By Hendrik Brackmann on Tuesday 16 February 2016
Last week, former financial regulator Lord Turner received a lot of attention with his statements that the peer-to-peer lending industry is going to suffer from big losses - he pointed to online credit underwriting of SME loans as the cause of the losses, stating that you need to visit a business in person to judge whether or not they can and will make repayments. I believe quite the contrary is true, and that an online approach has many advantages over more traditional methods.
While I have discussed the role of human involvement in the credit underwriting process before, I'm going to explain in a little more detail why 'kicking the tyres' of a business before deciding on a loan not only doesn’t help your ability to make a decision, but actively harms it.
The first problem with personal interactions is that we form a relationship with the customer, for example, we like someone because they are funny or attractive. While this is perfectly fine in our private life, it does tend to cloud our judgment in a business environment. A study by the American Psychological Association found that attractiveness can have a similar impact on your salary as intelligence. The anonymous online process is better since it prevents us from getting attached to irrelevant information.
The second problem is that the mood of the underwriter on the day he is meeting a business owner can create biases. Has the underwriter not had their coffee that morning? Are they tired and irritable? Or did their football team win last night? Mood impacts our decision making, and that creates room for error.
The third problem is that any communication made in person is hard to prove. One of the advantages of channeling communications online (and on the phone) is that all communication is tracked. Credit decisions usually take weeks, if not months, until they reach a conclusion, much longer than most of us can remember the details of a conversation or site visit. To be able to go back and analyze credit decisions requires a record of the communications for each case. It's hard to improve your underwriting if face-to-face interactions play a role. On the other hand, machines can learn and improve much more quickly.
The fourth problem is that most of the information captured in face-to-face interviews is not digitally available and so can’t be fed into any automated decision engines. So how can you account for it without arbitrarily over-riding the data? This is particularly problematic since –as pointed out earlier- the additional information gathered is much more likely to cause biases.
Lastly, while face-to-face meetings provide some sense of depth that is not available online, there is a whole different, hidden dimension behind online data sources. For example, we know exactly which forms our applicants have visited, how long they spend there and how fast they typed. Additionally, it is common to share valuable data between companies. For example, if you ever listed a fake room for rent on the internet, then this might affect your ability to receive credit in the future. Try learning that in person.
Even in a case of all things being equal, the higher costs associated with site visits and interviews might be enough to decide for an online version. By making underwriting cheaper and more effective we can reduce the cost of the middle man and get a better deal for lenders and borrowers.