Two Cheers for Fannie and Freddie Synthetic CDOs
My ears perk up whenever I hear the musical words "synthetic collateralized debt offering". (Bill Bratton and I did write the paper on history of these crazy things, after all....) So, it was with interest that I read a Wall Street Journal editorial decrying Fannie Mae and Freddie Mac's use of synthetic CDOs to transfer credit risk on mortgages to the private market through the STACR and Connecticut Avenue programs. Unfortunately, the WSJ piece does not accurately describe what Fannie and Freddie are doing and fails entirely to understand why unfiltered private capital is a recipe for financial instability in housing markets.
To review, a synthetic CDO is a securitization of a set of credit default swaps (CDS). The CDO entity sells a bunch of CDS protection on some assets, in this case Fannie and Freddie mortgages or MBS. Fannie and Freddie make periodic protection payments (insurance premia) until and unless there is a default under the CDS, in which case the CDO pays Fannie and Freddie. The CDO gets the money to pay Fannie and Freddie by selling structured (that is senior-subordinate tranched) securities, the proceeds of which it invests in very safe Treasury securities. So the credit risk on the mortgages or MBS is transferred synthetically to the investors' in the CDO securities without the mortgages or MBS ever leaving Fannie and Freddie's books.
As it happens, only a small slice of the synthetic CDOs are being offered to investors--some of the mezzanine tranches. The notional senior and junior tranches are not being sold. What this means is that the WSJ's claim that the GSEs are using synthetic CDOs for $800 billion of their $4 trillion book of business is incredibly misleading. The notional amount of the mortgages covered by the CDOs might be $800 billion, but private capital is backing only about $25 billion ($12.7B in STACR deals and $12B in Connecticut Avenue deals). The private capital in the synthetic CDO deals accounts for less than 0.625% of the credit risk held by the GSEs. In other words, this isn't that big of a deal, at least yet.
Still, synthetic CDOs have a long association with financial scandal, from Michael Milken to Enron to Abacus. That's enough to make one immediately suspicious. But synthetic CDOs aren't an inherently bad product. The problem is that they have often been used as a vehicle for regulatory capital arbitrage. Is there something to worry about here?
I don't immediately see a problem. It's not clear to me how, if at all, these synthetic CDOs are counted against the GSE's regulatory capital requirements, but at this point, they aren't big enough to worry about. (What we should worry about is what would happen if there's another housing downturn as the Fed can't drop interest rates any meaningful amount, and the GSEs' capital lifeline from Treasury might not be bottomless.)
In fact, I think the synthetic CDOs are probably a very good development for the GSEs. The fundamental policy tension in housing finance is between the relative problems and benefits of public and private capital, neither of which is ideal. There are obvious problems with public provision of housing finance--it can easily become politicized and result in the public incurring underpriced risk. This was a problem in the run-up to 2008, even if it did not (sorry WSJ) cause the financial crisis. (Note the irony that the WSJ wants to tar the GSEs with the evil buzzword of synthetic CDOs, which previously were only a private-market phenomenon.) The problems in GSE underwriting were far surpassed by those in the private-label market, and losses on GSE loans never came anywhere close to those on private-label securitized loans. But we did have a real political economy problem with the GSEs.
Private provision of housing finance is also problematic, and perhaps more so. Private capital markets aren't capable of providing anything more than a fraction of the capital needed for US housing markets without causing a wholesale collapse in housing prices--there just isn't the appetite for $10 trillion in credit risk at anything close to current mortgage rates. Moreover, private capital is likely to result in much less stable housing finance markets.
Private capital markets want risk-based pricing of mortgages. While we generally think risk-based pricing is a good thing, it isn't in housing finance markets because a major risk factor is location, and there is significant geospatial and serial correlation of housing prices. The price of my house is linked to those of nearby houses. This is a unique feature of housing (and real estate) as an asset class. The result is that if my neighbor is riskier (by whatever measure), it will affect my home price, and vice-versa. (The intermediate assumption here is that most homes are purchased on credit and the availability of credit will therefore affect home prices.)
Risk-based mortgage capital will mean tremendous geographic disparities in capital availability. That was the situation in the private, pre-GSE housing market, as there was much less capital available in the South and West. (We'll end up with effective red-lining too, with no capital in poor neighborhoods.)
Risk-based mortgage capital also means much more susceptibility to local bubbles and booms. Illinois has budget woes? Watch Illinois housing prices fall (which will in turn further exacerbate budget problems, etc.). The same is true on the way up. We will get self-reinforcing local booms and busts. That's a real problem given how much of household savings are tied up in housing as an asset class, and how much our state and local tax base (and thus government services) depends on real estate taxes.
Private mortgage capital is a recipe for financial instability, as is public mortgage capital. The challenge we face in housing finance is how to yoke together the best aspects of both without the downsides of either. Serious thinking about housing finance needs to address this without being shaped by ideological priors that favor either private or public capital. We are already living in a hybrid world (private markets bear interest rate risk, not credit risk for the GSEs), and any future housing finance system is going to be some sort of a hybrid.
The synthetic CDOs present a potential (and I emphasize potential) way of bringing in the benefits of private capital without the downsides. The synthetic CDO can essentially act as a filter against too-much risk-based pricing. Let's assume (and this detail isn't clear to me) that the synthetic CDOs are based on a blended bunch of GSE-held mortgages, without geographic information being made available to investors (I believe this is the case, but am not 100% sure). If so, there would not be geographically-risk-based pricing of the CDOs, and we'd get some of the benefits of private capital--namely lack of public assumption of risk, without the downside financial market instability. Yes, we'd lose some of the market discipline that comes with private capital, but the part we'd lose would be the part we don't really want to have. There's a limit to the synthetic CDOs--I don't think they will ever handle the majority of the GSEs' credit risk, but they're a step forward. So a tentative two cheers for Fannie and Freddie CDOs.
Update: Only one cheer for the Fannie/Freddie CDOs. It turns out that they are releasing some geographic information about the loan pools: state and (if I'm reading correctly) MSA-level data. That's not good. The pools are blended, however, so I don't think we've got much geographic risk-based pricing... yet. But it suggests that the doors are open for that to come in the future. (To be sure, some investors might want geographically diversified pools...but not all will.)
Also it's worthwhile noting something else. The WSJ suggests that it's better to have private capital loan directly than go through Fannie and Freddie. I've explained above why that may not be the case. But the question it begs is why is the private capital flowing here, and not to direct mortgage lending (which is certainly possible) or to private-label securitization? The WSJ suggests that it is because of an implicit government guaranty. I think that's nonsense. If there are credit losses that go up to the mezzanine tranches, the investors here will take their loses. They aren't getting bailed out because they total dollar figure at risk (which is more than likely losses) is too small to matter in the economy.
Instead, I think the answer relates to (1) Fannie and Freddie economies of scale and specialization, (2) Fannie and Freddie oversight of servicers (which while not great is better than in the private-label world, and (3) Fannie and Freddie underwriting standards. In particular, #2 and #3 combine in terms of enforcement of representations and warranties about the securitized loans. Fannie and Freddie have generally been better at enforcing repurchases than trustees for private-label securitizations. (I'm not saying that Fannie and Freddie have been as vigorous as they could be...) That matters to investors who want to be sure that they are incurring only those risks that they chose to incur. Simply put, unless reps and warranties are honored, securitization does not work. And that's the lesson with the failure of the private-label market. The reps and warranties were often breached in the first place, and then too often not enforced thereafter. That will kill a market.