A leading City of London law firm

Article

Reinsurance Spirals

By Rosalind Jones of Elborne Mitchell
First published: Global Reinsurance [1st September 2000]

A phenomenon known as the PA spiral is currently troubling reinsurers, particularly those in the North American market. Losses from workers' compensation and personal accident business written in 1993 and subsequent years appear to have spiralled to unexpectedly high levels of reinsurance protections. Those reinsurers who are bearing the brunt of these losses may now be questioning whether there are any parallels between the PA spiral and the LMX spiral which caused such massive losses in the London market in the late 1980's.

It is well known that, in general terms, a spiral effect occurs when excess of loss reinsurers extensively underwrite each other's excess of loss protections. This leads to a transfer of exposures to the same players within the market rather than their dispersal outside the market. The effect of the transfer of exposures was starkly demonstrated by the series of losses which occurred between 1987 and 1990, including UK Storm 87J, the loss of Piper Alpha, and Hurricane Hugo:

  • the catastrophe losses were concentrated in a handful of specialist excess of loss syndicates within Lloyd's rather than evenly dispersed in the market. The four Gooda Walker syndicates, for example, sustained 30 per cent of the entire market's losses in 1989.
  • because the losses were so concentrated, and there was so little "leakage" from the spiral by way of retentions and coinsurance, the catastrophe losses incurred by these specialist syndicates spiralled up and out of the top of their reinsurance programmes. For example, Gooda Walker syndicate 290 suffered a gross loss of US$385 million in respect of Hurricane Hugo against reinsurance protection purchased of US$170 million. [Footnote 1]
  • the catastrophe losses had a high ratio of gross to net loss, reflecting the extent to which the losses spiralled. For example, on a gross basis the Piper Alpha loss exceeded by a multiple of 10 the net loss that was covered on the London market.
  • traditional methods of rating, whereby the lower or working layers have a higher rate on line than the higher layers to reflect their greater exposure to losses, had no relevance because the higher layers of insurers protections were not significantly less exposed than the lower layers. In the litigation that followed the catastrophic losses, in which the Names on the LMX Syndicates accused their underwriters of negligence, the traditional rating methods adopted by the excess of loss underwriters were revealed as inappropriate to deal with the effect of the spiral.

Rates of 1 or 2 per cent on line were charged for the highest layers in the belief that they were very remote from loss whereas in reality these layers were far more exposed to loss than underwriters believed them to be.

These are all key spiral features, which are common to the PA spiral and LMX spiral alike. At first blush, however, the PA market seems an unlikely place to encounter similar spiral problems to those that happened in the property and casualty market and which caused such disastrous losses for the Gooda Walker and other LMX syndicates. The PA market, which insures the lives of individuals, albeit expensive individuals such as sports stars, is exposed to losses of a very different size. The death of the American basket ball player Drazen Petrovic in a car accident in 1993 was a large loss to the PA market of $10 million. By comparison, the insured loss from Hurricane Hugo, was estimated at US$4,100 million.

While it might be tempting to consider the PA spiral as a "dolls-house" version of the LMX spiral of the late 1980's there are important distinctions. Firstly, it took losses of some considerable magnitude - massive catastrophes, in fact - to exhaust the retentions of all the players in the London market and trigger the LMX spiral. By contrast, there has been nothing unusual about the pattern of losses suffered by the PA market either in terms of their frequency or magnitude. Various large single life losses have occurred and these have spiralled, for example the Petrovic and Ayrton Senna losses. The real damage, however, is being caused by a number of modestly sized losses which have spiralled.

It appears that what is exceptional about the PA spiral is not the loss pattern but the purity of the spiral itself. For the 1993 year, for example, there were only four core players in the PA spiral retaining a maximum of $5,000 each. Apart from the core players, the leakage from the spiral to other participants at lower levels was so minimal that any claim in excess of $55,000 would inevitably spiral to layers in excess of $500,000. At these higher levels, the non-core players such as North American reinsurers would participate in the spiral and provide "leakage" in the form of a retention substantial enough to absorb losses of the size which are spiralling. The practice in the PA market of granting unlimited (and often free) reinstatements meant that any number of small claims would spiral to be absorbed in the retention of the non core players. In this way, the damage to North American reinsurers in the PA market is being caused by the attritional effect of a large number of small and modestly sized losses rather than catastrophe type claims.

The existence of such a pure spiral does of course give rise to questions as to whether a spiral of such artificiality can occur naturally on its own. After all, if the core players in the spiral were of a mind to create a spiral with the aid of a participating broker, one of the features of the spiral which caused such difficulties to the London market in the 1980's, namely the irrational rating structure, could be turned to their advantage. Where the core players are only risk swapping at the lower layers, for similar rates, they are effectively buying cheap reinsurance from the North American reinsurer in the sure knowledge that it is he who will bear the loss. Leaving aside the question of the intention of the parties, the interesting question which arises is whether there is anything actually wrong with such underwriting practices.

