Traditionally, US federal tax law has treated life insurance quite liberally:
This is good public policy when we contemplate destitute widows and orphans as the beneficiaries, but not when a flimsy insurance veneer is tacked on to an investment in the hope of gaining a generous tax break. Internal Revenue Code §7702 and §7702A were enacted to temper these advantages for investment-oriented contracts, by imposing limitations on premiums and cash values in relation to death benefits. Three classes of contracts result:
Insurers are responsible for compliance with §7702 and, if the owner elects, §7702A. They face severe penalties[1] even for “slight” or “reasonable” mistakes, which are all too common due to the difficulty of understanding and implementing the requirements correctly. Unlike state nonforfeiture law, which is a reversible translation of straightforward actuarial formulas into English, §7702 and §7702A merely hint at the intricate calculations that they ordain. Dismay awaits anyone who delves into them hoping to come face to to face with an unambigous algorithm reflected through a glass darkly by the statutes. The legislative history is more illuminating, but cannot be distilled into one unequivocal specification: the drafters weren’t programmers or actuaries. Published commentary usually surveys the spectrum of permissible compliance methods without resolving all the quandaries that arise along any particular path. The goal of this monograph is to describe every detail of one sound path through the §7702 and §7702A thicket; there are other valid paths, but this is the one lmi follows.
These specifications apply only to contracts that were issued since §7702 and §7702A became effective, or that later become subject to them through loss of grandfathering. They do not address taxation of MECs: it is assumed that an existing admin system does that correctly. For the same reason, they do not address the §7702A(d) anticipation-of-failure rule, the §72(e)(11)(A)(i) anti-abuse rule that aggregates all MECs a company issues to a policyholder in a calendar year, or the §7702(c)(4) rule that aggregates all contracts of $10,000 or less.
1 Life insurance contracts must meet one of two tests prescribed by Internal Revenue Code §7702 in order to qualify for favorable tax treatment in the United States. One is the cash value accumulation test (CVAT); the other is the guideline premium and corridor test (GPT).
2 The CVAT has only one requirement: that the death benefit (DB) must never be less than a certain corridor factor times the §7702(f)(2)(A) “cash surrender value” (CSV). Departing from common industry usage, the statute defines CSV as account value (AV) ignoring any surrender charge, loan, reasonable termination dividend[2], or dividend accumulation[3]; some products include certain other items in CSV[4]. The CVAT corridor factor for each attained age is calculated at issue under assumptions prescribed by law (¶4/2), always assuming DBO A. It normally varies by gender, and sometimes by tobacco use and underwriting class.
3 The GPT has two requirements that must both be met at all times. The first is that the DB must never be less than a certain corridor factor times the §7702(f)(2)(A) CSV. The GPT corridor factors vary only by attained age, and not by gender, tobacco use, or underwriting class. The corridor factors are more liberal than for the CVAT because the GPT also imposes premium limits.
4 The second GPT requirement limits cumulative premiums at each moment. The limit is the greater of the guideline single premium (GSP) or the cumulative annual guideline level premium (GLP). Both guideline premiums are generally calculated according to actual contract mechanics, using assumptions prescribed by law. The minimum interest assumption is higher for the GSP than for the GLP. Unlike seven-pay premiums used for MEC testing, guideline premiums are loaded for current expense charges. The GLP varies by DBO, but the GSP always assumes DBO A. Guideline premiums are calculated at issue, and do not automatically change as attained age increases. They do generally change when benefits change.
5 A product can offer a choice of GPT or CVAT, but one test must be chosen for each contract when it is issued. The test chosen generally cannot be changed after issue[5]. Compared to the CVAT, the GPT is generally more conservative at early durations due to a stricter initial premium limit based on a higher statutory interest rate, but more liberal in later years because of lower corridor factors. The CVAT is the simpler and more flexible §7702 test, but MEC testing can be much simpler for GPT contracts. The GPT is preferred in the high net worth market where the net amount at risk (NAAR) may ultimately exceed available reinsurance.
1 Life insurance receives more favorable tax treatment than annuities. §7702A’s intention is to deny preferential treatment of living benefits on contracts whose early funding is deemed excessive, by defining them as MECs and exposing them to taxation under §72(e)(10), (e)(11), and (v). A life insurance policy becomes a MEC if it is issued in exchange for a MEC[6], or if premiums are paid at a rate more rapid than one seven-pay premium for each of the first seven years. Certain changes cause the policy to be treated as a new contract with an adjusted premium limitation for a new seven-year period.
2 §7702A would have no effect if the death benefit could be reduced without penalty the day after a large payment. Therefore, when the death benefit decreases, premium testing generally has to be repeated as though the lower death benefit had been in effect since the beginning of the most recent seven-year test period. For most policies, retesting is not necessary for decreases that occur beyond the end of the last seven-year test period.
3 Complex rules prescribe the §7702A treatment of other contract changes. Into every CVAT contract §7702A embeds a deemed contract. Its deemed cash value (DCV) accumulates like the actual cash value, but under prescribed assumptions. A premium is “necessary” to the extent it doesn’t cause the DCV to exceed the net single premium (NSP) that defines the CVAT corridor. Any further payment (which would cause the DCV to enter the corridor) is “unnecessary” premium. A “material change” occurs, e.g., when benefits increase for any reason (even if due solely to the corridor or to option B), but its recognition may be deferred until unnecessary premium has been paid. Notionally, when that happens, the cash value buys a paid-up policy, and we issue a new contract for the current death benefit minus the paid-up amount. The new contract is subject to a new seven-year premium limit. We don’t actually issue new policies; we just perform §7702A calculations as though we had. For GPT contracts, we calculate no DCV, but instead treat all premium as necessary and handle changes under our GPT rules (¶13/2–3).
1 Age means insurance age, nearest (ANB) or last (ALB) birthday as specified by the contract, as long as it is within one year of actual age[7]. For ANB, if two birthdays are equally near, either age may be used.
2 The issue age is the insurance age on the contract effective date. When a contract is backdated, this produces the correct result because it is the effective date that is backdated. §7702 does not allow artificial age adjustments exceeding one year, such as the setbacks for female or substandard that were common in the past, or the joint equivalent age method that is still sometimes used for survivorship contracts.
3 The insurance age at the beginning of the policy year defines the attained age, which is used for all off-anniversary changes. If an insured born on 1960-01-01 purchases an ANB contract on 2000-07-02 at insurance age 40, and changes it on 2001-07-01, the day before the first anniversary, then all §7702 calculations for the change use age 40, even though the transaction occurs 41 years and 181 days after birth.
4 Rates and values are looked up by attained age as defined above for §7702A calculations as well. Thus, when an off-anniversary material change causes the §7702A(c)(3)(A)(i) “contract year” not to coincide with any policy year, the attained age is still the insurance age defined in terms of the effective date given in the contract.
5 §7702 deems the contract to mature between ages 95 and 100 inclusive. Contracts maturing for a reduced endowment prior to age 95 are deemed to mature at age 95. If the contract does not specify any maturity age, then age 100 is used[8].
6 Riders mature when they expire by their terms, or at age 100 if they do not expire.
1 On the issue date, guideline and seven-pay premiums must be calculated based on the DB before any payment or §1035 exchange amount is applied; by definition, this DB cannot exceed the specified amount (SA). In order to achieve the most favorable guideline premiums, a contract should generally be issued at a SA no lower than its corridor DB (¶4/2) after the initial premium, dumpin, and any estimated §1035 exchange amount are recognized, net of any term rider. If this is precluded by sales considerations that demand a lower SA, then that lower SA becomes the DB for initial guideline and seven-pay premiums. Using any higher amount would expose the contract to §7702A difficulties: if poor investment performance caused the DB to decrease, then the seven-pay test would need to be reapplied retrospectively using the decreased DB. If the original seven-pay premium was paid on the issue date and such a decrease occurred more than sixty days after the first anniversary but before the eighth year, then the contract would become an irremediable MEC.
2 In certain sales situations, we may elect to use a lower amount for initial §7702 and §7702A calculations. For instance, if it is intended to take a $10,000 withdrawal from a $100,000 policy in the first seven years, we may choose to treat the initial death benefit as $90,000. Then we need not recognize a decrease or an adjustment event when the $10,000 withdrawal occurs. This requires adding an extra input parameter to existing systems.
3 In general, we should perform §1035 calculations at issue, using an assumed §1035 amount. The old contract is assigned to us and we surrender it, perhaps some months after the new contract was issued. When we get the actual funds, we redo the §7702 and §7702A calculations as though we had known the correct amount on the issue date. To prevent the later receipt of the rollover from causing a material change or an adjustment event, it is important not to underestimate the §1035 amount.
1 Cash surrender value (CSV) for all §7702 and §7702A purposes is the amount payable on full surrender, treating surrender charges and policy loans as though they did not exist. It equals account value (whether loaned or unloaned) plus any extra amounts payable on surrender other than dividend accumulations or reasonable termination dividends[9]. The specifications for some products deem certain additional amounts to be included in CSV for determining the corridor DB: for example, a refundable sales load or an experience-rating “reserve”. Fortunately, adjustment events and material changes do not cause recalculation of the add-on amounts in either of these particular examples, although §7702A(c)(2)(A) decreases do require such a recalculation[10]. Actuaries contemplating such add-ons for new products should strive to avoid any dependency that would cause such a recalculation.
2 Death benefit is the amount payable by reason of death. To satisfy the §7702 definition of life insurance, it must never be less than a corridor factor times the CSV. The GPT corridor factors are given in the statute[11]. The CVAT corridor factors are the reciprocal of the sum of the NSP and the present value of any qualified additional benefit (QAB) charges[12].
3 Decreasing benefits specified in the contract, for instance in the case of decreasing term, must be taken into account; lmi does not address such benefits at this time. Riders that terminate before maturity may be handled under the QAB rules.
