14 | FORBES ASIA AUGUST 2017
may notice that a cleaner “form” indi-
cates more likely repayment.) Addition-
ally, if the applicant agrees, fintech com-
panies gain access to activity data, such
as social media (WeChat moments) or
Tmall shopping records. The more appli-
cants, and more inputs, the surer the pat-
terns will be.
While China’s internet giants—Alib-
aba, Baidu and Tencent—all have their
own lending arms, these platforms are
usually exclusive to their existing user
base. Tencent’s Wei Li Dai is by invitation
only. Users need to show good account
history at Tencent’s WeBank to be invited
to apply for loans. Alibaba’s Jie Bei (“Just
Borrow”)—a counterpart to Hua Bei—
lends to approved Alibaba customers;
Baidu has You Qian Hua (“Has money
to spend”), a product heavily focused on
providing college loans.
In comparison, independent fintech
companies target a wider population by
developing the best credit scoring algo-
rithms for matching lenders with borrow-
ers (the unscorable 75%).
FORBES ASIA
RAPID-FIRE FINANCE
China Rapid Finance (CRF), China’s
largest consumer-lending marketplace by
number of loans facilitated, calls its target
users EMMAs (emerging middle-class,
mobile-active consumers), a group of 500
million underserved borrowers. These are
“typically young people aged 23-to-29-
years old, urban, employed, well-educated
and heavy smartphone users,” CRF tells
Forbes Asia. Founded in 2001 by Zane
Wang, who holds a Ph.D. in statistics, the
Shanghai company initially was only an
institutional risk manager but expanded
into the online-finance business in 2010.
The company employs a “low and
grow” strategy. It started with small loans
to prime and near-prime EMMAs, helping
them build up their credit history. With
repayment histories at the CRF platform,
many of these now qualify for larger and
longer-term loans. On July 10, the compa-
ny reached 20 million cumulative loans.
The total for the first half of 2017 was near
the number in all years leading to 2017.
Publicly traded since April, it expects to
exceed $100 million in revenues this year.Targeting roughly the same group of
users, three-year-old Yongqianbao of Bei-
jing distinguishes itself by a high loan ap-
proval rate—20% to 30% , it says. The in-
dustry average is in the single digits. Jiao
Ke, founder and CEO of Yongqianbao,
says the high approval rate is a sign of
an accurate AI risk-management model,
ICE (identification, calculation and eval-
uation). “Many people have the miscon-
ception that more data points will neces-
sarily lead to more accurate data models,
but that’s not exactly right,” Jiao says from
his Beijing headquarters. “For a machine
to learn, you also need to feed it with
enough loan outcomes and feed it fre-
quently.” Yongqianbao has fed 10 million
loan outcomes to ICE so far, and every
month 2 million more samples are added.
“While our loan approval rate is high,
our default rate is low,” says the Tsinghua
computer science graduate. Yongqianbao
wouldn’t disclose the exact default rate.
Back in October, Jiao told 36Kr, a Chinese
media outlet dedicated to entrepreneurs,
that Yongqianbao’s default rate was 60% of
the industry average. The company raised
$67 million this March from investors, in-
cluding Sinovation Ventures, Lee Kai Fu’s
venture capital firm.
Ning Tang, CEO of CreditEase, was a
pioneer in the Chinese fintech sector. He
founded the Beijing company in 2006; its
online consumer finance arm, Yirendai,
had the first IPO of a such a fintech com-
pany, in late 2015. Last year, it had reve-
nues of $488 million.
“When we first founded our compa-
ny, we didn’t even know what to name it
because there was no such word as inclu-
sive finance or [what’s also called] P2P
lending,” says Tang. “Over the past ten
years, we as an industry have made break-
throughs in individual credit rating, but
rating small and medium-size enterprises
is still an untapped field.”
To tackle that more complex challenge,
Tang says, fintechs must share their data.
Two years ago, CreditEase established Zhi
Cheng A Fu, a data-sharing platform, and
offered free access to its data points col-
lected over ten years. “Many say that we
may lose because of this,” he says. “But I
think if we don’t do this, the entire indus-
try will lose.” FFROM PORN FILTER
TO CREDIT RATER
Shing Tao, CEO and chairman of a small Las Vegas company, Remark Holdings,
sounds a bit cocky when he speaks of plunging into the huge Chinese fintech
space: “Our data model is trained by the amount of data they [the fintech com-
panies there] can never acquire. They no longer have to train their own data
models. Just buy ours.”
Remark has managed to collect data from almost every social media site on
Earth: 1.3 billion active user profiles, 10 billion images, 15 billion posts and
50 billion comments are gathered from Tencent, Alibaba, Facebook, Twitter and
others. Remark’s intelligence platform, KanKan, was assembled to analyze data
and build facial-recognition algorithms to help live-streaming companies filter
out pornography.
Now the New York-born Tao, 40, has decided that credit rating in China has
a more stimulating future. The NYU Stern graduate argues that a comprehensive
data set can generate algorithms for any domain. “Data is the new oil,” he says.
“Banks and P2P lending platforms provide us their historical loan outcomes.
We feed our models with those and sell our algorithms back to them,” adds
Chinese-born-and-bred Jason Wei, Remark’s chief technology officer. The com-
pany, with $59 million in revenues in 2016, claims to have signed up one of Chi-
na’s largest banks for KanKan.
“Many say that social data are not as relevant as financial data in determining
a person’s credit score,” Wei says. “Well, the connection may be weak, but human
beings are not to judge how weak the connection is, machines will.” —R.F.