June 2019 collateral performance in the U.S. auto loan asset-backed securities (ABS) sector weakened. Losses increased and recoveries declined month over month across both prime and subprime segments, which is typical beginning at this time of the year due to seasonality. Further, subprime 60-plus-day delinquencies were at their highest June level since we have started tracking them due to a greater presence of deep subprime auto loan ABS.

June Marks The Beginning Of The Traditional Loss Ascent

U.S. prime credit losses increased to 0.50% in June 2019 from 0.41% in May 2019; however, they improved annually from 0.52% in June 2018 (see table 1 and chart 1). Subprime losses increased month over month to 6.93% in June 2019 from 6.34% in May 2019 and increased seven basis points (bps) from 6.86% in June 2018. After netting out three deep subprime issuers, modified subprime losses increased by only 10 bps to 5.47% in June 2019 month over month. On an annual basis, the index increased slightly from 5.43% in June 2018.

Table 1

Santander (including its Santander Drive Auto Receivables Trust [SDART] and DRIVE deals) and AmeriCredit Automobile Receivables Trust transactions continue to make up most of the subprime index, at 61.35%, an increase of 188 bps from the prior month driven by increased share of DRIVE and AmeriCredit.

As a supplement to its subprime index, S&P Global Ratings created the modified subprime index, which excludes certain high-loss deep subprime issuers (DRIVE, American Credit Acceptance Receivables, and Exeter Automobile Receivables Trust). Santander's SDART transactions and AmeriCredit continue to make up most of the modified subprime index at approximately 55.03%. Still, this is lower than their combined 57.54% share a year ago: AmeriCredit's share declined to approximately 24.00% from 26.00%.

Recoveries Demonstrate Seasonal Weakness

Prime recoveries declined to 59.85% in June from 66.78% in May 2019, although they improved from 58.64% in June 2018 (see table 2 and chart 2). Subprime recoveries decreased to 44.10% in June 2019 from 48.90% in May 2019; annually, they decreased from 44.68% in June 2018. Subprime modified recoveries decreased to 44.41% in June 2019 from 46.77% in May 2019 and 45.05% in June 2018.

The monthly downtick in recoveries across the board is a June seasonal feature; however, this is the weakest recorded June recovery performance in this decade (barring June 2018 for prime).

Table 2

June Delinquencies Weaken Further

The prime 60-plus-day delinquency rate increased to 0.41% in June 2019 from 0.36% in May 2019 and 0.38% in June 2018 (see table 3 and chart 3).

The subprime 60-plus-day delinquency rate increased by 49 bps month over month to 5.00% in June 2019, and by 40 bps from 4.60% in June 2018. This was the highest June level since we started tracking this statistic, and is largely due to the greater concentration of deep subprime loans being securitized. On a modified basis after netting out three deep subprime lenders, the subprime modified 60-plus-day delinquency rate increased to 3.60% in June 2019 compared with 3.19% in May 2019 and 3.28% in June 2018. The modified subprime June 2018 delinquency rate was comparable to the level reported in June 2016.

Table 3

Collateral Characteristics

Prime loan terms length at record high levels

For the first six months of 2019, average loan terms lengthened to 66.5 months compared with 65.6 months for the first half of 2018. Considering the continuing lengthening of the loan terms, we have started tracking the share of loan terms with maturities of 72 months and higher in the pool mix. (Please see "U.S. Prime Auto Loan ABS Are Seeing More Back-Loaded Losses As Loan Terms Lengthen," published July 30, 2019, for more details on this trend.) As of June 2019, the share of 72-75 month loans stood at 13.5%, similar to the 2018 level.

We've also noticed that 84-month loans have become more prevalent in prime auto loan ABS. For example, Bank of the West's BWSTA 2019-1 had 68.7% with original terms of 73-84 months, with a significant portion of the loans concentrated in the 84-month bucket as the pool had a weighted average original term of 77.8 months. In addition, Toyota's recent TALNT 2019-1 five-year revolving transaction allows up to 15.0% of the loans to have original terms of 76-84 months, and Fifth Third's FTAT 2019-1 transaction included approximately 4.5% with original terms of 76-84 months. The weighted average original maturity of the Fifth Third and TALNT pool at closing were 69.4 months.

In addition to the increase in longer-term loans, the weighted average annual percentage rates (APRs) for the first half of 2019 for both sectors increased versus the same period in 2018. Consumers are borrowing more over a longer period of time as lending rates and average transaction prices have increased (see table 4A).

Table 4A

Subprime loan terms also continue to lengthen

In subprime, loan to value ratios (LTVs) increased to 111.4% from 109.9%, and loan terms of 73-75 months grew to 13.3% from 9.7% a year earlier (see table 4B).

Table 4B

Auto Loan Static Index (ALSI) Vintage Analysis

Prime performance continues to improve

Prime vintage static pool loss performance has improved with each passing vintage since 2016. The 2017 vintage, with 20 months of performance, has incurred 0.58% in cumulative net losses (CNLs), compared with 0.63% at the same seasoning point for the 2016 vintage (see chart 4). The first- half 2018 with 13 months of performance, stands at 0.32% losses, compared with 0.39% and 0.41% at the same point in 2017 and 2016 respectively. The second half of 2018 is showing higher losses; although with just seven months of performance, it's too early to predict its future course.

Subprime

The stabilization in subprime CNLs, beginning with the 2016 vintage, continued as of June 2019. For the 2016 vintage, CNLs through month 31 are 11.40%, slightly lower than 2015's 11.77%. For the 2017 vintage, CNLs through month 19 are 7.96%, close to 2016's CNL of 7.88% at the same point. The first half of 2018 is performing better than the 2015 and after vintages. For the second half of 2018, it's still too early to conclude whether this lower loss trend will continue.

Netting out the deep-subprime issuers, for the loans beginning in 2015, we use our modified subprime index, which excludes three large high-loss issuers (DRIVE, ACA, and Exeter). On this measure, CNLs for the modified 2015 and subsequent vintages are performing similarly overall to the 2013 and 2014 subprime index (see the modified lines in chart 5 and tables 7 and 8).

(For more on subprime performance, see "Subprime Auto Loan ABS Tracker: Losses Have Stabilized, But Renewed Growth Bears Watching," April 29, 2019.)

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 Auto Loan Static Index (ALSI) (see tables 5A and 5B).

Table 5A

Table 5B

In the prime segment, CarMax, Hyundai, and World Omni usually record higher-than-index losses. CarMax transactions of 2016 at month 31 are already very close to the 2015 vintage transactions at month 36. Hyundai 2016 transactions have already surpassed 2015 losses with 2017 transactions following them in line at month 19. World Omni issuances have also followed the same trend as Hyundai.

In the subprime space, the three largest issuers are performing as expected. While the Santander and AmeriCredit deals are performing better than the overall index, the DRIVE issuances are trending weaker. CPS' losses are trending lower than the subprime index for 2016 and 2017; however, they have longer terms than their peers. As a result, their losses tend to be more back-loaded than that of other subprime companies, and their deals usually end up losing more than the subprime index, as was the case in 2014 and 2015. We recently revised loss levels of all the CPS transactions upwards in June (see "Various Rating Actions Taken On 11 CPS Auto Receivables Trust Transactions," published June 14, 2019).

Revised Loss Expectations

We did not have any loss revision activity in July 2019.

Appendix I: ALSI Performance Data

Table 6

Table 7

Table 8

Table 9

Table 10

Table 11

Table 12

Table 13

Table 14

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 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 reaches its fourth month outstanding. 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 3.0% or less, average FICO scores of 700 or higher, and annual percentage rates (APRs) of 0.0%-5.0%.

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 7.5%, average FICO scores of less than 620, and APRs that exceed 14.0%.

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 Auto Loan Static Index 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 for which 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 (also known as CarNow), 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 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. Prime Auto Loan ABS Are Seeing More Back-Loaded Losses As Loan Terms Lengthen, July 30, 2019
• Various Rating Actions Taken On 11 CPS Auto Receivables Trust Transactions, June 14, 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
• Honor Automobile Trust Securitization 2016-1 Class B Notes Downgraded, Feb. 15, 2019
• Is There Extension Tension In U.S. Subprime Auto Loan ABS? Nov. 29, 2018