z., & Crane, D. (1997). Personal Investing: Advice, Theory, and Evidence. Financial Analysts Journal, 53(6), 13-23.

Bodie, Z., Merton, R. C., & Samuelson, W. F. (1992). Labour Supply Flexibility and Portfolio Choice in a Life Cycle Model. Journal of Economic Dynamics and Control, 16,427-449.

Brown, P., da Silva Rosa, R., & McNaughton, T. (2006). A portrait of managed fund investors.

Campbell, J. Y., Cocco, J. F., Gomes, F. J., & Maenhout, P. J. (1999). Investing Retirement Wealth: A Life-Cycle Model. Cambridge: National Bureau of Economic Research.

Carlson, G. (2006). Taking Aim at Target-Date Funds. Fund Spy Retrieved April 7, 2006, from http://news.momingstar.com/article/printArticle.asp?id=158394

65

Clare, R. (2005). Implications of choice of superannuation fund legislation for members, employers and funds. Sydney: Association of Superannuation Funds of Australia.

Cocco, J. F., Gomes, F. J., & Maenhout, P. J. (2005). Consumption and Portfolio Choice over the Life-Cycle. Review of Financial Studies, 18(2), 491-533.

Dougherty, C. (2002). Introduction to Econometrics (2nd ed.). Oxford: Oxford University Press.

Drew, M. E., & Stanford, J. D. (2003). A Review of Australia's Compulsory Superannuation Scheme After a Decade. Brisbane: University of Queensland.

Dunstan, B. (2006, April 8). Simplify life: set your sights on a target date. Australian Financial Review, p. 38.

Egan, L. (2006, March 8). Fifth of Aussies fail on retirement savmg. Money Management.

Financial Research Corporation. (2006). Lifecycle Fund Economics: Evaluating Next Generation Competitive Dynamics. Boston, MA.

Gallery, G., Gallery, N., & Brown, K. (2004). Superannuation Choice: The Pivotal Role of the Default Option. Journal of Australian Political Economy, 53, 44-66.

GESB. (2006). GESB Annual Report 2005/2006: Government Employees Superannuation Board.

HESTA. (2006). HESTA Super Fund 2005/2006 Annual Report: Health Employees Superannuation Trust

Huberman, G., & Jiang, W. (2006). Offering vs. Choice by 401(k) Plan Participants: Equity Exposure and Number ofFunds. The Journal of Finance, 61(2), 763-801.

Investment Company Institute. (2006). Research Fundamentals (Vol. 15). Washington, DC.

Iyengar, S., Jiang, W., & Huberman, G. (2004). How Much Choice is Too Much? Contributions to 401(k) Retirement Plans. In 0. Mitchell & S. Utkus (Eds.), Pension Design and Structure: New Lessons from Behavioral Finance (pp. 83-95). Oxford: United Kingdom: Oxford University Press.

Koijen, S., Nijman, T., & Werker, B. (2006). Dynamic Asset Allocation with Annuity Risk [Electronic Version] from http://ssm.com/abstract=897679

Kritzman, M. (1994). What Practitioners Need to Know ... About Time Diversification. Financial Analysts Journal, 50(1), 14-18.

66

Madrian, B., & Shea, D. (2001). The Power of Suggestion: Inertia in 401(k) Participation and Savings Behavior. Quarterly Journal of Economics, 116, 1149-1525.

Malkiel, B. G. (1996). A Random Walk Down Wall Street. New York: W.W. Norton & Company Inc.

Markowitz, H. (1952). Portfolio Selection. The Journal of Finance, 7(1), 77-91.

Markowitz, H. (1991). Foundations of Portfolio Theory. The Journal of Finance, 46(2), 469-477.

Marquez, J. (2005). Lifecycle Funds Can Help Companies Mitigate risk, Boost Employee Savings. Workforce Management, 84(4), 65-67.

McNaughton, T., Piggot, J., & Purcal, S. (1999). Growing Old Gracefully: Age-Phasing, Targets and Saving Rules. Retrieved 11 March, 2006, from actuary.web.unsw.edu.au/files/Growoldjppm.pdf

Merton, R. C. (1969). Lifetime Portfolio Selection Under Uncertainty: The Continuous­Time Case. Review of Economics and Statistics, 51(3), 247-257.

Mitchell, 0., Mottola, G., Utkus, S., & Yamaguchi, T. (2006). The Inattentive Participant: Portfolio Trading Behavior in 401 (k) Plans. Paper presented at the 14th Australian Colloquium of Superannuation Researchers, Sydney.

Morentsov, D. (2006). Asset Allocation Issues in Lifecycle Fund Selection and Design. Chicago: Mercer Investment Consulting, Inc.

Mossin, J. (1968). Optimal Multiperiod Portfolio Policies. The Journal of Business, 41(2), 215-229.

Poterba, J. M., Rauh, J., Venti, S., & Wise, D. (2005). Lifecycle Asset Allocation Strategies and the Distribution of 401 (k) Retirement Wealth: National Bureau of Economic Research, Inc.

Poterba, J. M., & Samwick, A. A. (1997). Household Portfolio Allocation Over the Life Cycle: National Bureau of Economic Research, Inc.

Reserve Bank of Australia. (2004). The Composition and Distribution of Household Assets and Liabilities: Evidence from the 2002 HILDA Survey: Reserve Bank of Australia.

Rice Walker Actuaries. (2006). Investment strategies in Australian Superannuation funds. Sydney.

Rubinstein, M. (2002). Markowitz's "Portfolio Selection": A Fifty-Year Retrospective. The Journal of Finance, 57(3), 1041-1045.

67

Russell Investment Group. (2005). Russell LifePoints Product Disclosure Statement. Sydney, NSW: Russell Investment Group.

Samuelson, P. A. (1969). Lifetime Portfolio Selection By Dynamic Stochastic Programming. The Review of Economics and Statistics, 51(3), 239-246.

Samuelson, P. A. (1989). A Case at Last for Age-Phased Reduction in Equity. The Journal of Economic Sciences, 86, 9048-9051.

Schooley, D. K., & Worden, D. D. (1999). Investors' Asset Allocations versus Life­Cycle Funds Financial Analysts Journal, 55(5), 37-43.

TIAA-CREF. (2005). TIAA-CREF Retirement Strategies: Teachers Insurance and Annuity Association-College Retirement Equities Fund.

UniSuper. (2006). UniSuper 2006 Report to Members: UniSuper.

Utkus, S. (2004). Selecting a Default Fund for a Defined Contribution Plan: The Vanguard Center for Retirement Research.

Wooldridge, J. (2006). Introductmy Econometrics: A Modern Approach (3rd ed.). Ohio: Thomson South-Western.

Yoo, P. S. (1994). Age Dependent Portfolio Selection: Federal Reserve Bank of St.Louis.

68

8.

