The reinsurance sector has undergone significant change over the last decade or so. During this time, the marked increase in the supply of dedicated reinsurance capital has been the predominant market driver and conditions have been generally favourable to reinsurance buyers. So even after the sector had clearly suffered one of its most expensive catastrophe loss years on record, JLT Re in last October’s Winds of change Viewpoint report was able to predict with confidence that the abundance of capital would prevent a return to the sort of pricing volatility and capacity constraints that had followed other prominent catastrophe loss years. And this is exactly what has happened. Three key factors have contributed to this outcome. It will come as no surprise that capital has been the primary driver for suppressed rate rises through 2018; but the lack, so far, of HIM claims development and loss activity in 2018 have also played important, albeit secondary, roles.
The abundance of capital has been crucial in suppressing rate increases after HIM. Figures 5 and 6 provide a carrier-level comparison of the relative impact of catastrophe losses in both 2005 and 2017.
Figure 5: Impact of 2005 Catastrophe Losses on Major Carriers (Source: JLT Re)
These charts clearly illustrate why KRW turned the property market whilst HIM merely brought about moderate rate firming. With roughly similar impacts to sector earnings, the key difference between the two periods was the relative amount of dedicated sector capital that was displaced by the catastrophes, with only three (re)insurers losing a quarter or more of their shareholders’ equity in 2017, compared to at least eighteen carriers in 2005. Twelve of the 30 carriers hit hardest in 2005 subsequently ceased trading as a direct or indirect result of the losses.
Figure 6: Impact of 2017 Catastrophe Losses on Major Carriers (Source: JLT Re)
Unlike in 2005, traditional carriers last year faced the losses from a position of capital strength. And when combined with other factors, such as significant losses filtering into alternative markets for the first time, (rating agency-led) diversification away from property-catastrophe business and the enhanced role of catastrophe modelling in the underwriting process, losses in 2017 mostly hit earnings. As a result, reinsurance capital was only marginally affected by HIM. Since then, sector capital has continued to grow strongly, with roughly USD 7 billion of new capital raised in the final four months of 2017 alone. Further growth during the first two quarters of 2018 meant sector capital was at record levels by the end of the first half of the year (see Figure 7).
Figure 7: Dedicated Reinsurance Sector Capital and Gross Written Premiums (Provisional) – 1998 to H1 2018 (Source: JLT Re, Swiss Re)
The proficiency of capital markets in reloading so quickly (and smoothly) after HIM was crucial to this outcome. This meant capacity levels were often found to be higher in 2018 than at corresponding renewals last year. Figure 8 shows some of the new reinsurance capital that has entered the alternative market since HIM made landfall, with substantial capital raising taking place for sidecar vehicles late last year and in time for this year’s mid-year renewals, in addition to significant new catastrophe bond issuance.
Figure 8: Selection of Announced New Reinsurance Capital – October 2017 to August 2018 (Source: JLT Re)
In fact, catastrophe bond issuance reached near-record levels of USD 8.5 billion during the first half of 2018, second only to H1 2017 (USD 9.7 billion).
This seemingly confirms a structural change in how capital is provided to the reinsurance market, with third-party capital entering the sector post-loss to fill gaps more or less immediately.
As with traditional property-catastrophe cover, much of this capacity has been deployed at a lower rate of return than has historically been the case, particularly in areas where ILS markets are most active, such as Florida.
Figure 9 shows how the spread between average catastrophe bond yields and expected losses narrowed during the first half of 2018 as strong investor appetite enabled cedents to transfer catastrophe risks at attractive pricing.
Figure 9: Catastrophe Bonds Coupon and Expected Loss – 2009 to H1 2018 (Source: JLT Re)
2. HIM DEVELOPMENT
An additional factor that has contributed to lower rate increases in 2018 has been claims development trends for HIM thus far. Losses for previous, costly events have often ended up considerably higher than original expectations, resulting in carriers having to make significant revisions to their initial loss estimates. As illustrated by Figure 10, loss expectations for expensive North Atlantic hurricane events have been subject to significant development. This graphic uses data from Property Claim Services (PCS) to show the development of claims up to 350 days after landfall for all North Atlantic hurricanes since 2000 that have incurred a total insured cost in the US of more than USD 10 billion.
