1

Information Cost and Member Choice in the Australian Superannuation Industry

Xiaowen Peng Bachelor of Commerce (Honours)

A thesis submitted for the degree of Doctor of Philosophy at The

University of Queensland in 2017

UQ Business School

2

Abstract

This thesis examines the determinants of superannuation members’ investment choices

across fund families, with a focus on the effect of information costs. Existing literature on managed

funds may not be easily generalised to the superannuation fund setting given the distinct attributes

and behaviour of superannuation investors. In contrast to prior research, this thesis uses an actual

decision-making measure, namely, member-nominated transfers, to examine investors who have

‘made a choice’ at the fund family level. While members’ information costs are not directly

observable or measurable, this thesis identifies two attributes that can reduce members’ information

search and processing costs, namely, superannuation fund families’ visibility and transparency.

Specifically, this thesis examines whether superannuation fund families’ marketing effort and the

quality of their Product Disclosure Statement (PDS) together with their performance, can be used

to attract investment flows. Based on a unique data set from the Australian Prudential

Regulation Authority (APRA) on regulated superannuation fund families,1 the results in Chapter 4

indicate that the asymmetric return-chasing behaviour commonly found in research on managed

funds does not exist in the superannuation context. The results show that superannuation investors

do not actively switch to better performing fund families but they do choose to leave poorly

performing fund families. A consistent positive relation is found between marketing efforts and

member choice. Further analysis shows that this result is mainly driven by retail funds, and it

is confined to fund families that conduct more extensive marketing. This thesis then provides

evidence on the current superannuation PDS disclosure practice in Chapter 5 using readability

(linguistic style) and extent of disclosure (linguistic content) under the new ‘short PDS regime’.

Results indicate that, on average, industry superannuation fund families provide better PDS

disclosure than their retail counterparts, in terms of both readability and the extent of disclosure.

In addition, the causes and consequences of superannuation fund families’ PDSs are studied in

Chapter 5 and Chapter 6. There is no significant relation between superannuation fund families’

performance and the quality of PDS disclosure. Increased investment inflows encourage

superannuation families to provide more information in an easier-to-understand manner in the

PDS, especially the industry families. And better quality/content of PDS disclosure can attract

inflows, especially for the retail families.

1 I thank the Australia Prudential Regulation Authority (APRA) for providing the proprietary superannuation data.

3

Declaration by author

This thesis is composed of my original work, and contains no material previously published or

written by another person except where due reference has been made in the text. I have clearly

stated the contribution by others to jointly-authored works that I have included in my thesis.

I have clearly stated the contribution of others to my thesis as a whole, including statistical assistance,

survey design, data analysis, significant technical procedures, professional editorial advice, and any

other original research work used or reported in my thesis. The content of my thesis is the result of

work I have carried out since the commencement of my research higher degree candidature and

does not include a substantial part of work that has been submitted to qualify for the award of

any other degree or diploma in any university or other tertiary institution. I have clearly stated

which parts of my thesis, if any, have been submitted to qualify for another award.

I acknowledge that an electronic copy of my thesis must be lodged with the University Library and,

subject to the policy and procedures of The University of Queensland, the thesis be made available

for research and study in accordance with the Copyright Act 1968 unless a period of embargo has

been approved by the Dean of the Graduate School.

I acknowledge that copyright of all material contained in my thesis resides with the copyright

holder(s) of that material. Where appropriate I have obtained copyright permission from the

copyright holder to reproduce material in this thesis.

4

Publications during candidature

No publications Publications included in this thesis No publications included

Contributions by others to the thesis No contributions by others

Statement of parts of the thesis submitted to qualify for the award of another degree None Research Involving Human or Animal Subjects No animal or human subjects were involved in this research

5

Acknowledgements

My doctorate has been a very fulfilling experience which allows me to develop as an

academic. I am grateful for all the help and encouragement I received along the way to accomplish

this PhD thesis.

I am deeply indebted to my supervisors Dr. Grace Hsu, Dr. Karen Alpert and Professor Peter

Clarkson for their consistent support, encouragement and stimulating suggestions. It has been a

great privilege and pleasure to study under the supervision of such patient, strong and supportive

advisors. I would also like to thank my PhD Committee Chair, Julie Walker, my readers, Associate

Professor Jacquelyn Humphrey and Honorary Associate Professor Irene Tutticci, and UQ Business

School staff. Thank you for the time devoted to the examination process.

I also wish to thank my parents for their unconditional love and trust as always, and my

husband Ronghong Huang for his company and support.

UQ Business School has been very special to me. I came to Brisbane 9 years ago to study at

UQ. I spent my 20s at UQ Business School, completing my Undergraduate and Honours degrees,

meeting my life partner and starting my first full-time job. The completion of my PhD thesis marks

another significant milestone in my life. I wish my memories at UQ will go on, and I am excited for

the challenges ahead.

6

Financial support

Financial support from the APRA, the Reserve Bank of Australia and the Brian Gray Scholarship is gratefully acknowledged.

7

Keywords

superannuation, choice, Product Disclosure Statement, Australia, disclosure Australian and New Zealand Standard Research Classifications (ANZSRC)

ANZSRC code: 150103, Financial Accounting, 50%

ANZSRC code: 150203, Financial Institutions, 50% Fields of Research (FoR) Classification

FoR code: 1501, Accounting, Auditing and Accountability, 50%

FoR code: 1502, Banking, Finance and Investment, 50%

8

Contents Abstract ................................................................................................................................................................ 2

Declaration by author ........................................................................................................................................... 3

Acknowledgements .............................................................................................................................................. 5

Chapter 1. Introduction ....................................................................................................................................... 10

Chapter 2 – Institutional Background ................................................................................................................. 13

2.1 Australian Superannuation ....................................................................................................................... 13

2.2 ‘Choice of Fund’ Policy ............................................................................................................................ 15

2.3 Superannuation Product Disclosure Statements and the Short-form Product Disclosure Statement Regime ................................................................................................................................................ 16

2.4 Managed Funds and Superannuation Funds ............................................................................................. 18

2.5 U.S. Pensions and Australian Superannuation .......................................................................................... 18

2.6 Fund Families and Individual Funds ......................................................................................................... 19

Chapter 3 - Literature Review ............................................................................................................................ 21

3.1 Behaviour of Defined Contribution Plan Members .................................................................................. 21

3.2 Optimal Dynamic Portfolio Allocation Theory ........................................................................................ 21

3.3 Flow-performance Relation ...................................................................................................................... 22

3.4 Information Costs, Marketing and Fund Flows ........................................................................................ 23

3.5 Fund Disclosure and Fund Flows ............................................................................................................. 25

Chapter 4 – Determinants of Member Choice .................................................................................................... 27

4.1 Introduction .............................................................................................................................................. 27

4.2 Hypothesis Development .......................................................................................................................... 29

4.3 Data and Methodology ............................................................................................................................. 30

4.3.1 Data Source ....................................................................................................................................... 30

4.3.2 Sample Selection Process .................................................................................................................. 31

4.3.3 Regression Model Specification ........................................................................................................ 32

4.3.4 Dependent Variable: Member-initiated Family Rollover .................................................................. 32

4.3.5 Alternative Member Choice Measures .............................................................................................. 33

4.3.6 Fund Family Performance ................................................................................................................. 34

4.3.7 Marketing Expense ............................................................................................................................ 34

4.3.8 Control Variables ............................................................................................................................... 35

4.4 Results ...................................................................................................................................................... 35

4.4.1 Summary Statistics ............................................................................................................................ 35

4.4.2 Effect of Performance ........................................................................................................................ 37

4.4.3 Effect of Marketing on Fund Family Flows ...................................................................................... 39

4.4.4 Control Variables ............................................................................................................................... 40

4.4.5 Alternative Choice Measures............................................................................................................. 41

4.4.6 Further Analysis on the Effect of Marketing on Fund Family Flows ................................................ 43

4.4.7 Robustness Tests ............................................................................................................................... 45

4.5 Conclusion .......................................................................................................................................... 47

9

Chapter 5 – The Short Product Disclosure Statement (PDS) Disclosure Practice ............................................. 49

5.1 Introduction .............................................................................................................................................. 49

5.2 Literature Review and Hypothesis Development ............................................................................... 51

5.2.1 The Need for Quality Superannuation Information .................................................................... 51

5.2.2 Experimental Studies on Superannuation Disclosure ................................................................. 51

5.2.3 Performance and Disclosure ....................................................................................................... 52

5.3 Data..................................................................................................................................................... 54

5.4 Methodology ....................................................................................................................................... 55

5.4.1 Readability Measures ................................................................................................................. 55

5.4.2 Content Analysis Score ..................................................................................................................... 56

5.4.3 An example of scoring: Smartsave versus Suncorp Super ................................................................. 58

5.4.4 Regression model for hypotheses testing ........................................................................................... 61

5.5 Univariate Results ..................................................................................................................................... 62

5.5.1 Summary Statistics ............................................................................................................................ 62

5.5.2 Disclosure Scores by Financial Year ................................................................................................. 65

5.5.3 Correlation .................................................................................................................................. 68

5.6 Multivariate results ............................................................................................................................. 69

5.6.1 PDS Disclosure and Performance ............................................................................................... 69

5.6.2 Additional Analysis .................................................................................................................... 71

5.7 Conclusion .......................................................................................................................................... 72

Chapter 6 - Product Disclosure Statement and Member Choice ........................................................................ 73

6.1 Introduction .............................................................................................................................................. 73

6.2 Literature Review and Hypothesis Development ............................................................................... 74

6.3 Research Method and Results ................................................................................................................... 76

6.3.1 PDS disclosure and Fund Family Inward Fund Flows ...................................................................... 76

6.3.2 Simultaneous Equations Model with Current Flows ......................................................................... 77

6.4 Multivariate Results ............................................................................................................................ 78

6.5 Additional Analysis and Discussion ......................................................................................................... 90

6.5.1 Lagged regression model ................................................................................................................... 90

6.5.2 Combined effect of marketing and PDS disclosure ........................................................................... 92

6.6 Limitations ................................................................................................................................................ 93

6.7 Conclusion ................................................................................................................................................ 93

Chapter 7 - Conclusion ....................................................................................................................................... 95

Bibliography ....................................................................................................................................................... 96

Appendix 1 ....................................................................................................................................................... 102

10

Chapter 1. Introduction

Superannuation is the term used in Australia to describe the setting aside of income for

retirement, which is generally known internationally as pension or retirement products. Because of

the privatisation of the Australian retirement saving system, superannuation has become the principal

retirement funding vehicle for Australians. Whether individuals have adequate savings for retirement

is crucial to the individuals’ well-being, and has important implications for a country’s social and

economic environments. The compulsory Superannuation Guarantee (SG) coupled with the

stipulation that benefits cannot be withdrawn until retirement has led to explosive growth in Australian

superannuation asset. Despite the social and economic significance of superannuation to Australians,

there are major concerns about investor inertia owing to factors such as the complexity of

superannuation regulation, the large number of choices of superannuation families and products,

investors’ lack of financial expertise, and the complexity/lack of clarity of superannuation disclosures.

Superannuation Product Disclosure Statements (PDSs) provide basic but important disclosure, but

they have been criticised for being difficult to understand.2 This partly occurs because superannuation

fund families treat PDSs as a ‘compliance tool’ to minimise their legal liability. The result is an

increase in information costs for investors and an increase in difficulties that investors face when

trying to make an informed investment decision. Such criticism has led to a review by the Australian

Securities and Investments Commission (ASIC) on the superannuation system in 2010 and the

introduction of the ‘Short-Form PDS Regime’ in 2011. Since these regulatory changes were

introduced, not much is known about the current disclosure practice of superannuation fund families.

To gain an understanding of superannuation members’ investment choices and the causes and

consequences of superannuation fund families’ PDS disclosure, the determinants of superannuation

members’ investment choices are examined with a focus on the effects of (1) performance, (2)

marketing efforts and (3) PDS readability/extent of disclosure.

The Australian superannuation system seems to suffer from an inherent contradiction (Cheah

et al., 2015). The system appears to assume that individuals are not prudently saving for their

retirement; thus the government mandates savings for the entire population of working Australians

through the compulsory SG scheme. However, individual members are required to choose where

and how the savings are invested, regardless of their ability to navigate the complexity of the system.

This tension in Australian superannuation presents a significant research opportunity. As a

mandatory system, superannuation avoids many selection bias problems. Since the system is

2 Superannuation fund families are required by the Financial Service Reform (FSR) Act 2001 and Corporations Act 2001 to provide comprehensive product information using PDSs.

11

complex and offers many choices to investors, it creates a fertile environment for exploring

superannuation members’ long-term financial decision making.

The first research question examines potential determinants of superannuation members’

investment choice, with a particular focus on fund families’ performance and marketing efforts.

Superannuation investors face significant information costs when they seek to understand a market

with complicated regulations, a wide array of investment choices, overly-technical disclosure and

information overload. Fund flow research overseas, particularly in relation to the U.S. managed

fund market, has found that fund performance is only one of several factors that affect fund flows.

Other factors, such as marketing spending and recommendations from star rating agents, have been

found to have an important influence on investor behaviour (Cronqvist & Nilsson 2005; Foster, Ng

& Wee 2015; Jain & Wu 2000). Therefore, research question 1 examines the determinants of

member choice in the Australian superannuation industry, with a focus on the role of fund family

performance and marketing efforts. There has been extensive research on U.S. managed funds to

study the drivers of fund flows. However, a study of the superannuation context will contribute to

the literature because Australian superannuation funds are very different from the managed fund

setting given its mandatory nature and the characteristics of unsophisticated investors. In addition,

this thesis uses a unique data set from the regulator APRA, which increases the sample size and

allows the use of new measures for member choice. The first research question addressed in

Chapter 4 is used as a building block for the following research questions.

For superannuation members who make active investment decisions, the investment

selection process is likely to start with the identification of a preferred superannuation fund family

and to then proceed with the search for the one product that is best suited to their investment needs.

This process likely requires investors to undertake at least some research in order to understand the

fund families and their products. The superannuation fund’s PDS is supposed to serve as a primary

communication mechanism between superannuation fund families and investors. Research

question 2 provides an exploratory analysis of the disclosure practice of Australian superannuation

fund families by examining the standard PDS under the new ‘short-form PDS regime’ introduced

by the government in 2011. To provide some preliminary evidence on regulators’ concerns about

the quality of superannuation PDSs, two dimensions of PDS disclosure are examined to obtain an

understanding of the readability and content (informativeness) of short-form PDSs under the current

regime. Additionally, arguments regarding managerial disclosure incentive are also tested to

determine whether performance affects superannuation families’ strategic use of PDS disclosure.

Research question 3 brings together the first two research questions and examines whether better

12

quality PDS disclosure facilitates superannuation member choice in Australia, i.e., whether the

readability and extent of PDS disclosure are associated with observable investment flows.

The remainder of this thesis proceeds as follows. Chapter 2 introduces the institutional

background of the Australian superannuation industry. Chapter 3 reviews relevant literature and

proposes hypotheses. The effects of fund family performance and marketing effort on superannuation

member’s choice are discussed in Chapter 4. Chapter 5 provides exploratory evidence on the current

disclosure practice of superannuation fund families using two attributes of their PDSs: readability

and content. Causes of superannuation PDS disclosure including performance and investment

inflows are investigated in Chapter 5 and Chapter 6. Chapter 6 examines the consequences of PDS

disclosure on superannuation member choice. Section 7 concludes the thesis.

13

Chapter 2 – Institutional Background

This thesis examines the determinants of superannuation members’ investment choice with a

focus on factors such as fund family performance, marketing effort and the quality and extent of

PDS disclosure. Therefore, this chapter provides an overview of the institutional background of the

Australian superannuation system. The importance of examining superannuation funds directly

despite the existing literature on managed funds is also discussed in section 2.4. Finally, the choice

of superannuation fund family as the unit of analysis is explained in section 2.6.

2.1 Australian Superannuation

Australia’s three-pillar retirement funding system consists of mandatory superannuation

contributions by employers under the government’s Superannuation Guarantee (hereafter SG),

supplementary voluntary contributions by employees to superannuation accounts, and the

government-funded age pension. 3 Owing to the incremental privatisation of the system,

superannuation has turned into Australians’ principal vehicle for retirement savings. Those who

have insufficient superannuation balances will be covered by the means-test (asset-tested and income-

tested) age pension. However, the public age pension (with the maximum level set at 25% of average

weekly earnings) is intended to be a bare minimum or a ‘safety net’ to supplement superannuation,

not to be an alternative. Bad decisions on superannuation arrangements can detrimentally affect

individuals’ long-term financial well-being, which, in turn, could mean a greater burden on taxpayers

to support retired individuals.

The cornerstone of the Australian superannuation system is the compulsory superannuation

contribution system, known as the SG, where employers are mandated to make superannuation

contributions on behalf of their employees at a proportion of their salaries and wages. The

employer’s mandatory SG payment is directed to employees’ nominated superannuation funds. To

ensure adequate retirement savings for the ageing population, the SG levy was gradually increased

in 2009 to 9.5% of an employee’s compensation from 3% at its inception in 1992. In 2021, the SG

rate is scheduled to increase by 0.5% per year until it reaches 12% on 1 July 2025,4,5 In addition, tax

incentives and government co-contributions are also available to encourage voluntary

3 Individuals can make an extra voluntary salary sacrifice and member top-up contributions to their superannuation accounts. Low income earners are encouraged to contribute through the government’s co-contribution scheme. 4 The compulsory SG is comparable to an auto-enrolment 401(K) plan in the US. 5 Source: https://www.ato.gov.au/rates/key-superannuation-rates-and thresholds/?anchor=Superguaranteepercentage#Superguaranteepercentage

14

superannuation contributions.6 When the SG was introduced in 1992, the superannuation system

held just under $150 billion in assets. Superannuation assets have then increased rapidly over the

past few decades, reaching $2.05 trillion as of June 2015, which is approximately the size of

Australia’s GDP (APRA 2016). By 2032, according to current projections, superannuation assets

are expected to reach $7.7 trillion. 7 7This superannuation system has turned Australia into a

‘shareholder society’, where Australians have more money invested in managed funds per capita

than individuals in any other economy.8

Australian superannuation funds are grouped into four categories, and the features of the

funds differ in each category. Retail funds are open to the public to join and are operated

commercially by financial institutions and fund managers. Profits are transferred to and retained by

the funds; hence, these funds are operated on a ‘for-profit’ basis. Industry funds are operated by

employer associations or unions, and they often nominated as default funds for employers who do

not operate their own fund. These funds were originally established for employees working in a

particular industry. Many industry funds are now public-offer funds, and they can offer membership

to the general public. Corporate funds are sponsored by a single private sector employer or a group

of similar employers for their employees. Public sector funds are created for employees of

governments or statutory authorities. These last three types of funds are operated on a ‘not-for-

profit’ or ‘profit-for-member’ basis.9

Superannuation in Australia is generally provided through a trust structure in which trustees

hold funds on behalf of members. Trustees owe members statutory fiduciary duties under

the Superannuation Industry (Supervision) Act 1993 (Cth). Superannuation funds are governed by

this Act, its regulations, trust deeds and governing rules. A registrable superannuation entity (RSE)

is a regulated superannuation fund or an approved deposit fund or a pooled superannuation trust but

does not include a self-managed superannuation fund. Trustees of registrable superannuation entities

(RSEs) must hold a Registrable Superannuation Entity licence (RSE licence) issued by APRA to

operate as public-offer superannuation entities.

6 Superannuation is the most tax-effective vehicle for retirement saving. Contributions, earnings and capital gains are concessionally taxed at 15% (a 10% capital gains tax applies to assets held for more than 12 months), instead of the investor’s marginal tax rate. More importantly, under reforms introduced in July 2007, superannuation benefits withdrawn after age 60 are tax free. 7 Estimate by the research firm Rainmaker assuming a growth rate of 9% per year. 8 Australia’s superannuation funds have experienced explosive growth at 11.7% per annum over the 10 years to 31 December 2014.This growth rate is among the fastest of 16 major pension fund industries studied by Towers Watson. Sources: Towers Watson, 2015 Global Pension Asset Study, February 2015. https://www.towerswatson.com/en/Insights/IC-Types/Survey-Research- Results/2015/02/Global-Pensions-Asset-Study-2015. 9 A fifth category of funds includes self-managed superannuation funds, which are generally established by a small number of individuals (up to 4 members) with larger amounts of superannuation assets or by family groups. These funds are subject to different regulations and their data are not publicly available. As such, they are beyond the scope of this study.

15

2.2 ‘Choice of Fund’ Policy

Along with the tremendous growth in the superannuation industry, the government’s policy

framework has undergone major structural transformations. When the SG was first enacted,

members of employer-based superannuation schemes have few opportunities in choosing the fund

family for their workplace contributions. Most employees were compelled to join the fund chosen

by their employer or established in the relevant industrial agreement.10 Moreover, most funds did

not offer members a menu of investment strategies.

Since 1 July 2005, with the implementation of the Superannuation Legislation Amendment

(Choice of Superannuation Fund) Act 2004 (Commonwealth of Australia 2005) many Australian

employees are able to nominate a fund family for their SG contribution.11,12 Investors can now choose

from hundreds of fund families. Concurrently, regulatory changes improved the portability of funds,

allowing balances to be rolled over between fund families more easily. Over time, product diversity

within fund families also increased to promote more choices and flexibility for members.

Funds have incentives to maximise assets and increase market share because fund managers

are usually compensated at a percentage of assets under management (Clark et al. 2013; Gaspar,

Massa & Matos 2006; Khorana & Servaes 1999). As the asset base grows, funds also enjoy the

benefits of economies of scale. In the superannuation context, this incentive is further amplified by

the persistence in superannuation fund flows, as investors regularly contribute to funds and

cannot withdraw their investment until the preservation age, early death or disability. Even at a time

when capital was scarce worldwide (i.e., the 2008-2009 Global Financial Crisis), the

superannuation industry continued to thrive.13

Despite the government’s reform to encourage active investment choices and despite the

proliferation of fund families and products, investor inertia appears to remain strong. That is,

superannuation members have a strong propensity to stay with the default fund family and fund. The

proportion of superannuation account balances transferred before and after the 2005 Choice

legislation was relatively unchanged at approximately 5-7% of total superannuation assets per annum

10 Before the ‘Choice of fund’ legislation, a degree of fund choice did exist for some employers. For example, in Western Australia, choice of fund was legislated for employees under state-based awards through the Industrial Relations (Superannuation) Regulations 1997 Act. 11For example, investors can change fund families if their current fund family is not available with a new employer, consolidate superannuation accounts to cut costs and paperwork, or change to a lower-fee, better performing, better service superannuation fund family. 12 The Choice of Fund Policy applies to all employees who are eligible for SG contributions, with the exception of those whose superannuation is paid under state awards or state industrial agreements, as well as members of certain public sector funds and defined benefit plans. 13 There was still approximately $2.2 billion flowing into superannuation funds weekly during the Global Financial Crisis. Source: DUNN, J. 2012. Footy, meat pies and super... Wealth.

16

(Fry, Heaney & McKeown 2007; Gerrans 2012). Nevertheless, the switch rate mentioned above still

involves huge sums of money and a multi-billion-dollar business. Given the importance of sound

superannuation investment choices to individuals and given the government’s apparent intention to

promote active choices, it is important to understand investors’ behaviour and the factors that are

more likely to drive active choices. Furthermore, the examination of member choice is motivated by

recent survey evidence, which suggests there is less default behaviour than expected. In surveying

1,031 members on their default behaviour, Butt, Adam et al. (2015) find that more members describe

themselves as active choosers than the authors expected. Only 36% of their sample stayed with

both the default option at both fund family and fund levels, meaning that 64% made at least one

active choice. In addition, about one-quarter of members in the default family and 9% of investors

in default funds deliberately choose the default option. Hence, the proportion of completely passive

investors with default funds in their sample is probably below one-third. Similarly, survey results

from Clark et al. (2013) suggest that although the tendency to choose the default option is influential,

it is not overriding.

In 2011, the Australian government initiated a low-cost and simple superannuation fund

product called MySuper for employers to choose as their default fund. However, even a well-chosen

default option may be undesirable if members have heterogeneous needs (Carroll et al., 2005). For

example, a young and cash-strapped member and an older member who needs to quickly build a

retirement nest egg will differ in their interests in superannuation arrangements and risk

preferences. The purpose of libertarian paternalistic retirement saving policies such as MySuper

should be to mitigate costs for passive investors without encroaching free-choice in the market

(Benartzi & Thaler 2001; Choi et al. 2002). Carlin, Gervais and Manso (2013) argue that the loss

of social learning from people relying on default options and not acquiring and sharing useful

information can be damaging to overall society. Active choices are socially optimal when

consumers have highly heterogeneous savings preferences and a strong propensity to procrastinate

(Carroll et al. 2009). These reasons further motivate this thesis’s investigation of the determinants

of active investment choices made by superannuation members. Understanding the factors that

affect member choice is also of significant interest to fund managers as the findings can help

them formulate strategies to attract/retain investments.

2.3 Superannuation Product Disclosure Statements and the Short-form Product Disclosure

Statement Regime

All financial services providers are required by the Financial Service Reform (FSR) Act

17

2001 and Corporations Act 2001 to provide comprehensive information about their available

products through a PDS. Financial advisors are also required to source their information from PDS

in their Statement of Advice when recommending customers to switch funds (ASIC, 2013). The

PDS is a direct source of basic information for all parties interested in superannuation funds,

including investors, financial advisors, research houses, rating agents and regulators and it is still

supposed to be the main channel of communication between superannuation fund trustees and

members. The objective of a PDS is to help members compare superannuation products and make

informed choices. While legislated disclosure requirements are primarily aimed at new members,

the PDS also provides all members with information about the fund’s investment options on an

ongoing basis that enables members to periodically review their investment options and decide

whether to switch their investment strategies.

An unintended consequence of FSR is the labyrinthine documentation that financial

providers produce to protect themselves against the legal risks associated with inadequate

disclosure. The Parliamentary Secretary to the Treasurer, Christopher Pearce, was quoted in 2005 as

acknowledging that ‘the end result of FSR has been that the disclosure documents themselves are

designed more to manage the liability of the service provider, rather than to inform the consumer’,

and consumers generally disregard these PDSs (Fear & Pace 2008). In numerous surveys,

consumers complained about the lack of accessible and understandable information – PDSs have

too much jargon, too many options, too much documentation, and too many different and non-

standard ways of describing products, fees, charges, and so on (Fear & Pace 2008; Fry, Heaney &

McKeown 2007).