AP

PE

ND

ICE

S

AP

PE

ND

IX A

Tab

le 1

U

S. L

ife-

cycl

e F

unds

Thi

s ta

ble

pres

ents

sum

mar

y in

form

atio

n fo

r th

e U

.S.

life

-cyc

le f

unds

as

cate

gori

sed

by M

orni

ngst

ar (

U.S

.) i

nclu

ding

, T

otal

Ass

ets

(tot

al a

sset

s un

der

man

agem

ent)

, N

et A

sset

Val

ue (

the

fund

's p

er-s

hare

pri

ce c

alcu

late

d as

tot

al a

sset

s o

f the

fu

nd

les

s fe

es a

nd e

xpen

ses

divi

ded

by n

umbe

r o

f sha

res

outs

tand

ing)

, Exp

ense

s (e

xpen

se r

atio

, fr

ont

load

fee,

and

def

erre

d lo

ad f

ee),

and

Ass

et A

lloc

atio

n (c

ash,

sto

cks,

bon

ds,

and

"oth

er")

.

To

tal

NA

V

Ex

pen

se

Fro

nt

Def

erre

d

Mo

rnin

gst

ar

Ass

ets

($)

Rat

io

Lo

ad

Lo

ad

Ass

et A

lloc

atio

n (a

s at

F

un

d F

amil

y

Fu

nd

Nam

e C

ateg

ory

{$

mil

lion

s}

23/0

8/06

} {%

2 {%

} {%

} {a

s at

24/

08/0

6}

Cas

h

Sto

cks

Bon

ds

%)

(%)

All

ianc

eBem

stei

n 20

00

Tar

get-

Dat

e A

LL

IAN

CE

BE

RN

ST

EIN

R

etir

emen

t S

tr A

~LTAAX)

2000

-201

4 0.

71

10.8

1.

10**

4.

25

-9.

5 48

.4

40.1

A

llia

nceB

emst

ein

2005

T

arge

t-D

ate

Ret

irem

ent

Str

A ~LTBAX2

2000

-201

4 0.

71

10.8

1.

10**

4.

25

-6.

7 57

.9

33.2

A

llia

nceB

emst

ein

2010

T

arge

t-D

ate

Ret

irem

ent S

tr A

~LTDAX2

2000

-201

4 12

10

.91

1.20

**

4.25

-

5.6

65.3

26

.7

All

ianc

eBem

stei

n 20

15

Tar

get-

Dat

e R

etir

emen

t Str

A ~LTEAX)

2015

-202

9 12

11

1.

20**

4.

25

-5.

1 71

.5

20.7

A

llia

nceB

emst

ein

2020

T

arge

t-D

ate

Ret

irem

ent S

tr K

~L T

AA

X2

20

15-2

029

15

11.0

8 1.

25**

4.

25

-4.

4 78

.3

14.4

A

llia

nceB

emst

ein

2025

T

arge

t-D

ate

Ret

irem

ent S

tr K

(L

TA

AX

2 20

15-2

029

14

11.3

3 1.

20**

-

-3.

2 85

.8

7.8

All

ianc

eBem

stei

n 20

30

Tar

get-

Dat

e 8

11.1

3 1.

25**

4.

25

-2.

3 92

.7

1.6

69

Oth

er

(%)

1.9

2.2

2.5

2.8

3.0

3.3

3.5

To

tal

NA

V

Ex

pen

se

Fro

nt

Def

erre

d

Mo

rnin

gst

ar

Ass

ets

($)

Rat

io

Lo

ad

Lo

ad

Ass

et A

lloc

atio

n (a

s at

F

un

d F

amil

y

Fu

nd

Nam

e C

ateg

ory

{$

mil

lion

s}

23/0

8/06

} {%

} {%

} {%

} {a

s at

24/

08/0

6}

Cas

h

Sto

cks

Bon

ds

Oth

er

Ret

irem

ent S

tr A

(L

TJA

X)

2030

+

All

ianc

eBem

stei

n 20

35

Tar

get-

Dat

e R

etir

emen

t S

tr A

~L

TK

AX

) 20

30+

6

11.1

8 1.

25**

4.

25

-2.

0 94

.4

0.0

3.5

All

ianc

eBem

stei

n 20

40

Tar

get-

Dat

e R

etir

emen

t S

tr A

~L TLAX~

2030

+

3 11

.27

1.25

**

4.25

-

2.0

94.4

0.

0 3.

5 A

llia

nceB

emst

ein

2045

T

arge

t-D

ate

Ret

irem

ent S

tr A

~LTAAX~

2030

+

2 11

.31

1.25

**

4.25

-

2.0

94.4

0.

0 3.

5

Am

eric

an C

entu

ry

LIV

ES

TR

ON

G 2

015

Inv

Tar

get-

Dat

e A

ME

RIC

AN

CE

NT

UR

Y

~ARF

IX~

2015

-202

9 10

9 11

.18

0.82

**

--

9.0

53.5

33

.9

3.5

Am

eric

an C

entu

ry

LIV

ES

TR

ON

G 2

025

Inv

Tar

get-

Dat

e ~ARWIX~

2015

-202

9 15

1 11

.59

0.87

**

--

7.7

65.0

24

.3

2.9

Am

eric

an C

entu

ry

LIV

ES

TR

ON

G 2

035

Inv

Tar

get-

Dat

e ~ARYIX~

2030

+

69

11.9

5 0.

92**

-

-3.

3 77

.0

17.5

2.

1 A

mer

ican

Cen

tury

L

IVE

ST

RO

NG

204

5 In

v T

arge

t-D

ate

~AROIX~

2030

+

40

12.0

6 0.

95**

-

-2.

8 82

.9

12.7

1.

6 A

mer

ican

Cen

tury

L

IVE

ST

RO

NG

Inc

Inv

T

arge

t-D

ate

~ARTOXL

2000

-201

4 34

10

.53

0.77

**

--

13.7

44

.4

38.1

3.

8

Bar

clay

s G

loba

l Inv

esto

rs L

P

Tar

get-

Dat

e B

AR

CL

AY

S

Ret

ire

I ~STLAX2

2000

-201

4 11

5 11

.46

0.85

**

--

5.2

38.0

56

.8

0.0

Bar

clay

s G

loba

l In

vest

ors

LP

T

arge

t-D

ate

(LP

-L

ifeP

ath)

20

10 I

(S

TL

BX

) 20

00-2

014

414

13.2

9 0.

85**

-

-4.

5 47

.3

48.2

0.

0

70

To

tal

NA

V

Ex

pen

se

Fro

nt

Def

erre

d

Mo

rnin

gst

ar

Ass

ets

($)

Rat

io

Lo

ad

Lo

ad

Ass

et A

lloc

atio

n (a

s at

F

und

Fam

ily

F

un

d N

ame

Cat

ego

ry

{$m

illi

ons}

23

/08/

06}

{%}

{%}

{%}

{as

at 2

4/08

/06}

C

ash

S

tock

s B

on

ds

Oth

er

%)

Bar

clay

s G

loba

l Inv

esto

rs L

P

Tar

get-

Dat

e 20

20 I

~STLCX2

2015

-202

9 73

2 16

.52

0.85

**

--

3.2

66.1

30

.7

0.0

Bar

clay

s G

loba

l Inv

esto

rs L

P

Tar

get-

Dat

e 20

30 I

~STLDX2

2030

+

470

16.1

8 0.

