Amid a crisis in confidence in AAA-rated asset-backed securities and growing turmoil in the asset-backed commercial paper markets, we saw for the first time prime mortgage borrowers--whose performance did not warrant a retrenchment in lending--swept into the subprime-initiated liquidity vortex. With the ability to fund much more important than the quality and performance of assets, even the highest quality originators like Thornburg Mortgage were forced to liquidate assets at substantial discounts. In the absence of a securitization exit, non-agency lenders began to originate only those loans they were willing to hold in portfolio, pushing the benchmark spread between the prime jumbo mortgage rate and the agency conforming rate to over 90 basis points--nearly six standard deviations above its historical mean. While this spread has retraced approximately one-third of its initial widening in response to the Federal Reserve rate cut on Sept. 18, illiquid conditions persist in the non-agency sector and issuance volumes have dropped precipitously.
From a housing market perspective, the pullback in prime non-agency lending undermines a market already weakened by the virtual elimination of subprime and Alt-A lending earlier in the year. While Fed actions have eased the liquidity crisis somewhat, we believe there is little the Fed can do to prevent the housing market from going through a significant correction. Indeed, subsequent to the Fed cut on Sept. 18, 30-year benchmark mortgage rates actually increased by 10 basis points. It is important to note that the illiquid housing environment is clearly evident in actual MBS prepayment rates where this summer--for the first time in the history of our database--we observed no discernible seasonal increase in MBS prepayment rates.
The fallout for MBS markets: a massive issuance migration from non-agency to agency MBS
With primary issuance in the non-agency sector on life support, the migration in issuance market share from the non-agency sector to the agency MBS sector (Fannie Mae, Freddie Mac, Ginnie Mae) will be immediate. We expect the agency market share of MBS issuance to increase from its all-time low of 44% ($905 billion) in 2006 to its all-time high of 86% ($1.2 trillion) by the end of 2008. Note that GNMA is expected to exhibit the largest increase in market share, moving from just 4% ($82 billion) in 2006 to a projected 12% ($169 billion) in 2008 on the back of its GNMA I, GNMA II and now FHASecure2 programs. In contrast, the staggering declines in non-agency issuance are compounded by a decline in both projected total issuance volumes and market share in 2008. For example, we expect subprime issuance to drop 90% from a near record $449 billion in 2006 to just $43 billion in 2008. While the shift in market shares is extreme, the final result will be an MBS landscape that looks very similar to its composition in 2001--well before the boom in affordability lending.
So what are the major implications for the mortgage market going forward?
1. A large scale reallocation of risk in the capital markets from credit to prepayment/interest rate risk.
This migration in issuance means that over time there will be a commensurate shift in investor exposure from credit to prepayment and interest rate risks (there is effectively no credit component of risk in FNMA, FHLMC, and GNMA securities). Although these markets are likely to experience supply pressures in the short-run we believe over time more and more mortgage investors will be crowded into this market looking for incremental yield in pass-through, interest only and structured product form. This will likely support agency structured product issuance going forward.
2. MBS convexity hedging will increase, but only modestly.
Although a sharp increase in agency fixed-rate MBS issuance might suggest that there will be a commensurate increase in MBS convexity hedging, higher volatility and wider swap spreads, we believe these effects will be modest at best for the following reasons:
* Duration swings in the mortgage market are dominated by interest rate changes, particularly when the market is near its refinancing elbow--a point of maximum prepayment sensitivity. Today mortgage rates are 70 basis points above that elbow and we expect to enter a period of extremely low prepayment volatility as the housing correction deepens.
* The two key MBS delta hedgers, the GSEs and servicers, are less active today. Portfolio caps have limited the ability of the GSES to grow their portfolios while servicers have less need to hedge prepayment risk in today's low prepayment volatility environment. Moreover, servicers are expected to utilize more discount collateral as a hedging vehicle given its liquidity and availability.
* Duration added from adjustable-rate mortgage to fixed-rate refinancing and slowing prepayments is relatively small (approximately $100 billion 10-year Treasury equivalents each). Moreover, this duration is added on a very incremental basis, not all at one time as in a rate move.
