Last year the Federal Housing Finance Agency (FHFA)—Fannie Mae’s and Freddie Mac’s regulator—announced a streamlined version of the federal government’s popular Home Affordable Refinance Program (HARP). The streamlined program will expand HARP eligibility to include mortgages originated on or after October 1, 2017.
Granting HARP eligibility to these new mortgages has important implications for Fannie’s and Freddie’s credit risk transfer (CRT) program—which consists entirely of loans originated after 2009 and were therefore not eligible for HARP. Starting October 1st, loans in the CRT program will now be eligible to refinance under HARP.
Last month the FHFA announced that going forward, CRT loans that refinance under HARP can be replaced in their original pools by the newly refinanced loans. An FHFA press release describes the rationale behind this new provision: “This will help preserve credit loss protection on the loans without unwinding the protection paid for through CRT transactions.”
HARP enables otherwise qualified borrowers (i.e. with a source of income and a good payment history1) with high-LTV loans owned by Fannie or Freddie to refinance their mortgages. HARP refinances are approved irrespective of credit score or debt-to-income ratio and even when the underlying property’s value may have fallen to a level where the resulting LTV ratio would not ordinarily support a conventional mortgage.
CRT Data and Market Implications of HARP Loans
Expanding HARP to include CRT loans has given rise to an exciting development for mortgage data enthusiasts. In response to the FHFA announcement, Fannie and Freddie have released a significant enhancement to their single family loan-level datasets that now include details about loans that have undergone a HARP refinance.2 This newly released information enables data and investment analysts to compare and contrast the performance of individual loans before and after they refinance through HARP. The ability to link a refinanced loan to the loan it replaced makes possible all kinds of interesting analysis that previously could not have been performed.
Here are some of the things analysts can do with the new HARP data.
- Positively identify which prepayments are associated with HARP refinances.
- Evaluate the performance of HARP loans—determine which pre-HARP characteristics and behavior drive post-HARP performance.
- Compare the performance of loans that refinanced through HARP with the performance of loans with similar characteristics—i.e., loans that might have refinanced through HARP—that did not refinance.
- Compare current LTVs before and after a HARP refinance.
The data will also enable analysts to assess loss-related assumptions that are incorporated into existing models. Today’s credit models likely do not take into account the new policy permitting HARP refinances to substitute for the prepaid loans they are replacing in the same CRT pool. Such models are likely to understate losses on future CRT deals because performance data on HARP refinances was previously excluded from the dataset. Machine learning techniques could contribute significantly to these and related analyses.
Cracking the Data
The Agencies have provided SAS and R code to assist analysts seeking to make pre- and post-HARP comparisons by appending the refinanced loan’s data to the original loan record, thus creating one continuous loan record. The code also allows individual records to be flagged for inclusion or exclusion. Links to Fannie Mae’s code can be found here. A general user guide to Freddie Mac’s enhanced dataset can be found here.
 Borrowers must not have missed any mortgage payments in the previous six months and must not have missed more than one payment in the previous 12 months. The refinance must also accrue some benefit (e.g., a lower monthly payment) to the borrower.
 Freddie Mac’s Single Family Loan-Level Dataset: General User Guide describes the new data as including “loans that were in Freddie Mac Single Family Loan Level dataset and went through the Relief Refinance program between 2009 and 2016, with an LTV at the time of re-finance above 80. Only loans that remained as Fixed Products following this refinance were included in this disclosure. HARP files are consolidated into one origination and one performance file and not separated by year and quarter.”