The team at European online lender October has published a blog post on credit risk analysis models including their construction and application.
October notes in an update that the Covid-19 health crisis has led to two situations: a rise in the number of new loan applications “related to the creation of state guarantees.” There’s also been an increase in “uncertainty about the probability of default.”
In order to address these risks, banking institutions and lending platforms should look into accelerating their transformation and also adopt a “reliable and fast risk scoring model,” October recommends.
The lending platform explains that they created their own risk scoring models. These are now accessible to third-parties with the introduction of October Connect, the company’s neolending tech for corporate finance.
Tejas Sherkar, Head of Data at October, says that a credit risk model is essentially a set of rules “to quantify the risk involved in extending credit to a borrower.” He adds that the rules and the data provided to them “determines the nature, complexity and the performance of the model.”
He also mentioned that the rating models mainly focus on reliably predicting the creditworthiness of the potential borrower. Meanwhile, scoring models can predict credit-worthiness as well as potential default, Sherkar explains.
“At October, we focus on building scoring models by means of predictive analytics. Once we have a clear vision of the question (or problem) to be solved, we can start building a model. In the case of October, the question was: how can we process loan applications in a fast, scalable and secure manner in order to help as many borrowers as possible while keeping our default risk low?”
Sherkar pointed out that “here, we are dealing with a binary (default vs non-default) classification problem.” We can begin by obtaining the relevant data from their data lake (a data store built in-house with enforced ACID properties) which includes existing firms in October’s portfolio and their repayment behavior, all historical or previous loan requests as well as their “associated financials, bank transaction data and default flags,” Sherkar added.
He also noted that this is usually “followed by a data cleaning step, where we look at the distribution of all data points related to historical loan requests, to treat outliers and missing values.” He further revealed that the main purpose of this exercise is “to understand our population, and build a representative dataset on which we can train our model.”
“At October, we use both linear or non-linear models trained on this representative dataset. Non-linear models are often considered to behave like a black-box, but we make use of SHAP to make non-linear models fully explainable.”
“After the model is trained and deployed in production, we monitor the data points (which the model uses for scoring) of the new loan requests over a period of time (usually 3-6 months).”
He further explained that if the statistical properties of these new data points have changed “significantly as compared to the last model training, it is likely we will re-train the model and deploy an improved iteration of it in production.”
However, this isn’t something “to be done lightly: we need to understand what changed in the population and the biases that were introduced,” he added.
Sherkar further noted:
“We are also on the lookout for new data points, either newly engineered from existing data or from suppliers, that could improve the performance of our model.”
He added that Magpie is an instant credit-risk scoring model which examines the financial (balance sheet + income statement) as well as behavioral information of the SME and provides a score “from 1 to 5 in the order of increasing probability of default.”
He also noted that Kea was introduced earlier this month. While commenting on the difference between Kea and Magpie scoring, Sherkar said:
“Under the hood, Magpie and Kea are built using the same class of machine learning models. However, they differ in the type of information analyzed and category of companies targeted.
Magpie looks at the borrowers financial and behavioral data to assess the probability of default (PD).” Meanwhile, Kea analyzes or looks closely at bank transactions and behavioral data to determine the borrower’s probability of default, he explained.
Responding to a question about what exactly Kea analyzes in the bank transactions of the company, Sherkar noted:
“Bank transactions provide a unique insight into the day-to-day operations of a company and Kea engineers many attributes to analyze the borrower’s ability and willingness to repay the prospective loan. The attributes range from whether costs are paid regularly, to existing loan repayment schedules, to late payments and bank balance trends over time.”
Addressing a question about the impact of the new DSP2 (open banking) regulation on the creation of this new risk model scoring, he said:
“The DSP2 regulation enables the client to securely share his/her company’s bank data with a lender (like October) via an API within seconds. A model like Kea (based entirely on bank transactions) can therefore analyse this data instantly, allowing for a safer and faster.”
Going on to explain what impact this risk scoring model might have on the credit process, he revealed:
“Risk scoring models like Magpie and Kea reduce the time to arrive at a credit decision and help in developing a scalable business. They also bring a certain predictability to the whole product offering, where we can let our partners and borrowers know early in the process about necessary steps to follow or documents to have handy.”
Addressing a question about whether there will be any tasks performed manually, Sherkar noted:
“While credit scoring is done automatically, we rely on the expertise of our Operations team to perform customer identification, some anti-fraud checks and due-diligence before funding the borrower.”
While commenting on the types of firms Kea scoring addresses, he added:
“At this moment, Kea scores micro-companies in France and in Italy. The loan amount can be up to €30k with or without state guarantee.”