Better mortality tables?

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
psteinx
Posts: 5785
Joined: Tue Mar 13, 2007 2:24 pm

Better mortality tables?

Post by psteinx »

Mortality tables/lifespan calculators are useful tools for planning in general, and also for evaluating the value of SPIAs and similar investments. The latter, in particular, is of interest to me at the moment.

The baseline mortality tables that many folks are aware of are those connected with Social Security. They are rather generic, with the only demographic/health factors being, IIRC, your current age and sex.

But of course, we know that a lot of other factors have strong connections with longevity, including somewhat obvious direct health indicators (smoking status, BMI), and less direct demographic factors - education, income, race, marital status. Some of these factors overlap- education and income say - so detailed analysis would need to be careful not to double-count.

Additionally, current health status can be particularly closely associated with lifespan. An 80 year old with no significant current health issues is quite different from an 80 year old with stage IV cancer or who has had two heart attacks already or whatever. That said, it is perhaps unrealistic to expect a widely accessible table or calculator to accurately incorporate such information.

A further complication is that the recent COVID pandemic likely abruptly shifted life expectancy for older folks, and with COVID's waning, one might prefer a table using pre-2020 data.

Finally, one might expect that medical outcomes will generally improve over time, so life expectancy for, say, a 40-year old, using 2023 (or 2019) mortality tables may underestimate, given likely healthcare improvements in coming decades. (I'm assuming that standard mortality tables work by using death rates for each age bracket, and ~compounding them.)

Anyways, while recognizing that the perfect resource may not exist, what are some of the better resources available?
Last edited by psteinx on Tue Jun 06, 2023 9:00 am, edited 1 time in total.
sc9182
Posts: 2165
Joined: Wed Aug 17, 2016 7:43 pm

Re: Better mortality tables?

Post by sc9182 »

Considered driving record, do you own/drive Red Sports car, how best credit score, drinking habits, love in South Chicago — among gazillion additional risk/reward factors.

I am sure someone has a calculator, or some factors that go into fine-tuning individual’s life expectancy figure.

Have you recently gotten medical underwriting toward one Term life policy !? Those guys be pretty close to figuring it - because their money depends on assessing as accurately possible Gotha - it may reflect in good premium $amount, but they may not tell you - your life expectancy they calculated
Second Round
Posts: 230
Joined: Thu Sep 30, 2021 8:16 am

Re: Better mortality tables?

Post by Second Round »

Are you an actuary? It sounds like you're considering trying your hand at the work.

Anyway - the professional actuarial body for annuities (and other life & health insurance) is the Society of Actuaries ... www.soa.org

You mention a lot of underwriting factors that go into adjusting mortality tables. That's all true, but unless you want to dig around your local state's department of insurance rate filings (and even then, sometimes key pricing info may be held back confidentially for competitive reasons), you're not likely to know how pricing actuaries apply those things in a rating program. What's more, I think you may have overlooked one of the biggest factors when choosing an appropriate mortality table: the fact that it's an annuity, vs. life insurance, vs. population-level stats.

People who buy annuities tend to live longer than people who buy life insurance. It's due to self-selection: they're protecting against the risk of living too long, outliving their assets. Whereas life insurance purchasers are protecting against the risk of dying too soon - they tend to be less healthy. And population level stats are another ball of wax entirely ....

On the face of it, you need an annuity table. And a lot of adjustment factors ... not just for the underwriting factors, but also adjusting for what used to be annual mortality improvement. Things have changed since Covid.

But at the core of what you're trying to get at, what you really need is a credentialed actuary, qualified to practice in the annuity area. Not many of them work for the general public, unfortunately.
Pops1860
Moderator
Posts: 1778
Joined: Thu Mar 14, 2013 4:05 pm

Re: Better mortality tables?

Post by Pops1860 »

This thread has been moved to the “Investing - Theory, News & General” forum. Moderator Pops1860
The power of accurate observation is often called cynicism by those who do not have it. ~George Bernard Shaw
Rex66
Posts: 2926
Joined: Tue Aug 04, 2020 5:13 pm

Re: Better mortality tables?

Post by Rex66 »

you just arent going to get better data bc better data is proprietary and purposefully not disclosed. It also isnt very actionable bc you can only take whatever spias are available as an example you mentioned. Lets just pretend they are all ripping you off. Nothing you can do about it if you want to buy a spia. You can buy the cheapest spia but none of the data actually helps you pick that. You can do that just based on price. SPIAs as a general rule dont have underwriting. There are a few which will include factors if you are really sick and push for it but that isnt the common purchase situation and a more limited market.
User avatar
Stinky
Posts: 14088
Joined: Mon Jun 12, 2017 11:38 am
Location: Sweet Home Alabama

Re: Better mortality tables?

Post by Stinky »

The Society of Actuaries, which is the professional education and research organization for life, health, and pension actuaries, has a tool on its website that gives access to more than 3,000 mortality tables. Here’s a link to the tool: https://mort.soa.org/

But even a tool as large as this doesn’t cover all of the mortality tables used by practicing actuaries. Most all life insurance companies have multiple proprietary mortality tables that they actually use for pricing their products. These “pricing” mortality tables are based, at least in part, on the company’s own experience. The very largest companies can construct pricing tables purely from their own mortality experience, while most companies need to blend in industry mortality experience at least some measure. So there are thousands and thousands more mortality tables that are not publicly available.

Presuming that you’re not a practicing actuary, there aren’t many uses for mortality tables. One use is to construct “life expectancy” tables, which can give a rough idea of the average future lifetime of a group of people of your age and sex.

You can get a rough idea of the life insurance company’s view of your mortality if you’ve been recently issued a level term life policy. Be aware that the mortality cost of term life is roughly half of the premium, with the other half going to expenses, commissions, taxes, cost of capital, and profit. So, let’s say that you’ve been recently issued a $1 million 20 year term policy at a premium of $1,000 per year. So your total premiums over the 20 years is $20,000, of which one half, or $10,000, will be used to pay death claims. So your chance of death over the next 20 years is ($10,000 / $1,000,000), or 1%. That’s not 1% per year - that’s 1% over the next 20 years.
Retired life insurance company financial executive who sincerely believes that ”It’s a GREAT day to be alive!”
bh1
Posts: 326
Joined: Thu Dec 30, 2021 1:49 pm

Re: Better mortality tables?

Post by bh1 »

I use mortality tables for investment decisions, so the suggestions here that mortality tables are only for actuaries are bogus.

I use the SS tables but adjust for age. Figure out how much healthier you are than the typical American and make yourself that much younger before doing a table lookup. It isn't hard to find data online for "smoking makes your life expectancy X years shorter" and converting that to the corresponding age shift by looking at the change in expectancy from test age shifts.
stan1
Posts: 14235
Joined: Mon Oct 08, 2007 4:35 pm

Re: Better mortality tables?

Post by stan1 »

If this is for you not a group of people, I'm not sure if a table is better than making a guess based on your own health and how long parents, siblings, aunts and uncles have lived (taking into account smoking, occupational risks, and advancements in medicine). So 80 (poor health/smoker), 90 (decent health), 100 (good health and long family longevity). Use different ages if you don't like mine for some reason.
Topic Author
psteinx
Posts: 5785
Joined: Tue Mar 13, 2007 2:24 pm

Re: Better mortality tables?

Post by psteinx »

The primary use case for me, at the moment, is whether to advise my elderly mother to convert her TIAA traditional to an annuity.

The rates within TIAA are better than what I saw on immediateannuities.com, so I'm not cross-shopping, just wondering how appealing (or not) it would be to exercise this conversion option.

That said, there are other instances in which I (and I presume others) would like to estimate longevity. There's a continuum, I'm sure, between the rather crude, simplistic approach of just using the Social Security tables versus something much more detailed and, perhaps, accurate. I'm not really aiming for perfection, not looking to be an actuary, and realize that any number is subject to some mental massaging from the general population to a specific individual.

