Deliverable 2015_2. Retirement savings trajectories: An analysis of the experience of fund members. Part One: Experience

Transcription

1 Deliverable 2015_2 Retirement savings trajectories: An analysis of the experience of fund members Part One: Experience Paul Gerrans 1, Maria Strydom 2, Carly Moulang 3, Jimmy Feng 2, Maurizio Fiaschetti 4, Gordon Clark 4,2 1 UWA Business School, The University of Western Australia (corresponding author paul.gerrans@uwa.edu.au) 2 Department of Banking and Finance, Monash University 3 Department of Accounting, Monash University 4 Smith School of Enterprise and Environment, University of Oxford Acknowledgements. Research for the paper was supported by the CSIRO-Monash Superannuation Research Cluster, a collaborative project between the Commonwealth Scientific and Industrial Research Organization (CSIRO), Monash University, The University of Western Australia, Griffith University, and The University of Warwick. Noted thanks to the invaluable research assistance of Jacqui Whale, UWA Business School. 1

2 1 Introduction Individuals confront a considerable array of choices when it comes to the occupation they pursue and, associated, the employer they select. Choices are constrained, however, by factors including individual characteristics (eg. education, age, skill) and external factors (eg. social norms, state of the economy). The consequence of these choices set the individual on a consumption and wealth accumulation trajectory over their life, such as stylised in the wellknown life-cycle model (Ando & Modigliani, 1963; Modigliani & Brumberg, 1954, 1980). Since 1992, all employees 1 in Australia have enjoyed a common condition of employment, namely an entitlement to payment of retirement savings contributions by their employer to a complying retirement savings (superannuation) fund. While this is a universal entitlement, individuals can alter their retirement savings trajectory, and ultimately their retirement standard of living, through choices they make, most notably through additional savings and the investment strategy applied to these savings. We are interested in the latter. Specifically, we are interested in the extent to which individual trajectories are influenced, or nudged, by demographic and social factors in retirement savings choices. This paper is the first of three examining investment strategy choice in superannuation. The focus of this first paper is descriptive, with the primary objective of documenting the level of investment choice using a large sample of members from a large, multiple employer, superannuation trust. A breakdown of investment activity is considered for a range of key member demographics. In addition, we examine the time variation in investment activity across the key demographics. Following this work, a second paper will examine the roles of moderating and mediating factors in member investment activity. The final paper will consider the consequences and outcomes that are associated with different investment choice. 1 With minor qualifications including age and amount earned. 2

3 Investment choice is not universal for all individuals in that investment choices made available are a feature of the particular plan they are enrolled in, and the provider of that plan. One key dimension of investment choice is whether changes to such choice are applied to future savings and/or to accumulated savings. These have previously been described as Contributions Investment Changes (CICs) and Balance Investment Changes (BICs) (Gerrans, 2012) and desired and actual allocations, respectively, by Agnew, Balduzzi & Sunden (2003) who note that the two coincide at the point of rebalancing. Agnew et al. (2003, p. 197) also suggest that desired allocations are most likely to reflect a participant s intentions. While it is easy to identify characteristics which may dictate that this would be true, it is also easy to identify the contrary. Those with small balances, in the first years of opening an account, with higher income and therefore contributions, are expected to be more likely to be concerned with investment changes to future savings or contributions. Those with large balances, with longer account memberships, and older members are expected to be more likely interested in the investment strategy applying to their accumulated savings. Given this, we prefer the neutral labels CICs and BICs to identify these two types of choices that are available to members. We investigate investment activity by making use of an administrative database provided by Mercer (Australia) Super Trust with details of 177 different employer sub-plans with a pooled membership over 258,113 individuals with up to ten years of observed behaviour. A feature of the trust structure is that while the overarching Mercer imprint is largely common across subplans (including investment choice menu and administrative processes for exercising investment choice) there remains variation among sub-plans. These variations may reflect negotiations as employers entered the trust (for example, incorporating an existing investment option into the regular menu of investment menu), but also include variation in the chosen default investment strategy. 3

