To borrow a phrase, a crisis as deep as the 2007-2008 collapse of U.S. housing finance is a terrible thing to waste. Yet, nearly eight years after investors shunned their debt, Fannie Mae and Freddie Mac remain in federal conservatorship. And there is no end in sight to the government’s dominant role in housing finance: securitizations by the GSEs and federal agencies still accounted for nearly 70% of originations in 2015 (with qualifying loan-to-value ratios as high as 97%). Despite this extensive government intervention in mortgage finance, the U.S. home ownership rate fell to 63.6% last year, its lowest level since 1966.
To say that U.S. housing finance is both inefficient and risky seems a dramatic understatement.
Two years ago, we explained why the U.S. federal government would not exit mortgage finance, despite a range of policy failures (see here). The constituencies supporting federal involvement are simply too numerous and politically influential to allow for withdrawal. There are the GSEs’ own employees and managers; real estate brokers; mortgage originators and servicers; builders and construction workers; and, of course, the current and future homeowners who can borrow up to $625,500 (the current limit for conforming loans in high-cost areas) at a below-market rate. Even homeowners without a mortgage benefit from the government’s debt subsidies through their impact on house prices.
Not surprisingly, the share of U.S. residential mortgages financed by the GSEs or by federal agency- and GSE-sponsored mortgage pools continues to set new highs, accounting for 61.5% of the nearly $10 trillion currently outstanding (see chart). At the same time, the private-label mortgage securities market remains virtually nonexistent: annual issuance peaked at more than $1 trillion in 2005 and 2006, but never rebounded following the crisis, totaling a mere $14 billion in 2015. Despite extensive efforts to overcome the numerous conflicts of interest that caused the private-label market to collapse, investors still lack confidence that private issuers and servicers will act in their interest, while the issuers and servicers fret over the potential costs—reputational and otherwise—of resolving delinquent mortgages (for an excellent update, see here).
Share of U.S. Residential Mortgages Held as Assets by GSEs or by Agency- or GSE-backed Mortgage Pools (Percent), 1990-201
On the margin, however, reforms are underway in three areas: (1) pricing of the GSE guarantees; (2) winding down of GSE mortgage holdings; and (3) transferring the risk from GSE guarantees to private investors. While progress remains limited so far, the trend is important. For example, recent research highlights that underpriced GSE guarantees are associated with more and riskier mortgage debt, while the resulting increase in financial system leverage makes it more crisis-prone and raises fiscal risks as well. In theory, progress on these three fronts will reduce the burden on taxpayers from a future collapse of housing finance and—by making the price of mortgage debt more commensurate with its risk—reduce threats to the financial system as a whole.
The most important and obvious example of progress is the first. In 2011 and again in 2012, the GSEs’ regulator—the Federal Housing Finance Authority (FHFA)—called on the GSEs to raise these “g-fees” (see history here). More recently, the FHFA left the overall fee unchanged, but proposed more risk-sensitive fees, such as higher charges on cash-out refinances and investment properties. As a result of these changes, the fees charged by the GSEs for new guarantees have more than doubled to the 50- to 60-basis-point range in 2015 from less than 25 basis points in the latter half of 2009 (see page 20 of the Urban Institute’s April 2016 monthly chartbook on housing finance).
The second area of progress is the wind-down of Fannie and Freddie’s own holdings of mortgages. In August 2012, Treasury announced plans to quicken the pace of reductions from 10% annually to 15% annually. As a result, the GSEs’ mortgage portfolios have plunged from peaks of more than $800 billion each to about $340 billion each in 2016, with further declines expected going forward (page 19 of the Urban Institute’s monthly chartbook).
Finally, beginning in 2013, the FHFA required the GSEs to sell a portion of the credit risk that they assume when providing guarantees to private investors. Since these “credit risk transfer” (CRT) programs began, Fannie and Freddie have issued securities that removed 21% and 28%, respectively, of their total book of business (page 21 of the Urban Institute’s monthly chartbook). The GSEs also have developed risk-sharing arrangements with originators and experimented with selling risk to re-insurers. As of 2016, the regulatory mandate for CRTs—recorded in the FHFA’s annual scorecard for assessing the GSEs—has risen to at least 90% of newly guaranteed single-family mortgages in major categories (including mortgages with terms greater than 20 years and loan-to-value ratios above 60%).
Over time, a sustained rise of CRTs would re-position the GSEs as insurers of catastrophic mortgage risk, with the private sector bearing the first losses and (presumably) charging a commensurate risk premium for the service. This outcome is consistent with two housing finance reform proposals that have languished in the Senate—the Corker-Warner (CW) bill of 2013 and the Johnson-Crapo (JC) bill of 2014—and with a recent private proposal to consolidate the two GSEs into one catastrophic insurer. The idea of limiting the government guarantee to catastrophic risk was one of the approaches to housing finance reform highlighted in the Treasury’s 2011 report to Congress. Importantly, both the CW and JC bills call for the private sector to cover the first 10% of mortgage losses—a level likely sufficient to weather the storm of even the awful 2007 vintage of GSE mortgages without taxpayer losses.
Of course, one ought to ask why there should be any involvement of the federal government in housing finance in the first place. Even the provision of catastrophe insurance could lead to a substantial misallocation of resources and expose taxpayers to sizable risk if it is not correctly priced—just think of the inadequate fees charged for federal deposit insurance in the years before the financial crisis. Moreover, if the federal government’s goal is to raise home ownership, its efforts have been remarkably ineffective. And the goal itself is at least questionable: few people doubt the prosperity of Austria and Germany even though their home ownership rates are low by advanced-economy standards (57% and 53%, respectively).
However, the case for a complete U.S. federal government exit from housing finance suffers from a problem of time consistency. Given its interventionist history extending back to the 1938 creation of Fannie Mae, we do not see how the U.S. government or the Congress could credibly claim that they would sit by idly as mortgage finance and housing values collapsed. That means that homebuyers, builders and financiers will all continue to act as if a backstop will always be there, regardless of what anyone might say or do. Consequently, the best way to provide proper incentives for the housing market today is for the government to charge mortgage lenders a fee commensurate with the cost of providing catastrophic insurance, rather than to provide it implicitly without any fee (Swagel has made a similar argument here).
Is the catastrophic insurance model viable in practice? Suppose the GSEs eventually were to sell off the most of their guarantee risk to the private sector, leaving them as the backstop in a catastrophe. We don’t really know whether there will be sufficient private capital to absorb the first losses associated with U.S.-style housing finance (namely, the country’s idiosyncratic reliance on 30-year fixed-rate mortgages) without either making the financial system riskier or raising the cost of housing finance beyond what homebuyers and policymakers will tolerate.
For example, if the CRT securities issued by the catastrophic risk insurer are acquired by highly leveraged intermediaries, then we may simply re-play the crisis of 2007-2009 at some future date when housing prices fall and mortgage losses undermine the capitalization of the intermediation system as a whole. So far, capital requirements have limited the willingness of banks to acquire GSE CRT securities, but if leveraged shadow banks become big investors in such instruments, regulators will merely have shifted, rather than reduced, the risks to the financial system. The problem is that—much like subprime debt—these CRT instruments are inherently systemic: they pay a stable return every year until, in a crisis, they all collapse at once. Only non-leveraged investors with deep pockets can bear this aggregate risk without weakening the financial system as a whole. (We are reminded of the discussions surrounding contingent capital and total loss-absorbing capacity for banks: the system cannot be made safer by having one bank hold the subordinated debt issued by another.)
Another challenge is for the GSEs to use the information coming from market CRT spreads to price their g-fees properly. In periods of diminished private risk appetite (“risk off”), those market spreads could surge. Passing on this higher cost of bearing guarantee risk in the form of a higher g-fee would diminish mortgage lending and homeownership, an unpopular outcome. Yet, failure to do so would again expose the taxpayer to potentially substantial downside risk. In theory, a stable g-fee that reflects the equilibrium cost across “risk-on” and “risk-off” periods might limit housing cyclicality. In practice, however, no one knows what that equilibrium cost is, and the temptation to underprice it in good times could be powerful. Again, the experience with FDIC insurance pricing comes to mind.
So, what’s the bottom line? Seven years after the financial crisis wound down, U.S. housing finance remains fundamentally the same as it was in 2009. That said, at the margin, important changes underway are gradually trimming taxpayer guarantees. And, while there’s still a long way to travel along this road—and there are many uncertainties about how far we can or will go—these changes could become the foundation for a broad reform of housing finance.
Acknowledgments: We thank our friends and colleagues, Stijn van Nieuwerburgh and Larry White, for very helpful conversations and for a range of background materials regarding GSEs and housing finance reform.