The main purpose of this webpage is to summarize assumptions and methods needed for the Office of the State Actuary’s 2020 Valuation Projections Model. A reader can find the assumptions and methods specific to the Projections Model in their respective sections below. We also provide a high-level background and disclosures section to help clarify the differences between an Actuarial Valuation Report and the Projections Model.

Background and Disclosures

Background

On an annual basis, we produce an actuarial valuation report (AVR) and projections model. Each serve their own purpose as described below.

The AVR relies on participant and financial data from a single point-in-time and set of best estimate assumptions. This is known as “closed group” analysis meaning the model assumes no new members join open plans after the measurement date. These valuations are the basis for monitoring funding progress and determining on-going contribution rates consistent with the state's funding policy. A single point-in-time measurement is helpful in tracking changes and trends across multiple AVRs. However, the AVR is limited by only providing results for a single measurement date. For estimates beyond the measurement date, we rely on our Projections Model.

Our Projections Model is like the AVR but includes additional assumptions for the demographics and number of new members joining open plans. This allows the Projections Model to estimate funding progress and contribution rates at future measurement dates. In other words, the model estimates future annual AVRs. We generally rely on two modes for the model: “deterministic” and “stochastic”. The deterministic setting assumes all future experience is consistent with our best estimate assumptions. This tells us the state of the retirement systems on an “expected” basis. The stochastic model helps us understand the impact to the retirement systems if future experience does not match our assumptions. Our stochastic modeling includes 2,000 simulations of investment returns, plan funding, and inflation. We summarize the results of our stochastic modeling in our annual Risk Assessment.

Disclosures

The Pension Funding Council (PFC) and Law Enforcement Officers’ and Fire Fighters’ (LEOFF) Plan 2 Retirement Board adopted new economic assumptions in fall of 2021. The PFC also adopted language for a contribution rate phase-in during their October 2021 meeting (Link: Pension Funding Council - Adoption of Economic Assumptions). Given the materiality of the change in economic assumptions, we incorporated the new assumptions, and a placeholder contribution rate phase-in, in our 2020 Valuation Projections Model. Typically, these changes would be reflected in the upcoming Actuarial Valuation Report. In addition to new economic assumptions, our 2020 Projections Model assumes $800 million will be appropriated to TRS 1 by the General Fund-State on June 30, 2023. Please see Economic Assumptions and Methodology sections below for additional information.

Unless noted otherwise, we rely on assumptions, methods, data, and assets from the 2020 Actuarial Valuation Report (AVR) for our Projections Model. However, the AVR does not contain some assumptions and methodology that are required to run the Projections Model such as new entrant [or members hired after the measurement date] assumptions or stochastic analysis assumptions. These assumptions and methods are generally updated annually and disclosed below. We disclose the remaining assumptions and methods in the 2016 Risk Assessment Assumptions Study (RAAS) since they are not updated as frequently. Please see the RAAS report for additional supporting information.

We prepared the Projections Model to support our work products as well as proactively address stakeholder questions related to expectations for the future. We believe the set of actuarial assumptions, methods, data, and assets used for the work products that rely on the Projections Model are reasonable but use of another set may also be reasonable and might produce different results. We expect this model will be updated during fall of each year. Kyle Stineman (ASA, MAAA) meets the Qualification Standards of the American Academy of Actuaries; he served as the reviewing and responsible actuary for the 2020 Valuation Projections Model.

Demographic Assumptions

Demographics of New Entrants

The assumed demographic characteristics of future new entrants in the Washington State retirement systems are summarized in the New Entrant Profiles tables.

Plan Growth

For systems that offer both plan 2 and plan 3 we need to develop specific growth assumptions for each plan. The Assumed Growth Rate by Plan table uses both the system growth for open plans and plan election assumptions to provide assumed plan growth rates, by year, for each open plan. Please see the RAAS report for additional detail on the general methods used for setting this assumption.

Please note that while an individual plan may show a negative growth rate, we still assume the plan has new entrants each year according to the Plan Election assumptions. For instance, the number of current members assumed to exit Teachers’ Retirement System (TRS) Plan 3 is greater than the number of assumed new entrants into that plan, but overall the entire TRS system is expected to grow.

Economic Assumptions Back to Top

As noted earlier, some economic assumptions may not match assumptions used in the 2020 AVR.

Discount Rate

The discount rate, or valuation interest rate, is the rate we use to determine the present value of future salaries and present value of future benefits. The discount rate assumption for projections is 7.0 percent for all systems.

Return on Assets

Our deterministic and stochastic projections rely on the total commingled trust fund (CTF) return for fiscal year 2021. For fiscal year 2021, we assume a dollar weighted return of 31.54 percent for all Washington State retirement plans. For deterministic projections, we assume a 7.0 percent return on investments each fiscal year beyond 2021.

Stochastic projections rely on a distribution of possible investment returns in order to calculate potential outcomes. Our stochastic projections use a downside log-stable distribution of investment returns provided by the Washington State Investment Board (WSIB). This distribution reflects WSIB’s 2021 Capital Market Assumptions with the commingled trust fund asset allocation adopted by WSIB in October 2021. The Investment Return distribution provides a summary of the range of investment returns we might expect in a single year. The range of returns we expect over a longer period will vary from the distribution of returns for a single year.