4 2 Background and Choice Literature 2.1 Australian Retirement Savings System Australia s current retirement savings system architecture was established in 1992 through the Superannuation Guarantee (SG), which had its origins in workplace agreements of the 1980s. 2 The SG mandates the payment of contributions by an employer on behalf of an employee to a complying fund. 3 Though the system has been in place since 1992, the contribution rate was initially modest (three percent), gradually becoming a more significant nine percent only in 2002, and is currently 9.5 percent for the foreseeable future. 4 The first cohort of workers that will have received at least nine percent contributions over their working life, therefore, won t retire, or at least be able to access 5 their retirement savings until at least The current system, therefore, contains a heterogeneous mix of retirement savers who may have had more than a decade accumulating retirement savings at a rate of nine percent or more, and some with two decades of accumulation at the lower and higher rates. Those with fractured employment histories, notably women, will have equally fractured contributions histories. This heterogeneity is reflected in the rich database available for investigation. Before examining patterns of overall investment choice within the available database, it is worthwhile to review recent legislative changes which, while they occur after the period of data available and therefore don t impact the analysis per se, may impact the relevance of conclusions that can be drawn. 2 See (Coates, 2004) for a discussion of the history. Superannuation existed much earlier primarily for white-collar employees but widespread coverage was formalised in Exclusions apply to the requirement including if monthly earnings less than $450 or if under 18 years of age and working 30 hours or less per week. 4 A previous commitment to increase the contributions rate to 12 percent has been stalled and remains a distant prospect to be in place from 1 July 2025 ( 5 Superannuation savings are preserved until at least age 60 years for those born from 1 st July

5 In the original system design, choice was assumed to play a central role in driving the efficiency of the overall superannuation system and, by extension, beneficial outcomes for members. The review of the superannuation system (Commonwealth of Australia, 2010) made clear that this foundation had underpinned the system since its inception. As a consequence of the review, however, a clear philosophical change in approach was argued: If members want to engage and make choices, then the system ought to encourage and facilitate them doing so. If members are not interested, then the system should still work to provide optimal outcomes for them. (Commonwealth of Australia, 2010) As a consequence of the review s recommendation, the Federal Government legislated that from January 1 st, 2014 Superannuation Guarantee contributions could only be made to a new type of superannuation product, designated as an approved MySuper product which became the new default option, for those who had not actively chosen their investment strategy. Note here that the distinction is also made between ongoing contributions and the accrued balance that a member has within a fund. At present the requirements regarding the new MySuper product relates to contributions, while a member s accrued balance in the fund s default option must be transferred to a MySuper product by 1 st July Consistent with the view that enhanced protection was needed for those members who made no choice, the MySuper product impacts trustees more directly than members. Trustees are, arguably, more explicitly responsible to members who remain in a MySuper product. Trustees are constrained in the types of fees they are permitted to charge in the MySuper product as well as the insurance they are required to include. Notwithstanding potential benefits that may flow from the new MySuper product to members, which becomes an empirical question given the passage of time, substantively the member s investment choice options 6 This distinction has relevance in the analysis in this paper as investment strategy choices can be made to contributions only, existing balances only, or both. More is discussed in section three. 5

6 remain unaffected. In a MySuper regime, a member retains the ability to choose to exercise an investment strategy choice and move out of, and into, the MySuper product. If a member s investment choice retains a MySuper product as one of her choices that portion of her balance left in the MySuper product remains under MySuper rules. If no balance is left in the MySuper product the member is no longer covered by the MySuper rules. From a financial asset perspective it is just another investment strategy option, typically a balanced or diversified option with exposures to equity, property, cash, etc. as previously. As such the analysis conducted in this paper remains very much of relevance to fund trustees concerned for the behaviour of their members as well as government and regulators who are interested in overall engagement in retirement savings investment behaviours and the differences attributable to member and plan characteristics. Two important choices for individuals to make for their retirement savings are whether to make additional savings, that are beyond the compulsory employer contributions, and whether to change from the default investment strategy for those contributions. The former choice has been examined with the current dataset by Feng and Gerrans (2014) who document a significant decline in voluntary contribution participation over time. The latter choice is available within 68 percent of Australian superannuation funds who account for 98 percent of industry assets (Australian Prudential Regulation Authority, 2014). As noted earlier, investment strategy choice can be applicable to future contributions, hereafter referred to as Contributions Investment Changes (CICs), and/or the accumulated balance, hereafter referred to Balance Investment Changes (BICs). Data is not available on the availability of investment choice at this level across the industry. Within the Mercer Super Trust members can make both CICs and BICs, with few limitations or direct cost. We examine the incidence of these choices and the explanatory role of a rich set of individual and social 6

7 (including sub-plan level) characteristics which serve to nudge members into making such choices. 7

