After the first round of results and feedback sessions regarding the proposed Financial Condition Reporting (FCR) standards for the short-term insurance industry, several players are reeling from the results. Many have asked: “How can this be? We have always viewed ourselves as conservative and well capitalised”. The answer to this and other similar questions may be found not only in the proposed new structure, but also in the application of the structure.
The results as they have been presented were generated from data gleaned from the STAR returns to the Financial Services Board (FSB). These returns were never designed to incorporate data to the required level of detail for this exercise. This implies the use of several assumptions.
The two principal concerns concentrate on the choice of distribution and the incorporation of non-proportional reinsurance. Other minor considerations will also be raised during the course of the discussion.
The proposed level of confidence at which the total capital base should be set is the 99.5th percentile of the total ultimate loss distribution of the insurer. At this high percentile, distributions are notoriously volatile, with minimal variation in parameter values introducing significant fluctuations in results. This renders the choice of distribution and parameters critical in the calculation of the capital requirements.
With the potential impact of estimation errors, complete, detailed and accurate base-data would be a prerequisite in the modelling process. Therefore, use of the STAR returns as base-data, would increase this risk, thus reducing the statistical significance of the results materially.
The second main concern focuses on the incorporation of non-proportional reinsurance. According to feedback received, the basis for inclusion was on a Rand-for-Rand basis. This implies that such reinsurances are priced at the 99.5th percentile. This would render such reinsurances too expensive. The result of this allowance is that the capital requirements are effectively double-counted to the extent that the premium differs from the 99.5th percentile for the specific risk assumed.
Ancillary to this is the fact that the approach of the FCR standards is volatility-based. Since non-proportional reinsurance is used to reduce underwriting volatility, its use serves to significantly alter the ultimate loss distribution from which the 99.5th percentile is calculated as a capital requirement, linking this concern back to the estimation errors alluded to previously.
The impact of these concerns is most noticeable in the cell-captive and captive market results. This is partially due to the greatly increased use of non-proportional reinsurance in most programmes written through these entities. The resulting double-counting of capital requirements would ultimately result in an increase in premiums to policy-holders without a commensurate increase in protection.
Other concerns which came to the fore during presentations and discussions include the risk proxies used in the calculations, allowances for risk aggregations, lack of incorporation of programme structures and maximum loss exposures and the existence of differentiated internationally accepted capital requirements for cell-captives and captives
In the study, significant attention was paid to reduction in capital requirements due to increased premium volumes. The rationale is that as business volumes increase, stability is created, reducing the relative capital requirements. This is intuitively correct. However, in some of the results, there seems to be cases where the converse applies. This is intuitively wrong and may indicate that the parameters selected do not describe the risk adequately and that there are specific features confounded in the data used which cannot be sufficiently isolated with the approach implemented.
The sophistication of these adjustments contrast with the lack of allowance for programme structures, including maximum losses and risk aggregations. The limitation of the information contained in the STAR returns is the principal reason for this, but excluding these features would further increase the estimation errors.
Furthermore, the 99.5th percentile is not the only international standard. In the global captive environment, it is commonly accepted that the capital requirements are not as onerous as for the conventional market. A captive/cell captive is a financial risk management tool available to a corporate entity to assist in the effective management of their risks. This contrasts with a pay-away arrangement as found in the conventional market.
The use of such vehicles should thus be viewed in a business context rather than a pure conventional insurance perspective. This does not imply poor financial health or imprudent financial regulation. It simply highlights the considerable differences that exist between the conventional insurance industry and the captive/cell-captive industry.
Of the more commonly used international captive standards, the range of capital requirements vary between the 75th and the 95th percentile, dependent on several factors such as the type of risks, the maximum exposures and financial health of the parent company.
In summary, considering the monetary values involved in this type of exercise, it cannot be approached from a minimalist perspective. A possibly preferred approach would be to conduct a representative sample throughout the industry and completely analyse several programmes to isolate all relevant factors and relationships. These results should then be compared to the information in the STAR returns to ensure that the elements not reflected in the returns are accurately allowed for.
The final formulae can be based on the STAR returns information only, with some adjustments and additions to the standard format, once the relationships between actual risk-based data and the returns are fully understood and isolated. Proper treatment of unique financial, corporate and risk structures can then be adequately allowed for. In addition, industry involvement should be an integral part of the exercise from the outset.