HomeMy WebLinkAbout2009 01 27 Handout Called 'Funding Public Pension Plans' Distributed to Board prior to the Meeting by Chairperson FairDate: January 27, 2009
ATTACHED HANDOUT CALLED
`FUNDING PUBLIC PENSION PLANS'
WAS DISTRIBUTED TO THE BOARD
PRIOR TO THE MEETING BY
CHAIRPERSON FAIR.
''!~
T~~13E~~~~' ~ MARTIN MCCAULAY
Funding Public Pension Plans
OST PUBLIC PENSION PLANS areingoodshape,andtraditionalactuarialmethodsseemtowork
relatively well under many conditions. But they may not be sufficient in all cases, such as when
the funded share falls below 60 percent. To strengthen these plans, there's a need for rules that
require plan sponsors to make progress toward full funding without actually mandating it.
For mature public plans that are less than 60 percent
funded, the amortization payment should be based on a
descending period, with a minimum payment of the in-
terest on the unfunded actuarial accrued liability (UAAL).
As a starting point for discussion, consideration should
be given to requiring an overriding minimum contribu-
tion (OMC) and a recommended contribution that is the
greater of the amount determined using traditional actu-
arial methods and the OMC.
To ensure that public pension plans are more accurately
reviewed for fiscal soundness, I would recommend the fol-
lowing for regulating cost, asset, and amortization:
> The cost method should calculate liabilities on a pro-
jected basis, including pay increases and cost-of-living
adjustments (COLAs). It should calculate benefits on
a projected basis and reflect all plan provisions in the
liability calculations. Unlike projected-unit-credit and
entry-age-normal methods, the unprojected-unit-credit
method doesn't reflect projected benefits in the liability
and isn t a reasonable actuarial cost method for ongoing
pay-related plans with active members. Any method that
excludes benefit features such as deferred retirement op-
tion plans, overtime included in the benefit calculation,
or regularly granted COLAs shouldnt be considered
reasonable. It would be preferable to have liabilities for
contingent COLAs based on investment gains included
in the liabilities for funding purposes, but if that isn t the
case, the liability for contingent COLAs should still be
calculated and disclosed.
> The asset method should minimize the smoothing of in-
vestment gains and losses. I would recommend limiting
the period for smoothing gains and losses to five years
and setting the value range from 80 to 120 percent of
mazket value.
> The amortization method should provide for amortiz-
ingthe UAAL over a closed or fixed period or by a fixed
date. Otherwise, the UAAL grows to infinity and the re-
sponsibilityfor paying it is passed on to future taxpayers.
MARTIN MCCAULAY serves as the deputy executive
director for the Texas State Pension Review Board in Austin
and can be reached at martinmccaulay@hotmail.com.
The point is to make progress on funding the UAAL.
Traditional public plan methods work for closed groups,
but the payroll growth assumption in open groups,
combined with constant resetting of the amortization
period, can lead to perpetual negative amortization in
which payments are never sufficient to pay the inter-
est on the UAAL. This is analogous to a nontraditional
mortgage product in which a borrower makes payments
that aze less than the interest, except that the mortgage
has a property serving as collateral for the debt while the
UAAL has no such underlying asset. Also, most mort-
gages that did allow negative amortization had limits.
Many of these mortgages allowed negative amortization
for only five years or less and had terms to recast the pay-
ment to afully amortizing schedule if the principal grew
to apre-specified amount, such as 110 to 125 percent of
the original balance.
Determining the OMC
Contributing either the annually required contribution
(ARC) or the recommended contribution under traditional
actuazial methods will not necessarily ensure a plan's ac-
tuarial soundness. Guidance from the Governmental
Accounting Standards Board and the Actuarial Standards
Board allows for methods and assumptions that can result
in no progress being made on the amortization of the UAAL
50 Contingencies ~ JAN/FEB.09
rRADECRP.FT
•
0% s FR < 50% NC + BP
50% 5 FR NC + [(1- FR)/FR] BP,
not less than 0
and a deterioration of the plan's funded ratio.
