How Alternative Data is Changing the Credit Risk Modeling Landscape?
Many Indians struggle to get a loan because of insufficient credit history and the absence of a credit score. This is because credit scoring models have primarily considered traditional data to derive the creditworthiness of a customer or a business. By traditional data, it means the data from a lender own payment history, credit utilization, credit history, types of credit, and new credit. This creditworthiness is formed into a 3 digit number called a consumer credit score that attempts to measure the person credit payment pattern. All this is achieved using traditional data.
However, as mentioned before, a considerable population does not have a credit history, and are practically invisible to the credit bureaus – hence they fall into the category of Credit Invisibles?. For calculating the creditworthiness of such people, there is a need to consider what is called Alternative Data. Alternative data is any data that is not directly related to a consumer credit behavior. This data can be derived using a number of unconventional sources. Using this data, credit risk management models can develop a credit score so that lenders can safely extend credit to an entirely new customer base, and consumers and small businesses can get access to the credit they deserve.
Where can we find Alternative Data?
There are various sources of alternative data that can be used for credit risk assessment to build a credible personality of a consumer. Let us take a look at the apparent few.
Retailer/ Purchase data:
In today times of retail malls and online transactions, much of the customer data is recorded in detail. This data is best to provide behavioral insights. Purchases detailed on a loyalty card, for instance, would give a fair idea of a particular family income and structure. However, this data is hard to retrieve outside marketing purposes as retailers do not generally sell such information to credit bureaus. But also, using certain tools, this data can still be accessed, albeit anonymously without personal information using only details such as banking transactions, browser cookies, and device IDs.
Social Media & Web Pages:
Your LinkedIn profile can tell about your employment history and professional connections. Your Instagram and Facebook profiles can indicate a lot about your lifestyle choices and spending habits. Although, not all information portrayed on social media can be taken on its face value. The value of this data can be understood to be far lower than the value of data with a stronger credit connection. How a customer navigates through your website, how long they stay and where they click can be predictive of various parameters too.
Telecom, Rent, and Utility bills:
Rental payments, electricity bills, water/gas bills, telephone bills, and all such periodic payments are among the most reliable alternative data sources. Monthly payments provide an accountable history of consistent behavior and financial responsibility. Telecom & electricity bills can reveal multiple traits about a customer such as full name, billing address, payments due and payments done, etc. All these points reveal whether the customer pays their dues on time and in full or otherwise.
Use of Machine Learning to Extract Useful Information
While it is true that there can be enough data sources, a mechanism like Machine learning will surely be required to extract only sensible & usable information from the sea of data. With large, unstructured data sets, the right use of ML & AI can help identify data patterns that relate specifically to credit risk.
The credit industry now needs to acknowledge the importance of financial inclusion and the lending gap it needs to close. Credit information companies have started seeking credit scoring model that include alternative data to establish a credit score. According to market insight, there are an estimated 3 billion adults worldwide who do have credit records. Tapping that market should be a priority for lenders.