8 2.2 Retirement Planning and Relevant Characteristics Countries have adopted various blends of state and individual responsibility in ensuring an adequate income and standard of living in retirement. The relative importance of the three (state pensions, employment linked savings, private savings) or four (plus the family home) pillars of a chosen retirement incomes policy captures the various expected contributors to retirement income provision in a country. More recently, greater emphasis has been placed on the second and third pillars with individuals compelled, and incentivised respectively, to forego some of their current earnings to save for the future event of their retirement. In Australia it is suggested that only 53% of couples and some 22% of individuals are on track to achieve a sufficient level of retirement income (Burnett, Davis, Murawski, Wilkins, & Wilkinson, 2014). Furthermore, inertia is widely reported in retirement savings across the world (Choi, Laibson, Madrian, & Metrick, 2002; Madrian & Shea, 2001; Mitchell, Mottola, Utkus, & Tamaguchi, 2006). Several potential explanations have been offered for this lack of appropriate retirement planning. Thaler and Shefrin (1981) describe an internal struggle between planning and doing and suggest that self-control plays a role in the theory of individual inter-temporal choice. The conflict between the planner s long-term concern for maximising utility over the lifetime is pitted against the doer s short-term focus. The government s Superannuation Guarantee (SG) mitigates this somewhat given that at least 9.5% of total salary is deposited into an individual s account for the long term outcome. Aging and changed labour supply circumstances are two key factors suggested as a basis for a need to adjust portfolio asset allocations given their relationship with the capacity to tolerate risk (Bodie, 2003; Bodie, Merton, & Samuelson, 1992; Merton, 1969; Samuelson, 1991, 1994). Aging also correlates positively with the size of savings which is expected to increase the likelihood of attention to appropriate investment strategy given the size-of-bet effect (G. Clark, Caelewy-Smith, & Marshall, 2009). Agnew et al. (2003) report that age is associated with an 8

9 increased likelihood of individuals reducing their equity exposure (and thus risk) in their 401(k) plans and rebalance more frequently. Mitchell, Mottola, Utkus, and Tamaguchi (2006) find that among those who do trade, it is likely to be older, wealthier males though they highlight that for the large majority of 401(k) plan members age does not induce portfolio changes. Agnew (2006) similarly shows that high-salaried individuals are more likely to trade than those earning lower salaries. Gender has been previously investigated in relation to the likelihood of individuals making changes in their retirement savings plans. It has been found that in later years women are more likely to accelerate their savings activity compared to that of men (Whitaker, Bokemeiner, & Loveridge, 2013), however it is also commonly accepted that women tend to be more risk averse in terms of their superannuation investment choices compared to men (Bajtelsmit & Bernasek, 1996; Bernasek & Shwiff, 2001; Dwyer, Gilkeson, & List, 2002; Gerrans & Clark-Murphy, 2004; McNaughton & Watson, 2007). This gender difference, however, also appears to be mitigated by factors such as income, knowledge and confidence (R. L. Clark, d'ambrosio, McDermed, & Sawant, 2004; Dwyer et al., 2002; Lusaardi & Mitchell, 2008; McNaughton & Watson, 2007). 9

10 3 Data and Method Data is sourced from Mercer (Australia) and the Corporate Division of the Mercer Super Trust. The full database contains 187 employer sub-plans of a wide range of private sector employers providing defined contribution plans. 7 We only consider the part of a member s membership when they remain with their employer s sub-plan. When a member leaves an employer they are transferred to the Personal Division of the MST. Analysis also excludes changes that are administrative, or sub-plan, initiated so that all remaining changes are member initiated. We exclude sub-plans where membership is reduced to fewer than five members. We also limit the sample to individuals aged at least 18 years, with at least one year membership length, and with at least one contributions record over their observed membership period. The one-year membership criteria is imposed to reduce excessive noise (Agnew et al., 2003) in the data. We distinguish observed membership and sub-plan membership when imposing the membership length criteria. A number of employers enter the Mercer Super Trust from existing plans with existing members but the prior history of these members prior to joining is not in the administrative data available. When measuring membership length we use sub-plan membership and also recognise those who were existing members. A final criteria applied is the requirement of a minimum balance of $1,000 accumulated at the end of June 2012 or exit date to ensure that non-trivial accounts are considered. Application of these criteria results in a total sample of 258,113 members. An overview of individual members is presented in panel A of Table 1. The average member in the sample is 39 years old, has $62,829 in accumulated contributions (in 2011/2012 dollars) and has employer contributions of some $7,938 per year. Our sample has more men (61%) than women and some 28% of members make additional contributions with an average membership 7 A number of sub-plans originally offered defined benefit (DB) benefits to employees though no sub-plans now offer DB benefits to new members. Those retaining a DB membership are not included in the analysis. 10

