### Teachers Appreciation Week: Public Teacher Mortality

by meep

Okay, before I begin. A few videos I did on my favorite teachers:

- On Dr. John Kolena, my high school physics teacher
- Mr. Dan Teague, who taught me about applied math
- Dr. Sue Collins who gave me and told me about opportunities in math

Those are all positive. In case you want to escape from the following.

I’m going to be actuary-ing all over the place below, because it’s Mother’s Day, and dangit, I’m indulging myself.

**ALL TEACHERS MUST DIE**

(eventually)

In my prior post on teachers pensions, I mentioned that teachers tend to be longer-lived than many others who have defined benefit pensions, which is why their pensions are more expensive for the same nominal $ amount of benefits: they simply live longer.

To the extent that teachers’ longevity is reflected in valuation assumptions, you will see this come out in terms of what the various costs are — whether normal cost or the actuarially “required” contribution, which covers the normal cost and an amortized amount of the unfunded liability.

Of course, if the longevity is low-balled compared to the actual experience coming out, you’ll see this in rapidly increasing contribution requirements. To be sure, the discount rate is an even bigger factor here, but if both mortality and discount rate (and payroll increase rate and…) are off, it really starts to bite. But we’ll get into that in the next post.

So let’s start digging into what the Society of Actuaries has found in preliminary research on public pension mortality.

**TEACHERS MORTALITY VS OTHER GENERAL DB PENSION MORTALITY**

The Society of Actuaries published Pub-2010 Public Retirement Plans Mortality Tables was released in 2018 and then some revisions were made and it was republished in February 2019.

The full report is here, and I’ll excerpt the executive summary:

1.1 Purpose

As a result of comments received on the prior RP-2014 Mortality Tables (RP-2014) study, which included

only data from private pension plans, the Society of Actuaries (SOA) and the Retirement Plans Experience

Committee (RPEC or “the Committee”) initiated a mortality study of public pension plans in January 2015.

The primary focus of this study was a comprehensive review of recent mortality experience of public

retirement plans in the United States. The objectives of this study were the following:1. Develop mortality tables based exclusively on public-sector pension plan experience.

2. Provide new insights into the composition of gender-specific pension mortality by factors such as

job category (e.g., Teachers, Public Safety, General), salary/benefit amount, health status (i.e.,

healthy or disabled), geographic region and duration since event.….

All of the deferred annuity values shown in the following tables were developed using amount-weighted

mortality rates, a pre-retirement discount rate of 7.0% and a post-retirement discount rate of 5.0%. The

7.0% rate was chosen to be broadly representative of discount rates recently used in the funding

valuations of public-sector retirement plans, and the “spread” of 2.0% broadly representative of recent

post-retirement cost-of-living adjustments.

Here is the comparison table specifically for teachers:

I will try to keep this simple so that non-actuaries can understand. The annuity factor, on a monthly deferred-to-62 basis, means that if you’re projecting that at age 62 one gets, say, $60,000 per year in monthly payments (assuming asset returns of 7% per year, and annual COLAs of about 2% per year), the actuarial present value of that benefit equals $60,000 times the annuity factor at the current age. (If you’re over age 62, you assume that benefits are already being paid)

So, if we use the Pub-2010 tables for teachers, and a female teacher is currently 55 years old, the actuarial present value of $60,000 per year at age 62 is 9.1655*$60,000 = $549,930.

The annuity factors are the real thing one should compare, not life expectancies nor mortality rate ratios.

Note on the right side of the table, comparisons are made against fairly common mortality assumptions. For that age 55 female teacher, the Pub-2010 assumptions give a cost ranging from 4% to 7% higher than standard assumptions.

Back to the summary:

The amount-weighted deferred annuity values for Teachers are consistently larger than those for Public

Safety and General, and, in fact, they are considerably larger than even those developed using the White

Collar version of the projected RP-2006 table.

…..

Multivariate analysis indicated that salary (for Employees) and benefit amount (for nondisabled

Annuitants) were the most statistically significant predictors of mortality differences within individual

gender/job classifications. As a result, the Committee produced Above-Median and Below-Median

versions of the Employee, Retiree and Contingent Survivor tables. In general, the impact of moving from

the total dataset table to either the Above- or Below-Median tables is considerably smaller for Teachers

than for Public Safety or General, and the impact for males in each of the three job categories is

considerably larger than that for females.

Can’t say I’m surprised about the salary effect for males and non-effect for females. That’s for another time, though.

Here’s just mortality ratios for female and male teachers against the standard assumptions — ratios below 1.0 mean that the mortality rate for teachers at those ages is lower than that from the other DB mortality tables.

These are retiree mortality rates compared. There are also graphs for active employees, and similarly, we see ratios well below 1.

Mind you, for all the ages below 70 or so, mortality rates are very low. So even if the new tables show even lower mortality, it doesn’t necessarily have a huge effect.

Again, why we see increases in pension values in the single digit percentage range.

Of course, if the “true” discount rates are a lot less than 7%… then the bite would be a lot bigger.

