Climate change and insurance study Executive Summary CLIMATE RISKS: IMPACT ON NATURAL HAZARDS INSURANCE BETWEEN NOW AND 2040

Transcription

1 Climate change and insurance study Executive Summary CLIMATE RISKS: IMPACT ON NATURAL HAZARDS INSURANCE BETWEEN NOW AND 2040

2 SUMMARY 4 Foreword by Jean Jouzel 6 Introduction 7 Study methodology 8 LOOKING BACK 10 PROJECTION TO PERIL BY PERIL ANALYSIS DROUGHT FLOOD COASTAL FLOODING STORM OVERALL CONCLUSION

3 FRENCH INSURANCE FEDERATION CLIMATE RISKS: IMPACT ON NATURAL HAZARDS INSURANCE BETWEEN NOW AND 2040

4 FOREWORD BY JEAN JOUZEL By the end of 2015, the eyes of all those concerned with the future of our climate were on Le Bourget, where the COP21 Agreement was signed. Almost all countries on our planet took part in COP21: it is to be hoped this will be the true starting point for the journey towards a low carbon world, one in which the long-term rise in temperatures has been controlled and, if possible, limited to 2 C, compared to the pre-industrial period. This is indispensable if we want today s young people and future generations to be able to adapt, at least in all essential ways, to a climate that will inevitably change. If action is not taken to achieve a rapid and ambitious reduction in our greenhouse gas emissions, it will be difficult to deal with the impact of a temperature rise that could, on average, reach 4 C to 5 C between now and 2100 and keep rising after that. These consequences, which the Intergovernmental Panel on Climate Change has classified in five categories, will affect almost all sectors of the economy. One of those categories concerns extreme climate events: droughts, heat waves, floods, hurricanes, etc. While the IPCC remains very cautious in attributing recent changes in these events to human activity, the diagnosis becomes clearer when we look at the future. In a warmer world, almost all these extreme events will become more frequent and/or more intense. This will be the case, for example, with heat waves and heavy rain over most continental regions in the mid-latitudes and humid tropics. While the number of hurricanes is not expected to rise, specialists predict an increase in the maximum wind speeds and rainfall associated with these events. These extremes are at the origin of a significant part of the costs insurers have to meet and it is therefore important for the industry to establish as accurately as possible the scale of the risks that will be associated with such events in the future. 4

5 That is the aim of this study, Climate Change and Insurance to 2040 (Changement climatique et assurance à l horizon 2040), which draws very heavily on the work of the scientific community and takes as its starting point the report Climat de la France au XXI e siècle (France s Climate in the 21 st Century) prepared by researchers at Météo-France and the Institut Pierre Simon Laplace (IPSL). It focuses on France and on the period up to 2040, a timescale over which climate change depends little on greenhouse gas emission scenarios. That is as true for the planet as a whole as it is for France which, by 2040, will see average temperatures rise by between 0.6 C and 1.3 C, and perhaps, in summer, by between 1.5 C to 2 C in the south-east of the country. Starting from these projections, the study draws heavily on the work on extreme events led by my colleague Pascal Yiou at the Environmental Science and Climate Change Research Institute (Laboratoire des Sciences du Climat et de l Environnement), using an innovative methodology that overlays economic projections and climate projections. It uses a peril by peril approach (floods, coastal flooding, drought, wind) and takes uncertainties fully into account, in particular with regard to the effect of climate change on storms. It provides valuable insights, supported by figures, and underlines the fact that, over the next 25 years, these hazards which will be less and less natural will have a cost. This study highlights the importance of adapting prevention policies, starting now, and of developing a risk awareness culture in our country; it represents a significant contribution by the insurance industry to raising the awareness of actors in the run-up to COP21. JEAN JOUZEL (Director of Research at the French Nuclear Energy Commission), (Vice-chairman of the IPCC from 2002 to 2015). 5

