I’d like to thank George Popescu, publisher of the Lending Times, for allowing me to respond to his critique, contained in his July 13th digest, of a few points within my recent American Banker BankThink opinion piece titled “There Are No Universal Truths in Marketplace Lending.”
The first critique is as follows (quoted text is from the opinion piece; highlighted is in the original by Lending Times):
“For example, given the “success” of automating the origination and underwriting of consumer loans, other lending products are trying to automate their entire processes from A to Z, with varying levels of success. Small business loans seem to be benefitting from such automation, real estate not so much.” [Comment: I don’t understand why the author claims that real estate is not a success from automation. I entirely disagree. ]
My statement doesn’t say that “real estate is not a success from automation.” It merely states that real estate has not benefited as much as other asset classes. At some point this is semantics between George and myself, but it raises a much bigger issue which should be addressed. When somebody states that marketplace lending’s strength is in the processes, including automation, the question that needs to be answered is: What is being automated? In this I see two phases of automation, the collection and verification of data; and the automation of a credit decision. When people hear “automation + marketplace lending” they jump to the conclusion that credit decisions are being automated. With regards to real estate, nothing is further from the truth. You can jump on a consumer unsecured loan site and get approval within minutes. That doesn’t happen in real estate. In fact, if one were to dig into the real estate model of marketplace lenders (especially commercial real estate) you will find a lot of humans making a lot of judgment calls regarding the credit decision. In fact, even the origination of the loans for the majority of the real estate marketplace lenders employ highly seasoned real estate professionals to pass judgment on whether a loan will ever get to application, let alone underwritten, approved, and funded.
If real estate marketplace lenders truly benefited greatly from automation (to the same extent as other product types) they would not be struggling to ramp up production of quality assets. We are in the middle of a real estate lending boom, so now is the best time to let automation talk hold. The reality is, in my opinion, real estate marketplace lenders are mortgage banks that have employed technology to make the investment process simpler for the investor as well as the tracking of investments/loans after loan closings. This in and of itself is nice to have, but not as revolutionary as say what Lending Club and OnDeck have implemented. It doesn’t mean technology doesn’t have a place in real estate, it is just that all lending currently uses technology to some extent and I believe the art of real estate lending will not be usurped by the “science” of technology (maybe machine learning will make me change my mind). For purposes here, I exclude residential mortgage banking firms (nonbanks) that originate half the conforming and jumbo loans in the U.S. These firms are not considered marketplace lenders as currently defined. Their use of technology is now skyrocketing and mortgage banks have a good chance of making a real difference in terms of borrower experience and decisioning via technology.
The second critique is:
“What is happening now is that everybody is clustering around the same ideas — automated, instant, and convenient. These are lofty goals. The last decade’s subprime residential loans could arguably be described as having these same attributes of automated (i.e., everybody was approved), instant and convenient — and we are still paying the price.” [Comment: I believe that the author is mixing here blind fraud and automated systems which have many more verification points than any paper process was ever able to. Once again, I really disagree with the author here.]
Let’s set aside the idea of “blind fraud” (not sure what that means) and deal with just decision making. The subprime residential loan crisis was the result of bad underwriting policies and procedures. Just because borrowers may have fudged on their income doesn’t let the lender off the hook. The process, whether automated or manual, should require the lender to verify borrower submitted information. To the extent the residential subprime lenders didn’t verify information, the decisions did become automated—everyone was automatically approved. One of the critiques of the currently unsecured consumer marketplace lenders is that they don’t verify income of all the borrowers. To do so is understandably cost prohibitive and time consuming and would reduce the borrower satisfaction experience. Assuming we are not talking about identity verification, how is this any different than taking a subprime residential home loan borrower at his word regarding his income?
It is true that automation does look at many more data points when underwriting a loan. I would argue that outside of the basics of underwriting any loan—income (person or property), cash flow, etc.—most of the online automation is used to prevent identity theft type fraud. And this is as it should be. If you are going to remove the person from the front end of the process, then you have to replace it with a lot of data points to verify identity.
So, to come full circle, George started his critique of my piece with this statement:
A good pretext for a frank discussion about marketplace lending.
The whole point of an opinion piece is to abet discussion. Opinion pieces by their very nature are limited in scope (to do otherwise would be to create rambling musings of a pedantic windbag). I think it is time, as I say in my piece, that we should be critically evaluating each of the models as they apply to each of the asset classes. The responsibility for this falls firmly into the hands of loan investors.
Again, thank you, George, for this opportunity to respond to your critique. I challenge your readers to continue the conversation and help make the Lending Times ground zero for a long overdue evaluation of all aspects of the marketplace lending model…or shall I say, models?