4 Decreasing benefits not specified in the contract need not be taken into account. Where an elective decrease is intended, we should reduce the initial death benefit (¶3/2).
5 Increasing benefits can be taken into account prospectively only if they are specified in the contract itself, and then only to the extent they do not increase the NAAR. This includes the usual option B when it is elected. Increases not specified in the contract are always ignored. It makes no difference that such increases follow a specific pattern specified in advance, as in a letter from the owner or an illustration signed by the owner, because even such manifest intentions are not bound to be performed. It makes no difference that the contract contains a provision allowing such increases to be elected at the owner’s option.
6 The §7702A CVAT DCV always uses the actual death benefit option, but the usual return of premium (ROP) death benefit option is generally treated as option A for guideline premium purposes (but for ROP forceouts see ¶6/4). Therefore, a change from option A to ROP or ROP to A is not an adjustment event. In theory, for ROP we could take into account increases up to the option B amount, but only to the extent they are certain to occur; yet this is awkward or impossible to apply when the premium is flexible.
7 Increasing benefits specified in the contract that do increase the NAAR must be taken into account as they occur. For instance, a “jumping juvenile” contract that provides a death benefit of $1000 per unit through age 20, and $5000 per unit from age 21 on, has an adjustment event at age 21. It is treated exactly the same as a level contract with an elective increase at age 21.
8 §7702 and §7702A calculations take an endowment benefit into account. But if the contract specifies a reduced endowment benefit (as some retirement income policies do), then the endowment benefit cannot exceed that reduced amount (lmi does not support this). Initially, the endowment benefit is the SA at issue. For CVAT contracts, it is reset to the new SA upon each material change. For GPT contracts, it is reset to the new SA upon each adjustment event, but only with respect to the seven-pay premium and the quantity B in the A + B − C formula (¶5/4); the quantities A and C use the SA immediately prior to the adjustment event[13].
1 Guideline premiums are recalculated whenever certain
changes (“adjustment events”) occur.
Adjustment events include
changes in SA, if and only if DB also changes,
changes in death benefit option,
changes in QABs,
reductions in substandard table ratings or flat extras, including rider ratings, and
changes in the §7702 mortality or interest basis.
These transactions are adjustment events only if they change
factors or values actually used in the guideline calculations;
for instance, if the implementation conservatively ignores
substandard charges and QABs, then no adjustment event arises
when they change.
The server must
nevertheless be notified of all such changes, because an
insurer might amend its current practices at any time.
Automatic adjustments in the DB of an “integrated” (¶11/8)
term rider due to the corridor are not adjustment events if
the total DB does not change; early termination of the rider
and voluntary changes in its SA are adjustment events if the
total DB changes.
Since an adjustment event may be triggered
by any transaction that affects the SA, we must test all
withdrawals.
Refer all cases of misstatement of age or gender to the actuarial
department[14].
2 The owner cannot directly change the DB, because the contract’s elective increase and decrease provisions apply specifically to the SA only. A change in SA is not an adjustment event if the DB does not also change—for example, a $10,000 SA increase on an option A contract with a $100,000 SA and a $150,000 corridor DB is not an adjustment event. But there may be an adjustment event in this example due to some other cause, for instance if the DBO is changed on the same date.
3 A DBO change is an adjustment event (but see ¶4/6) even if it doesn’t cause the SA or the DB to change, because it directly affects the GLP calculation (¶14.3/1). When a DBO change is the only adjustment event, GSP does not change, because by definition it is independent of DBO. In that circumstance, recalculating GSP is a superfluous step that leaves the value unchanged. It may be performed anyway if that is simpler than writing code to anticipate and avoid this special case, or it may be omitted.
4 Recalculations follow the
A + B − C
method.
The formula is:
A = guideline premium before change
B = guideline premium at attained age for new DB and new DBO
C = guideline premium at attained age for old DB and old DBO
new guideline premium = A + B − C
This formula is applied to both the GLP and the GSP (which
latter never varies by DBO).
The new GLP and GSP apply until
the next change, and are used as the quantity A above in
calculating the effect of any subsequent change.
5 Special care must be taken in defining the benefit amount for B and C above. Some older published actuarial commentary, citing §7702(f)(7)(A), suggests using the SA (thus, significantly, ignoring the corridor), but the drafters considered a contract in the corridor as fully funded for the corridor amount, making any premium increase with respect to the corridor redundant. For instance, consider an option A contract with a $100,000 SA and a $150,000 corridor DB. A $1 SA increase should not cause the guideline limit to increase substantially if at all. A $1 SA decrease should not cause the guideline limit to increase at all. Following this reasoning, IRS might look askance at a strategy of increasing the SA in exact anticipation of each corridor increase.
6 Instead we define B and C according to §7702(f)(3): “the term ‘death benefit’ means the amount payable by reason of the death of the insured”. This ensures that no extra allowance is given for extra death benefit that has arisen due to the corridor.
7 The “old DB” is captured before the day’s transactions are applied, using the current day’s corridor factor and the latest unit values available at that moment. It is not altered as a result of the day’s transactions. For instance, if SA is increased on the same day as a premium payment, then the potential adjustment event is processed ignoring the premium. The premium does not affect the “old DB”, even if it would otherwise drive the contract into the corridor[15]. Neither does it affect the “new DB” because adjustments must be processed before premium can be accepted (¶5/8).
8 Events occurring on different days are never combined. All adjustment events and material changes that occur on the same date are combined together and processed as one single change. This is done as soon as all transactions that potentially create adjustment events have been applied, and must be done before any new premium is accepted because an adjustment can affect the guideline premium limit. This aggregation is impossible in the case of a DB increase due to a payment under the ROP DBO[16] because the payment changes the SA and potentially the DB, but other adjustment events must be processed before the payment is accepted. Furthermore, any such increase is not a material change due to the necessary premium exception. Therefore, we ignore DB increases arising out of the normal operation of the ROP benefit, treating them no more liberally than corridor DB increases. This means that the guideline limit would be higher on an otherwise identical option A contract with elective SA increases tailored to match the ROP DB pattern.
9 If the guideline limit becomes negative, ensuing forceouts may eventually reduce CSV to zero[17]; in that case, §7702(f)(6) might justify maintaining the contract effectively as term insurance (¶6/7). Alternatively, administrative practice can simply forbid any change that would reduce the guideline limit to an amount less than zero.
10 Off-anniversary changes never amend past history.
They have only a prospective effect on guideline premiums, which
are linearly interpolated between A and
(A + B − C)
according to the proportion of the policy year completed at
the time of the change.
For example, if a contract with a
$10,000 GLP is changed 249 days after its anniversary in a
leap year, such that its new GLP by the
A + B − C method
is $46,600, then the GLP limit for that policy year is $21,700:
$21,700 = 10,000 × [249 ÷ 366] + 46,600 × [1 − (249 ÷ 366)]
Any fraction of a cent must be discarded.
lmi does not perform the interpolation because
illustration systems generally do not allow such changes.
11 Less favorable taxation (under §72, without considering the exception in §72(e)(5)) applies to cash distributions accompanied by benefit reductions within the first fifteen years from the issue date actually shown in the contract. This issue date is not adjusted when §7702 or §7702A merely deems the contract to be reissued or exchanged. But for a §1035 exchange, it is the date shown in the new contract. Cash distributions include forceouts, dividends, and withdrawals, but do not include loans or amounts returned (¶6/2) to preserve the §7702 or §7702A status of the contract. A benefit reduction is any decrease either in the death benefit or in any QAB benefit below the level assumed on the issue date. A cash distribution accompanies a benefit reduction if it occurs on the same date as the reduction or within the preceding two years[18].
12 The amount subject to LIFO taxation due to benefit reductions in the first fifteen years is the actual cash distribution, limited to a so-called “recapture ceiling” that may be thought of as the excess of actual CSV before the reduction, over allowable CSV after the reduction. The recapture ceiling is thus in the nature of a forceout amount, except that funds are not necessarily forced out of the contract. If the benefit reduction occurs during the first five policy years, the recapture ceiling for a CVAT contract is the excess of CSV immediately before the reduction over NSP immediately after the reduction. If the benefit reduction occurs during the first five policy years, the recapture ceiling for a GPT contract is the excess of CSV immediately before the reduction over corridor cash value (DB times corridor factor) immediately after the reduction; or the excess of premiums paid immediately before the reduction over the guideline limit immediately after the reduction: whichever is more restrictive. If the benefit reduction occurs after the first five years but during the first fifteen years, then the recapture ceiling is the excess of CSV over corridor cash value using GPT corridor factors for both GPT and CVAT contracts. If the recapture ceiling is zero or negative, then no adverse taxation applies. An admin system that doesn’t actually perform tax reporting must nevertheless flag such reductions for manual attention. The recapture ceiling is irrelevant to a MEC.
13 Unilateral changes the insurer makes in current interest, mortality, expense, or QAB[19] rates are neither adjustment events nor material changes; they are reflected immediately in the CVAT DCV, but otherwise only when an adjustment event or material change later occurs, and only with prospective effect in either case. Changes in rate class that are initiated by the owner[20] (such as smoker-to-nonsmoker changes and reductions in substandard table ratings or flat extras) are adjustment events and material changes to the extent they affect §7702 and §7702A calculations, respectively; typically, such changes are subject to underwriting and affect guaranteed as well as current charges.
14 For §7702 and §7702A purposes, a substitution of insured is treated as a taxable exchange[21] rather than an adjustment[22]. A new contract is deemed to be issued. Any value in the old contract is taxed as a full surrender, and the remainder is rolled into the new contract. The effective date of the new contract is deemed to be the effective date of the exchange. However, a substitution performed under a binding obligation in the contract does not disturb grandfathering or restart the fifteen-year clock (¶15/12).
1 For §7702 purposes, premiums paid are all payments—no matter who pays them—less forceouts, nontaxable withdrawals[23], amounts returned under ¶6/2, and charges for non-qualified additional benefits that are not deducted from AV. However, amounts the insurer pays into the contract under a waiver benefit are excluded[24].
2 Amounts returned in order to preserve the §7702 qualification[25] of the contract reduce premiums paid, as long as they are returned with taxable interest within sixty days of the end of the policy year of payment. Amounts returned in like manner to prevent a MEC under §7702A[26] are treated the same way, because they are in the nature of forceouts. The interest that is obligatorily added to such amounts does not reduce premiums paid.
3 No admin system should accept any premium (or process any other transaction) that makes a contract a MEC[27] without the owner’s prior consent. No system—not even an illustration system—should ever accept any premium that violates the guideline limit; the great majority of vendor admin systems refuse to, and so does lmi. The alternative of accepting premium unconditionally and processing a countervailing transaction later is unreliable. Across the industry, many MEC failures have resulted from reliance on this alternative, when the countervailing transaction is not processed in time.
4 If a post-issue change reduces the guideline limit below cumulative premiums paid, the excess is forced out of the contract. Any such distribution is subject to the normal rules of taxation[28], particularly including those described in ¶5/11–12. A forceout is otherwise similar to a withdrawal transaction except that it is not subject to any fees or limits that apply to voluntary withdrawals. Customarily, withdrawals reduce the SA, but forceouts do not. If a forceout reduces the amount of premium considered under the terms of an ROP benefit, then a cyclical formula results. This can be resolved by applying the formula repeatedly until it yields no further forceout of even a fraction of a cent[29].
5 Products for the high net worth market typically require involuntary withdrawals[30] (with no fee and no effect on SA) to keep NAAR constant when it would otherwise increase beyond reinsurance capacity due to cash value growth. This is neither a taxable event under §7702(f)(7)(B–E), nor an adjustment event, nor a material change. It’s like the payout of a dividend: instead of putting the increment into the cash value, we mail it to the owner.
6 When a withdrawal decreases the SA, that decrease is an adjustment event if the DB also changes. Any transaction fee charged on a withdrawal is deducted from the CVAT DCV, but is not included in the expense charges that enter the guideline premium calculation; however, to the extent that it decreases the SA, it flows into the adjustment event calculation. Insofar as such a fee is part of the withdrawal, it reduces basis, §7702 premiums paid, and §7702A amounts paid.
7 The GPT always permits payment of the minimum premium required to keep a contract in force until the end of any contract year, even if it would exceed the guideline premium limit, as long as the CSV (ignoring any surrender charges or loans) will be zero at the end of that year[31]. To obviate the premium solve this implies, admin systems ought instead to calculate each month the least amount that will prevent the contract from lapsing before the next monthiversary[32].
1 Product specifications must always state guideline premium interest assumptions. The remainder of this section describes how these assumptions are determined, and may be skipped by programmers.
2 §7702 prescribes the interest basis for all §7702 and §7702A calculations as the interest rate actually guaranteed in the contract, or a statutory rate if greater. The original 1984 statutory rate was 4% for GLP, 6% for GSP, and 4% for all CVAT and §7702A calculations except that the necessary premium for guideline contracts is defined in terms of the guideline limit. Beginning in 2021, H.R. 6800 provided for those rates to vary[33], depending on certain published values.
3 The §7702 net rate is determined in two steps. First, the guaranteed interest rate is determined from the contract, and the statutory rate is used instead if it is greater. This operation is performed separately for all periods with different guaranteed rates[34]. For example, if the guaranteed rate is 4.5% for five years and 3.5% thereafter, and the original 1984 statutory rates apply, then the GLP interest rate is 4.5% for five years and 4.0% thereafter, while the GSP rate is always 6.0%. For products such as pure variable UL that offer no explicit guarantee, the statutory rate is used. For variable products that offer a general-account option, the guaranteed gross rate must be no less than the general-account guaranteed rate.
4 Even short-term guarantees at issue must be reflected in the GSP, the CVAT NSP, and the §7702A NSP, seven-pay premium, and DCV. They may be ignored as de minimis in calculating the §7702 GLP[35], but only as long as they last no longer than one year. Only guarantees that either last longer than one year or are present on the issue date are taken into account: a guarantee subsequently added for a future period lasting no longer than one year is a dividend, not an adjustment event. Here, “issue” excludes cases where the contract is merely deemed by statute to be reissued[36].
5 Second, any current asset-based charges specified in the contract are deducted if we wish. The interest rate remains what it is; the net rate that results from subtracting asset-based charges is merely a computational convenience that simplifies the formulas. In fact, the full interest rate (never less than statutory) is credited, and then asset-based charges are subtracted from the account value. Therefore, this adjustment affects only the §7702 guideline premiums and the §7702 DCV, because those quantities reflect expenses. It must not be taken into account when calculating the §7702 CVAT NSP or CVAT corridor factors, or the §7702A NSP or seven-pay premium, because those quantities do not reflect expenses.
6 Asset based charges can be deducted only if they are specified in the contract itself; charges imposed by separate accounts cannot be deducted unless they are specified in the life insurance contract proper, since any charge not so specified is deemed to be zero[37], and naturally do not reduce the interest rate for contracts that offer a general-account option. They also must not exceed the charges reasonably expected to be actually imposed[38]. If the schedule page announces a charge of “up to 100 basis points” and we actually charge 50 bp and expect to keep charging that, then we can use 50 bp; but if we ever charge less than 50 bp, an adjustment event results.
7 It is critical that the result be rounded up if at all, and never rounded down or truncated. The GPT is a bright-line test, and truncation at, say, eight decimal places may have an effect of more than a dollar per thousand[39] at a later duration. Special attention must be paid to the exact method the administration system uses (e.g. beginning of period versus end of period), to be sure that the resulting charge is what will actually be imposed. A §7702(f)(8) waiver granted in one actual case that was pennies over the limit cost tens of thousands of dollars in filing and attorney’s fees.
8 Thus, an account value load that is deducted from the account value at the beginning of each month, before interest is credited, may be reflected in GPT calculations. We could calculate it as a monthly load in order to follow the precise contract mechanics, but that would require a significant modification of Eckley’s formulas, which do not contemplate a load on AV. Instead, we net the account value load against the §7702 interest rate; as explained in ¶7/5, this is a mere computational convenience that does not change the actual interest rate. A proof that this alternative is conservative is given in the Actuarial Addendum (¶B/8).
9 On the other hand, it is not clear that a classical mortality and expense charge (M&E) can be reflected, because it is part of the daily unit value[40] calculation. The effect of this M&E on monthly interest is a function of the ratio of successive unit values, and the actual charge approaches zero when the unit values decrease quickly. If a definite charge were clearly and unconditionally deducted at the beginning of each day, before crediting interest, then we might take it into account by adding daily commutation functions to the Formulas section and extending the proof in ¶B/8 accordingly; but lmi ignores such charges.
10 Multiple guaranteed rates may result, for instance in the case of a variable contract with a general-account option and a distinct guarantee for loaned funds. The highest such rate is used, because that produces the most conservative guideline premium limits.
11 A higher rate guaranteed in a side letter must be reflected as in ¶7/4, as though it were written in the contract. For products that guarantee a rate tied to an index, the §7702 interest rates in the first guarantee period must be at least as high as the rate determined by the index when the contract is issued. Such guarantees must be taken into account even if they arise indirectly or contingently, for instance in the case of an unloaned credited rate that is guaranteed to be no less than 50 bp below an indexed loan rate.
12 For calculating mortality charges, most UL products discount the NAAR for one month’s interest at a rate specified in the contract. §7702 and §7702A calculations must use the §7702 rate instead whenever that is higher than the contractual rate. This affects all premium rates and also the CVAT DCV and corridor factors. Whenever this rate is converted to a monthly equivalent, the result must be rounded up if at all. If the contract specifies no such discount and none is actually applied, then a discount rate of zero may be used.
13 The interest rate guaranteed by the contract is the greater at each duration of the guaranteed loan credited rate or the rate otherwise guaranteed. If a fixed rate is elected, then the guaranteed loan credited rate, if not stated explicitly, is the fixed rate charged on loans minus the guaranteed loan spread if any. If the contract guarantees neither the loan credited rate nor the loan spread, then a fixed loan rate has no §7702 or §7702A effect.
14 There is a concern if a variable loan rate (VLR) is elected. Section 3.D of the VLR model regulation provides that “the maximum rate…must be determined at regular intervals at least once every twelve (12) months, but not more frequently than once in any three-month period”. There is no rate guarantee after the first anniversary, because the VLR rate may change by that time. However, since the maximum VLR is fixed for at least three months at issue, there is a short-term guarantee that must be reflected as in ¶7/4 if the rate actually credited on loans is too high. The complications that ensue may be avoided by actually crediting a loan rate no higher than §7702 otherwise requires during the first loan rate determination period, or simply by forbidding loans during that period.
1 Product specifications must always state §7702 mortality assumptions. The remainder of this section describes how these assumptions are determined, and may be skipped by programmers. The Actuarial Addendum (¶B/9) addresses the conversion of annual mortality rates to monthly.
2 §7702 prescribes the use of reasonable mortality that does not exceed the charges the insurer actually expects to impose, except for the safe harbor that IRS Notice 2006-95 provides. The mortality tables specified in the safe harbor vary by gender (except that unisex rates may be used for females, but only where required by state law[41]) and, if prescribed in the contract, by tobacco use.
3 For insureds with no table rating, lmi uses guaranteed mortality, limited to 100% of safe-harbor mortality at each duration. The 100% limit applies even if the guaranteed mortality is greater (as for some guaranteed-issue and simplified-issue contracts). lmi does not support contracts that guarantee mortality lower than the safe harbor[42].
4 For insureds with a substandard table rating, we may compare current mortality reflecting the rating to 100% of safe-harbor mortality, and use the greater of these two values in each year. We may choose to ignore ratings due to foreign residence. We may indeed choose to use 100% of safe-harbor mortality in all cases as an administrative simplification, in which case changes in table ratings or flat extras are not adjustment events; that is the only option lmi currently supports. As a consequence, a contract may lapse early even though the CVAT NSP is paid as a single premium, the seven-pay premium is paid annually for the first seven years, or the GLP is paid annually forever.
5 Any flat extras may be added to the mortality rates for all applicable durations, provided that we actually expect to impose the total charge that results (as for table ratings, above). However, for the GLP, it may be better to add any temporary flat extra as an explicit charge in every applicable year, in order to avoid an adjustment event when the flat expires or is reduced or forgiven. This follows the Blue Book discussion of QABs; even though a flat extra doesn’t otherwise look like a QAB, it’s a reasonable thing to do for a temporary flat. But this approach gives no relief for flat extras that are permanent: an adjustment event occurs if they are later reduced or forgiven.
6 At any rate, for UL contracts, it is probably better to ignore flat extras altogether for all §7702 and §7702A purposes; that is what lmi does. Flat extras normally flow through the AV, and IRS has expressed concern that they may therefore earn tax-advantaged interest. While we realize that flat extras increase the premiums necessary to achieve a given AV, it is not transparent to IRS that they make the contract less investment oriented.
7 Some policy forms “blend” mortality by gender or tobacco use, specifying (at issue, and immutably unless there is a misstatement of age or gender) the applicable safe-harbor mortality tables. Their treatment of current mortality may pose daunting practical problems if it is desired to reflect substandard ratings in §7702 and §7702A calculations, for example if the blending percentages for current mortality are revised periodically to reflect case demographic changes. We have seen a contract that blends mortality by interpolating on qx instead of tpx, and IRS might assert that this departure from generally accepted actuarial practice ignores the expected improvement over time as the proportion of nonsmokers and females increases. At best, any use of blended current mortality could be expected to produce unwelcome adjustment events. It does not affect any §7702 or §7702A calculation unless substandard mortality is reflected, so substandard mortality should certainly be ignored for such contracts.
1 Current expense charges that the insurer actually expects to impose are reflected in guideline premiums and the CVAT DCV. Higher guaranteed charges are disregarded. Originally, guaranteed charges were allowed, but the statute was changed in 1988[43] due to perceived abuses. Expense charges are ignored for the CVAT net single premium and for the net single and seven-pay premiums used for MEC testing. However, they are always taken into account for necessary premium calculations. Furthermore, charges for QABs[44] may always be reflected in all calculations.
2 For products with charges that are tiered by premium or by total assets (which can change on every valuation date), it is convenient to select a charge amount low enough that it will never need to be changed, or conservatively to disregard the charge altogether[45]. lmi applies the lowest current tiered rate (cf. ¶7/6) for all guideline premium calculations, but uses the actual current charge for the CVAT DCV.
3 Some products pass the actual premium tax through as a load; it’s treated like any other load. If the insurer adapts it to changes in premium-tax rates, that’s treated like any other unilateral change (¶5/13).
4 To account for the possibility that current expense charges have changed, the admin system must provide the current GLP and GSP to the inforce illustration system.
1 For products with a secondary cash value guarantee, the §7702 interest, mortality, and (if applicable) expense basis at each duration must be at least as conservative as the basis that actually determines the guaranteed value at each duration[46]. An exact calculation must be done for each month and for each possible combination of issue age, underwriting class, and so on[47]. The practical difficulty of these calculations might be mitigated by using the more conservative basis for all durations, and by using the CVAT instead of the GPT[48].
2 Consider a product that is guaranteed to remain in force with a CSV of at least zero as long as the GLP is paid each year. This guarantee may be disregarded under §7702 to the extent that the guarantee premium exactly equals the GLP; but if it is calculated in any other way[49], then ¶10/1 applies, requiring potentially onerous calculations.
1 Riders and other additional benefits are either “qualified” or not. QABs receive more favorable treatment (¶11/2–7), which extends to CVAT NSP and corridor factors, and to §7702A NSP, DCV, and seven-pay premiums, as well as to guideline premiums, except where specifically noted otherwise. Currently, lmi does not treat any eligible additional benefit as a QAB, but it does support the term rider described in ¶11/8.
2 QABs can be prefunded through the AV. A QAB’s benefit is deemed to be the stream of current charges for the QAB, but only over the period during which such charges are payable as established in the contract[50]. Thus, the QAB’s charges create a level positive increment to the guideline premiums for that charge period, with appropriate effects on the other quantities in ¶11/1. At the end of that period, the increment goes away, even if the QAB’s benefits continue; this is neither an adjustment event, nor a material change, nor a §7702A(c)(2)(A) decrease in benefits[51]. Thus, QABs may not be funded through contract maturity unless their charges continue through maturity.
3 Both benefits and charges for non-qualified additional benefits are completely disregarded if their charges cannot flow through the AV, in which case the charges are paid in cash and do not count as premium[52]. Otherwise, payments to support them count against the premium limit, and their charges decrease the AV (and, hence, the CSV and corridor DB) without directly increasing any payment limit—although they do decrease the DCV[53]. That outcome can be avoided by stipulating that the charges must be paid in cash and cannot be deducted from the AV. At any rate, no change in a non-qualified additional benefit is an adjustment event, a material change, or a §7702A(c)(2)(A) decrease[54].
4 Only the following are QABs according to §7702(f)(5)(A)
(but see ¶11/5):
guaranteed insurability,
accidental death or disability benefit,
spouse and child term riders,
disability waiver benefit, or
other benefits prescribed under regulations (of which there are none).
The seven-pay premium calculation disregards any QAB whose
benefits last fewer than seven years.
Otherwise, that calculation can include a QAB’s charges for
the period those charges continue, even if that is less than
seven years.
5 A term rider on the main insured whose benefit terminates prior to age 95 is treated as a QAB for §7702, but as death benefit for §7702A unless it expires within the first seven contract years. If the benefit is guaranteed at least until age 95, then it is treated as death benefit for both §7702 and §7702A. Death-benefit treatment is the most favorable outcome: it values the benefit using §7702 mortality rather than current rider charges. For a term rider treated as death benefit, §7702 calculations reflect the actual benefit duration, but §7702A calculations deem the benefit to last until maturity[55].
6 Increasing or decreasing the benefit amount of a QAB (such as a term rider) is an adjustment event. Terminating a QAB is an adjustment event and a §7702A(c)(2)(A) decrease, whether the termination is elective or required by the terms of the contract; but expiry at a date set in the contract at issue is not (¶11/2). In particular, there is an adjustment event when a spouse or child term rider terminates due to their death[56], but not when such a term rider terminates due to the family member’s attainment of a specified expiry age[57]. Revising or removing a rating on a QAB (for example, an occupational accidental death rating) is an adjustment event unless the rating was disregarded in §7702 calculations. Neither disability, nor recovery therefrom, is an adjustment event for a waiver benefit.
7 A rider that waives monthly deductions (as opposed to paying a stipulated premium), with a charge proportional to the actual monthly deduction, poses a definitional difficulty. The charge for the benefit is indeterminate because AV growth generally changes the NAAR and hence the mortality charges. In theory the waiver charge attributable to determinate amounts such as the policy fee or load per $1 of SA could be reflected. But those amounts are probably almost negligible, so it is better to treat such a waiver benefit as a non-qualified additional benefit—which is what lmi does.
8 Consider an “integrated” term rider that automatically adjusts to offset the corridor and automatically converts to the base contract at, say, age 70. While it is in force, its DB is the term SA minus the portion of the base policy DB due solely to the corridor, but never less than zero. This rider is appropriately treated as death benefit, ¶11/5 notwithstanding, and not as a QAB because the automatic conversion preserves the total DB with no action on the owner’s part and a positive election is required to obtain any different outcome[58]. Termination of this rider, e.g. because there is insufficient AV to pay its monthly cost, is an adjustment event and a §7702A(c)(2)(A) decrease, unless its DB simultaneously converts to the base.
1 Product designs that impute an addition to AV when calculating the corridor DB should address the concerns described in ¶4/1, preferably in such a way that the add-on amount does not change when an adjustment event, material change, or decrease occurs.
2 Pursuant to the discussion of ROP above (¶4/6), a modified ROP death benefit option might provide for the greater of the usual ROP death benefit and the option B death benefit. With this option, we could calculate GLP (but not GSP) on an option B basis, perhaps tripling the GLP that would otherwise apply. However, in durations where the option B death benefit governs, higher death benefits and COI charges would result.
3 When interest rate guarantees are added after a contract is issued, it is preferable to limit them to one year. Otherwise, they adversely affect the interest rate used for calculating the GLP (¶7/4).
4 If it is desired to reflect account value loads (such as M&E charges) in guideline premium calculations, then it is important to specify and implement them in such a way that they can legitimately increase the guideline premiums (¶7/8–9).
5 During the first VLR rate determination period, VLR contracts should avoid short-term guarantees by either forbidding loans or crediting loan interest at the §7702 rate (¶7/14).
6 Consideration might be given to moving flat extras outside the AV, by billing for them and requiring them to be paid directly on a current basis. This should remove any IRS concern that they could be prefunded, thus making it safe to reflect them in guideline premium calculations (¶8/6). Flat extras are particularly important for certain survivorship product designs.
7 §7702 (f)(5)(C)(ii) allows us to exclude the charges for non-qualified additional benefits from premiums paid, so that they do not count against the guideline limit, if they are “not prefunded”. The only way to preclude prefunding is to prevent the charges for these benefits from flowing through the AV by requiring them to be paid in cash (¶11/3). It may be beneficial to structure such benefits that way.
