S&P’s ratings of Mapfre and Generali have been among the most contentious major insurance group ratings of recent years, due to the impact of S&P’s sovereign ratings for Spain and Italy respectively.
While some other ratings were clearly and specifically dragged down by their direct exposure to sovereign debt, for these two giants S&P’s concern – at least in part – was a more nuanced and less easily defined concept around the wider impact on them of the default risk of their largest country exposure.
Since the asset and business risks associated with a given country of operation are already captured in the underlying rating analysis, the application of a further ‘sovereign’ constraint on the rating appeared to be ‘double-counting’.
The agency’s recently updated set of criteria for this has now thrown more light on its approach. The outcome is that Mapfre’s rating has been restored to ‘nearly’ its stand-alone (Group Credit Profile; GCP) level for the financial strength ratings of its core carriers (‘A-,’ vs a ‘GCP’ of ‘A’). However, Generali remains on credit-watch negative with a downside rating risk of two notches (from ‘A’- to ‘BBB’ for its core insurance carriers) to the Italian sovereign rating level.
S&P’s updated approach is to run a ‘stressed’ scenario of its capital model for a rated insurer as if the sovereign had defaulted. This includes substantially reduced values for all major domestic asset classes not just those directly linked to sovereign debt, along with a liquidity stress test.
If the insurer then still achieves the minimum level of regulatory capital after this stress, a rating above the sovereign level can be assigned. If not the sovereign rating becomes the de-facto rating ceiling.
As we note below, while there is a clear credit rating logic to this approach in terms of the relative level of ratings, we would suggest that it also implies a significant inconsistency in the agency’s wider approach to the impact of stressed scenarios on insurer ratings overall.
In summary the process for the adjustment of the rating level of a group (the GCP) to reflect the sovereign rating is as follows. If the group’s GCP is above that of the sovereign rating AND it can survive the sovereign default stress as above, then it can be assigned a rating at its GCP level up to 4 rating notches above the sovereign (for non-life) or 2 notches above (for life). However, if it does NOT survive the stress test then it cannot be rated higher than the sovereign since failure of the sovereign would imply failure of the group (or at least the breach of its regulatory capital requirement since financial strength ratings rate to the risk of regulatory intervention).
Looked at from a purely credit rating comparison point of view this application of a sovereign rating stress seems to make perfect sense, but it begs an important question. Ratings are not ‘worst case’ or even ‘reasonable worst case’ scenario analyses (if they were, many high profile re/insurance groups would be rated far lower)*.
In addition the failure of a currently BBB rated sovereign is a 1 in 500 year event; by most standards a worst case test.
So why apply that degree of stress to an insurance group’s rating for a ‘rare event’ sovereign default scenario but not for other rare financial events? At that degree of rarity profound hits to asset prices and market turmoil are easily envisaged (the South Sea Bubble, Dutch Tulips, the Wall Street Crash etc.) within even the highest rated sovereign environment (and the cause can have nothing to do directly with sovereign creditworthiness).
Such rare event stresses are also easily envisaged on the liability side of the balance sheet (for example the emergence of the ’next asbestosis’ cause of slow-developing latent disease). Yet an ‘A’ rated casualty insurer does not get its capital stressed for a 1 in 500 year adverse development scenario.
The answer to that, we presume, is that it is simply too counter-intuitive for the agency to rate a group above the sovereign rating when it believes the group would not maintain its required minimum capital in the event of a sovereign default. However for those groups impacted, this is imposing a case specific stress test on them not seen in other parts of S&P’s analysis.
Stuart Shipperlee, Analytical MD
March 24th 2014
*It should be noted that reviews of catastrophe exposures as part of re/insurer ratings, while often couched in rare event terms (1 in 250 etc.), are describing the rarity of the scale of that individual loss event to the re/insurer, not the rarity of the re/insurer experiencing such a loss across its entire book.