bigtech credit in china - piie · 2020. 11. 24. · china leads in some digital financial...
TRANSCRIPT
HUANG YipingPeking University
Financial Statement Series, Peterson Institute of International EconomicsNovember 24, 2020
Bigtech Credit in China
First Seller: CUI WeipingYokohoma, Japan
First buyer: JIAO ZhenzhongXi’an, China
First customer officer: LIANG BinHangzhou, China
The second-hand Fujifilm Camera
On October 18, 2003, Jiao paid RMB750 to Alipay to buy the second-hand camera from Cui, intermediated by Liang, on Alibaba’s online shopping platform Taobao.
How did Alipay start?Paypal was created in December 1998, and Alipay came online in December 2004. Why did Alibaba need Alipay?
Rapid development
100190
270
450520
900
355
500
697
889
989
1097
0
200
400
600
800
1000
1200
2013 2014 2015 2016 2017 2018
Alipay WeChat
Active users of Alipay and WeChat pay (million)
Convergence of digital financial inclusion
‡ Peking University Digital Financial Inclusion IndexRed, orange, yellow and green represent, from most to least developed, levels of digital financial inclusion development in relevant municipality.
The Hu HuanyongLine
China leads in some digital financial businesses, especially mobile payment, Big tech credit, online investment, digital insurance and central bank digital currency. And regional gaps narrowed rapidly between 2011 and 2018.
China is the largest Bigtech lender
Fintech lending volumes are diverging1 Big tech credit is booming in Asia, the US and Africa2 Index, Q1 2017 = 100 USD mn, logarithmic scale
CN = China, JP = Japan, KR = Korea, US = United States, KE = Kenya, ID = Indonesia. 1 Data are based on five platforms for Australia and New Zealand, all platforms covered by WDZJ.com for China, 49 platforms for Europe, 34 for the United Kingdom and five for the United States. Volumes are reported in local currency. 2 Figures include estimates.
Source: Cornelli et al (2020), “Fintech and big tech credit: a new database”, BIS WP 887.
Distinction between Fintech lending (P2Ps) and Big tech credit (MYbank, WEBank)
Bigtech credit in China
• Bigtech credit refers to loans extended to SMEs and individuals by Bigtech companies, taking advantage of Bigtech platforms and big data credit risk assessment
• Speed: Alibaba’s micro finance created the “3-1-0” lending model in 2010: it takes no more than 3 minutes to fill online loan application; if approved, the fund is in the account within 1 second; and there is 0 human intervention
• Contact-free: Bigtech credit continued during COVID-19
• Quantity: Each of the three virtual banks, WeBank, MYbank and XWBank, with 1000-2000 employees, can now grant around 10 million loans annually
• Quality: Non-performing loan ratiobefore COVID-19MYbank: 1.6%Banks’ SME loans: 3.2%SME loans (<5 m): 5.5%
MYbank’s NPL ratio (%) in 2020
1.57 1.66 1.72 1.742.01 2.05 2.02 1.98
1.73
0
1
2
3
Jan Feb Mar Apr May Jun Jul Aug Sep
D30 D90
A new credit risk management framework
Customer acquisition Risk identification Repayment management
Predicting ability and willingness to repay loans
Fintech credit risk assessment models
Machine learning models
Central bank credit registry information
Credit history
Digital footprint
Repayment records
Real-time monitoringLong-tail customers
Adverse selection problem Moral hazard problem
Punishing default
BigTech Platform
Big Data
Big data credit risk assessment
Horizontal axis: proportion of normal loans falsely predicted as default loans; vertical axis: proportion of correctly identified default loansSample size: 771594 (training set), 1053748 (test set)
Var12 Var3 Var123 Var1234 Var5 Var12345
ScorecardModel
0.70 0.58 0.72 0.72 0.69 0.76
Machinelearningmodel
0.67 0.54 0.76 0.80 0.76 0.84
Area Under The Curve (AUCs)
Variable groups: (1) Asset and financial information(2) Credit history (information of defaults)(3) Vendor-specific information(4) Local economy information (for provinces and cities)(5) Proprietary information on borrower behavior
Receiving Operating Characteristic (ROC) Curves
Traditional model (A): traditional variables + scorecard model→ Big data model (D): big data information + machine learning model(A) → (B): information advantage; (A) → (C): model advantage
Significance level: ** p<0.05; *** p<0.01.
Macroeconomic implications
• Bernanke, Gertler and Gilchrist (1999) once coined the term “financial accelerator”
• Bigtech credit relying on data instead of collaterals weakens financial accelerator mechanism
• But would this give rise to a new “cash flow accelerator”?
0.055
0.581***
0.209***
0.041** 0.018
0.147***
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Big tech credit Secured bank credit Unsecured bank credit
Elasticity with respect to House PriceElasticity with respect to GDP
Bernanke, Ben S. & Gertler, Mark & Gilchrist, Simon, 1999. "The financial accelerator in a quantitative business cycle framework," Handbook of Macroeconomics, in J. B. Taylor & M. Woodford (ed.), Handbook of Macroeconomics, edition 1, volume 1, chapter 21, pages 1341-1393, Elsevier.
Bigtech credit survived COVID-19?
1.57 1.66 1.72 1.74
2.01 2.05 2.02 1.981.73
0
0.5
1
1.5
2
2.5
3
Jan Feb Mar Apr May Jun Jul Aug Sep
D30 D90
Mybank’s proportion of overdue loans, 2020 (%)
Thanks 感谢