In the second part of his series, Youssef Al Kareh provides recommendations on how securitisation of insurance risk can be a creative alternative to the traditional insurance business paradigm
Securitization of insurance risk is the most prominent example of the convergence of the capital and insurance markets. This process is relatively new dating back to the aftermath of Hurricane Andrew in 1992 (that left US$ 20 billion of insured losses in inflation-adjusted terms.1) In 1994, the Northridge earthquake resulted in insured losses of US$ 15 billion in 2002 money2. These two events drained the available capacity in the traditional catastrophe3 reinsurance markets. Consequently, this supply side-effect increased cat reinsurance prices to unprecedented levels causing insurers to look for alternative more cost-effective solutions.
One such solution was to draw upon the resources of the much more liquid capital markets. Swiss Re (2001) estimated that the market value of the publicly traded stocks and bonds stood at US$ 60 trillion and at the same time estimated that an earthquake on the New Madrid Fault in the US with an 8.5 magnitude on the Richter scale would cause insured losses close to US$ 115 billion, a mere 0.19 per cent of the US$ 60 trillion in securities. In fact, a US$ 250 billion fluctuation in the securities markets is a normal daily occurrence.
However insignificant this 0.19 per cent might be, compared to an estimated US$ 200 billion capacity of the US Property & Casualty insurance industry, of which only US$ 20 billion belong to reinsurers (Swiss Re Sigma No. 5/1996), it can wipe out an entire industry in one single event.
Hannover Re kick-started the market for cat bonds in 1994 with its US$ 85m KOVER issue. At the time of writing their Sigma No. 3/2001, Swiss Re estimated the total issue volume of insurance risk securities to be US$ 5 billion, including cat bonds, cat swaps/options, option to issue, and life bonds. In the May 2002 issue of "Reactions", the 2001 cat bonds issue alone was estimated at US$ 1 billion and Judith Klogman of Swiss Re Capital Markets estimates that for the year 2002, US$ 2 billion of issues were expected. All these signs point to the fact that activity in this sector is taking roots, especially after the catastrophic losses of September 11 which were expected to consume close to US$ 50 billion of reinsurance capacity.
Against this backdrop, let us examine more closely what are the driving forces behind this promising activity. But before we do that, let us first examine how securitization works in practice.
Typical Securitization Structure
An insurance company sets up a Special Purpose Reinsurance Vehicle (SPRV) in a tax haven jurisdiction. Its sole purpose is to engage in the securitization transaction by providing reinsurance coverage to its parent company (just like a captive). With the help of an investment bank the SPV issues bonds to the capital markets.
Payment of coupon and principal on these securities is linked to the occurrence of a catastrophic event; in fact the same event that the sponsor insurance company is seeking protection against. The proceeds from the sale of the securities are kept in a trust fund and invested in highly liquid securities (e.g. T-Bonds or highly rated commercial paper). These funds can only be accessed when the coverage is triggered by the occurance of the catastrophic event, in order to settle claims to the insurer and upon payment of interest and principal to the bondholders. At the same time, the funds are ring-fenced against any liquidation threat of the sponsor company to ensure that the vehicle does not become bankrupt. The bonds can be sold in tranches to appeal to different risk appetites. Each tranch is rated except the equity tranch. Rating of the bonds is not possible without the contribution of catastrophe modeling firms4 that specialize in calculating the likelihood of a catastrophic event occurrance, its magnitude and impact on the exposed regions.
Of course, each transaction is unique and subject to lengthy negotiations. As a result, the structures may be more complicated. However, it is not the purpose of this article to analyze the complexity of these structures or to investigate all of the alternatives. Eventually, solutions of this sort are limited only by the imagination of the transaction engineers. The purpose of the above structure therefore, is to highlight the flow of funds in the transaction and the various contributing parties.
For example, often there is a reinsurance company between the insurer and the SPRV, or we might have tranches with protected principal. When the principal is partially or totally protected, the bond issue raises an amount larger than actually needed to provide the reinsurance support. The balance is kept in a safe account to be used in the situation where there is an event triggering the coverage. Here, the balance is used to purchase deeply discounted T-Bills to repay the principal, though with a considerable delay commensurate with the maturities of the T-Bills (e.g. after 10 years). This feature is called defeasance.
Drivers of the Transaction
There are many interests, on the supply as well as on the demand side, that make the securitization of insurance risk possible. The issuers (insurance companies) are encouraged by several factors including access to new capacity, reduced credit risk, leverage, price, multi-year protections, maximizing shareholder value, and other strategic issues. Investors on the other hand are attracted to these issues because of two main considerations, higher yields compared to similar rated bonds and a great diversification effect. Let us now consider each one of these factors in turn.
