Project Topics
Get Complete Project Material File(s) Now! »
IN-DEPTH REVIEW OF EARTHQUAKE INSURANCE SOLUTIONS
The International Disaster Database (CRED), shows that the number of damaging earthquakes (i.e. more than 100 people affected) and the consecutive economic loss are increasing since 1960. Analysing the consequences of natural disasters occurred since 1970s at the worldwide scale, Ghesquiere and Mahul (2010) have identified three different phases in post-disaster management: the Relief (1-3 months), the Recovery (3-9 months) and the Reconstruction (>9 months). They have also quantified the financial amount required, as illustrated in Figure 2.1 which highlights the time scale of the post-disaster requirements.
Ghesquiere and Mahul (2010) listed 11 financial sources, sorted in three categories: donations, public debt and insurance. Each is characterized by a disbursement period, a quantity of funds available, and the issuance period (either before or after the disaster occurrence). Figure 2.2 shows that insurance solutions (parametric insurance, alternative risk transfer and traditional insurance), donor support and budget contingencies can cover the second and the third months post disaster period (Ghesquiere and Mahul 2010), corresponding to the late Relief phase and the beginning of the Recovery phase (Fig. 2.1).
The financial amount that donor support and budget contingencies is estimated by Ghesquiere and Mahul (2010) at an Uncertain and Small quantity, respectively. Furthermore, these two kinds of financial sources are determined after the disaster and, consequently the amount available is known only after the disaster. At the opposite, insurance can provide a Large funding capacity, depending on the insurance market size. About claim amount and payment pattern, they are agreed at the insurance policy issuance and therefore, known before the disaster. Between the 4th and the 6th month after a disaster, insurance and external credit provide both a Large financial assistance (Fig. 2.2). However, funds from external credits are allocated to affected people by public grants, the amount dedicated to each affected people depends on public policies set after the event and does not necessarily meet the affected people’s needs. At the opposite, since insurance policy can be underwritten by households, this financial assistance can be directly allocated to affected people accordingly to each insurance policy.
According to Ghesquiere and Mahul (2010), insurance is a key financial instrument for financing post-disaster needs during the late Relief and the Recovery phases (2nd – 6th months following the disaster) as illustrated in Figures 2.1 and 2.2.
Current earthquake insurance solutions are significantly different from one country to another depending on several variables like the country’s wealth or the experience of natural disasters. In this study, four different areas are considered, representing typical combinations of country’s wealth and hazard level: France (developed country with a low seismic hazard), India (developing country with a low seismic hazard outside Himalaya), Indonesia (developing country with a high seismic hazard) and California (developed country with a high seismic hazard). Recent earthquakes affecting these countries depict well the wide range of insured loss as reported by the Global Facility for Disaster Reduction and Recovery (Dani 2012): [0%; 10%] for the 2006 Yogyakarta (Indonesia) and the 2001 Gujarat (India) earthquakes and [30%; 40%] for the 1994 Northridge (USA) earthquake. In France, no severe earthquake has occurred since 1982, the creation of the CAT-NAT insurance plan covering earthquake risk, among other perils. Nevertheless, in France, earthquake and windstorm insurance covers are both compulsory in housing insurance policy. So, the share of insured loss can be estimated from the 1999 Lothar and Martin windstorms at [80%; 90%]. While the very low rate of insured loss in India and Indonesia is consistent with a Medium Human Development Index, the share at [30%; 40%] for California compared to [80%; 90%] for France is more surprising. Indeed, California is more exposed to earthquake risk than France, but seems to have a less protective earthquake insurance system.
In California earthquake insurance has existed since the early 1900s. A lot of large events have occurred over the 20th century, including the 1994 Northridge, the 1989 Loma Prieta, the 1933 Long Beach, the 1971 San Fernando and the 1925 Santa Barbara earthquakes. These damaging events (Alquist et al. 2009) have forced the insurance scheme to react. In the same time, risk awareness arose, thanks to an active scientific field. California earthquakes are also constantly monitored by the USGS and their consequences are calculated (e.g. PAGER alerts). From the economic side, the California Department of Insurance (CDI) provides an open access to data about earthquake insurance market. For all of these reasons, the California earthquake insurance constitutes an excellent case study.
