WfS Zs t) POLICY RESEARCH WORKING PAPER 25 60 Catastrophe Risk In providing support for disaster-prone areas such as Management the Caribbean, the development community has Using Alternative Risk Financig tDbegun to progress from Using Alternative Risk Financling disaster reconstruction and Insurance Pooling Mechanisms assistance to funding for investment in mitigation as an John D. Pollner explicit tool for sustainable development. Now it must enter a new phase: applying risk transfer mechanisms to address the financial risk of exposure to catastrophic events that require funding beyond what can be controlled solely through mitigation and physical measures. The World Bank Latin America and the Caribbean Regioni Finance, Private Sector, and Infrastructure Sector Unit H F_. _._ -Caribbean Mid-% | 2 / / '-_ < |-4- OECS Mid-50%/ l 3.0% _____ 2.D% x' 0.0% 1991 1992 1993 1994 1995 1996 1997 1998 Source: Benfield Greig Excess of loss reinsurance rates (rates-on-line)' followed the same pattern both in the Caribbean as a whole and in the OECS countries as seen in the above chart. High volatility was experienced in excess of loss premiums at the middle (50%) layer of reinsurance cover, that is, at the midpoint in the range of excess-of-loss coverage layers. The period between 1993-95 was the peak with rates on line reaching 7%. At the lower layers of cover which would be accessed more often (e.g.: first 5%), rates were much higher during 1993-95, in some countries reaching up to 16% of loss amounts insured. 8 The term "rate on line" refers to the premium dollar price as a percentage of the loss level (in $ damages) to be covered under an excess of loss contract. In this sense it is akin to cost of capital or the interest cost of capital. Rate on line should not be confused with the premium as a % of total insured value. Since rate on line is the price of compensation for exact losses measured in dollar terms, such loss levels are generally a sub-set of the total insured value (and based on a probabilistic application to expected and maximum losses). - 25 - Average Caribbean Catastrophe Rates - Commercial Properties 180 160 140 120 f or caatoh-rmr-rmu ae n omrilpoet hr oaiity DRs Global Carib- OECS Barbados Domin. Trinidad The Jamaicaarbado XL Bahamas so (0 (%Jamaica E ' I -i- Cayman *j60 40 20- on 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 est. This patter was experienced across the wider Caribbean as the above chart shows for catastrophe primary premium rates on commercial property where volatility was particularly apparent in rates charged in countries such as Jamaica but also Bahamas and Cayman Islands. The volatility parameters are summarized below: Volatility Indicators of Global and Caribbean Reinsurance and Primary Insurance Rates Global Carib- OECS Barbados Domin. Trinidad The Jamaica Cay- XL bean XL XL Republic & Bahamnas man Rate Rates Rates Tobago Islands Price (50% (50% Commer- Index mid- mid- cial Comm. Comm. Comm. Comm. Comm. point) point) Rates Rates Rates Rates Rates Rates Standard 1.3% 1.6% 2.1% 0.13% 0.17% 0.16% 0.34% 0.45% 0.34% Deviation Rate Mean 2.7% 5.2% 4.4% 0.65% 0.59% 0.6% 0.92% 0.83% 0.79% Standard Deviation 49% 31% 48% 20% 29% 27% 37% 54% 42% Normalized to Mean T'he key indicator to take account of above, is the standard deviation normalized to the mean (coefficient of variation). This shows the actual percentage deviation of the rates from the mean over the period. As can be seen, the global excess-of-loss reinsurance (XL) price index had a percentage standard deviation of 49% around the - 26 - mean and OECS XL rates were similar at 48%. Primary rates in Jamaica were at record highs showing a percentage standard deviation of 54%, in part due to both the effects of hurricane Andrew in 1992, and previously hurricane Gilbert in 1988 which directly hit Jamaica. The above analysis shows the large volatility in insurance pricing which affects the Caribbean. Following hurricane Andrew, additional reinsurance capacity was created in the Bermuda market as traditional major reinsurers such as Munich Re exited the Caribbean market and world reinsurance prices increased. This additional capacity along with the global development of capital market instruments such as catastrophe bonds financed in the capital markets, have helped stabilize volatility. Nevertheless, future volatility cannot be ruled out, and even a fraction of that seen in the past would be destabilizing to Caribbean insurance markets and the availability of coverage needed against natural hazards. Within this context, it is observed that a type of 'contagion' effect occurs globally when major disasters use up significant portions of reinsurer capacity. While this contagion does not occur in the same manner as emerging market jitters occur during economic crises, it has a similar effect and essentially reflects supply constraints which result in higher pricing to reach a new supply/demand equilibrium. Such adjustments are widely recognized by insurers and reinsurers alike, and reflect a 'recovery' effort by both insurance and reinsurance providers for capital lost or depleted due to large events. In this sense, catastrophe reinsurance contains elements of risk financing ('finite insurance') to fund the insured party, with compensatory payback obtained through temporarily higher rates. It is widely stated in the industry, therefore, that due to large catastrophe exposures, the low rates seen in the last two years will not continue. Supply Shift and Demand Effect after Catastrophes Post Event Supply Function Pre-Event Supply Function N = New Price & Capacity Equilibrium Demand Function Quantity (Reinsurance Capacity) Another factor affecting the stable supply of catastrophe reinsurance relates to the manner in which the market is organized. As expressed in Goldman Sachs' Fixed Income Research on Insurance Linked Securities: "Natural catastrophes such as hurricanes and earthquakes reduce private wealth just as do economic losses on diversified portfolios of stocks and bonds. Nevertheless, a substantial portion of this exposure is borne by private corporations (privately held reinsurance companies) and small numbers of wealthy individuals (Lloyds unincorporated names). This inefficient sharing of risk has resulted in high reinsurance rates and has motivated many primary insurance companies to maintain very large retained exposures and to cease issuing new - 27 - or renewal policies in selected areas. A more efficient risk sharing procedure would allow the capital markets to spread the risk among large numbers of investors to whom this exposure is a very small portion of their total risk exposure" (Goldman Sachs: Fixed Income Research Series on Insurance Linked Securities). Professor Kenneth Froot of Harvard University, similarly points out some aspects of these market inefficiencies and why catastrophe risk is distributed inefficiently: "Another reason institutional arrangements may be inefficient is that the lack of objective information acts as a kind of barrier to entry. When objective information is costly to assemble, a greater investment is required to get into the underwriting business. Indeed, when objective information is in short supply, markets tend to be organized around relationships and reputation. By contrast, when objective information is plentiful, markets tend to be organized around transactions, with the players being more interchangeable. Because newcomers are discouraged from entering the market, the incumbents who specialize in underwriting cat risks, such as cat-risk reinsurers, can more easily charge high prices" (The Market for Catastrophic Risk: A Clinical Examination, August 1999). However, information can be improved through institutional strengthening to increase the collection and dissemination of critical information. In the U.S. for example, the Insurance Services Office enables insurers to achieve and share economies of scale in the collection and analysis of data. Such institutions contribute to improved underwriting and pricing decisions on the part of individual insurers by making pooled data and analyses of pooled data widely available. Global Reinsurance Capacity and the Caribbean Market 1999 was the third highest loss year in terms of insured losses from catastrophes ($18 billion) with associated economic losses of $80 billion. Similarly, 1998 was the fourth highest loss year in terms of catastrophes causing global insured losses amounting to $15 billion and overall economic losses of $94 billion. This compares to catastrophe insured losses of approximately $27 billion in 1992 and $18.4 billion in 1994, the two years of highest losses to date. Of the $15 billion in losses in 1998, almost 80% were due to windstorm disasters. Catastrophe excess of loss reinsurance cover globally amounted to about $75 billion, and this represented on average, 40% of total coverage with the remainder held by primary insurers or under quota share treaties. The earthquake share of insured and total economic losses in 1999 was substantial, representing over 35% of total losses. Thus, recent annual losses worldwide averaging around $20 billion would represent about 27% of the current excess of loss capacity and 1 1% of total available insurance and reinsurance before considering the net surplus capital in the industry. If future windstorms (hurricanes, typhoons, cyclones) increase, or if property affected were in a high value area such as Miami, Florida, there would likely be a squeeze on available reinsurance capacity as additional coverage to protect against large events would reach into the limits of global reinsurance capacity. While potential devastating events globally could in aggregate cause up to $200 billion in insured losses, this scenario will, statistically speaking not occur in any single year or closely consecutive years. However, because of these constraints and the potential for worldwide catastrophes to significantly 'bite' into the reinsurance market's - 28 - capacity, the consideration of alternative risk transfer and financing instrument makes sense in the context of providing additional available risk financing via the world's capital markets. The table below indicates that, in the period following hurricane Andrew, the rates for ESC nations were generally in line or higher than those for Miami, but much higher than the rates for the less hurricane exposed cities of Tampa and Tallahassee. Thus the international insurance pricing equally affected those areas affected by the hurricane (Florida, Bahamas) and those Caribbean countries which were not affected. Base Property Rates (on insured value) for ESC Countries Compared with Coastal Areas in the U.S. (1994 - post hurricane Andrew) ESC and other CARICOM countries Price in % per $0.1 mn. limit Antigua 1.10 Bahamas 1.20 Barbados 1.00 Belize 0.90 Dominica 0.90 Grenada 1.02 Montserrat 1.00 St. Kitts & Nevis 0.95 St. Lucia 1.15 St. Vincent 1.05 Trinidad & Tobago 0.77 Turks & Caicos (associate member) 1.11 United States (Florida) Miami 0.99 Tampa 0.59 Tallahassee 0.43 The pricing for excess of loss reinsurance, i.e., the rate-on-line, varies by type of risk. A primary company writing in a low catastrophe exposed area will pay the lowest rates on line. Rates also vary by retention level. The rate on line for $15 million in coverage above a retention level of $10 million will be much higher than for $15 million above a $100 million retention level9. This is because there is a higher probability of losses exceeding $15 million than exceeding $100 million. Rates on line for ESC countries, vary from 25-30 percent at the lowest levels of retention (with highest exposure and probability) to 2-3 percent at the top levels. An average level would be 9 percent. Retention levels for the ESC countries are much lower on average than for U.S. insurers, but exposure to hazards are higher. Initially this makes sense from the perspective of 'transferring the risk' outside of the affected 9 Retention level is the insured amount that is 'retained' by the original/primary insurer after it has 'reinsured' or 'ceded' a portion of its portfolio to the reinsurer. In the East Caribbean, primary insurers reinsure on average 80% of their portfolio, thus effectively retaining only 20% as an insurance risk on their books. -29 - countries into the international market, however, as the subsequent analysis will show, excessive transfer abroad with very little retention can also be sub-optimal from a cost- effectiveness and risk-sharing point of view. A different approach to addressing the issue of adequacy of prices is to compare rates on line with "pure" rates developed by models which project hazard frequencies and their intensity in terms of damage functions. This report addresses this issue under the hazard and financial modeling sections. The methodology used for this purpose includes the following steps: Stochastic Module: Hazard Module: Generation of hazard Characteristics of hazard event and statistical > intensity (windspeed, frequency distributions earthquake ground motion), (historical data, scientific hazard 'shock' distributions analysis, expert opinions). at different intensities. Damage Function: Financial Module: Calculation of structural Loss quantification in $ damage, vulnerability terms. Insurance contract coefficients for different pricing based on hazard intensities exceedance probabilities of (engineering expertise, $ loss levels. Determination damage experience). of capital requirements. Under the damage module, a vulnerability function for property structures is established against levels of hazard intensity. This is the 'real sector' input used for subsequent financial modeling for insurance: Damage Severity (expressed in $) 0 50 100 150 200 Hazard intensity (e.g.: wind speed) The above reflects the typical "S" shape damage/severity function for hurricanes. At the lower wind speeds, the damage function increases more slowly in proportion to the increase in wind speed. As the winds surpass 1 00 mph, the damage level begins increasing proportionately at a higher rate. After reaching top winds, the damage - 30 - function decreases in its 'acceleration' since the incremental wind speed does little additional damage to the gross destruction already done. To provide some perspective of the context of this risk market, it is estimated that the ESC countries generate currently about $150 million in insurance premiums. Most ESC countries face a high level of risk. The islands are in the front line of Atlantic storm activity. A number of the islands have significant earthquake risk as well. A few countries have a volcano risk, most notably Montserrat, where an eruption in 1996 caused extensive damage. Projections by risk management firms show that the East Caribbean region can expect 2.5 storms every year. Severe hurricanes, defined as category 3, 4 and 5, have, up until the late 1 990s been less common. Category 3 hurricanes can be expected to occur every second year, and Category 5 storms every fifth year. Reviewing recent hurricane experience, the following records are obtained: Hurricanes which Affected Caribbean Countries Between 1970-1999 Year Hurricane Country 1974 Carmen Jamaica 1974 Fifi Belize, Jamaica 1979 David Dominica 1980 Allen Barbados, St. Lucia, St. Vincent, Jamaica 1984 Klaus Dominica 1985 Gloria Antigua, St.Kitts, Turks & Caicos, Anguilla 1987 Emily Grenada, St. Vincent 1987 Floyd Bahamas 1988 Gilbert Dominica, Jamaica, G. Cayman, St. Lucia 1988 Joan Tobago 1989 Hugo Antigua, Dominica, Guadeloupe, Montserrat, St. Kitts 1992 Andrew Bahamas 1994 Debbie St. Lucia 1995 Iris Antigua, Dominica, St. Vincent, T & T 1995 Erin Bahamas 1995 Marilyn Dominica 1995 Luis Anguilla, Antigua, Dominica, Montserrat, St. Kitts & Nevis 1998 Georges St. Kitts & Nevis, Antigua 1999 Floyd Bahamas 1999 Jose St. Kitts, Antigua 1999 Lenny St. Lucia, Grenada, St. Kitts, Dominica, St. Vincent - 31 - The experience of the twenty-year period, suggests that the Caribbean countries would be damaged by two tropical storms each year and by a hurricane every two years. In general, climatologists view the 1969 to 1989 period as below average in terms of cyclonic activity in the Atlantic. So far, nine hurricanes have hit the Caribbean islands in the 1990s, or an average of almost one per year, double the rate of the prior 20 years. There are significant regional differences in exposure, however. Northern and eastern islands are more exposed than southern islands. The Leeward Islands including Antigua, Montserrat, St. Kitts and Nevis are viewed as being most exposed, while Trinidad and Tobago have minimal exposure. Insurance loss modeling thus, involves three main steps: An assessment is made of the probability of the underlying natural phenomena, either hurricane storms or earthquakes and their intensities. This information is developed from historical meteorological and seismic data. The second step is to assess the exposure level and vulnerability of insured property to differing hazard intensities (as per the graph above). The third step is to combine the information from the first two steps and produce a probable loss distribution in dollar terms. To illustrate, consider an extremely simple example of a small island, subject to only a category II type hurricane. The probability of such a hurricane striking the island is 10 percent. In other words, it can be expected that such a hurricane will hit the island once every ten years. The value of insured property on the island is $100 million. It is estimated that a category II storm hitting the island will cause $20 million in insured loss. This information can then be used to calculate appropriate annual insurance rates and to decide on the maximum capital ($20 million) needed in any given year to pay losses. Insurance loss modeling develops data on these two main figures -- annual average loss and probabilities of maximum losses -- from data on multiple meteorological or seismological events. The Nature of Catastrophic Risk Catastrophic risks in primary markets tend to be spread temporally more than spatially. For example, prior to Hurricane Andrew, insurance companies added a 14 percent catastrophic property factor to insurance premiums in Florida. With residential insurance premiums totaling $1.2 billion in 1992, this resulted in the catastrophe portion of premiums of $168 million per year. This factor was based on loss data for 30 to 40 years. The assumption was that catastrophe losses would average out to 14 percent of the premium over 40 years. In any single year, a large hurricane could cause losses many times the $168 million in premium. However, the companies assumed that over a 30 to 40 year period they would break even. The pricing of catastrophe reinsurance is based on the notion of spreading risk over time. The reinsurance company promises to indemnify the primary company for losses above a retention level. For example, a primary insurance company might purchase reinsurance for catastrophe losses between $20 million and $30 million. In pricing this $10 million layer of insurance, the reinsurer will first estimate the probability of loss in this range. Let us assume that this probability is 10 percent. In other words, the reinsurer believes that once every ten years the primary insurer will need to pay losses of - 32 - $20 million or more. If the reinsurer charges a rate of 10 percent it will break even over time. This rate is known as the "pure premium". To come up with the final rate, the reinsurer adds a factor for expenses and profit. The final rate is expressed as a percent of the $10 million layer of insurance provided and is called the rate on line, as discussed earlier. There is a third way that risk is spread by insurance, which is of some relevance to the ESC region. We can label this way as "speculative." Insurance organizations provide insurance for somewhat exotic risks such as satellite launches. The pricing of these risks are calculated on a slim if any actuarial basis and they are not spread over space or time. It could be argued that if enough of these event risks are taken on, they involve a spreading of the risk. In other words, if Lloyd's insures 100 such diverse risks, and by the laws of probability only one of them results in a claim, then the premiums for the 99 are paying for the loss of the one. Some risks in the ESC region could be viewed as falling into this "speculative" category. A reinsurance or insurance company with excess capital might take a risk from the ESC region on a "speculative" basis. The net effect of treating ESC reinsurance risks in this fashion is to create increased volatility in the marketplace. In a period of excess capacity in worldwide reinsurance markets, reinsurance will be readily available to primary companies in the ESC region. In periods of tight capacity, reinsurance will be in short supply. There is also the danger of a major or even total withdrawal of reinsurance capacity to the region. The state of Hawaii faced this problem following Hurricane Iniki in 1992 and in Fiji in the mid-1980s. Such a withdrawal would have disastrous consequences for the economies concerned. Without insurance, the financing of construction would be greatly diminished. The fact that catastrophe risks are spread over time has two important implications for ESC markets: Reinsurance or risk transfer relationships need to be long-term in nature: Major primary insurers around the world have long-term relationships with their reinsurers. So if a reinsurer has a large loss in a particular year, the "loss" will be made up by payments from the primary company in future years. If reinsurers do not have a long-term relationship then they view a large loss as more in the speculative area where they are making a one-year bet. As a result they will require a higher profit rate for this risk, as has occurred in the ESC region. - 33 - Insured and Uninsured Losses from Weather Related Natural Disasters 100 90 ....... . ?