Issuer's Perspective
New capacity in the capital markets has been the actual propeller of this industry. In 1996, Swiss Re estimated a catastrophe coverage gap5 of between US$ 20 and 30 billion. More important is the ever increasing property values and concentration of population in cat-prone areas, which is constantly driving up the levels of exposure6. If there is to be any hope of providing adequate coverage to this increased exposure, it has to come from the capital markets.
Moreover, for most AAA or AA rated (re)insurers, traditional reinsurance capacity will have to come from lower rated companies, an uncomfortable situation by all means. Ensuring quality of the securities in the capital markets is the only solution in this case.
Reduced credit risk: In the aftermath of a major catastrophe, the robustness of the reinsurance industry is often put to the test. It is in these difficult times that insurers have the greatest need for their reinsurance programs to absorb the sustained losses in their portfolios. Unfortunately, it is exactly during this critical period that most reinsurers fail: and as a consequence many insurers. For this reason, a transfer of cat risk to the capital markets is a much safer proposition because such a transaction virtually eliminates the credit risk exposure of the insurer.
This is seen in the above example of a typical securitization structure in the fact that the proceeds from bonds sales are invested in high quality AAA-rated securities. These same assets are used to bolster the potential liabilities and are locked into a secured trust with a sole purpose of paying claims and servicing of outstanding debt. Not only that, but in this case liabilities are collateralized up front and dollar for dollar unlike in the case of traditional reinsurance where the reinsurer is only requested to keep risk capital, sometimes only a fraction of the total exposure of the company. The latter point is the reason why many companies fail in the aftermath of a major catastrophe where the aggregation of losses from the same event overwhelms the capital base of the insurer.
Leverage: Reinsurance capacity has always been used by insurers to leverage their balance sheets to allow them to write more risk at a given level of capital. If the market for cat risk securitization develops and matures, then it might offer even greater latitude for insurers to use leverage. Firms with a competitive advantage in underwriting might engage in aggressive sales and marketing of particular types of insurance policies, package these policies and sell them to an intermediary who later transfers the risk to the capital markets via securitization. As such, insurers use their core competency to leverage their ability to write more business knowing there is a suitable warehouse of risk at the end of the chain.
This might encourage the formation of a completely different insurance business paradigm than the one existing at present. Whereas now the market consists of end users, insurance companies and their reinsurers, in the future, and thanks to a mature securitization market, we might have something like the end user, an underwriting agency, an intermediary (transformer of insurance risk) and the capital markets as the end warehouse for cat risks.
Price: The insurance industry has always suffered because of the whims of the underwriting cycle. At times of strained capacity, high demand drives insurance prices up. Investors lured by the smell of profits flood the market with capital. Competition intensifies for the same business and prices drop (soft market) until a major event (or a series of consecutives events) wipes out sufficient capacity from the market. The resulting reduced supply of reinsurance capital drives up prices again (hard cycle), sometimes to really prohibitive levels.
Soft markets inhibit the development of the markets for securitization of insurance risks where the spreads on cat bonds become much higher than the price of traditional reinsurance. However, in a hard market, securitization might become a cheaper source than traditional reinsurance and its pricing more lucrative. This is another reason for insurance companies to use this method of risk transfer.
Multi-year protection: Bonds are issued for a minimum of three years. This feature of a multi-year protection relieves insurance companies from the instability in pricing in reinsurance markets, making financial planning much easier and certain, a characteristic cherished by CFOs.
Maximizing shareholder value: Investors in general are compensated only for systematic risk inherent in the equity markets. If cat risks (which are clear, discernable and transparent) were not diversified, the returns received by shareholders of insurance companies would not be commensurate with the level of risk they are carrying. As such, the pooling and transfer of cat risks is one way of maximizing the share price of insurance stocks.
In other words, it can be said that if insurers decide to self-insure cat risks, they would be requested to lock risk capital commensurate with the added exposure. By laying off this risk, insurers can release the capital for a more economical use and as such increase the value of their stock. (Marsh & McLennan Securities 1998)
Strategic issues: If the market for securitized cat risk evolves, companies who have been proactive will be the first to benefit from any niche opportunities that become available. Also, early participation might project an image of innovativeness and leadership.
At the same time, rating agencies and regulators might be comforted by the fact that an extra layer, over and above the traditional Cat XOL exists, in the form of an insurance linked security.
