1. General remarks
C1 This Chapter of the PRICL deals with the question of whether a reinsured may or must aggregate two or more separate losses for the purpose of presenting a claim to its reinsurers. Legal certainty in the matter of aggregating losses is of fundamental importance as the structure of the coverage forms a vital element in calculating the premium to be charged as well as the capacity needed.
C2 Aggregating two or more losses and treating them as a single loss may lead to the aggregated loss exceeding the deductible (retention; excess point; attachment point; priority), thereby triggering the reinsurer’s duty to indemnify the reinsured. Depending on the sums involved, an aggregation of losses may also lead to a sum in excess of the reinsurance coverage limit, for which the primary insured will not be able to claim indemnification. Determining whom an aggregation of losses will benefit depends on the structure of the coverage (retention/coverage limit) as well as the structure of the separate losses in a specific case (see William 3.2).
C3 An aggregation of losses may be undertaken on the basis of a particular period of time in the sense that all losses that can be attributed to this period (depending on whether the reinsurer’s liability is triggered by the underlying “risk attaching”, the underlying “losses occurring” or “claims made” by the primary insured) are to be added together for the purpose of determining whether the deductible is exceeded or the cover limit is pierced. The so-called aggregate policies do not require that the losses to be aggregated arise from a common unifying factor.
C4 By contrast, so-called event-based or cause-based policies require that there is some sort of causal link between the individual losses and the respective unifying factor. This Chapter exclusively deals with event-based and cause-based aggregation mechanisms. If the parties agree to aggregate two or more losses based on an aggregating factor other than either an “event” or a “cause”, the PRICL does not contain any direct rule for this situation but may still provide some guidance.
C5 Generally, the parties are free to agree on whether they want to treat two or more separate losses as a single loss, and if so, under which conditions they wish to do so. If they agree to aggregate separate losses to one single loss, they will generally also agree on a unifying factor of some kind. If a contract does not provide for an aggregation mechanism, no loss aggregation is possible.
C6 In this Chapter, it will not be possible to resolve all of the uncertainties arising with regard to the aggregation of losses based on an “event” or a “cause”. By making conscious decisions as to what constitutes an “event” or a “cause” under the PRICL, this uncertainty may, however, be reduced considerably (cf. William 4.166, 5.36). To this end, it is intended that the PRICL provides sensible definitions of the notions of “event” and “cause” in order to make them fit for their purpose in respect of aggregating losses in contracts of reinsurance. Hence, it may well be that the meanings the PRICL attaches to these notions deviate from the meanings they have in ordinary speech. It should further be noted that the rules set forth in the PRICL do not necessarily align with judicial or arbitral decisions under any particular applicable law. To the extent that the PRICL is inconsistent with a legal precedent, it prevails when it has been selected by the parties to the contract of reinsurance.
2. Only individual losses allocated to the same reinsurance period can be aggregated
C7 Under the PRICL, individual losses will first be allocated to a reinsurance period (see Chapter 4). Only individual losses that are allocated to the same reinsurance period may be aggregated in application of Article 5.2 or 5.3. Conversely, if multiple individual losses are allocated to different reinsurance periods they are not to be aggregated even if they arise from the same event or the same cause.
Illustration
I1. Equipment is stored at a port and is vandalized through a series of individual acts of pilferage during a period of some 18 months. During this time, multiple individual losses occur. The individual losses all result from one originating cause (Article 5.3(2)), i.e. the port’s failure to put in place an adequate system to protect the stored goods. Three consecutive contracts of reinsurance are in force during the 18-month period. Contract 1 is in force during the first 3 months, contract 2 during the 12 months following and contract 3 during the last 3 months (facts based on Municipal Mutual Ins Ltd v Sea Ins Co Ltd [1998] CLC 957, 966 et seq.).
Paragraph (1) establishes that only individual losses that occur when either contract 1, 2 or 3 is in force can be aggregated. By contrast, individual losses that are allocated to different reinsurance periods cannot be aggregated, even though they may all originate from the port’s failure of putting in place an adequate system of protecting the stored goods.
