paryy-marchapril

Get your Credit Program in Shape

In September of every year, the Information Management Network puts on an asset-backed securities conference in Miami Beach, Florida, at the world famous Fontainebleau Hotel. The conference events are fantastic; the hotel is amazing and the beach is truly something to behold. For many years, I attended that conference faithfully in order to support my company’s efforts at promoting their bonds (cough, cough).

My business partner and I would always head to the conference early so that we could enjoy the surroundings before we were stuck in meeting rooms for 10 hours a day. Of course, going to South Beach is a lot of pressure in and of itself. With the obvious exception of the conference attendees, everyone walking around there is aesthetically perfect. I would plan my diet and workouts meticulously, and as the year progressed they would look something like this:
• January – “If I can lose 2 pounds per week, I will be in great shape by conference time.”
• April – “If I can lose 5 pounds per week, I should be able to come close to my conference goal weight.”
• August – “If I can drop 50 pounds by the end of this workout, I might make it.”

I suppose that is why Mark Twain said, “New Year’s Day–Now is the accepted time to make your regular annual good resolutions. Next week you can begin paving hell with them as usual.”

While I joke about my prior resolutions, being in good physical condition is extremely important. Adipose tissue (fat) requires blood flow to maintain, creating a resource drain on the body, which can lead to high blood pressure. Excess weight puts tremendous stress on the joints, leading to a higher risk of osteoarthritis. In addition, obesity can cause problems with the performance of the heart and brain.

In much the same way, an out of shape credit program can cause serious ripple effects throughout the whole organization. Many lenders looked to 2018 as a year where they would shore up the bottom line, meaning they put the focus on profitability over volume. Like those who are now reassessing their New Year’s resolutions, smart lenders will evaluate which activities are helping and hindering the pursuit of their goal.

How fit is your program?
An out of shape credit program leads to a number of negative consequences that erode profitability, not to mention dealer satisfaction. Recognizing that your credit program has deficiencies requires that the lender be able to to connect application data from the loan origination system to performance results stored in the servicing system.

With that link in place, data may be analyzed that confirms or refutes the value of each component of the program. There are generally three factors at the root of a sub-optimal credit program. These are irrelevancy, redundancy and imbalance. They coalesce to produce poor closure rates, bloated expense and inconsistent credit performance.

Irrelevancy
Irrelevant factors are those rules in the credit program that are believed to be important, but in fact do little to help the company predict performance. Without good data and analytics, this is the most difficult error to overcome. Most judgmental buying programs emanate from the lender’s resident credit expert, who is typically a senior manager, or founder, with a tremendous amount of lending experience.

Many of the rules developed by the company expert are both intuitive and valid. However, many others do not hold up when compared to actual performance. I will borrow from Mark Twain again and quote, “The trouble with the world is not that people know too little; it’s that they know so many things that just aren’t so.”
The following examples are drawn from actual sub-prime auto loans with 24 months of credit performance. The data comes from a random sample of credit bureau records with the following characteristics:
• Credit score < 600
• Lender type: Finance company (as opposed to a bank or captive)
• APR ≥ 18 percent
• Amount financed between $5,000 and $30,000

This profile is typical of sub-prime paper below a 600 score, and demonstrated a 24 percent default rate over the first 24 months on books. The sample contained 300,000 records of the credit and performance attributes of actual auto loans.

A classic case of irrelevancy is found in deferred student loans. While it is true that, historically, deferred student loans would have the effect of inflating the credit score, it is not true that people with a large number of student loans, or deferred loans, are a materially worse default risk than those who do not have them (refer to the student loans chart). Certainly, if the only credit an applicant has consists of deferred student loans, the lender should be careful; however, there is no meaningful drop off in credit performance as the number increases.

Another example is the number of charged-off accounts in the last 12 months (for consumers with no bankruptcy on file). Those with no charge-offs are modestly better than those with only a few, but for consumers with four or more, the charge-off rate is lower. Does this mean lenders should pursue applicants with many recent charge-offs? Of course, not – the consumers with more charge-offs also have many more years in the file and have many other accounts that have paid.

One of the most powerful attributes in the credit bureau file, as least as it relates to auto loan charge-offs, is the Average Months in the Credit Bureau File. Loosely defined, this characteristic measures the average number of months that all trades have been on record. As can be seen in the chart, those with fewer average months in file have a dramatically higher default rates, and vice versa.