85**

-

-2.

2 80

.3

17.4

0.

0 B

arcl

ays

Glo

bal I

nves

tors

LP

T

arge

t-D

ate

2040

I (

ST

LE

X2

2030

+

303

19.2

7 0.

85**

-

-1.

5 92

.1

6.4

0.1

Ban

e o

f Am

eric

a R

etir

emen

t T

arge

t-D

ate

CO

LU

MB

IA

2005

A ~BAAAX2

2000

-201

4 0.

04

10.2

1.

02**

-

-N

/A

N/A

N

/A

NIA

B

ane

of A

mer

ica

Ret

irem

ent

Tar

get-

Dat

e B

ane

of A

m R

etir

emen

t 20

10 A

~BFBAX2

2000

-201

4 0.

04

10.2

3 1.

03**

-

-N

/A

N/A

N

/A

N/A

B

ane

of A

mer

ica

Ret

irem

ent

Tar

get-

Dat

e 20

15 A

~BFCAX2

2015

-202

9 0.

04

10.2

1 1.

04**

-

-N

/A

N/A

N

/A

N/A

B

ane

of A

mer

ica

Ret

irem

ent

Tar

get-

Dat

e 20

20 A

~BFDAX2

2015

-202

9 0.

04

10.1

8 1.

04**

-

-N

/A

N/A

N

/A

N/A

B

ane

of A

mer

ica

Ret

irem

ent

Tar

get-

Dat

e 20

25 Z

~BFEZX2

2015

-202

9 0.

06

10.1

5 0.

83**

-

-N

/A

N/A

N

/A

N/A

B

ane

of A

mer

ica

Ret

irem

ent

Tar

get-

Dat

e 20

30 Z

~BFFZX)

2030

+

0.06

10

.14

0.84

**

--

N/A

N

/A

N/A

N

/A

Ban

e o

f Am

eric

a R

etir

emen

t T

arge

t-D

ate

2035

Z ~BAGZX2

2030

+

0.06

10

.1

0.87

**

--

N/A

N

/A

N/A

N

/A

Ban

e o

f Am

eric

a R

etir

emen

t T

arge

t-D

ate

2040

Z ~BFHZX2

2030

+

0.05

10

.07

0.89

**

--

N/A

N

IA

N/A

N

/A

Tar

get-

Dat

e D

WS

-SC

UD

DE

R

DW

S T

ar!li

et 2

008 ~KRFGX2

2000

-201

4 21

10

.63

1.31

5.

00

-0.

3 20

.5

79.3

0.

0 T

arge

t-D

ate

DW

S T

arge

t 201

0 (K

RF

AX

) 20

00-2

014

50

9.16

1.

03

5.00

-

0.5

24.1

75

.1

0.3

71

To

tal

NA

V

Exp

ense

F

ron

t D

efer

red

Mo

rnin

gst

ar

Ass

ets

($)

Rat

io

Lo

ad

Lo

ad

Ass

et A

lloc

atio

n (a

s at

F

un

d F

amil

y F

un

d N

ame

Cat

egor

y {$

mil

lion

s}

23/0

8/06

} {%

} {%

} {%

} {a

s at

24/

08/0

6}

Cas

h S

tock

s B

onds

O

ther

t%2

~%)

(%2

(%)

Tar

get-

Dat

e D

WS

Tar

get2

011

(KR

FB

X)

2000

-201

4 79

10

.63

1.05

5.

00

-0.

4 26

.0

73.2

0.

4 T

arge

t-D

ate

DW

S Ta

r~et

201

2 ~KRFCX2

2000

-201

4 62

9.

06

1.10

5.

00

-0.

1 31

.7

67.7

0.

4 T

arge

t-D

ate

DW

S Ta

r~et

201

3 ~KRFDX2

2000

-201

4 43

9.

22

1.18

5.

00

-0.

4 30

.1

69.0

0.

4 T

arge

t-D

ate

DW

S Ta

r~et

201

4 ~KRFEX2

2000

-201

4 44

7.

62

1.18

5.

00

-0.

3 26

.4

73.0

0.

4

FID

EL

ITY

F

idel

ity

Adv

isor

Fre

edom

T

arge

t-D

ate

INV

ES

TM

EN

TS

20

05 A

~FFAVX2

2000

-201

4 49

11

.14

0.92

**

5.75

-

9.6

48.8

40

.4

1.4

Fid

elit

y A

dvis

or F

reed

om

Tar

get-

Dat

e A

dvis

or

2010

T ~FCFTX2

2000

-201

4 39

2 11

.53

1.18

**

3.50

-

8.5

49.0

41

.3

1.2

Fid

elit

y A

dvis

or F

reed

om

Tar

get-

Dat

e F

reed

om

2015

A ~FFVAX2

2015

-202

9 40

2 11

.58

0.95

**

5.75

-

5.4

59.0

34

.7

0.9

Fid

elit

y A

dvis

or F

reed

om

Tar

get-

Dat

e 20

20 T

~FDTFX2

2015

-202

9 81

4 12

.44

1.23

**

3.50

-

3.0

69.1

27

.3

0.7

Fid

elit

y A

dvis

or F

reed

om

Tar

get-

Dat

e 20

25 A

~FATWX2

2015

-202

9 31

2 11

.98

0.98

**

5.75

-

2.7

73.3

23

.4

0.6

Fid

elit

y A

dvis

or F

reed

om

Tar

get-

Dat

e 20

30 T

~FTFEX)

2030

+

489

12.9

6 1.

25**

3.

50

-2.

3 82

.1

14.0

0.

6 F

idel

ity

Adv

isor

Fre

edom

T

arge

t-D

ate

2035

A ~FATHX2

2030

+

154

12.2

5 1.

00**

5.

75

-2.

3 82

.7

14.3

0.

7 F

idel

ity

Adv

isor

Fre

edom

T

arge

t-D

ate

2040

T ~FTFFX2

2030

+

399

13.2

9 1.

76**

3.

50

-2.

2 85

.1

12.0

0.

7 F

idel

ity

Adv

isor

Fre

edom

T

arge

t-D

ate

Inc

T (

FT A

FX

) 20

00-2

014

79

10.4

1.

09**

3.

50

-24

.7

20.5

51

.6

3.2

72

To

tal

NA

V

Exp

ense

F

ron

t D

efer

red

M

orn

ing

star

A

sset

s ($

) R

atio

L

oad

L

oad

A

sset

All

ocat

ion

(as

at

Fu

nd

Fam

ily

Fu

nd

Nam

e C

ateg

ory

{$

mil

lion

s}

23/0

8/06

} {%

} {%

} {%

} {a

s at

24/

08/0

6}

Cas

h

Sto

cks

Bon

ds

Oth

er

(%

(%)

Fid

elit

y F

reed

om 2

000

Tar

get-

Dat

e ~FFFBX2

2000

-201

4 15

64

12.3

5 0.

52**

-

-30

.9

25.7

41

.7

1.7

Fid

elit

y F

reed

om 2

005

Tar

get-

Dat

e (F

FFV

X2

2000

-201

4 60

7 11

.27

0.62

**

--

13:5

46

.7

38.6

1.

2 F

idel

ity

Fre

edom

201

0 T

arge

t-D

ate

~FFFCX2

2000

-201

4 10

892

14.2

3 0.

62**

-

-11

.1

49.0

38

.8

1.1

Fid

elit

y F

reed

om 2

015

Tar

get-

Dat

e ~FFVFX2

2015

-202

9 31

91

11.7

3 0.

67**

-

-7.

6 58

.4

33.0

0.

9 F

idel

ity

Fre

edom

202

0 T

arge

t-D

ate

~FFFDX2

2015

-202

9 14

414

14.8

9 0.

70**

-

-4.

0 69

.8

25.4

0.

7 F

idel

ity

Fre

edom

202

5 T

arge

t-D

ate

~FFTWX2

2015

-202

9 25

60

12.1

5 0.

72**

-

-3.

8 72

.6

22.8

0.

8 F

idel

ity

Fre

edom

203

0 T

arge

t-D

ate

~FFFEX)

2030

+

8868

15

.23

0.74

**

--

2.8

82.1

14

.3

0.8

Fid

elit

y F

reed

om 2

035

Tar

get-

Dat

e ~FFTHX2

2030

+

1422

12

.43

0.75

**

--

2.8

82.4

14

.0

0.8

Fid

elit

y F

reed

om 2

040

Tar

get-

Dat

e ~FFFFX2

2030

+

4174

8.

96

0.76

**

--

2.6

84.2

12

.3

0.9

Fid

elit

y F

reed

om 2

045

Tar

get-

Dat

e ~FFFGX2

2030

+

11

9.91

0.

79**

-

-N

/A

N/A

N

/A

N/A

F

idel

ity

Fre

edom

205

0 T

arge

t-D

ate

iFF

FH

X)

2030

+

10

9.9

0.79

**

--

N/A

N

/A

N/A

N

/A

Fid

elit

y F

reed

om I

ncom

e T

arge

t-D

ate

(FF

AX

) 20

00-2

014

2163

11

.38

0.51

**

--

31.9

21

.0

45.0

2.

1

HA

RT

FO

RD

MU

TU

AL

H

artf

ord

Ret

irem

ent I

ncom

e T

arge

t-D

ate

FU

ND

S

A~HTRAX)

2000

-201

4 0.

81

9.42

1.

25**

5.

50

-12

.5

28.8

57

.8

1.0

Har

tfor

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 2

9.38

1.

30**

5.

50

-8.

4 48

.0

42.8

0.

8

73

To

tal

NA

V

Exp

ense

F

ron

t D

efer

red

M

orn

ing

star

A

sset

s ($

) R

atio

L

oad

L

oad

A

sset

All

ocat

ion

(as

at

Fu

nd

Fam

ily

F

un

d N

ame

Cat

ego

ry

{$m

illi

ons}

23

/08/

06}

{%}

{%}

{%}

{as

at 2

4/08

/06}

C

ash

S

tock

s B

onds

O

ther

(%

) 20

10 A

(H

TT

AX

) 20

00-2

014

Har

tfor

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

20 A

~HTW A

XL

20

15-2

029

1 10

.02

1.35

**

5.50

-

7.5

61.7

29

.9

0.9

Har

tfor

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

30 A

(H

TH

AX

L

2030

+

0.73

8.

92

1.40

**

5.50

-

6.1

76.3

17

.0

0.7

JPM

orga

n S

mar

tRet

irem

ent

Tar

get-

Dat

e JP

Mo

rgan

20

10 I

~JSXIXL

2000

-201

4 9

15.2

1 0.

70**

-

-N

/A

N/A

N

/A

N/A

JP

Mor

gan

Sm

artR

etir

emen

t T

arge

t-D

ate

~SmartRetirementl

2015

I ~JSFIXL

2015

-202

9 11

15

.14

0.75

**

--

N/A

N

/A

N/A

N

/A

JPM

orga

n S

mar

tRet

irem

ent

Tar

get-

Dat

e 20

20 I

~JTTIXL

2015

-202

9 35

15

.07

0.80

**

--

N/A

N

/A

N/A

N

/A

JPM

orga

n S

mar

tRet

irem

ent

Tar

get-

Dat

e 20

30 I

psM

IX)

2030

+

40

15.0

5 0.

90**

-

-N

/A

N/A

N

/A

N/A

JP

Mor

gan

Sm

artR

etir

emen

t T

arge

t-D

ate

2040

I ~SMTIXL

2030

+

41

15.0

5 0.

90**

-

-N

/A

N/A

N

/A

N/A

JP

Mor

gan

Sm

artR

etir

emen

t T

arge

t-D

ate

Inco

me

I (J

SIIX

L

2000

-201

4 0.

41

15.2

1 0.

60**

-

-N

/A

N/A

N

/A

N/A

Mas

sMut

ual

Sel

ect D

esti

n T

arge

t-D

ate

MA

SS

MU

TU

AL

R

etir

e 20

10 A

~MRXAXL

2000

-201

4 14

4 10

.75

1.18

**

5.75

-

26.0

35

.3

38.5

0.

2 M

assM

utua

l S

elec

t Des

tin

Tar

get-

Dat

e R

etir

e 20

20 L

(M

RT

LX

L

2015

-202

9 50

5 11

.14

0.99

**

--

17.0

52

.8

30.1

0.

1 M

assM

utua

l S

elec

t Des

tin

Tar

get-

Dat

e R

etir

e 20

30 L

~MYRLX)

2030

+

360

11.7

2 1.

08**

-

-11

.6

74.7

13

.6

0.0

Mas

sMut

ual

Sel

ect D

esti

n T

arge

t-D

ate

Ret

ire

2040

L ~MRFLX)

2030

+

215

11.9

9 1.

13**

-

-3.

6 95

.4

0.9

0.1

Mas

sMut

ual

Sel

ect D

esti

n T

arge

t-D

ate

259

10.4

0.

76**

-

-29

.3

22.6

47

.9

0.2

74

Tot

al

NA

V

Exp

ense

F

ron

t D

efer

red

Mo

rnin

gst

ar

Ass

ets

($)

Rat

io

Lo

ad

Lo

ad

Ass

et A

lloc

atio

n (a

s at

F

un

d F

amil

y F

un

d N

ame

Cat

egor

y {$

mill

ions

} 23

/08/

06}

{%}

{%}

{%}

{as

at 2

4/08

/06}

C

ash

Sto

cks

Bon

ds

Oth

er

(%)

(%

%)

(%

Ret

ire

Inc

Y (

MD

RY

X)

2000

-201

4

MFS

Lif

etim

e 20

10 A

T

arge

t-D

ate

MF

S

~MFSAX)

2000

-201

4 4

10.4

1 1.