C8 Consequently, the parties cannot extend the reinsurance period by means of an aggregation clause. Rather, if the parties wish to extend the reinsurance period to include losses that result from a cause or event that was initiated during the reinsurance period but occurred after the latter has elapsed, they may provide for an “extension of protection clause”. Where the contract provides for such a clause, individual losses occurring during the extension of the reinsurance period may be aggregated with losses occurring before the extension.
Illustration
I2. Reinsurance Contract 1 is taken out against the peril of hurricane and is in force from 1 January 2018 to 31 December 2018. Contract 2 is taken out against the same peril and enters into force on 1 January 2019. Both contracts contain an “extension of protection clause” to the effect that if an event develops during the reinsurance period, causes individual losses, is still in progress when the reinsurance expires and thereafter causes further individual losses, the reinsurance period is extended to also cover those latter losses. Moreover, they both contain an event-based aggregation clause.
On 31 December 2018, a hurricane develops and causes multiple individual losses that day. The hurricane continues to sweep over town until the morning hours of 1 January 2019. Accordingly, some of the losses occur on 1 January 2019. Based upon the “extension of protection clause” in Reinsurance Contract 1, hurricane losses that occur on 1 January 2019 are covered under it. As all individual losses – regardless of whether they occurred on 31 December 2018 or 1 January 2019 – are allocated to Reinsurance Contract 1, they are to be aggregated under Article 5.2(1).
3. Deductible, retention, priority, excess point, attachment point
C9 Both in facultative excess of loss reinsurance and in treaty excess of loss reinsurance, a reinsured takes out reinsurance only on an amount in excess of a specific sum. Hence, the reinsured retains liability for the entire loss below that excess point but transfers liability for the part of the loss above it. The amount of the loss for which liability is retained by the reinsured is termed a “deductible”, “retention” or “priority”. The upper limit of a deductible or retention is referred to as an “excess point” or “attachment point” (see William 2.72 et seq.). Excess or attachment points are regularly defined in relation to any losses that arise from a unifying factor, such as an event or a cause. In other words, the reinsurer promises to indemnify the reinsured against the aggregate of all losses arising from a particular event or cause above the excess or attachment point.
4. Cover, limit, cover limit
C10 Contracts regularly contain limits on liability. More specifically, reinsurers only reinsure their reinsureds up to a certain sum, the so-called “coverage limit”, preventing the reinsureds from claiming indemnification for any loss amount above this “limit” (see William 2.86 et seq.). The parties to a contract regularly define their coverage limit in relation to the aggregate of all losses arising from a unifying factor, such as an event or cause. In other words, the reinsurer promises to indemnify the reinsured against any losses arising from a particular event or cause up to a fixed amount of money.
5. Loss
C11 In the context of loss aggregation, the word “loss” generally refers to the monetary damage suffered by the primary insured. By contrast, in the context of loss allocation (see Comment 1 to Article 4.2), the word “loss” refers to the reinsured’s obligation to reimburse the policyholder under the underlying contract (see William 2.19 et seq.).
C12 With regard to life insurance policies, the word “loss” refers to the primary insured’s death and not to a monetary detriment caused by his death.
C13 Hence, the notion of “loss” does not refer to the event or “occurrence out of which a claim arises, for loss suffered by the original insured, such as storm damage, flood damage or the like, or in the case of professional [liability] losses, the negligent act or omission of the insured” (see Caudle v Sharp [1995] CLC 642 (CA) 648).
6. Unifying factor
C14 Individual losses may be aggregated if they all result from one common unifying factor. The terminology used to describe such unifying factors varies considerably. In fact, the industry has used a variety of terms, such as “event”, “occurrence”, “happening”, “accident”, “catastrophe”, “disaster”, “calamity”, “cause” and “source”, to define it.
