There are several standard methods available for the assessment of risk. Methods may include, but are not limited to, scenario tests, sensitivity tests, stochastic modeling, and stress tests.

• Scenario Test – A process for assessing the impact of an event on a plan’s financial condition. This may be one possible event, or several simultaneous or sequentially occurring possible events.
• Sensitivity Test – A process for assessing the impact of a change in an actuarial assumption or method on an actuarial measurement.
• Stochastic Modeling – A process for generating numerous potential outcomes (i.e., thousands) by allowing random variables in one or more inputs over time for the purpose of assessing the distribution of those outcomes.
• Stress Test – A process for assessing the impact of adverse changes in one or relatively few factors affecting a plan’s financial condition.

In determining a method for risk assessment, actuaries use their professional judgment to select methods that best account for the nature, scale, and complexity of the plan while also considering the usefulness, reliability, timeliness, and cost efficiency of the methods.

The Office of the State Actuary uses the methods noted above and combines the methods in some cases. Below are two examples.

Using our Interactive Reports, a user can assess how the plan’s present value of future benefits, funded status, and contribution rates change when the user selects an assumed rate of investment return or asset valuation method that varies from the best estimates used for the measurement.

These reports provide users with (1) a sense for how a plan’s financial condition or funding requirements may change if our assumptions prove inaccurate, and (2) an understanding of how our actuarial measurements would change if we applied different assumptions or methods at that same measurement date. For example, if we replaced the actuarial (or smoothed) value of assets with the Market Value of Assets (MVA), then how much would a plan’s funded status or contribution rates increase or decrease from current actuarial measurements? As another example, if we assume a lower or higher long-term assumed rate of investment return than our current best estimate, then how much would a plan’s funded status or contribution rates change from our current actuarial measurements? See our Interactive Reports to access these sensitivity tests.

We highlight the impact of the long-term rate of investment return assumption due to the significant impact of that assumption on plan funding. Over the past 20 years across all plans, investment returns have comprised approximately 70 percent of the pension fund’s total income, with the remaining 30 percent coming from employer contributions and employee contributions (split approximately 20 percent and 10 percent, respectively).

Our current best estimate measurements assume that future investment returns will continue to cover a significant share of future pension costs. However, past performance does not guarantee similar, future performance and our assumptions, in the long term, may prove to be inaccurate.

As required under the ASOPs, we assess all assumptions (economic or demographic) each year for reasonability. Consistent with the Revised Code of Washington, we recommend economic assumptions every two years as part of our Report on Financial Condition and Economic Experience Study. We also perform a demographic experience study approximately every six years as part of our Demographic Experience Study.

In 2010, we created a custom stochastic model to perform Risk Assessments periodically on our largest state pension plans. Every five to six years, we review the assumptions and methods used in those assessments as part of our Risk Assessment Assumptions Study.

Stochastic modeling is the most complex and time-consuming method for assessing risk. Unlike sensitivity analysis where you replace a single expected outcome with a different individual outcome under an alternative assumption or method, stochastic modeling provides a distribution of outcomes under numerous alternative assumptions or methods. This type of modeling provides the user with a better sense of the range of potential outcomes and the potential likelihood of those outcomes based on the assumptions and methods from the stochastic model. Because this modeling also includes very pessimistic outcomes, it represents a form of stress testing as well.

In Washington State, we combine scenario testing with stochastic analysis to assess two key risks for our state pension plans: (1) contribution risk – defined in this context as the potential for contributing entities to contribute less than the full actuarially required amount, and (2) benefit enhancement risk – the potential for the Legislature to enact future enhancements to current benefit provisions.

Why are these risks to the systems? Significant and persistent underfunding weakens a plan’s financial condition and leads to significantly higher future contribution requirements. This can create a cycle of funding shortfalls as demonstrated in the graphic below.

Significant one-time benefit enhancements or the practice of regularly enacting smaller benefit enhancements over time can also weaken a plan’s financial condition and lead to unsustainable contribution requirements.

To assess these risks, we perform stochastic modeling under the following two cases:

• Current Law – Where we assume all plans receive the full, actuarially required contribution amount (subject to certain assumed maximums) and benefit provisions remain unchanged over time, and
• Past Practices – Where we assume all plans receive a percentage (less than 100 percent) of the actuarially required contribution and the Legislature enhances benefit provisions in the future consistent with past practices.

By comparing the stochastic measurements from these two cases, users can see how the plan’s solvency and affordability risks change when the plans receive full funding and benefit provisions remain unchanged or when past practices continue in these areas.

For each case, we display select risk measurements and graphs generally, for the plans as a whole. The purpose of this information is to help users understand how much certain outcomes can vary from our best estimates and the likelihood of those outcomes based on the assumptions and methods from our stochastic modeling.

