Ensuring that retirees have sufficient financial resources to support their health and well being in retirement represents one of the major challenges for society in our age. Australia is an ageing society, with retirees representing an ever-growing share of the overall population. Over the next 10 years, Australia’s baby boomer demographic will move into the retirement phase and become the healthiest, wealthiest retirees in Australian history. At the same time, the rising cost of health care and growing size of the retired population is straining the public purse. The question of how to reduce the call on public finances while ensuring that retirees have sufficient income to support their expected standard of living is a hotly contested area in the today’s public policy debate.
The CSIRO-Monash Superannuation Research Cluster commenced in 2013 with the aim of undertaken deep research to help inform this important area of public policy. The Cluster undertakes this work through 11 work streams that represent research that cover the role of superannuation in the economy; the technical aspects of funds management in the superannuation sector; and social or behavioural considerations for Australians over 60.
Now halfway through its 3-year plan, the Cluster has produced more than 30 papers, with a number of papers already accepted for publication in highly rated academic journals. Original datasets being developed under the Cluster are starting to yield results, as are the streams undertaking modelling work.
With the research now starting to bear fruit, this year’s Symposium showcased some exciting work. The 2nd annual CSIRO-Monash Superannuation Research Symposium was convened on 1 December 2015 by the Australian Centre for Financial Studies. The conference provided a forum for academics to share findings from the CSIRO-Monash Superannuation Research Cluster with stakeholders, policymakers, researchers and industry participants.
Issues that have arisen in the public debate arose throughout the conference, including a re-examination of the purpose of the superannuation as either a wealth accumulation vehicle or a system to produce retirement income; approaches to fund management for retirement income; the development and pricing of retirement income products; the financial implications of rising health care costs for ageing Australians; gender issues in superannuation; financial literacy amongst retirees and financial product design; and many more.
Below we provide an overview of the presentations and discussion at the Research Symposium. As we head into the final year of the current Cluster, we are excited to see the finalisation of the current work streams and to engage with industry practitioners, regulators and policymakers on the research findings so as to inform policy and practice.
How should we design and assess institutions which deliver retirement income?
In his plenary address Professor Clark discussed the need for governments and policymakers to define the overall objective of their pension system. Pension delivery institutions (superannuation funds) also need to define and understand their purpose. This purpose may seem nebulous – a sustainable flow of retirement income is hard to measure – but it should dictate the structure of the institution. Defining a singular purpose is difficult as objectives may conflict, for example a fund may want to ensure capital is never reduced through losses while earning high returns and operating at a low cost. These may be mutually incompatible objectives; for example, earning and high and stable return on assets over time may require retaining a higher-cost fund manager. Obtaining sufficient information on fund members to design appropriate products may require significant capital expenditure on modernising internal reporting and IT systems.
The board has a vital role to play in identifying and agreeing on a shared purpose for the organisation, and embedding this through the organisation. This can be difficult when board membership is granted as a political perk. An agreed purpose is required if boards are to make necessary, long-term decision, such as investing in skills and experience, and infrastructure.
This institution’s purpose should also be used to determine salaries of senior executives (rather than benchmarking against external financial services firm) and manage claims for power and prestige within the fund.
Professor Giesecke presented his computable general equilibrium (CGE) model of the Australian economy with a new module that accounts for the financial sector. Modelling of the financial sector is an important step forward because, in Professor Giesecke’s words, “most economic models have been completely silent on what’s happening in the financial markets”. This model, developed at the Centre of Policy Studies, allows financial instruments (i.e. bonds, cash, deposits and loans, equity, and gold and special drawing rights) to be incorporated as the assets and liabilities of a range of different economic agents. The superannuation sector is modelled as one of a number of financial agents mediating the supply and demand for financial instruments.
Professor Giesecke simulated a 1 per cent increase in the superannuation guarantee rate. The consequences of this shock can be decomposed into two components: an intermediation effect (an increase in the proportion of household savings routed into the superannuation sector) and a savings effect (a rise in the household savings rate). The simulation predicts a decrease in the current account deficit due to the rising savings rate, but also a depreciation of the Australian dollar as increased intermediation of funds through the superannuation sector would see a rise in outbound investment. Employment would suffer in the short run as a higher savings rate would reduce consumption in the first 1 to 3 years following the change, but higher employment would result after this period as the economy adjusted to the higher savings rate and benefitted from growth in the capital stock, with mining and construction anticipated as being the biggest winners. However, the simulation suggests that a higher contribution rate does not materially affect macroeconomic stability. The datasets being used for the modelling predate the significant growth of self-managed super funds in Australia, which may be an area for future inquiry. Overall, these findings provide valuable guidance to policymakers and regulators.