A starting point that is somewhat more precise/narrow than "83 year old female" might be helpful...
User avatar
Watty
Posts: 28813
Joined: Wed Oct 10, 2007 3:55 pm

Re: Better mortality tables?

Post by Watty »

I would be cautious thinking that you can predict lifespan with any precision even with a complex model and lots of historical data.

The reason is that medical care is a lot different than it was 25 years ago and may things like heart attacks and cancer are much more survivable now than they were a generation ago.
psteinx wrote: Mon Jun 05, 2023 5:49 pm Anyways, while recognizing that the perfect resource may not exist, what are some of the better resources available?
They may or may not be better but here are some are different resources.

A huge factor for a couple is that their individual life expectancy is not nearly as important as their joint life expectancy. This calculator is fairly simplistic but it will help you look at questions like "My wife and I are both 65, how likely is it that at least one of us will live to be 95?"

Another different one is a this retirement calculator which tries to look and your financial model and then factor in the odds of you actually living to live to a certain age.

"Rich, Broke, or Dead"

https://engaging-data.com/will-money-last-retire-early/

For example if your retire at 65 the classic safe withdrawal rate studies would suggest that you should start with a safe withdrawal rate of around 4%. If you instead start out spending 8% you will likely be heading towards running out of money before the 30 year time period is up but most of the time you will die before you actually run out of money.
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Better mortality tables?

Post by McQ »

psteinx wrote: Mon Jun 05, 2023 5:49 pm Mortality tables/lifespan calculators are useful tools for planning in general, and also for evaluating the value of SPIAs and similar investments. The latter, in particular, is of interest to me at the moment.

The baseline mortality tables that many folks are aware of are those connected with Social Security. They are rather generic, with the only demographic/health factors being, IIRC, your current age and sex.

...

Finally, one might expect that medical outcomes will generally improve over time, so life expectancy for, say, a 40-year old, using 2023 (or 2019) mortality tables may underestimate, given likely healthcare improvements in coming decades. (I'm assuming that standard mortality tables work by using death rates for each age bracket, and ~compounding them.)

Anyways, while recognizing that the perfect resource may not exist, what are some of the better resources available?
For your purposes, please do not use the SS tables. Use the IRS tables, which assume an annuitant population, i.e., life expectancy achieved by the kind of individual who would actually dare to give an insurance company a large amount of money in return for a promise of life time income.

Here at Bogleheads, when evaluating SPIA and the like, it is important to use annuitant life expectancy. That means the life expectancy for elderly folks who don’t use tobacco, are not couch bound, did not subsist on burgers, fries, and milkshakes, and who can reasonably hope to live a long time based on health, lifestyle, and family history. Else, they would never dare give an insurance company a large sum of money.

The IRS publishes a unisex mortality table for this population: https://www.federalregister.gov/documen ... ng-minimum. It has live footnotes that take you to the actuary sites that provided the source and the method (including updated mortality estimates).

Kitces publishes a joint life expectancy spreadsheet: https://www.kitces.com/joint-life-expec ... alculator/

You can paste the IRS table into both male and female columns of that SS to probe joint life expectancy, i.e., the likelihood that one of you will still be alive 25, 30, 40 years from now.

You may also find this site useful for a definition of terms:https://lifeexpectancy.org/lifetable.shtml Life expectancy is not necessarily defined according to a common sense interpretation of those two terms; the formal definition is given there (LE is not to be confused with median age of survival).

As a teaser, annuitant LE runs 3-4 years more than SSA LE. Makes a huge difference in deciding whether an SPIA maximizes utility.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
User avatar
Stinky
Posts: 14088
Joined: Mon Jun 12, 2017 11:38 am
Location: Sweet Home Alabama

Re: Better mortality tables?

Post by Stinky »

As mentioned above, a "consumer" use for mortality tables is to get an idea of personal longevity.

