Milliman on Mortality: Hope You Die

The Milliman report criticizing assumptions used by Detroit’s erstwhile actuaries Gabriel Roeder Smith & Company (GRS), whose reports showed Detroit’s plans to be among the healthier public plans in the nation, kicked off with a scathing indictment of mortality assumptions used by GRS which Milliman characterizes as too ‘optimistic’.

You tell a layman that mortality assumptions are ‘optimistic’ and the assumption would be that people are living longer but in the perverted world of public pension funding an optimistic assumption (and they’re all optimistic) is one that lowers costs and having retirees die sooner is good news.  Having all 20,000 Detroit retirees contract the plague would, in the world of big benefits and low costs, be cause for unbridled celebration by all concerned (except of course the 20,000 decedents).

Quoting from the Milliman report:

While we have not received experience data to perform any tabulation, the assumptions indicate to us that the mortality rates of Detroit public employees are significantly worse than national averages.  While it is possible that the current assumptions are appropriate, as a VRPG for item D in the table above, the liability has been increased by 10% as an adjustment to reflect unbiased mortality rates.

In general for all actuarial assumptions, the more optimistic the current assumption is, the higher the likelihood that the assumption will not be met, leading to higher costs later as actuarial losses accumulate.

The mortality assumption used by GRS in their June 30, 2011 General Retirement System report (page E-1) that Milliman found so objectionable:

The mortality table used to measure retired life mortality was 110% of the RP-2000 Combined Table for males and 110% of the RP-2000 Combined Table set back 2 years for females. These tables provide a margin for mortality improvements of approximately 15%.

The mortality assumption used by Milliman in their June 30, 2012 valuation of the New Jersey Teachers’ Pension and Annuity Fund (page 61):

Rates of mortality vary by age, gender and type of retirement. A generational approach is applied using Scale AA to apply for future mortality improvement for non-disabled annuitants. The base year is 2000 for males and 2003 for females.

Notice much difference?  What if you do a VRPG of the retiree populations in each report.

In the Detroit General Retirement System there are 11,555 retirees getting $219 million in annual payouts which GRS values at $1.984 billion which comes out to an average annuity factor of 9.06.

In the New Jersey Teachers Pension and Annuity Fund there are 89,308 retirees getting $3.46 billion which Milliman values at $32.4 billion which comes out to an average annuity factor of 9.36.

Bumping up that GRS factor by 10% would bring it to 9.97, far above what Milliman actuaries are assuming for at least one public plan in their regular day jobs.

13 responses to this post.

  1. Posted by Anonymous on August 8, 2013 at 9:35 am

    Hey John, isnt it really a race to see which government fails first, city, state or federal. I know that neither you nor I want to face the fact that the federal is close to collapsing completely.. That is difficult to face.


  2. John, I’m not and actuary so I don’t know what an VRPG is and I don’t understand what the mortality factor means. Could you explain these for interested readers outside the field? Many thanks


    • AS far as I know VRPG makes it’s first appearance in this Milliman report (might even be patented). There isn’t even a real-world acronym:

      What it seems to mean (and it could well catch on after this) is a CYA move for when you look at a number and decide it’s probably wrong in relation to other numbers based on your gut instinct (no calculations involved). In the Milliman report it’s defined as “Very Rough Preliminary Guesstinmate”

      The Mortality assumption is simply a statistical estimate of the number of deaths expected to occur over a given period of time.

      A pension annuity factor is the present value of a cash flow that produces $1 in income over so many periods of time, generally for life. A mortality assumption is an important aspect
      of an annuity factor since the sooner you die the lower that annuity factor is. Though the interest rate is usually a more important factor – that is, the more money you expect to make on trust earnings the lower the annuity factor also needs to be which is the more visible debate these days where GASB is wanting that interest rate to be 4.5% while the public plan community is comfortable with their 8%.


  3. Posted by Tough Love on August 8, 2013 at 11:20 am

    I’m sure Millman would say that the mortality assumptions they used in their NJ Valuation is ……. within the range of reasonable assumptions.

    I’m also sure they would say the same thing with respect to their work in reviewing GRS’s assumptions in Detroit’s pension Plan valuations.

    With the 10% bump, they differ by about 6.5% (9.36% vs 9.97%). In the world of mortality assumptions and for groups this large (and similar), they would NOT reasonably be able to justify this differential, which suggests that it must be influenced by the “desired” results of their employers.


  4. Posted by Robert Mitchell on August 8, 2013 at 3:23 pm

    A mortality study for Detroit retirees should consider the actual to expected ratios for deaths compared to the tables that appear in the valuations. Consider also that census records have a racial component, and minority populations generally have lower life expectancy rates, as do poorer vs wealthier populations. Without an actual experience study, this truly is a WAG, the red-neck version of VRPG.


  5. Posted by Robert Mitchell on August 8, 2013 at 3:23 pm

    I meant VPRG… oops.


    • No you were right the first time. It is VRPG – Very Rough Preliminary Guesstimate – which doesn’t spell something you can say in one syllable which may be the only reason it wouldn’t catch on. Perhaps with a little more thought they might have had something.

      You have to keep ‘Rough’ and ‘Guesstimate’ since they’re the core but ‘Very’ and ‘Preliminary’ can be replaced without changing much of the meaning. I would suggest ‘Fantastically’ and ‘Initial’, respectively and you’re in business.


  6. Posted by John Langhans on August 8, 2013 at 3:40 pm

    You cannot draw any conclusions from a comparison of the 9.06 and 9.36 average annuity factors because the Detroit and NJ retires have different age and benefit distributions.


  7. Posted by Pension Actuary on August 8, 2013 at 3:45 pm

    You are assuming that Detroit retirees and NJ retirees have the same average age. Do they? Did you check that out?


    • Good point but even if they were the same average age there’s also the issues of newer retirees with higher payouts, impacts of COLAs, and forms of benefit.
      This, after all, is only a FRIG and the point is that those assumptions and methodologies used by Milliman in NJ and GRS in Detroit aren’t that different and Milliman would be criticizing their own NJ reports if anyone ever hired them to do so.


    • Posted by Tough Love on August 8, 2013 at 4:18 pm

      Yes, very good point. Life annuity factors in the low 9s corresponds to a (unisex) attained age in the low 70s, and (at that age), each 1 year of difference in age changes the annuity factor by about 0.35.

      An difference in NJ vs Detroit early retirement options or programs could also sway the attained ages of the 2 retiree groups.


  8. Posted by eatingdogfood on August 8, 2013 at 11:05 pm

    To All Public Servants And Retirees; Repeat After Me : ” EVERYBODY GETS NOTHING “! Repeat 1000 times! Now, Doesn’t That Feel Better!


    • Posted by bpaterson on August 9, 2013 at 10:35 am

      cruel, just cruel……………….many of them are low paid, hard workers with their head in the sand and were taken advantage of just like the voters who elected these public officials who fleeced us all.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: