A New Theory of the Role of the GSEs in the Housing Bubble

Bill Black has an interesting new take on the role of Fannie and Freddie in the housing bubble. He sees their investment in non-prime mortgages as being driven by executive compensation, rather than a fight for market share against investment bank securitization conduits or govt affordable housing policy. The government affordable housing policy point has been repeatedly debunked (and Susan Wachter and I have a new paper that adds to this debunking via an examination of the commercial real estate bubble, where there was no government involvement whatsoever). Black is not, however, able to disprove the market share theory. What he does point to is that the GSE's involvement with nonprime mortgages was as whole loans kept in portfolio, rather than securitized (and also via purchases of MBS), which he says was a move to increase the short-term yield for the GSEs and thus maximize short-term executive compensation.

I think this is an interesting theory, but there are a few data points necessary to make it work, and I'm skeptical that they all support Black. 

According to Black, the GSEs' involvement in nonprime mortgages grew out of their earlier accounting scandals. Black sees these scandals as the result of efforts to increase executive compensation. He argues that Fannie and Freddie first tried to goose their returns by increasing the size of their whole loan portfolios and thus taking on massive interest rate risk. Fannie bet the wrong direction on interest rates and this resulted in the Fannie accounting scandal as it tried to cover up its losses. Freddie got the bet right, but then tried to set up a "cookie jar" to cover future losses in order to inflate future earnings. The SEC was having none of it, and forced out the CEOs and mandated accounting restatements. In the wake of these scandals, OFHEO (the GSE's regulator) made the GSEs limit their hedging activities and reduce the size of their portfolios.  

Black argues that this boxed in the GSEs in terms of how their could up their returns. In order to increase the returns from their portfolio, they shifted it to non-prime products, and moved more of the prime products into MBS. He further argues that the only thing that kept the GSEs from getting into really poorly underwriting non-prime products was that their risk managers and underwriters had a superior culture of responsibility and fought harder to maintain standards.

I don't doubt that executive compensation could (and probably did) play a role in the GSEs' purchases of nonprime mortgages. But Black hasn't presented a convincing story. His story relies on three pieces of evidence that he has not produced: first, that the GSEs' kept the nonprime in portfolio and securitized only prime; second, that this execution was more profitable in the short term than securitizing the nonprime mortgages; and third, that the GSEs' had a superior culture of underwriting and risk-management that kept things in check.

The first point should be reasonably easy to verify. But I'm less sure about the second, and the third, is entirely speculative. Regarding the second point, remember that the GSEs' bear the credit risk on mortgages regardless of whether they are in portfolio or securitized (via a guarantee). The only difference is whether the GSEs bear the interest rate risk, which they do for portfolio, but not for securitized loans. Unless there was a real difference in yield based on interest rate risk (which is possible) between a securitized and a portfolio nonprime loan, there'd be no reason to securitize the prime and keep the non-prime in portfolio because of higher yields. That matters because Black's argument that the GSE involvement in non-prime was because of higher yields which boosted executive compensation, and if the form of their involvement didn't track with yields, this story falls apart. Certainly the investment banks decided that securitization rather than portfolio was the better execution, and it's hard to believe that the economics would have been different for the GSEs. 

Again, an executive compensation angle is quite possible, but I think there's a lot to be said for the market share theory. Executive compensation (and tenure) depended on GSE share price, and GSE share price was dependent upon market share, which was falling. What happened, then was an insurance rate war. The GSEs are best understood as mono-insurance companies, like MBIA (or AIG): like insurance companies they have an investment portfolio (MBS or whole loans) and they issue bond insurance, in the form of the guarantees on their MBS. The only difference is that the GSEs securitize the products that they insure.

A major insurance regulation concern is preventing rate wars for market share, as a rate war will leave all competitors undercapitalized for paying out future claims. The GSEs got into a rate war with private-label MBS. They did this not by lowering the G-fee that they charge for their guarantee, but by holding the G-fee constant and lowering underwriting standards. The rate that the GSEs charge needs to be understood as a risk-adjusted rate, and while the stated rate stayed constant, the risk-adjustment did not, resulting in a lower risk-adjusted rate.

Bottom line here is that we don't have definitive evidence supporting for any of the theories of the GSEs' involvement in the housing bubble. There's strong evidence against some theories, but little affirmative evidence. The GSEs are themselves sitting on the best evidence; it would be really nice to see FHFA (or the FHFA IG) put out a definitive report on the role of the GSEs in the housing bubble.  

Comments

[mort_fin]

prepayment risk on prime was thought to be widely understood in 2005/2006, so you wouldn't expect much of a risk adjusted spread for holding the market risk on prime. prepayment risk on subprime, and especially alt-a, was not as widely understood, so you would expect a larger risk adjusted return from holding that risk on your books. add in the fact that the gse's could convince themselves that they were the smartest guys in the room with respect to prepay risk on mortgages, and you have the makings of Black's argument that they could juice their roi more by holding the market risk and credit risk on non-traditional, while holding just the credit risk on traditional, given that they had an overall portfolio cap.

but they started ramping up non-traditional product in late 2005, but i don't think the portfolio cap hit until 2006 (but i might be wrong about that - was there a temporary cap before the permanent one went in in 06?)

[Thomas Wicklund]

After having read about a dozen book length discussions of the 2007-2008 crisis (including the FCIC report and dissents), about the only thing clear to me is there is no single smoking gun cause of the crisis.

Our financial system is a complex system. As one finds with physical disasters (the Bhopal India disaster, Three Mile Island, etc) it is not a single event which causes the disaster. Rather it is a combination of events which results in the unforeseen disaster. Some combination of events occurs which was not anticipated, "independent" systems turn out to be "dependent", etc.

[Richard Davet]

can't see the forest for the trees ?

If you can't see the forest for the trees, you can't see the whole situation clearly because you're looking too closely at small details, or because you're too closely involved.

Take a step back and view the big picture. It is THE GSE BUSINESS MODEL......"FATALLY FLAWED".

Every PLAYER in the Model is working the franchise to their own end.

[Virginia]

Look at their patents. You'll be able to see the systematic changes. The alterations in MindBox are also very telling. While short term compensation was probably a side benefit... It looks like survival was the driving force. It appears the GSEs (including FHLB) knew your theory that rescission was eminent because they knew that these mortgage loan documents had not been properly assigned and transferred to the trusts. Their short-sightness and lack of sense of consequence allowed their dellusion to control better judgment. Between "we're too big to fail", "we'll never get caught because this is too convoluted" and IBG-YBG... there was no chance of reality. They also knew that 2 versions of the note existed and one version was being used for leasing, collateral, overnight trading... Follow the patents.

[mort_fin]

rechecked my history. fnma portfolio cap went in place in 2004, freddie in 2006. so did fannie start holding the crazy stuff in portfolio first? if so, would tend to confirm black's suggestion.

virginia - your post makes no sense at all. no one's raised issues of bad assignments into trusts for gse paper. it's all the pls stuff where paperwork wasn't done. one of the things the gse's remained careful about was paperwork for securitization (paperwork for foreclosure is a different matter).

[Virginia]

Mort_fin - when you get to the driving force behind the engine - you'll be into the patents. I'm not sure what you call careful - maybe well-planned? But everything - the entire scheme is patented and the GSEs developed many of these gems. Put another way they knew what was and wasn't being done with the securitzation. The securitzation is only as good as the people following the controlling documents. FYI Prof. Levitin raised the assignment issue before Congress 10/18/10... It is part of the securitization debacle - and the GSEs knowingly bought empty paper.

[bozo]

Agree with mort_fin; GSEs foresaw minimal returns on prime and was tempted with the spiked juice in subprime. This was coupled with a foolish arrogance that their understanding of and ability to hedge prime credit risk would translate to subprime.

One thing I've never understood is the existence of T-deals... I'm still waiting for a good article to explain why there's a non-Agency neg-am deal out there that's got a Freddie guarantee on the senior tranches. Neither market share chasing behaviour nor executive compensation really explains this IMHO.

[Adam Levitin]

The Freddie Mac T-series are a really interesting episode. I don't think they make Black's case, however--they were done too early. They started in the late 1990s and ended in 2001 or so. They weren't being done during the bubble. The T-Series was Freddie's attempt to grab the monolines' business by leveraging the lower funding cost via the implicit guarantee. So at most this shows that Freddie was greedy earlier. Was that a move to generate false alpha to maximize bonuses? Maybe, or maybe it was just a dumb move. Were the T-Series subject to the usual G-fee or was there bespoke pricing? If the latter, then it might be real alpha.

I'm not totally sure why Freddie stopped the T-Series, but my guess is that they concluded that it was more profitable to invest in the AAA-tranches with of the non-Agency deals that had a monoline wrap than to provide the wrap themselves. Basically the investment portfolio was the better execution than the insurance arm.

I didn't realize that any of the T-Series were neg-am. Do you have a particular deal's name or CUSIP?

[Adam Levitin]

I was wrong--having done a bit of research, it seems that Freddie didn't stop the T-series in 2001 or even 2003. The first deal was in November 1995. There have been 82 T-Series deals to date--the majority (58 or 71%) were closed before 2004. It seems that there was some change in 2003 in terms of what Freddie would wrap--it became less aggressive on T-Series wraps and more aggressive in what it bought for its MBS investment portfolio.

There's an old American Banker piece that makes the really interesting point that the T-Series opened the door for major banks to get involved in subprime lending--they didn't have to worry about warehouse risk, as they could securitize any kind of crap if it had a Freddie wrap. And it probably helped legitimize subprime securitization with investors. Yet Fannie doesn't seem to have joined the party. Wonder why?

If you want to see a neg-am deal, look at T-065, which is a Freddie wrap on a bunch of WaMu mortgages with neg am limits up to 125%. The latest T-series deal is T-082, a BoA deal from April 2009.

[bozo]

The Fannie Mae Grantor Trust (FNGT) deals are probably the closest thing, though these seem like true 'experimental' programs compared to Freddie's T-deals. These are also pretty interesting: I can't think of any good reason why, for example, Fannie would be wrapping deals collateralized by FHA/VA loans.

Would you have the link or title of that American Banker article?