Introduction
This brief is the second of a series of articles exploring actuarial considerations related to public long-term care (LTC) programs using a social insurance framework. We will use the recently proposed federal program—the Well-Being Insurance for Seniors to be at Home (WISH) Act—as reference point to help highlight considerations. Public LTC programs are complex in nature with many different questions to answer when constructing program features and benefits. Plan-specific parameters, geographic assumptions, and program eligibility requirements are just a few examples that influence the required revenue to fund public LTC programs.
Brand new public LTC programs, such as the proposed WISH Act,1 face considerable modeling challenges given the lack of prior experience. Sensitivity testing of assumptions is an important part of an actuarial framework to help inform and prepare for the unknown, from both a risk-monitoring and a rate-setting perspective. This article explores the variability for one component of the WISH Act, the required revenue (i.e., the tax rate). We illustrate the potential sensitivity of the needed tax rate by testing changes in various key modeling assumptions.
Figure 1 below summarizes some of the WISH Act’s key program parameters. For detailed background on the proposed WISH Act and its benefit design, please see the first article in this series, “Setting the stage: A journey on public LTC program design.”
Figure 1: Summary of the WISH Act’s key program parameters
| Proposed WISH Act | |
|---|---|
| Covered Population | Mandatory federal program, where all workers would help fund the program. Individuals can receive benefits if: (a) they contribute to the program for a set number of years, (b) they reach Social Security retirement age, and (c) they meet a benefit eligibility trigger. |
| Key Program Parameters | “Catastrophic” benefit design includes: $3,600 monthly cash benefit indexed to wages, unlimited benefit period, and an elimination period of one to five years (depending on lifetime income earned). |
| Program Revenue | 0.6% payroll tax, with 0.3% from employees and 0.3% from employers. |
Our testing indicates the WISH Act tax rate could range from 0.35% to 1.10%.
We observe the tax rate ranging from approximately 0.35% to 1.10% by sensitivity-testing several key assumptions in modeling projected revenue and expenses for the program. This range uses the WISH Act’s proposed payroll tax of 0.6% as a “central point” for illustration. We then examined the incremental impact to the tax rate by testing changes to select modeling assumptions one at a time. A larger range is possible to the extent the best estimate is not 0.6% and different sensitivity tests are performed (such as testing other key modeling assumptions, testing larger changes to assumptions, and testing changes to multiple assumptions at the same time).
For example, our high-level testing of the tax rate implies we might expect the best estimate to be higher than the 0.6% included in the proposed legislation. In that case, we would expect the range from sensitivity testing to be higher than the aforementioned range and to center around the higher best estimate tax rate.
Figure 2 shows the results of the tested assumptions changing one at a time in an effort to isolate the impact of each assumption. In reality, we might expect assumptions to be correlated, such that if one assumption varies from the best estimate assumption, other assumptions will be correlated (either negatively or positively), and also vary from the best estimate. Furthermore, the assumptions tested and presented in Figure 2 do not represent an exhaustive list of sensitivities. For this article we primarily focused on economic assumption tests, given their often direct impact on the program’s needed revenue. Other important assumptions to test include demographic and morbidity assumptions, such as mortality, mortality and morbidity improvement, fertility rates, and migration, among others
Figure 2: Sensitivity testing of key modeling assumptions one at a time
Testing of economic assumptions
We tested two economic assumptions for illustration: wage growth and investment rate of return. The results of these tests are presented in Figure 3.
Figure 3: Sensitivity testing of economic assumptions
| Sensitivity Test | Revenue Wage Growth Assumption | Benefit Inflation Assumption | Ultimate Investment Rate of Return Assumption | 75-Year Payroll Tax | Relativity to Best Estimate |
|---|---|---|---|---|---|
| “Best Estimate” | 3.6% | 3.6% | 4.7% | 0.60% | - |
| Low Benefit Inflation | 3.6% | 2.3% | 4.7% | 0.35% | 0.59 |
| High Benefit Inflation | 3.6% | 4.8% | 4.7% | 1.10% | 1.83 |
| Low Revenue Wage Growth | 2.3% | 3.6% | 4.7% | 0.90% | 1.50 |
| High Revenue Wage Growth | 4.8% | 3.6% | 4.7% | 0.40% | 0.67 |
| Low Revenue & Benefit Wage Growth | 2.3% | 2.3% | 4.7% | 0.50% | 0.83 |
| High Revenue & Benefit Wage Growth | 4.8% | 4.8% | 4.7% | 0.70% | 1.16 |
| Low Investment Rate of Return | 3.6% | 3.6% | 3.6% | 0.70% | 1.17 |
| High Investment Rate of Return | 3.6% | 3.6% | 5.8% | 0.50% | 0.83 |
Under the proposed WISH Act legislation, wage growth influences the program in two ways:
- Benefits are “indexed to wages,”2 meaning that, as wages grow, the program’s benefits grow.
- The program is funded through a payroll tax, so as wages grow the program’s revenue base grows.
What is not immediately apparent in the legislation is whether benefits are indexed to total wages or perhaps to some subset of wages that might follow cost of care trends more closely, such as caregiver wages. To help illustrate the impact of this distinction, we performed the following tests:
- For the “Benefit Inflation” tests, we modeled only a change to the inflation applied to the benefit payments and not wage growth, which has a large impact on the tax rate (increasing the annual benefit inflation by 120 basis points and increasing the tax rate by 83%).
