Posts tagged ‘Manju Puri’

Why You Need To Know Your Member’s Digital Footprint

Regardless of how big or small your credit union is, its  ability to capture and understand the digital footprint of its members  and use it to make lending decisions is going to be a key determinant of its future success or failure.

This point was crystalized for me Wednesday by Manju Puri of Duke University and the FDIC, who summarized research she conducted which demonstrates just how powerful and important digital information is becoming. Using ten easily accessible variables she and her fellow researchers were able to determine with accuracy comparable and at times exceeding the credit bureaus whether or not an individual is a good lending risk. The relatively limited dataset  included information seemingly  unrelated to lending such as the time of day a website was accessed; the email service provider  used; the email address chosen by the user and my personal favorite, whether a user consistently uses lowercase when writing email. Other more technical information included whether the consumer used a tablet, phone or computer to go online; the operating system they used and if their customer uses settings authorized device tracking.

Using this data, the researchers concluded that the simple variables are reliable proxies “for income, character and reputation and are highly valuable for default prediction.” Furthermore, “variables that proxy for character and reputation are also significantly related to future behavior.” This data set is important because it is information that all but the most technology resistant Luddite is going to generate.

For me, this research crystalizes many of the unique issues involving credit unions and the digitalization of marketing and lending. One of the biggest advantages that credit unions and true community banks have always had: They literally know their members and  have  used anecdotal highly personal information to make judgments about lending decisions which big lenders could never justify based on credit scores alone. Big data wipes out this advantage. It allows the biggest banks to account for character and other intangibles in a way they never could before.

This also  underscores the importance of data access. Imagine if the biggest companies and banks are allowed to maintain a monopoly of the most predictive data? This is a huge barrier to entry and competition. As an industry,  we have to broaden our focus so that we not only have a position on greater data security but on ensuring greater access to data for all institutions.

The final point that comes to mind is how far ahead the computer programmers are  , particularly when it comes to fair lending. Every day, data miners are refining increasingly complex algorithms that will predict precisely who will want and qualify for a loan and under what conditions. Simply put, the days of ecstatic limited set of lending criteria with well understood correlations to race are over. For example, I don’t believe the researchers who discovered a correlation between capitalization and lending are seeking out tricky ways to discriminate but the reality is no one knows if this and similar data will disproportionally exclude otherwise qualified individuals from getting loans on the basis of race. In addition, we certainly don’t know whether more benign criteria could be substituted in its place. For those of you who find this aspect of the debate as interesting as I do, here is another paper that was presented at the conference.

April 26, 2019 at 9:26 am 1 comment

Authored By:

Henry Meier, Esq., Senior Vice President, General Counsel, New York Credit Union Association.

The views Henry expresses are Henry’s alone and do not necessarily reflect the views of the Association. In addition, although Henry strives to give his readers useful and accurate information on a broad range of subjects, many of which involve legal disputes, his views are not a substitute for legal advise from retained counsel.

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 775 other followers