Financial Simulation of the Oregon PERS Plan

1
Financial Simulation of theOregon PERS Plan
Version 1-11/21/02Subject to Revision

• Summary Version

Jim VoytkoExecutive DirectorOregon PERS
2
An Investigation, Not a Declaration
A Work in Progress
3
Project Objectives

• Using
• Actual System Data
• Current Actuarial Assumptions
• Explicitly Defined Policy Choices
• to dynamically model the complete OPERS
  financial structure on a system-wide, aggregate
  basis to
• Forecast the most probable future trends in
  financial outcomes
• Assess the financial consequences of alternative
• Policies
• Environments
• Structural Plan Features
• Participant Behavior

4
The Simulation Project Team
5
Model Overview
Investment Return Futures
OPERS Plan Rules OPERS Environment
Financial Outcomes
6
Model Overview

• Assumes
• Sequential Crediting
• Estimated 12/31/2002 Ending Balances
• Current Actuarial Assumptions (as of summer 2002)
• Actual PERS Membership and Employer Data
• Current Statutory and PERS OAR Provisions
• Assumed future OAR revisions to accommodate
  sequential crediting

7
Model Overview

• Full output and documentation available on
  request
• If Board so directs, can be run with stakeholder
  inputs on a resource/time available basis
• Model cannot be provided to parties lacking
• Software licenses issued by Palisade, Inc. and,
• Simulation hardware (each run currently takes
  40-48 hours of computation time)

Subject to PERS OARs regarding public records
requests
8
The Base Case
9
What is the Base Case?

• Four Categories of Inputs
• Environment
• Investment
• Economic
• Participant
• Rule Set
• Statutory
• OAR
• Actuarial

• Policy
• Reserve Funding Policy
• Reserve Goals
• Assumed Rate
• Mortality Policy
• Sequential Crediting
• Membership Behavior
• Variable Exposure
• Years of Membership
• Retirement Rates
• Propensity to Refund
• Use of One-Time Variable Transfer
• Deactivation Rates

10
Base Case

• Environment
• Investment
• Long Run Portfolio Return Frank Russell Co.
• Volatility Forecast for OPERS Fund Frank
  Russell Co.
• Correlations Frank Russell Co.
• Economic
• Inflation of 3.25 - Milliman, USA
• Wage Growth of 4.25 - Milliman, USA
• Participant
• Tier 1 Tier 2 Retirement Rates Milliman, USA
• Membership Composition PERS Database

11
Base Case

• Rule Set
• Current statutory provisions and OARs
• Sequential Crediting
• Current Actuarial Assumptions (e.g. amortization
  methods)

12
Base Case

• Policy
• Board 30-month reserve goal for pre-funding Tier
  1 Member Guarantee shortfalls
• All of Tier 1 Member earnings in excess of the
  assumed/guaranteed rate immediately available to
  fund Tier 1 Member Rate Guarantee reserve (after
  higher priority deductions and if necessary)
• Assumed Rate 8 Tier 1 Member Guaranteed Rate
• Mortality Policy Current

13
Base Case

• Member Behavior
• Variable participation by active members
  consistent with past trends and dynamically
  linked to stochastic investment returns
• All members execute one-time conversion of
  variable earnings to regular at retirement
  (assumes rational self interest)
• Retirement/Deactivation/Refund Probability
  consistent with actuarial assumptions

14
Base Case

• Focus on Sequential Crediting
• Tier 1 Rate Guarantee Reserve Deficit
  Reserve Structure
• Only Tier 1 Member Accounts Receive Annual
  Guarantee
• Tier 1 Member account excess earnings tapped
  first for deficit account elimination and Rate
  Guarantee reserve funding
• All other excess earnings (excluding Tier 2,
  Variable, ER Lump Sum) tapped for deficit account
  elimination if Tier 1 member excess earnings are
  insufficient

15
And, What is a Conclusion?

• Deterministic
• If X, then Y.
• Stochastic
• Given X, the distribution of possible values for
  Y is
• 5
• 10
• 15
• 95
• Stochastic Forecast Best Estimate of Future
  Outcomes

16
Ten Notable Conclusions from the Base Case
17
Conclusion 1-1

• There is a low probability that Tier 1 members
  can be credited with earnings in excess of the
  assumed rate over the next 10-15 years.
• Probability of at least 1 Tier 1 earnings
  crediting 8
• 2002-2006 15
• 2007-2011 45
• 2012-2016 50
• However, the likelihood of a call eliminating the
  Tier 1 deficit followed by the aggressive
  reserving policy assumed in the forecast allows
  that probability to rise for very young Tier 1
  members expected to retire after 2016.