11 of 4.43 years. A comparison of a subset of characteristics aggregated at sub-plan levels is presented in panel B. 3.1 Member choice behaviour groups As previously identified, two types of investment strategy choices are available to members. 8 The first is a change to the investment strategy for future contributions (covering mandatory employer contributions, pre-tax and post-tax employee voluntary contributions) labelled a CIC. The second is a change to the investment strategy applied to the member s accumulated balance labelled as a BIC. The two changes open up four possible choice combinations. The first are those who make no investment strategy changes at all over their history (no CIC or BIC). The second group only makes a change to their future contributions (CIC only) and the third makes only a change to their accumulated balance (BIC only). The final group contains those who are observed to make a change to the investment strategy for both their contributions and accumulated balance (CIC and BIC). An important caveat to the analysis is that we do not examine the intensity of investment activity. That is, we do not consider how many times a member makes a CIC or a BIC, only the fact that they have over their membership. Intensity is limited to whether a member makes both a CIC and a BIC. 3.2 Member characteristics A member s age (Age) is included as at their last observed date in the data, being either the exit date or 30th June 2012 which is the last review date in the data. Four age groups are constructed: 20 s and younger; 30 s, 40 s, and 50 s and older. A member s account balance 8 Consistent with Mitchell, Mottola, Utkus, and Yamaguchi (2006) we find a considerable amount of activity to be administrative in origin due to the closure of an investment options or account adjustments due to administrative reconciliations. All administrative driven changes are excluded from the analysis. 11

12 (Balance) is included as at the last observed review date (June 2012) or at exit date for those that have left the sub-plan. Values are adjusted to an equivalent June 2012 value using the consumer price index. Employer contributions (Contributions) are aggregated for each financial year of the member s history, adjusted to 2012 values using the wage price index and the average of these values employed. 9 Both balance and contributions are expected to be positively related to the likelihood of making both types of investment changes. However, for the CIC only and BIC only choices a differential impact is expected with contributions and balance, respectively, expected to be more important. The number of years the member is observed in the data is examined (Membership) as an exposure measure for the opportunity to make a decision. The final member characteristic included is gender. 9 Financial year is 1 July to 30 June. The criteria applied requiring a minimum of one year observed membership for inclusion can mean contributions records span two financial years, each for less than a full year. In this case whichever financial year has the most number of membership days for the member is used. This will be at least six months and the contributions received in this part financial year are nominally adjusted to an equivalent financial year value. 12

13 4 Analysis & Results 4.1 Any CIC, or any BIC Activity Overall, the unconditional incidence of a member having made either a CIC or BIC in the sample is 18.9 percent. A first breakdown of investment choice activity examines whether a member made a CIC, and separately whether they made a BIC at any point over their observed membership. That is, not considering whether they made both or either by itself. These results are presented in the first two columns of Table 2 (presented at the end of the paper) and presented graphically in Figure 1. Figure 1 indicates that members are more likely to make CICs (14.50 percent) compared with a BIC (13.2 percent). A relatively large gender gap is evident in both choices with percent (10.87 percent) of males (females) having made a CIC and percent (11.54 percent) a BIC. The proportion making a CIC or a BIC increases monotonically with age. The proportion of those in their 20s who make a CIC (BIC) is 5.09 percent (8.57 percent) whereas for members in their 50s and above the proportions are percent and percent respectively. A similar pattern is evident by membership length given it s expected, but not consistently correlation with age. The disproportionate amount of activity belongs to those in the top third of those ranked by the level of balance or contributions. 13

14 Figure 1 Overall Any CIC, any BIC Activity 14

15 4.2 CIC only or BIC only activity Members are not restricted to one type of investment activity and the next section examines this. Three mutually exclusive choice groups can be identified: those who make a CIC only, those who make a BIC only; and those who make both a CIC and a BIC. This breakdown is presented in columns three to five of Table 2 and presented graphically in Figure 2. Though there are three mutually exclusive choice groups there remain three degrees of freedom as a member can also do nothing, a fourth no-choice group. The largest proportion of members makes both a CIC and BIC (8.8 percent) compared to 5.7 for a CIC only and 4.3 percent for a BIC only. The previously noted gender difference in CICs and BICS is reflected most in those who make a CIC and BIC together (10.4 percent versus 6.3 percent). While a larger proportion of males make a CIC only, a larger proportion of females make a BIC only. The increasing age and activity trend is reflected in the CIC and BIC grouping as well as those making a BIC only. The CIC only age profile is humped with the larger proportion being for those in their 30s and the lowest for those in 50s. When broken down by membership length, the largest proportion of those who make a CIC and a BIC together are those with a minimum five year membership (14.9 percent) nearly double that of those with a four to five year membership (8.1 percent). For CIC only, the proportion declines with membership length and for a BIC only a humped shape relationship is suggested, peaking for those with a four year membership. A member s balance is a clear discriminator for those who make a CIC and a BIC with 18.9 percent of those in the top third of balances having made both choices compared with only 1.5 percent of those in the bottom one-third having done so. The same pattern, though less pronounced, is revealed when comparing activity by contributions. When examining those who make a CIC only the proportion increases with contributions but is largely the same across levels 15

