How Artificial Intelligence is changing the Credit Scoring Markets?

Access to credit is the single most important aspect of an emerging economy. Credit scoring helps individuals and businesses take financial loans by which they can invest, buy, and access incredible amounts of benefits that upheave the economy and create a better quality of life.

However, with a largely unbanked population as is the case in many of the emerging markets, this can cut access to finances and hinder financial inclusion initiatives. Furthermore, since many of the businesses that operate in these markets are highly dependent on cash-based transaction, this leads to difficulties when small businesses try to get access to financing to grow. This is where the power of AI can have a transformative role, enabling such economies to grow and prosper by making credit accessible.

So, what is a credit score?

A credit score is a metric by which institutions and lending organizations can measure the credit worthiness of a certain individual or business, by which they make sure that a financing service can be paid back.

The Egyptian bureau for credit (I-score) has the following weights for each feature to calculate the credit score for a certain individual

https://www.i-score.com.eg/ar/information-for-individuals/home/

The definition of each category is defined according to the Egyptian Credit Bureau is as follows:

Payment history (35%)

Is the amounts due; either paid on time or there are any delays reported; that are shown in the data reported with number of days where payment was delayed and how frequent delays are reported. This in addition to if there is any negative data reported from The Central Bank of Egypt.

Outstanding Debt (30%)

Percentage of available credit card balances availed to the customer for usage.

Credit History Length (15%)

Your Score takes into account:

• How long your credit accounts have been established, including the age of your oldest account, the age of your newest account and an average age of all your accounts.

• How long specific credit accounts have been established?

• How long it has been since you used certain accounts.

Pursuit of new credit (10%)

• Inquiries: Number of recent inquiries for the last twelve months.

• New Facilities: Number of facilities opened in the last year.

Credit Mix (10%)

What is the mix of credit product types?

• Revolving Credit: Number of credit cards

• Installment Credit: Percent of accounts that are installment loans

These are the attributes and their respective weight in scoring an individual credit. This score is then shared with credit facilitators, and based on their internal lending policies, they either approve or reject an individual and build their financing terms (Payment Schedule, Amount of loan, etc.).

The goal is to ensure that the borrower is willing and capable to pay back the loan. Certainly, there is always risk involved, although, the goal is to decrease the chance of an individual unable to pay back their dues.

As you can see, for an unbanked population you might see that there is a bit of a problem in providing such a score. To that matter, many financial institutions, have their own investigative activities by which they can assess a person’s credit worthiness, through interviews, home visits, or Inquiring in-person.

For Example, one of the most important features for banks, and is fairly common across most institutions is income. In emerging economies most people rely on ‘free work’, which means that they might not have proof of regular or consistent income, such as having a monthly salary. This kills their chances of ever getting a loan. Moreover, many applicants might be sharing assets, while, others can be business owners with no regular income, all of which might unfairly score an individual.

Businesses also follow a similar trend when asking for loans. They are usually analyzed on their finances, expenditures, outstanding contracts etc. When it comes to Small and Medium businesses, they usually operate in unfamiliar sectors, not having enough assets as collateral, work in informal business, such as with cash only transactions with no invoices for example.

Furthermore, the inquiry process takes so long, and the amount to be financed is not that big, at least from the perspective of the lending facility, with more probable risk. So, many lending facilities might be more interested in big ticket loans, for their safer "bets", besides, bigger tickets have bigger upsides and are more appealing.

Understanding this dilemma, now comes the role of Artificial Intelligence to provide an unprecedented opportunity to make credit accessible to both, individuals and Small & Medium businesses.

Artificial intelligence is a type of software engineering, where algorithms are designed to find patterns in the data, to generate a logic, or intelligence as output. AI doesn’t only take in quantitative data; it can take qualitative data as well, such as images, text, social media, phone, macro-economic data, research reports, etc.

Furthermore, an A.I. model trained on the right datasets can be extremely powerful in predicting the willingness and capabilities of someone’s payback potential. Not to mention, that Artificial intelligence can also be very dependable in detecting if someone is lying on their application, by flagging abnormal and inconsistent combinations of filled fields in an application for example.

Usually, the way to go about this is to train an A.I. algorithm on a large dataset of approved, rejected, defaulted, and delinquent transactions. The more examples of each, the better. If the data is coming from a certain company, then the A.I. logic in this case might be biased on the credit policies of the institution that owns the data. Although, this approach can be accurate in predicting defaults for the lending company, the same might not work on a more general population, in a different country for example.

Furthermore, business to business financing solutions, can also utilise qualitative data to predict a business performance, such as management profiles, their LinkedIn network, macroeconomics data, CRM or ERP data, news mentions, product segments, business location and so much more data to produce an accurate business performance trajectory.

Indeed, this will expedite the inquiry processes, as per an institution’s policy, which can add to customer experience and encourage more people to choose the company for the different financing services.

The data can also be augmented with more features, such as applicant social media data for example to allow the A.I. to base their assessment on social media data as well. Phone data is especially important when it comes to alternative credit scoring. More than 14,000+ variables can be accessed from a phone, some of which can be powerful predictors of one’s willingness and capabilities to pay back their loans. With this data, an A.I. can validate a person’s living place and maybe can find a correlation between the capability to pay, the frequency of one charging their phone and how many calls they do per day (This is just for clarification purposes, and does not reflect reality of the matter) .

One of the pitfalls when working on such project is the inherent biases in the data. An A.I. model is dependent on the data it is trained on, and if the data is biased against a certain race, sex, location, or occupation, A.I. can be counterproductive.

For the first time in modern history, Artificial Intelligence is creating an unprecedented opportunity for the credit market. With all the digital Transformation initiatives and financial inclusion activities currently happening in Egypt, there is a lot of data being Generated, and the adoption of Artificial Intelligence in Egypt can uplift the Credit Market big time.

A.I. has the power to put emerging countries on a socioeconomic trajectory if the credit problem is solved. We live in a time where the computing power and the algorithms for such use case are readily accessible, not to mention, the data explosion happening every minute, with the growth of internet and mobile technologies all over the world.

This is one more example of how A.I. can change the world

Can't get enough?

Here's an in depth case study on how we leveraged an AI powered credit scoring algorithm to redefining credit risk: https://www.synapse-analytics.io/credit-scoring

Or contact Us to know more about our A.I. powered Credit Scoring Services.