Last Monday Fannie Mae and Freddie Mac announced plans to begin disclosing six new loan/borrower attributes in their MBS pools. The additional pool level disclosure items were recommended in a study conducted by a joint task force
of the Department of the Treasury, the Securities & Exchange Commission (SEC) and the Office of Federal Housing Enterprise Oversight (OFHEO).
The six items are credit score, original LTV, loan purpose, property type, occupancy status and servicer information. After the release of the study, both GSEs announced plans to disclose information on the six new MBS items recommended in the report. Industry observers hailed the event.
"We view this as an extremely important development for MBS investors and the mortgage market," wrote Dale Westhoff, senior managing director at Bear Stearns, in a report titled Agency MBS Disclosure: A New Era Dawns. "The refinancing waves during the last two years have served to highlight the need for additional disclosure as market participants have increasingly recognized loan characteristics that reduce prepayment risk."
He added that the evolution of risk-based pricing and automated underwriting systems have allowed GSEs to target the underserved segments of the borrower population, as well as to broaden their underwriting guidelines to include a more diverse borrower universe in agency pools. "Full disclosure allows us to understand and model the complete risk profile of the borrowers backing MBS pools," Westhoff said.
Fannie Mae stated in a release that effective for April 2003 issuances, it would start disclosing the six items in its MBS prospectus supplements. The Agency also stated that it would change the timing on several existing disclosures from post-settlement to pre-settlement. These elements include: distribution of loan coupons within pool by quartile; distribution of remaining terms to maturity within pool by quartile; weighted average loan age as well as distribution of loan age within pool by quartile; weighted average loan term as well as distribution of original loan terms within pool by quartile; distribution of original loan sizes within pool by quartile; seller identity; and a table of loan origination years.
Freddie Mac, on the other hand, has not stated the specific time when it would start the disclosures, but representatives from the Agency said that three to six months is probably a good estimate of how long it might take. However, unlike Fannie, Freddie has committed to making the disclosures on existing as well as new pools.
"We've already begun the process of making system adjustments, and we'll have a better idea within the next few weeks of when we'll start disclosing the information," said Sharon McHale, spokesperson at Freddie. "It's just that we want to do it in a systematic and orderly way. We particularly want to make sure that the transition is orderly so that our lenders and market investors make the transition without any difficulties."
Impact on prepayment
Street analysts said that the access to more information on agency pools is a crucial and beneficial development in prepayment analysis. Such access would, for example, provide a seamless analysis between the agency and non-agency sector. According to Bear's Westhoff, "The expansion of agency underwriting criteria across credit, loan size and other loan attributes has blurred the lines between the agency and non-agency sector."
For instance, he explained that in the case of conforming balance Alt-A loans that find their way into agency pools, once the transition into the agency sector is made, analysts lose the important information about the non-standard characteristics of the borrower; therefore the ability to analyze the pools that contain those loans is compromised.
"Having access to this full disclosure now will allow us to analyze those pools just like we do in the non-agency sector, where we have access to full attribute information," said Westhoff. He added that with the access to the information, valuation results on agency pools would be consistent with similar loans that are in the non-agency sector.
Andrew Davidson, president of the Andrew Davidson & Co., Inc., said he welcomes any additional information provided by the GSEs. However, providing the information on a pool-level basis is not necessarily the best form of disclosure, he noted. The ideal form would be to get the loan-level information every month. This level of disclosure is currently available in the jumbo market.
"I'm glad that this staff report came out and that the GSEs are responding to the request to provide more data," said Davidson. "However, even more data would be welcome. This is still not bringing the agency MBS market up to the level of detail available to the non-agency market, and I would consider getting to that level a minimum goal. I hope that Fannie and Freddie would look to become proactive in trying to provide that type of information to the investor community."
Though market participants are not really clear as to which form the disclosure would take, such as which exact tapes it would go on, early indications show that the information would come in the form of weighted averages and quartile distributions.
