Using Personal Data to Better Estimate Retirement Savings

Christine Benz: Hi, I'm Christine Benz for Morningstar. How well do people do at estimating the lengths of their retirement, and how can they do better? Joining me to discuss some research on that topic is David Blanchett. He's head of retirement research for Morningstar Investment Management. David, thank you so much for being here.

David Blanchett: Thanks for having me.

Benz: So, David, let's just talk about why it's so mission-critical when you're creating a retirement plan to have some sense of how long your retirement will be. Why is that such an important number?

Blanchett: Well, I mean, the length of retirement really defines its cost. So, if you have to live in 10 years in retirement versus 40, the estimate is very different. I did some research that came out last year, all about how to estimate when retirement's going to start. So, the new research is focused on the end of retirement, figuring out what really is an appropriate in-date for retirement.

Benz: So, how do people tend to do at this? How good are they at estimating essentially what is their life expectancy?

Blanchett: If you get 1000 people together, the average estimate's pretty good, but at the individual level, there's huge deviations in the actual value. Some fun statistics are, there's actually a panel data set where they ask individuals, How long do you think you'll live? And then they track those individuals to see if they survive to the point in time. Those that said they had a 0% chance of surviving had about a 50% chance, and those that said a 100% chance had about an 80% chance. So, people do get certain things right--they do incorporate their health status. But looking at these estimates from probabilities or even just point estimates makes me think I don't want to ask someone how long they're going to live. I'd rather have them take an online questionnaire and use that number for a financial plan.

Benz: Let's talk about the implications if someone makes a faulty estimate of how long they'll live--if they assume that they will live, say, a shorter time than they are actually likely to live.

Blanchett: Again, if you're setting the retirement period, whether it be 20 years or 40 years, and you're saving based upon that, you may over- or under-save based upon what it ends up being. If you're in retirement, you may end up over- or under-spending. So, I think that, again, no one knows how long they're going to live, but there are things that are related to life expectancy, like income, health, smoker status, all these things. I think that, at least on the margin, we should think about those when doing a financial plan or just estimating how long to actually plan for in retirement.

Benz: How customized do financial planners tend to be when estimating clients' life expectancies? Your research took a look at that as well, or you made some inferences about that.

Blanchett: I had information on about 30,000 financial plans, which is a pretty good number, but there's probably millions of them done every year. One thing that I found that was interesting is about 97% of plans used a multiple of five, so 90 or 95 or 100. About 70% used age 90 and about 20% used age 95. While that's good, it doesn't suggest that it's a lot of personalization. I don't think that planners that are at least using this software are asking questions to say, "Hey, I should use maybe age 87 for you and age 93 for you." They're just kind of using the same age for everyone, and I think that's going to lead to imprecise estimates about the cost of retirement.

Benz: So, you mentioned that ideally, someone would go through some kind of a questionnaire. What would be some of the key questions that would come up? You mentioned smoker status, health status, and so forth. What else should be in the mix when considering longevity?

Blanchett: There's some incredibly complicated calculators out there and there's some relatively simple ones. I think that anything that at least gets to health, ideally gets to income, incorporates things like gender, that's great. You can ask 25 questions or you can ask four, but I mean, to be honest, the evidence that I saw suggests that a lot of planners don't actually use any questions. They don't deviate their client assumptions at all based upon their client need and health status.

Benz: So in your research, you came away with some guidelines that you think planners or individuals thinking about their own life expectancy can use. Let's talk about how single people might approach this as well as how married couples might approach this topic.

Blanchett: Sure. So, when selecting the length of retirement, it's important where there's different levers that you can use to affect the outcome. You shouldn't really ever just use life expectancy. If you use life expectancy, there's a 50% chance you could live beyond the target period, so you've got to add on a margin of safety. What I did in the analysis is say, "Okay, let's assume that you actually are going to use life expectancy from an online personalized tool. You're going to go get a good estimate for you. What should you tack on to create a reasonable period?"

What I found was that you should add on five years for a single individual, or you should add on eight years to the longer of the two life expectancies for a joint couple. So, if you both go out and fill out an online questionnaire and one person says they're going to live 22 years, one person says 20, you'd use 30. The net out of all of it was that a 30-year retirement period for a married couple aged 65 is actually a pretty good starting place today, at least for most households.

Benz: David, you've spent a lot of time with these various calculators. Are there any that you really like and that you'd recommend?

Blanchett: There's a ton of them out there. The one that I used for the research is called the Longevity Illustrator. It's by the Society of Actuaries, and it only asks a few questions. It asks things like when you're going to retire, what is your health status, do you smoke? But for most people that actually gets down to the heart of the matter, and why I really like their calculator is because it shows you the probability of living to different ages. So, again, life expectancy is that 50-50 middle point. It doesn't tell you that there's a 10% chance you could live to age 105, but the Longevity Illustrator tool does. So, to me, tools that give you a more comprehensive perspective on how long you could possibly live, I think that that's better than just saying, "Hey, odds are you live to 95," and with no additional information.

Benz: And that's a free tool--I used it last night after reading the paper.

Blanchett: Yes.

Benz: Okay, David, really interesting research. Thank you so much for being here.

Blanchett: Sure.

Benz: Thanks for watching. I'm Christine Benz for Morningstar.