December 2024 - The Anatomy of Loan Defaults

In this month’s edition of The LoanStreet Beat, we examine how different loan parameters impact the likelihood of charge-off and what credit unions should focus on when pricing loans. In addition, we recap an active economic news cycle and share our observations on loan trading. Enjoy, share and please comment below!

LoanStreet Market Commentary

November has been a volatile month for Treasury yields. Yields surged post election, with the 2-year Treasury moving from around 4.20% to a peak of 4.40%. The yield has since come back down to 4.16% at the end of November. At the November Fed meeting, the Fed decided to cut the fed funds rate by another 25bps, after cutting 50bps in September. The market expects another 25bps cut in December. Despite the November cut and the expectation for another cut in December, the Fed, along with the market, have sounded a bit more dovish when it comes to future cuts. The latest inflation readings have pointed to slower progress on getting inflation down to the 2% target.

On the labor side, while the October job report showed only 12k in jobs added, vs the 100k estimate, the weakness was largely attributed to the impact of hurricanes and strikes. The November report, scheduled to be released December 6th, might provide a clearer picture of where the labor market stands. The latest JOLTS report (https://www.bls.gov/news.release/jolts.nr0.htm), released December 3rd, showed higher than expected job openings paired with slower hiring and a higher quits rate. This data showed conflicting metrics as more openings and a higher quits rate suggests the labor market remains hot, while slower hiring points to a cooling market. Lastly, the ratio of vacancies per unemployed worker is now at 1.1 to 1, far from the peak of 2 to 1 back in 2022.?

Loan Trading Trends and Implications

The end-of-year push is in full effect as both buyers and sellers are trying to right their balance sheet by year-end. Auto loans continue to remain in high demand. Despite the volatility in the Treasury market, auto loan yields have remained stable over the last month. Strong demand from buyers has put a ceiling on how high yields need to be while the amble supply has limited the downward pressure on yields. Due to the lack of supply, sellers looking for best execution should focus on direct auto. Further, buyers remain premium adverse, meaning sellers who are looking for the most favorable pricing should consider taking a higher servicing fee in exchange for a lower premium as this will result in buyers willing to accept a lower yield on the pool.?

Outside of auto, we are seeing strong demand for residential. This strong demand is being met with even stronger supply. Residential buyers are highly focused on their buy-box, usually min credit score of 700, max DTI of 43% and max LTV of 80%. Given the available supply of residential loans, we are generally able to meet that buy-box. Buyers are able to find favorable execution on more seasoned residential loans as they are priced at a discount due to the low coupons. These pools have a lower probability of prepayment, given the low coupon, and should they prepay quicker than expected, it would positively impact the yield given the discount pricing.?

Based on feedback from our clients, we expect this level of activity to continue into the new year. We are already evaluating deals which are scheduled to launch in January while multiple buyers have cited the fact that they are waiting until January to start buying again.?

Deep Dive: The Anatomy of Loan Defaults

In this month’s deep dive we will take a look at LoanStreet’s database of loans to evaluate what loan parameters drive losses. For this post, we will focus strictly on auto; our data set will include a total original balance equal to approximately $19B across 628k loans. The parameters we decided to evaluate are: credit score, DTI, LTV, term, vehicle vintage, new/used, and direct/indirect. To evaluate the impact of each parameter, we looked at what percent of loans within a given parameter range, for example 120-140% LTV, ultimately charged off relative to the total portfolio balance within that parameter range. It is important to note that we isolated each parameter without controlling for the other. Meaning that the loans with lower credit scores might have also had lower LTVs which muted the losses.?

Credit Score: As expected, the lower the credit score, the higher the losses. The biggest jump in losses can be seen when moving to sub 700 credit scores. Elsewhere, migrating from the 760-779 bucket to the 740-759 bucket also results in a relatively large increase in charge-offs.?

LTV: The LTV parameter paints an even clearer picture than credit score as losses pick up significantly above an LTV of 120%.?

Term: The story is similar with loan terms, losses escalate on loans with terms 72 months and greater. It’s worth noting that the original term of 72-96 month buckets make up about 80% of the portfolio despite the relatively higher losses compared to the shorter terms.?

New/Used: While used auto makes up about double the size of new on the portfolio, it accounts for three times as many losses.?

Direct/Indirect: Similarly to used auto, indirect auto is a little more than three times the size of direct on the balance sheet. In this case the difference in losses is even greater, as indirect losses are almost seven times that of direct.?

We also looked at DTI and vehicle vintage but they didn’t provide much valuable information. Most pools are capped at a 50% DTI, and losses on DTI’s below 50% did not vary from bucket to bucket. Vehicle vintage seemingly did not have any impact on losses as the percentage of loans charged-off within each bucket was closely related to the percentage of total loans within that bucket. This may reflect that vehicle vintage is already taken into account within the LTV parameter.?

Running a logistic regression on the data yields us similar results. LTV and credit score have the largest impact on the probability of charge-off for a given auto loan. Loan terms, new/used and direct/indirect also have a significant impact but to a lesser degree. It is important to note that the data does heavily skew towards loans above 72 month term, used and indirect, making the data imbalanced. That being said, the above data provides insight into which variables should be the focus when pricing loans. For this reason, and as discussed in a previous LoanStreet Beat article (https://www.dhirubhai.net/pulse/august-2024-power-positive-thinking-selection-loanstreet-llc-5bpmc/?trackingId=s8INSoySRueSPLbbZy%2FU4w%3D%3D), buyers should pay attention to the threshold filters applied to a given pool they are evaluating.?

This article was authored by Matt Rudzinski, Director of Sales and Trading.

For more market commentary and to learn more about LoanStreet's solutions, visit www.loan-street.com

Disclaimer

LoanStreet is not a Registered Exchange, Financial Planner, Investment Adviser, or Tax Adviser. The information provided herein is for general informational purposes only, and does not, and is not intended to, constitute legal, financial, investment, or tax advice.