Presentation is loading. Please wait.

Presentation is loading. Please wait.

BUSINESS CYCLES AND THE IMPACT OF NATURAL HAZARD EVENTS A Parochial Reinsurance Market View David Simmons: Managing Director Analytics, Willis Re Imperial.

Similar presentations

Presentation on theme: "BUSINESS CYCLES AND THE IMPACT OF NATURAL HAZARD EVENTS A Parochial Reinsurance Market View David Simmons: Managing Director Analytics, Willis Re Imperial."— Presentation transcript:

1 BUSINESS CYCLES AND THE IMPACT OF NATURAL HAZARD EVENTS A Parochial Reinsurance Market View David Simmons: Managing Director Analytics, Willis Re Imperial College: 26 th March 2013

2 Catastrophe Reinsurance Pricing The “Traditional” Model  Catastrophes are, by their nature, rare events – Before the “modelled age” pricing was based upon recent loss history and required return – Pricing at near return period dictated by recent history (burning cost) – Pricing at far return periods set by minimum return requirements (minimum rate on line)  Concept of “the bank” and “payback” prevailed – When loss occurred reinsured was in effect calling in their “bank” of premiums paid in clean years – If bank insufficient then rates in future years increased so that reinsurer was paid back over a fixed time period – But these arrangements were non-contractual, market practice only  Result was that catastrophe reinsurance pricing was very reactive – When losses occurred prices increased steeply – In period of no losses prices tended to drift down due to market pressure – Exacerbated by tendency for some reinsurers to exit post-loss and new entrants emerge when rates are high 2

3 Catastrophe Reinsurance Pricing 1990s UK Catastrophe Example  Catastrophe Market in 1990 was already stressed – Large “1 in 100” windstorm loss in 1987 - 87J – USD 3.1m (original values per Munich Re) – Other market losses: Piper Alpha and Hurricane Gilbert (1988), Hurricane Hugo Exxon Valdez tanker (1989) tested catastrophe and specifically the Lloyd’s market  Storm 90A or Daria in January caused insurance losses event greater than 87J – USD 5.1m – Followed by a series of other smaller storms including Vivian in February costing USD 2.1m  In 1991 UK catastrophe prices reinsurance prices spiked in reaction to these losses – Prices more than tripling on average (source Willis Re) – Prices continued to increase in 1992 (impact of Hurricane Andrew) and 1993 as the LMX spiral, partially caused/revealed by this sequence of losses reduced ability of reinsurers to protect themselves so further reducing capacity – Prices peaked in 1994 with UK catastrophe reinsurance rates over 5 times 1990 levels 3

4 The New Modelled World  1991/2 saw the first UK windstorm models – Concepts outlined by Don Friedman in 1984, put into practice by Karen Clark in the late 80s for US Hurricane  Prevailing view was that new modelling would damp reinsurance pricing movements – Pricing now technical rather than reactive – New market entrants in Bermuda aggressively predicated their offering on this new technical approach – Beginning of breakdown of old bank/payback model, Insureds were tempted by lower prices of new technical reinsurers, breaking gentlemen’s payback agreements  Threat of Capital Markets entry to market was widely believed to further constrain pricing – New Bermuda capital could leave as fast as it arrived, triggering price increases? – But capital market players, with “infinite capital” attracted to new zero beta class would stay/pile in post loss? – Prices declines steadily from 1994 to 2000 as confidence In the modelling increased and memory of 1990 weakened, helped by a benign period for European Storms and the catastrophe market 4

5 But shocks still have an impact  9/11 in 2001 provided an unexpected shock to the system – Not a UK loss, not a natural catastrophe, but a major threat to the health of reinsurers – Market Loss circa USD 32m, over 50% higher than the highest natural catastrophe, Hurricane Andrew – P&C insurers suffered real losses to their capital (chart below source Insurance Information Institute) – Price impacts were felt throughout the market, UK prices jumped despite there being no underlying change to the assessed UK catastrophe risk and no actual UK catastrophe losses – The reactive kick-up in pricing was not limited to the UK – all markets showed a similar picture – Although not a model failure, the multi-class nature of loss caused reinsurers to question their base assumptions 5

6 Post 9/11 A series of disappointments  The catastrophe market has proven to be very resilient in the current millennium despite a series of major events, each revealing a flaw in underlying modelling assumptions – Hurricane Katrina: Levee burst/flood not modelled – Hurricanes Katrina/Rita/Wilma: Hurricane clustering – Sichuan Earthquake: Missed fault – Japanese Earthquake: Tsunami not modelled, intensity of earthquake on fault – New Zealand Earthquake: Liquifaction impacts, intensity of earthquake on fault, aftershocks – Australian Flood: Unmodelled, scale/intensity, classification (riverine vs flash flood) – Thai Flood: Unmodelled, contingent business interruption claims, scale  But the re/insurance industry remained resilient to all of these despite modelling flaws – Why? Despite problems with catastrophe models, their introduction has lead to as greater appreciation of risk, portfolio development, aggregate control and data quality  Capital market involvement in reinsurance is growing BUT not reason for stability – Capital markets took fright after “model error” of Katrina, retreated from indemnity deals to parametric trigger – Now back, largely driven by seeking any asset with a return with low correlation to market risk – Ironically, it was market risk that caused the biggest impact on re/insurers, the 2008/2009 asset crash, but no significant long-term casualties (other than AIG) 6