Investor's Perspective
Higher yields: Swiss Re (1999 & 2001) suggests several reasons why cat bonds offer higher returns to investors than similarly rated bonds or asset-backed securities. One reason is the 'newness premium' to compensate investors for the non-traditional nature of the securities. This premium is expected to disappear or at least shrink once the market in cat bonds matures. Even then, investors still have the chance to earn higher returns as compensation for model risk (risk that the modeled losses are less than expected actual losses), the illiquidity of cat bonds (so far a secondary market in these instruments has not fully developed).
When hard markets drive reinsurance prices towards sky-high levels, insurers might be willing to compensate investors more generously than under normal market conditions. There is also an opportunity for the shrewd investor to discern undervalued issues and as such realize a higher than average return by arbitraging this discrepancy in pricing.
Goldman Sachs (1996) compared the returns of a high-yield bond (single-B rated) to those of a hypothetical Cat bond issue (under two different statistical estimation techniques, log normality and bootstrap7). The latter was judged similar to the high-yield bond because of its high coupon rate and the exposure to its principal. "The average rate of return on a single-B bond is 5.61 per cent, whereas the average rate of return from simulations using log normality is 7.47 per cent and 7.88 per cent, based on the bootstrap. The minimum return on the CAT bond is 0.6 per cent, whereas the minimum return on a high yield bond is -1.38 per cent". Simulations were used because it is statistically proved that relying on historical data for rarely occurring cat events is a misleading approach.
Figure 2 below is a comparison showing the returns for three types of bonds of similar risk: a Ba2 rated industrial bond, a Ba2 rated sovereign Argentina bond8, and four different issues of Cat bonds (Parametric, Trinity, Residential, and Mosaic). The higher yield of the latter (though of similar risk) is very obvious. The industrial and sovereign bonds have similar maturities to the Cat bonds and they are swapped to LIBOR to enable comparison.
Figure 3 shows that over time, cat bonds have consistently provided returns in excess of Baa rated corporate bonds. The trend line represents the returns on Baa corporate bonds, while the individual dots show the returns on the following issues, respectively: Residential Re, SR EQ Fund, Parametric Re, Trinity Re, Trinity Re II, Mosaic Re, Residential Re II, Mosaic Re II, Domestic Re, Residential Re III, Juno Re, Namazu Re, Seismic Re, Alpha Wind, and Res Re 2000.
2. Diversification effects: On top of the attractive yields that they provide cat bonds are a textbook example of the diversification effect on investment portfolios. Modern finance theory tells us that the inclusion of securities, which are not or are lowly correlated with the existing securities in a particular portfolio, reduces the riskiness of the portfolio by decreasing its overall standard deviation. Cat bonds do just that because of their statistically insignificant (sometimes negative) correlation with stocks and fixed income securities (Tables 1 and 2 below). In other words, including cat bonds in a portfolio of bonds and stocks for example opens opportunities for even more efficient portfolio selection.
The addition of uncorrelated securities (cat bonds) shifts Markowitz's 'efficient frontier'9 upwards, that is to say the new frontier (including cat bonds) dominates10 the old one (without cat bonds). This is a goal for any portfolio manager (Figure 4 below). The enhancement effects of cat bonds on a portfolio of securities are seen in Table 3 below. Note that though the inclusion of cat bonds somewhat lowers the overall return of the portfolio, the latter's volatility drops sharply, i.e. the reward-to-risk ratio11 increases .
Types of Triggers
The transfer of insurance risk to the capital markets is a genuine one; it resembles closely a typical Cat XOL reinsurance policy which works by compensating the primary insurer once its Ultimate Net Loss (UNL)12 breaches a certain threshold (the deductible), up to a certain maximum (the limit), this is the indemnity principle of insurance. Most cat bond issues work in a similar fashion when the SPV compensates the insurer once its UNL exceeds the pre-agreed limit, again up to a certain limit (as per the terms of the reinsurance policy between the SPV and the insurer).
When a cat bond payment is triggered in this manner, we say that the issue uses an indemnity trigger. This has the advantage of tailoring the protection to the individual portfolio of the insurer and as such providing it with a complete protection.
The downsides of this method are moral hazard (after reinsurance is bought there is less incentive on the part of the insurer to mitigate losses), adverse selection (the insurer might be trying to cede exactly those risks that are most vulnerable to losses) and lengthy claims development periods (the latter aspect locks in investors' money for lengthy periods of time).
Another drawback, always from the investor's point of view, is that with indemnity triggers, careful assessment of the underlying risk becomes a necessity and is often a strenuous job because investors operate at arms' length from the insurer and so are not well equipped to perform this type of analysis. This is true in the case of one issue of cat bonds. Now try to imagine the amount of analysis needed if an investor is interested in assembling a portfolio of these cat bonds.
To overcome these investor concerns, several other types of triggers have been devised; index-linked triggers, physical or parametric triggers, and modeled loss triggers.