The next session saw presentations by three CSIRO staff. Dr Reeson analysed superannuation withdrawal patterns using Australian Taxation Office (ATO) data. Interestingly, he found little evidence of ‘decumulation’ for either APRA funds or self-managed superannuation funds (SMSFs) – in fact, both types of fund continue to grow in value (on average) throughout people’s lives, with only a modest decline after the age of 70. The implication of this finding is that today’s retirees are unlikely to utilise the majority of their superannuation during their lifetime. He suggested at least two motives: one is the desire to leave a bequest, and the second is the psychological challenge of switching from saving throughout one’s life to consuming in retirement. In addition, minimum drawdown rates may be acting as a ‘default’ that retirees hold in mind as a benchmark for appropriate expenditure to mitigate against longevity risk.
Dr Zhu presented a cascading multi-factor model to forecast future economic scenarios, and a ‘simulation with uncertainty for pension analysis’ (SUPA) model to forecast superannuation fund balances. He outlined a framework for retirement life-cycle management whereby retirees use dynamic strategies to decide if they should, say, invest in a life annuity or invest in the share market (or progressively switch from one to the other). Dr Shevchenko examined retirement product design and the pricing of variable annuities with guarantees. He noted that pricing these products, which are affected by equity risk, interest rate risk, systematic mortality risk and human behaviour, is a difficult numerical problem. Current industry pricing techniques are computationally intensive. CSIRO is developing prototype software to accurately price such products in real time.
Associate Professor Lajbcygier presented on the use of a nonparametric ‘generalised additive model’ (GAM) to measure market impact costs, i.e. the cost associated with a large trade moving the price of a security with low liquidity. Associate Professor Lajbcygier and Mr Pham suggested that market impact costs are particularly severe for alternative index and ‘smart beta’ products that are not anchored to price. While alternative indexing might appear to perform better than market cap-weighted indexing (MCWI) without costs, market impact is a large hidden cost of trading that is often overlooked. As the turnover of the average active fund is 100 per cent per annum (versus just 15 per cent for index funds), market impact costs tend to add up over time.
Dr Tupitsyn discussed passive hedge funds. His research found that two-thirds of hedge funds exhibit linear factor exposures, one-fifth exhibit nonlinear factor exposures, and one-fifth do not have significant linear or nonlinear exposures to alternative risk factors. Surprisingly, nonlinear active funds underperform linear passive funds. If understanding nonlinear active fund performance as a proxy for “beta” in a portfolio, the findings suggest that active fund managers do not outperform passive fund managers over time. This adds to evidence suggested in other studies that it is challenging for active fund managers to consistently produce superior returns over time. This may be an important consideration for superannuation fund managers when mandating funds to external fund managers.
Associate Professor Bianchi examined the predictive performance of conventional asset pricing models in forecasting infrastructure returns. He noted that it makes sense that superannuation funds should be attracted to infrastructure, given the long-term nature of its cash flows. Using 16 years of monthly return data, he found that simple, fixed excess return models tend to outperform the Capital Asset Pricing Model (CAPM) and the ‘Fama-French’ three-factor model in predicting listed infrastructure returns over a number of time horizons. In other words, the simple historical mean return seems to outperform conventional asset pricing models. Future research work will examine unlisted infrastructure.
Associate Professor Bianchi also presented on the question of whether infrastructure is a unique asset class. Some academics and industry professionals believe that it is an asset class, while others believe it is a subset of asset classes that are already available. Applying an approach known as the ‘Merton zero-intercept criterion’ to seven MSCI world and regional infrastructure indices, he found that global and European infrastructure index returns can be replicated with a linear combination of world stocks, world utilities, large-cap returns and growth-related returns. These findings suggest that listed infrastructure cannot be defined as a separate asset class. One exception is listed infrastructure returns in the Asia-Pacific region, where returns could not be replicated utilising the methodology that was successful in other jurisdictions.
Professor Harris opened this session by outlining the demographic shift as the ‘baby boomer’ generation moves into retirement and life expectancies increase. While longer life expectancy is a good thing it poses challenges and costs to individuals, government and society. Chief among this is growing healthcare expenditure. Currently, these costs are not well understood, with no robust data showing total healthcare spending by age. Despite this, healthcare costs rise significantly in later years, as health declines. Professor Harris’ work models the effect of increasing life expectancy on healthcare costs. This primarily depends on whether increased lifespan results in an increase in the number of ‘healthy’ years – in which case large healthcare costs are simply deferred – or an increase in the number of ‘frail’ years – which leads to large health costs being incurred for a longer period and significant increase in total health expenditure. Healthcare costs are also expected to increase as wealth grows. Traditionally, healthcare pricing in Australia supply-driven, and it will be interesting to see how providers respond to the increasing age and wealth of their customers. This may result in greater choice and increased variability in costs of products and services being offered.
