Auto lenders have had two issues on their minds entering 2023, which are managing a rise in delinquency and planning for declining used vehicle values. The first installment of this series addressed the delinquency issue by drawing attention to properly supporting the collections operation. This article deals with the second issue, managing recoveries in an environment of declining auction values.
By the time delinquency issues become obvious, the lender already has a bubble of worse paper that must work its way through the pipeline. This is why it is critical to have early warning systems in place with regard to credit quality, and a strong executive to push meaningful changes down to buyers in a timely manner. In the case of declining vehicle values, the issue is partially mitigated by the fact that the lender does not take all of their repos to auction the same day values drop.
I frequently do market updates with large groups of bond and equity investors in the auto lending space. The participants on these calls often have a chicken little view of vehicle values, and they fear their specific investments will take a 30 percent hit, or more, when it comes to net loss rates. The term they use to describe this is a “recovery cliff”. I point them, and any equally concerned lenders, to history to put things in perspective.
Prior to the pandemic, we saw two sudden shocks to vehicle values. In the first quarter of 2003, rental car companies were dumping their fleets and manufacturers were offering employee pricing. This created a glut of used vehicles in the auction markets, causing lenders to see recovery rates drop by as much as twenty percent for a two to three month period, before they quickly rebounded. The second occurred in late 2008, when the capital markets effectively shut down in the turmoil caused by the subprime mortgage crisis. Vehicle values rapidly dropped due to the inability of consumers to get loans, but again, values rebounded within a few months.
Today’s environment is similar to what occurred in 2009, when vehicle values rose thirty percent over a two year period. This increase in values was due to massive cuts in production during the Great Recession, and the subsequent impact of a tsunami hitting Japan in 2010, causing an additional reduction in new vehicle inventory. For those who remember that period, everyone was expecting a precipitous drop in values under the idea that what goes up must come down.
At that time, estimates of millions of vehicles coming off lease were the basis for countless predictions of an impending recovery value crisis. Of course, that never happened. The reason for this was that a massive number of consumers had been sidelined during the recession and there was tremendous pent-up demand.
Sudden shocks to either supply or the capital markets tend to resolve quickly, minimizing the impact to a lender’s total portfolio. Protracted production issues tend to cause values to rise or fall over a multi-year period. In addition, the majority of an auto lenders defaults tend to be distributed across a thirty six month window, depending on credit niche and loan term. This means the impact to the bottom line is spread out, with new originations booked at values that are adjusted to the current state of the market, offsetting a significant portion of the risk at a portfolio level.
Fast forward to 2023. We came through a pandemic, which actually resulted in a jump in auto sales from mid-2020 to mid-2021; however, the chip-manufacturing crisis caused an unprecedented drop in inventory. New vehicle values increased thirty percent, or more, due to limited supply. This resulted in pushing customers to the used market, which resulted in a similar effect due to demand.
Subprime auto lenders that were formerly originating at an average $18,000 amount financed, for example, are now booking similar vehicles near $27,000. While terms increased to help keep the payment affordable, the average used vehicle car payment increased by over $100 from early 2021 to late 2022. The combination of high prices and significant economic pressure on consumers is in part to blame for increased delinquency, which has resulted in a drop in new and used sales and a modest tightening by lenders who are focused on credit quality issues. Fewer sales and contracted lending capacity always results in a drop in recovery values, which we are seeing presently.
Here are some key data points to consider:
• The Manheim Used Vehicle Index rose 10 points to 234 in February 2023, but is down substantially from 258 in January of last year. Likewise, the Used Vehicle CPI dropped from 207 to 190 in the last six months. In addition, Black Book’s Vehicle Retention Index dropped from approximately 200 to 170 in the last six months.
• While new vehicle sales showed a modest uptick in seasonally adjusted annual rate for January 2023, the last year has been essentially flat (Federal Reserve TOTALSA trend). Retail Sales at Used Car Dealers have dropped approximately twelve percent since this time in 2022 (Federal Reserve MRTSSM44112USN trend).
• Auto Inventory to Sales Ratio has increased slightly in the last year, but sits at only .684 units for every one sold (Federal Reserve AISRSA trend). In normal periods this value is between 2.5 and 3 units on the lot for every one sold.
