Jessica Leber over at Fast Company reports on a number of startups that aim to use Big Data analytics as a decision support tool to completely replace the FICO score.
AvantCredit, a Chicago based start up uses many more factors to evaluate credit worthiness, including an applicant’s use of social media and prepaid cell phones. The company believes that its use of analytics will allow AvantCredit to offer better rates to people who are considered to be higher credit risks, or making loans to people who could not get one from a traditional bank at all.
Since it was founded a year ago, AvantCredit has made more than 5,500 loans for a total of $15 million, says co-founder John Sun. It recently raised $20 million to expand to more states. The company targets middle-income families with credit scores in the 580 to 700 range, just below what is considered a “good” score by traditional banks. “Because these applicants are often turned down by banks, some may be forced to pay higher rates than they deserve with payday lenders, says Sun. On AvantCredit, the application is done fully online and an automated decision–and the rate they’ll pay–is made within minutes. Transparency is the goal, and the company has skin in the game because it backs its own loans.”
Current credit scoring is slow, limited, and often based on dated or incorrect information. Corrections and updates can take several weeks to post, and lenders are constantly working with old information. The scoring itself is based on formulas the industry will not disclose, leaving no way to judge its accuracy or relevance. Data is limited to static reports on past accounts that often supply dirty data, and the credit reporting agencies do not use real predictive analysis. Evidence of this can be found in the recently reported on Equifax’s recent dirty data blunder in which a judge awarded the plaintiff $18.6 million.
A Federal Trade Commission study of 1,001 consumers who reviewed 2,968 of their credit reports found 21 percent contained errors. The survey, which is required as part of a 2003 law, found that 5 percent of the errors represented issues that would lead consumers to be denied credit.
All data is credit data,” said Shawn Budde from Zest Finance. The company’s software tabulates thousands of variables that it has found to correlate with the likelihood of repayment even down to whether a person uses proper capitalization on the web form.
Banks can be slow to change, but better credit models might extend loans on fairer terms to a large swath of society. Payday lenders, which can trap people in cycles of debt, cost the economy nearly $1 billion in 2011 alone, according a recent report by the Insight Center for Community Economic Development.
Read the Full Story.