Mitigating Recovery Risk

Lenders have been foregoing repossession and deferring accounts in large numbers since the COVID-19 explosion in March of this year. In the coming six months, these same lenders will have to put many vehicles through the auction, which will create a glut of inventory and downward pressure on recovery rates. Dealing with this risk involves two issues: mitigating vehicle risk in new originations, and managing the existing portfolio to minimize losses. For managers to successfully address these risks, they must identify trends and embrace tools that allow them to act before problems become apparent.


The process of underwriting credit can generally be reduced to assessing the risk of the consumer (frequency of loss) and the collateral (severity of loss). Credit programs tend to focus overwhelmingly on aspects related to the consumer with only minor attention paid to loss severity; however, in times of declining vehicle values, companies must ensure that collateral risk is minimized.

While servicing strategy and the economy can influence credit performance, the biggest drivers of quality have to do with what happens at the point of origination; therefore, the most important thing companies can do to buffer against the risk of falling collateral values is to shore up their underwriting standards. This means that a lender must be able to effectively identify risk by vehicle, and understand how that risk interacts with the risk from the applicant.

While there are still a few companies out there that tell the dealer “send me everything, we’ll figure it out”, the standard approach to designing an originations credit program is to break approvals into tiers (sometimes called products) based on the perception of how risky the credit is. This is most effectively done using custom score ranges, but many lenders achieve good results using a combination of the credit bureau score combined with a variety of rules. Elements related to risk, such as down payment, amount financed and LTV are typically restricted by tier, so that the least risky applicants get the most leeway, and vice versa.

In preparing for lower recoveries, lenders should review how their programs deal with the following:
Good LTV versus Bad LTV – Research has shown that consumers with a quality warranty or service contract have a lower incidence of default. In addition, back-end products are usually refundable if the loan terminates early. Negative equity or a high front-end LTV is simply air – it is lost if the loan defaults. Lenders should limit exposure to front-end LTV and make certain that any back-end products are reasonably priced and truly add value to the consumer.

The impact of a default in the lower credit tiers is much worse than the higher tiers. Worse credit generally defaults earlier, meaning that little principal has been paid and the charge-off amount is higher. Early defaults tend to have higher recovery rate than later stage defaults, but does that matter? Would you rather have a 70 percent recovery rate which results in a $15,000 loss, or would you prefer a 20 percent recovery rate that results in a $5,000 loss? Additional risk comes from the fact that the lower credit tiers may be comprised of less expensive, older collateral. During periods of falling used vehicle values, older cars are also disproportionately impacted. Program limits must reflect these differences.

On average, vehicles depreciate roughly 20 percent in the first year, and 15 percent thereafter. On a $15,000 car, that would be $250 per month in the first year and $200 per month in the second year. Of course, some makes and models could have twice that amount. Lenders should at the very least consider limiting advance, by tier, for the riskiest collateral.

I have often said that LTV only matters if you have to go pick up the car, alluding to the fact that companies can afford to give higher advance rates to less risky customers – but there are limits. Financially astute people do not put themselves in absurdly upside-down contracts. At some point, extreme loan structures present a substantially increased frequency risk, regardless of credit score. Even the highest tiers need sensible limits.

Portfolio Management

Tightening up one’s credit program to allay future collateral risk is important, but it doesn’t do anything for a lender’s existing portfolio. Fortunately, managing portfolio risk is comprised of the same factors that are used to manage originations risk – frequency and severity. Unfortunately, traditional collections strategies are lacking when it comes to managing the severity side of the equation.

That may sound counter-intuitive, as collections managers commonly think of the state of the portfolio almost exclusively in dollars delinquent, but the amount of the payment (or the gross balance) may have nothing to do with the lender’s exposure to collateral risk. For example, delinquent accounts are frequently called based on potential to roll, which is a method that sorts loans by the number of days until they roll to the next worse stage of delinquency. Rather than calling each account the same number of times (by delinquency bucket), calls are prioritized based on how many dollars are going to roll in the near term. While this is an effective strategy, it focuses on balances and does not consider which accounts have the greatest net loss potential. Likewise, metrics such as right party contact rate, promises made/kept, and dollars collected per hour are exclusively based on balances and payment amounts.

