Most individual invest-ors have one major investment goal — to save enough for retirement. Spending out of their portfolio is usually minimal in the years when they are working. Then spending becomes essential at the time of retirement. For this reason, there is a life cycle to investing. In the years when wealth is being accumulated, asset allocation is much more aggressive than when the investor nears retirement.
Investment firms have begun to formalize this process by which asset allocation changes over time. These firms have created target retirement funds, which change continuously as the investor gets closer to retirement. The funds are usually defined relative to the year of retirement. So in 2010, a 52-year-old might invest in a 2025 retirement fund because that investor intends to retire at 67 (for full Social Security benefits). Consider the 2025 target retirement fund offered by [The] Vanguard [Group Inc.]. The 52-year-old is initially invested in a 75%/25% stock/bond portfolio (diversified between domestic and foreign stocks). By the time of retirement 15 years later, this investor will have only 50% invested in stocks and 50% invested in bonds, with a 10% allocation in inflation-protected bonds (to help protect against inflation in retirement).
Target retirement funds are designed to model the life cycle of investing, beginning with the early years of working, when very aggressive allocations are called for. Until the investor reaches 25 years before retirement, Vanguard chooses a 90%/10% stock/bond allocation. The fund begins to increase its allocation to bonds until the investor finally reaches retirement age. Even after retirement, the allocation continues to shift. Five years after retirement, the stock/bond allocation is at 36%/64%. Experts can debate whether these specific allocations are optimal, but the figure shows clearly how the proportion of riskier assets depends on the distance from the age of retirement.
Vanguard is only one of the firms that offer such target retirement funds. The stock/bond ratio varies from 68%/32% in the Fidelity [Investments] program to 79%/21% in the T. Rowe Price [Group Inc.] program. And the percentage of stocks invested in foreign markets varies from 30% in the [The Charles] Schwab [Corp.] program to 20% in the Vanguard program. But it's clear that for all such programs, the asset allocation is governed by savings and spending decisions for retirement.
LONGEVITY
Many Americans really don't understand how long their retirement could be. Life expectancy has increased steadily over the past 50 years, while at the same time, the age of retirement has fallen. According to the Labor Department, the median age of retirement for both men and women is about 62. That's down from an average of between 66 and 67 in the 1950s. Americans at 62 often can look forward to 20 or even 30 more years of life in retirement. Yet few Americans have a coherent plan to make sure their resources will last that long. Savings are often inadequate and spending often is too high to be sustainable. Investment decisions, moreover, often are inconsistent with spending rates.
Some Americans are fortunate enough to have guaranteed pensions that provide them with a steady income throughout retirement. These are the old-style defined-benefit pensions that once were quite common in corporate America (and are still provided by many state and local governments). The pensions provide a guaranteed income to the employee, and often to the employee's spouse in the event of the death of the employee. Sometimes, the income is indexed to inflation, rising with the cost of living during retirement. Today, the balance has shifted away from defined-benefit pension plans to defined-contribution pension plans, like the 401(k) plan, where workers contribute part of their salaries to the plan, with firms often matching or supplementing the employee contributions. According to a Labor Department study, in 2003, only 24% of Americans in the private sector still had defined-benefit pensions, compared with 48% that had defined-contribution plans. Employees with the latter type of pension plan are, in a sense, responsible for their own retirement. If they save enough during their careers and invest wisely, they can enjoy a comfortable retirement.
How much is enough? That depends on how much they hope to spend in retirement and how much income they can derive from their portfolios. This chapter will explore both investing and spending in retirement. Decisions that Americans make about investing and spending can make a big difference in determining how financially secure they are in retirement.
In considering these issues, it will be helpful to know just how long our savings must last. For a 62-year-old man, the median age of death is estimated to be 85, with 25% of his cohorts likely to live to be 92. For a 62-year-old woman, the median age is 88 and the 25% point is reached at 94. For a married couple at 62, the relevant statistic is the life expectancy of the surviving spouse. The median age of death for the surviving spouse is 92. So the nest egg accumulated for retirement must last a long time.