6 INTRODUCTION The increasing frequency and intensity of natural hazards is often cited as an early indicator of climate change. Insurers, using past figures for insurance payments for damage caused by natural hazards having occurred in metropolitan France, have attempted, with the study Climate Change and Insurance (Changement climatique et assurance), to reply to the following question: IS IT POSSIBLE TO PUT A FIGURE ON THE IMPACT OF CLIMATE CHANGE ON INSURANCE BETWEEN NOW AND 2040? A remark should be made about the time horizon used for this study. Twenty-six years is a short period in climate terms. Nevertheless, this horizon was chosen because putting a figure on the impact requires climate projections to be considered together with projections of social and economic assets and activities at risk. In terms of these latter projections, 26 years is a long time and indeed represents the limit for a credible projection. The period chosen is therefore a compromise. REMARKS The geographical scope of this study is limited to the territory of metropolitan France. The overseas territories and départements are exposed to specific hazards which must be analysed using other models than those used in this study. In addition, this study focuses on the direct damage caused to property by natural hazards, including operating losses. It does not include bodily injury and the damage caused to farm crops that have not been harvested and placed in store. The time reference period of the study is from 1988 to 2013 (for the Looking Back section) and from 2014 to 2039 (for the Projection section). 6

7 STUDY METHODOLOGY The study was based on a projection of the socio-economic data (housing and other assets and activities exposed to natural hazards) on the one hand and a climate projection on the other. For the socio-economic projections, the method consists of developing a scenario of the growth in wealth and its exposure in the next 26 years in France. For most of the criteria (number and size of dwellings, number and size of businesses, growth in geographical concentration) the most up-todate projections of the French National Institute of Statistics and Economic Research (INSEE) were used. They were supplemented by specific data on hazard areas collected by the Natural Risks Mission (MRN Mission Risques Naturels) 1. For the climate projections, the study called on the services of the climate and environmental science laboratory Laboratoire des Sciences du Climat et de l Environnement (LSCE). The laboratory provided the data from two of its climate models covering the period , both based on scenario RCP8.5 of the Intergovernmental Panel on Climate Change (IPCC). The two climate models used were a global model of the Institut Pierre Simon Laplace (hereinafter IPSL) and a regional model of météo France (hereinafter MF). The results of these two models were taken as the extremes of a range, with the study presenting the average of that range, except where stated otherwise. An array of climate indicators provided by these models was analysed over the last 25 years to test their relevance in terms of correlation with the annual amounts paid out by insurers. Based on their observed relevance, four indicators were used as references in the study s future climate projections. These were: MAXIMUM DAILY WIND SPEED MAXIMUM DAILY RAINFALL CUMULATIVE DAILY RAINFALL DAILY TEMPERATURE (MEAN, MIN, MAX) 1 MRN : Association created by the French Insurance Federation tasked with providing better assessment and prevention of natural hazards. 7

8 LOOKING BACK The history of insurance payments made following the occurrence of natural hazards is summarised in the following graph: In millions of current Euros NÎMES 288 M 1990 DROUGHT 355 M 1990 DARIA M FLOOD STORM DROUGHT 1992 VAISON LA ROMAINE 240 M 1999 LOTHAR & MARTIN M 2002 GARD 700 M 2003 RHÔNE 670 M 2003 DROUGHT M 2009 KLAUS & QUINTEN M 2010 VAR + XYNTHIA M 2010 XYNTHIA 735 M 2011 DROUGHT 800 M 2014 HAIL 850 M The breakdown of the total amount paid out by insurers over the last 25 years ( ) by type of peril is as follows: 50 % STORM, HAIL AND SNOW 34% FLOOD 16% DROUGHT The detailed breakdown of these figures, by amount and number of claims by category of insured parties (private individuals and business/professional) is shown in the table below: From 1988 to 2013 Number of claims paid Amount paid out (billions of constant Euros 2013) FLOOD ,6 Private individuals ,2 Business / professional ,4 STORM (SHS) ,1 Private individuals ,3 Business / professional ,8 DROUGHT ,6 All perils ,3 8

9 was the year that saw the lowest level of payouts, at 760 million euros, while 1999 saw the highest level, at 13.1 billion constant 2013 Euros claims paid per year 1,86 Billions of Euros/year During the last 25 years, insurers have paid out on an average of 431,000 property damage claims each year, amounting to 1.86 billion Euros, for property damage. 0,76 Billions of Euros The average amount paid per claim was 4,310 Euros. The breakdown of the average cost per claim by type of peril and category of insured party (private individuals and business/professional) is shown in the table below: AVERAGE COMPENSATION PAID PER CLAIM ( ) Private individuals 13,1 Billions of Euros Business / professional Total FLOOD STORM DROUGHT NS* Average all perils * Until now drought claims (shrinkage-expansion of clay) have only concerned private individuals 9