The problem with pool-level disclosure, Davidson said, is that there is still a mix of loans within a particular pool, so it does not really make the use of pool-level data for forecasting prepayments particularly easy. There is the additional factor of making sure that those same variables are available to users when they run prepayment models. Further, before these numbers could be used, it will take some time to build up some historical data.
"One of the key features of mortgages is that many of these factors create non-linear effects," said Davidson. "So to the extent that pools are large and have many different types of loans, pool-level information is still not the ideal form."
For instance, Davidson explained that if a pool that had some borrowers with 90% LTV and some with 70% LTVs, the weighted average of the pool might be 80% LTV. However, the performance of a pool that had a combination of borrowers with 90% LTV and 70% LTV is very different from a pool where all the mortgages have 80% LTV. So even though this quartile information is provided, it would be very difficult to incorporate it into standard modeling. "This is something we are going to look at to see how we are going to take advantage of that distribution data," said Davidson.
He noted that the additional data to be provided by the GSEs would serve as enhancements to his firm's prepayment models. "We have models for jumbo loans and subprime loans that already include these types of variables. With the additional information, we are going to have the opportunity to extend some of the analysis we've been able to do in the areas where loan level data is available to the agency sector."
Other researchers said that with the new information available, agency prepayment models will be much more precise across a wide range of credit and loan attributes. Bear's Westhoff stated that for the very first time mortgage investors would have an almost complete understanding of the risk profile of the mortgages in agency pools.
"I think that it's extremely important that we incorporate this new information into our prepayment models," he said. "The best way we can serve our customers is to build more accurate prepayment models with this information. They can then use them for valuation purposes and hopefully have a much better understanding of the risks that they already own as well as make better relative value decisions going forward."
Issues with the TBA market
Much of the hesitation about having the GSEs divulge more pool-level information is centered around the TBA market, where observers say the increased access to information - which usually leads to adverse selection in TBAs - would hurt the liquidity of this sector.
In the joint report, the three agencies stated that "MBS market participants also agreed, almost without exception, that the significant changes in disclosure did not affect the highly liquid nature of the GSE and Ginnie Mae pass-through and to-be-announced markets, and MBS markets generally operate reliably and efficiently."
Analysts from Merrill Lynch agreed that increased GSE disclosure has not impacted the liquidity of TBAs. They added that in terms of widened spreads on TBAs, it is fairly hard to prove that the adverse selection phenomenon has widened spreads in this sector considering all the other variables that impact mortgage spreads.
However, some buysiders say that having the TBA market so thoroughly picked through is going to make it more difficult for money managers, especially those managing versus the Index, to make money.
"The Lehman Index is priced like a TBA, " said an MBS observer. He said that if you look at 2002 production 6.5s, for instance, they get lumped into one big pool, but if investors want to buy this collateral in the market, which is priced at TBA, they would get 80% of the worst production. "If you want to buy the collateral with the better characteristics, you have to pay up for that," he said. "So it's almost like you're paying to match the Index."
Adverse selection would also hurt liquidity in the higher coupons, he added. If rates back-up and everything is at a discount, liquidity will be fine. However, it becomes tricky when people want to move up in coupon, like a large seller of 5.5s going into 6.5s, which is a good example of collateral that is getting picked though. "When you want to do that trade, you don't want to do it TBA for TBA," he said. "You are going to want to find the right pool to buy 6.5s. and adjust it for the cost. So it would be a much more difficult time for the big mortgage investor to move up in coupon."
Nadine Cancell, vice president at the Bond Market Association, said that it is the nature of the TBA market to have securities that are cheapest to deliver. She added that the additional information that the GSEs are going to provide will be used less for cherry picking collateral than it would be for improving prepayment analysis. A lot of the recent objections to increased disclosure from the GSEs are similar to those heard in the 1980s, when Fannie and Freddie first started releasing information on WALAs (weighted average loan ages) as well as some other quartile information.
Copyright 2003 Thomson Media Inc. All Rights Reserved.