Targets help members to see, at any point in time, if they are adequately covered for retirement or need to supplement their savings. But what the target should be, is a complex issue. Needs vary widely. More importantly, savings targets don’t address the quality of protections an individual has in the course of their financial journey. In an ideal world, we need to design a multi-tiered target framework. To make these targets meaningful in the real world of high employee mobility, we need to establish a standardised approach that we can apply to all funds.
At the heart of employee benefits sits the employee: a person with needs, plans and dreams. And deciding how much to save for retirement is a big part of the financial planning that goes into meeting these needs and pursuing these aspirations1. When the defined contribution (DC) model was introduced, some boards of trustees believed this ‘financial planning’ problem simply transferred to the member.
The member alone knew what that ‘right amount’ for retirement would be for them. But if there has been one important lesson we’ve learnt from the DC experience, it is that the ability to define those targets was most definitely beyond the scope of most members. It became clear that to meet the needs of the broadest number of members, boards needed to establish targets, and manage funds around those targets.
What has also becoming increasingly apparent is that providing members with targets, and helping them understand whether they are achieving them over time, is an important catalyst for getting members to take action that might improve their retirement readiness2.
The graph below shows the improvement in retirement readiness an individual has achieved by using an online calculator to work out their target. The results are given for four different income quartiles and the graph shows that online calculators seem to be the most useful for lower-income earners– those very individuals who may not be able to afford a financial planner. So how do we decide what an appropriate target would be for a fund?
In setting a target at fund level, there are a few desirable characteristics that the target must fulfil, like: Targets must be: Realistic, Communicable, Balanced against other needs and Based on income requirements.
Targets must be: 1. Realistic 2. Communicable 3. Balanced against other needs 4. Based on income requirements
Trustees cannot set targets independently of the current design of the retirement fund. Contribution rates, expenses, normal retirement age and other pertinent factors are required to project the possible retirement funding outcomes that members could achieve. Sensitivity testing under a range of investment return assumptions can help to provide the range that the target must fall into. If members’ projected outcomes are clearly inadequate, an exercise of this sort may also provide an indication that trustees may need to redesign the fund.
Short service periods indicate that members may be accumulating their retirement funding from a number of different sources. As a result, it may make sense to communicate goals in terms of multiples of salary given that these are mathematically simpler to manipulate than replacement ratios3. This is the wealth-earnings ratio we describe later. Current estimates suggest that 75% of single men need 12.9 times annual salary at age 654. The comparative replacement ratio is 90.9%5. We argue later on in this chapter that trustees may find a goal that reflects a multiple of a member’s annual salary far easier to communicate to their membership.
In an effort to maximise their retirement funding outcome, individuals face conflicting objectives. They can increase their savings rate, but they will have to sacrifice take-home pay. They can allocate the highest possible portion of their retirement fund contributions to the investment vehicle, but that would be at the expense of important protections of their income and families. Setting a target realistically would involve finding that balance and looking at the problem holistically.
Individuals will have variable views on how to spend their retirement savings. Some may choose an inflation-linked income while others may choose a fixed, guaranteed income for life. We may need to change the way we calculate the target to accommodate these differences. For instance, if members all intend to buy an inflation-linked income, then we should calculate the replacement ratio based on inflation-linked annuity rates rather than a fixed, guaranteed annuity.
Perhaps the most practical suggestion is for trustees of DC funds to have a goal that includes both the total amount an individual needs to save, as well as some broad indication of what their savings will translate into at retirement to provide that post-retirement income. The goal may not, however, fulfil all of a member’s income requirements and trustee boards should communicate this to members. Armed with that information, members can then at least determine for themselves what their total picture is likely to be and whether they need to find alternative ways to address potential shortfall issues.
Members may have to address potential shortfalls between fund targets and their unique targets.
The replacement ratio is one of the more widely used targets for South African funds. Defined as the ratio of income in the year after retirement to the income (usually pensionable income) in the year before retirement, South African boards of trustees use replacement ratios as a way to determine whether fund members are achieving adequate outcomes. Replacement ratios originated in defined benefit (DB) funds because of how benefits were calculated:
Accrual rate x final salary x number of years of service
The pension benefit accrual rate usually varied around 2% and the resulting pension benefit replaced between 60% and 80% of the member’s pre-retirement income, based on between 30 and 40 years of service.
Replacement ratios can be difficult measures for individuals to understand. Wealthearnings ratios provide individuals with the multiple of current pensionable salary they would need for an adequate retirement income6. Wealth-earnings ratios share many similarities with replacement ratios, but interpretation of the outcome on the part of the member is made much easier with the use of multiples rather than percentages. This means that members with multiple retirement savings vehicles could easily work out the total collective outcomes of these funds to work out whether they will achieve an adequate income level.
