Risk analysis question is about more than just disclosure: Samir Shah

This is a commentary by Samir Shah, ex-Head of Insurance Capital Markets at AIG, on our article titled ‘Disclosure of risk analysis in ILS offerings benefits investors: SIFMA speakers’, published March 11, 2016, in which he explains that the question should not be whether to disclose, but rather how to make disclosure efficient and reliable.

March 13, 2016

The question has never been about whether or not to provide risk analysis to investors but rather about what is the most efficient and reliable way to provide that information.

The current practice of providing risk analysis performed by an independent party in the offering docs was established when cat bonds were primarily indexed to industry losses – and it made sense in that context. However, against the current practice of indemnity triggers, the traditional approach reveals some important weaknesses.

It may seem as if a third party is indeed providing an independent review; however, let’s take a closer look. For indemnity triggers, the two biggest components of the risk analysis are the data and the model. The cedant typically provides detailed exposure data to the modeler who accepts it as a given, other than performing some high-level checks for reasonability and consistency. There is no independent validation of whether the data was complete or whether it appropriately captures the coverage terms of the insurance policies in the subject portfolio. The modeler can only say what model it used (and specify the assumptions, or “switches”). This is a huge gap if the objective is independently developed risk analysis.

The other issue is the cost inefficiency of having to rerun the data through the same exact model (with the same “switches”) that the company has already run for its own benefit. The modeler also has to charge for the liability that it takes on (not an insignificant portion of their fees) for providing analysis that is part of the offering docs, even though it is being indemnified by the cedant.

There are some alternatives that can improve the reliability and/or the cost.

One approach is to have the risk analysis provided to the rating agency as part of its own rating – a more logical place for risk analysis – rather than include it in the offering docs. This eliminates the liability to the modeler and therefore a large portion of its fees. (AIG’s 2015 Compass Re II bond used this approach). Also, several rating agencies already gather this information as part of their rating of the company and are therefore in a better position to validate the results, particularly the completeness and accuracy of the data, if they also rated the cat bond.

An alternative approach is to allow the cedant to perform the risk analysis and have that work be audited by an independent party. Quite frankly, cedants are in the best position to provide the most reliable analysis of the risks underlying their portfolio.

What’s needed is governance around this to control for errors, omissions and fraud. There is robust governance over this work internally by risk and audit functions and externally by auditors as the results are reported to underwriting, management, board, investors, rating agencies and regulators and are relied on for risk selection and pricing, risk and capital management, risk-adjusted performance analysis, external rating, statutory capital requirements, etc. Of course, the disclosure itself is also subject to strict securities regulations.

Keep in mind that under the current approach, the third-party modeling is only validating that a certain model was used, not the data. It’s easy to audit the use of a specific model to achieve the same level of comfort for investors at much lower cost.

More importantly, this approach would allow the third-party to also review the exposure data more closely. Again, several rating agencies that already gather this information are in a good position to provide a second level of validation if they were also rating the cat bond.

This “audit” approach is similar to how companies typically convey information to investors for equity and debt offerings. Companies do not provide all of their financial transactions to the auditor and have them independently develop the financial statements. They produce their own analysis that is audited by a reliable independent party.

MBS and its variants offer yet another alternative. Here, the sponsors typically provide the exposure data (loan tape) to investors and allow investors to develop their own risk analysis – there is no risk analysis provided in the offering docs. However, unlike insurance, the underlying risk factors are macroeconomic and investors are generally skilled in modeling these factors since they are common to most asset classes in the capital markets. For insurance, if investors had to do their own risk analysis, it’s likely that the market would remain a domain of specialists. Yet, it may not constrain growth in ILS investments by those specialists – it hasn’t for MBS.

It’s clear that there are alternatives worth exploring. The SIFMA discussion seems to have focused only on whether or not to provide the risk analysis in the offering docs. It would be valuable to have a broader discussion on ways to improve both the reliability and efficiency of providing risk analysis to investors, and to standardize it to allow investors to easily compare across securities. I am sure that we can come up with better alternatives.