16 of balance. For BIC only the pattern is reversed. The proportion increases with balance level but is largely the same across contributions levels. 16

17 Figure 2 Overall CIC only, BIC only, CIC & BIC Activity 17

18 4.3 CIC and BIC Activity over Time A further analysis can be made of choice activity over time. Here the focus is restricted to those who make a CIC and separately those who make a BIC in the financial year their membership spanned. That is, we don t consider the time pattern in the mutually exclusive choice groups (CIC only, BIC only, CIC and BIC). Table 3 and Table 4 present the statistics for CICs and BICs respectively. Whilst the difference in the rate of CICs between men and women remains fairly stable over the sample period, it is evident in Figure 3 that in 2006/07 there was no difference in the proportions making BICs. This appears to be due to a disproportionate rise in female members changing investment strategy. The analysis here does not extend to the type of change (ie. asset allocation). There are at least two possible explanations. The first is linked to the change in government policy to allow a one-off non-concessional contribution of up to $1 million following the reduction in older members contributions limits. A second is linked to reactions to the onset of the global financial crisis. A preliminary analysis suggests the former is the more likely explanation. Average member contributions doubled in 2006/07 compared with 2005/06, though the increase was bigger for males in fact. However, the increased balance appears to explain the increase in females making a change to their investment strategy. 18

19 Figure 3 CIC, BIC Activity by Gender and Financial Year Analysis of BIC activity by age over the sample period, presented in Figure 4, links to the previous suggested explanation for the increase in proportion of female members in 2006/07. The increase in BICs is much larger for the 50s and older age group. More generally the previously identified increase in activity by age widens over time for the 50s and over age group. The gap widens from 2006/07 and is evident for CICs as well. 19

20 Figure 4 CIC, BIC Activity by Age and Financial Year A closer look at the pattern in BICs highlights a similar time profile across age groups. The 2008/09 increase in the proportion of members changing their BIC can be linked with the global financial crisis. The increase is large relatively, in the order of 40 to 50 percent, but in absolute terms the proportion remained relatively small. For example BICs increased from 5.0 percent in 2007/08 to 8.5 percent in 2008/09 for the 50s and older group, and from 3.1 percent to 4.7 percent for the 40s age group. A relative peak in CICs is only found for the 40s and 50s age groups. In combination, the global financial crisis appears the catalyst for investment change for those with closer proximity to retirement, and possibly, larger balance sizes. However, considering the absolute proportion change levels, it is stability in overall behaviour rather than change which is more characteristic of member behaviour. 20

21 Figure 5 indicates that CIC activity was largely constant until 2008/09 where a relative peak for all but the two shortest membership lengths is observed. Post global financial crisis, CIC activity declines before increasing in 2011/12. The two peaks in BIC activity noted previously are again observed across all membership lengths. The 2008/09 peak is markedly higher (8.53 percent) for the 4 years membership group, nearly three times the level of the other membership length groups including those with the longest membership length. Figure 5 CIC, BIC Activity by Membership Length and Financial Year Finally, we consider investment activity by member balance and contributions (employer contributions). A positive correlation between balance and contributions is expected, though results are not as uniform as one may expect. The correlation is stronger among those in the top and bottom third of balance size, less so among those in the middle third of balances. Knowing 21

22 that a member was in the middle of the balance distribution is less informative as to their contributions. Knowing that a member is in the top third of the contributions level is more helpful in predicting their balance. In sum, the two variables are not perfect substitutes for each other. Figure 6 presents the CIC and BIC breakdown by contributions levels. The previously noted increasing activity level with increasing contributions level is evident. The overall trend in CIC activity over the sample period by contributions level (low, middle and top thirds) is consistent. The previously noted peaks in BIC activity in 2006/07 is observed across contributions levels. The increase in BICs in 2008/09 is largely restricted to the highest contributions level. Figure 6 CIC, BIC Activity by Contributions and Financial Year 22

23 Finally, Figure 7 provides a breakdown of CIC and BIC activity over time according to balance. There is less distinctive CIC activity by balance level in the first three years. From 2006/07 those with the highest balance level have a higher CIC activity level. Turning to BIC activity, the increasing activity with balance level is more consistent by balance level with the previously identified peaks in 2006/07 and 2008/09 evident. Figure 7 CIC, BIC Activity by Balance and Financial Year 23

