As part of the ongoing support for their credit risk transfer program, in March 2016 Freddie Mac released updated single family loan-level data containing origination and performance records on 21.8 million loans originated between January 1999 and March 2015. Using RiskSpan’s data discovery tool, RS Edge, the origination characteristics of the loans originated from 1999 to 2014 were analyzed.
- Purchase volume grew almost two times since 2011 and surpassed refinance volume in 2014 for the first time in data history.
- First time home buyer exposures were trending up among purchases while investor purchases reached a new high in 2014.
- FICO declined in 2013 and 2014, but remained higher than pre-recession levels.
- LTV was 76.6% in 2014 - a new high since 2000, largely due to the increase of purchase volume over refinances.
- DTI went up in 2013 and 2014, however the concentration in the highest DTI bucket (45-65) remained below the pre-recession average.
Figure 1 depicts origination volume by loan purpose. Figure 2 displays purchase loan originations by occupancy.
Exhibit 1: Original UPB by Loan Purpose
Exhibit 2: Purchase Distribution
- In conjunction with the largest mortgage rate drop (105bps) since 2001, refinance volume soared in 2009 to over $300 billion.
- Refinance volume fell below purchases in 2014 for the first time in data history.
- Purchase volume grew 195% since 2011 to 125B in 2014 – the highest in data history.
- The percentage of first time home buyers (FTHB) increased from 28.2% in 2012 to 32.9% in 2014, while non-FTHB and second home buyer percentages fell.
- Origination volume on investment properties among purchases reached a new high of $7.1B in 2014.
Figures 3 and 4 depict year-over-year FICO by distribution and by loan purpose.
Exhibit 3: Year-Over-Year FICO Distribution
Exhibit 4: Year-Over-Year FICO by Loan Purpose
- Although FICO declined from its peak of 767 in 2012 to 752 in 2014, it remained above the pre-recession average in 720s.
- Although the distribution of stellar credits (750 and above) shrank in 2013 to 57.4% in 2014, that was still higher than pre-recession levels in the 30% range.
- Historically during the observed periods, FICO on purchase loans appeared less volatile than refinances.
Figures 5 and 6 are year-over-year LTV by distribution and loan purpose.
Exhibit 5: Year-Over-Year LTV Distribution
Exhibit 6: Year-Over-Year LTV by Loan Purpose
- Average LTV went up between 2009 and 2014 regardless of loan purpose, which may be a result from a relaxation of underwriting standards. Nonetheless, they remained in line with pre-recession averages.
- LTV is correlated to purchase volume in a given year since purchase loans generally carry higher LTV than refinances at origination.
Figures 7 and 8 depict year-over-year DTI by distribution and loan purpose.
Exhibit 7: Year-Over-Year DTI Distribution
Exhibit 8: Year-Over-Year DTI by Loan Purpose
- DTI remained flat between 2009 and 2012 despite a continued drop in WACs. Combined with an upward swing in LTV in the same period (Figure 6), this may be an indication of borrowers taking out larger loans.
- Despite the upward trend in DTI since 2013, regardless of loan purpose, they remained below pre-recession levels.
As the economy continued to recover, Freddie Mac’s loan-level origination data shows a marked increase in the volume of purchase loans, and to a lesser extent investment properties. The data also indicates originators cautiously taking more risks as FICO, LTV and DTI metrics loosened from the recession years, although tighter than pre-recession levels. This trend in credit quality and the associated risks should be monitored. Our next post will focus on the performance of the different vintages.