A discussion of index-based natural catastrophe protections by Jean-Paul Conoscente, Jean-Yves Nouy, Christophe Gaudron and Emmanuel Dubreuil.

Indices have been used in many different markets to establish the basis of remuneration of a contract. An obvious example is a basket of investments referenced to the NASDAQ or the NYSE indices. Such approach has also been used over the past ten years in the insurance and reinsurance industry, mainly under two forms, catastrophe (CAT) bonds and weather derivatives. In such contracts, the loss experience of the cedant has no reflection on whether a loss is paid out or not. Instead, payment is triggered only when the contractual index exceeds a pre-agreed threshold. Typically, the amount paid is also a function of how much the index exceeds the threshold, with a pre-agreed cap or limit.

The risk transfer can be carried out in two different ways:

  • a reinsurance contract, in which case loss must be proven and the indemnity received must be less than the loss suffered; or

  • a financial contract.

    In the case of CAT bonds, both types of contracts are used, since a financial contract is passed between investors and a special purpose vehicle (SPV) set up for the transaction, and a reinsurance contract is passed between the cedant and the SPV.However, to date, both CAT bonds and weather derivatives have had limited success.

    Bonds and derivatives
    CAT bonds seemed to be the answer to the hard reinsurance market worldwide after the large natural catastrophe losses between 1989 and the mid-1990s, including the San Francisco earthquake of 1989, the European windstorms of 1990, the Japanese typhoon Mireille of 1991 and hurricane Andrew in 1992. CAT bonds were devised to transfer natural catastrophe insurance risks to the financial markets, through financial SPVs.

    While a number of transactions have been successfully placed over the past decade, their development remains very limited because of the large frictional costs involved (legal investment banker catastrophe modelling fees, etc.). These costs are typically of the order of 1.5% to 2% of the capacity placed. Indeed, in spite of the current hard reinsurance market, the number of CAT bond transactions and the capacity available in 2001 and 2002 remain very limited.

    Similarly, weather derivatives were developed in the US in the 1990s after energy deregulation. Companies, such as energy producers, were able to protect their balance sheets against seasonal changes that might affect their revenue, while keeping their costs constant. Weather derivatives were developed to trigger when excess variations around a seasonal average temperature occurred, therefore protecting the ceding (or issuing) company against a warm winter or a cool summer, which would reduce the demand for electricity.

    In such transactions, no loss needs to be proven; the cedant receives a payment once the contractual index exceeds a certain threshold. This market has developed substantially in the US, primarily with energy-related companies. However, its development outside the US energy market remains limited, again because of the important frictional costs involved (intermediary fees and correlation study fees) and because the deregulation of most energy markets in Europe and Asia still lags behind the US.

    Transaction of interest?
    Despite these apparent setbacks, cedants and risk takers remain interested by an index-based concept for transferring natural catastrophe risks. Cedant companies are interested because the product is a bridge between the financial and the reinsurance markets, it does not require any clause or event definition and the payment is available very quickly.

    The major drawback for a cedant remains the correlation or basis risk, i.e. the risk that the money received will be less than a company's actual loss. A detailed comparative analysis of the proposed index and a company's portfolio loss can mitigate this risk, as well as adapting the index as closely as possible to a company's loss profile.Similarly, risk takers are interested in underwriting index-based transactions because the quality of data available is typically better than the information usually available in traditional CAT placements, the transactions carry no run-off and they are very simple to settle. Given the hardening insurance and reinsurance markets and the remaining interest in such transactions, a number of solutions have been proposed using physical measurements of natural catastrophes with as little transaction frictional costs as possible.

    Structure for windstorm
    An example of such cover is the index-based reinsurance protection Benfield Greig placed for a large French insurance company to protect itself against French windstorm.

    There are three steps required to structure such a transaction:

  • selection of the meteorological stations to be used;

  • correlation analysis to decide the weighting factors assigned to each station; and

  • derivation of the probability distribution of the derived index;

    An example of the selection of meteorological stations is presented in Figure 1, to construct a wind index for a pan-European portfolio. This selection is representative both of the exposure distribution of the cedant's portfolio, of the vulnerability to damage of the risks insured to the peril under consideration.

    In a second step, weighting factors are assigned to each station, in function of the cedant's regional loss potential, including exposure and vulnerability. For example, windstorm damage to individual homes is typically higher as a percentage of the exposure than storm damage to apartment buildings. This can be reflected in the weighting factors assigned to each station, according to the distribution of such risks in the cedant's portfolio.

    Using the measured wind speed at each station and the assigned weighting factors, a portfolio index is derived, taking into account the spatial correlation between the different stations (i.e. what is the wind speed at station A given that the wind speed at station B is X). An example of the spatial correlation between different French meteorological stations is shown in Figure 2. This shows, for example, that the wind speed measured near Marseille (Marignane) on the Mediterranean is highly correlated to the wind speed measured further west along the coast at Sète and highly independent of the wind speed measured in Paris (Athis-Mons).

    The transaction is then designed to trigger for a given index value (or attachment point) and the cover is exhausted once the index exceeds a certain threshold (or exhaustion point). This type of non-proportional agreement is the same one used in traditional reinsurance transactions, except that in this case, the attachment and exhaustion points are expressed in kilometres per hour instead of a monetary value. Then, as with weather derivatives, the payment received is a function of the wind speed payment in excess of the attachment point and of the monetary value assigned per kilometre per hour.

    Conclusion
    Despite the lack of previous enthusiasm for index-based transactions, both cedants and reinsurers remain interested in such an approach, especially if the structural costs are kept to a minimum and the prices asked by reinsurers and investors remain competitive with traditional reinsurance. A number of market players, including Benfield Greig, believe that the further growth of this market will benefit all participants and are actively pursuing its development.

    By Jean-Paul Conoscente, Jean-Yves Nouy, Christophe Gaudron and Emmanuel Dubreuil

    Jean-Paul Conoscente, Jean-Yves Nouy, Christophe Gaudron and Emmanuel Dubreuil are with Benfield Greig in London.