The first performance month after the COVID-19 pandemic began, March 2020, showed slightly weaker performance relative to February, quite unlike the normal tax season trend. The most poignant indicator of potential consumer distress materialized with a spike in extension rates. Collateral performance in the U.S. auto loan asset-backed securities (ABS) sector also worsened in March on a year-over-year basis, reporting higher monthly losses and delinquencies with lower recoveries. In our view, it is still too soon to analyze the full performance effects of COVID-19 including April's unemployment level of 14.7%; these will become more evident over the next few months. Despite weakening performance and high extension levels, U.S. auto loan deals have numerous structural protections that, in our opinion, will protect these transactions from near-term liquidity-related bond defaults (see "The Potential Effects Of COVID-19 On U.S. Auto Loan ABS," published March 26, 2020).

S&P Global Ratings acknowledges a high degree of uncertainty about the rate of spread and peak of the coronavirus outbreak. Some government authorities estimate the pandemic will peak about midyear, and we are using this assumption in assessing the economic and credit implications. We believe the measures adopted to contain COVID-19 have pushed the global economy into recession (see our macroeconomic and credit updates here: www.spglobal.com/ratings). As the situation evolves, we will update our assumptions and estimates accordingly.

Losses Rise On A Year-Over-Year Basis

Prime net losses increased to 0.60% in March 2020 (the highest March level since March 2010's 1.03%) compared with 0.57% in February 2020, and 0.54% in March 2019 (see table 1 and chart 1). The losses increased month over month because of higher losses from Toyota, CarMax, Honda, Nissan, and Ally, which together make up 57.00% of the prime index.

While subprime net losses decreased 26 basis points (bps) to 8.37% in March 2020 from 8.63% in February 2020, this represented the highest March level since March 2009's 9.05% and was an increase of 126 bps from 7.11% in March 2019. The year-over-year increase in subprime losses was primarily due to higher losses reported in the following three issuing platforms: Exeter Automobile Receivables Trust (Exeter); Santander Consumer USA's Santander Drive Auto Receivables Trust (SDART); and Santander's Drive Auto Receivables Trust (DRIVE). These three shelves together made up approximately 48% of the overall subprime index as of March 2020. After netting out three deep subprime issuers (American Credit Acceptance, Exeter, and DRIVE), modified subprime losses decreased by 21 bps to 6.52% in March 2020 from 6.73% in February 2020, but increased 75 bps from 5.78% in March 2019.

Table 1

Recoveries show weakness due to servicing disruptions caused by the pandemic

Prime recoveries declined 426 bps to 53.89% in March 2020 (the lowest since March 2009's 53.67%) from 58.15% in February 2020, and 701 bps from 60.91% in March 2019 (see table 2 and chart 2).

Subprime recoveries declined 186 bps to 39.46% in March 2020 from 41.32% in February 2020. These are the lowest subprime March recoveries seen yet, lower even than March 2009's 45.59%. Year-over-year, rates decreased 961 bps from 49.07%. Subprime modified recoveries decreased 162 bps to 39.48% in March 2020 from 41.10% in February 2020, and decreased 887 bps from 48.35% in March 2019.

Recovery rates took a beating for a number of reasons: 1) lenders opted to work with delinquent customers rather than try to repossess and sell vehicles into a depressed and only partially functioning used vehicle wholesale market; 2) many states banned repossessions and the garnishment of CARES Act payments while their state is under a state-of-emergency; 3) to the extent vehicles were actually sold at auction during the second half of March, they garnered lower recovery rates than earlier in the month.

Recovery rates are expected to be potentially worse in April due to a full month of COVID-19 related auction shutdowns in many locations. Also, according to Manheim's Used Vehicle Value Index, wholesale used vehicle prices (on a mix-, mileage-, and seasonally adjusted basis) decreased 11.41% month-over-month in April, and decreased 9.2% year-over-year to an index level of 125.8, the lowest level in three years.

Table 2

Delinquencies increased substantially on a year-over-year basis

The prime 60-plus-day delinquency rate remained essentially flat at 0.41% in March 2020 versus 0.40% in February 2020, but it increased seven basis points from 0.34% in March 2019 and was at the highest March level since March 2009's 0.54%) (see table 3 and chart 3).

The subprime 60-plus-day delinquency rate increased to 5.01% in March 2020 from 4.94% in February 2020, and from 4.34% in March 2019. These are the highest subprime March 60-plus-day delinquency rates seen yet. On a modified basis, after netting out three deep subprime lenders, the subprime modified 60-plus-day delinquency rate increased to 3.55% in March 2020 from 3.52% in February 2020, and from 3.15% in March 2019.