This chapter reviews the current earthquake insurance models and highlights their strengths and weaknesses. In the first part, the California insurance scheme, for which the most data is available, is analysed. Next, it is compared to the earthquake insurance solutions in India, Indonesia and France from an economic perspective.
Earthquake insurance market in California
Historical background
Earthquake insurance in California started in the aftermath of the 1906 San Francisco earthquake but was initially unpopular (Buffinton 1961; Meltsner 1978; Goltz 1985; Muir-Wood 2016a). Indeed, most of loss related to this event was due to consecutive fires and, therefore, damage was already covered by the fire insurance cover (Goltz 1985; Yeats 2004; Gioncu and Mazzolani 2011). At the opposite, loss caused by the 1925 Santa Barbara earthquake was due to ground shaking, and therefore not insured (Goltz 1985; Alquist et al. 2009). This event boosted drastically the demand for earthquake insurance, as illustrated in Figure 2.3 by the total amount of premium collected by insurance companies.
After the 1925 Santa Barbara earthquake, Great Depression started, stopping the appeal. The occurrence of two other damaging earthquakes in 1933 (Long Beach) and 1940 (Imperial Valley) triggered the attractiveness of earthquake insurance cover, until 1957 (Muir-Wood 2016a; Meltsner 1978). Indeed, is observed (Fig. 2.3) a decreasing trend in the premium amount collected between 1957 and 1970. This evolution is surprising considering solid economic US growth during that period (2.5% average annual growth rate of the GDP per capita between 1950 and 1973, against 2% between 1870 and 2007, according to Jones 2016) and the significant earthquakes occurred the six previous years: 1) the 1952 Kern County earthquake (M7.3): the epicentre was located at 40km of Bakersfield and 90km of Santa Barbara, and it caused $2.8bn (USD 2005) (Alquist et al. 2009) damage (i.e. similar to the 1925 Santa Barbara event); 2) the 1955 Terminal island earthquake (M3.5): occurred in Los Angeles and caused $3m (USD 1955) (Vranes and Pielke 2009) despite the very low magnitude; 3) the 1957 San Francisco earthquake (M5.7): the biggest in this area since the 1906 event, despite a very limited damage estimated at $27m in USD 2005 (Alquist et al. 2009). According to Buffinton (1961) and Meltsner (1978), insurance companies did not record significant losses from these events (the global insured loss ratio is below 20%). Furthermore, the earthquake engineers’ community and public officials congratulated themselves about the efficiency of the seismic retrofitting codes settled during the 1940’s (Geschwind 2001). Except the damage, they even communicated that there was « no cause to fear an earthquake like 1906 » (Geschwind 2001), despite objections raised by earthquake researchers led by Charles Richter. As a likely consequence, people ignored the risk and cancelled their earthquake insurance policy.
The 1971 San Fernando earthquake caused a $6.6bn (USD 2005) damage but less than 10% were covered by insurance companies (Meltsner 1978; Alquist et al. 2009). This event stands out from the previous ones because the commercial and industrial sectors were heavily affected (almost equal to the residential losses in Los Angeles City), and a large share (62%) of buildings affected collapsed or was heavily damaged (Meltsner 1978; Alquist et al. 2009). As a consequence, Goltz (1985) mentions that professionals bought earthquake insurance products, boosting the sector at an unprecedented level (Fig. 2.3). The demand for earthquake insurance was even more triggered by the devastating 1983 Coalinga earthquake ($120m. in USD 2005) and the consecutive Assembly Bill AB2865 (McAlister 1984; Fig. 2.4).
This legislative act mandated insurance companies to offer an optional earthquake coverage complementary to the dwelling insurance. Consequently, after the 1994 Northridge earthquake, the consecutive loss for insurance companies reached $11.4bn (USD 1995), i.e. three times Source: after Goltz 1985.
higher than the $3.4bn (USD 1995) earthquake insurance market premiums collected since 1970 (Snyder 1995). Even if no company was declared bankrupt, claims overpassed the maximum loss assessed by the contemporary actuarial models (RMS 2004; Insurance Information Institute 2016). Forced at that time by the Assembly Bill AB2865 (McAlister 1984) to propose an earthquake coverage in residential insurance policies, most of the insurance companies (≈ 90%) decided to restrict or even to stop selling new residential insurance cover in California (California Earthquake Authority 2016a).