In 2010, the Panel of the government-launched Super System Review concluded that the

superannuation system lacks transparency, comparability and accountability with respects to costs,

fees and investment returns.14 The ASIC conducted a review of 200 superannuation PDSs in 2010.

This review outlined the ASIC’s concerns about superannuation PDS disclosure and identified ways

to potentially improve superannuation PDSs. Specifically, the ASIC recommends that superannuation

entities should issue timely PDSs and improve both the quality of the information provided and the

ease of understanding (ASIC, 2013). The comments in the ASIC’s report are also relevant to the shorter

and simpler PDS regime, which was implemented toward the end of the ASIC’s review.15

14 The review is more commonly known as the Cooper Review, after its chairman, Jeremy Cooper, the former deputy chairman of the ASIC. 15 Mandated by the Corporations Amendment Regulations 2010 (No. 5), the shorter PDS regime is designed to make superannuation PDSs shorter and simpler to improve comparability among products. The new regime commenced fully on 22 June 2012, although issuers of new products have been required to comply with the regime since 22 June 2011 and other issuers have been able to voluntarily opt-in.

18

2.4 Managed Funds and Superannuation Funds

Although there is an extensive literature on managed funds, the results are unlikely to be

generalisable to superannuation funds because superannuation funds differ from managed funds in

fundamental ways in terms of investor attributes and their decision-making process. First, contribution

to superannuation is compulsory under the SG for almost all working Australians, whereas investment

in managed funds is at the investors’ discretion. Hence, managed fund investors voluntarily make a

decision to invest in managed funds, and their investment decisions are thus more likely to be informed

decisions based on information about the funds. These investors are also expected to be more actively

engaged in monitoring and managing their investments than superannuation investors, for whom the

investment process is largely passive. As such, managed funds investors are unlikely to be

representative of the superannuation investor population.

Australian employees are mandated by law to have a certain percentage of their compensation

deducted by employers and invested in a superannuation fund, often in a default option of a default fund

family chosen by their employer with a prescribed investment horizon. As members cannot access

benefits until retirement, they tend to be less attentive to these locked-in savings. In general, employees

must have contributions directed to only one fund family, while managed fund investors can easily

invest in multiple fund families concurrently. Thus, superannuation investors tend to ‘set and forget’,

such that inertia and disengagement in superannuation investment decisions are major concerns that do

not apply to managed funds investors. The compulsory nature of the SG system, the limits on benefits

withdrawal, the wider coverage of individual and unsophisticated investors and investors’ stronger

tendency to choose the default option suggest that findings from the managed funds literature cannot be

directly applied to the superannuation context.

2.5 U.S. Pensions and Australian Superannuation

Superannuation in Australia is comparable to US 401(k) plans. The primary difference is that

superannuation is compulsory in Australia, with a minimum level of contribution set by the

government. There is an established shift towards greater individual responsibility for investment

decisions with the growing dominance of Defined Contribution plans.16 Over time, Defined Benefit

Superannuation plans have been largely replaced by Defined Contribution plans. Hence,

16 A defined benefit plan provides a specified payment amount in retirement, which can be based on the employee's salary, years of service, discount rate or a number of other factors. Employees have little control over the funds before retirement. A defined contribution plan allows employees and employers to contribute and invest funds over time to save for retirement. In a defined contribution plan, the employer no longer has any obligation on the account's performance after the funds are deposited, which means employees bear the investment risks to make investment decisions and grow the funds adequate for retirement.

19

superannuation investors face increasing responsibilities in making their own investment decisions,

and bad decisions can detrimentally affect individuals’ long-term financial well-being. In 2015,

Australia has 87% of superannuation assets in defined contribution plans, whereas the US and UK

have 60% and 32% respectively. In response to this global trend, there is increasing investment choice

offered by 401(k) plans in the US and continued debate on the introduction of individual accounts as

part of possible Social Security reforms. In addition to the dominance of Defined Contribution plans,

Australia is also among the top-performing countries in retirement saving by multiple rating. 17

Therefore, the examination of superannuation member choice in Australia has international

implications for the understanding of retirement saving behaviour and the development of

retirement saving systems.

2.6 Fund Families and Individual Funds

Superannuation funds are not stand-alone entities, as most of them are members of fund

families. A fund family is a collection of all funds managed by the same sponsoring management

company. The family structure offers economies of scale, and many important strategic decisions

are made at the fund family level, e.g., marketing and product mix decisions.

Although fund families offer a large range of products that can significantly differ in returns,

fund returns are more closely correlated within rather than between fund families (Elton et al.,

2007). This correlation is attributed to overlapping stock holdings and similar exposure to economic

sectors or industries. Hortacsu and Syverson (2004) find a significant positive correlation between

flows into the S&P 500 fund and the performance of the non-S&P 500 funds in the family, which is

consistent with their argument that a performance spillover effect exists in the family setting. It is

also widely documented that the superior performance for individual funds has star spillover effects

(Nanda, Wang & Zheng 2004; Warner & Wu 2011),whereby other funds in the same family also

enjoy increased fund flows. Hence, the family structure under which constituent funds share the

same research analysis and prescribed investment style justifies this study’s focus on the fund

family as the unit of analysis.

Prior studies also find that major decisions, such as marketing and PDS disclosure decisions,

17 Australia’s retirement income system was ranked 3rd among 25 countries by the Mercer Global Pension Index, and 6th in Finances in Retirement by the Nataxis Retirement Index in 2015. See http://www.mercer.com/content/dam/mercer/attachments/global/Retirement/Melbourne-Mercer-Global-Pension-Index- 2015/Report.pdf and http://durableportfolios.com/docs/398/951/2016%20Natixis%20Global%20Retirement%20Index%20Report_Final.pdf.

20

are made at the fund family level. Previous research suggests that mutual fund families use family-

level strategies to maximise the benefits for the overall family. Gaspar, Massa and Matos (2006)

find that fund management companies actively pursue a direct family strategy of cross-subsidising

the performance of high-value funds (i.e., those more likely to generate fee income or extra

investment inflows) at the expense of low-value funds. In addition, assets exhibit stickiness within

the fund family, i.e., investors are reluctant to withdraw funds from the fund family. Accordingly,

fund families expand fund offerings as a way of retaining assets. Massa (2003) and Khorana and

Servaes (2004) argue that fund families can employ category proliferation as a way to of countering

competition and increasing market coverage. Product differentiation is also a tool for exploiting star

spillover effects, since fund families are well rewarded in terms of new cash inflows for producing

stellar funds. Khorana and Servaes (1999) find hat families with star funds tend to expand product

lines. Mamaysky and Spiegel (2002) develop a model of mutual funds in which fund families do

not specialise; rather, the optimal strategy is for families to offer their products in more diversified

fund categories. As a matter of random luck, having more funds in a fund family will increase the

likelihood of having at least some stellar funds in the family (Nanda, Wang & Zheng 2004). Previous

studies have also examined fund families’ product policies (Khorana & Servaes 2004; Mamaysky &

Spiegel 2002; Siggelkow 2003) and advertising decisions (Gallaher, S, Kaniel, R & Starks, LT

2006). Since marketing and disclosure decisions are coordinated at the family level, it is more

appropriate for thesis to use the superannuation fund family as the unit of analysis.18

18 The discussion on PDS disclosure decisions at the fund family level is provided in chapter 3.

21

Chapter 3 - Literature Review

3.1 Behaviour of Defined Contribution Plan Members

Theoretical models in prior research, such as those in the managed fund literature, typically

assume a significant degree of investor sophistication and learning ability (Berk & Green 2004; Dangl,

Wu & Zechner 2006; Huang, J, Wei & Yan 2007; Lynch & Musto 2003). However, in realistic

environments, an overwhelming amount of empirical evidence suggests that average pension

participants display low levels of financial literacy and investment savviness in their financial

decisions. Several studies have highlighted the behavioural biases and financial literacy constraints

that hinder decision making, showing, for example, that pension participants use naive asset allocation

strategies and exhibit inertia in rebalancing (Agnew, Balduzzi & Sunden 2003; Benartzi & Thaler

2001; Huberman & Jiang 2006) and they are are subject to framing and default effects in investment

options (Benartzi & Thaler 2001; Karlsson & Nordén 2007). The observed investor apathy toward

superannuation investments likely arises because of the nature of the clientele of superannuation funds:

largely retail investors with limited financial sophistication and cognitive biases in decision making.

However, two streams of studies support the notion that changes in retirement savings behaviour

can be motivated simply by the power of suggestion. Madrian and Shea (2001), Choi et al. (2002) and

(Choi et al. 2005) show that automatic enrolment in a 401(k) plan dramatically increases participation

and contribution rates. The findings of these studies indicate that that many employees interpret the

default option as investment advice on the part of the company. Duflo and Saez (2003) highlight the

role of social interactions on retirement savings decisions, and their evidence suggests that peer effects

strongly influence the decision to participate in retirement plans.

3.2 Optimal Dynamic Portfolio Allocation Theory

Research in finance, by analogy with other parts of economics, can be normative or positive.

Normative research prescribes what investors should do to maximize their welfare, whereas positive

describes the choices investors currently make. The discrepancies between observed and ideal

behaviour, or investment mistakes, are often central to the field of household finance (Campbell 2006).

The seminal work of (Merton 1971, 1973) introduces a conceptual framework for life cycle

portfolio allocation given time-varying investment opportunities. Merton emphasizes that long-term

investors must consider not only shocks to their wealth, but also shocks to any state variable that predicts

expected returns on wealth. It was not until the 1990s that empirical versions of the Merton model

emerged. One branch of the literature concentrates on shocks to the real interest rate, assuming that risk

22

premia are constant through time (Campbell & Viceira 1998; Wachter 2003). A second branch of the

literature focuses on the equity premium, assuming that it follows an exogenous time-series process

such as an AR(1) (Campbell & Viceira 1999; Kim & Omberg 1996). More recent work allows for

general multivariate processes that drive both interest rates and risk premia on multiple assets (Brennan,

MJ, Schwartz & Lagnado 1997; Campbell 2006; Lynch & Musto 2003).

Researchers often use the life cycle theory to generate the distribution of retirement outcomes,

which is usually evaluated through simulations (Bodie, Merton & Samuelson 1992; Farhi & Panageas

2007; Gomes, Kotlikoff & Viceira 2008; Kingston & Fisher 2014). This body of research recognizes

that optimal dynamic portfolio allocation models are highly individualized and identifies various factors

for the optimal design of retirement savings products, including asset allocation strategy over the life

cycle; return assumptions and uncertainty over returns; human capital and contribution levels; flexibility

around the work/leisure/retirement choice, as well as around contributions and withdrawals; the

availability of social security benefits; existence of other assets; life risks such as health and longevity;

bequests; and risk aversion.

Given the complexity of their financial planning problem and the often confusing retirement

saving products that are offered to them, it is not surprising that some investors make investment

mistakes. Potential remedies can be provided to reduce the incidence and welfare costs of these

mistakes, such as consumer regulation, disclosure rules and financial education. Therefore, the work on

these topics offers a powerful practical rationale for the empirical study of superannuation investment

behaviours (Campbell 2006), and, in particular, the study of superannuation choice and the effect of

disclosure.

3.3 Flow-performance Relation

A convex flow-performance relation at the individual fund level is a well-documented

phenomenon in the managed funds literature, where investors flock disproportionately to recent

winners but do not punish poorly performing funds proportionately by withdrawing their

investments (Chevalier & Ellison 1997; Goetzmann & Peles 1997; Gruber 1996; Huang, J, Wei & Yan

2007; Lynch & Musto 2003; Sirri & Tufano 1998). This asymmetric flow-performance relation is

commonly found among retail investors, as they infer managers’ skill or future performance from

past performance, and chase after superior performance in a simplistic way (Bailey, Kumar & Ng

2011). Further, Navone (2012) demonstrates that for their allocation decisions, inexperienced retail

investors heavily rely on a fund’s past performance when making resource allocation decisions,

since other information (such as that on expenses) is more difficult to obtain. James and Karceski

23

(2006) find that retail and institutional investors differ in their reactions to performance. They

find evidence that institutional investors do not chase returns in the same way as retail investors,

since institutional funds employ more sophisticated performance measures than retail investors.

Huang, JC, Wei and Yan (2012) use the dampening effects of performance volatility and fund age

on flow-performance sensitivity to infer the relative level of sophistication among mutual fund

investors. The volatility of performance introduces noise into investors’ learning process, whereas

longer track records enhance informativeness. Consistent with the idea suggested by (Bailey, Kumar

& Ng 2011), Huang, JC, Wei and Yan (2012) hypothesise that the dampening effects on flow-

performance sensitivity are stronger for funds with more sophisticated investors. In comparing load

versus no-load funds, institutional versus retail funds, and star versus non-star funds, they find

evidence supporting their investor learning hypothesis. However, in the context of superannuation

funds, the average member tends to be a naïve investor without superior investment knowledge;

thus, heightened flow-performance sensitivity is expected for superior performance.

In contrast, some recent studies show that for pension funds, the fund flow-performance

relation tends to be flat owing to clientele differences. Sialm, Starks and Zhang (2015) find that

participants in defined contribution plan neither invest more in high performance funds nor

significantly pull out of low performance funds. They show that the flow-performance sensitivity

documented in previous studies is primarily driven by the plan trustees’ actions in adding and

deleting funds on the plan menu. Using a dataset covering savings in Sweden’s Premium Pension

System, Dahlquist and Martinez (2015) find that pension investors do not seem to react to past fund

performance owning to inattention and inertia. Given the difference in prior research results for

different settings, this thesis examines the relation between performance and fund flows in the

Australian superannuation setting in order to gain an understanding of the impact of fund family

performance on Australian superannuation investors’ investment decisions.

3.4 Information Costs, Marketing and Fund Flows

This thesis also examines marketing efforts as a method that superannuation fund families

can use to reduce superannuation members’ information search costs and thus attract fund flows.

Studies on mutual funds have often recognised the information barrier that investors encounter in

allocating their investment among different funds. Sirri and Tufano (1998) and Huang, J, Wei and

Yan (2007) argue that fund flow should be related to mutual fund investors’ information search

costs. Information search and processing costs affect investors’ ability to acquire and understand

24

information, and they partially determine the extent to which investors are informed. Hortaçsu and

Syverson (2004) examine the retail S&P 500 index funds segment, where all funds are characterised

by nearly homogeneous return patterns, and they find that this sector exhibits large share price

dispersion (i.e., these companies are priced very differently despite their homogeneous return

patterns). To address this puzzle, they study the possible roles of informational (or search) frictions

that deter investors from finding the fund that offer the highest utility, and they suggest that the

influx of novice mutual fund investors during their sample period underlies the observed shift in

sector assets to more expensive funds. It is worth noting that the authors assume that these novice

investors are typically those with high information costs, which links back to the discussion in the

previous section on unsophisticated investors.

In the presence of investor demand for information and substantial information costs,

mechanisms or conditions that reduce information search and processing costs can have a material

effect on investors fund choices. Several papers have inferred that advertising efforts reduce investor

search costs and facilitate fund flows (Gallaher, S, Kaniel, R & Starks, LT 2006; Jain & Wu 2000;

Karlsson & Nordén 2007). Some mutual fund studies also have examined the relation between

individual fund flows and proxies for advertising efforts. For example, Sirri and Tufano (1998) use

total fees charged as a proxy for marketing and distribution expenditures. They find no relation

between the flow-performance relation and this proxy, except in the case in which they separate

mutual funds into those with high fees and those with low fees. In this case, they find that funds

with higher fees, which the authors assume are funds with greater marketing efforts, have greater

flow-performance sensitivity. However, because they are forced to employ a coarse proxy for

marketing efforts, they cannot ensure that their results do not result from confounding factors; for

instance, funds with higher service levels (associated with higher fees) may attract greater flows.

Jain and Wu (2000) use a dummy variable approach to compare the flows of individual mutual funds

with advertisements in one of two magazines during a month with the flows of funds without

advertisements in these two magazines. Over their sample period of July 1994 to June 1996, they

find that advertised funds have higher net inflows, after they control for prior performance, lag

flows, and fund size. Cronqvist and Nilsson (2005) examines a number of issues concerning

the advertising of Swedish 401(k) type funds, including what funds advertise, which types of

advertising affect investors’ allocation decisions, and whether fund advertising is a signal of

future performance. These studies have focused on the role of advertising in individual funds, but it

is important to keep in mind that the advertising expenditure decision is a fund family decision,

not an individual fund decision. Thus, this thesis tests the hypothesis that advertising affect flows at

25

the fund family level, by using more precise measures of marketing expense and fund flows.

3.5 Fund Disclosure and Fund Flows

As discussed in the previous section, higher information-processing costs impede investors’

ability to make informed decisions. Disclosure by superannuation funds with greater readability is

easier to interpret and process and therefore saves investors time and effort in identifying and

understanding relevant information. Prior studies in the accounting literature show that improved

readability positively affects the reactions of small investors. Miller (2010) shows that 10-K filings

with lower readability are associated with lower overall trading volumes, and this relationship

appears arise from a reduction in trading activities among small investors. Rennekamp (2012) uses

a controlled experiment and finds that, when constant length and information content are held

constant, disclosures with greater readability lead to stronger reactions from small investors; thus,

changes in valuation judgements are more positive (negative) when press releases convey good

(bad) news.

Prior literature has mostly examined information disclosures by companies. However,

information asymmetry and agency costs also exist between superannuation fund managers and

members.

Previous research within the field of mutual funds helps provide some insights regarding

how shorter and easier-to-understand information may influence the responses of investors. In

examining the impact of certain types of summary information on individuals’ attitudes towards

mutual funds, Kozup, Howlett and Pagano (2008) find that summary information increases subjects’

sensitivity to prior fund performance. Beshears et al. (2009) use an experiment to estimate the effect

of the SEC’s Summary prospectuses, which simplifies mutual fund disclosure, on participants’

portfolio allocation choices. Contrary to Kozup, Howlett and Pagano (2008)’s results, they find

that returns of investment portfolios are statistically indistinguishable across the participants

receiving the statutory prospectus and participants receiving the summary prospectus. However,

the participants receiving the summary prospectus spend less time on their investment decision

without adverse effect on portfolio quality. Hence, simpler disclosure does not appear to be a

useful channel for making mutual fund investors more informed. However, it makes the decision-

making process easier and less time-consuming.

Based on an Australian sample, Cheah et al. (2015) analyse focus group discussions regarding

long-run (retirement) financial decisions. A key implication of these focus groups is that simpler

26

and more meaningful forms of disclosure may improve the attraction of and confidence in the

superannuation system (Cheah et al. 2015). Using a laboratory experiment, Bateman et al. (2015)

study how university students and staff participants choose retirement savings investment options

using five information items in the new short-form PDS prescribed by regulators. The results indicate

that participants were most responsive to asset allocation information, and preferred options with

more asset classes and with equally size weights. Bateman et al. (2016) use a discrete choice

experiment to test how the alternative presentations of risk affect respondents’ decision making.

However, the impact of superannuation fund disclosure on fund flows remains unexplored.

An interesting question is whether the quality of PDS disclosure affects investors’ resource

allocation decisions among superannuation funds. Fund families differentiate themselves in terms of

non-performance related characteristics so that they do not need to compete solely on the basis of

performance (Massa 2003). PDSs provide a relatively easy and inexpensive way to communicate

such non-performance-related fund features to attract investments. Concurrently, the PDS is an

important direct source of information for investors, and the quality of such disclosure can

substantially affect investors’ information costs and level of understanding of the fund family and

its products. Therefore, this study examines both (1) the quality of Australian superannuation PDS

disclosure and (2) the relation between PDS disclosure quality and fund family flows.

27

Chapter 4 – Determinants of Member Choice

4.1 Introduction

Given the economic and social significance of superannuation to individual Australians and

Australian society, it has become ever more important to gain a better understanding of whether and

how these mostly novice, small, retail superannuation members make their investment choices.

Consequently, this chapter examines the determinants of member-nominated transfers across

superannuation fund families in order to elucidate the role of fund families’ performance and

marketing efforts in members’ decisions to switch fund families.

To explore the determinants of superannuation member choice, this chapter first examines

the effect of fund family performance, since it is one of the most salient features of funds that

investors can observe. An extensive line of the managed funds literature has found asymmetric

return-chasing behaviour among investors. However, findings from managed funds research may

not directly apply to superannuation funds considering the differences in the investment behaviour

and financial sophistication of superannuation members.

Research on managed funds has provided mechanisms or conditions that reduce investors’

information search and processing costs facilitate fund flows, as investors tend to purchase funds

that are easier or less costly to identify (Huang, J, Wei & Yan 2007; Jain & Wu 2000; James &

Karceski 2006; Sirri & Tufano 1998). One mechanism to reduce information costs for investors

is enhanced fund visibility through marketing, as marketing provides investors with information

regarding the fund’s existence and characteristics. Similarly, the attention (or familiarity)

hypothesis implies that marketing can increase investors’ awareness of a fund. With almost full

coverage of the workforce, the cohort of superannuation members is much broader and is likely to

have a lower level of financial sophistication on average, relative to managed fund investors. These

unsophisticated superannuation members, on average, incur higher information costs when making

investment decisions than their managed fund counterparts. This chapter thus examines whether and

how marketing affects superannuation member choices.

To determine how investors respond to the strategies pursued by funds, fund flow is widely

used in the managed fund literature. This is because fund flow reflects the choices made by

investors, that is, the revealed preferences of investors. Previous studies approximate fund flows by

using the fractional net flow specification. However, net flow is less informative in the

superannuation setting because a large portion of the new inflows is mandated by the SG, and

because outflows from a superannuation fund largely consist of benefit payments to members in

28

retirement. Hence, net flow in superannuation is not a precise proxy for member choice. Namely,

every flow in managed funds represents a decision made by investors; however, because of the

superannuation contribution and withdrawal rules, only member-initiated fund transfers capture

superannuation members’ active investment decisions. Therefore, this thesis uses the APRA-

reported rollover from one superannuation entity to another superannuation entity after adjusting for

flows from winding-up funds.19 The derived member-initiated transfers provide a more precise

reflection of superannuation members’ active investment choice.

A potential complication facing any fund analysis is the fact that virtually all funds are

affiliated with fund families. Superannuation members have two possible levels of choices: the

choice of fund family (or plan) to receive contributions and the choice of an individual fund

(investment strategy or option) within a particular fund family. While the effect of fund family

characteristics on individual fund flows has been recognised in prior studies, few studies have

focused on family flow. Although superannuation entities are certainly interested in the level of

fund flow to each of their individual funds, they view these funds as a series of products, with the

central interest being the aggregate flow to the entire family of funds (Bhattacharya, Lee & Pool

2013; Chevalier & Ellison 1997). Employees generally are allowed only one fund family to direct

their SG contributions to. From investors’ perspective, the investment choice usually starts with

fund family brand recognition, which precedes fund product selection. Moreover, the relative ease

of switching funds within a family (in terms of fees and search costs) suggests that the choice of

fund family is an important decision before a specific fund is selected (Massa 2003). In addition,

the active choices at the family level rather than at the individual fund level that drive the

competition and improve the overall efficiency of the superannuation system (Commonwealth of

Australia 2012). Considering the reasons above, this study takes a broader perspective and uses the

superannuation fund family as the unit of analysis for both member choices and their determinants.

Multivariate analysis is performed on two main potential determinants of superannuation

member choices: fund family performance and marketing efforts. The results, first, show that

performance is not a key determinant of superannuation family inward flows, though poor

performance is significantly associated with outflows, suggesting that poor performers need to

improve returns in order to avoid outward flows. Second, the results indicate that marketing efforts

attract superannuation investors into superannuation families, but this effect is limited to fund

families with high marketing expense ratios. In addition, this result is driven by retail funds. In

19 This measure captures the rollovers of funds that already exist in the superannuation system, or members’ consolidation decisions.

29

other words, extensive marketing does not appear to be a useful strategy for industry funds to attract

investors. The results are robust to the use of fixed effects models and the adoption of a dynamic-

panel Generalised Method of Moments (GMM) estimator (Arellano and Bond, 1991). These results

contribute to existing literature in a number of ways. First, they provide insight into the

superannuation investment decisions of average retail investors. A large and growing segment

of the superannuation fund market is targeted towards retail clients. Winning new business entails

capturing members (and their assets) from another superannuation fund family. Hence, evidence

on how individual investors hold and switch between superannuation fund families is important

for fund managers and regulators. This study also uses a more precise measure of investor

choice, namely, member-initiated fund transfers at the family level, whereas prior literature has used

the net flow fractional approximation. In addition, this study is consistent with the spirit of the

government’s ‘Choice’ policy (Commonwealth of Australia 2005) which encourages

superannuation fund members to exercise choice among fund families. The understanding of the

determinants of fund family flow precedes that of fund flow, and findings on the determinants of

flows at the family level can help superannuation fund families to better understand the members

that they serve and accordingly form strategic decisions. Finally, this study’s empirical analysis

using the universe of Australian regulated-superannuation funds advances typical studies on

retirement saving, which rely on small and specialised datasets using surveys or experimental

research designs.

4.2 Hypothesis Development

Mixed evidence of the flow-performance relationship from prior research on managed funds

has been discussed in Chapter 3. Performance is one of the most salient fund features that investors

can observe, and return-chasing behaviour is prevalent among unsophisticated investors.

Nonetheless, superannuation investors may not react to past performance as strongly as other

investors, given the long-term nature of superannuation investment and the inattention and inertia of

superannuation members. Whether family flows are associated with prior family performance in the

superannuation context remains unexplored. To address this question, this thesis examines the

nature of the performance-flow relation for superannuation fund families, as stated in the following

alternative form:

H1: There is a positive flow-performance relation in superannuation fund families.

