Can Networks combined with Analytics make better loans?


A recent study by National Consumer Law Center regarding Payday loans finds Big Data is not yielding better loan decisions. The author of the report, Persis Yu said, "The big-data algorithms do not appear to lead to the development of better loan products.”

This comes at a time when Silicon Valley is quite interested in funding initiatives around P2P lending.  They know that there is huge demand for these receivables in a zero based investing climate.  In fact, the American Banker recently did an article on small business loans emerging as the new frontier of P2P lending.

So do we conclude that all analytics are not helpful.  That is nonsense.  What this study focused on was generally the unbanked in society, or the 68 million Americans that have a poor credit history.  These new modeling companies can come up with all types of innovative approaches to analyze large data sets, including rent records, social media sites, data from prior payday loans, etc.  Will the idea that weighing thousands of data variables let them better predict creditworthiness?

More data does not necessarily make better decisions.  It reminds me of my friend, an avid horse handicapper, who can find all types of variables and data to choose from when picking a horse – trainer, how the horse runs in mud, who the horse was sired by, recent finishes, etc.   But perhaps it’s really one or two variables that really make the difference – such as is the horse on medication.

With Supplier Networks, more and more third parties are looking to analyze the data on trade flows within these networks.  What some who have stated investing in these networks tell me is that transactional data is more predictive than any other external trade credit information or any other credit related underwriting information.  What’s critical is to ensure the data is transactional down to a deep level.  For example,

  • Does it include dispute and dispute resolution?
  • Can you actually track an invoice to a payment?

That’s the challenge, but to come out and claim algorithms don't lead to better lending products, I would say don’t throw the baby out with the bathwater, supplier networks, in the right situation, present opportunities.

Discuss this:

Your email address will not be published. Required fields are marked *