Aug 13

Was the NBER Correct on RMBS/CMBS Ratings?


A rather fascinating, detailed mathematical treatise is out from the National Bureau of Economic Research.  Its basic conclusions are:

(1)    Using PIMCO/Blackrock methodology for RMBS/CMBS ratings resulted in a huge decrease in required risk capital (over $15B less as an industry, including $3B combined for Met and TIAA).

(2)    This new system only considers current (expected) losses and does not provide any buffer for future losses.

Well, I guess this new system did exactly what it was designed to do. 

Using NRSRO ratings on securities that are seldom (sometimes, if ever) reviewed for updated ratings cannot be called a ‘best practice’ from either the insurers’ or the regulators’ viewpoints.  Thus, a different way was considered

And, let’s not forget that although that capital savings may look like a regulatory giveaway to, primarily, the largest insurers (and a proportionately larger giveaway to the usually lesser capitalized largest life insurers), there is another factor to consider.  In order to reduce the carrying value of a security, the insurer must run a charge through its statutory capital directly (in the case of P/C companies) or indirectly (through the Asset Valuation Reserve).  So, to some extent capital was already impaired so there should be some reduction in required capital (no double counting).

Of course, some insurers may be gaming the system by writing down a security only to the highest price that would provide the desired NAIC rating (this may very well be quite true in a few cases).  And, some insurers may be gaming the system by perhaps using some excess AVR reserves to cushion those write downs.

But, generally, if played fairly, the current system can be a better indicator of required capital (assuming reasonably worthwhile modeling skills at PIMCO at Blackrock) than using slow to change NRSRO ratings.  If played fairly…

As for the second point, once again the new system did what it was designed to do.  It was designed to include the impact of losses expected today (and it is updated annually).  However, I do not believe it was designed to provide a buffer for future unexpected losses.  That, of course, is the idea behind annual updating.  As better information is obtained, models are revised, recalibrated and recalculated.  It is as simple as that.

However, I must congratulate the researches at the NBER.  They are tackling a rather controversial topic.  Not because it relates to a minority of holdings at insurers; and not because it tackles one of the requirements of Dodd-Frank legislation (to find a better alternative to NRSRO ratings).  It is because as NRSRO ratings continue to be pummeled in both legislative and other corners, the insurance regulators at the NAIC may move toward assigning NAIC ratings based upon the pricing of, for example, corporate bonds in the future.  After all, why should the same bond rated ‘A’ by the NRSRO’s but held at 85 (when rates were high) by one insurer have the same severity of loss as one held at 105 (when rates were low)?

Makes one want to reconsider what the true credit risks are on an insurer’s books – from the perspective of the regulators, rating agencies, financial analysts, actuaries and even the insurers themselves.