09**

5.

75

-16

.1

36.8

47

.0

0.1

MFS

Lif

etim

e 20

20 R

3 T

arge

t-D

ate

~MFLGX)

2015

-202

9 12

10

.45

1.66

**

--

3.1

75.5

21

.0

0.4

MFS

Lif

etim

e 20

30 R

3 T

arge

t-D

ate

~MLTGX2

2030

+

6 10

.38

1.73

**

--

2.0

96.2

1.

4 0.

3 M

FS L

ifet

ime

2040

R3

Tar

get-

Dat

e ~MLFGX2

2030

+

4 10

.36

1.73

**

--

2.0

97.7

0.

0 0.

3

PR

INC

IPA

L

Pri

ncip

al I

nv L

ifeT

ime

2010

T

arge

t-D

ate

INV

ES

TO

RS

In

stl ~PTTIX)

2000

-201

4 92

1 12

.73

0.70

**

--

7.9

39.3

47

.7

5.1

Pri

ncip

al I

nv L

ifeT

ime

2020

T

arge

t-D

ate

Inst

l ~PL W

IX)

2015

-202

9 16

56

13.0

7 0.

70**

-

-6.

4 56

.7

31.3

5.

5 P

rinc

ipal

Inv

Lif

eTim

e 20

30

Tar

get-

Dat

e In

stl ~PMTIX2

2030

+

1367

12

.92

0.70

**

--

5.7

67.1

23

.3

3.8

Pri

ncip

al I

nv L

ifeT

ime

2040

T

arge

t-D

ate

Inst

l ~PTDIX)

2030

+

613

13.0

8 0.

70**

-

-5.

3 74

.3

17.6

2.

7 P

rinc

ipal

Inv

Lif

eTim

e 20

50

Tar

get-

Dat

e In

stl ~PPLIX2

2030

+

283

12.6

4 0.

70**

-

-4.

6 83

.8

9.8

1.8

Pri

ncip

al I

nv L

ifeT

ime

Str

at

Tar

get-

Dat

e In

stl ~PLSIX2

2000

-201

4 36

3 12

.34

0.77

**

--

9.3

23.6

61

.4

5.7

Rus

sell

Lif

ePoi

nts

2010

T

arge

t-D

ate

RU

SS

EL

L

Stra

te~

R3 ~RJLDX2

2000

-201

4 7

10.5

9 1.

35**

-

-11

.7

41.1

44

.5

2.7

Rus

sell

Lif

ePoi

nts

2020

T

arge

t-D

ate

Stra

tes;

:l: R

3 (R

LL

DX

2 20

15-2

029

12

10.7

8 1.

42**

-

-10

.6

50.2

36

.5

2.7

Rus

sell

Lif

ePoi

nts

2030

T

arge

t-D

ate

11

10.9

5 1.

48**

-

-9.

6 58

.6

29.1

2.

6

75

To

tal

NA

V

Exp

ense

F

ron

t D

efer

red

M

orn

ing

star

A

sset

s ($

) R

atio

L

oad

L

oad

A

sset

All

ocat

ion

(as

at

Fun

d F

amil

y F

un

d N

ame

Cat

ego

ry

{$m

illi

ons}

23

/08/

06}

{%}

{%}

{%}

{as

at 2

4/08

/06}

C

ash

S

tock

s B

onds

O

ther

(%

(%

) (%

) S

trat

egy

R3

(RR

LD

X)

2030

+

Rus

sell

Lif

ePoi

nts

2040

T

arge

t-D

ate

Str

ateg

y R

3 (R

XL

DX

) 20

30+

8

11.0

9 1.

51 *

* -

-8.

9 66

.0

22.5

2.

5

Sch

wab

Tar

get 2

010

Tar

get-

Dat

e S

CH

WA

B

~SWBRX2

2000

-201

4 51

10

.9

0.91

**

--

5.6

60.4

27

.5

6.4

Sch

wab

Tar

get 2

020

Tar

get-

Dat

e ~SWCRX2

2015

-202

9 67

10

.99

0.92

**

--

3.0

68.4

22

.8

5.8

Sch

wab

Tar

get 2

030

Tar

get-

Dat

e ~SWDRX2

2030

+

43

11.0

6 0.

93**

-

-3.

0 74

.7

17.5

4.

8 S

chw

ab T

arge

t 204

0 T

arge

t-D

ate

(SW

ER

X)

2030

+

25

11.1

7 0.

99**

-

-3.

2 81

.0

11.9

3.

8

Sel

igm

an T

argE

TF

und

Cor

e T

arge

t-D

ate

SE

LIG

MA

N

A~SHVAX2

2000

-201

4 35

7.

47

1.33

**

4.75

-

7.5

62.6

29

.6

0.2

Sel

igm

an T

argE

TF

und

2015

T

arge

t-D

ate

A ~STJAX2

2015

-202

9 19

7.

64

1.35

**

4.75

-

0.7

88.4

10

.7

0.3

Sel

igm

an T

argE

TF

und

2025

T

arge

t-D

ate

A(S

TK

AX

) 20

15-2

029

17

7.66

1.

35**

4.

75

-0.

6 98

.8

0.5

0.1

Sta

te F

arm

Lif

ePat

h 20

10 A

T

arge

t-D

ate

ST

AT

EF

AR

M

Le~acr (

SA

TA

X)

2000

-201

4 34

9 12

.31

1.24

5.

00

-11

.9

43.8

44

.3

0.0

Sta

te F

arm

Lif

ePat

h 20

20 A

T

arge

t-D

ate

Le~acr ~SAW AX

2 20

15-2

029

556

13.3

4 1.

22

5.00

-

14.1

58

.7

27.1

0.

0 S

tate

Far

m L

ifeP

ath

2030

A

Tar

get-

Dat

e Le

~acr

~SAY AX

2 20

30+

39

9 14

.08

1.21

5.

00

-15

.7

69.1

15

.2

0.0

Sta

te F

arm

Lif

ePat

h 20

40 A

T

arge

t-D

ate

Leg

acy

(SA

UA

X)

2030

+

281

14.7

5 1.

21

5.00

-

10.0

84

.2

5.7

0.0

76

To

tal

NA

V

Exp

ense

F

ron

t D

efer

red

M

orn

ing

star

A

sset

s ($

) R

atio

L

oad

L

oad

A

sset

All

ocat

ion

(as

at

Fu

nd

Fam

ily

Fu

nd

Nam

e C

ateg

ory

{$

mil

lion

s}

23/0

8/06

} {%

} {%

} {%

} {a

s at

24/

08/0

6}

Cas

h

Sto

cks

Bon

ds

Oth

er

%

(%)

(%)

(%)

Sta

te F

arm

Lif

ePat

h In

com

e T

arge

t-D

ate

A Les;ac~ ~SLRAX2

2000

-201

4 13

8 11

.48

1.26

5.

