No credit scoring model will be able to tell you with a 100% certainty whether a borrower will return the funds or not. But for the lack of a better system, lenders rely on the existing models, use alternative ones, a combination of the two or develop their own, proprietary algorithms. The reliability and accuracy of the credit scoring model is the gatekeeper of your lending business. It’s what determines how much business you’ll get and how profitable it’ll be.
A while back credit scores became a commodity sold by big companies like FICO and Vantage Score Solutions. But the whole financial system is undergoing tectonic changes right now with the monopoly of banks being questioned by alternative players. And in order to reach new audiences and locations, those players often choose to use different factors like borrower’s education, behavioral factors, psychological assessment, social media accounts, personal savings, and investments. All this data wasn’t previously used but is perfectly suited to assist in loan decisioning. And now it’s being factored in by savvy lenders who are willing to question the status quo.
The thing is that each lending operation is different and therefore the approach to evaluating borrowers should differ as well. So an uncustomizable scoring provided by old-school systems doesn’t cut it anymore. That is why a whole plethora of alternative lenders has gone one of two ways. They either use a custom scoring model, like Affirm, Upstart or SoFi or go with a customizable AI-powered scoring model provided by a lending automation company, like that of TurnKey Lender.
And the stats say that it works out very well for them, both as a selling point and as a business benefit resulting in lower risks and therefore better returns. For example, Upstart’s model shows 75% fewer defaults than that of a traditional bank while reaching 173% more approvals. And the AI-driven solution for lenders by TurnKey Lender runs a safe risk evaluation in 30 seconds where it takes conventional banks days.
The importance of a scoring model
If you’re running a lending business, there’s no way to eliminate risks altogether. The best you can do is have a scoring model reliable enough to weed out as many borrowers who won’t return the money as you can and approve as many loans to those who will as possible. Throughout the history, the scoring models were far from optimal or safe which would result in either lender declining people who would pay them back or approving people with higher risk scores which leads to ramping up the interest rates to balance the system out.
But leaving money on the table the way traditional banks used to do approving just 26% of small business loans, is no way to run a lending operation. Thus the need for more sophisticated models came to be. And since alternative digital lenders are far more agile and willing to use any secret weapons they can, they became the ones experimenting with the new technologies. And to show just how bad using unreliable scoring models can turn out, we can look at the recent meltdown of the peer-to-peer lending industry in China.
A reliable yet flexible and adjustable scoring model lets lending business outperform the competition, stimulate portfolio growth, increase client lifetime value, and offer better rates while reducing risks and working at a profit.
Existing scoring systems and why you need something better
Existing scoring models used by the big banks are tailored for the large volumes of clients and extremely low risks. Or they were. But the times changed and the development of technology has given us access to a ton of data that simply didn’t exist before. Here are the factors that traditional lenders take into account when deciding whether to issue a loan or not, from most influential to least.
For as long as the modern credit scoring exists, credit history used to be one of the key factors determining whether you’ll get a loan or not. So a lender would get a three-digit number evaluation for example from FICO and would base their loan decision on that. But the generations change and millennials are on the verge of becoming the largest paying generation. And in terms of loans, younger people think differently. For example, they are less likely to buy a house or use credit cards. And if they don’t there’s no way they can build that impressive of a credit history while still being suitable for a loan.
Even the providers of traditional scoring models are in on this trend with FICO taking a more modern approach and data sources like their checking and savings accounts in some of its solutions.
It wouldn’t be fair to say that it’s time to completely forget about the traditional evaluation models. But for an alternative lending operation to work more effectively they sure should be complemented by additional factors and considerations.
Credit bureaus and credit analytics companies
Many lenders still use companies like FICO and their local credit bureaus in their borrower data collection and evaluations. The difference between the two is that credit bureaus simply collect and sell your credit information while FICO uses its models to process borrower data and sell lenders the score.
And for the lack of an alternative, this approach was unquestioned. But ever since the alternative lending came to be, proprietary algorithms became more common. And then the FinTech companies joined in on the race and started offering lenders reliable bank-grade borrower evaluation software at a fraction of a price. So right now, you can make up your own scoring model using the data from credit bureaus in your calculations or you may buy scores from FICO and further use them to come up with your own evaluation.
