17th An Approach to Pricing Natural Perils Tim Andrews David McNab Ada Lui Finity Consulting Pty Ltd 2010
Why is everyone talking about the weather?
The Melbourne and Perth storms were further evidence that severe weather is becoming more frequent and severe, according to Mike Wilkins, CEO of IAG. His actuaries calculated the likelihood of two $1 billion events occurring in quick succession to have a 1 in 1400 year probability (Sep 2010) "Climate change, we believe, is a fact. It triggers natural disasters and the number of these natural disasters has more than doubled since the 1980s Ernst Rauch, Munich Re (Jul 2010) We ve seen more losses in areas in terms of threat and severity that we haven t seen before Robert de Souza, Aon Benfield (Nov 2010)
Historical Estimated Underwriting Ratio Home 20% 10% 0% 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009-10% -20% -30% -40% Source: ISC and APRA to 2001, ISA from 2002 onwards
Breakdown of Weather and Attritional Losses - Home 140% 120% 100% Loss Ratio 80% 60% 40% 20% 0% Mar-02 Sep-02 Mar-03 Sep-03 Mar-04 Sep-04 Mar-05 Sep-05 Mar-06 Sep-06 Mar-07 Sep-07 Mar-08 Sep-08 Mar-09 Sep-09 Mar-10 Attritional Weather & Events Source: ISA, Insurance Council Disaster List
Our Objective Use examples of bushfire, storm and cyclone to illustrate approaches that may be used to estimate cost of natural perils Focused on Home
Roadmap Background What do natural perils cost? Growing availability of data What overall cost to allow for Examples: Bushfire, Storm, Cyclone
Background
What do natural perils cost? On an annual basis, estimate of industry ground up average cost for Home: Peril AAL ($m) Storm $800m to $1,000m Bushfire $100m to $150m Cyclone $150m to $250m Earthquake $50m to $100m Total $1,100m to $1,500m Average annual cost over last 5 years: Approximately $3.0 billion (all Home costs)
Available data is expanding all the time Insurance Council Catastrophe listing Flood database Geographic data Some of this already compiled by third parties Bureau of Meteorology wind and rain data Insurer s own experience, or pooled industry data (eg Insurance Statistics Australia) Catastrophe modelling results Competitor rates.but there are still a lot of gaps
What overall cost should we allow for?
Natural Perils Cost Build Up Earthquake RI Premium Cyclone Bushfire Loss Ratio (%) Average Cost of Retention Loss Ratio (%) Flood (nonriverine) Hail Attritional Events Storm
Example 1: Bushfire
Drivers of Bushfire Risk Broad geographic factors Humidity Temperature Factors within region and potentially address-specific Site specific landscape (eg slope of land and nearby topography) Proximity to fuel Relatively simple approaches can probably give significant gains for many insurers
Bushfire Average Annual Loss 1,400 Inflated Insured Cost (2010$) 1,200 1,000 800 600 400 200 Hobart Adelaide Hills Ash Wednesday Canberra 0 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 Black Saturday Source: Insurance Council disaster list, adjusted for population growth and price inflation Historic average annual loss of around $85 million Allowance for some events not already in data Is climate change impacting? Could assume around $100 million in average annual loss
Risk Index Benefits of having a risk index Compare for an identical risk on a zone by zone basis Assists with technical costing Can compare indices based on different sources Can be useful for checking from an intuitive point of view
Variations by State 1,400 Inflated Insured Cost (2010$) 1,200 1,000 800 600 400 200 State Insurance Council PerilAus Other Selected ACT 14.5 0.8?????? NSW 1.0 1.0?????? NT 0.0 0.2?????? QLD 0.0 0.5?????? SA 4.2 0.4?????? TAS 8.0 1.2?????? VIC 4.1 0.6?????? WA 0.0 0.6?????? 0 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 Note: PerilAus index based on analysis of PerilAus scores by postcode ACT NSW NT QLD SA TAS VIC WA Source: Insurance Council disaster list, adjusted for population growth and price inflation
Zone 34 Dandenong Ranges
Bushfire Risk - Kinglake VIC 2009 Bushfires Around 900 residences Most are fairly close to bushland
17th 7 10 November 2010 Sheraton Mirage, Gold Coast Bushfire Risk Boronia Around 12,000 addresses Only a small number are close to bushland
Bushfire Top Down Approach VIC $40m (12%) Zone Rel: 2.0 Zone Rel: <1 ICA Zone 34 (Dandenong ranges) $5m (24%) P/C Rel: 2.5 P/C Rel: 0.2 3763 Kinglake $0.1m (60%) Premium loadings up to 100% State Rel: 2.0 Australia $100m (5% of gross building prem) 3155 Boronia Other Zones $0.01m (4%) Other states $35m Premium loadings from 0% up Note that these numbers are purely illustrative $60m State Rel: <1 State relativities States with substantial risk are VIC, NSW, SA, TAS Mapping of number of houses ISA data PerilAus ICA relativities Combination of State and Postcode approaches Postcode relativities Index based on number of houses proximity to bushland
Example 2: Storms
Key Challenges for Storm What is the true annual cost given the historical experience is so volatile? How should this cost be allocated within the portfolio?
Average Annual Loss Has it been more or less stormy than average? 200 National Storm Index 180 National Storm Index 160 140 120 100 80 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 8 Year Average Annual Score Note: Indices based on a subset of Bureau of Meteorology rain and wind data to 30 June 2010. Index set to a base of 100 at June 2010 Storm scores could be used to assess whether last 3, 5, 10 years of actual data is reflective of the longer term
Storm Risk Index Top down approach to allocation What can be used to establish geographic relativities? Historical data BoM rainfall and wind storm scores Elevation data Other interesting risk measures PerilAus, NFID Index (NSW Average set at 100) 160 140 120 100 80 60 40 20 0 Storm Score Index (8 Year Rolling Average) 1949 1952 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 ACT NSW NT QLD SA TAS VIC WA Note: Indices based on a subset of Bureau of Meteorology rain and wind data to 30 June 2010. Index set to a base of 100 for the average of NSW.
Example 3: Cyclone
Cyclone 4,000 3,500 Inflated Insured Cost (2010$) 3,000 2,500 2,000 1,500 1,000 500 0 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 Source: Insurance Council disaster list, adjusted for population growth and price inflation Sparse losses in historical data More appropriate to use a bottom up approach
Cyclone Bottom up approach Reliance on catastrophe modelling Need to specify the right risks to input Different use of reinsurance models Not PML, rather cost per risk
Cyclone Risk
Summary of conclusions
Summary Build up of natural perils cost Top down/bottom up approach Use as much data as possible Construction of a risk index
Questions?