8 Term riders that are not “integrated” with the base policy for §7702 purposes (¶11/8) receive more favorable treatment if they last at least until age 95. On the other hand, New York has non-extraterritorial rules that make it difficult to continue a term rider past age 70, and even more difficult past age 80, in that state. Consider extending term riders to the base contract’s maturity age outside New York.
1 The choice of §7702 test affects MEC testing under §7702A. The latter section inherits definitions and calculation rules from the former; they ought to be consistent for each definitional test. An event-driven algorithm for MEC testing is given in Addendum A.
2 MEC testing can be simpler for GPT than for CVAT contracts. It is advantageous to make the necessary-premium and guideline-premium limits identical so that the former need not be calculated separately: then no unnecessary premium can be paid without failing the definitional test[59]. To this end, the server offers two different methods. In order to avoid certain difficulties that might otherwise occur[60], the first method treats every adjustment event as a material change, and therefore applies both the rollover rule and the reduction rule to decrease adjustments[61]. The second method simply assumes that such difficulties do not arise; it treats adjustment events as reductions when they decrease benefits, and as material changes otherwise.
3 Decreases are subject to retrospective seven-pay testing, whether or not an adjustment event has occurred. Premature termination[62] of a QAB is treated as a decrease.
4 A material change resets the “contract year” so that it might not coincide with any policy year. Underwriting procedures should guard against abuse such as a series of trivial changes on successive days[63].
1 The determination of guideline premiums requires a considerable amount of floating point calculations. It is important to minimize the cost of these calculations in the case of systems that administer a great number of contracts, or illustration systems that are judged on their speed.
2 Commutation functions as defined by Eckley [TSA XXXIX, page 19] are used in order to perform the calculations rapidly while following the actual mechanics of a UL contract. This implementation trades perfection for speed by conservatively ignoring the corridor. It appears that this affects only the GLP for DBO B, which could be increased somewhat at a significant expense in complexity and run time.
1 The following monthly effective rates must be specified in
product-specific addenda; they are level within any policy
year.
qc §7702 mortality rate
ic §7702 interest rate, less asset charge if applicable
2 The following parameters are derived from the usual
product specifications.
ig monthly effective death-benefit discount rate for NAAR
mChgPol monthly charges per contract
aChgPol annual charges per contract
mChgSA monthly charges per dollar of specified amount
ChgSALimit maximum specified amount to which mChgSA applies
LoadTgt premium load up to target
LoadExc premium load on excess over target
QabRate total monthly rate for all QABs combined (ADD is the only QAB reflected today)
QabLimit maximum specified amount to which QabRate applies
3 The following variables are already calculated by systems
that use these specifications.
SpecAmt specified amount
TgtPrem target premium
DeathBft death benefit
4 The formulas below handle the general case where all these parameters and variables vary by policy year. The duration subscripts may be disregarded for those that do not.
1 Initialize:
μ = maturity duration = maturity age minus issue age
aD0 = 1.0
2 For each annual duration t in [0,…μ−1]:
f = qct × (1.0 + ict) ÷ (1.0 + igt)
g = 1.0 ÷ (1.0 + f)
q = f × g
i = (ict + igt × f) × g
if(DBOt is B)
i = i − q
end if
v = 1.0 ÷ (1.0 + i)
p = 1.0 − q
vp = v × p
vp12 = vp12
ma = (1.0 − vp12) ÷ (1.0 − vp)
mCt = ma × aDt × v × q
mDt = ma × aDt
aDt+1 = aDt × vp12
3 Variables with no subscript need not be saved across iterations. The variable aDt has a value at the maturity duration so that it can be used to discount the endowment benefit payable at the end of that year; other vector variables are shorter by one.
4 Recalculation after issue can be obviated by calculating Ct and Dt for all durations and storing the results for both DBO A and B. They cannot change after issue unless the mortality or interest basis changes.
1 GSP calculations always use DBO A commutation functions. GLP calculations use commutation functions for the DBO in effect (for ROP, see ¶4/6).
2 Calculate trial values of
GLPt and
GSPt:
Chargest =
mDt × mChgPolt
+ aDt × aChgPolt
+ mDt × mChgSAt × min(SpecAmtt, ChgSALimit)
+ mDt × QabRatet × min(SpecAmtt, QabLimit)
+ mCt × DeathBftt
NetPaymentFactort = aDt × (1.0 − LoadTgtt)
Endowment = aDμ × endowment benefit [¶4/8]
| GLPt = | ( | Endowment + |
μ
∑ i = t |
Chargesi | ) | ÷ |
μ
∑ i = t |
NetPaymentFactori |
| GSPt = | ( | Endowment + |
μ
∑ i = t |
Chargesi | ) | ÷ | NetPaymentFactort |
3 If GLPt ≤ TgtPremt, then accept the trial value. Otherwise, recalculate it using the same formula, but with the modified Charges and NetPaymentFactor that follow.
4 If GSPt ≤ TgtPremt, then accept the trial value. Otherwise, recalculate it using the same formula, but with the modified Charges and NetPaymentFactor that follow.
5 Modified values to be used only for premiums that exceed
target:
Chargest =
Chargest [as calculated above]
+ TgtPremt × (LoadTgtt − LoadExct) × aDt
NetPaymentFactort = (1.0 − LoadExct) × aDt
6 The recalculation with modified values is performed only for premiums that exceed the target. For example, if GLPt < TgtPremt < GSPt, then only GSPt is recalculated.
1 Although the statute uses the term “corridor” only with respect to the GPT, I use it in a parallel sense to indicate the minimum death benefit on a CVAT contract.
2 I use “contract” to refer to a policy or certificate of insurance. But I say “policy year” because “contract year” is a term of art in §7702A that need not equal policy year.
3 For ease of reference, acronyms are spelled out here
unless they’re obvious (e.g., IRS, FIFO, lmi).
At its first occurrence, each is spelled out and, where
necessary, defined.
ALB age last birthday
ANB age nearest birthday
AV account value
CSV cash surrender value, defined to disregard loans and surrender charges
CVAT cash value accumulation test
DB death benefit: the amount payable by reason of death on the date of death
DBO death benefit option: A (level), B (increasing), or ROP
DCV deemed cash value for CVAT MEC testing
GLP guideline level premium
GPT guideline premium test
GSP guideline single premium
LDB least death benefit in the most recent seven-year test period
M&E mortality and expense charge (an interest spread on variable contracts)
MEC modified endowment contract
NAAR net amount at risk
NSP net single premium used in MEC testing and in setting the CVAT corridor
QAB qualified additional benefit
ROP return of premium (death benefit option)
SA specified amount
TSA Transactions of the Society of Actuaries
VLR variable loan rate
“Contract year” does not always mean policy year. Policy year is one plus the number of full years since issue. Contract year is one plus the number of full years since the last material change, or since issue if there has been no material change. Contract year restarts at one on the date of any material change, which may not be a policy anniversary.
Least death benefit (LDB) is the lowest death benefit (including any QABs) in the most recent seven-year test period[64].
Other terms (e.g., CSV) are as defined in our GPT specifications.
Calculate the seven-pay premium as
7Px × DB + amortized QAB charges[65]
using the initial DB defined in our GPT
specifications[66].
Round it down if at all.
Start a new seven-year test period on the day of issue.
Record the initial death benefit as the LDB.
For CVAT contracts, set DCV to zero before recognizing the initial premium or any §1035 exchange.
A new life insurance contract is a MEC if it is issued as a §1035 exchange from a MEC.
A §1035 exchange, while technically a material change,
receives special handling.
Process all §1035 exchanges before any other payment.
At issue, treat the anticipated
exchange amount, net of all premium-based loads including
any premium-tax load, as a non-premium increment to CSV.
For CVAT contracts, increase DCV by this increment.
Do not test this increment against the seven-pay premium.
Recalculate the seven-pay premium with the formula
7Px+t × (DB − CSV ÷ Ax+t) + amortized QAB charges
used for material changes, with t naturally equal to zero.
Test for unnecessary
premium[67].
When the exchanged funds are ultimately received, reissue the policy reflecting the actual §1035 amount[68], and retest from the original issue date.
During every seven-year test period, test each “amount paid”[69] (except any §1035 amount) against the seven-pay premium limit. The limit, measured on each day during that period, is cumulative seven-pay premiums minus cumulative amounts paid. If any amount paid exceeds the limit, then the contract becomes a MEC. Otherwise, test for unnecessary premium.
Our GPT specifications ensure that no unnecessary premium can ever be paid. Skip the rest of this section for GPT contracts. But perform these calculations for CVAT contracts:
NSP is the net single premium rate for the attained age times LDB, plus the present value of QAB charges, reflecting the benefit amount of each QAB in the first contract year. Round it down if at all.
Necessary premium is NSP adjusted for the cash value.
Normally the adjustment equals DCV.
Whenever CSV is lower than DCV, use CSV
instead[70],
but don’t change DCV.
If the adjustment is less than zero, then use zero instead.
Thus:
Adjustment = max[0, min(DCV, CSV)]
Net necessary premium is NSP minus the adjustment for cash value.
Round it down if at all.
If the result is less than zero, then use zero instead.
Thus:
Net = max(0, NSP - Adjustment)
Gross necessary premium is net necessary premium adjusted
for premium-based load (PL) including any premium-tax load.
If PL is a scalar:
Gross = Net ÷ (1 − PL)
Sometimes PL is tiered such that one load PLTarget applies up
to target premium and a different load PLExcess on any excess.
In that case:
Gross = [Net + Breakpoint × (PLTarget − PLExcess)] ÷ (1 − PLExcess)
In either case, round the result down if at all.