Table 3

Extension tension

One of the early and most significant effects of COVID-19 on auto loan performance is the spike in extensions, which is clearly seen in the March performance.

Prime

As reported in our article, "While Stay-At-Home Orders Clear Traffic, U.S. Auto Loan Extensions Rise," published May 1, 2020, March extensions on a dollar basis spiked to 4.12% of outstanding loans, 12x February's level of 0.33% (see table 4). The data for this analysis came from the loan level Regulation AB (Reg AB) II loan level data that most prime issuers file with the SEC. For the month of March, the data we pulled from this site included $65 billion in prime auto loans across 14 issuers.

Subprime

For subprime issuers, we augmented the Reg AB II extension data provided in four issuing shelves (Santander's SDART and DRIVE, AmeriCredit Automobile Receivables Trust (AmeriCredit), and World Omni Select) with the monthly servicing report data for those subprime issuers that issue 144a transactions and are exempt from providing loan level data. Across these two groups, Reg AB II and 144a, which total 144 deals and $42.3 billion, the extension rate jumped to 6.93% in March 2020 from 2.40% in Feb 2020 (see table 4). Using only the Reg AB II data, the extension rate spiked to 6.82% from 1.53% (see table 4).

Table 4

Collateral Characteristics: First-Quarter 2020--The Last Of The Pre-COVID-19 Deals

First-quarter 2020 consisted primarily of loans originated prior to the pandemic and the stay-at-home orders that came with it and the resulting post-depression record unemployment level of 14.7% for April. Issuance dropped considerably on a year-over-year basis for the first quarter so compositional factors played a part in the overall collateral averages. Prime-rated deals equaled $6.9 billion in collateral across five deals compared with $11.6 billion across eight deals a year earlier. Similarly, the total amount of subprime collateral included in our tracker data declined to $3.7 billion across eight deals, which included three small issuers that typically include a significant percentage of loans to individuals who've had a previous bankruptcy. This was a substantial drop from the $7.6 billion in subprime auto loans securitized in the first quarter of 2019 across 10 deals, which included five deals in excess of $1 billion-–two DRIVE deals, and one deal a piece from SDART, AmeriCredit, and Westlake Automobile Receivables Trust.

Prime

The collateral characteristics for prime transactions issued in 2020 weakened with the inclusion of a larger percentage of longer-term loans, which increases tail risk (see table 5A). Mitigating this somewhat, the weighted average FICO score improved to 755 in first-quarter 2020 from 752 in first-quarter 2019, although remained essentially flat to the 2019 vintage as a whole. In addition, the weighted average loan-to-value ratio (LTV) declined slightly to 95.08% from 95.30%.

Also, the percentage of used vehicles financed increased to 42.30% from 24.33%, which was driven by CarMax Auto Owner Trust and Capital One Prime Auto Receivables Trust. Loans with an original maturity greater than 60 months increased to almost their highest level at 60.49%. Beginning in 2019, we started to track loan terms greater than 72 months. In first-quarter 2020, the percentage of loans securitized with an original term of 73-75 months was 10.86% compared with 6.75% for first-quarter 2019 (see "Speed Bump Ahead: As Auto Loans Accelerate Toward 84 Months, Caution Is Warranted," published Sept. 18, 2019).

Table 5A

Subprime

Collateral credit weakened slightly in first-quarter 2020 compared with first-quarter 2019.

• The weighted average FICO score decreased to 569 from 585 due to issuance composition, including issuers that include a significant amount of loans to individuals with prior bankruptcies.
• The weighted average LTV increased to 116.88% from 110.49%.
• Both the used percentage and annual percentage rate (APR) have increased to 77.92% and 19.39% from 65.48% and 17.62%, respectively.
• As noted above, in 2019 we started to track loans with terms of 73 months and longer. Loans with original maturities of 73-78 months represented 3.79% of the loans securitized, which was down from the 9.30% in first-quarter 2019.

Table 5B

Auto Loan Static Index: Vintage Analysis-–Too Early To Detect COVID-19 Impact

Prime

Auto loan ABS issuers' tighter underwriting standards and more selective criteria for their 2017 and subsequent securitizations are still generating positive results. The 2018 vintage, with 16 months of performance, has incurred 0.45% of cumulative net losses (CNLs) compared with 0.48% at the same seasoning point for 2017 vintages. In addition, the first-quarter 2019 has incurred 0.39% of CNLs, which is in line 2017 but three bps worse than 2018. For second-, third-, and fourth-quarter 2019, we have very limited data.

