New regime or huge blip? Get in by Feb 15!

As we have talked with many clients over the last few weeks about 2020 annuity policyholder behavior changes, three important questions have emerged:

     1. Did these changes persist through year-end?

     2. Should this be interpreted as a new behavioral regime or a huge blip?

     3. How should this be reflected in annual assumption reviews for stat and GAAP?

We are here to help you answer! Get your order forms in for 2021 FIA and VA studies by February 15 for lower pricing if you have not done so already, including VA mortality study and analysis of COVID-19 impact.

In case you missed it, check out the video and notes of our 2020 study results and this article to help you get internal approvals. (Hint: make a business case instead of asking for budget dollars).

Video: highlights of 2020 variable annuity industry studies

If you were not able to join our call on January 21, here is a video of the presentation and discussion, and notes are available here.  For lower pricing, get those order forms in for 2021 studies by February 15 (June ETA, with data thru full-year 2020).


Ruark Releases 2020 Variable Annuity Study Results

Indications that COVID-related market activity and disruptions affected policyholder behavior

Ruark Consulting, LLC today released the results of its 2020 industry studies of variable annuity (VA) policyholder behavior, which include surrenders, income utilization and partial withdrawals, and annuitizations.

“We expected that 2020 behavior would be different,” said Timothy Paris, Ruark’s CEO. “By looking at industry-level data, we are better able to identify and quantify those differences – especially on current-generation products.”

Ruark’s 2020 study spanned the period from January 2008 through June 2020. The study period was designed to capture early effects of COVID-19 and related market movements. COVID-related findings include:

  • Extreme market activity in the first half of 2020, and disruption to policyholders’ usual communication patterns with advisors and agents by COVID-related social distancing, appear to have affected VA surrender and income commencement behavior
  • Contracts with guaranteed lifetime withdrawal benefits (GLWB) persisted at greater rates than expected, as current-generation products exhibited greater sensitivity to 2020 market movements than they did in the past
  • Surrender rates fell uniformly on older product types in 2020; this is suggestive of a new, unique surrender regime, distinct from the regimes observed before and after the 2008 financial crisis
  • GLWB commencement rates were depressed in 2020 among contracts with the highest propensity to exercise the benefit: in-the-money contracts following the end of the deferral bonus period. Both the level and sensitivity of commencement rates were reduced.

Study data comprised 89 million years of exposure and 13.9 million policyholders from 20 participating companies, with $675 billion in account value as of the end of the study period. The study’s in-the-money exposures on GLWB contracts were 23% higher than in Ruark’s 2019 study (and 40% higher for deep in-the-money contracts). Among contracts issued since 2011, deep in-the-money exposure increased to 9% of total exposure, up from 6% in 2019. The study contained over 740,000 exposure years prior to withdrawal commencement for contract durations 11 and beyond, more than doubling the comparable exposure in Ruark’s 2019 study.

Other study highlights include:

  • GLWB deferral incentives appear to be effective. Income commencement rates are low overall; about 13% in the first year and falling to about half of that in years 2-10. However, commencement rates more than double in year 11 with the expiration of common 10-year bonuses for deferring income, before falling to an ultimate rate. After commencement, GLWB continuation rates are over 85%.
  • Income commencement rates increase when GLWBs are more in-the-money, that is, the benefit base exceeds the account value. This effect is quite pronounced after the expiration of common 10-year deferral incentives, with commencement rates ranging from low single digits to nearly 30% depending on moneyness.

  • Annual withdrawal frequency rates for GLWB and GMIB have continued to increase and have become more efficient with approximately 65% of recent experience at the full guaranteed income amount.
  • Free partial withdrawal amounts increase after the end of the surrender charge period, similar to the familiar “shock” in surrender rates. Excess withdrawal amounts on GLWB and GMIB increase as well.
  • Contracts with GLWB and GMIB have much lower surrender rates, and this effect is even more pronounced for those limiting their partial withdrawals to the guaranteed income amount.

