A Big Problem for Big Data

In order to get a personal loan, credit card or mortgage, you usually need a credit history. If you’re one of the 64 million Americans without one, your limited or non-existent access to these products can be incredibly frustrating.

That’s why some groups have put together alternative consumer reporting models, which attempt to measure a person’s creditworthiness by looking at different data than credit card payments and loan applications. Instead, these groups use some of the data we each generate on a daily basis through social media, email, online shopping and more.

Big Data says it is swooping in to save the day by making up for the shortcomings of the established credit reporting industry. But according to a recent report from the National Consumer Law Center, Big Data is nothing but a big disappointment, at least as far as consumer reporting goes.

Is More Better?

Imagine you’re browsing an online shopping marketplace for new shoes. You’re trying to decide which pair you should buy, so you compare other shoppers’ reviews, read the item specifications, click through pictures and so on. If your search returns little information — no image, no user reviews, no detailed description — you’ll probably wouldn’t think twice about moving on to the next result. It helps to have a lot more information to consider, right?

Keeping with the shoe-shopping analogy, you’d really appreciate the five users who gave reviews, saying how well the shoes held up, and you’d be so thankful they said the shoe runs small, so you should order up a size.

Of course, you’d be really annoyed if you bought the shoe and it was way too big or fell apart after a week.

For obvious reasons, you want the information on which you base decisions to be accurate.

Credit scoring works a bit like your shoe shopping adventure, with your credit report serving as the reviews and information online: The scoring model takes your credit report, weighs the information in it according to an algorithm and spits out a number that helps a lender decide whether you’ll be a good borrower.

Well, guess what: The online-shopping review system isn’t perfect, and neither is credit reporting. Errors can appear on your credit report, and those errors can impact your credit scores if you aren’t vigilant in monitoring your credit. (You can get free credit reports every year with the major credit bureaus, and you can monitor your credit scores every month using the Credit Report Card, a free tool that updates two of your credit scores every month.)

The Consumer Financial Protection Bureau started accepting consumer complaints on credit reporting in October 2012, and about 75% of complaints have involved credit report inaccuracies. About 11% mentioned frustration with the disputing process.

But like shoes without any photos or reviews, millions of Americans are without credit histories — they’re not necessarily bad credit risks, but the lender has no information to make a decision.

Here’s where data companies want to step in.

Can Big Data Fill the Void?

Everything we do on Internet-enabled devices generates massive amounts of information, and data collectors can piece together portraits of us based on things they tie to our email addresses, social media accounts and Web searches. While the scale of this data is impressive, it’s also a burden.

“If data aggregators or data analysts harvest data from dozens of sources, inaccuracies are harder to detect and the source of the error can be difficult to identify,” said the report from the National Consumer Law Center called “Big Data: A Big Disappointment for Scoring Consumer Credit Risk.”

The NCLC had volunteers request their information from five big data companies (Acxiom, eBureau, ID Analytics, Intelius and Spokeo), and they generated all sorts of errors: incorrect email addresses, education levels, income, relatives, occupations, social media accounts and more.

“Big Data only generates better results if the algorithm is predictive and if the data that feeds it is accurate,” the report says. “At present, there is no mechanism in place to ensure the integrity of credit scores generated by Big Data.”