With lifetimes this long, investment horizons must be just as long. In fact, they need to be longer because you may live longer than the average person your age. Yet Americans entering retirement often choose portfolios appropriate for retirees of their grandparents' generation, who typically lived only a few years after they retired. Retirees of that generation used to invest in bonds during retirement. Investing in bonds surely seems the safe thing to do. It helps us to sleep at night if we avoid stocks and other volatile investments. That's all well and good for emotional wellbeing, but does the average American realize how little can be spent if a portfolio is weighted heavily toward bonds?
[More: Investment Portfolio allocations: Is 70/30 the new 60/40?]
SPENDING RULES
There is a key concept in retirement planning that most Americans have not even heard of. That is the concept of a spending rule, a rate of spending in retirement that can be sustained through time. Foundations have spending rules that guide their activities through time. So must retirees, since they also must live off of their endowment — the wealth they have accumulated for retirement. If the retiree has a defined-benefit retirement plan, spending can be tied to the income from that plan (plus Social Security). Most of us are not fortunate enough to have such a plan. For the many Americans with only defined-contribution retirement plans, there is no guaranteed income from those plans, and retirement spending must depend on returns from accumulated wealth. So a spending plan is necessary.
Like foundations, some retirees base their spending in retirement on the income from their bonds and stocks. They choose their portfolios so as to maximize the coupons from their bond portfolios and dividends from their stock portfolios. This strategy may or may not be ideal as an investment strategy, but it should not be the basis of a spending rule. Retirees should be willing to use both income and principal from their portfolio if the spending can be sustained.
Like foundations, retirees have to worry about the volatility of their portfolio returns. We can measure that volatility using standard deviations, but somehow that doesn't fully capture investment risk as perceived by retirees. After all, a foundation can go out of business if it draws bad returns. A retiree must struggle on. The timing of returns certainly matters. Bad returns may occur early in retirement. A bear market like that experienced in 2001 or 2008 can cripple a retirement. To take into account such bad scenarios, we use simulation methods where a large range of different outcomes can be examined. By considering many trials drawn from a sample with given average returns and volatilities, we can try to model the investment uncertainties facing retirees. Each trial will consist of at least 1,000 simulations drawn randomly from a statistical distribution.
Returns are not the only source of uncertainty for the retiree. Central to investing and spending in retirement is longevity risk. If we knew for certain when we were to die, we could have a strategy for using up our capital before death. This is the principal behind annuities that guarantee income up until death.
LONGEVITY RISK
To create proper model spending rules, it's important to incorporate longevity risk directly into the simulation. The simulation software we will use, Financeware [from Financeware Inc.], does just that. Whenever a simulation is run, it draws from the mortality distributions developed by the Society of Actuaries. In any given simulation, the man or woman may die early in retirement or live long past the median age of death for that cohort.
By the time of retirement, the investor should have reduced the proportion of stocks in the portfolio way below that chosen during earlier working years. As discussed above, the Vanguard Target Retirement Fund shifts the investor from a 90% stock portfolio when the investor is 25 years from retirement to a 50/50 portfolio at retirement. The 50/50 retirement portfolio is commonly chosen, at least early in retirement.
Because risk is central to the success or failure of spending rules, we will try to reduce risk by diversifying the portfolio. In the case of bonds, the returns are based on the Barclays Aggregate Bond Index. Both foreign and domestic stocks are included, with foreign stocks being one-fifth of the total stock allocation (as in the Vanguard fund). Foreign stocks are represented by the Morgan Stanley [Capital International] EAFE Index, while U.S. stocks are represented by the Russell 3000 Index. So a 50%/50% bond/stock portfolio consists of 50% in the Barclays Aggregate index, 40% in the Russell 3000 and 10% in the MSCI EAFE Index.
The returns on these three indexes are obtained using the premium method. The premiums then are applied to the historical returns on the basic capital market assets, the S&P 500 and medium-term Treasury bond measured over the period since 1951. The compound real returns on these assets were 2.4% for the Treasury bond and 6.7% for the S&P 500. The premiums are 0.3% for the Barclays Aggregate and MSCI EAFE indexes, and 0% for the Russell 3000 Index. Standard deviations and correlations were calculated over the 1979 to 2009 sample period.
TWO CHEAPER THAN ONE?
We begin with two sets of simulations that will investigate differences between spending rules for single individuals and married couples. The first simulation will be for a 62-year-old man who has just retired. The second simulation will be for a 62-year-old married couple, also newly retired.
