After nearly 30 months of conservatorship, the new House and the Administration are expected to signal just how they plan to reform Fannie Mae (OTCQB:FNMA) and Freddie Mac (OTCQB:FMCC), the housing government-sponsored enterprises (GSEs). Given how dominant Fannie and Freddie are in terms of market share today, reform of these institutions will largely determine the future of the $11 trillion market for residential mortgage finance.
On Wednesday, the House Financial Services Committee will hold its first hearing of the 112th Congress on the subject of “ GSE reform: Immediate Steps to Protect Taxpayers and End the Bailouts.” In addition, there are grumblings that the Administration will finally release (either this week or next) its long awaited reform plan for overhauling the nation's housing finance policy.
The analytical challenge presented by reform is that the most egregious excesses of the previous GSE model are not what precipitated all the taxpayer losses. For example, the first instinct of many reformers would be to ensure that the GSEs (or their successors) are never again able to build big mortgage portfolios. The second instinct would probably be to strictly limit the mortgages that would qualify for purchase (or guarantee) by the new GSEs.
While both make sense and would have made for sound reform in 2005, focusing on these two issues now more or less ignores the big lessons from the 2006 - 2010 market meltdown. Of the GSEs’ combined $226 billion in losses (pdf), over $166 billion (73%) came from the guarantee business. The investment portfolio accounts for just $21 billion (9%) of losses. Had the investment portfolios been eliminated in 2005, the GSEs would have still suffered losses from guaranteed mortgages that would have wiped out their capital base several times over. It would be disingenuous to therefore claim that just eliminating the investment portfolios now would protect taxpayers.
But, in seeing that over 70% of the losses came from mortgage guarantees, one might reasonably ask why wouldn’t better limits on the types of mortgages that are accepted be the right way to go to protect future taxpayers. This may be true in the short-term, but it’s important to acknowledge that even good steps in this direction now would not address the fundamental problem of how to accurately price the insurance – or what amounts to the cost of providing a government guarantee.
Put another way – for many, the challenge ahead seems to be designing a strategy to maintain a government guarantee for mortgage credit risk while attenuating some of the more egregious elements of the old GSE model. The problem with operating under this framework is that it was the mispricing that arose from the government guarantee (itself) that really turned out to be the big source of the taxpayer losses.
The management of both GSEs used their ability to issue implicitly guaranteed debt to build massive portfolios of the same mortgage-backed securities (MBS) they issued. Once a pool of mortgages was converted into GSE-guaranteed MBS notes, there was no need for them to then issue additional debt to repurchase the guaranteed MBS.
As argued by Fed Chairman Greenspan and others, these portfolios served “no credible purpose” aside from serving as a profit center for GSE shareholders and management. The profits came from the huge gap between the yields on mortgages and the interest rate Fannie and Freddie paid on their own borrowings, which was just slightly greater than Treasury rates thanks to government sponsorship.
Worse than doing nothing to reduce mortgage rates or improving liquidity, the large investment portfolios created massive risks. For every $100 of mortgages added to the portfolio, the GSEs committed just $3 of equity capital, borrowing the remainder. Even this $3 share was suspect because the GSEs could count deferred tax assets and “temporary” reductions in the market value of securities as capital. The risk was that a sudden increase in interest rates could wipe out the GSEs’ capital, or a sudden fall in interest rates could set off a wave of refinancing, causing the interest income on the new mortgages to fall below the cost of existing borrowings.
The future of housing finance should not involve the GSEs (or their successors) building up big portfolios. Yet, removing them from the current equation would limit GSE operations to buying and guaranteeing mortgages, converting the mortgage payments to guaranteed cash flows from MBS notes, and standardizing MBS notes to enhance investor acceptance and market liquidity. In this world, the investors in GSEs (or the government) would capture the guarantee fees on the mortgages and pass through the rest of the mortgage payments to investors in the MBS. Yes, the risk of the portfolios and excess net interest margins captured by shareholders and management would be eliminated, but the mortgage guarantee (and corresponding pricing problem) would survive intact.
The Pricing Problem
Increasing it looks like would-be reformers will focus on implementing strict limits on the mortgages that would qualify for purchase by the new GSEs. This seems to be precisely what the Obama Administration has in mind, as the maximum mortgage that would qualify for purchase would be $625,000, down from the largely non-binding current cap of $729,000. In addition, the reforms would include underwriting guidelines that would limit the kinds of mortgages that could be purchased so that the GSEs could not accumulate hundreds of billions of dollars of exposure to Alt-A and subprime loans. By eliminating the investment portfolios and limiting the types of mortgages that would qualify, these reforms would (1) ensure the government guarantees could not be exploited for windfall profits and lavish executive compensation; (2) limit the subsidies to middle-to-upper-middle class homeowners; and (3) deny high-risk mortgages from inclusion in MBS pools to further discourage their origination.
