"If the agencies were to exit the credit business but wrap the remaining risk, as we think the current plan contemplates, our estimate is that [mortgage] yields would rise by about 50-75 bps from current levels." [Probably a permanent increase in the mortgage rate spread]
GSE Reform Implications
Over the past few weeks, there has been some renewed attention from both Congress and the President on how to reduce the dominant role the Government Sponsored Enterprises play in the US housing market. With the housing market recovery underway and the GSEs returning to relative health, the question of what to do with the US mortgage agencies has been intensifying. Balancing a reduced role versus the impact on household borrowing is a difficult exercise due to the sheer size of the agencies relative to the total stock of debt in the US. With nearly $8 trillion of guarantees and balance sheet assets, the GSEs are one of the dominant credit providers in the US economy. The agencies play this role not by making direct loans but mostly by insuring about 60% of the US mortgage market. This in turn allows agency mortgages to be transformed to risk-free assets, giving a subsidy to borrowers and leaving the GSEs and ultimately the US taxpayers on the hook for losses. There now seems to be broad consensus that the GSEs, while keeping some role in creating a liquid market for mortgage financing, need to reduce their role as credit providers. Shifting this massive credit burden from the public to the private sector will be very difficult because of the sheer size of the risk, as the plans to use the securitization market to transfer the credit risk from the GSEs to the private sector will require the private sector to take on record amounts of risky US mortgage debt. So that transition will both raise mortgage rates and also significantly affect the aggregate credit capacity of the US financial system, and it will take many years, even decades, to occur in an orderly manner. It is also worth noting this transition will create a significant liquid market for credit risk on prime mortgages, which will help to more transparently and accurately price mortgage risk in the US.
The role that the GSEs play can be broken down to credit and funding, with some overlap between the two. Currently, the GSEs guarantee all of the credit risk of conforming mortgages by "stamping" mortgage pools that are originated by the banks. These "stamped" mortgages are sold to investors at near risk-free rates. And an important consequence of underwriting all of the credit risk is that the government effectively creates a multi-trillion dollar pool of standardized AAA-assets that is extremely liquid and easy to finance. Our read of recent reform proposals is that the government wants to remove the credit risk from the agencies but maintain an extremely liquid and risk-free market that will appeal to the current buyers of US agencies. In this way, the government could potentially reduce taxpayers' exposure to losses while keeping spreads from rising too much.
What It Means For the GSEs to Exit the Credit Business but Stay in the Liquidity Business
The plan (as noted in the Corker-Warner bill and referenced in a recent Obama speech) is to use securitization to offload credit risk from the agencies while maintaining a standardized, liquid market for risk-free mortgages. Instead of guaranteeing all the mortgage risk in a pool, the new plan would involve separating mortgage pools into two sections: a first-loss bond and a senior bond. The first loss would be junior to the senior bond (i.e. the senior bond would only take losses once the first-loss bond is completely wiped out). The first-loss bonds would be sold to private investors without an agency guarantee and would likely carry risk similar to a high-yield corporate bond. The senior bonds would still be guaranteed by the US agencies, so they would essentially be risk-free for investors, and would look very similar to the current agency MBS bonds. As we note below, US prime mortgages have relatively low credit risk. Therefore, if the US agencies sell a relatively small first-loss piece it would drastically cut down the risk of losses reaching senior bonds (i.e. rising to the point that the agencies would need to pay on their guarantee of these bonds). That would amount to the agencies offloading nearly all credit risk while maintaining a liquid agency market. The current plan is to sell first-loss pieces equal to 10% of the mortgage balance, which seems reasonable as it would roughly ensure the agencies will have low losses on their guarantees, yet it will allow 90% of mortgages to be funded by the private sector at low spreads.
This plan would likely have a notable, but contained impact on mortgage rates (particularly if done over a slow enough time frame for market participants to absorb the new supply). As one indication of this negative impact on rates, currently the agencies are undercutting the private sector in the pricing of mortgage risk. The GSEs currently charge 55bps to guarantee and stamp conforming mortgage pools. At this price, banks are retaining few conforming mortgages, largely because they are finding it a better trade to originate mortgages and sell them to the agencies and in many cases buy back agency bonds. This implies that the 55 bps that the agencies charge for credit protection (which has recently been raised by 30 bps) is still less than what banks think is an appropriate level. Another data point that banks are uncompetitive is that banks are charging 25 bps higher rates for loans they can't sell to the agencies (non-conforming loans). It is interesting to note that this nonconforming spread has tightened of late, likely due to the limited supply of non-agency loans available for banks to bid on.
If the agencies were to exit the credit business but wrap the remaining risk, as we think the current plan contemplates, our estimate is that yields would rise by about 50-75 bps from current levels. This largely reflects the expense of selling a high-yielding first loss piece to private investors. We are penciling in a 10% first-loss piece but of course this could change as political and economic circumstances evolve. If the agencies wrap 95% instead of 90%, this would require less public capital and probably result in lower spreads. For perspective, if the agencies were to exit the market altogether, our rough guess is that yields would rise by roughly 150 basis points, largely based on where current jumbo RMBS deals are priced. These numbers are not meant to be overly precise, but to give a range of the choices that can be made and the likely consequences. And of course, this effect will likely be smoothed in over a number of years.