At retirement, it is assumed that the 62-year-old man will choose a spending rule that is to rise with inflation. For example, a 5% spending rule for a retiree with $1 million will permit the retiree to spend $50,000 (before tax) the first year. With a 2.5% inflation rate, spending will rise to $51,250 the second year, and so on. Later simulations will allow part of the spending to fluctuate with the size of the portfolio rather than being a set dollar amount (adjusted for inflation).
The man is assumed to want to use his wealth to support his retirement. That is, he has no plans to leave a bequest, so his wealth can be used up during his lifetime. This will allow him to raise his rate of spending higher than in the case in which his aim is to keep a given wealth level intact. Later simulations will allow for a specific bequest. Of course, he does not know his age of death ahead of time, so the challenge will be to adopt a spending rule that will keep his wealth positive throughout the remainder of his life.
This retiree must choose a spending rule low enough so that he does not run out of money before death. In the presence of uncertainty, however, it is difficult to eliminate all possibility of running out of money. So the aim is to choose a spending rule low enough so that the probability of running out of money (failure) is low. So we will be asking the following question: What is the probability of failure if a specific spending rule is adopted?
Simulations are run for spending rules ranging from 4% to 6% of initial wealth. The results for a single man at 62 are moderately encouraging. A spending rule of 4.5% has only a 4% probability of failure. That is, in 4% of the simulations, the 62-year-old man runs out of money before his death. A 5% spending rule raises the failure rate only to 10%.
Consider how these results are changed if the family consists of a married couple rather than a single man. The results will be different for two reasons. First, there are two lives to worry about rather than one, so the median age of death of the surviving retiree generally will be later than that of a single retiree. Second, the second person involved is a woman with a longer life expectancy.
Consider first a 5% spending rule. The probability of failure for the couple is 19% rather than the 10% found for a single man. If the spending rule is lowered to 4.5%, the probability of failure becomes 9% rather than 4%. The moral of the story seems to be: reduce your spending in order to enjoy marital bliss. Isn't it true that two can live more cheaply than one?
There are two features of the simulations discussed above that need to be investigated. First, the simulations assume the retirees are willing to use up wealth during their lifetimes. Some retirees may want to leave a bequest to charity or to their heirs. By introducing a planned bequest, the retiree also provides a cutoff point for the spending plan before wealth is more seriously depleted by market events. After all, not many retirees will adhere to a spending plan that completely impoverishes them. Second, the simulations assume that the retirees will keep spending at a given rate regardless of how high or low their investment returns are. We will investigate changing both features.
Consider first the bequest motive. The simulations might be designed so that the target level of wealth at death is some fraction of the original wealth level (adjusted for inflation). The retirees may choose this target for two different reasons. As stated above, the higher target will provide a bequest after death. But the higher target may also be chosen because retirees regard a decline in the portfolio anywhere near 100% as a disaster. Instead of a target level of zero wealth at death, suppose we assume that the retirees have a target level equal to 50% of the initial wealth.
VARIATIONS IN SPENDING
Raising the target level of wealth will raise the probability of failure, since now failure is defined as having assets fall below 50% of the initial wealth. So it's important to find some other way to mitigate the effects of a bad sequence of returns. Realistically, retirees are not going to keep spending the same amount if their wealth has fallen drastically. And they are not likely to keep their spending constant if they have had a whole string of good returns. So the second modification we will make is to have spending vary with current wealth.
It may not make sense for all spending to vary with wealth. (A proportional spending rule of 5% would cut the dollar amount spent in half, adjusted for inflation, if wealth falls by 50%). Perhaps a reasonable plan is to make half of the spending vary with wealth and to make half of it fixed (in real terms) over time. We will consider a plan where wealth is allowed to drop to 50% of its initial level and where half of the spending is tied to current wealth. So, for example, a 5% spending rule for retirees with $1 million will be split into two holdings: $25,000 that is held fixed over time (adjusted for inflation) and $25,000 that will vary with the level of wealth. (The fixed allocation might be designed to cover fixed expenses). This flexibility in the spending rule will make it easier to keep wealth above the target level.