Instead, I would recommend that all contributions be tested
against an OMC calculation (and that this calculation be applied
automatically to all plans that are less than 60 percent funded).
Doing this, plans could avoid unreasonable payroll growth
assumptions and perpetual negative amortization from the ex-
tension or resetting of amortization periods. Because it would use
the mazket value of assets (MVA), it would avoid the distortions
caused by excessive smoothing of investment gains and losses.
An OMC calculation that is independent of amortization and
asset-smoothing methods can also help preserve funded ratios.
The proposed required contribution would be the greater of the
amount determined using traditional methods and the OMC.
For example, the OMC for a public pension plan can be
expressed as a function of the normal cost (NC), the beginning-
of-year funded ratio (FR), and the expected annual current year
benefit payments (BP), as shown in Figure 1.
The OMC is the NC plus the FR factor times the BP. The FR
factor is the ratio of 1 minus the FR divided by the FR, with a
maximum of 1. The NC and expected BP would include interest
adjustments to match the timing of the contributions. The FR
would be based on the ratio of the MVA to the actuarial accrued
liability (AAL).
The formula for the OMC is derived by assuming that when
the FR is less than 100 percent, the rate of asset growth should
equal or exceed the rate of liability growth to avoid an actuarially
unsound scenario. The ratio of the MVA at the end of the yeaz,
MVA,, to the MVA at the beginning of the year, MVAc, must be
greater than the ratio of the AAL at the end of the yeaz, AAL,, to
the AAL at the beginning of the year, AALa
For example, a plan with $2 million in liabilities and a mazket
value of assets of $1 million is 50 percent funded at the beginning
of the year. Suppose the NC is $100,000, the BP are $200,000, the
cash flows aze at the beginning of the year, the interest-rate as-
sumption is 8 percent, and the contributions determined under
traditional methods are $140,000. The FR would be expected to
decline to 49 percent at the end of the yeaz; based on an AAL,
of $2,052,000 and an MVAI of $1,015,200. Using the OMC of
$300,000 (the NC plus the BP for a plan that is 50 percent fund-
ed), the expected FR would be 58 percent, based on an AALI of
$2,052,000 and an MVAI of $1,188,000.
Using the same numbers as in the previous example but with
stazting assets of $1.6 million and contributions, determined un-
dertraditional methods, of $110,000, the FR would be expected to
decline from 80 percent to 79 percent at the end of the yeaz, based
on an AALI of $2,052,000 and an MVAI of $1,630,800. Using the
OMC of $150, 000, the NC plus 25 percent of the BP for a plan
that is 80 percent funded, the expected FR would be 82 percent,
based on al AALI of $2,052,000 and an MVAI of $1,674,000.
This formula was developed to make sure that the contribu-
tion doesnt allow the funded ratio to deteriorate for plans with
unfunded liabilities. As a starting point for discussion, consider-
ation should be given to making the recommended contribution
for funding purposes subject to an OMC calculation.
It's not necessary to tazget 100 percent funding as the ultimate
goal. The UAAL amortization period doesn t need to descend to
zero but could descend for 10 yeazs and then become a rolling
10-year amortization period.
There are several reasons for not funding at the 100 per-
cent level. When a plan is below 100 percent funding, there's
no question about ownership of any excess. The contributions
carry through the plan to cover the last remaining member. Also,
pressure to improve benefits once the funded ratio is 100 percent
or greater can reduce a plan sponsor's interest in increasing the
funding. I would recommend that plan sponsors be required to
make progress toward 100 percent funding, without requiring
them to do so.
Selecting aBest-Estimate Range
Risk is measured using the standard deviation. Actuarial Standard
of Practice No. 27, Selection of Economic Assumptions for Mea-
suringPension Obligations, defines the best-estimate range as the
narrowest range within which the actuary can reasonably antici-
pate that the actual results, compounded over the measurement
period, are most likely to fall. I would recommend determining
the long-term compounded annual rate of return by taking the
expected arithmetic annual return and adjusting it for variance
drain or volatility drag, resulting in a reduction of about one-half
of the variance.
