2. Utilizing interest assumptions, in calculating reserves and the expenses related to future benefit payments, that are well below the yield on investments experienced in recent years and anticipated in the future 3. Utilizing mortality tables derived from periods when death In response to these concerns, both the A. M. Best Company and the American Institute of Financial Analysts developed methods of using information available in statutory statements to adjust net gain and balance sheet figures. The two approaches are quite similar, and we have used them with certain simplifications to calculate adjusted earnings and balance sheet figures as components in possible early warning tests. Our method of adjusting earnings is illustrated graphically in Exhibit I. We adjusted net gain in two ways. First, we used the sum-of-the-yearsdigits method to allocate individual life and annuity acquisition expenses over a five-year period. These acquisition expenses include first-year commissions, medical examination fees, inspection report fees, agency expense allowances, and agency conferences. The five-year period for charging off these expenses represents a simplification of the Best and AIFA approach, in which the period for charging off acquisition expenses depends upon the lapse rate. Since the distribution of lapse rates was found to be the same for insolvent and solvent companies, we did not believe this simplification would alter the results of the analysis. We also felt it would be most practical to begin with five years' information and build up the data base over time. ments. Second, we adjusted the aggregate increase in reserves for policies and contracts with life contingencies (excluding group insurance) for the difference between the current average reserve interest assumption and the fiveyear average investment yield. For the solvent companies, the five-year yield was assumed to be one-half of 1 percent less than the current yield; for the insolvents, the five-year yield was calculated from the annual stateThe reserve increase adjustment was made according to the 10-for1 rule adopted by Best and the AIFA. This rule holds that for every percentage point of difference between the interest assumption and the investment yield, there should be a 10-point adjustment in the reserves and the increase in reserves. In order to limit the impact that this adjustment might have on companies with unusual financial results (which includes a number of the insolvents), we limited this adjustment factor to plus or minus 30 percent. Of course, if the yield is higher than the interest assumption, The adjusted and unadjusted tests of change in surplus are about equally effective 2 years before insolvency, but the unadjusted test is better in the final year.... the reserves and the increase in reserves are adjusted downward, and vice versa. Following the precedent set by Best and the AIFA, we made no adjustments for conservative mortality assumptions. Balance sheet adjustments, illustrated in Exhibit II, were made in a similar manner. Assets and surplus were increased by the amount of any acquisition expense yet to be charged off, and liabilities and surplus were adjusted for the reserve interest assumption using the 10-for-1 rule. In addition, the mandatory securities valuation reserve was deducted from liabilities and added to surplus. We used these adjusted earnings and balance sheet figures to calculate three tests: (1) adjusted earnings to total income; (2) change in adjusted surplus; and (3) liquid assets to adjusted liabilities. In the latter, liquid assets means cash and invested assets minus any excess of investments in real estate over 5 percent of adjusted liabilities. Liquid assets might also exclude investments in affiliates, but this information was not available for most of the insolvent companies. Exhibit III compares the ratio of adjusted earnings to total income with the ratio of net gain to total income. The adjusted earnings test is moderately superior both when the insolvents are taken in the third year before insolvency (as shown in Exhibit III) and in other years. are The two tests of the change in surplus adjusted and unadjusted shown in Exhibit IV. They are of approximately equal effectiveness when insolvents are taken in the second year before insolvency. The change in stated surplus is more effective in the insolvents' final year. For both tests, any surplus paid in during the year is deducted from the change in surplus. The effectiveness of the ratio of liquid assets to adjusted liabilities is shown in Exhibit V. This test has no parallel among the current tests; a similar ratio using stated liabilities would have no discriminating effectiveness, as shown in the exhibit. The ratio based on adjusted liabililities, however, is moderately effective. EVALUATION OF ADJUSTED TESTS In order to determine whether these adjusted tests should be employed in the Early Warning System, the significant question is not ow well they 99-073 078 - 22 |