**WHY NO COMBINED TABLES?**

Before I go into this, there are already defined benefit mortality tables out there. They were developed on private pensions, and many times single employer pension plans. They do have breakouts by white collar/blue collar/intermediate.

This is what the SOA had to say in their own report:

13.1 Rationale for No “Combined” Public Table

The Committee did consider publishing a “combined” public plans table that included all of the data

received for the study from each of the three job categories. Ultimately, it was decided that this would

not be done. In addition to the statistically significant differences in mortality by job category discussed in

Section 4, it was determined that “combined” rates at various ages were often more reflective of the

relative concentrations of the component subpopulations (Teachers, Safety, General) than of underlying

mortality characteristics. The covered populations in many public retirement plans have demographic

characteristics (including job category) that are quite different than that of the population that would

have been used to develop any combined Pub-2010 table. Therefore, it would be better for the actuary

with knowledge of the specific member demographics either to segregate the populations and use

appropriate tables for each, or to construct a custom combined table using appropriate weighted

averages of the job category and Above- or Below-Median rates from this study.Last, many public-sector retirement programs specifically cover Teachers, Safety or General employee

populations. Even those that cover multiple populations often provide different benefit features and track

census data separately by job categories.

I looked into the report to see the comparison by job type… there was the data, but not a nice graph. So I made my own, using the data from Appendix D, tables D.4 which gives life expectancies from specific ages.

Most people find life expectancy from age X confusing, so I’m going to graph expected age at death given current age.

Females:

Males:

So some things to note: teachers, for both females and males, the life expectancy is higher by a year or two compared to the other job categories for most working ages, and the difference gets smaller in retirement ages (which one would expect).

There’s not much difference between safety and general male public employees mortality, and there is a bigger difference for females.

In any case, teachers are a fairly long-lived bunch.

Therefore, teachers pensions at the same benefit levels are going to be more expensive (in experience) than general public employees and safety officers.

**BETTER MORTALITY REFLECTED IN FEW PLANS**

Now, I have mentioned this is a new set of mortality tables, but that doesn’t mean the difference in mortality wasn’t already recognized by actuaries.

For smaller plans, they’d likely use a standard table and perhaps adjust it to allow higher life expectancy.

But, luckily, I don’t have to investigate this myself. The SOA already did.

U.S. Public Pension Plan Mortality Assumptions

Compared to Pub-2010 Mortality Tables

Here are comparisons of annuity factors — yes, under a variety of discount rate/COLA assumptions. But they’re all being compared on the same discount rate/COLA basis.

The black lines are the annuity factors using the Pub-2010 tables.

That most of the data points are under the black lines means that if the plans switch to these updated mortality tables, the recognized cost on the balance sheet will be higher (again, single-digit percentages higher). The higher the discount rate, the less this change matters.

Of course, many plans are having to lower their discount rates now to reflect current reality.

**PROBLEM TO BE FIXED?**

The only problem to be fixed, given this new information, is to check that mortality assumptions used for valuations make sense for public plans. In the above comparison, it’s not just teachers plans potentially low-balling longevity, it’s also general public employees and public safety plans. The largest plans, as in California or Texas, will likely have good enough credibility to develop their own tables. That’s fine.

But for the smallest plans? They really need to pick one of the standard tables.

There was an actuary who was way off compared to population averages, which was known by lots of folks. That’s partly why he was hired, after all — compared to other actuaries, he developed lower cost requirements, because he had mortality assumptions better suited to a population from 20+ years earlier, and not now. He was suspended by the American Academy of Actuaries for two years effective August 2018, the first I’ve seen for using inappropriate mortality assumptions in public plans.

I’m sure others have taken that lesson. His particular assumptions were WAY off, as opposed to using one of the current private plan mortality tables, which is off by single-digit percentages for annuity factors.

Still.

It’s good to see the professional actuarial organizations stepping up and investigating what is going on in public pensions, just as they have for private pensions (which have federal oversight… and public pensions have essentially none.)

Either way, teachers pensions, all else being equal, are more expensive than safety officer and general public employee pensions. And to the extent that actuaries have been reflecting it, various governments have often short-changed these plans by deliberately undercontributing to these plans.

I suppose they think the “It’s for the children!” argument will hold. For retired teachers. Who are no longer teaching. And cannot strike. And many of the teachers pensions are for relatively modest amounts, given their base salaries generally being lower than late career safety officers, and given that they don’t do much overtime to spike their pensions.

But if the teachers live longer, and retire about the same ages as other public employees, then the COLAs will bite and they can still end up costing more than other employee categories. Also, they outnumber all the other categories in many states and localities.

This upcoming week, I’ll be looking at what’s going on in particular states — Oregon, Kentucky, and more. In most of these cases, teachers pensions are a huge portion of the pension problem.

I have a lot of sympathy for all involved, but if we are to really solve these issues, we need to reflect the reality of what’s going on in these plans.

Related Posts

Geeking Out: On Life Expectancy and Conditional Expectation

Mortality Monday: Broken Heart Syndrome

Memory Monday: The Spanish Flu Peak in October 1918 and More