Press Release - Fall 2015 VA Study

Key Observations

  • Fairly Stable Surrender Rates over the past six years. Surrenders at the shock duration (the year following the end of the CDSC period) have remained fairly stable since the beginning of 2009.  This stability followed a steep decline from nearly 30% at the onset of the economic recession.  As the VA industry retrenched their product offerings in the wake of market volatility and low interest rates, contract owners no longer had ever-richer guarantee options within VA’s or attractive vehicles outside of VA’s to move to.
  • Effect of the moneyness and presence of a living benefit is notable. Contracts which are “in-the-money” (on an actuarial or nominal basis) with a GMIB or lifetime GMWB (GLWB) have much better persistency than those “out-of-the-money” or with other types of guarantees. GMIB’s have surpassed GLWB’s in this regard only in the past several years, with rates now a point and a half lower at the shock and nearly a point lower post-shock.
  • What is less important? Factors that are less significant for assumptions include attained age, gender, and contract size.  Even when surrender rate differences by these measures appear, they are explained away once the more significant factors of surrender charge and living benefit presence and value are accounted for.
  • Annual withdrawal frequency rates have been increasing over the past several years. Some of this change is due to demographics: frequencies go up with age.  However, even rates within age buckets have increased. GLWB riders are riskiest to the writing companies when contract owners take the full maximum annual withdrawal amount.  Utilization of the withdrawal feature continues to be less efficient in this way than initially expected, although efficiency is slowly increasing.  Overall, a bit less than half of annual withdrawals are at this maximum amount, while a third takes less than that.  The remaining 20% of withdrawals are in excess of the max, which degrades the guarantee for future years.
  • Principal drivers of GMIB annuitization are the relative value of the rider (“in the money-ness”) and attained age. Rider forms that allow partial dollar-for-dollar withdrawals have much lower exercise rates than those that reduce the benefit in a pro-rata fashion. The latter form emphasizes the availability of guaranteed income while retaining control of the account value, and so is more akin to a lifetime GMWB rider than a traditional GMIB.

Participating companies

AIG Life & Retirement

Allianz

AXA

Commonwealth

Genworth

Guardian

John Hancock

Mass Mutual

Met Life

Nationwide

New York Life

Ohio National

Pacific Life

Penn Mutual

Phoenix

Protective

Prudential

Security Benefit

Voya

Industry overview and methodology

Ruark’s studies include mortality, surrender, partial withdrawal, and GMIB annuitization –  policyholder behaviors that are vitally important to the long-term financial performance of variable annuity products, particularly with the burgeoning popularity of lifetime income guarantees such as GLWBs. Through the experience study results report and discussions with us, participating companies gain a detailed understanding of complex industry results and comparison to their own-company results for benchmarking purposes.

The number of participating companies and volume of data studied has grown dramatically. For example, Ruark’s GMIB Annuitization study captured 60% more exposure than 2014’s study.  Each company provided seriatim data files for the period January 2008 through June 2015.  We thoroughly scrubbed the data files and validated them with each company.  The 19 participating companies, representing approximately 70% of the industry, contributed 54 million policy years of data for the surrender study, and over 20 million policy years of data for the partial withdrawal study.

About us

Since 2007, Ruark has completed dozens of annuity experience studies that facilitated vital understanding of industry policyholder behavior.  We are recognized leaders in annuity risk management, policyholder behavior analytics, and reinsurance brokerage and administration.
Contact:  Joel Lagan | 860 930 5069 | joel@ruark.co | www.ruark.co


Variable & Indexed Annuity Reinsurance

While effective hedging of investment risks has rightly been the focus of variable annuity companies for the last few years, the enormous longevity risk implicit in living benefit guarantees has gone largely unnoticed and unmanaged.  This longevity risk is due to the fact that if and when living benefit guarantee claims are triggered, they typically take the simple form of a life annuity.  While this helps retirees mitigate the risk of outliving their assets, variable annuity companies risk that long-term increases in human longevity will outpace the level of longevity priced into the living benefit guarantee.

Three facts exacerbate this risk:

  1. There is no industry experience for living benefit guarantees in the payout phase;
  2. Industry mortality experience in the accumulation phase does not follow standard mortality tables;
  3. Demographers have a long history of severely underestimating mortality improvements, by as much as 5 years life expectancy at birth.[1]

Let’s quantify with a simple example.  With the illustrative assumptions of a male age 60 buyer and that claims are very unlikely to be triggered in the first ten years, the corresponding underestimation of life expectancy in the payout phase is 2.1 years[2].  For a $10 billion premium block of 5% living benefit guarantees, we would expect perhaps 67% still inforce after 10 years.  So the additional 2.1 years of payments would cost the variable annuity company $700 million ($10 billion x 5% x 67% x 2.1 years)!  This is equivalent to an additional cost of 1.40% annually on the declining asset balance, for a feature with only about a 1.00% policyholder charge.

We do not think that this level of longevity risk is within the risk appetite of many variable annuity companies.  How can it be managed?  Longevity reinsurance.  Similar to longevity swap products in the pension market, the variable annuity company and reinsurer would essentially swap the contingent living benefit guarantee payments modified for longevity deviations relative to a negotiated benchmark.

Some modifications would be necessary for the variable annuity living benefit guarantee market.  For variable annuities, we would expect the benchmark to reflect a customized blend between industry mortality experience in the accumulation phase and standard mortality tables in the payout phase, such as the Ruark Mortality Table and 2012 Immediate Annuity Table.  The reinsurance volume and coverage period would be set at the start of the transaction in order to mitigate policyholder behavior risk.  For example, the variable annuity company might expect that a $10 billion premium block of 5% living benefit guarantees would likely have 55-67% still inforce when the earliest claims are triggered after 10-15 years, so they might seek longevity reinsurance for $275 million ($10 billion x 5% x 55%) of annual lifetime payments triggered in that period.  Longer deferrals of the coverage period would naturally result in more conservative pricing, and reinsurers would likely require a modest premium stream as compensation for the risk that the payout phase is never triggered.

This type of longevity reinsurance is a creative extension of our expertise in the development, placement, and administration of mortality reinsurance.  We believe that as variable annuity companies recognize the enormous longevity risk embedded in living benefit guarantees, longevity reinsurance will join hedging programs and mortality reinsurance as indispensable modern tools for the management of their investment and insurance risks.  If you would like to learn more about how we can design and implement a longevity reinsurance program to meet your company’s needs, please contact:

Tim Paris, FSA, MAAA
Chief Executive Officer
(860) 866-7786
timothy.paris@ruarkonline.com

[1] Brian C. O’Neill, Deborah Balk, Melanie Brickman, and Markos Ezra, “A Guide to Global Population Projections”, Demographic Research, 4, p. 203-288, 2001.  Chris Shaw, “Fifty Years of United Kingdom National Population Projections:  How Accurate Have They Been?”, Population Trends, 128, Office for National Statistics, 2007.

[2] Timothy Paris, “Modern Variable Annuity Risk Management”, p. 6, 2012.