The second potential determinant of superannuation members’ choice is fund families’

30

marketing efforts. Marketing is potentially an important mechanism for superannuation

funds/members owing to the members’ passivity/inertia; thus, marketing represents a way to force

brand recognition and provide information. The role of marketing is likely important considering

that superannuation members are unsophisticated and passive (i.e., they are less likely than managed

fund investors to actively conduct extensive research of superannuation funds in order to make

switching decisions). Repeated messages provided by advertisements are likely to attract their

attention and make their information search/processing decisions easier. The industry

superannuation families in 2005 began their now-famous ‘compare the pair’ advertising campaign

in response to the choice of fund regime.20 Their advertising campaign reached such a level in 2014

that it prompted the Financial Planning Association (FPA) and Association of Financial Advisers

(AFA) to question whether the industry funds were really acting in the best interests of their

members. 21 The Industry SuperFunds justified their marketing spending by stating that

advertisements have demonstrated to be an effective way to grow and retain members and to

compete with their retail superannuation counterparts. Thus, this thesis examines whether a positive

relation exists between marketing and superannuation fund flows. To the extent that a mandatory

superannuation system will increase the coverage of those with lower financial literacy, a stronger

relation between marketing and fund flows for superannuation funds may be expected to exist.

H2: Marketing is positively associated with fund family flows.

4.3 Data and Methodology

4.3.1 Data Source

The APRA supervises regulated superannuation funds in Australia, Approved Deposit Funds and

Pooled Superannuation Trusts, all of which are regulated under the Superannuation Industry

(Supervision) Act 1993.22 The first data source is the APRA Superannuation Fund Level Profiles and

Financial Performance statistical publications, which provide data on investment rollovers and other

superannuation fund family characteristics, including expenses, rate of return, number of investment

options, member age segmentation, personal/employer contributions and proportion of default assets.23

20 An example of the advertisement is where two men sit on a park bench discussing their superannuation. Man A has a retail superannuation fund, and Man B has an industry superannuation fund. The industry super fund wins out. 21 See http://www.moneymanagement.com.au/expert-analysis/editorial/industry-super-funds-advertising-onslaught-incomparable 22 Self-Managed Superannuation Funds (SMSF) are supervised by the Australian Taxation Office. Member-nominated outflows can be directed to a SMSF, which is not reportable to the APRA. 23 Some items in the public publications are masked for privacy reasons. The full dataset is offered by the APRA to perform the analysis in this study.

31

APRA collects these data from mandatory periodic Superannuation Reporting Forms (SRFs) submitted

by APRA-regulated superannuation fund families. The sample period spans from 2005 to 2014, when

the ‘Choice’ policy was in place. The terms ‘fund’ and ‘product’ used in the APRA’s publications

correspond to ‘fund family’ and ‘fund’ in common usage. To be consistent with the prior literature,

these terms are used throughout this thesis.

Considering the significant consolidation in the superannuation industry over the past decade,

there is a need to adjust for successor transfers, based on the information reported in the proprietary

SRF 250.0 Superannuation Entity Profile. Section 4 of this form is used to identify the name of the

winding up entity, the date of wind up, and the successor entity. Other proprietary data provided by

APRA include the number of new members and the break-down of expense items from a series of

reporting forms.

4.3.2 Sample Selection Process

The sample selection process is reported in Table 4.1. The initial sample comprises 4,166

superannuation fund family years for the period from 2005 to 2014. Corporate, industry and retail

families together account for almost 95% of the total observations. The sample selection process

begins with the elimination non-public offer families. Public offer superannuation fund families are

open to the public to join, and they promote competition in the industry by providing alternatives

for members choosing superannuation fund families. Non-public offer funds are excluded because

they are not free for all investors to join; thus, they are not valid investment options for all

superannuation investors. This step results in the exclusion of 94% of corporate families, 40% of

industry families, and 80% of public sector families from the sample. The majority of retail families

are public offer families, as only 335 of 1,963 retail families are non-public offer families.

Exclusively Defined Benefit (DB) families are deleted, since DB members are not responsible for

managing their own investments and DB plans are not as portable as Defined Contribution (DC)

plans. Most superannuation fund families have a year-end of 30 June. A number of superannuation

families with year-ends other than 30 June are removed in order to provide a consistent basis for

comparison. Observations with missing or erroneous values (e.g. negative inward rollover) are

deleted. The final sample comprises 1,802 fund family years for the period from 2005 to 2014. Over

half (67.5%) of these families are retail superannuation families. The second largest type is industry

superannuation families, which accounts for 19.6% of the sample.

32

Table 4.1 Sample Selection Process This table reports the sample selection process. TNA represents the total net assets under management. MktExp denotes the marketing expense. The

number of investment options is denoted as InvOpt. Return is the APRA-reported family return. The flow measures (net rollover, inward rollover and

outward rollover) are APRA-reported transfers among superannuation entities.

Sample Selection Criteria/Superannuation Fund Family Type Corporate Industry Public Sector Retail Retail-ERF Total Initial Sample 1386 597 220 1815 148 4166 Delete non-public offer families 1301 239 176 325 10 2051 Delete Defined Benefit families 6 0 0 49 0 55 Delete families with year-end other than 30 June 10 0 0 107 0 117 Delete families with missing net rollover 0 0 0 0 0 0 Delete families with negative or missing inward rollover 0 0 0 11 0 11 Delete families with negative or missing outward rollover 0 0 0 0 0 0 Delete families with missing return 3 0 0 19 0 22 Delete families with 0 or missing TNA 2 1 0 41 3 47 Delete families with negative or missing MktExp 0 2 0 5 0 7 Delete families with non-positive or missing InvOpt 0 0 0 0 0 0 Delete families with non-positive or missing Above50 9 1 0 41 3 54

Final Sample 55 354 44 1217 132 1802

4.3.3 Regression Model Specification

Determinants of superannuation investment rollovers are examined using the following

regression model. The regression variables are discussed in more detail in the next few sections, and

the variable definitions are provided in Table 4.1.

𝐶𝐶ℎ𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑖𝑖,𝑡𝑡 = 𝑜𝑜 + 𝑏𝑏1𝑜𝑜𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑃𝑃𝑜𝑜𝑜𝑜𝑃𝑃𝑖𝑖,𝑡𝑡−1 + 𝑏𝑏3𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡−1 + 𝑏𝑏4𝐼𝐼𝐿𝐿𝐼𝐼𝐼𝐼𝑀𝑀𝑀𝑀𝑖𝑖,𝑡𝑡 +

𝑏𝑏5𝐿𝐿𝑏𝑏𝑜𝑜𝐼𝐼𝑜𝑜50𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡

(1)

where 𝐶𝐶ℎ𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 is one of three fund flow measures (Net Rollover, Inward Rollover and

Outward Rollover) of APRA-reported transfers among superannuation entities. MktExp is the marketing

expense ratio as proxied by other operating expenses. For performance (𝑃𝑃𝑜𝑜𝑜𝑜𝑃𝑃), we use the APRA-

reported family return. 𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿 is the natural logarithm of total net asset under management. 𝐼𝐼𝐿𝐿𝐼𝐼𝐼𝐼𝑀𝑀𝑀𝑀

stands for the number of investment options/funds within the fund family, and 𝐿𝐿𝑏𝑏𝑜𝑜𝐼𝐼𝑜𝑜50 is the proportion

of members above the age of 50.

4.3.4 Dependent Variable: Member-initiated Family Rollover

Not having access to exact fund flows, previous managed fund studies approximate net flows

by using fund total net assets and fund returns as follows (Chevalier & Ellison 1997; Del Guercio &

Tkac 2002; Sirri & Tufano 1998):

33

𝐿𝐿𝑜𝑜𝑀𝑀𝑁𝑁𝑁𝑁𝑜𝑜𝑁𝑁𝑖𝑖,𝑡𝑡 = 𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡 − 𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡−1(1 + 𝑅𝑅𝑖𝑖,𝑡𝑡)

𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡−1 (2)

where 𝐿𝐿𝑜𝑜𝑀𝑀𝑁𝑁𝑁𝑁𝑜𝑜𝑁𝑁𝑖𝑖,𝑡𝑡 is the net flow growth for fund i during year t; 𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡 is the total net asset of fund i

in year t; 𝑅𝑅𝑖𝑖,𝑡𝑡 is the return of fund i at time t.

The application of this measure to the superannuation setting is problematic due to its failure to

isolate Superannuation Guarantee (SG) mandated contributions (which leads to a high autocorrelation

in the flows) and payments of member benefits, both of which are not member-initiated investment

decisions.

This thesis thus uses APRA-reported investment rollovers, which are transfers between APRA-

regulated superannuation entities. Hence, the inward rollover does not include amounts that are new to

the superannuation system, e.g., the mandatory contributions under the SG. Similarly, the outward

rollover excludes benefit payments. When a fund winds up, its rollovers are not member initiated.

Section 4 of Superannuation Reporting Form (SRF) 250.0 is used to identify the name of the winding

up entity, the date of wind up, and the successor entity. Fund flow due to fund merger/consolidation is

not initiated by individual members. Therefore, the outward rollover from a family wind up is deleted.

Inward rollover of a successor fund family from winding up families, using the outflow of those funds

leaving the industry in the year in which they wound up is also removed. If these data are not available,

this thesis uses assets of those superannuation fund families in the year prior to wind up. Thus, this

thesis can separate flows nominated by employers or trustees.

Admittedly, investors who switch fund families may do so because they change jobs (Fear

& Pace 2008). While the change of employer cannot be controlled for because of lack of data, a job

change actually provides a chance for investors to make an active choice regarding whether to switch

their fund family. Most employees are free to direct their superannuation contributions to any public-

offer fund, or their own SMSF and are not constrained by the employer’s choice of default fund.

Furthermore, the APRA-reported rollovers include transfers from job changes only when the

member decides to consolidate their existing superannuation balance into the new employer’s default

superannuation fund family. In such a case, the investor has made an active investment choice.

4.3.5 Alternative Member Choice Measures

To complement the main member’s choice measure, alternative measures, including the

employee’s personal contribution, the proportion of assets in the default investment option and the

number of new members, are also used. Members’ personal contribution is a voluntary after-tax

34

superannuation contribution, which includes contributions from both new and existing members.

The proportion of default asset can be used to examine the level of active choices made within the

fund family to determine the proportion of members who opt for a passive (i.e., just using the

default option) or active investment style. The number of new members represents the number of

personal and employer-sponsored members who joined the superannuation fund family during the

financial year.

4.3.6 Fund Family Performance

Fund family performance measure used in this study is the APRA-reported rate of return,

which is calculated as a fund family’s net earnings after tax divided by the cash flow-adjusted total

net assets under management. As opposed to individual fund returns, such family returns measure

the combined earnings of superannuation assets towards fund members’ retirement benefits in a

superannuation fund family. The Superannuation Industry (Supervision) Act 1993 requires that

superannuation trustees must ‘formulate, and give effort to, an investment strategy that has regard to

the whole of the circumstances of the entity and in the best interest of its members’. APRA claims

that its rate of return is a useful measure to assess a superannuation trustee’s ability to deliver on the

fund’s investment strategy for the benefit of all members over time (APRA 2013). In addition, the

APRA-reported return is widely used by superannuation disclosure documents (such as PDSs and

member statements) as well as superannuation comparison websites such as SuperGuide24 and

SuperRatings25. It is therefore expected to be observed and used by investors when they select

superannuation fund families. To avoid concerns relating to reverse causality, I follow Sirri and

Tufano (1998)’s approach of using the family returns over the preceding year for the regression

analysis.

4.3.7 Marketing Expense

Various measures have been used to capture managed funds’ marketing effort and the

resultant on fund flows. Khorana and Servaes (2004) and Barber, Odean and Zheng (2005) measure

marketing expenses at the individual fund level through 12b-1 fees. Sirri and Tufano (1998) and

Huang, J, Wei and Yan (2007) use a fund’s total fee ratio, defined as the annual expense ratio plus

one-seventh of the up- front load fees. Jain and Wu (2000) find that an advertisement in one of two

business periodicals is associated with larger flows to the advertised managed fund than to a

24 See http://www.superguide.com.au/. 25 See http://www.superratings.com.au/.

35

matched fund in a control group. Gallaher, S, Kaniel, R and Starks, LT (2006) examine the effect of

monthly print advertising expenditures of mutual fund families over the 1992-2001 period on

investment flows.

Marketing decisions originate at the fund family level, so this thesis uses the superannuation

fund family as the unit of observation for marketing activities. A close examination of the expense

items in SRF200 reveals that marketing expenses are not reported as a separate expense category

but is reported under ‘other operating expenses’, which is used as a proxy for marketing costs in this

study.26 As fund advertising has no persistent effect (Gallaher, S, Kaniel, R & Starks, LT 2006), this

thesis uses current rather than lagged other operating expenses as a proxy for marketing expenses in

this study.

4.3.8 Control Variables

Several control variables that are likely to affect investor choice in the regression are also

included. Fund family size reflects economies of scale and scope. The natural logarithm of TNA (total

net asset) is taken to represent brand recognition and resources controlled by the family. Again, to

account for endogeneity concerns, the previous year’s value of TNA is used. The natural logarithm of

the number of investment categories available to members is used because a greater variety of

investment options is expected to attract a broader set of members who have different performance

targets and risk appetites. The proportion of members above the age of 50 is included given anecdotal

evidence that investors approaching retirement age are more aware of and more engaged in their

superannuation arrangements. They are more likely to make active switching choices, and thus affect

fund flows. Family type and year fixed effects are also controlled for.

4.4 Results

4.4.1 Summary Statistics

Considering that this study is the first empirical investigation of this kind, this chapter first

presents some descriptive statistics to provide a better understanding of Australian superannuation

fund families. Table 4.2 provides a breakdown of the sample for each year. Over time, the number

of superannuation fund families has decreased, which reflects the trend of consolidation in the

superannuation industry. The consolidation together with new contributions has resulted in an

increase in family size (measured by TNA). Marketing cost is a proxy for superannuation fund

26 This has been checked and confirmed by the APRA statistics team.

36

families’ visibility to the public. The average amount of marketing costs increased owing to the

adoption of the ‘Choice’ policy in 2005 until 2008, and it fell in 2009 and 2010 before it started to

increase again in 2011. The average number of investment options per fund family has increased

dramatically since 2005 and has remained over 100 since 2009. Net rollover is not zero owning to

restrictions applied in the sample selection process. In general, the inward rollover is about 7-8% of

assets under management. Fund family returns coincide with economic cycles (with negative returns

in 2008, 2009 and 2012).

Table 4.2 Means for the Full Sample This table reports means for the full sample from 2005 to 2014. At the end of each year, I calculate the cross-sectional mean value of the following

superannuation fund family characteristics: total net asset under management (TNA), marketing expense (MktExp), investment options (InvOpt),

APRA-reported family return (Return) and the proportion of members above the age of 50 (Above50). The flow measures (net rollover, inward

rollover and outward rollover) are APRA-reported transfers among superannuation entities.

Year N TNA MktExp InvOpt Return Above50 Net Rollover Inward Rollover

Outward Rollover

(in millions) (in millions) (%) (%) (in millions) (in millions) (in millions)

2005 223 883.77 1.12 51.74 9.88 37.23 50.01 133.75 82.94

2006 203 1394.64 1.77 61.66 11.95 36.78 78.68 223.62 141.37

2007 201 1883.08 2.46 67.12 13.03 37.18 47.77 218.82 183.21

2008 190 2601.77 2.55 99.19 -7.97 36.05 27.79 246.07 204.43

2009 191 2668.43 2.27 104.46 -10.21 38.90 23.55 189.60 165.84

2010 177 2502.60 1.80 115.37 7.44 39.00 45.39 230.53 168.15

2011 165 3172.61 1.96 147.82 6.64 39.79 36.43 229.33 194.48

2012 158 3770.27 1.92 150.31 -0.39 41.05 6.25 223.41 232.96

2013 150 3957.86 2.52 161.87 11.58 43.23 10.93 271.09 250.89

2014 144 5210.54 271.97 9.44 23.12 314.27 307.18

Panels A and B in Table 4.3 provide summary statistics for retail and industry

superannuation fund families, respectively. The consolidation of fund families in the full sample

seems to be driven by retail superannuation fund families, whereas the number of industry fund

families has increased during the same period because more of them have become public-offer

families.27 The average size of industry families is larger than that of their retail counterparts. The

performance of industry families is better, except during the years 2008 and 2009. Industry families

spent more money on marketing than retail families in terms of both the absolute dollar amount

27 A quick analysis reveals that, despite the significant consolidation in the industry (the number of superannuation funds decreased by 91% from 3,720 in June 2001 to 336 funds in June 2012), the level of industry concentration measured by the Herfindahl- Hirschman Index (HHI) is as low as 2.7 %. No single superannuation fund has a dominant market share of more than 4%. The largest five funds by assets in 2012 comprised 16% of the market share of the superannuation industry. By comparison, the four major banks comprised about 79% of banking industry assets in June 2012. Hence, there are still a large number of funds competing for members’ business.

37

of expenses and the expense ratio (divided by total net assets). Many more investment options

are offered in retail families. Retail families provide about 300 options, on average, while

industry families provide approximately 13 options. The amount of both inward and outward

rollovers is greater for retail families.

Table 4.3 Means for the Subsamples

This table reports means for the retail/industry superannuation fund families in the sample from 2005 to 2014. At the end of each year, I calculate

the cross-sectional mean value of the following superannuation fund family characteristics: total net asset under management (TNA), marketing

expense (MktExp), investment options (InvOpt), APRA-reported family return (Return) and the proportion of members above the age of 50

(Above50). The flow measures (net rollover, inward rollover and outward rollover) are APRA-reported transfers among superannuation entities.

Panel A: Retail Families

Year N TNA MktExp InvOpt Return Above50 Net Rollover Inward Rollover

Outward Rollover

(in millions) (in millions) (%) (%) (in millions) (in millions) (in millions)

2005 161 892.38 0.93 68.93 10.01 42.68 57.42 158.03 99.50

2006 138 1406.54 1.59 86.59 12.05 43.77 80.30 252.15 179.10

2007 142 1686.67 2.05 91.95 12.94 44.06 50.28 249.52 216.46

2008 129 2526.93 2.23 142.38 -8.68 42.67 31.32 298.77 247.05

2009 131 2439.80 1.77 148.69 -9.84 46.31 16.52 216.62 199.79

2010 117 2265.16 1.12 170.50 7.60 46.60 64.92 290.51 205.95

2011 106 2944.85 1.55 225.73 6.42 47.93 27.53 262.05 236.99

2012 100 3299.44 1.33 232.61 -0.85 49.76 9.43 263.77 279.31

2013 98 3255.88 1.53 242.85 11.50 52.36 22.30 305.18 268.70

2014 95 4305.58 405.58 8.81 57.99 343.14 316.33

Panel B: Industry Families

Year N TNA MktExp InvOpt Return Above50 Net Rollover Inward Rollover

Outward Rollover

(in millions) (in millions) (%) (%) (in millions) (in millions) (in millions)

2005 29 1625.14 2.93 12.69 11.76 23.81 53.13 122.97 69.84

2006 38 1855.14 3.26 13.21 13.67 23.08 121.98 246.39 79.07

2007 36 2971.31 5.08 10.33 15.16 22.01 61.04 190.86 129.82

2008 37 3474.97 4.99 11.24 -6.81 23.72 29.53 175.64 146.12

2009 38 3899.94 4.46 10.82 -12.02 23.75 51.30 167.84 116.54

2010 36 3815.54 4.76 11.08 7.80 23.66 0.60 144.53 124.21

2011 36 4536.78 3.85 10.61 8.03 25.24 51.41 199.47 148.06

2012 37 5448.09 3.58 10.78 0.41 27.04 7.09 192.46 185.37

2013 34 6149.54 5.29 11.65 13.46 26.05 2.06 259.40 257.34

2014 33 7959.79 16.24 11.71 -57.12 310.07 347.33

4.4.2 Effect of Performance

While the term ‘chasing performance’ has no standard definition, it is loosely used to

indicate that investors gravitate towards high-return investment strategies. Gruber (1996) and Sirri

and Tufano (1998) propose that investors infer managerial skill from past returns and therefore

38

chase returns. While such performance-chasing behaviour generally holds at the individual fund

level for mutual funds, much less is known about family performance in the superannuation context.

Table 4.4 reports multivariate regression results regarding the determinants of fund family

flows for the overall sample and two sub-subsamples of retail and industry fund families. The first

column from Table 4.4 shows a statistically significant and positive relationship between fund

family performance and net rollovers. The untabulated results for effect sizes reveal that 0.8% of the

variance unexplained by other factors is explained by past performance; in this way, it provides

evidence of a positive flow-performance relation at the family level in the superannuation industry.

Similar results are found for the sub-samples of both retail families and industry families. These

results provide support for H1. However, after partitioning the rollover into inward and outward

rollovers, no statistically significant relation is found between performance and inward rollover;

rather, a negative association is found with outward rollovers. These findings be explained by the

passive investment style of superannuation investors and the long-term nature of their

investment. Dahlquist and Martinez (2015) infer that pension investors face a greater risk of being

caught in poorly performing funds because of their inattention to past performance. However, the

results in the study show that superannuation members punish families with poor performance by

directing investments to other families. Members can be inattentive to their superannuation

investments until poor performance cues indicate that a change is necessary. This finding suggests

that investors will alter their superannuation choice only when they are sufficiently dissatisfied with

their current superannuation investments. The observed flow – the performance relationship with

outward rollovers – provides a contrast to the findings in the managed fund literature, which

again highlights the importance of studying investment flows in the superannuation context.

39

Table 4.4 Determinants of Investment Flows

This table examines the determinants of superannuation fund family investment flows. The flow measures (Net Rollover, Inward Rollover and Outward

Rollover) are APRA-reported transfers among superannuation entities. All the flow measures are scaled by superannuation fund family size. Each year,

fractional performance ranks (Perft−1) ranging from zero to one are assigned to superannuation fund families according to their return in last year.

Marketing Expense (MktExp) is scaled by TNA. LnTNAt−1 is the lagged natural logarithm of net asset under management. LnInvOpt is the natural

logarithm of the number of investment options offered by a superannuation fund family. The proportion of members above the age of 50 is denoted as

Above50. Robust p-values are reported in parentheses. ***, ** , and * denote significance at the 1%, 5% and 10% level, respectively.

Sample All Families All Families

All Families

Retail Families

Retail Families

Retail Families

Industry Families

Industry Families

Industry Families

Choice Measure

Net Rollover

Inward

Rollover

Outward Rollover

Net

Rollover

Inward

Rollover

Outward Rollover

Net

Rollover

Inward

Rollover

Outward Rollover

Perf t-1 0.057*** 0.024 -0.032*** 0.062** 0.029 -0.030* 0.032 0.024 -0.010*

(0.009) (0.262) (0.006) (0.034) (0.327) (0.056) (0.164) (0.360) (0.061)

MktExp 16.876*** 18.912*** 1.358 21.930*** 23.984*** 1.226 -1.977 -2.587 -1.513

(0.002) (0.000) (0.150) (0.001) (0.000) (0.283) (0.672) (0.616) (0.333)

LnTNA t-1 -0.002 0.002 0.004 -0.003 0.003 0.006 0.002 -0.004 -0.006***

(0.763) (0.694) (0.206) (0.693) (0.667) (0.100) (0.877) (0.709) (0.000)

LnInvOpt 0.010* 0.004 -0.006 0.009 0.002 -0.007 0.032 0.034 0.002

(0.054) (0.407) (0.129) (0.134) (0.715) (0.115) (0.131) (0.130) (0.508)

Above50 0.062 0.097** 0.044* 0.049 0.089** 0.050* 0.205* 0.150 -0.064***

(0.146) (0.020) (0.088) (0.286) (0.044) (0.082) (0.093) (0.243) (0.000)

Constant -0.204*** -0.110*** 0.094*** -0.114*** -0.031 0.081*** -0.043 0.061 0.112***

(0.000) (0.001) (0.005) (0.004) (0.439) (0.000) (0.439) (0.310) (0.000)

Observations

1,332

1,332

1,332

897

897

897

267

267

267

Adjusted R- squared 0.094 0.123 0.141 0.126 0.148 0.055 0.081 0.058 0.416

Year Fixed Effect Yes Yes Yes Yes Yes Yes Yes Yes Yes

Family Type Controlled Yes Yes Yes No No No No No No

4.4.3 Effect of Marketing on Fund Family Flows

The second potential determinant of fund family flows examined is marketing efforts. Table 4.4

shows a strong positive relation between marketing expense and net rollover, which suggests that

investors pay attention to marketing when selecting a family in which to invest funds. Most

superannuation investors have no formal financial training. There are thousands of funds for investors

to choose from, far more than any investor can carefully consider. The significant positive relation

between marketing and net rollover supports the premise of this thesis that marketing helps attract

investors’ attention as these fund families are easier and less costly for investors to identify. Analysis

40

of the effect size suggests that 2.8% variance is unexplained by effects other than MktExp is explained

by the marketing effect.

Next, when net rollovers are divided into inward rollovers and outward rollovers, I find that

marketing expense increases inward rollovers but does not have a significant impact on outward

rollovers. As the information cost of new investors is higher than that of existing members, enhanced

visibility produced by marketing efforts attracts more inward rollovers, but does not affect existing

members to the same extent.

The above analysis shows that fund families with high marketing expenses attract more

investment rollovers. To explore whether this relation applies to different types of fund families, this

study re-runs the tests for retail and industry fund families respectively, as these two types of funds

account for the majority of the sample families. Similar results are found for the retail superannuation

fund family sub-sample. As for industry families, despite the fact that they spend more on marketing

compared to retail families (Table 4.3), the relationship between marketing and fund flow is not

statistically significant.28 In summary, it appears that retail or for-profit funds are responsible for

driving the significant positive relation between marketing and inward rollover for the overall sample.

No significant effects are found for industry funds. The results suggest the higher amount of

marketing expenditure incurred by industry families does not appear to attract eligible members to

switch fund families.

Industry superannuation funds advertising campaigns reached a level in 2011 that it prompted

the Financial Planning Association (FPA) and the Association of Financial Advisers (AFA) to question

whether the industry funds were really acting in the best interests of their members. Industry funds

justify advertising by saying they advertise to attract more members to achieve cost efficiencies that

provide benefits to members. 29 Nevertheless, no evidence from this chapter shows marketing is a useful

strategy for industry funds to compete with their retail competitors to pursue growth and retain members.

And the benefit conferred (lower fees) is also questionable.

4.4.4 Control Variables

Turning to other control variables, size (LnTNA) is often used to proxy for economies of scale

in raising fund visibility. No statistically significant relation is found between family size and

rollovers, except for the negative relation between fund family size and industry families’ outward

28 Superannuation investors may perceive industry funds as funds they cannot choose if they do not belong to that particular industry. 29 See news articles from http://www.moneymanagement.com.au/features/editorial/industry-super-funds-advertising-onslaught-incomparable and http://www.news.com.au/finance/superannuation/industry-superfunds-rejects-claims-advertising-using-members-money-not-delivering-value/news-story/c5da303ddac8d782c5eb6058883b95b4.

