Experts generally agree that a system of surplus distribution should, to the extent possible, be equitable. As explained earlier, equity is best served by a dividend formula that allocates to each policy its share of surplus in the proportion it has contributed to that surplus. To attain equity most companies in the United States and Canada employ a system of surplus distribution called the contribution plan. For reasons of simplicity noted above, consideration is usually limited to the three major sources of surplus: mortality savings, excess interest, and loading savings. A distribution plan that recognizes only these sources of surplus is called the three-factor contribution plan.

The interest factor of the dividend formula seeks to credit each policy with its share of a company�s investment earnings above the sum needed to meet its obligations. The evolution of computer technology has expanded the extent to which a particular policy�s interest contribution can be traced. The strict equity of considering the different times of premium payments and different interest rates apportioned to the cash value portion of premiums paid under universal life contracts is approximated in participating whole life contracts. For example, because policy loans can cost the insurance company much of the investment return it could otherwise earn, some companies vary the dividend on an individual policy depending on whether a policy loan was outstanding over the dividend year. Loans are very popular during periods of high interest rates, and they can be particularly costly at those times because they preclude the insurer from reinvesting the funds at the higher yields available. Thus if each policyowner receives dividends according to identical scales, those who do not borrow subsidize those who do borrow.

The growing use of direct recognition dividend scales since the late 1970s is an excellent example of attempts to increase the equity of dividend distributions. Direct recognition means that policy choices made by an individual policyowner are reflected in the policy dividends returned to that policyowner.

However, total equity cannot be attained in reality. Mortality experience, for example, varies widely with duration, occupation, amount of insurance, and plan of insurance. It is impractical to distinguish among these factors in computing the individual contributions to surplus from favorable mortality. Many factors affect the rate of expense, including the size of the policy and the variation in the premium tax rates among the different states. Recognizing all these factors would unduly complicate the dividend formula.

Also the divisible surplus will not, except by accident, equal the total "profits" or aggregate contributions to surplus. It is affected by considerations unconnected to the sources of surplus that cannot be related to individual policies or even to classes of policies. Therefore a method of distribution that calls for the computation of individual contributions to surplus would require modification to equal the sum available for distribution. Such modification is likely to disturb the relationship between the assumed contributions from different sources.

Thus under practical conditions it is not possible to refund the excess payments of individual policyowners exactly. In such matters policies must be dealt with as groups or classes. The system of computing refunds or dividends aims to approximate equity between classes and among individual policies within classes. As a minimum standard of equity, surplus distribution should consider the sources from which surplus arises and should not be oversimplified to cause injustice to any group of policyowners.

Just as gross premiums and surrender values are tested by asset share calculations prior to adoption, so are proposed and existing dividend scales tested. Such a test combines a given set of gross premiums and surrender values with realistic assumptions about mortality, interest earnings, expenses, and voluntary policy terminations. The result shows whether the accumulated asset shares for the various plans, ages at issue, and durations meet the requirements of both adequacy and equity. An existing dividend scale should be tested periodically to assure that it meets the same objectives.

Over the life of any block of policies the aggregate dividends distributed will be somewhat less than the amount contributed to surplus. This is necessary if the company is to accumulate and maintain a contingency reserve sufficient to protect it from its liabilities. That is an objective of well-managed companies. As reserves increase, whether from the sale of new policies or the natural progression under old policies, the absolute size of the contingency reserve must also increase. Apart from interest earnings on the contingency reserve or "free surplus," the only source of such funds is the current earnings from policies. Therefore even over the long term, something less than the net additions to surplus from all blocks of policies will be returned to policyowners as dividends. Equity demands that each group of policies bear a share of this cost. This is just another way of saying that a policy�s asset share should eventually exceed the reserve and that management expects each policy to make a permanent contribution to the company�s surplus.

The dividend process described here is simplified to deal with only the major sources of surplus. Some elements of cost and price�such as policy size, smoking behavior, and gender�are commonly included today in premium computations. Should experience vary for these populations, some adjustment may be needed in the dividend scale. Furthermore, this discussion ignores several refinements that may be introduced in the interest of equity or under the pressures of competition.

Ordinary (immediate) annuities present unique problems in pricing and dividend policy. First, expense loading from any single premium contract can arise only in the first year. Second, as reserves under such policies decline with duration, excess earnings are likely to diminish each year. Finally, unless an actuary provides for future improvement in annuitant mortality, as with the use of projection factors, declining death rates among annuitants erode any margins in the mortality assumptions and may eventually produce mortality losses, offsetting the declining gain from excess interest. Thus unless the margins in the actuarial assumptions are very conservative, which might produce noncompetitive rates, dividends are seldom used in annuity contracts. When they do appear, they are likely to be small and to become smaller over time.