The limitations of traditional Credit Scoring
Credit scoring, as a tool of lending, has been in vogue for over a generation. However, the problem with conventional models is that they restrict themselves to information pertaining to the formal banking sector. This excludes young people with no credit history in developed countries and poor people who are getting into the formal sectors for the first time in developing countries. Further, credit score is often a poor judge even when there is some credit history given a large portion of a person’s financial behavior isn’t included in conventional credit scores. For example, someone who earns a significant amount of money and spends it on a frequent basis often doesn’t have that showing up on his or her credit history. Nor do people who earn a significant amount of money and not spend it at all, show up! Further, a credit worthiness report isn’t merely a report that helps in lending money to a given individual. It can be used for a wide variety of applications, including estimating point of equilibrium between default and loan off take, if it’s done right.
What’s the alternative?
Enter “Enhanced Credit Score”. This method of scoring does not restrict itself to the conventional credit score, which is a one size fits all score for all consumers. Instead it optimizes the score for each purpose and enriches the data with external data elements to enable an accurate prediction for a specific purpose. For example, a person with a lifestyle that’s flamboyant and includes risky behavior correlates with greater risk than a person with similar income levels who’s got a quieter lifestyle. Crawling social media for lifestyle choices and studying its impact on actual default is one of the best ways to expand the base for lending while at the same time decreasing the rates of default. Behavioral credit scores of existing customers can be used in the early detection of high-risk accounts and enable targeted interventions, for example by pro-actively offering debt restructuring. Behavioral credit scores also form the basis for more accurate calculations of the total consumer credit risk exposure, which can result in a reduction of bad debt provision.
Similarly, the probability of recovering a loan that’s been defaulted on can be estimated based on enhanced scoring techniques. This helps lenders optimize their collection effort, focusing on maximizing recovery.
Enhanced Credit Scoring Models
There are a variety of model types, such as scorecards, decision trees or neural networks. When you evaluate, which model type is best suited for achieving your goals, you may want to consider criteria such as the ease of applying the model, the ease of understanding it and the ease of justifying it. At the same time, for each particular model of whatever type, it is important to assess its predictive performance, i.e. the accuracy of the scores that the model assigns to the applications and the consequences of the accept/reject decisions that it suggests. The best model will, therefore, be determined both by the purpose for which the model will be used and by the structure of the data set that it is validated on.
The traditional form of a credit-scoring model is a scorecard. This is a table that contains a number of questions that an applicant is asked (called characteristics) and for each such question a list of possible answers (called attributes). One such characteristic may, for example, be the age of the applicant, and the attributes for these characteristics then are a number of age ranges that an applicant can fall into. For each answer, the applicant receives a certain amount of points – more if the attribute is one of low risk, less vice versa. If the application’s total score exceeds a specified cut-off amount of points, it is recommended for acceptance. This is less of a model and more of a heuristic. It fails to learn and does poorly when the goal is to maximize the loan amount without default. It merely tries to avoid default.
Regression and Decision Trees:
These are both examples of other conventional techniques that are adopted by the lenders. However, there is a fundamental problem associated with these models: they diminish the richness of information that the organization can collect on the applicants and thereby erode the basis for future modeling. With the decision tree, we could see that there is such thing as a decision rule that is too easy to understand and thereby invites fraud.
A way forward is using is using a more generic model, such as Neural networks. But Neural Networks require a lot more features for them to work. Neural networks are extremely flexible models that combine combinations of characteristics in a variety of ways. Their predictive accuracy can, therefore, be far superior to scorecards and they don’t suffer from sharp ‘splits’ as decision trees do. However, it is virtually impossible to explain or understand the score that is produced for a particular application in any simple way. It can therefore be difficult to justify a decision that is made on the basis of a neural network model. A neural network of superior predictive power therefore is best suited for certain behavioral or collection scoring purposes, where the average accuracy of the prediction is more important than the insight into the score for each particular case.