CNNMoney reported yesterday that companies are now looking at your social networks to determine your creditworthiness, or in other words social underwriting is here. Social underwriting promises to uncover things that traditional underwriting techniques can not. Between yesterday and that tomorrow lies today where we really need to think about what this means.
While I may have lead in with the idea of social underwriting, I think the issue is much bigger than that. Social data is just one aspect of the big data collection craze going on. Big data is all about analyzing large unstructured data sets. Social becomes just one part of that data set. What are the other sources of data for big data? Loyalty program data, item purchase data, travel data, etc. Pulling all this together allows for huge insights to be gained. Should those insights be a value add for users or a value detractor?
Here is my problem with this idea. It is too early, for one. And using data in this manner has the possibility of gutting the interactions that make social so compelling for users. If you know that friending certain people will keep you from getting a loan, will you still friend them? What about talking honestly about your problems with supportive people? How much would you share if a loan officer was going to look at that at some point? Is it better to evaluate credit risk or to help people be better people?
Don’t get me wrong, if I am on the business side I want as much data as I can get for every decision I can make. The problem is that the folks that are innovating around this are not of the social generation. They see the promise of business process but they don’t understand why social is there. Social is by the user, of the user, for the user, right? Have the folks who are rolling out these ideas been active members of social, sharing things with people, connecting with people as much as the consumers they are rating.
Does Social Underwriting Even Work?
This is where I think it is too early. I am skeptical of whether this works at all right now. There is good data there, don’t get me wrong, but what about all the fluff. How many false negatives are out there waiting to trip you up. If I friend people who are allegedly delinquent or behind on payments, is that because I am like them, or that I am supporting them as they try and put their life back together after a job loss, or something else.
How do you filter out these false negatives? Is one association enough to deny me credit? Two? What about other things you get evaluated for.
Life Insurance? Say you have alot of friends who are in cancer-support groups. Should I charge you a higher life insurance premium? What about being a fan of lots of fast food chains?
Auto Insurance? Have lots of friends who talk about accidents they have been in? What about risky behavior? What about pictures next to a dented car? Or talk about going 5,000 miles past the suggested oil change date? What is truly a data point, and what is false?
Regulators are going to have a field day, folks
In the United States, there are regulations that govern what data can be used, and how it can be used with respect to credit decisions- the Fair Credit Reporting Act (FRCA). I don’t think a system like this passes the smell test in the United States. There is just too much room for abuse. I also fail to see how you can truly explain why you got denied a loan and somebody similarly situated got one. I like the idea of crowdsourcing and vouching for someone. That is a value add. When you start using things like this as a value detractor (who shouldn’t get a loan) you start to look bad to me.
Do The Right Thing
If you are going to incorporate social into your business processes, you need to always keep in mind what the right thing is. Also, who the right thing is for. If you are in the business of making money off of loans to people, focus on the positive aspects of data, not the negative aspects of data. Focus on the connections that can be made through social. Use social to find friends of good creditors. This all just seems like a big gotcha to me.