CLUSTER PROJECT 11: CSIRO MULTI-DISCIPLINARY RESEARCH PROJECT

Project Overview:

This multi-disciplinary project is structured into five themes: Government Data Analysis, Retirement Income Forecasting,  Retirement Investment Life-Cycle, Retirement Product Design and Behavioural Economics and Digital Services.

Principal Researchers: Dr Andrew Reeson (CSIRO Data61), Dr Zili Zhu (CSIRO Data61)

Outcomes:

Theme 1: Government Data Analysis

ATO Data Analysis on SMSF and APRA Superannuation Accounts, September 2015, CSIRO, Zili Zhu, Thomas Sneddon, Alec Stephenson, Aaron Minney

Currently, over $500 billion of assets are controlled by SMSFs. To gain more insight into SMSFs, it is useful to compare SMSFs with conventional superannuation funds that are regulated by Australian Prudential and Regulatory Authority (APRA). The Australian Tax Office (ATO) provided CSIRO with a large dataset of individual and self-managed superannuation fund (SMSF) annual return information to facilitate the comparative analysis of APRA-regulated superannuation funds and SMSFs. The analysis of SMSF and APRA fund activity can shed light on how older Australians behave in relation to withdrawal, contribution and maintenance of their superannuation entitlements. This study represents the first time the original raw ATO return data has been used directly as evidence. In this report, we provide the outlines on the analysis of this ATO data, and also some insights into the behaviour of older Australians in relation to their superannuation fund entitlements.

This paper provides a longitudinal study of withdrawals from account-based pensions from superannuation savings to provide a better understanding of drawdown patterns in retirement. Our analysis indicates that most retirees in their 60s and 70s draw down on their account-based pensions at modest rates, close to the minimum amounts each year. It also suggests that most retirees would die with substantial amounts unspent if these drawdown rates were to continue. These findings are consistent with existing empirical evidence which indicates that retirees are inclined to draw down their wealth relatively slowly.

Summary | Working Paper

Theme 2: Retirement Income Forecasting

This study examines the likely effect of the new schedule for compulsory superannuation contribution rates introduced in September 2014. The CSIRO Simulation of Uncertainty for Pension Analysis Model (the ‘SUPA’ model) has been developed to assist superannuation- related research within the CSIRO-Monash Superannuation Research Cluster. We use the SUPA model to study retirement outcomes under both the new superannuation contribution rate regime and the previous schedule.

Currently, there is no consensus on the minimum superannuation amount required for retirees to maintain a reasonable living standard throughout retirement. An unambiguous target is important as it can help inform individuals as well as the government when it is setting superannuation contributions rates. Using CSIRO retirement income modelling analytics, this study calculates the minimum superannuation fund balance needed to maintain a comfortable or modest retirement lifestyle.

Given the high level of uncertainty involved  in estimating future investment  returns and    the required expenditure for a certain retirement standard, a logical approach would be to also provide the probability of success in meeting a particular retirement living standard. For example, if a 95 per cent probability of success is regarded as the acceptable risk tolerance level, the simulation results suggest that $851,000 is a sufficient superannuation balance at retirement age to achieve a comfortable retirement, and there is only a 5 per cent chance of exhausting this balance during the retirement phase. When eligibility for  the full age pension is met, a super balance at retirement of only $73,000 is needed to provide a modest retirement income, with a 95% probability.

The CSIRO SUPA model is also available in a software form for use by stakeholders as part of the deliverables.

At this stage, there has been no consensus on the minimum superannuation amount required for retirees to maintain a reasonable living standard throughout retirement. An unambiguous target is important as it can help inform individuals as well as the government when it is setting superannuation contributions rates. Using CSIRO retirement income modelling, this study calculates the minimum superannuation fund balance that is needed to maintain a comfortable or modest retirement lifestyle.

Summary | Publication

This study examines the likely effect of the new schedule for compulsory superannuation contribution rates introduced in September 2014. We use the SUPA model to study retirement outcomes under both the newly legislated superannuation contribution rate regime and the previous schedule. The CSIRO Simulation of Uncertainty for Pension Analysis Model (the ‘SUPA’ model) has been developed to assist superannuation-related research within the CSIRO-Monash Superannuation Research Cluster.

Summary | Publication

Theme 3: Retirement Investment Life-Cycle

On retirement at age 65, a retiree may not feel comfortable that their balance is sufficient to fund their retirement due to longevity risk. To mitigate this risk, retirees may purchase an inflation-linked annuity, which pays a guaranteed amount each year until death. However, they need to decide what is the optimal payout period for the annuity and how much of their superannuation balance should be used to buy an annuity that ensures they live comfortably in retirement.

We model these two choices facing retirees using a real-option approach. First, we find the optimal portfolio of financial instruments that maximises the terminal wealth of the superannuation balance over a 30-year planning horizon.

Then we find the optimal sequence of annuity purchases using funds from the chosen superannuation balance to ensure that the retiree has sufficient funds to cover their living costs over the 30-year planning horizon. Based on least-squares Monte Carlo simulation, we also develop an efficient method for solving dynamic portfolio selection problems in the presence of transaction costs, liquidity costs and market impacts. The CSIRO optimal portfolio algorithm-based on the real-option approach is available to stakeholders.

Theme 4: Retirement Product Design

To take advantage of market growth and protect savings, many asset management companies worldwide are offering investment protection options such as capital protection options. US insurers began including these types of options in their variable annuity products in the 1990s and these products have become popular in Europe, the UK and Japan and, more recently, in Australia. Pricing of these contracts is similar to pricing variable annuities with guaranteed minimum withdrawal benefits (GMWB), however, there are several important differences making it a more challenging task.

We develop a new method to more accurately price variable annuities. We extend    the methodology for pricing variable annuities to account for stochastic interest rates, stochastic life tables and stochastic volatility. We also model causes of death (including long-term forecasts) in Australia and account for this in the valuation of annuity portfolios. Note that here we develop new modelling methodology different from the time series approach.

Theme 5: Behavioural Economics

Behavioural Economics and Digital Services, Andrew Reeson, Claire Mason, Murni Greenhill, Elizabeth Hobman, Sorada Tapsuwan, Iain Walker

One of the biggest issues facing the superannuation sector is low member engagement. A review of the behavioural and psychological literature has identified how digital service delivery might help enhance engagement, through promoting perceptions of

competence, relatedness and autonomy, and hence building intrinsic motivations towards superannuation. In order to improve our understanding of how to better engage members we are running a set of survey studies to investigate how people respond to automated financial services. A first survey with 138 respondents provides us with some key insights into the current attitudes towards automated advice, indicating that acceptance of automated advice decreases with age and increases with income. Follow-up work will test how people respond when exposed to digital advice and primed for perceived competence, relatedness, autonomy and trust.