ASIAN ECONOMIC OBSERVATORY NETWORK A Proposal on Data-Driven Modeling of Asian Economies

Transcription

1 ASIAN ECONOMIC OBSERVATORY NETWORK A Proposal on Data-Driven Modeling of Asian Economies Siew Ann CHEONG Division of Physics and Applied Physics School of Mathematical Sciences Nanyang Technological University

2 Nanyang Technological University Nanyang Technological Institute Nanyang University Nanyang Technological University 1991 School of Physical and Mathematical Sciences Nanyang Technological University Singapore University of Technology and Design National University of Singapore Singapore Management University Former Central Library Presently Chinese Heritage Centre 2

3 The Asian Property Bubble Giant bubble can implode anytime Interventions so far ineffective Meltdown cannot be controlled at country level 3

4 Learning From the Impending Crisis After so many market crashes, markets still crash Did we learn nothing? Inevitable? No coordinated research in Asia by economists or econophysicists Set up regional network of teams Asian Economic Observatory Network (AEON) Aim to understand and control crisis What knowledge and tools do we need? 4

5 Economics From theory-driven data-driven Big Data Not overhaul, but upgrade Still aggregate data Raw data available for hypothesis testing Policy-driven economics Policy problems hard, generally avoided Involve policy from the get go Measurements + milestones Development of tools + evidence-based intervention 5

6 The Need for a Network of Observers Each observatory sees a difficult cultural facet of Asian economy Difficult & unethical to perform experiments Each observatory sees a different facet of night sky Impossible to perform experiments 6

7 The AEON Proposal CHINA JAPAN Zhou Wei-Xing Hideaki Aoyama ECUST Kyoto University HONG KONG Chuo University Michael Wong (?) National Defense Academy of Japan Yuji Aruka Akira Namatame HKUST KOREA Sharing of data & methods Collaboration & personnel exchange Kim Seunghwan POSTECH Lee Duk Hee KAIST Plug-in to global network of research groups TAIWAN Chen Shu-Heng INDIA Bikas K. Chakrabarti Saha Institute of Nuclear Physics SINGAPORE Cheong Siew Ann National Chengchi University Li Sai-Ping Academia Sinica Nanyang Technological University 7

8 The AEON Program United States Japan Prof H. Eugene Stanley (Boston University) AEON Prof Didier Sornette (ETH Zurich, Switzerland) FCO China Korea FOC Prof Guido Caldarelli (IMT Lucca, Italy) CRISIS India Taiwa n Singapore Financial Labour Europe Model PI: Li Baowen (NUS) Co-PI: Huang Weihong Data PI: Low Buen Sin Consumer Financia l Microeconomics Prof J. Doyne Farmer (Oxford University, UK) Analytics PI: Cheong Siew Ann Calibratio n Validatio n Forecasting Experiments PI: Yohanes Eko Riyanto Co-PI: Walter Edgar Theseira Individual Choice Social Interaction s 8

9 The AEON Approach further hypothesis testing Hypothesis+ Hypothesis Hypothesis Hypothesis+ Hypothesis Model Data Re-analysis mining/machine learning Model Hypothesis Hypothesis+ Hypothesis Hypothesis+ Data (Macroeconomic, Microeconomic, Financial) Hypothesis Data Generate predictions Model (Macroeconomic, Microeconomic, Financial) Hypothesis Hypothesis+ Model Hypothesis+ Hypothesis Hypothesis+ Hypothesis Model Hypothesis Hypothesis+ empirical laws agent-based models further predictions 9

10 Challenges Data Different availability conditions, reporting standards, data formats Create standard data format? Model No track records for very-large-scale ABM Endorsements from European collaborators Government involvements Different political priorities Use each other as leverage 10

11 Sharing Data & Methods: Example 1 Scheffer et al., Nature 461, 53 (2009) vegetation fishery Scheffer et al., Science 338, 344 (2012) neurology climate socio-economic systems? 11

12 The Subprime Crisis ( ) Triggered by subprime loan defaults Credit default swaps (CDS) sold in tranches by big banks Bought by many funds in US & Europe What is a prime loan? Given to borrowers with good credit ratings Good income & middle class Consistent history in paying back car & personal loans What is a subprime loan? Given to borrowers with poor credit ratings Low income, poor job stability No history of loans or history of delinquency & defaults 2.5% or 4.0% teaser rate for first 23 years ~10% adjustable rate subsequently Will Scheffer s universal early warning indicators work? 12

13 US Housing Market Data US Census Bureau metropolitan statistical areas (MSAs) Jan 1995 to Nov 2012 monthly housing permit Zillow.com about 3000 US cities Sep 1997 to Dec 2012 home sold, median sale price, % of homes sold for profit 13

14 Red Shift in Power Spectrum normal Subprime Crisis Technology Bubble concentration of power in lower frequencies = slowing down of time scale Lehman Brothers Asian Financial Crisis Subprime Loans 14

15 Spatio-Temporal Dynamics 15

16 Transition Cross Sections Tan & Cheong, Eur. Phys. J. B 87, 38 (2014) Not able to find data and collaborators for comparative study! 16

17 Sharing Data & Methods: Example 2 bubble? 17

18 Singapore Housing Market Data Urban Redevelopment Authority (URA) Real Estate Information System (REALIS) Individual transactions Jan 1995 to Dec 2012 Data fields Property Name Transacted Price ($) Tenure Postal Sector Address Unit Price ($/m2) Completion Date Postal Code No. of Units Unit Price ($/ft2) Type of Sale Planning Region Area (m2) Contract Date Purchaser Addresser Planning Area Indicator Type of Area Property Type Postal District 18

19 Log CDF of Home Prices exponential power-law another power-law 19

20 The Singapore Housing Bubble Log CDF of Condominium Prices for Year Log Probability 2 2 housing bubble! what is this? 4 4 why this spatial distribution? same spatial distribution! Property market liberalized in Prices 4000 investment districts!

21 The Taiwan Housing Bubble also effect Sai-Ping Li of transport Academia hubs Sinica, Taiwan Singapore Taiwan Look for scaling collapse Effect of new transport hubs Connection with wealth distribution 21

22 The Way Forward A team in each country Japan has a large team already Singapore building up More efforts elsewhere Collaborations Scale interesting local projects to much more interesting regional projects Try each other s methods on data Comparative approach to generate insights Endorsements International collaborators Governments 22