An issue with an index-linked trigger compensates the insurance company only when an index of industry losses (as measured by an independent third party agency) in a particular exposure region exceeds a certain level. The most notable of these indices are the PCS13 indices which, following a natural catastrophe, estimate the losses to the insurance market in any of nine geographical areas in the US14 using survey data from the major insurers involved in writing P&C15 insurance business in the affected area. Another index on the rise is the GCCI16 index covering seven geographical areas17.
While this type of trigger removes the negative aspects of indemnity triggers, it leaves the insurer with a possibly serious basis risk exposure18; the reason being a low correlation between the losses that the insurer had suffered and those captured in the index. Even though the industry may have not suffered enough losses to trigger the coverage, the individual insurer's own portfolio might have been severely hit.
A prominent transaction using this type of trigger (PCS index) is the Swiss Re's Redwood Capital II in April 2002 which provided Swiss Re with an US$ 200 of protection against California earthquake.
A parametric trigger does exactly the same job as an index trigger. Simply, this type of trigger compensates an insurer once the magnitude of a natural phenomenon exceeds a certain threshold as measured by an independent agency such as a Japanese earthquake higher than seven on the Richter scale (as measured by the Japan Meteorological Agency), or a Class four hurricane in Florida. (June 2003, Zenkyoren raised US$ 193 million through Phoenix Quake LTD to protect against Japanese EQ).
In such a case however, there is still an important risk that an insurer's portfolio may be seriously affected by an EQ of magnitude six or by a Class three hurricane. This situation might easily arise if the insurer's portfolio is concentrated close to the epicenter of the EQ or right on the trajectory of the hurricane. Therefore basis risk is a major feature of parametric triggers.
The last type of triggers used is the one linking the trigger to a modeled loss. Catastrophe models are used to rate cat bonds. They are based on simulations of a set of events that may impact a portfolio of insured risks, estimating the likelihood of such an event occurring in a particular region, its magnitude and its likely impact on the insured portfolio taking into account the vulnerability of the insured properties, the distribution of the insured values with respect to location and risk class, and insurance conditions applying to the original cover.
These same models can be used to estimate losses once a certain event takes place. Naturally this method, just like the preceding two, reduces the influence of the issuer on the recovery potential but at the expense of a sizeable basis risk if the model fails to predict the exact amount of the loss. No matter how well conceived models may be, they are still subject to uncertainty as they predict very complex and volatile phenomena. (In April 2002, Lloyd's Syndicate 33 (Hiscox) raised US$ 33 million through its St. Agatha Re issue, the cover was multi-peril).
We note that the use of parametric and modeled losses is very attractive to investors (as shown in Figure 5) for several other reasons than the management of moral hazard and adverse selection. These triggers are transparent, certain and easy to understand. They also provide a straightforward method of claims settlement reducing development periods and providing investors with the opportunity to reallocate their invested principal as soon as possible.
Last but not least, physical triggers in particular provide hope for the standardization of cat bonds' transactions where investors can re-use their analysis of a previous transaction to decide on future ones that use the same parameter as a trigger, making life for investors much easier.
Figures 5 and 6 below show the percentage usage of the different types of triggers and the amount of basis risk associated with each type of these triggers respectively.
Life Insurance Securitization
Catastrophe risks, though the dominant element in the insurance risk transfer market, are not the only type of insurance risk that can be securitized and sold to the capital markets. Life insurance portfolios have always been another interesting candidate. In December 2001, New Jersey-based Prudential Financial Inc. completed the largest such transaction by securitizing a closed block of life insurance policies, the amount of the issue neared US$ 2 billion ($1.9 billion exactly). This transaction is believed to provide the impetus for an added activity in this sector19.
The structures for life insurance securitization are very similar to those used in cat bonds transactions. Our aim is not to elaborate on the intricacies of these structures, so what we will explore in this section are only the differences that separate life from cat risks securitization. There are a number of elements worth the mention as such:
Cat bonds are genuine risk transfer mechanisms and it is this risk transfer that drives the transactions. Life insurance securitization on the other hand has been driven by totally different purposes, notably regulatory arbitrage20 and more efficient capital management, which has become ever necessary with the demutualization trend and consequent shareholder pressures for higher returns on capital;
Cat bonds are liability-backed transactions; the performance of the bonds is linked to the claims on the insurance portfolio. Whereas in life insurance, it is the embedded value21 (an asset) which is actually monetized, a typical asset-backed securitization transaction;
The legal constraints in life insurance securitization are predominant. In the US for example, regulators are reluctant to allow the true sale of assets to non-licensed insurance companies such as Special Purpose Vehicles. This aspect is a major deterrent to investors whose main comfort is the bankruptcy remoteness of the transformer that cannot be realized unless the legal ownership of the assets backing the transaction is actually transferred. The Prudential deal mentioned above provides one way around this legal hurdle by establishing an in-house transformer, a separate trustee-controlled debt service account, and by credit wrapping the higher tranches of the transaction. Still, many investors feel that this is not enough. Unless a genuine solution is found, this legal issue will stand in the way of what could be a very promising sector.