The Fair Access to Insurance Requirements (FAIR) is a state-managed insurance syndicate gathering all California insurance companies. Since 1968, it provides in last resort a basic insurance cover to households that insurance companies do not want to cover. Thus, to limit the threat of a shortage in property insurance products, the FAIR plan launched in 1994 a basic earthquake insurance cover (Mulligan 1994). Constrained by the sharp decrease of new insurance policies offer, customers rushed to subscribe the FAIR plan product (Sanchez 1996). This unforeseen popularity generated fears among the authorities about the capacity to face a major earthquake loss. It resulted in a strict limitation of the FAIR plan house insurance subscription on June 1st, 1996 to very poor zones and highly exposed to brush fire risk (Sanchez 1996; Reich 1996a; Reich 1996b). The FAIR plan reopened 5 months later while the California Earthquake Authority (CEA) was created as a response to the earthquake insurance crisis (Reich 1996a; Reich 1996b).
Initiated by law in 1995, the CEA aims at providing an earthquake insurance for households (Consumers Union 1997; Knowles 1997), called the Mini-policy because of the low guarantees provided. After the commitment of more than 75% insurance companies to sell it (later referred as CEA insurance company members) and the purchase of the reinsurance cover imposed by such risk, the CEA was officially launched on December 2nd, 1996 (Consumers Union 1997; Knowles 1997). The new insurance conditions of the Mini-policy were less attractive than the FAIR plan cover because more expensive and more restrictive as summarized in Table 2.1.
Consequently, many people were no longer insured against earthquake risk and, despite a significant premium amount increase (Tab. 2.1; Fig. 2.4, Significant EQ premiums increase) it resulted in a drop of the total written premium amount (Fig. 2.4). From 31% in 1996, the share of people covered against earthquake falls to 19.5% in 1997 as illustrated in Figure 2.5 (California Department of Insurance database).
After 1997, the number of earthquake insurance policies decreased slightly until 2003 (the year of the San Simeon earthquake) and then has been increasing until now (Fig. 2.5a). However, this increase is slower than for the number of housing insurance policies, resulting in a slight decrease of the ratio between earthquake and housing insurance policies (Fig. 2.5b). Nowadays, the CEA has some competitors representing 25% of the policies and 35% of the premium amount related to the earthquake dwelling insurance market (California Department of Insurance database). The difference between these two values reflects that the CEA protects more low-value houses than its competitors (assuming that all insurance companies use similar pricing models and policies).
Focus on the California Earthquake Authority
Until now, the CEA has faced only small losses despite the occurrence of several moderate earthquakes (Fig. 2.6). Indeed, only the 2003 San Simeon and the 2014 Napa earthquakes caused claims above $2m. (USD 2015). The total claim amount paid during the period 1997-2015 is equal to $19m. (USD 2015), which corresponds to only 2‰ of the total premium amount for the same period. Although the period is too short to draw any conclusions on the premium amount adequacy, the issue of the collected premium allocation is raised: does it increase the CEA claim-paying capacity to face more and more devastating earthquakes?
The claim-paying capacity is the maximum amount that an insurance company can pay. This amount is equal to the sum of the company’s reserves and the cash-flow that it can benefit from Figure 2.7: CEA’s claim-paying capacity according to the source of funds: the CEA’s capital, the CEA reinsurance cover, and the Industry Assessment Layer (IAL) corresponding to the funds provided by the CEA’s insurance company members. The lines correspond to the loss incurred by the CEA if the 1906 San Francisco, the 1989 Loma Prieta or the 1994 Northridge earthquakes occur today. The ‘1994 Northridge x2’ corresponds to a hypothetical earthquake causing a direct economic loss twice higher than the 1994 Northridge earthquake. Source: after CEA Financial Statements.
