STUMP » Articles » UPDATES and REPOST: Public Pension Watch: How Important is the Mortality Assumption? And Other Assumptions? » 26 September 2014, 17:16

[originally posted on 21 Sept 2014, I have corrections/updates below]

Okay, before I get into this:

I am not a pension actuary. My experience is with life insurance, annuities, and reinsurance.

The way I got into the whole pensions brou-ha-ha in the first place is that, as someone who worked in retirement annuities at TIAA-CREF, I realized how … generously… public pensions were valued compared to annuities. I was annoyed. To understate the situation.

But let me get to the reason for this post — Instapundit linked to a post on an out-of-date mortality assumption:

BOOKS, COOKED: Pension Funds Using 45-year-old Mortality Table To Make Obligations Look Smaller.

Ugh. I want to be fair. Let me link to the post:

If you were obligated to pay pensioners for their entire lives, would you estimate how much you would need using life expectancies from the 1960s for people born between 1914 and 1918?

Jean Lotus, a reporter at the Forest Park Review, just published a splendid article about a few Illinois municipalities recently doing just that. Outdated mortality tables are one way for governments to hide their pension problems from voters.

For more than a decade up to a couple years ago, according to the article, the actuary for Forest Park and some other cities,

“had been using a group annuity mortality table called the GAM-1971. As its name implies the table was created in 1971 using mortality data from police officers and firefighters collected between 1964 and 1968. Life expectancies on the tables tracked public safety workers who, at age 50, would have been born between 1914 and 1918.”

I need to disclose something. I have written with some of the people involved with respect to the use of this mortality table. I gave some advice back then. It is continuing, obviously.

Anyway, the mortality assumption being used, GAM-1971, is pretty off from just general mortality rates and life expectancies, forget about life expectancy for a working population (the general population tends to have a worse mortality, because those who don’t work are usually due to disability, poor health, or other things associated with low life expectancy. Yes, there are some rich people in the general population who don’t work, but they are vastly outnumbered by those with worse mortality.)

Let’s go to to the article: (link fixed)

UPDATE: CREDIT — this graph comes from Foster & Foster Dept. of Insurance Actuarial Experience Study. It is a comparison of the GAM-1971 table against actual mortality from recent years.

[original: I believe that’s a comparison of GAM-1971 against RP-2000. I will do a life expectancy comparison in a later post. UPDATE: I was wrong about that, and I will definitely have a few more follow-ups to this post.]

One of the other people involved in the matter, Tia Goss Sawhney, comments in the post:

Tia Goss Sawhney September 11th, 2014 at 7:05 pm

The use of GAM 1971 GAM is not award-winning for Illinois actuarial work. See page 82 of the 2013 Comprehensive Annual Financial Report of the Metropolitan Water Reclamation District (MWRD) — http://www.mwrdrf.org/media/1048/cafr2013_v1.pdf. The disability “rates are from the Hunters Disability Rates Transactions of the Actuarial Society of America, vol XII
pp. 44-71″. That sounds reasonable enough until one realizes that the Actuarial Society of America no longer exists and last published its Transactions in 1949 — https://www.soa.org/about/history/about-library.aspx. While I have been unable to determine exactly how old these rates are, they are for sure, at least 65 years old! The actuary is Jason Franken of Foster & Foster.

The ASA was a precursor to the org that both Sawhney and I are members of: the Society of Actuaries.

But here’s the deal: that might not make a huge difference… because the higher the discount rate, the less other assumptions matter. And with respect to disability — the old rates chosen may be more conservative (in that it makes the liability look bigger) than current, but I don’t know. I assume they had higher disability rates back in the day… but then you didn’t have to worry as much about paying a disability pension, because they also died sooner.

You see the problem — biasing it one way or another can confound the results in different ways, and not in the direction you expect.

A lot of these assumptions interact with each other, which is why it’s safer to try to use an assumption set as close to expected as possible, and then add any margin for comfort at the very end.

Still, while using GAM-1971 for a 2014 population is questionable, I do want to address one of the issues: that we’re using people long dead to develop current mortality assumptions and that means we have to be way off from reality.

In these various actuarial tables, not only for pension use, but also life insurance and annuities, we don’t just leave the base rates be. We have mortality improvement scales that are applied to old data to get the info more up to date. And in projecting into the future, if we’re looking at products (like annuities) with longevity risk, we project further improvement. The mortality rates we use differ based on whether it’s for annuities, life insurance, pensions, workers compensation, or many other items. Usually a margin is put in for comfort — if you’re doing a valuation for a private company or insurance products, that is. Insurers are required to sock away more than the value of the expected promises.

Pensions are not.

I don’t know what mortality assumptions are used in public pensions in terms of what is most often used. Insurance companies do not have a choice as to the mortality tables (currently… but even under regulatory change being considered, any change of table would have to be backed up by lots of data.)

With regards to discount rates, and perhaps other assumptions, the actuary valuing public pension liabilities is often given them from their public pension fund clients. In some cases, the assumptions (usually discount rate) are written into state law. The actuary doesn’t have a choice for the basis of valuation.

But I believe that few plans dictate the choice of mortality assumption. I have a bad feeling that someone set up their valuation system once, and have never updated their code. For decades.

We are given absolute amounts by which the valuations changed due to a change in mortality assumption:

Forest Park’s police and fire pension systems took a hit this year because a simple actuarial change recalculated how long safety personnel can be expected to live. Actuary Timothy W. Sharpe, of west suburban Geneva, changed one element of his calculations last year, revealing a $104,000 shortfall in the police pension fund and a $94,000 shortfall in the fire pension fund.
…..
When Sharpe changed to the updated mortality tables for the municipalities he worked for, tax bumps took place all over the state. Aurora’s police and fire pension funds discovered a $3 million shortfall. Rockford’s gap was $1.1 million. Highland Park discovered an $800,000 shortfall. Streamwood’s was $300,000 and Oswego’s was $200,000. River Forest, LaGrange and Western Springs all replaced Sharpe with another actuary.

Problem is, I can’t tell how material these changes are for the individual plans.

I mean, I could start digging into the Comprehensive Annual Financial Reports, but I am not getting paid to do this. Just because one number is bigger than another up there does not mean it’s more material for the plan in question. These cover different numbers of employees, just as a starter.

But here’s the bottom line: pretty much every other sphere of actuarial work outside of public pensions and state-run insurance, etc., there is substantial oversight because it’s understood any of these systems can fail.

Except for public pensions.

There is essentially no oversight for them, except on the edges, when entities like the SEC cannot ignore that some really bad things are going on (and even then, it doesn’t fix the pension being underfunded. They just ask that info being disclosed when you issue new state bonds.)

As I have written before, by assuming that public pensions can’t fail, people make public pensions even more frail.

As with public pensions chasing returns, risking much bigger losses, being loose with valuation because hey, taxpayers will always top it off, it may make taxpayers even more pissed off to think they were lied to for decades as to the real costs of the pensions being promised.

If, from the beginning, there was the consciousness that the pensions could fail, perhaps the trustees and the actuaries would have been a little more careful about the valuations.

And maybe we wouldn’t be in this current mess.

Compilation of Illinois posts