The retirement puzzle

Abstract

The present-day retirees may not be as well off as they expect to be during their retirement. Given the current state of the world – higher life expectancy, close to zero real interest rate and the economic turmoil caused by Covid-19 pandemic, a superannuation balance as large as $545,000 may not be enough to support a comfortable lifestyle. Besides, the means-tested age pension results in cases where spending at retirement does not increase proportionally with assets, providing incentives for a broad spectrum of people to deplete their superannuation in the early years of retirement, with significant negative consequences for government finances.

JEL Classification: G17, G18, G51, H55

Keywords

Comfortable retirement longevity means-tested age pension superannuation

1. Introduction

The Australian retirement system is considered one of the world’s best retirement systems. It ranks fourth following the Netherlands, Denmark and Israel (Mercer, 2020a). Nevertheless, there is scope to make significant improvements in the system if some of the anomalies and inconsistencies in the current system are adequately addressed (Cohen et al., 2019; Cohen and Ruthbah, 2019).

The Australian government requires superannuation (SA) funds to develop and offer a Comprehensive Income Product for Retirement (CIPR) to help retirees meet their necessary expenses with ease by 2022 (The Treasury, 2018). In response, the SA and investment funds in Australia are proposing different strategies for their clients. In September 2019, the government commissioned a review of the retirement income system recommended by the productivity commission. This review aims to assess the three pillars of the Australian retirement income system – their role in supporting retirement spending, the distributional impact and the fiscal burden imposed by the existing system.¹

Australia’s retirement income system rests on three pillars – the age pension, compulsory SA, and voluntary savings.² With its mandatory super laws and tax incentives for voluntary savings, the government aims for as many Australians as practicable to achieve partial or total financial independence in retirement without the need to rely on a full or part pension. The current retirement system, with the interactions among its different pillars and rules, is quite complex. As a result, there is a wide range of opinion regarding its adequacy, sustainability, equity and efficacy. The structure of the system and its implications are puzzling for practitioners and researchers in this industry, let alone the vast number of retirees who do not have adequate financial literacy (Cole et al., 2014; Gharghori et al., 2008; Lusardi and Mitchell, 2006; Willis, 2011). While providing a safety net for the retirees, the means-tested age pension discourages others from working after the retirement age or saving more – consequently adding burden to the system (Asher et al., 2017). While the tapering rate aims to ensure equity, it can make the effective tax rate for some working retirees as high as 82.5% (Cohen and Ruthbah, 2019). Some findings imply that retirees do not need to worry about running out their nest egg and lead a frugal life as they have more than enough savings (Daley et al., 2018). Hence, they can afford to withdraw more than the minimum required (5%) from their super account (Ravin et al., 2019). Others point out that about half of the non-retired households and a third of the retired households believe that they do not have enough savings to finance a comfortable retirement (Bray and Gray, 2016; ME Bank, 2019). Despite being highly ranked, the system with all its goals, structure, policies and caveats produces, in many cases, somewhat puzzling outcomes.

This article addresses two critical issues related to the Australian retirement system’s adequacy and equity, which are products of its intricate composition. The first one explores whether the retirees in Australia can afford to have a comfortable life given the current economic conditions. The study finds that many Australians may not be well placed, particularly given the advent of coronavirus and its economic fallout. The second one, related to the first pillar of the Australian retirement system – the means-tested age pension, finds that the marginal worth of retirement savings is not the same for everyone. Spending at retirement depends both on the level of SA balance and the age pension. This article finds that there is a range of SA balance where the reduction in age pension for the same amount of increase in assets is higher than at other levels. This finding has profound implications for both the working-age and the retired individuals’ saving and spending behaviour. When the age pension declines at a higher pace, the retiree’s spending capability will not increase by the same amount as the increase in SA balance. There is a range of SA balance where each additional dollar saved today will translate into less than a dollar of consumption in retirement. This will provide working-age individuals incentive not to save beyond some level if their SA balance falls in that range. It also means that there is a range of SA balance at which retirees will have an incentive to invest part of their savings in assets excluded from the asset tests or spend it at the beginning of their retirement. For example, they can invest in their principal residential home or buy a prepaid funeral. They can thus improve their net worth as it makes them eligible for a higher age pension. Therefore, this finding can explain, to some extent, why some retirees might have the incentive to tailor and contrive their spending patterns to maximise pension entitlements, with critical negative implications for the federal budget – particularly in the longer term as the population continues to age.

The question of adequacy generated mixed responses not only in Australia but also in other developed countries (Munnell et al., 2014a). Beshears et al. (2019) find that in the United States absolute level of equivalent income for retirees aged 70 has increased but has declined for those aged 80 below the median, while the share of households totally dependent on social security has increased in the 21st century. Research by Chen et al. (2018) suggests that nearly half of all households in the United States will fall short of the target replacement rate. A 2014 study finds that retirees of almost all income groups could achieve the target replacement rate under the then prevailing market conditions in Iceland (The Financial Supervisory Authority in Iceland, 2014). However, the future retirees with less than 40 years of contribution may find themselves in financial stress, given the current low-interest rate, low growth environment (OECD, 2020). One reason behind the mixed results is that sometimes researchers generalise the problem of adequacy while this is a concern mostly for the poor and the ‘precarious middle’ (Poterba et al., 2011; Statman, 2014) or it can vary by factors such as marital status (Hurd and Rohwedder, 2012) and homeownership (Callaghan et al., 2020).

Discussions on the Australian retirement system’s equity implications are mainly focused on the tax treatment of super balance, the super guarantee rate and age pension (ASFA, 2012; Callaghan et al., 2020; Clare, 2001). In the Retirement Income Review, Callaghan et al. (2020) find that the tax concessions on SA increase inequality among retirees while the age pension offsets some of it across income groups and gender. However, the exclusion of residential home from the age pension entitlement creates some inequalities between the homeowners and renters. The pension tapering thresholds and rates also have different implications for those at the margin of the thresholds and those receiving a part pension (Boal, 2020). This article discusses one of those implications.

This article’s findings contribute to the debate on the policy reforms proposed or required to make Australia’s retirement system better in two ways. First, it substantiates the need for comprehensive retirement income products that would provide adequate income flows during retirement. The industry body and the government may also need to update their retirement adequacy standards so that retirees are aware of the changing market environment and its implications. Second, the issue with equity further raises concern regarding the mean-testes age pension. One way to address the unintended nudges it provides is to simplify the system or make it universal. Universal age pension would reduce the complications and cost of administering the system at the expense of increasing the government’s fiscal burden. However, many stakeholders view that designing it as part of the SA system (BetaShares, 2020; Cohen and Ruthbah, 2019) or with appropriate tax structure (Mercer, 2020b) can offset the additional fiscal burden.

The rest of this article is organised as follows. Section 2 describes the methodology used in this article. The main findings are presented in section 3. Section 4 touches on the robustness of the results presented in section 3. Section 5 projects the size of the SA asset required for a comfortable retired life, and finally, section 6 makes some concluding remarks.

2. Methodology

We use Monte Carlo simulation and historical risk-return data for our analysis. This is a standard methodology applied in retirement planning literature (Alonso-Garcia and Sherris, 2019; Bengen, 1994; Ganegoda and Evans, 2015).³ An individual can expect to live up to 30 years after retirement. The goal is to spend an inflation-adjusted $43,687 a year as long as he lives.⁴ His retirement spending is funded by his SA balance, which he invests in a 60/40 (Australian) portfolio and age pension when eligible.⁵ While there is no conclusive evidence on which type of retirement product – annuities or equity portfolios, better serves the retirees (Alonso-Garcia and Sherris, 2019; Doyle et al., 2004; Horneff et al., 2010a, 2010b; Milevsky and Kyrychenko, 2008), we focus our analysis on a 60/40 portfolio. The article’s primary goal is to analyse the welfare of the retirees under the current market conditions not to point out the relative benefits of different products. The incoherent outcomes of the age pension system that we find are robust to various other asset allocation strategies. In projecting the possible spending trajectories, we consider risk sequencing, which is a genuine concern, given the development in the world economy since last year due to COVID-19. All our analyses are carried out for a single male retiree who owns a home.

Our planning horizon of 30 years from retirement, while longer than benchmarks in other studies, takes account of ever-increasing life expectancies (Cocco and Gomes, 2012).⁶ Male Australians born in 1954/1955 are expected on average to live 86.4 years, while those born today are predicted to reach 92.⁷ So 30 years is a reasonable planning horizon. The analysis in section 4 shows that increasing the planning horizon to 40 years does not change the inferences drawn in this article.

The projection of future outcomes depends on assumptions made regarding the overall economic conditions – interest rates, market performance, inflation, the timing of investment, the model parameters – longevity, portfolio allocation, drawdown strategy, and above all policy changes. The assumptions made in simulating the success rates, pension and spending trajectories in our analysis are as follows:⁸

Number of simulations: 10,000.

The retiree’s goal is to spend an inflation-adjusted $43,687 each year which he withdraws from his SA.

If he is eligible for the age pension, then the withdrawal amount is adjusted accordingly.

A single retiree, who owns a home, qualifies for a full age pension – $24,554 per annum – if his assets (excluding his family home) are worth less than $268,000. Above this level, the age pension declines by $78 for each $1000 increase in assets. The pension cuts out entirely at an asset level of $583,000.

The age pension declines by $0.5 for each $1 income earned over $172 per fortnight.

The deemed rate of return is 0.25% for assets up to $53,000 and 2.25% above this level.⁹

Retirees hold all their assets in SA and therefore are exempt from any tax consequences. Retirees can keep up to $1.6 million tax-free in SA, and we only consider assets less than or equal to $0.8 million in this article.¹⁰

The minimum drawdown rate from SA is 5%.

The steps we follow in the simulation are provided in Appendix 1.

3. Analysis

3.1. A comfortable retirement

According to Australia’s peak SA industry body, the ASFA (2012), a single male retiree would need to spend $43,687 a year to have a comfortable retirement.^11,12 According to their Super Guru website, a male retiree with a life expectancy of 86 years can live a comfortable retired life with a SA balance of $545,000.^13,14 The Australian Securities and Investments Commission’s (ASIC) moneysmart calculator adopts a similar comfortable retirement income benchmark. It suggests that a retiree, retired at the age of 65 with a $580,000 (a SA balance little less than the pension tapering asset threshold) SA balance, could spend $43,7516 until age 91. After 91, he would have to depend entirely on the age pension ($24,554 p.a.).

However, our simulation model predicts that a $580,000 portfolio with a 60/40 allocation can have a 49% chance of running out of money after 30 years into retirement and a 31% chance of having to entirely rely on age pension after 25 years (Panel (a), Figure 1). The point is, even a $580,000 super balance may not be enough to guarantee a comfortable retired life all through. Panel (a) of Figure 1 shows the probabilities of running out of money for a 60/40 portfolio at any given year of the retiree’s life for different SA balances. We find that retirees who are not eligible for age pension (with a SA balance of $600,000) at the beginning of their retirement life (as their asset is above the maximum threshold of $583,000) also have a 12% chance of falling back solely on age pension 20 years into retirement.

Figure 1.

The success rate of a 60/40 portfolio, spending and age pension over 30 years of retirement: (a) success rate of a 60/40 portfolio, (b) spending and pension at SA balance $260,000 and (c) spending and pension at SA balance $580,000.

Most retirees in Australia have a SA balance far less than the ASIC recommended $545,000. The average SA balance of a retiree aged 64–75 in 2017–2018 was $402,600, and the median was $225,200.¹⁵ As shown in Figure 1, a retiree with a $400,000 SA balance has a 40% chance of drawing down his or her assets to zero 20 years into retirement, which goes up to more than 60% as he turns 90. The picture is grimmer for the majority of the Australian retirees whose SA balance is just below the pension tapering threshold ($268,000 at time of publication). A portfolio of $260,000 is predicted to be out of money at the end of 15 years with a probability of 61%.

These observations have severe implications for individual retirees, and how they should plan their spending trajectories, as well as for the Australian government’s finances. An increase in the number of retirees becoming eligible for the age pension ahead of predictions would generate unanticipated stress on government financial resources. In Figure 1, Panels (b) and (c) show the assumed spending trajectories of retirees with SA balances of $260,000 and $580,000, respectively, and their relative reliance on their SA savings and the age pension over time.¹⁶

The analysis of the success rates, spending and age pension trajectories for retirees with different SA balance leads to the second set of results/observations presented in the next sub-section.

3.2. The distributional effect

A higher SA balance at the beginning of the retirement results in higher spending during retirement. However, we find that the changes are not necessarily proportional. A given increase in SA balance translates into different amounts of increase in the present value (PV) of spending at various levels of SA balances. We also find that the relationship between SA balance changes and changes in the PV of age pension entitlements is not linear and regressive.

The PV of retirement spending and the age pension entitlements at different asset (SA balance) levels (as predicted by our simulation) are presented in Table 1 and Figure 2.¹⁷ They show the net PVs of a retiree’s spending and age pension with a SA balance ranging from $0 to $800,000.¹⁸

Table 1.

The present values (PV) of spending in retirement and age pension entitlements over 30 years at different SA balances (in $000).

Category	PV in $000			Changes in PV from
	PV in $000			Previous category		Category 1
	SA balance	Spending	Pension	Spending	Pension	Spending	Pension
	(1)	(2)	(3)	(4)	(5)	(6)	(7)
1	0	711.64	711.64
2	50	734.34	711.64	22.70	0.00	22.70	0.00
3	100	782.05	711.64	47.71	0.00	70.41	0.00
4	150	836.08	711.64	54.03	0.00	124.43	0.00
5	200	896.65	711.55	60.57	−0.09	185.01	0.00
6	250	961.52	710.40	64.87	−1.15	249.88	−0.09
7	300	1020.80	704.09	59.28	−6.31	309.15	−1.24
8	350	1070.93	691.63	50.14	−12.47	359.29	−7.55
9	400	1111.35	672.68	40.41	−18.95	399.70	−20.02
10	450	1141.93	647.71	30.58	−24.97	430.28	−38.97
11	500	1165.46	618.30	23.54	−29.41	453.82	−63.94
12	550	1183.80	585.78	18.33	−32.52	472.15	−93.34
13	600	1198.78	551.30	14.99	−34.48	487.14	−125.86
14	650	1211.40	515.51	12.61	−35.78	499.75	−160.35
15	700	1221.45	478.86	10.05	−36.66	509.80	−196.13
16	750	1229.46	442.07	8.01	−36.78	517.81	−232.79
17	800	1235.95	405.82	6.49	−36.25	524.30	−269.57

PV: present value; SA: superannuation.

All figures represent present values in $000 – thus, in column 2, for example, the number 711.64 denotes $711,640. The present values are calculated using the 30-year (real) long bond rate of 0. 224%. This was the average yield rate of the 30-year inflation-indexed Australian government bond in the past year.

The sum of SA balance and the present value of pension does not add up to the present value of spending for any of the categories except for category 1 where the SA balance is 0. There are three reasons why this might happen.

First, for the two to add up to the present value of spending, at the end of 30 years, the retirees should not have any SA balance left. As Figure 1 shows, retirees with SA balance of $260,000 have a 12.23% chance of not running out the nest egg at the end of 30 years. This means that of the 10,000 simulations, 12.23% of the time there was some SA balance left with the retirees. Therefore, the present value of spending will be less than the sum of SA balance and present value of the age pension.

The second reason why they might not add up is that in calculating the age pension, we use the rate of return that the ATO uses (the deeming rates). These rates could be (and in reality they are) very much different from the returns that retirees earn on their investment. Since we used historical risk-return data in generating our simulated portfolio returns, in most cases, the average return that retirees earned is either higher or lower than the deeming rates used by the ATO. Therefore, the retirees get more/less pension than what they would have received if the ATO used the actual return in calculating the age pension. The larger the SA balance the retiree has, the larger will be the difference. If the portfolio’s average return is higher than the deeming rate, then the present value of spending will be higher than the sum of SA balance and the present value of the age pension.

Third, the discount rate used to calculate the present value of spending and the age pension is different from the returns SA balance earns under various asset allocations. The point Table 1 tries to make is that there is a range of SA balance where, for every additional $50,000, spending increases up to so some level and starts to decline afterwards. This implies that the marginal contribution of current savings (savings in SA balance) to future consumptions is not the same at all levels of SA balance. At some levels, a dollar sacrificed today generates more than a dollar worth of spending during retirement and less at other levels. This is true for a range of discount rates. This is shown in Table 2.

Thus, the final result, whether the present value of spending is higher or lower than the sum of the SA balance and the present value of age pension will depend on the combined effects of the above three.

Figure 2.

Present values (PV) of spending and age pension over 30 years of retirement with a 60/40 portfolio: (a) PV of spending and age pension at different SA balance (in $000), (b) marginal contribution of $50,000 SA balance to spending and age pension (in $000) and (c) spending and age pension of a retiree compared to one who has no SA balance (in $000).

Columns 1–3 of Table 1 present the retiree’s SA balance, the PVs of projected spending and age pension entitlements at those SA levels, respectively, over a 30-year time frame. Columns 4 and 5 show the change in the PVs of spending and age pension entitlement as SA balance increases by $50,000. For example, as SA balance increases from $200,000 (category 5) to $250,000 (category 6) the retiree can increase the PVs of his retirement spending by $64,867 while the PV of his age pension declines by $1151. Columns 6 and 7 present changes in the PVs of spending and age pension as SA balance increases from zero (category 1). For example, as SA balance increases from 0 to $150,000, the PV of spending increases by $124,433 but change the PV of the age pension is $0 as a retiree with no balance is eligible for the same amount of age pension as a retiree who has $150,000 SA balance.

Table 1 shows that, as SA balance increases the PV of spending increases and that of age pension decreases (Columns 2- and 3). However, the additional spending the retiree can afford for each additional $50,000 saved in SA varies at different levels of SA balance. As column 4 shows, between $0 and $250,000, the marginal contribution of additional $50,000 in SA balance to spending is increasing, and beyond that level of SA balance its marginal contribution to spending starts declining. As SA balance grows beyond $350,000, each additional $50,000 SA balance generates less than $50,000 worth of consumption during retirement. Similarly, the decline in eligible age pension for each addition $50,000 increases in SA balance is not the same (Column 5). Between SA balance $0 to 150,000, there is no change in the amount of age pension. Beyond $150,000, age pension declines and the amount of reduction increases with the level of the SA balance, but the amount of reduction is less than $50,000. This makes the system economically and socially regressive.

Table 1 also indicates that the relationships of pension and spending to SA balance are non-linear. This becomes even clearer if we look at Panels (b) and (c) in Figure 2. Panel (b) shows that retirees can improve their net worth by lowering their SA balance at some asset level. This provides incentives to those just above this asset level to invest (or consume) a part of their SA balance into something that does not count towards the asset test (e.g. on the renovation of their principal residence that increases its value) and enjoy a higher age pension.

For example, the PVs of spending a retiree with SA balance $400,000 can afford is $1,111,346, and the PV of age pension he is eligible for is $672,678. If at the beginning of his retirement, he spends $50,000 on the home renovation or prepaid funeral, that is, he starts his retirement with $350,000 in his SA account. He is then eligible for $691,625 as age pension, and the PV of this total spending (or net worth) becomes $1,120,934 (=$50,000 +$1,070,934). This is true for any SA balance beyond which the marginal contribution to spending is declining. Thus, retirees in some range of SA balances will have an incentive to spend their nest egg on unproductive or unnecessary consumption – expenses that they would not have incurred in the absence of the pension tapering mechanism; and thus qualify for full/part age pension (Andreasson et al., 2017; Bütler et al., 2017). This adds significantly to the burden of the age pension system and on public finances.

This also has implications for working-age individuals. Under the circumstances where the sacrifice of current consumption to improve the SA balance does not lead to an equivalent increase in the future spending power in retirement, some working-age individuals may save less for retirement putting additional stress on future fiscal balance. This burden will only increase over time as the population ages, and as future governments have to deal with the economic fallout from coronavirus.

4. Robustness

The analysis in the preceding section is based on a 60/40 portfolio allocation (60% equities and 40% defensive interest-bearing securities) and used historical return data and prevailing market conditions to arrive at our assumptions regarding the portfolio’s risk-return profile. The real expected return for a 60/40 portfolio is assumed to be 3.29% per annum with a volatility of 13.20%.¹⁹ These returns are calculated from annual historical returns data (DMS Global returns data). A description of the historical risk-return premiums in the equity and bond markets in five developed counties of the world is presented in Appendix 1 (Historical (inflation-adjusted) return in the equity and bond markets (1900–2019)). Historically, the equity bond premium has been 5.01% in Australia. The yield rate of 30 years inflation-linked bonds has fluctuated around −0.236% to 1.391% in the last year, with an average of 0.224%.²⁰ For our core analysis, we assumed a real return of 0.224% for 30-year bonds.

The assumptions in this article diverge from the current asset allocation strategies of most large super funds which have diversified beyond equities and bonds into other asset classes – ranging from the very liquid (alternative assets) to those with zero liquidity in short to medium run (private equity and infrastructure). The default or balanced investment options in the largest super funds have a target of CPI + 3.5% to 4.0% return with a varying level of risk (low to high) and 5–10 years investment horizon (HESTA, 2019; QSuper, 2019).²¹ Their default/core investment strategies allocate 16%–30% of the pot to assets others than equity, bonds and cash. The big funds have their own return targets and assumptions that differ from those made in this article in the years leading up to the COVID-19 crash. These other asset classes are excluded from the current analysis because they do not have a sufficiently long history of risk-return data to enable reliable conclusions to be drawn. However, it is possible that with an astute asset manager, these portfolios may achieve a higher return at a lower risk than assumed here, and the portfolio of the retiree will have a lower rate of failure (running out of money) than predicted in this study. To allow for this possibility, one of the scenarios presented in the next section involves a portfolio generating a higher return at a low volatility over the retiree’s lifetime. It also tests the strength of the findings in the preceding section by varying the assumptions on longevity. However, changing the long-run market expectations or asset allocation strategy or the assumption on longevity does not change the main inferences drawn in this article.

4.1. Sensitivity analysis

4.1.1. Discount rate

The trajectories of spending and age pension and their PVs depend on the discount rates along with the other parameters. The above analysis assumes a discount rate of 0.224%, which was the average yield rate of the 30-year inflation-linked bond in 2020. This section shows that the inferences drawn above are robust to a wide range of discount rates. The yield of 30-year inflation-indexed bond fluctuated between −0.236% and 1.391% in the past year. Here is a comparison of our analysis in section 3.2 with those that correspond to these discount rates.

Columns 1 and 2 of Table 2 present the corresponding columns 4 and 5 of Table 1, that is, the marginal contribution of $50,000 savings in SA balance to the PVs of spending and age pension when the long bond rate is −0.236%, and columns 3 and 4 show the same for the rate 1.391%. We find that there is a range of SA balance where the change in the PV of spending is bigger than the change in the SA balance and lower at other levels at both positive and negative long bond yields (or discount rates).

Table 2.

The marginal contribution of $50,000 SA balance to present values (PV) of spending in retirement and age pension entitlements at different discount rates.

Discount rate		–0.236%		1.391%		0.224%
Category	SA balance	Spending	Pension	Spending	Pension	Spending	Pension
Category	(1)	(2)	(3)	(4)	(5)	(6)	(7)
1	0
2	50	22.56	0.00	34.03	0.00	22.70	0.00
3	100	47.23	0.00	48.17	0.00	47.71	0.00
4	150	53.48	0.00	55.15	0.00	54.03	0.00
5	200	59.70	−0.05	61.83	−0.30	60.57	−0.09
6	250	64.57	−0.82	63.33	−2.43	64.87	−1.15
7	300	60.48	−5.49	53.04	−9.36	59.28	−6.31
8	350	52.68	−10.95	40.57	−16.99	50.14	−12.47
9	400	43.71	−17.11	29.44	−24.00	40.41	−18.95
10	450	34.34	−23.12	20.58	−29.73	30.58	−24.97
11	500	27.11	−27.64	14.62	−33.39	23.54	−29.41
12	550	21.55	−30.99	10.55	−35.66	18.33	−32.52
13	600	18.00	−33.23	8.10	−36.82	14.99	−34.48
14	650	15.46	−34.74	6.51	−37.13	12.61	−35.78
15	700	12.42	−36.00	4.65	−36.67	10.05	−36.66
16	750	10.24	−36.58	3.49	−35.37	8.01	−36.78
17	800	8.46	−36.53	2.67	−33.45	6.49	−36.25

SA: superannuation.