For lenders facing declining resale values, they must account for that risk with a more balanced approach to collections decisions. This involves integrating vehicle specific collateral risk in strategies related to:

  • Behavioral Scoring – Many lenders prioritize collection calls using behavior scores, which predict the customer’s likelihood of rolling to the next stage of delinquency. But likelihood alone is not enough, as a particular vehicle’s collateral risk may entirely overshadow modest differences in default probability. A 30 percent chance of a $10,000 loss should be more of a concern than a 40 percent chance of a $5,000 loss (an expected value of $10,000 x .3= $3,000 vs. $5,000 x .4=$2,000).

  • Payment Extensions – Most companies have a finite number of deferments to grant, and so must make wise choices in how to use them. The decision to grant a deferment is often a judgmental one, based on issues related to the customer’s specific situation, as well as where the company’s delinquency is as it approaches month-end. More sophisticated lenders treat the deferment as a true credit decision, and employ scoring models to evaluate the customer’s likelihood of getting back should they receive an extension. As with the prior examples, collateral risk should figure in the mix. Consider the case where two customers qualify as good candidates, but only one deferment may be granted. Both customers have a similar gross balance, but completely different vehicle values. If a default for the first customer represents a $10,000 net loss while the second one a $5,000 net loss, it would be far better to rehabilitate the first customer. Deferments should be spent where they can do the most good.

  • Timing of Repossessions – A number of lenders have strategies designed to avoid the charge-off as long as possible in hopes of turning the customer around. For these lenders, deferments are routinely done in the late stages of delinquency, and repossessions are sometimes not assigned until nearly 100 days past due. Depending on the customer’s payment history, recovery activity for accounts with high-risk vehicles should be accelerated as the monthly depreciation may be more than the overdue payment amounts.

Taking Advantage of Data

All of the aforementioned strategies sound good, but they hinge on the ability of the lender to effectively recognize varying degrees of vehicle risk. Many would be doing well merely to sort their historical recovery results into 4 or 5 vehicle make/class groupings, and then flex strategies based on that. Although this a simple approach, it is far superior to treating all accounts the same. The good news is that rich data sources are available that lead to precise and predictive solutions.

Lenders use credit scoring models to account for the complexity in how hundreds of factors interact with relation to credit risk. Lenders have used these models to simplify, optimize and automate their platforms with tremendous success. Well-built models consider not only what a lender has experienced internally, but also include data from customer performance outside of what the lender has seen in order to produce the most robust prediction.

In the same way, lenders can optimize decisions related to collateral risk using a data-driven approach. Unlike scorecards, though, a company’s internal history with a particular vehicle might in no way represent future risk. Collateral recovery models must include external data containing robust leading indicators, such as:

  • Auction trends by class and region
  • Historic depreciation trends by make and model across multiple credit cycles
  • Economic trends involving fuel prices, consumer prices and consumption, expansion/contraction of credit, interest rates and employment
  • Supply and demand data, which include prior year’s vehicle sales by class, dealer inventory, off-lease projections and disposal rates

As auto finance lenders face the prospect of declining used vehicle values related to the COVID-19 crisis, they must modify strategies to mitigate the risk on each individual vintage of loans. The ability to develop and deploy sophisticated models to address this issue is not limited to large organizations with a sizable investment in analytics and technology. Any lender, regardless of scale, can take advantage of powerful tools that can considerably limit exposure to vehicle value risk. As for companies with well-established modeling teams, don’t get fooled into thinking the answer is solely contained in your internal data. Seek to balance your models by partnering with subject matter experts that understand leading trends.

Daniel Parry is co-founder and CEO of TruDecision Inc., a fintech company focused on bringing competitive advantages to lenders through analytic technology. He was previously co-founder and CEO of Praxis Finance, a portfolio acquisition company, and co-founder and former chief credit oficer for Exeter Finance. Prior to this, he was senior vice-president of Credit Risk Management for AmeriCredit Corp (GM Financial). If you have questions, you may reach Daniel at [email protected].