00

-12

.8

34.6

52

.6

0.0

Sun

Am

eric

a 20

10 H

igh

Tar

get-

Dat

e S

UN

AM

ER

ICA

W

ater

mar

k C

~HWICX2

2000

-201

4 29

0 10

.34

2.30

-

1.00

26

.6

0.0

73.4

0.

0 S

unA

mer

ica

2015

Hig

h T

arge

t-D

ate

Wat

erm

ark

A ~HWF A

X2

2015

-202

9 14

6 11

.05

1.65

5.

75

-45

.5

0.0

54.5

0.

0 S

unA

mer

ica

2020

Hig

h T

arge

t-D

ate

Wat

erm

ark

A (

HW

KA

X)

2015

-202

9 53

11

.16

1.65

5.

75

-53

.3

0.0

46.7

0.

0

TIA

A-C

RE

F I

nsti

tuti

onal

T

arge

t-D

ate

TIA

A-C

RE

F

Life

c~cl

e 20

10 ~TCLEX)

2000

-201

4 40

10

.79

0.60

**

--

3.6

46.5

49

.5

0.3

TIA

A-C

RE

F I

nsti

tuti

onal

T

arge

t-D

ate

Life

c~cl

e 20

15 ~TCLIX)

2015

-202

9 39

10

.83

0.61

**

--

3.2

53.5

42

.9

0.4

TIA

A-C

RE

F I

nsti

tuti

onal

T

arge

t-D

ate

Life

c~cl

e 20

20 ~TCLTX2

2015

-202

9 30

10

.93

0.61

**

--

2.9

58.4

38

.3

0.4

TIA

A-C

RE

F I

nsti

tuti

onal

T

arge

t-D

ate

Life

c~cl

e 20

25 ~TCLFX2

2015

-202

9 25

10

.98

0.61

**

--

2.6

63.4

33

.6

0.4

TIA

A-C

RE

F I

nsti

tuti

onal

T

arge

t-D

ate

Life

c~cl

e 20

30 ~TCLNX)

2030

+

20

11.0

2 0.

61 *

* -

-2.

3 68

.3

29.0

0.

4 T

IAA

-CR

EF

Ins

titu

tion

al

Tar

get-

Dat

e Li

fec~

cle

2035

~TCLRX2

2030

+

14

11.0

8 0.

61 *

* -

-2.

0 73

.3

24.3

0.

5 T

IAA

-CR

EF

Ins

titu

tion

al

Tar

get-

Dat

e L

ifec

ycle

204

0 (T

CL

OX

) 20

30+

14

11

.15

0.61

**

--

1.7

78.2

19

.6

0.5

T.R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

T.R

OW

E P

RIC

E

2005

~TRRFX2

2000

-201

4 49

1 11

.3

0.67

**

--

11.2

54

.9

33.2

0.

7

T.R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

2085

15

.18

0.71

**

--

8.0

63.4

27

.8

0.8

77

To

tal

NA

V

Ex

pen

se

Fro

nt

Def

erre

d

Mo

rnin

gst

ar

Ass

ets

($)

Rat

io

Lo

ad

Lo

ad

Ass

et A

lloc

atio

n (a

s at

F

un

d F

amil

y

Fu

nd

Nam

e C

ateg

ory

{$

mil

lion

s}

23/0

8/06

} {%

} {%

} {%

} {a

s at

24/

08/0

6}

Cas

h

Sto

cks

Bo

nd

s O

ther

(%

) (%

(%

) (%

) 20

10 (

TR

RA

X)

2000

-201

4

T .R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

2015

~TRRGX2

2015

-202

9 13

44

11.7

0.

75**

-

-6.

2 69

.9

23.1

0.

9

T.R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

2020

~TRRBX2

2015

-202

9 27

96

16.3

3 0.

80**

-

-4.

6 76

.2

18.2

0.

9 T

.Row

e P

rice

Ret

irem

ent

Tar

get-

Dat

e 20

25 (

TR

RH

X)

2015

-202

9 11

76

12

0.82

**

--

4.0

81.7

13

.5

0.9

T.R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

2030

~TRRCX)

2030

+

1835

17

.29

0.84

**

--

3.8

87.5

7.

9 0.

8

T.R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

2035

~TRRJX2

2030

+

566

12.1

8 0.

84**

-

-3.

8 89

.0

6.4

0.8

T .R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

2040

~TRRDX2

2030

+

857

17.3

7 0.

84**

-

-3.

8 89

.0

6.3

0.8

T.R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

2045

~TRRKX2

2030

+

120

11.3

6 0.

81 *

* -

-3.

8 89

.0

6.4

0.8

T.R

owe

Pri

ce R

etir

emen

t T

arge

t-D

ate

Inco

me

(TR

RIX

) 20

00-2

014

697

12.6

6 0.

57**

-

-23

.4

41.7

34

.4

0.6

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e V

AN

GU

AR

D

2005

~VTOVX)

2000

-201

4 90

7 11

.17

0.20

**

--

2.8

32.2

65

.0

0.0

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

10 (

VT

EN

X2

2000

-201

4 5

20.5

7 0.

21 *

* -

-N

/A

NIA

N

/A

N/A

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

15 ~VTXVX2

2015

-202

9 33

54

11.8

3 0.

20**

-

-1.

4 48

.2

50.1

0.

4

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

20 ~VTWNX)

2015

-202

9 6

20.6

4 0.

21 *

* -

-N

/A

N/A

N

/A

N/A

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

25 (

VT

TV

X)

2015

-202

9 36

01

12.2

0.

20**

-

-1.

3 58

.3

40.0

0.

4

78

To

tal

NA

V

Exp

ense

F

ron

t D

efer

red

M

orn

ing

star

A

sset

s ($

) R

atio

L

oad

L

oad

A

sset

All

ocat

ion

(as

at

Fu

nd

Fam

ily

F

un

d N

ame

Cat

ego

ry

{$m

illi

ons}

23

/08/

06)

(%}

{%)

(%)

{as

at 2

4/08

/06}

C

ash

S

tock

s B

onds

O

ther

(%

V

angu

ard

Tar

get R

etir

emen

t T

arge

t-D

ate

2030

~VTHRX2

2030

+

4 20

.7

0.21

**

N/A

N

/A

N/A

N

/A

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

35 ~VTTHX2

2030

+

2310

12

.84

0.20

**

--

1.1

75.6

22

.7

0.6

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

40 ~VFORX2

2030

+

6 20

.58

0.21

**

--

N/A

N

/A

N/A

N

/A

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

45 ~VTIVX2

2030

+

1055

13

.25

0.20

**

--

1.1

86.9

11

.3

0.7

Van

guar

d T

arge

t Ret

irem

ent

Tar

get-

Dat

e 20

50 (

VF

IFX

) 20

30+

4

20.6

8 0.

21 *

* -

-N

/A

N/A

N

/A

N!A

V

angu

ard

Tar

get R

etir

emen

t T

arge

t-D

ate

Inco

me

(VT

INX

) 20

00-2

014

799

10.4

5 0.