If one does a Google search on "how long will I live", the results will show dozens of websites that purport to estimate your personal longevity.

My personal favorite website for this purpose is www.longevityillustrator.org. This website, which is a joint effort of the Society of Actuaries and the American Academy of Actuaries (the two primary professional actuarial bodies in the United States), accepts user inputs as to current age, sex, smoking status, and self-reported health to develop an estimate of longevity.

"Life expectancy" is often reported as a single, apparently definitive number. And that is what it may be for a large population of similarly situated lives. But for an individual, it's arguably more useful (and certainly more informative) to see "what are the chances that I live X more years", with numerous values of "X" being provided to the user.

The results produced by this tool vary considerably by the input parameters. For example, using a male who is exactly 70 years old -
--- If he reports as being a non-smoker in excellent health, the tool reports that there is a 50% chance he will live to age 88, a 10% chance he will live to age 98, and a 2% chance he will live to age 105.
--- If he reports as being a smoker in poor health, the tool reports that there is a 50% chance he will live to age 76, a 10% chance he will live to age 85, and a 2% chance he will live to age 92.

The tool also allows the input of data on two individuals, to show (for example) the chances of one or both of the individuals living to a certain age. This can be especially useful for a married couple to plan their retirement withdrawal strategy.

Based on my personal and professional experience, this tool is the one that I would recommend for consumer use, because of the credibility of the groups that developed the tool and the inputs that were used.
Retired life insurance company financial executive who sincerely believes that ”It’s a GREAT day to be alive!”
Rex66
Posts: 2926
Joined: Tue Aug 04, 2020 5:13 pm

Re: Better mortality tables?

Post by Rex66 »

I have patients come in all the time who thought they were either unhealthy in their 60s and now in their 90s or who are in their 60s but unfortunately just developed a terminal condition as their first real problem. The precision just isn’t possible for an individual. That’s why insurance companies insure so many as well as pad the stats a tad. They know that even they can’t get it right on an individual basis to a degree that’s reasonable.
RubyTuesday
Posts: 2236
Joined: Fri Oct 19, 2012 11:24 am

Re: Better mortality tables?

Post by RubyTuesday »

I’ve often read posters writing that annuity purchasers often live longer than the general population due to self selection by healthier than average buyers.

I have some background working with companies to transfer their defined benefit pension obligations to insurers by purchasing annuities for the beneficiaries.

Those “purchasers” would not be self selecting and would not have any obvious bias toward being more healthy.

And it’s a very large market: 2022 market over $48 Billion.

Source
US Pension Plan Buyouts Reach Record Volume 2022

And

Pension risk transfer market off to strong start this year
While the first quarter estimate of $6 billion is significant, the second quarter has already exceeded that total with a single transaction. AT&T Inc., Dallas, announced on May 1 it had completed an agreement to purchase group annuity contracts from two Athene Holding subsidiaries to transfer $8.1 billion in U.S. pension plan liabilities. In August, Athene Annuity & Life Co. and Athene Annuity & Life Assurance Co. of New York will take on the responsibility for paying benefits to about 96,000 AT&T participants and beneficiaries.
So I wonder what portion of the annuity market is related to pension risk transfers versus individual self-selecting annuitants.

I imagine insurers would price individual policies much less competitively than they would price group annuities to a plan sponsor, whose participants may even be less healthy than average populations.

At any rate, I’m veering a bit off topic at this point.
“Doing nothing is better than being busy doing nothing.” – Lao Tzu
RubyTuesday
Posts: 2236
Joined: Fri Oct 19, 2012 11:24 am

Re: Better mortality tables?