Figure 10 shows that significant adverse development has been a consistent theme for the majority of these hurricanes, with loss expectations for Sandy, Wilma and Ike undergoing substantial revisions during the first 250 days after landfall. Estimates for Hurricane Katrina were also adjusted up during the same timeframe, albeit to a lesser extent (in percentage terms). This is, of course, to be expected: loss expectations undergo significant revisions as more up-to-date information emerges, particularly for more complex events which often bring unforeseen and unmodelled consequences.
However, the lack of comparable development for HIM within 250 days after landfall was striking, especially given the levels of complexities involved for all three storms. Whilst loss revisions for Harvey since then seem to have stabilised (not too far off original estimates), those for Irma and Maria have been adjusted up, although they continue to remain significantly below the levels reached by Sandy, Ike and Wilma in percentage terms. In fact, nearly a year on, there is still some uncertainty as to whether aggregate losses from HIM will reach some of the higher-end ranges predicted by the modelling companies and others immediately after landfall. This has been reflected by a number of notable HIM-related reserve releases by some insurers and reinsurers so far this year.
Figure 10: Claims Development for Costly North Atlantic Hurricanes – 2000 to 20173 (Source: JLT Re, PCS)
3 Released with the permission of PCS.
A combination of conservative initial loss picks and improved loss estimation methodologies may have helped prevent adverse development to the degree witnessed during previous large-loss years.
It is nevertheless important to note that HIM-related releases have not been universal and, perhaps more importantly, that claims continue to develop. Losses from previous hurricanes have typically taken less than 350 days to reach close to their final tallies but it currently looks as if last year's losses will take longer to develop fully. Irma and Maria in particular look set to have longer tails than usual for hurricane events: in both instances the desire to settle and close claims quickly has led to a high number of cases being re-opened.
Figure 11: Evolution of Estimated Insured Losses in Florida for Hurricane Irma (Source: JLT Re, FLOIR)
This is causing the cost of claims to increase beyond some (re)insurers’ original expectations, particularly in Florida where carriers are continuing to encounter long-standing assignment of benefits (AOB) issues in what is an aggressive legal environment. Unexpectedly large loss adjustment expenses (LAE) are compounding the situation after Harvey inflated the cost of claims adjusters for subsequent events in 2017, including Irma and Maria.
All this is also seemingly reflected by claims data for Irma collected by the Florida Office of Insurance Regulation (FLOIR). Figure 11 shows how the FLOIR’s estimated cost of reported claims has evolved since Irma's landfall. Whilst the cost of claims initially conformed to the expected pattern of an immediate spike followed by more gradual increases as the number of new claims declined, the continuation, and even acceleration, of this trend beyond 330 days may imply loss creep linked to the reopening of reported cases. Indeed, a number of specialist insurers in Florida have recently reported adverse development for Irma. Given that these carriers typically buy high levels of reinsurance cover which trigger at low attachment points, reinsurers look set to incur the bulk of the costs.
3. 2018 LOSS ACTIVITY
Benign loss activity so far in 2018 has been another factor in stifling price increases (see Figure 12). Despite an above-average number of events occurring during the first half of the year, insured losses were slightly below long-term trends at just over USD 20 billion. This has provided some respite and allowed carriers to restore capital buffers. A repeat of losses sustained during the first half of 2011, when (re)insurers paid out more than USD 75 billion, would likely have tested carriers’ capital resilience.
Much, of course, could still change should the wind blow again this year. Of the top 15 most costly losses ever to impact the (re)insurance sector, two thirds have been caused by hurricane landfalls in the United States and Caribbean during the peak months of August, September and October.All eyes are therefore firmly fixed on this year’s hurricane season. Most season forecasters now expect below average activity, having downgraded earlier predictions to account for unseasonably cold waters across the tropical Atlantic and Caribbean and the possibility of a developing El Niño towards the end of the season. But history has shown it only takes one large event to wreak major destruction. Whatever the outcome, the reinsurance market is strongly positioned to deal with most potential loss scenarios.
Figure 12: Global Insured Catastrophe Losses by Quarter – 2014 to H1 2018 (Source: JLT Re)