Accept payments up to the gross necessary premium. Any remaining payment is unnecessary premium. If there is any unnecessary premium, process a material change before accepting it. The material change happens after the necessary premium was applied to CSV and DB was updated if necessary to reflect any corridor, ROP, or similar increase due to necessary premium.
Update DCV for CVAT contracts. Accumulate it like AV, but using §7702 assumptions. Reflect the actual death benefit option. Use current charges and loads[71]. Use §7702 interest and mortality. Reflect all transactions, but ignore loans[72]. Disregard charges for nonqualified additional benefits[73]. Add any extra amounts payable on surrender such as refundable sales loads but do not accumulate those amounts at interest. Round it up if at all. Whenever it is negative, set it to zero. Ignore this step for GPT contracts, which have no DCV.
Process material changes as of the very day they take effect, and not as of any other date[74].
Start a new seven-year test period on the day of the material change, even if that is not a monthiversary. §7702A treats the contract as though it were issued on that day.
For CVAT contracts, set DCV equal to CSV immediately prior to the material change.
Calculate a new seven-pay
premium[75]
as
7Px+t × (DB − CSV ÷ Ax+t) + amortized QAB charges
where CSV reflects all necessary premium paid, but no
unnecessary premium, and DB reflects the corridor factor
times this CSV and any ROP increase due to necessary
premium, as well as any increases.
Round it down if at all.
If the new seven-pay premium is negative, set it to zero. In this case, the contract does not become a MEC, but no premium can be paid for seven years[76].
Record the values of CSV and DB for use in handling any later decrease.
Test any unnecessary premium against the new seven-pay premium limit. If it exceeds the limit, then the contract becomes a MEC.
Examine the death benefit and every QAB benefit each day to see whether they decrease for any reason. If neither the death benefit nor any QAB benefit decreases below the level originally assumed at the beginning of the most recent seven-year test period, then skip this step.
For individual life and first-to-die contracts, skip this step if the most recent seven-year test period has already ended. But for survivorship contracts, proceed even if that period has ended.
Update LDB and each QAB’s benefit amount to reflect decreases only[77]. If one of these items increases while another one decreases, then process a material change and skip the rest of this step[78].
Calculate a new seven-pay premium as
7Px × (LDB − CSV ÷ Ax) + amortized QAB charges
LDB reflects the decrease.
CSV is as of the beginning of the
first contract year, and may often be zero.
The stream of QAB charges begins in the first contract year, and
its present value is as of the first contract year.
If the decrease occurs within the first seven policy years, then
CSV reflects any §1035 exchange.
If the new seven-pay premium is negative, set it to zero.
Round it down if at all.
Using the new seven-pay premium, retest each payment made throughout the most recent seven-year test period. Never look back before the latest material change. Stop testing at the end of the seven-year period, even for survivorship contracts. If premium exceeds the new seven-pay limit at any time during the retrospective seven-year test period, then the contract becomes a MEC as of the current date[79].
Notify all appropriate parties immediately. The admin system ought to have prevented an inadvertent MEC (¶6/3).
All or a portion of any payments can be returned within sixty days after the end of the contract year in which they were paid. It may be possible to return enough to prevent a MEC. Subtract the payments returned, without interest, from amounts paid, and redo the §7702A calculations using the revised amount paid. Return the payments to the owner with taxable interest.
Unless it is cured this way, once a contract is a MEC, it is always a MEC. IRS can cure it, but normally will not except under a formal remediation program. There is no other way to cure a MEC. For instance, a retroactive change in specified amount does not correct a MEC.
Whenever a contract becomes a MEC, apply less-favorable MEC taxation to all distributions made within the preceding two years (¶5/11), even though they were not so taxed when they occurred.
Correction of a misstatement of age or gender is a material change. Consult the actuarial department when this occurs. An irremediable MEC may result[80].
Removing a substandard rating is a material change if the rating was reflected in §7702A calculations. So is a change from smoker to nonsmoker, if that distinction was reflected in §7702A calculations[81].
Any fundamental modification such as a change in the §7702 mortality or interest basis is a material change. For substitution of insured, see ¶5/14.
Increasing SA whenever a MEC would otherwise occur is a feasible if somewhat impractical illustration strategy when every SA increase is recognized as a material change. But when payment of unnecessary premium is the only material-change trigger, that strategy no longer works very well[82]. Unnecessary premium paid during a seven-year test period typically violates the seven-pay limit, producing a MEC. This strategy is inefficient anyway: it is generally much better to purchase a new policy.
Another common strategy reduces each successive premium as required to satisfy the seven-pay test and avoid a MEC. But paying the full input premium might increase AV when the reduced premium would let AV decrease. Thus, reducing the premium on an option 2 contract can cause the DB to decrease and trigger retrospective retesting that could produce a MEC. That problem might be avoided by manually increasing SA whenever necessary to avoid a decrease that would produce a MEC.
Furthermore, reducing each successive premium to the seven-pay limit still allows unnecessary premium to be paid outside a seven-year test period. The ensuing series of material changes might produce a stream of payments totaling less than could be paid by avoiding unnecessary premium.
nPx+t is the net n-pay premium per dollar of death benefit for a life issued at age x at duration t, reflecting actual monthiversary mechanics.
Ax+t is the net single premium per dollar of death benefit for a life issued at age x at duration t, reflecting actual monthiversary mechanics.
x is the insurance age as of the beginning of the first contract year (which is not necessarily the first policy year).
t is the number of full contract years since the beginning of the most recent seven-year test period. For example, if a contract was materially changed two years and three months ago, then t is two today. Wherever t occurs in this addendum’s formulas, it always happens to have the value zero, because the contract is deemed to be reissued upon material change; it is written only to emphasize that the insurance age as of the first contract year is meant.
1 This section discusses some technical issues and is intended only for actuaries who desire a more detailed treatment. Others may skip it.
2 The power of Eckley’s UL commutation technique is inadequately appreciated. It permits the calculation of any isolated annual value without iteration in many important cases[83]. Account values can be conveniently determined by applying the prospective formula for terminal reserves.
3 I have calculated the account values associated with guideline premiums (akin to a guaranteed maturity fund) in lmi, computing values iteratively each month, and compared them to results from a spreadsheet that uses monthly commutation functions. I selected issue age 20 so that the calculations would span many years, and ran both systems to endowment at age 100 with the same assumptions and no rounding. The absolute value of the largest relative error in AV in any year was 0.00000000012, or one hundred twenty parts per trillion. This discrepancy arises because of the way floating point numbers are stored and manipulated by computer hardware.
4 One much coarser method uses classic annual curtate commutation functions made approximately semicontinuous by an i ⁄ δ adjustment. It is not appropriate to apply such an adjustment to the formulas in this monograph, because they already reflect the monthiversary mechanics of a typical UL contract (but see ¶B/9 for an adjustment to q that is appropriate). At any rate, monthly functions are nearly equal to continuous: for i = 4%,
| i ⁄ i(n) | n | |
| 1.009901951 | 2 | semiannual |
| 1.014877439 | 4 | quarterly |
| 1.018203509 | 12 | monthly |
| 1.019814474 | 365 | daily |
| 1.019869268 | ∞ | continuous |
5 It would be a good idea to work the examples in Desrochers’s paper (TSA XL) and compare and contrast our results with his.
6 We ought to analyze the conservatism that results from ignoring the corridor in guideline premium calculations, particularly with respect to option B.
7 When calculating guideline premiums two ways—above and below target premium—some practitioners always perform both calculations and assume that the lower is the one that should be used. It would be interesting to know whether this is always correct. But it does not seem to suggest any way to speed up our program.
8 Here is a proof of the assertion (¶7/8) that netting an account value load against the §7702 interest rate is a conservative alternative to reflecting the AV load exactly. Assuming that the account value load is deducted before other monthly charges, we have two cases:
Case 1: DBO A
Eckley’s equation (1):
(1) [(0V+P)−Q(1/(1+ig)−(0V+P))](1+ic)=1V
Eckley’s equation rewritten with an AV load, r:
(1a) [(0V+P)(1−r)−Q(1/(1+ig)−(0V+P)(1−r))](1+ic)=1V
Eckley’s equation rewritten with an interest decrement, r:
(1b) [(0V+P)−Q(1/(1+ig)−(0V+P))](1+ic−r)=1V
Rewrite (1a) and (1b) in terms of (1):
(1a) = (1) − r(0V+P)(1+Q)(1+ic)
(1b) = (1) − r[(0V+P)(1+Q)−Q/(1+ig)]
AV Load ≥ interest decrement if Q ≥ 0, ig > −1, ic ≥ 0
Case 2: DBO B
Eckley’s equation (14):
(14) [(0V+P)−Q((1+0V+P)/(1+ig)−(0V+P))](1+ic)=1V
Eckley’s equation rewritten with an AV Load, r:
(14a) [(0V+P)(1−r)−Q((1+(0V+P)(1−r))/(1+ig)−(0V+P)(1−r))](1+ic)=1V
Eckley’s equation rewritten with an interest decrement, r:
(14b) [(0V+P)−Q((1+0V+P)/(1+ig)−(0V+P))](1+ic−r)=1V
Rewrite (14a) and (14b) in terms of (14):
(14a) = (14) − r[(0V+P)−Q(0V+P)/(1+ig)+Q(0V+P)](1+ic)
(14b) = (14) − r[(0V+P)−Q(1+0V+P)/(1+ig)+Q(0V+P)]
AV Load ≥ interest decrement if Q ≥ 0, ig > −1, ic ≥ 0
Q ≥ 0 because cost of insurance rates are never negative.
ig > −1 because guaranteed interest always exceeds −100%.
ic ≥ 0 because the §7702 interest rate is never negative.
