14th June 2019

The Sweet Science of Credit Scoring

Credit scoring refers to a person’s ability/likelihood of repaying a debt. This score is a tool usually used by banks and lending organizations to hedge their “investment” and assess the possibilities of reclaiming the principal and interest amount.

The components that go into this magic formula are multiple and may vary from provider to provider. There are however, a few components that are rather standard such as historical financial behavior, current income, debt and types of credit history.

Think of risk scoring as the CV, the track record, the progress card of your financial history and behavior. The higher the score, the more are the chances your loan application is accepted.

The exact algorithm and weighted average of the aforementioned is not set in stone and differs amongst providers but the core concept is the same.


FICO, originally Fair, Isaac and Company, is a data analytics company based in San Jose, California focusing on credit rating services. The company, introduced the first credit risk score in 1981 and has since gained an industry-standard related name, a benchmark for risk scoring. Whilst many people refer to FICO as a single score, the reality is that FICO is the brand and there are multiple scores under that umbrella.

The Basic FICO score that most people refer to and use is broken down into five categories:

  • Payment history (35%)
  • Amounts owed (30%)
  • Length of credit history (15%)
  • Credit mix (10%)
  • New credit (10%)

How’s Credit Scoring Evolving?

The principle and implementation of risk scoring has been around for quite some time but like with most things, change was about to come sooner or later. The rapid growth of tech and data aggregation methods and techniques have been two of the biggest proponents in the evolution and development of risk scoring.

The biggest hurdle for risk scoring has always been information asymmetry. Some lending scenarios do not offer the necessary information for the credit score algorithm to generate a sufficient and believable score.

Take a university student wanting a loan for their studies who doesn’t have enough credit history. What about people in rural areas? Farmers? People that are not part of the traditional banking grid. The information on such demographics is obviously not enough to feed the algorithm.

In one of our previous posts we explored how artificial intelligence is changing the face of banking and even though credit scoring was not part of the discussion, it most certainly is one of the areas affected by AI. Machine learning and neural networks allow banks and other lending establishments to harvest, analyze and extract deeper insights from a wider range of data.

Each loan application is unique in its own right and an algorithm with standard variables does not allow flexibility or wiggle-room for that 1 in 10,000 type of credit behavior. Neural networks can help identify rare patterns and incorporate them in the algorithm’s memory, widening its scope.

Those five simple variable we mentioned before are a very basic way to assess the creditworthiness of a person. AI will allow the algorithm to “learn” on the go and incorporate new information in real time. The variables and data sets will increase and broaden considerably allowing for unique cases to be treated fairly.

credit score

Open Banking

The new EU directive, PSD2 (Revised Payment Service Directive), is seen as the beginning of open banking – an era where all banks must release data in a secure, standardized form and consumers have more control over their personal data and information.

This newly-found access to data allows for more transparency and a shift of focus in the credit scoring realm. Credit scoring has traditionally been about the past, the history of your transactions and the story they tell about your financial health and habits. Open banking allows for a shift to the presence.

Empowering customers to authorize permission to third-party providers to read real-time transaction data would add an extra layer of accuracy to the algorithm. In addition, by having control over their information, customers now can choose how much of it they want to share in exchange for better deals from lenders. Because of open banking, customers now have some leverage in the negotiations with a better chance of approval, better rates and more.

Transactional data is becoming the big takeaway of the open banking era by enhancing accuracy, diminishing risk and improve efficiency in decision making. Not only do they help identify risk behaviours, patterns and trends but they can also provide the necessary insight for customer-centric products.

Alternative credit scoring

What’s the old saying? “There’s more than one way to skin a cat.” A bit gruesome for our purposes but you get the point.

Apart from building on and developing the current algorithm-based way of scoring someone’s ability to repay a loan, alternative methods of assessments are also making their appearance on the horizon. These methods usually employ non-financial data.

Enter psychometric tests.

Psychometric testing bases its credit scores on a line of questioning analyzing personality and behavior, rather than credit history. These tests craft a very specific line of questioning aimed at investigating traits such as conscientiousness, extroversion, agreeableness and neuroticism. One might say such an approach is ambiguous and vague, not really focusing on facts but the truth is there is a solid argument to be made in favor of this alternate approach.

What is credit scoring trying to assess in the first place? Someone’s ability to repay a loan. If you reverse-engineer the decision-making process behind that what you’ll find is that the bedrock, the foundation of it all is built on someone’s values that propel them to make decisions.

The proof is in the pudding and there are institutions that have taken the plunge and tested the waters. Take Sovcombank for example. The Russian bank is already using psychometric testing and their take on it is interesting.

“We are using the EFL score to give young people a new way to get their first credit card,” said Sergey Khotimskiy, deputy CEO at Sovcombank. “Moving forward, we intend to use the score to rate the risk of people who don’t have a FICO Score because they are new to credit, and so don’t have a credit history to score. This is an innovative way for us to grow our market.”

The idea and prospect is exciting but it is very important to understand the parameters and challenges of this new method. Language and educational level of the test-taker are big points of contention as testing assumes people are of a certain demographic. Taking into consideration that psychometric tests are mainly used to reach the unbanked, or those without a financial track record, literacy is educational background will significantly vary.  

Tests should be tailored to each demographic and take into consideration cultural, topical and geographical parameters that might affect a participant’s test results.

credit score

Power to the consumer

Early last year, FICO announced a brand-new score, the UltraFICO. This new score will evaluate consumers by measuring their checking, savings accounts and other additional information to possibly boost their credit scores. This elevated power of what you choose to share with lenders will allow you to improve your credit score and make it considerably easier to apply for credit.

Here are some of the specifics that the UltraFICO will be looking at:

  • The amount of time your accounts have been open
  • How frequently you use your account
  • History of recent overdrafts
  • Your account balance

The purpose of the UltraFICO is twofold: help people improve their credit score and allow lenders to enter a market of 130 million consumers who are essentially excluded from the credit system because they don’t meet the current qualifications.

The future of credit scoring does not stop here. In an ideal world, consumers won’t have to provide information or even request a loan. Technology will be able to identify, allocate and categorize all the relevant information to build an accurate credit profile.

Their credit score won’t suffer based on their decisions as their decision will be made by the algorithm. When you decide to purchase something, the algorithm will decide which means of funding suits your current financial position, cash or credit, and make the decision accordingly so your score does not take a hit.