Using the standard deviation of aportfolio, abest-estimate
range of the long-term compounded rate of return would be de-
veloped and expressed as a range between the 25th and 75th
percentiles of the expected results, with the range around the
expected compounded return plus or minus one-tenth of the
standazd deviation.
The compounded return would be lower than the expected
annual return because of volatility. The reasons for the volatil-
ity drag are explained in an article by James D. MacBeth in his
1995 Financial Analysts Journal article "What's the Long-Term
Expected Return to Your Portfolio?" The vaziance drain would be
about half the variance of the portfolio.
To improve on the estimate for the variance drain, a factor of
0.46 could be used instead of one-half. Using the 0.46 factor re-
producesthe results from an exact formula (developed by Olivier
de La Grandville in his 1998 Financial Analysts Journal article
"The Long-Term Expected Rate of Return Setting It Right°) with-
52 Contingencies ~ JAN/FE6.09
TRADECRAFT
~~ ~- * fir.
i
~~
in about two basis points for portfolios with standard deviations
less than 17 percent. Standard deviations that are greater than 17
percent are found in portfolios that are more than 90 percent in-
vested in equities, an uncommon occurrence in pension funds.
Annual Compounded Rate of Return
=Expected Annual Retarn - [0.46 (Variance)]
z
= µP - .46 aP
For example, a sample portfolio of 100 percent equities with an
annual expected rate of return of 10 percent and a standard devia-
tion of 20 percent would have variance drain of about 200 basis
points, and the compounded expected return would be about 8.2
percent before expenses.
The best-estimate range of the expected return is the range
in which the results are more likely to fall. This is the range from
the 25th to 75th percentile, or the inter-quartile range. The inter-
~ -~fi- _r4
F~.~~ r O ~~~ r 3 1f ea rs
o e
~ Serving as. The
., ~ Confidential Intermediary
~~.~~: '
Mid America. Search®
www.midamericasearch.com
~~9 <~
JP
~:"
quartile range would be constructed using the long-term expected
standard deviation, which will decrease over time. The inter-
quartilerange for the 50-year compounded rate of return would be
the return plus or minus the standard deviation, divided by 10.
A formula exists for the amount of reduction in the standard
deviation over time. After 50 trials, the standard deviation will
be reduced by dividing the original standard deviation by 50%.
Using Chebyshev's theorem, the half of the results that fall in the
inter-quartile range must be within 1/(2'r~) standard deviations
of the mean. The formula for the corridor for the inter-quartile
range after 50 years is 1 divided by the product of 50'~ times 2f,,
or 1 divided by 100r~, 100'h equals 10.
Based on the asset mix, the expected compounded-return
assumption before expenses can be developed by taking the ex-
pected annual return and subtracting about one-half, or 0.46, of
the variance. The inter-quartile range can be developed by using
a corridor within one-tenth of a standard deviation around the
expected compounded return.
Best-Estimate Range of Annual Compounded Rate of
Return
= Expected Annual Return - [Variance/2] t [Standard
Deviation/10]
For example, for a sample portfolio with 60 percent in equi-
ties and 40 percent in fixed-income investments, with an annual
expected return of 8 percent, a standard deviation of 12 percent,
and expenses of 30 basis points, the net expected compounded
return is in the 5.8 to 8.2 percent range, with a best estimate of
about 7 percent.
To sum up, we should consider making the recommended
contribution for funding public plans subject to an OMC calcu-
lation. When determining the actuarial value of assets, I would
recommend limiting the period for smoothing gains and losses.
I would also recommend limiting the corridor or range around
the market value to prevent excessive smoothing from distort-
ing the results. Regarding actuarial cost methods, all of the plan
provisions should be reflected in the liabilities. I believe that the
OMC would be sufficient to keep plans progressing toward full
funding. ~
This article is solely the opinion of its author. It does not express
the official policy of the American Academy of Actuaries;
nor does it necessarily reflect the opinions of the Academy's
individual officers, members, or staff.
54 Contingencies ~ JAN/FE6.09