Contingencies Online - July/August 2011

When a company has major pension obligations,
even this seemingly minor proposed shift in assumptions can affect
its financial outlook noticeably.
deaths assumed to trigger spousal sur-
vivor benefits. The longevity of widows
is somewhat reduced, but to a much
lesser extent than for males who lose
their spouses.
fine-tuning the Model
Since my proposed approach implic-
itly breaks future mortality rates down
by marital status, I was concerned that
the longevity adjustment could conflict
with funding rules that specify future
mortality rates.
I posed the question to the IRS, and
it responded in Question 12 of the 2009
Gray Book. In its response, the IRS dis-
couraged reflecting shorter life spans
for widow(er)s even implicitly when
making calculations under the Pension
Protection Act of 2006. As a result, my
suggested assumption set includes a spe-
cial percentage for accounting.
I don’t believe that it’s worthwhile
(or fully appropriate) to include a lon-
gevity adjustment for people for whom
survivor benefits already have been
triggered. This approach would tend to
make results slightly conservative. This
still prompts the question: Are other ad-
justments appropriate?
I also don’t propose any change in
mortality rates for benefits payable due
to turnover, disability, or retirement. Val-
uation results for those benefits would
be unaffected. Keep in mind that some
current participants can be expected to
be widow(er)s already, so mortality ex-
perience for all marital states should be
reflected. The overall mortality rates in
standard tables might reflect some non-
working widow(er)s, but because of this
group’s age characteristics, that proba-
bly leads to only a slight understatement
of liabilities for benefits triggered by
decrements other than death. In
addition, many participants cur-
rently married are implicitly and
appropriately expected to become
nonworking widow(er)s at some point.
Most of the co-annuitants for benefits
paid in the joint and contingent form
of payment are married, so there might
be a slight understatement of their li-
abilities versus applying mortality rates
adjusted for marital status. But since a
higher-than-normal proportion of oth-
er retirees probably is unmarried, that
group most likely is overvalued. Given
the seemingly microscopic net bias (of
unknown sign), I wouldn’t propose any
further refinement.

A Call to Action
I approached a few of the largest pen-
sion plans in this country to review my
assumptions. None chose to perform an
analysis of their preretirement deaths.
When a company has major pension
obligations, even this seemingly minor
proposed shift in assumptions can af-
fect its financial outlook noticeably. I
looked at the actuarial assumptions for
one major employer with a strong com-
mitment to traditional pension plans,
for example, and estimated that usage
of the modernized assumptions would
increase its price per share by about 1
percent. That’s based on a current price-
earnings ratio that is roughly half of
the historical average of approximate-
ly 16, so 2 percent might be the better
long-term measurement. Its sharehold-
er equity in its 10-K filing would have
improved by about 6 percent.

TOM SCHRYER is an actuary with
Findley Davies Inc. in Cleveland. He can be
reached at tschryer@findleydavies.com.
MCININCH / DREAMSTIME
JUL | AUG. 11 CONTINGENCIES 67