All figures represent present values in $000 – thus, in column 2, for example, the figure 22.56 denotes $22,560.

Figure 3 compares the marginal contributions of $50,000 SA balance to PVs of spending and age pension at different discount rates. It shows that at a higher discount rate (1.391%) the range of SA balances for which the retiree (or a working-age individual) has incoherent incentives expands making the implications for future fiscal balance more acute.

Figure 3.

Marginal contribution of $50,000 SA balance to present values of spending and age pension at different discount rates.

4.1.2. Longevity

The simulation results presented in the previous section depend on the retiree’s planning horizon, as the distribution of returns for 40 years will be different from that of 30 years. As a 100-year lifespan might be possible for millennials (Gratton and Scott, 2016; Vernon, 2017), this section presents the simulation results for a retiree expected to live 40 more years. The new simulation shows that increasing longevity does not change the results presented above. As Figure 4 shows, the retiree starting with an SA balance of $400,000 has a 45% chance of running out of assets at 20 years into retirement and an 80% chance of falling back entirely on age pension at the end of 30th year. The probability of failure increases with the planning horizon.

Figure 4.

The success rate of a 60/40 portfolio, spending and age pension over 40 years of retirement: (a) success rate of a 60-40 portfolio, (b) spending and age pension at SA balance $260,000 and (c) spending and age pension at SA balance $580,000.

Changes in the PV of spending and age pension with assets also are non-linear as we find above. However, the range of SA balance for which spending increases faster than SA balance is slightly different under the assumption of a 40-year retirement. The range shifts towards the right, creating a disincentive (as discussed in the preceding section) for a larger group of retirees, as shown in Figure 5.

Figure 5.

Present values (PV) of spending and age pension over 40 years of retirement with a 60/40 portfolio: (a) PV of spending and age pension at different SA balance, (b) marginal contribution of $50,000 SA balance to spending and age pension (in $000) and (c) spending and age pension of a retiree compared to one who has no SA balance (in $000).

4.1.3. Choice of portfolio/asset allocation

An aggressive allocation of all assets into equity increases the rate of return at the cost of higher volatility. As expected, higher volatility reduces the success rates and increases the probability that the retiree will fall back on the age pension, increasing the pressure on fiscal balance.

We also find that higher volatility in asset returns makes retirees with higher levels of SA balance eligible for more age-pension at early years of retirement. However, higher returns enable them to increase their spending without adding to the burden of the age pension system.²² At SA balance $600,000, the retiree with a 100% equity portfolio is eligible for $694,450 in age pension and can enjoy $1,505,713 worth of spending (Panel (d), Figure 6). In contrast, the retiree with a 60/40 portfolio is eligible for a higher age pension ($785,500) but enjoys a lower level of spending ($1,479,617) (Panel (a), Figure 5).

Figure 6.

The success rate of a 100% equity portfolio, spending and age pension over 40 years of retirement. The horizontal axes in Panels (a), (b) and (c) measure years into retirement and those in Panels (d), (e) and (f) measure superannuation balance. In Panels (d), (e) and (f), superannuation balance, spending and pension are measured in $000: (a) success rate of a 100% equity portfolio, (b) spending and age pension at SA balance $260,000, (c) spending and age pension at $580,000, (d) PV of spending and age pension at different SA balance, (e) marginal contribution of $50,000 SA balance to spending and age pension and (f) spending and age pension of a retiree compared to one who has no SA balance.

4.1.4. Market expectations

Figure 7 shows the same results for a portfolio allocation under very favourable market conditions – earning a real return of 5.1% with an annual volatility of 10.1%. Higher returns and lower volatility reduce the failure rates of the portfolio, but they are still significantly high. Compared to a 60/40 portfolio, this allocation reduces the probability that a $400,000 portfolio will run out of money at the 25th year of retirement from 66% to 25% (Figure 7, Panel (a) vs Figure 4, Panel (a)). Similarly, the probability of failure of $580,000 portfolio declines from 35% to 4%.

Figure 7.

The success rate of a (5.1%, 10.1%) portfolio, spending and age pension over 40 years of retirement. The horizontal axes in Panels (a), (b) and (c) measure years into retirement and those in Panels (d), (e) and (f) measure superannuation balance. In Panels (d), (e) and (f), superannuation balance, spending and pension are measured in $000: (a) success rate of portfolios, (b) spending and age pension at SA balance $260,000, (c) spending and age pension at $580,000, (d) PV of spending and age pension at different SA balance, (e) marginal contribution of $50,000 SA balance to spending and age pension and (f) spending and age pension of a retiree compared to one who has no SA balance.

As Panels (d) and (f) of Figure 7 show, the patterns in spending and age pension eligibility and the non-linear and regressive relationships of SA balance with spending and age pension still prevail.

Next, we consider another set of capital market expectations that are more optimistic than the one just discussed. We show that even if a prudent asset manager succeeds in achieving a 3.5% return at 6.5% volatility, retirees starting with an average SA balance will still run a significant risk of running out of money. Those who are currently not eligible for age pension will still receive a considerable amount of government support and retirees at certain levels of SA balance will have the incentive to spend more than otherwise planned to remain eligible for part or full pension. The probability that a $400,000 portfolio will run out of money after the 30th year of retirement is 63% even under this optimistic market scenario (Panel (a), Figure 8).

Figure 8.

For a portfolio with a 3.5% return and 6.5% volatility over 40 years of retirement. The horizontal axes in Panels (a), (b) and (c) measure years into retirement and those in Panels (d), (e) and (f) measure superannuation balance. In Panels (d), (e) and (f), superannuation balance, spending and pension are measured in $000: (a) success rates of portfolios at different SA balances, (b) spending and age pension at SA balance $260,000, (c) spending and age pension at $580,000, (d) PV of spending and age pension at different SA balance, (e) marginal contribution of $50,000 SA balance to spending and age pension and (f) spending and age pension of a retiree compared to one who has no SA balance.