It would of course be possible to generalize Eckley’s work to reflect the load exactly, gaining a slight advantage at the cost of much extra work. The algebra would be complicated enough that questions might arise as to its rigorous validity. We prefer to use Eckley’s work as he presents it, pointing out that we can accurately reproduce the numerical examples in his paper.
If instead the account value load is deducted after other
monthly charges, then the validity of the assertion can seen
immediately: both Eckley’s equations (1) and (14) are of the
form
(AV+P−deductions)(1+ic)=1V
which, for a load s deducted after other charges, becomes
(AV+P−deductions)(1+ic)(1−s)=1V
and it is conservative to apply the net rate (1+ic−s)
because
1+ic−s < 1+ic−s−(s×ic) = (1+ic)(1−s)
where s and ic are both necessarily positive.
When the load (whether deducted before or after other monthly charges) is netted against the interest rate in this fashion, the annual effective rates are simply subtracted. It is important to understand that we are not deducting the spread from the statutory rate to produce a lower §7702 interest rate. Rather, the §7702 interest rate remains what it is, and we are netting the AV load against it as a conservative mathematical simplification.
9 Tabular annual safe-harbor (¶8/2) mortality rates (q) may
be converted as follows for use as monthly effective rates
in ¶14.1/1.
DB = death benefit at beginning of month
E = expense charges deducted at beginning of month
COI = cost-of-insurance deduction
AV = account value at beginning of month, before deduction of E or COI
i = annual effective death-benefit discount rate for NAAR calculation (¶7/12)
q = annual mortality rate to be converted to monthly
mq = 1 − (1 − q)1⁄12
mv = 1 ÷ (1 + i)1⁄12
Deducting the COI charge at the beginning of the month increases
the amount actually at risk, suggesting an equation that has the
COI term on both sides:
COI = [DB × mv − (AV − E − COI)] × mq
Rearranging:
COI = [DB × mv − (AV − E)] × mq + COI × mq
COI × (1 − mq) = [DB × mv − (AV − E)] × mq
COI = [DB × mv − (AV − E)] × mq ÷ (1 − mq)
suggests the definitions:
NAAR = DB × mv − (AV − E)
which is what contracts normally specify, and:
mortality-charge rate = min[limit, mq ÷ (1 − mq)]
= min[limit, (1 − (1 − q)1⁄12) ÷ (1 − q)]1⁄12
for some limiting value such as 1 ⁄ 12.
Parameters for lmi’s “sample” product:
qc §7702 mortality rate:
1980 CSO converted as described in ¶B/9,
limited to 1⁄12
ic §7702 interest rate:
0.00327373978219891 in all years for GLP
0.00486755056534305 in all years for GSP
ig death-benefit discount rate: same as ic
[1] See, for example, Revenue Ruling 91-17.
[2] The reasonableness of a termination dividend depends on historical practice. The legislative history cites New York’s rules as an example and notes that termination dividends have historically been modest: DEFRA Blue Book, page 647. In the 1990s, one company marketed a survivorship product with a termination dividend on the order of $400 per thousand; that would obviously be part of CSV for §7702 and §7702A.
[3] Because, for instance, interest on dividend accumulations is taxed when it accrues. But the cash value of paid-up additions is included in CSV.
[4] ¶4/1.
[5] Except that a GPT contract can switch to the CVAT when a nonforfeiture option is elected. This may make a reduced paid-up benefit easier to implement, although it may increase the amount of that benefit.
[6] This is so even if the new contract is issued in exchange for several old contracts, only one of which is a MEC.
[7] DEFRA Blue Book, page 651. But see footnote 57 on page 655: “The use of the date on the policy would not be considered the date of issue if the period between the date of application and the date on which the policy is actually placed in force is substantially longer than under the company’s usual business practice.”
[8] While the legislative history refers to the end of the mortality table, the statute literally says age 100, so age 100 applies even to the 2001 CSO table, which extends to age 121.
[9] See footnotes to ¶1/2.
[10] With respect to MEC testing only; §7702A(c)(2)(A) has no effect on the definitional test. For example, an experience-rated group policy might apportion its experience reserve across certificates according to their most recent yearly COI deductions, which affects the amount of the deductions; but a decrease does not change historical deductions. Or consider a contract with a sales load refund equal to some percentage of commissions. That calculation would be different if the policy were reissued, even though commissions aren’t drawn back when §7702A(c)(2)(A) deems the contract to be reissued at a lower benefit. Therefore, a reduction during the sales load refund period may create a retrospective MEC.
[11] §7702(d)(2).
[12] Table-driven systems normally disregard this adjustment, as lmi currently does, yet it is explicitly permitted by the DEFRA Blue Book, page 649: “Finally, the amount of any qualified additional benefits will not be taken into account in determining the net single premium. However, the charge stated in the contract for the qualified additional benefit will be treated as a future benefit, thereby increasing the cash value limitation by the discounted value of that charge.”
[13] §7702(e)(1)(D) permits treating a UL contract’s cash value at maturity as an endowment benefit, up to the “least amount payable as a death benefit at any time under the contract”. This means that NSP is not simply Ax, but rather Ax:angle(100−x): not Mx ⁄ Dx, but rather (Mx + D100) ⁄ Dx. The purpose of the computational rule is to forbid any coefficient higher than unity on the D100 term. Thus, IRS Notice 2009-47, section 3.02(b), says that NSP calculations under the 2001 CSO “would assume an endowment on the date the insured attains age 100”. The DEFRA Blue Book, at page 652, says that 7702(e)(1)(B) “will generally prevent contracts endowing at face value before age 95 from qualifying as life insurance”; it specifically does not rule out contracts that endow later for an amount no higher than “face value”. For contracts that have undergone benefit changes, this amount must be adjusted. Page 653 says that for CVAT contracts, it “must be computed treating the date of change, in effect, as a new date of issue”; for GPT contracts, “the date of change for increased benefits should be treated as a new date only with respect to the changed portion of the contract”—i.e., B in the A + B − C formula—so that A, B, and C all apply their respective benefit amounts to D100 as they do to Mx. This is what DesRochers’s paper in TSA XL does, e.g. on pages 240 and 262–263. The endowment amount is limited, as in ¶3/1, to the specified amount. (It’s okay if ¶5/5 happens to overstate the endowment amount for C, because that is a subtractive term.)
[14] Some contracts change the death benefit to what would have been purchased by the actual premiums paid at the correct age and gender: the intention is to satisfy the Code, although the required recalculation may be difficult. Other contracts take the ratio of what the most recent monthly deduction should have been to what it actually was, and multiply the death benefit by that; the intention is to simplify the calculation, and a likely result is an irremediably failed contract. Many contracts have an overriding provision empowering the company to do anything necessary to reform the contract so as to comply with §7702 (but less often §7702A).
[15] This appears unlikely but is not impossible. For instance, a sufficiently large policy fee might cause the GSP to exceed the reciprocal of the corridor factor. Even so, we would ignore the corridor increase in determining old and new DB for purposes of this paragraph.
[16] The legislative history supports not treating any such DB increase as an adjustment event anyway. DEFRA Blue Book, page 654: “no adjustment shall be made if the change occurs automatically, for example, a change due to…the payment of guideline premiums or changes initiated by the company.”
[17] Forceouts are not limited to the tax basis: income can be forced out as well. Thus, §7702(f)(1)(A) premiums paid can become negative. §7702(c)(1) requires that “the sum of premiums paid…does not at any time exceed the guideline premium limitation as of such time”, so premiums paid must be negative whenever the guideline limit is.
[18] 730 days; 731 if an intercalary day is comprised. Some would look back only to the beginning of the current calendar year, arguing that regulations implementing the two-year lookback of §7702(f)(7)(E) and §7702A(d)(2) have never been issued.
[19] OBRA House Report, page 1438: “an increase in the charge for a qualified additional benefit is not a material change…. An addition of, or an increase in, a qualified additional benefit, however, is a material change”. The legislative history does not address decreases in a scale of QAB charges. Ignoring them is consonant with the apparent intent.
[20] 128 Cong. Rec. S10943, 1982-08-19: “Such adjustments are only to be made in two situations: First, if the change represents a previously scheduled benefit increase that was not reflected in the guideline premiums because of the so-called computational rules; or second, if the change is initiated by the policy over [owner] to alter the amount or pattern of the benefits.” See also OBRA House Report, page 1438.
[21] “If a life insurance policy provides the policyholder with an option to change the insured, the exercise of the option is a sale or other disposition under section 1001 of the Code and section 1035 does not apply…. Section 1.1035-1 of the regulations expressly excludes from the application of section 1035 exchanges of policies that do not relate to the same insured and thus prevents policy owners from deferring indefinitely recognition of gain with respect to the policy value.” Rev. Rul. 90-109.
[22] Some might treat a substitution of insured as both an adjustment event and a material change, relying on the DEFRA Blue Book, page 656: “A substitution of insured…pursuant to a binding obligation will not be considered to create a new contract”. However, IRS would more likely enforce Rev. Rul. 90-109.
[23] Withdrawals are generally nontaxable up to basis under §72(e) unless the contract is a MEC. But see §7702(f)(7)(B–E). The server assumes that its client distinguishes taxable from nontaxable withdrawals.
[24] PLR 9106050: “A waiver of the monthly deduction under the disability waiver rider does not affect Taxpayer’s ‘investment in the contract’ under section 72(e)(6) of the Code or the ‘premiums paid’ under section 7702(f)(1)(A) for the Contract.” Andrew Strelka’s “Taxing the Disabled” in the Spring 2007 issue of Richmond Journal of Law and the Public Interest discusses this matter in depth.
[25] §7702(f)(1)(B).
[26] §7702A(e)(1)(B).
[27] It may be impossible to avoid a MEC in extraordinary circumstances (e.g., misstatement of age) or pathological cases (e.g., an investment return approaching −100% on a VUL contract with a negative seven-pay premium).