  • Policyholders that take systematic withdrawals on GLWB and GMIB exhibit a select-and-ultimate effect, with very low surrenders in the first systematic withdrawal year and increasing thereafter. In the fourth systematic withdrawal year and beyond, surrender rates are comparable to those of contracts that have not taken any withdrawals.
  • When calculating relative value for GLWBs, use of a “nominal” moneyness basis (account value relative to the GLWB benefit base) can be deceiving, since it fails to reflect important aspects of the product’s economics. Therefore, it may be preferable in many cases to use an actuarial basis that incorporates interest and mortality rates. Surrenders exhibit a dynamic relationship to moneyness, whether measured on a nominal or actuarial basis. On a nominal basis 80% of GLWB exposure is in-the-money, whereas on an actuarial basis only 12% is in-the-money.

  • Surrender rates vary little across distribution channels, once other drivers of surrender behavior are accounted for. The exception is where companies cannot ascertain whether a policyholder has an ongoing relationship; where the distributor-policyholder relationship is weak, surrenders are as much as 50% higher.

  • Annuitization rates for GMIBs are in the low single digits and continue to decline. “Hybrid” versions that allow partial dollar-for-dollar withdrawals have much lower rates than traditional versions which reduce the guarantee in a pro-rata fashion, especially in the first year of eligibility. Factors influencing annuitization rates include age, duration, last year of eligibility, death benefit type, contract size, and moneyness.

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.

How to make the case for buying the data you need

In October 2019, I was one of the speakers at session 134 of the SOA Annual Meeting -- How To Get Real Results in Policyholder Behavior Modeling.  There were several Qs in the Q&A, and one of them was such a killer that my glib attempted A missed the deeper point.  Hand raised again, then killer Q was rephrased.  Gulp.  This time I got the point.  I think that words came out and we eventually moved on, but I knew and the asker of the killer Q knew that I did not produce a worthy A.

This wrecked my day.

It gnawed at me for months.  And it got me thinking about what we do from a much broader perspective.  Whether you are a potential buyer or seller of data and data-based insights, you are in the data business, so you need to know how to make the pitch and close the deal for data.  Here is the A that the killer Q deserved.  I hope you find it helpful...

From the November 2020 Issue of The ActuaryDemonstrating the Value of Data

Sympathy for the Actuary: Improving your model of FIA GLIB income commencement

Discussion slides are available for download here:  Ruark - BAAC 20201022 - Sympathy for the Actuary

The next meeting of our Behavioral Analytics Advisory Council will be:

Sympathy for the Actuary: Improving your model of FIA GLIB income commencement
We will discuss problems for annuity companies and actuaries with traditional modeling approaches, limitations of using only own-company data, and most importantly, an approach that we have developed and are using with clients that can quantifiably improve your model and risk management. Good for your company and good for you.

We look forward to your constructive feedback and an interesting discussion. You might even find this valuable for continuing education credit.  Please contact me if you have not already received an invitation with access info.

Thursday October 22 at 2:00-3:00pm Eastern

Best regards,


2021 Behavioral Analytics Plans

To aid your planning and budgeting for the rest of this year and 2021, here is a 1-page summary of our plans.


Thanks to so many of you who shared your ideas and suggestions at our most recent Behavioral Analytics Advisory Council call in June and in other discussions this year. We expect that our plans for 2021 will meet your needs. The inclusion of FIA and VA experience data from the 2020 crisis, the mortality study, and the other changes noted are worth discussion, so please plan to join our next call.  Please contact me if you have not already received an invitation with access info.


VA folks -- friendly reminder that data collection is in progress for our 2020 VA studies, with data through Q2 2020, target completion in November. Don't miss your chance!


Thank you for your ongoing support. Please contact me with any questions.


Best regards,



T minus 7

Hello friends of Ruark,

7 days to go!

If you have not done so already, reminder to please return your order form to us by June 30 before heading off for summer vacation. Thanks in advance, as this will help us plan for a very busy next few months. (And VA clients, prices increase 25% after June 30). We all will soon learn a lot about crisis-behavior for recent vintage products, which should mean better experience studies and better modeling.

If you still need to make your case with the budget hawks or higher-ups:

Even for large blocks, relevant industry data in a credibility-based framework has a quantifiable and compelling cost-benefit. Here is a short case study demonstrating this for GLIB/GLWB income modeling for a company that has 35% market share. If you are not using the industry data this way, why not?