4.2. An allocation targeted for retirees

The analysis so far assumes that the retiree has the same asset allocation all through his retired life. Traditional wisdom suggests retirees move towards safer assets as he progresses through retirement. In this section, we present an analysis where the asset allocation varies depending on retirees’ age. We use the asset allocation suggested by Vanguard Target Retirement Fund. This fund increases the share of defensive assets as the retiree ages, starting with a 50.61% bond, 48.37% equity and 1.02% cash portfolio at the age of 65. It increases its share of the bond to 65.02% when he is 67 and 69.33% when 71.

Figure 9 presents the success rates, spending trajectories and the way spending and age pension moves with SA balance for this portfolio. We find that the success rates of this target retirement fund of different sizes decline compared to the 60/40 portfolio described in section 3 as it earns a lower average return with lower volatility. However, as panel (e) of Figure 9 shows, the relationships between the marginal contribution of SA balance to PV of spending and that of the PV of age pension are still non-linear. Thus, the implications for individual’s savings behaviour and fiscal balance remain the same.

Figure 9.

The target retirement fund. The horizontal axes in Panels (a), (b) and (c) measure years into retirement and those in Panels (d), (e) and (f) measure superannuation balance. In Panels (d), (e) and (f), superannuation balance, spending and pension are measured in $000: (a) success rate of a 60/40 portfolio, (b) spending and pension at SA balance $260,000, (c) spending and pension at SA balance $580,000, (d) PV of spending and age pension at different SA balance (in $000), (e) marginal contribution of $50,000 SA balance to spending and age pension and (f) spending and age pension of a retiree compared to one who has no SA balance.

The Vanguard Target Retirement Fund (VTRF) or the different retirement funds or annuities provided by some of the largest institutional investors vary the allocation depending on the age of the retiree, not the level of savings. Accordingly, this analysis does not explore this avenue – whether retirees’ investment decisions should vary according to their SA balance or risk aptitude at different age in detail. However, to see how the outcomes might vary if retirees move to a safer allocation as they grow old and the size of their nest egg shrinks, we consider another allocation where, after 20 years into retirement, those with less than $100,000 is SA balance put all their money into cash and earn zero real interest. We find that this move does not change the average pension entitlement of retirees and would remain unchanged for a threshold less than the lower asset test bound. The PV of spending changes only marginally for retirees with higher asset balance. For example, those with an SA balance of $800,000, the PV of spending declined by only $6080 with a marginal decline in the portfolios’ success rate. After 20 years into retirement, the success rate declines compared to the VTRF as retirees keep drawing down their asset while earning no returns.²³ It may be noted that Estrada (2020) also finds that when retirees need to adjust their retirement strategy, changing the portfolio allocation helps little.

4.3. The impact of Covid-19

The risk-return profiles used in this article are based on historical data. Even though the world was struck by another pandemic 100 years ago, extrapolating that experience in today’s world would not be possible as the extent of financial and economic integration was much less in 1919, and it was also the aftermath of the World War I. Nevertheless, it is too early to predict or measure the impact of the Covid-19 pandemic on the economy and the financial market. Given these constraints, one possible way to account for the effects of the current pandemic on the financial market is to sequence risk – that is, to assume that the investment return on SA balance for the first year of an individual retiring today would be the lowest of all the coming years. This means the retirees will see a sharp decline in their portfolio’s value at the end of the first year of investment. We maintain this assumption in all the analysis in this article. The implication for this risk sequencing is that retirees would run out money sooner than later. This is shown in Figure 10, which compares the retiree’s spending and age pension in a world with and without risk sequencing (or Covid-19). Had the world not hit by this pandemic, the retiree could have spent a little more than what he would now. However, the impact on the age pension is much larger. A retiree with $800,000 SA balance and 60/40 asset allocation would get $294,902 in age pension in a no Covid-19 world while after Covid-19 he would be eligible for a higher amount ($405,819) as he would become eligible for the age pension much earlier (Panel (a)). Another way to deal with the uncertainty created by the Covid-19 pandemic is to assume that the market would be very volatile in the first few years of retirement before going back to its historical risk-return trend.²⁴ This is also shown in Panel (a) of Figure 10. Even with the second scenario with highly volatile early years, we find that retirees’ pension entitlement with an SA balance of $800,000, who are not currently eligible for it, increases (by $45,890) compared to a no Covid-19 scenario. This is true irrespective of the asset allocation strategy. As Panel (c) shows, in case of the target retirement fund, a retiree with $800,000 SA balance would be eligible for $140,405 more in age pension in a Covid-19 hit world compared to one without Covid-19.

Figure 10.

Comparing with a no Covid-19 world: (a) PV of spending and age pension at different SA balance (in $000) for a 60/40 portfolio, (b) changes in spending, age pension and SA balance from the previous category (in $000) for a 60/40 portfolio, (c) PV of spending and age pension at different SA balance (in $000) for a target retirement fund and (d) changes in spending, age pension and SA balance from the previous category (in $000) for a target retirement fund.

5. How much is enough when it comes to super?

It is clear from our analysis that Australian retirees who start with an average-sized SA balance are at significant risk of being unable to enjoy a financially comfortable retirement. This raises the question: how much in savings would someone retiring now need to live a comfortable life, particularly given increased longevity, uncertain future investment returns and the initial loss of assets resulting from the financial market fallout caused by Covid-19?

The size of savings required for a comfortable retired life depends on several factors other than the expected span of retired life and the real market rate of returns – the desired spending rate, the availability of social security (age pension in case of Australia), an individual’s degree of risk aversion and so on. It may also vary by country or region. Munnell et al. (2014b) find that a retiree in the United States requires USD538,000 to finance 48.57% of their retirement spending, where the social security funds the rest of the expenditure. According to ASFA’s consumer website Super Guru, a single retiree needs $545,000 SA balance to have a comfortable retired life (spend an inflation-adjusted $43,787 a year).²⁵ However, when we consider the current turmoil in the financial market and the uncertainty involved in investing in assets, we find that $545,000 might not be enough to live comfortably for 34 years in retirement. These findings will be subject to further refinement, or perhaps substantial revision, in future after markets settle into new patterns following the economic mayhem caused by the Covid-19 pandemic. For now, however, they demonstrate that projections by the super industry and the Australian government about what constitutes an adequate retirement nest egg may have been significantly underestimated.