21 *

* -

-6.

7 20

.6

72.6

0.

2

--

V A

NT

AG

EP

OIN

T

V an

tage

poin

t Mil

esto

ne

Tar

get-

Dat

e F

UN

DS

20

10 ~VPRQX2

2000

-201

4 36

10

.62

0.86

**

--

5.1

45.1

49

.7

0.1

V an

tage

poin

t Mil

esto

ne

Tar

get-

Dat

e 20

15 ~VPRPX2

2015

-202

9 60

10

.83

0.89

**

--

3.9

60.6

35

.4

0.1

V an

tage

poin

t Mil

esto

ne

Tar

get-

Dat

e 20

20 ~VPROX2

2015

-202

9 44

10

.93

0.91

**

--

3.3

69.4

27

.3

0.1

Van

tage

poin

t Mil

esto

ne

Tar

get-

Dat

e 20

25 ~VPRNX)

2015

-202

9 33

11

.03

0.93

**

--

2.8

76.4

20

.8

0.1

V an

tage

poin

t Mil

esto

ne

Tar

get-

Dat

e 20

30 ~VPRMX2

2030

+

18

11.1

2 0.

95**

-

-2.

7 81

.7

15.5

0.

1 V

anta

gepo

int M

iles

tone

T

arge

t-D

ate

2035

~VPRLX2

2030

+

11

11.1

8 0.

97**

-

-2.

8 86

.5

10.7

0.

1 V

anta

gepo

int M

iles

tone

T

arge

t-D

ate

2040

~VPRKX2

2030

+

7 11

.17

0.97

**

--

2.8

87.5

9.

7 0.

1

V an

tage

poin

t Mil

esto

ne

Tar

get-

Dat

e 18

10

.45

0.81

**

--

6.5

29.5

63

.9

0.0

79

Mor

ning

star

T

otal

A

sset

s N

AV

($

) (a

s at

Exp

ense

R

atio

F

ron

t L

oad

Def

erre

d L

oad

Ass

et A

lloc

atio

n

Fu

nd

Fam

ily

Fu

nd

Nam

e C

ateg

ory

($m

illi

ons)

23

/08/

06)

(%)

(%)

(%)

(as

at 2

4/08

/06)

WE

LL

S F

AR

GO

Ret

irem

ent I

ncom

e (V

PR

RX

)

Wel

ls F

argo

Adv

anta

ge D

J T

arge

t 201

0 A

(S

TN

RX

) W

ells

Far

go A

dvan

tage

DJ

Tar

get 2

020

A (

ST

TR

X)

Wel

ls F

argo

Adv

anta

ge D

J T

arge

t 203

0 A

(S

TH

RX

) W

ells

Far

go A

dvan

tage

DJ

Tar

get 2

040

A (

ST

FR

X)

Wel

ls F

argo

Adv

anta

ge D

J T

arge

t Tod

ay A

(S

TW

RX

)

2000

-201

4

Tar

get-

Dat

e 20

00-2

014

Tar

get-

Dat

e 20

15-2

029

Tar

get-

Dat

e 20

30+

T

arge

t-D

ate

2030

+

Tar

get-

Dat

e 20

00-2

014

207

12.9

5 1.

25

378

14.3

1 1.

25

240

15.0

2 1.

25

262

16.6

9 1.

25

85

10.2

8 1.

25

Cas

h S

tock

s B

onds

(%

) (%

) (%

)

5.75

-

0.6

44.9

54

.5

5.75

-

0.8

63.1

36

.1

5.75

-

1.2

76.9

21

.9

5.75

-

0.6

89.2

10

.2

5.75

-

0.5

35.5

64

.1

80

Oth

er

(%)

0.0

0.0

0.0

0.0

0.0

AP

PE

ND

IXB

Tab

le 1

D

t;{z

nitio

ns o

fVa

ria

ble

i8 E

mpl

oyed

in R

egre

ssio

n A

naly

sis

Var

iabl

e N

ame

Equ

ity

Par

tici

pati

on

Equ

ity

All

ocat

ion

Age

Mal

e E

rCon

tsL

ast1

2Mth

s M

arit

al S

tatu

s R

isk

Pro

file

Edu

cati

on

Sha

res/

MIF

N

et W

ealt

h (e

xcl.

Sup

er)

Abb

revi

atio

n S

H A

LL

OC

SH

AR

ES

AG

EU

36

AG

E36

42

AG

E43

48

AG

E49

54

AG

E0

55

M

AL

E

ER

C

ON

TS

MA

RR

IED

DE

FA

CT

O

LO

W T

O A

VE

RA

GE

RIS

K

--

-H

IGH

RIS

K

HIG

H S

CH

OO

L

CE

RT

TR

AD

E D

IP

--

BA

CH

EL

OR

DE

G

PO

ST G

RA

D D

EG

-

-SH

AR

ES

MIF

LO

WE

R

WE

AL

TH

A

VE

RA

GE

WE

AL

TH

H

IGH

WE

AL

TH

Des

crip

tion

=

1 i

f mem

ber

has

an e

quit

y al

loca

tion

in

thei

r su

pera

nnua

tion

acc

ount

, 0

othe

rwis

e P

erce

ntag

e al

loca

tion

to e

quit

y in

sup

eran

nuat

ion

acco

unt

(%)

Dum

my

vari

able

s in

dica

ting

age

of m

embe

r at

the

tim

e o

f cho

ice:

=

1

if m

embe

r is

les

s th

an 3

6 yr

s ol

d, 0

oth

erw

ise

=

1 if

mem

ber

is b

etw

een

36 a

nd 4

2 yr

s ol

d, 0

oth

erw

ise

=

1 if

mem

ber

is b

etw

een

43 a

nd 4

8 yr

s ol

d, 0

oth

erw

ise

=

1 if

mem

ber

is b

etw

een

49 a

nd 5

4 yr

s ol

d, 0

oth

erw

ise

=

1 if

mem

ber

is m

ore

than

55

yrs

old,

0 o

ther

wis

e =

1

if m

embe

r is

mal

e, 0

if f

emal

e T

otal

mem

ber

cont

ribu

tion

am

ount

in

2004

($)

=

1

ifM

arri

ed o

r in

De-

fact

o re

lati

onsh

ip,

0 ot

herw

ise

Dum

my

vari

able

s in

dica

ting

the

mem

ber'

s vi

ew o

f ris

k:

=

1 if

take

Low

to A

vera

ge r

isk

expe

ctin

g to

ear

n lo

w-a

vera

ge r

etur

ns,

0 ot

herw

ise

=

1 if

take

Hig

h ri

sk e

xpec

ting

to e

arn

abov

e av

erag

e re

turn

s, 0

oth

erw

ise

Dum

my

vari

able

s in

dica

ting

mem

ber'

s le

vel

of e

duca

tion

: =

1

if c

ompl

eted

som

e le

vel

of H

igh

scho

ol,

0 ot

herw

ise

=

1 if

com

plet

ed C

erti

fica

te/T

rade

/Dip

lom

a le

vel,

0 ot

herw

ise

=

1 if

com

plet

ed B

ache

lor

degr

ee,

0 ot

herw

ise

=

1 if

com

plet

ed P

ost-

Gra

duat

e de

gree

, 0

othe

rwis

e =

1

if m

embe

r in

vest

s in

sha

res

or m

anag

ed in

vest

men

t fun

ds,

0 ot

herw

ise

Dum

my

vari

able

s in

dica

ting

mem

ber'

s le

vel

ofw

ealt

h (

excl

udin

g su

per)

: =

1

if b

etw

een

Low

er w

ealt

h, 0

oth

erw

ise

=

1 if

bet

wee

n A

vera

ge w

ealt

h, 0

oth

erw

ise

=

1 if

bet

wee

n H

igh

wea

lth,

0 o

ther

wis

e

28 W

here

ther

e w

as m

ore

than

one

dum

my

vari

able

in

a pa

rtic

ular

cat

egor

y, o

ne d

umm

y va

riab

le w

as e

xclu

ded

from

the

reg

ress

ion

and

used

as

a re

fere

nce

grou

p (i

.e.

the

grou

p ag

ains

t whi

ch c

ompa

riso

ns a

re m

ade)

. U

sing

all

dum

my

vari

able

s w

ould

res

ult i

n ov

er s

peci

fica

tion

, in

trod

ucin

g pe

rfec

t col

line

arit

y.

81

Cal

enda

r Y

ear

CA

L YR

* Tho

se d

umm

y va

riab

les

in b

old

are

the

refe

renc

e gr

oups

.

Cal

enda

r ye

ar o

f cho

ice.

Dum

my

vari

able

s to

cap

ture

pos

sibl

e ti

me

effe

cts

(199

7 -

2004

). F

or e

ach

obse

rvat

ion:

=

1 in

a p

arti

cula

r ye

ar,

0 ot

herw

ise

82

Table 2 Sue.erannuation Fund Investment Strategy Choices

HESTA STA GESB UniSuper

Cash Plus Balanced Cash Plan Cash High Conservative Capital

Core Pool Growth Plan Stable Sustainable Balanced Conservative

Shares Plus Balanced Plan Balanced Readymade

Options Eco Pool Stable Growth Plan Balanced O'seas Share Capital Pool Guaranteed Growth Aust Share Pool SR Shares

Shares Cash Cash Cash Cash

Fixed Interest; Inflation

AustFixed Linked Interest; Bonds;

Fixed Int'l Fixed Global Interest Fixed Interest Interest Bonds

Aust Property; Int'l

Property Property Pro2erty Property Asset Aust Shares;

Classes Int'l Shares; Aust Sustainable Shares; Int'l

Aust Shares; Sustainable Aust Shares; Shares Int'l Shares Shares Int'l Shares

Absolute Return Absolute Strategy Return Funds; Strategy Commodities; Funds; Private Private Equity; Equity;

Other Infrastructure Infrastucture

83

APPENDIXC

Table 1 Average Asset Allocations by Group- Full Sample Excluding Readymade Option This table presents statistics for average member asset allocations, showing the average percentage of the superannuation held in cash, fixed interest, property, shares, and 'other'. Observations are sorted by age only, and gender and age. Excludes observations where members adopt only one readymade option as opposed to a do-it-yourself strategy, or a mixture ofboth the readymade oEtion and the do-it-yourself strategy.

Obs Cash F.Interest Pro,eerty Shares Other ALL (Excluding 53233 9.84 15.58 15.25 53.02 6.32 Readl:_made) AGE: <36 21666 10.35 15.30 13.99 54.16 6.21 36-42 9891 8.28 16.04 14.98 54.94 5.76 43-48 6319 7.43 13.01 17.87 55.22 6.46 49-54 5893 8.87 17.50 16.57 50.99 6.06 55+ 4371 14.21 21.13 15.51 43.72 5.42 Unknown 5093 11.01 12.07 16.12 52.07 8.74 GENDER: Male 21108 8.80 17.83 14.45 53.39 5.53

<36 9485 8.97 17.06 13.39 54.67 5.90 36-42 3363 7.33 16.40 14.63 56.40 5.24 43-48 3637 7.15 17.54 15.96 53.98 5.36 49-54 2499 8.50 19.06 15.72 51.52 5.20 55+ 2123 13.53 22.56 14.86 44.05 5.00

Age unknown 1 2.95 10.40 12.95 66.55 7.15

Female 27022 10.43 14.48 15.70 52.91 6.48 <36 12173 11.43 13.92 14.45 53.75 6.45 36-42 4260 8.69 12.81 16.20 55.74 6.57 43-48 4948 8.32 13.62 17.13 54.31 6.62 49-54 3393 9.15 16.35 17.20 50.60 6.70 55+ 2248 14.85 19.79 16.13 43.40 5.81

Gender unknown 5103 10.99 12.07 16.12 52.09 8.73

84

Table 2 Average Asset Allocations by Group -Restricted Sample Excluding Readymade Option This table presents statistics for average member asset allocations, showing the average percentage of the superannuation held in cash, fixed interest, property, shares, and 'other'. Observations are sorted by age only, and gender and age. Excludes observations where members adopt only one readymade option as opposed to a do-it-yourself strategy, or a mixture of both the readymade option and the do-it-yourself strategy.

Obs Cash F.Interest Pro,eerty Shares Other ALL (Excluding 490 8.84 17.37 22.44 42.75 11.45 Readl:made) AGE: <36 93 5.76 14.92 19.96 48.98 15.56 36-42 96 6.43 13.39 26.35 45.59 13.42 43-48 101 7.02 13.97 24.41 46.77 10.42 49-54 98 5.34 19.68 23.01 43.75 10.74 55+ 102 19.07 24.51 18.53 29.45 8.97 GENDER: Male 213 8.72 14.70 19.86 47.55 12.53 <36 47 4.82 14.32 18.35 52.03 16.99 36-42 43 3.20 11.94 22.45 49.32 16.08 43-48 52 5.28 13.31 23.05 53.19 7.47 49-54 32 4.20 11.81 16.33 57.62 15.30 55+ 39 27.78 22.42 17.45 24.42 8.83

Female 272 9.06 19.71 24.31 39.00 10.41 <36 46 6.71 15.55 21.62 45.87 14.29 36-42 53 9.05 14.57 29.51 42.55 9.53 43-48 49 8.86 14.66 25.85 39.96 13.07 49-54 66 5.89 23.50 26.25 37.03 8.97 55+ 58 14.69 27.68 18.18 31.74 7.98 Unknown 5 2.00 4.00 31.00 42.00 21.00 MARITAL STATUS: Married/De-facto 343 9.69 17.01 21.52 43.91 10.68 Unmarried 145 6.84 17.99 24.76 39.97 13.27 Unknown 2 7.50 35.00 12.50 45.00 0.00

85