- For the “Revenue Wage Growth” tests, we modeled only a change to wage growth on the revenue side and not benefit inflation, which has a large impact on the tax rate (decreasing wage growth by 120 basis points and increasing the tax rate by 50%).
- For the “Revenue & Benefit Wage Growth” tests, we modeled consistent changes to wage growth impacting both benefits and revenue. Under these tests, the impact on the tax rate is dampened because while the revenue is decreasing or increasing, the value of benefits is similarly decreasing or increasing.
Figure 3 presents the impact of testing the investment rate of return, where using a lower rate of return increases the required revenue by 17% and using a higher rate of return decreases the required revenue by 17%. The investment rate of return determines the level of investment income earned when the program has a fund balance. Public LTC programs may accumulate fund balances when revenue exceeds benefits and administration expenses. As the interest rate earned by the program fund increases, the necessary revenue funded through payroll tax decreases. Alternatively, if interest rates decrease, less investment income is earned on the program fund, requiring increased funding through payroll taxes.
For the purposes of the sensitivities tested in this article, we modeled the 2020 Federal Old-Age and Survivors Insurance and Federal Disability Insurance (OASDI) Trustees Report intermediate, low-cost, and high-cost assumptions.
Testing of other assumptions
We tested two noneconomic assumptions for illustration: benefit payments and vesting. The results of these tests are presented in Figure 4.
Figure 4: Sensitivity testing of other assumptions
| Sensitivity | 75-Year Payroll Tax | Relativity to Best Estimate |
|---|---|---|
| "Best Estimate" | 0.60% | n/a |
| Benefit Payments -20% | 0.50% | 0.83 |
| Benefit Payments +20% | 0.70% | 1.17 |
| Lower Vesting (5% to 10% reduction in vested individuals) |
0.55% | 0.92 |
| Higher Vesting (5% to 10% increase in vested individuals) |
0.65% | 1.08 |
As seen in Figure 4, adjusting benefit payments has almost a one-to-one impact on the tax rate, where changing benefit payments by 20% changes the tax rate by approximately 20%. Actual program benefit payments will be sensitive to many morbidity assumptions including the likelihood a program participant will start needing care (incidence rate), how long a beneficiary will need care (length of stay), the level of care needed (utilization rate), and the likelihood someone survives to a given age (mortality rate).
Figure 4 also presents the impact of sensitivity testing the vesting assumption. Vesting generally refers to a period of time an individual has to pay into the program to become eligible to receive benefits. In the case of the WISH Act, an individual must pay payroll taxes for a minimum requirement of five quarters to be eligible to receive partial benefits from the program or work a minimum of 40 quarters (or 10 years) to be eligible to receive full benefits from the program.
There are many reasons there might be uncertainty surrounding the vesting assumption, such as:
- This is a concept that is dissimilar to participant requirements of other LTC programs (such as private long-term care insurance). As such, it may be challenging to find data available to project expected vesting. As part of our sensitivity modeling, we used detailed Social Security data with individual, deidentified work history to project the percentage of the population to vest by age, gender, and projection year.
- When using historical work history data, it is important to consider how work histories have changed in the past 75 years, particularly by age and gender, and to consider how work patterns may continue to change into the future.
- It is also important to consider whether a program such as the WISH Act could have a behavioral impact on the vesting rates. For example, if the WISH Act was passed, would individuals be incentivized to work one more year than they otherwise would have if it means they can earn more benefits or become fully vested in the program?
As shown in Figure 4 above, assuming that 5% to 10% more individuals become vested in the program than what was assumed as part of the “best estimate” could require an 8% increase to the payroll tax to maintain program solvency.
Other important noneconomic assumption sensitivities to consider include: healthy life mortality, disabled life mortality, fertility rates, migration, immigration, incidence rates, lengths of stay, and utilization rates, as well as future calendar year changes to those assumptions. Given that the WISH Act’s design covers catastrophic benefits, tests on disabled life mortality and mortality improvement will be important to understand, as they could have large impacts on the amount of benefits paid to individuals.
Ending
Public LTC program results can be highly sensitive to even small changes in certain assumptions, as demonstrated in this article by testing the needed tax rate for the WISH Act. Sensitivity testing is an important and valuable tool that should be included when managing a public LTC program. Some advantages of sensitivity testing include:
- Assumptions can be difficult to project, especially given that many public LTC programs are new, without any direct existing experience to help inform estimates. Sensitivity testing will help understand which assumptions have the largest impact on program results.
- A program might have limited capacity to collect experience and data upon implementing a program, especially during the early years of the program. Sensitivity testing can help prioritize the type of data that would be useful to collect to monitor results.
- Actual expenses and related required revenue will inevitably vary from expectations, particularly for projections extending many years into the future. The variability of results and the ability of a program to quickly react and “course correct” will factor into how much margin the program should consider in its tax rate, premium rate, and fund level. For example, if a program could monitor results in real time and react immediately, the need for margin would be lower (all else equal). In cases where programs cannot react to adverse experience quickly (for example, a program change that can only be implemented through a statute change), sensitivity testing can be used to help inform the level of margin and establish a risk management framework consistent with program goals.
Next up in our article series, we will consider how changes to some key program features of the WISH Act could impact the needed tax rate.
1 The full text of the WISH Act is available at https://www.congress.gov/bill/117th-congress/house-bill/4289/text?q=%7B%22search%22:%5B%22WISH+Act%22%5D%7D&r=1&s=1.