See our Risk Assessment webpage to access the results of our latest stochastic and scenario analysis.

Using this same risk assessment model, we also project employee and employer contribution rates for the next two biennia under the current law case. We then summarize the output of our modeling and compare the expected outcome with both pessimistic and optimistic outcomes. This information can be found on our Contribution Rates Projections webpage.

Thus far, we primarily focused our risk discussion on economic outcomes. However, demographic risks exist in our actuarial measurements as well.

Two significant demographic assumptions that affect the timing and amount of future pension benefit payments are rates of mortality (including rates of mortality improvement) and rates of retirement. For example, if retirees live significantly longer than currently assumed (longevity risk) or members eligible for subsidized early retirement retire significantly earlier than currently assumed, the plan’s financial condition would weaken and current contribution requirements could be inadequate.

This point is illustrated by the table below, which compares the funded ratios of the retirement plans if we double or remove our assumed rate of future mortality improvement. A funded ratio represents the portion of a plan’s actuarial accrued liability that is covered by its current actuarial assets. It serves as one of many measures that helps to explain the health of a pension plan. The below mortality improvement sensitivities are for illustrative purposes only and may not reflect reasonable assumptions for funding purposes.

In general, open plans (i.e., Plans 2/3) see larger changes in funded ratios than their closed plan counterparts (i.e., Plans 1). This is because the open plans have younger populations, and thus will have a longer exposure to mortality improvement over their remaining lifetimes.

We currently reduce the impacts of these demographic risks by reviewing annual experience and updating these assumptions, at a minimum, every five to six years where appropriate after the completion of a comprehensive demographic experience study. In the area of mortality improvement, we reduce longevity risk by applying assumed mortality improvement on a generational basis. Using this basis, we adjust currently assumed rates of mortality downward slightly each year in the future. For example, we would expect a 70-year old in 2021 to live longer than a 70-year old in 2020. The projected rates of mortality improvement we use are based on nationally published tables and informed by professional judgment. See our most recent Demographic Experience Study Report for further information on the development of our current demographic assumptions.

Often the easiest way for users to understand how measures can change over time may be to review a history of past actuarial measurements. See the Historical Data webpage on our website for select historical actuarial measurements from our past actuarial valuation reports.

Some risks can emerge just by the nature of a pension plan growing or maturing over time. For example, the more members and annuitants a pension plan has (i.e., as its obligations or contribution requirements grow), the larger it can become as a share of the government enterprise that supports it. In other cases, the larger and more mature a pension plan becomes (i.e., as its assets grow relative to the payroll of contributing members), the more volatile the contribution requirements can become. Both of these outcomes can present “enterprise risk” to the governing entities that sponsor the plan.

To illustrate this point, consider the Public Employees’ Retirement System (PERS), which is the largest Washington State pension plan. Both the state (primarily through the State General-Fund [GF-S]) and local governments support PERS through employer contributions (tax dollars). If employer contribution requirements for PERS grow too large over time, they could “crowd out” other government services. Alternatively, if contribution volatility becomes too significant, it may lead to unaffordable contribution requirements and funding shortfalls. Any funding shortfalls would also serve to increase future contribution requirements.

We measure the affordability of Washington State pension plans and the associated enterprise risk by comparing estimates of historical and projected pension contributions paid from the GF-S to the total GF-S budget. In the following graph, the line represents estimated GF-S pension contribution as a percentage of GF-S budget. The vertical bars correspond to the estimated dollar amount paid from the GF-S budget.

The historical percent of GF-S budget that is attributable to estimated GF-S contributions (green line) has grown over the past couple biennia and is projected to peak around fiscal year 2020. As discussed in the 2021 Report on Financial Condition and Economic Experience Study, the peak coincides with the highest collected contribution rates for most systems. GF-S pension contributions as a percent of the GF-S budget are projected to become a smaller percent of the state budget as the contribution rates decline. As an example, we expect this ratio to approximately halve for the state when the Plans 1 Unfunded Actuarial Accrued Liability (UAAL) attain 100 percent funded status and the UAAL payments are no longer required. As of the measurement date, the state contributes more than double the long-term expected percent of GF-S budget.

It is also worth noting that this graph shows the estimated GF-S pension contributions across all plans as a percent of the total GF-S budget. This percentage varies by plan and even varies over time. This is illustrated by the pie charts below which highlight, among other things, the arrival and maturation of the newer Plans 2/3 and the dissolvement of the closed Plans 1 once there are no members remaining. In addition, the charts display TRS exceeding PERS in percent of GF-S budget due to our higher fund splits assumption even though the retirement system is smaller.

