Banking in the age of social media
I happened to be in Guangzhou last week, and attending an interesting round table discussion during a banking technology event.
One of the guests was from WeBank, sharing his experience working in a technology company that happens to hold a banking license in China.
He shared the fact that they started as a technology company even before they had the license, developing a banking platform that is now used by 8 local banks in China in a BaaS (Banking as a Service) model.
The most interesting part of the conversation originated from a question, asked by an attendee who works in the Credit Risk department of a big “traditional” bank. The surprise and awe that followed his answer is a clear proof of the different culture of the two companies.
The question followed a presentation, in which was mentioned that — as expected — WeBank uses data from its parent company Tencent’s service WeChat to make credit decisions. The question was asked by a person who does not use WeChat wallet — the company’s mobile payments system — and shops almost exclusively on websites (e.g. taobao) outside of Tencent’s “reach”. The question was how can they make an instant credit decision (56000 RMB of credit in this particular case — versus an average amount 10000 RMB), given that they don’t have any financial-related data about her.
The answer listed a few of the methods WeBank uses to ‘infer’ your financial health… some of them are related to one’s life financial sphere, some are not.
WeBank uses the following info to categorise users into “segments”, and they then make credit decisions based on those:
- WeChat wallet payment transactions: how many tx, average amount, what type of online and offline retailers, …
- What type of company service accounts one follows on WeChat
- What type of articles one reads on those service accounts
- Which pictures (what is the content of the pics) one like on Moments
- The city one lives in (1st, 2nd, 3rd, rural, …)
- The area of the city one works (e.g. if it’s CBD or industrial/factory area)
- The area of the city one spend leisure time e.g. evenings and weekends
- The neighbourhood one lives (based on location during night hours)
It was clarified that the company does not look at the chat content (claim that we can’t verify, although it is known that chats are not end to end encrypted because of the “censorship” function that is used from time to time to block messages with “sensitive“ content…).
One funny anecdote is their claim they can identify when a company is in financial troubles (e.g. laying off people) by the frequency of new chats that happens — during office hours — between group of people with heavy overlaps… they claim it’s symptom of colleagues, who may already be connected in “work” chat groups, creating various separate chats with subset of colleagues in a frenzy to understand if their job is safe or not.
We can believe it or not, but one thing we know for sure: that financial decisions (e.g. lending money) that used to be made only on the basis of few proxies (e.g. collaterals like cars, apartments, a steady job, …) are now made using a lot more data points that, although unrelated to finance at first sight, can help determining a person’s lifestyle and spending power.
I definitely want to read and learn more on the subject!