According to the California Earthquake Authority (2014), the company has a claim-paying capacity large enough to face the loss of some largest historical earthquakes (1989 Loma Prieta, 1994 Northridge and 1906 San Francisco) if they occur again. The largest event sustainable by the CEA is a two Northridge-size event (Roth 1997), estimated at a 400y return period (California Earthquake Authority 2018a). Figure 2.7 shows also that the CEA’s claim-paying capacity is made of its own reserves (dark grey), the reinsurance capacity bought on financial markets (grey) and the Industry Assessment Layer (IAL) which is an additional reinsurance coverage provided by the CEA insurance company members (light grey). While the capital increases as a consequence of the premium collected and the losses recorded (Fig. 2.6), the IAL is decreasing. It results in a constant claim-paying capacity since 1997 when calculated in $2015. Consequently, Figure 2.7 highlights that part of the premiums collected since 1997 and not used to pay claims (Fig. 2.6) is used to decrease the contribution of the participating insurers in the earthquake coverage. As the IAL is a reinsurance cover free of charge for the CEA (Marshall 2018), this decrease has no impact on the premium amount. However, the excess of premium collected allowed also the CEA to apply several premium discounts since 1997: -11% in 1997, -23% in 2006, -12.5% in 2012 and -10% expected in 2016 (California Earthquake Authority 2015a).
The earthquake premium calculated by the CEA considers several building characteristics, as a proxy of the earthquake vulnerability. For instance, the online CEA premium calculator indicates that the premium amount for a one-story modern house is between 3 and 4 times less expensive than an old one-story house built in earthquake vulnerable materials (for instance unreinforced masonry). In addition to the premium scale, the CEA and the California Governor’s Office of Emergency Services launched in September 2013 the prevention plan called Earthquake Brace+Bolt (California Department of Insurance 2015). This initiative aims at promoting earthquake retrofit for houses and mobile homes of CEA’s clients by financing the work up to $3,000. Table 2.2 draws a global picture of the initiative through different figures.
First, the need for retrofitting in California is huge: among the 13,987,625 housing units in 2015 (US Census), 1,200,000 are particularly at risk and would benefit from this initiative (Lin II 2015; Xia and Lin II 2016). The Earthquake Brace+Bolt initiative has been launched cautiously with a budget of $1.8m., with 600 houses retrofitted by the end of year 2015 as objective. Between 2015 and 2017, the funds increased from $1.8m to $6m driving the Earthquake Brace+Bolt initiative growth. In 2017, 2,000 houses are targeted to be retrofitted over 140 postal codes mostly in big urban areas (Los Angeles, San Francisco, Eureka and Riverside). Despite the two years-old initiative Earthquake Brace + Bolt is expanding quickly, it is still in its infancy. Only 3.5‰ of the highly vulnerable houses have been retrofitted after 3 years of existence. However, the public authorities and the CEA rely on the Earthquake Brace+Bolt initiative to increase the public awareness of the risk and to urge households to retrofit their house by themselves (Lin II 2015).
In this section, the earthquake insurance models in force in California, France, India and Indonesia are compared based on several economic metrics: the premium amount, the loss allocation between the insured, the insurance companies and public funds and the solvency of insurance companies.
The insurance premium
Comparing the premium rate (equal to the premium amount divided by the insured value of the house) aims at highlighting the affordability of earthquake insurance considering both the insurance development and the seismic hazard level. Figure 2.8 shows the spatialized premium rate for each studied area.
The building characteristics considered for calculating the premium amount (i.e. modern one storey house built in unreinforced masonry) have been chosen because it is assumed to be the most similar from one country to another. Consequently, for a given area, the premium rate illustrated in Figure 2.8 is not representative of the market when the selected conditions are not characterizing the buildings taxonomy.
Figure 2.8 shows first that for each country the premium rate depends on the location. The premium is the lowest in India (shiny pink colour) excluding the Himalayan area, the Gujarat state and the west of Maharashtra state, exposed to induced seismicity after the Koyna dam construction, according to Phadnis (2016). In France, the premium rate appears to be low (below 0.25‰) especially because the premium covers flood and subsidence risks in addition to earthquake risk (Caisse Centrale de Réassurance 2011). In France, this premium rate is established by law (as part of the CAT-NAT plan) and consequently, does not necessarily reflects the risk level. Furthermore, the French State offers under the CAT-NAT plan an unlimited reinsurance cover. It means that in case of extreme loss, the French State will pay part of insurance claims. Hence, the State budget dedicated to the CAT-NAT plan can be assimilated to an additional premium. Nevertheless, the premium rate is regularly increased (Caisse Centrale de Réassurance 2019a): +64% in 1985 and +33% in 1999. Furthermore, the ongoing revision of the CAT-NAT plan could also include a new premium increase (Bonnefoy 2019). California is the country with the highest premium rate, especially along the San Andreas Fault and in the vicinity of big cities like Los Angeles or San Francisco. It is also the country where the location has the highest impact on the premium amount. Indonesia shows spatial variation in premium rate that can be explained by the different levels of seismic hazard.