- - 80 70 60 e ;- - C NN '-Uninsured 50 Insured 40 30 10 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 Source: Munich Re, Guy Carpenter, WorldWatch The above chart shows the economic (unisured) losses and insured losses from 1980 to 1999 for weather related disasters, globally. From weather related disasters alone (excluding earthquake), 1998 was the year with the highest total economic losses worldwide. 1999 was the third highest year for total economic losses as well as weather related insured losses. Generally, hurricane and windstorm losses in a given year represent about 70% of total losses, however, in 1999, earthquakes represented almost 40% of insured losses from events in Colombia, Turkey, Greece and Taiwan. 1999 was also a year with significant flood damage in Central Europe and South East Asia as well as hailstorms in Australia and gale storms in Europe (Lothar, Martin). Windstorm damages include for the U.S. and the Caribbean those from hurricane Floyd, the Orissa cyclone in India, tropical cyclone Bart in Japan, as well as major tornadoes in Oklahoma. Total economic losses from weather related and earthquake events combined approximated $100 billion in 1999 compared to $93 billion in 1998, and for all events, insured losses were higher in 1999 as well ($22 bn.)compared to 1998. The intensifying loss trends, requiring matching insurance risk capital, can be attributed to increased urbanization, population growth, rising growth of both number and value of properties in catastrophe prone areas, as well as climatological and environmental changes. Given the context of global disaster risks as well as the risk exposure in the Caribbean, the following chapter examines the operation of the property insurance market in the Caribbean. Based on the nature of catastrophic risk, the organization and structure of the domestic market in the Caribbean represents the first line of defense for successful risk management. The following chapter thus identifies some of the domestic industry and regulatory factors which could constrain effective capacity in the management of such catastrophic risks. - 34 - -35 - II. THE PROPERTY INSURANCE MARKET IN THE CARIBBEAN The structure of the Caribbean insurance industry has to-date depended primarily on a strategy of substantial transfer of insurance risks to the international reinsurance markets. While such a strategy has allowed a reduction in financial exposures, it has also reduced the incentives for domestic underwriting and risk management capacities, and has linked the local market development cycle to external market movements. The large number ofplayers in the local domestic markets along with the practice of extremely modest retentions of risk, does not permit sufficient flexibility at the country level to optimize the risk management, risk transfer and premium pricing options in the most cost effective manner. The high vulnerability of individual countries calls for improved risk pooling strategies, some of which have been attempted in limited ways in the past. These can help lower the insurance cycle risk while promoting regulatory changes to both strengthen domestic insurance suppliers and promote improved loss reduction measures. - 36 - II. THE PROPERTY INSURANCE MARKET IN THE CARIBBEAN Large catastrophic events like hurricanes have direct costs as they lead not only to large losses of capital stock and inventories but also indirect costs in lost income, employment or services which result from lost productive capacity. These events jeopardize internal and external macroeconomic stability leading to larger than anticipated public sector and balance of payments deficits. For example, the direct effects of hurricane Gilbert on Jamaica in 1988, amounted to US$956 million representing 27 percent of GDP, with half from losses in agriculture, tourism and industry, 30 percent in housing, and 20 percent in economic infrastructure. As a result of the hurricane, losses in export earnings were estimated at US$130 million representing 14 percent of exports, and the government incurred US$220 million in additional expenditures while the public sector deficit increased from an earlier expected 2.8 percent of GDP to 10.6 percent, also fueling inflation. While 80% of the gross property insurance premiums are transferred to reinsurers, the actual remittance flow is reduced by the reinsurance commission paid by reinsurers (e.g.: 30% of 80% = 24%). A year with abnormally high property claims experience can of course, result in a net remittance inflow from reinsurers. Tight reinsurance conditions, i.e., a high priced market, can give very good results for well managed Caribbean companies as high reinsurance costs materially boost reinsurance commission incomes. Policy coverage restrictions are generally designed and imposed by foreign reinsurers, and their effect falls on the policyholders rather than on the insurance companies. Historically, there have been very mixed feelings as to the industry's role for a proactive involvement in promoting hazard and vulnerability mitigation measures. While not denying the inherent benefit of such measures, the insurance companies' concerns center on the implementation complexities and costs, particularly as reinsurers are seen as unlikely to share in these costs. Companies view the leadership role for mitigation measures as lying with their governments. The insurance markets are intensively competitive for the property insurance classes -- a competition primarily seeking reinsurance commission revenues rather than underwriting, or 'risk taking' profits. The larger Caribbean insurance markets contain insurance companies (with sizeable markets shares), forming part of broader commercial groups. It is estimated, however, that several insurance companies are under capitalized and markets are saturated. New regulations expected to come into effect in Barbados, the OECS and Trinidad & Tobago will significantly increase capital requirements thus likely leading to mergers and buy-outs within the local industry. - 37 - Structure of the East Caribbean General/Property Insurance Market (1998) Antigua Barbados Dominica Grenada St. Kitts St. St. Trinidad Total and and Lucia Vincent & Barbuda Nevis & the Tobago Grendns No. of 16 22 13 19 9 21 12 29 141 General Insurance Companies % Foreign 88% 45% 95% 76% 67% 86% 85% 12% 66% Companiesa Property $ 10.6 $ 43.0 m. $ 5.8 m. $8.6 m. $6.3 m. $11.9 $ 9.6m. $ 52.1 m. $148 Insurance million m. mn. Premiums (Gross) Gross 1.8% 3.0% 2.4% 2.7% 2.4% 2.1% 3.3% 0.9% -- Premiums % of GDP . a. Of these, approximately /4 are companies from Trinidad, Barbados and Guyana. Insurance Sector Regulatory Characteristic in the East Caribbean Markets ($US Dollars Equivalent) Antigua Barbados Dominica Grenada St. Kitts St. Lucia St. Vincent Trinidad and and Nevis and the and Barbuda Grenadines Tobago Minimum $0.07mn $1.5 mn $0.07 mn. $0.09 $0.07mn. Local $0.07 mn. $0.16m Capital mn. $0.10m Requirement Foreign $0.20m Solvency Min. $0.25mn. Min. $0.06m Min. $0.07m or Min. Min. Margin capital or 25% capital or or 10% capital or 20% of capital. Or capital. or (Assets - or 10 % of 10 % of of 10% of premium 10% of 20% of Liab.) above of prem. premium premium premium premium Income premium premium premium Income income income income income income Income income Reserve na 40% of 30% na 10% of 40% of na 40% of Requirement annual annual annual annual premium premium premium premium premium income income income income Premium tax na 5% None na 5% Local: 3% 3% 6% foreign: _____ 5% Corporate 40% 40% 30% 30% 38% 33.3% 40% 35% tax Observations on East Caribbean Market Characteristics The proportion of residential and commercial properties in the Caribbean covered by insurance is significantly higher than in most developing countries, on account of both the susceptibility to natural disasters but also on the influence of tourism and the requisite insurance of tourist facilities. In comparison with the insurance density in the U.S. (3.3% of GDP for the property and casualty business), the average for the OECS, Barbados and Trinidad & Tobago is 2.3% of the combined countries' GDP. - 38 - The number of general/property insurance companies in the East Caribbean, however, is large. The ratio of premiums earned ($149 mn.) to number of primary companies writing property business (numbering 145) is just a little over $1 million, i.e., the average premium written per company. In contrast, the average premium written per company in the U.S. (2,500 companies in total), is $112 million or a multiple of one hundred times that of the East Caribbean. However, such a comparison may not be necessarily meaningful given the different levels of development and insurance markets. Nevertheless, if one takes the relative populations serviced by insurance companies, the results are: 14,000 inhabitants are served per each insurance company in the East Caribbean versus 107,000 inhabitants in the U.S. served per each company. This would suggest a potentially over extended industry in the East Caribbean which implies inefficiencies of scale in terms of both operating costs and risk management. A significant share of companies, however, effectively function as agencies with little desire to operate as genuine risk underwriters. While most of the East Caribbean countries do not admit non-registered insurance companies, the actual insured base in each country is likely higher than reported due to non-admitted providers. In those countries where non-registered companies are allowed to conduct business, large commercial and tourist properties are directly insured abroad. Due to the non-reporting nature of these businesses at the Caribbean level, figures on this market are difficult to estimate although via the tallying of the total market value of insurable assets and comparison of these with the premiums collected annually, one could, by process of elimination deduce the quantity of assets insured directly abroad. On average, 75% of the OECS market is held by Trinidadian and Barbadian companies. Thus, the East Caribbean insurance market (prior to reinsurance) is effectively a Barbados and Trinidad dominated market. Regulatory Developments Solvency margin regulations generally follow the EU regulatory systems whereby such margins are indicated by capital requirements of at least 20% of premium income. In Trinidad & Tobago, features of the Canadian system were adopted. However, new Canadian norms recently developed, include additional safeguards such as the provision for catastrophe reserving for earthquakes, a practice which would equally apply to the hurricane and wind storm risks in the Caribbean. Expense ratios in the Caribbean are relatively high, between 30%-40% of premium income compared to the U.S. average expense ratio of between 26%-28%. Expense ratios reflect the costs of business acquisition, brokerage fees, underwriting fees, administrative costs and overhead as a percentage of annual premiums earned. In the Caribbean, due to the relatively small size of companies and diseconomies of scale, expense ratios constitute a higher percentage of annual premium income, despite the fact that administrative and staff costs are generally lower than in the U.S. Such costs as well as low retentions of underwriting risks prevent a quicker build up of capital reserves/surplus, than could othervise be achieved. A new OECS Insurance Act is under review by those member governments, and this would substantially raise minimum capital requirements as in Trinidad. In Trinidad - 39 - & Tobago, the minimum capital requirement will be changed to $1.6 million to be phased in within 5 years. The new OECS Act would require minimum capital of $0.9 million for local companies and a requirement of $1.8 for foreign companies. The differentiation is meant to provide a competitive adaptation period for indigenous companies to reduce costs and increase their scale economies, and for foreign companies, to ensure that sufficient capital for operations and claims settlement, is maintained. In Grenada and St. Lucia existing regulations already provide for differentiated capital requirements for foreign companies. The new Insurance legislation, however, will attempt to harmonize the regulatory requirements across all of the OECS to include uniform solvency margins (e.g.: 20% of premium). In OECS countries, all insurers must be registered locally whether foreign or domestic. In contrast, Barbados and T&T have separate regulations for non-registered foreign companies, although there is concern about monitoring such business more carefully given the non-reporting of such entities. In terms of exchange controls, approval by most central banks is required for transfers above specified level (as in the banking sector) and in some cases, a remittance charge of 1% is applied to premiums ceded to reinsurers. However, these regulations are being relaxed in line with WTO standards, and thresholds requiring approval are being increased. Reserve requirements are generally high and require prescribed investments of surplus of companies, a factor which might reduce overall profitability if other more high yielding investment opportunities become available. Premium (sales) taxes, corporate taxes, and stamp taxes vary widely across the East Caribbean - factors which also will require harmonization if further market integration and pooling structures are to be considered. Catastrophe Insurance Reserves Insurance company reserves fall into two basic categories. First are the shareholders' capital and free 'surplus' reserves, and second, the insurance or 'technical' reserves. The latter are customarily tax deductible being constituted for known liabilities such as payment of reported but unpaid claims and the unexpired portion of pre-paid, (e.g. annual), premiums. In effect, the capital and free reserves represent the solvency margin and are intended as the last asset resource should the technical reserves prove inadequate. Relating these considerations to natural hazard peril insurance, there are two particular factors. First, although catastrophes are accepted as severe but infrequent events, in accounting terms (Generally Accepted Accounting Principles - GAAP), they cannot be precisely forecasted as to timing or amount. Secondly, over 80% of the insured catastrophe liabilities fall under reinsurance contracts placed mostly outside the Caribbean. These two factors prompt a local insurer to first determine how to prudently reserve for the net liability retention (approximately 20%) before the catastrophe event occurs, and second how to ensure that the reinsurers are financially secure enough to meet their liabilities fully and on time for the lion's share (over 80%) of the liabilities. - 40 - On the first challenge, Caribbean insurers were, until recently, discouraged by existing tax laws to have incentives to set up specific reserve provisions for catastrophe perils before a catastrophe event. Several Caribbean countries are now perrnitting tax deductibility for such dedicated reserves. Without this dispensation, very little of premium paid becomes available to meet future catastrophe claims liabilities under issued policies. In the absence of a natural hazard event, less than 20% of premiums customarily see their way to increasing free reserves. Operating expenses can characteristically consume 40% of premiums, income tax about 3 5%, and a 40% dividend policy on the remainder would use up a further 10 %, thereby leaving approximately 15% passing on to free reserves. Hence, in the policyholder's interest, there should be allowed dedicated, properly monitored, tax deductible catastrophe reserves. Some larger and special risk categories (e.g.: power utilities), have also over recent years found it impossible to obtain full, and in some cases, any, affordable insurance. On occasions such risks have voluntarily devised very high self-insured deductible levels aimed to cover the expected loss damage potential and separate self- insurance funding for business interruption. These risk management arrangements have served to attract greater levels of insurance/reinsurance cover for the higher, less exposed, risk levels. Furthermore, the insurance cost and availability difficulties have prompted trade associations (e.g. Caribbean Hotel Association - CHA), to employ risk management techniques and/or off-shore captive insurance company arrangements to buy reinsurance on a group basis. A catastrophe insurer's gross potential liabilities (potential losses on sums insured under natural hazard peril policies), can run into hundreds of millions, if not billions of dollars. The gross liabilities are reducible to net liabilities though reinsurance. Provision also needs to be made for so called second event 'reinstatement' scenarios as reinsurance contracts customarily vary from the primary policies' provisions by limitations on protection amounts available for second (and subsequent) catastrophe occurrences during any one reinsurance contract period. At present, a majority of primary insurers' reinsurance contracts cover second event catastrophe losses. Clearly, a company's capital and free reserves, plus any dedicated catastrophe reserves (and corresponding assets), need to be readily realizable for these purposes. The assets held to cover a company's insurance or 'technical' reserves should not be considered as available for catastrophe claim payment purposes as these technical liability reserves are constituted for their own specific purposes - customarily to cover prior known reported outstanding claims and the unexpired portion of pre-paid (e.g. annual) premiums. These observations argue for an insurance company's capital and free reserves plus any dedicated catastrophe reserves, to be seen as strong enough, and readily available to meet rationally estimated net of reinsurance catastrophe liabilities. The sections that follow also show how the pooling of catastrophe risks also improve the efficient frontier for a more optimal capital/exposure ratio. An insurance company's investment policy should also select investment instruments which best meet the purposes for which policyholders purchased their insurance, i.e., the invested funds be secure and available for meeting aggregated claims at unknown future dates. One consideration has to be the secondary potential of natural hazard events to adversely affect local financial markets at the time asset realization is required. Specific regulations should support the view that preferred instruments are - 41 - found in hard currency financial markets least likely to be impacted by natural catastrophes. Likewise, instruments should be placed in open financial markets entirely on an 'ann's length' basis (without any strings attached as opposed to some existing regulations which allow up to 20% of assets being held in intra-industry group holdings). Regulatory Improvements and Mitigation of Financial Risks The catastrophe insurance market has a parallel with the overall financial services regulation, whereby the 'hazard' to be mitigated may be a liquidity crunch with a potential for a run on a banking system. A natural hazard could provoke a run on an insurance system. Contagion effects, albeit for different reasons, can occur, and in the insurance case this is particularly evident following catastrophes wherein supply of capital has been compromised and demand for coverage surges. In both the banking and insurance examples, there is the potential for a secondary 'hazard' of public recriminations if the regulatory perfornance before the event is perceived inadequate. Development of strong effective and harmonized insurance regulation, therefore, is recommended to include: (a) minimum capitalization requirements for local carriers and brokers, (b) solvency & liquidity levels, and adequacy of technical reserves; (c) adequate asset/liability management (including maturity and currency matching where applicable), as well as reinsurance credit risk, (d) incentives and requirements (including tax concessions) to build-up catastrophic reserve funds up to minimum required levels, (e) minimum standards for non-ceded retention of local coverage, (f) accurate verification and valuation of companies' balance sheet entries to ensure adequate financial capacity to cover claims, (g) increased allowance for overseas investments of insurance assets, (h) industry entry requirements including admission of suitable foreign competitors, (i) requirements to verify security and reliability of overseas reinsurers who take on portfolios of local coverage, () linkage of insurance regulation to building code compliance prior to providing insurance coverage, or to discounts based on vulnerability reduction measures, (k) monitoring and inspection techniques, (1) conditions for revoking licenses and shutting down operations; and (m) Consolidated regional & institutional harmonization of regulators/supervisors. Recent Limited-Basis Pooling Initiatives For the most part, government physical assets such as buildings, schools, libraries, roads, and some hospitals are uninsured or underinsured. Exceptions include Barbados with the government owned Insurance Corporation of Barbados responsible for insuring - 42 - public assets. Additional exceptions include properties owned by statutory corporations such as port and airport authorities, as well as utility companies that can independently access the insurance markets. Utility companies in particular, to reduce insurance cost, have been actively considering a regional self-insurance program with the Caribbean Development Bank (CDB) and the Caribbean Electric Utilities Service Corporation (CARILEC). Although in the conceptual stage, the program's principal concepts include a backstop credit line in the initial years of premium accumulation. As the fund grows, the utility companies should rely less on the line of credit until it eventually becomes a standby support to be used only after the fund is depleted because of claims arising from a catastrophe event at the uppermost loss levels. The Barbados Light & Power Company Limited (BLP) started its own self insurance program, due to unavailability of Transmission and Distribution (T & D) catastrophe insurance coverage in 1993, and the subsequent extraordinary high rates to obtain the coverage, estimated at 25% of the estimated losses of assets in a catastrophic event. The fund is composed of cash and committed lines of credit. Similar to the regional approach, the lines of credit are to be used only after the cash portion of the fund is depleted. BLP is now able to look at its insurance needs from a more reasonably priced excess of loss layer, above the level of its self-insurance fund. The company estimates very significant annual premium cost savings. In 1993, The Caribbean Hotel Association (CHA) retained a US-based risk management firn to perform a pan-Caribbean study regarding wind storm risks to its members' properties to see if there was some way of reducing the upward spiraling costs of insurance. The computer-generated wind study performed by a sub-contractor of the risk management firm provided a probable maximum loss profile of the region and divided the Caribbean into 6 different risk zones. The study suggested that there appeared to be enough diversification of risks among these zones to allow a regional insurance company for the CHA properties to survive a 1.3% probability of a major storm disaster event. Using the expected loss (EPL) information as the starting point, and based on their own financial modeling capabilities, the risk management firm determined a capitalization figure for a regional insurance company to sell 'all risks' property insurance to each of the 1,000 CHA or so members. The risk management firm then created, and today manages a Bermuda insurance company whose exclusive clientele are members of the CHA. Incentives and Disincentives to Risk Based Pricing For many homeowners, the answer to soaring insurance rates, during the 3 years following 1992, was to reduce or eliminate their coverage. With the lack of statistics, it is difficult to measure the degree to which capital stocks are underinsured though 30% is estimated in the insurable housing sector and in most ESC countries few government assets and buildings are insured. From an insurance industry perspective, underinsurance is business lost which would otherwise carry little incremental acquisition cost. Furthermore, policyholders who terminate their insurance seldom return. Insurance, is a discretionary expense, and once terminated, becomes excluded from buying patterns. Local governments are, of course, concerned since the greater the non-insurance (or - 43 - underinsurance) by a population, the longer it takes for the local economy to recover from catastrophic events. Two local trends, as yet slowly evolving, may suggest some change in strategy in the markets. First is the setting up of tax deductible dedicated catastrophic reserve funds by a few insurers. This, at current understood levels, will take a long time to justify reduced reliance on reinsurance. Second, there is some increase in the buying of 'excess of loss' reinsurance versus the traditional 'proportional' or 'quota share' reinsurance. The former is perceived more costly up front, making this strategy affordable only for the financially stronger companies, however, its use can permit better risk management and accumulation of capital for well managed companies. Further trends include several nations' regulatory reforms which embrace very much needed increased minimum capital requirements and tighter solvency ratios for insurance companies. Regrettably, from an overall disaster mitigation standpoint, there has existed for Caribbean catastrophe insurance companies a strong economic rationale inhibiting the spread of vulnerability reduction measures via insurance incentives. As earlier discussed, blanket reinsurance pricing across a whole portfolio works against encouraging insurance companies' adoption of individual risk discriminatory premium pricing. This argument, on the surface, is valid and is exacerbated by the market's fierce competition for reinsurance commission revenues. Insurance companies therefore have perceived that their primacy lies in obeying the reinsurers' imposed broad pricing methodology and defending their reinsurance commission revenues. They fear that significant premium discounts for the better protected risks (however well merited), cannot be balanced by surcharging the poorer risks and they would end up paying the shortfall in reinsurance premium from their own pockets. Given that characteristically the reinsurers' share of catastrophe policies' premiums exceeds 80%, one can appreciate the basis of insurers' attitudes. One could argue that an insurance company should be better off with a portfolio of the better risks, to which discounts had been awarded and would result in fewer claims -- this argument, however, has little acceptance because such a strategy would imply loss of the reinsurance commission from the poorer risks no longer insured. This is not, however, necessarily seen as an incentive to reinsurers who take on the bulk of overall catastrophe liabilities, i.e., both the good and poorer risks, since it would imply closer and better monitoring costs, and more highly differentiated risk pricing. In addition, Caribbean primary insurer attitudes were also influenced until some 3 years ago by the earlier tariff market mechanisms which worked to limit price competition. One particular segment of the insurance industry, however, has for over a century been practicing vulnerability reduction as part of its underwriting practices: insurance companies participating in the Factory Mutual (FM) system and the Industrial Risks Insurers (IRI) consortia based in the U.S. impose stringent design criteria and supervise construction as preconditions for underwriting a property. As a consequence, participating companies, some of which have had links to business in the Caribbean, have a much better understanding of the true risk to which a property is subject and are able to offer much lower premiums than traditional insurance companies. However, as mentioned, the two inhibitors to encouraging vulnerability reduction measures via premium incentives suffer themselves from being non-discriminatory from - 44 - a reinsurer's standpoint. Both the blanket portfolio reinsurance premium rate and the reinsurance commission being commonly priced on all ceded premiums do little to promote overall risk quality in a reinsurer's accepted portfolio of risks. There are some profit commission arrangements, very limited for 'catastrophe' reinsurance specifically, whereby the earned reinsurance commission rate varies by the loss ratio experienced - i.e. from the claims incurred. Such arrangements still, do not directly address portfolio risk quality improvement. The adaptation of the working of reinsurance premium pricing and reinsurance commission mechanisms is feasible to meet the objective of encouraging improvement in risk quality of portfolios and hence allowing discriminatory premium pricing for vulnerability reduction measures by policy holders. What is first required is a set of meaningful and workable risk quality criteria mutually agreed by an insurance company and reinsurers. The implementation of risk-based premium pricing, to be successful, would therefore require a comprehensive hazard mapping effort, the inventorying of all typologies of structures and contents with respect to vulnerability, and certification by engineers of individual risk characteristics. Coupled wvith the actuarial estimates of event probabilities and intensities, and identifying market values which can be used to calculate probable losses, such an exercise would thus set the base for a more accurate catastrophe risk accounting methodology in the Caribbean insurance industry. This would include classifications of: i. Hazard mapped locations ii. Building type & structure vulnerability characteristics iii. The same as above for building contents iv. Engineering certification for individual risk characteristics - needed to retain discipline and deter discounts being given for 'sales' reasons. v. Application of the above to actuarial/statistical distributions of event probabilities and intensities. vi. Deterrnination of fair market value of properties to combine the above factors in establishing expected loss amounts. To implement such a process which reflects preconditions for improvements in the insurance markets, it is recommended that a Caribbean privately funded insurance services office be established. Such an institution which could operate at a regional or sub-regional level, would be well positioned to develop insurance plans by asset class and advise on prospective loss costs taking into account individual risk characteristics. This would permit primary insurers, at their option, to use such standardized class plans and loss costs in pricing their products. At the request of insurers, such an office would also provide reinsurers with pertinent data which could facilitate the negotiation of reinsurance contracts. Additionally, functions such as inspections of individual properties would be conducted when requested by individual insurers. A key function would also be to maintain updated insurance and risk data on the local industry. Factors Affecting Catastrophe Insurance Demand Insurance policies cover the 'natural' catastrophe perils of earthquake, volcano and storm; the last embraces declared hurricanes (>75 mph sustained winds), windstorm, flood and stormn surge. Volcano and flood cover is practically unattainable in some areas recognized as particularly prone to these events (e.g.: Montserrat). -45 - Since the late 1 980s, full value coverage for catastrophe perils is generally limited by claims deductibles expressed as percentages of the full insurable values. For earthquake these deductibles are characteristically 5% and for hurricane 2%. Soft market conditions in some limited areas have reduced the latter to 1%. The impact of claims deductibles can be severe to policyholders. As an example, a dwelling valued at $100,000 would only recover above the $2,000 deductible. This amount can be sufficient to meet repair costs for characteristic partial roof damage but a very material amount when related to average disposable income levels. Policy conditions also provide for the amount of claim to be reduced to the extent the amount of insurance purchased (sum insured) is less than the full insurable value at the time of loss (the 'average clause'). Some variations of this exist, e.g., the adjustment is only made if the policy sum insured is less than 85% of the full market value. Policyholders in Caribbean nations with weak currencies and/or high inflation are especially vulnerable to these provisions. Socioeconomic and behavioral factors also merit consideration on both the supply and the demand sides of the catastrophe insurance market and it is helpful to segment property assets into their ownership classes: PrivateDwellings Small/Indigenous, Medium, Large Private Businesses Small/informal, Medium, Large Public Ownership: Building Structures, Utilities, Public Infrastructures With regard to private dwellings and business properties, the owners' disposable income levels comprise an important factor in the demand for catastrophe insurance the annual cost of which is near 1% of a structure's value. This substantiates the observation that the small/informal sectors only purchase catastrophe insurance to the extent of lending institutions' requirements (and these can be limited to the amount of outstanding loan principal balances). It is estimated that between 25% and 40% of dwelling stock is uninsured with the small/indigenous segment being the least insured. Furthermore, insurance premiums are higher (less affordable) in areas such as the OECS Leeward islands which suffer frequent storm events. Medium and large dwelling owners, without the same affordability constraints almost universally carry catastrophe insurance, as is the case with medium and large business property owners. In the case of the latter, business interruption insurance (stoppage as a consequence of catastrophe perils), is seldom purchased although this could mitigate employers' loss of income and employees' loss of pay while a workplace was idle. Larger businesses especially have access to insurance brokers (as opposed to insurance companies' own agents), who are adept at placing covers with 'foreign' insurance companies, i.e., operating without local registration, characteristically from the U.S. or Europe. Such arrangements are especially prevalent in those Caribbean nations where foreign currency is readily accessible for premium remittances. - 46 - Given that small property owner segments in the less advantaged sectors do not partake of insurance coverage, it is interesting to note that the Barbados government recently created the Poverty Alleviation Fund under the responsibility of the Minister of Social Transformation. In concept this could perhaps serve as an agency for simple building structure vulnerability reduction measures being implemented (possibly with hands-on assistance from Defense Force members). To some small dwelling and business property owners, traditional insurance is a little understood or respected mechanism. For these sectors, a non-traditional approach via the property tax mechanism could offer a worthwhile level of pre-event risk funding. Simultaneously, governments as a matter of ongoing policy should undertake public education programs and publicity on risk management and the prudent use of insurance for protection. A modest property tax surcharge could fund a flat amount for storm damage essential repair costs (possibly geared to property square footage). Access to the program could be conditioned on property owners' compliance with elemental storm protection measures (e.g., roof straps), the materials for which could be provided for free. A side effect would be that the arrangement, being independent of the catastrophe insurance market mechanism, would not disturb the available market levels of insurance or reinsurance capacity. As illustrated in the subsequent sections of this report, a sub- regional insurance pool might accumulate surplus capital out of which periodic limited 'dividends' could be paid out to support complementary financing for repair costs in the above mentioned sectors which are generally 'uninsurable' from the industry's standpoint. This concept would appear to very directly address, at modest cost, the goals of actualizing vulnerability reduction measures and providing pre-event funding for essential repairs. Accomplishments towards these combined goals would diminish both the direct and indirect socioeconomic consequences of storm events. Such a program could also be enabled by a stand-by credit arrangement from an international funding agency as already exists for the OECS countries at the country levello. Competitiveness of the Industry Most of the indigenous/local insurance companies originated some 30 years ago, hitherto having operated as general agencies or branches of foreign (mainly U.K.) insurers. For the most part, local companies continue to day as insurance units of trading and financial companies whose profit strategies center on their role as agents for a wide range of products and services. Three primary classes of insurer transact catastrophe insurance (as opposed to reinsurance) in the Caribbean: 1. Locally Registered Insurance Companies regulated by the insurance supervisors of each nation. 2. Regional Insurance Companies registered in each home country but also conducting business in other Caribbean countries in which they operate. 10 OECS Emergency Recovery and Disaster Management Programn, funded by IBRD/IDA to undertake mitigation investments, post emergency reconstruction investments, and institutional disaster preparedness. -47 - 3. 'Foreign' Insurance Companies - so called because they are regulated and based outside the Caribbean, mainly in Europe and U.S.A. It is estimated that the first two have the large preponderance of small and mid- sized dwelling and commercial business; while the latter has a significant and growing share of the larger commercial risks segment. Of the 'foreign' companies, several are exempted from most local regulation and taxation. Bermuda and Barbados are the largest centers for exempt companies. World Trade Organization developments and CARICOM trade liberalization generally are progressively blurring the original demarcations between the three primary categories of insurer. Today, foreign companies have little practical difficulty in accessing the business they want in nations with reasonable access to strong currencies. They can either issue a 'foreign' policy (e.g. from the US, Canada, or UK), or they can make a 'fronting' arrangement with a local company under which almost all the risk bearing is reinsured back. The selection of method employed generally depends on the relative ease and cost of purchasing foreign currency for premium payment. Foreign companies are adept at merging Caribbean risks into multinational package programs especially for Caribbean subsidiary operations of multi-national corporations. Foreign companies tailor-make policy coverage and insuring large structures often using insurance risk management techniques seldom encountered in local Caribbean markets. The larger local companies realize the need to compete on technique as well as price but can be deterred from such initiatives because of the relative rigidity of their reinsurance programs which are primarily geared to run-of-the mill heterogeneous small and mid-sized risks. Most local companies' strategies are primarily geared to generating the normally higher margin reinsurance commission levels available from their conventional insuring practices. Foreign insurers, on the other hand, primarily seek profit from their individual risk underwriting expertise acknowledging that expense elements of the premium need to always be kept competitively low. The practice of insurance in the Caribbean is advanced in comparison to most emerging economies and this most likely reflects the recognized exposure to natural hazards as well as the influx of the tourism industry which demands insurance services to protect its capital investments. However, the local insurance industry, while effective in transferring risks, requires needed improvements in risk underwriting and the ensuing ability to discriminate between risk ratings of property risks to encourage pre-insurance loss mitigation measures. These will, in the long run reduce the costs of insurance for both recipients and suppliers. The next chapter discusses methods whereby local insurers, policyholders, property owners and governments can significantly lower the overall price of insurance, and therefore increase competitiveness through physical vulnerability reduction measures. Such actions should represent a basic underlying strategy for reducing risk in disaster prone regions. - 48 - - 49 - III. MITIGATION, SELF INSURANCE &VULNERABILITY REDUCTION MEASURES The natural risk exposure of Caribbean economies implies that even while exploiting the insurance mechanism in the most cost effective manner, this will still result in higher relative premiums compared to other regions in the world. Besides the advantages of risk pooling (discussed in chapter VI) which allows diversification of risk and better leverages the available capital, another major source of both premium and risk reduction, consists of physical measures to reduce the structural vulnerability of properties and critical assets. This chapter shows how very modest investments in mitigation such as roof straps to protect against hurricane force winds can reduce the loss exposure substantially, sometimes by as much as 50%. Such measures need to be accompanied by regulatory enforcement of building codes as well as insurance-based incentives which can be achieved through the standardization of data on construction code ratings and property risk characteristics. These resources can also be harnessed to develop publicly sponsored quasi insurance schemes for vulnerable low income communities, in order to instill a framework and awareness regarding the costs and benefits of both physical and financial risk management tools. - 50 - III. MITIGATION, SELF INSURANCE &VULNERABILITY REDUCTION MEASURES While the insurance mechanism can be considered a strategic financial instrument for transferring risks which are too intense to bear by individual property holders, both public and private sector, the insurance tool should be used only once all other 'manageable' mitigation measures have been exhausted. As confirmed in the section below, the manageable part of the risk, which can be reduced via physical measures has a great potential for reducing the underlying structural risk of physical assets, with the ensuing effect of dramatically reducing the potential 'loss value' of properties at risk. This in turn has the benefits of commanding lower premiums for such exposures for whose residual non-manageable risks should be transferred via the insurance market. The specifics of such measures are discussed in this section, as well as the hazard information requirements needed to establish the underlying date for classifying insurance risks properly. In addition, these concepts are also applied to managing risks for low income communities where a policy of explicit but limited government insurance can be coupled with mitigation activities that begin 'transferring' such risks away from the public sector. The mitigation of catastrophe loss (direct loss, ensuing reconstruction debt, and other adverse consequences) includes the following strategies which can reduce structural and financial vulnerability under pooled or non-pooled insurance arrangements, alike: a. Vulnerability Reduction Physical Measures b. Use of Building Codes and Other Regulatory Measures. c. Generation and Availability of Information on Hazards, Vulnerability and Risk. d The Contributions of the Insurance Industry. Vulnerability reduction measures are broadly accepted as the highest-impact mechanisms to reduce catastrophe loss as they operate before the event to neutralize the impact of physical forces. The insurance mechanism, on the other hand, is limited to providing