18
Conclusion 1-2

• Tier 1 Replacement ratios are highly likely to
  rise dramatically from their current levels over
  the next 15 years, then begin to decline.
• Mean/Median Replacement Ratios3
• Actual1 (Tier 1 2 combined)
• 1990-1995 44.83
• 1996-2002 58.96
• Expected2 (Tier 1)
• 2002-2006 55.9
• 2007-2011 71.6
• 2012-2016 94.0
• 2017-2021 124.2

1 From an analysis of PERS Replacement Ratios
8/29/2002
2 50th percentile of expected mean. Expected
ratios are consistent with but not strictly
comparable to 8/29/02 study mean values
3 Does not include health insurance subsidy,
social security or PERS COLA
19
Conclusion 1-3

• Employer rates for PERS coverage (before pickup)
  are highly likely to rise sharply over the next
  5-10 years and probably will remain at higher
  than historic levels for the next 30 years.

Employer Rate - of Payroll(Before Pickup)
15.56
1990-2002 Average9.73
20
Conclusion 1-4

• There is a material chance of a call, a
  considerably smaller chance of a second one, and
  the probability of any call declines sharply
  after 15 years.

• Likely Magnitude of a Call Between 1.5 Billion
  and 2.5 Billion (PV)

PV present value in todays dollars
21
Conclusion 1-5

• The systems UAL is highly likely to rise sharply
  in the next 5 years and decline slowly over the
  next 20-25 years
• Rates may need to be kept in the 16 to 19 range
  after this period to keep the UAL stable in
  nominal terms (and declining as a percentage of
  total plan assets)

90
System UAL(Billions of s)
System Funded Ratio
80
70
60
50
22
Conclusion 1-6

• The relative contribution funding burden (ratio
  of employer contributions to employee
  contributions) rises steadily over the next 10
  years from approximately 3.4 to one to roughly
  5.0 to one, then declines over the next 25 years.
• The ratio of employer contributions to employee
  contributions should trend toward approximately
  2.7-3.1 to one sometime after 35 years.

2.36
23
Conclusion 1-7

• An acceleration of Tier 1 member retirements or
  terminations while the system has a Tier 1
  deficit account balance raises the probability of
  a call and may shift the periodic cost of
  deficit elimination more toward employers.
• Slower than anticipated Tier 1 Member retirements
  allow Tier 1 account balances to grow longer thus
  enlarging the costs of Tier 1, both at time of
  retirement and in the 15-20 years which follow.
  These longer lived Tier 1 accounts, however,
  retain assets whose excess earnings can continue
  to contribute to deficit elimination.

24
Conclusion 1-8

• A continuous stream of annual 8 investment
  returns is not sufficient to
• eliminate the likelihood of employer rates
  reaching at least 22-24 before pickup, nor
• eliminate a possible call in excess of 2
  billion, nor
• prevent a persistent but small UAL, even beyond
  the Tier 1 era.

25
Conclusion 1-9

• At this point in time, a less aggressive Board
  policy for funding the Tier 1 Rate Guarantee
  reserve has almost no effect on the probability
  of multiple system calls, but slightly
  increases their size.

26
Conclusion 1-10

• Misestimating the assumed rate has significant
  consequences for employer costs

1 ARAssumed Rate2 CMRCapital Market Returns
27
What Does the Simulation Suggest About Tier 2?
28
Conclusion 2-1

• Tier 2 Becomes Important With Respect to
• Recruitment 100 controlling as of 1997
• PERS Membership 55 of PERS Members are Tier 2
  as of 2005
• Pre-Retirement Assets 55 of member non-retiree
  assets are Tier 2 as of 2018

29
Conclusion 2-2

• Tier 2 is not Full Formula It is Money Match
  Driven under the Base Case Assumptions.
• However, there will always be some members who
  get Full Formula based benefits
• Specifically Tier 2 produces average
  replacement ratios approximately 10 percentage
  points higher than the Full Formula..
• Why?
• An 8 investment return coupled with the expected
  duration of member service consistently produced
  sufficient account growth to ensure Money-Match
  based benefits exceed the Full Formula for most
  members.

