Bloomberg has a delightful story today about a new JPMorgan RMBS transaction, its first non-agency deal since the crisis. Specifically about this:
The bonds are made riskier by the New York-based bank and other originators of the mortgages offering weaker promises to repurchase misrepresented loans than those on similar deals, Fitch Ratings said today in an e-mailed report. Lenders and bond sponsors have been seeking to trim potential liabilities in such deals as the market revives after suffering billions of dollars of losses from debt sold before the collapse in home prices.
The value of the so-called representations and warranties in the JPMorgan transaction is “significantly diluted by qualifying and conditional language that substantially reduces lender loan breach liability and the inclusion of sunsets for a number of provisions including fraud,” New York-based Fitch analysts including Roelof Slump wrote in the presale report.
So naturally the deal is limited to an Aa rating, as it would be at Moody's based on those sort of rep and warranty weaknesses, right? Errr not so much:
The classes of the deal expected to receive top credit ratings carried loss buffers of 7.4 percent as Fitch said it adjusted its analysis to reflect the greater investor dangers created by the weaker contracts, according to the report.
So 92.6% of the deal will be AAA rated at Fitch and Kroll, the other rating agency on the deal. Here's the cap structure from Kroll's report:
The rep and warranty provisions are pleasingly mechanical: per the Kroll report, a third-party reviewer, PentAlpha, will review certain loans (those that are 120 days delinquent, where the servcer stops making advances, or that liquidate at a loss), test certain reps and warranties depending on the reason for review, and
determine, in its sole judgment, whether the facts giving rise to a R&W breach materially increased the loan's credit risk, resulted or are expected to result in a material increased loss in connection with a loan's liquidation, or materially impaired the Issuer's ability to enforce payment obligations on the loan, in each case with consideration given to whether the loan nevertheless complied with underwriting guidelines, whether the loan would have been similarly at risk without the defects, and whether certain life events (including the death, disability, life-threatening illness, divorce, or loss of job of the borrower) were the cause of the actual or projected loss. ... The specificity of the review process provides a straightforward and transparent mechanism for breach reviews. By its nature, however, it also limits the scope of the review process such that potential breaches may be less likely to be discovered than under a less prescribed process.
Also the reps sunset after 3-5 years.
We talked a little bit about how Moody's came to the - fairly reasonable - conclusion that things like this drastically reduce JPMorgan's risk of buying back these loans. Or, put another way, that they drastically increase the risk that RMBS buyers are buying loans based on fake credit scores, made-up incomes, fraudulent appraisals, etc. Got a mortgage based on a fraudulent appraisal, lost your job, and defaulted? Well, wasn't that life event - the job loss - the cause of the loss? Then you're off the hook for the fraudulent appraisal. Got your 790 FICO included in the pool's 769 weighted average FICO, when it was really a 720? That's okay - as long as a 720 FICO satisfied the underwriting requirements, it doesn't matter that it was mis-reported.
So this is all a horrible fraud! Nah, not really. This is a pool of jumbo loans to high-credit-quality borrowers, almost half of them originated by JPMorgan itself, all of them independently reviewed, with average/max LTV of 65% and 80%. That parade of fraudulent underwriting horribles I laid out in the last paragraph seems pretty unlikely. You can see why Kroll and Fitch don't care that much.
And you can see why JPMorgan is pushing on the reps and warranties with this pool. You sound silly if you get too worried about them, so you don't. Do a few of these, and the next subprime RMBS securitization anyone does, in like 2026 or whatever,1 will have the same documentation. "It's just market standard," they'll say.2
This is a good deal for the banks, who after all are selling these mortgages because they want to pass on the credit risk - they don't want them lingering around, as quasi-covered-bond putback claims, if any trouble hits. That's not evil or fraud or whatever, necessarily: a main goal in selling off these mortgages is gaining certainty that they won't come back, and investing up-front in third-party reviews and significant credit enhancement and so forth is a great trade-off in exchange for getting more of that certainty.
And it's a good deal for Fitch and Kroll, for the sort of obvious reason that they'll give this deal AAA ratings, and get paid for it. It's not so hot for Moody's, who, according to their own bluster, won't. That's a risky position for Moody's to take: their bluster, presumably, was meant to persuade investors that they were right, and that no one should treat a mortgage-backed security as AAA with weak reps and warranties. Perhaps they'll win, and investors won't be that into deals rated mainly by Fitch and Kroll. But if they lose - if investors shrug and these terms become market standard - then Moody's is frozen out of a lot of deals.
Of course Moody's could change its mind. But since the DOJ's S&P lawsuit, that seems a lot harder. Remember that the DOJ accused S&P of, essentially, changing its ratings criteria so they'd be more in line with Moody's, for competitive reasons. Moody's - who were more aggressive to begin with - avoided trouble.
For ratings agencies, it's better to be wrong first: you just look dumb. Moving toward other people's wrong answers - or, never mind wrong, just moving toward more liberal bestowals of Aaa ratings - is what gets you in trouble with the Justice Department. Moody's has staked out its Aa maximum for these sorts of deals, and now it's kind of stuck there.
1.No, obvs, short memories, in like September.
2.And the fact that here it worked because it was traded off against other things - high credit quality, low LTVs, etc. - won't mean much. Documentation terms are hard to quantify and so hard to trade off: once it's market standard, it's standard everywhere, even in places where it was never meant to go.