Climate change derivatives : a potential solution to the insurance coverage gap for catastrophic events

MASTERS IN INTERNATIONAL FINANCE

THESIS

CLIMATE CHANGE DERIVATIVES: A POTENTIAL

SOLUTION TO THE INSURANCE COVERAGE GAP FOR

CATASTROPHIC EVENTS

Prepared By

Anna Brink

Date

August 2016

Student Number

11081724

ABSTRACT

The increased prevalence of natural disasters in the past three decades has fostered a

growing insurance coverage gap for economic damages. Damages not covered by the

insurance industry are primarily reliant on ex-post government aid. The financial markets

have attempted to solve this problem; however, the market has remained relatively small due

to lack of liquidity and accessibility and in some instances too narrowly focused or not

well-defined derivatives contracts. It has been suggested that governments and the insurance

companies should work together to reduce the insurance coverage through the use of

climate change derivatives. These derivatives would mimic Mortgage Backed Securities by

using premiums obtained from customers and creating a diversified risk pool. Not only would

this reduce the coverage gap but also it would promote investment in risk mitigation

measures to reduce post disaster recovery costs, and reduce moral hazard. Ultimately, the

goal of this paper is to determine if climate change derivatives can be a financially viable

solution to the catastrophic insurance gap, as well as be a mechanism to determine the cost

of climate change.

TABLE OF CONTENTS

INTRODUCTION  ...  4  

LITERATURE REVIEW  ...  8  

THEORETICAL FRAMEWORK  ...  15  

RESEARCH METHODS & TECHNIQUES  ...  18  

1) APPROPRIATE INDEX  ...  18

2) RISK MITIGATION TECHNIQUES  ...  21

RESULTS  ...  22  

1) APPROPRIATE INDEX  ...  22

CATASTROPHIC DISASTER DATA ... 22

UNITED STATES HOUSING MARKET ... 26

DETAILED ANALYSIS: FLORIDA ... 28

2) RISK MITIGATION MEASURES  ...  30

DISCUSSION & CONCLUSION  ...  31  

BIBLIOGRAPHY  ...  36  

APPENDICES  ...  39  

APPENDIX 1 ... 40

APPENDIX 2 ... 42

APPENDIX 3 ... 43

APPENDIX 4 ... 45

APPENDIX 5 ... 47

APPENDIX 6 ... 49

APPENDIX 7 ... 51

APPENDIX 8 ... 53

APPENDIX 9 ... 55

APPENDIX 10 ... 57

APPENDIX 11 ... 59

APPENDIX 12 ... 61

APPENDIX 13 ... 63

APPENDIX 14 ... 65

APPENDIX 15 ... 67

APPENDIX 16 ... 69

APPENDIX 17 ... 71

APPENDIX 18 ... 73

APPENDIX 19 ... 75

APPENDIX 20 ... 77

APPENDIX 21 ... 79

INTRODUCTION

A disaster is defined by the United Nations Office for Disaster Risk Reduction (UNISDR) as

“a serious disruption of the functioning of a community or a society involving widespread

human, material, economic or environmental losses and impacts, which exceeds the ability

of the affected community or society to cope using its own resources.” Natural disasters, on a

global scale, have significantly increased in frequency over the past three decades. With the

combination of higher occurrences of catastrophic events and increased infrastructure and

population density in prone regions it has exacerbated the post-disaster recovery costs (See

Figure One). UNISDR estimates between $200 to $250 billion dollars are spent annually on

disaster costs, and this trend is expected to increase (United Nations Office For Risk

Reduction, 2015).

Figure One: Global Natural Disasters By Frequency and Total Dollars of Damage* Scaled to 2014

(EM-DAT 2016)

In most cases only a small percentage of the annual damages are actually covered by

insurance companies. The insurance industry covered less than half of the damages incurred

for 198 natural disasters events in 2015 (Munich Re, 2015), and even in light of this they are

struggling to keep up with the payouts. Insurance companies thus rely on reinsurance

companies to help reduce the overall risk; however, even the reinsurance companies are

 $-­‐        

 $50,000,000    

 $100,000,000    

 $150,000,000    

 $200,000,000    

 $250,000,000    

 $300,000,000    

 $350,000,000    

 $400,000,000    

0  

50  

100  

150  

200  

250  

300  

350  

400  

450  

1960  

1962  

1964  

1966  

1968  

1970  

1972  

1974  

1976  

1978  

1980  

1982  

1984  

1986  

1988  

1990  

1992  

1994  

1996  

1998  

2000  

2002  

2004  

2006  

2008  

2010  

2012  

2014  

reluctant to invest in certain areas that are very prone to catastrophic events. Therefore, this

leads to insurance coverage not always being available in all areas. According to the Global

Reinsurance Forum, a panel that includes the eleven leading global reinsurers, reinsurers

can only bear between 40-65% of large catastrophic insured losses (GRF, 2014). The large

portion of post disaster recovery not covered by insurance is funded by governments and aid

organizations, resulting in a large insurance coverage gap (See Figure Two). This ultimately

creates moral hazard issues where governments fill the void of insurance but premiums are

not paid commensurate with risk.

Figure Two: Global Natural Disasters Overall Losses Versus Insured Losses (Munich RE 2015)

For instance, Hurricane Sandy, which plagued the east coast region of the United States

during 2012, led to a significant amount of economic damages due to its large concentration

of assets in a densely populated area. In the end, over 650,000 houses were damaged or

destroyed, a record surge height of 3.5 metres above mean sea level was recorded, 20,000

flights were cancelled, and the New York Stock Exchange was closed for two days. The total

economic damage from Hurricane Sandy was estimated to be around $70 billion, of which

$30 billion was covered by the reinsurance industry. New York represented a large portion of

the Hurricane Sandy’s economic damages with around $19 billion dollars. According to

Swiss Re’s Economics of Climate Adaptation study if the hurricane occurred in the year

2050, the damages would increase to $90 billion for New York if no action were taken.

Overall, the average annual loss from storm surges and windstorms is predicted to grow from

around $1.7 billion to $4.4 billion in 2050 in today’s dollars (GRF, 2014). Therefore, it is

evident that the economic damages from natural disasters will only continue to increase at an

alarming rate.

There is a need for risk management practices to be implemented to reduce the costs of post

disaster recovery (i.e. stronger building codes, levies, etc.). Both governments and insurance

companies are looking into risk management techniques to reduce ex-post expenditures.

Nevertheless, reinsurance companies see the biggest impediment in implementing risk

management practices is the lack of participation between governments and the private

sector. They believe that a public-private partnership could be a key driver in reducing the

uninsured gap in catastrophic coverage (GRF, 2014).

Moral hazard is also a significant contributor to the insurance coverage gap. The main

participants in natural disaster management (mitigation, preparedness, response, and

recovery) are government, media, and nongovernmental organizations. Notably absent are

the consumers. According to Baker (2009), knowing natural hazard risks in an area does not

prevent from settling in those areas, nor does it give them a desire to relocate, as people

become attached to places of familiarity. Furthermore, people who have never experienced

natural disasters will often underestimate the likelihood of their occurrence, and even if they

do happen they know that the government will often bail them out. This is especially evident

with mega catastrophes, where it falls on the government to be the de facto “insurer” of last

resort. For instance, federal aid for Hurricane Katrina amounted to over $100 billion,

accounting for a significant portion of the total disaster relief. A large amount of these costs

could have been covered with insurance but were not. Providing federal aid was required in

this case but it will set precedent for future mega catastrophes unless mitigation measures

are put into place to reduce future disaster expenditures, and people are incentivized to

purchase catastrophic insurance (Litan, 2006).

Catastrophic events will likely be impacted significantly by climate change. According to the

International Panel on Climate Change, an increase in greenhouses gases in the

atmosphere will result in higher temperatures in most regions of the globe. Not only will

average temperatures increase, but also the variance of temperatures will increase, with

extreme temperatures becoming more prevalent. Possible outcomes of more extreme

temperatures are higher likelihoods of droughts and increased intensities of storms (i.e.

higher wind speeds for tropical cyclones, increased precipitation for Asian monsoon’s, and

more intense mid-latitude storms). Climate change will also have an impact on sea levels, as

they are likely to rise with increased temperatures in the poles. Increased sea levels could

lead to deadlier storms. Especially with coastal flooding, as it is predicted to increase in

severity when storms materialize (NASA, 2005). It should be noted that not all natural

disasters are exacerbated by climate change. Other natural disasters such as earthquakes

and tsunami’s are not often linked to climate change due to their occurrences being

associated with seismic activity. Nevertheless, according to Liggins et al (2010) climatic

factors could have an impact on the levels of geological and geomorphological activity in the

future. This is due to the fact that increased ocean masses from the melting of ice caps may

alter the stability of submarine slopes. In addition, this may elicit volcanic and seismic activity

in coastal and island areas. This will result in a formation of volcanic landslides, submarine

landslides, and tsunami’s.

Reducing greenhouse gases will be the most vital solution in the long term to reduce the

effects of climate change; however, in the short to medium term governments urgently need

to strengthen their regions disaster resiliency. It is expected that half the global population

will migrate to within 100km of the coast in the next 25 years; with sea levels rising this will

only worsen the current problem (GRF, 2014). Coastal assets tend to have higher property

values, which could lead to higher economic damages after a catastrophic event. The main

ways in which governments can improve their disaster resiliency is to utilize adaptation

strategies such as improvements to infrastructure and technological developments. This is

especially prevalent for Hurricane Katrina. It is estimated that over $100 billion dollars in

damages could have been avoided if the levee built between the northern border of New

Orleans and Lake Ponchatrain was constructed to endure a strong Category 3 hurricane

(Litan, 2006). According to McKinsey, up to 65% of post disaster damages could be averted

from implementing risk mitigation methods. Unless these measures are adapted the

accessibility and affordability of catastrophic insurance protection will significantly diminish

(Economics Of Climate Adaptation Working Group, 2009). Therefore, there is a strong need

for the public and private sector to work together to help combat the issues of the insurance

coverage gap, as well as promote investment into risk mitigation techniques to help reduce

future costs.

In this paper, it will explore a potential strategy for addressing the insurance coverage gap

through the use of climate change derivatives between the public and private sector. The

main goal will be to determine whether this strategy could be a financially viable solution to

the current problem. In addition, this paper will address whether the costs of natural disasters

have the potential to be accurately assessed by the market through price discovery, to

ultimately gauge the costs of catastrophic events. In the first section, the catastrophic

insurance market will be evaluated to determine the risks faced by insurance and

reinsurance companies, and the impediments to the current catastrophic derivatives. It will

conclude with how a partnership between the pubic and private sector could drastically

reduce the problems of moral hazard associated with natural disasters. The following section

will describe the theoretical framework of the climate change derivatives. This includes how

the derivative will be structured and how these techniques could be applied to the global

market to obtain price discovery. Afterward, the research methods & techniques are

described, with the main goal of developing an appropriate index for the derivative in the

United States market. Risk mitigation techniques will also be briefly explored. The two final

sections of this paper will explain the results of the study and how these findings could be

applied to the catastrophic market. Ultimately, it will determine whether climate change

derivatives could be a financial feasible solution to the catastrophic insurance coverage gap

problem.

LITERATURE REVIEW

As evidenced by Figure One in the previous section, the frequency and economic damages

of natural disasters is on the rise. Even if the frequency of catastrophes were not on the rise,

the increased population density and assets within highly susceptible areas results in

significantly higher post disaster recovery costs. These increasing costs may lead to

insurance companies avoiding catastrophic risk due to fear of financial insolvency. For

instance, studies have been conducted on the United States property-casualty insurance

industry in regards to testing their ability to payout for large catastrophic events during the

late 1990’s. The nationwide resources of the industry, including those of disaster prone

regions such as Florida, were examined. It was determined that if a large catastrophe

equating to $100 billion were to occur, 90% of the damages could be covered by insurance

companies but it would result in the bankruptcy of 140 companies (Cummins, 2006).

Reinsurance companies are also more reluctant to take on the risk of catastrophes.

According to Standard and Poor’s 2015 Global Reinsurance Assessment Review, “In our

view, an increased focus on catastrophe risk weakens a reinsurer’s risk position by

increasing volatility in earnings and on the balance sheet [and thus] most reinsurers display

little appetite for increased balance sheet exposure to catastrophe risk.” In addition, there

could be an issue with insurance companies concentrating their assets at the micro level,

especially with spot transactions. For example, insurance companies could provide contracts

to a high concentration of houses in one neighbourhood. This can affect the loss distribution

and increase the probability that reinsurance payouts will be triggered by a catastrophic

event (Lewis, 1996).

Ultimately, the distribution of losses from catastrophes have fat tails, which can lead to

insurance companies actually increasing their risk by covering a bigger pool of natural

catastrophes. According to Edward Liddy, the President of All State (an American insurance

provider), “The insurance industry is designed for those things that happen with great

frequency and don’t cost that much money when they do. It’s the infrequent thing that costs a

large amount of money to the country when it occurs—I think that’s the role of the federal

government” (Jaffee, 2006). This evidently causes a problem within the market for

catastrophic risk since the ability to predict within a degree of accuracy the overall risks that

could surface is very challenging.

Seeing the apparent need for diversification of catastrophic risk, insurance companies in the

1990’s turned to the financial markets. The Chicago Board of Trade (CBOT) shortly after

Hurricane Andrew in 1992 launched catastrophic futures that were dependent upon the

losses of a pool of various property insurance companies policies. A national index and three

regional indexes were created, and if a catastrophic loss were to occur, the underlying index

would increase. Therefore, an insurer could take a long futures position and reduce a

significant portion of their catastrophic risk (Niehaus, 2002). Within in a few years, the market

introduced additional catastrophic risk protection through the implementation of Catastrophe

(CAT) bonds in 1994. CAT bonds are high-yield bond debt instruments that are triggered

when a particular catastrophe occurs. This would result in a loss or delay in the interest

payment or principal for the investor. Overall, CAT bonds provide fully collateralized

coverage for insurers with clearly defined risks on excess of loss basis (See Figure Three

Figure Three: Structure of a CAT Bond (Swiss Re, 2012)

Nevertheless, since the financial crisis the market has remained relatively small with $5.92

billion USD in 2015 in trading volume (See Figure Four). This may be due to the fact that the

conditions of the agreements are often too specific in relation to when the payment will be

triggered. For instance, CAT bonds were created for the Japanese market amid concerns of

the potential of earthquakes. These contracts, however, only encompassed the region of

Tokyo. The earthquake that hit the rural areas of Japan in early 2011, with damages totalling

over $300 billion dollars, were not considered to be a trigger for these CAT bonds. Of the

$1.7 billion dollars worth of CAT bonds issued for Japan, only one $300 million CAT bond

was triggered (Bloomberg, 2011). Ultimately, CAT bonds have proved to be much more

lucrative for investors than the issuers of the bonds, with returns of 60 percent over the past

five years (2006-2011) according to the Swiss Re Cat Bond Total Return Index. The industry

can be described as a “hole-in-one insurance”. According to Tom Keatinge, managing

director in JPMorgan Chase's (JPM) insurance capital management team, "typically, for a

CAT bond to trigger, you need a bull's-eye to be hit instead of a general shot in the right

direction" (Bloomberg, 2011). Therefore, this market will likely remain small, as it is far less

risky for insurance companies to diversify their risk through traditional insurance.

Figure Four: Catastrophe Bond Risk Capital Issued and Outstanding (Guy Carpenter & Company,

LLC, 2015)

Since their implementation in the 1990’s, catastrophic securities now include an array of

different products besides CAT bonds. There are also Catastrophe Collateralized Risk

Obligations (CROs), CAT-linked derivatives, and Industry Loss Warrants (ILW’s). CRO’s are

essentially a form of a collateralized debt obligation (CDO’s), where catastrophic risk can be

pooled among multiple financial investors. A Specialized Purpose Vehicle (SPV), overseen

by a risk manager, is created which includes a portfolio of risks combining traditional

reinsurance and CAT linked securities. Similar to CAT bonds, CRO’s are also fully

collateralized and provide the sought after benefits of portfolio diversification (OECD, 2009).

CAT-linked derivatives initially when implemented in the 1990’s were fairly unsuccessful due

to their lack of liquidity and awareness by the market. This was due to the Chicago Board of

Trade (CBOT’s) failure at generating insurer and investor interest in these derivatives, as

well as there were concerns with the accuracy of the indexes used in some of the derivatives

contracts. In the original implementation of the catastrophe derivate, the trigger for the index

was determined by “a ratio of total quarterly losses reported by a sample of U.S. insurance

companies to an estimated industry property premium number.” The flaw in the system was

that only 26 companies reported their loss and premium estimates to PCS (division of ISO),

notably absent were two major insurance companies All State and State Farm. When

Northridge earthquake hit in 1994, the ratios had to be altered so that it would be reflective of

the industry. Subsequently, the ISO Index significantly underestimated losses to the

insurance industry (Bouriaux & Tomas, 2014).

Recently, however, attempts have been made to revive the CAT-linked derivatives market

with Deutsche Bank sponsoring event loss swaps and NYMEX, the CME, and the IFEX

introducing exchange-traded futures for catastrophic events. Event loss swaps behave

similarly to Credit Default Swaps (CDS’s), where the buyer pays an initial premium to the

seller of the catastrophic protection, and if a particular trigger is reached in regards to

insurance industry losses then the seller is required to pay out the full notional value of the

swap. NYMEX’s options and futures, on the other hand, are standardized contracts with an

index based off of Property Claims and Services (PCS) data. They were created through a

partnership between NYMEX and Aon Insurance Company. CME hurricane futures are

similar to NYMEX in that it uses an index; however, it is based upon the Carvill Hurricane

Index (“CHI”) that uses parametric conditions such as maximum wind velocity and size

(radius) (OECD, 2009). Industry Loss Warranties (ILW’s) are a relatively new form of

catastrophe risk protection, where payouts are dependent upon industry wide catastrophic

losses rather than one’s own losses. They can either be reinsurance indemnity-based

contracts or through derivatives. Usually, it is in the best interest of the reinsurance company

that they mimic the market since the triggered amount is based upon an industry wide

threshold. Outliers will have a significantly higher risk that their losses will not be similar to

that of the industry average (Ishaq, 2005).

Overall, the drawbacks with the current forms of catastrophe risk protection is that it is not

very accessible to the entire market preventing accurate pricing of these mechanisms.

Access is fairly restricted to only institutional investors, where the general public can only get

exposure through mutual fund companies. Nevertheless, there are only a small handful of

mutual fund companies that will invest in catastrophic risk. Furthermore, since the ratings of

these securities are often within the B to BBB range it may attract hedge funds or CAT funds

but it is unlikely to attract investors who are seeking highly rated securities in the A to AAA

range. These investors are the key players within other securitized markets such as the

Asset Backed Securities (ABS) or Mortgage Backed Securities (MBS) markets. CAT-linked

securities, unlike ABS or MBS, are not correlated with the market (zero-beta assets). This

may not necessary always be true with extremely costly catastrophes, which could transfer

into financial markets (OECD, 2009).

It could be possible for the catastrophic market to obtain ratings similar to that of the ABS or

MBS market. For instance, Fitch has recently assigned a rating of AA with a stable outlook to

a Florida Catastrophe Fund valued at $1.2 billion (Fitch, 2016). Nevertheless, CAT bonds still

usually receive below investment grade ratings due to their high default probability,

historically averaging 1.4% (CAIA, 2015). With governments participation with climate

change derivatives it could significantly reduce the default probability and thus increase the

overall rating for these securities. In addition, it would be advantageous for an individual

investor to invest in the catastrophic market for portfolio diversification especially in light of

the lack of diversification seen during the most recent financial crisis.

Governments in most cases have been reluctant to invest in the catastrophic market as a

form of protection against catastrophic risk. They have traditionally relied on ex-post relief for

uninsured losses from natural disasters. This may prove to be effective in high-income

countries; however, in low-income countries this can have a disastrous impact. Hurricane

Katrina in 2005 accumulated $150 billion dollars in economic damages representing 1.2% of

the United States GDP. On the other hand, Hurricane Ivan that hit the coast of Grenada with

$900 million in damages represented 200% of their GDP. Not only are economic damages a

burdening factor for countries but also human losses. For instance, the major earthquake

that hit Haiti in 2010 was responsible for 300,000 direct and indirect human losses

attributable to poor post-disaster medical treatment (Michel-Kerjan et al., 2011). To avoid the

risk of these sudden financial burdens, some governments have, however, issued CAT

bonds ex-ante to help alleviate the financial costs. In 2006, the Mexican government amid

concerns of the significant costs associated with earthquakes issued a $160 million CAT

bond for particularly high-risk zones in their country (Härdle & Cabrera 2010). The World

Bank in 2014 issued $30 million dollar CAT Bond to help mitigate the risks associated with

tropical cyclones and earthquakes in sixteen Caribbean countries (World Bank, 2014). Both

Mexico and the Caribbean countries received assistance from World Bank to help structure

their CAT bond deals. Michel-Kerjan et al. suggest that often countries do not have the

in-house expertise to facilitate these deals, and thus that could be the reason governments

have been reluctant to invest in CAT bonds.

It has been suggested that creating a partnership between the private and public sector may

be the best solution to reduce the insurance coverage gap. Rather than being reliant on

governments to deal with the large financial burden of post disaster recovery, an

(Litan, 2006). Federal governments have the advantage over insurance companies since

they do not have the similar bankruptcy constraints. Private insurers are hindered by high

bankruptcy costs insofar as it prevents them from being able to diversify catastrophic risks in

an adequate manor. Thus, federal governments could charge a significantly smaller premium

than traditional reinsurance companies for upper levels of catastrophic risk. As of now, the

upper levels of risk have remained relatively untouched preventing market equilibrium to be

achieved within the catastrophic securities market (Lewis, 1996). In addition, the federal

government, unlike private insurers, does not have a “timing risk”, meaning there is not a gap

in time between when premiums are paid and when losses are incurred. This is due to their

capacity to print and borrow money on an efficient basis (Litan, 2006). Reducing both

bankruptcy and timing risk would allow catastrophe premiums to be more reflective of the

true costs without unnecessary risk premium charges. Premiums would not only become

more affordable to the average consumer but also insurance would be more readily available

for homeowners in disaster prone areas.

Governments, in addition, will be vital in creating a large platform for this market to promote

significant growth. Even in the initial stages of the securitization market in the 1970’s the

federal government in the United States was a major participant. Ginnie Mae (Government

National Mortgage Association), a government sponsored agency, pooled mortgage loans

and sold a single-class collateralized MBS to the market. Shortly after, in the early eighties,

two other federal agencies, Freddie Mac (Federal Home Loan Mortgage Corporation) and

Fannie Mae (Federal National Mortgage Association), followed suit with a multiple class MBS

issuance. Financial institutions soon joined the market, and it was not until 2005 that private

issuance surpassed government-sponsored agencies in the securitization market (OECD,

2009).

Ultimately, for this climate change derivative to be fully functioning a complete disclosure of

spending patterns between the insurance companies and government agencies needs to

occur. Asymmetry of information can often be an impeding factor to market growth. This was

especially prevalent in the most recent financial crisis with securitization. Many market

participants were not aware of the risks of mortgage-backed securities, which ultimately lead

to their collapse due to flight of capital. It is very important that within the catastrophic market

participants are aware of the risks to avoid moral hazard and adverse selection. Moral

hazard in terms of the risk transferor not being incentivized to decrease risk, and adverse

selection in terms of the risk transferor not being able to monitor his risk and thus may

securitize an unattractive part of their portfolio (OECD, 2009).

As long as insurance is made available without government disaster relief, creating a

partnership between the public and private sector will significantly reduce the moral hazard

risk that is currently present in the market. The general population is often reluctant to

purchase catastrophe insurance since they know that the government will bail them out if a

disaster were to occur. Also policyholders tend to not be as rational as insurance companies

believing that the likelihood of catastrophes occurring is much less (Jaffee, 2006). With the

public and private insurance sector working together it is likely to increase the adoption of

risk mitigation practices. If insurance is more easily accessible to customers and they are not

able to be reliant on governments to payout for post disaster recovery, they will more likely

adopt better methods (such as improving the structural components of their houses or not

investing in new properties in disaster prone regions). Warnings by government agencies are

not enough to convince the general population to implement risk mitigation measures. For

instance, even with warnings from FEMA (Federal Emergency Management Agency) in the

United States, only 10% of earthquake and flood-prone households have adopted mitigation

strategies (Shreve & Kelman, 2014). Therefore, there needs to be some sort of monetary

incentive for citizens to adopt risk mitigation practices.

THEORETICAL FRAMEWORK

Figure Five: Diagram of Catastrophic Derivative

To expand on Litan’s idea of the government acting as the reinsurance industry for

catastrophes, I suggest the creation of a climate change derivative between the public sector

and the insurance industry. Insurance companies could buy a climate change derivative from

the public sector using a certain percentage of the annual insurance premiums they receive

Insurance Companies

Government Agencies

% Of Customer Premiums

used by governments to invest in risk management practices with the goal to reduce future

expenditures on post disaster recovery. This would not only be limited to natural disasters

due to climate change but could also apply to other natural disasters such as earthquakes,

tsunamis, etc. (any sort of disaster relief could be covered under these derivatives).

An index would need to be determined for the derivative (it could be based on billions of

dollars of damage accumulated or another factor), as well as a trigger point when the

government would have to pay out to the insurance companies. Insurance companies would

be responsible for a certain amount of dollars in damage prior to reaching the trigger.

Utilizing the strategies already used by Asset Backed Securities (ABS), the derivate could be

structured in a similar way. Insurance companies could create a diversified pool of premiums

with varying levels of risk that they could sell directly to the government. The insurance

companies would be similar to the banks with Mortgage-Backed Securities (MBS), where

they would transfer their catastrophic risk to the market, as banks do with mortgages. While

the government would initially play the role of the institutional investors eventually it would be

transferred to the entire market (See Figure Six For Structure of a Catastrophic Derivative).

Figure Six: Diagram of Catastrophic Derivative Similar to a Mortgage Backed Security

In the United States, regulation of the market could be conducted by the already established

agency FEMA (Federal Emergency Management Agency). Credit ratings of the assets could

be undergone by the credit rating agencies, as they do with CAT bonds, or create new

agencies with experience in catastrophe modeling. Catastrophe risk could be diversified

either through different regions across a nation, or even on a global scale with a multitude of

different countries. Senior tranches would hold catastrophic premiums in areas that have a

lower risk of natural disasters occurring, while junior tranches would encompass the areas

that have a higher likelihood and thus have a higher yield. Ultimately, investors who are

Premiums

Premiums

Premiums

Premium Pool

Unsecured

Mezzanine

Sr. Secured

Premiums

Expected

Return

12%

10%

8%

seeking higher risk-return, such as hedge funds, could invest in the junior tranches of the

securities.

In order for the system to function, governments would need to require that households buy

catastrophic insurance with a purchase of a large asset, as they do normally with regular

insurance in regards to automobiles, and state publicly that they will not provide disaster

relief to incentivize people to purchase insurance. The issue of moral hazard that is currently

present would be drastically reduced by these measures. Not only would this reduce moral

hazard but it would also incentivize parties to invest in risk mitigation measures. Customer’s

premiums could be reduced if they invest in disaster prevention techniques, such as

improving the structures of their commercial or residential properties. Insurance companies,

since they are responsible for payouts to their customers up to a certain threshold, will want

to aid the government with risk mitigation strategies to reduce post disaster recovery

expenditures. In essence, the risk management practices should reduce the overall costs of

post disaster recovery benefiting the government, insurance companies, and asset owners

through cost savings.

Once the rules of the derivatives contract and metrics for paying are determined then market

forces can gauge the overall value of the derivative. These derivatives could be traded on the

open market and would therefore be accessible to other parties thus providing price

discovery. The determinant on the size of the market would be dependent on whether it

could be open to investment by other companies in the private sector. For complete price

transparency it would be ideal for the derivatives to be openly traded on an exchange.

Climate change sceptics could even sell these derivatives, while companies who are

drastically impacted by weather related events could buy them. Corporate headquarters that

are located in the United States may be susceptible to damages if they are high-risk areas.

Tourism is another industry that could be financially impacted by catastrophic events, and

they too could benefit from purchasing climate change derivatives. Therefore, there are many

potential participants that could benefit from this market.

Ultimately, the goal of this paper is to determine if climate change derivatives between the

public and private sector can be a financially viable solution to the growing insurance

coverage gap for catastrophic events, and a mechanism to convey market signals to all

players so that rational investment decisions can be made.

RESEARCH METHODS & TECHNIQUES

As stated above, the overall goal is to investigate if it will be financially feasible to create a

climate change derivative between the government and the insurance industry. Since this

paper will explore climate change derivatives on a theoretical basis, an exact pricing

mechanism will not be sought after; rather it will be ascertained if these have the potential to

be applied to catastrophic events. The key component that needs to be explored is whether it

is possible to develop an appropriate index, followed by a brief section on the potential risk

mitigation practices that could be implemented to reduce post-disaster expenditures.

1) APPROPRIATE INDEX

The major component that first needs to be determined is how the derivative will be

structured and when payouts will occur. For this paper, the focus will be on the United States

since they are in a region that is highly susceptible to a multitude of catastrophic events, and

they have a well-established and functional economic system. First, the structure of

catastrophic securities will be explored to give a general indication of how this derivative

could function. The techniques of Asset Backed securities will be analyzed, as well as

already present catastrophic securities, such as catastrophic bonds. Next, developing an

appropriate index will require a significant amount of post disaster recovery data to gain an

indication of the overall costs that need to be accounted for. The Emergency Events

(EM-DAT) Database created by the Centre for Research on the Epidemiology of Disasters

(CRED) will be used to determine frequencies and annual spending on post disaster

recovery for the past seven decades (See Figure Seven). This will be broken down by

natural disaster to determine the proper cost allocations and risk mitigation techniques, as

well as by state to determine which are the most susceptible and therefore would require

higher premiums.

The EM-DAT Database classifies an event as a natural disaster according to these

requirements (at least one needs to be fulfilled):

• Deaths: 10 or more people

• Affected: 100 or more people affected/injured/homeless

• Declaration/International Appeal: declaration by the country of a state of emergency

and/or an appeal for international assistance

The next factor that needs to be explored is how disaster relief expenditures are divided up

between the public and private sector. Since the availability of data is scarce for the private

sector the public sector will be the main focus. Historical expenditures of the public sector will

be analyzed through the disaster relief assistance database from FEMA, a United States

government body that was created to provide financial assistance with catastrophic events. It

should be noted that FEMA only accounts for a small percentage of government spending on

natural disasters, state and local governments are usually responsible for a significant

amount of financial assistance that is not noted in this database. Analysis on a state-by-state

level would be required in order to gain the most accurate depiction of spending. This paper,

however, is more broadly focused and therefore only uses FEMA data. Furthermore, access

to state and local government spending on catastrophic events can be difficult to obtain.

Figure Seven: Frequencies of Natural Disasters for the United States (EMDAT, 2016)

To gain an overall perspective on the current market of the insurance industry, the private

sector industries annual reports and market outlooks will be evaluated. Unfortunately, access

to catastrophic spending by insurance companies is quite limited. Munich RE provides a

general indication of insured versus uninsured losses on an annual basis for the United

States (See Figure Two in the Introduction). Nevertheless, detailed statistics on a state level

are not accessible. This will be one of the main benefits of creating a partnership between

insurance companies and the government so that a free flow of information can be

transferred between both parties. Ultimately, once a general indication of the spending

0

5

10

15

20

25

30

35

40

1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

EMDAT Frequencies of Natural Disasters for the United States

patterns of each sector is determined then it will aid the process of determining the

appropriate index between the two parties.

In order to determine an estimate of the premiums that could be charged to the general

public, the overall asset exposure to catastrophic events needs to be analyzed. Asset

exposure will be ascertained by using average home price data from Lincoln Institute of Land

Policy and housing unit data from the U.S. Census Bureau, both on a national and state-wide

level.

!"#$"%&'("  !""#$%&  !"#  !""#$  !"#"$   =  

!"#$  !"#$#%&'  !""#"$%&'(

!"#$%  !"#$%"$#&'  !""#$  !"#$%&'(  

!"#$%&$  (!"#$  !"#  !"#$) =

!"#$  !"#$#%&'  !""#"!"#$%

!"#$%&  !"  !"#$%&'  !"#$%

Dividing FEMA disaster assistance by total structural asset exposure will give an indication of

the structural assets exposed to catastrophic risk by state on a per asset level. Forecasts of

future total structural asset exposure will also be determined. Premiums (cost per unit) that

could be charged to the average household annually are obtained by dividing FEMA disaster

assistance by the number of housing units. Although overall catastrophic losses would be

better suited to use than FEMA data, the EMDAT database does not provide data past the

national level. An annual premium can be determined with EMDAT data for the United States

as a whole, but not on a state level. Furthermore, included in this analysis will be a

forecasted population dispersion and accompanying assets data, which will give a gauge of

how much of the population is projected to migrate to disaster prone areas in the United

States and how many assets will be a risk.

Next, to get a better indication on how the derivate would be structured on a state level,

Florida data by county will be analyzed. Data was obtained from the U.S. Census Bureau

and FEMA database. FEMA disaster assistance was analyzed by each county, including

assistance applied statewide. Statewide assistance was distributed to each county based on

their counties total market value of assets percentage of the entire Florida housing market.

Total market value county data was obtained through the Florida Department of Revenue

database. To determine the average house price, total market value data was divided by

housing unit data from the U.S. Census Bureau. The same process as above for determining

premiums by each state was used to determine the premiums charged to each Florida

County. A further breakdown in each county would need to be conducted to get an

appropriate premium charged to each household. Again it should be noted that only FEMA

disaster assistance data has been analyzed and it will not be fully representative of the entire

premium a customer would have to pay. Total economic damages by Florida country are not

currently available from the EMDAT database.

2) RISK MITIGATION TECHNIQUES

The second component that needs to be explored is risk mitigation strategies. The ultimate

goal is to determine the potential cost savings from utilizing particular methods, and whether

there are certain risk mitigation techniques that are universal to all disasters, such as

improving building structures, or more specific to a natural disaster, such as levies for

flooding or hurricanes. In this section, however, it will only focus on the vulnerability of

building structures in residential areas. The age of housing units within a particular region

can affect the amount of damage that can occur after a natural disaster. In most cases,

houses that were built a significant amount of time ago are more likely to have weak

structural components and fewer protective measures (result of stricter building codes and

improved technology). States that have housing infrastructure that is older could be deemed

higher risk, while the opposite is true for States that have newer structures. Housing data

from the U.S. Census Bureau will be analyzed to determine the potential vulnerabilities in

housing structures. The year in which housing units are built, ranging from 1939 or Earlier to

2010 or Later, will be broken down at the state and national level by decade. Overall, this

information can aid in determining which risk mitigation strategies are the most cost effective

and what will be the most beneficial in reducing the ex-post recovery expenditures. The mere

charging of premiums and a policy of no government aid could result in fewer new structures

being built in risk prone areas.

RESULTS

1) APPROPRIATE INDEX

The results section is broken down into three different components. First, both EMDAT

frequencies and economic damages, and disaster assistance by FEMA were analyzed to

give an indication of which disasters have been the most costly and which states were the

most impacted by catastrophic events. Next, the housing market in the United States was

assessed to determine which states have the most total assets at risk. Finally, both FEMA

assistance and total structural asset exposure are utilized to determine an annual

catastrophic risk premium per average household by state. The section will then conclude

with a detailed analysis into natural disaster exposure in the state of Florida.

CATASTROPHIC DISASTER DATA

_Frequency

(1948 - 2015)

Frequency

(2000 - 2015)

Average

(1948-2015)

Average

(2000-2015)

Disaster Type

Storm

559

219

8.22

13.69

Flood

167

79

2.46

4.94

Earthquake

34

7

0.50

0.44

Drought

13

9

0.19

0.56

Extreme Temperatures

35

13

0.51

0.81

Other (i.e. Wildfire)

76

52

1.12

3.25

TOTAL

884

379

13.00

23.69

Figure Eight: United States Frequency of Catastrophic Events from 1948-2015 [EMDAT Database]

To begin, the frequency of catastrophic events were analyzed to get an indication of which

events were more likely to occur. Storms were responsible for over 63% of total frequencies

from 1948 to 2015, with an annual average of 8.22. From 2000 to 2015 this average

significantly increased to an average of 13.69 events per year. Floods were the second

largest contributor representing around 19% of the total, followed by the other category and

extreme temperatures. Overall, 884 events were recorded by the EMDAT database from

1948 to 2015. The number of events annually averaged at 13, while over the latest time

period it increased to over 23 events annually (attributable to a large increase in storm and

flood events). Even compared to the time period of 1980 to 2015, the annual average

frequencies of natural disasters from 2000 – 2015 have increased by 1.8 events per year

_{EMDAT Economic}

Damages (1948-2015)*

Average (1948-2015)

Average (2000-2015)

Disaster Type

Storm

$821,341,554,985

$12,078,552,279

$34,292,639,740

Flood

$105,502,381,455

$1,551,505,610

$2,269,384,106

Earthquake

$76,569,349,954

$1,126,019,852

$243,404,962

Drought

$45,869,981,465

$674,558,551

$2,623,843,580

Extreme Temperatures

$37,089,737,379

$545,437,314

$232,371,746

Other (i.e. Wildfire)

$29,298,413,647

$430,859,024

$1,210,486,942

TOTAL

$1,115,671,418,885

$16,406,932,631

$40,872,131,075

*Damages Scaled to 2015 $USD

Figure Nine: United States Catastrophic Economic Damages from 1948-2015 [EMDAT Database]

Total economic damages by disaster type were then assessed for the United States on a

national basis. As noted above, this data was obtained from the EMDAT database. Since

1948, catastrophic losses have amounted to $1.1 trillion with an average of $16.4 billion in

losses annually (See Figure Nine). In the last seventeen years, the average annual loss has

increased by two and half times ($40.8 billion), and the frequency of catastrophic events

have increased by 75%, now averaging 9.5 events annually (See Appendix 1 for Detailed

Table).

Disaster Type

FEMA Disaster

_Assistance

(1999-2015)*

Average

(1999-2015)

EMDAT

Economic

Damages

(1999-2015)**

% Of Total

Economic

Damages

Hurricane

$48,292,093,779 $2,840,711,399 $380,614,049,861

12.69%

Severe Storm

$13,531,909,120

$795,994,654

$21,334,571,633

63.43%

Fire

$7,549,201,611

$444,070,683

$19,478,654,361

38.76%

Severe Ice Storm

$2,170,818,637

$127,695,214

Flood

$2,123,208,760

$124,894,633

$36,310,572,494

5.85%

Snow

$1,085,582,399

$63,857,788

Coastal Storm

$642,285,529

$37,781,502

Earthquake

$397,089,997

$23,358,235

$3,894,479,393

10.20%

Tornado

$286,179,175

$16,834,069

$87,280,489,808

0.33%

Typhoon

$176,651,234

$10,391,249

Other

$192,780,896

$11,340,053

Unclassified

$126,075,574,610

TOTAL

$76,447,801,136 $4,496,929,479 $674,988,392,161

11.33%

*Damages Scaled to 2015 $USD

**Empty Values Due To Different Classifications of Catastrophes

Figure Ten: United States FEMA Disaster Assistance & EMDAT Economic Damages by Disaster Type

from 1999-2015

Next, FEMA’s disaster assistance spending by disaster was analyzed to determine if there

were any spending trends (See Figure Ten). Over the time period from 1999-2015, $76.4

billion dollars has been spent by FEMA, averaging $4.5 billions dollars per year. As a

percentage of total economic damages over that same period, disaster assistance accounted

for 11.33%. It should be noted that when comparing FEMA disaster assistance to EMDAT

economic damage, the two databases have different classifications for disaster type, and

therefore not all categories can be accurately assessed. Categories that have data in both

categories are listed in Figure Ten above; empty values represent categories that are under

a different classification system. Similar to EMDAT’s economic damage, hurricanes account

for 63% of the disaster assistance provided by FEMA, followed by severe storms with 18%

and fire with 10%. The other categories such as floods and tornados have received minimal

assistance, representing the remaining 9% (See Appendix 2 for Detailed Table).

State**

FEMA Disaster

Assistance

(1999-2015)*

Average

(1999-2015)

Top Three Disaster Types

New York

$21,567,344,329

$1,347,959,021

Hurricane, Fire, Severe Storms

Louisiana

$16,677,473,611

$1,042,342,101

Hurricane, Fire, Severe Storms

Florida

$6,437,085,607

$402,317,850

Hurricane, Coastal Storm, Fire

Texas

$4,782,227,990

$298,889,249

Hurricane, Coastal Storm, Severe Storms

Mississippi

$4,177,961,849

$261,122,616

Hurricane, Severe Storms, Severe Ice Storm

TOTAL TOP

FIVE

$53,642,093,386

$670,526,167

Delaware

$44,351,850

$2,771,991

Hurricane, Severe Storms, Snow

Utah

$36,813,300

$2,300,831

Flood, Severe Storms, Other

Nevada

$36,141,486

$2,258,843

Severe Storms, Fire, Snow

Wyoming

$12,983,731

$811,483

Severe Storms, Flood, Severe Ice Storm

Idaho

$11,939,423

$746,214

Flood, Severe Storms, Hurricane

TOTAL

BOTTOM FIVE

$142,229,790

$1,777,872

*Damages Scaled to 2015 $USD

**Excluding Territories of the United States

Figure Eleven: United States FEMA Disaster Assistance by State from 1999-2015

To further the analysis, FEMA disaster assistance was then broken down by state (excluding

territories) to give an indication of which regions in the states are receiving the most funds

(See Figure Eleven). It should be noted that this is not indicative of total economic damages

in each area rather based on assistance provided. New York has received the most disaster

assistance totalling $21.6 billion over seventeen years with an average of $1.3 billion

annually, representing 28% of the total assistance provided nationally. A significant portion is

associated with hurricanes (64%), followed by fires (29%). Louisiana is also responsible for

22% of overall damage assistance with $16.7 billion in damages, averaging a little over $1

billion per year. Similar disaster types to New York have plagued the region. Idaho has

received the least amount of assistance over the time period totalling $11.9 million,

averaging only $746,000 annually (See Appendix 3 for Detailed Table by Disaster Type).

Over the seventeen-year time period, 2005 was the worst year for catastrophes

accumulating $193 billion in economic damages and $23.6 billion in disaster assistance. This

accounted for 29% of total economic damages, and 31% of FEMA assistance. The states

that were largely impacted that year were Louisiana (64% of disaster assistance), Mississippi

(16%) and Florida (10%). The following year had the least amount in economic damages

with only $7.6 billion in comparison; however, it was not the smallest amount in terms of

disaster assistance provided by FEMA, 2015 only had $764 million. (See Appendix 4 for

Detailed Table on Annual Spending by State)

State**

Frequency Of

Disaster

Declarations

State**

Average FEMA

Spending Per

Disaster

Declaration

New York

45

Louisiana

$694,894,734

Oklahoma

44

New York

$479,274,318

Kansas

41

Florida

$195,063,200

Texas

34

Mississippi

$160,690,840

Florida

33

Texas

$140,653,764

TOTAL TOP FIVE

197

AVERAGE TOP FIVE

$334,115,371

Nevada

9

Delaware

$3,695,987

Idaho

8

District of Colum

$3,623,263

Michigan

8

Maine

$3,598,662

Utah

8

Wyoming

$2,596,746

Wyoming

5

Idaho

$1,492,428

TOTAL BOTTOM FIVE

38

AVERAGE BOTTOM FIVE

$3,001,417

TOTAL U.S

1141

AVERAGE U.S.

$67,000,702

*Damages Scaled to 2015 $USD

**Excluding Territories of the United States

Figure Twelve: United States FEMA Disaster Declarations by State & Average FEMA Spending Per

Disaster Declaration from 1999-2015

Frequencies of FEMA disaster declarations were examined to give an indication of which

states were more susceptible to disasters (See Figure Twelve). New York had the highest

number of disaster declarations with 45 events; around half of them were due to severe

storms and a quarter due to snow. Oklahoma and Kansas high frequencies were mainly

attributable a large number of severe storms. Wyoming had the fewest number of disaster

declarations over the seventeen time period with only five, followed by Utah, Michigan, and

Idaho with eight events. Overall, the average number of events per year for the United States

was 71.31 with severe storms accounting for over 53% of all disaster declarations (613

events). Floods were responsible for 170 events (15%) and snow for 103 events (9%). (See

Appendix 5 for Detailed Table on Frequencies by State)

Total FEMA disaster assistance was divided by frequencies to get a gauge of the average

spending per disaster declaration (See Figure Twelve). It should be noted that frequencies of

disaster numbers do not necessarily translate to frequencies of natural disasters. In most

cases this is true, however, some larger events may have more than one disaster number

due to occurring over a larger time period. Louisiana had the largest average spent per

disaster with almost $695 million; this was largely attributable to hurricane disaster

assistance with over $2 billion per disaster for eight hurricane events. The second top sate

was New York with an average spending of $479 million. Similar to Louisiana, the average

spending per hurricane was $2.7 billion over the seventeen year time period but there were

fewer frequencies with only five events. Although hurricanes have been higher in frequency

for New York and thus represent more off the overall spending, fires had the highest average

FEMA spending per disaster declaration. This was associated with fires that occurred on

September 11

_{, 2001. Overall, the top five states average equalled $334 million, while the}

bottom five states equalled $3 million. The smallest average spent per disaster was Idaho

with only $1.5 million, then Wyoming and Maine with $2.6 million and $3.6 million

respectively. The United States’ average was $67 million per disaster declaration, with an

average of $284 million per hurricane and $222 million per fire. Even though, frequencies for

severe storms make up the majority of total events, only an average of $22 million is spent

per event. (See Appendix 6 for Detailed Table on Average Annual Spending by Disaster

Declaration)

UNITED STATES HOUSING MARKET

Next, the United States structural housing market data was assessed to determine the total

asset exposure by state (this excludes the land value) (See Figure Thirteen). The top five

states in terms of number of housing units account for 35% of the entire United States

market; while the bottom five only represent 1%. Average structural property values for the