30
Conclusion 2-3

• If Tier 2 benefits do become predominately Full
  Formula driven in the period 2020-2040, employer
  rates will be already beginning a new and
  substantial rise beyond that forecast in the base
  case.
• Why?
• While the bulk of members retiring during this
  period are Tier 2, the system will also be
  financing a considerably larger Tier 1 driven BIF
  whose fixed financial demands rest on an assumed
  rate of 8. If investment returns have been low
  enough during 2002-2020 to kick in Full Formula
  as the dominant benefit and effective benefit
  floor, they will also have been too low to
  finance the then much larger BIF without even
  higher employer contributions

31
Conclusion 2-4

• Tier 2 is less costly than Tier 1

• Base Case Tier 2 Mean Replacement Ratio
  Equilibrium
• Approximately 50 - 52 of FAS2

1 From an analysis of PERS Replacement Ratios
8/29/2002
2 Does not include health insurance subsidy,
social security or PERS COLA
32
Conclusion 2-4 cont.

• Tier 2 can still produce 65 to 100 replacement
  ratios for long service Tier 2 members

(a) Under the new entry termination scenario,
these ratios reflect those expected under money
match for current Tier 2 members who retire in
these periods.
(b) These ratios are a blend of the long service
Tier 2 employees already in service, plus post
2002 shorter service hires that are also
projected to retire at these times
33
Conclusion 2-5

• The Full Formula Floor
• Immunizes Tier 2 member benefits from the
  possibility of low investment returns yielding
  low Money Match driven benefits.
• The guaranteed floor adds approximately 1.2
  billion to the present value of system costs
  versus Tier 2 Money Match alone

34
Conclusion 2-5 cont.

• Why does the Full Formula guaranteed floor
  generate 1.2 Billion in additional costs versus
  Money Match alone?
• All the benefit outcomes in the lower portion of
  the distribution get shifted upward to the Full
  Formula outcome

Observed Full Formula Floor at approximately
40 Replacement Ratio
Theoretical Tier 2 Money Match Replacement Ratio
Distribution
35
Conclusion 2-5 cont.

• producing a new, more costly benefit structure
  that drops the low outcomes but retains the high
  ones.

36
Conclusion 2-6

• Employer rates during the Tier 2 era of the
  simulation (2036-2051) range from 16.2 to 21.5.
  
• This rate reflects amortization of continuing UAL
  and the normal cost of Tier 2.
• Simulation estimate of the Tier 2 Era (2042-2051)
  Rate Components

All rates are before pickup
37
Question -

• Why is there so large an abnormal add-on to the
  Tier 2 normal rate in this period?
• This outcome is not yet fully understood
• Until we understand the dynamic completely, we
  will neither conclude this is a solid estimate of
  2041-2051 era outcomes nor that it is a modeling
  anomaly
• We have already begun to investigate this outcome
  and have learned that

38
Possible Answers

• Possible answers may include
• Contribution rates were too low during the
  2026-2041 period?
• Volatility in the investment returns has a
  tendency to produce residual UAL when benefits
  are annuitized at the theoretical assumed rate?
• Increasing presence of retiree obligations alters
  the financial dynamics of the system when
  combined with Money Match features and the COLA?
• All of the above?

39
Volatility is Almost Certainly a Factor

• Annual fluctuation in Tier 2 investment returns
  interacts with the money match methodology, the
  full formula floor, and the increasing retiree
  reserve to produce rates above normal costs

40
The Role of Volatility

• That volatility plays at least some role is borne
  out when it is completely removed

41
Our Current Hypothesis
Volatility in Investment Returns
ROI
T2 FF Floor
The Downside Removed by the Full Formula Floor
for Benefits Only
but with
combined with
tends to produce
T2 MM
UAL
Small but Persistent Slices of New Unfunded
Liability
Money Match Investment Driven Benefits
42
What if

• So why not just invest in zero volatility
  securities?
• Generally speaking, they dont exist in the
  quantity and credit quality PERS would need
• The returns for low volatility investments are
  substantially lower than our current portfolio
  much too low to fund current and projected member
  benefits
• 10 yr US Treasury Note 4.000
• 30 yr US Treasury Bond(inflation adjusted)
  3.375
• It would cause employer rates to rise explosively
  (due to lower projected investment returns)

43
What Does the Simulation Suggest About Outcomes
Under a Partitioned System?
44
What is Partitioning?

• Partitioning describes a structure under which
  the OPERS assets are divided into separate pools
  dedicated to serving specific PERS liabilities
  (Tier 1, Tier 2, BIF) and invested in a manner
  most consistent with those liabilities.

