STUMP » Articles » The Return of Mortality Monday and RIP, President GHW Bush » 5 December 2018, 05:44

(Yes, I know it’s Wednesday, but I’ve been having internet access issues at home)

I may have a few more Spanish flu pandemic posts — actually, they will tie in nicely with what I plan on writing about for Mortality Monday — but now I will be returning to my original love: mortality trends.

There are two things that are bringing me back, but I’m going to be taking a much more basic approach.

RIP, PRESIDENT BUSH

I have no particular memories to share about President Bush, but I want to point out one thing I had to say to other people: I expected him to die this year, for two reasons. The first reason: because his wife died in the spring.

Back in 2017, I wrote about broken heart syndrome, which is a real phenomenon:

Someone wrote an email to me recently, asking about mortality assumptions when pricing annuities:

For pricing a joint annuity is it normal to assume mortality of the two lives are independent? I’d guess that there might be some positive correlation (e.g. b/c of life style), but I’ve never seen any reference made to that.

Here is the answer I wrote:

For annuities, yes.

For life insurance, no.

Let me explain why — there is a correlative effect, in that if one dies, the other has a heightened mortality, at least in the short run [there really is a “broken heart” syndrome.]

Actuaries developed copulas originally for pricing Joint Last Survivor life insurance, because the results were bad (from the insurer point of view) if you used the independence assumption. The copulas had some built in positive correlation.

With the built in positive correlation, the joint life status has a shorter life expectancy. The issue with independence is that the joint life status for the same base mortality tables will give you a longer life expectancy — bad for life insurance.

You don’t want to assume the joint lives will have shorter lifespan when pricing annuities. That’s not being conservative. The independence assumption is conservative for joint annuities.

Thing is, my explanation there, while correct, may be confusing if you’re not used to calculating mortality and survival rates.

Over the next several weeks, I will build up the concepts used in life actuarial work (which, of course, is all about death). I will start with core concepts: q_x, omega, life expectancy, choice of mortality tables, building mortality tables, and more.

These will not be enough info to pass any actuarial exams, but you may better understand reasoning behind mortality.

Oh, and I also really expected President Bush to die this year, not only because of his wife’s death (which would have boosted his mortality rate), but that the mortality rate for males age 94 in recent calendar years is about 25% (according to Social Security 2015 calendar year tables). That is really high as a mortality rate.

To contrast, mortality at age 65 for males is about 1.6%. While broken heart syndrome can boost mortality at younger ages, doubling the one year rate from 1.6% to 3.2%, say, is noticeable to actuaries, but it’s not like 25% getting boosted to 50% (and no, the effect is not necessarily doubling. I was just showing that the same multiplier will end in something that is fairly probable at a high age compared to something that is still low probability at a much younger age.)

ON LIFE EXPECTANCY DROPS

You may have heard last week about life expectancy dropping in the U.S. for the third year in a row.

WSJ: U.S. Life Expectancy Falls Further

‘We are losing too many Americans, too early and too often, to conditions that are preventable,’ CDC head says

Life expectancy for Americans fell again last year, despite growing recognition of the problems driving the decline and federal and local funds invested in stemming them.

Data the Centers for Disease Control and Prevention released on Thursday show life expectancy fell by one-tenth of a year, to 78.6 years, pushed down by the sharpest annual increase in suicides in nearly a decade and a continued rise in deaths from powerful opioid drugs like fentanyl. Influenza, pneumonia and diabetes also factored into last year’s increase.
…..
The U.S. has lost three-tenths of a year in life expectancy since 2014, a stunning reversal for a developed nation, and lags far behind other wealthy nations.

Life expectancy is 84.1 years in Japan and 83.7 years in Switzerland, first and second in the most-recent ranking by the Organization for Economic Cooperation and Development. The U.S. ranks 29th.
…..
White men and women fared the worst, along with black men, all of whom experienced increases in death rates. Death rates rose in particular for adults ages 25 to 44, and suicide rates are highest among people in the nation’s most rural areas. On the other hand, deaths declined for black and Hispanic women, and remained the same for Hispanic men.
……
Earlier this century, the steady and robust decline in heart-disease deaths more than offset the rising number from drugs and suicide, Dr. Anderson said. Now, “those declines aren’t there anymore,” he said, and the drug and suicide deaths account for many years of life lost because they occur mostly in young to middle-aged adults.

While progress against deaths from heart disease has stalled, cancer deaths — the nation’s No. 2 killer — are continuing a steady decline that began in the 1990s, Dr. Anderson said. “That’s kind of our saving grace,” he said. “Without those declines, we’d see a much bigger drop in life expectancy.”

Drug-overdose deaths skyrocketed between 2015 and 2017, particularly for adults between ages 25 and 54. The main culprit was fentanyl and other synthetic opioids that became pervasive in illicit drug supplies in the U.S. around that time.
…..
Suicides rose 3.7% in 2017, accelerating an increase in rates since 1999, the CDC said. The gap in deaths by suicide widened starkly between cities and the most rural areas between 1999 and 2017, the data show. The rate is now far higher in rural areas. “There’s a much wider spread,” Dr. Hedegaard said.

But there’s something a bit misleading about capturing these mortality changes by talking about life expectancy (from birth).

I will give you a hint: the “life expectancy” for women in calendar year 1920 was 13 years less than the life expectancy of women actually born in 1920.

I will explain this more in the coming weeks, and show how I know this… and why this difference exists.

I will be posting these explanations here on STUMP, but I will also do the explanations via video on my Meep’s Math Matters YouTube channel. Right now, I’m primarily reviewing books in my Mornings with Meep series, and to separate that from the coming series, that will be called Mortality with Meep. But that will start next week (Happy Holidays!)

Enjoy!