A perfect storm?
A 'perfect storm' describes an unusual combination of circumstances that will result in extreme outcomes. The spate of major catastrophes that the Asia Pacific region has encountered since 2010 has certainly been unusual, the question remains as to the outcomes that will result.
Readers will need little reminding of the recent events; unprecedented hailstorms in Melbourne and Perth in March 2010, a damaging earthquake in Christchurch NZ in September, and flooding in Queensland, Australia culminating in the Brisbane floods. ANZ then saw tropical cyclones Anthony and Yasi, flooding in Victoria, and an even more damaging second earthquake in Christchurch. Japan then witnessed the Tohoku earthquake, and catastrophic tsunami.
In total (including the Chilean earthquake) Swiss Re estimates the insured catastrophe losses for this period as between USD 30 – 50bn, with the reinsurance market burning 2 – 3 times the annual global property catastrophe premium of about USD 12bn.
These events have uncovered shortcomings in most if not all of the loss models reinsurers use to price business.
Lessons to date
Both the Perth and Melbourne hailstorms, costing around USD 1bn each, were dramatically larger events than anticipated – Perth had not previously even been considered as a major hail exposure.
These underline the general inadequacy of expected loss estimates for unmodelled perils and scenarios. Most underwriters rely either on a general loading to cater for these exposures, or a post loss 'burning cost' approach – both of which would probably have underestimated the impact of these events.
Liquefaction in Christchurch was much greater than catered for in the modelling, and modelling error seems to have arisen in the areas of construction quality, business interruption and the interaction between private insurance and the government Earthquake Commission. All this was compounded by the two losses occurring so close together.
From the Tohoku earthquake, the learnings seem to be in the business interruption area. Japan is a market with comparatively low business interruption penetration. Contingent business interruption losses outside Japan however might be very significant, although sublimited. In the recent market conditions, sublimits have increased, and can now be as much as USD 1bn per insured. Both Tokyo and Christchurch remain exposed to aftershocks with a higher short term frequency than probabilistic loss models calculate.
In several of these events, the impact of catastrophe losses on the results of Surplus treaties has been dramatic – providing cover with unlimited reinstatements, Surpluses generate insufficient margins to fund low frequency losses.
At least one insurance company appears to have incurred losses to Christchurch which significantly exceed its reinsurance cover – boards and regulators throughout Asia Pacific might be prompted to re-evaluate the adequacy of reinsurance limits and modelling assumptions.
Asia Pacific catastrophe exposures have been seen as generally diversifying for global reinsurers – this works only when margins are generated over the long term. The modelling gaps in these events also show the dangers inherent in diversifying exposures – by nature, underwriters know them less well!
In loss-free periods, underwriters generate an optimistic bias, tending to assume that the absence of loss equals absence of exposure, and overlook the potential for bad news. Post event, many over-correct.
Outlook for the market
So what is the likely outlook? The arguments for a global market hardening have been building, as unexciting investment returns shift focus back to underwriting results. Solvency regulations are under review, increasing the capital burden for re/insurers. Run off gains from prior years' underwriting are slowing. For catastrophe business, a recent model revision in the US will increase both demand, and expected price for several key scenarios, while recent losses have burnt multiples of the global catastrophe premiums.
Conversely, re/insurers have been able to rebuild capital bases, and the industry is now recapitalised to pre- GFC levels. The market is finely balanced between these conflicting pressures, with each event shifting the scales towards a hardening market.
In Asia Pacific regardless of global trends, these events can be expected to exert pressure on reinsurance pricing, both from modelling, and an increased margin expectation. For example, the April Japan renewals saw up to 50% increases on earthquake programmes, and up to 10% for wind-flood. Capacity for retrocession, and low rate on line layers was in shorter supply.
In all regions we should expect underwriters to consider the learnings from these losses, and challenge any optimistic bias – these might include: earthquake exposures especially business interruption for more exposed industries, unmodelled perils, broad policy coverage, and infectious disease extensions. A sharper focus on understanding the underlying exposures, querying modelled output, and unmodelled exposures should result.
The events of 2010 and 2011 might not yet constitute the perfect storm to push the global cycle past the tipping point – time will tell, but as we enter a North Atlantic Cyclone season which has the potential to remove doubt as to the direction of the market, in Asia Pacific we can expect some corrections.
Published May 2011