24 5 Summary This paper has provided an overview of member initiated investment changes to their superannuation accounts using a large sample of members drawn from a diverse set of employers. We distinguish between changes to how future contributions (CICs) and the accumulated balance (BICs) are invested. Our analysis examined key member characteristics over a 10 year period in order to provide an insight into the characteristics associated with CIC and BIC activity as well as their consistency over time. Overall, we find that approximately one fifth of investors make some sort of change (a CIC or a BIC) over the period examined. Whilst consistent with prior literature in reporting low levels of activity, we note the caution of Bateman et al. (2014) and do not claim that this lack of revealed behaviour by the majority is necessarily an indication of a lack of engagement. The analysis here is significant given that it tracks members over a substantial period of time (up to 10 years). We document large gender differences across both BICs and CICs over the sample period. Specifically, men consistently make more CICs and BICs than women. Our analysis of membership suggests that the incidence of changes to investment strategy increase with time in the sub-plan. Finally, the analysis of balance and contributions indicates that those with higher balances and larger contributions are more likely to make changes. However, this appears to be disproportionately among those with much larger contributions and balances. Finally, the policy changes announced in May 2006 appear to be a significant event associated with an increase in investment activity across a relatively large proportion of members. This study is the first of three that will focus on the investment activity of superannuation fund members. The focus has been deliberately descriptive presenting a predominantly bivariate analysis of investment activity. It does, however, provide a starting point for a more in depth analysis of the potential roles of moderating and mediating factors in member investment activity and the consequences of this activity on member outcomes. 24

25 6 References Agnew, J. (2006). Do behavioural biases vary across individuals? evidence from individual level 401(k) data. Journal of Financial and Quantitative Analysis, 41(4), Agnew, J., Balduzzi, P., & Sunden, A. (2003). Portfolio choice and trading in large 401(k) plan. The American Economic Review, 93(1), Ando, A., & Modigliani, F. (1963). The" life cycle" hypothesis of saving: Aggregate implications and tests. The American Economic Review, Bajtelsmit, V. L., & Bernasek, A. (1996). Why do women invest differently than men? Financial Counseling and Planning, 7, Bateman, H., Deetlefs, J., Dobrescu, L. I., Newell, B. R., Ortmann, A., & Thorp, S. (2014). Just interested or getting involved? An analysis of superannuation attitudes and actions. Economic Record, 90(289), Bernasek, A., & Shwiff, S. (2001). Gender, risk, and retirement. Journal of Economic Issues, XXXV(2), Bodie, Z. (2003). Thoughts on the future: life-cycle investing in theory and practice. Financial Analysts Journal, 59(1), Bodie, Z., Merton, R., & Samuelson, W. F. (1992). Labor supply flexibility and portfolio choice in a life cycle model. Journal of Economic Dynamics and Control, 16(3-4), Burnett, J., Davis, K. T., Murawski, C., Wilkins, R., & Wilkinson, N. (2014). Measuring Adequacy of Retirement Savings Melbourne Institute Working Paper (Vol. 5/14). Choi, J., Laibson, D., Madrian, B., & Metrick, A. (2002). Defined contribution pensions: Plan rules, participant decisions, and the path of least resistance. In J. Porteba (Ed.), Tax Policy and the Economy (Vol. 16, pp ). Clark, G., Caelewy-Smith, E., & Marshall, J. C. (2009). Solutions to the Asset Allocation Problem by Informed Respondents: The Significance of the Size-of-Bet and the 1/NHeuristic. Risk Management and Insurance Review, 12(2), Clark, R. L., d'ambrosio, M. B., McDermed, A. A., & Sawant, K. (2004). Sex differences, financial education and retirement goals. In O. S. M. S. P. Utkus (Ed.), Pension design and structure: Lessons from behavioural finance (pp ). Oxford: Oxford University Press. Coates, N. (2004). Superannuation policy: commentary on an interview with Paul Keating, former Prime Minister. Journal of Australian Political Economy, 53(June), Commonwealth of Australia. (2010). Super System Review Final Report: Part One Overview and Recommendations. Dwyer, P. D., Gilkeson, J. H., & List, J. A. (2002). Gender differences in revealed risk taking: evidence from mutual fund investors. Economic Letters, 76, Feng, J., & Gerrans, P. (2014). Patterns of voluntary worker retirement savings: A longitudinal analysis. Gerrans, P. (2012). Retirement savings investment choices in response to the global financial crisis: Australian evidence. Australian Journal of Management, 37(3), doi: / Gerrans, P., & Clark-Murphy, M. (2004). Gender differences in retirement savings decisions. Journal of Pension Economics and Finance, 3(2), doi: /S Lusaardi, A., & Mitchell, O., S. (2008). Planning and financial literacy: how do women fare? American Economic Review, 98(2), Madrian, B. C., & Shea, D. F. (2001). The power of suggestion: inertia in 401(k) participation and savings behavior. The Quarterly Journal of Economics, CXVI(4), McNaughton, M., & Watson, J. (2007). Superannuation investment choice - gender difference. The Journal of Superannuation Management, 1(1), Merton, R. C. (1969). Lifetime portfolio selection under uncertainty: The continuous-time case. Review of Economics and Statistics, 51(3),