Subprime

Subprime CNLs, which rose considerably with the 2015 vintage due to the deterioration in Exeter's performance and the inclusion of Santander's inaugural DRIVE deals, have shown consistency in performance on a weighted average basis for the 2016 and 2017 cohorts (see chart 5). Some of this can be attributed to compositional factors with the DRIVE deals representing a lower share of issuance in 2016 than in 2015. In fact, as evidenced in table 6B, most subprime issuers reported higher CNLs on their 2016 pools than for those in 2015. CNLs for the aggregate 2018 vintage is trending below the three prior annual vintages (see chart 5), at least for the time being. The first-quarter 2019 cohort, which had been trending below 2018, is now tracking on top of the 2018 vintage, and in our view will likely start suffering higher losses due to higher charge-offs and lower recovery rates as it falls victim to COVID-19.

To normalize the subprime index to account for the greater concentration of deep subprime loans, we used our modified subprime index, which excludes three large high-loss issuers (DRIVE, ACA, and Exeter). On this measure, CNLs for the modified 2017, modified 2018, and modified first-quarter 2019 vintages are demonstrating year-over-year improvement in losses. (See the modified lines in chart 5, and table 10).

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 6A and 6B).

A number of prime issuers reported higher losses in 2016 at month 36 than their 2015 and 2014 vintages. CarMax's 2016 vintage experienced 2.06% losses, whereas its 2015 pools averaged 1.91% at month 36. Similarly, World Omni Auto Receivables Trust's (World Omni's) losses rose to 2.33% for its 2016 vintage compared with 1.70% in 2015. However, World Omni's prime pool performance has recently improved due to the following: beginning in 2017, it added a minimum FICO score of 650 to its prime pools; beginning in 2018, it excluded non-Toyota vehicles; and in 2019, it created a separate issuing shelf, World Omni Select Auto Trust, to securitize its nonprime/subprime auto loans.

Most prime issuers are reporting satisfactory results for their 2017 pools. However, Bank of the West's 2017 vintage shows weakening in performance with losses of 1.42% at month 28 compared with 0.52% at month 36 in its 2015 vintage. Similarly, Nissan's 2017 vintage at month 28 has already surpassed its 2016 vintage's losses of 0.69% at month 36. This trend continued for Nissan for its first-quarter 2019 vintage, which at 12 months has experienced higher losses than its 2018 deals at the 15th month mark.

Most of the subprime issuers' 2016 vintages performed worse than those from 2015. Their 2017 vintages are generally performing in line-to-better than their respective 2016 vintages.

Table 6A

Table 6B

Historical Ratings Activity

Through April 30, 2020, our reviews resulted in 73 upgrades and no downgrades. However, on May 12, 2020, we placed 33 classes of nonIG subprime auto loan ABS on CreditWatch with negative implications, see "Thirty-Three U.S. Subprime Auto ABS Ratings From 26 Transactions Placed On CreditWatch Negative," published May 12, 2020.

Table 7 shows S&P Global Ratings' historical ratings activities for U.S. ABS auto loans.

Table 7

Appendix I: ALSI Performance Data

Table 8

Table 9

Table 10

Table 11

Table 12

Table 13

Table 14

Table 15

Table 16

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 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 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

• Thirty-Three U.S. Subprime Auto ABS Ratings From 26 Transactions Placed On CreditWatch Negative, May 12, 2020
• Infrastructure: What Once Was Lost Can Now Be Found--The Productivity Boost, May 6, 2020
• U.S. Jobs Report: April Is The Cruelest Month, May 5, 2020
• While Stay-At-Home Orders Clear Traffic, U.S. Auto Loan Extensions Rise, May 1, 2020
• Tourism-Dependent U.S. States Could Face Credit Pressure From COVID-19's Outsized Effects On The Industry, April 27, 2020
• Economic Research: The Escalating Coronavirus Shock Is Pushing 2020 Global Growth Toward Zero, March 30, 2020
• It's Game Over For The Record U.S. Run; The Timing Of A Restart Remains Uncertain, March 27, 2020
• The Potential Effects Of COVID-19 On U.S. Auto Loan ABS, March 26, 2020
• Assessing The Coronavirus-Related Damage To The Global Economy And Credit Quality, March 24, 2020
• How The Wave Of Negative Rating Actions On Global Automakers Has Affected U.S. Auto ABS Ratings, Feb. 13, 2020
• Speed Bump Ahead: As Auto Loans Accelerate Toward 84 Months, Caution Is Warranted, Sept. 18, 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