To go further in the analysis of the premium amount, Figure 2.9 plots the premium rate against the hazard level extracted from the GSHAP hazard map and corresponding to the return period 475y (Giardini et al. 1999).
Table of contents :
CHAPTER 1: General introduction
1.1. Problématique
1.2. Plan de la thèse
CHAPTER 2: In-depth review of earthquake insurance solutions
2.1. Introduction
2.2. Earthquake insurance market in California
2.2.1. Historical background
2.2.2. Focus on the California Earthquake Authority
2.3. Main differences with earthquake insurance models in France, India and Indonesia
2.3.1. The insurance premium
2.3.2. The loss allocation
2.3.3. Insurance companies’ solvency
2.4. Challenges and opportunities in earthquake insurance market development
2.4.1. A long-term profitable market with extreme loss
2.4.2. A need for being full-covered
2.4.3. A large untapped market despite new insurance solutions
2.5. Conclusions
CHAPTER 3: Limits of earthquake insurance solutions
3.1. Introduction
3.2. California earthquake insurance unpopularity: the issue is the price, not the risk perception
3.2.1. Introduction
3.2.2. Data collection and processing
3.2.3. Model development for the period 1997-2016
3.2.4. Evolution of the homeowners’ risk perception since 1926
3.2.5. Understanding the current low take-up rate
3.2.6. Conclusions
3.3. Assessing the performance of the French « CAT-NAT » insurance plan
3.3.1. Introduction
3.3.2. Review of past CAT-NAT declarations following an earthquake
3.3.3. The CAT-NAT procedure following the 2003 heatwave
3.3.4. An empirical model for the declaration of municipalities in CAT-NAT situation
3.3.5. Modelling the performance of the French CAT-NAT plan in case of extreme earthquakes
3.3.6. Conclusions
3.4. A maturity scale for earthquake insurance development based on the California experience
3.4.1. Introduction
3.4.2. Level “Emerging”: the birth of the earthquake insurance (California 1906 – 1925)
3.4.3. Level “Standard”: An empirical insurance model (California 1926 – 1994)
3.4.4. Level “Advanced”: an insurance model designed to face extreme events (California 1995 – Today)
3.4.5. Level “Sustainable”: current initiatives for a sustainable insurance model (unreached level)
3.4.6. The maturity scale for earthquake insurance
3.4.7. Conclusions
3.5. Summary
CHAPTER 4: Improving the risk modelling
4.1. Introduction
4.2. Comparing probabilistic seismic hazard maps with ShakeMap footprints for Indonesia
4.2.1. Introduction
4.2.2. Dataset
4.2.3. The testing method
4.2.4. Testing PSHA maps for Indonesia
4.2.5. Conclusions
4.3. Assessing the performance of existing repair-cost relationships for buildings
4.3.1. Introduction
4.3.2. The Earthquake Damage Database
4.3.3. Using the Earthquake Damage Database to test existing damage-cost relationships
4.3.4. Example: the Nepal Mw7.8 earthquake
4.3.5. Testing some existing damage-cost relationships
4.3.6. Conclusions
4.4. Summary
CHAPTER 5: A new insurance model
5.1. Abstract
5.2. Introduction
5.3. Example of the CEA insurance model
5.4. A life insurance mechanism to increase affordability
5.5. Case studies on cities of San Francisco and Los Angeles
5.6. Leveraging on building retrofitting works for a risk reduction
5.7. Involving homebuilder companies and public authorities in the insurance scheme
5.8. Conclusions
CHAPTER 6: General conclusions and perspectives