These simulations are once again for 62-year-old married couples who are newly retired. For comparison purposes, the simulations based on no bequest are shown again. The results are quite distressing. Spending rules of 5% are downright dangerous with a 31% failure rate. (Would you want a nearly one-third chance of dropping 50% below your initial wealth?) Spending rules of 4.5% result in an 18% failure rate (compared with 95% when no bequest is made). The bottom line is that even spending rates as low as 4.5% appear to be problematic if the retirees want to make sure that the portfolio stays at least 50% intact.
Do retirees have portfolios large enough to live on less than 4.5% of wealth? Some do, of course. But for many American families, retirement in this new age of defined-contribution plans may be bleak.
Many retirees relying on their savings to finance a retirement envy the financial security of those with defined-benefit plans. Such plans guarantee an income flow for the rest of the retiree's life. In some cases, the income flows are indexed to inflation. The great advantage of such plans is that they insure against the most important risk in retirement — longevity risk.
Retirees relying on defined-contribution plans can try to create a defined-benefit plan ex-post. This income stream is obtained by investing a portion of the initial wealth in an immediate fixed annuity. The annuity works by pooling a large group of retirees of the same age in the same pool. Some will die early and will end up not capturing as much income as the average member of the pool. Others will die much later and will capture much more than the average member of the pool. By joining a pool of other retirees of the same age, retirees can guarantee that they never run out of money.
Why does that increase income in retirement? The reason is that the wealth invested in the annuity can be deliberately exhausted before death. It's always possible to increase the income on a bond portfolio if we are willing to use up the capital in that portfolio. The immediate fixed annuity does just that. Individual retirees cannot take the chance of using up all of their capital unless they know the date of their death in advance. But a pool of retirees can use up their capital jointly, since actuaries can predict fairly accurately the longevity of a large pool of retirees. To protect the individual retiree further, an insurance company not only organizes the pool but guarantees annuity payments in the event that the actuaries underestimate annuity commitments.
There is one big drawback of many annuities offered to retirees: They do not protect against inflation. For the same reason that we base spending rules on real returns so that nominal spending can rise with inflation, we should want to invest in annuities that are indexed to inflation. The income from these inflation-indexed annuities will naturally be lower than in the case of a nominal annuity. But investing in an inflation-indexed annuity will protect against two major risks in retirement — longevity risk and inflation risk. This is exactly what the current Social Security system does. It provides us lifetime income that is indexed to the CPI.
It's not really necessary to run new simulations to illustrate outcomes if such annuities are purchased. Consider a retiree who decides to invest a third of the portfolio in immediate annuities indexed to inflation. That annuity will provide a floor on retirement income very much like Social Security does. Given that floor, the retiree can consider simulations like those already analyzed but where disaster leaves the retiree with some floor level of income.
FOCUS ON RISK
Americans are retiring in their early 60s and living long lives in retirement. Many of these Americans lack the luxury of a defined-benefit plan providing them income in retirement. They might have accumulated wealth to carry them through their retirement years, but many do not understand how to invest that wealth and how to make sure that it lasts a lifetime. That's why it is so important to address the issue of spending rules in retirement.
No set of simulations can give you a spending rule that is the correct one. So let's summarize the key issues that those contemplating retirement must address. First, you need to focus on the risk that really matters in retirement — the risk of running out of money. Second, you need to base your spending rate on the returns you expect to earn on your portfolio after inflation has been taken out. Third, the spending rate has to be lower than the expected real, or inflation-adjusted, return because otherwise, your risk of failure will be too high. Fourth, you have to recognize that we are not sure about what average real returns will be in the future, so we may have to be even more conservative than past returns would indicate.
If investors want to raise spending in retirement beyond the spending rules analyzed above, then it makes sense to annuitize some of the retirement portfolio. But there is one other suggestion that may make sense to investors — postpone retirement. The benefits of working a few more years are multiple. First, Social Security benefits increase each year that retirement is postponed. A 62-year-old gains an extra 7% or more per year in benefits by delaying retirement until 66 (the normal retirement age for those nearing retirement). Second, the investor has a few more years to save for retirement. Third, the investor can allow the portfolio to grow further before beginning to draw it down with retirement spending. Fourth, when the investor finally does retire, annuities will provide even better returns than before. That's because the annuity tables work in favor of older investors.
Excerpted from “Portfolio Design: A Modern Approach to Asset Allocation” by Richard Marston (John Wiley & Sons Inc., 2011).
For archived columns, go to
InvestmetnNews.com/avisersbookshelf.
