Collateral performance was generally positive for the month of September in the U.S. auto loan asset-backed securities (ABS) sector. Losses and 60-plus-day delinquencies decreased month over month as well as year over year in both the prime and subprime segments, although recoveries in both segments declined compared to the prior month. Lower recoveries are typical as we approach year-end and next year's model vehicles start crowding the showrooms.

Prime And Subprime Losses Improved

U.S. prime credit losses improved to 0.53% in September 2019 from 0.58% in August 2019 and 0.61% in September 2018 (see table 1 and chart 1). The top three issuers--Toyota, CarMax, and Honda--make up approximately 42% of the index, and showed stable or improved losses on both a monthly and annual basis.

Subprime losses decreased 35 basis points (bps) to 8.40% in September 2019 from August 2019, and 17 bps from September 2018. Santander Drive Auto Receivables Trust's (SDART's) losses worsened on a monthly basis but improved annually, whereas Drive Auto Receivables Trust's (DRIVE's) losses worsened both monthly and annually. These two platforms make up approximately 44.00% of the subprime index. Aside from the Santander-related transactions, most other issuers reported lower losses month over month. After netting out three deep subprime issuers (including DRIVE), modified subprime losses decreased by 71 bps month over month to 5.88% in September 2019. On an annual basis, the modified index improved year over year from 6.44% in September 2018 and stood at the lowest September level since September 2012.

Table 1

Recoveries Declined Monthly But Improved Annually

Prime recoveries declined considerably by 588 bps to 53.11% in September 2019 from 58.99% in August 2019, although they improved from 52.01% in September 2018 (see table 2 and chart 2). Subprime recoveries decreased to 40.70% in September 2019 from 42.95% in August 2019, but improved from 39.42% in September 2018. Subprime modified recoveries decreased to 38.91% in September 2019 from 42.41% in August 2019, and weakened from 40.42% in September 2018.

Recoveries typically decline over the course of the second half of the year, bottoming out around the end of the year or the beginning of the next. September's decline was further magnified by the cool-off in used vehicle prices after an unusually strong summer.

Table 2

Delinquencies Improved Across The Board

The prime 60-plus-day delinquency rate decreased slightly to 0.40% in September 2019 compared to 0.42% in both August 2019 and September 2018 (see table 3 and chart 3).

The subprime 60-plus-day delinquency rate declined to 5.22% in September 2019 from 5.25% in August 2019, and 5.36% in September 2018. On a modified basis, after netting out three deep subprime lenders, the subprime modified 60-plus-day delinquency rate decreased to 3.59% in September 2019 from 3.82% in August 2019, and 3.78% in September 2018.

Table 3

Collateral Characteristics

The percentage of longer-term prime loans continues to increase

For the nine months ended September 2019, average loan terms increased to 66.13 months compared to 65.73 months for the same period in 2018, and the percentage of loans with an original term greater than 60 months (60.31%) surpassed the level in 2018 (57.78%). In addition to lengthening terms, the weighted average annual percentage rate (WAAPR) has increased along with the percentage of used vehicles. Mitigating this, however, are a high weighted average FICO of 755 and lower weighted average LTV (95.48% compared to 95.94% a year earlier).

We have started to track the share of loan terms with maturities greater than 72 months, which year to date through September was 7.80% (6.50% in 73-78 months plus 1.26% in 79-84 months). While the popularity of larger and more expensive vehicles has endured among consumers as gas prices remain low, the affordability issues thereby created continue to be addressed by lengthening loan terms (see "U.S. Prime Auto Loan ABS Are Seeing More Back-Loaded Losses As Loan Terms Lengthen," published on July 30, 2019).

Table 4A

Subprime loan terms also continue to lengthen

The percentage of subprime loans with an original term greater than 60 months has fluctuated between approximately 83% and 85% for each one-year period from 2015 through 2018. The percentage for the nine months ended September 2019 is roughly in the middle of that range at approximately 83.5%, albeit that represents an increase from 81.3% for the same period in 2018 and 83.0% for the full-year 2018. In addition, the percentage of loans with terms greater than 72 months has increased noticeably to 7.38% for the nine months ended September 2019 from 3.80% for the same period in 2018.

In recent years, the WAAPR has steadily increased, as has the percentage of used vehicles, which has been somewhat in flux since the Great Recession. The weighted average FICO and LTVs have also experienced fluctuations over the years but have most recently worsened (see table 4B).

Table 4B

Auto Loan Static Index (ALSI) Vintage Analysis

Prime performance continues to improve

Prime vintage static pool loss performance continues to improve with each passing vintage since 2016. The 2017 vintage has 23 months of performance data with losses of 0.64%, which is lower than the 2016 vintage at 0.70% with the same seasoning. Both the 2018 vintage (11 months of performance) and the first-quarter 2019 vintage (seven months of performance) are currently tracking the 2015 vintage.

