TEST SCORES FOR INSOLVENT COMPANIES NOT IDENTIFIED BY SCREENING SYSTEM

interpret the test results, as well as examining the patterns of test scores produced by insolvent and sound companies. Three principal conclusions emerged:

1. Most insolvent companies fell outside the usual range on many
tests by wide margins.

2. Experienced users base their evaluation more on the number of tests on which the company fell outside the usual range than on the amount by which test results exceeded the bench marks or on the particular combination of test scores.

3. Although each user considers some tests more important than others, there is little agreement on relative priorities except that the reserve tests, taken as a group, are important.

In addition, we tested the effectiveness of a wide variety of screening systems, which varied in the number of tests included, the bench marks used, and the complexity of the scoring system. For each approach tested, we calculated the percentage of all companies that would be classified as priority companies in 1972 and compared this with the percentage of insolvent companies that would have been so classified one and three years prior to insolvency. Our conclusion was that one of the simplest and most straightforward screening systems is the most effective: the number of tests on which the company fell outside the usual range.

We decided that the priority designation would be given to those companies falling outside the usual range on four or more of the eleven new tests at the new bench marks. This system would have classified as priority companies approximately 11 percent of all companies, as compared with 82 percent of insolvent companies in the third year prior to insolvency and 96 percent of insolvent companies in the year prior to insolvency (Exhibit XIII). Test results for the insolvent companies not identified by this system are shown in Exhibit XIV. Of four companies not identified three years prior to insolvency, two (Members and United Bonding) had changes in ownership or management during their last three years. Citizen's Casualty, which would not have been identified in its last year, was clearly identified by the system five years before insolvency (by seven tests) and three years before insolvency (by nine tests).

Exhibit XV shows the percentage of all companies and of insolvent companies falling outside the usual range on two to seven tests. The four-test rule for selecting priority companies provides the best balance between insolvent and solvent companies. If the rule were tightened to pick up

companies with three or more scores outside the usual range, the percentage of all companies identified would almost double, with little increase in the percentage of insolvent companies identified. On the other hand, loosening the standard to five tests would reduce the percentage of insolvents identified in their last year by almost 10 points.

Exhibit XVI compares the effectiveness of this new system with a system based on the eleven old tests and bench marks. About half the improvement in effectiveness - as indicated by the distance from line C to line A in Exhibit XVI is due to the two new tests (investment yield and change in writings), while the other half of the improvement is due to eliminating two of the old tests (asset change and reserve adequacy Test 3) and changing the bench marks on the remaining reserve tests from 10 percent to 25 percent.

The new screening system is compared to more complex and more simple systems in Exhibit XVII. The dotted line, which is the same as line B in Exhibit XVI, represents the chosen system without the two new tests. Line D represents the Bailey scoring system, under which from one to five points is assigned for each test result outside the usual range, depending on the amount by which the test result exceeded the bench mark. This system would, on balance, be only slightly less effective than the chosen system, but it would also be significantly more complex. Line E represents a typical screening system based on a smaller number of tests (old Tests 1, 2, 4, 5, 7, and 8). It is significantly less effective than the chosen system. No system involving a reduced number of tests was found to be more effective than the chosen system, which uses all eleven tests.

IM PROVING

REPORT FORMAT

In order to make the early warning reports easier to understand and use, we decided that summary reports would be produced specifically for each state, and that companies would be listed in the following groups:

Priority domestic companies

Priority foreign companies

Nonpriority domestic companies

Nonpriority foreign companies.

Thus, each state would receive an action-oriented document tailored to its own requirements. In addition, we decided that the report should identify the name and usual range for each test and that all scores falling outside the usual range should be indicated by an asterisk.

In addition to a more action-oriented summary report, the states will also receive a data sheet on each priority company, which will provide:

Test results for three years

A breakdown of the four more complex tests into their component parts

Key annual statement information.

This data sheet is intended to assist the states in analyzing test results and preparing for examination of priority companies. The provision of threering binders will permit the states to insert each new summary release and priority company data sheet in its proper place as it is received.