Banding Together.

Reinsurance: Index-based pooling provides a way for insurers with different geographic concentrations of risk to swap standardized catastrophe exposures.

Catastrophe risk in different geographies is largely independent, with catastrophic damage often confined to relatively small areas. Thus, an optimal portfolio of catastrophe risk--one that minimizes loss variation--would cover many geographic areas and consist of an equal amount of risk in each area.

Unfortunately, several factors conspire against insurance companies that would like to create such portfolios:

* property exposures in the United States tend to be concentrated along the coasts;

* regulation often makes it difficult or unattractive to write insurance in all states; and

* a company's distribution channels may tend to concentrate their property risk geographically.

Although it is nearly impossible for an insurer to take on just the exposures it would like, companies typically rebalance their portfolios by buying and selling reinsurance, which increases their underwriting capacity and helps them achieve more stable earnings. Nevertheless, price and capacity fluctuations in the reinsurance market have made it desirable for primary insurers to explore other avenues of risk transfer, including reinsurance pooling.

By jointly agreeing to share a subset of their combined loss experience, insurers can ensure that risks are spread evenly among pool members. More importantly, reinsurance pools tend to have a price advantage since they make more effective use of members' existing capital rather than attempting to find a new source of capital to fund this risk.

Despite these benefits, reinsurance pools are often considered a last resort because pool administration can be costly, complex and difficult to understand. Managing a pool generally requires a staff of professionals to review claims, perform reserve analysis and maintain the pool's books and records.

There also is the possibility that some members will benefit at the expense of their peers. Typically, member companies are responsible individually for risk-selection and claim-payment functions. This sets the stage for potential disagreements, since some pool members will do a better job of underwriting and mitigating losses. Since losses are shared equally regardless of individual efforts, there is an inherent disincentive for superior performance. Ultimately, standards suffer for all pool members.

In addition to moral hazard, the lack of standardization and the complexity of a reinsurance pool can create problems for everyone associated with one. The lawsuits and losses that followed the failure of the recent Unicover workers' compensation pool show what can go wrong when it is not clear what risk is being assumed and ceded.

Pooling Solutions

Index-based pooling overcomes these weaknesses by using a catastrophe index instead of individual company losses. Management risk is eliminated since companies that beat the index can receive more recoveries than they had in insured losses, in effect, creating a strong incentive for good performance. Also, use of an index eliminates the need for the pool administrator to audit claims or perform reserve analysis. This makes the calculation of pool settlements relatively simple, completely transparent and objective.

A geographic risk pool provides a mechanism for homeowners insurers with different geographic concentrations of risk to swap standardized catastrophe exposures. As more members become participants in the pool, both the loss experience of the pool and of each pool member begin to look more like the industry as a whole.

Before the start of each risk period, members must agree on a definition of pool losses; that is, specific attachment points and coverage amounts in terms of the index for each geographic area covered by the pool. For example, members might decide that the pool would cover the 90th to the 99th percentile of loss experience. Establishing this range equalizes the risk of loss in different areas. Thus, the likelihood of a member receiving a pool payment would be the same regardless of where its exposures are located.

For each geographic area included in the pool, the expected index value in the pool coverage range is multiplied by each member's insured exposures to determine how much risk each member has contributed to the pool in total. Members fund pool losses in proportion to how much risk they have contributed to the pool. For example, if a company contributes exposures amounting to 10% of the pool's overall risk, it will be responsible for paying 10% of the pool's ultimate losses. After the risk period is over, published index values are used to determine the extent of pool losses, and payments to members are based on their proportionate share of property exposures in each affected area.

Since total member assessments equal the total payments made to members, the pooling mechanism is simply a way of homogenizing the geographic component of each member's catastrophe risk. While the pool does not change the expected value of a member's catastrophe losses, it helps stabilize each member's underwriting losses. By providing another source of capital to fund catastrophe losses, the pool also acts as a means of diversifying credit risk.

To test the concept, EQE International's U.S. Wind Catastrophe Model was used to estimate the range of loss experience that would constitute a coverage layer equaling the 90th to the 99th percentile of expected loss due to hurricanes in each sectional center for every coastal state from Texas through Maine. (A sectional center is a geographic area that includes all the ZIP codes that share the same first three digits.) Using actual exposure information and modeled loss information from 22 companies, it became clear that the geographic risk pool could substantially reduce the underwriting volatility of small state or regional companies.

