What would your reaction be if you wanted to get a loan and your bank asks to go through your Facebook profile? In China, this is already happening on a large scale, but it’s not banks that are doing the rating—it’s the country’s burgeoning fintech companies. And it’s not Facebook they are looking at—its social platform WeChat and shopping website Taobao.
Social credit scoring analyses data from non-traditional sources: social media, online shopping, payment apps, cell phone accounts, and more. This type of scoring is meant to fill a gap for people who want a loan but don’t have any way of proving they can repay one. In order to gauge whether you are creditworthy or not, the score can take into account a number of variables: who your friends are, what you buy, whether pay your bills on time or even how much time you spend reading the user agreement. It’s like FICO but decidedly more creepy.
“The data sources and the scoring algorithm are of course the two key components in scoring—they go hand in hand and both need to be strong for accurate credit assessment,” said Sahil Chugani, a recent Cheung Kong Business School graduate and former employee at Goldman Sachs’ Fintech and Asset-Backed Securities. Chugani is currently exploring how China’s online lenders, including Alibaba’s Ant Financial, are using social credit scoring.
“For some apps that are engrained in your day-to-day, like [Alibaba’s payment app] Alipay, datasets are readily available there: what you pay for, who you receive money from, spending habits like your average purchase size and frequency; these are some of the obvious ones,” he said. “Other companies partner with third-party databases that also have user-consented data verified by their ID cards. Other datasets are publicly available.”
The idea of social credit scoring is not new—in fact, Facebook once had a similar plan. The company secured a patent for assessing Facebook users’ ability to repay a loan based on their social network. At some point, Facebook gave up on the idea, but other companies like Baidu and JD-backed ZestFinance based in the US have been popping up to offer financial services based on social network big data.
Social credit scoring is one of the factors that has helped China become the world leader in fintech adoption. According to consulting firm Mckinsey, the local fintech explosion was ignited by a unique landscape, including highly developed e-commerce and online payments, regulatory support, and, as always, the possibility of making a lot of money. But what really pushed fintech into a boom was the fact that traditional banks were simply not lending money to individuals because they lacked a reliable way to assess credit scores—only 25% of the population have a credit history. This often made borrowers turn to shadow banking or unregulated borrowing.
The untapped pool of borrowers has prompted tech companies such as Alibaba, Tencent, and many more to develop alternative ways to assess creditworthiness as well as new forms of financing such as peer-to-peer lending.
“These financial big data firms such as [Alibaba’s] Sesame Credit or those that score and engage in lending have a huge responsibility in harmonizing customer data to more accurately assess who is not biting off more than they can chew. The traditional Western credit assessment has failed at this on many occasions, see ‘07 for more!” said Chugani.
Alibaba was once a kind of shadow lender too. The company first started building its own credit scoring model to provide loans to Taobao vendors. For this, it relied solely on the platform’s ability to gather big data—transactions, user ratings, market positioning, and others.
Today, Alibaba offers several financial services under its financial arm Ant Financial. It has also built up China’s biggest social credit database, the controversial Sesame Credit. But according to Alibaba, the system is not used for credit assessment at all, although similar data is used for offering financial services.
Sesame Score (screenshot above) tracks five areas: identity information, such as information on users’ education and work, ability to keep financial obligations, credit history, behavioral preferences like shopping, money transfers, and connections with other people. In return, it offers deposit-free bike and power bank rentals as well as other benefits.
“For Ant Financial’s credits and loans we have a different set of algorithms. One factor might play a bigger role in assessing whether to give a loan than deciding how high is the Sesame score,” a spokesperson for Ant Financial told TechNode.
The actual data Ant Financial uses for credit assessments is much more complicated and not too different from banks, according to Ant Financial. As Chugani explained, this is a common trait for most companies using social credit scoring.
“The main thing to note here is that the weighting in the credit scoring algorithms allocated to the purely social data ( e.g. your LinkedIn network, or your Alipay friends) is not very high,” he said. “Your average ticket size, geolocation data, are much more powerful indicators of creditworthiness.”
Ant Financial claims that Sesame Score is used for something less Orwellian than critics had mooted: to offer deposit waivers and fast-tracks to certain services in exchange for gathering more data. According to the company, the system is purely commercial and doesn’t belong to China’s government-backed Social Credit System, even though until recently, the project was used as one of the test pilots meant to assist building it. However, this does not mean that social credit scoring systems such as these do not deserve close attention.
Computer says “no”
Although it was created as a tool for giving more access to credit for those who need it, be it for education, starting a business or buying the newest iPhone, social credit scoring bring its own set of issues connected to big data, as illustrated by the European Commission report on the topic.
The obvious issue is security. Even if we rule out hackers, there are other concerns to consider when it comes to our valuable data. Many fintech companies do not own data centers, they rent out cloud services from other companies we know nothing about or in countries that have weaker privacy regulation.
Tests have also proven that data and algorithms are not neutral, they reflect our own biases. These tools may perpetuate and intensify existing biases by scoring consumers on the basis of race, gender, religion, politics and other factors. And even when the algorithms aren’t biased there is a possibility we can make them biased in our favor: people online are already sharing tips on tweaking FICO and Sesame scores.
Lastly, social credit scoring is far from transparent. From a user’s perspective, it is difficult to gauge whether your score went lower because you bragged how you got “wasted” last Friday online or because you liked “Fully automated luxury gay space communism” on Facebook. In Chinese online space where the Chinese government is even more involved, this question gets even more difficult.
Fintech in China has indeed made life easier for millions of people and is truly realizing the ideal called “inclusive finance.” Chugani believes that increased functionality and quality of life outweighs privacy in China: “In particular, with better machine learning, the marginal benefit of giving up where you are, or what you bought on Taobao is ever-increasing.”
But as Pamela Kyle Crossley, history professor at Dartmouth College recently noted, “while in the United States we associate government data collection with passive surveillance and regard the voluntary surrender of huge amounts of personal information to commercial entities as some other kind of thing, in China there generally is no illusion that such a distinction exists.”