Friday, January 4, 2008

When is a policy really insurance?

An operational definition of insurance is that it is

  • the benefit provided by a particular kind of indemnity contract, called an insurance policy;
  • that is issued by one of several kinds of legal entities (stock insurance company, mutual insurance company, reciprocal, or Lloyd's syndicate, for example), any of which may be called an insurer;
  • in which the insurer promises to pay on behalf of or to indemnify another party, called a policyholder or insured;
  • that protects the insured against loss caused by those perils subject to the indemnity in exchange for consideration known as an insurance premium.

In recent years this kind of operational definition proved inadequate as a result of contracts that had the form but not the substance of insurance. The essence of insurance is the transfer of risk from the insured to one or more insurers. How much risk a contract actually transfers proved to be at the heart of the controversy.

This issue arose most clearly in reinsurance, where the use of Financial Reinsurance to reengineer insurer balance sheets under US GAAP became fashionable during the 1980s. The accounting profession raised serious concerns about the use of reinsurance in which little if any actual risk was transferred, and went on to address the issue in FAS 113, cited above. While on its face, FAS 113 is limited to accounting for reinsurance transactions, the guidance it contains is generally conceded to be equally applicable to US GAAP accounting for insurance transactions executed by commercial enterprises.

[edit] Does the contract contain adequate risk transfer?

FAS 113 contains two tests, called the '9a and 9b tests,' that collectively require that a contract create a reasonable chance of a significant loss to the underwriter for it to be considered insurance.

9. Indemnification of the ceding enterprise against loss or liability relating to insurance risk in reinsurance of short-duration contracts requires both of the following, unless the condition in paragraph 11 is met:

a. The reinsurer assumes significant insurance risk under the reinsured portions of the underlying insurance contracts.

b. It is reasonably possible that the reinsurer may realize a significant loss from the transaction.

Paragraph 10 of FAS 113 makes clear that the 9a and 9b tests are based on comparing the present value of all costs to the PV of all income streams. FAS gives no guidance on the choice of a discount rate on which to base such a calculation, other than to say that all outcomes tested should use the same rate.

Statement of Statutory Accounting Principles ("SSAP") 62, issued by the National Association of Insurance Commissioners, applies to so-called 'statutory accounting' - the accounting for insurance enterprises to conform with regulation. Paragraph 12 of SSAP 62 is nearly identical to the FAS 113 test, while paragraph 14, which is otherwise very similar to paragraph 10 of FAS 113, additionally contains a justification for the use of a single fixed rate for discounting purposes. The choice of an "reasonable and appropriate" discount rate is left as a matter of judgment.

[edit] Is there a brightline test?

Neither FAS 113 nor SAP 62 defines the terms reasonable or significant. Ideally, one would like to be able to substitute values for both terms. It would be much simpler if one could apply a test of an X percent chance of a loss of Y percent or greater. Such tests have been proposed, including one famously attributed to an SEC official who is said to have opined in an after lunch talk that a 10 percent chance of a 10 percent loss was sufficient to establish both reasonableness and significance. Indeed, many insurers and reinsurers still apply this 10/10" test as a benchmark for risk transfer testing.

It should be obvious that an attempt to use any numerical rule such as the 10/10 test will quickly run into problems. Implicit in the test is keeping the 10/10 that either are upper bonds for the comment made by the SEC official therefore, the rest of this paragraph doesn't apply. Suppose a contract has a 1 percent chance of a 10,000 percent loss? It should be reasonably self-evident that such a contract is insurance, but it fails one half of the 10/10 test.

It does not appear that any brightline test of reasonableness nor significance can be constructed.

Excess of loss contracts, like those commonly used for umbrella and general liability insurance, or to insure against property losses, will typically have a low ratio of premium paid to maximum loss recoverable. This ratio (expressed as a percentage), commonly called the rate on line for historical reasons related to underwriting practices at Lloyd's of London, will typically be low for contracts that contain reasonably self-evident risk transfer. As the ratio increases to approximate the present value of the limit of coverage, self-evidence decreases and disappears.

Contracts with low rates on line may survive modest features that limit the amount of risk transferred. As rates on line increase, such risk limiting features become increasingly important.

[edit] "Safe harbor" exemptions

The analysis of reasonableness and significance is an estimate of the probability of different gain or loss outcomes under different loss scenarios. It takes time and resources to perform the analysis, which constitutes a burden without value where risk transfer is reasonably self-evident.

Guidance exists for insurers and reinsurers, whose CEO's and CFO's attest annually as to the reinsurance agreements their firms undertake. The American Academy of Actuaries, for instance, identifies three categories of contract as outside the requirement of attestation:

  • Inactive contracts. If there are no premiums due nor losses payable, and the insurer is not taking any credit for the reinsurance, determining risk transfer is irrelevant.
  • Pre-1994 contracts. The attestation requirement only applies to contracts that were entered into, renewed or amended on or after 1 January 1994. Prior contracts need not be analyzed.
  • Where risk transfer is "reasonably self-evident."



No comments: