Voya 2013 Annual Report Download - page 187

nonperformance risk adjustment was based on the CDS spreads of ING Verzekeringen N.V. (“ING V”), a
subsidiary of ING Group and, previously, our indirect parent, and applied to the risk-free swap curve in our
valuation models. As a result of the availability of our own market observable data following our issuance of the
2022 Notes in the third quarter of 2012, we changed the estimate of nonperformance risk as of the beginning of
the third quarter of 2012 to incorporate a blend of observable, similarly rated peer holding company CDS
spreads, adjusted to reflect the credit quality of our individual insurance subsidiary that issued the guarantee as
well as an adjustment to reflect the priority of policyholder claims. The impact of the nonperformance risk
adjustment on the fair value of these liabilities was a reduction of approximately $377 million and $924 million
as of December 31, 2013 and 2012, respectively.
UL and VUL: Reserves for UL and VUL secondary guarantees and paid-up guarantees are calculated by
estimating the expected value of death benefits payable and recognizing those benefits ratably over the
accumulation period based on total expected assessments. The reserve for such products recognizes the portion of
contract assessments received in early years used to compensate us for benefits provided in later years.
Assumptions used, such as the interest rate, lapse rate and mortality, are consistent with assumptions used in
estimating gross profits for purposes of amortizing DAC.
Assumptions and Periodic Review
We have only minimal experience on policyholder behavior for our GMIB and GMWBL products and, as a
result, future experience could lead to significant changes in our assumptions. Our GMIB contracts have a ten-
year waiting period before annuitization is available, with most of these GMIB contracts issued during the period
2004 to 2006. These contracts first become eligible to annuitize during the period 2014 to 2016, but contain
significant incentives to delay annuitization beyond the first eligibility date. As a result, to date we have only a
statistically small sample of experience used to set annuitization rates. Therefore, we anticipate that observable
experience data will become statistically credible later this decade, when a large volume of GMIB benefits begin
to reach their maximum benefit over the four-year period from 2019 to 2022. It is possible, however, that
policyholders may choose to annuitize soon after the first annuitization date, rather than delay annuitization to
receive increased guarantee benefits, in which case we may have statistically credible experience as early as the
period from 2014 to 2016.
Similarly, most of our GMWBL contracts are still in the first four to six policy years, so our assumptions for
withdrawal from contracts with GMWBL benefits may change as experience emerges. In addition, like our
GMIB contracts, many of our GMWBL contracts contain significant incentives to delay withdrawal. We expect
customer decisions on annuitization and withdrawal will be influenced by customers’ financial plans and needs
as well as by interest rate and market conditions over time and by the availability and features of competing
products. If emerging experience deviates from our assumptions on either GMIB annuitization or GMWBL
withdrawal, such could have a significant effect on our reserve levels and related results of operation.
We also make estimates of expected lapse of these products, which is the probability that a policy will not
remain in force from one period to the next. Lapse rates of our annuity products may be significantly impacted by
the value of guaranteed minimum benefits relative to the value of the underlying separate accounts (account value
or account balance). In general, policies with guarantees that are “in the money” (i.e., where the notional benefit
amount is in excess of the account value) are assumed to be less likely to lapse. Conversely, “out of the money”
guarantees are assumed to be more likely to lapse as the policyholder has less incentive to retain the policy.
Lapse rates would also be adversely affected generally by developments that affect customer perception of
us. Our lapse rate experience of these products has varied significantly over the period from 2006 to the present,
reflecting among other factors, both pre- and post-financial crisis experience. During the early years of this
period, our lapse rate experience was higher than our current best estimate of policyholder lapse behavior would
have indicated; in the later part of this period, after mid-2009, it was lower. Management’s best estimate of lapse
behavior incorporates actual experience over the entire period, as we believe that, over the duration of the
policies, we will experience the full range of policyholder behavior and market conditions.
177