C15 In Articles 5.2 and 5.3, there are rules for aggregation either on the basis of an “event” or a “cause”. These two concepts are distinguished by different causation requirements between the individual losses on the one hand and the respective unifying factors on the other hand (see Comments 18 et seq. to Article 5.1).
7. Causation
a. General aspects
C16 Normally, the parties agree to aggregate losses that have a “common origin in some act or event or cause” (see Edelman & Burns 4.57). Hence, there must be a causal link between the unifying factor and all the losses to be aggregated (see William 3.16 et seq.).
C17 The degree of causality required in relation to loss aggregation is most controversial. As the workings of causation are infinite, the parties must decide how far back in the chain of causation a unifying factor may lie so that it is still justified to aggregate the losses that arise from it (cf. Lloyd’s TSB Gen Ins Holdings Ltd v Lloyd’s Bank Group Ins Co Ltd [2003] UKHL 48, 17).
C18 Traditionally, contracting parties have used “event-based” and “cause-based” aggregation language. The parties agree on an event-based aggregation if, for example, they agree to aggregate “each and every loss […] arising out of one event” (see Caudle v Sharp [1995] CLC 642 (CA) 644) or similar wordings. They agree on a cause-based aggregation if, for example, they agree to aggregate losses “arising from one originating cause” (see Cox v Bankside Members Agency Ltd [1995] CLC 671 (CA) 679) or similar wordings.
C19 As to the difference between event- and cause-based aggregation, it has been held that an event is something which happened at a particular time, at a particular place, in a particular way. A cause is a less constrained concept and has a much broader connotation than the notion of event (Axa Reins (UK) Ltd v Field [1996] 1 WLR 1026, 1035).
C20 Consequently, cause-based aggregation allows the parties “to look further back” in the chain of causation and to “use a remoter common” unifying factor for the purpose of determining the aggregate loss than event-based aggregation (Louw & Tompkinson 11). The question is, however, where exactly in the chain of causation each unifying factor comes to lie.
b. Causation: Unities test is inappropriate
C21 Applying English law, arbitral tribunals and courts determine the required degree of causation between the relevant event and the separate losses by means of the so-called “unities test”, which was developed in the Dawson’s Field Arbitration. The test assesses whether multiple losses “involve such a degree of unity as to justify their being described as, or as arising out of one” event. “In assessing the degree of unity regard may be had to such factors as cause, locality and time and the intentions of the human agents” (cited to and quoted in Kuwait Airways Corp v Kuwait Ins Co SAK (No 1) [1996] 1 Lloyd’s Rep 664).
C22 It is undisputed that there must be a causal link between the relevant event and the separate losses (unity of cause). This aspect does not, however, deal with the required strength of these causal links; it only establishes that there must be causal links. It is on the basis of the further aspects of “unity of locality”, “unity of time” and “unity of intentions” that the unities test is meant to assess the strength of the causal links. However, the fact that separate losses occur in close spatial proximity (unity of location), in close temporal proximity (unity of time) and may be the product of related human intentions (unity of intentions) has no bearing on the required level of causation but rather on the level of correlation. Therefore, the unities test can be considered a mixed test of causation and correlation.
C23 It is certainly possible to test whether separate losses may be aggregated on the basis of a mixed test of causation and correlation. However, testing the unity of location and the unity of time is only sensible if the meaning of proximity in space and time is defined from the outset. The lack of a clear understanding of what spatial and temporal proximity means can be seen as the gravest deficiency of the unities test, as it allows for result-oriented judgments (see William 4.72 et seq.).