Although we expect volatility to remain elevated, we attribute most of this increase to greater uncertainties around economic, housing, and credit fundamentals rather than from increased MBS delta hedging.
3. Non-agency reform.
Major non-agency market reforms will be a necessary precursor to any future growth of the non-agency market. The three key areas of reform center around underwriting, brokers, and rating agencies. Most of the underwriting reform has already taken place, as evidenced by the much improved collateral attributes of the late 2007 vintage. Broker reform has centered on the creation of a national registration database as a means of creating some level of accountability for brokers. Finally, rating agency reform has focused on establishing greater transparency and consistency in ratings across asset classes and rating agency compensation structure.
Are subprime resets the next shoe to fall?
Much of the recent focus on capital hill has been on the issue of subprime resets. Although $20$30 billion subprime loans have been resetting every month, until now most of these borrowers have been able to refinance because of accumulated equity. That all begins to change rapidly as the share of subprime borrowers at reset with combined loan-to-value ratio above 90% increases from 12% to 25% in October to over 40% by the spring of next year assuming stable home prices. If we assume home prices fall 3% over this period, that share increases to above 50%. The high LTV share is rising because of both weak home prices and a growing share of 80/20 loans at reset. Therefore, even though many of these borrowers have demonstrated an ability to make monthly payments, they have no refinancing alternatives at reset in an era of much tighter underwriting standards including few new loans originated with LTVs above 90%. With the political spotlight shining brightly on this issue, three potential outlets have developed for subprime borrowers facing reset:
* GSE Expanded Approval Programs;
* GNMA's FHASecure Refinance Program;
* Servicer loan modification.
We expect servicers to filter their loans through these options in the coming year bifurcating remaining borrowers between those able to qualify for a loan modification and those expected to default. Of the three options, we expect the FHASecure program to offer the most immediate relief to subprime borrowers facing reset.
A perfect storm for record slow prepayments
We think one of the enduring legacies of the 2007 liquidity event will be a record slowdown in MBS prepayment speeds. Our first indication of the impact of these events came this summer when for the first time in the history of our prepayment database there was virtually no seasonal increase in prepayment rates. With winter fast approaching, we think absolute speeds have further to go. More importantly, we believe 2008 is shaping up to be a continuation of this theme culminating in the slowest year for MBS prepayments on record. The combination of declining home prices and tighter underwriting guidelines across the entire credit spectrum of borrowers has shut down the repeating cycle of cash-out refinancing and housing turnover that propelled the entire prepayment curve to unprecedented levels in 2005.
Moreover, the current illiquid state of the housing market that will likely persist through all of 2008 is expected to pressure baseline housing turnover to levels not seen in over 12 years. In general, while prepayment models have been adjusted to reflect these conditions, 2008 will mark a period with no statistical precedent that will likely require modelers to make multiple downward adjustments to their models through the year.
While much has been made of the impact that slower speeds will have on mortgage duration extension, we believe these concerns are overstated. We estimate that weak housing conditions and slow prepayments could add approximately $100 billion in 10-year Treasury equivalents to total market durations. While this is not a trivial number, it is important to note this duration is added in incremental steps, in contrast to rate-driven negative convexity that happens immediately. The one caveat to our scenario is if the broader economy begins to slip into recession, driving long-term rates down 75 to 100 basis points. This would push the market through the refinancing elbow and trigger a major refinancing event shortening aggregate MBS duration by $850 billion 10-year Treasury equivalents.
 |
 |
| Dale Westhoff | V.S. Srinivasan |
This week's Learning Curve was written by Dale Westhoff and V.S. Srinivasan analysts at Bear Stearns. It was originally published in a short-term prepay estimates publication. The research analysts who prepared this research report hereby certify that the views expressed in this research report accurately reflect the analysts' personal views about the subject companies and their securities. The research analysts also certify that the analysts have not been, are not, and will not be receiving direct or indirect compensation for expressing the specific recommendation(s) or view(s) in this report.