[28] The DEFRA Blue Book, page 654, says that such a “distribution will be taxable to the policyholder as ordinary income to the extent there is income in the contract.” The 1986 statute changed that (§7702(f)(7)(B–E)). Often, a forceout is a tax-free return of premium.
[29] Convergence of the resulting series could be hastened by conservatively adding one cent to each forceout iterand. Alternatively, the geometric series theorem could be applied. The server does not do this because it is desired that the client perform the forceout calculation.
[30] We call these “involuntary withdrawals” to distinguish them from the “forceouts” of ¶6/4. The concepts are somewhat similar, but their treatment under §7702(f)(7)(B–E) differs.
[31] §7702(f)(6).
[32] The monthly approach is recommended, since interest credited during the year cannot be known in advance.
[33] Prevalent industry opinion holds that statutory interest should be treated as an aspect of plancode: determined on the issue date, and never changing thereafter (except in the case of a "deemed exchange": see Adney et al., "They Go Bump in the Night", Society of Actuaries, Taxing Times Supplement May 2012). This is the way mortality is conventionally treated. For example, the interest and mortality basis of a 2001 CSO contract with a 4% CVAT rate would not change to 2017 CSO or 2% due to an adjustment event or material change in 2021.
[34] DEFRA Blue Book, page 648.
[35] DEFRA Blue Book, page 649.
[36] For example, §7702A(c)(3)(A)(i).
[37] §7702(c)(3)(D)(i).
[38] §7702(c)(3)(B)(ii).
[39] This would not be a “reasonable” approximation: DEFRA Blue Book, page 653.
[40] In all likelihood, unit values reflect quarterly fund expenses that are applied to average assets in a way that varies from one fund to the next, and such expenses aren’t specified in the contract (¶7/6).
[41] And also in Norris cases, where state law must be read to conform to Title VII of the Civil Rights Act of 1964.
[42] Guaranteed mortality lower than the safe harbor might raise state approval issues, for instance in the area of nonforfeiture.
[43] Public Law 100-647 of 1988 (TAMRA) amended §7702(c)(3)(B)(ii), which formerly permitted “any charges…specified in the contract”, to allow only “reasonable charges…reasonably expected to be actually paid”.
[44] §7702(b)(2)(B).
[45] Asset-tiered charges could be reflected exactly at the cost of extra complexity. Ignoring them is safe even under the old-fashioned interpretation that any change in current charges is an adjustment event.
[46] DEFRA Blue Book, page 649.
[47] This exhaustive approach is not necessary if a formulaic approach suffices to cover every possible case.
[48] Alternatively, Desrochers [Transactions of the Society of Actuaries (TSA) XL, page 209] suggests setting a load equal to the difference between the GLP and the nonforfeiture premium. But the rules on reasonable charges that came after that paper was published rendered this approach of little or no practical applicability.
[49] For instance, by using an approximate calculation such as a table lookup, or by using a different mortality table.
[50] A MEC testing server will need to know the maximum funding duration and maximum benefit duration for each QAB.
[51] Neither is there any such consequence if a QAB is terminated, or its benefits increased or decreased, after the funding period, as long as the current charges remain zero. One could imagine a QAB term rider, fundable over a shorter term than its benefits last, whose benefit is conditionally reduced whenever poor investment performance would cause NAAR to increase beyond reinsurance capacity, in a context similar to ¶6/5.
[52] §7702(f)(5)(C)(ii).
[53] Charges for a non-QAB are treated as any other amount deducted from a contract: they’re potentially taxable. Consider a single-premium life contract with a long term care benefit funded by withdrawals. The withdrawals generate taxable income, just as if the base policy and the non-QAB were separate entities.
[54] A non-qualified additional benefit is not a benefit at all under §7702(f)(5)(C)(i).
[55] Perhaps this means that the endowment benefit (¶4/7) could include the term amount for §7702A only, but we disregard that reading.
[56] TEFRA Blue Book, page 371: “the guideline premiums are to be adjusted…if a qualified additional benefit ceases for any reason, including the death of an individual (such as the insured’s spouse) insured thereunder, this is considered a change in benefits requiring an adjustment of the guideline premiums.” Thus, if a family member covered under a QAB dies within the first seven contract years, and the full seven-pay premium reflecting the QAB has been paid each year, then the death triggers a §7702A(c)(2)(A) decrease and a retrospective MEC.
[57] Presumably the QAB was not funded past its expiry date, so the quantities B and C would cancel.
[58] PLR 9513015, PLR 9519023, PLR 9741046.
[59] Thus, for instance, increases due to payments under the ROP death benefit option are not material changes because of the necessary premium exception. Cf. ¶5/8.
[60] Treating as material changes only adjustment events that increase the guideline limit can allow payment of unnecessary premium under a GPT contract, for instance if endowment benefits are reflected, or if guideline premiums are not strictly nondecreasing by duration for a given issue age, as might occur with high expense charges applicable in the first year only.
Furthermore, consider a change from an increasing to a level death benefit option that is not accompanied by any change in death benefit. The guideline limit decreases. Page 654 of the DEFRA Blue Book implies that this is as much a material change as any other §1035 exchange. “Further, under prior and present law, for the purpose of the adjustment rules, any change in the terms of a contract that reduces the future benefits under the contract will be treated as an exchange of contracts (under sec. 1035)…. This provision was intended to apply specifically to situations in which a policyholder changes from a future benefits pattern taken into account under the computational provision for policies with limited increases in death benefits to a future benefit of a level amount (even if at the time of change the amount of death benefit is not reduced).” (The enactment of §7702(f)(7)(B–E) superseded this part of the legislative history with respect only to certain changes during the first fifteen years.)
[61] Applying both rules to decrease adjustments is conservatively less advantageous to the taxpayer, but seems so extraordinary on the face of it that some elaboration is in order. The rollover rule must be applied because that is the way this method keeps the necessary-premium and definitional guidelines synchronized. The reduction rule must be applied because recognizing a material change does not satisfy §7702A(c)(2)(A). We handle the reduction first, using its seven-pay premium for retrospective testing; then process the material change, using its seven-pay premium for prospective testing.
[62] Other than on the QAB’s normal expiry date: see ¶11/6.
[63] To prevent systematic abuse, a pro-rata portion of the amount paid in the old contract year could be deducted from the seven-pay limit in the first new contract year (cf. ¶5/10). For example, suppose a $1000 premium is paid on anniversary, and then the seven-pay premium becomes $2000 due to a material change nine months later. Only nine-twelfths of the $1000 payment is attributable to the completed portion of the old contract year, so the other $250 = 1000 × [1 − (9 ÷ 12)] must be attributed to the new contract year about to begin. Thus, for the twelve months following the material change, premiums are limited to $2000 − 250 = 1750. If this adjusted limit is negative, then the premium limit for that year is zero, consistent with page 1439 of the OBRA House Report; but no forceout is required. Any such rule would be added to the procedures for CVAT MEC testing as well.
[64] §7702A(c)(3)(B)(i) says “the lowest level of the death benefit and qualified additional benefits payable in the 1st 7 contract years”. But §7702(c)(2)(A) forces that to equal the level in the first contract year. As long as the benefits are properly updated, LDB is always the benefit as of the beginning of the first contract year.
[65] Amortized QAB charges means the present value of QAB charges divided by a seven-year annuity-due factor, with the annuity period duly reduced if it would otherwise extend past maturity.
[66] ¶3/1–2
[67] If the net 1035 amount exceeds the necessary premium, we declare the contract a MEC, though some might hold that it is not—reasoning, perhaps, that the CVAT corridor saves it. Cf. the general recommendation in ¶3/1.
[68] ¶3/3
[69] ¶6/1–2
[70] TAMRA Conference Report, page 105, footnote 3.
[71] ¶9/1
[72] TAMRA Conference Report, footnote 3. Policy loans affect AV in that loaned and unloaned funds generally earn different rates of interest, but that has no effect because the DCV interest rate is prescribed by statute.
[73] ¶11/3.
[74] TAMRA Conference Report, page 98: “as of the date that the material change takes effect”. For instance, it is not permissible to delay recognition of a material change to the next monthiversary or anniversary.
[75] If the restriction suggested in a footnote to ¶13/4 is desired, apply it here. I.e., in the first contract year only, reduce the new seven-pay premium by a pro-rata portion of the amount paid in the partially-completed former contract year.
[76] OBRA House Report, page 1439.
[77] This means that decreases occurring outside the seven-year test period on a CVAT contract are ignored.
[78] It is unduly harsh to process the decrease while ignoring the increase, even though the necessary premium exception may permit that. In order to recognize the increase, a material change must be declared.
[79] Another interpretation is that it becomes a retrospective MEC as of the (past) failure date; the 7702A(d) anticipation-of-failure rule might also apply. At any rate, it is most likely too late to refund the offending premium.
[80] If the misstatement is discovered prior to death, the premium or mortality charge might be adjusted instead of the benefits, and some would hold that this is not a material change.
[81] Some would hold that underwriting liberalizations are not material changes.
[82] Insurers could manually administer cases for which it behooves them to recognize a material change.
[83] Irregular premium, withdrawal, and SA patterns are easily handled. Option changes and loans can be handled with more work. Calculations that depend on other calculated values, such as waiver of monthly deductions, corridor death benefits, banded COI rates, and tiered loads are considerably more difficult, but can be handled by iterating vector commutation-function equations. Exact rounding of intermediate values cannot be modeled at all.
Copyright © 2009, 2010, 2011, 2012, 2013, 2014, 2015, 2016, 2017, 2018, 2019, 2020, 2021, 2022, 2023 Gregory W. Chicares. This program, including its documentation, is free software. Read the terms under which you can redistribute and modify it.
Maintained by Gregory W. Chicares. The latest version of this file can be found at the lmi website.