10 PROJECTION TO 2040 Is it possible to put a figure on the impact of climate change on insurance between now and 2040? In order to answer this question, the first stage of the study was to produce a socio-demographic projection for the period and link it to a climate projection, making it possible to project the amounts that insurers will pay out between now and The results were analysed to identify the impact of different factors in order to isolate the part that can be strictly attributed to climate change. This was done for each individual peril, identifying the impact on businesses and professionals (traders, artisans, businesses, local authorities, farms) on the one hand and the impact on private individuals (essentially housing) on the other. The different factors that can explain the difference between the facts observed over the last 25 years and the projections to 2040 are of two types: SOCIO-EONOMIC FACTORS Wealth Increasing wealth produces greater concentrations of businesses and housing. Regional and local authorities have more infrastructure. The consequences of a natural event of exactly the same intensity, occurring in exactly the same place, at an interval of 20 years, will increase in proportion to the rate of growth of wealth. This wealth factor includes the number of establishments factor (greater concentration of housing or businesses) and the magnification factor (each unit is more expensive). Distribution Independently of the overall rate at which the country s wealth grows, the geographical distribution of that growing wealth has consequences for the overall vulnerability to natural hazards. Clearly, a département having an overall vulnerability to floods that is twice the national average and experiencing twice the national average rate of growth of wealth creates a multiplier effect with regard to overall vulnerability. The same multiplier effect will also be seen if, within that département, the increases in wealth are concentrated, for example, along river banks. The diagrams below illustrate this phenomenon. They show the growth of an urban area between 1999 and This area of the municipality has been entirely built in an area prone to flooding, shown in blue. 10

11 1999 FLOODED AREA CLIMATE FACTORS Climate change This factor manifests itself in a change in the frequency of occurrence and intensity of natural events in the country. The study analysed projections of extreme events and those of more common events, referred to as background noise, separately. For extreme events we considered changes in the return period 2, while for more common events we considered frequency of occurrence. This factor is strictly linked to climate change AREA FLOODED IN 1999 PROPERTIES FLOODED IN 1999 NEW PROPERTIES BUILT SINCE 1999 Natural climate hazard Over the last 25 years, France has experienced significant natural events with a return period in excess of 25 years. On the other hand, there are events of this type which we have not experienced in the last 25 years but which could very well occur in the next 25 (major flooding of the Seine or the Loire, for example). The study therefore assigned a probability of occurrence to these events in the projections in order to obtain a better estimate, in actuarial terms, of future insurance payments. This climate hazard effect was treated as being separate from climate change because it reflects the natural variability of the climate as it has always existed. Thus, an overall growth in wealth of 10% can increase the vulnerability of a municipality by 50% or 75% if that growth takes place in a vulnerable area. We call this phenomenon the distribution factor. This distribution factor can be further broken down into the migration factor, which concerns migrations of wealth from a less vulnerable to a more vulnerable département, and the hazard area factor, which takes account, at a more local level, of developments in risk areas (flood-prone areas as in the previous example or areas of clay soil for drought risk). COASTAL FLOODING Coastal flooding is a peril that is not taken into account by the climate models used in the study. While this may be an effect linked to climate change, its origin lies in the rise in sea levels, which itself lies outside the regional climate indicators used in the study. In particular, the risk of coastal flooding associated with storms is excluded from the study. Consequently, this peril is dealt with in a separate chapter in the study, as it requires a different methodology to the flood, storm and drought risks. 2 Return period of a natural event: statistical time between two occurrences of a natural event of a given intensity. This period represents a probability, not a forecast. Thus, if the return period of a natural event of a given intensity is determined as being 10 years, that does not mean that it will recur every 10 years, but that it has a 10% probability of occurring each year. 11