What one may reasonably conclude from this information is that the drop in used vehicle prices has less to do with an influx in inventory than it does with economic pressures putting a damper on both demand and lending capacity. This is good news for several reasons. First, tax refund season is upon us which will result in more sales and pressure on limited inventory. Second, there is pent-up demand, which if the economy improves will result in a longer period of increased demand pressure that will offset some of the natural drop in used vehicle values resulting from the return of vehicle supply.
Perhaps the most important issue related to these facts is that we have not yet begun to see the expected drop in values do to a meaningful rise in inventory. This means managers still have time to take steps to mitigate recovery rate risk in their portfolios. Effectively managing this risk requires lenders to:
– Acquire the right data
– Make surgical adjustments
– Challenge the status quo
Acquire the Right Data
There are typically between 350 and 400 combinations of make and model of vehicle produced each year. Multiply this by, say, six to ten model years and you have thousands of individual units, each with its own trajectory for depreciation. Layer on top of this how recoveries change based on seasonality, the state of the auction markets and where we are in the credit cycle and you have a dizzying array of things to forecast. I applaud any lender that seeks to apply quantitative analysis to learn from their historical data, but it is highly likely that the data you need is not fully contained within the data you have. There are several top data providers available to auto lenders. Acquiring this data in as granular a form as possible is critical to refining recovery estimates that should be used to modify both originations and collections strategy. Many lenders will use high level recovery estimates based on program tier, however, across vehicle class actual recovery rates can vary substantially (refer to the depreciation graph on this page) Furthermore, specific units within a single vehicle class may vary wildly. Remember, it is not what the vehicle is worth at the point of origination, but rather what it will be worth in fifteen or so months when you are likely to repossess it.
Make Surgical Adjustments
Frequently, lenders wait until a crisis is evident to take action. This tends to result in bold sweeping changes that impair the company temporarily when things improve. A case in point is where many companies were laying off buyers and shrinking originations at the outset of the pandemic. Within a few months, volume was on the rise and the spike in defaults predicted by many failed to materialize. Lenders who over-corrected missed a window to originate more volume at higher yields in a fairly non-competitive market. The lesson in this is to cut surgically in order to have the greatest impact on risk with the least impact to volume and profitability. For more than two decades I have observed that more than a third of delinquency and credit loss can be isolated to roughly ten percent of originations through the application of data analytics. As it relates to recovery risk, the place to start is in the credit program. Using robust data sources, lenders may either upcharge or cut advance on the most risky vehicles while getting more aggressive on callbacks for units that are likely to retain the greatest value over time. Small changes can make a strong impact without disrupting your ability to take advantage of an improving environment later on.
Challenge the Status Quo
One of the primary reasons lenders get surprised by credit performance issues is the status quo. Unfortunately, things that work in a stable environment (i.e., most of the time) have unreliable outcomes during market or economic stress. This may be further complicated by levels of management that are incented on activities or targets that are suboptimal given the present environment. It is common for many collections managers to target delinquent loans with the highest principal balance for either increased call activity or for repossession. This is done in order to have the greatest potential impact on dollars delinquent at month-end. This may also be very shortsighted for several reasons. For subprime auto lenders, the roll rate for accounts that are one payment past due is typically around twenty percent. Effective behavior scores may easily identify accounts within that delinquency category that have a roll likelihood ranging from five to seventy percent. Failing to prioritize the highest risk accounts due to lower balances will result in more dollars in late stage delinquency and charge-off. Additionally, data-driven recovery forecasts may be used to accelerate repossession activity on the highest risk vehicles, minimizing the potential net credit loss. Finally, lenders offer various forms of forbearance to customers in hopes of rehabilitating the account – although they may only do so on a limited basis such as in the case of deferments. Part of the equation must be the potential dollar loss if the customer does not get back on track. Again, this involves factoring in a solid recovery forecast into the decision process. This is more of a net present value approach to collections strategy that balances the cost of action today versus next month. For executives, such a mindset will only be effective if collections leaders at all levels have incentives that are revised to target the right outcomes.
We may or may not be headed into a recession. Vehicle values may increase with demand, or continue to decline. There are a variety of factors that influence auto portfolio performance, and many are outside of the lender’s control. Having said that, managers who focus on the right data, precise adjustments to strategy and optimal performance targets will significantly minimize the credit performance risks we face in the coming year.
