Can Social Media Activities Influence Your Credit Scores?
Three companies recently made headlines by claiming to use social media data as a basis for determining credit risk. The companies, Lenddo, Kreditech and Kabbage, are currently extending credit to consumers or businesses using non-traditional data, including social media activity as well as PayPal and eBay account information. The question is…is this legal and does it actually work?
The theory behind the use of social media data is “birds of a feather flock together.” Lenddo’s CEO suggests that “humans are really good at knowing who is trustworthy and reliable in their community.” As such, Lenddo will penalize you if you’re friends with someone on Facebook who has been late paying back a Lenddo loan. Seems more like “guilt by association.”
In the past it was safe to assume that a lender would only consider your risk when you applied for a loan, unless you applied jointly. Today, however, it seems like how your friends manage their credit is an interesting enough variable that the spotlight is being somewhat refocused and expanded.
There are a variety of problems with using social media data as a risk assessment tool. The first, and probably the most significant, is U.S Federal law. The ECOA (Equal Credit Opportunity Act) prohibits the use of a scoring system for lending unless that system is empirically derived and statistically and demonstrably sound.
What all of that means in English is that any scoring system has to be built using scientific methods and it has to actually work, meaning it has to be proven to rank order consumers based on risk odds. If your scoring system doesn’t actually work then it’s illegal to use, and ineffective.
While nobody is suggesting the models that use social media data aren’t empirically derived, the jury is still out on the second requirement, which is the demonstrably sound aspect of the ECOA. Traditional credit scores have long proven to be demonstrably sound, which means consumers with better scores default less often than consumers with lower credit scores. In fact, FICO makes available validation summaries when they develop or redevelop their scoring systems, thus making the job easier for lenders to prove that it’s in ECOA compliance.
Social media models may or may not be demonstrably sound, we just don’t know. There is no data available that proves your Facebook friends list or eBay “Feedback” rating is predictive of credit risk. That’s why the use of this social data is either being used for non-U.S lending or being used in the U.S in conjunction with more traditional underwriting tools, like credit reports and FICO scores.
Another problem with using social media data for underwriting is the ease with which it can be manipulated and controlled by the potential borrower. You have full control over your Facebook activities, LinkedIn contacts, and your Tweets. You don’t have full control over your credit reports and credit scores.
Imagine what would happen if large U.S banks started using Facebook as a tool to assess your risk. It would only be a matter of time before consumers would friend or de-friend accordingly based on the impact on their social media credit risk score. Think of it like credit repair or piggybacking in the non-credit report world.
There’s also a potential issue with access to your social media information. While these small lenders don’t seem to have a problem accessing your social media networks, the networks may have a problem with them accessing your data. Can you imagine the impact to Facebook if it were widely announced that your friend’s list could hurt your ability to get a loan?
Social media networks could make it either impossible or difficult for these lenders to gain access to your information on the large scale. That would essentially put them A) out of business or B) out of the social media evaluation business. It would be the equivalent of a lender losing access to credit reports and credit scores…not good.
Still, while all of the talk about social media and lending is interesting and certainly thought provoking, it’s doesn’t seem to be the wave of the future. The data that will be used by lenders to assess risk will end up being the data that’s most predictive of risk. That has always been true.