Viewpoint Using Capital Markets to Develop Private Catastrophe Insurance 19928 John D. Pollener The global catastrophe insurance market exhibits inherent cyclical risks. Disaster-prone countries can improve their protection against catastrophic risk and premium volatility by using capital markets to boost the capacity of the private sector to absorb and spread the risks-both domestically and internationally. This Note proposes two mechanisms for more efficient management of catastrophic risk: pooled insurance coverage supported by liquidity and credit enhancement facilities, and hazard- indexed bonds to securitize risk. Support from multilateral development institutions can play a catalytic role in the development of these mechanisms while still preserving actuarially fair premiums. Some small countries may require multilateral support (a credit enhancement) to gain access to capital markets. Or perhaps all that is needed is for the government and local industry to recognize that the new mechanisms can both increase cover and stabilize premiums. Such an assessment requires scarce financial modeling skills-something multilaterals can help organize and finance. Catastrophic events are unique among insur- Also in the United States an options market hased ance risks: while traditionally insurable risks on insurecl loss indexes has been set up at the occur with preclictahle frequency and relatively Chicago options exchange. These options allow low losses, catastrophes occur infrequently insurers to hedge some of their catastrophic risk but with high losses. Three mechanisms have in thc capital market. In Japan. Europe, and thc been developed in recent years to better man- United States markets for 'cat honds" offer age these risks by tapping capital markets. another alternative for transferring risks. in these In the United States public authorities in markets investors hbLy high-yield bonds that are earthquake-prone California and in hurricane- backed by premiums on the underlying insurance prone states such as Florida and Hawaii col- portfolio, btut also contain provisions that trig(er laborate with the private sector to address loss of interest and principal if a major catastro- insurers' fears of large losses (and thus poten- phe occurs during the life of the bond. Investors tial market flight). These public-private efforts are attractecl to these honds because of their lack combine catastrophe coverage under special- of risk correlation with global financial markets. purpose pooled funds with outside capital (such as long-term loans and bonds that secu- Some of these risk management methods coulC ritize insurance risks), increasing the avail- he adapted for use in developing countries. ability of insurance. The pooling arrangement achieves efficiencies of scale while giving all High and volatile premiums parties the confidence that contractual obliga- tions will be met. This approach is also used in In catastrophe-prone developing countries the Europe. domestic insurance industry generally reinsures The World Bank Group * Finance, Private Sector, and Infrastructure Network Using Capital Markets to Develop Private Catastrophe Insurance Reinsurance prices are quoted as "rates on line." Assume, for example, that a primary insurance company wishes to purchase reinsurance for catastrophic losses between US$15 million and US$25 million. If losses in a catastrophe are less than US$15 million, the company retains all the loss. This ceiling is referred to as the retention level. If the loss is more than US$15 million, the reinsurer pays Transferring risks abhroad also means that because the excess up to the USS25 million-the excess-of-loss level. If the reinsurer the domestic insurers retain little of the underwrit- charges US$1.5 million for this US$10 million of coverage, the rate on line-the ing risk, they face louw incentives to monitor com- price expressed as a percentage of the coverage-is 15 percent. pliance with structural codes or to promote Rates on line vary by type of risk. A primary insurance company doing busi- measures to mitigate losses or improve industry ness in an area at low risk of catastrophes will pay the lowest rates on line. Rates efficiency that in the long run might lover the cost also vary by retention level. The rate on line for USSI0 million in coverage above a of insurance. Whe0n all but very low levels of risk retention level of US$15 million would he much higher than that for US$10 million Of insured all t verancew coverage is are reinsured abroad, the reinsurance coverage is above a US$100 million retention level. The reason is that there is a higher proba- gener bility that losses will exceed US$15 million than that they will exceed US$100 mil- th ally expensive because of the high likelihoot lion. A typical average for rates on line is 13 percent. The rates for hurricane-prone th igh ereds o ntractinguld inow rane at countries are high relative to worldwide catastrophe rates. But retention levels for much higher levels of loss would lower the pre- small, disaster-prone countries are much lower on average than for U.S. insurers. miums beCauSe of the lower probability that losses would reach those levels. Thus a key cost issue is finding the right balance over time between retain- its local portfolio through international insur- ing risk locally ancl transferring it abroad. ance or reinsurance companies that can better bear the risk of catastrophic losses through Transferring catastrophic risks to the global diversification. While this reinsurance capital market transfers the risk abroad, it may not provide the kind of coverage that