monetary compensation after loss has been incurred. Furtherrnore, only part of the insurance premium dollar is available to compensate losses as over half is customarily consumed by operating expenses and taxes. The dollar expended on vulnerability reduction (e.g. roof strap retrofitting) retains a very lasting physical protection value. Empirical data showing vulnerability reduction cost/benefit yield potential is not maintained. However, regional civil engineering experts estimate that on average, expending 1% of a structure's value on vulnerability reduction measures can characteristically reduce the probable maximum loss (PML)" from windstorm (CAT III, 120 mph+) by at least a third. For example a $1 00,000 dwelling's pre-retrofitting PML of $10,000 would become $3,000 after retrofitting. Such retrofitting is customarily 1 l PML differs from EPL in that PML reflects an estimated 'maximum' loss from an event with a specified intensity and estimated probability, and which is used as a 'reference' benchmark to calculate insurance capital requirements and potential losses within a reasonable time frame. EPL on the other hand, reflects the mean expected loss over a range of events and frequencies. In this regard, PML can be seen as identifying a 'tail' on the loss distribution function which is considered the reference point at which insurers and property owners may consider the need for undertaking preventive risk management actions and holding sufficient capital reserves to cover such an eventuality. -51 - focused on roofs, openings and claddings and often capable of effective installation by a property owner. Similarly, it was reported from Trinidad that relatively simple measures (tying walls to foundations, walls to walls, walls to roof supports, and roof supports to roof sheeting), provide effective structural cohesion to withstand high winds and flash flooding. Past hurricanes illustrate all too well the cost of indifference to vulnerability reduction measures. Nearly a third of the 1988 Jamaica hurricane Gilbert direct losses (totaling $956 million) were from dwellings primarily caused by the loss of roofs; most of which would have been prevented by effective and inexpensive roof strapping. Hurricane Andrew in 1992 severely ravaged dwellings in Florida as a result of construction code violations, shoddy workmanship, and inadequate inspection enforcement. International aid and development funding agencies, besides sharing consternation at delays, disruptions, and increased costs, have the strong view that wisely planned hazard and vulnerability reduction efforts and funding before a catastrophe pay excellent dividends in reducing economic impacts. Mitigation expenditures are a very small fraction of the funds spent on reconstruction in the aftermath of catastrophes. The use of building codes and other regulatory measures capable of encouraging mitigation can be divided into two categories: Non-Structural: Identification of hazard-prone areas and limitations on their use Land-use allocation and control Incentives Structural: Use of building codes and materials specifications Retrofitting existing structures Use of protective devices. Non-Structural /Regulatory/Risk Rating Measures Non-structural measures in the Caribbean are in use but on a very limited basis. Hazard mapping, although resource and time intensive, has to be considered the fundamental underpinning of any meaningful strategy for catastrophe mitigation. Hazard mapping in particular has critical application to a rational catastrophe insurance market mechanism and hence to rational reinsurance pooling arrangements. To set the base for providing reliable planning inforrnation, governments of the Caribbean should consider as a priority, the mapping of hazard locations (and appropriate zoning regulations) throughout each country, and the inventorying of all properties and physical assets (including building code classifications) to be reflected on such maps and information databases. Such measures not only serve for hazard planning purposes, but also provide the actuarial basis for localized insurance pricing of assets in all zones. - 52 - The establishment of comprehensive 'risk data' information bases, also permits the insurance industry to participate in loss mitigation incentives by providing tools to better understand and price local risks. Such information allows taking into account detailed characteristics such as building code ratings when determining potential loss costs to be used in rating insurance coverage for individual properties. This results not only in fairer prices for insurance but also provides an economic incentive for communities to strengthen their building codes and enforcement. Insurers can take a number of steps to help promote investment in vulnerability mitigation measures and to reduce potential loss profiles on underwritten risks. For this purpose, industry service functions could be established under a specialized industry sponsored office (similar to the Insurance Services Office in the U.S.), in order to assign 'grades' to communities based on the quality of their building codes and their enforcement. For this, a grading schedule reflecting building code enforcement can be established for qualifying communities which may be eligible for premium discounts. Under such a program, grades would be assigned to local communities based on structural factors as well as public protection services (e.g.: disaster preparedness, fire fighting capability, etc.) available. Loss cost estimation and associated premium levels should, in addition to the above factors, be also based on the characteristics of individual properties including location and risk exposure in accordance with the hazard mapping informnation discussed above. Structural Measures Structural measures have also made but limited headway in the Caribbean. The Caribbean Unified Building Code (CUBIC) has now progressed into useable codes in several nations but, as was recently observed, these improved codes lack any effective enforcement practices. Limited progress on retrofitting can be attributed primarily to the lack of incentives and concerted leadership in the promotion of benefit features and practices. A fee based approach which might result in more effective enforcement actions would be to set up a Bureau de Controle type firm at the regional level, with the responsibility to inspect building during the construction cycle. While political decisions would need to be made to permit such an entity to operate as a regional private concern with fees charged to either property owners or their financiers, the benefits would be direct given the economic incentives for such a firm to ensure that code compliance was undertaken. Such entities could also enter into cost sharing arrangements with banks and insurance companies as well as property owners, to share the costs more broadly. For structural measures, information exists on demonstrably worthwhile mitigation technologies. However, a major exception relates to catastrophe event return period probabilities which are essential in determining the cost/benefit trade offs of different types of mitigation actions. For this, historical patterns are limited to a few centuries; also changing and/or unmeasured patterns (e.g., Greenhouse effect and El Niflo) exacerbate the challenges of forecasting frequency probabilities. However, even the extensive information available is too seldom articulated in layman language of practical use by those who should derive the most implementation benefits. These include development decision-makers as well as professionals and - 53 - artisans in the construction (and retrofitting) fields. Recommended reading includes: Managing Natural Hazards to Reduce Loss (OAS, Disasters, Planning and Development: Managing Natural Hazards to Reduce Loss, 1990). Strategies Linked to Insurance Usage For non-structural technologies, clear and durable on-the-ground markers could disseminate valuable information. Such could display floodwater and wave surge expected levels. Structural information techniques could include sample skeleton structures (perhaps mobile) which would display practical retrofitting installations. More than anything however, is the need for government, civic and private, (including insurance companies) to grasp initiatives for mitigation implementation. There is no empirical data on self-insurance usage. Public utilities (CARILEC & Barbados L&P) have such programs which include stand-by credit mechanisms to provide liquidity during build-up of the self-insurance funding amounts. An argument can be made for tax deductibility of such dedicated funding, as the considerations are analogous to the tax deductibility allowed insurance companies for essentially the same purposes. So-called 'Captive' insurance company arrangements contain self-insurance features. Here industry affinity groups, e.g., Caribbean Hotels Association, enter into insurance purchasing combinations. These can include the joint capitalization of an offshore captive insurance company, which could have the function of carrying the 'claims deductible' exposure of shareholder property owners in the manner described earlier. This is an example of risk and reinsurance pooling. In the wider socioeconomic context, the broad-brush practices used for risk rating, serve to discourage building structure vulnerability reduction measures. Outside the Caribbean such measures are very prevalent and of proven effectiveness to mitigate catastrophe destruction and disruption. Insurance companies, with full concurrence of their reinsurers, offer significant premium rate differentials conditioned on the specific location and physical risk characteristics of properties insured. Empirical data showing the vulnerability reduction (VR) cost/benefit yield potential is not standardized. However, a respected ESC civil engineering firm (CEP Engineering Ltd.) compiled the following table to summarize estimates of the VR measure expenditures required to reduce by 50% the Probable Maximum Loss (PML) from a Category III (120 mph+) hurricane: - 54 - Characteristic Expenditure on VR Measures Required to Reduce CAT III Hurricane PML by 50 % Expressed as % of Overall Building Value New Construction Retrofittink Construction Categories: Dwellings A. Reinforced concrete & masonry 1.2 1.8 B. Lightweight roof & masonry walls 1.5 2.2 C. All lightweight construction 2.2 3.4 Commercial/ Industrial/Public A. Reinforced concrete & masonry 1.3 2.0 B. Lightweight roof & masonry walls 1.7 2.8 C. Steel structure & lightweight cladding 2.8 3.8 As an example, a $ 1,000,000 value Class B commercial structure, without deliberate VR measures, could typically be assessed at a 10% ($100,000) PML for a Category III hurricane. Estimates suggest that VR measures costing $17,000 (new) or $28,000 (retrofitted) would reduce the PML by 50% to $50,000. Hurricane Vulnerability Reduction (VR) Measuresfor Structures Cost/Benefit Yield Estimates The methodology for determining VR cost estimates to yield a 50% reduction in Probable Maximum Loss (PML) from a Category III hurricane (Saffir-Simpson scale, above 120 mph wind speed) is illustrated below. This category storm is generally recognized by insurance markets for damage exposure estimations. The approach used'2 was first, to express the Category III maximum wind speed as one minute pounds per square foot pressure. As the mean value, the figure of 2.25% was employed, representing the estimate of the incremental VR measures' cost for a new dwelling structure. This figure was deduced from 20 years' of professional experience on both design and retrofitting consulting as well as many on-the-spot surveys of post- hurricane damage scenarios. Original estimates were compared and adjusted using other information sources including Munich Re's published (worldwide) materials. In particular findings showed: 12 Source: CEP Engineering (T. Gibbs, Barbados). - 55 - 1. Not unexpectedly VR retrofitting proves more expensive than VR measures implemented at the time of original construction. 2. Caribbean industrial purpose structures, as opposed to regular commercial structures, customarily have the least storm resistive designs and materials. 3. From insurers' observations, storm damage to contents characteristically runs twice the damage cost to the respective structures. Following that, the 2.25% mean was scaled to VR cost estimates for the several construction and occupancy use classes. The methodology and judgement incorporates reasonable confidence on the potential of VR measures, at the indicated cost levels, being able to reduce hurricane damage (and PMLs) by 50%. This confidence is founded on post-hurricane surveying and information from clients who have adopted similar VR measure recommendations. It should be made clear that estimates shown in the table are necessarily broad averages for the classes shown. Individual structures have individual structural, exposure, and location characteristics which can vary their VR performance from the averages displayed. Illustration of Impact of Hurricane Vulnerability Reduction (VR) Measures on the Risk Premium : Risk Amount Ratio Assumptions: 1. Building Structure Hurricane PML: Before VR retrofitting 10 % After VR retrofitting 5 % 2. Policy Claim deductible is 2% x sum insured 3. Sum Insured is $ 500,000 4. Premium rates (%) are: Full rate Catastrophe (CAT) Element Without VR measures 1.00 .60 With VR measures 0.75 .45 Risk Premium : Risk Amount Ratios Before VR Measures After VR Measures $ $ Sum Insured 500,000 500,000 PML % 10% 5% PML $ 50,000 25,000 Claim Deductible 2 % 10,000 10,000 Net Insurance Liability 40,000 15,000 - 56 - CAT premium rate % 0.60% 0.45% CAT Premium $ 3,000 2,250 Ratios: a. Risk Premium/Risk Amount 7.50 % 15.0 % (3,000/40,000) (2,250/15,000) b. l/a 13.3 6.7 The illustration shows that implementation of VR measures, after a CAT premium rate reduction of 25 % (0.60 to 0.45), has served to decrease the Risk / Premium ratio ('b') from 13.3 to 6.7 -- an improvement of almost 50%. Clearly, the above measures, if disseminated and implemented across the capital stock assets of Caribbean countries, would not only dramatically reduce post disaster economic strain on the government and on private individuals, but would also permit much broader usage of the insurance mechanism as a less costly policy and risk management tool. While the analysis of pooling and risk transfer in the subsequent chapters is based on existing premiums based on current but limited VR practices, the incorporation of such VR practices on a broad scale could potentially halve the price of insurance protection and thus broaden its usage on a much wider scale. One sector, however, which requires special attention and which to-date is not equipped to afford insurance or be insurable, are the low income communities which remain the responsibility of regional governments in terms of disaster assistance and recovery. Developing Insurance for Low Income Communities One of the primary challenges for governments in disaster prone regions, is the protection of low income communities and the development of incentives for community participation in risk management measures. The properties, particularly the housing stock of low income communities is generally fragile, and this fact makes them difficult to insure even if affordability was not an issue. However, affordability is also an issue, therefore, governments have the dual challenge of (i) promoting structural measures to reduce the vulnerability of low income housing assets and (ii) improve economic welfare of such communities to begin allowing a phased in process of risk management using the insurance mechanism. Due to the above factors, it is imperative that governments in disaster prone countries such as the Caribbean, exploit to the fullest extent the range of tools available for hazard mitigation in order to reduce exposures of low income dwellings, while at the same time making transparent what the government risk liability consists of and what risk liabilities should be borne by the communities themselves. This sort of strategy calls for a two-track approach whereby public funds for 'mitigation works' are made available to these communities in exchange for an explicit public insurance policy which is limited in its coverage but which allows low income communities to be adequately protected if they follow the appropriate practices of vulnerability reduction. Non-participating individuals or communities, while not completely exempt from aid in the event of a major catastrophe, would nevertheless receive second priority than those homeowners who had taken pro-active measures at mitigation and who thus would be explicitly covered for insurance purposes and up to a specified limit, by the government. Such a system would engender inter-community competition in vulnerability reduction measures and at the - 57 - same time raise the level of awareness regarding the nature of insurance policies and risk sharing. It should be noted that the implementation of a public insurance scheme such as the one outlined below, is a medium to long term endeavor which requires a process of not only self management of risks, but also an educational process to ensure that affected communities are well versed in the risk framework (both physical and regulatory) in which they operate. Initial financing for the required mitigation measures under such a scheme can be provided via public and international donor funds, as well as through potential savings generated through insurance pooling mechanisms using alternative risk transfer instruments as described in chapter V. In the latter case, an explicit 'social compact' with the insurance industry would be required for such contributions to be viable, although such a compact would also help to promote the practice and understanding of insurance in low income communities for eventual usage; income levels and structural standards permitting. Below is diagrammed one potential financial structure for a public insurance fund which would combine the aspects of mitigation with education and promotion of insurance. Outline of a Public Insurance Scheme for the Protection of Low Income Communities Public Insurance Schemes for Lower Income Communities (4- - *l Notional Mitigation Market Entry Insurance Rating Program Accounts Program l (Gov't. collect (Premium paid w/coverage premium or by Government. ______ -adiustment insurance co. Cotkigw Dfined Limited Value Coverage (e.g.: 60%)| Dwellngs Unsabe Communities '~ ~~~~~~~~~~~~~~~~~~~~~~~ > The above scheme relies on a three phase approach which begins with the essential element of mitigation and incentives for vulnerability reduction. As mentioned earlier, the financing for such a phase will likely require public contributions in terms of funding for materials and in some cases skilled technicians. However, the labor cost, for - 58 - the purposes of risk sharing, would be borne by the communities themselves, as this would be necessary to assure buy-in of the program with the subsequent incentives for obtaining explicit government insurance protection. The government insurance program would attempt to explicitly define the levels of loss covered by the public sector in the event of a major catastrophe. In this regard, it can be compared to central banking/deposit insurance policy in the financial sector whereby moral hazard is reduced by having the government provide some insurance (in that case to depositors) while limiting the upside loss amounts. In the case of catastrophe insurance, the same principle would apply, i.e., following the verification of structural mitigation measures undertaken by such communities (which would reduce asset risk exposure), the govemrnent would provide explicit coverage in an amount that for example, could reach 40%-60% of the property value if this was lost on account of a natural disaster. To support the seriousness of the policy, the government would issue insurance policies to all such homeowners or dwellers showing the level of coverage and the notional actuarially fair premium to be fully subsidized by the government at the outset. The initial notional premium would therefore constitute no charge to the policy holder, however, the government would make it explicit from an accounting standpoint in order to disseminate the nature and cost of such an insurance policy. The government would also specify in the policy statement what the structural rating grade of the property was since that would provide an input into the notional premium price paid by the government. Of course the government would not be actually paying a premium unless it deemed it prudent to build up reserves for such an event (an option which could be viable in some jurisdictions), but it would hold the liability to pay for damages in the event of a disaster. The rating category shown on the policy holder's statement would also be reviewed periodically based on progress in the mitigation programs undertaken, and adjusted accordingly. For those properties undertaking significant mitigation measures, the government could safely increase the coverage for the same notional premium (or reduce the premium for the same coverage), an adjustment which would be budget neutral from its standpoint, but which would provide incentive signals to low income dwellers to continue undertaking vulnerability reduction measures if they wish to partake of increased government insurance protection. Eventually, as income levels increased in such communities or parts thereof, the government could begin phasing in cost recoveries on its notional premiums so that property owners themselves began paying for some of their insurance. At such a stage, property owners would also have the option to use insurance from the private markets if deemed to be on more favorable terms. A key question which might be raised about such a scheme relates to the possible moral hazards, that is, if a given property holder decided to do nothing at all, wouldn't he/she naturally count on government aid anyway, in the event of a natural disaster? The response