12 PERIL BY PERIL ANALYSIS DROUGHT PAGE 13 FLOOD PAGE 16 COASTAL FLOODING PAGE 20 STORM PAGE 24 12

13 DROUGHT DROUGHT THE DROUGHT PERIL CONCERNS THE PHENOMENON OF SUBSIDENCE, I.E. THE EFFECT OF DAMAGE CAUSED TO BUILDINGS (ALMOST EXCLUSIVELY INDIVIDUAL HOUSES) OWING TO SHRINKAGE AND THEN EXPANSION OF SUBSOILS ASSOCIATED INITIALLY WITH INSUFFICIENT RAINFALL. This phenomenon occurs mainly on clayey subsoils, which are particularly vulnerable to subsidence. Such subsidence occurs mostly in some thirty départements of France (cf 1). CLAYEY SUBSOILS 1 SOCIO-ECONOMIC FACTORS INVOLVING CHANGES IN ASSETS AT RISK BETWEEN NOW AND 2040 The analysis of the projection over the next 26 years of assets at risk (private housing at risk of subsidence) can be summarised in map 2 below: Départements such as Morbihan or Cantal show a high growth of assets at risk even though they are not included in the list of départements at risk because of clayey subsoils. This is because, although they are not on that list, they do contain localised areas of clayey subsoils that coincide with areas where the forecast growth of such assets is very high. In total, ten départements have projected growth rates of more than 10% for assets vulnerable to drought peril in the period to SOIL TYPE CLAYEY SOILS NON-CLAYEY SOILS AVERAGE ANNUAL CHANGE 2 < 1% > +5% 13

14 DROUGHT CLIMATE FACTORS The two climate indicators used to project the drought peril over the next 26 years were levels of rainfall (cumulative and maximum daily rainfall) and temperatures (mean, maximum and minimum), which were studied in the framework of the two models: IPSL and MF1. The correlations obtained between these variables and drought events giving rise to insurance payments for losses give the following frequency changes between now and 2040 (3 & 4) : In the Météo-France model 3, the annual average change is above 3% in Brittany and above 2% on the north-west coast and in the Pyrenees. The change is clearly lower in the north-east region (blower than +1.5 % a year). For the IPSL model 4, the highest changes are in the south-west, extending to Brittany, and also include the Mediterranean region. This model too shows the north-east experiencing lower increases. For extreme events, the reference provided by the 2003 drought allows us to observe a temperature higher than the 90th quantile associated with levels of rainfall below the median. On this basis, according to the models we can estimate changes in the return periods for such an event as follows: CHANGE APPLIED TO EXTREME EVENTS Average change MF model IPSL model Annual average change + 5,6 % + 4,3 % Change over 13 years % + 73 % Past return period 20 years Future return period 10 years 12 years ANNUAL AVERAGE CHANGE APPLIED TO THE FREQUENCY OF BACKGROUND NOISE FOR EACH DÉPARTEMENT 3 MF Model 4 IPSL Model Indicators used: rainfall levels lower than the median and temperatures higher than the 3 rd quantile < 0% > +3% < 0% > +3% 1 The results of the IPSL (Institut Pierre Simon Laplace) and CNRM (Centre National de Recherches Météorologiques) simulation are referred to as IPSL and those of the WRF (Weather Research & Forecasting Model) and Météo France simulation are referred to as MF. 14

15 DROUGHT CONCLUSION CONCERNING THE PROJECTED IMPACT OF THE DROUGHT PERIL AND ITS DIFFERENT CONTRIBUTING FACTORS BETWEEN NOW AND 2040 Billions of Euros COST BASED ON LAST 25 YEARS DISTRIBUTION NATURAL CLIMATE WEALTH VARIABILITY CLIMATE CHANGE Billions of Euros The study concludes there will be a STRONG UPWARD TREND in the drought peril with regard to DAMAGE FOR WHICH INSURERS PAY COMPENSATION Billions of Euros Billions of Euros For example, THE RETURN PERIOD for a drought on the scale of the drought we experienced in 2003 would change FROM 20 YEARS TO 11 YEARS The share of this change accounted for by CLIMATE CHANGE is significant, accounting for 60%, I.E. 8 BILLION EUROS Our model finds there will be a CUMULATIVE COST of 21 billion Euros and that, between now and 2040, the ANNUAL AVERAGE COST OBSERVED TODAY WILL ALMOST TRIPLE 15