Although the concept of using a target to help with financial planning and retirement fund design is a simple one, finding the quantum for the target can be difficult. South African savers and trustees may be familiar with the rule of thumb of aiming for a 75% replacement ratio. Indeed, academic literature from both the US and UK suggests targets (gross of tax) of 65%7 to 89%8. But perhaps we need to interrogate this number more fully.
90% of South African households will need more than 75% of their actual pre-retirement income
Using consumption smoothing as a way to estimate needs.
One way to estimate how much South African households would need is to estimate the lump sum at retirement that would allow pensioners to smooth their consumption into retirement and then divide this by the earnings at retirement to give a wealth-earnings ratio9. In practice, we can calculate targets for individual households by taking the value of goods and services consumed and then projecting this through to retirement.
At retirement, this consumption may change slightly. Research shows that most South African households don’t radically change their consumption at retirement although some households may start to spend more on healthcare and less on other things10. During retirement, health expenditure rises faster than other consumption until it reaches a critical level when the pensioner reins in their health expenditure by, for example, switching to state services and generic medicines. Other expenditure rises with inflation. Consumption may also increase if the pensioner is widowed and no longer benefits from shared living expenses. Consumption smoothing means that consumption can change at retirement as a result of lifestyle changes, but pensioners shouldn’t have to cut back due to lack of funds.
Once we’ve estimated this consumption stream, we can calculate the amount of income needed to provide for this consumption. This involves adjusting for the fact that the person’s property may be mortgage-free in retirement and that they would need to pay tax on income before they can spend it. When we’ve calculated the income stream, we can estimate the value of the lump sum required to provide for that income stream11. If you consider targets for real South African households using this method, it becomes clear how inadequate the 75% replacement ratio target is.
Data collected from households over the course of 2010 and 2011 suggested that 90% of households will need more that 75% of their actual (and not pensionable) pre-retirement income12. This was not an isolated result, as repeating the exercise for data collected in 2006 and 2007 produced a similar result13. These studies also showed that targets differ from household to household for a number of reasons and that the targets change over time because of changes in the economy and taxation14. This means that, at best, trustees can aim to get members within the ballpark of their target, but members will need to take responsibility for determining and reaching their own unique target.
Job mobility makes the replacement ratio convention particularly problematic. Most members are unlikely to retire with 40 years of service from one employer, so it becomes even more difficult to assess if their financial position at retirement is attributable to the retirement goal that any particular fund used along the way. It is arguable that a goal designed to be met over 40 years is of little practical use in a world where the average service period of an employee at any particular company is three years15.
Rather than standardise the number that all funds should target, we should standardise the approach to calculating the target.
One way to address the impact of high employee turnover is to have a standard target that all funds will aim to achieve for their members. This is not a new concept since many South African funds now target a replacement ratio of 75%. However, complications arise when we consider the fact that pensionable salary definitions vary across funds. This means that if we consider the target as a percentage of an individual’s pensionable salary, the true amount that funds are targeting can end up being quite different from one fund to the next. Setting an arbitrary target of 75% of pre-retirement earnings also ignores the consumption and expenditure pattern of fund members.
We therefore propose that rather than standardise the number that all funds should target, we should standardise the approach to calculating the target. A consumption smoothing model can be used to solve simultaneously for the contribution rate and the target that the fund will aim to achieve. The advantage of a consumption smoothing model is that it looks at the spending patterns of members and aims to give them a post-retirement income that will continue to support that spending. The added value of the consumption smoothing model is that it also solves for the required contribution rate.
Rather than using replacement ratios as our measure for the target, we propose the use of a wealth-earnings ratio. Targets like this are easily communicated and easily aggregated across different funds. And if all funds use the same approach to setting the target, the investment strategy and fund design used to achieve that target should not vary too much between funds. The consumption smoothing model will require information about fund salaries, salary increases over time and the normal retirement age. And when we consider the different income requirements of individuals we find that the wealth-earnings ratio that they require can vary from person to person. For instance, a person with higher housing values may have a higher target16. As such, the model may have to be run for distinct categories of workers, making the target even more specific.
However, this specific approach looks solely at the outcome the retirement fund must deliver and does not aim to optimise the allocation of contributions towards protection and saving at each point in a person’s life cycle. For this we will need a more holistic measure of a person’s needs. So in summary, a consumption smoothing model helps us to establish a target. That target can be in terms of a replacement ratio or a wealth-earnings ratio. But because wealth-earnings ratios are easily communicated to ordinary fund members, we suggest the use of this measure when calculating targets.