And here is an even shorter one in terms of hedge breakage. We have seen that relevant industry data can consistently improve A/E results by several percentage points, which can avoid $ millions of losses due to hedge breakage.

...Data is not a budget question, it is an enterprise risk management question. Please, for your sake, our sake, and that of the actuarial profession, do the math before you decide. Let us know if we can be of any further assistance as you do. If you can find something that has better bang for the buck, we would love to hear about it.

Thanks for your ongoing support, stay safe, and have a great summer!

Kind regards,



Online meeting May 20 - Highlights from 2020 FIA Industry Studies

Thanks to the many of you who joined this online meeting!

Discussion slides are here - Ruark BAAC 20200520 - 2020 FIA Industry Studies.


FIA folks, this one is for you.  (And yes, as shown above, GLIB income commencement is sensitive to moneyness).

We thank those of you who have already purchased our 2020 FIA industry studies of policyholder behavior. We have held online meetings with many others over the last couple months to share highlights from our 2020 FIA industry studies including our first-ever FIA mortality table. This meeting on May 20 was intended for those with whom we had not yet had a chance to meet (online). Please let me know if you would like to schedule a separate meeting to aid your purchase decisions.

Lots of data:

  • Total exposure years grew to 23 million, a 21% increase over last year's study
  • Double the exposure years for GLIB contracts past the end of the surrender charge period
  • GLIB exposure constituted 44% of exposure overall, and 49% of exposure in the last 12 study months
  • 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

Discussion slides are here - Ruark BAAC 20200520 - 2020 FIA Industry Studies.

Kind regards,


Timothy Paris, FSA, MAAA
Chief Executive Officer
Ruark Consulting, LLC
530 Hopmeadow Street
Simsbury, CT 06070
Mobile 860.866.7786
Connect with me and follow Ruark on LinkedIn®


Revised plans for 2020 VA industry studies

To download the discussion slides from April 24, click here.

On behalf of our team we hope that you and your families are doing well.

On April 17, we announced revised plans for our 2020 VA industry studies of policyholder behavior. We made this decision with great thought and care following discussions with most of our VA clients over the preceding few weeks. Thanks to the constructive feedback and enthusiastic support of so many, we believe that this change will not only meet your needs but provide you with far more value than what we were originally planning.

  • Our revised plan is to perform the 2020 VA studies with supplemental data through June 30 2020, target completion in November.
  • In the interim, to aid your annual assumption review process, we will make the industry-only portion of last year’s studies available now if you submit your order form for the 2020 VA studies by June 30. After that, the studies will be subject to an additional 25% charge.

We look forward to supporting you in this important work in 2020 and beyond. Please contact me with any questions -

Online meeting March 23 - Market turmoil: What does it mean for annuity policyholder behavior?

Thanks to the nearly 100 of you who participated in this online meeting!

Given the turmoil in financial markets and hearing the concerns of our annuity company clients and others in recent weeks, we expanded what would otherwise have been our Behavioral Analytics Advisory Council to an open meeting for anyone interested.  Discussion slides are here - Ruark BAAC 20200323.

Our primary focus was the brief report "Market turmoil: What does it mean for annuity policyholder behavior?" published last week (link here) by our COO Eric Halpern. Here we have real FIA and VA data and insights 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.

With all of this great data, we are uniquely positioned to help you better manage risks, especially now.  Our insights from this data allow us to build more accurate models of policyholder behavior, with potentially less hedge breakage for VA and FIA -- how much is 1% A/E improvement worth to you?  Our predictive analytics techniques use company- and industry-level data to help you improve your 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 you can do if limited to your own data.

Links to a brief case study and highlights from our recent work are below:

Let me know if you would like to arrange a call or online meeting to discuss our work products and modeling services.

Please stay tuned as we are planning other ways to keep in touch by phone and online meeting over the next few months.  Let me know if you would like us to add you to our email distribution list.  And you can continue to find updates here on our website, LinkedIn, or social media pages using the links below.

Thanks for your continued support, and we hope that you and yours stay well.


Timothy Paris, FSA, MAAA
Chief Executive Officer
Ruark Consulting, LLC
530 Hopmeadow Street
Simsbury, CT 06070
Mobile 860.866.7786
Connect with me and follow Ruark on LinkedIn®