Have you ever wondered how INPRS knows how much money to set aside for your retirement?
Keep reading to discover how INPRS predicts how long pension benefits will be paid.
As Chief Actuary at INPRS, I have the privilege of serving on the Retirement Plans Experience Committee (RPEC) of the Society of Actuaries to bring the latest news in mortality trends to INPRS and to further my own professional development. RPEC develops and publishes mortality tables and improvement scales for retirement plans to use to project future benefit obligations.
Retirement actuaries often use RPEC tables as a baseline for mortality experience studies and to set appropriate mortality assumptions for measuring a plan's liabilities. In May 2025, RPEC published a new report on public pension plan mortality. This article will overview the new study and provide background on pension mortality assumptions in general.
Why mortality tables?
Before we dive in, it’s worth remembering why mortality tables are important to INPRS and other public pension systems. With a defined benefit, INPRS will continue to make monthly payments for the rest of a retiree’s life, and in some cases, a beneficiary’s life as well. To collect necessary contributions for these future benefits, INPRS estimates the value of all of those payments. This means estimating how long the payments will continue into the future – in other words, how long the recipients will live.
Actuaries don’t use life expectancy directly in these calculations. Instead, they predict the probability that a payment is being made each year using mortality rates. Near the beginning of a member’s retirement, there is a high chance that the payment will be made that year. As a retiree ages, the probability decreases until there is a 0% probably forecasted at age 120.
Where mortality tables come from
Mortality tables are tables of expected one-year rates of death by age, sex, and other categories. They typically come from studies of large sets of data. The Pub-2016* Public Retirement Plans Mortality Tables Report reflects approximately 58 million life-years of data submitted from 41 public pension systems across the United States. Contributors were asked to submit up to seven years of experience ending in 2020. This data takes time to request, submit, clean, validate, and analyze, so there is a significant delay between when the mortality is observed and when published rates are available.
Pub-2016 is not just one mortality table, but rather a family of mortality tables. As part of this analysis, RPEC determines how the data should be segmented. For example, women typically outlive men, so it makes sense to produce separate rates of death for males and females. However, there are other divisions where distinct mortality patterns emerge: employees’ job classifications, healthy vs. disabled retirees, contingent survivors, juveniles, and income level.
There are also different ways to weight the mortality experience. Most pension plans use amounts-weighted mortality. In other words, their mortality tables are not predicting the probability that a person dies at a member’s forecasted age, but rather the probability that a dollar dies at that forecasted age. Other types of benefit plans use headcount-weighted tables, so RPEC produces mortality rates with both types of weightings.
All of these segments and weightings lead to a lot of different mortality rates for different subsets of public pension plan populations. Ultimately, these mortality tables from the Pub-2016 study are a potential starting point for the mortality assumptions that a public pension plan’s actuary needs to make to estimate the plan’s liabilities.
Mortality in Motion
The initial mortality rates are only one part of the assumption that INPRS uses to predict future payments. As mentioned above, there is a delay between when the data is observed and when the tables are published. The “2016” in Pub-2016 is because that is the central, or average, year of the mortality experience data in this study.
RPEC also produces mortality improvement scales that can be used to adjust mortality rates from the past. Mortality improvement scales are likewise developed using large data sets and analytical techniques, but it takes a lot more data. In fact, the mortality improvement scales can only be developed from national mortality sources, such as the Social Security Administration, Centers for Disease Control and Prevention, the US Census Bureau, and the Centers for Medicare and Medicaid Services.
The mortality assumption starts by assuming the mortality rates are as of 2016 and then uses these improvement scales to project forward. The projection is not only to the current year, but also many years beyond. In other words, the mortality assumption INPRS uses expects that members will live longer in the future. A 65-year-old in 2026 will have a given probability of death, but a 65-year-old in 2056 will have a lower probability of death than the one in 2026.
Plan-Specific Adjustments to Mortality Rates
Both mortality rates and mortality improvement scales may be adjusted from what is published in their initial studies.
If a public pension plan is large enough, actuaries may be able to adjust the mortality tables to better fit the specific population of that plan. For example, the shape of the mortality table may match the pattern of what the pension plan experienced, but the overall rates may be too high or too low, so the actuary will adjust.
These types of adjustments are done in a process called an experience study. As of the publication of this article; INPRS’s most recent experience study was finished in 2025. INPRS completes a new experience study every five years. At the time of the 2025 experience study, the Pub-2016 tables were not yet finalized. INPRS will evaluate the Pub-2016 tables now that they have been published.
Mortality improvement also incorporates a degree of professional judgment about how mortality rates will improve going forward. The future is uncertain, and actuaries may adjust the published mortality improvement scale to better fit what they expect to occur to the population they are working with.
COVID-19 impact and what’s next
You may have noticed that the Pub-2016 data observation period went into 2020, when the COVID-19 pandemic began in the United States. RPEC examined this data but ultimately excluded anything in calendar year 2020 from the Pub-2016 mortality tables.
Although the US and world experienced higher-than-expected mortality during the COVID-19 pandemic, it is unclear how mortality may have changed going forward from the pandemic. It will take several more years of data gathering and analysis before we get a clear picture of post-pandemic mortality. The RPEC 2025 Mortality Improvement Update report suggests that the first post-pandemic improvement scale may be released around 2027-2029. The next public plan mortality study is expected to be released sometime around 2030.
* “Pub” for public pension plan mortality, as opposed to “Pri” for private pension plan mortality.