to this is yes, but a very qualified yes. In other words, the government would not be able to completely ignore a low income property dweller following a disaster, simply because he/she had not signed up for the mitigation program and the public insurance policy. However, the government could prioritize post-disaster reconstruction and give preference to those communities who had undertaken pro-active measures and whose 'insurance policies' would therefore be honored in a timely manner. This measure -59- in itself, if made explicit by the government could encourage inter-community competition to sign up for the mitigation/insurance program in order to receive funding first following a disaster. While this system does not solve all of the issues of free riders and moral hazards, it does initiate a process of education about risk management and insurance, which in the long run could provide its own societal incentives for communities to reduce their passive reliance on government help, particularly if such help and its timeliness will be prioritized according to risk sharing measures undertaken a priori by other communities. Mitigation and Insurance as Simultaneous and Sequential Strategies This chapter has discussed the powerful tool of mitigation as a core strategy in a comprehensive risk management program. It has shown how mitigation measures and insurance policies go hand in hand as two sides of the same coin in the process of risk management. Nevertheless, once all mitigation measures have been exhausted, countries exposed to natural disasters still require mechanisms to "hedge" the remaining risks which constitute economic or financial risks which would be difficult to avoid even by undertaking the most comprehensive vulnerability reduction measures. Such hedging mechanisms which are embodied in the established insurance practice can be structured in different configurations for the purpose of managing and transferring risk. The following chapter describes how the insurance sector in the Caribbean has structured its insurance arrangements with the international market for this purpose. It examines the implications of these arrangements in terms of managing the financial impacts of potential natural catastrophes, including the coverage benefits and premium costs of various strategies. - 60 - - 61 - IV. FINANCIAL STRUCTURE OF REINSURANCE CONTRACTS IN THE ESC The transfer of riskfor potential catastrophic liabilities constitutes a keyfinancial strategy in the economic management of disaster prone countries. Risk transfer in the insurance sector generally takes the form ofreinsurance contracts taken by local insurers with international (re)insurers who can better absorb large risks. This chapter examines the current practice and structure of reinsurance arrangements used in the Caribbean and the alternative approaches usedfor structuring reinsurance contracts. In particular the modalities ofproportional reinsurance treaties versus excess-of-loss treaties or their combination are explored to show the advantages and limitations of each. Such financial arrangements set the basis for examining the range of alternative risk transfer instruments which can be deployed under pooled catastrophe funds to manage large loss risks effectively. The structure of these contracts point to the development offuture strategies which can deploy a more optimal balance between capitalfor domestic risk retention and capitalforfinancing the cost of risk transfer. - 62 - IV. FINANCIAL STRUCTURE OF REINSURANCE CONTRACTS IN THE ESC This section starts by setting out the components of 'proportional' or 'quota share ' reinsurance arrangements. These have by far the most prevalent use by Caribbean insurance companies for their catastrophe reinsurance purposes. Also displayed is a simplified income statement using the described components. There follows a discussion of Excess of Loss (XL) reinsurance with an explanation of why this mechanism, in wide use elsewhere, is not favored in the Caribbean. The section continues with a discussion of some fundamentals of reinsurance portfolio risk management in the context of Caribbean catastrophe insurance. Sample Caribbean Insurance Company's Property Insurance Portfolio and Reinsurance Arrangements Assumptions: $ Million 1. Insurer's total property (Fire etc. & Catastrophe perils) policy sums insured, i.e., total gross exposure liabilities. 100.00 2. Hurricane peril Estimated Expected Loss (EPL) assessed at average 20% x sums insured = Gross EPL 20.00 3. Average policy claims deductible is 2% times sums insured 2.00 4. Net of deductible aggregate portfolio catastrophe Gross EPL 18.00 5. Average all peril policy premium rate is 0.8% times sums insured - Producing portfolio premiums (Original Gross Premium-OGP) 0.80 6. Gross premium rate (# 5) is comprised of: Non-catastrophe element (Fire perils, etc.) 0.32% Catastrophe peril element 0.48% Total 0.80% 7. Company's direct (before reinsurance) administrative & acquisition expense ratio for property insurance is 30% times OGP (# 5). 8. Reinsurers allow reinsurance commission rates of: On non-catastrophe premiums ceded 32.5 % On catastrophe premiums ceded 25.0 % Average reinsurance commission rate 28.0 % 9. Proportional Reinsurance (i.e. not Excess of Loss) across whole portfolio is structured as 80% reinsured and 20% retained. 10. Respective Catastrophe Liabilities of primary insurer and reinsurers are therefore allocated as: - 63 - Primary Insurer Reinsurers Total $ million Gross Sums Insured 20.00 80.00 100.00 Catastrophe aggregate EPL (net of deductibles) 3.60 14.40 18.00 $ 11. Gross Claims are: Non-Catastrophe $ 250,000 Catastrophe - Gross $ 500,000 Less deductibles $ 100,000 = $ 400,000 $ 650,000 12. Investment Income is assumed at 10% of net premium + reinsurance commissions + carry over liquid investment balances = $44,000 Simplified Income Statement for Primary (Reinsured) Insurance Company Property Portfolio Income Expense Net Balance Assumption Entries I Gross Premiums Received 800,000 7 Direct Expenses 240,000 9 80% Reinsurance Premium 640,000 8 28% Reins. Commission 179,200 12 Investment Income 44,000 143,200 11 Incurred Eligible Claims 650,000 9 Claims recoveries from reinsurers (80%) 520,000 13,200 The above example displays the fundamental financial parameters in most common use in the Caribbean for risk pooling utilizing the proportional (quota share) reinsurance mechanism. XL (Excess of Loss) Reinsurance employed in Conjunction with Proportional Reinsurance Caribbean insurers with proportional reinsurance contracts customarily purchase an 'inner' layer of XL reinsurance to protect their catastrophe peril only net retention. This XL reinsurance covers approximately the upper 20% layer of the retained (co- insured) risk under the main proportional reinsurance contract. Reinsurance costs for - 64 - proportional covers are essentially a function of the gross premium rates and the reinsurance commission rates allowed as related to the direct administrative costs (assumption # 7 in the example) of the insurance company. For the sake of simplicity, the inner XL catastrophe reinsurance cover is not shown in the above income statement example. An insurance company, to select the threshold point for an inner XL cover, will primarily consider its net worth strength as well as the additional cost of the XL cover. As an example using the sample shown above, the company had (net of reinsurance) property insurance premiums of $160,000 (20% of gross), which could suggest the company having notional capital for allocation to property business of some $40,000 (a 4:1 premium/capital ratio). The company's remaining capital and any free reserves (retained profits, etc.) could also be considered. Conceivably this company could have selected a $120,000 threshold, above which, it would utilize an XL catastrophe reinsurance cover which would thus allow ceding 25% of the retained risk. In the above example, a $120,000 threshold retention would have allowed invoking $10,000 in excess of loss cover (gross claims of $650,000 times 20% = $130,000 retained claims, or $10,000 above the XL reinsurance attachment point). XL Reinsurance Employed as a Principal Mechanism for Catastrophe Reinsurance The Excess of Loss (XL) reinsurance mechanism is widely used outside the Caribbean (by relatively larger insurers) for their catastrophe reinsurance needs. Under XL structures the reinsurers' liabilities are triggered above a given retention amount and customarily for 100% of aggregated claims above the retention amount. The XL reinsurance price is normally expressed as a 'rate on line' signifying the dollar reinsurance premium as a percentage of reinsured loss limit. This rate on line is assessed by judging the extent to which a reinsured portfolio's aggregate EPL would penetrate the reinsured layer's limits. For example, a reinsurance rate on line of 30% ($300,000 per million of limit) would suggest a fully penetrated (exposed) layer and an approximate 3- year notional payback period; i.e., a high frequency and severity probability to reinsurers. As the reinsured limit layers get higher and further removed from anticipated full exposure penetration, the rates on line diminish and the pay back periods increase. Reinsurance commission is not a feature of XL covers. The accurate and reasonable pricing of XL reinsurance calls for meticulous assessment of a primary portfolio's aggregate catastrophe EPL liabilities. Short of having full information a reinsurer will price to cover the 'doubt' factor and resulting quotations will often deter prospective primary company customers. Primary companies can on occasion be dazzled by the front end cash flow benefits of XL reinsurance without prudently assessing the potential back end liabilities represented by having full liability for the retention limit - a liability actualized by a catastrophe event. XL reinsurance is generally considered as more of a 'bet' than proportional reinsurance under which the respective claims liabilities between reinsured and reinsurer are clearly apportioned from the lowest claims level to the highest. Thinly capitalized primary companies as seen in the Caribbean are understandably averse to full XL reinsurance. Second Event Reinsurance (Reinstatement Contracts). This is primarily a feature of XL reinsurance contracts and refers to the basic premise that the reinsured - 65 - limits are available once during a single contract period (normally annual). Once reinsured limits are 'consumed' by a catastrophe event or other claims, the limits require reinstating for 'second (or subsequent) event' via payment of additional reinsurance premium. Most primary companies in the ESC purchase second event coverage reinstatement as included in their original XL contract negotiations as this is generally both cheaper and more likely to be available than having to face reinsurers' quotations in the aftermath of a major claims event. Customarily, 'second event' protection is not an issue under proportional reinsurance contracts where reinsurers contract to reinsure a portfolio's claims for the full contract period. Reinsurance Portfolio Risk Management Considerations This report has discussed the proportional catastrophe reinsurance mechanism from the standpoint of the primary insurer. From the standpoint of the reinsurer, the parameters are essentially the converse with the reinsurer entering reinsurance premiums net of commissions, any claims incurred, and any income derivable from invested funds - primarily unexpired premium reserves or reserves for reported, but yet to be paid, claims. However, having a portfolio of reinsurance contracts allows the reinsurer to adopt risk management practices analogous to those practiced by banks on loan portfolios. To obtain an optimum match between risk and return, the reinsurer can reinsure further ('retrocede') to other reinsurers (perhaps with reciprocity), as well as consider securitization to contain liability. However, throughout the chain of risk bearing (from policy issuing insurer, to reinsurer, to retrocession carrier, to securitization), the key parameter is the relationship, or the ratio of Risk Premium to Risk Amount. In the context of catastrophe insurance, the value of 'Risk Amount' is comprised of: 1. Severity element - the Probable Maximum Loss (PML) and 2. Frequency Element - the probability of a catastrophe event. For the severity element, proven technologies exist to obtain reasonably accurate quantification. These require detailed hazard mapping and awareness of distinctive building structure methods and materials. The frequency element assessment has yet to attain a similar stature of reliable precision. Actuarial approaches to catastrophe event return periods can be reasonably viewed as volatile for reinsurers' short term practical purposes. Furthermore, the 'Greenhouse' and El Nifno theories suggest that historic natural disaster frequency patterns are unlikely to be repeated. Thus the forecasting of catastrophe frequencies has yet to mature to a reliable level in calculable risk probability terms. These factors -severity and frequency - are those which underpin the catastrophe reinsurance markets' cycles. In effect, Risk Premium levels prove almost invariably either too high or too low, and only fortuitously are rarely in equilibrium to the Risk Amount exposure. In the Caribbean, the existing severity (PML) assessment techniques are very broad brush. Characteristically, a portfolio's aggregate PML is established by reinsurers' view of hurricane storm tracks and impact potentials. Wide geographic areas containing - 66 - several nations are customarily swept into a single PML category without regard for the topographical features and structure resistance distinctions propounded by regional and international experts. Furthermore, Caribbean insurers' catastrophe premium rates are also very broad brush without discrimination to reflect topographical hazard and structure resistance distinctions. It can therefore be logically held that the poorer risk quality properties are subsidized in premium rate terms by the better risk quality properties. These broad brush reinsurance and policy rating practices may have produced administrative simplicity but cannot claim effectiveness as regards accuracy in assessing the Risk Premium: Risk Amount ratio. Capital market alternatives, whether they be in the form of securities such as bonds or in the form of loans, can potentially reduce or at least stabilize the prices of coverage. The reasons are twofold: the capital markets comprise approximately $42 trillion in assets currently compared with insurance industry capital estimated at $0.9 trillion, and second, investors have shown portfolio preferences in purchasing catastrophe bonds given that their yields are both higher than the market and uncorrelated with global financial market movement, thus providing a diversifying hedge. Although such bonds have 'default' characteristics in the sense that if a disaster strikes, the bonds are liable to lose interest and principal, the probability of such events is generally lower than similarly rated sovereign bonds which have higher probabilities of country default risk. Before considering in the next chapter, the cost effectiveness in the pricing of alternative risk transfer instruments such as catastrophe bonds or contingent credit lines, an illustration of the catastrophe insurance coverages and their structures is shown: Combined Proportional (Quota Share) Treaty with Catastrophe Excess of Loss (XL) Reinsurance Structure Insured/retained by primary insurer 4Exhaustion Point (e.g. $200 inn.) Propor- Reinsured layer tional sharing by minsurer u (eg. 30%) ._____ ____ _ Attachment Point (e.g.: $50 mn.) Insured/retained by primary insurer The above example shows a reinsurance structure which combines the elements of proportional treaty coverage (used widely in the ESC) and excess of loss (XL) covers. - 67 - As mentioned above, local insurance companies in the ESC traditionally use proportional treaties (also known as quota share treaties) to share risks with reinsurers. The above diagram implies that after the initial retention of risk by the primary local insurer at the lower levels, the primary insurer then cedes about 70% of the coverage (and premiums) to the international reinsurer. For any claim based on damage, the insurer would thus pay 30% and the reinsurer 70% (past the initial retention level and before the exhaustion point). Generally in the East Caribbean, the proportional treaties do not have upper limits as in the above diagram, although more recently proportional reinsurers have begun limiting their upper layer liability. As discussed above, local companies in the ESC, have begun utilizing excess of loss (XL) cover on what they retain, i.e., on the 30% portion in this case. As a matter of practice, however, XL cover as a form of insurance/reinsurance is not based on proportion of total claims submitted but rather on a specified level of quantified losses (independent of the claim amount). The XL reinsurance cover begins paying after the 'attachment point' and stops paying if losses exceed the 'exhaustion point' or cumulative limit. The above structure combines both methods in that claims are paid by the reinsurer based on attachment and exhaustion limits, though within those limits, the primary reinsurer also shares a small proportion of the risks, or what is also known as 'coinsurance'. These structures provide strong incentives for all parties to be concerned about loss reduction measures and to avoid adverse selection problems. However, as is discussed later, the ESC catastrophe reinsurance structure is different from the above model in that an even greater proportion of risk is ceded or transferred (see diagram below). While this makes sense from a macroeconomic perspective, it can lead to disincentives for loss reduction measures if very little risk is retained/managed, and because of the high dependence on reinsurance, it leads to more potential volatility in the pricing of catastrophe insurance which is subject to the global market supply availability. Typical Caribbean Reinsurance Treaty Structure Excess of loss reinsur. Reinsurance contract: Proportional/quota Initial co- treaty (70% reinsured / insured co-insured). 30% retained, before XL Policy Holder Deductible The determination of whether pooled structures coupled with alternative risk transfer instruments can be more cost effective, requires an analysis of the current structure of reinsurance arrangements in the East Caribbean market. As mentioned - 68 - above, the primary reinsurance medium consists of the quota or proportional treaty which absorbs the larger portion of the risks written in the region. In order to evaluate the suitability of instruments (either risk financing or risk transfer) which utilize excess of loss type covers (XL), the pricing of the existing quota treaties needs to be compared on equivalent terms with XL rates. Quota treaties are priced based on primary insurance rates, i.e., the premium over the sum insured is charged to the primary insurer minus the commission received by that primary insurer from the reinsurer. Conceptually, however, the 'layers' of loss coverage implicit in a quota treaty arrangement should be similarly priced to those under XL arrangements after taking into account differential transaction costs. The table on the page following, illustrates this approach for the East Caribbean, i.e., using rate averages currently in effect in the OECS, Barbados and Trinidad & Tobago: At the left side of the table, a proportional/quota treaty based on a EPL of 10% and a weighted average premium of 0.59% of sums insured is illustrated. Applying the premium cost over the product of EPL & value insured, this yields a rate-on-line (ROL) premium equivalent of approximately 5.9% before taking into account commissions paid. After subtracting commissions to cover business acquisition and administrative costs (28% of premium ceded), the result is a net 4.2% ROL equivalent. The 10% EPL is used based on the more diversified reinsurer's portfolio which further minimizes portfolio risk to below the individual company average EPLs in the region (15%-20%). As can be observed, the gross ROL figure derived from the quota treaty is equivalent to the average of all ROL layers shown at the middle and left of the table. Here, each level of loss is shown along with its respective probability of occurrence and ROL premium. At the rightmost side, the weighted sum of the various ROLs yields an average ROL of 5.9% as well. - 69 - Analysis of ROL Rates and Proportional Treaty Equivalents - Eastern Caribbean Insured Val. Coverage Coverage Weighted Treaty a/ EPL Prem. Point Layer Prob. ROL Spread ROL Rate $ 100.00 10% 0.6% 5.7% 5.7% 15.00% 17.00% 2.0% 1.0% 10.0% 4.3% 11.00% 14.00% 3.0% 0.6% ROL Eqv. 5.9% 14.3% 4.3% 7.50% 10.79% 3.3% 0.5% 21.4% 7.1% 5.30% 8.30% 3.0% 0.6% less rein. 28.6% 7.1% 3.50% 6.64% 3.1% 0.5% commis. 4.2% 35.7% 7.1% 2.50% 5.80% 3.3% 0.4% 42.9% 7.1% 1.50% 4.90% 3.4% 0.4% 50.0% 7.1% 1.20% 4.20% 3.0% 0.3% 57.1% 7.1% 0.75% 3.90% 3.2% 0.3% 64.3% 7.1% 0.70% 3.80% 3.1% 0.3% 71.4% 7.1% 0.35% 3.50% 3.2% 0.3% 85.7% 14.3%/, 0.18% 3.40% 3.2% 0.5% 100.0% 14.3% 0.15% 3.35% 3.2% 0.5% 100.0% 5.9% a/ Assumes after deductible under proportional/quota share reinsurance treaty. A summary of the above ROL pricing structure at selected probability levels is illustrated below: Prob. Rate X; over indabove 65%losses 0J.35% - .0 1.50)% 7 v vg > ..., ia,,9__, , 4.90% - - E o ad above-i / s Imp- 7.50% 11.0% 15.0% .0% The above, thus, serves to show that reinsurance rates-on-line are consistent with primary premium rates under proportional/quota treaties. Nevertheless, it also shows that well managed low-cost local companies stand to gain from reinsurance commissions - 70 - received when these more than cover their expenses. However, it also means that local ESC companies may have biases towards higher premiums as these in turn receive higher reinsurance commissions, all else being equal. The above (second illustration), however, also shows that reinsurance arrangements at various levels of risk can be tailor-made to improve the coverage/cost ratio of insurers and policyholders depending on the level of risk transferred abroad and the amount retained as "own" capital. Since premium rates decline at the higher (less probable) layers of loss, this provides a number of financial structuring opportunities to optimize pricing, assuming a larger retention of risk at the lower levels where reinsurance is most costly. Recently the capital markets (both bond and credit markets) have entered the reinsurance business through the provision of alternative risk transfer and financing instruments which are geared towards providing special purpose excess of loss covers, particularly at the highest loss levels. These developments are examined further in the next chapter to demonstrate how, under certain risk sharing arrangements, such instruments can provide efficiently priced sources of capital to fund insured risks. In addition, they can allow domestic insurers who pool their capital and increase risk reserves, to leverage retained capital in ways that can help to increase local insurance capacity. This would permit a reduction in the sensitivity of local insurance coverage to external supply crunches. - 71 - V. TRANSFERRING CATASTROPHIC RISKS TO THE CAPITAL MARKET A number of events have prompted capital market institutions to enter the catastrophe reinsurance business during the last decade. In particular, some of the colossal disasters experienced in the 1990s demonstrated the vulnerability of even the largest global insurers, and the observation that reasonable "reference " scale disasters which could be expected, still have the potential to disrupt the insurance markets if events occur at frequencies slightly above the norm. It was almost inevitable given the sheer size of the capital markets, that capitalfrom such sources would eventually be channeled to fund insurance risks directly. However, since capital market institutions are not typically risk underwriters in the classic insurance sense, the instruments developed to fund or 'securitize' risk sometimes have particular properties for measuring risk which are tailored towards the investor community that provides such capital. Capital market participation first manifested itself by way of credit suppliers that provided backing for government/industry partnerships via the financing of the highest layers ofpotential losses which insurers in disaster prone regions had become reluctant to underwrite. These innovations have appeared primarily in the developed world markets, nevertheless, they can provide immediate opportunities for catastrophe prone countries. In the Caribbean some of these tools can be incorporated as part of their package ofpolicy measures in order to implement sound and sustainable risk management & disaster funding strategies. - 72 - V. TRANSFERRING CATASTROPHIC RISKS TO THE CAPITAL MARKET In response to the capital constraints that might affect global reinsurers (and therefore emerging market insurance sectors) during periods of recurring high frequency catastrophes, and in response to domestic risk aversion by primary insurers in the wake of major disaster events, the need for additional 'risk bearing capacity' and long term funding availability has generated a supply of risk capital from non traditional sources, i.e., the capital and credit markets. While the reinsurance capacity has also increased by way of new industry players such as in the Bermuda market, the catastrophes of the last decade demonstrated that world funding reserves for potentially major events needed to be assured in ways that would not only avoid past cyclical supply/demand capacity constraints and associated price volatility, but also assure the future solvency of major insurance and reinsurance players. The response, therefore of the capital and credit markets, far from being an indication of a replacement of reinsurance, has shown more precisely that both reinsurance and capital market instruments need to work together to best manage large event risks in ways that optimize risk sharing among property owners, primary insurers, reinsurers and capital markets. The new instruments for managing catastrophe risk are not only used by those wishing to be insured in the traditional sense, but also by major reinsurers themselves who see the advantages of distributing risks to other markets while putting less of their own core capital at risk. Those 'other' markets whose capacities are practically unlimited in terms of financing the potential range of global catastrophic losses, have also the advantage of absorbing financial shocks with little or no 'bounce back' effect through the insurance markets, a characteristic which is particularly beneficial for countries which can be severely affected from the insurance industry cycles following natural disasters. Paradoxically though, the fact that such alternative instruments have entered the world stage, may in-and-of-themselves, serve to dampen premium volatility which in the past was caused mainly by the absence of alternative capital sources. However, this implies that volatility in pricing may require a permanent presence and utilization of alternative sources of risk capital by insurance industries throughout the world. This would be to maintain the supply of such capital available while global insurance capital grows incrementally until it reaches a new plateau where catastrophic events might be more fully insured and funded. Both governments as well as private sector players in emerging economies can partake of such tools by pooling sufficient capital to allow accessing these alternative forms of risk financing and risk transfer. Risk Financing Arrangements One of the initial forms in which the capital markets got involved in paying for catastrophic losses was through financing arrangements for pooled structures. This took the form of a line of credit that was syndicated to banks, insurance companies and other lenders. The primary benefit of such an arrangement was that it made large sums of funds immediately available and was meant as a supplemental source to pay for the potentially very large losses. This was especially helpful in areas where reinsurers had high exposures and additional capacity was an issue. Another benefit of a credit facility was its relative cost in terms of having funds available to pay for losses. The - 73 - commitment fee ranged from 0.25% to 0.375% per annum and the borrowing cost had a range between 0.75% to 1.5% over LIBOR for a term of up to 7 years. The disadvantage with a financing arrangement is that any amounts availed of, would have to be repaid together with interest. In order to enter into a financing arrangement, a distinct source of repayment had to be established and pledged to the lender(s) as security. This distinct source of repayment took the form of an assessment or surcharge to policyholders of a pool or, more often, spread to a broader base, typically all policyholders in a state. The California Earthquake Authority had the former, while the Hawaii Hurricane Relief Fund and the Florida funds adopted the latter approach (see last section of this chapter). The order in which the credit facility was utilized in relation to the other claims paying sources (cash on hand, reinsurance, member company assessments, etc.) was a function of the objectives of the pool. In general, the credit facility was placed at the topmost layer (last to be utilized) since the pools wished to avoid imposing assessments or surcharges to its constituents. Another form of credit financing that has been employed are pre-event notes. These are bonds that are issued prior to any catastrophic event. The cash proceeds are placed in a trust and can only be utilized to pay for losses. The added advantages of issuing bonds relative to a line of credit are that the bonds have a longer maturity (10 to 30 years), fewer loan covenants, and less pre conditions for the drawing of funds. The disadvantage is that the bonds are generally more expensive than a line of credit facility and require the immediate payment of interest. The Market for the Securitization of Catastrophe Risk The securitization of catastrophe risk, that is, the packaging of insurance risks into marketable financial securities, and which is opening new options for catastrophe coverage, is manifested in the private capital markets in the U.S., Europe, and Japan under the following instrumentalities: Catastrophe Bonds: Catastrophe bonds pay investors high yields, but are subject to default on all or part of principal and interest if a catastrophic event occurs during the life of the bond. Thus investor appeal is based on the high returns with low probabilities of default while the insurer's interest is in obtaining additional reinsurance capacity which is made available for claims payments in the event of a disaster. Funds obtained from bond proceeds are normally invested in risk free instruments which also help the insured to lower the eventual net cost of the interest coupon payments. In order to accommodate the desire of insurers and reinsurers to treat protection attained through catastrophe bonds and derivatives in the same fashion as traditional insurance, the approach in the capital markets has been to create a reinsurance contract between the ceding Insurer and a special purpose vehicle (SPV) which then effectively securitizes the contract on the market13. Such entities issue catastrophe bonds to investors, and as a separate transaction, sell reinsurance to interested insurers who can then take advantage of the more favorable 13 In most jurisdictions, without using an SPV, a regulatory drawback of catastrophe bonds is that, unlike straight reinsurance whose premium cost can be deducted from an insurer's gross written premium, the cost of securitizing catastrophe risk cannot be deducted in the same manner. Thus, when considering securitized risk transfer instruments, a risk based capital criteria would appear to favor a company's net surplus position when using straight reinsurance versus having the same coverage using catastrophe bonds. Nevertheless, as discussed, financial markets have devised methods for overcoming such impediments, primarily through the establishment of offshore special purpose reinsurers. - 74 - regulatory accounting treatment. The insured party is not subject to any reinsurer 'credit risk' either, as the insurance coverage is fully collateralized. Contingent Surplus Notes: These financial Notes are essentially 'put' rights which allow the insured party to issue debt to pre-specified buyers in the event of a catastrophic event. The Notes are a risk financing (verus risk transfer) mechanism but under regulatory norms can be considered as part of the insurer's available surplus or capital. The issuance of notes can be done in exchange for cash or liquid assets which are received from investors. Such liquid securities are kept in a trust which, in the event of a disaster, are exchanged via a financial intermediary, for the surplus debt notes issued by the insurer to finance catastrophe loss claims. Exchange Traded Catastrophe Options: The property claim service (PCS) options which trade on the Chicago Board of Trade as investment instruments, are mechanisms which provide the right of the purchaser to demand payment under an option contract, if the claims index surpasses a pre specified level (the strike price). The indexes used cover different areas of the U.S. (East, MidWest, West) and reflect insurance industry aggregate reported claims which are converted into an industry index. A national index also exists, but the regional indexes permit hedging against large risks In specified areas. The market does not yet constitute a large segment of the insurance market but Investors can profit from selling the options in diversified territories which are unlikely to suffer losses simultaneously (with corresponding option pay outs). Use of an index rather than specific claims experience also can result in 'basis risk', i.e, the risk that specific claims obligations do not necessarily exactly match the compensation amount from the option pay out. A drawback, however, is that insurance companies cannot deduct these costs as premium equivalents. Catastrophe Equity Puts: Equity puts are also a form of an option which, for a pre-paid fee, permits the insurer to sell equity shares on demand to investors, as a means of funding claims in the wake of a major disaster. Such instruments as in the case of contingent surplus notes, are risk financing instruments and do not actually perform the traditional insurance function of risk transfer, though they provide immediate liquidity. Catastrophe Swaps: Catastrophe swaps are another method of paying premiums for catastrophe reinsurance. Such swaps use capital market players as counterparties. In a catastrophe swap arrangement, an insurance portfolio with potential payment liabilities are swapped for a security and its associated cash flows. An insurer would take on the obligation to pay an investor periodic payments on a specified security (or portfolio of securities) which the investor was liable for, while conversely, the investor would take on the potential liabilities under an insured portfolio, for example, by making payments for catastrophe losses based on agreed measures of magnitude or severity (e.g.: a catastrophe loss index). For the insurer, payments made on the investor's securities are equivalent to a reinsurance premium. Weather Derivatives: Weather derivatives are contracts against weather change triggers generally take the form of premium payments for contracts which provide payouts to the "insured" in the event that a pre-defined number of days with a specified time period reach temperatures above or below the trigger point. Farmers in areas subject to crop freeze for example can purchase weather options whereby payment is made if the number of cold days below a certain point go past a pre-defined period and temperature level. Similarly for drought or heat affected areas such weather derivatives can be purchased to invoke payment if hot weather remains in -75 - force for a longer forecast period (e.g: for utility, electric or gas companies). Parties in opposite weather areas can swap positions to insure interested counterparties for such transactions, although speculative investors can also benefit from the premium income eamed in the case no trigger is invoked. One of the earliest examples of successful securitization of catastrophe risks as well as one of the largest amounts of securitized risk done to date, involved the USAA Insurance Company's transaction (7/97 to 6/98) with Residential Re, which was a special purpose reinsurer set up specifically for this purpose. The outline of the structure is shown below: Bond proceeds Residential A n Reinsurance l *~~~~~P Class A-2 Premiums + Libor + 5.76 Investors USAA Trust Account b Ru reinsurance two flowsBof bond proceeds shown for the A-i investors reflecef h i Class A-1 Investor Class A-1 Collateral Account w Investors Libort+ 2.73 As can be observed, the USAA deal offered cat bonds to Class A-1I and A-2 investors. Class A-1 received a lower bond yield but received protection on the principal in the event of a catastrophe exceeding the 'attachrnent' point of the contract (i.e.: loss levels reaching up through the reinsurance layer which was covered by the bond). The two flows of bond proceeds shown for the A-2 investors reflect the fact that additional funds beyond reinsurance needs had to be raised in order to place part of the bond proceeds in a Collateral Account where they would be invested to return the principal to those investors. The remainder ofthe A-I bond proceeds were kept in the trust account, subject to use in the event of a disaster. The additional interest cost of those extra bond proceeds plus those used for reinsurance ftrding, was offset by the lower interest rate offered to the class A-id investors. For the A-2 investors, the yield was much higher as they were subject to losing all principal. Residential Re was the special purpose reinsurer set up to collect the 'premiurn' payrnents from USAA and pay for reinsurance 'claims'. The bonds had a dual trigger which included both hurricane strength and actual loss levels. As a comfort against moral hazard, USAA kept a high retention of risk ($1 bn.) as well as a proportional / quota share reinsurance treaty above that level. Disaster index contracts or parametric hazard triggers as illustrated below, can also provide an efficient means of setting up insurance contracts since they depend on objectively measurable events to trigger indemnification. This can be useful for the protection of govermnent assets for example, since it skips the 'insurance loss adjustment' stage which requires site-by-site evaluation of damages. One drawback, however, is the 'basis' risk, i.e., the risk that the event 'trigger' does not necessarily correlate with losses on - 76 - the ground. Nevertheless, this was successfully implemented under Parametric Re, a special purpose reinsurer used by Tokio Marine Insurance to obtain securitized reinsurance against earthquake risk. The 'parametric trigger' was based on earthquake intensity as well as on two grids rings specified around Tokyo and other Japanese cities (an inner and an outer area grid) to determine what reinsurance would be provided in the event of an earthquake. Parametric Re: Earthquake Intensity & Grids as used as Payout Triggers Earthquake level: Richter Scale 7.1 7.2 7.3 7.4 7.5 7.6 7.7 % Bond loss: Inner City Grid Epicenter 25% 40% 55% 70% 85% 100% % Bond loss: Outer Grid Epicenter 25% 44% 63% 81% 100% Grid Areas for Epicenter Determination Outer Grid Inner Grid Tokyo Yokohama Chiba Insurance/Reinsurance Flow Structure 4-.-- = Reinsurance Coverage - = Premium/bond payment Tokio Maritne Swiss Re Parametric Re Insurance Co. *Company *(Cayman 4 nvestors\ Islands) (Reinsures with ReinsurersiRe- Reinsures Purchase Swiss RE.) >trocedes w, > Swiss Re /> Cat Bonds Pararnetric Sells cat bonds\ Source: Goldman Sachs In line with this type of 'parametric trigger' design, the following illustrates a hypothetical catastrophe bond structure and its issuance terms in which a 'multi-hazard bundle' is incorporated into the risk profile of the bond. While such bonds have not yet been issued to date, investors wishing to diversify against not only financial market movements, but also against concentrations of single hazards, may find such an issue attractive since some of the hazards (rainfall versus drought) are mutually offsetting. -77 - Structure of a Multi-Peril Catastrophe Bond/Note Using weather index triggers Terms of Note: US $100,000,000 Issue Maturity: 3 years Coupon Interest (no event): LIBOR + 4.99 Expected (probability-adjusted) value of coupon interest: LIBOR + 4.93 Coupon Interest (post-event): LIBOR - 2.40 Expected Loss: 0.79% of Face Value (probability adjusted) Expected value of yield given principal/coupon interest adjustments: LIBOR + 4.68 Index triggers for coupon/principal adjustment: Earthquake: Richter 7+ quake (specified locations/cities - first event trigger) Hurricane: Windspeed 120+ m.p.h.(specified coastal locations - 2nd+ event trigger) Flood: Accumulated Rainfall (defined period) 40+ cm.