Lenddo (http://www.lenddo.com/) has a 5 year track record in using social, email and mobile-phone data to underwrite unsecured personal loans in Philippines, Mexico and Colombia. Their underwriting model quality measured with the gini coefficient is between 0.32 and 0.39.
Prosper and Lending Club are using thousands of variables to determine whether to fund prime & super prime borrowers. They have an approval rate of around 5-10% only. Even though they are drawing data from people who have usually long banking histories, high FICO scores and other financial & repayment info which creates many financial data points which are analysed and researched by hundreds of data scientists, the companies still have a default rate in excess of 5%. Compare this to micro finance lenders in developing countries who serve the under banked poor and are able to cap their default rates at around 2%. What is the micro lenders secret sauce, which is not being captured by the algorithms of tech wizards in Silicon Valley? The answer to that has been known since Grameen Bank dazzled the finance world with its business model; the key is to understand society and its Social nuances. The micro lender had pioneered a group lending system where it lends to an individual only when he or she belongs to a 5 member group. The group was not the guarantor for its member’s loans but further credit was not extended to a group when any of the members defaulted. This created societal pressure on the delinquent to repay the loan and reduced fraud. The important question to emerge is how to bring these social nuances into play via technology at a larger scale; how to reduce operating costs and to quantify the social elements for better risk taking. Lenddo has emerged as the pioneer in the science to evaluate all these unique factors and has leveraged social media and email to bring about a massive upheaval in the market.
Lenddo was started in 2011 as an online lender in Philippines by serial entrepreneur Jeffrey Stewart and management consultant Richard Eldridge. The company extended its lending operations to Colombia and Mexico in the next 2 years. The company launched LenddoScore, a proprietary credit rating system without any traditional banking inputs in October 2013. In January 2015, after having proven that the company can underwrite loans successfully without using traditional data, the company started offering its algorithm to third parties for lending and verification purposes . The company has started working with brick and mortar banks, p2p lenders, telecom companies and other financial institutions that are looking to make better and faster decisions for under banked clients. It is leveraging new data sources to address risk by not focusing on reducing default rates but by managing default to increase profitability. The company raised a combined 14 million dollars in Series A in 2012 and 2013. It closed an undisclosed Series B in October 2015 led by AT Capital and Life.SREDA. Blumberg Capital, Golden Gate Ventures & Omidyar Network were investors in both Series A & B. Accel, iNovia Capital, Kickstart Ventures, Lumia Capital and Skype founder Toivo Annus were investors in Series A. The company has raised capital from cream of the VC world and has added Vince Passione (Founder CEO of Lendkey) and John Elton of Greycroft Partners as Board members along with the other investor representatives.
In an interview with Lending-Times.com, Florentin Lenoir of Lenddo explained that “LenddoScore is a numerical value ranging from 1 to 1000, which measures the customer’s potential credit risk at the point of application. It is also a predictor of a customer’s character or willingness to make payments. The score is based exclusively on non-traditional data derived from the customer’s social data and online behaviour. The formula is Lenddo’s Intellectual Property and is tested and monitored regularly by the Lenddo’s Data Science team.” It is developed by maintaining an elaborate Trustgraph which maps relationships between more than 120 million social media profiles on Facebook, Email, Linkedin, Twitter, Yahoo etc. It analyses 2.6 billion individual pieces of communication for decision making and for a typical loan applicant examines more than 12,000 data points. Lenddo has invented a graph variable methodology- Archano Score; it is a proprietary version of a PageRank-like scoring algorithm, incorporating attributes of each member and their social group. Its software goes through the mails of the applicant to better understand the applicant and his chances of default. Social Media analysis is used as a verification tool to detect fraud. Though it might seem that very few applicants would be comfortable in providing access to their personal mail, reality is that the opportunity of credit is far more important to the unbanked than privacy .The company is leveraging android data from the applicants smart phone where it does not have access to emails.
Lending was carried out using exclusively non-traditional data for a borrower group in Philippines. The country had a sample Gini Coefficient of .25 to .32, whereas the Lenddo model had a Gini of .32 to .39 using non traditional data. (Gini Coefficients are used in the banking world to evaluate the predictive power of credit scoring tools. A Gini Coefficient can help a lender understand how good its credit score is at predicting who will repay and who will default on a loan. The better a credit score, the better it is at giving lower scores to riskier applicants, and higher scores to safer applicants. A Gini Coefficient is merely a scale of predictive power from 0 to 1. A higher Gini means more predictive power, a lower Gini means less predictive power). The company used the model developed for Philippines on Mexico, without any country customization and still was able to get better results in Mexico than Philippines. The company claims that by combing LenddoScore with credit scores available through traditional agencies, it is able to increase the approval rate by 50% and reduce the risk by 12%, thereby saving on precious marketing dollars. This was proven by testing LenddoScore by providing more than 2,000 loans in Colombia.
Present and the Future
The company has now more than 20 clients in 15 countries across the globe and around 10,000 applicants are onboarded towards financial inclusion every month due to approvals via LenddoScore. Two p2p lenders in India, LendingKart and i-Lend are partnering with the start-up to co-opt the non traditional credit scoring. The company charges for consulting, credit risk analysis and helps clients understand how various variables interact in the credit model. Secondly it charges for verification and credit scoring, which starts from 1-2$ per user and can go down to 50 cents on high volumes. The company is underwriting more than 10,000 loans per month and have integrated their API with enterprise software solution providers-Cloud Lending and Mambu. Its social underwriting experience is not only limited to social networks like Facebook, especially since its rumoured that Facebook is not sharing its API with lenders as it aims to launch its own lending module. The company’s only competitor even on radar seems to be a “FriendlyScore” based out of London and specialising only in social media analysis. It has raised only 400,000$ till date. Lenddo has opened an office in New York and is looking to serve the 45 million under & unbanked residents of America. The company is on the path of revolutionizing how credit bureaus operate across the world and would force them to incorporate non traditional social indicators to present a more reflective credit score. Though there are many p2p lenders’ leveraging social media to understand the behaviour of borrowers, Lenddo is the only one to exclusively use such social data points to provide lending. More importantly it has proven its model by not just creating an algorithm and stress testing but by actually giving out thousands of loans across developing countries to come out ahead on comparison to traditional credit score based loans. It is impertinent to note that Lenddo’s growth will definitely see copycat attempts to match its success, but the company seems to have pulled far ahead of any potential rivals. It will be interesting to see whether the company is able to do to FICO what Uber did to taxi companies.
Author : Heena Dhir and George Popescu