Figure 11 shows the probabilities that portfolios of size $650,000 and $800,000 will last at different time horizons under different risk-return assumptions assuming that the first year of retirement will be the worst in terms of portfolio returns.

Figure 11.

Success rates of portfolios under different risk-return scenarios. The optimistic portfolio has an expected real return of 5.1% with an annual volatility of 10.1%. The Optimistic 2 portfolio has 3.5% real return and 6.5% annual volatility: (a) success rate of a $650,000 portfolio and (b) success rate of a $800,000 portfolio.

It shows that even a portfolio of size $650,000 has a very high probability (45%) of running out of money at the 30th year of retirement if the market follows the historical risk-return pattern in the coming years. At these risk-return trajectories, a retiree would require a SA balance of at least $800,000 to last up to 30 years with a probability of 70%. However, the requirements are less stringent under more favourable market expectations. Nevertheless, the historical performance of the equity and bond markets in the developed countries suggests that the prospect of a favourable market condition may not be huge. Even if the volatility of the market tones down in the next 40 years as indicated by the more recent data, the returns are also likely to be revised downwards. The very optimistic picture of the future market (3.5% return with 6.5% volatility) also portrays that a portfolio of $650,000 has a 25% risk of running out at the 40th year.

6. Conclusion

The retirement system in Australia is one of the best in the world and at the same time, not very simple. All the rules of age pension, income and asset tests, the minimum drawdown of super and income deeming provide substantial support to the retirees who are not as well off as others and at the same time makes the relationship between retirement savings and spending non-linear. Spending at retirement does not increase proportionally with assets. For some range of SA balances, the net PV of spending could increase more than the increase in assets as the net PV of the age pension does not decline at par with increases in assets. In other asset levels, the opposite happens. This provides incentives to some retirees to remain eligible for part or full age pension by drawing down their super account at a rate that would otherwise be deemed irrational. The distributional impact of the age pension is unequal across different asset levels and is regressive to some extent.

In addition, accepted wisdom about how much super individual Australians need to enjoy a comfortable retirement may not stack up; modelling for this study shows that people with super balances of between $545,000 and $580,000 – the current industry and government benchmarks for comfortable, largely self-funded retirement – may in fact find themselves well short, and relying heavily on the age pension in later years.

These findings have potentially drastic implications, not just for individuals, but for future Australian governments and their finances. As more retirees become eligible for the age pension than anticipated, in part due to the system’s incentives, other Australians will be adding to financial pressures on the pension system just by living longer than their parents. And these financial demands will be building at a time when governments are likely to be still dealing with the as-yet unquantifiable economic fallout from the COVID-19 pandemic.

Footnotes

Appendix 1 Acknowledgements

I am grateful to Professor Deep Kapur for encouraging me to explore the adequacy of our retirement system and his guidance throughout this research project. My colleagues Nga Pham, Bei Cui and Yolanda Wang provided helpful comments and suggestions. I appreciate the comments from the reviewers, which helped to improve the clarity of the paper. Finally, I would like to thank Dr Roger Cohen, who helped me to understand the different aspects of Australia’s superannuation and age pension system.

Final transcript accepted 13 April 2021 by Tom Smith (AE Finance).

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship and/or publication of this article.

ORCID iD

Ummul Ruthbah

Notes

References

Alonso-Garcia

Sherris

(2019) One size fits all? Drawdown structures in Australia and The Netherlands. The Journal of the Economics of Ageing 13: 14–27.

Americks

Veres

Warshawsky

(2001) Making retirement income last a lifetime. Journal of Financial Planning 14: 60–76.

Andreasson

Shevchenko

Novikov

(2017) Optimal consumption, investment and housing with means-tested public pension in retirement. Insurance: Mathematics and Economics 75: 32–47.

Arnott

(2004) Sustainable spending in a lower-return world. Financial Analysts Journal 60: 6–9.

ASFA (2012) The equity of government assistance for retirement income in Australia. ASFA research paper, ASFA, Sydney, NSW, Australia. Available at: https://www.superannuation.asn.au/ArticleDocuments/359/ASFA_Equity-of-government-assistance_Feb2012.pdf.aspx?Embed=Y

Asher

Meyricke

Thorp

, et al. (2017) Age pensioner deaccumulation: Responses to incentives, uncertainty and family need. Australian Journal of Management 42: 583–607.

Bengen

(1994) Determining withdrawal rate using historical data. Journal of Financial Planning 7: 171–180.

Beshears

Choii

Laibson

, et al. (2019) Trends in retirement income adequacy: Evidence form IRS tax data. National Bureau of Economic Research, RDRC Centre paper NB19-06. Available at: https://www.nber.org/sites/default/files/2020-04/NB19-06%20Beshears%20et%20al%20Summary.pdf

BetaShares (2020) Retirement income review submissions. Available at: https://treasury.gov.au/sites/default/files/2020-02/betasharescapital310120.pdf

10.

Boal

(2020) Spending in retirement and the taper rate. Firstlinks. Available at: https://www.firstlinks.com.au/spending-in-retirement-and-the-taper-rate

11.

Bray

Gray

(2016) Ageing and Money – Public Opinion on Pensions, Inheritance, Financial Wellbeing in Retirement and Caring for Older Australians. Canberra, ACT, Australia: ANU. Available at: https://csrm.cass.anu.edu.au/sites/default/files/docs/ANUPoll-ageing-money-feb-2016_0.pdf

12.

Bütler

Peijnenburg

Staubli

(2017) How much do means-tested benefits reduce the demand for annuities? Journal of Pension Economics and Finance 16: 419–449.

13.

Callaghan

Ralston

Kay

(2020) Retirement income review – Final report. The Treasury. Available at: https://treasury.gov.au/publication/p2020-100554

14.

Chen

Munnell

Sanzenbacher

(2018) How much income do retirees actually have? Evaluating the evidence from five national datasets. CRR WP 2018-14. Chestnut Hill, MA: Centre for Retirement Research at Boston College.

