Market turmoil: What does it mean for annuity policyholder behavior?

To download the full report, click here.

Executive Summary

We offer insights on expected annuity policyholder responses to recent financial market turmoil, gleaned from our studies of annuity policyholder behavior since 2007.

Variable annuity writers should expect:

  • Greater persistency overall, but elevated surrenders for at-the-money GLWB
  • Greater income utilization, especially for GLWB after the deferral incentive period and “hybrid” GMIB
  • Greater GMIB annuitization elections, especially on traditional “pro-rata” benefit forms

Fixed indexed annuity writers should expect:

  • Greater persistency for GLIB, and lower persistency without GLIB
  • Greater income utilization for GLIB

COVID-19 impact on mortality:

  • Will likely depend on the level of containment among the general population at retirement ages, with potential differences between those with and without living benefit guarantees

Ruark is uniquely positioned to help as risk management takes center stage:

  • We have the data from past times of crisis –monthly policyholder behavior and mortality data going back to 2007 covering about 70% of the market with over $1 trillion of current account values
  • We have developed predictive analytics techniques that use company- and industry-level data to help our clients improve their annuity pricing, valuation, and risk management models. Our approach is rigorous, transparent, and tailored to each company, allowing for quick implementation and quantification of improvement in financial risk profile from what they can do if limited to their own data.

To download the full report, click here.

Ruark Releases 2020 Fixed Indexed Annuity Studies

Increasing data exposure in key areas

Ruark Consulting, LLC today released the results of its 2020 industry studies of fixed indexed annuity (FIA) policyholder behavior, which include surrenders, income utilization and partial withdrawals. Ruark’s FIA studies cover products both with and without a guaranteed lifetime income benefit (GLIB).

“With new data contributors, and rapid growth in the FIA market, data exposures in key areas continue to increase,” said Timothy Paris, Ruark’s CEO. “More data enables us to do more detailed analysis, identify new patterns, and – critically – help our clients achieve meaningful risk reduction in their models.”

Among the notable increases in exposure:

  • Total exposure years grew to 23 million, a 21% increase over the 2019 study
  • Double the exposure years for GLIB contracts past the end of the surrender charge period
  • A 39% increase in lifetime income withdrawals, to $5.4 billion

The study data comprised over 4 million policyholders from 17 participating companies spanning the 12-year period from 2007-2019, with $296 billion in account value as of the end of the period. GLIB exposure constituted 44% of exposure overall, and 49% of exposure in the last 12 study months.

Highlights include:

  • Lifetime income commencement rates are low, 6% overall in the first year following the end of the waiting period and then falling to the 2-3% range in years 3 and later. There is evidence of a spike in utilization after year 10, particularly where the benefit is structured as an optional rider rather an embedded product feature. Age, tax status, and contract size all influence commencement rates.
  • Lifetime income utilization increases sharply when policies are in the money, that is, the benefit base exceeds the account value. After normalizing for age, tax status, and contract duration, contracts that are 25% in the money or more exercise at a 10% rate. In contrast, when contracts with lifetime withdrawal benefits are out of the money, at the money, or modestly in the money, policyholders exercise at a base rate of about 2%.
  • FIA contracts typically offer the opportunity to take 5-10% of account value annually in penalty-free withdrawals, often following a 1-year waiting period. This is the case for contracts with and without a GLIB, though free partial withdrawal frequencies and amounts are somewhat lower on contracts with a lifetime income guarantee.
  • Free partial withdrawal activity is influenced by age and required minimum distributions, as well as contract size and the presence of a waiting period. Notably, withdrawal sizes spike in the year following the end of the surrender charge period, when all partial withdrawals become penalty-free; average withdrawal sizes for contracts without GLIB double to 20% of account value in the year immediately following the end of the surrender charge period.
  • Surrender rates continued to climb in 2019, particularly among contracts past the surrender charge period. The increase is broadly consistent with the rise in FIA sales that has been reported across the industry. While net sales have grown, a certain proportion of the increase in gross sales is likely attributable to exchanges of one FIA product for another.
  • Contracts with a guaranteed lifetime income benefit have much better persistency than those without, and among contracts that have begun taking income withdrawals, surrender rates are even lower. Persistency appears insensitive to nominal moneyness (the relationship of account value to the benefit base), but when an actuarial moneyness basis which discounts guaranteed income for interest and mortality rates is applied, we see that persistency is greater when the economic value is higher, as should be expected.
  • The relationship between surrender charges and surrender rates can be quantified. The study examines the relationship of persistency to the effective surrender charge, that is, the difference between account value and cash surrender value.
  • Surrenders are sensitive to external market forces as well as the absolute level of credited interest rates. Contracts earning less than 2% exhibit sharply higher surrenders than those earning more. As competitive market interest rates increase, so do surrenders, though there is some indication that a higher credited rate tempers the increase. In contrast, equity returns are negatively correlated with surrenders. Where cash surrender values are subject to market value adjustment, surrender rates for policies with a positive market value adjustment exceed those for policies with a negative adjustment. In the aggregate, policyholders act as though a positive MVA is a bonus, rather than a mechanism to make both parties whole, while surrender rates for contracts with negative MVA are similar to those for contracts with no MVA feature.

Detailed study results, including company-level analytics and customized behavioral assumption models calibrated to the study data, are available for purchase by participating companies.

Stat and GAAP: raising the bar for data analysis and policyholder behavior modeling

Whether VM-21 for variable annuities, GAAP LDTI, or the prospect of VM-23 for fixed indexed annuities, regulatory changes are raising the bar for data analysis, use of relevant industry data, and policyholder behavior model development.  Let's discuss how our industry studies, benchmarking, and customized model development services can help you.