26 Mitchell, O., Mottola, G. R., Utkus, S. P., & Tamaguchi, T. (2006). The inattentive participant: portfolio trading behavior in 401(k) plans. Working Paper, University of Michigan Retirement Research Centre(WP ), Mitchell, O., Mottola, G. R., Utkus, S. P., & Yamaguchi, T. (2006). The Inattentive Participant: Portfolio Trading Behavior in 401(k) Plans. University of Michigan: Michigan Retirement Research Center Working Paper No Modigliani, F., & Brumberg, R. (1954). Utility analysis and the consumption function: an interpretation of cross-section data. In A. Abel (Ed.), The collected works of Franco Modigliani (Vol. 6, pp. 3-46). Cambridge, MA: MIT Press. Modigliani, F., & Brumberg, R. (1980). Utility analysis and aggregate consumption functions: an attempt at integration. In A. Abel (Ed.), The collected works of Franco Modigliani (Vol. 2, pp ). Cambridge, MA: MIT Press. Samuelson, P. A. (1991). Long-run risk tolerance when equity returns are mean regressing pseudoparadoxes and vindication of businessman s risk. In W. Brainard, W. Nordhaus, & H. Watts (Eds.), Money, Macroeconomics and Economic Policy. Cambridge, MA.: MIT Press. Samuelson, P. A. (1994). The long-term case for equities and how it can be oversold. Journal of Portfolio Management, 21(1), Thaler, R. H., & Shefrin, H. M. (1981). An economic theory of self-control. Journal of Political Economy, 89(2), Whitaker, E. A., Bokemeiner, J. L., & Loveridge, S. (2013). Interactional associations of gender on savings behaviour: showing gender's continued influence on economic action. Journal of Family and Economic Issues, 34, doi: /s

27 Table 1 Data Summary Mean Median Standard Deviation Panel A: Individual Members (n=258,113) Age (years) Male (percentage) Male Balance (2011/12 dollars) $ 62, $ 25, $ 115, Contributions (2011/12 dollars) $ 7, $ 6, $ 6, Additional Contributions (percentage) Existing members (percentage) Membership (years) Panel B: Sub-Plan Averages (177 sub-plans) Age(years) Male Balance (2011/12 dollars) $ 62, $ 60, $ 31, Contributions (2011/12 dollars) $ 7, $ 7, $ 2, Total Members Difference BIC-CIC (percentage) Additional Contributions (percentage)

28 Table 2 Choice Group Activity Overall CIC BIC CIC only BIC only BIC & CIC only Overall 13.15% 14.51% 5.71% 4.34% 8.81% 258,113 Gender Male 14.60% 16.40% 6.00% 4.20% 10.40% 100,288 Female 10.87% 11.54% 5.24% 4.57% 6.30% 157,825 Age 20's 5.09% 8.57% 5.58% 2.11% 2.99% 60,635 30's 10.90% 13.40% 6.40% 3.81% 7.09% 83,876 40's 14.69% 15.70% 5.86% 4.80% 9.88% 61,624 50's 24.36% 21.65% 4.55% 7.26% 17.10% 51,978 Membership 1 to < 2 years 5.65% 10.00% 6.48% 2.13% 3.53% 62,693 2 to < 3 years 8.29% 11.64% 6.19% 2.83% 5.45% 44,311 3 to < 4 years 11.00% 13.87% 5.82% 2.96% 8.05% 32,691 4 to < 5 years 16.65% 13.68% 5.55% 8.53% 8.13% 32,137 5 years 20.61% 19.83% 4.92% 5.70% 14.91% 86,281 Balance Bottom third 3.17% 6.64% 5.19% 1.72% 1.45% 86,038 Middle Third 10.60% 12.52% 6.41% 4.49% 6.11% 86,038 Top Third 25.68% 24.38% 5.52% 6.82% 18.86% 86,037 Contributions Bottom third 9.10% 8.85% 4.44% 4.69% 4.41% 86,038 Middle Third 11.98% 12.99% 5.81% 4.17% 7.81% 86,038 Top Third 18.38% 21.70% 7.50% 4.17% 14.21% 86,037 n 28

