September/October 2009

Managing Savings to Last a Lifetime
By Chuck Yanikoski
Aging Well
Vol. 2 No. 4 P. 32

Question: “How much can I safely withdraw from my retirement savings each year?”

Answer: “Don’t ask!”

By that, I don’t mean that you shouldn’t ask because the number is too frighteningly low—though sadly, that’s the case for many people. Rather, I mean that this question should not be asked because it is not pertinent to sensible money management.

The reason why it’s not pertinent is because most people who are retiring have a remaining life expectancy of at least 20 years. Some have a life expectancy closer to 30 years, and many people, regardless of life expectancy, will live as long as 40 years, with some living even longer—or their spouses will. What are the odds that cash flow needs will not change substantially during that long a span of time? Close to zero.

There are four reasons why the need to withdraw money will not remain constant in retirement. First, and most commonly, expenses shift upward over time. Even if the calculation of the regular withdrawal amount takes inflation into account, inflation rates can lurch unpredictably, as those of us who remember the early 1980s know well. But expenses can go up for other reasons. Medical costs such as hospital visits, expensive drugs, assisted living, and nursing home costs can, at different stages of one’s life, ratchet upward—and never come down. Income taxes could increase over time, as many people believe they will. A family need could arise. As we all know, new expenses somehow continue to crop up.

Second, other expenses fortunately can shift downward. Most older adults these days retire with a mortgage still in place that some day will go away, or they eventually move to apartments or smaller houses. Those who live into their 80s and 90s will probably start spending a lot less on entertainment, travel, expensive meals, home furnishings, nice clothes, and other discretionary items.

Third, some events create temporary spikes in expenses. It could be a medical emergency, property damage, or helping fund a grandchild’s education. It could be a binge of travel, home improvement, or an expensive hobby that doesn’t go on forever. Life is full of opportunities for unanticipated expenses.

And fourth, sometimes we get temporary spikes in income. An inheritance or a big life insurance settlement arrives or we sell a house or other property. Maybe we get lucky and a child or other family member finally pays off the loan they took from us.
For people who are married, one other factor is that one spouse usually dies before the other, which can cause a host of changes in all of these categories, including adjustments to Social Security and pension income.

Given all this, it is clearly foolish to even ask the question “How much can I safely withdraw from my retirement savings every year?” Yet amazingly, this is the question that many financial advisors, and most large financial institutions, continue to want to answer for people.

Unless they have a model that can take into account all of these future changes—and none of them currently does—the results they produce are worthless. In reality, they are less than worthless because they come from sophisticated people using sophisticated models, so clients actually believe them. This is a serious problem because the answers are often either way too high or way too low, so believing in them is dangerous.

If the answers are too high, then people withdraw too much and risk running out of money far too soon, especially in economic times like these. If the answers are too low, people withdraw less than they could, so they scrimp too much, impairing the quality of their lives and even the length.

It is also worth noting that these fancy models that do such a bad job have other serious flaws. They tend to be Monte Carlo models, which sound exotic and fun—or perhaps just mysterious and intimidating—but this simply means that instead of doing a single projection of a retiree’s situation using a single set of assumptions, hundreds or even thousands of alternative scenarios are randomly generated, and the advice is based on a combined analysis of all of them. This is intended to take into account the uncertainty of the future.

There is nothing wrong with the Monte Carlo technique, but it works correctly only when the risks that the model takes into account are fully understood. For example, suppose your money was invested some place where every year some neutral party flipped a coin, and you earned 10% if it came up heads but lost 5% if it came up tails. Since we know the odds on a coin flip, we could build a Monte Carlo model that accurately took this risk into account.

But no one fully understands the risks inherent in the financial markets. All we have to go on is about one long lifetime’s worth of historical data. Yet using that data implicitly assumes that the next 20 to 40 years will have the same underlying financial characteristics as the last 80 years or so. There is no reason to believe this will be the case.

On the contrary, the fact that baby boomers are aging and will soon be retiring in huge numbers will undoubtedly have large and unprecedented effects on the financial markets. Changes in regulations, tax structures, and economic policies will make a big difference. The rise of China and India as potentially dominant economic players could have a huge impact. A single suitcase-size nuclear bomb or a devastating health epidemic could change everything. The future is simply unknown and, while the financial models that pretend we understand the underlying risks are doing the best they can with the data they have, they are wrong to pretend that this is actually good enough and that the results have any real validity.

So what should we do? First, do not deal with any models that fail to take into account the four kinds of changes to future cash flows—no exceptions. It’s fair to legitimately question the expertise of anyone who is using or promoting such a model.

Second, if you or your client has access to one of the rare models that does take account of most changes in cash flows over time, it is important to revisit it every year to see whether things have changed. The retiree’s own circumstances may have changed—and, at the very least, the retiree is a year older, which should have some effect. But the financial markets may have changed in a year, other external factors may have changed, and the model itself may have been improved (made more sophisticated and with errors fixed).

Or what is better for many older adults is to take a completely different approach. The traditional way of dealing with retirement finances—before all the “financial geniuses” came onto the scene—was to invest conservatively, live on the investment income (in combination with Social Security and any pension payments that might be available), and not touch the principal until the end of life drew near and money may be needed for extra medical costs, living assistance, hospice, and burial. This approach still makes a great deal of sense, but even so, it will work best if future changes in income and expenses are taken into consideration. So a model that takes robust account of cash flow changes over time still has value.

Another worthwhile strategy for many retirees is to convert some of their savings into guaranteed income using a privately purchased annuity. This is especially helpful for older adults who do not have a pension of the traditional kind (i.e., paying benefits for life) but instead have 401(k)s, 403(b)s, IRAs, or other similar plans with big balances but no guarantees. Annuities become a better deal as people get older, though, so this is usually not as good a solution for elders retiring in their 50s or early 60s. And again, some understanding of likely future cash flow requirements is highly desirable in making this decision.

The bottom-line message here is that there are no safe, easy answers. But we can start by asking the right questions instead of the wrong ones. And we can seek out advisors and tools that use detailed, even cumbersome methods that are thorough rather than easy ones that spit out dangerously simplistic advice.

It’s always important to assume that all projections of the future are wrong and to revisit any analysis frequently, even if an individual’s own life doesn’t seem to be changing because, at the very least, the world is changing around us.

— Chuck Yanikoski is president of Still River Retirement Planning Software, Inc. in Harvard, MA, and principal founder of the Association for Integrative Financial and Life Planning.