Using transactional behaviour to calculate your credit score is not a new phenomenon. In fact, in a bid to make credit scoring fairer and more accessible to all, it is one of many alternative methods to traditional credit scores.
Photo credit: Chris Liverani
A growing area of alternative scoring is using transaction data to predict a person’s behaviour. These behaviour types might include when and where a person uses a credit card, or how many direct debits a person has. These types of datasets are becoming more widely available too, thanks to third-party banking APIs, PSD2, and GDPR.
There is value in using transaction data to predict behaviour since studying a person’s history can say a lot about a person’s financial health and how stable their income is. This includes how well a person can pay off any debts or financial obligations, now or in the future, which is essential information for lenders.
Traditional credit scoring already uses some behaviours to affect your score
You may be aware that many financial behaviours in fact already affect a person’s credit score. For example, if you have missed a credit card payment, this is likely to have a negative impact on your score. The number and frequency of new credit cards you apply for will have an effect, too. Applying for too many cards indicates an increased credit risk for lenders.
As well as activity on credit and debit cards, tracking behavioural transaction data can be found in other types of accounts, too. For example, utility or phone bills could provide an excellent source of information to a lender. Do you pay your bills on time? Have you been a responsible long-standing customer? These types of accounts are already a form of credit history and are already being used by some credit bureaus.
New types of behaviours that could affect your score
To modernise traditional credit scoring methodology, many companies are looking at ways to alternatively assess an individual's candidacy for credit. One of the leading trends at the moment is analysing financial transaction behaviours, and using this information to make predictions on how likely a person is to pay back a loan.
There are many different types of transactional behaviour, but some of them might include:
- Frequency of transactions: including how many transactions you undertake in any given period.
- Timing of withdrawals: Timing, amount and frequency of cash withdrawals. For example, high volumes of late night cash withdrawals may not work in your favour, since this type of behaviour is often linked to someone who is unemployed.
- Cash spend: spiky cash spending could signal unpredictable and unreliable behaviour to lenders.
- Regular payments/expenses: Regular payments are a great indicator to a lender of your financial responsibility. Watch out for too many large inflexible expenses though, e.g., electricity or rent. A lender may be wary of your ability to take on even further financial responsibility because it is difficult to get rid of these commitments.
- Retailer types and locations: do you regularly go to casinos? Lender’s don’t like to deal with gambling types so if you do gamble often, this may go against your credit score.
- Customer service calls: surprisingly the number of times you call customer services may say a lot about your personality type. Someone who excessively needs help may not be the best suited to taking on the responsibility of a loan.
This is not an exhaustive list, just an illustrative example of what could be used to generate a wide range of predictive characteristics to estimate a person’s creditworthiness.
Photo credit: Artificial Photography
It’s worth noting that many of these behaviours may not necessarily correlate with someone’s credit score. For example, you could have an excellent score yet frequently enjoy gambling during the small hours of a Friday night.
“Actually, when we looked at the relationship between the salary and the amount spent on gambling, we saw that our models started to perform better" Joacim Strand, Data Scientist.
Analysing the types of behaviour above is one of the crucial methods to making the scoring system fairer and more accessible for people with no credit versus traditional methods.
Case studies of transactional data usage
While time-consuming, there are undoubtedly many benefits to using transactional data for future credit scoring. For example, Instantor has helped clients gain a 4-7 percentage uplift in GINI coefficients. FICO and Westpac found $6 million per year in increased revenue and reduced bad debt expense.
Many online lenders across the globe are already embracing using transaction data to aid lending decisions, including BBVA Madiva in Spain and Inbank in Latvia. In addition to banks embracing this new type of alternative credit scoring, several institutions with a licence are also starting to jump on the bandwagon.
Analysis of transactional data is now a reality. These alternative credit scoring models are outperforming traditional credit scoring models and proving their value, providing any company which uses them a substantial competitive advantage. Using transactional bank data and other sources of data are a crucial methods to reliably understand whether a person is likely to default or not, which is the ultimate question lenders are looking to answer.