(specified locations - 2nd- event) Drought: Accumulated (defined period) 2- cm. (specified locations - Ist event trigger) Coupon strip payment Coupon strip payment Coupon strip payment Coupon strip payment Earthquake Risk Winsdstorm (hurrica- Flood (rainfall) ris Drought (rainfall) risk LIBOR + 5.5 ne risk) LIBOR+5.0 LIBOR + 4.6 LIBOR + 4.9 (no event) (no event) (no event) (no event) Post Event Terms: Post Event Terms: Post Event Terms: Post Event Terms: LIBOR - 3 LIBOR - 2 LIBOR - 2.5 LIB3OR -2 Prob.: 0.4% Prob.: 1.2% Prob.: 1.1% Prob.: 0.3% Weight: 25% Weight: 25% Weight: 30% Weight: 20% Weighted coupon interest payment above LIBOR (pre-event): 0.25 *(5.5)+0.25*(5.0)+0.25 *(4.6)+0.20*(4.9) (no loss of principal) = 4.99 % Coupon Weighted Expected value: coupon interest above LIBOR, taking into account event probabilitiestrip weight rate Earthquake risk: ((0.004)*(-3.0)+(1-0.004)*(5.5)) 5.47 0.25 1.37 Hurricane risk: ((.012)*(-2.0)+(I-0.I2)*(5.0)) 4.92 0.25 1.23 Flood risk: ((.01I)*(-2.5)+(1-.011)*(4.6)) = 4.52 0.30 1.36 Drought risk: ((.003)*(-2.0)+(1-.003)*(4.9)) 4.88 0.20 0.98 1.00 4.93 Weighted coupon interest payment below LIBOR (post-event): 0.25*(-3)+0.25*(-2)+0.3*(-2.5)+0.2*(-2) (with loss of principal) = -2.40 % Expected discount on principal, taking into account event probabilities (over 3-yr. period): Loss of Principal Earthquake risk: 0.004* 100*0.25 0.10 Hurricane risk: 0.012* 100*0.25 0.30 Flood risk: 0.011* 100*0.3 0.33 Drought risk: 0.003* 100*0.2 0.06 Total ($US millions) 0.79 Expected yield on note/bond taking into account probabilities of loss of interest/principal (post event): (note: event time is not specified, but averaged out over bond repayment period - assumes annual coupon) Years: 0 1 2 3 Cash Flows: -100 4.93 4.93 4.93 ($US mn.) 99.2 Total: -100 4.93 4.93 104.1 Yield (IRR): Libor +4.68% - 78 - Cost/Benefit Factors in Utilizing Capital Market Instruments A simple numerical example shows how capital markets can supplement insurance and reduce costs. Assume that a primary insurance company calculates the probability of a loss of more than US$15 million but less than US$25 million at 1O percent. If the primary insurer purchases reinsurance at this rate, it will break even over time. Adding administrative costs, operating costs, and a risk 'load', the reinsurer might charge a premiurn of 14 percent (10 percent + 4 percent). Alternatively, the primary insurer could issue a US$10 million bond to investors, then put the US$10 million in U.S. treasury notes paying, say, 5 percent. The investors' principal of US$10 million would be put at risk as part of the contract. If a catastrophe with losses exceeding US$25 million occurs, the investors might lose all their principal. For putting their principal at risk, the investors would demand at least a 17 percent return: 5 percent as risk-free interest, 10 percent for the "pure" risk of losing their principal (akin to a default risk), and 2 percent as an additional risk load. Net of the investment in treasury notes, the insurer's total financing cost would approach 12 percent, compared with the 14 percent for traditional reinsurance at that given level of risk. Institutional/Financial Structure for a Catastrophe Bond Scheme Shareholder Bond Guarantor Contributions (for contract comp liance) pay V V losses Bonds Natural Disaster Recipient/benefi- Catastrophe Fund Ciaries of catastro- Investors phe coverage Cash transfers 11 11 bond proceeds v v Defeasance Account Investment Account (Trust) In the above example, a "Catastrophe Fund" can be privately or publicly owned. In the international private markets as mentioned, this usually takes the form of a "special purpose vehicle" which serves as a legal/financial entity to invest proceeds obtained from catastrophe bonds and pay investors the bond coupon. In the event of an actual catastrophe, funds would be paid to the insured recipients. The defeasance account is used to accumulate funds for repayment of the bond principal if investors are to receive some level of 'principal protection' in the event of a catastrophe event - in this instance, as discussed earlier, the funds raised usually exceed those needed for indemnification since some are set aside solely for the purpose of repaying principal. However, some of these bonds are structured so that the investor loses part or all of the principal and some interest in the event of a catastrophe. The compensation is that the bonds usually pay yields much above the market level for similar 'default risks'. In yet another option, however, the insurer could arrange a standby credit of US$10 million with a 2 percent commitment charge and an interest rate of 12 percent that - 79 - kicks in if the loan is needed. If a catastrophe occurs, and assuming a ten-year repayment period for principal (yielding a combined principal plus interest "insurance" cost equivalent to 1 8 percent), the expected financing cost would be 3.6 percent (0.1 [1 8 percent] + 0.9 [2 percent]), much lower than with direct reinsurance. These capital market schemes to supplement insurance have many possible variations. These range from full risk transfer with no financing (where the full principal is at risk, just as in reinsurance) to zero risk transfer with full financing (full repayment 6f principal). The following diagram illustrates the typical 'full risk transfer' catastrophe bond arrangement showing the pre-event and post-event cash flow payments (post disaster event flows indicated by the dotted lines). Premium (4.5%) | _ ~~~$100 mn. _ _0 milio No Disaster: * _h i -- { | i LIBOR + 4 (coupon) |----- $100 mn. (at maturity) Post Disaster: LIBOR - 2 Loss Event can be Defined by Objective Parameteri suc s WinSee or Eathquke Inest Implicit in the above structure, is that disaster-prone countries might be sufficiently creditworthy to issue 'cat' bonds for investors to purchase. Support by IBRD in the form of a contractual risk guarantee would enhance the credit rating of such bonds and potentially improve their marketability. The attractiveness of such bond issues could also be enhanced through the use of 'parametric' indicators to trigger eligibility for payment. Such triggers mean that measurable physical events such as wind speed or earthquake intensity at a defined location could trigger the 'insurance payment' instead of using the traditional loss adjustment process based on a structure-by-structure analysis of damage. As discussed in this report, parametric triggers imply some basis risk. It is difficult to precisely pinpoint why the private sector does not always fully respond to such market needs. The competitive marketplace is a dynamic, ever-changing world where demands and supply are changing constantly in response to underlying societal and natural phenomena. In the case of catastrophe insurance, a new era was formed following Hurricane Andrew which is still in the infant stage of its development. - 80- The slowness of the response is partly explained by the extensive legal and regulatory barriers that had grown around the insurance industry over the past century, and which required some modifications. For example, in the case of Cat Bonds, even in developed economies, substantial time was spent obtaining opinions and rulings from various jurisdictions stating that investors purchasing such bonds would neither be acting as unlicensed insurers or violating anti-gambling laws. Substantial additional time elapsed working with insurance regulators to assure that such bonds purchased by insurers would be assessed their risk based capital on bonds, and not as higher risk securities which carry a more punitive rate. Other regulatory issues pertained to disclosure, listing & timing requirements, withholding taxes, and revisiting the definition of insurance; aspects where the multilateral development institutions can be crucial in accelerating market development. A factor which could greatly assist the development of an ESC reinsurance pool would be the recent legislation in some Caribbean countries, e.g., Trinidad & Tobago, and Barbados allowing for a tax deductible catastrophe reserve. An ESC reinsurance pool would benefit from this provision. Catastrophe Programs Around the World Japan, New Zealand, France, the United Kingdom and the states of Hawaii, California and Florida in the U.S. have all adopted plans to deal with catastrophic risk. In Japan, New Zealand, and California the risk in question is earthquake. For France the main concern is flood. For Hawaii and Florida, the risk is hurricanes. For the U.K., the risk is terrorism. Special Megacatastrophe Programs In the U.S., the states of Hawaii, Florida, and California have special programs to deal with megacatastrophes. Hawaii. The Hawaii Hurricane Relief Fund works as follows: Insurance companies are allowed to write homeowners insurance with a hurricane exclusion. Each participating insurance company then acts as a servicing carrier for the HHRF, issuing the insured a separate hurricane policy and collecting a separate premium that is then forwarded to the HHRF. The HHRF receives ongoing revenue from hurricane premiums, insurance company assessments on property business, and mortgage recording fees. The plan envisages providing insurance to cover a megacatastrophe costing up to $2.1 billion. This is over 1.5 times the cost of residential damage caused by Hurricane Iniki, which hit Hawaii in 1992. Under the Hawaii scheme, the first 10% in loss from a megacatastrophe is assumed to be bome by homeowners through deductibles. The next loss layer would be paid by private insurance companies participating in the HHRF. The following loss layer comes from reinsurance purchased by the fund. The last $750 million comes from a line of credit, which is secured by future surcharges on all property/casualty premiums. If losses exceed the total coverage amount, claims are paid on a prorated basis. Coverage occurs when the National Weather Service announces a hurricane watch and ends 72 hours after the - 81 - hurricane ends. TIhe HHRF, however is now being phased out as private insurers have begun returning to the market. Aggregate Coverage $1.45b Line of Credit $750m $700m Reinsurance Single Year XOL $175m $400m Maximum Industry Loss Assessment Maximum $B00m - Estimated Actual $400m HHRF - Sources of Cash Flow to Pay or Finance Claims Premiums from On-Goincg Windstorm Coverage i$1 .40-$1 .751$1000J Annual Assessment 3.75% All Mortgage Recording Fees 0.10% Assessment on Event-Triggered Gross Premiums Participating Potential (Max. $500mper Event)Insurers Order Assessment on of Gross Premiums 11.25%) All Collectio Surcharge on Insurer il 2 _ Thus, for example if St. Lucia's portion of the retention layer was $10 million, and the country suffered a loss totaling $30 million, it would be eligible to collect $16 million from the pool. This reflects $20 million above its retention minus the 20% co- insured retained portion under the proportional treaty reinsurance layer. At the initiation of the pool, the annual premium income would be $111 million less 15% commissions which equals $94 million. Reinsurance premium costs to the pool would amount to $67 million ($15 mn. + $29 mn. +$21 mn. + $2 mn. loan commitment fee). However on the proportional treaty, the $15 mn. of premiums ceded would accrue a commission to the pool, which, assuming a 15% rate, would provide an additional $3 million. Therefore, the pool's initial annual income would amount to $97 million, leaving a surplus of $30 million to cover administrative costs and to accumulate reserves. It should be noted that the XL reinsurance premiums have been 'conservatively' estimated at rates which are actuarially fair but are three times the current market rates, to take account of potential future price fluctuations and potential variations in the estimated expected loss percentage. Therefore a significant cushion exists for additional savings if these 'adverse' events do not develop. Nevertheless, since the pool assumes a retention of at least $100 million in cat risk, this amount would be required as initial capital. This could be provided by private sector shareholders, strategic investors, or other means. Resulting Financial Income & Expenditure Flows for Pool ($ millions) Financial Income Catastrophe Premiums Received collected from Primary Insurers $11 1 Commissions from Reinsurers (under Quota Share Treaty) _3 Total $114 Financial Expenses Commissions paid to primary insurers $17 Net Quota Share premium paid $15 XL first layer premiums $29 XL second layer premiums $21 Credit commitment fee $2 Total $84 Net Income/Flow $30 - 113 - Defining a Catastrophe Poolfor Public Assets While the above analysis has combined both public and private sector assets in the pool structures as defined, governments could also choose the option of establishing a purely government pool to insure public assets and infrastructure. While such a pool would have less capital than the combined private/public pool, it could nevertheless be considered as an option if inclusion of private insured assets proved to be unwieldy or less desirable to the domestic insurance industry even if it implied no reduction in net income flows. Below, a potential structure for a sub-regional government insurance pool is outlined as well as the financing role of multilaterals in order to assure initial liquidity for the pool and provide a topmost layer of excess-of-loss protection in case of a major catastrophic event. The pool described would follow the same financial structure of the industry/public assets pool described above. The assets included in the pool would be government buildings and properties, schools, hospitals, and critical infrastructure facilities such as electricity and water transmission systems. A key aspect to consider in a public asset insurance pool, is each government's tolerance for the payment of premium from its fiscal budget. Therefore, such a structure would first need to determine the level of premium 'subscriptions' for a specified range of assets, commensurate with the affordability constraints and risk management objectives of each government. An additional feature of a government assets pool would be the provision of a low layer credit line from the World Bank or other multilateral agencies, to protect the pool from losses if they occurred early in the cycle before the premium build-up provided a sufficient capital cushion to absorb risks before reinsurance layers were invoked. This lower end protection becomes more feasible in a government versus a private sector pool since the credit line becomes a public 'general' liability while in a private sector pool such a liability at the lower layer could constitute too high of a debt burden for corporate statutory standards. However, in the public sector case, since losses at the subsequent layer would be covered by reinsurance, the debt burden is strictly limited to the difference between available capital for 'retained' risk and the required full retention amount before the reinsurance could be invoked. - 114 - Outline of an Insurance Pool Structure for Public Assets Upper Reinsurance Layer |Government | Public Critical j _ 4 ~~buildings Hospitals T Infrastructure| and other and (electricity, Assets Schools water, etc.) Initial Risk Exposure Assessment and Fiscal Tolerance (Premium) Iits The structure also includes a top layer risk financing facility as in the private sector pool, which would only be invoked at the lowest probability event levels which generated losses above the last reinsurance layer. As will be shown in the next section of this chapter, such financial instruments prove to be more cost effective in the long run than adding an additional topmost reinsurance layer, since they require little or no cost when not deployed, and they allow repayment over a long period when used (which can be funded by future premiums). Results Obtained from Testing Risk Transfer and Risk Financing Mechanisms This section provides the methodology and results for testing viability of alternative risk transfer mechanisms for financing catastrophic risks. Capital market products such as loans of different maturities and pricing (commercial, IBRD, IFC) were tested for specified event probabilities, and compared with existing available reinsurance pricing. Commercial loan financing was tested utilizing credit enhancement instruments such as the IBRD Guarantee which was also used to test viability of catastrophe bond products for capital market issuance. Event scenarios at the 5% probability level (reflecting relatively frequent but less loss intensive hazards) were first tested over a 200 year simulation period using IBRD loan terms of 15 year maturities and 5 years grace at an interest rate of 6.5% (above the - 115 - current IBRD variable rate). The market premium for excess of loss reinsurance at this level is 7-8% of the loss layer covered, though a 7% rate was used. The same exercise was also applied to IFC loan terms using a 10 year maturity, 2 year grace, and an interest rate of 10.5% reflecting the high end of IFC pricing which ranges between LIBOR + 2 to LIBOR + 4. Commercial credits were also tested using maturities of 6 years with 1 year grace at 11% interest cost. A 5% risk free rate was used as the discount factor. This rate is extremely conservative and tends to bias results in favor of reinsurance cover when compared to utilizing an opportunity cost of capital rate which might be double the risk free rate and which would tend to bias results towards credit instruments. For commercial financing, an IBRD guarantee was assumed in order to provide credit enhancement for private sector lenders to participate. The type of guarantee is not specified because both a partial credit guarantee would apply (e.g.: to ensure a minimum maturity period of the loan or loans) or a partial risk guarantee (to ensure contractual commitments on the debt service obligations). In both the cases of risk transfer (reinsurance) versus risk financing (lending), the net present values for the "insured" were negative, that is, after taking into account the indemnity payments received following a disaster, the policy holder or primary insurer would not "profit" from the coverage, which is consistent with the use of insurance for protecting against cash flow shocks and drastic reductions in asset wealth. The results are as follows: Testing Alternative Risk Financing Instruments: IBRD, IFC and Commercial Loan Terms Using 5% Catastrophe Event Probability Reinsurance IBRD Terms IFC Terms Commercial Simulation period 200 years 200 years 200 years 200 years Event probability 5% 5% 5% 5% Premium Rate 7% Interest Rate na 6.5% 10.5% 11.0% Commitment fee na 0.75% 0.5% 1% Front end fee na I% 1% 1% Maturity/grace na 15/5 10/2 6/1 Guarantee fee na 0.7% Financial Results PV Cost Ratio: Financing / Reins. 1.0 0.6 1.4 2.0 Ratio of: avg. debt service / Premium 1.0 1.9 2.5 3.5 The above results show that in present value terms, the cost of utilizing IBRD financing would, over a 200 year period constitute only 60% of the conservatively estimated cost of reinsurance for the specified level of risk. As seen, however, for both IFC and commercial lending (the latter including a small guarantee fee of 0.70%), the long term cost is substantially higher in present value terms (1.4 and 2.0 respectively). The second ratio shown, attempts to demonstrate during any given period of utilizing loan proceeds, what the ratio of debt service (P+I) would be during such period versus having used and paid reinsurance premiums. Such a ratio is useful to determine the level of reserves needed at any time to ensure sufficient liquidity and solvency to repay the required debt service. The following table shows a 400-yr. simulation period for an event probability of 1% (i.e., less frequent but potentially of much larger loss magnitude) using - 116- the same financing terms as above, except that the corresponding reinsurance premium at that level would be 4% of the loss coverage layer, given the lower probability of use. Testing Alternative Risk Financing Instruments: IBRD, IFC and Commercial Loan Terms Using 1% Catastrophe Event Probability Reinsurance IBRD Terms IFC Terms Commercial Simulation period 400 years 400 years 400 years 400 years Event probability 1% 1% 1% 1% Premium Rate 4% Interest Rate na 6.5% 10.5% 11.0% Commitment fee na 0.75% 0.5% 1% Front end fee na 1% 1% 1% Maturity/grace na 15/5 10/2 6/ 1 Guarantee fee na 0.7% Financial Results PV Cost Ratio: Financing / Reins. 1.0 0.2 0.3 0.4 Ratio of: avg. debt service / Premium 1.0 3.5 4.8 6.5 The second scenario above provides a much more favorable cost/benefit outcome in terms of relative PV costs of using financing versus reinsurance. An additional benefit would be the absence of volatility in the 'reinsurance' rate due to pre-set financing terms. For this purpose, the IBRD loan facility has the additional option of being converted to a fixed rate loan following disbursement through a currency swap/conversion agreement which, under current IBRD policy would only charge a 0.125% fee. At a LIBOR rate of 5.6%, for example, and IBRD's loan spread of 0.55%, the total cost before fixing the rate, would amount to 6.15%. The additional fixed rate cost plus the transaction fee would raise this rate higher, therefore, the 6.5% rate used in the simulation is reasonable and within expected bounds. In addition, the commitment fee assumed is 0.75% while present waivers only require a 0.25% fee payment currently, and in the foreseeable future. As can be observed, when the lending proceeds are utilized (1% probability), the relative debt service ratio compared to reinsurance premiums is much higher than in the first scenario with 5% probability, and if the loan repayment were annuitized, it would be equivalent to a 13% rate on line. The main reason for this is that, at the 5% probability level and leaving the loan termns unchanged, the premium itself is substantially higher (7% vs. 4%) and therefore the relative debt service ratio in the first case appears more favorable. However, even if this were not the case, there exists another key problem with engaging risk financing at the 5% probability level: At 5% event probability, the excess of loss reinsurance cover represents a low to middle layer cover among the possible layers of cover. At this level, this constitutes a key segment of the XL reinsurance market and therefore using credit facilities, particularly at favorable terms, could constitute market interference. It is therefore not recommended, unless private commercial parties are willing to provide financing at terms similar to IBRD, to engage credit financing for catastrophe risk at the 5% probability layer. However, the same argument does not apply to the 1% probability layer. This is because at the highest (least likely) but potentially more destructive event probabilities, it - 117- is where the global reinsurance market can experience capacity/supply crunches as occurred following hurricane Andrew in 1992. At these upper levels it is precisely where the cat bond market has developed since additional 'capacity' in the capital market exists at these levels. Thus, a more promising use for alternative risk financing and credit enhancement instruments would be at the highest but least probable loss levels which pose inherently risky scenarios for insurers and reinsurers alike. Based on the current insured loss exposure for the OECS, Barbados and Trinidad & Tobago under a pooled structure ($750 mn.), the backstop credit financing requirement would amount to $150 mn. based on the 20% upper level currently reinsured under XL contracts in the region. However, since the debt service usage under such scenarios implies much higher cash flow requirements (3.5, 4.8, 6.5 as multiples of reinsurance premium for IBRD, IFC and Commercial terms respectively) any such entity, pool or fund utilizing such instruments must have within its financial structure sufficient initial capital to cover the required debt service even if long run costs are lower in PV terms. The latter fact, however, provides the possibility of accumulating reserves at a faster rate so that initial extra capitalization might only need to ensure coverage for the first instance of loan utilization. Nevertheless, because of the higher debt service costs at the 1% event probability level, in relation to reinsurance premium, the case still needs to be made to determine if use of credit at such levels is indeed more cost effective not only from a cost/benefit viewpoint but also from an annual operating scenario. To do this, the 'savings accumulation' factor was calculated based on the lower cost of utilizing credit in the long run, but also taking into account how quickly those 'savings' accrue before the debt service comes into play. The following illustrates the results of this calculation: Accumulation of Additional Reserves Using IBRD Loan Facility To Compensate of Higher Debt Service Costs at 1% Probability Assuming