Subprime

The subprime cumulative net losses (CNLs) have shown a stable trend beginning with the 2016 vintage--which, together with the 2017 vintage, tracks the 2015 vintage relatively closely--while later vintages currently exhibit incremental improvement. For the 2016 vintage, CNLs through month 35 are 12.28%, which is lower than 2015's 12.70%. For the 2017 vintage, CNLs through month 22 are 9.07%, which is in line with 2016's 9.03%. The 2018 vintage has 11 months of data that average 3.86% in CNLs, which is considerably less than the 2015-2017 vintage loss levels of 4.17%-4.26%.

Beginning in 2015, we use S&P Global Ratings' modified subprime index, which excludes three large high-loss issuers (DRIVE, ACA, and Exeter). On this measure, CNLs for the modified 2017 vintage at month 22 (6.76%) are below the modified 2016 vintage losses (7.13%) at the same time, and CNLs for the modified 2018 vintage at month 11 (2.97%) are lower than the modified 2016 vintage (3.19%) and modified 2017 vintage (3.15%) at the same point. Most subprime lenders tightened their credit standards in 2017, and this is becoming evident in their most recent vintage performance.

Issuer-Specific Cumulative Net Loss Index Data

We track CNLs by vintage for a number of issuers and compare those losses to the prime and subprime ALSI (see tables 5A and 5B).

Table 5A

A number of prime issuers are reporting higher losses on their 2016 vintages at month 34 than their 2015 vintages at month 36. The 2016 vintage was an extremely competitive period in the auto finance industry, characterized by slightly weaker standards and certain lenders moving rapidly into new markets. For example, CNLs on California Republic Bank's (CRB's) 2016 transactions at month 34 (3.12%) have already surpassed losses for the 2015 transactions at month 36 (3.07%). We recently reviewed CRB's transactions and revised our expected CNLs on their 2015 and 2016 transactions to between 3.45%-3.90% and 3.55%-4.50%, respectively. We also revised our expected CNLs on their 2017 and 2018 transactions to 4.30%-4.50% and 4.50%-4.80%, respectively (see table 8).

CNLs on Hyundai's 2016 pools are also trending higher than 2015, but our revised expected CNLs on their 2016-B pool of 1.65%-1.85% is only slightly higher than our original expected CNL of 1.50-1.70% (see table 9).

Table 5B

In the subprime space, CNLs through month 34 on First Investor's and SDART's 2016 transactions have already surpassed the levels on their 2015 transactions at month 36, but they remain close to 2014's levels. Losses on Prestige's and Westlake's 2016 pools are exceeding not only those on their 2015 transactions, but also those on their 2014 transactions.

Revised Loss Expectations

In October 2019, we revised our loss expectations for the following transactions:

• Four Westlake transactions backed by subprime auto loan receivables;
• Three TCF Auto Receivables transactions backed by prime auto loan receivables;
• Seven California Republic Auto Receivables transactions backed by prime retail auto loans;
• Three Hyundai Auto Receivables Trust transactions backed by prime retail auto loans;
• Four Drive Auto Receivables Trust transactions backed by subprime auto loan receivables; and
• Two CarMax Auto Owner Trust transactions backed by prime retail auto loans.

Of the 23 transactions we reviewed, we lowered our expected CNLs for 17, maintained them for two, and increased them for four.

Table 6

Table 7

Table 8

Table 9

Table 10

Table 11

Historical Ratings Activity

Through October 2019, our reviews resulted in 325 upgrades and five downgrades.

Table 12

Appendix I: ALSI Performance Data

Table 13

Table 14

Table 15

Table 16

Table 17

Table 18

Table 19

Table 20

Table 21

Appendix II: Auto Tracker Methodology And Definitions--Frequently Asked Questions

Effective with our U.S. auto loan ABS tracker report, "U.S. Auto Loan ABS Tracker: Full-Year 2017 And December 2017 Performance," published on Feb. 22, 2018, we modified our methodology for calculating the loss, delinquency, and recovery rates reported. Under the new methodology, we do not incorporate a transaction's performance into the composite results until it has been outstanding for four months. We have applied the new methodology to transactions that closed in December 2005 and thereafter.

How do you define prime auto loan ABS?

We generally categorize prime auto loan ABS transactions as those backed by loan pools with initial expected CNLs of 3.00% or less, average FICO scores of 700 or higher, and annual percentage rates (APRs) of 0.00%-5.00%.

How do you define subprime auto loan ABS?

We generally categorize subprime auto loan ABS transactions as those backed by loan pools with initial expected CNLs of at least 7.50%, average FICO scores of less than 620, and APRs that exceed 14.00%.