16 FLOOD FLOOD EXCLUDING COASTAL FLOODING THIS SECTION DEALS ONLY WITH THE CONSEQUENCES OF FLOODS CAUSED BY WATER COURSES OVERFLOWING, EXCLUDING COASTAL FLOODING. AS INDICATED AT THE BEGINNING OF THE STUDY, THE QUESTION OF COASTAL FLOODING IS CLOSELY ASSOCIATED WITH THE RISE IN SEA LEVELS, WHICH IS NOT INCLUDED IN THE CLIMATE INDICATORS USED IN OUR MODEL. THE ISSUE OF COASTAL FLOODING WILL THEREFORE BE DEALT WITH IN THE FOLLOWING CHAPTER. SOCIO-ECONOMIC FACTORS INVOLVING CHANGES IN HOUSING AND OTHER ASSETS AND ACTIVITIES AT RISK BETWEEN NOW AND 2040 According to the INSEE s latest research, the current projections of the number of homes and businesses over the next 26 years and their distribution across the country are as follows: ANNUAL AVERAGE CHANGE IN NUMBER OF ESTABLISHMENTS OF BUSINESSES AND PROFESSIONALS 1 ANNUAL AVERAGE CHANGE IN NUMBER OF HOMES OF PRIVATE INDIVIDUALS 2 < -0,3% > +0,1% < 0% > +2,0% 16

17 FLOOD The changes used for flood hazard areas were those observed over the period and were as follows 3: An increase in the proportion of housing in flood-prone areas means the increase in housing in hazard areas is higher than the overall increase in housing. Estimates of the growth in wealth and its distribution across the country over the next 26 years show: For housing: an increase in number of + 1.1% a year and in size of 0.4% a year with higher rates in a crescent extending from the départements of Brittany to the northern part of the French Alps, passing through a wide band of the south of France. For businesses: an increase in number of % a year and in value of 1.7% a year. Apart from the Île-de-France, the regions with higher growth are in the south and all along the Alps. Particularly high hyper-growth in hazard areas is observed on the Atlantic coast, in the Île-de-France and on the Mediterranean coast. CHANGE IN PROPORTION OF HOUSING IN HAZARD AREAS 3 < -0,3% > +0% 17

18 FLOOD CLIMATE FACTORS The two climate indicators of the IPSL and MF models that were used because of their past relevance were: Maximum daily rainfall by département. Cumulative daily rainfall by département. The climate indicator that correlates best with amounts paid out by insurers in the past is the 90% quantile of the cumulative amount of rainfall of the two models used. This led us to apply the following annual changes to the underlying frequency of floods over the next 26 years: MAJOR EVENTS For extreme events, the 99% quantile of the cumulative amount of rainfall was used for the two models. This allowed the following changes in return periods to be established for major events: CHANGE APPLIED TO EXTREME EVENTS MF model Average change IPSL model Annual average change + 0,7 % + 0,7 % Change over 13 years 9,3 % 9,5 % Past return period 50 years Future return period 46 years 46 years Past return period 75 years Future return period 68 years 68 years Past return period 100 years Future return period 91 years 91 years 90% QUANTILES 4 MF model 5 IPSL model < 0% > +0,5% < 0% > +0,5% 18

19 FLOOD CONCLUSION CONCERNING THE PROJECTED IMPACT OF THE FLOOD PERIL, EXCLUDING COASTAL FLOODING, AND ITS DIFFERENT CONTRIBUTING FACTORS BETWEEN NOW AND 2040 Billions of Euros COST BASED ON LAST 25 YEARS DISTRIBUTION CLIMATE NATURAL CHANGE CLIMATE WEALTH VARIABILITY Billions of Euros We note that, unlike in the case of drought, the CLIMATE CHANGE FACTOR ACCOUNTS FOR LITTLE OF this change (1 billion Euros i.e. 6 % of the difference between the two periods) Billions of Euros Billions of Euros The models show in particular that THE RETURN PERIODS FOR MAJOR FLOODS WILL REMAIN ALMOST UNCHANGED Apart from the wealth factor THE COST ASSOCIATED WITH THE DISTRIBUTION OF THAT WEALTH IS SIGNIFICANT: 5 billion Euros. This cost underlines the fact that, if there is no change, the geographical distribution of growth will increase the overall vulnerability to flooding. The study concludes that there will be an UPWARD TREND IN THE FLOOD PERIL (excluding coastal flooding) leading to insurers paying out a total of 34 billion Euros, compared to the 16 billion Euros paid between 1988 and 2013 I.E % 19

