Artificial Intelligence in Financial Lending

I remember the 90s when I wanted to get a home loan and it took me 3 months to complete the process from providing all the hard copies of my income, tax returns, identity proofs then bank checked my creditworthiness & provided the approval.

Today everybody has some kind of loans like home loan, auto loan, education loan, two wheeler loan or even loan to buy appliances like HD TV and Refrigerator.

How do they assess your creditworthiness? There are so many cases of defaulters, which keeps increasing and hence established banks or lenders constantly looking for ways to improve the returns or proactively identify risks.

Lenders traditionally make decisions based on a loan applicant’s credit score, a three-digit number obtained from credit bureaus such as the TransUnion, Experian, and Equifax.  But these credit scores are based solely on credit-history and do not take into account rich data available, which can potentially give lenders access to data points as varied as online purchases, the strength of social connections and travel patterns. When viewed this data holistically, lenders can get a complete picture of potential borrowers & can significantly improve their ability to predict loan defaults.

Today digital transformation has changed everything. While the interest rate and closing costs on loans are still primary considerations, the speed, simplicity, transparency and customer service of the entire process is important.

As the purchasing power among millennials & gen Z continues to increase, they tend to purchase property and acquire assets that will provide stability & generate wealth.

The ability to cross-sell to these customers on loan products drives a significant portion of new loans. The difference for a digital-first customer is that they do their shopping online and may select an alternative provider based on the right combination of cost and ease of process.

Artificial Intelligence is used today, to determine the creditworthiness of those who don’t have any credit history like students or immigrants etc. It also helps to improve customer experience, e.g. by showing pre-approved loan amount. AI makes loan approvals quick and easy, reduce operational costs and these savings can then be extended to customers in the form of lower rates. Artificial Intelligence can process large amounts of data that human underwriters would simply not be able to make sense of.

Machine learning streamlines the process, drastically reduces the likelihood of errors and significantly cuts down the time it takes to approve a loan and disburse funds to the borrower, thereby enhancing the customer experience.

AI & Machine learning also helps to detect fraud by comparing customer behavior with the baseline data of normal customers and removing outliers.

Today apart from credit score and income, lenders are also looking at the digital footprint, payment data from other sources, purchase history, professional reputation from LinkedIn and other sources.

This is called alternative data sourcing. The use of machine learning to analyze this alternative data in loans and credit rating is going to raise some privacy, ethical, and legal concerns.

The future of digital lending will reduce the friction associated with the borrowing process, eliminating paperwork and moving all of the steps of the customer journey to an online and mobile capability. AI and Machine learning will become an inherent part of financial lending.

Machine Learning - The brain of Digital Transformation

We are all familiar with machine learning in our everyday lives. Both Amazon and Netflix use machine learning to learn our preferences and provide a better shopping and movie experience.

Artificial intelligence (AI) has stormed the world today. It is an umbrella term that includes multiple technologies, such as machine learning, deep learning, and computer vision, natural language processing (NLP), machine reasoning, and strong AI.

Organizations are using machine learning for various insights they want to know about consumers, products, vendors and take actions which will help grow the business, increase the consumer satisfaction or decrease the costs.

Here are some top use cases for machine learning:

·     Predicting & preventing cyber-attacks: With WannaCry making havoc in many organizations, machine learning algorithms have become extremely important to look for patterns in how the data is accessed, and report anomalies that could predict security breaches.

·     Algorithmic Trading: Today many of financial trading decisions are made using algorithmic trading at higher speed, to make huge profits.

·     Fraud Detection: This is still one of the key issues in all the financial transactions. With the help of deep learning/artificial intelligence, the identification and prediction of frauds have become more accurate.

·     Recommendation Engines: In this digital age, every business is trying hyper-personalization using recommendation engines to give you a right offer at right time.

·     Predictive Maintenance: With embedded sensors as the Internet of Things, many of the heavy industrial machinery manufacturers are applying machine learning to predict the failures in advance, to avoid the costly downtime and improve efficiency.

·     Text Classification: Machine Learning with NLP is used to detect spam, define the topic of a news article or document categorization.

·     Predict patient’s readmission rates: By taking into consideration patient’s history, length of stay in hospitals, lab results, doctor’s notes, hospitals now can predict readmission to avoid penalties and improve patient care.

·     Imaging Analytics: Machine learning can supplement the skills of doctors by identifying subtle changes in imaging scans more quickly, which can lead to earlier and more accurate diagnoses.

·     Sentiment Analysis: Today, it is important to know consumer emotions while they are interacting with your business and use that for improving customer satisfaction. Nestle, Toyota is spending huge money and efforts on keeping their customer’s happy.

·     Detecting drug reactions: With Association analysis on healthcare data like-the drugs taken by patients, history & vitals of each patient, good or bad drug effects etc; drug manufacturers identify the combination of patient characteristics and medications that lead to adverse side effects of the drugs.

·     Credit Scoring & Risk Analytics: Using machine learning to score the creditworthiness of cardholders, defaulters, and risk analytics.

·     Recruitment for Clinical Trials: Patients are identified to enroll into clinical trials based on history, drug effects

With today’s advanced cognitive computing capabilities, image/speech recognition, language translation using NLP has become a reality which is used in very innovative use cases.

Machine learning is nothing new to us but today it has become the brain of digital transformation. In future, machine learning will be like air and water as an essential part of our lives.

Credit Score Cards

 

Information Technology as an industry has grown up in leaps and bounds. You may not find any organization on the planet which does not have any IT involved.  This has given rise to a lot of jobs supporting the IT functions. Salaries have increased tremendously in IT compared to other business areas. The overall economy had gone up which increased the tendency of people to afford & buy more & more.

This has increased the usage of Credit in everyday life. “Buy now pay later” syndrome became common. Everyone started using the credit cards and also started availing credit or loans for big purchases like home, car etc.

Eventually, this resulted in many people avoiding or defaulting on the payments. This is where applying analytics for assessment of the risk of providing the credit came along and the birth of credit scoring.

Credit Risk is the risk of losing a bank or credit giving company will incur when Customer does not repay the mortgage, unsecured personal loan, auto loan, credit card amount, overdraft etc.

In the early days of lending businesses used to judge borrowers based on 5 Cs:

• The character of the applicant
• The capacity of the applicant to borrow
• Capital as backup
• Collateral as security for credit
• Conditions which were mostly external factors

Then Credit Scoring was introduced by Fair Isaac which is now commonly known as FICO score.

Credit Scoring in simple terms giving some numbers to customers based on certain parameters like age, earnings, accommodation type (owned or rented), expense history & payment history etc.

There are 3 types of Scorecards which are currently used:

Application Scorecard:  This is mainly used in scoring the customer's applications for credit. This tries to

predict the probability that the customer would become "bad". The score given to a customer is usually a three or four digit integer which is finally used to approve or reject the credit application of the customer. This is where you get messages from Banks that you have pre-approved loans or Credit cards.

Behavioral Scorecard: This is mainly used to identify or predict which of the existing customers are likely defaults on the payment so alternative measures can be taken to contact the customers & ensure that payments are received on time.

Collection Scorecards – This is mainly used to arrive at how much loss the company will incur, due to nonpayment from groups of Customers.

How businesses are using Credit Scorecards:

• Banks are using them to separate good borrowers from bad borrowers
• Financial institutions are using it to determine credit limits
• Early detection of high-risk account holders to reduce potential losses
• Improved debt collection
• Insurance companies are using it for the cost of insurance product for a Customer

Today applying analytics to the data to get such insights is of prime importance.