Post by RubyTuesday »

RubyTuesday wrote: Tue Jun 06, 2023 6:45 am
So I wonder what portion of the annuity market is related to pension risk transfers versus individual self-selecting annuitants.
From LIMRA I learned:
Improved interest rates have spurred growth in the income annuity market. Single premium immediate annuity (SPIA) sales were $3.1 billion in the fourth quarter, a year-over-year increase of 94%. In 2022, SPIA sales were $9.1 billion, 44% higher than 2021 results. In the fourth quarter, deferred income annuity (DIA) sales rose 59% to $720 million. For the year, DIA sales were $2.1 billion, up 24%. LIMRA forecasts income annuity sales to experience steady growth through 2026.
So the Pension Risk Transfer market would seem to be significantly larger than the individual specific premium immediate and deferred annuity markets.
“Doing nothing is better than being busy doing nothing.” – Lao Tzu
User avatar
House Blend
Posts: 4877
Joined: Fri May 04, 2007 1:02 pm

Re: Better mortality tables?

Post by House Blend »

psteinx wrote: Mon Jun 05, 2023 9:50 pm The primary use case for me, at the moment, is whether to advise my elderly mother to convert her TIAA traditional to an annuity.

The rates within TIAA are better than what I saw on immediateannuities.com, so I'm not cross-shopping, just wondering how appealing (or not) it would be to exercise this conversion option.
Cross-shopping or not, it is worth noting that one factor favoring TIAA in this case is that their payout rates are unisex. (Consequences of ERISA.)

So, all else being equal, females seeking a one-life annuity will get a better deal from TIAA than from a provider whose payout rates are different for male vs. female.
Rex66
Posts: 2926
Joined: Tue Aug 04, 2020 5:13 pm

Re: Better mortality tables?

Post by Rex66 »

Many pension people will chose lump payment if unhealthy.
itsmeagain
Posts: 224
Joined: Sat Jan 09, 2016 6:24 pm

Re: Better mortality tables?

Post by itsmeagain »

Stinky wrote: Tue Jun 06, 2023 4:11 am As mentioned above, a "consumer" use for mortality tables is to get an idea of personal longevity.
...
My personal favorite website for this purpose is www.longevityillustrator.org. This website, which is a joint effort of the Society of Actuaries and the American Academy of Actuaries (the two primary professional actuarial bodies in the United States), accepts user inputs as to current age, sex, smoking status, and self-reported health to develop an estimate of longevity.
...
The tool also allows the input of data on two individuals, to show (for example) the chances of one or both of the individuals living to a certain age. This can be especially useful for a married couple to plan their retirement withdrawal strategy.
...
Thank you, Stinky, for this very helpful link. I found it simple to use, with clear and useful output.

My analysis made me both happy and sad. Happy because it reinforced the excellent deal I got on a TIAA Traditional SPIA a few months ago. But sad reflecting on the probabilities of survival (each individual, both, neither), which reminds me that time is the ultimate limiting resource.
User avatar
MJS
Posts: 1309
Joined: Sat Aug 05, 2017 10:55 pm

Re: Better mortality tables?

Post by MJS »

Did you already look at the Bogleheads list of "Life expectancy calculators" at https://www.bogleheads.org/wiki/Tools_and_calculators ?
Ipsa scientia potestas est. Bacon F.
User avatar
Watty
Posts: 28813
Joined: Wed Oct 10, 2007 3:55 pm

Re: Better mortality tables?

Post by Watty »

RubyTuesday wrote: Tue Jun 06, 2023 6:45 am I’ve often read posters writing that annuity purchasers often live longer than the general population due to self selection by healthier than average buyers.

I have some background working with companies to transfer their defined benefit pension obligations to insurers by purchasing annuities for the beneficiaries.

Those “purchasers” would not be self selecting and would not have any obvious bias toward being more healthy.
It gets tricky.

A pool of people that are healthy enough to get hired and successfully stay employed for long enough to qualify for a pension will likely be healthier than the general population.
Wrench
Posts: 1046
Joined: Sun Apr 28, 2019 10:21 am

Re: Better mortality tables?

Post by Wrench »

RubyTuesday wrote: Tue Jun 06, 2023 7:12 am
RubyTuesday wrote: Tue Jun 06, 2023 6:45 am
So I wonder what portion of the annuity market is related to pension risk transfers versus individual self-selecting annuitants.
From LIMRA I learned:
Improved interest rates have spurred growth in the income annuity market. Single premium immediate annuity (SPIA) sales were $3.1 billion in the fourth quarter, a year-over-year increase of 94%. In 2022, SPIA sales were $9.1 billion, 44% higher than 2021 results. In the fourth quarter, deferred income annuity (DIA) sales rose 59% to $720 million. For the year, DIA sales were $2.1 billion, up 24%. LIMRA forecasts income annuity sales to experience steady growth through 2026.
So the Pension Risk Transfer market would seem to be significantly larger than the individual specific premium immediate and deferred annuity markets.
True. But the insurance companies almost certainly don't use the same mortality tables for those two different markets so that fact is probably irrelevant to pricing for individual SPIA purchasers.

Wrench
ryman554
Posts: 1634
Joined: Sun Jan 12, 2014 8:44 pm

Re: Better mortality tables?

Post by ryman554 »

McQ wrote: Mon Jun 05, 2023 10:34 pm For your purposes, please do not use the SS tables. Use the IRS tables, which assume an annuitant population, i.e., life expectancy achieved by the kind of individual who would actually dare to give an insurance company a large amount of money in return for a promise of life time income.

Here at Bogleheads, when evaluating SPIA and the like, it is important to use annuitant life expectancy. That means the life expectancy for elderly folks who don’t use tobacco, are not couch bound, did not subsist on burgers, fries, and milkshakes, and who can reasonably hope to live a long time based on health, lifestyle, and family history. Else, they would never dare give an insurance company a large sum of money.
Is it me, or when I look at the IRS unisex mortality table (your link is broken) https://www.irs.gov/pub/irs-drop/n-22-22.pdf, the annuitant is actually *more* likely to die than the non-annuitant? Pick an age where it matters (say 70). It's strange that there is a crossover where the non-annuitant is more likely to die at age 50. I thought reading this table -- if I get to age <N>, what's the probability I die in a year.

Why would it crossover?
McQ
Posts: 1414
Joined: Fri Jun 18, 2021 12:21 am
Location: California

Re: Better mortality tables?

Post by McQ »

ryman554 wrote: Wed Jun 07, 2023 8:54 am
McQ wrote: Mon Jun 05, 2023 10:34 pm For your purposes, please do not use the SS tables. Use the IRS tables, which assume an annuitant population, i.e., life expectancy achieved by the kind of individual who would actually dare to give an insurance company a large amount of money in return for a promise of life time income.

Here at Bogleheads, when evaluating SPIA and the like, it is important to use annuitant life expectancy. That means the life expectancy for elderly folks who don’t use tobacco, are not couch bound, did not subsist on burgers, fries, and milkshakes, and who can reasonably hope to live a long time based on health, lifestyle, and family history. Else, they would never dare give an insurance company a large sum of money.
Is it me, or when I look at the IRS unisex mortality table (your link is broken) https://www.irs.gov/pub/irs-drop/n-22-22.pdf, the annuitant is actually *more* likely to die than the non-annuitant? Pick an age where it matters (say 70). It's strange that there is a crossover where the non-annuitant is more likely to die at age 50. I thought reading this table -- if I get to age <N>, what's the probability I die in a year.

Why would it crossover?
Apologies for the broken link. Try: https://www.federalregister.gov/documen ... ng-minimum

If that doesn't work search on Federal register required minimum distributions.

Your link went to a mortality table for valuing pensions. It is beyond my paygrade to answer your question (aka, I'm not an actuary). The link I intended governs required minimum distributions from IRAs etc. Under statute, the IRS must from time to time bring its life expectancy estimates for taking withdrawals from retirement plans up to date, as it did here.*

*If serious about using this document as a guide to annuitant life expectancy, read until you find "11/24" in a sentence, and adjust the numbers in the life expectancy tables accordingly.
You can take the academic out of the classroom by retirement, but you can't ever take the classroom out of his tone, style, and manner of approach.
Post Reply