41

rollovers. This means industry fund investors are less likely to leave the family if the family is larger

in size. Next, superannuation families can employ category proliferation to limit competition and

increase market coverage, but the number of products offered (LnInvOpt) only has a marginal positive

impact on the net rollover for all families (column 1 of Table 4.4). The results also show that members

above the age of 50 (Above50) are generally more active in terms of both inward and outward

rollovers (for the overall sample and retail sub-sample). This is in line with the survey evidence of

Clark et al. (2013) and Butt, Adam et al. (2015) that older members are more likely to make active

choices as account balances increase and retirement approaches. In contrast to retail families, a

significant negative coefficient is shown for industry families in the outward rollovers column,

indicating that the propensity to mover funds out (outward rollovers) is lower for members over 50s

in the industry funds. A possible explanation for this is that members above 50 are less likely to

change job or move industry.

4.4.5 Alternative Choice Measures

In this part of the analysis, the determinants of investor choice using alternative dependent

variables including personal contribution, proportion of default asset and number of new members are

revisited. This analysis is used to complement the main analysis as these measures are not pure choice

measures that capture active switching decisions. For example, both the personal contribution and the

percentage of default asset fail to disentangle the effect caused by existing and new members. The

percentage of default option reflects choice at individual fund level. The number of new members

includes those automatically enrolled by their employers. However, these alternative measures may still

provide interesting supplementary evidence for the level of investor engagement. The results are

presented in Table 4.5.

Table 4.5 results reveal no statistically significant relation between performance and all choice

measures, except a marginally significant positive relation between performance and the number of new

members, suggesting that better performance is associated with more new members. The results for

marketing expenses show that marketing is associated with more personal contributions. In addition,

employer contribution is also tested. Results from column 2 in Table 4.5 show no significant

coefficients. This is because employer contribution is determined by wages and salaries, and does not

represent the choices made by investors. This also supports this study’s argument that the traditional net

flow approximation is not an appropriate choice measure because the measure includes employer

contributions. Next, it is found that the proportion of default asset decreases as marketing expense

increases. This result carries two potential implications: more active investors are less likely to choose

42

the default option, or marketing can also encourage existing members to make active switching choices

within the fund family. The effect of marketing is further supported by its positive and significant

relation with the number of total new members. When the sample is split into retail and industry sub-

samples, results from Table 4.6 are similar to the main results discussed in this section.

Table 4.5 Determinants of Member Choice – Alternative Measures

This table examines the determinants of superannuation fund family choices, using the alternative choice measures. PerCon is the scaled personal member

contributions. EmplCon represents the employer contributions, which is also scalded by TNA. DefAsset is the proportion of fund asset in the default option.

TotNew is the number of new members each year. Each year, fractional performance ranks (Perft−1) ranging from zero to one are assigned to

superannuation fund families according to their return in last year. Marketing Expense (MktExp) is scaled by TNA. LnTNAt−1 is the lagged natural

logarithm of net asset under management. LnInvOpt is the natural logarithm of the number of investment options offered by a superannuation fund family.

The proportion of members above the age of 50 is denoted as Above50. Robust p-values are reported in parentheses. ***, ** , and * denote significance at

the 1%, 5% and 10% level, respectively.

Sample All Families All Families All Families All Families

Choice Measure

PerCon

EmplCon

DefAsset

TotNew

Perf t-1 -0.000 6.609 0.009 0.021*

(0.975) (0.291) (0.789) (0.091)

MktExp 2.304*** 714.754 -11.868*** 8.437***

(0.003) (0.299) (0.003) (0.002)

LnTNA t-1 0.003** -1.626 -0.012 0.000

(0.032) (0.288) (0.146) (0.929)

LnInvOpt 0.000 0.327 -0.091*** 0.011***

(0.981) (0.345) (0.000) (0.003)

Above50 0.042*** -5.888 -0.027 -0.107***

(0.001) (0.287) (0.733) (0.000)

Constant -0.041*** 8.155 0.795*** 0.182***

(0.000) (0.295) (0.000) (0.003)

Observations

1,332

1,332

1,331

1,320

Adjusted R-squared 0.236 0.059 0.450 0.136

Year Fixed Effect Yes Yes Yes Yes

Family Type Controlled Yes Yes Yes Yes

43

Table 4.6 Determinants of Member Choice – Alternative Measures Cont.

This table examines the determinants of superannuation fund family choices, using the alternative choice measures. PerCon is the scaled personal member

contributions. EmplCon represents the employer contributions, which is also scalded by TNA. DefAsset is the proportion of fund asset in the default option.

TotNew is the number of new members that join the family each year. Each year, fractional performance ranks (Perft−1) ranging from zero to one are

assigned to superannuation fund families according to their return in last year. Marketing Expense (MktExp) is scaled by TNA. LnTNAt−1 is the lagged

natural logarithm of net asset under management. LnInvOpt is the natural logarithm of the number of investment options offered by a superannuation fund

family. The proportion of members above the age of 50 is denoted as Above50. Robust p-values are reported in parentheses. ***, ** , and * denote

significance at the 1%, 5% and 10% level, respectively.

Sample Retail Families

Retail Families

Retail Families

Retail Families

Industry Families

Industry Families

Industry Families

Industry Families

Choice Measure PerCon EmplCon DefAsset TotNew PerCon EmplCon DefAsset TotNew

Perf t-1 -0.003 8.132 -0.006 0.025 -0.005 -0.010 0.044 0.005

(0.782) (0.283) (0.896) (0.124) (0.449) (0.190) (0.269) (0.704)

MktExp 3.139*** 915.095 -15.692*** 12.495*** -1.031 -2.539 -24.665 -1.674

(0.000) (0.293) (0.001) (0.000) (0.403) (0.245) (0.208) (0.716)

LnTNA t-1 0.004** -2.155 -0.021** 0.002 0.001 -0.000 0.036 -0.002 (0.036) (0.281) (0.035) (0.517) (0.396) (0.833) (0.102) (0.710)

LnInvOpt -0.000 0.501 -0.088*** 0.010** -0.003 0.001 -0.128*** 0.021**

(0.961) (0.327) (0.000) (0.014) (0.436) (0.730) (0.009) (0.038)

Above50 0.036*** -6.509 0.027 -0.109*** 0.085*** -0.188*** -0.393*** -0.110**

(0.008) (0.287) (0.749) (0.000) (0.001) (0.000) (0.010) (0.026)

Constant -0.020** 12.111 0.825*** 0.078*** 0.013 0.186*** 0.999*** 0.108*** (0.023) (0.279) (0.000) (0.004) (0.178) (0.000) (0.000) (0.004)

Observations 897 897 896 894 267 267 267 259

Adjusted R-squared 0.216 0.078 0.323 0.166 0.467 0.593 0.198 0.136

Year Fixed Effect Yes Yes Yes Yes Yes Yes Yes Yes

4.4.6 Further Analysis on the Effect of Marketing on Fund Family Flows

Gallaher, S, Kaniel, R and Starks, L (2006) find that advertising affects fund flows in a non-

linear fashion with convexity at the upper end. Sirri and Tufano (1998) find that if funds spend more,

which they assume are funds with greater marketing efforts, then investors are more sensitive to the

fund’s salient features. Thus this study employs a specification that allows for this non-linear relation.

Marketing expense is normalized on a [0,1] interval analogous to the Sirri and Tufano (1998)

normalization procedure. The fractional rank for superannuation fund families in the bottom marketing

expense quintile (LowMkt) is defined as min (MktExp, 0.2). Families in the three medium marketing

expense quintiles (MidMkt) is defined as min (MktExp – Low, 0.6). The rank for the top quintile (High)

is defined as MktExp – MidMkt – LowMkt. For example, a fund family in the 10th percentile would

44

experience flows of 0.1 × 𝑏𝑏1 if all the other covariates were equal to zero. On the other hand, a fund

family in the 90th percentile would experience flows of 0.2 × 𝑏𝑏1 + 0.6 × 𝑏𝑏2 + 0.1 × 𝑏𝑏3 if all the other

covariates were zero. A two-kink continuous piecewise-linear specification is then used in the manner

of Sirri and Tufano (1998) and others, which allows for different flow-marketing sensitivities in

different marketing expense ranges.

Earlier results from Table 4.4 show that marketing expense is positively associated with inward

rollovers for retail families. The results from Table 4.7 highlight a non-linear relationship where families

with higher marketing expenditures enjoy disproportionately larger fund inflows. That is, the magnitude

of the flow coefficient on the top advertiser’s marketing expense is substantially larger. For example, a

10 percentile increase in the marketing expense rank increases the net flow by 8.62% for the top quintile.

On the other hand, the families in the middle range of marketing expense ratio have smaller outflows.

The fund families who spend the least have no significant relation between their marketing expense and

fund flows. High relative levels of advertising are significantly related to high net flows and inflows for

retail families. Therefore, it appears that for marketing to matter, the fund family must ensure that it is

one of the top advertisers on a relative basis.

In summary, it appears that a threshold of marketing expenditure relative to competitors’

marketing expenditure exists before the marketing efforts have a significant effect on flows into the

family. Given that the results indicate marketing has a significantly positive impact only at the top end,

marketing decision can be a costly strategy for fund families.

45

Table 4.7 Piecewise Regression Results

This table presents the results of piecewise regressions of investment flows on different marketing expense ranges. The flow measures (Net Rollover,

Inward Rollover and Outward Rollover) are APRA-reported transfers among superannuation entities. All the flow measures are scaled by superannuation

fund family size. Each year, fractional performance ranks (Perft−1) ranging from zero to one are assigned to superannuation fund families according to

their return in last year. The fractional rank for superannuation fund families in the bottom marketing expense quintile (LowMkt) is defined as min(MktExp,

0.2). Families in the three medium marketing expense quintiles (MidMkt) is defined as min(MktExp– Low, 0.6). The rank for the top quintile (High) is

defined as MktExp – MidMkt – LowMkt. For example, a fund family in the 10th percentile would experience flows of 0.1 × b1 if all the other covariates

were equal to zero. On the other hand, a fund family in the 90th percentile would experience flows of 0.2 × b1 + 0.6 × b2 + 0.1 × b3 if all the other

covariates were zero. LnTNAt−1 is the lagged natural logarithm of net asset under management. LnInvOpt is the natural logarithm of the number of

investment options offered by a superannuation fund family. The proportion of members above the age of 50 is denoted as Above50. Robust p-values are

reported in parentheses. ***, ** , and * denote dignificance at the 1%, 5% and 10% level, respectively.

Sample All Families

All Families

All Families

Retail Families

Retail Families

Retail Families

Industry Families

Industry Families

Industry Families

Choice Measure

Net Rollover

Inward Rollover

Outward Rollover

Net Rollover

Inward Rollover

Outward Rollover

Net Rollover

Inward Rollover

Outward Rollover

Perf t-1 0.054** 0.023 -0.030*** 0.058** 0.027 -0.028* 0.029 0.022 -0.009*

(0.015) (0.304) (0.008) (0.046) (0.364) (0.069) (0.164) (0.360) (0.082) LowMkt 0.184 0.131 -0.074 0.160 0.016 -0.149 -0.233 -0.163 0.079*

(0.131) (0.283) (0.245) (0.337) (0.928) (0.102) (0.338) (0.535) (0.078)

MidMkt -0.085** -0.045 0.050** -0.102* -0.045 0.066** 0.020 0.009 -0.021*

(0.041) (0.264) (0.024) (0.084) (0.428) (0.039) (0.633) (0.849) (0.089)

HighMkt 0.862*** 0.889*** -0.042 1.274*** 1.297*** -0.063 -0.056 -0.064 0.003 (0.004) (0.002) (0.631) (0.003) (0.002) (0.609) (0.590) (0.554) (0.904)

LnTNA t-1 -0.005 -0.001 0.004 -0.007 -0.000 0.006* 0.002 -0.004 -0.006*** (0.381) (0.877) (0.175) (0.367) (0.961) (0.082) (0.874) (0.693) (0.000)

LnInvOpt 0.010* 0.004 -0.006 0.008 0.002 -0.006 0.032 0.035 0.002 (0.058) (0.395) (0.145) (0.167) (0.727) (0.141) (0.122) (0.129) (0.648)

Above50 0.054 0.093** 0.049* 0.037 0.082* 0.056* 0.203* 0.147 -0.065***

(0.177) (0.020) (0.070) (0.403) (0.057) (0.063) -0.233 (0.238) (0.000)

Constant -0.171*** -0.081* 0.088** -0.063 0.026 0.082*** -0.009 0.085 0.100*** (0.004) (0.075) (0.019) (0.221) (0.623) (0.001) (0.865) (0.122) (0.000)

Observations 1,332 1,332 1,332 897 897 897 267 267 267 Adjusted R-squared 0.059 0.077 0.146 0.077 0.090 0.064 0.077 0.053 0.438

Year Fixed Effect Yes Yes Yes Yes Yes Yes Yes Yes Yes

Family Type Controlled Yes Yes Yes No No No No No No

4.4.7 Robustness Tests

To supplement the main fixed effect tests, this section modifies the unexpected fund flow

measure using the methodology in Warther (1995) to provide Generalized Method of Moments (GMM)

analysis. Fund flow is known to be highly autocorrelated, meaning a portion of the fund flow is

46

predictable. Warther (1995) uses Box-Jenkins diagnostics to identify the time-series properties of fund

flows and uses time-series models to estimate the unexpected component of the flows. Warther (1995)

uses time-series analysis at aggregate fund flow level, whereas data in this study are panel data with a

relative short time period. To derive the unexpected flow, lagged dependent variables need to be

included in the model. This leads to the application the system Generalized Method of Moments (GMM)

procedure of Arellano and Bond (1991) in the dynamic panel data. This chapter regresses the standard

fractional flow measure and add lagged dependent variables (lagged flows) as long as the lagged values

are significant. Further lagged flows are used as level and difference instruments.30 It turns out that the

level instrument is lagged by 2 years and the difference instrument is lagged by 3 years. The resulting

equation is assessed by Sargan test, and the Sargan P value suggests the over identifying moment

conditions are valid. In addition, Arellano-Bond test is run to ensure residuals are not autocorrelated.

Both order 1 and order 2 tests statistics indicate there is no serial correlation in the idiosyncratic errors.

The residual value in the model is then used as the unexpected flow for the current period. The results

reported in Table 8 are qualitatively the same as earlier analysis.

Robustness checks are also performed by including the non-public offer families, as they account

for more than half of the initial sample. Another sensitivity test is to run the test with dummy variables

for banks and insurers who offer financial products other than superannuation, because they tend to

have higher levels of visibility compared to other entities that offer superannuation products only. The

results (untabulated) for these tests are similar to the main results.

30 The Xtdpd command in Stata is used to estimate the system GMM. Xtdpd enables separate specifications for each instrument.

47

Table 4.8 Determinants of Member Choice – Using Unexpected Flow

This table examines the determinants of superannuation fund family choices, using the alternative choice measures. Unexpected Flow is the derived

residual of a dynamic panel data regression for fractional net flows. Each year, fractional performance ranks (Perft−1) ranging from zero to one are

assigned to superannuation fund families according to their return in last year. Marketing Expense (MktExp) is scaled by TNA. LnTNAt−1 is the lagged

natural logarithm of net asset under management. LnInvOpt is the natural logarithm of the number of investment options offered by a superannuation

fund family. The proportion of members above the age of 50 is denoted as Above50. Robust p-values are reported in parentheses. ***, ** , and * denote

significance at the 1%, 5% and 10% level, respectively.

Sample All Families Retail Families Industry Families

Choice Measure Unexpected Flow Unexpected Flow Unexpected Flow

Perf t-1 0.005 0.001 0.016

(0.504) (0.901) (0.253)

MktExp 4.544*** 5.899*** -0.894

(0.004) (0.002) (0.522)

LnTNA t-1 -0.004* -0.005* -0.006 (0.051) (0.073) (0.286)

LnInvOpt 0.004** 0.004* 0.017

(0.032) (0.065) (0.110)

Above50 -0.005 -0.010 0.050

(0.743) (0.498) (0.251)

Constant -0.038*** -0.001 -0.065*** (0.005) (0.927) (0.001)

Observations 1,332 897 267

Adjusted R-squared 0.693 0.641 0.833

Year Fixed Effect Yes Yes Yes

Family Type Controlled Yes No No

4.5 Conclusion

Superannuation is very important to working Australians in terms of both its economic and

social implications. Australians are offered various levels of choice in their superannuation

arrangements. While much of the literature focuses on the investment strategy (individual fund)

choice, very little analysis is available on investors’ choices at fund family level. This study

examines the choice of superannuation fund families, i.e. the choice to rollover accumulated savings

to another superannuation family, using a more precise measure of active investor choice. In contrast

to the managed funds setting, results show the superannuation investors appear to be insensitive to

higher returns but do choose to switch fund family if their fund family performs poorly, and

48

superannuation fund members appear to withdraw assets from fund families with poor performance.

The implication for superannuation fund families is that performance-maximization may not be the

optimal strategy to attract new investors. However, poor performers need to improve return to avoid

outward rollovers and to retain members.

The results also show that there is a strong positive association between a family’s rollovers

and its relative levels of marketing expenditures with a significant effect for advertisers with largest

marking expenditure only. The results suggest that marketing is effective in attracting

unsophisticated superannuation investors into retail superannuation families. In contrast, marketing

does not appear to be a useful strategy for industry families to attract new investments.

49

Chapter 5 – The Short Product Disclosure Statement (PDS) Disclosure Practice

5.1 Introduction

This chapter examines the Product Disclosure Statement (PDS) disclosure practice of Australian

superannuation fund families, and the relation between PDS readability/content and superannuation

family performance. These issues are important given the serious concerns raised by regulators about

the usefulness and understandability of superannuation PDS disclosure, which resulted in the

introduction of the short form PDS regime in 2011 to make it easier for superannuation members to

locate important information in the PDS and compare information across superannuation products

(ASIC, 2011). The regulatory change aims to improve the readability and comparability of PDS

disclosure but the restrictions of PDS documents to a maximum of 8 pages with prescribed section

headings may also reduce the fund families’ ability to provide useful details to distinguish themselves

from other fund families. These restrictions, when coupled with the concern that many superannuation

fund families produce PDS documents mainly to comply with regulations and minimise their liabilities,

raise concerns about whether the short-form PDSs are too standardised to provide sufficient fund family

specific information to help investors choose among fund families. Therefore, this study first

investigates the short-form PDS from 2012 to 2015 to understand the current landscape of the PDS

disclosure practice. In addition, prior research on corporate disclosure shows the readability and extent

of managerial disclosure are affected by company performance, so this study also examines whether

fund family performance affects the readability and extent of PDS disclosure (Lang & Lundholm 1993;

Li 2008; Schrand & Walther 2000).

As discussed in chapter 4, marketing serves as a tool to reduce superannuation members’

information costs by increasing superannuation families’ visibility. Superannuation PDS is another

mechanism that can help reduce information costs, as it directly provides more detailed information to

investors, and is likely less costly than marketing. To attract investors, superannuation fund families

may provide more information (i.e., the extent of disclosure) or improve the readability of their

disclosure to facilitate investors’ decision-making process. Drawing on the voluntary disclosure

incentive literature in corporate disclosure (Lang & Lundholm 1993; Li 2008; Schrand & Walther

2000), superannuation fund families may also be more forthcoming in the disclosure of information

when their performance is good.

The need for disclosure generally arises when there are information asymmetry and agency

problems. In the absence of professional analysts following, fund families are expected to use

disclosure as their primary means to assist investors to make informed investment decisions. The

50

PDS conveys the most basic but most important information about the fund family, and is often one

of the first sources a potential investor or a financial adviser uses in making investment judgements.

The average superannuation fund member is not trained in financial analysis and often lacks the

expertise to fully process the financial information available to them. Hence the role of informative

and easy-to-understand PDS can be crucial to investors’ understanding and choice of superannuation

products. Nevertheless, in the absence of definitive rules governing presentation, superannuation

fund families have a fair amount of discretion in how they frame fund specific information,

provided that the minimum disclosure requirements specified in Corporations Act are satisfied.

Furthermore, it appears that superannuation funds often see disclosure as, at least partly, a liability

minimisation exercise. Before the introduction of the short form PDSs in year 2011, most PDSs were

too long and complex so were unsuitable for general consumption (PJC, 2007). As a result, the

Investment and Financial Services Association (IFSA) called for shorter documents that are presented

well and written in plain English, which was later implemented through the ‘Short PDS Regime’

(IFS, 2007).

This chapter is motivated to examine the short PDS documents under the new short form PDS

regime to gain an understanding of the current disclosure practice of superannuation fund families.

The sample period therefore spans from 2012 to 2015 when the ‘Short PDS Regime’ is in place. Useful

information that helps reduce information interpretation costs needs to (1) contain all important

information that are most relevant to investors, and (2) be presented in an easy-to-understand manner.

Therefore, this chapter assesses both the extent and readability of PDS disclosure in the Australian

superannuation setting.

This is the first study that investigates the attributes of a large sample of Australian

superannuation PDSs in a systematic fashion. Two disclosure measures are used: a subjective

measure (content disclosure score) and an objective measure (readability index). Moreover, much

of the literature on fund disclosure has relied on the research methodology of experiments to

manipulate the content and format of the information disclosed (Foster, Ng & Wee 2015; Kozup,

Howlett & Pagano 2008), and used qualitative methods to document key themes and provide insights

for follow-on quantitative studies (Cheah et al. 2015). In comparison, this thesis uses the actual

disclosure documents and quantitative methods to provide direct evidence on superannuation

disclosure practice. The results carry important implications for investors and regulators with regards

to the current and contentious issue of superannuation disclosure’s transparency and informativeness.

The evidence also helps provide a better understanding of mechanisms that help lower investors’

information costs, and at the same time provide a new perspective to examine whether fund families’

51

managerial incentives affect superannuation disclosure.

5.2 Literature Review and Hypothesis Development

5.2.1 The Need for Quality Superannuation Information

The privatization of the retirement savings system in Australia has increased the general

public’s awareness of the need to make wise decisions related to the choice of superannuation

funds in order to meet long-term personal financial needs. A nationally representative sample of

1,022 full time working Australians aged 25 to 65 years were surveyed in 2009, among which

only 16% rated their knowledge of superannuation as strong, while 54% want to achieve a strong

level of knowledge (Mercer, 2009). Unsophisticated superannuation members have a high level of

demand for informative and understandable information.

The large amount of complex superannuation disclosures, coupled with the proliferation of

investment choices, have made investing in a superannuation product a difficult and confusing

decision-making process. The short form PDS regime was introduced in 2011 due to serious concerns

about the understandability of PDS disclosure. However, transparency was still raised as a serious

concern in the national survey conducted by the Financial Service Council (FSC) and ING Direct in

2013. Among the 1000 surveyed working Australians with superannuation, two-thirds (66%)

agree that superannuation funds are not transparent enough. The less engaged account holders

simply did not know or understand how their funds and fees worked, and the more engaged account

holders felt there was not enough disclosure (FSC, 2013). The effectiveness of superannuation fund

communication and the ability of investors to make informed decisions based on the information

provided by superannuation funds remain a serious concern. All these survey results point to the

importance of effective superannuation fund disclosure.

5.2.2 Experimental Studies on Superannuation Disclosure

Surprisingly, despite the importance of superannuation disclosure to regulators and

superannuation members, there is little empirical evidence on the disclosure practice of

superannuation fund families. A number of prior studies conducted experiments to examine the

impact of attributes of displayed information on individuals instead of examining actual disclosure by

superannuation funds. For example, Kozup, Howlett and Pagano (2008) examine the impact of certain

types of summary information on individual’s attitude towards mutual funds. They find that

summary information increases an individual’s sensitivity to past fund performance. Foster, Ng and

Wee (2015) examine the effect of presentation format of retirement funds’ information on people’s

52

ability to assess and use information to make retirement savings decisions. They find the

incremental benefit of placement of information in more prominent positions is limited to individuals

with moderate numeracy skills, and the authors conclude that there is value accruing from financial

education as advocated by regulators. Both of the aforementioned studies use experimental design

to manipulate the format and location of the information disclosed.

Given the serious concerns about PDS disclosure quality and the importance of disclosure for

informed investment decisions, this thesis extends the literature by examining all short-form

superannuation PDSs available from financial year 2012 to 2015, with two main purposes: First, to

provide exploratory descriptive evidence on current superannuation PDS disclosure quality as

reflected in the readability and extent of the PDS under the short PDS regime, and second, to examine

whether the quality of PDS disclosure is associated with management incentives as proxied by

fund family performance.

5.2.3 Performance and Disclosure

As mentioned above, the second purpose of this chapter is to gain a further understanding of

the current PDS disclosure practice of superannuation fund families by examining the relation between

superannuation fund family performance and PDS disclosure attributes because the propensity to

disclose may be greater for better-performing superannuation families, presumably to promote the

family to attract more return-chasing investors. Prior research on discretionary corporate disclosure

finds that company managers are more forthcoming in making disclosures when their companies are

performing well (Schrand & Walther 2000), but they underplay or ignore negative information

(Brennan, NM, Guillamon-Saorin & Pierce 2009). Miller, GS (2002) examines a comprehensive set of

voluntary company disclosures and finds an increase in the extent of disclosure during periods of

increasing earnings. As the companies begin to report earnings declines, their disclosure decreases and

eventually returns to the level provided in the period of flat earnings performance.