Mean EPL Savings from Loan Ratio of accumulated Facility as Percentage of premiums paid versus Previously Paid Premiums debt service under under Reinsurance using loan facility per Contract - 100 yr. period 100 yr. Period a. Savings accumulation in PV terms immediately 81% 5.3 before 1% event b. Savings accumulation in PV terms immediately post 82% 5.6 1% event, before debt serv. c. Savings accumulation in PV terms, post event and 81% 5.2 after debt servicing costs The above shows, as per the earlier analysis of differential present value costs, that the savings from utilizing a stand by credit facility are sufficient to accumulate at a rate which would compensate for the periodic higher debt service payments using a loan facility. Since under the IBRD option, the ratio of debt service to premium was 3.5, the level of savings should be sufficient to build up reserves to pay for eventual loan - 118- servicing costs even after including other incidental costs such as administrative expenses. In comparison to the percentage standard deviation of catastrophe reinsurance prices in the last decade, the savings accumulation rate exceeds that significantly, which means that the savings more than compensate for the higher debt service costs (versus reinsurance). Since the savings accumulation provides the necessary funding in disaster years, and since the loan facility maintains a stable cost of financing, this facility effectively reduces the volatility in reinsurance pricing (such reinsurance pricing also rises after major disasters, as discussed previously). It should be clarified that the above figures assume a constant premium rate for the XL coverage at the 1% risk layer. However, as documented earlier, rate volatility is inevitable and by historical standards, 1998 and 1999, the years upon which the reinsurance rates are based in this exercise, is one of the lowest rate periods in the last decade for the East Caribbean market. The use of credit financing swapped into fixed rates at the highest loss layers, would therefore lock in more stable rates. In this context, multilateral institutional support could only be considered as a catalytic function since permanent support for such a scheme would be expected to be borne by the private markets . It should also be clarified that given the risks that the 1% probability occurs 'early' in the expected cycle, the pool would need to be adequately capitalized initially in the event that the distributions of actual catastrophes become more concentrated than the mean values suggest. The use of debt in this regard has to be measured against the future repayment capacity taking into account the need for adequate initial reserves to accommodate early occurring events. Incremental PV Savings: Credit vs. Reinsurance Cost 7000 6000 E 4000 302C00 | i i; ;1~ *Twice Expec Pob EEppected PYb 2000 E 1000 -1000 -- Elapsed Years To test a more severe scenario, the probability of the catastrophic event was set at double its actuarially estimated frequency so that use of the credit would be twice as frequent over the 400 year simulation period. In this case, even though the 'savings' rate diminishes, the initial accumulation is still sufficient to more than compensate for the cost of debt service following the event. The chart below shows the difference between - 119- savings under the mean expected probability scenario and a scenario with a 100% increase in probability. In the above chart, the "expected probability" (1%) shaded area is of the same magnitude as the 'twice expected' series area except at periods in which the 'twice' series requires additional use of debt service. In such cases, the savings become negative. However, even with this worse case scenario, the initial accumulation of savings (vs. reinsurance costs) is more than sufficient to offset the "dip" years. Therefore, provided that the savings obtained are maintained in a trust or investment account (rather than used for expenditures or paid out as dividends), they would maintain the credit-based cost of reinsurance at a low rate and continue allowing protection against exogenous rate-related volatility. Eventually, credit support would need to be borne by the commercial market which as shown above, could represent savings of up to 60% in long term premium costs based on a 1% actuarially based event projection. The build up of additional reserves from such savings would also lower the capital 'stress' caused by debt service repayments during 'event' years and this could potentially be ameliorated further by extending the maturity and/or lowering the interest rate further supported by additional credit enhancement mechamisms. Feasibility of Utilizing Catastrophe Bonds as Reinsurance The use of catastrophe bonds to support reinsurance capacity for a pool is also considered. At the outset is should be mentioned that catastrophe bonds usually require a number of preparation costs which can amount to 1%-2% or more of the issuance value. These costs pertain to bond underwriting costs and cover the legal expenses, setting up of special purpose vehicles & trusts, ratings analysis, investor marketing costs, and other necessary costs as part of the capital market based transactions. For multi-year bonds, however, these costs can be amortized. If well structured, catastrophe bonds can provide another source of reinsurance capacity with the added benefit of having rates based on the financial markets (e.g.: LIBOR) rather than on global reinsurance capacity alone. In contrast to contingent credit lines, bond coupon interest is payable from the start. In this respect, catastrophe bond payments are more akin to ongoing reinsurance premiums. As discussed in the earlier chapters, catastrophe bonds to-date, require an above market rate of interest for similarly rated risks. For the upper XL layers under the 1% event probability, cat bonds are being currently priced at LIBOR + 4 (or 9.6%). At the layer below (e.g.: 1%-2% event probabilities) the price is LIBOR + 5.5 (11%). While for investors, the lower probability (upper loss) levels are more attractive given that they imply lower 'default' possibilities, for the 'insured' using such bonds for reinsurance at extremely low probabilities may not be worthwhile since the rate will not necessarily be lower commensurate with the risk (due to the fixed costs of issuing the bond). Bonds which do not protect the principal from loss in the event of a catastrophe that generates losses above the attachment point, are priced substantially higher than those which have principal protection. In the case of a pooled catastrophe insurance structure for the East Caribbean, it is assumed that credit enhancement support would be required to assure a successful bond issuance for purchase by international investors. As per the earlier analysis, a World - 120- Bank guarantee for this purpose would add a 0.7% annual fee to the cost. Therefore, assuming that a catastrophe bond would be used for reinsurance cover at the 0.6%-1 % probability level, the total cost to the pool would be LIBOR + 4.7, or at current rates, 10.3%, substantially higher than an equivalent reinsurance premium. However, as per the earlier discussion, the proceeds of a catastrophe bond are to be invested in a trust account in risk free securities. The current U.S. treasury bill rate is approximately 4.9%. Thus, this investment would offset the cat bond financing costs by that amount and provide an approximate net cost of 5.4% = LIBOR + 4.7 - 4.9. So far we have excluded underwriting fees assumed to be an additional one time expense of 2%. If we use the cat bond form of reinsurance for a $100 million loss limit, the total annual coupon payment 'premium' or rate-on-line equivalent would amount to $5.4 million, and the first year underwriting fee $2 million. In the financing structure of the pool listed earlier, a reinsurance premium of $21 million as a maximum was estimated, even though the current market rate-on-line for the same, would be $3.5 million, i.e., below the cat bond net interest payment. Therefore, based on the projected financial income and expenditure of the pool, a cat bond would still be affordable as an alternative method of reinsurance. As discussed earlier, though, investors might only be receptive to purchasing such instrunents if the loss trigger was clearly defined and transparent. The current loss adjustment process and determination of indemnity payments in any given country might not be perceived as infornation which is symmetrical in its availability to both the investor and the insured. Because of this, a parametric trigger to determine the claims eligibility is proposed as the only feasible option even though this can generate potential basis risks. Nevertheless, with significant modeling based on hazard intensities at given sites and related dollar losses, a determination of the basis risk could be made. This would require use of meteorological measurements of windspeed and/or earthquake in islands which have suffered from disasters. For improved 'payment triggering' under a catastrophe bond arrangement, heavy duty wind measuring devices for example, could be purchased and set up in selected spots on each island participating in the pool. These could be connected via wireless transmission to a central database administered by the pool as well as to an independent body such as the Miami Hurricane Center. Once an above threshold wind measurement was recorded, this would serve as the basis for declaring a pre-specified 'loss payment' from the bond, i.e., the bond proceeds would be used to indemnify for affected damages, in which case the investors would lose part of their principal and future interest. While it would theoretically be possible for the pool administrators in conjunction with the domestic insurers to report exact losses on the ground, and thus allow the cat bond's loss on its principal be pro-rated accordingly, the institutional infrastructure for developing such a reporting system would not only take substantial time to implement, but as mentioned above, might not be attractive as a reporting system to the bond investors. Therefore, loss payments from cat bond proceeds might be structured parametrically as follows: - 121 - Hazard Intensity Investor Loss on Threshold Bond for Indemnification Category 5 Hurricane 80% of Principal Earthquake Richter 7.5 1.5 % on Interest Category 4 Hurricane 60% of Principal Earthquake Richter 6.5 1.0 % of Interest Category 3 Hurricane 40% of Principal Earthquake Richter 5.5 0.5% of Interest The hazard event would need to be measured at that level for a specified time (e.g.: 60 seconds or more) at one of the pre-defined sites. The loss of principal and interest, however, would only apply to that portion of the loss pertaining to the country or countries affected. Thus, as in the pool structure shown above, if a category 5 hurricane hit only one country which represented 10% of the total pool exposure, then the loss of the cat bond's principal would be 0.1*(80%), or 8% of face value. An additional feature that might be built into a catastrophe bond contract is a swap agreement to convert LIBOR based interest payment into fixed rate payments so as to maintain rate stability for the insured and the SPV during the life of the bond. This would also add to the initial funding costs, but as earlier discussed, would still be affordable based on the pool financials. Therefore, in the catastrophe bond instance under such arrangements, a cost versus variance trade off would need to be examined further to determine whether the higher financing costs might indeed offset rate volatility in the reinsurance markets. Investors have proved to be receptive to catastrophe bonds despite their apparent riskiness. Nevertheless, when compared to similarly (or lower) rated securities in the financial markets, their return versus risk profile is actually much more favorable. Below are shown the relative characteristics of catastrophe bonds with certain attachment probabilities along with other rated securities: - 122 - Comparison of Default Rates on Cat Bonds versus Other Bonds Bond Default Probabilities (or equiv. Cat Event Prob.) Spread above LIBOR Below Investment Grade Bonds Ba2 0.60% 1.10% Ba3 2.70% 1.36% B I 3.80% 1.84% B2 6.79% 2.00% Principal at Risk Cat Bonds Res Re '97 1.00% 5.76% Parametric 1.02% 4.30% Trinity 1.53% 4.36% Res Re '98 0.87% 4.00% Principal Protected Cat Bonds Res Re ' 97 1.00% 2.73% Parametric 1.02% 2.06% Trinity 1.53% 1.49% Source: Goldman Sachs In addition to the higher return/risk profile that many catastrophe bonds can offer investors, they also improve the investors overall portfolio risk given that catastrophe bond performance is not linked to global financial market prices since they are based on natural event triggers versus economic/financial factors. Thus, besides the benefits of higher returns, the lack of correlation of cat bond performance with the broader financial markets, improve the efficient investment frontier: Change in Efficient Investment Frontier from Inclusion of Cat Bonds in Portfolio A B Risk The addition of cat bonds to a portfolio of assets with significant global price correlation, would reduce expected portfolio volatility due to the risk diversification aspect of cat bond returns versus global financial market returns (e.g.: U.S. markets, - 123 - European markets, emerging markets, in fixed income securities or equities). Thus an equivalent portfolio rate of return can be achieved with a lower risk profile (curve A) compared to a similarly valued portfolio before adding uncorrelated cat bond securities (curve B). In conclusion, it should therefore be stated that a number of configurations and options for managing catastrophe risk exist, and these can exploit recent development in risk transfer instruments as well as different capital structures for managing and holding sufficient catastrophe reserves. The structures presented previously for deployment in the ESC case represent what is seen as the most applicable and least complex to implement given the financial participants, country situations, and affordability constraints particularly in terms of fiscal resources available for the payment of insurance premium. However, in the process of developing a risk management framework for an intra regional group of countries, the risk exposure, insurability, and pricing characteristics may vary once data is collected based on actual conditions and level of resolution used. Such an exercise may well point to alternative structures and finding mechanisms for assets at risk which may need to be seriously considered as additional choices. This report has attempted to show, however, that under reasonable and conservative assumptions, the case for improved risk management by using a package of policy and financial instruments in the ESC area, can prove to be a valuable policy tool to improve long term development prospects of such nations, as these periodically suffer the devastating effects of natural disasters. - 124 - - 125 - Bibliography CARICOM Working Party, "Working Paper on Insurance & Reinsurance and Catastrophe Protection in the Caribbean", prepared with OAS, World Bank, USAID, May 1996. Canabarro, E. and Finkemeier, M., Goldman Sachs Fixed Income Research, "Analyzing Insurance Linked Securities", October 1998. Chase Manhattan Bank, "Financing Natural Catastrophe Risk Exposures", June 1996. Chase Manhattan Bank / Chase Securities, "Disaster Risk Management for Central America", April 1996. Cholnoky, T., Zief, J., Werner, E., Bradistilov, R., Goldman Sachs Insurance, "Securitization of Insurance Risks", April 1998. Dominica Registrar of Insurance Office, "Commonwealth of Dominica: Report of the Registrar of Insurance on the Administration of the Insurance Act for the year ended December 31, 1997". Eastern Caribbean Central Bank, "1993/1995 Insurance Review". Eastern Caribbean Central Bank, Draft of "OECS Insurance Act". Evans, A., "The Insurance and Reinsurance Market in the Caribbean", June 1999. Fernando, E., Managing Director, Chase Manhattan Bank, "Banking and Governmental Interface in Catastrophic Risk Management", September, 1998. Froot, K., "The Limited Financing of Catastrophic Risk: An Overview", Havard Business School, and NBER, September 1997; Froot, K. "The Pricing of U.S. Catastrophe Insurance", Harvard Business School and National Bureau of Economic Research, March 1997. Froot, K. "On the Pricing of Intermediated Risks: Theory and Application to Catastrophe Reinsurance", Harvard Business School and NBER, April 1997. Froot, K., "The Market for Catastrophic Risk: A Clinical Examination", National Bureau of Economic Research, August 1999. Gibbs, Tony, CEP Co., Civil Engineering. Estimates of Vulnerability Reduction Measures. Goldman Sachs, "Risk Securitization - Opportunities in a Developing Asset Class", September 1998.. Goldman Sachs, "Overview of Investing in Risk-Linked Securities", August 1999. Guy Carpenter Monitor, "The World Catastrophe Reinsurance Market 1999", October 1999. Insurance Services Office, Inc., U.S., "Managing Catastrophe Risk", May 1996. Insurance Services Office, Inc., U.S., "Insurer Financial Results: 1998", June 1999. - 126- Insurance Services Office, Inc., U.S., "Financing Catastrophe Risk: Capital Market Solutions", January 1999. International Monetary Fund, "International Financial Statistics", December 1999. Litzenberger, R., Beaglehole, D., and Reynolds, C., Goldman Sachs Fixed Income Research: "Assessing Catastrophe Reinsurance-Linked Securities as a New Asset Class", July 1996. Marsh & McLennan Securities/Guy Carpenter, "The Evolving Market for Catastrophic Event Risk", August 1998. Millete, M., VP Goldman Sachs, "Introduction to Capital Markets Instruments for Catastrophe Risk Management", September 22, 1998. Mooney, S., Insurance Infornation Institute, and World Bank, "Managing Hurricane Catastrophic Disasters in the Caribbean Region through Mitigation and Financial Risk Sharing", June 1997. Paterson, D. and Mitchell, A., Beinfield Greig Ltd., Technical Notes on Risk Exposures and Pooling Analysis for the OECS, Barbados and Trinidad & Tobago. Pollner, J., "Proposed Strategy for a World Bank Role in a Small States Regional Catastrophe Insurance Program", February 1999. Poliner, J., World Bank Viewpoint, No. 197, "Using Capital Markets to Develop Private Catastrophe Insurance", October 1999. Pollner, J. and Clarke, L., World Bank and CCMS: CGCED Report, "Wider Caribbean Financial Sector Review", May 1998. St. Lucia Registrar of Insurance Office, "Report of the Registrar of Insurance for the year ended December 315', 1997". St. Paul Re., "Average Gross Fire Rates (including Catastrophic Perils) in the Caribbean Region, 1989- 1997". Stubbs, N., "Estimation of Building Damage as a Result of Hurricanes in the Caribbean", OAS/USAID, June 1996. Stubbs, N., "A Probable Maximum Loss Study of Critical Infrastructure in Three Caribbean Island States", World Bank and OAS, June 1999. Swiss Re., Sigma, No. 7/1997, "Too Little Reinsurance of Natural Disasters in Many Markets", 1997. Trinidad & Tobago Supervisor of Insurance Office, "Report of the Supervisor of Insurance for the year ended 31 st of December, 1996". Policy Research Working Paper Series Contact Title Author Date for paper WPS2542 Checks and Balances. Private Philip Keefer February 2001 P. Sintim-Aboagye Information, and the Credibility of David Stasavage 37644 Monetary Commitments WPS2543 When Do Special Interests Run Philip Keefer February 2001 P. Sintim-Aboagye Rampant? Disentangling the Role in 37644 Banking Crises of Elections, Incomplete Information, and Checks and Balances WPS2544 The Uniqueness of Short-Term Leora Klapper February 2001 A. Yaptenco Collateralization 31823 WPS2545 Financing the Future: Infrastructure Marianne Fay February 2001 A. Francois Needs in Latin America, 2000-05 37841 WPS2546 Gender Dimensions of Pension Paulette Castel February 2001 J. Smith Reform in the Former Soviet Union Louise Fox 87215 WPS2547 The Design of Incentives for Health Jeffrey S. Hammer February 2001 H. Sladovich Care Providers in Developing William G. Jack 37698 Countries: Contracts, Competition, and Cost Control WPS2548 International Provision of Trade Alan V. Deardorff February 2001 L. Tabada Services, Trade, and Fragmentation 36896 WPS2549 Measuring Poverty Dynamics and Erzo F. P. Luttmer February 2001 C. Wodon Inequality in Transition Economies: 32542 Disentangling Real Events from Noisy Data WPS2550 Measuring Equity in Health Care Adam Wagstaff February 2001 A. Maranon Financing: Reflections on (and 38009 Alternatives to) the World Health Organization's Fairness of Financing Index WPS2551 Infrastructure Coverage and the Kristin Komives February 2001 S. Minovi Poor: A Global Perspective Dale Whittington 30012 Xun Wu WPS2552 Inventories in Developing Countries: J. Luis Guasch February 2001 J. Troncoso Levels and Determinants-a Red Flag Joseph Kogan 37826 For Competitiveness and Growth WPS2553 The Value of Relationship Banking Giovanni Ferri February 2001 A. Yaptenco during Financial Crises: Evidence Tae Soo Kang 31823 from the Republic of Korea In-June Kim Policy Research Working Paper Series Contact Title Author Date for paper WPS2554 Administrative Costs and the Estelle James February 2001 A. Yaptenco Organization of Individual James Smalhout 31823 Retirement Account Systems: Dimitri Vittas A Comparative Perspective WPS2555 Implicit Pension Debt, Transition Yan Wang February 2001 A. Datoloum Cost, Options, and Impact of Dianqing Xu 36334 China's Pension Reform: Zhi Wang A Computable General Equilibrium Fan Zhai Analysis WPS2556 Household Strategies for Coping with Michael M. Lokshin February 2001 P. Sader Poverty and Social Exclusion in Ruslan Yemtsov 33902 Post-Crisis Russia WPS2557 Decentralization and Accountability: Stuti Khemani February 2001 H. Sladovich Are Voters More Vigilant in Local than 37698 in National Elections? WPS2558 Growth, Inequality, and Poverty: Martin Ravallion February 2001 P. Sader Looking beyond Averages 33902 WPS2559 Deposit Insurance as Private Club: Thorsten Beck February 2001 P. Sintim-Aboagye Is Germany a Model? 38526