The rising popularity of larger, more expensive vehicles and new vehicle technology has driven growth in the average transaction price and financed amount of new automobiles. Higher financed amounts coupled with increased borrowing costs (according to the Federal Reserve's G.19 report, the average annual percentage rate [APR] for a 60-month bank-originated auto loan increased to approximately 5.4% as of fourth-quarter 2018 from an average of 4.2% from 2014 to 2017) are fueling upward pressure on monthly payments, creating an affordability issue for consumers. Auto lenders have been addressing this concern by lengthening loan terms, which serve to lower monthly payments. While loan tenors of 61-72 months remain the most common term (40% of new and 44% of used vehicle loans), loans with an original term of 73-84 months have grown to approximately 31% and 19% of all new and used vehicle loans, respectively, as of second-quarter 2019, according to Experian.

Chart 1

These 84 month loans have been making their way into U.S. auto loan asset-backed securities (ABS), primarily in the prime space. This year marked the first time a U.S. automotive captive finance company securitized 84-month loans in a public U.S. ABS term transaction. In June, Toyota Motor Credit Co. completed its inaugural five-year revolving transaction, which allows up to 15% of the pool to include 84 month loans (Toyota Auto Loan Extended Note Trust [TALNT] 2019-1). While this is a new development in the U.S. auto captive ABS market, U.S. banks have been including 84-month loans in their ABS pools for at least 10 years. Also, two years ago Canadian banks and certain Canadian captives started to include a limited portion of these in their ABS.

These longer-term loans pose additional risk to investors and can lead to higher cumulative net losses (CNLs) due to a number of reasons. Because new vehicles depreciate rapidly, auto lenders are often "upside-down" in the vehicles they finance, which means that the loan balances are higher than what the vehicles could be sold for in the wholesale market if they were repossessed. Thus, as loan terms lengthen, their principal balances amortize more slowly, and loss severity escalates. The slower amortization also delays the point at which a borrower starts to build equity in the vehicle, giving the borrower less incentive to remain current on their obligation if they encounter financial difficulty. The probability of this increases as the term lengthens. Furthermore, depending upon the lender, some of these longer-term loans are extended to weaker credit quality obligors (as reflected by lower FICOs and/or higher payment-to-income [PTI] ratios) and/or with higher loan-to-value (LTV) ratios. The combination of these factors is often referred to as layered risk.

Depending upon the lender, these longer term loans often carry higher APRs than shorter-term loans. To the extent the lenders have priced the risk appropriately, these loans can be quite profitable for the lenders. Additionally, the captives provide 84-month loans to help their auto manufactures sell more vehicles, although these could potentially lengthen the ownership trade-in cycle.

To measure the degree to which longer-term loans have higher losses than their shorter-term counterparts, we analyzed the origination static pool net loss data of three prime auto issuers from 2008 through 2014 (for vintages that had at least 48 months of performance; see table 1) and through 2013 (for vintages with at least 60 months of performance; see table 2). We found that, on average, CNLs were more than four times higher for loans with terms in the 73-month to 84-month range than those with terms of less than or equal to 60 months, although there are variances across the issuers. The CNL data displayed below for each of the annual vintages is a simple average of the CNLs for each of the issuers.

On an issuer-specific basis, the relative performance we observed for 84-month loans relative to 60-month loans (averaged across the 2008-2013 vintages) ranged from two to nine times worse. While nine times appears quite steep, keep in mind that losses can be quite low on 60 month loans. For example, if an issuer has experienced an average of only 25 basis points (bps) of losses on its 60 month loans, 2.25%, nine times that average, is still a prime loss level.

It's important to point out that our analysis used origination data for only certain issuers; others could have different results. Further, we used historical origination statistics and not securitization performance. Securitizers are likely choosing their collateral from more recent originations, which could differ in credit quality and performance compared to the loans originated in the years shown below.

Table 1

Table 2

By examining performance through both month 48 and month 60 for the above vintages, we noticed that a meaningful portion of losses still occur on 72-month and longer loans after the fourth year, especially during economically weak periods. For example, for the 2008-originated 73-84 month loans, their average CNL increased 50 bps to 3.1% during their fifth year outstanding from 2.6% at the end of their fourth year.