One response to this question is that an underwriter of spiral business cannot complain if he loses the amount of his retention, even if, as in the case of PA underwriters, he loses it again and again on numerous small losses. Spiral business is known to be "opaque", in other words, the underwriter simply has no means of knowing what is in the book of business he is reinsuring or to what extent it is exposed to any one loss. Because of this opacity, it is a mistake for an underwriter to believe that he is less exposed to risk at higher levels of his cedant's programme. An underwriter wishing to underwrite spiral business, should therefore assess his maximum exposure to any one loss, decide how much he is prepared to retain of it, and reinsure out the rest. He should be prepared to lose what he has elected to retain.

This fits with the underwriting approach formulated by Mr Justice Phillips, the Judge who presided over the litigation between the Names and the Gooda Walker syndicates, as being the proper way to underwrite spiral business. In fact, Phillips J went further, saying that the only way to competently underwrite spiral business was to take advantage of the disparity of rates for low level and high level layers of business ie to "arbitrage" the risks.

Phillips J's approach in Gooda Walker was subject to some criticism at the time because it is questionable whether in the true sense of the word any underwriter can "arbitrage" because of course no one knows from where or from which books of business or combinations of books, the losses are going to come. In any event the Phillips J model inevitability relies on some underwriters taking "advantage" of rate disparity - or in practice taking advantage of another underwriters' ignorance of rates or ignorance of the nature of the market. While in other financial markets there may undeniably be such an element, "arbitraging" on the whole involves taking advantage of market idiosyncrasies or inconsistencies that are freely available to (and are transparent to) all players in that market and usually in circumstances where the particular position is taken in the light of that individual players' trading position. Arbitraging, as a concept as accepted by Phillips J, does not fit neatly with uberrimae fidei, the duty of dealing in the utmost good faith.

There is an important distinction, however, between the market in which the Syndicate underwriters were operating in the late 1980's and the PA market. Although the market of the late 1980's was an incestuous one, there was no question of back to back reinsurances being written between a small number of core players at similar rates. Similarly, there was no question of the LMX spiral arising other than through a combination of circumstances which existed in the London market at the time.

The question of whether there is anything wrong with the underwriting practices which brought about the PA spiral is therefore a different one to that which was considered by Phillips J in the Names' litigation. The answer essentially turns on two issues; whether those underwriting back to back reinsurances were aware of this fact and, assuming for these purposes that they were, whether that is something which they ought to have disclosed to their reinsurers.

Under English law [Footnote 2] the assured, and his broker, are under an obligation to disclose to the insurer prior to the conclusion of the contract every material circumstance which is known to him. At what point does the existence of back to back reinsurances become "material" to the underwriter presented with the risk? It is possible that if one risk is passed from one reinsurer to another and back again at a low level this would not be material to a reinsurer much higher up the programme. Where, however, back to back reinsurances exist which will effectively funnel the claims up to the reinsurer, and make a nonsense of his rating of the risk, then that would surely be material to a prudent underwriter and ought to be disclosed by the assured, assuming that he is aware of it.

Another issue which has been highlighted by the PA spiral is the position of the foreign insurer doing business with the London market. Under English law relating to non disclosure, it is not necessary for the assured to disclose to the insurer any circumstance which is known or presumed to be known to him and this includes "matters of common notoriety or knowledge". [Footnote 3] However, circumstances which are common knowledge in the rumour mill of the London market are not necessarily known to a reinsurer situated North America. For example, where back to back reinsurances are being placed, the slips taken around Lloyd's for the syndicate underwriter to scratch will reveal to them the number of participants regularly participating in a particular line of business and the extent to which syndicates were underwriting on each other's programmes. These slips would not normally be shown to the foreign reinsurer at the time of placement. The foreign reinsurer is clearly at a disadvantage to his London counterparts. It is, however, open to the courts to allow for this disadvantage and hold that a matter of common knowledge to one insurer is not of common knowledge to another, foreign, insurer. This could lead to the somewhat curious result that a broker placing a risk to a foreign insurer must disclose more to him at the time of placement than to his London counterpart underwriting the same risk.

Finally, one of the reasons why the PA spiral caused such damage to the US reinsurers was that they often carried significant net retentions in marked comparison to their London cedants. Partly this reflected the US reinsurers' perception of themselves as being remote from risk. It probably also reflected an ignorance as to the minimal amount being retained by their cedants. It is surprising how often in practice reinsurers do not ask their reassureds how much of the risk they are retaining net, despite the obvious importance of this information. If all the participants in the PA spiral had carried even moderately sized retentions, this would have provided enough leakage to counteract or at least reduce the spiral effect.

Footnote 1: Figures as at 31.12.92. The Judgment records that the figures were not agreed between the parties.

Footnote 2: S18 and 19 of the Marine Insurance Act 1906.

Footnote 3: S18(3)MIA 1906.