Your Tech Habits May Someday Get You Approved (or Declined) for a Mortgage

When you go online, you leave a trail of digital bread crumbs. And if new research catches on, mortgage lenders could someday use those crumbs against you (or for you).

The U.S. National Bureau of Economic Research (NBER) has published a study showing that creditworthiness is linked to a borrower’s “digital footprint.”

NBER studied nine digital footprint variables:

1. the borrower’s device type (for example, desktop, tablet or mobile)
2. the operating system (for example, iOS or Android)
3. the channel through which a customer comes to a lender’s website (for example, search engine, rate comparison website, etc.)
4. whether the customer enabled “Do not track” (to prevent tracking of device, OS, etc.)
5. the time of day of the loan application (for example, morning, afternoon, evening or night)
6. the email service provider (for example, Gmail, Yahoo or a paid or company email)
7. the email address format chosen by the user (e.g., did they use their first and/or last name, a number, a period…)
8. whether the customer consistently used lowercase letters when writing
9. whether the customer made typos.

Researchers then compared the power of these variables to predict loan default to a standard credit score.

“Our results suggest that even the simple, easily accessible variables from the digital footprint proxy for income, character and reputation and are highly valuable for default prediction,” the NBER concluded.

The results are eye-opening. Among their findings:

• “…The difference in default rates between customers using iOS (Apple) and Android (for example, Samsung) is equivalent to the difference in default rates between a median credit score and the 80th percentile of the credit score.”
• “…Customers coming from a price comparison website are almost half as likely to default as customers being directed to the website by search engine ads…”
  • The study authors said this finding is consistent with marketing research documenting the importance of personality traits associated with impulse shopping.
  • On a side note: It’s no secret to mortgage rate comparison websites that their clientele are generally more qualified than the average Canadian borrower. From our anecdotal discussions with lenders, it appears the industry is slowly starting to realize this. Eventually, lenders will market more aggressively to rate site borrowers as a result.
• “Customers arriving on the homepage through paid ads (either clicking on paid Google ads or being retargeted after prior Google searches) exhibit the largest default rate.”
• “Customers having their names in [their] email address are 30% less likely to default.”

In essence, credit scores alone were only 68.3% effective in predicting default. Using a borrower’s digital footprint was more effective (69.6%).

That’s remarkable given how much time, money and effort has gone into perfecting credit scores since the very first “FICO” score in 1989.

Digital footprints today can even “forecast future changes in a borrower’s credit score,” NBER adds. And it’s also possible they could better predict fraud.

Despite these findings, mortgage lenders won’t be abandoning credit scores. For one thing, NBER doesn’t specifically test these findings in the mortgage market. (Although it does link them to other loan types.)

Study co-author, Tobias Berg, tells the Spy: “Mortgage lenders can use the digital footprint as well…It seems reasonable to assume that digital footprints work similar for mortgages, but we cannot provide direct evidence with our data at hand.”

“One of the key difference is that mortgages are typically larger…so most banks also [use] human beings (and not only an algorithm) to decide about granting a loan or not.”

Moreover, NBER says, “digital footprint complements rather than substitutes for credit bureau information.” That’s because digital footprint variables and credit score have as low as a 10% correlation.

For that reason, “…using both the credit bureau score and the digital footprint variables significantly exceeds the [predictive] power of models that only use the credit bureau score or only use the digital footprint variables.” We’re talking 73.6% accuracy in predicting default versus 68.3% for credit scores alone.

Those five percentage points may not sound like much, but on a big bank mortgage portfolio that’s measured in tens of billions of dollars, this tech could potentially help lenders avoid millions in losses every year.

And there’s one more consumer benefit. “…Digital footprints can facilitate access to credit when credit bureau scores do not exist,” NBER writes. For borrowers new to Canada and/or those with “thin” credit, this technology may someday help them get approved for a mortgage—whereas they otherwise wouldn’t have been.