Replacement ratios and wealth-earnings ratios measure retirement readiness. But they don’t show how retirement readiness, and other employee benefits, fit into an individual’s journey to financial health. So, we now introduce a new measure, one which looks at an individual’s ability to achieve financial health over the course of their lives.
Individuals will have needs at each point of their life. Consider a 30-year-old man who is employed as a construction site engineer and who is about to be married. His employer has provided an arrangement that offers group life cover, disability insurance and retirement savings. However, his spouse earns less than he does and in the event of his death, he wants to ensure that she is able to comfortably maintain her lifestyle. In his view, the answer is to take out additional life cover.
He is still young and at high risk of becoming disabled while at work, but he chooses to forgo additional disability protection in favour of the life cover. What happens if he becomes disabled? Does he have enough protection or additional savings in place to maintain his standard of living? Is it more important to have life cover or disability cover? Individuals need tools to help them address these issues.
Achieving wellness at an individual member level involves considering mental, physical and financial needs. In Benefits Barometer 2013 we defined financial health as having enough income and net assets to meet your needs throughout your life, with enough protection in place to ensure that you can maintain this, irrespective of what life throws in your path. This means catering for immediate financial needs as well as preparing for both certain and uncertain events.
Events such as death are certain to occur, but the timing is not certain. Other events like disability or retrenchment are uncertain in both the timing and probability of occurrence. We need to make appropriate allowances for all these events, through savings (pre-funding) or buying insurance cover, to ensure we remain financially healthy throughout our lifetime. To measure financial wellness, we focus on what individuals need and compare this to what they have. We also consider the means available to an individual. Having less doesn’t have to mean being less prepared, but it does mean having to make more careful choices.
To carry out an appropriate assessment of current and future financial health, we need to at least consider:
If we can identify the events that may affect an individual at a particular stage of their life, we can match up the required protection that will help them cope.
This is a benchmark against which individuals can assess their current position to determine whether or not they have enough protection against various ‘shock’ events. Please note this is not about how much money an individual will have or make, but rather about the choices they make with the resources available and how what they have compares to what they need.
How can we use this measure?
A primary objective in measuring an individual’s financial health is to enable them to understand where their potential shortfalls are so that they can take corrective action. Equally important is knowing when there is too much cover and to adjust for this as well. While these scores have been designed to provide individuals with important feedback, employers or boards of trustees can aggregate individuals’ scores to create a score that is split by specific demographic characteristics such as age, gender, occupation or province.
This could provide a useful insight into which areas might require more focused communication. For example, if a company were to find that the majority of its 20- to 30-year-old married males were under-insured in terms of life cover, they could target this group through interventions or default structures and encourage them to increase their life cover. This is not only useful for trustees to identify which groups are over- or underinsured, but should also help them to determine the best form of education and advice.
Trustees could target under-insured groups through interventions or default structures and encourage them to increase their cover. This is not only useful to identify which groups are over- or underinsured, but should also help them to determine the best form of education and advice.
Under a defined contribution model, individuals are being asked to assume more responsibility for ensuring they meet their goals. Deciding how much to save for retirement is a function of the total sum of money an individual needs to provide for an adequate income during retirement17. For some, this may be less than their preretirement earnings. For others, a need for comprehensive medical cover in retirement may result in a required income that is greater than their pre-retirement earnings. The rest may need a post-retirement income that is equal to their pre-retirement earnings.
We have determined that, despite their flaws, first-tier targets like replacement ratios and wealth-earnings ratios, are a necessary part of the financial planning process in defined contribution funds. But we need to understand how employee benefits fit into the holistic well-being of the individual over their life cycle. For this reason, we need to increasingly consider second-tier targets, both as employers and individuals.
1 Bernheim, Forni, Gokhale & Kotlikoff (2000). Tacchino & Saltzman (1999)
2 Goda, Manchestwe and Sojourner (2012)
3 Butler & Van Zyl (2012)
4 Butler (2011)
5 Butler (2011)
6 Moore & Mitchell (1997)
7 Burns & Widdows (1990)
8 Greninger, Hampton, Kitt & Jacquet (2000)
9 Butler & Van Zyl (2012b)
10 Butler & Van Zyl (2012a). Butler (2013)
11 Butler & Van Zyl (2012b). Butler (2011)
12 Butler (2011)
13 Butler & Van Zyl (2012b)
14 Butler (2013
15 Member WatchTM 2012 data set
16 Butler (2013)
17 Mitchell & Moore (1998)