The smallest companies in the sample tended to benefit from the geographic risk pool, with companies having less than 1% market share enjoying an average reduction in volatility of 13.2%. (See the table "Risk Pool Benefit," right.) On average, the largest insurers did not benefit much from the pool. It seems, in general, that their greater geographic spread of risk makes them much better able than smaller companies to cope with catastrophe losses. But some companies were much more or less diversified than others in their peer group, indicating that size alone does not guarantee exposure diversification.

To reduce loss volatility for any individual member, the pool must have more diversified exposures than the potential pool member. If the member's exposures are more geographically diverse than the pool exposures, participation in the pool will increase the member's loss volatility. Thus, the geographic risk pool works best when members do not have a significant amount of overlapping exposures.

To test this hypothesis, we constructed a group of state-specific companies with exposures contained within each coastal state from Texas through Maine. Exposures for each hypothetical company were compiled by using index exposures within 10 miles of the coastline.

The results showed that pooling can significantly reduce volatility for state-specific insurers by diversifying the geographic component of their catastrophe risk. (See the table "Risk Pool Benefits for State-Specific Insurers.") Although there is considerable variation in how effectively this mechanism works, average volatility was reduced by 26% for this group of companies. When compared with the insurers' loss experience in the Risk Pool Benefit table, it is clear that the geographic risk pool can offer significant benefits to companies whose exposures complement each other well.

Reinsurance Comparison

While this type of index-based pool acts only to diversify the geographic portion of a company's catastrophe risk, it differs from traditional reinsurance programs in several significant ways. The most striking difference is that pool members pay no premiums, only their share of pool losses. This insulates the pool from the reinsurance pricing cycle.

The pool also protects members against certain predetermined levels of industry loss experience, as specified by the index in certain geographic areas, rather than against a certain amount of overall company dollar losses. This implies that pool membership could be especially useful in funding the cost of events that are not significant enough to trigger recoveries under excess-of-loss reinsurance. It also implies that there will be some amount of basis risk, since most companies will not track the index exactly, creating the possibility of a mismatch between pool losses and a member's actual loss experience.

Basis risk--the random variation between the member's underlying loss experience and the pool recovery--will be an important consideration for potential pool members. Although basis risk has an expected value of zero, insurers will want to weigh the possibility of a significant recovery shortfall against the cost of other reinsurance alternatives.

Nevertheless, accepting a small amount of basis risk is worthwhile if companies can eliminate most of the operating and transaction costs that accompany other risk-management solutions. Also, the geographic risk pool uses each member's existing capital base more efficiently so no new capital is required to fund pool losses, Considering both the greater capital efficiency and the reduced administrative costs, this type of pooling may cost up to 30% to 40% less than reinsurance under even the most favorable market conditions.

Accounting Treatment

When evaluating how members should account for their participation in the pool, three factors should be kept in mind:

* members have no collateral or security deposits at stake;

* members pay no premium to participate in the pool; and

* membership does not change any member's expected net loss.

These allow the pool to overcome typical accounting issues concerning asset valuation and income and expense recognition that have been problematic for other alternative risk-transfer products. Since the pool is not based on its members' actual losses, it is likely that regulators will not view it as an insurance product. Therefore, one would expect that net amounts due to and from the pool would be recognized when members are notified of their pool positions and classified as either "other income or expense" or as "investment income or loss."

Into the Pool

Index-based pools offer homeowners insurers a new way of diversifying the geographic component of their catastrophe risks via a standardized swapping mechanism. While this may not be helpful for large insurers with well-diversified exposures, it offers an opportunity for small insurers to band together in the risk-management equivalent of a school of small fish.

Bruce Thomas is president, Insurance Indices LLC, Enfield, Conn. Xin Cao is a statistician with Hartford Steam Boiler Inspection & Insurance Co., Hartford, Conn.

                             Risk Pool Benefit
     Average      Number of  Average
     Market       Companies Volatility
      Share                 Reduction
 [less than] 1%      14       13.2%
     1% - 2%          4        5.4%
2% [greater than]     4        3.4%
                            Risk Pool Benefits
                        For State-Specific Insurers
        Volatility
 State  Reduction
  AL      27.7%
  CT      16.1%
  DE      19.1%
  FL      19.2%
  GA      34.3%
  LA      24.4%
  ME      21.2%
  MD      26.8%
  MA      21.4%
  MS      41.9%
  NH      23.0%
  NJ      27.0%
  NY      14.0%
  NC      27.4%
  RI      21.4%
  SC      23.0%
  TX      54.2%
  VA      26.5%
Average   26.0%