The insurance company�s cost of providing annuity benefits is based on the probability of survival rather than the probability of death. In itself this fact would seem to have no greater significance than that the insurance company actuaries, in computing premiums for annuities, have to refer to the actuarial probabilities of survival rather than probabilities of death. As a matter of fact, however, writing annuities poses a unique set of actuarial problems.

In the first place, insurance companies have found that the mortality among persons who purchase annuities tends to be lower, age for age, than that of people who purchase life insurance. There may be several reasons for this, including the peace of mind that comes with an assured income for life, but certainly one of the most important is the selection practiced against the company. Individuals who know that they have serious health impairments rarely, if ever, purchase annuities. In fact, many persons contemplating purchasing immediate annuities subject themselves to a thorough medical examination to make sure that they have no serious impairments before committing their capital to annuities. On the other hand, people who know or suspect that they have an impairment usually seek life insurance. Whatever its origin, the mortality difference between life insurance insureds and annuitants is so substantial that special annuity mortality tables must be used for the calculation of annuity premiums.⁶

Second, the trend toward lower mortality that has been such a favorable development with respect to life insurance has been very unfavorable with respect to annuities. Many annuity contracts run for 60 to 75 years, counting the accumulation period, and rates that were adequate at the time the contract was issued may, with the continued increase in longevity, prove inadequate over the years. All mortality tables, of course, contain a safety margin�which, for life insurance mortality tables, means higher death rates than those likely to be experienced, and for annuity mortality tables, lower rates of mortality than anticipated. While a long-run decline in mortality rates increases the safety margin in life insurance mortality tables, it shrinks the margin in annuity mortality tables, sometimes to the point of extinction. Therefore an annuity mortality table that accurately reflects the mortality among annuitants at the time it was compiled gradually becomes obsolete and eventually overstates the expected mortality.

Finally, a high percentage of annuitants are women, who as a group enjoy greater longevity than men, which has intensified the first two factors mentioned above. It also forced companies to introduce a rate differential between male and female annuitants long before a rate differential based on sex was applied to the sale of life insurance policies. However, court decisions have required insurers to base some group annuity contracts on unisex mortality rates.

Life insurance companies cope with these problems or complications in various ways. In the first place, they compute annuity considerations on the basis of mortality tables that reflect the annuitants� lower mortality. A number of annuity tables have been constructed and used since the 1937 Standard Annuity Table that was in common use until the 1950s. First, the Annuity Table for 1949, or some modification thereof, was widely used for writing individual annuities, but it has been supplanted by three newer tables: the 1955 American Annuity Table, then the 1971 Individual Annuity Table, and currently the 1983 Individual Annuity Table on which all annuity calculations in this book are based.

For many years companies dealt with the specific problem associated with the continuing decline in mortality rates among annuitants by using age setbacks. In other words, a person was assumed to be one, 2, or 3 years younger than his or her actual age. Thus a person who was actually 65 and had a life expectancy of an individual that age was presumed for the purpose of premium calculations to have the life expectancy of a person aged 64, 63, or even 62, thereby increasing the premium for a given amount of income. Ages for females were usually set back 4 or 5 five years in addition to the setback for males in recognition of the sex differential in mortality. Thus if male ages were set back one year, female ages were set back as much as 6 years. If the reduction in mortality had been reflected equally at all ages, the setback technique could have been utilized indefinitely without serious distortion of the equities among annuitants at different ages. However, the reduction is not at a uniform rate at various ages, which has definitely limited the efficacy of the setback technique.

Ironically the 1983 Individual Annuity Table was derived from separate male and female experience and has gender-distinct probabilities even though legal and social events have prompted many insurers to base their annuity products on unisex mortality. Comparison of the male and female mortality rates indicates that the practice of deriving female tables using an age setback of male tables as the only adjustment was a very crude approach to a very complex relationship. The practice should have entailed varying setbacks at each age to accurately adjust male mortality rates to represent longer-lived females.

The most recent approach to the "problem" of declining mortality is an annuity table that, by means of projection factors, attempts to forecast and make suitable adjustments for future reductions in mortality rates. For example, the Annuity Tables for 1949, 1955, 1971, and 1983 all contain a set of projection factors that can be used to adjust the mortality assumptions for all ages from year to year or, in lieu of that, to project the basic rates of mortality to some future date. The projections make allowances for anticipated future reductions in mortality.

Historically insurance companies attempted to hedge future improvement in annuitant mortality by using an unrealistically low interest assumption in the premium formula. The rates were substantially lower than those used in the calculation of life insurance premiums. The effectiveness of this technique can be judged by the fact that an interest margin of .25 percentage point (25 basis points) is capable of absorbing a general reduction in mortality of 6 or 7 percent. Intensified competition among insurance companies and between insurance companies and investment media, however, has caused companies to adopt interest assumptions much closer to the level of their actual investment earnings. Considerations for individual deferred annuities are generally being computed today on the basis of interest assumptions running from 2.5 to 5.5 percent, while immediate annuities may be priced on the basis of slightly higher interest assumptions.

A final approach to adjusting annuity mortality for anticipated future increases in life expectancy is to compute the premiums (or considerations) on a participating basis, which permits conservative assumptions (safety margins) with respect to all factors entering into the computations. Annual-premium annuities issued by mutual companies are almost invariably participating during the accumulation period and may be participating on some basis during the liquidation period. Some stock companies also issue annuities that are participating during the accumulation period. Single-premium immediate annuities, whether written by mutual or stock companies, are usually not participating.