I have seen many lenders scratch their heads when an account with recent charge-offs receives a high credit score. The reason for this is that the thickness of the credit file, combined with positive past history, compensates for some of the risk of a recent default. If one takes two applicants with a recent credit default, the risk is entirely different based on the total body of their credit histories. Refer to the chart on the bottom right of the graphic shown on page 15.

Including factors in your credit program that don’t correlate very well with actual performance leads to unacceptably low approval rates, and ultimately poor closure rates. Worse yet, forcing good scores (or other highly correlated factors like Average Months in File) to conform to irrelevant rules will quickly degrade credit quality.

Redundancy
I consulted with many lenders over the past few years, and the most common problem I see in credit programs is that they are far more complicated than they need to be. Most credit attributes are not independent of other factors, meaning that people with more tradelines also have been in the file for a longer period of time. Applicants with higher revolving credit limits have earned them by paying bills consistently. Consumers with numerous medical collections often tend to have an unstable employment history. Almost everything that can be measured on a consumer is in some way related to a large number of other attributes.

The purpose of a lending program, essentially, is to predict credit performance. Each tier has its own pricing and loan structure parameters, set with the expectation of making money. No rational lender would knowingly set a price that would leave the company under water. Unfortunately, this happens all the time because either the lender does not know what their losses should be, or the program itself is volatile.

Credit programs become overly complex for several reasons. First, credit managers want to account for every scenario, leading to dozens of redundant factors. Second, the desire to keep up with what the dealer says the competition is buying leads to additional rules designed to fish out the best of the deals the buyers were formerly turning down (or putting in lower tiers). Finally, when lenders start to grow, the number of exceptions naturally increases. Poor execution usually results in credit surprises down the road.

As a rule, the more factors that go into a predictive model (particularly redundant ones) the less robust it will be. I have seen programs where lenders have specific rules about the number of 30, 60 and 90 day past due tradelines within a specific period. That is completely unnecessary, as they are all measuring varying degrees of the same problem. The company should determine which one specific attribute is the most correlated to default, and simplify down to a single rule. Other lenders have rules about the payment patterns of auto loans, all installment loans and revolving trades. Again, this is likely overkill. Find the best predictor and simplify the program to achieve more stable results.

Imbalance
An imbalanced program is one where too much focus is placed on some factors, while not enough is placed on others that significantly affect net losses. For example, I often see programs where little is known about how many consumers are likely to default in each tier, but tremendous attention is given to vehicle risk. This is backwards, as the recovery value only matters if the company has to repo the car.

Consider the case where the lender has a 30 percent default rate and a 40 percent recovery rate. Assuming no principal paid down, a $10 million portfolio would have $1.8 million in net losses ($10mm x 30% x 60%). Improving recoveries by 500 basis points would change the net losses to $1.65 million ($10mm x 30% x 55%). However, improving unit losses by 500 basis points would change the net loss figure to $1.5 million ($10mm x 25% x 60%).

The default rate has the most influence on losses, and must be given the greatest attention in the credit program. Other factors, such as limits on LTV, mileage and down payment should be scaled based on the default rate in each program tier. In other words, give the less risky consumers more latitude – which will lead to a more competitive program and a higher closure rate.

Getting in shape
The benefits of weight loss and cardiovascular fitness are well documented. Increased mobility and strength, reduced risk of disease and improved mental ability are but a few of the benefits. Likewise, a fit credit program requires fewer resources in terms of headcount and system capabilities. Furthermore, approval rates and decision speed will increase. Finally, the company will see both reduced (and more predictable) credit losses. All of this leads to lower operating expense and higher profitability. Now is the time to review your program in order to eliminate irrelevant rules, select the best of the redundant ones and place the emphasis on the factors that contribute most to the bottom line. As for me, I will focus on getting back into conference shape.

Daniel Parry is co-founder and CEO of TruDecision Inc., a fintech company focused on bringing competitive advantages to auto dealers and lenders. He is also co-founder and CEO of Praxis Finance, a portfolio acquisition company, and co-founder and former chief credit officer for Exeter Finance Corp. For questions or inquiries, please email danielparrynaf@live.com visit www.trudecision.com.