Do you wish to apply for your 1st loan? Well, if you don’t have a Bureau Score, your application may be rejected. But wait, how well do you fare on Social Media?
Let us answer this question by taking a very simple example. Assume there are 4 individuals and we know their social connections as given by the directed graph below:
Now, let’s calculate the importance of each of these individuals in terms of their social connections. In other words, calculate the importance or “Ranks” of A, B, C and D.
- Let us assume that the initial importance of each of these individuals as ¼.
- To calculate rank of A, we have to look at all the people with connections to A i.e. C
So, rank of A = (rank of C)/#outbound connections from C
- Similarly, rank of B = (rank of A)/#outbound connections from A
(rank of C)/#outbound connections from C
= (1/4)/2 + (1/4)/3 = 5/24
Person C has the highest rank as it has the highest number of connections to other people (A,B and D), at the same time, it has many connections from other people (A and D).
So, rank of a person is how well-connected (to and from) other persons. This is most clear when you compare the rank of B and D. Both B and D have 2 connections each from other people, however, D is more isolated as it has no connections to other people compared to B. Hence, the rank of B is higher than D. This defines the strength of the connections not just in terms of the number of connections but also the quality of those connections.
If we now were to use this model for underwriting a loan applicant, we could add variations to create a more sophisticated model:
• Focus on Linkedin connections;
• Give a higher weightage to connections working in the same company;
• Elements like education levels, designation, geographical location can also be factored in for calculating the final rank;
• Finally, once each individual has a rank, we need to decide on the cut-off score beyond which applications will be safely declined. This would again be based on historical data and making minor adjustments to segments and cut-offs in a test-control environment.
In addition, we need to combine strategies that segment out possibilities of ‘intent to fraud’ the social ranking. Example: Linked Account should be created atleast 1 year before loan application Date or should have a minimal number of connections, etc.
What is Social underwriting?
Traditionally, this assessment is done using the applicants credit history through past financial records, bank statements and other collateral provided by the customer. Banks also use credit bureau information that provides credit risk scores for individuals based on their personal information, financial data and other data taken from a variety of sources.
However, for some segments like the young, unbanked, lower income population, getting information is a challenge as they may not be very financially active i.e. not involved in organized financial sectors like having a bank account, insurance policies etc. This is where social underwriting shows a lot of promise.
What is social underwriting?
The basic principle of social underwriting is to tap into social networks of applicants and underwrite on the basic principles of character, community and trust. And that is where the story of underwriting changes. An applicant’s social connections and personal/professional interests can help us uncover aspects not only related to “ability to pay” (financial stability) but also “willingness to pay” (character) – both equally important to understand while extending credit.
The concept of social underwriting is not new. Grameen Bank founded by Professor Muhammad Yunus, the Nobel Peace Prize winner in 2006,extends credit to the poorest of the poor in Bangladesh and their underwriting is based on principles of character and community instead of financial collaterals. Thus, started social underwriting where wisdom of the crowd matters more than an individual/organization. Grameen Bank has been a very big success – As of September 2017, Grameen Bank has 8.92 million borrowers, 97 percent of whom are women. The system’s cornerstone is “social collateral”, where each borrower must belong to a five-member group. Although the group is not required to give any guarantee for a loan to its members and it is the sole responsibility of the individual to repay the loan, but because a group is involved, it oversees that loans are repaid. Since Grameen Bank does not extend further credit to a group in which a member defaults, group members contribute the defaulted amount and collects the money from the defaulted member later. This kind of societal pressure ensures loan repayment rates are high.
The role of Social Media in Social underwriting
With the advent of social media, social underwriting hastruly taken wings. Social media sites, where people themselves are providing information on their social networks, are a goldmine of information. Such information can be leveraged for social underwriting where people’s connections, their interests and their personal/professional profiles can reveal their character and financial stability. In fact, many start-ups like Lenddo has developed algorithms utilizing non-traditional data – social media activity, browsing behaviour, geolocation and mobile smartphone data to gain insights into consumer’s behaviour, network, and strength of their social connections and converting to a credit risk score that predicts people’s willingness and ability to repay a loan.
The biggest concern with social underwriting, specifically the usage of social media data, is that it could lead to discriminatory lending practices. Assessing a person’s financial stability based on his/her social connections may lead to discrimination on racial and geographical lines.
It might also appear absurd that financial standing is being judged on the many dubious online connections one develops on social media.
Social underwriting as a concept is here to stay. While skeptics may be concerned about discriminatory lending practices and privacy, the benefits of social underwriting can far outweigh these concerns. Social underwriting has enabled financial access to many segments of society, who otherwise would been considered too risky. For example, unbanked, underbanked and millennials, who might not have a lot of financial data available on them, thus making it difficult for traditional methods of underwriting to apply. However, using their social profiles and connections, newer methods of underwriting like social underwriting can help in their financial inclusion. With appropriate privacy and non-discriminatory practices, social underwriting can evolve into a very powerful tool for the future.
By Jahnavi Mahanta (Founder, @Deeplearningtrack)