45
What is the Partitioned Base Case

• Assets dedicated to Tier 1 member liabilities are
  invested more conservatively than the rest of the
  OPERS fund.
• Specifically

• All other Base Case factors remain unchanged

46
Four Notable Conclusions From the Partitioned
Base Case Stochastic Simulation
47
Conclusion 3-1

• Moderate Partitioning Lowers System Costs (and
  therefore employer costs) by over 4 billion.

48
Conclusion 3-2

• Moderate Partitioning Lowers the Tier 1 Member
  Assumed Rate, Guaranteed Rate, and Replacement
  Ratios

49
Conclusion 3-3

• Moderate partitioning moderately lowers employer
  rates over the next 30 years

50
Conclusion 3-4

• Partitioning has no effect on current retirees or
  Tier 2 members active or retired

51
What Does the Simulation Suggest About the
Consequences of a Maturing OPERS?
52
Changing Financial Dynamics

• From a financial dynamics point of view, the PERS
  we knew then (1975-1990), or know now
  (1991-2002), is not exactly the PERS we must
  manage in the years ahead

53
Relative PERS Liability Structure
In 1980

• Ratio of active members to retirees 3.71 to 1
• Ratio of BIF assets to active assets .58 to 1
• Ratio of Benefits Outflows to Contributions
  Inflows .41 to 1

54
Relative PERS Liability Structure
In 2001

• Ratio of active members to retirees 2.53 to 1
• Ratio of BIF assets to active assets .75 to 1
• Ratio of Benefits Outflows to Contributions
  Inflows 1.58 to 1

55
Relative PERS Liability Structure
In 2040

• Ratio of active members to retirees .82 to 1
• Ratio of BIF assets to active assets 1.38 to 1
• Ratio of Benefits Outflows to Contributions
  Inflows 1.96 to 1

56
Maturity Matters

• The Systems Financial Maturation is Inevitable
  and Inexorable Absent a Dramatic and Sustained
  Growth in Public Sector Employment and
  Compensation

57
Consequences of PERS Maturation

• There are Many.
• A few examples are
• An increase in the COLA formula becomes far more
  expensive
• Why? Substantially greater flows affected by
  COLA change
• Employer Rates become more sensitive to
  investment returns and accurate estimation of the
  assumed rate
• Why? Retiree (BIF) obligations are fixed and
  inflexibly dependent on investment returns at
  least meeting the assumed rate

58
What Does the Simulation Suggest About Equal vs.
Sequential Crediting?
59
Well

• The Simulation does not allow direct quantitative
  comparison of Sequential vs. Equal crediting
• We believe from inspecting the earnings crediting
  paths under each statutory theory and the models
  results that
• the financial outcomes to Tier 2 members and
  member variable accounts are identical
• Employer pre-retirement balances in aggregate
  should stay solvent provided the system can flex
  s between the BIF and employer accounts as
  necessary
• sequential crediting may shift some Tier 1 member
  deficit costs from Tier 1 members to employers
  but the change is highly likely to be marginal

60
What Does the Simulation Suggest About
Contraction in the Relative Size of the Public
Sector?
61
Zero Growth

• Zero absolute growth in PERS employment (and
  therefore a steady decline in the relative size
  of the sector) lowers pension costs but the
  savings is modest until the current UAL is paid
  off

62
What Does the Simulation Suggest About the
Consequences of a New Entry Plan Termination?
63
Conclusion 4-1

• Terminating new entry to the PERS plan frees up
  approximately 11.5 billion in employer costs
  versus the base case forecast

64
But

• but the cost savings emerge gradually over time
• and we assume there will be cost associated with
  providing pension coverage to new hires under a
  successor plan.