By the Numbers
Quality 2020 loans shift prepay risk into lower coupons
Brian Landy, CFA | February 19, 2021
This document is intended for institutional investors and is not subject to all of the independence and disclosure standards applicable to debt research reports prepared for retail investors.
Conventional mortgage originations reached record levels in 2020 as many borrowers took advantage of low interest rates to refinance. The Covid-19 pandemic also helped fuel home purchases, as many borrowers looked for larger homes to make it easier to work and attend school from home. Borrowers that took out loans in 2020 have much higher average credit scores than previous vintages, which has shifted prepayment and extension risk out of older vintages and into 2020 vintage MBS pools.
Tracking burnout through credit scores
The culling of stronger borrowers out of older vintages has helped drive prepayment burnout—some borrowers taking advantage of lower rates, leaving behind the borrowers that could not refinance as easily. These remaining borrowers stand to face frictions to refinancing in the future, which improves the convexity of older vintages. Pools in the 30-year 3% and higher coupons should benefit from this shifting of credit, since most seasoned pools are in these coupons.
The current reported credit score on pools issued prior to 2020 likely overstates the true credit score of those borrowers. MBS investors do not know the current credit score of a borrower since the agencies only report the score when the loan was originated. This makes credit score less informative as a pool seasons.
In 2020 the credit score of new originations was much higher than that of the loans that paid off. Most new originations in Fannie Mae and Freddie Mac pools, whether a refinance or a repeat home buyer, resulted in the payoff of a loan that was also in a conventional MBS pool. This means there was credit improvement in the loans that prepaid and implies the credit score of the loans that did not prepay is likely lower than indicated by the stale MBS disclosure. The sustained fast prepayment speeds during the pandemic amplifies this discrepancy.
A credit score gap opens up between refi’d loans and existing loans left behind
One factor that plays a significant role in the credit quality of a new loan delivered to the GSEs is the overall level of refinance activity. Low FICO pools typically command a pay-up since those borrowers tend to prepay slower in a refinance environment but prepay faster in a discount (turnover) environment. This can be seen by comparing the average FICO score at the start of each calendar year to the average FICO scores of loans that prepaid and to newly originated loans (Exhibit 1). Only Freddie Mac data is used since the agency reports history on pools issued starting in December 2005, whereas Fannie Mae’s data does not begin until the 2013 vintage.
Exhibit 1: Credit score trends in Freddie Mac pools
Outstanding credit score is measured at the start of the year. Prepaid credit score is the average credit score of loans that paid off that year. Only owner-occupied loans are included.
Source: Freddie Mac, eMBS, Amherst Pierpont Securities
The heaviest refinancing years—2012 and 2020—show that borrowers with higher credit scores tend to refinance faster. The average credit scores of loans that prepaid in those years was four points to six points higher than the average credit score outstanding. But in the slowest refinance years—2014, 2017, and 2018—the average credit score of prepaid loans was five points to six points lower than the average outstanding loan.
Those credit scores do not represent a borrower’s current credit score, however. Comparing the average credit score of new originations to the average credit score of the loans that prepaid demonstrates that there has been credit improvement among the borrowers that refinanced. For example, in 2020 the average credit score of new originations was 760, which was six points higher than the credit score of the loans that prepaid. This difference is largest in periods of heavy refinance activity but shrinks in turnover environments.
The repeat buyer typically has a higher credit score than the loans that prepaid. Their credit score has been stable over time, although tends to be a little lower in years with less refinance activity. It was also higher in 2011 and 2012 when the credit box was tighter following the financial crisis. The average credit score of refinances changes a lot over time, and borrowers that refinance in a discount environment tend to have credit scores below the average that prepaid. This may be an indicator that those refinances are dominated by borrowers with weaker credit that need to cash out some home equity even though rates are not low.
This data suggests that lower credit borrowers are unable to take advantage of refinance opportunities, which is why they provide prepayment protection. In a discount environment they may be more likely to move or be forced to cash-out home equity, which bolsters discount prepayment speeds and provides extension protection.
The credit score of first-time buyers also jumped in 2020, another contributor to worse convexity in that vintage. The average credit score jumped 5 points from 2019 and 8 points compared to 2018. Typically, the credit score spread between repeat and first-time home buyers is 10 to 12 points. However, that spread narrowed to 8 points in 2020.
Seasoned pools should benefit from burnout, in part due to prepayments of the best credit borrowers in those pools. This leaves behind lower-credit borrowers, which have demonstrated they do not refinance easily but also tend to prepay faster when rates are higher.