How do you calculate the monthly net loss rate?

The monthly net loss rate is annualized. It equals each transaction's net loss rate weighted by the transaction's ending pool balance for the current month over the aggregate ending pool balance of all transactions included in the index.

We only allow a transaction to enter the composite starting in its fourth month outstanding. Transactions usually have zero or low losses during their first three months, which dilutes the composite figures.

How do you calculate the monthly recovery rate?

We calculate recoveries by taking the recovery amount reported (which typically includes all recoveries, including disposition proceeds, post-disposition proceeds, and any other reported recoveries) over the gross loss amount for the current month. We then weight each transaction's recovery percentage by the transaction's ending pool balance for the current month over the aggregate ending pool balance of all transactions included in the index.

We only allow a transaction to enter the index starting in its fourth month outstanding. During a transaction's first three months, unusually high or low recoveries are reported, leading to a spike in the composite figures. Previously, we often excluded recovery rates in the first two months of a deal's life because of negative recovery rates (resulting from recoveries exceeding gross losses).

How do you calculate the monthly 60-plus-day delinquency rate?

We calculate delinquencies by taking each transaction's 60-plus-day delinquency amount over the ending pool balance for the current month. We then weight each transaction's 60-plus-day delinquency percentage by the transaction's ending pool balance for the current month over the aggregate ending pool balance of all transactions included in the composite.

We only allow a transaction to enter the composite starting in its fourth month outstanding. During the transaction's first three months, zero or lower delinquencies are reported, which dilutes the composite figures.

What is the ALSI?

Our ALSI monitors the credit performance of securitizations that were originated in the same year on a weighted average basis. The number of months displayed for each vintage is generally determined by the last point that all securitizations for that time period have a data point. We calculate the prime and subprime ALSI CNLs by taking the weighted average of the CNLs of the transactions that were completed in the same time period (generally a year). Each transaction's CNL is weighted by its initial pool balance over the aggregate initial pool balance of all the transactions included in the index for that period. In the subprime ALSI, transactions from Byrider Finance LLC (doing business as CarNow Acceptance Corp.), Credit Acceptance Corp., and DriveTime Automotive Group Inc. are excluded because they do not have the typical indirect auto loan business model.

Which transactions are included in the prime, subprime, and modified subprime composites and indices?

For a list of the transactions included in our prime, subprime, and modified subprime composites and indices, see "U.S. Auto Loan ABS Tracker: March 2019," published on May 23, 2019. However, note that we subsequently added S&P Global Ratings-rated transactions that have since closed, most prime transactions that closed and were not rated by S&P Global Ratings from 2016 through the present, and the Santander transactions S&P Global Ratings did not rate.

Related Research

• U.S. Auto Sector Faces Bumpy Roads Ahead With Rising Recession Odds And Falling Demand, Oct. 16, 2019
• Twelve Ratings Raised, Eight Affirmed On Four Westlake Automobile Receivables Trust Transactions, Oct. 16, 2019
• Eight Ratings Raised And Three Ratings Affirmed On Three TCF Auto Receivables Owner Trust Transactions, Oct. 15, 2019
• Fourteen Ratings Raised And Ten Affirmed On Seven California Republic Auto Receivables Trust Transactions, Oct. 10, 2019
• Three Ratings Raised, 12 Ratings Affirmed On Three Hyundai Auto Receivables Trust Transactions, Oct. 4, 2019
• Fourteen Ratings Raised, One Affirmed On Four Drive Auto Receivables Trust Transactions, Oct. 2, 2019
• Six Ratings Raised, Seven Affirmed On Two CarMax Auto Owner Trust Transactions, Oct. 2, 2019
• Four Ratings Raised, One Affirmed On Two Credit Acceptance Auto Loan Trust Transactions, Oct. 4, 2019

• Speed Bump Ahead: As Auto Loans Accelerate Toward 84 Months, Caution Is Warranted, Sept. 18, 2019
• Six Ratings Raised, Two Affirmed On Two Santander Drive Auto Receivables Trust Transactions, Sept. 17, 2019
• Four CPS Auto Receivables Trust Ratings Lowered And Removed From CreditWatch Negative, Sept. 9, 2019
• U.S. Prime Auto Loan ABS Are Seeing More Back-Loaded Losses As Loan Terms Lengthen, July 30, 2019
• Subprime Auto Loan ABS Tracker: Losses Have Stabilized, But Renewed Growth Bears Watching, April 29, 2019
• The Severity Of Subprime Auto Loan Delinquencies Is In The Eye Of The Beholder, March 18, 2019
• 10-Year Retrospective: Changes In U.S. Auto ABS In The Decade Since The Great Recession, Feb. 15, 2019
• Is There Extension Tension In U.S. Subprime Auto Loan ABS? Nov. 29, 2018