Timothy Paris


How much is 1% A/E improvement worth to you?

For deferred annuities, minimizing hedge breakage is a key risk management objective.  Here is a simple example showing how a seemingly modest 1% improvement in actual-to-expected ratios can dramatically reduce hedge breakage, even for small- to medium-sized blocks.  How to do it?  By expanding on your own company's experience data to use relevant industry data and credibility theory to improve policyholder behavior models.

This is what we do.  Our work is not an expense, it is an investment in risk management with quantifiable benefits.  Let's discuss exactly how this can work for you.


Timothy Paris


Case study - modeling FIA GLIB income commencement

Download the case study here:  Ruark - case study - FIA income commencement using credibility theory and PA

Quantifying the benefits of using your company data, industry data, and credibility theory in a predictive analytics context.  This case study is focused on FIA GLIB income commencement but the approach works similarly well for other products, riders, and policyholder behaviors.  Our experience is that the financial benefits can be 1000x greater than the costs.  Let’s discuss exactly how this can work for you.


Timothy Paris


New VA and FIA mortality tables, splits for benefit type and durational anti-selection

I am very pleased to announce that we have released new industry mortality tables for variable annuity (VA) and fixed indexed annuity (FIA) products. Building on the industry studies and tables that we have produced since 2007, the new tables are derived from our 2018 studies of VA and FIA mortality and are an expansion of this work for specific VA rider types and for FIA. They include a table for VA contracts with lifetime withdrawal benefits (“RVAM-LB”); a table for VA contracts without living benefits (“RVAM-DB”); and a table for FIA (“RFIAM”) in aggregate. All are single-life mortality tables.

• The RVAM-LB table incorporates 34 million exposure years and 320,000 deaths on VA contracts with guaranteed lifetime withdrawal benefits (GLWB) or hybrid GMIB. The table is calibrated to experience in contract durations 3 and later, with select factors for the earlier durations. This reflects key findings from our 2018 study - GLWB and hybrid GMIB mortality is lower than average at issue and rises to an ultimate level over time.

• The RVAM-DB table incorporates 29 million exposure years and 523,000 deaths on VA contracts without living benefits. The table is a select-and-ultimate table with a 5-year select period. This reflects key findings from our 2018 study - VA contracts without living benefits, primarily with death benefit (DB) only, have higher mortality than average at issue and the magnitude of anti-selection varies by issue age.

• The RFIAM table incorporates 16 million exposure years and 265,000 deaths on FIA contracts, both with and without lifetime income riders. Similar to RVAM-LB, the RFIAM table is calibrated to experience in contract durations 3 and later, with select factors for the earlier durations reflecting lower mortality consistent with findings from our 2018 study.

These new tables are purpose-built for deferred annuities, and are demonstrably better than standard industry tables for VA and FIA valuation -- they reflect not only the effects of age and gender, but also differences by product type and contract duration which are important components of mortality anti-selection.

We are making the new tables immediately available, free of charge, to clients who have already purchased our respective 2018 VA and FIA mortality studies. New purchasers of the these studies will also receive the tables.

Detailed study results, including company-level analytics, benchmarking, and customized behavioral assumption models calibrated to the study data, are available for purchase by participating companies.

Please contact us if you would like to learn more.

EBIG Conference: using predictive analytics to model annuity policyholder behavior

Here is our presentation from session 2B of the EBIG Conference in November 2019.

It includes an exploration of drivers, cohorts, and dynamics for policyholder behavior for VA and FIA based on industry experience data, including changes in recent years and the emergence of long-term data in key areas.

Moreover, we discuss critical elements to developing a sustainable and coherent framework to translate this complex experience data into assumption models.

Ruark 2019 top news

  • How much is 1% A/E improvement worth to you? As you can see, we estimate that it can be hundreds of millions or even billions, but we recommend that you do the math for yourself. And yes, this 1%+ is what we can typically do using the data from our VA and FIA industry studies of policyholder behavior (over $1 trillion current account value) in a credibility-based predictive analytics framework. If you are aware of a better cost-benefit anywhere, we'd love to hear about it! We're trying to make it as easy as possible for you to fend off the budget hawks.


  • VM-21 exposure draft looks set to raise the bar on policyholder behavior data analysis and modeling assumptions. Fortunately, we've got you covered there too.


  • 2019 FIA and VA studies are still available for purchase, along with our most recent (triennial) 2018 mortality studies, if you have not done so already.

Other goodies --

  • Data gathering is in flight for 2020 FIA studies and we will turn to VA data after year-end. And we are working to bring some new data contributors aboard too. If you're not in yet, let's fix that.


  • As previously mentioned, we also plan to gather data in H1 2020 for a first ever GMIB post-annuitization mortality study. Longevity anti-selection? Comparisons to GLWBs or SPIAs? We want to know too, and the data is now emerging.


  • Looking ahead to 2020, our plans are here.  The industry studies feed into our more customized engagements which range from assumption review to development of predictive models and related assumptions to complete process management.



Please contact me if you would like to discuss.


Ruark: delighted to assist with the SOA Pri-2012 Private Retirement Plans Mortality Tables

While we still love our individual annuity work, we were delighted to assist the SOA in the compilation and processing of pension mortality data for the Pri-2012 mortality tables.


Policyholder Behavior is the focus at this SOA seminar

We will have a lot to do there -- 🔥case study on Fixed Indexed Annuity income utilization using credibility theory in a predictive analytics context, a review of policyholder behavior experience data findings across the industry including some newly emerging data and interrelationships in key areas, and much more.

Hope to see you there!  Bridging the Gap seminar on Nov 10, which leads into the Equity-Based Insurance Guarantees Conference Nov 11-12.