Understanding financial strength ratings and default rates

Financial strength ratings are widely used, but many industry executives admit to being foggy on many of the nuances underpinning them. Stuart Shipperlee, head of analytics at Litmus Analysis, explains what they really mean and how they relate to default rates.

The term financial strength rating (FSR) was adopted for insurance ratings, as related to the credit risk taken by policyholders, by Standard & Poor’s (S&P) in the mid-1990s. Previously it had used the term claims-paying ability (CPA).

In the late 1990s AM Best also adopted the term. Other rating agencies now also tend to also use this, or add ‘insurer’ to it.

An FSR is a forward-looking opinion as to the future ability of a re/insurer to meet its financial obligations to policyholders. It is typically assigned only to a legal entity that issues insurance policies, with the exception of the Lloyd’s market FSR.

Unlike debt ratings, FSRs refer only to ‘ability to pay’ and not ‘willingness to pay’, reflecting the difficulty that can occur in agreeing the validity of claims.

Only an opinion

Ratings are opinions about the future based on historical information and future projections; they are forecasts. Agencies make it very clear that they are intended as no more than forward-looking opinions.

For ratings users this means that the agency is describing its view of the most likely (not the only) outcome for the future financial strength of the re/insurer through its rating.

While international re/insurance market participants have become very familiar with ratings symbols (‘AAA’, ‘A-’, etc) over the last 20 years, the symbols themselves obviously don’t actually define the degree of credit risk being taken—for example, how risky is a rating of ‘A-’? To help address this issue, S&P, Moody’s and Fitch have long published annual studies of the frequency of default observed for different rating levels over time across all industries, to show for example how often in practice does an ‘A-’ rated entity actually default.

These cover all the entities globally that they rate, examining how many companies within different ratings categories have defaulted by the end of the year.


Crunching the numbers

For global corporates (including insurers) S&P’s data show, on average, that after one year roughly one in 1,000 ‘A-’ rated firms defaults. Over 15 years we see that roughly one in 40 typically defaults (a 2.5 percent probability). The 15-year number does not mean they are still rated ‘A-’ when they default, just that they were rated ‘A-’ at the start of the first year of the 15-year period.

Even down at the ‘CCC’ rating level, near-term survival is the most likely outcome, with one-year ‘CCC’ default risk being around 27 percent (ie, nearly three out of four ‘CCC’ rated issuers do not default in the following year).

These data also give us a partial way of benchmarking rating outcomes to the Solvency Capital Requirement (SCR) published under Solvency II requirements. A 100 percent SCR equates to an expectation of a one-in-200 default rate, roughly equivalent to an S&P ‘BB+’ default rate, although in practice there are lots of differences between SCRs and ratings that market participants should be aware of.

The data also tell us that we should not view the fact of default by an ‘A-’ rated firm within a year of having that rating as unexpected. It should be rare (in that an individual one-in-1,000-year event is rare), but if we have 500 ‘A-’ rated firms then it would happen, on average, once every two years. It is like cat risk: any given case is rare, but cumulatively they are quite common.

We might expect that default-causing events impact several companies at once, therefore we won’t actually see a regular ‘once every two years’ default pattern from our 500 ‘A-’ rated firms, but rather periodic clusters of defaults. Nonetheless, the data say we should not be surprised when it happens.

Default data can be a useful guide for re/insurance market participants when interpreting re/insurer FSRs. But there are some important caveats.

The vast majority of default examples in the data reflect non-financial corporations defaulting on bonds. Typically, that is something that is easily observable, happens on a given date, is common enough to provide useful data and is not impacted by regulatory action. None of these characteristics tends to apply to rated re/insurers defaulting on policyholder obligations.

Rated re/insurers will usually be operating in domiciles where regulators seek to intervene before they default on policyholder obligations. AM Best uses the term ‘impaired’ for this and publishes ‘impairment rates’ by rating category for the US.

However, since impairment can happen to re/insurers who will never actually default, impairment rates can be far higher than default rates, especially if the earlier phase, precautionary, regulatory interventions are counted in the data.

That has been Best’s approach to date but it is now adding research which focuses on where a re/insurer impairment can indeed seem to be a good proxy for default.