In contrast, the ‘management obfuscation hypothesis’ proposes that managers strategically

increase the processing cost of adverse information in order to reduce or delay its incorporation into

stock price, because market reaction may be delayed or muted by the difficulty of extracting and

understanding information contained in complicated disclosure (Bloomfield 2002). That is, managers

make negative information more complex and difficult to understand. Similarly, the ‘incomplete

revelation hypothesis’ by Grossman and Stiglitz (1980) and Bloomfield (2002) also suggest that

mangers may strategically hide adverse information through less transparent disclosures. Smith and

Taffler (1992) suggest poor readability is associated with poor performance while ease of readability

53

is associated with financial success. The inference is that firms actively signal positive performance

but attempt to obscure messages which convey poor performance. However, some empirical research

produces conflicting findings on this issue. Courtis (1998) assigns readability scores to the Chairman's

Address of Hong Kong companies in 1994/1995, and find evidence consistent with the

obfuscation hypothesis regarding the readability scores of 30 companies with the highest

percentage increases in annual profitability and 30 companies with the largest percentage decreases

in annual profitability. In contrast, Clatworthy and Jones (2006) investigate the narrative reporting

practices of 50 highly profitable and 50 highly unprofitable listed UK companies, and find no

such relation. These two papers are based on samples of extremely profitable or extremely

unprofitable companies and thus only represent the tails of the distribution. Li (2008) provides the

first large-sample evidence that loss firms prepare annual reports that are both longer and less

readable than firms with positive earnings. Among firms with positive earnings, firms with more

transitory earnings prepare longer and less-readable annual reports.

While the influence of performance on disclosure is a fundamental issue in the accounting

disclosure literature, prior literature on corporate disclosure does not provide specific guidance on how

fund family performance affects disclosure by fund managers. The studies discussed above generally

support the management obfuscation hypothesis, but the positive relation observed between disclosure

readability/content and company performance may not apply to the same extent to superannuation fund

families due to a number of major institutional differences between companies and superannuation fund

families. First, superannuation fund family level performance is not directly comparable across fund

families. This is because the investment strategy (e.g. asset allocation, investment horizon and risk

profiles) and the use of performance measures differ among superannuation families. Besides, it is not

compulsory for superannuation families to disclose past performance in their PDSs. Much of the PDS

disclosure involves forward-looking information such as expected performance rather than historical

performance. Hence there is less need to hide adverse historical performance information using more

complicated PDSs for superannuation fund families. Moreover, superannuation fund family’s PDS

disclosure practice tends to be persistent over time. Regulation 7.9.11M of the Corporations Amendment

Regulations 2010 (No.5) states that the shorter superannuation PDS has been designed in a manner that

allows information updates to be incorporated into the PDSs by reference. Superannuation PDSs thus

only have to be amended occasionally and only if there are major changes to the products. Finally,

superannuation is a long-term investment, and members tend to be less sensitive to short term

performance relative to company shareholders, because longer-term performance and investment risks

are more important (as evidenced by the results from chapter 4). In addition, despite the introduction of

54

the ‘Choice of Fund Policy’ in 2005, which enables and encourages superannuation investors to make

more active choices of their superannuation arrangements, investor inertia is still widely observable.

Therefore, superannuation members are much less likely to switch fund families frequently whereas

company shareholders trade in shares much more frequently. This means superannuation fund managers

are under less pressure for short-term performance compared to corporate managers whose job security

and compensation are usually highly dependent on short to medium-term profit and share price

performance, which results in greater incentive for manipulation of disclosures. On the other hand,

superannuation families also need to compete for new investments, which affect fund managers’

compensation (which is determined by asset under management). This means there are still some

incentives for disclosure manipulation although the incentives are not as strong as those for corporate

managers.

The discussions above suggest the relation between superannuation family performance and

PDS readability/content is an empirical question. To examine this issue, this study tests the

management obfuscation hypothesis that superannuation fund families signal superior performance

(obfuscate inferior performance) by greater (less) transparency in the presentation of information

and more (less) content disclosed in their PDSs. The hypotheses are stated as follows:

H3: The extent of PDS disclosure is positively associated with superannuation fund family’s profit

performance.

H4: The readability of PDS disclosure is positively associated with superannuation fund family’s

profit performance.

5.3 Data

Superannuation funds’ PDSs are not readily available from a central archive or repository.

They are hand-collected from multiple sources at annual intervals during the sample period when

the Short PDS Regime is in place from 2012 to 2015. The latest PDSs are downloaded from

superannuation entities’ websites. Superannuation families are contacted by e-mail and phone to

collect historical PDS. When the superannuation families fail to respond to the requests, their

historical PDSs are retrieved from archived webpages where possible. Data on superannuation

family flows and characteristics are obtained from APRA’s public and proprietary superannuation

statistics publications and reports.31

For the purpose of this thesis, it is necessary to select funds with accumulation and hybrid

31 The author would like to acknowledge APRA’s assistance in providing proprietary superannuation data.

55

style benefits, where members are responsible for their own investment decisions. As discussed in

Chapter 2, superannuation families coordinate actions across funds in the family. Disclosure is a

family-level decision, as evidenced by consistent disclosure format and content across different funds

in the same fund family. Therefore, this thesis again focuses on the role of superannuation family

in this chapter and examines family disclosure strategies instead of fund-specific incentives.

5.4 Methodology

5.4.1 Readability Measures

This study first uses conventional computational linguistic readability tests to obtain the Flesch

Kincaid Ease Test Score (Flesch Score) and Gunning Fog Index (Fog Index). 32 The use of

computational linguistics indices in recent literature offers several advantages. First, these indices

enable the study of a large and diverse sample as opposed to small bodies of text or case-by-case analysis

adopted in prior research. Second, is the indices provide an objective measure generated by

computational linguistic software, rather than surveys or researcher’s opinions. The Flesch Score is the

most commonly used measure when a readability test is specified in laws or regulations in the U.S.

Higher scores indicate material that is easier to read. The formula for the Flesch Score test is:

𝑁𝑁𝑁𝑁𝑜𝑜𝑜𝑜𝑜𝑜ℎ 𝑆𝑆𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 = 206.835 − 1.015 ∗ �𝑁𝑁𝑜𝑜𝑜𝑜𝑤𝑤𝑜𝑜

𝑜𝑜𝑜𝑜𝐿𝐿𝑀𝑀𝑜𝑜𝐿𝐿𝑜𝑜𝑜𝑜𝑜𝑜�− 84.6 ∗ �

𝑜𝑜𝑠𝑠𝑁𝑁𝑁𝑁𝑜𝑜𝑏𝑏𝑁𝑁𝑜𝑜𝑜𝑜𝑁𝑁𝑜𝑜𝑜𝑜𝑤𝑤𝑜𝑜 � 】(3)

where Flesch Score between 90 – 100 is interpreted as ‘easily understood by an average 11-year-old

student’; 80 – 90: ‘easy to read. Conversational English for consumers’; 70 – 80: ‘fairly easy to read’;

60 – 70: ‘plain English easily understood by 13– to 15–year–old students’; 50 - 60: ‘fairly difficult to

read’; 50 – 30: ‘best understood by university students’; and 0 – 30 indicates ‘very difficult to read. Best

understood by university graduates’.

The Fog Index has been widely used in recent accounting studies to examine the readability of

company annual reports in connection with earnings persistence (Li 2008), investment efficiency

(Biddle, Hilary & Verdi 2009), and analyst behaviour (Lehavy, Li & Merkley 2011). The Fog Index is

interpreted as the number of years of formal education required for a person of average intelligence to

read the document once and understand it. The Fog Index is calculated with the following algorithm:

32 Two other popular measures of readability are: Coleman and SMOG scores. Sensitivity tests are performed using these alternative measures and the empirical results are similar to those based on the Fog Index and Flesch Score, so they are not reported for the sake of brevity.

56

𝑁𝑁𝑜𝑜𝐹𝐹 𝐼𝐼𝐿𝐿𝑤𝑤𝑜𝑜𝑀𝑀 = 0.4 ∗ 【𝑁𝑁𝑜𝑜𝑜𝑜𝑤𝑤𝑜𝑜

𝑜𝑜𝑜𝑜𝐿𝐿𝑀𝑀𝑜𝑜𝐿𝐿𝑜𝑜𝑜𝑜𝑜𝑜+ 100 ∗ �

𝑜𝑜𝑜𝑜𝑐𝑐𝑀𝑀𝑁𝑁𝑜𝑜𝑀𝑀 𝑁𝑁𝑜𝑜𝑜𝑜𝑤𝑤𝑜𝑜𝑁𝑁𝑜𝑜𝑜𝑜𝑤𝑤𝑜𝑜 � 】 (4)

where complex words are defined as words with at least three syllables. The relation between Fog

Index and reading ease is as follows: as follows: Fog Index ⩾ 18 means the text is unreadable; 14-18

means the text is difficult; 12-14 means the text is ideal; 10-12 means the text is acceptable; and 8-10

means the text is childish.

To avoid distortion of the readability scores, this study follows (Li 2008) and removes white

spaces, headings, paragraphs of less than one line, all tables and tabulated text from the PDS before

computing readability scores. The calculation of PDS readability is based on the text rather than on

the numbers or tables. The file after the editing is then analysed using the Fathom package in Peal

to generate typical text statistics.

Prior studies also describe disclosures as less readable if they are longer (Li 2008; Rutherford

2003). Hence, the related readability studies also use the length of the disclosure document to

measure the complexity of the disclosure. However, the short PDS is legally required to have eight

pages, so the length does not vary across fund families. Hence, this thesis focuses on the two

linguistic readability measures described above.

5.4.2 Content Analysis Score

Linguistic style captures the methods used to convey meaning to the audience, whereas

linguistic content of disclosure is the literal meaning of the information conveyed, or the concrete

facts contained in a disclosure. A high-quality PDS not only needs to be easy to understand but also

needs to contain information that is useful to investors.

Corporations Regulations 2001 section 1013C (3) of Part 7.9 states that ‘the information

included in the PDS must be worded in a clear, concise and effective manner’. There was no limit

on the length of a PDS until the Corporations Amendment Regulations 2010 (No.5) prescribe a

maximum length of eight (A4) pages for superannuation PDSs. These regulations amend the Principal

Regulations with the aim to make PDSs of superannuation products substantially shorter, simpler

and more readable. In addition, the regulations prescribe the section headings and basic content

of the shorter PDS to ensure that consumers are provided with sufficient relevant information to

make an informed investment decision. The prescribed content covers key information such as

fund features and benefits, risks, taxations, investment options, and fees and costs. The specific

form and content of a superannuation PDS are prescribed in Schedule 10D of the Principal

57

Regulations.33

Accordingly, to capture the extent of PDS disclosure, this study uses content analysis to

compile a disclosure index based on the extent and quality of information disclosed in the PDSs.

The selection of the items included in the index is guided by Corporations Amendment Regulations

2010 (No. 5) Schedule 10D, the Australian Securities and Investment Commission’s (ASIC)

Regulatory Guide 168 and Information Sheet 155, the Financial Services Council’s (FSC) standard

risk measure guidance and Super System Review, Chapter 4. The items included in the disclosure

index reflect information identified by regulators as useful in superannuation investment decision

making.

The validity of the disclosure items selected in the disclosure index has been checked by a

team of experts from the Australian Securities and Investment Commission (ASIC). ASIC is the

regulatory body that monitors and assesses all PDSs issued by superannuation entities, and it

provides further guidance through the ASIC regulatory guides. Comments from the ASIC and some

superannuation fund families are also taken into consideration in the development of the disclosure

index.

The disclosure index consists of ten broad categories, consistent with Corporations

Amendment Regulations 2010 (No. 5) Schedule 10D (see Appendix 1). A list of the ten parts is as

follows: 1. About XX super; 2. How super works; 3. Benefits of investing with XX super; 4. Risks

of super; 5. How we invest your money; 6. Fees and costs; 7. How super is taxed; 8. Insurance in

your super; 9. How to open an account; 10. Other. Overall, 111 criteria (items and sub-items) from

the scorecard are used to assess each superannuation family’s PDS in the sample. Complete content

analysis scoring is available from the author upon request.

A score of zero is coded for non-disclosure and one point is given for meeting the minimum

compliance standard prescribed by regulations. One additional point is given if the information

provided is ‘hard information’ that exceeds the minimum requirements, such as a numerical example

of age-based asset allocation and investment strategy for the default option, or industry-

benchmarked past performance data. Multiple points are not awarded for multiple references to the

same disclosure item. Specific quantitative information is weighted more heavily in the disclosure

index (i.e., it is worth an additional point) than general qualitative information because precise

information is usually more useful and credible. Overall, 199 superannuation PDSs in the sample

are manually scored based on this scoring system. To reduce the subjectivity of the assessment, all

33 See Appendix 1 for detailed information.

58

items for every superannuation family are rated twice and cross checked by different raters.

5.4.3 An example of scoring: Smartsave versus Suncorp Super

To provide some insights into the scoring procedure, this section demonstrates the PDS

scoring system using the 2014 PDSs of Smartsave and Suncorp Super. Both Smartsave and Suncorp

Super are public retail superannuation fund families. As of 30 June 2014, Smartsave (Suncorp

Super) has $280 million ($6,409 million) in assets and 3,873 (Smartsave) and 193,725 (Suncorp

Super) members. Part 2 (How super works) from both superannuation families’ PDSs and the

disclosure index for part 2 of the PDS, together with the scores awarded, are shown as follows:

59

Figure 5.1 Part 2 of Smartsave’s PDS in 2014

Figure 5.2 Part 2 of Suncorp’s PDS in 2014

60

Figure 5.3 PDS Disclosure Index for Part 2

Part 2. How super works Smartsave Suncorp

1. Superannuation is in part compulsory and is a way of building savings for retirement

1

1

2. There are different ways to accumulate superannuation funds

2.1 Employer contributions (compulsory Superannuation Guarantee) 1 1 2.2 Voluntary contributions

2.2.1 Salary sacrifice 1 2.2.2 Voluntary contributions from after-tax salary (regular or lump-sum

)

1 2.3 Government co-contributions 1 1 2.4 Spouse contributions 1

2.5 Self-employed contributions

2.6 Investment growth 1 2.7 Rollover from other superannuation fund families

3. Why superannuation should be boosted 1 4. Tax savings are provided by the government 1 1 5. There are limitations on contribution each financial year to receive a tax concession

1

1

6. Limitations on contributions 1 1 7. Investors have a choice of fund family into which they direct their superannuation guarantee contributions

1

1

8. Superannuation funds consolidation

8.1 Investors can consolidate their superannuation accounts 1 8.2 Why they need to consolidate (e.g., less fees etc.) 1 8.3 How to consolidate (e.g., help from customer service staff, alone using

an online account, etc.)

1 8.4 Things to consider before consolidation decisions(e.g., exit fees, tax

situation, benefits such as employer contribution lost, etc.)

1 9. Limitations on withdrawals 1 10. Types of withdrawals: annuity and lump-sum withdrawals 1 1 Total score for part 2 9 17

The type of information disclosed in this section (How super works) is general in nature, but

differences are observed in the content analysis scores. Smartsave is awarded a score of 9 with

some basic information about what is superannuation (1), contributions to (2.1, 2.3, 2.4, 4,5 ,6, 7)

and withdrawals from (10) superannuation accounts. It is noted that Smartsave’s information is

described in manner that is almost identical to the ASIC’s regulatory guidance. In contrast, the same

items (1, 2.1, 2.3, 4, 5, 6, 7, 10) are rephrased and discussed in a question and answer (Q&A) format

by Suncorp Super. In addition, Suncorp has provided information about how to grow one’s account

balance (2.2.1, 2.2.2, 2.6) and why it is important to boost superannuation assets (3). Moreover,

Suncorp Super provides detailed information (8.1, 8.2, 8.3, 8.4) about the consolidation of super

61

accounts, which can be used not only to encourage existing members to consolidate multiple

accounts and thereby minimise fees (8.2) but also to attract potential investors to join Suncorp

Super given the ease of consolidation (8.3). As a result, Suncorp Super achieves a total score of 17

for the information disclosed, whereas Smartsave received a total score of 9.

Turning to the readability of the information, although a higher level of information is

provided by Suncorp Super, understandability of disclosure is not compromised. At a glance,

Suncorp Super adopts the Q&A format to address technical superannuation issues in an interactive

and intuitive way. When readability tests are performed on the two PDSs used in this section,

Smartsave obtains a Flesch score (Fog index) of 59.6 (10), whereas Suncorp Super has a Flesch

score (Fog index) of 75.1 (7.4), indicating that part 2 of Suncorp Super’s PDS is both more

informative and easier-to-understand.

5.4.4 Regression model for hypotheses testing

As discussed above, the second objective of this chapter is to examine the relation between

fund family performance and disclosure readability/extent in a multivariate regression setting.

Hypotheses 3 and 4 are tested using the following equation:

𝑅𝑅𝑜𝑜𝑜𝑜𝑤𝑤𝑜𝑜𝑏𝑏𝑜𝑜𝑁𝑁𝑜𝑜𝑀𝑀𝑠𝑠 𝑆𝑆𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑜𝑜𝑜𝑜 𝐶𝐶𝑜𝑜𝐿𝐿𝑀𝑀𝑜𝑜𝐿𝐿𝑀𝑀 𝑆𝑆𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜 𝑖𝑖,𝑡𝑡 = 𝑜𝑜 + 𝑏𝑏1𝑃𝑃𝑜𝑜𝑜𝑜𝑃𝑃𝑖𝑖,𝑡𝑡−1 + 𝑏𝑏2𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖

+ 𝑏𝑏3𝐼𝐼𝐿𝐿𝐼𝐼𝐼𝐼𝑀𝑀𝑀𝑀𝑖𝑖,𝑡𝑡 + 𝑏𝑏4𝐿𝐿𝐼𝐼𝑀𝑀𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑃𝑃𝑜𝑜𝑜𝑜𝐶𝐶𝑜𝑜𝐿𝐿𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝐿𝐿𝑜𝑜𝑀𝑀𝐿𝐿𝑜𝑜𝑁𝑁𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (5)

Fund family performance ( 𝑃𝑃𝑜𝑜𝑜𝑜𝑃𝑃 ) is measured in the same way as described in section 4.3.6.

Ex ante, there are many factors that might affect PDS readability and content. It is important and useful

to both regulators and investors to gain a better understanding of the factors (including fund families’

characteristics) that are associated with the quality of PDS disclosure. This study focuses on the role

of fund performance, as return chasing behaviour has been widely observed among unsophisticated

retail investors in management funds literature. On the other hand, it is not known whether

superannuation families’ institutional characteristics are key determinants of PDS disclosure quality, as

PDS disclosure is mandatory and there is a large amount of standard prescribed information. In addition,

a lot of the information contained in the PDS document is ‘soft’ information that can be imitated by

peers easily.

Control variables are included in the regression model as follows:

Size of fund family (𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿) - size is calculated as the logarithm of the total net assets of the

superannuation family to capture aspects of a superannuation fund family’s operational environment.

The natural logarithm is used due to the diminishing returns to scale effect. In corporate disclosure

literature, size can be viewed as a proxy for litigation risk or for other factors such as economies of scale

62

in raising fund visibility and the amount of resources available for making disclosure (e.g., Lang and

Lundholm, 2000). Ex ante, it is expected that large superannuation families to have higher-quality PDS

(in terms of both readability and extent).

Breadth of investment options (𝐼𝐼𝐿𝐿𝐼𝐼𝐼𝐼𝑀𝑀𝑀𝑀) – This factor represents the diversity of the products

offered by a fund family and affects the amount and complexity of disclosure required to explain the

products. This study uses the natural logarithm of the total number of products offered by a fund family

as a control for product complexity in light of the large variation in the number of product offerings

across different superannuation fund families. Although the marginal cost of preparing information

decreases when the span of products is expanded, more complex products are inherently more difficult

to describe in fewer and more readable words. Thus, this study does not form a directional expectation

on the relation between the breadth of offerings and PDS readability. A positive relation is expected

between product offerings and the extent of disclosure.

Other control variables, namely, Total operating expense (TOE), Personal contribution

(PerCon) and Number of new members (TotNew). Total operating expense represents resources

that can be devoted to operating activities. It is also used as a proxy for marketing efforts. 34This

variable is expected to be positively associated with disclosure readability/content. Members’ personal

contributions and growth in new members are included as control variables because they are also

proxies for fund family size and the demand for better quality disclosure, so a positive relation is

expected between these variables and disclosure readability/content. In addition, year and fund family

type (e.g., retail and industry fund family types) effects are also controlled for as their features differ.

5.5 Univariate Results

5.5.1 Summary Statistics

This section presents the descriptive statistics on superannuation fund families’ PDS

disclosure readability and content to help provide an understanding of current PDS disclosure

practice under the short PDS regime. The information provided in this section is the first of its kind

to examine the landscape of Australian superannuation product disclosure practice.

This section presents the descriptive statistics on superannuation fund families’ PDS

disclosure readability and content to help provide an understanding of current PDS disclosure

34 Chapter 5 and Chapter 6 do not use the same marketing expense measure as Chapter 4, due to APRA’s change in its superannuation reporting framework.

63

practice under the short PDS regime. The information provided in this section is the first of its kind

to examine the landscape of Australian superannuation product disclosure practice.

Panel A of Table 5.1 presents the summary statistics of all sample superannuation PDSs.

The mean Flesch scores fall into the band of 50 - 60, which suggests that the material is ‘fairly

difficult to read’. The average Fog index (12.43) indicates that reading ease is ‘ideal’ (between 12 -

14). Text for a wide audience generally needs a Fog index of less than 12 to be considered easily

readable. The standard deviation of the Flesch (Fog) scores of the PDSs in the sample is 5.93 (1.22).

This suggests some variation in the readability scores for sample PDSs. The Flesch score ranges

from 37.6 (best understood by university students) to 65.7 (plain English easily understood by 13-

to 15-year-old students). The Fog score ranges between 10.1 (acceptable) and 15 (difficult to read).

To provide a benchmark for comparison, the difference in the Flesch scores between Reader’s

Digest and TIME magazine is approximately 2.

64

Table 5.1 Summary statistics of PDS readability and content analysis scores

This table reports summary statistics of readability scores/content analysis scores of superannuation fund families’ PDSs for the entire sample

period from financial year 2012 to financial year 2015. Two readability measures, the Flesch Kincaid Ease Test Score (Flesch) and the Gunning

Fog Index (Fog), are used for all superannuation fund families/retail superannuation fund families/industry superannuation fund families. A

higher Flesch Kincaid Ease Test Score (Flesch) indicates material that is more readable; a lower Gunning Fog Index (Fog) indicates material that

is easier to read. A higher content analysis score (measuring disclosure content) indicates that the content disclosed is of better quality. T-tests are

conducted for univariate differences between the means for the retail and industry subsamples. ***, ** , and * denote significance at the 1%, 5%

and 10% level, respectively.

Panel A: All superannuation fund families

PDS disclosure measure N Mean Std. Dev. Min Max

Flesch 199 52.85 5.93 37.6 65.7

Fog 199 12.43 1.22 10.1 15

Content 217 55.48 13.79 0 86

Panel B: Retail superannuation fund families

PDS disclosure measure N Mean Std. Dev. Min Max

Flesch 94 50.57 6.6 37.6 65.7

Fog 94 12.82 1.28 10.2 15

Content 115 48.22 13.44 0 81

Panel C: Industry superannuation fund families

PDS disclosure measure N Mean Std. Dev. Min Max

Flesch 76 56.05*** 3.84 47.4 63.4

Fog 76 11.87*** 0.87 10.1 14.3

Content 89 63.80*** 8.91 45 86

Panels B and C of Table 5.1 present the readability scores of the PDSs of retail

superannuation fund families and industry superannuation fund families. Retail superannuation

families’ PDSs are slightly more difficult to read than the overall sample, with the mean Flesch

(Fog) being slightly lower (higher) than the overall sample at 50.57 (12.82); however, these scores

still fall into the same categories as the overall sample, i.e., the disclosure is relatively difficult to

read based on its Flesch score, and ideal based on its Fog score. The variation in the retail families’

Fog scores (1.28) is similar to that for the entire sample (1.22). The industry families’ PDSs are

easier to read than either both the overall sample or the retail families with a higher Flesch score of

56.05 (fairly difficult to read) and a Fog score of 11.87 (acceptable). The variation in the industry

families’ PDS readability is much lower than that for the entire sample and for retail families with a

standard deviation of 3.84 (0.87) for the Flesch (Fog) score.

Turning to the extent of disclosure, the number of observations for content analysis (217) is

65

higher than that for readability scores (199) because some of the PDSs files are ‘read-only’ files and

cannot be edited by the linguistic software. The mean content score is 55.48 out of 111 marks for all

the PDSs. A comparison of means between the retail and industry subsamples reveal that industry

families prepare better quality PDSs with a mean content score of 63.80 compared to retail families’

PDSs, which have a mean content score of 48.22. Both readability scores and content scores are

statistically significance between retail and industry families. As is the case with the readability

scores, industry families show less variation in the quality of disclosure than the overall sample and

retail families, with a standard deviation of 8.91 compared to 13.79 and 13.44. This suggests

industry families more uniformly provide better quality PDS disclosure than retail families.

5.5.2 Disclosure Scores by Financial Year

Table 5.2 reports the means of the Flesch and Fog scores of the PDSs for the sample

superannuation families by financial year. There are fewer observations in financial year 2012

because the use of short PDS is not compulsory unless the superannuation fund families issue new

products in the first year of the short PDS regime. Interestingly, there is a statistically significant

decrease in the Flesch (increase in Fog) scores in the following year when the ‘short PDS regime’

was fully enforced. This indicates that the first group of short PDSs (i.e., those issued in 2012) are

easier to read than the subsequent group of short PDSs. Both Flesch and Fog scores are fairly stable

for the overall sample from 2013 to 2015, with only minor fluctuations in the Flesch score,

whereas the Fog score improves very slightly over the same time period. Similar patterns are

observed for the industry families. In comparison, both Flesch and Fog scores improved slightly

from 2013 to 2015 for retail families. In each year, the industry families have better readability

scores than retail families, e.g., the Fog scores for industry families fall into the category of

‘acceptable’, whereas the Fog scores for retail families fall into the more difficult-to-read category of

‘ideal’.

66

Table 5.2 Mean Readability Scores by Financial Year

This table reports mean readability scores of superannuation fund families’ PDSs by year from financial year 2012 to 2015. Two readability

measures, the Flesch Kincaid Ease Test Score (Flesch) and the Gunning Fog Index (Fog), are used for all superannuation fund families/retail

superannuation fund families/industry superannuation fund families. A higher Flesch Kincaid Ease Test Score (Flesch) indicates material that is

more readable; a lower Gunning Fog Index (Fog) indicates material that is easier to read. T-tests are conducted for univariate differences between

the means for the retail and industry subsamples. ***, ** , and * denote significance at the 1%, 5% and 10% level, respectively.

All Families

All Families

All Families

Retail Families

Retail Families

Retail Families

Industry Families

Industry Families

Industry Families

Year N Flesch Fog N Flesch Fog N Flesch Fog

2012 15 55.01 11.87 4 52.53 12.05 9 56.70 11.70

2013 54 52.59* 12.58*** 27 49.56 13.02** 19 56.72 11.93

2014 69 52.55 12.52 33 50.50 12.90 26 55.83 11.92

2015 61 52.87 12.34 30 51.29 12.66 22 55.46 11.83

Table 5.3 presents the mean level of content analysis scores for each part of the PDS and the

entire PDS. A higher content score indicates that more and better (e.g., more precise) information is

disclosed. Panel A shows that for the overall sample, total content score decreased from 58.95 in

2012 (with a smaller sample size) to 55.42 in 2013 and, then fluctuated slightly around 55 and 56

during 2014 and 2015.

The results from Panel B show that the amount/quality of content disclosed for the retail

families decreased over the sample period as the content score decreased from 51.86 to 48.92,

indicating that retail families’ PDSs are providing less information over time. This decreasing trend

appears to be attributable to the decrease in scores for part 1 (About XX super) and part 5 (How we

invest your money) over the sample period.

Panel C shows that industry families’ PDSs content scores are better than those of retail

families’ for all parts of the PDS, except for part 4 (Risks of Super) and part 9 (How to open an

account). Total content score for the industry families is higher than for the retail families throughout

the sample period, with the difference in total scores ranging from 10.97 points in 2012 to 15.76

points in 2015. This is because the total content score for the retail subsample showed a consistent

decreasing trend, whereas the content disclosed by industry families improved from

62.77 in 2013 to 64.68 in 2015, suggesting that the industry families are providing more (or better

quality) information in their PDSs over the sample period. This general trend of improvement is

evident from 2013 to 2015 in several parts of the PDSs (i.e., parts 1, 2, 3, 4, 6 and 10).

67

Table 5.3 Mean Scores by Financial Year

This table reports mean content analysis scores of superannuation fund families’ PDSs by year from financial year 2012 to 2015. The content

analysis scores of each part (p1, p2, p3, p4, p5, p6, p7, p8, p9 and p10) in the PDS and the total content analysis score (Content) for all superannuation

fund families/retail superannuation fund families/industry superannuation fund families are reported. A higher score indicates more information or

better quality information is disclosed.

Panel A: All superannuation fund families Year N p1 p2 p3 p4 p5 p6 p7 p8 p9 p10 Content

2012 22 4.14 11.45 7.23 2.95 10.73 6.36 5.45 6.14 2.95 1.55 58.95

2013 56 3.3 10.45 6.25 2.96 10 6.16 5.84 5.88 2.98 1.59 55.42

2014 74 3.5 10.85 6.72 3 9.7 6.52 5.62 6.04 3.01 1.5 56.49

2015 65 3.51 10.86 6.48 2.89 9.51 6.46 5.72 5.72 3.02 1.62 55.78

Maximum score 11 20 13 6 21 9 10 12 5 4 111

Panel B: Retail superannuation fund families

Year N p1 p2 p3 p4 p5 p6 p7 p8 p9 p10 Content

2012 7 3.29 9.57 5.86 3 9.86 6.43 5.29 4.43 2.86 1.29 51.86

2013 31 2.45 9.32 5.13 3.03 9.13 6.19 5.55 4.68 2.97 1.39 49.84

2014 38 2.42 9.24 5.63 3.18 8.53 6.58 4.95 4.61 3 1.24 49.37

2015 36 2.39 9.39 5.5 3.03 8.22 6.42 5.25 4.36 3.03 1.33 48.92

Maximum score 11 20 13 6 21 9 10 12 5 4 111

Panel C: Industry superannuation fund families

Year N p1 p2 p3 p4 p5 p6 p7 p8 p9 p10 Content

2012 12 4.5 12.83 7.92 2.75 10.92 6.25 5.83 7.17 2.92 1.75 62.83

2013 23 4.35 11.87 7.61 2.74 11 6.17 6.35 7.48 2.96 1.91 62.77

2014 31 4.71 12.39 7.97 2.74 10.74 6.47 6.39 7.65 2.97 1.9 64.22

2015 25 4.96 12.56 8.04 2.76 11.08 6.52 6.32 7.44 2.96 2.04 64.68

Maximum score 11 20 13 6 21 9 10 12 5 4 111

Table 5.4 reports the standard deviation of PDS readability and content scores by financial

year. During the transitory period (from 2012 to 2013), as more families started to prepare short-

form PDSs, the variation of readability increased, whereas the variation of content scores decreased

for the overall sample. From 2013 onwards, there is a general trend of decline in the variation of

PDS readability (reflected in both Flesch and Fog scores) for the entire sample, whereas the

variability of content scores fluctuated between 11.49 and 14.70. These patterns are not observed

for the retail families, which have increasing variability in their Flesch scores (increasing from 4.19

in 2012 to 6.92 in 2015) and decreasing variability in their content scores (with a substantial

decrease from 19.88 in 2013 to 11.17 in 2015). These results suggest retail families’ PDS disclosure

readability has become more diverse over time, whereas PDS content and quality have become

much more uniform. In contrast, the standard deviation of industry families’ Flesch scores decreased

from 4.36 in 2013 to 3.70 in 2015, indicating fewer variations in these families’ PDS readability.

68

This is consistent with the trend observed for the overall sample. The Fog score is more stable and its

variability only fluctuated slightly between 1 and 0.82 between 2013 and 2015. The variability of

industry families’ content scores shows a declining trend that is similar but more moderate than

that of the retail families, decreasing from 9.30 in 2013 to 8.56 in 2.15. These trends suggest that both

retail and industry families’ PDSs are becoming increasingly alike, raising concerns that the

standardised format and prescribed content of the short PDSs may have substantially decreased the

extent of fund family specific information available to investors.

Table 5.4 Standard Deviations of Scores by Financial Year

This table reports the standard deviations of readability scores (Flesch and Fog) and content analysis scores (Content) for all superannuation fund

families/retail superannuation fund families/industry superannuation fund families’ PDSs by year from 2012 to 2015.

All superannuation fund families Std. Dev. by year

Measures 2012 2013 2014 2015

Flesch 4.41 6.39 5.97 5.79

Fog 0.80 1.28 1.23 1.20

Content 15.25 11.49 14.70 12.58

Retail superannuation fund families Std. Dev. by year

Measures 2012 2013 2014 2015

Flesch 4.19 6.41 6.80 6.92

Fog 0.47 1.23 1.36 1.29

Content 18.94 19.88 14.79 11.17

Industry superannuation fund families Std. Dev. by year

Measures 2012 2013 2014 2015

Flesch 3.63 4.36 3.74 3.70

Fog 0.60 1.00 0.82 0.96

Content 9.18 9.30 9.15 8.56

5.5.3 Correlation

Pearson correlations for all the variables specified in Chapter 5 and Chapter 6 are reported in

Table 5.5. There is a statically significant correlation between the Flesch and Content scores with a

correlation coefficient of 0.483, which suggests PDSs with more information provided are also easier to

read. Overall, the correlations among the test variables range from -0.484 to 0.485, except for the high

69

correlations between Personal Contribution (PerCon) and Inflow (0.595), and the percentage of default

asset (DefAsset) and the number of investment options (LnInvOpt) (-0.535). PerCon and Inflow are not

used as independent variables in the same equation, so multicollinearity is not an issue for them. Results

with DefAsset and LnInvOpt are robust to sensitivity tests of excluding either one of them in the

regressions.

Table 5.5 Correlation Matrix The Pearson correlations are reported in the lower diagonal and the correlations reported in bold are significant at the 5% level or less. Inflow is APRA-

reported inward transfers scaled by superannuation fund family size. Flesch represents Flesch Kincaid Ease Test Score. Content represents the Content

Analysis Score. All the other variables are defined in Table 4.5.

Inflow Flesch Content Perf LnTNA LnInvOpt TOE PerCon TotNew DefAsset Above50

Inflow 1

Flesch -0.188 1

Content -0.113 0.483 1

Perf -0.060 0.003 0.122 1

LnTNA -0.108 0.291 0.305 0.262 1

LnInvOpt 0.133 -0.300 -0.333 0.164 0.417 1

TOE -0.067 0.273 0.405 -0.002 -0.167 -0.073 1

PerCon 0.595 -0.251 -0.227 -0.002 0.111 0.382 0.061 1

TotNew -0.040 0.238 0.194 0.082 0.449 0.104 -0.026 -0.061 1

DefAsset -0.191 0.339 0.412 -0.028 -0.120 -0.535 0.057 -0.310 0.117 1

Above50 0.222 -0.484 -0.459 -0.037 -0.081 0.320 0.006 0.485 -0.244 -0.363 1

5.6 Multivariate results

5.6.1 PDS Disclosure and Performance

Table 5.6 presents the results of regressing the Flesch (Content) scores on fund family profit

performance. Flesch score is used in the analysis as a higher Flesch score represents more readable

material (i.e., higher readability), which is easier to interpret than the Fog index (where a higher score

indicates the material is less readable). The interpretation for Flesch score is also consistent with the way

content scores are interpreted. In all regressions, the variables discussed in Chapter 5.6.2 as potential

determinants of PDS disclosure (other than fund family performance) are included as control variables.

The results without these control variables are not reported but are of similar economic magnitude and

statistical significance. All the standard errors are clustered at fund family type level in order to control

for within-family-type correlation of the PDS disclosure quality.

70

Table 5.6 Regression of the relation between fund family performance and PDS disclosure readability/content

This table examines the determinants of superannuation fund family PDS readability/extent of disclosure. Flesch Kincaid Ease Test Score (Flesch) is used

to measure the readability of the material. Content analysis score (Content) is used to measure the extent of disclosure. A higher Flesch indicates the

material disclosed is more readable. A higher Content indicates the material disclosed is of better quality. Return is the APRA-reported superannuation

fund family level annual return. LnTNA is the natural logarithm of net asset under management. LnInvOpt is the natural logarithm of the number of

investment options offered by a superannuation fund family. Total operating expense is reported as TOE. PerCon is the personal member contributions

scaled by TNA. TotNew is the number of new members during the year. Robust p-values are reported in parentheses. ***, ** , and * denote significance

at the 1%, 5% and 10% level, respectively.

Current return and PDS readability Current return and PDS content

(1) All (2) Retail (3) Industry (4) All (5) Retail (6) Industry

Dependent variable Flesch Flesch Flesch Content Content Content

Perf 10.461 -8.091 40.800* -35.554 -74.148 76.525 (0.414) (0.609) (0.097) (0.422) (0.164) (0.482)

LnTNA 0.524 1.193* (0.133) 1.598 1.928 1.376

(0.317) (0.093) (0.880) (0.170) (0.207) (0.580)

LnInvOpt -1.137** -1.688*** -0.395 -1.314 -1.676 -0.765

(0.02) (0.004) (0.828) (0.256) (0.222) (0.839)

toe 0.012 0.026 -0.094 -0.001 0.003 -0.004 (0.303) (0.282) (0.185) (0.976) (0.875) (0.960)

PerCon -34.777 -27.027 -383.298** -52.279 -55.185 321.259 (0.180) (0.236) (0.014) (0.373) (0.367) (0.555)

TotNew 0.000 0.000 0.000* 0.000 0.000 0.000 (0.440) (0.56) (0.075) (0.444) (0.669) (0.812)

Constant 52.306*** 50.052*** 59.767*** 49.338*** 42.178*** 39.743* (0.000) (0.000) (0.000) (0.000) (0.000) (0.067)

Observations 133 61 52 146 75 61

Adjusted R-squared 0.292 0.150 0.171 0.315 0.094 0.085

Year Fixed Effect Yes Yes Yes Yes Yes Yes

Family Type Effect Yes No No Yes No No

Contrary to the prediction of hypotheses 3 and 4, fund families’ profitability (Perf)is not

statistically significantly associated with Flesch scores (column 1) or Content scores (column 4).

Contrary to the corporate disclosure literature which tends to find a strong relation between profit

performance and disclosure attributes, the results here suggest that superannuation fund families’ profit

performance does not affect the readability or the quality of content of PDS disclosure. Separating the

PDSs into retail and industry subsamples shows that the relation between performance and the Flesch

score is marginally significant at the 10% level for industry PDSs with a positive coefficient of 40.8.

This provides some support for hypothesis 3 which suggests better fund family performance is

associated with better PDS readability. As for the control variables, a significant negative relation is

found between the number of investment options (LnInvOpt) and Flesch scores for the overall sample

71

and the retail families, which confirms the expectation that more complex products are more difficult

to describe. Little other statistically significant relation is found between PDS disclosure and

superannuation families’ characteristics except for the negative and significant relation between PerCon

and Flesch scores for the industry families. This negative relation is inconsistent with the expectation

that larger fund families (as proxied by more personal contributions) produce easier to read disclosure.

The number of new members (TotNew) is statistically significant at the 10% level for industry families

but the coefficient of 0 lacks economic significance. For the regression tests for performance and PDS

content (column 4 to 6), no statistically significant relation is found for any of the variables. One possible

explanation for these results is that PDS documents have become highly standardised and therefore

lacks sufficient variations for fund family characteristics to affect PDS content.

In summary, inconsistent with the expectation that superannuation fund families intentionally

highlight superior performance to magnify good news or obfuscate poor performance in order to

mitigate negative reactions to bad news, there is no association between performance and PDS

disclosure except for the finding that better performance is associated with greater readability for

industry fund families. This result is different from the findings of prior research on company disclosure.

5.6.2 Additional Analysis

Similar to most studies on disclosure, it is difficult to establish causality between current

performance and PDS readability/content. Superannuation families may have incentives to make the

PDSs more difficult to read or to hide important information if good performance of the current period

is transitory or if poor performance is consistent. To address this issue, this study regresses one-year

ahead return on the current year’s return, the Flesch and content scores, and their interactions using a

sample of all family-years with positive profit in both year t and year t-1. The analysis is limited to

profitable superannuation fund families due to insufficient number of observations for loss-making

superannuation families. The interaction term captures the change in performance persistence as PDS

readability/content changes. The results are not tabulated as marginal evidence is found that PDS

disclosure quality affects the persistence of profits. Similar results are observed when the performance

level is replaced with a profit/loss dummy, which equals 1 if a superannuation fund family reports a

positive return and 0 otherwise. Using the ranking of performance instead of raw performance also

yields similar results.

72

5.7 Conclusion

This chapter examines the current super PDS disclosure practice as reflected in PDS readability

and quality under the new short form regime given the importance of effective PDS disclosure for

investment decisions. This is the first empirical investigation of superannuation fund PDS. Specifically,

the lexical feature of the PDSs and the extent of information disclosed in the PDSs are examined using

computational readability scores and content analysis index.

The results show that the overall reading ease of the PDSs analysed are classified as either

‘ideal’ (which is less readable than ‘acceptable’) or ‘fairly difficult to read’ based on commonly used

readability measures. On average, industry superannuation fund families provide better PDSs as

opposed to their retail counterparts, in terms of both readability and the extent of disclosure. Since

the short PDSs were first introduced, retail superannuation fund families PDSs are found to have

deteriorated in terms of both readability and content from 2012 to 2013 when the Short PDS

Regime was first introduced. The readability of retail superannuation families PDSs then improved

slightly from 2013 to 2015. Industry superannuation fund families’ PDSs have improved in both

dimensions of readability and content, except the Flesch score which deteriorated slightly over time

for industry families. Nevertheless, a trend worth noting is that superannuation families PDSs are

becoming increasingly alike, as indicated by decreasing variation in the two disclosure measures for

the PDSs. Contrary to the ‘management obfuscation hypothesis’ or ‘good news hypothesis’, this

thesis finds no significant relation between superannuation fund family’s performance and the

readability and extent of PDS disclosure.

73

Chapter 6 - Product Disclosure Statement and Member Choice

6.1 Introduction

This chapter examines the empirical relationship between the attributes of superannuation

fund family PDS disclosure and member’s investment choice. The accounting disclosure literature

has extensively studied corporate managers’ voluntary disclosure decisions in the presence of

investor demand for information and capital market consequences of the disclosure (Healy et al.,

2001). However, there is a paucity of empirical evidence on the role of information disclosure on

superannuation members’ behavior. The PDS is the first document investors and advisors are likely to

look at when trying to understand/select a fund family. The PDS is also regarded as an important

communication channel by regulators, as evidenced by the introduction of regulation to improve the

readability of PDSs. Therefore, the next logical question is whether the PDS does affect member’s

choice?

The research question is interesting because it is not known whether fund families simply

prepare standardised PDS to comply with legislation and minimize their own liability, or whether funds

families also regard the PDS as a useful mechanism to provide important basic information and attract

potential investors. Before the short PDS regime, there were major concerns about whether the long

and complex PDS documents cause information overload for investors because simply providing a lot

of difficult-to-understand information does not mean users can use the information effectively. It is

therefore interesting to examine whether PDSs under the short form regime, which is supposed to have

improved readability of these documents, are actually associated with members’ investment choices.

In addition, whether the fund families under the short PDS regime have made the PDS documents

highly standardised so that these documents do not contain sufficient variations to affect member’s

investment choices is another issue to be explored.

This thesis employs direct measures of the superannuation disclosure document - readability

and content of the PDS - to provide an understanding of the possible consequences of

superannuation fund family’s PDS disclosure. The provision of information typically can provide

important consumer benefits such as improved decision making, enhanced product quality, and lower

prices (Mazis et al. 1981). However, in order for that information to have a positive impact on the

consumer’s decision-making process, the information must be easily accessible and presented in

a clear and understandable format.

There are a number of factors that likely contribute to information overload for investors. One

factor could be how the information is presented to the investors. Simply providing more information

about investment options may not be enough to help investors make good decisions. As predicted in

74

the economics of information literature, consumers will tend to use information more extensively if

it costs less time and/or money to acquire and process (Nelson 1970; Stigler 1961). This suggests that

when information is easier to obtain and evaluate, consumers are more likely to use it when making

decisions or choices. If information about investment options is presented in a simpler and clearer

way, investors should suffer less information overload and be more likely to use the information

productively when making asset allocation decisions.

On average, superannuation members are ill-equipped to handle their increasing financial

responsibilities due to limited financial sophistication, an overwhelming array of family and fund

choices, and complicated superannuation regulation. There are over 300 superannuation entities

with more than 270 investment options on average. The complicated and ever-changing regulations

and the PDS have also contributed to the information ‘overload’. As a result, the ASIC proposed

and subsequently adopted a new simplified disclosure requirement. Superannuation funds now are

required to use short-form PDS that cannot exceed eight A4 pages.

This thesis contributes to the literature by examining the usefulness of the short PDSs and by

using more direct measures of information costs and member choice. To my knowledge, there has been

no direct empirical investigation of how the short PDS affects superannuation members’ choices, even

though the effectiveness of the new short PDS in helping investors select fund families should be a

fundamental and important question for regulators and investors. Therefore, this thesis contributes to

the literature by examining the association between the attributes of short PDS and their behavioral

implications on superannuation member’s investment choice. In addition, this thesis quantifies the

extent and quality of information contained in PDS documents by using both content analysis and

computational linguistic readability tests. The disclosure scores are then used to explore the

relationship between fund disclosure and superannuation members’ choices. This approach

distinguishes this thesis from previous studies that use various fund characteristics related to fund

visibility to indirectly proxy for the reduction in information costs.

The impact of corporate disclosure in capital market research is often measured by the firm’s

cost of capital where voluntary disclosure is expected to reduce information asymmetry and thereby

reducing the firms’ cost of capital (Healy and Palepu, 2001). Cost of capital is not directly

observable for superannuation funds, so fund flow provides a direct measure of investors’ choice

and helps assess the value investors attach to superannuation fund families’ disclosure.

6.2 Literature Review and Hypothesis Development

Innovations in information technology and financial services, as well as increased financial

75

disclosure requirements mandated by regulators, have created an environment replete with

information for superannuation members. However, navigating through the plethora of information

is not easy for the average superannuation member. Members used to face long superannuation PDSs

crammed with information regarding fund performance, fees and costs, as well as details of product

features. Given the sheer quantity of information available to investors and the thousands of

products available for choice, selecting the fund that most appropriately matches the superannuation

member’s needs is a challenging task.

From the perspective of superannuation fund families, PDS disclosure offers three advantages

as a potential non-performance-based strategy to attract investors. First, unlike disclosure of portfolio

holdings in hedge funds, which may expose the funds to front-running or free-riding activities by other

market participants, PDS disclosure only updates information annually, which is less frequently than

the disclosure of portfolio holdings. Second, communication through PDS is more cost - effective

compared to marketing, as PDS disclosure is required anyway. In addition, (Gallaher, S, Kaniel, R &

Starks, LT 2006) provide evidence that advertising expenditures have a significant positive effect on

managed fund’s investment flows only for advertisers with high expenditures relative to others.

Similarly, results from Chapter 4 of this thesis also imply that a threshold of advertising expenditures

exists before advertising has a significant effect on fund flows. Finally, PDS disclosure can be superior

to other non-performance strategies (e.g. star-creating strategy) because the use of quality PDS does

not sacrifice the fund family’s performance. Fund families may be induced by spill-over effects to

pursue strategies which take advantage of the cash flow response to a star performer in the same fund

family. However, Nanda, Wang and Zheng (2004) find cross-fund return standard deviation is

associated with significantly poorer family level performance. This suggests that a star-creating

strategy, presumably targeted at less informed investors, impairs fund family’s overall performance.

From the investor’s perspective, disclosure can lower investors’ search and processing costs in

making financial decisions. Studies in the US show that employees increase their participation in 401(k)

retirement plans when their employers offer financial education programs, whether in the form of

brochures or seminars35, supporting the argument that reduction in investors’ information costs facilitate

their decision-making process.

Superannuation funds’ PDS can serve as a useful mechanism to reduce information acquisition

and processing costs because it provides both historical and forward-looking information on

superannuation products’ existence and characteristics. However, complex disclosures that are difficult

to understand increase the difficulty of making informed financial decisions. Recent studies in the

35 Hecklinger, Richard E. Deputy Secretary-General of the OECD speaking January 9, 2006 at The Smith Institute, London.

76

accounting literature show that complexity in qualitative disclosures is associated with weaker market

reactions to news (Rennekamp 2012; You & Zhang 2009) and lower trading volume among small

investors (Miller, BP 2010).

A good PDS (e.g., in terms of readability or informativenss) may attract investors sufficiently to

make them seek more information about the fund while a poor quality PDS (e.g., difficult to undertand)

may deter investors from choosing or switching to the fund. That is, better quality PDS disclosure in

terms of readability and content should reduce investor’s information costs more than poor quality PDS

disclosure. Consistent with prior empirical evidence that lower information costs are associated with

more investment inflows (Jain & Wu 2000; Massa 2003; Nanda, Wang & Zheng 2004; Sirri & Tufano

1998), it is expected that superannuation fund families with superior PDS disclosure practice attract

more fund flows than families with relatively inferior PDS attributes.

The hypotheses on the relation between superannuation PDS disclosure and member choice are

stated as follows:

H5: The level of superannuation fund family inflow is positively associated with the readability of

PDS information.

H6: The level of superannuation fund family inflow is positively associated with the extent of PDS

information disclosed.

In this chapter, the readability of PDS information is proxied by the Flesch readability score

while the extent of PDS information is proxied by content analysis scores.

6.3 Research Method and Results

6.3.1 PDS disclosure and Fund Family Inward Fund Flows

Consistent with earlier chapters, this chapter uses the fund family as its unit of analysis.

Disclosure strategies originate at the fund family level, as evidenced by the standard disclosure

format applied to all funds within a fund family. In addition, the use of fund level flow is

problematic because it captures some intra-family flows that are not caused by differences

in disclosure among fund families. In addition, the primary goal of superannuation fund entities

is to maximize total sponsor profits at the family level. Thus, to fully understand the motivation

and impact of PDS disclosure, the fund family, not individual funds, should be used as the

unit of measurement.

Several prior studies have documented many beneficial capital markets effects associated

77

with improved disclosure, including lower equity and debt costs ((Botosan 1997; Sengupta

1998), narrower bid-ask spreads (Healy, Hutton & Palepu 1999; Leuz & Verrecchia 2000; Welker

1995) and greater stock price responsiveness to earnings (Price 1998). However, none of these

measures are directly applicable to superannuation funds. The main purpose of the

superannuation PDS and the regulatory reform introducing the short PDS are to help investors

make better investment decisions. Therefore, this study uses the member- initiated inward fund

transfer as the measure of member choice to evaluate the effectiveness of PDS disclosure.

6.3.2 Simultaneous Equations Model with Current Flows

Most theoretical papers recognize the endogeneity of disclosure choices and their

consequences (Dye 2001; Verrecchia 2001). If better disclosure attracts more investors, fund flows

will translate into larger fund size. However, larger funds are often regarded as having more

resources available to make better disclosures. The growing number of members also represents an

increasing demand for high-quality information. In addition, larger families are likely to have

higher political costs and higher visibility, so they experience more pressure to make quality

disclosures. This implies that the empirical model must simultaneously incorporate both disclosure

decisions and investment decisions. Empirical models that fail to account for the endogeneity of

disclosure choices will likely generate biased estimates. To mitigate potential bias caused by

endogenous decision-making, this chapter uses a simultaneous equations model to test the

relationship between better disclosure and investment choice. A three-stage least squares (3SLS)

regression model (see (Zellner & Theil 1962)) is employed to estimate a system of structural

equations, where some equations contain endogenous variables among the explanatory variables.

Typically, the endogenous explanatory variables are dependent variables from other equations in

the system. The disturbance is correlated with the endogenous variables, violating the assumptions

of OLS. The 3SLS model uses an instrumental-variables approach to produce consistent estimates

and generalized least squares (GLS) to account for the correlation structure in the disturbances

across the equations.

The 3SLS can be thought of as producing estimates from a three-step process. Step 1 is to

develop instrumented values for all endogenous variables. These instrumented values can simply be

considered the predicted values resulting from a regression of each endogenous variable on all

exogenous variables in the system. This stage is identical to the first step in 2SLS and is critical for

the consistency of the parameter estimates. In step 2, a consistent estimate is obtained for the

covariance matrix of the equation disturbances. These estimates are based on the residuals from a

78

2SLS estimation of each structural equation. In step 3, a GLS-type estimation is performed using

the covariance matrix estimated in the second stage and with the instrumented values in place of the

explanatory endogenous variables.

To apply the 3SLS model in this thesis’s context, the readability/extent of disclosure in the

PDS (Readability /Content score) relates to investment inflows (Inflow), superannuation family

performance (Perf), and other institutional attributes. Simultaneously, investment inflows (Inflow)

depend on PDS readability/content (Readability /Content score), performance (Perf) and those fund

family characteristics that affect the level of inflows (e.g., the breadth of product offerings, members’

age profile). In this system, Readability /Content score and Inflow will be endogenous variables,

with some of the fund family characteristics being exogenous. All exogenous variables in the model

are used as instruments. The first equation examines the causes of PDS disclosure. Specifically,

readability and content scores (Readability /Content score) are regressed on the growth in the

superannuation family inflows and other control variables. Contemporaneous disclosure and flow

measures are used in the model, as superannuation families usually update their PDSs on an annual

basis and only provide potential investors with the latest PDSs, the usefulness of the old PDS is

minimal once a new PDS is issued. The first equation examining causes of PDS disclosure

(Readability /Content score) is as follows:

𝑅𝑅𝑜𝑜𝑜𝑜𝑤𝑤𝑜𝑜𝑏𝑏𝑜𝑜𝑁𝑁𝑜𝑜𝑀𝑀𝑠𝑠 𝐼𝐼𝐿𝐿𝑤𝑤𝑜𝑜𝑀𝑀/𝐶𝐶𝑜𝑜𝐿𝐿𝑀𝑀𝑜𝑜𝐿𝐿𝑀𝑀 𝑆𝑆𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑖𝑖,𝑡𝑡 = 𝑜𝑜 + 𝑏𝑏1𝐼𝐼𝐿𝐿𝑃𝑃𝑁𝑁𝑜𝑜𝑁𝑁𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑃𝑃𝑜𝑜𝑜𝑜𝑃𝑃𝑖𝑖,𝑡𝑡

+ 𝑏𝑏3𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡−1 + 𝑏𝑏4𝐿𝐿𝐼𝐼𝑀𝑀𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝑃𝑃𝑜𝑜𝑜𝑜𝐶𝐶𝑜𝑜𝐿𝐿𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝐿𝐿𝑜𝑜𝑀𝑀𝐿𝐿𝑜𝑜𝑁𝑁𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (6)

The variables in the model are defined in Section 5.4.4.

Using the second stage of the simultaneous equations, this thesis investigates the potential

impact of better PDS disclosure on fund family inflows while controlling for the potential effects of

superannuation family characteristics on family disclosure decisions. More specifically, family

inflow (Inflow) is regressed on PDS readability or content analysis score (Readability/Content

score), fund family performance and other control variables.

The ‘consequence of disclosure’ equation is constructed as follows:

𝐼𝐼𝐿𝐿𝑃𝑃𝑁𝑁𝑜𝑜𝑁𝑁𝑖𝑖,𝑡𝑡 = 𝑜𝑜 + 𝑏𝑏1 𝑅𝑅𝑜𝑜𝑜𝑜𝑤𝑤𝑜𝑜𝑏𝑏𝑜𝑜𝑁𝑁𝑜𝑜𝑀𝑀𝑠𝑠 𝐼𝐼𝐿𝐿𝑤𝑤𝑜𝑜𝑀𝑀/𝐶𝐶𝑜𝑜𝐿𝐿𝑀𝑀𝑜𝑜𝐿𝐿𝑀𝑀 𝑆𝑆𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑜𝑖𝑖,𝑡𝑡 + 𝑏𝑏2𝑃𝑃𝑜𝑜𝑜𝑜𝑃𝑃𝑖𝑖,𝑡𝑡

+ 𝑏𝑏3𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝐿𝑖𝑖,𝑡𝑡−1 + 𝑏𝑏4𝐿𝐿𝐿𝐿𝐼𝐼𝐿𝐿𝐼𝐼𝐼𝐼𝑀𝑀𝑀𝑀𝑖𝑖,𝑡𝑡 + 𝑏𝑏5𝐿𝐿𝐼𝐼𝑀𝑀𝑖𝑖,𝑡𝑡 + 𝑏𝑏6𝐷𝐷𝑜𝑜𝑃𝑃𝐿𝐿𝑜𝑜𝑜𝑜𝑜𝑜𝑀𝑀𝑖𝑖,𝑡𝑡 + 𝑏𝑏7𝐿𝐿𝑏𝑏𝑜𝑜𝐼𝐼𝑜𝑜50𝑖𝑖,𝑡𝑡 + 𝜀𝜀𝑖𝑖,𝑡𝑡 (7)

6.4 Multivariate Results

Table 6.1 and Table 6.2 report the results of simultaneous equations analysis of the causes

79

and consequences of superannuation fund families’ PDS quality. Three-stage-least-square (3SLS)

regression is used to estimate the system to mitigate potential biases attributable to endogeneity in

the system equations.

Panel A of Table 6.1 reports results from the first stage of the simultaneous regressions

which examines the potential impact of family inflow on PDS readability. The positive and

significant coefficient of Inflow (coefficient of 63.406 with a p-value of 0.003) suggests that

superannuation families disclose more readable information when the family inflow is high. The

retail and industry fund families are then examined separately. The results show that the relation

between PDS readability and inflow stems mainly from the retail subsample (Inflow is positive and

significant at the 5% level). One plausible reason can be that industry funds are established by

members of a particular industry. This fixed member base may explain why industry funds are less

responsive to investment inflows when formulating disclosure strategies.

Panel B of Table 6.1 presents the results for the second stage of the simultaneous regressions

which examines the potential impact of PDS readability on family inflow. In contrast to the results

in Panel A, Panel B shows that there is only a marginally significant positive relation between

inflow and PDS readability, supporting Hypothesis 6 and suggesting that more readable PDSs result

in higher inflows. This provides some evidence supporting the usefulness of higher quality PDS

disclosure in attracting fund flows. No significant results are found between inflow and PDS

readability when the same analysis is performed for the retail and industry subsamples. Most of the

control variables are not statistically significant for explaining fund family inflows, except for

Above50, which is marginally significant for the overall sample, suggesting that a higher investment

inflow is observed in fund families with more members approaching retirement. This result makes

logical sense, because superannuation members who are closer to retirement age are more likely to

contribute larger amounts of money to their superannuation funds.

80

Table 6.1 Three-stage-least-square Regression Results

This table provides simultaneous Equation Analysis (3SLS) of PDS readability (Flesch) and fund family inflow (Inflow). Inflow is APRA-reported

inward transfers scaled by superannuation fund family size. DefAsset is the proportion of fund asset in the default option. The proportion of members

above the age of 50 is denoted as Above50. All other variables are defined in Table 5. Robust p-values are reported in parentheses. ***, ** , and *

denote significance at the 1%, 5% and 10% level, respectively.

Panel A: Causes of PDS readability

(1) All (2) Retail (3) Industry

Dependent variable: Flesch

Inflow 63.406*** 60.462** 20.637

(0.003) (0.048) (0.869)

Perf -0.309 0.082 1.681

(0.922) (0.986) (0.373)

LnTNA 1.019* 1.283 -0.656

(0.058) (0.109) (0.424)

LnInvOpt -0.834 -1.231 0.361

(0.161) (0.120) (0.723)

TOE 0.012 0.023 -0.024

(0.395) (0.423) (0.650)

PerCon -269.690*** -247.022* -521.515***

(0.005) (0.065) (0.006)

TotNew 0.000 0.000 0.000

(0.650) (0.719) (0.455)

Constant 38.614*** 45.853*** 64.541***

(0.000) (0.000) (0.000)

Observations

135

63

52

Year dummies Yes Yes Yes

Family type dummies Yes No No

81

Panel B: Consequences of PDS readability

(1) All (2) Retail (3) Industry

Dependent variable: Inflow

Flesch 0.094* 0.087 -0.002

(0.093) (0.220) (0.313)

Perf 0.260 0.231 -0.002

(0.409) (0.613) (0.834)

LnTNA -0.068 -0.086 -0.002

(0.197) (0.315) (0.672)

LnInvOpt 0.095 0.117 0.001

(0.156) (0.301) (0.855)

TOE -0.001 -0.001 0.000

(0.324) (0.371) (0.174)

DefAsset 0.03 0.061 -0.005

(0.720) (0.712) (0.836)

Above50 0.933* 0.792 -0.05

(0.089) (0.217) (0.193)

Constant -4.635 -4.593 0.173

(0.102) (0.248) (0.181)

Observations

135

63

52

Year dummies Yes Yes Yes

Family type dummies Yes No No

The same regression model is used to examine the relation between investment inflow and the

content disclosed in PDSs (as measured by the content analysis scores using this study’s self- constructed

disclosure index). The results reported in Panel A of Table 6.2 show positive and significant

coefficients for inflows across all samples (subsamples), suggesting that more inflows appear to cause

superannuation families to disclose more/better quality information in PDSs. Relative to retail families,

the disclosure content scores of industry superannuation families appear to be more responsive to inflows

as reflected by the coefficient of 651.699, which is almost 4 times of the coefficients for the overall

(172.387) and retail (174.579) samples.

Panel B of Table 6.2 reports the impact of the quality of PDS content on investment inflows. In

contrast to earlier results for the readability scores in Table 6.1, the positive and significant coefficients

for all samples in Table 6.2, Panel B show that higher PDS content scores appear to attract more inflows.

However, the positive fund flow consequence of disclosing more and better content in the PDS is lowest

for industry families (Column 3 of Panel B). These results provide support for Hypothesis 7.

Turning to the results for control variables in Panel B of Table 6.2, again it appears that most

control variables are not statistically significant determinants of inflows except for Above50, consistent

82

with the results found in Panel B of Table 6.1 (the impact of PDS readability on investment flows)

that superannuation families with a higher proportion of members approaching retirement age receive

more investment inflow.

Table 6.2 Three-stage-least-square Regression Results

This table provides simultaneous Equation Analysis (3SLS) of PDS content quality (Content) and fund family inflow (Inflow). Inflow is APRA-reported inward

transfers scaled by superannuation fund family size. DefAsset is the proportion of fund assets in the default option. The proportion of members above the age of

50 is denoted as Above50. All the other variables are defined in Table 5. Robust p-values are reported in parentheses. ***, ** , and * denote significance at the

1%, 5% and 10% level, respectively.

Panel A: Causes of PDS content quality

(1) All (2) Retail (3) Industry

Dependent variable: Content

Inflow 172.387*** 174.579*** 651.669*

(0.001) (0.007) (0.061)

Perf 6.445 2.900 2.185

(0.325) (0.788) (0.653)

LnTNA 1.023 1.196 0.549

(0.395) (0.500) (0.782)

LnInvOpt -1.402 -1.638 -0.649

(0.267) (0.353) (0.779)

TOE 0.027 0.027 -0.044

(0.373) (0.516) (0.518)

PerCon -708.552*** -735.747*** -446.193

(0.003) (0.009) (0.332)

TotNew 0.000 0.000 0.000

(0.534) (0.899) (0.771)

Constant -0.585 -6.885 52.486***

(0.981) (0.764) (0.001)

Observations

148

77

61

Year dummies Yes Yes Yes

Family type dummies Yes No No

83

Panel B: Consequences of PDS content quality

(1) All (2) Retail (3) Industry

Dependent variable: Inflow

Content 0.034*** 0.038** 0.003*

(0.002) (0.020) (0.086)

Perf -0.039 0.245 -0.008

(0.773) (0.387) (0.589)

LnTNA -0.020 -0.029 -0.001

(0.422) (0.411) (0.805)

LnInvOpt 0.046 0.049 0.002

(0.146) (0.228) (0.782)

TOE 0.000 0.000 0.000

(0.577) (0.697) (0.655)

DefAsset 0.003 0.054 -0.005

(0.967) (0.683) (0.541)

Above50 0.792*** 1.073** 0.026

(0.005) (0.027) (0.509)

Constant -0.268 -0.212 -0.144

(0.615) (0.666) (0.152)

Observations

148

77

61

Year dummies Yes Yes Yes

Family type dummies Yes No No

To provide further understanding of the relation between disclosure Content and Inflow, the

content analysis index classifies each PDS into 10 parts (based on the section headings required by

the Corporations Act). Simultaneous equations are run for each of the 10 parts of the PDS documents

to assess whether each part’s disclosure content is associated with fund family inflows. Table 6.3

reports the results for the determinants and consequences of the content disclosed in each of the 10

sections in the superannuation PDSs. Because stage 1 of the simultaneous equations (causes of

content disclosed) is not the main test for the regression analysis in this chapter, for the sake of

brevity Table 6.3 Panel A only presents the results for the key variable Inflow without listing

the results of the individual control variables. The results from Panel A Table 6.3 show that when

superannuation families receive more investment inflows, the content disclosure score (which

indicates disclosure informativeness/quality) of almost all parts of the PDS are significantly higher,

except for Part 9 (How to open an account). This result further supports the finding from Table 6.2,

Panel A that higher levels of investment flows cause more and better content to be disclosed in the

PDS.

Table 6.3 Panel B presents the main test results of the potential impact of disclosure quality

on inflow for each of the 10 parts of PDS documents (a full list of the 10 parts can be found in

84

section 5.4.2 and in Appendix 1). The results indicate that a higher content score for the majority of

the 10 parts of PDS documents is useful/relevant for attracting more investment flows. The three

exceptions are part 4, part 8 and part 9, which are ‘Risk of super’, ‘Insurance in you super’ and

‘How to open an account’, respectively. These three parts usually contain general/standardised

information with a high level of similarity among fund families. Part 8 may also contain information

about the other products of the superannuation fund family (i.e., insurance).

Table 6.3 3-stage-least-square Regression Results This table provides simultaneous Equation Analysis (3SLS) of content quality of each PDS section (p1, p2, p3 etc,.) and fund family inflow (Inflow). Robust p-values

are reported in parentheses. ***, ** , and * denote significance at the 1%, 5% and 10% level, respectively.

Panel A: Causes of content disclosed in each section of PDS

Dependent variable p1 p2 p3 p4 p5 p6 p7 p8 p9 p10

Inflow 20.760*** 60.990*** 34.684*** 13.145*** 42.112*** 10.550*** 31.093*** 24.059** 2.079 11.452*** (0.000) (0.000) (0.001) (0.000) (0.000) (0.006) (0.000) (0.032) (0.130) (0.000)

Control variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 148 148 148 148 148 148 148 148 148 148

Year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Panel B: Consequences of content disclosed in each section of PDS Dependent variable: Inflow PDS part P1 P2 P3 P4 P5 P6 P7 P8 P9 P10 0.114*** 0.067** 0.138*** -0.058 0.075*** 0.227** 0.114*** -0.064 0.370 0.319*** (0.000) (0.014) (0.000) (0.264) (0.000) (0.023) (0.001) (0.119) (0.212) (0.000)

Control variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations 148 148 148 148 148 148 148 148 148 148

Year dummies Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

It is surprising that the section 4 (Risk of Super) is not associated with fund flows as risk is one

of the most important factors superannuation investors need to consider in making financial decisions.

The result is likely caused by fund families using general or standardised descriptions of risks in the PDS

instead of providing fund family specific or product specific information about risks.

To provide an example of section 4 disclosure, ‘Part 4: Risk of Super’ is extracted from the 2014

PDSs of Smartsuper and Suncorp Super as shown below in Figure 6.1 and 6.2. The content of part 4 are

highly similar between the two examples, as the two superannuation families simply copy and use the

prescribed description of risks in Schedule 10D of the Principal Regulations of Corporations Act (see

appendix 1).

85

Figure 6.1 Smartsave 2014 PDS Part 4: Risks of Super

86

87

Figure 6.2 Suncorp Super 2014 PDS Part 4: Risks of Super

88

The use of the Standard Risk Measure (the number of negative returns expected in 20 years for

a superannuation product) is advocated by regulators to improve the transparency and comparability

of superannuation products. Nevertheless, the discussion of this measure is very broad and general

in the risk section of PDS. In other words, Part 4: Risks of Super is used to provide only general

information about the risk of superannuation investments. It is observed from the scoring process that

specific product risk information is often provided in other parts of the PDS. An example is taken from

‘Part 5: How we invest your money’ in Suncorp Super’s 2014 PDS: various standard risk measures

are provided for each of the Lifestage Fund products (shown in Figure 6.3 below), which suggests

that the disclosure of risk information is not limited to Part 4 of the PDS and is often more specific and

meaningful when associated with individual superannuation products.

89

Figure 6.3 Suncorp Super: Table of ‘How we invest your money’

90

6.5 Additional Analysis and Discussion

6.5.1 Lagged regression model

To provide a complete analysis and to further address the potential endogeneity issue, an

alternative panel regression model with lagged readability/content scores is used to study the causal

impact of PDS disclosure on fund family flows. Family inflow for year t is regressed on disclosure

scores of PDS documents that have already been issued by year t (i.e., disclosure score is based on

financial year t-1). The results of the lagged independent variable model are reported in Table 6.4

and Table 6.5. Little significance is found for the potential relationship between PDS disclosure and

inflows in Table 6.4. The same regression model is then performed between inflows and the content

score for each of the 10 parts of the PDS. The results in Table 6.5 show that the most of the 10 parts of

the PDS are useful for attracting investment flows, apart from Parts 4, 6, 8 and 9, which is largely

consistent with the simultaneous regression model results in Table 6.3. Nevertheless, the use of the

lagged regression model is less desirable, because superannuation families only provide potential

investors with the most recent PDSs. In addition, the usefulness of the old PDS in the previous year is

minimal once a new PDS is issued.

91

Table 6.4 Multivariate Regression Results

This table examines the relationship between superannuation fund family inflow (Inflow) and superannuation fund family PDS Readability/Content. Robust p-values

are reported in parentheses. ***, ** , and * denote significance at the 1%, 5% and 10% level, respectively.

PDS readability and fund family inflow PDS content and fund family inflow

(1) All (2) Retail (3) Industry (4) All (5) Retail (6) Industry

Dependent variable: Inflow

Flesch𝑡𝑡−1 -0.002 -0.001 0.000

-0.146 -0.483 -0.647

Content𝑡𝑡−1 0.001* 0.001 0.000

-0.07 -0.127 -0.204

Control variables Yes Yes Yes Yes Yes Yes

Observations 60 25 28 71 33 34

Adjusted R-squared 0.39 0.214 0.143 0.541 0.411 0.031

Year Fixed Effect Yes Yes Yes Yes Yes Yes Family Type Fixed Effect Yes No No Yes No No

92

Table 6.5 Multivariate Regression Results

This table examines the relationship between superannuation fund family inflow (Inflow) and the content of each PDS section (P1, P2, P3 etc,.).

Robust p-values are reported in parentheses. ***, ** , and * denote significance at the 1%, 5% and 10% level, respectively

Dependent variable: Inflow

P1𝑀𝑀−1 0.006***

(0.004)

P2𝑀𝑀−1 0.003** (0.045)

P3𝑀𝑀−1 0.003** (0.041)

P4𝑀𝑀−1 0.001 (0.916)

P5𝑀𝑀−1 0.004** (0.020)

P6𝑀𝑀−1 0.001 (0.833)

P7𝑀𝑀−1 0.007** (0.034)

P8𝑀𝑀−1 0.001 (0.492)

P9𝑀𝑀−1 0.01 (0.297)

P10𝑀𝑀−1 0.009*

(0.069)

Control variables Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Observations

71

71

71

71

71

71

71

71

71

71

Adjusted R- squared 0.532 0.521 0.505 0.487 0.528 0.487 0.531 0.492 0.496 0.502

Year Fixed Effect Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes

Family Type Fixed Effect

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Yes

6.5.2 Combined effect of marketing and PDS disclosure

One might expect PDS disclosure to affect the relation between fund flow and marketing

efforts in that PDS content or presentation could mitigate or magnify the importance of marketing

(which is examined in Chapter 4). A high-quality PDS may reduce the need for marketing, whereas

and a low-quality PDS may require more marketing efforts to compensate for the deficiency. In

contrast, fund families that make the effort to market themselves may also take more care in

preparing their PDS documents. To explore these possibilities, interaction between

readability/content scores and total operating expense (TOE), which serves as the proxy for

93

marketing costs, is added to the regression to investigate this possibility. The unreported results

show that none of the interaction variables is statistically significant. The lack of results may be

driven by at two possibilities. First, the TOE appears to be merely a coarse proxy for marketing.

Other costs such as management fees are also included in the TOE, which is expected to have the

opposite (i.e., a negative) impact on investment flows. The measure we use for marketing in this

chapter (i.e., the TOE) is different from the marketing measure used in Chapter 4 (Other Operating

Expense, under which most of the marketing expense is reported), because the latter measure is not

available for 2015. Second, there can be no relation between marketing and PDS attributes, i.e.,

these are separate decisions by fund families and are not interrelated.

6.6 Limitations

Admittedly, even if optimal disclosure and investor education programs are successfully

implemented, there will always be some individuals who either do not have the ability or are

unwilling to acquire the necessary knowledge and skills to become informed investors. This thesis

does not deny the existence of investor inertia. Rather, this thesis focuses on investors who have

made a choice (proxied by inflows) to examine the influence of PDS disclosure.

Several caveats are in order. First, because of data constraints, it is not possible to distinguish

investor switches between fund families attributable to a change of employer or a conversion

to a pension product. Second, the impact of superannuation fund family choice extends far beyond

visible switching. An employee who chooses to remain in his or her current fund is exercising

choice in a manner that does not add to the switching statistics. Third, individual investors’ behaviour

bias and financial literacy can lead to different investment reactions. This thesis assumes that on

average, superannuation fund members are generally naïve investors and not highly financially

literate, and it examines them as a group.

6.7 Conclusion

This chapter provides empirical evidence on the fund flow consequence of PDS disclosure

properties using the linguistic properties of superannuation fund family PDSs. More specifically,

this chapter studies the implications of the readability and extent of PDS disclosure for investment

inflows. The empirical findings can be summarized as follows. First, superannuation fund families

provide more readable PDSs when there are more inflows, but there is only weak evidence for the

overall sample that investors are attracted to superannuation families because of the ease of reading

of PDSs. One possible reason is that the overall PDSs are already relatively easy to understand

under the short-form PDS regime (see discussion in Chapter 5.5.1). Second, our results show that

94

more inflows encourage superannuation families to provide more and better information in the PDS,

especially the industry families. In support of Hypothesis 6, the quality and extent of PDS disclosure

is positively associated with fund inflows. This relation is especially strong for the retail families.

Viewed collectively, this provides important evidence that superannuation families with better

PDS content do attract more inflows.

95

Chapter 7 - Conclusion

This thesis examines the determinants and consequences of superannuation investors’

investment choice with a focus on the effects of superannuation families’ performance, marketing

efforts and PDS disclosure quality on the investment choices made by superannuation members.

These issues are very important given the economic and social significance of superannuation to all

working Australians, along with investors and regulators’ concern about the quality of PDS

disclosure for making informed investment decisions, as evidenced by ASIC’s review of the

superannuation system and the introduction of the ‘short PDS regime’ in 2011. The results of this

thesis suggest that return-chasing behaviour does not exist in the superannuation market. Marketing

is an effective channel through which to direct members’ search and reduce their recognition costs;

however, little information is provided in advertisements. To study the effect of members’

information processing costs, this thesis assesses PDS disclosure on two dimensions—(1) ease of

understanding and (2) the extent of detailed reporting—to see whether they can be used to mitigate

members’ costly interpretation efforts. Marketing increases investors’ awareness of the

superannuation family, and the amount and quality of information disclosed provide additional

understanding. Both strategies are found to result in more investment flows.

The results from this thesis suggest that superannuation members are not chasing returns.

Accordingly, this thesis provides insights into the strategy to be pursued by superannuation fund

families to attract investors (i.e., the use of marketing and informative PDSs). It also sheds light on

regulation formulation to guide and encourage informative and understandable PDS disclosure.

Finally, it adds to the literature of superannuation and more generally, the literature of retirement

savings, by providing empirical evidence in addition to experimental study evidence and inferences

drawn from survey or focus groups with respect to investors’ long-term saving arrangements.

96

Bibliography

Agnew, J, Balduzzi, P & Sunden, A 2003, 'Portfolio choice and trading in a large 401 (k) plan', The American Economic Review, vol. 93, no. 1, pp. 193-215.

APRA 2013, Annual Superanuation Bulletin June 2012.

—— 2016, Annual Superannuation Bulletin June 2015.

Arellano, M & Bond, S 1991, 'Some tests of specification for panel data: Monte Carlo evidence and an application to employment equations', The Review of Eeconomic Studies, vol. 58, no. 2, pp. 277-97.

Bailey, W, Kumar, A & Ng, D 2011, 'Behavioral biases of mutual fund investors', Journal of Financial Economics, vol. 102, no. 1, pp. 1-27.

Barber, BM, Odean, T & Zheng, L 2005, 'Out of sight, out of mind: The effects of expenses on mutual fund flows', The Journal of Business, vol. 78, no. 6, pp. 2095-120.

Bateman, H, Dobrescu, LI, Newell, BR, Ortmann, A & Thorp, S 2016, 'As easy as pie: How retirement savers use prescribed investment disclosures', Journal of Economic Behavior & Organization, vol. 121, pp. 60-76.

Bateman, H, Eckert, C, Geweke, J, Louviere, J, Satchell, S & Thorp, S 2015, 'Risk presentation and portfolio choice', Review of Finance, vol. 20, no. 1, pp. 201-29.

Benartzi, S & Thaler, RH 2001, 'Naive diversification strategies in defined contribution saving plans', American Economic Review, vol. 91, no. 1, pp. 79-98.

Berk, JB & Green, RC 2004, 'Mutual fund flows and performance in rational markets', Journal of Political Economy, vol. 112, no. 6, pp. 1269-95.

Beshears, J, Choi, JJ, Laibson, D & Madrian, BC 2009, How does simplified disclosure affect individuals' mutual fund choices?, NBER Working Paper No. 14859.

Bhattacharya, U, Lee, JH & Pool, VK 2013, 'Conflicting family values in mutual fund families', The Journal of Finance, vol. 68, no. 1, pp. 173-200.

Biddle, GC, Hilary, G & Verdi, RS 2009, 'How does financial reporting quality relate to investment efficiency?', Journal of Accounting and Economics, vol. 48, no. 2, pp. 112-31.

Bloomfield, RJ 2002, 'The “incomplete revelation hypothesis” and financial reporting', Accounting Horizons, vol. 16, no. 3, pp. 233-43.