could ensure economic A simple numerical example shows how capital recovery and fiscal stability in the wake of a nat- markets can replace or supplement insurance ural catastrophe. Excessive risk transfer might and reduce costs. Assume that a primary insur- also imply low capital among local insurers. ance company calculates the probability of a loss of more than USS15 million but less than US$25 Insurance for natural disasters is generally avail- million at 10 percent. If the primary insurer pur- able for even small developing countries. But fol- chases reinsurance at this rate, it will break even lowing an unusual series of major losses, over time. Adding administrative and operating domestically or globally, insurance could costs, the reinsul-cr might charge a premium of become scarcer and pricier. Thus relying on rein- 12.5 percent (10 percent + 2.5 percent). surance from foreign companies affects the price of disaster insurance (box 1). Prices have been Alternatively, the primary insurer could issue a dropping since 1995. But they could rocket if a US$10 million ondl to investors, then put the major catastrophe occurs and catastrophe rein- US$10 million in U.S. treasury notes paying, say, surance markets tighten. In 1990 the rate for 5 percent. The investors' principal of US$10 mil- home insurance in parts of the Caribbean wvas lion would be put at risk as part of the contract. 0.4 percent. After Hurricane AndrewI hit Florida If a catastrophe wi ith losses exceeding USS25 mil- in 1992 prices more than tripled, reaching 1.3 lion occurs, the investors might lose all their prin- percent in 1994, then declined to 0.8 percent in cipal. For putting their principal at risk, the 1997. By contrast, Hurricane Mitch, which dev- investors would demand at least a 15 percent astated parts of the Caribbean and Central return-5 percent as risk-free interest and 10 per- America in 1998, had little effect on insurance cent for the 'pure" risk of losing their principal prices because much of the property it damaged (akin to a default risk). Net of the investment in was uninsured. Thus the price of insurance treasury notes, the insurer's total financing cost tripled for property owners in the Caribbean at would approach 10 percent, compared with the a time when there had been no change in their 12.5 percent with traditional reinsurance. underlying risk. Instead, the price fluctuations reflected changes in the supply of catastrophe In yet another option the insurer could arrange reinsurance following a disaster elsewhere. a standby credit of tJSS10 million with a 2 per- cent commitnient charge ancl an interest rate of cover wotuldl also serve as a partial buffer 12 percent that kicks in if the loan is n e(tdcl. If aga,inst tuLctuations in international reinsur- a catastrophe occurs, and assuming a ten-year ance pricing, since the loan temns wvould repayment period for principal (yielding a corn- remain unchanged. hined principal plus interest insurance" cost of * Once Such arrangements prove hinancially 18 percent), the expected financing cost wVouldC viahle, local financial markets or international he 3.6 percent (0.1 [18 percent] + 0.9 [2 percent]), commercial lenders coulci offer liquiditysupport much lower than with direct reinsurance. facilities. Development of these instrulmtents Nou'ld he catalyzed hby the credit provided hy These capital market schemecs to replace or sup- multilateral development institutions. plement insurance have many possible vatria- tions. These range from frill risk transfer with no While this mechanism would finance rather than financing where the full principal is at risk, just transfer risk, if structLred xvith proper temris and as in reinsurance) to zero risk transfer with frill approprialte levels of excess-of-loss coverage. itcotild financing (full repayment of principal). provide more cost-effective coverage and longer- temr price stability thalnt traditional reinsurance. Applying new insurance technologies Hazard-indexed catastrophe bonds Two compatihle financial structuLres COLldd he used to address the challenges of c.atastrophe Weathaer-indexecl catastrophe honds. hased on insurance in disaster-prone countries, separately payouts linkedl to measurable Weather events (as or as a joint mechanism. reflected in wveather inclexes or pararnetric inea- shires). have the advantage of heing relatively easy Pooled coverage supported by liquidity and to implement once a reliable wvealther merasure- credit enhancement facilities ment mechanism is identified. Thiey hypass the traditional insurance loss adjustment." wlhicll A mechanism in whichi liquidity and creclit en- requLires site-hy-site evaluation of losses before hanceimlent facilities suppol)n insuranCe coVerage' indemnity is 1)po\ icled. The payout is sini)ply ba.sed against catastrophic risks could fuinction as followvs: on the wveather incdex reaclhing a certain range. * The domestic instirance incdustry xvould trans- Payouts for the Tokyo Marine Insurance cat hood, fer catastrophic coverage (through premiuLml for example. are basecd on specifiecd Riclhter me.t- cessions) to a central fund regulated by the gov- stores of earthquake intensity and damnages witlhin ernment and operatecl by the insurance indLus- a specified radius arouncl Tokyo. try. The risks covered wouLld not he reflected on the balance sheets of local insurers hut would T'he main risk, with hazard-indlexecd instnurments is instead be liabilities of the pooled fund. The hasis risk-the risk that the basis for triggering the international insurance industry could then loss payment (such as a high windspeed, excessive reinsure