 |
Objectives, Scope and Methodology |
- We examined private agency ratings of life/health insurers at the request of Chairwoman Cardiss Collins of the Subcommittee on Commerce, Consumer Protection and Competitiveness, House Committee on Energy and Commerce. Specifically, our objectives were to (1) compare the rating systems of the major raters of life/health insurance companies--Best, D&P, Moody's, S&P and Weiss and (2) determine which raters were first to report the vulnerability of financially impaired or insolvent insurers.
- To obtain information about life/health insurer ratings and related studies, we reviewed relevant articles, spoke with insurance experts and interviewed representatives from the five major rating agencies. We obtained rating information from the raters.1 Although earlier rating data was received from most of the raters, we compared ratings from August 31, 1989, the date Weiss began rating life/health insurers, to June 30, 1992. Because (1) results based on data from one period may not be applicable to a future period and (2) at least one rater has made changes to its rating scale subsequent to our analysis, our results are strictly applicable only to the period analyzed.
- Most of the rating data supplied by Best from their computer data base did not contain the exact date of rating changes. To obtain the dates, as Best suggested, we consulted their published ratings and press releases, which they also supplied. Although data from the other raters also required some preparation and cross-checking, it was of a lesser degree than that required for data supplied by Best.
- We used NAIC data to determine the universe of existing life/health insurers, their assets and unique identification of these insurers. However, NAIC could not assure us that its records contained all life/health insurers because the states do not require all insurers to file with NAIC. In addition, because insurers may stop filing statutory financial reports when they become financially impaired, NAIC's files may have gaps in financial information. We interpolated asset data that was missing from NAIC files. Although our analysis assumed insurers' statutory financial data to be comparable, we are aware that this assumption may not be true in all cases due to differing state accounting practices.
- We compared the agencies' rating scales, descriptions and methodologies, as well as industry coverage and actual ratings for our analysis period. We used the raters' descriptions of their ratings to establish a correspondence among rating levels so that we could compare the assignment of actual ratings. We confirmed the division of ratings into "secure" and "vulnerable" categories with the raters and noted the cases where they did not use this characterization.
- We examined the agencies' coverage of the life/health insurance industry by determining the percentage of life/health insurers they rated. This was done by comparing the life/health insurers rated by a particular agency to the total of life/health insurers, both in number and weighted by asset size. We also compared the agencies' distribution of ratings both for (1) insurers rated by each agency individually and (2) insurers rated in common between agencies, two agencies at a time.
- In addition, we compared the raters' timing in reporting the vulnerability of financially impaired or insolvent insurers by comparing the date they assigned "vulnerable" ratings to the date the first public regulatory action was taken.2 We defined financial impairment or insolvency in the same manner as state insurance regulators and NAIC. We limited our comparison to the period from August 31, 1989, to June 30, 1992 and constrained the number of days we credited by starting the count from August 31, 1989. We removed from comparison insurers that had a "vulnerable" rating from more than one rater on August 31, 1989. In addition, if an agency had assigned ratings prior to August 31, 1989, but did not rate on or after this date, we did not use these ratings in our comparison. And, finally, if a rater assigned ratings only after the first regulatory action occurred, we categorized this as "not rated" for the purpose of our timing comparison. This had no effect on D&P's, Moody's, or S&P's results. However, Best's date of "vulnerable" assignment was constrained to be no earlier than August 31, 1989, in the 7 cases where Best was first in assigning a "vulnerable" rating. In 5 of these 7 cases, Best assigned a "vulnerable" rating prior to August 31, 1989.
- In doing this analysis, we tried to take the point of view of insurance consumers. By comparing raters' timing in assigning "vulnerable" ratings to insurers that became financially impaired or insolvent, we placed the most value on reducing the likelihood that an insurer would be rated "secure" when it should have been rated "vulnerable." We realize that this placed less value on reducing the likelihood that an insurer would be rated "vulnerable" when it should have been rated "secure."
- To do our timing comparisons, we constructed a file of state regulatory actions against insolvent or financially impaired life/health insurers. Because NAIC obtains its information about insolvent or financially impaired insurers from voluntary reporting by state regulators, we could not be assured that NAIC's information was comprehensive. In fact, we identified a number of such insurers from other sources--NOLHGA, Best, S&P and the author of a related study.3 In a number of instances, we called individual state regulators to verify confusing or conflicting information. In several cases, we obtained information about additional public regulatory actions taken against the insurers, in some cases at earlier dates than we had obtained from previous sources. Thus, even though the information we have about public state regulatory actions is more comprehensive than any of the individual sources we used, we still cannot be assured that it is fully comprehensive.
- We used the date of the first public action taken by any state against a financially impaired or insolvent life/health insurer as the reference point from which we measured a rater's timing in detecting financial problems. The earliest type of action contained in our sample ranged from a cease and desist order to a liquidation notice. However, because different states may take the first public action at different stages of financial impairment, no public action may be recorded in some states if, for example, an insurer is sold to or merged with another insurer.
- We did our work between January 1992 and September 1994 in accordance with generally accepted government auditing standards. Because we used information supplied by the rating agencies, we asked them to review the facts contained in a draft of this briefing report. We received responses from Best, Moody's, S&P and Weiss, who generally agreed with the factual information presented.
- Best, S&P and Moody's also provided other comments critical of the report. The three agencies pointed out that rating approaches and philosophies vary among the agencies and that, because of this, alternative methodologies covering longer periods of time could produce different results. We do not disagree. We did not use data prior to August 31, 1989, because we were asked to compare all five rating agencies and Weiss had no ratings prior to that date. Thus, we carefully limited our analysis to a description of events that occurred over a period of slightly less than 3 years. Our results are not projectible and should not be construed as applying to any other time period.
- A second common criticism was that we should have presented the percentage of life/health insurers in each category that later became financially impaired. The agencies believed that this would have counteracted a tendency in the report to concentrate on (in Moody's terminology) "Type I' error (too high a rating on a company that defaults)" rather than "Type II' error (too low a rating on a company that is financially stronger than indicated by the rating)." In our timing comparison, we looked only at the ratings of life/health insurers that became financially impaired. The period we could use to compare all five agencies was less than 3 years--August 31, 1989 to June 30, 1992. We believe that the results of the additional analysis suggested by the rating agencies would be inconclusive over such a short time period. The failure rate of insurance companies overall is relatively low. Because the sets of insurers rated by the agencies vary substantially, the set of institutions rated in common by all raters is quite small. Since the numbers of both companies and associated failures that would be available for comparison are small, obtaining a statistically valid result would require a longer sample period than was available to us.
- Finally, S&P believes that we should have included their Qualified Solvency Ratings in our analysis. We excluded these ratings because S&P began their publication in April 1991, so the comparisons we could have made with other ratings were limited.
|
|