As explained at the beginning, this would have been considered a solid reform agenda back in 2005. But, the data and lessons from 2006 on demonstrate that this issue is more complex – and the debate has to end with a discussion on the more fundamental issue of how to price the insurance or what’s known as the mortgage guarantee.
When the GSEs rapidly expanded acquisitions of subprime and Alt-A mortgages in 2005-2007, the conforming mortgage limit was $417,000. The average mortgage purchased by the GSEs is about $217,000 (pdf) (page 84). While it makes sense to limit the government involvement to more modest mortgages, this is about avoiding subsidies for the “rich,” not really about protecting taxpayers.
Perhaps the best way to understand the pricing problem is to review the credit risk of some of the best mortgages (based on FICO, downpayment, etc). To unpack the data, a good place to start is Fannie Mae’s monthly funding summary for November 2010 (pdf). It shows that 4.5% of all loans were 90 days past due. Based on the 2010-Q3 credit supplement (pdf), 28.2% of all loans had some non-traditional feature and these loans had an 11% delinquency rate. This means that the 71.8% of traditional, prime, 30-year fixed rate mortgages must have a serious delinquency rate of 1.95%.
While this seems low, consider that the GSEs charge only 20 basis points to guarantee these mortgages. If we assume that in a default situation, there is a 30% loss to account for foreclosure costs and decline in property value, this means that credit losses on the highest quality mortgages will be about 58 basis points, or three times as much as the fee charged to guarantee them. (It is important to remember that today’s delinquency rate is down by one-fifth from the peak of 5.6% recorded in 2010, so losses during the worst months of the crisis would have exceeded income by an even larger magnitude.) Finally, this calculation is very conservative, as losses on prime interest-only loans and prime loans with an LTV greater than 90 with mortgage insurance, in addition to all subprime and Alt-A loans are excluded.
How much would the government entity have to charge to cover these costs? Perhaps about 75 basis points to 1 percentage point – once overhead costs and risk premiums are included, or more than three-times the rates it normally charges. The bottom line is that this analysis reveals just how difficult (if not impossible) it is to solve the pricing problem once and for all.
Existing GSE Business
Finally, reformers have to deal with the fact that the GSEs’ remaining book of business is hugely underwater. According to Fannie Mae, it’s current net worth is -$130 billion (pdf) (page 65) which means its assets are worth $130 billion less than its liabilities and it could not possibly fund itself without continued taxpayer backing. Freddie Mac’s liabilities exceed its assets by a somewhat less startling $56 billion (table 19.6). In total, taxpayers are on the hook for $186 billion in addition to the $148 billion already injected into the GSEs. The $186 billion hole is what it would cost to liquidate them right now, which is probably higher than the cost of a gradual wind-down that allowed the mortgages to pay off the liabilities. So, while the ultimate losses may very well end up lower, this $186 billion figure is still the best current measure of the ongoing credit support that will be required by Fannie and Freddie. As a default on Fannie and Freddie debt or MBS would be nearly as traumatic as a U.S. government default, reform cannot be about cutting off taxpayer funding for past mistakes, as distasteful as that may be. Reform has to be oriented towards protecting taxpayers from future debacles.
In the end, it appears as though the most promising path for Congress is to commit to a credible strategy that puts the GSEs in receivership and liquidate their operations over a 5 to 7 year period. Any shortfalls would be covered by taxpayers so no creditor losses anything in a wind-down or is tempted to sell their securities. In the future, Congress would keep Federal Housing Administration (FHA) mortgages available for borrowers under certain income and mortgage loan thresholds and leave the rest of the market to the private sector. The likely result is higher mortgage costs, as the old guarantees would now be paid for by mortgage borrowers instead of taxpayers. If Congress wants to offset some of this cost increase, it does have options – even though some may not be popular with consumers. For example, mortgages could be made fully recourse to the borrower so homeowners’ assets would be at risk in the event of default. (For more information on this, take a look at Canada’s housing system.) Many would view recourse mortgages as anti-consumer and difficult to implement, but the change would help offset some of the cost increase. If Congress is particularly worried, it could also explore ways to explicitly subsidize low-income borrowers through mechanisms like interest rate swaps (See Charles Calomiris (pdf) and Raj Date; Josh Rosner has other ideas regarding the mortgage interest rate deduction that are worthy of review).
As the Administration and Congress weigh these issues, they will likely wrestle with the fundamental issue that continued credit support for housing – at the current magnitude, at least – is probably not in the long-run interests of the economy. After all, housing is a form of consumption and its continued subsidization diverts capital from other more productive uses.
Despite many of the aforementioned points, the most likely outcome of this reform debate is that the government still stays involved – and in a big way. Mortgage rates will be kept artificially low because credit risk is under priced and subsidized by taxpayers. Sadly, it’s these same taxpayers who’ll likely also be called on again someday to write large checks to cover for the losses of whatever institutions succeed Fannie and Freddie.