Bodie, Z, Merton, RC & Samuelson, WF 1992, 'Labor supply flexibility and portfolio choice in a life cycle model', Journal of Economic Dynamics and Control, vol. 16, no. 3-4, pp. 427-49.

Botosan, CA 1997, 'Disclosure level and the cost of equity capital', Accounting Review, vol. 72, no. 3, pp. 323-49.

Brennan, MJ, Schwartz, ES & Lagnado, R 1997, 'Strategic asset allocation', Journal of Economic Dynamics and Control, vol. 21, no. 8, pp. 1377-403.

Brennan, NM, Guillamon-Saorin, E & Pierce, A 2009, 'Methodological Insights: Impression management: developing and illustrating a scheme of analysis for narrative disclosures–a methodological note', Accounting, Auditing & Accountability Journal, vol. 22, no. 5, pp. 789-832.

97

Butt, A, Donald, MS, Foster, FD, Thorp, S & Warren, G 2015, Delegation, Trust and Defaulting in Retirement Savings: Perspectives from Plan Executives and Members CIFR Paper No. 065/2015; FIRN Research Paper No. 2638998, <Available at SSRN: https://ssrn.com/abstract=2638998 or http://dx.doi.org/10.2139/ssrn.2638998>.

Campbell, JY 2006, 'Household finance', The Journal of Finance, vol. 61, no. 4, pp. 1553-604.

Campbell, JY & Viceira, LM 1998, Who should buy long-term bonds?, NBER Working Paper No. w6801.

—— 1999, 'Consumption and portfolio decisions when expected returns are time varying', The Quarterly Journal of Economics, vol. 114, no. 2, pp. 433-95.

Carlin, BI, Gervais, S & Manso, G 2013, 'Libertarian paternalism, information production, and financial decision making', The Review of Financial Studies, vol. 26, no. 9, pp. 2204-28.

Carroll, GD, Choi, JJ, Laibson, D, Madrian, BC & Metrick, A 2009, 'Optimal defaults and active decisions', The Quarterly Journal of Economics, vol. 124, no. 4, pp. 1639-74.

Cheah, KK, Foster, FD, Heaney, R, Higgins, T, Oliver, B, O’Neill, T & Russell, R 2015, 'Discussions on long-term financial choice', Australian Journal of Management, vol. 40, no. 3, pp. 414-34.

Chevalier, J & Ellison, G 1997, 'Risk taking by mutual funds as a response to incentives', Journal of Political Economy, vol. 105, no. 6, pp. 1167-200.

Choi, JJ, Laibson, D, Madrian, BC & Metrick, A 2002, 'Defined contribution pensions: Plan rules, participant choices, and the path of least resistance', Tax Policy and the Economy, vol. 16, pp. 67-113.

—— 2005, 'Saving for retirement on the path of least resistance', Rodney L White Center for Financial Research-Working Papers-, vol. 9.

Clark, G, Duong, H, Gerrans, P, Lajbcygier, P, Moulang, C, Strydom, M, Vaz, J & Wickramanayake, J 2013, A Review of Retirement Savings Investment Behaviours: Theory and Evidence, Working Paper, CSIRO-Monash Superannuation Research Cluster, Melbourne.

Clatworthy, MA & Jones, MJ 2006, 'Differential patterns of textual characteristics and company performance in the chairman's statement', Accounting, Auditing & Accountability Journal, vol. 19, no. 4, pp. 493-511.

Commonwealth of Australia 2005, Superannuation Legislation Amendment (Choice of Superannuation Funds) Bill 2004.

—— 2012, Super System Review Final Report, Canberra, <http://www.psaonline.com.au/files/docs/smsf/cooper_review_smsf.pdf>.

Courtis, JK 1998, 'Annual report readability variability: tests of the obfuscation hypothesis', Accounting, Auditing & Accountability Journal, vol. 11, no. 4, pp. 459-72.

Cronqvist, H & Nilsson, M 2005, 'The choice between rights offerings and private equity placements', Journal of Financial Economics, vol. 78, no. 2, pp. 375-407.

Dahlquist, M & Martinez, JV 2015, 'Investor inattention: a hidden cost of choice in pension plans?', European Financial Management, vol. 21, no. 1, pp. 1-19.

98

Dangl, T, Wu, Y & Zechner, J 2006, 'Market discipline and internal governance in the mutual fund industry', The Review of Financial Studies, vol. 21, no. 5, pp. 2307-43.

Del Guercio, D & Tkac, PA 2002, 'The determinants of the flow of funds of managed portfolios: Mutual funds vs. pension funds', Journal of Financial and Quantitative Analysis, vol. 37, no. 4, pp. 523-57.

Duflo, E & Saez, E 2003, 'The role of information and social interactions in retirement plan decisions: Evidence from a randomized experiment', The Quarterly Journal of Economics, vol. 118, no. 3, pp. 815-42.

Dye, RA 2001, 'An evaluation of “essays on disclosure” and the disclosure literature in accounting', Journal of Accounting and Economics, vol. 32, no. 1, pp. 181-235.

Farhi, E & Panageas, S 2007, 'Saving and investing for early retirement: A theoretical analysis', Journal of Financial Economics, vol. 83, no. 1, pp. 87-121.

Fear, J & Pace, G 2008, Choosing not to choose: making superannuation work by default, Australia Institute.

Foster, FD, Ng, J & Wee, M 2015, 'Presentation format and financial literacy: accessibility and assessability of retirement savings statements', Journal of Consumer Affairs, vol. 49, no. 3, pp. 519-49.

Fry, T, Heaney, R & McKeown, W 2007, 'Will investors change their superannuation fund given the choice?', Accounting & Finance, vol. 47, no. 2, pp. 267-83.

Gallaher, S, Kaniel, R & Starks, L 2006, 'Madison Avenue meets Wall Street: Mutual fund families, competition and advertising', Competition and Advertising (January 2006).

Gallaher, S, Kaniel, R & Starks, LT 2006, Madison Avenue meets Wall Street: Mutual fund families, competition and advertising, <Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=879775>.

Gaspar, JM, Massa, M & Matos, P 2006, 'Favoritism in mutual fund families? Evidence on strategic cross‐fund subsidization', The Journal of Finance, vol. 61, no. 1, pp. 73-104.

Gerrans, P 2012, 'Retirement savings investment choices in response to the global financial crisis: Australian evidence', Australian Journal of Management, vol. 37, no. 3, pp. 415-39.

Goetzmann, WN & Peles, N 1997, 'Cognitive dissonance and mutual fund investors', Journal of Financial Research, vol. 20, no. 2, pp. 145-58.

Gomes, FJ, Kotlikoff, LJ & Viceira, LM 2008, Optimal life-cycle investing with flexible labor supply: A welfare analysis of life-cycle funds, NBER Working Paper No.w13966.

Grossman, SJ & Stiglitz, JE 1980, 'On the impossibility of informationally efficient markets', The American Economic Review, vol. 70, no. 3, pp. 393-408.

Gruber, MJ 1996, 'Another puzzle: The growth in actively managed mutual funds', The Journal of Finance, vol. 51, no. 3, pp. 783-810.

Healy, PM, Hutton, AP & Palepu, KG 1999, 'Stock performance and intermediation changes surrounding sustained increases in disclosure', Contemporary Accounting Research, vol. 16, no. 3, pp. 485-520.

99

Hortaçsu, A & Syverson, C 2004, 'Product differentiation, search costs, and competition in the mutual fund industry: A case study of S&P 500 index funds', The Quarterly Journal of Economics, vol. 119, no. 2, pp. 403-56.

Huang, J, Wei, KD & Yan, H 2007, 'Participation costs and the sensitivity of fund flows to past performance', The Journal of Finance, vol. 62, no. 3, pp. 1273-311.

Huang, JC, Wei, KD & Yan, H 2012, 'Investor Learning and Mutual Fund Flows', paper presented to AFA 2012 Chicago Meetings Paper.

Huberman, G & Jiang, W 2006, 'Offering versus choice in 401 (k) plans: Equity exposure and number of funds', The Journal of Finance, vol. 61, no. 2, pp. 763-801.

Jain, PC & Wu, JS 2000, 'Truth in mutual fund advertising: Evidence on future performance and fund flows', The Journal of Finance, vol. 55, no. 2, pp. 937-58.

James, C & Karceski, J 2006, 'Investor monitoring and differences in mutual fund performance', Journal of Banking & Finance, vol. 30, no. 10, pp. 2787-808.

Karlsson, A & Nordén, L 2007, 'Home sweet home: Home bias and international diversification among individual investors', Journal of Banking & Finance, vol. 31, no. 2, pp. 317-33.

Khorana, A & Servaes, H 1999, 'The determinants of mutual fund starts', The Review of Financial Studies, vol. 12, no. 5, pp. 1043-74.

—— 2004, Conflicts of Interest and Competition in the Mutual Fund Industry, <Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=240596>.

Kim, TS & Omberg, E 1996, 'Dynamic nonmyopic portfolio behavior', The Review of Financial Studies, vol. 9, no. 1, pp. 141-61.

Kingston, G & Fisher, L 2014, 'Down the retirement risk zone with gun and camera', Economic Papers: A Journal of Applied Economics and Policy, vol. 33, no. 2, pp. 153-62.

Kozup, J, Howlett, E & Pagano, M 2008, 'The effects of summary information on consumer perceptions of mutual fund characteristics', Journal of Consumer Affairs, vol. 42, no. 1, pp. 37-59.

Lang, M & Lundholm, R 1993, 'Cross-sectional determinants of analyst ratings of corporate disclosures', Journal of Accounting Research, vol. 31, no. 2, pp. 246-71.

Lehavy, R, Li, F & Merkley, K 2011, 'The effect of annual report readability on analyst following and the properties of their earnings forecasts', The Accounting Review, vol. 86, no. 3, pp. 1087-115.

Leuz, C & Verrecchia, RE 2000, 'The economic consequences of increased disclosure', Journal of Accounting Research, vol. 38, pp. 91-124.

Li, F 2008, 'Annual report readability, current earnings, and earnings persistence', Journal of Accounting and Economics, vol. 45, no. 2, pp. 221-47.

Lynch, AW & Musto, DK 2003, 'How investors interpret past fund returns', The Journal of Finance, vol. 58, no. 5, pp. 2033-58.

Madrian, BC & Shea, DF 2001, 'The power of suggestion: Inertia in 401 (k) participation and savings behavior', The Quarterly Journal of Economics, vol. 116, no. 4, pp. 1149-87.

100

Mamaysky, H & Spiegel, M 2002, 'A theory of mutual funds: Optimal fund objectives and industry organization', Yale School of Management working paper.

Massa, M 2003, 'How do family strategies affect fund performance? When performance-maximization is not the only game in town', Journal of Financial Economics, vol. 67, no. 2, pp. 249-304.

Mazis, MB, Staelin, R, Beales, H & Salop, S 1981, 'A framework for evaluating consumer information regulation', The Journal of Marketing, vol. 45, no. 1, pp. 11-21.

Merton, RC 1971, 'Optimum consumption and portfolio rules in a continuous-time model', Journal of Economic Theory, vol. 3, no. 4, pp. 373-413.

—— 1973, 'Theory of rational option pricing', The Bell Journal of Economics and Management Science, vol. 4, no. 1, pp. 141-83.

Miller, BP 2010, 'The effects of reporting complexity on small and large investor trading', The Accounting Review, vol. 85, no. 6, pp. 2107-43.

Miller, GS 2002, 'Earnings performance and discretionary disclosure', Journal of Accounting Research, vol. 40, no. 1, pp. 173-204.

Nanda, V, Wang, ZJ & Zheng, L 2004, 'Family values and the star phenomenon: Strategies of mutual fund families', The Review of Financial Studies, vol. 17, no. 3, pp. 667-98.

Navone, M 2012, 'Investors’ distraction and strategic repricing decisions', Journal of Banking & Finance, vol. 36, no. 5, pp. 1291-303.

Nelson, P 1970, 'Information and consumer behavior', Journal of Political Economy, vol. 78, no. 2, pp. 311-29.

Price, RA 1998, Price Responsiveness of Informed Investors to Increases in Financial Statement Disclosure Quality MS No. 98-3052, <Available at SSRN: https://ssrn.com/abstract=137957 or http://dx.doi.org/10.2139/ssrn.137957>.

Rennekamp, K 2012, 'Processing fluency and investors’ reactions to disclosure readability', Journal of Accounting Research, vol. 50, no. 5, pp. 1319-54.

Rutherford, BA 2003, 'Obfuscation, textual complexity and the role of regulated narrative accounting disclosure in corporate governance', Journal of Management and Governance, vol. 7, no. 2, pp. 187-210.

Schrand, CM & Walther, BR 2000, 'Strategic benchmarks in earnings announcements: The selective disclosure of prior-period earnings components', The Accounting Review, vol. 75, no. 2, pp. 151-77.

Sengupta, P 1998, 'Corporate disclosure quality and the cost of debt', Accounting Review, vol. 73, no. 4, pp. 459-74.

Sialm, C, Starks, LT & Zhang, H 2015, 'Defined contribution pension plans: Sticky or discerning money?', The Journal of Finance, vol. 70, no. 2, pp. 805-38.

Siggelkow, N 2003, 'Why Focus? A study of intra‐industry focus effects', The Journal of Industrial Economics, vol. 51, no. 2, pp. 121-50.

Sirri, ER & Tufano, P 1998, 'Costly search and mutual fund flows', The Journal of Finance, vol. 53, no. 5, pp. 1589-622.

101

Smith, M & Taffler, R 1992, 'Readability and understandability: Different measures of the textual complexity of accounting narrative', Accounting, Auditing & Accountability Journal, vol. 5, no. 4.

Stigler, GJ 1961, 'The economics of information', Journal of Political Economy, vol. 69, no. 3, pp. 213-25.

Verrecchia, RE 2001, 'Essays on disclosure', Journal of Accounting and Economics, vol. 32, no. 1, pp. 97-180.

Wachter, JA 2003, 'Risk aversion and allocation to long-term bonds', Journal of Economic Theory, vol. 112, no. 2, pp. 325-33.

Warner, JB & Wu, JS 2011, 'Why do mutual fund advisory contracts change? Performance, growth, and spillover effects', The Journal of Finance, vol. 66, no. 1, pp. 271-306.

Warther, VA 1995, 'Aggregate mutual fund flows and security returns', Journal of Financial Economics, vol. 39, no. 2, pp. 209-35.

Welker, M 1995, 'Disclosure policy, information asymmetry, and liquidity in equity markets', Contemporary Accounting Research, vol. 11, no. 2, pp. 801-27.

You, H & Zhang, X-j 2009, 'Financial reporting complexity and investor underreaction to 10-K information', Review of Accounting Studies, vol. 14, no. 4, pp. 559-86.

Zellner, A & Theil, H 1962, 'Three-stage least squares: simultaneous estimation of simultaneous equations', Econometrica: Journal of the Econometric Society, vol. 30, no. 1, pp. 54-78.

102

Appendix 1

Corporations Amendment Regulations 2010 (No.5) Schedule 10D

Item [17] also inserts a new Schedule 10D, Form and content of Product Disclosure Statement -

superannuation product, to which Subdivision 4.2B of Division 4 of Part 7.9 applies, into the

Principal Regulations containing the detailed regulations prescribing the form and content of a

simplified superannuation product PDS.

Clause 1 - Length and font size for Product Disclosure Statement for superannuation product

This clause ensures that the superannuation product PDS does not exceed a maximum page limit,

while providing for alternative formats which deliver the equivalent content. Requirements relating

to font sizes also are included to ensure that the PDS is readable.

The total length of the PDS (not including any information incorporated by reference) must not

exceed:

(a) 8 A4 pages of content; or

(b) 16 A5 pages of content; or

(c) 24 DL pages of content; or

(d) if in any other format, as long as it fits into 8 A4 pages.

The font size must not be less than:

(a) for the name, address, ABN and/or ACN and AFSL—8 points;

(b) for body text—9 points.

Further, the standard requirements under subsection 1013C(3) of the Act still apply requiring that

the PDS must be worded and presented in a clear, concise and effective manner.

Clause 2 – Minimum content of Product Disclosure Statement for superannuation product

103

This clause states that the superannuation product PDS must be made up of a number of sections

which must be numbered, ordered and titled as prescribed.

The reason for prescribing headings is to ensure comparability. Allowing issuers to vary the

headings may unnecessarily confuse consumers.

The section relating to insurance cover provided through the superannuation product can be omitted if

the product does not offer such cover and relevant sections can be renumbered accordingly.

Subclause (3) prescribes that the sections as numbered and titled in subclause (1) must be set out in a

table of contents. Given that not all potential customers may have access to the internet, simply

providing a website link means that the incorporated material is not considered to be reasonably

accessible. Therefore, subclause (3) also requires the inclusion of a telephone number in the PDS

which can be used to request a hard copy of the PDS and any incorporated information.

Subclause (4) also provides that the PDS must advise the reader that it contains references to

important information, and that consumers should consider that information before making a final

decision to invest in the product. The warning also states that the information provided is general

information only and does not take account of the client’s personal financial situation or needs, and

that the reader should obtain financial advice tailored to their own personal circumstances. This

statement must be displayed in a prominent style and position at or near the beginning of the

document.

Subclause (5) allows the PDS to also include other sections and information at the discretion of the

responsible person, but these must fit within the prescribed page length. Any additional sections

need to be located at the end of the prescribed sections.

Subclause (6) allows information that is required to be included in a PDS, to be included where it is

applicable only. Further, no reference needs to be made if a particular requirement is not applicable.

Clause 3 - Contents of section 1 (About [name of superannuation product])

This section provides a short summary of the superannuation fund and the products it is offering.

104

Clause 4 - Contents of section 2 (How super works)

This section provides important prescribed statements summarising key elements of how

superannuation works, as follows:

• that superannuation is in part compulsory and is a way of building savings for retirement;

• that there are different types of contributions that can be made by an investor (e.g. employer

contributions, voluntary contributions, government co-contributions);

• that there are limitations on contributions and withdrawals;

• that tax savings are provided by Government; and

• that investors have a choice into which fund to direct their superannuation guarantee contributions.

Further detailed information on each of these matters may be provided by incorporating information

by reference or by including a reference to a website operated by or on behalf of the

Commonwealth. It is expected that the PDS refer to ASIC’s consumer website, FIDO, or equivalent.

Clause 5 - Contents of section 3 (Benefits of investing with [name of superannuation product]

This provision prescribes what should be included in section 3. It needs to start with a summary of

the relevant superannuation product and include a summary of its significant features and benefits.

Further information on the features and benefits of the super fund or superannuation in general can

be provided through incorporation by reference.

Clause 6 - Contents of section 4 (Risks of super)

This item prescribes the content in section 4 of the PDS. Section 4 needs to include a number of

statements addressing certain general risks of investing, including that all investments carry risk;

that different strategies carry different levels of risk; and that higher long-term returns often carry

the highest short-term risks.

Clause (2) states that section 4 is also required to include a summary setting out the significant risks

105

of the particular superannuation product to which the PDS applies.

Section 4 has to contain a number of statements addressing the key risks of investing in super

generally (provided these risks have not already been addressed in clause (2) above), including the

risk that:

• the value of investments will vary;

• the level of returns will vary, and that future returns may differ from past returns;

• returns are not guaranteed and members may lose some of their money;

• superannuation laws may change;

• a person’s superannuation savings may not adequately provide for their retirement; and

• the relevant level of risk for each member will vary depending on a range of factors including

their age, investment time frame, the member's other investments and their individual risk tolerance.

Additional information about significant risks can be provided by incorporation by reference.

Clause 7- Contents of section 5 (How we invest your money)

This provision provides the content requirements for section 5. Section 5 must contain a summary

description of the investment options offered by the fund, and what happens if the member does not

make a choice of where to invest, i.e. whether there is a default option. There must also be a

warning to the member to consider the likely investment return, risk and their investment timeframe

when choosing which option to invest in.

Under subclause (3) prescribed information needs to be set out for at least one investment option, as

follows:

• the name and a short description of the option, including which type of investor it is best suitable

for;

• its asset mix and the strategic allocation of the asset classes - this may be done in the form of a

range;

106

• the investment return objective of the fund. A short explanation of this measure should be

provided;

• the minimum timeframe for holding the investment - this can be done in the form of a range; and • a

summary description of the risk level of the option. This could, for instance, be done in the form of

a risk meter.

If the fund has a default option this section requires that this information must be provided for that

default option, regardless of what other information has already been provided for any other

investment option. If there is more than one default option, the information must be provided for the

default option which has the most funds invested.

If the product has no default option, it must disclose the required information for a balanced

investment option (the definition of a balanced option is taken from the enhanced fee regulations in

the Principal Regulations - refer item 101 of Schedule 10). If neither a default option or a balanced

investment option exist, the information must be provided for the investment option under which

the superannuation trustee has the most funds invested.

The information prescribed in subclause (3) must be provided for all other investment options as

well, but may be included by using the incorporation by reference mechanism.

Under subclause (9), the superannuation trustee is also required to provide information about

switching investments and the extent to which investment options can be changed and how.

Information about the extent to which labour standards or environmental, social or ethical

considerations are taken into account in the investment activity relating to the product are also

prescribed. Where labour standards or environmental, social or ethical considerations are not taken

into account, this must be stated.

The information prescribed under subclause (9), as well as further information about investment

options, can be incorporated by reference.

Clause 8 - Contents of section 6 (Fees and costs)

The provisions in section 6 regulate the disclosure of information relating to the fees and costs of

the product. In general, these provisions are modelled as closely as possible on the enhanced fee

disclosure regulations in Schedule 10 of the Principal Regulations. This is intended to minimise the

107

compliance cost for industry, as they are already providing this information in their current PDSs.

Section 6 needs to set out summary fees and cost information in a prescribed format. The format is

essentially taken from the enhanced fee disclosure regulations in Schedule 10 to the Principal

Regulations. It has to be provided for each investment option presented in detail in section 5.

The section also has to start with a consumer warning regarding the potential impact of fees and

costs on the final balance for a member, illustrated with a concise example. The format for the

warning and example is taken from Schedule 10 of the Principal Regulations.

This must be followed by a table summarising the prescribed main fees and costs information for

the option(s) disclosed in Section 5. The rules in completing the table are taken from Schedule 10 to

the Principal Regulations, with some minor amendments to ensure that the information provided is

clear and simple. Certain information relating to fee changes also has to be provided as prescribed

in Schedule 10 to the Principal Regulations.

A statement has also to be provided that the information in the table can be used to compare fees

and costs between different funds. A further short general statement explaining that fees and costs

can be paid directly out of the member’s account or deducted from investment earnings also needs

to be included.

Section 6 also has to provide a worked example for one investment option, which must be the

default option in the product. If there is no default option, the worked example must cover the

balanced option, and if no balanced option exists, then the option with the most funds under

management. The rules for providing the worked example are taken from Schedule 10 to the

Principal Regulations.

A reference to ASIC’s calculator on ASIC’s consumer website, FIDO, (or its equivalent) has to be

provided, with an explanation that these can be used to calculate the impact of fees and costs on

account balances. A reference to the fund trustee’s own calculator on its website (if available) may

also be provided.

The section must contain a warning that additional fees may be paid to a financial adviser if

applicable, and a reference to the Statement of Advice must be provided where details of these fees

can be obtained.

The superannuation trustee must incorporate information on fees and costs for all investment options

calculated and presented as prescribed in Schedule 10 to the Principal Regulations and may use the

108

incorporation by reference mechanism to do so.

If the trustee wishes to provide further information in relation to fees and costs, this information

may be incorporated by reference.

Clause 9 - Contents of section 7 (How super is taxed)

This provision provides for the content requirements of section 7. Section 7 must contain a summary

of tax information relating to superannuation funds, including how tax amounts are paid and the

main taxes payable for contributions, fund earnings and withdrawals.

Subclause (2) requires that section 7 include a warning that the member’s tax file number (TFN)

needs to be provided and a statement of the consequences if it is not, as well as a warning of the tax

consequences of exceeding the contribution caps, where this is relevant.

Additional information about taxation matters relating to superannuation can be incorporated by

reference.

Clause 10 – Contents of section 8 (Insurance in your super)

This item in subclause (1) provides for the inclusion of information in section 8 of the PDS on any

insurance cover included in a superannuation product. If no insurance is included this section can be

omitted.

Subclause (2) says that if insurance cover is provided, section 8 needs to summarise the main types

of insurance cover available and how members can apply. There also needs to be a statement that

there are costs associated with the insurance cover and describe in summary who is responsible for

paying the costs and how they have been calculated.

If default insurance cover is provided, subclause (3) requires that a summary description must be

provided of the type and level of cover, the cost expressed in actual dollars or as a range, who is

responsible for paying the costs, whether members can decline or cancel the cover, and how they

can do this; and whether the insurance cover can be changed and how.

A further warning needs to be included to the effect that unless members cancel their default cover,

109

the cost thereof will be deducted from their account or from their contributions (whichever is

applicable).

Further information has to be provided about eligibility for and cancellation of the insurance cover

and any conditions and exclusions.

Subclause (4) requires that if insurance cover is an option rather than a default, the section must

include information about the level and type of cover available, the cost expressed in actual dollars

or as a range, eligibility rules, cancellation of cover and any conditions and exclusions applicable.

Other significant matters in relation to the insurance cover also need to be included - for example,

how a person can apply or cancel the insurance.

Under subclause (5), the issuer may incorporate by reference information about eligibility for and

cancellation of insurance and any conditions and exclusions where the insurance cover is provided

by default. If this information is provided by incorporation by reference, the PDS must include a

warning in section 8 that the matter may affect a person’s entitlement to cover and that the

information should be read before deciding whether the insurance is appropriate.

If information about the level and type of insurance cover available, the cost or the range of costs or

any other significant matter is provided by incorporation by reference, the PDS in Section 8 must

include a warning that the information needs to be read before deciding whether the insurance is

appropriate.

Clause 11 - Contents of section 9 (How to open an account)

This clause prescribes information for inclusion in section 9. Section 9 must provide information,

where applicable, on how to open an account with the superannuation provider, explain the cooling-

off period that applies and how to make a complaint.

Subclause (2) provides that further detailed information may be provided about cooling offer

periods, complaints and dispute resolution and that this may be done by incorporation by reference.

Where the previous section is not applicable, that is insurance is not offered in the superannuation

product, this section can be presented as section 8 (refer Schedule 10D, Item 2(2)).