Are Auto Loan ABS Issuers Mitigating Higher Loss Risk?

In efforts to soften the blow of potentially higher losses from 84-month loans, some auto lenders may be more stringent in their underwriting requirements for these loans. Among the six prime issuers that we examined that are currently securitizing 84-month loans, Bank of the West has the highest concentration with 73- to 84-month loans representing 68.7% of its 2019-1 pool. It appears that the company somewhat offsets its high concentration of long-term loans by maintaining high FICOs on these loans. In their 2019-1 transaction, for example, the weighted average FICO score of loans with terms of 73 months to 84 months was 740, marginally higher than the pool's overall average of 736.

Table 3

For other lenders, however, we observed weaker collateral characteristics for their 76-84 month loans as compared to their overall pool. For Toyota, the weighted-average LTV ratio and weighted-average FICO on these loans were 112.6% and 732, respectively, compared with 101.2% and 754 for the overall pool. Similarly, Fifth Third had a higher weighted-average LTV ratio and lower weighted-average FICO on its 76-84 month loans compared to its overall pool.

While publicly available data is limited with respect to the 73- to 75-month loan term, initial data seems to indicate that these loans could be classified as a "stretch" product, especially when the lender also provides an 84-month loan. For example, in Fifth Third's 2019-1 transaction, the 73- to 75-month loans had lower FICOs and higher LTV and PTI ratios than the 84-month loans. As a percentage of the pool these 73- to 75-month loans represented only 1.4% of 2019-1, down from 35.9% for its 2017-1 transaction. Further, even with the relatively high percentage of these "stretch" loans in its 2017-1 transaction, the pool is trending toward CNLs of less than 70 bps.

Credit Implications For Rated Auto Loan ABS

As the presence of longer-term loans increases in U.S. auto loan securitizations, S&P Global Ratings may adjust its expected cumulative net loss (ECNL) levels upward, which could result in higher credit enhancement levels. For example, when assuming a "worst case" pool mix for Toyota's revolving TALNT 2019-1 transaction, which allows up to 37.5% in 73- to 84-month loans (76- to 84-month loans are capped at 15%) as well as other flexible eligibility criteria, our base case CNL level increased to 1.80% from approximately 0.60% on its traditional amortizing term deals, which exclude loans with original terms greater than 72 months. Credit support also increased commensurately.

Our ECNL levels will continue to be informed by issuer-specific static pool performances broken out by loan term. Most of the lenders securitizing 84-month loans today have a wealth of performance data on how these loans have performed, including through the last recession. In some cases this data is further stratified by credit score and LTV ratio. Our rating analysis also includes an examination of the collateral characteristics of these longer loan terms to determine if there is layering of risk, which exacerbates the credit risk associated with these loans.

Longer-term loans also pose greater back-end risk in auto loan ABS than their shorter-term brethren as defaults on these loans will have higher loss severities. Given the sequential pay structure of auto loan ABS, the most subordinated classes bear the brunt of this risk. S&P Global Ratings has observed that the increase in longer-term loans in rated U.S. prime retail auto loan ABS has also caused securitization losses to become more back-loaded (see "U.S. Prime Auto Loan ABS Are Seeing More Back-Loaded Losses As Loan Terms Lengthen," published July 30, 2019).

The lengthening of loan terms, while a risk, is in our view a manageable one that can be addressed in a number of ways. First, the portion of longer-term loans can be limited, especially if the issuer has a short track record of making these loans. We saw this when 72-month loans became the industry standard. In general, lenders slowly added these to their pools, many keeping the percentage in a range of 10%-15%. In addition, credit enhancement can address the higher risk associated with these loans. Finally, today's environment is similar to the early 1990s when loan terms lengthened to 60 months from 48 months, and 2001 when they started to stretch to 72 months. At the same time, vehicles are lasting longer than they did 20-25 years ago, with many still on the road after 10 years. The auto finance and auto loan ABS market survived the earlier lengthening of loan terms, and we believe it will weather the evolution to 84 months, although it's not a welcomed trend.

The author would like to thank Dan Daley, Aaron Dalal, and Sahay Senathikagu for their contributions to this report.