and Servicer identification provides additional information about underwriting standards and servicer actions that could impact prepayments.

This type of information is currently utilized in our loan level models of jumbo and sub-prime loans. Until now, we have primarily incorporated information about the loan contract into our pool models. We also indirectly incorporate information about the borrower based on the initial spread (as described above), and information about collateral based on a national measure of home price appreciation.

The staff report also recommended additional disclosures for each pool of:
           Loan purpose
           Original LTV
           Standardized credit scores of borrowers
           Servicer of the pool
           Occupancy Status
           Property Type

Beginning with April 2003, Fannie Mae will provide the recommended disclosures. They will also change the timing on some existing disclosures. In addition to the new disclosures, Fannie Mae publishes information about loan coupons, remaining term, weighted average loan age, weighted average loan term and original loan terms, original loan size, and seller identity for each pool. Freddie Mac will start to provide the recommended disclosures beginning with the June 2003 factor release date. Freddie Mac also provides similar disclosures to Fannie Mae.

Unfortunately, the task force did not recommend and the GSEs have not agreed to loan level disclosure. The task force found "that market participants generally sought aggregate pool information - as opposed to loan level information - perhaps expressed in quartiles or other standardized breakdowns." It would be surprising to us if modelers of prepayments did not request loan level detail. Currently, such detail is available in the non-agency market. If we merely relied on pool information, we would find it much harder to construct prepayment models for jumbo and sub-prime mortgages.

For example, if a pool contains half 70% LTV loans and half 90% LTV loans, it will have different prepayment characteristics than a pool with all 80% LTV loans. Even if quartile or other data is presented, it is difficult to determine whether the high LTV or low LTV loans prepaid in any given period. With loan level data, it is possible to clearly see the impact of the varying LTVs. Loan level >>>