Despite an uncertain economy, the number of small businesses has
soared. And that translates to big opportunity for lenders who
leverage unique data, according to Patrick Reily, co-founder of Uplinq Financial Services.
In our Market
Pulse podcast episode, “How Data Helps Lenders Navigate an
Uncertain Market,” Reily provides specific examples of how lenders can
use unique data to build powerful models that better predict risk
behavior. He said Uplinq alone connects to over ten thousand
authoritative data sources, adding up to about 12.5 million different
data series in the U.S.
Listen to the complete podcast interview or
continue reading for excerpts from our interview.
How can lenders find opportunity? Are there specific
examples that you can provide?
I would say that probably the big gap right now is most
traditional lenders tend to use what I would call a decision tree
type approach. It’s a whole collection of yeses and no’s. You're
passing through a gauntlet as a lone applicant. And as long as you
keep hitting “yes,” you can get all the way through the other end.
But if at any point along the way you hit a “no,” you're out.
What's the problem with a system like that? As a math guy I
can tell you, the problem is that you’ll say no to a lot of really
good people in order to get to the portion that is reliably good for
you. So, if you can use a more dynamic method where you can balance
the positives and the negatives to arrive at a conclusion, then you
can say yes, a lot more often.
If you think about it in the unregulated space, even some of
the riskiest lending areas of commercial lending have loss rates
that are rarely above about 20-percent. That means 80-percent of
those customers are still great credit customers for a bank. So,
what we can do is manufacture a process to reach those segments, and
we do that with multi-variant modeling. It balances the positives
and the negatives to arrive at a view of what we should do.
You refer to unique data. What do you define as unique data?
We look at sources that can be well-vetted,
well-authenticated in which we can have high confidence. So, what
are we talking about? Things like bank account data, transactional
and statement information, building information, government sources,
utility, reporting, and other sorts of similar information sources.
They can give us high confidence about what's going on.
Some quick examples of that include things like location
specific, aggregated mobile device activity. We don't have to know
who's going up and down the street. That's not important to us. But
to understand the volume of traffic that moves in front of a store
front can be incredibly informative about the likelihood of success
of that restaurant or that shopkeeper or that business. So now we're
not just dependent on government data that tells us where people
live and where they work. We can see the river they flow down when
they're going between all their points in life.
Another great example is tax return information. We can
aggregate peer groups that are reflective of the type of business
we're considering. We can adjust for the age of the business and the
size of the business, the marketplace the business exists in, and so
on, and so forth to arrive at a view of how well this company is
performing relative to its peer group.
Are there any industries that you think in terms of
opportunity that would be better to focus on the others? And does
Yes, and yes. But it's not only a good location and bad
location or a good industry or bad industry category. It's a lot
about fitment. And in geek speak, we talk about
interactions between data. So, when we're looking, we're not only
trying to determine if a community is growing or contracting.
We’re looking at whether it’s a resource rich environment for this
type of business to build within. Most importantly, how well does
Subscribe to Market Pulse Podcast
The Market Pulse podcast is available wherever you listen to
podcasts. This episode was a continuation of our conversation of the
February Market Pulse webinar. You can access
it on-demand and sign up for our next webinar.