Illustration
I3. In Aioi Nissay Dowa Ins Co Ltd v Heraldglen Ltd [2013] EWHC 154, 23, the court affirmed the arbitral tribunal’s findings that the New York Twin Towers were located in close proximity to one another but that this proximity “did not give rise to a sufficient degree of unity for them to conclude that” the losses that occurred to them 16 minutes apart from one another can be said to have arisen out of one event. Similarly, the court affirmed that although “there were similarities in the timing of the events from the commencement of the flights to contact with the Towers […] these were not such as to lead to the conclusion” that the losses can be said to have arisen out of one event. Furthermore, the court upheld the tribunal’s finding that, although the attacks on the Twin Towers could be said to be the “execution of a dastardly plot to turn each [aircraft] into a guided missile”, there was no basis “for concluding that there was any factor amounting to an event of sufficient causative relevance to override the conclusion that two separate hijackings caused separate loss and damage (Aioi Nissay Dowa Ins Co Ltd v Heraldglen Ltd [2013] EWHC 154, 21).
I4. In the Dawson’s Field Arbitration, by contrast, three aircraft were hijacked and flown to a remote desert airstrip in Jordan. Thereafter, all three aircraft were blown up one after another “in close proximity more or less simultaneously, within the time span of a few minutes, and as a result of a single decision to do so without any one being able to approach the aircraft between the first explosion and their destruction.” All the losses that resulted from the destruction of the three aircraft were held to have arisen out of one event (Dawson’s Field Arbitration, quoted in Aioi Nissay Dowa Ins Co Ltd v Heraldglen Ltd [2013] EWHC 154, 9).
C24 The two Illustrations above demonstrate that the result of the unities test very much depends on how it is conducted. In fact, in the case of the attacks on the Twin Towers, the time span deemed relevant was “the whole period from check-in and passenger scrutiny [i.e. even before the hijacking of the aircraft] to the collapse of each Tower and not just from the time each flight took off” (Aioi Nissay Dowa Ins Co Ltd v Heraldglen Ltd [2013] EWHC 154, 22), whereas in the Dawson’s Field Arbitration the relevant period was considered the time between the blowing up of the first and the blowing up of the third aircraft and not the whole period from the hijacking of the individual aircraft to their destruction.
C25 In light of the above, for the purposes of the PRICL, the unities test is considered inappropriate for determining whether a plurality of losses can be aggregated.
c. Causation: Unfortunate event test is inappropriate
C26 Applying New York law, courts use the so-called “unfortunate event test” to determine whether multiple losses have arisen out of one single or of multiple events. The application of the unfortunate event test is, however, limited to third-party liability insurance cases (cf. World Trade Center Properties, LLC v Hartford Fire Ins Co, 345 F3d 154, 190 (2d Cir 2003)).
C27 The unfortunate event test determines whether multiple losses result from one event of an unfortunate character that takes place without one’s foresight or expectation. The question at the core of the test is whether the liability arises out of one single or multiple “incidents”. If liability arises out of multiple incidents, they may still be considered one event if they occur in temporal and spatial proximity and are part of the same causal continuum.
C28 There is no hard and fast rule determining a particular number of seconds or minutes that must elapse before two incidents are distinct events. Instead, it is relevant whether the relative timing of the various incidents has played a role in causing any of the other incidents. Equally, spatial proximity appears to be relative. Furthermore, to be part of the same event, the operative incidents must be part of the same causal chain. Once an incident occurs and that incident does not, in turn, cause one or more of the subsequent incidents, the causal chain is broken.
C29 Unlike the unities test, the unfortunate event test not only requires that the various losses arise out of the same event but additionally that one liability triggering incident causes another such incident.
Illustrations
I5. Reinsured A provides primary automobile liability insurance to Primary Insured C. Primary Insured C collides with a car in one incident “but an instant” before striking another car in a second incident. The second incident is a consequence of the first incident.