Inflation

For deterministic projections, we assume 2.75 percent inflation for all systems.

Stochastic projections rely on a distribution of possible inflation rates in order to calculate potential outcomes. The Inflation distribution provides a summary of the range of inflation rates we might expect in a single year. The range of inflation we expect over a longer period will vary from the distribution of inflation for a single year.

General Salary Growth

The general salary growth assumption is 3.25 percent for deterministic and stochastic projections for all systems.

Nominal Revenue Growth

Nominal revenue growth is used to project General Fund-State (GF-S) revenues over the next 50 years.

We use the Short-Term GF-S Forecast table to project revenue into the future. We relied on the Economic and Revenue Forecast Council's GF-S Forecast as the basis of our assumption.

Our long-term projections for GF-S revenue are assumed to grow as a result of three of our assumptions [inflation + population growth + real revenue growth = nominal revenue growth] following fiscal year (FY) 2025. Please see the RAAS report for additional detail.

Miscellaneous Assumptions Back to Top

Assumed Funding Sources

Where an individual works determines how their contributions are allocated from the various sources used to fund retirement benefits. The Projected GF-S Fund Splits table summarizes our assumption for what percent of total employer contributions, for each system, will be allocated from General Fund-State, Non General Fund-State, and Local Government accounts. The PERS, PSERS, and WSPRS assumptions were developed using data from Office of Financial Management (OFM), Washington State Patrol, and Department of Retirement Systems. TRS and SERS assumptions were informed by input from Senate, House, and OFM staff.

Projected State and Local Fund Splits table 

These Fund Split assumptions were most recently reviewed during the 2020 Legislative Interim. This table should not be used to reflect actual obligations for any funding source. We use long-term assumptions to produce our short-term budget impacts. Therefore, our short-term budget impacts will likely vary from estimates produced from other short-term budget models.

Methodology Back to Top

Approach to UAAL in LEOFF 1 Retirement Plan

Under current law, no contributions to LEOFF Plan 1 are required when the plan is fully funded. When LEOFF 1 was not fully funded, any unfunded actuarial accrued liability (UAAL) was required, under prior law, to be amortized by June 30, 2024.

A LEOFF 1 UAAL can occur beyond June 30, 2024 under our stochastic projections. When this occurs within the model, we assume the UAAL will be amortized, through contributions from the GF-S exclusively, over a ten-year rolling period of total LEOFF retirement system salary. This assumed funding method is similar to the current funding method for PERS 1 and TRS 1 except we do not assume a minimum contribution rate for LEOFF 1.

PERS 1 and TRS 1 UAAL

Under current law, employers of PERS, SERS, and PSERS contribute to the PERS 1 UAAL while employers of TRS contribute to the TRS 1 UAAL until the plans attain a 100 percent funded ratio. When this occurs, our Projections Model assumes all UAAL contributions, including the supplemental 10-year fixed contributions for past Plan 1 benefit improvements, will no longer continue.

RCW 41.45.060 is silent on what happens if a UAAL re-emerges so our model assumes UAAL contributions may resume. Our model assumes the employers would resume contributions based on current funding policy until the plan once again attains a 100 percent funded ratio.

Approach to Phase-In of Contribution Rates

During their October 2021 meeting, the Pension Funding Council adopted language to phase in contribution rates over six years. This action included a cap on 2023-25 biennium rates. (Link: Pension Funding Council - Adoption of Economic Assumptions). The Council did not further define the parameters of the phase in. To acknowledge this action, we held contribution rates in the 2023-25 biennium constant with what is being collected during the current biennium (2021-23 biennium). Beyond the 2023-25 biennium, we allow our Projections Model to calculate contributions rates consistent with current funding policy.

The phase in used within the Projections Model is a placeholder and should not be considered our recommendation for this approach. We expect to work closely with the Pension Funding Council beginning in the 2022 Legislative Interim to develop this approach.

$800 Million Appropriation to TRS 1

Under Engrossed Substitute Senate Bill (ESSB) 5092, the General Fund-State is scheduled to appropriate $800 million to fund the TRS 1 UAAL. Given this is in current law, our Projections Model assumes this payment will occur on June 30, 2023 and will be an additional contribution, over and above, the standard normal cost and UAAL rates collected over retirement system salaries.

Plan 3 Total Allocation Portfolio (TAP) Annuities

The Projections Model includes assets, benefit payments, and liabilities for Plan 3 TAP annuities that have been purchased by members. The model does not forecast future purchases of Plan 3 TAP annuities.

Comments on Projections Model

As required under ASOP 56 - Modeling, we share comments in the 2020 Actuarial Valuation Report related to our reliance on the ProVal® software developed by Winklevoss Technologies. Please see the 2020 AVR for disclosures related to the software created by an external source.

Using Microsoft Excel, our office created a Projections Model for developing measurements as of a future point in time. The model relies on output from ProVal software as well as assumptions, methods, data, and assets disclosed on this webpage. The output was reviewed for reasonableness by comparing results to simplified estimates done in Microsoft Excel. A limitation of our deterministic Projections Model is if experience is materially different from our expectations. Stochastic analysis can be run in the Projections Model to demonstrate how the investment experience and choices made by policy makers can deviate from our expectations; however, the model does not perform analysis on potential changes in demographic experience.


Last Reviewed: 01/07/2022

Last Updated: 01/07/2022