Conclusion
The hard market post September 11 was supposed to send the Cat bonds market towards unprecedented levels as insurers seek to replace lost traditional capacity and to cushion phenomenal price increases.
However, a fresh injection of capital into the traditional markets, mostly into the Bermudan market (to the order of US$ 12 billion) appeared to have halted this expected explosion in cat bonds issues. Very long gestation periods for cat bonds however might account for this slow take off and many deals are said to be in the pipeline.
Whatever its volume and growth, the transfer of insurance risk to the capital markets is here to stay.
Though it is suggested that it will never replace traditional catastrophe XOL reinsurance, which has many irreplaceable qualities, it will however remain a strong complement with a strong potential for growth once the modeling, rating, and legal expenses become more mainstream and less expensive.
Regional Applications
Fortunately, the GCC region is blessed with an almost complete absence of natural catastrophe activity of the sort that cat bonds are designed to tackle.
Having said that, other man-made catastrophes are increasingly being modeled today with the advances in modeling software, chaos mathematics, etc. One prominent example is terrorism risk which is a potential threat to the many mega institutions in the region with assets wo billions of dollars especially in the energy field (refineries, oil and gas carriers, etc.). It is said that any event that can be modeled can be securitized.
So who knows, maybe at some stage a 'terrorism bond' could be an effective replacement or complement to the expensive traditional reinsurance cover. Governments and regional syndicates may be the ultimate sponsors of such an issue.
Contributed by Mr Youssef Al Kareh
Mr Al Kareh is the General Manager of Ensurion, the first licensed insurance manager in the Middle East. He has a wide range of experience in (re)insurance, brokerage, captive insurance, and alternative risk transfer/finance (ART/F) with the Arab Insurance Group (ARIG) in Bahrain and with Marsh in London. M Al Kareh is a fellow of the Chartered Insurance Institute in London (FCII), a member of the Securities & Investment Institute (MSI), and an Associate Member of the Association of Corporate Treasurers (AMCT). He holds an MSc in Finance from the ISMA Centre at the University of Reading where he graduated with top distinctions.
Footnotes:
Global Reinsurance, July/August 2002 issue, p l;
ibid;
Henceforth in the paper, the word cat is used interchangeably with catastrophe. This is a recognized convention in the relevant markets;
The most prominent of these firms which are actually monopolizing this market are: Applied Insurance Research (AIR), EQE and Risk Management Solutions (RMS);
Cat risk exposure not met by the traditional capacity in the Cat XOL (Excess of loss) reinsurance market;
According to the July/August 2002 issue of Global Reinsurance, the population living in the hurricane exposed areas of the US is projected to increase 15.3% by 2025;
The use of the historical realized Adjusted Historical Loss Ratios (AHLR) as a discrete probability distribution is referred to as a bootstrap;
Of course before Argentina defaulted on its debt in December 2001;
Markowitz's efficient frontier is the universe of all possible investment opportunities open to an investor to realize maximum reward for a given level of risk or minimum risk for a given level of reward; When frontier X dominates frontier Y, we have a situation where for each level of reward, the corresponding risk on X is lower than that on Y, or for each level of risk, the relevant reward on X is higher that that on Y;
Reward to risk ratio = (realized returns - risk free return) / standard deviation;
UNL means the actual loss, including loss adjustment expense, paid or to be paid by the insurance company on its net retained liability after making deductions for all recoveries, salvages, subrogation and all claims on inuring reinsurance, whether collectible or not;
Property Claims Service;
Northeast, Southeast, East Coast. Midwest, West, California, Florida, Texas and National;
Property & Casualty;
Guy Carpenter Catastrophe Index;
Northeast, Southeast, Gulf Area, Midwest, Florida, Texas (subject to data availability) and National;
Basis risk is the risk that the insurer's protection will not fully compensate it for the losses it suffers;
For full details on the Prudential transaction, sec the May 2002 issue of Reactions, pp 32-33;
Releasing regulatory capital for a more efficient use, for example the funding of new business acquisition costs and/or diversification projects;
Embedded value in a life insurance portfolio is the present value of the future profits on the existing book of business, augmented by the present value of the future profits on new business.
© POLICY 2006