7 Pricing trends from 4 major markets 7

8 Current Research Impact on Economy of Cat Events  What impact does a major catastrophe have on the economy of a country? – Recent losses have given us some evidence to chew on – Consider the Kobe (Great Hanshin) Earthquake of 17 th January 1995  Kobe Earthquake – 6,500 dead – Economic Loss circa USD 100bn (Insurance Losses though of order of only USD 3bn) – By end 1996  manufacturing back to 98% of pre-earthquake trend  All department stores and 795 of city stores reopened  Import trade through port fully recovered  Export trade through port recovered to 85% below 1994 level – Both the Nikkei 225 and the USD/JPY rate broadly recovered to pre-earthquake levels after 9 months – Only significant broader financial impact was collapse of Baring’s Bank due to Nick Leason’s fraudulent dealings around movements in the Nikkea index  What impact does an extreme catastrophe have on the global economy? – Consider the San Francisco Earthquake of 1907 and potential of a Beijing Earthquake 8

9 San Francisco Eathquake 1906 impacts  Consider Paper by Odell and Weidenmeier “ Real Shock, Monetary Aftershock@ the 1906 San Francisco Earthquake and the Panic of 1907”, published in 2004  Causal chain proposed as follows: 1. Earthquake strikes San Francisco. 2. Claims are paid in gold by English insurers. 3. Gold flows into the US, thus increasing liquidity in the country. 4. Bank of England reacts to gold outflow by raising interest rates (and also taking other monetary restriction measures, such as ban on discounting US bills). 5. Gold flows from New York back to London. 6. Liquidity squeeze results in New York, causing stock market crash and drop in industrial activity. 7. Short term interest rates rise in the US, thus offsetting Bank of England’s action.  The world today is very different: – Thankfully claimants don’t demand to be paid in gold – Arguably, financial instruments are more liquid (if not under stress) – But could a similar chain occur today? – My colleague at Willis Re, Giorgio Brida, has proposed a potential 21st Century equivalent, a Beijing Earthquake 9

10 Updated Odell-Weidenmeier model Beijing Earthquake  Assume a major Beijing Earthquake at a point in the near future (one where insurance penetration rates in China begin to approach those of the West) 1. A very big earthquake strikes in Beijing 2. Foreign reinsurers are required to cover a significant amount of claims 3. Foreign reinsurers have to sell a significant amount of foreign assets, since they could not cover their Chinese risks with Chinese investments 4. Let us suppose that reinsurers sell T-bonds. Treasury yield spike in the US 5. The Fed cannot “sterilize” the sell-off because it does not have enough room for manoeuvre, due to the huge expansion in its balance sheet following the continuing financial crisis 6. Recession in the US, possibly aggravated by liquidity squeeze on shadow banking system. … could be offset by monetary inflows from other countries … but then US dollar appreciates, thus hampering US firms international competitiveness …  A related scenario have also been discussed for a large Tokyo earthquake – It envisages a large repatriation of Japanese funds from overseas investments in additional to a call on foreign reinsurers – The Kobe incident was large but well within the margins of Japan’s GDP (loss estimated at 2% to 2.5% of GDP) – But a repeat of the 1923 Great Kanto Earthquake in the Tokyo bay area could have a global impact 10

11 Lessons learnt  Reinsurers are pretty resilient – Very few recent failures – Catastrophe modelling has improved risk understanding but has not performed well when tested – Greater awareness now of unmodelled risks and perils – Focus on policy wordings (eg to avoid contigent BI issue for Thai Flood) and exposure control  Financial shocks can hurt as much as insurance ones – But insurers now largely de-risked – Implies lower investment returns to act as buffer to insurance cycles/catastrophe impacts?  Potential systemic risk from model use? – Regulators (eg Solvency II) are avoiding endorsing a model or models (like Florida) rather encouraging companies to take their own view of risk – BUT in practice difficult to be the one different from the others – “Don’t get sacked for buying IBM” = “Don’t get sacked for using RMS?” – Need contrarians to ensure robustness?  Be wary of surprises – Many Japanese insurers suffered more form the Thai Floods than the Japanese Earthquake/Tsunami – It’s the unknown unknowns that hurt every time 11

12 A Parochial Reinsurance Market View David Simmons: Managing Director Analytics, Willis Re BUSINESS CYCLES AND THE IMPACT OF NATURAL HAZARD EVENTS

Download ppt "BUSINESS CYCLES AND THE IMPACT OF NATURAL HAZARD EVENTS A Parochial Reinsurance Market View David Simmons: Managing Director Analytics, Willis Re Imperial."

Similar presentations

Ads by Google