Applying the unfortunate event test, the two collisions are to be considered one single event due to their temporal and spatial proximity. Furthermore, as one incident was consequential for the other one, the two incidents were part of the same causal continuum (facts based on Hartford Accident & Indem Co v Wesolowski, 33 NY2d 169, 907, 910, 305 NE2d 907 (NY 1973)
I6. Reinsured A provides construction insurance to Building Company C. Building Company C constructs two independent protecting walls located blocks apart from each other at different job sites. During heavy rainfall, one wall collapses under the water pressure and destruction pours into a building. Almost an hour later, the second wall gives way and water floods another building. The failure of the first wall did not cause the failure of the second. Furthermore, the rainfall in itself did not damage the walls. Rather, the walls collapsed primarily due to faulty construction (facts based on Arthur A Johnson Corp v Indem Ins Co of N Am, 7 NY2d 222, 227 et seq. (NY 1959)).
Applying the unfortunate event test, the two collapses are to be considered two events as the failure of the first wall did not cause the failure of the second one, and the collapses occurred at different locations and at different times.
C30 The connection between the criteria of temporal and spatial proximity as well as the criterion of causal continuum has not been clearly established. Therefore, it is not easy to predict the impact of the different criteria in a specific case. It has been said that courts have reached inconsistent results even when applying the test to similar fact patterns (Murray 1502).
C31 In light of the above, for the purposes of the PRICL, the unfortunate event test is considered inappropriate for determining whether a plurality of losses can be aggregated.
d. Causation: Proximate cause test is inappropriate
C32 Particularly under Californian law, courts determine the degree of causation between the unifying factor and the separate losses by means of the so-called “proximate cause test”. This test should not be confused with cause-based aggregation under Article 5.3 or cause-based aggregation under English law (cf. Various Eateries Trading Ltd (formerly Strada Trading Ltd) v Allianz Ins Plc [2024] EWCA Civ 10 [53] et seq.).
C33 The principal rationale underlying the proximate cause test is that any losses resulting from a single proximate, uninterrupted and continuing cause may be considered as having arisen from one single event. In law, an act or omission is the proximate cause of a loss if, in a natural sequence unbroken by any new efficient cause, it produces that loss, and in the absence of which the loss would not have occurred. If, on the other hand, a cause is interrupted and replaced by another intervening cause, then the chain of causation is broken, resulting in two or more events depending on the number of intervening causes. Any assessment of whether there is proximate cause is based on reasonability.
C34 The proximate cause test is similar but not identical to the approach taken under the PRICL. The similarity lies in the fact that, under both tests, the unifying factor must be the “direct” cause of the separate losses in order to aggregate them. The difference is that, under Article 5.2, the unifying factor must be an “event” rather than any “cause”. The concept of “event” is a narrower one than the concept of “cause” (see Comments 2 et seq. to Article 5.3).
C35 In light of the above, for the purposes of the PRICL, the proximate cause test is considered insufficient for determining whether a plurality of losses can be aggregated.
8. No presumption of back-to-back aggregation mechanisms
C36 There is no presumption that aggregation clauses in contracts of reinsurance are to be interpreted in compliance with aggregation clauses in the primary insurance policies (see William 7.23 et seq.). In fact, the Principles have not adopted a presumption of back-to-back presumption (see Comment 33 to Article 6.1.1), subject to that an incorporated matching term (as defined in Article 6.1.1(2)) is given the same interpretation as its corresponding term in the underlying insurance contract in accordance with Article 6.1.3.
Illustration
I7. The aggregation clause in the underlying errors and omissions insurance contract reads: “any one occurrence or series of occurrences arising from one originating cause”. Hence, the primary insurance contract provides for a cause-based aggregation mechanism (facts based on Cox and Ors v Bankside Members Agency Ltd and Ors [1995] CLC 671 (CA) 679).
The primary insurer has taken out non-proportional reinsurance in the matter. The relevant contract provides for the following aggregation clause: “For the purpose of this reinsurance the term ‘each and every loss’ shall be understood to mean each and every loss […] arising out of one event”. By contrast to the primary insurance contract, the contract of reinsurance provides for an event-based aggregation mechanism (facts based on Axa Reins (UK) Ltd v Field [1996] 1 WLR 1026, 1031 et seq.).
The two contracts’ aggregation clauses cannot be said to be back-to-back and thus merit autonomous interpretation.