Morningstar - Q4 2020 - 63

EXHIBIT 2

EXHIBIT 3

Actual Survival Probability Versus
Self-Reported Probability of Living to
Age 75

Financial Advisors Aren't Adjusting Assumptions by Client Age or Gender
Male Retirement End Age
Assumptions

Actual
Actual
Probability
Probability
of of
Survival
Survival
(%) (%)
Actual
Probability
of (%)
Survival
100100 100
7575

75

5050

50

Red
Red
dots
dots
show
show
responses
responses
Red
dots
show responses
by by
decile
decile
group,
0 -010
-.10. 0 - 10. 2525
bygroup,
decile
group,

25

00
0
00
4040
6060
8080 100100
02020
20
40
60
80 100
Self-Reported
Self-Reported
Probability
Probability
of of
Living
Living
to to
Age
Age
75to75
(%)
(%) 75 (%)
Self-Reported
Probability
of
Living
Age

Female Retirement End Age
%%
of of
Total
Total
%%
of of
Total
Total
% of TotalAssumptions
% of Total
Financial
Financial
Financial
Financial
Financial
Financial
Plans Plans 100100
Plans Plans
YrsYrs10095Yrs
95 9090
100100
YrsYrs10095Yrs
95 9090
95 8585
90 8080
85 7575
80 Plans
75
95 8585
90 8080
85 7575
80 Plans
75
100100 100
100100 100
20%
20%
Average
Average
20%
20%
Average
Average
20% Average
20% Average

70%
70%
Average
Average
70% Average

5050
505555
556060
606565
Retirement
Retirement
Start
Start
Age
Age
Retirement
Start Age

657070

7575

75

5050

50

2525

25

00
70

Source: Health and Retirement Study.

Source: Anonymous Financial Planning Software.

personalized mortality factors when establishing
the retirement end date in a financial plan?
To determine this, data is obtained from a financial
planning platform available in Canada. A total
of 31,211 plans met the required filters from
an initial data set of 65,535 plans, which contain
a total of 48,923 retirement end age estimates.
(The number of retirement ages exceeds
the number of plans because some plans were
for two people.)

How to Model the End of Retirement
Even when suitable mortality rates have
been determined, there is no consensus approach
to estimate the length of retirement in a
financial plan. In other words, planners using the
exact same mortality table could reach very
different conclusions about the suitable length
of retirement for planning purposes.

E X H I B IT 3 includes the distribution of retirement
end age assumptions in financial plans by
gender. The vast majority (97.2%) of retirement
end age assumptions were multiples of 5, between
the ages 75 and 100, which is why only these
multiples are included in E X H I BI T 3 .

The distribution of retirement income ages is very
similar by gender and age, as well as by
assumed retirement start age. Approximately 70%
of retirement end ages were 90, whether male
or female, and approximately 20% of retirement
ages were age 95. The average assumed
retirement end age only increases by approximately
two years for men, moving from a start age of
50 to 70 (89.8 to 91.9) versus 0.9 years for women
(91.1 to 92.0). These findings suggest financial
advisors aren't personalizing retirement periods
for clients, at least in this data set.

In the original paper, I conducted an analysis
to determine how long the retirement period
should last once a personalized life
expectancy estimate has been determined.
In theory, considering the full distribution
of mortality rates, like in EXHIB IT 1 ,
would be the best way to go, but because the
information is not always made available
in online tools, life expectancy is used as the
base value.
The analysis suggests that adding five years
to a personalized life expectancy estimate
would be a way to estimate an appropriate
retirement period for a single individual. For joint
households, eight years should be added
to the longest life expectancy of the two members
or three years if the retirement period
considers the actual joint survival probabilities.
(For additional context on these results, I suggest
reading the original paper.)

70%
70%
Average
Average
70% Average

0
5050
505555
556060
606565
Retirement
Retirement
Start
Start
Age
Age
Retirement
Start Age

7575

75

5050

50

2525

25

00
70

0

657070

Estimate Accuracy Has a Considerable Impact
Personalized mortality estimates can have
a considerable impact on the results of a financial
plan. The assumed retirement period not only
affects things like required savings and
optimal spending, but also important decisions,
such as when to retire. Decisions about how to
fund retirement-whether to delay claiming
Social Security benefits, to purchase an annuity,
and others-will also be affected. Evidence
suggests that while subjective estimates may
be relatively accurate on average and that
households appear to do a relatively good job
considering various objective factors (such
as health status), there are often significant errors
in individual forecasts, and households do
not appear to correctly consider all the relevant
objective factors (such as income). Therefore,
financial planners need to educate themselves on
how to better model and personalize mortality
assumptions into financial plans. K
David Blanchett, Ph.D., CFA, CFP, is head of retirement
research with Morningstar Investment Research. He is a
member of the editorial board of Morningstar magazine.
Reference
Chetty, R., Stepner, M., Abraham, S. et al. 2016. "The
Association Between Income and Life Expectancy
in the United States, 2001-2014." Journal of the American
Medical Association, Vol. 315, No. 16, PP. 1,750-1,766.

morningstar.com/lp/magazine

63
http://www.morningstar.com/lp/magazine
Morningstar - Q4 2020

Table of Contents for the Digital Edition of Morningstar - Q4 2020
