PROBABLE MAXIMUM LOSSES OF INSURED ASSETS FROM NATURAL AND MANMADE CATASTROPHES IN AUSTRALIA WILLIAM GARDNER BEc OCTOBER 2006 A THESIS PRESENTED TO THE SCHOOL OF PHYSICAL GEOGRAPHY OF MACQUARIE UNIVERSITY IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE PHD DEGREE IN RESEARCH Declaration I, William Gardner, hereby declare that this thesis has not been submitted for a higher degree to any other university or institution, and to the best of my knowledge, represents my own work, except where due reference is made and acknowledged. October 2006 ii Abstract In a first attempt to quantify the national catastrophe exposure of insured property, this thesis uses a range of models of varying complexity to estimate the Australian industry’s exposure to natural perils. These perils include cyclones, earthquakes, hail, bushfire, flood and severe thunderstorms. The focus is exclusively on the insurance industry and so wider community economic losses are ignored. To achieve this goal, the author first developed a national database of residential and commercial property exposure at the postcode level of resolution. The next major tasks were to develop a stochastic national tropical cyclone loss model, implement a published earthquake model and combine the results of these models with extreme value statistics of losses from other perils. An Australian terrorism loss model developed by the author in the course of this thesis is also presented here to highlight differences between the modelling of man-made perils and those arising from natural hazards; the most notable difference being the difficulty in defining the probability of an attack compared with the return period of natural hazards of varying severity. The key results for natural perils are expressed as Probable Maximum Losses (PML) at given Return Periods and are given in the Table below. Return Period Cyclone Earthquake Other Perils Combined (years) ($m) ($m) ($m) ($m) 1,000 11,945 42,226 5,953 42,551 250 4,699 7,639 3,622 10,530 100 2,164 2,884 2,360 5,421 Probable Maximum Loss by Return Period These numbers can be used by insurers and international reinsurance companies to price catastrophe risk. The thesis quantifies the level of uncertainty in these results and provides commentary on how the information provided can be used by the insurance industry, the government and the public to provide a better understanding of natural peril risk to insured property in Australia. iii Acknowledgements I wish to thank the following people who have contributed to the production of this thesis: • my supervisors over the years: Professors John Pollard, Russell Blong and John McAneney; • my employers: Aon Re Australia, Impact Forecasting LLC and Aon Re Inc.; • along with others who have assisted me in various aspects of the thesis including Dr George Walker, Dr Siamak Daneshvaran, Phil Heckman, Dr Andres Mendez, Dr Pierre Wiart and John Aquino; and • most importantly, my wife Donna and my family who have patiently allowed me to work on and complete this research and document. iv Table of Contents DECLARATION .................................................................................................................... II ABSTRACT............................................................................................................................III ACKNOWLEDGEMENTS ................................................................................................. IV TABLE OF CONTENTS ....................................................................................................... V CHAPTER 1 - INTRODUCTION...........................................................................................1 Catastrophes in Australia .....................................................................................................................................1 Insurance Companies Catastrophe Concerns .....................................................................................................2 Insurance Company Need for Catastrophe Models............................................................................................3 Insurance Regulation in Relation to Catastrophe Models .................................................................................4 Guidelines in the United States.............................................................................................................................5 Objectives of Thesis ...............................................................................................................................................6 CHAPTER 2 - CATASTROPHE MODELS..........................................................................8 Definition ................................................................................................................................................................8 Probable Maximum Loss ......................................................................................................................................9 Model Complexity and Uncertainty ...................................................................................................................10 CHAPTER 3 - EXTREME VALUE ANALYSIS ................................................................13 Methodology.........................................................................................................................................................13 Analysis of Australian Catastrophe Data ..........................................................................................................13 Other Methods of Fit ...........................................................................................................................................15 Distribution of nth Largest Value........................................................................................................................17 Conclusions...........................................................................................................................................................18 CHAPTER 4 - EXPOSURE DATABASE............................................................................20 Overview...............................................................................................................................................................20 Census data...........................................................................................................................................................21 Insurer Data .........................................................................................................................................................24 Average Value ......................................................................................................................................................24 Commercial Risks................................................................................................................................................25 Motor Risks ..........................................................................................................................................................25 Construction Material .........................................................................................................................................26 Building Age.........................................................................................................................................................26 Total Exposure.....................................................................................................................................................28 Proportion Insured ..............................................................................................................................................28 v Data Quality .........................................................................................................................................................28 CHAPTER 5 - CYCLONE MODEL ....................................................................................30 Overview...............................................................................................................................................................30 History of Tropical Cyclone Modelling..............................................................................................................30 Modern Models of Cyclone Risk ........................................................................................................................32 Cyclone Structure ................................................................................................................................................33 General Methodology ..........................................................................................................................................37 CHAPTER 6 – DEVELOPMENT OF A CYCLONE EVENT SET..................................41 Historic Event Database......................................................................................................................................41 Event Start Position Generation.........................................................................................................................43 Starting Central Pressure....................................................................................................................................51 Maximum Potential Intensities...........................................................................................................................54