15.

Clare

(2001) Equity and Retirement Income Provision in Australia. Sydney, NSW, Australia: ASFA Research Centre. Available at: https://www.superannuation.asn.au/ArticleDocuments/359/equity.pdf.aspx?Embed=Y

16.

Cocco

Gomes

(2012) Longevity risk, retirement savings, and financial innovation. Journal of Financial Economics 103: 507–529.

17.

Cohen

Ruthbah

(2019) Pitfalls in the Retirement System and Some Thoughts about Risk, Reward and Remediation (White Paper 01-19). Melbourne, VIC, Australia: Monash Centre for Financial Studies.

18.

Cohen

Chen

Zhu

(2019) The Retirement Trap. Canberra, ACT, Australia: CSIRO Technical Report.

19.

Cole

Paulson

Shastry

(2014) Smart money? The effect of education on financial outcomes. Review of Financial Studies 27: 2022–2051.

20.

Cooley

Hubbard

Walt

(2003) Does international diversification increase the sustainable withdrawal rates from retirement portfolios? Journal of Financial Planning 16: 74–80.

21.

Credit Suisse (2020) Summary Edition Credit Suisse Global Investment Returns Yearbook 2020. Zurich, Switzerland: Credit Suisse Research Institute. Available at: http://credit-suisse-global-investment-returns-yearbook-2020-summary-edition.pdf

22.

Daley

Coates

Wiltshire

, et al. (2018) Money in Retirement: More Than Enough. Melbourne, VIC, Australia: Grattan Institute.

23.

Doyle

Mitchell

Piggott

(2004) Annuity values in defined contribution retirement systems: Australia and Singapore compared. Australian Economic Review 37: 402–416.

24.

Estrada

(2020) Managing to target (II): Dynamic adjustments for retirement strategies. The Journal of Retirement 7: 28–38.

25.

Ganegoda

Evans

(2015) The Australian retirement lottery: A system failure. Australian Journal of Management 42: 3–31.

26.

Gharghori

Sujoto

Veeraraghavan

(2008) Are Australian investors smart? Australian Journal of Management 32: 525–544.

27.

Gratton

Scott

(2016) The 100-Year Life Living and Working in an Age of Longevity. London: Bloomsbury Information.

28.

HESTA (2019) Investment Choices. Parramatta, NSW, Australia: HESTA.

29.

Horneff

Maurer

Rogalla

(2010a) Dynamic portfolio choice with deferred annuities. Journal of Banking Finance 34: 2652–2664.

30.

Horneff

Maurer

Mitchell

, et al. (2010b) Variable payout annuities and dynamic portfolio choice in retirement. Journal of Pension Economics and Finance 9: 163–183.

31.

Hurd

Rohwedder

(2012) Economic Preparation for Retirement in Investigations in the Economics of Aging. Chicago, IL: University of Chicago Press, pp. 77–113. Available at: http://www.nber.org/books/wise11-2

32.

Lusardi

Mitchell

(2006) Financial literacy and planning: Implications for retirement wellbeing. DNB Working papers 078. Amsterdam: Netherlands Central Bank, Research Department.

33.

ME Bank (2019) Household Financial Comfort Report. Melbourne, VIC, Australia: ME Bank. Available at: https://www.mebank.com.au/getmedia/899ccb18-7fe5-4868-b51b-b64ad2b467cf/216317_HFCR-Jan-2019_V11-WEB.pdf

34.

Mercer (2020a) Mercer CFA Institute Global Pension Index. Melbourne, VIC, Australia: Mercer. Available at: https://www.monash.edu/__data/assets/pdf_file/0009/2375748/MCGPI-2020-full-report-1.pdf

35.

Mercer (2020b) Submission to the retirement income review. Available at: https://treasury.gov.au/sites/default/files/2020-02/mercer030220.pdf

36.

Milevsky

Kyrychenko

(2008) Portfolio choice with puts: Evidence from variable annuities. Financial Analysts Journal 64: 80–95.

37.

Milevsky

Robinson

(2005) A sustainable spending rate without simulation. Financial Analyst Journal 61: 89–100.

38.

Munnell

Rutedge

Webb

(2014a) Are retirees falling short? Reconciling the conflicting evidence. CRR WP 2014-16. Chestnut Hill, MA: Centre for Retirement Research at Boston College.

39.

Munnell

Webb

Hou

(2014b) How much should people save? Issues in brief ib2014-11. Chestnut Hill, MA: Centre for Retirement Research at Boston College.

40.

OECD (2020) OECD Pensions Outlook 2020. Paris: OECD Publishing. Available at: https://doi.org/10.1787/67ede41b-en

41.

Poterba

Venti

Wise

(2011) The drawdown of personal retirement assets. NBER working paper 16675. Available at: https://www.nber.org/digest/jun11/drawdown-personal-retirement-assets#:~:text=On%20average%2C%20households%20aged%2060,their%20account%20balances%20each%20year.

42.

QSuper (2019) Investment Choice Guide. Brisbane, QLD, Australia: QSuper.

43.

Ravin

Liu

Rooyen

, et al. (2019) Spend Your Decennial Age: A Rule of Thumb Form Retirement. Sydney, NSW, Australia: Institute of Actuaries of Australia.

44.

Statman

(2014) Retirement income for the wealthy, middle and the poor. The Journal of Retirement 1: 31–36.

45.

The Financial Supervisory Authority in Iceland (2014) Retirement savings adequacy: Measurement in Iceland. Available at: https://en.fme.is/media/news/Retirement-Savings-Adequacy—Iceland.pdf

46.

The Treasury (2018) Retirement Income Covenant Position Paper: Stage One of the Retirement Income Framework. Canberra, ACT, Australia: Australian Government.

47.

Vernon

(2017) Life Planning in the Age of Longevity: An Action Plan. Stanford, CA: Stanford Centre on Longevity, Toolkit Series Brief. Available at: http://162.144.124.243/~longevl0/wp-content/uploads/2016/10/Life-Planning-Action-Plan.pdf

48.

Willis

(2011) The financial education fallacy. American Economic Review 101: 429–434.