20 COASTAL FLOODING COASTAL FLOODING COASTAL FLOODING IS AN EMERGING PERIL FOR FRANCE. THE RECENT TRAGIC CONSEQUENCES OF THE COASTAL FLOODING ASSOCIATED WITH STORM XYNTHIA REMAIN FRESH IN OUR MEMORIES. IS THIS TYPE OF EVENT LIKELY TO BE REPEATED BETWEEN NOW AND 2040? METHODOLOGY USED The methodology used by the study to evaluate the amounts that insurers might be required to pay out in connection with coastal flooding was completely different from the methodology used for the other perils. Coastal flooding is the result of a conjunction of a large number of natural factors: winds, rain, swell, tide, sea level, coastal topography, etc. These factors lie well outside the models described above. Consequently, this study used the following to make projections about this peril: The past frequency of coastal floods 1 Contrary to what one might believe, our country has suffered events of this type in the past: 84 over the last 30 years to be precise. With the exception of Xynthia, most had no major consequences. The mapping of Approximate Envelopes of Potential Flooding (Enveloppes Approchées des Inondations Potentielles) associated with coastal flooding drawn up by the Environment Ministry 2 On the basis of an accurate analysis of the topography, the Ministry of Ecology, Sustainable Development and Energy has mapped maximum flood extents taking account of a sea level rise of 1 metre by This envelope, constructed to evaluate exposure of housing and other assets and activities to flood risk in the same way across the whole country, must not be confused with the mapping of floodprone areas under the Risk Prevention Plans (Plans de Prévention des Risques) and Flood Area Atlases (Atlas de Zones Inondables) or of Areas at Significant Risk of Coastal Flooding (Territoires à Risque Important d inondation littoraux). Finally, analysis of the housing and other assets and activities at risk in the hazard zones thus defined, as well as their changes over time. 20

21 COASTAL FLOODING HISTORY OF COASTAL FLOODING OVER THE LAST 30 YEARS BY GEOGRAPHICAL SECTOR 1 Number of coastal flooding events during the last 30 years by geo-sedimentary area (Source MRN events database) 12 CLIFFED COAST ROCKY COAST LOW SANDY COAST DUNES SALT MARSHES COASTAL FLOODING HAZARD HYPOTHESIS FOR ESTIMATION OF EXPOSURE OF HOUSING AND OTHER ASSETS AND ACTIVITIES 2 (Source Ministry of Ecology, Sustainable Development and Energy) This is an approximate and maximum envelope developed on the basis of prospective methodology. ÎLE DE RÉ LA ROCHELLE ÎLE D OLÉRON ROCHEFORT 21

22 COASTAL FLOODING On the basis of these data we transposed the impact of future events lineally in time, defining their occurrence on the basis of frequencies observed in the past. We estimated that these flood extents could be associated with events with a return period of between 500 and 1,000 years. Within this temporal linearity, we took a snapshot of the situation in 2040, the horizon of our study. Finally, we assumed that the future growth of housing and other assets and activities in hazard areas would be similar to that observed in the past. RESULTS The estimate obtained in this way shows an additional cost of between 3.2 and 4.2 billion Euros over 26 years. This compares to a cost of coastal flooding for insurers of 1 billion Euros (including 800 million Euros for Xynthia) over the last 25 years. It should be noted that this cost relates only to coastal flooding in the true sense of the term. Rising sea levels also raise the question of coastline management (progressive erosion making it necessary in certain parts of France to move housing or activities). That management also has a cost (which will certainly be very high) which falls under public prevention policies, but which is not included in this study. Consequently, the flood risk including coastal flooding is estimated as follows: (see conclusion on the next page). 22