[related-solutions]
Should you use a custom scorecard
At TurnKey Lender, dozens of people contribute to ongoing improvements and upgrades to the scoring and risk evaluation models we provide our clients with. The artificial intelligence algorithms are getting more and more sophisticated and the data processing software makes huge strides forward with each new release. Such investments into the scoring model are a must if you want to mitigate risks and increase the safety of issuing loans. But luckily lenders don’t have to go to such lengths and can simply choose the software with the best suitable model built-in and customize it. This way you get the benefits of having bank-grade software which is constantly being worked on and still get it custom-tailored for your own needs.
Scoring model factors to consider
When building your own scoring model or customizing the one of your LaaS solution, other than the traditional factors, you may take into account things like:
- Financial behavior;
- Social media signals;
- Smartphone data;
- Education;
- Work history;
- Personal savings;
- Any accessible bills;
- Psychological factors.
Psychometrics
One of the hottest trends in scoring right now is Psychometrics. It’s a branch of psychology which allows evaluating psychological properties of a person in a numerical value. Applied to lending, this would mean accessing things like borrowers:
- Honesty;
- Family relations;
- Punctuality;
- Responsibility;
- Money management and financial planning;
- Self-esteem;
- Confidence;
- Trust;
- Likelihood to take risks;
- Organizational skills;
- Consistency;
- Thinking speed.
The applications of this approach are already real and they are especially useful when working with locations and demographics who haven’t had access to proper credit products in the past. Of course, psychometrics will work best only when paired with other factors, but savvy lenders should already have it on their radar and strongly consider it.
Using alternative factors to evaluate loans lenders not only can reduce their risks but also serve the global community by not incentivizing taking loans just to build up a credit history.
Big data and AI in credit scoring
With each borrower comes a ton of data which lenders up until recently lenders haven’t been using in their evaluations. But as Artificial Intelligence started to become more and more of a real thing, algorithms became able to process lending data, learn on it and evaluate more of the right borrowers faster and safer. This is a completely new word in borrower scoring automation. And we’ll talk more about it next, in the section about the scoring TurnKey Lender provides its clients with since we’re the ones pioneering in this domain.
TurnKey Lender credit scoring
One of the major factors that make TurnKey Lender stand out among the competitors is the company’s meaningful use of AI combined with big data.
Ignoring the need for radical changes in scoring models and origination leaves space for human error, lengthy loan approval process, and weak fraud protection. TurnKey Lender solved this problem by implementing advanced AI algorithms which analyze large sets of consumer data, learn the behavior patterns, and make risk evaluations and credit scoring based on this data.
TurnKey Lender mainly uses AI for credit decisioning and risk evaluation. The system is developed by a team of PhDs that did machine learning and scoring projects for Boeing, LG, Bank of America, and Stanford University in the past. In our software, we utilize machine learning with self-learning scoring models. Working with the client data, the system learns to use prediction, classification, clustering, and association to process loan applications. For safety purposes, the system doesn’t just use the data client is providing but also pulls the available information from the credit bureaus it’s synchronized with. All the data is processed by the TurnKey Lender algorithms and is then presented in the form of a risk evaluation.
Having worked with lending businesses for several decades, key executives of TurnKey Lender used their experience to build a scorecard that helps achieve radical reduction of credit risks associated with borrowers in different business areas. We made sure that our built-in scorecard allows lenders to benefit from TurnKey Lender’s professional expertise joined with the world’s best practices of lending and credit scoring.
TurnKey Lender’s software lets you:
- Streamline your lending process;
- Improve loan officers’ efficiency;
- Reduce human bias in the lending decision;
- Adjust the credit policy according to your risk classification;
- Better quantify expected losses for different risk classes of borrowers.
You can either rely on our AI-driven scorecard or you can customize it. We made sure that all the Scoring Characteristics and Risk Segments can be adjusted manually.
Final thoughts
The importance of having a safe and intelligent scoring model can’t be overvalued for a lending business. And while you can rely on the traditional models, the industry has long moved forward so there are a lot of options on the table which can and will result in better returns and lower risks. If you’d like to give TurnKey Lender’s all-in-one lending automation platform a try and test our scoring engine for yourself, feel free to reach out to our sales staff at [email protected].