Dr Kassenboehmer’s paper focuses on the relationship between ‘locus of control’ (a psychological concept that captures individuals’ beliefs of the causal relationship between their own behavior and life events) and savings behaviour. Households with an internal locus of control tend to save more and accumulate more wealth. They also hold less financial wealth and significantly more pension wealth. Her findings suggest that perception of control is as important as human capital and cognitive skills in explaining wealth accumulation and portfolio allocations. The work also has policy implications, suggesting households with low perceptions of control are a sensible group to target for intervention as they tend to save less and allocate less wealth to their pensions.
Professor Gerrans’ research explores investment strategy choice in superannuation. Most employees make few explicit decisions regarding their retirement savings trajectory, instead relying on the default superannuation fund nominated by their employer and its default investment strategy. Analysis of the Mercer administrative database (up to 10 years of observations for almost 260,000 individuals covering 177 employer sub-plans) shows one in six employees changed how either their savings balance or future contributions were invested, and one in 12 changed how both were invested. However, it is important to note that lack of action does not necessarily imply employees are not engaged with their superannuation. Investment changes are more likely among men, employees with high balances, and in response to policy changes. Peer group effect on investment behaviour is also observed.
There is a large gap between the superannuation savings of Australian men and women, explored in research presented by Professor Whiteside and Dr Feng. The gap is larger than the pay gap (due to the effect of compound interest) and has significant budget implications as most women are reliant on the Age Pension. One possible cause is interrupted work patterns – research suggests 5-6 years out of the workforce leads to superannuation balance 17-25% lower at retirement. The superannuation system assumes approximately 10% contributions from full-time work over 40 years. As more employees become self-employed, work part-time and experience career breaks, low superannuation balances may become more common.
The Superannuation Guarantee is scheduled to increase to 12 per cent by 2020 – how will this impact retirement income adequacy, and could the same gain be achieved through investment strategy alone? Professor Drew’s research finds increasing contributions improves retirement adequacy in expectation, however it also increases sequencing risk (especially when coupled with static asset allocation), and the proportion of retirees who will outlive their savings is not significantly reduced. A dynamic asset allocation to manage sequencing risk exposure could achieve the same improvement in retirement adequacy without any increase in contributions.
How should you set expenditure levels in retirement to minimise the risk of financial ruin, and how do major health- or aged-care cost affect this? Dr Walk’s research shows unexpected health costs which increase longevity pose the greatest risk to income stability in retirement (that is, lead to the greatest risk of financial ruin). Optimal withdrawal (expenditure) rates in retirement are highly sensitive to the timing of health costs and, to a lesser extent, to later-life aged care expenses. In response to large health costs that threaten income security, risk of ruin can be mitigated using a dynamic lifecycle investment strategy.
In discussion, the authors noted that dynamic investments strategies are not a silver bullet to remove all risks to retirement income. Instead, higher contributions, insurance products and investment strategies should all be used together to minimise risk.
Professor Lindley discussed the large diversity of situations and pathways to retirement for older workers. Individuals’ decisions about retirement depend on preferences, labour market opportunities, other activities (such as caring, volunteering and leisure) and savings. Policies encouraging the employment of older workers should carefully asses the value, and costs, of them remaining in the workforce. Employers and government will play a role in extending working lives. For employers this may include offering training and development opportunities, and internal mobility (easier for larger organisations). For government this will include changing attitudes about the value of older workers, but should be careful that this does not lead to ‘ageist’ outcomes which penalise younger workers.
Professor Taylor echoed this point, and noted that empirical evidence highlights ageism against younger, not older workers. His work surveyed employers to gauge their reaction to the ageing workforce, particularly in light of the economic slowdown. He found employers are experiencing labour supply challenges, however there is little evidence that older workers are suffering because of this.
- The social construction of retirement and evolving policy discourse of working longer Philip Taylor and Catherine Earl
Mr Duffield remarked that the conference had done a terrific job of bringing together a lot of research that is relevant to the superannuation industry. He described the presentations as “interesting, insightful, innovative and intelligent.” Mr Duffield suggested to industry attendees that many of the academics may be willing to present at their offices, helping to build useful academic-practitioner partnerships. Professor Ralston, lead investigator on the CSIRO-Monash Superannuation Research Cluster, noted that the CSIRO-Monash Superannuation Research Cluster is an iterative program and there would be opportunities to shape the research as it evolves.
Mr Bone commented that he found something interesting in almost every presentation, and saw great potential for some of the research to be expanded. For example, the model presented by Professor Giesecke could be extended to simulate changes in taxation or other policy changes. Mr Bone thought that Associate Professor Lajbcygier and his team did a very good job of “debunking a lot of what the industry would like you to believe,” and that Dr Reeson’s presentation on withdrawal patterns raised important questions about the lump sum payout system. Mr Duffield added that research into retirement income streams had only just begun and yet was already revealing surprising new insights. Mr Keary added that the Cluster will play an important role in building the evidence-base behind superannuation policy. A number of papers presented highlighted how few data are available on important issues such as drawdown rates and healthcare costs in retirement.