We Need to Do Better at Predicting the End (of Retirement)
It takes more than a sense of the average length of retirement.
Let’s be honest. No one really wants to talk about the “end” of retirement, because the end of retirement means the end of, well, us. However, bleak as it may be, thinking about how long retirement will last and metrics like life expectancy after retirement are critical when it comes to financial planning and making it our goal to ensure people can enjoy their life and retirement to the fullest. So, “the end” of retirement is something we should do our best to measure as accurately as possible.
That’s why I decided to take a deeper dive into some key concepts and issues relating to the retirement period decision-making process and reported my findings in a recent white paper: Estimating ‘The End’ of Retirement. This article summarizes a few key points from the research, but I’d recommend reading the full paper if you want the entire scoop.
1. People aren’t great at estimating their own life expectancy.
For starters, I found that people aren’t very good at estimating their own mortality. Using responses from a public survey (the Health in Retirement Study), I was able to gauge the accuracy of predictions related to the probability of an individual living to age 75.
I found that while the average response across all households was reasonably in line with standard mortality predictions (think "wisdom of the crowds"), there were significant errors at the individual household level.
For example, individuals who said they had a 0% probability of surviving to a given age (75) actually had about a 50% chance of surviving, and those who said they had a 100% probability only actually had about an 80% chance.
This means personal opinions aren’t that useful when it comes to estimating life expectancy, and that using objective information like health status, in my opinion, is a better way to go.
2. There’s no average length of retirement--rather, retirement periods should be personalized.
Retirement periods need to be personalized based on each household’s situation. A number of attributes can have significant impacts on life expectancy, such as income and health status (especially whether someone smokes).
Incorporating this information can result in retirement periods that vary by more than 15 years, which can significantly affect required savings and/or optimal spending levels in retirement. For example, the average person who overestimates their life expectancy after retirement would be forecast to save 38% more than required; the average person who underestimates their life expectancy after retirement would be forecast to save 30% less than required.
In other words, the potential costs associated with getting the estimate wrong can be significant.
3. Many planners apply a “one size fits all” approach to identify an average length of retirement.
In my opinion, financial planners don’t appear to be personalizing retirement periods for clients. After reviewing life expectancy assumptions in 31,211 financial plans, I found that about 70% of plans used a retirement end age of 90 and about 20% of plans used age 95. This suggests that most planners are using a “one size fits all” approach and identifying an average length of retirement, rather than personalizing retirement period estimates for each client.
On average, financial advisors also didn’t appear to be incorporating the additional “tail risk” associated with the longer potential retirement periods for married couples (that is, planning for longest survival).
There’s a Better Way to Plan Than Going Off the Average Length of Retirement
To avoid some of these common shortcomings, I built a model to estimate a reasonable retirement period.
It starts with what we believe to be an accurate estimate of someone’s life expectancy after retirement based on their situation. When estimating the length of retirement, though, it’s important to also consider the shortfall aversion metric (for example, probability of success) to ensure recommendations are not overly conservative.
I found that adding five years to projected life expectancy for a single household and eight years to the longest life expectancy of either member of a joint household (or to each member if separate end ages are used) at retirement is a retirement-appropriate end age assumption. This approach suggests a retirement period of 30 years (to age 95) is a reasonable assumption for the average 65-year-old male/female couple retiring today.
However, retirement period assumptions should be revisited regularly to ensure they are timely, similar to other key assumptions in a financial plan.
I often refer to retirement as the “great unknown.” We don’t know when it’s going to start, when it’s going to end, or how much we’re going to spend each year--all we can do is guess. This is why the key is making educated guesses and not only relying on an average length of retirement. Hopefully this new research will help you make a more informed decision about when retirement could end.
Morningstar Investment Management LLC is a registered investment adviser and subsidiary of Morningstar, Inc. The information is the proprietary material of Morningstar Investment Management. Reproduction, transcription, or other use, by any means, in whole or in part, without the prior written consent of Morningstar Investment Management, is prohibited.