Random Thought: Evaluating Credit Worthiness Using Offline Behaviour.

I am quite surprised by a colleague’s sharing recently that, in Southeast Asia, especially rural area, it is very common for friends and families to borrow money from each others. Evaluating credit worthiness is the heart of lending business. Traditional financial institute relies on credit bureau data, and documents submitted by the applicants, such as income proof, to evaluate the credit risk. In developed countries where majority of their people participates in the financial system and general default risk is low, the method works well. Still, there are people who do not have access to financial system, especially in less developed countries. Companies then come up with a new way to evaluate credit worthiness by using Non Credit Bureau Data. For example, e-commerce giants launch their financial products and use e-commerce data to evaluate credit risk. However, there are users who do not participate in financial system and also the online world. This poses a challenge to further serve the under-bank.

There is a common plot line in evaluating credit worthiness of people. Financial institute wants to serve as many people as possible given a certain trade off between risk and return. They started with wealthier individuals where data is easier to get (e.g. Credit Bureau) and slowly expand to subprime populations. Data availability determines the fundamental capability to evaluate credit worthiness, from using bureau data, self-own online data, to third-party online data. In the digital world, once you are connected, your footprint will be left somewhere in the Internet. But, a lot of the world populations are not active in digital world or they don’t trust it. Indeed, they tend to be poorer and have a higher default risk. But wealth is not the cause of low trustworthiness. We see a lot of rich people with low integrity. How can we serve trustworthy people and provide them with the finance they needed even they come from less developed countries?

Is there any way to record offline behaviour? The lending between friends and families; The dishonesty encountered in their daily live. I think in our mind, we have a rough sense of trustworthiness of our friends. Can we make use of that to underwrite more users?