catastrophic coverage Linder the funcd rainfall, or earthquake intensity) is not clirectly uip to a specified loss limit. linked to actual loss (such as damage to a specific * Multilateral development institutions mighlt house or building). A loss payment may he nuaed provide contingent credit at the next highest (with the bondholder losing interest or principal) loss level, supporting the liquidity of the ftund even though the insured experiences no loss. Or in the event of immediate large losses in the the insured may experience a loss but receive no initial years of operation. The credit wvouldl inclemnity because the index is not triggerecd (a., a eventually he repaid and securecl through result of a lower-than-threshold xvindspeet). future premium collections by the fond. The extended repayment period would provide Most catastrophe boncds-SUch as those issuecl in optimal risk spreading over time. This layer of EIurope and the United States-are triggered by Using Capital Markets to Develop Private Catastrophe Insurance reportecL losses and inclemnification claims in the the catastrophe insurance pool, serving as one industry rather than weather indexes. But of the upper layers of coverage. Such risk trans- investor appetite for such honds issued in devel- fer mechanisms-involving capital market oping countries might he low because of lack of investors under multiperiod contracts-can fur- clirect knowledcge of the local insurance industry. ther reduce the potential volatility of insurance l3onds hased on easily verifiahle weather indexes and reinsurance prices. They would also enahle should he more attractive. Aclding to their attrac- the government to insure public property at rea- tiveness are the opportunities they w;ould offer sonahle prices. Ancd they would enahle the local international investors for portfolio diversifica- insurance industrv to extend coveragc to such tion, since natural disasters have little or no cor- hard-to-insure sectors as small farmers, public relation xvith glohal financial market trends. infrastructure, and low-income communities. What would the financial structure of a scheme A catalytic rote for development hased on hazard-indexed honds look like? institutions * Catastrophe honds could cover public infra- stnuLcturie or provide financing to a private pooled Financial support from miltilateral development fttnd. as descrihed in the previous section. The institutions to create an insurance pool and honds would payv igher-than-average yields but hazard-indexed honds, separately or together, woould aLso carTy a risk of a significant drop in would meet sevcral ohjectives. The support Viewpoint is an open the coupon rate or a loss of principal in the event woould help reclicc potential market failures due forum intended to of a catastrophe that leads to loss payments. to historical premium volatility that result in encourage * A multilateral instittition or affiliate could guar- lapses in coverage. It wouldc also help overcome dissemination of and debate on ideas, antee the contractual payments of bond suboptimal coverage resulting from the scale dis- innovations, and best couipons and any principal due investors. In a economies in small countries' insurance markets, practices for expanding fully private arrangement the hond payments and the lack of incClntives for measures to miti- tie private sector. The wol edt e ul eue views published are would need to he fullv secured by the premi- gate catastrophic losses. And because capital those ofthe authors and urns collected in the common fund. market-basecl arranngements could increase the should not be attributed - The hasis for triggering loss payments couldi insttred asset hase in developing economies of its affiliated or ganiza be information from weather station tracking while promoting reliahility in economic com- t ons. Nor do any of the equipment wvith satellite links to global record- pensation folLowing natural disasters, the sup- conclusions represent ing centers or from weather monitoring sys- port could encourage participation hy the official policy of the World Bank or ofrita tems of the U.S. National Weather Service, international reinsturance industry. Executive Directors or which tracks high-altitude hurricane activity in the countries they the Pacific ancd Atlantic regions. Earthquake Such initiatives go hand in hand with the needed represent. risk couldl likewise he measured. restructuring of local insurance industries. Thus To order additional the involvement of multilateral institutions could copies please call Before hazard-indexed honcls are introdticed, help strengthen the domestic insurance industry 2t2458 lit, ordcontac historical data on hazarcd events and associated and improve hazard mitigation. Where regional Room F11K 208, losses wxottld need to he compiled and analyzed rather than nationaL arrangements are optimal, The World Bank, to ensuire a suff'icientlv strongy correlation mtiltilateral institutions could facilitate inter- lRtt H Street, NW, Washington, D.C. 20433, hetwveen index-triggered payments and actual country policy diiatogue. or Internet address losses. These data are essential for structuring ssmith7@worldbank.org. ancl pricing suchi insurance contracts. John D. Pollner (/)ollner@n'orldbank.oqgj The series is also Senior Financial Sector Specialist. Latin availiabl e on hnieo (vwvw.worldbark.org/ A joint mechanism Amnerica and the Caribbean Region, Finance, html/fpd/notes/). Private Sector and Infrastricture Department @ Printed on recycled White a hazard-indexed hond couild he devel- papert oped on its own. it could also he comhined with