A Template for MBS Settlements and How Safety-and-Soundness Regulati...

Over the past couple of years, the Massachusetts Attorney General's office has reached settlements with a number of major banks regarding mortgage securitization. These settlements has received very little notice in the press, but I think they provide a real template for future AG settelements and are worth examining. 

1. The Mass AG has gotten a good deal of principal reduction as part of these settlements. Of the $102M paid by Morgan Stanley for its New Century dealings, $52M went to principal reduction. Similarly, check out the remedy chart (below) for the settlement with Goldman Sachs. Granted, it only covered around 700 loans, but it should be the template for any AG settlement.  I haven't attempted to calculate what the cost would be for doing that nationwide (it'd be huge), but we'd have a real estate market that was immediately cleared.  It might require the recapitalization of our biggest financial institutions, but I've got to think that there are plenty of funds to invest in them once their balance sheets are cleared up. 

2. The Morgan Stanley settlement is a particularly fine illustration of the involvement of Wall Street in subprime mortgage lending. Morgan Stanley never made any of the subprime loans itself. But it provided the warehouse line of credit for New Century (including wet funding when New Century was about to fail and could no longer able to gin up any liquidity itself), purchased the loans for securitization, and underwrote the securitizations. Absent the backer of an institution like Morgan Stanley, it's questionable whether New Century would have even been in business.  

3. The Morgan Stanley settlement is a settlement that helps both homeowners and investors. A critical part of the Mass AG's allegation is that New Century/Morgan Stanley were securitizing loans that didn't meet the underwriting standards the investors were told to expect. Compare that with the proposed investor BoA settlement. More $ to investors, but nothing that clearly assists homeowners (and why should we expect it, since it is an investor settlement). I hope that the NY AG intervenes in the settlement to demand that in addition to the payout to investors that there also be real relief for homeowners (see the Goldman settlement chart above...). 

4. There's absolutely no reason to think that Morgan Stanley or Goldman Sachs were in any way unique in their practices. What's amazing, then, is not these settlements in Massachusetts, but that we haven't seen this elsewhere. Part of the reason, I believe is because of National Bank Act preemption. The Massachusetts AG was able to get the dirt (particularly in the form of loan tapes) on Morgan Stanley and Goldman Sachs because they weren't able to raise preemption as a defense. But for preeemption, I'd go dollars to donuts that we'd have seen similar settlements with BoA, JPM, Citi, and Wells.

So, where is the OCC with all of this? Shouldn't they be doing the same sort of investigation? Instead, the only thing I've heard out of the OCC is the claim that national banks didn't really do subprime lending. Well sure. Maybe the national bank itself didn't, but they surely did it through their non-bank subsidiaries and affiliates (remember Watters v. Wachovia?) and provided the warehouse lines and securitization conduits that made the whole system possible. 

All of this goes to a simple point: safety-and-soundness regulation is fundamentally incompatible with law enforcement. Prudential regulators aren't interested in law enforcement.  They're interested in preserving quiet and stability and that sometimes means papering over problems and looking the other way.

Consider a very present example of the conflict beteen safety-and-soundness regulation and law enforcement. The FDIC, as receiver for WaMu, sue LPS and CoreLogic over appraisal practices on WaMu loans. The FDIC is alleging $150M+ in damages on just around 1% of the loans it has sampled in a roughly 200,000 loan pool. If that's correct, we're talking about $15B in actual damages. Of course there's no way that the FDIC actually goes for the jugular on this, although if the complaint is to be believed, the appraisal fraud is staggering. FDIC's duties as receiver, however, conflict with its duties as a prudential regulator. A $15B judgment destroys LPS. But the problem is that LPS is too big to fail. LPS is a central node for a huge percentage of mortgage foreclosure work and provides the standard mortgage servicing software platform. It's in a position of a critical single-sourced supplier for the entire automotive industry. If it fails, the whole foreclosure system grinds to a halt. So what is FDIC supposed to do? Clearly settle the case for enough money to save face on the receivership front, but not cripple LPS.

In the bankruptcy context, it's inconceivable that this sort of conflict of interest would be permitted, but we allow it for bank receivership.  This suggests that whenever we have the next round of financial regulatory reform, that receivership duties be separated from prudential regulation. Or maybe our willingness to accept this sort of conflict tells us something else about bank regulation, namely that we tolerate different standards for banks because they are quasi-utilities. But if that's true, then there are a lot of implications that no one seems to eager to explore.

Put differently, we want to believe that we have a free market financial system, but in fact we have a system that exists and works only because of major government intervention. If government is going to bear the risk in the financial system, it would seem obvious that it should exercise much more control over the system (yes, that raises all sorts of other problems). But it's clear we aren't going to go there. Instead, we're going to pretent that Dodd-Frank means that the government doesn't bear the risk anymore and therefore continue with a system of private gain facilitated by public intervention. When seen on this level, the changes wrought by Dodd-Frank all look somewhat irrelevant.