The Reinsurance Association of America has recently published its latest biennial study on historical loss cost developments in certain liability classes in the US.

Most of the talk in recent weeks has been of a rapidly-hardening market, with huge rate rises, high deductibles and much stricter terms and conditions. From the claims side, the focus has been on WTC-related payments, with estimates inevitably heading upwards. These have, however, masked the regular market underlyings, including loss developments.

Last year saw some US re/insurers releasing estimates of claims deteriorations relating to asbestos exposures. Insurance analysts soon came out with some pretty horrendous figures, suggesting that the global industry may be exposed to the tune of some $200bn. Back in June, consultant Tillinghast-Towers Perrin noted that the number of claims filed against some asbestos defendants in the previous year had increased up to threefold. "A significant majority of the new claims do not relate to individuals with mesothelioma or lung cancer, but rather to non-malignant diseases. The propensity to sue for non-malignancies has clearly increased. Over 450,000 claims have been filed to date and we project that close to 1,000,000 claims will be filed before the litigation ends, absent some type of federal legislation," said Mike Angelina, a consulting actuary with Tillinghast-Towers Perrin and co-author of the study.

The study's authors estimated the US industry would find itself bearing claims between $55bn and $65bn. Adding in environmental claims would bring the total estimates for the US industry to between $85bn and $105bn; at that point, the industry had paid $41bn for the exposures and held about $23bn in reserves resulting in a shortfall of up to $41bn.

Of course, exposures such as asbestos and pollution are the `crisis' end of the claims spectrum, but being able to project loss developments in other liability classes is just as critical. Historical data becomes an intrinsic part of helping project forward loss developments, and the Reinsurance Association of America (CRAA) has recently issued the 2001 edition of its Historical Loss Development Study.

Published every two years, the study assesses and compares loss developments in automobile liability, general liability, medical malpractice and workers compensation.

The RAA takes great pains to point out that the study aims to "reinforce awareness of historical loss development patterns": it does not attempt to project future loss developments.

Data for the study, which looks at historical loss development patterns in casualty excess reinsurers and primaries writing high deductible or umbrella business, comes from 24 members of the RAA plus three other companies writing casualty excess reinsurance for the four identified lines.

Since the 1991 edition - the study was first produced in 1969 - asbestos and environmental liability losses have been excluded from the data wherever possible. The authors have also attempted to take out mass tort claims such as those relating to silicone implants, Dalkon Shield and lead poisoning.

Certain factors around the base data may skew the loss development. These include:

  • sudden changes in the volume of reported losses, resulting in extremely volatile development. This may be due to one or two large adverse judgments or very late emerging claims;

  • the data consist of losses excess over retentions of ceding companies. These retentions vary widely in size between companies and over time;

  • most of the data was submitted net of retrocessions. Use of retrocession cover has increased over the data period (about 40 years) as a general trend, though it, too, varies between companies and over time;

  • some of the companies providing data were unable to submit information in its full specifications of net of assumed and ceded loss portfolio transfers, assumed and ceded commutations, and aggregate provisions;

  • most data were submitted on an accident year basis, though some was provided on an underwriting year basis;

  • although the study covers just the US, excess loss settlement patterns may not be uniform across the country because of differing populations and legal climates;

  • underwriting rules vary between companies providing data for the study;

  • discounting practices vary between companies submitting data for the study;

  • loss development patterns can imply late reported claims or a change in the aggregated estimated cost of claims already reported, varying by line and by company;

  • reinsurers may have adopted new contract provisions affecting the loss development data, such as index clauses, late reporting clauses, loss ratio caps, sunset clauses and commutation clauses; and

  • the study does not include data from failed reinsurers.


  • Automobile liability
    This is the fastest developing type of business in the study. About 25% of the ultimate losses are known to reinsurers at the end of a given accident year, and by the fifth year, about 90% of the ultimate losses have been identified. Five years later, all but 4% have emerged.

    Workers compensation
    This is much slower to emerge, with only 12% of ultimate losses known at the end of the first year. This has increased to 45% by the fifth year, and just over 55% by the tenth. The study suggests that the slow development may partly be due to the use of discounted reserves on life pension cases. "Losses for this line are paid out very slowly, with payments on some claims extending over the remaining lifetime of the claimant," it explains. "In addition, a portion of the development might stem from the use of outdated mortality tables in the past, and actual medical inflation rates that were higher than anticipated at earlier valuations."

    General liability
    Again, this area has been characterised by particularly slow development; by the end of the first year, just 8% of the ultimate losses have been reported. This does, however, increase faster than the workers compensation losses, lifting to 71% by the fifth year and about 88% five years after that. The rest of the losses emerge over the ensuing 20-25 years. "It should also be kept in mind that the actual historical development of general liability in total was much slower than these statistics indicate, due to late emerging asbestos, environmental and other mass tort losses that were excluded from the data," notes the report.

    Medical malpractice
    The medical malpractice development follows that of general liability particularly closely, with around 5% of ultimate losses reported in the first year, rising to just over 65% by the fifth year and just under 90% by the tenth. "In reviewing the medical malpractice development, it should be noted that the data is a mixture of claims made and occurrence coverages," comments the report. "The proportion of claims made coverage included in the data is unknown, but is likely to be substantially greater since 1975, when claims made coverage was widely introduced."

    As well as looking at the historical loss development in the four categories, the study details the changes in historical loss development. It details the marked differences in reporting patterns between excess reinsurers and primary insurers, and compares historical loss development between treaty and facultative business. There are further sections on the importance of attachment points, the range of variation in historical loss development, and the impact of asbestos, environmental liability and other mass torts. This is then augmented by almost 200 pages of data and graphs.

    The Historical Loss Development Study can be ordered via the RAA's website, , at a cost of $250.