23 COASTAL FLOODING CONCLUSION CONCERNING THE PROJECTED IMPACT OF THE FLOODING AND COASTAL FLOODING PERIL AND ITS DIFFERENT CONTRIBUTING FACTORS BETWEEN NOW AND 2040 Billions of Euros COST BASED ON LAST 25 YEARS WEALTH DISTRIBUTION CLIMATE CHANGE NATURAL CLIMATE VARIABILITY Billions of Euros FLOODING COASTAL FLOODING COASTAL FLOODING on its own accounts for 18% OF THE ADDITIONAL COST OF FLOODS 34 Billions of Euros + 4Billions of Euros As the emergence of this peril is associated with climate change, we can put the total COST OF CLIMATE CHANGE IN RESPECT OF THE FLOOD PERIL AT 5 BILLION EUROS I.E. 1/4 OF THE INCREASE INCLUDING COASTAL FLOODING IN FLOOD RISK INCREASES BY 4 BILLION EUROS THE PROJECTED ADDITIONAL COSTS of the flood peril. 23

24 STORM STORM OVER THE LAST 25 YEARS, STORMS HAVE BEEN THE MOST EXPENSIVE PERIL FOR INSURERS. WHEN THEY OCCUR, THEY AFFECT A LARGE PART OF THE COUNTRY AND LEAD TO A VERY LARGE NUMBER OF CLAIMS. SPECIFIC DIFFICULTIES POSED BY THE STORM PERIL The storm peril, from a strictly climate-related point of view, poses a dual problem: Firstly, in the analysis of past events, the most relevant climate indicator provided by the models used (highest wind speeds recorded for 10 minutes) has the weakest correlation of all the perils analysed with the amounts paid by insurers. No clear correlation was observed between this indicator and the frequency of storms experienced over the last 25 years. Secondly, the projection of exceptional events (which account for most of the damage for which insurers pay out) gives very different results depending on whether the IPSL or the MF model is used. Using the 99% quantile gives the following annual average changes depending on the model: 99% QUANTILES 1 IPSL model 2 MF model < 0% > +0,5% < 0% > +0,5% 24

25 STORM This is too uncertain a basis for a credible climate projection over the next 26 years. This accords with the observations of many climatologists concerning the uncertainties in measuring the impact of climate change on wind. Thus, Jean Jouzel states in his August 2014 study Climat de la France au XXI e siècle (France s Climate in the 21 st Century): // The conclusions presented in this section are based on a study of a maximum winter wind index excluding gusts. As they relate only to the two Aladin-Climat and WRF models, they are not accompanied by considerations of multi-model uncertainty. Over the period , the strongest winter winds occurred near the Channel coasts and Brittany, along part of the Atlantic coast, and near the Mediterranean coasts. Initial estimates show that, according to the Aladin-Climat model, the intensity of the strongest winds could reduce in the late 21st century over the whole of the country, whatever the RCP scenario. While the WRF model also seems to show a reduction in strong winter winds in the south of the country, it simulates an overall increase in strong winds in the north of the country. This initial analysis therefore gives an indication about the strongest average winds, with different results for the north of the country depending on the model. However, it does not permit conclusions to be drawn concerning the frequency and intensity of winter storms, which will require a specific study. // FACTORS USED IN THE STUDY ON THE BASIS OF THESE UNCERTAINTIES Given these inconsistencies and uncertainties, the study took the following approach to projecting the storm peril: Neutralising the climate change factor. In simple terms, the study starts from the principle that, given the current state of knowledge, it should be assumed that climate change will not lead to either a decrease or an increase in storms in France between now and Taking account of the exceptional nature of the Lothar and Martin storms in Given the particularly long return periods for these 2 events (70 years for each event) and the proportion of past costs they accounted for (13.4 billion Euros), they were probabilised in the future projection, which therefore includes a negative climate hazard factor of 1.3 billion Euros. The hazard areas used to project the housing and other assets and activities at risk ( wealth and distribution factors) were kept the same as for the last 25 years. 25

