At Funding Circle, we take a data-driven approach to assessing the risks associated with lending. We have numerous methods of building statistical models that guide our underwriting by assigning a probability of default to each loan we review. Yet, we also understand the limitations of models in evaluating loans. This is where our team of experienced underwriters contribute their knowledge and bring the human touch that we consider so crucial to making sound credit decisions.
With the wealth of data surrounding us today, it’s tempting to think that fully-automated underwriting is an optimal approach. It’s not. Rigorous underwriting relies on a carefully crafted combination of algorithms and human input. Here’s why:
When we assess a loan application, we are evaluating not only a borrower’s ability to make payments but also their willingness to do so. One of the “5 Cs of Credit,” Character is perhaps the most important and, unfortunately, the most difficult to get right. Models and algorithms, sophisticated as they may be, simply can’t judge character as well as human experience can. The numbers may make all the sense in the world, but if the intent is off, so too will be our credit decision.
Messy financial statements tend to be a pretty strong indication that a business owner doesn’t really have a pulse on what is going on in their business, and if a borrower can’t clearly depict the performance of their own business, then it is all the more difficult for an underwriter to do so. While our models will evaluate the quantitative information contained in financial statements – value of assets, outstanding debt amounts, cash flow and margins – it is our underwriters who will put this into context and determine “Does this really make sense?”
Algorithmic underwriting does a great job of analyzing historical trends. What it fails to do is give credit to a business’ future plans for growth and expansion and what that business needs in order to realize those plans. As a lender, we need to understand these ambitions; they are often the driving force behind a deal and, importantly, the source of our repayment.
Human credit knowledge provides significant value to the underwriting process. This is not to say, however, that models aren’t vital – they are very important. Models highlight the most significant factors to a loan, ensure consistency in underwriting and make the whole process more efficient.
Funding Circle’s credit models are a significant part of our competitive advantage, and so is our team of seasoned underwriters who know how to combine our computers’ complex algorithms with the material details that emerge when digging deeper into the human side of the equation. The future of underwriting will not be taken over by algorithms and robots, but will be predicated on combining a large and rich list of data points and information with the right balance of human and algorithmic interaction.
By: Vittoria Reimers, Director of Credit