29 Table 3 Choice Activity Over Time: CIC Activity 2003/ / / / / / / / /12 Overall 4.62% 3.92% 3.25% 3.64% 3.45% 4.02% 3.13% 2.55% 3.84% Gender Male 5.07% 4.46% 3.80% 4.43% 4.19% 4.89% 4.01% 3.35% 4.87% Female 3.84% 2.99% 2.31% 2.37% 2.26% 2.61% 1.70% 1.26% 2.17% Age 20's 3.96% 3.74% 2.82% 2.93% 2.78% 2.28% 1.73% 1.57% 1.59% 30's 4.64% 3.94% 3.13% 3.64% 3.19% 3.12% 2.58% 2.07% 2.58% 40's 5.05% 3.89% 3.58% 3.74% 3.47% 4.10% 3.29% 2.49% 3.73% 50's 5.08% 4.25% 3.77% 4.66% 5.00% 8.14% 5.65% 4.55% 8.31% Membership 1 to < 2 years 5.56% 5.43% 3.79% 4.31% 3.94% 3.57% 2.82% 3.44% 4.24% 2 to < 3 years 5.13% 4.55% 3.50% 3.47% 3.10% 3.18% 3.44% 3.25% 2.13% 3 to < 4 years 4.89% 4.27% 3.26% 3.28% 3.26% 4.63% 3.35% 1.65% 2.62% 4 to < 5 years 4.90% 4.07% 3.27% 3.77% 4.34% 5.10% 2.45% 1.86% 3.11% 5 years 4.16% 3.34% 3.08% 3.60% 3.31% 3.85% 3.27% 2.65% 4.61% Balance Bottom third 4.73% 4.73% 3.13% 3.05% 3.10% 2.60% 1.76% 1.59% 1.57% Middle Third 3.62% 2.65% 2.39% 2.59% 2.15% 2.56% 2.02% 1.71% 2.46% Top Third 5.41% 4.18% 4.12% 5.17% 4.94% 6.86% 5.43% 4.23% 7.42% Contributions Bottom third 3.40% 2.44% 1.87% 1.59% 1.44% 1.71% 1.11% 0.88% 1.21% Middle Third 4.24% 3.52% 2.69% 2.93% 3.01% 3.35% 2.33% 1.97% 3.06% Top Third 6.12% 5.60% 5.07% 6.29% 5.74% 6.96% 5.77% 4.68% 7.18% 29

30 Table 4 Choice Activity Over Time: BIC Activity 2003/ / / / / / / / /12 Overall 6.40% 2.33% 2.65% 4.93% 2.61% 4.02% 2.52% 1.95% 3.33% Gender Male 5.11% 1.63% 1.76% 4.77% 1.34% 2.59% 1.22% 0.85% 1.73% Female 7.14% 2.75% 3.18% 5.03% 3.40% 4.90% 3.33% 2.63% 4.33% Age 20's 5.01% 1.30% 1.62% 3.15% 1.22% 1.52% 0.99% 0.75% 1.05% 30's 6.37% 2.15% 2.61% 3.67% 2.18% 3.12% 1.87% 1.43% 2.09% 40's 7.37% 2.91% 3.28% 5.03% 3.10% 4.66% 2.74% 2.02% 3.25% 50's 7.45% 3.67% 3.62% 10.17% 5.00% 8.50% 5.30% 4.15% 7.80% Membership 1 to < 2 years 6.00% 1.46% 2.16% 3.55% 1.49% 2.54% 1.00% 0.90% 1.96% 2 to < 3 years 6.17% 2.37% 2.59% 3.66% 1.62% 3.10% 1.66% 1.20% 1.94% 3 to < 4 years 6.15% 2.25% 2.42% 3.68% 2.11% 3.11% 2.04% 1.43% 2.51% 4 to < 5 years 7.28% 2.41% 2.84% 4.38% 2.03% 8.53% 2.18% 1.71% 2.90% 5 years 6.47% 2.51% 2.78% 5.62% 3.06% 3.42% 3.02% 2.45% 4.31% Balance Bottom third 4.31% 0.92% 1.05% 3.17% 0.68% 1.02% 0.59% 0.48% 0.84% Middle Third 6.33% 2.00% 2.48% 4.64% 1.84% 2.80% 1.76% 1.44% 2.19% Top Third 8.57% 4.07% 4.43% 6.98% 5.31% 8.23% 5.20% 3.92% 6.98% Contributions Bottom third 3.73% 1.05% 1.18% 4.52% 0.91% 1.74% 0.91% 0.67% 1.16% Middle Third 7.03% 2.08% 2.43% 3.69% 2.06% 3.55% 1.83% 1.44% 2.59% Top Third 8.46% 3.86% 4.35% 6.59% 4.86% 6.76% 4.81% 3.73% 6.25% 30