26 STORM CONCLUSION CONCERNING THE PROJECTED IMPACT OF THE STORM PERIL, AND ITS DIFFERENT CONTRIBUTING FACTORS BETWEEN NOW AND 2040 NATURAL CLIMATE VARIABILITY COST BASED ON LAST 25 YEARS DISTRIBUTION WEALTH 33 Billions of Euros Billions of Euros The exceptional nature of the storms LOTHAR ET MARTIN in 1999 means it is necessary to include A NEGATIVE CLIMATE HAZARD FACTOR of 1 billion Euros over the next 25 years. On this basis, the study concludes there will be WEAK GROWTH IN DAMAGE CAUSED BY STORMS COMPARED TO THE PERIOD (a cumulative amount of 9 billion Euros, i.e % over 26 years). However, this projection was produced by a cautious methodology in view of the HIGH LEVEL OF SCIENTIFIC UNCERTAINTY SURROUNDING THE EFFECTS OF CLIMATE CHANGE ON THIS PERIL Wind is the peril for which climate models give the most conflicting results concerning future changes in our country. In view of this, the study took the approach of NEUTRALISING THE CLIMATE CHANGE FACTOR IN ITS PROJECTION FOR STORMS BETWEEN NOW AND

27 ENSEMBLE DES PÉRILS OVERALL CONCLUSION Bringing together all the perils studied, the projections provided by the study concerning the quantified consequences of natural hazards for insurers between now and 2040 give the following results: 27

28 ENSEMBLE DES PÉRILS CONCLUSION CONCERNING THE PROJECTED COST BASED ON LAST 25 YEARS WEALTH Billions of Euros THE SECOND MOST SIGNIFICANT FACTOR IS DIRECTLY LINKED TO CLIMATE CHANGE and accounts for 30% of the projected increase. It is estimated that climate change will cost 13 billion Euros between now and THE TOTAL INCREASE IN WEALTH OF OUR COUNTRY ESTIMATED AT 19 BILLION EUROS (density and average value of housing, businesses and assets of regional and local authorities) ACCOUNTS FOR 43 % OF THIS INCREASE and is the leading contributory factor. IT IS EXPECTED TO MAINFEST ITSELF MAINLY THROUGH THE DROUGHT PERIL with the amount of additional damage estimated at 8 billion Euros. COASTAL FLOODING IS SECOND IN THE LIST OF PERILS THROUGH WHICH CLIMATE CHANGE WILL MANIFEST ITSELF with a risk of significant increases in the amounts paid out by insurers in the medium term. The projections produced estimate that the damage caused by this peril in the next 25 years will be 4 billion Euros, compared to 1 billion Euros in the last 25 years (mainly Xynthia). This peril presents the strongest growth dynamic and, all things being equal, will become very significant beyond the 26 year period considered in this study. 28

29 ENSEMBLE DES PÉRILS IMPACT OF ALL PERILS BETWEEN NOW AND 2040 DISTRIBUTION NATURAL CLIMATE VARIABILITY CLIMATE CHANGE Billions of Euros Billions of Euros Billions of Euros The study projects that THE CUMULATIVE AMOUNT OF DAMAGE CAUSED BY NATURAL HAZARDS WILL BE 92 BILLION EUROS between now and UNFAVOURABLE SPATIAL PLANNING is the third most significant contributory factor to the projected future increase, accounting for 18% of the projected increase. It is estimated that it will cost 8 billion Euros between now and 2040, 60% of which relates to flood peril. There is currently no scientific consensus concerning THE IMPACT OF CLIMATE CHANGE ON WIND. The study therefore neutralised this impact. While it may also lead to additional costs in the next few decades, it is not possible to put a figure on those costs. i.e. an INCREASE OF 44 BILLION EUROS in constant Euros. CLIMATE CHANGE IS THE SECOND MOST IMPORTANT CONTRIBUTORY FACTOR IN THIS INCREASE, ACCOUNTING FOR 13 BILLION EUROS BETWEEN NOW AND

30 30

31 31

32 32

33 33

34 34

35

36 Décembre 2016 / Conception graphique : Vanessa Vansteelandt / Crédits illustrations : Shutterstock 26, boulevard Haussmann Paris - Cedex 09 CONTACT PRESSE FFA Benoît Danton b.danton@ffa-assurance.fr