Note: Plans not displayed comprise less than 0.05% of the GF-S budget.

To estimate GF-S contributions, we apply our historical and projected fund split assumptions to the applicable year’s total employer contribution. Our most recent fund splits assumption can be found on our Projections Model and Assumptions webpage. Where available, our comparison relies on GF-S budget information found in the June 2020 Washington State Economic and Revenue Forecast Report.

We measure the volatility risk of pension contributions by determining the ratio of the MVA to active participant payroll. In general, the higher the ratio, the more sensitive (or volatile) contribution rates are to asset returns that are lower or higher than expected (sometimes referred to as investment losses or investment gains).

To demonstrate this, we use the following example, which is conducted across the pension plans as a whole. At the current ratio for 2019, if assets earn 10 percent less than assumed in a single year without a subsequent and offsetting investment gain, the loss would require an additional contribution of about 2.9 percent for 20 years. If the current ratio were to double in the future, the same asset investment loss would require a contribution of about 5.8 percent for 20 years.

The graph below illustrates that the historical ratio between MVA and Active Participant Payroll across all plans has grown from about 2.8 in 1995 to 4.3 in 2019 and is projected to hold around 5.2 in 2050 (about a 20 percent increase from what was observed in 2019).

The above graph displays this measure for all plans but the measure can significantly vary between retirement plans. For additional information, we provided the ratio of MVA to active participant payroll for the Washington State pension plans that are open to new participants. Based on this measurement date, LEOFF 2 and WSPRS appear to have the most contribution rate volatility due to investment return experience that does not match expectations. The historical fluctuations in contribution rates for WSPRS reinforces the below chart; however, we do not believe LEOFF 2 is as high of a contribution risk due to its higher funded status and funding policy, which includes a higher minimum rate floor than the other plans.

As a pension plan matures, more members move from active to inactive status. While in active status, members generally contribute to the pension plan. While in inactive status, members stop contributing to the plan and either are in receipt of a pension benefit (drawing down plan assets) or have a vested right to receive a future pension benefit.

The relative maturity of a pension plan in this area can have implications on future contributions in two areas:

1. A reduction in the number of active members and their employers, relative to inactive members, subsidizing the cost of the plans. Dependent upon plan experience, this can increase or decrease the future contributions of plan members/employers. For example, consistent lower/higher than expected investment returns will be shared over fewer contributing members/employers and increase/decrease future contributions.
2. A reduction in the plan’s cash flow (annual contributions minus annual benefit payments), thus increasing the liquidity needs of the plan. The need for increased liquidity may lead to a revised asset allocation. The asset allocation for a mature plan could have an emphasis on asset classes with shorter investment horizons. An asset allocation with more liquid assets would have a lower expected return as a tradeoff for the less investment risk. Additional information on liquidity can be found further below.

Generally, these risks increase for closed plans (closed to new members) and have less relevance for open plans unless they become significantly underfunded.

The Washington State Investment Board monitors the on-going cash flow needs of the Commingled Trust Fund (CTF) and sets the asset allocation for the CTF with consideration of the expected cash flows. One common measure of liquidity needs is the ratio of (contributions less benefit payments) to the MVA, which we look at across all retirement plans in the graph below. The vertical bars in this graph correspond to the left axis and are generally negative dollar amounts which indicates benefit payments exceed contributions, while the line on the graph corresponds to the right axis and denotes this same dollar amount as a percentage of the MVA.

The CTF cash-flow measurement in the above graph does not account for investment returns which, as discussed in the Risk Measurements for Washington’s Public Pension Plans section, have comprised approximately 70 percent of the pension fund’s total income over the past 20 years. When you exclude expected investment returns from this measure, we observe a negative cash-flow measure for the CTF. If the measure becomes significantly more negative, as a percentage of the MVA, it indicates increased liquidity needs for the CTF. That outcome would likely result in a change in asset allocation, a lower assumed rate of investment return, and an increase in future contribution requirements.

When considering liquidity risk, it is also important to note that the open plans in the CTF provide liquidity for the closed plans invested in the same CTF. More specifically, the closed plans are shrinking and need to sell assets to pay benefits whereas the open plans are growing, currently taking in more in contributions than benefit payments, and need to buy more illiquid assets in the CTF. In this circumstance, the open plans can buy the closed plans shares/units of the CTF invested in illiquid assets. This allows the closed plans to remain invested in the CTF and benefit from higher investment rates of return without the liquidity risk they would face if invested in a separate trust fund with the same asset allocation.

The graphs below help to illustrate this dynamic by comparing the contributions, benefit payments, and net cash flow (or contributions less benefit payments) between the closed plans and the open plans.