Extending Credit or Reducing Fraud: The Story Behind the Data

How well do consumer lenders know their customers?

I love it when a company gets to know me to make my life easier. Amazon suggests great books I should read. Spotify delivers new playlists to me daily.

Consumer credit companies do this too. They get in touch with me when they recognize I’ve reached a new milestone. When I got my first job out of college, my mailbox immediately started to fill with a steam of pre-approved offers. They also quickly flag any suspicious expenses that look fraudulent, like my credit card showing my buying gas in Siberia at 3am when I’m actually in my apartment on the East Coast.

However, unlike other consumer-oriented brands, consumer lenders often don’t know us as well as they should. This affects their ability to extend the right credit limit or to accurately identify fraudsters. The types of data they rely on to make these decisions has a lot to do it with it.

Read a case study on enabling a payment provider to harness more data and reduce fraud.

Let’s start with credit.

I was a little surprised recently when, after my friend took a new job with a much better salary, it took several months for her bank to adjust her credit line to her new professional situation. Because my friend was making more, she was spending more. Specifically, she was charging more to her cards. In industry parlance, this means she was revolving a higher fraction of her credit utilization. This all makes sense. But why didn’t her bank pick up on her increased usage and step in to increase her credit limit?

As it turns out, people who have sudden and significant increases in their revolving credit utilization generally fall into two categories: the recently promoted and the recently unemployed. These two groups behave in a remarkably similar manner until about four months since their initial utilization spike. After four months, the recently promoted continue their new spending patterns without any problems. But the unemployed start to run into real financial difficulties, which are only compounded by their added debt levels.

The problem here is obvious: how do we tell these groups apart? The data that card companies have internally, such as transaction patterns, can’t be used to distinguish between these customers.

The answer for these firms is also obvious: look at other types of data. Have the people you’re worrying about been shopping around for a new car? A new house? People looking to make big purchases are probably financially stable. But if you notice that they’re no longer buying lunch with their colleagues or that they’ve recently changed their job title to “dynamic storyteller”, you might have a problem.

What about fraud?

The same sorts of issues apply to cases of synthetic fraud — the creation of fictitious identities based on a combination of real and fabricated information. The criminals who do this are part of the reason why it can be so difficult to open a bank account or sign up for a credit card. The more real a synthetic person appears, the more successful money fraudsters are at stealing from a bank. The people running these scams are so sophisticated that they’ll set up fake social media accounts, spoof income statements, and have their friends pretend to be their employers. Plus, looking for one fake account is like trying to find a very annoying needle in a haystack.

Thankfully the heavy burden placed on fraud investigators can also be eased through data. It turns out that some things, like the fact that 89 people all with the same income and employer are all applying for the same card, are very hard to fake if you know where to look to get a whole a picture. There are some harder to miss examples too, such as a credit history that only stems from particularly phone and utility companies who have lax standards. Using these kinds of insights, investigators can get access to relevant information to help separate real customers from synthetic ones.

So what’s next?

These sorts of challenges are much more pervasive than people may realize, and make it so much harder for honest consumers to access financial services. So it’s increasingly important for lenders to look in all the right places.

The data is out there and a regulatory framework that protects individuals’ privacy is in place. It’s up to institutions to make the most of these opportunities for their customers’ sakes.

Start exploring the types of data you can use to boost lending or reduce fraud.

By Tynan Daly, Data Scientist



Share on facebook
Share on twitter
Share on linkedin

More to explore

SME Business Prospecting

Prospecting SMEs within a data-driven model Demyst has helped several data-driven firms crack the code on how to find and rank the biggest and newest businesses within a particular geography, to identify which companies might

Read More »

The Demyst Approach to “Agile Data”

Analogous to the contrast between waterfall and agile methodologies in software development, Agile Data is focused on achieving a minimum viable releasable improvement. Ignore the scrum masters for a moment; what allows organizations to follow

Read More »

Just Click Here

In today’s marketing landscape, the function of lead generation has expanded beyond the traditional mandate of procuring contact details and now requires that companies have an in-depth understanding of their customers’ firmographic data. Just Click

Read More »
Close Menu