THE CONNECTEDNESS OF ASIAN

COUNTRIES: NETWORK ANALYSIS

15.4 1
DAVAAJARGAL Luvsannyam (Ph.D)
Director, Research Division, The Bank of Mongolia

15 0.75
KHUSLEN Batmunkh
Research and Statistics department, The Bank of Mongolia

14.6 0.5

TSOLMON Otgonbat
Research and Statistics department, The Bank of Mongolia
14.2 0.25

13.8 0

The views in this paper are those of the

authors and do not necessarily reflect the
views of the Bank of Mongolia.
 INTRODUCTION
-Importance
-Literature Review

METHODOLOGY
-Diebold-Yilmaz Index
-Network Connectedness
CONTENT
THE CONNECTEDNESS OF
ASIAN COUNTRIES
EMPIRICAL RESULTS
-Data
-Spillover Index

CONCLUDING REMARKS
 Importance Literature Review

Introduction Understanding how economies are connected and

Figure 1. Economic growth by continents, ✔ shocks are transmitted is a crucial and modern issue
for policymakers and researchers.
1960-2017
14
Asian countries tend to play fundamental role in
12
✔ Global economy.
10 • For instance, shocks from some Asian countries spread
Methodology complex effects to neighboring economies, and apparently
8 to other economies across the globe during the financial
crisis of 1997-98.
6

4 Recent activities in economies also prove that

emerging market economies in Asia can become a
2 ✔ major source of shocks that transmitted widely, as well
Empirical 0 as their increasing economic importance.
results 1960's 1970's 1980's 1990's 2000's 2010's
East Asia & Pacific Euro area
European Union Latin America & Caribbean
Middle East & North Africa North America
South Asia Sub-Saharan Africa
Therefore, studying the economic connectedness
✔ of Asian countries is an important and an early
action for policy decisions.
Concluding
remarks
 Importance Literature Review

Introduction Diebold “Measuring financial asset return and

& Yilmaz volatility spillovers, with application to
(2009) global equity markets”
Spillover Plot, Global Stock Market Returns
• They provide a simple and intuitive measure of and Volatility, 11/1995–11/2007
interdependence of asset returns and/or volatilities.

Methodology
Diebold
“Measuring the Dynamics of Global
& Yilmaz
Business Cycle Connectedness”
(2015)
• Using a connectedness-measurement technology
fundamentally grounded in modern network theory, they
measure real output connectedness for a set of six
Empirical developed countries,1962-2010.
results
Davaajargal “Diebold-Yilmaz index”, Eviews add-
(2018) ins

• The add-in estimates the Diebold-Yilmaz index of

spillover using forecast error variance decomposition
method of VAR model.
Concluding
remarks
 Importance Literature Review

Introduction Diebold “Measuring financial asset return and

& Yilmaz volatility spillovers, with application to
(2009) global equity markets”
Dynamic Total Connectedness, G-6 Industrial
• They provide a simple and intuitive measure of Production, 1958:01- 2011:12
interdependence of asset returns and/or volatilities.

Methodology
Diebold
“Measuring the Dynamics of Global
& Yilmaz
Business Cycle Connectedness”
(2015)
• Using a connectedness-measurement technology
fundamentally grounded in modern network theory, they
measure real output connectedness for a set of six
Empirical developed countries,1962-2010.
results
Davaajargal “Diebold-Yilmaz index”, Eviews add-
(2018) ins

• The add-in estimates the Diebold-Yilmaz index of

spillover using forecast error variance decomposition
method of VAR model.
Concluding
remarks
 Importance Literature Review

Introduction Diebold “Measuring financial asset return and

& Yilmaz volatility spillovers, with application to
(2009) global equity markets”
• They provide a simple and intuitive measure of
interdependence of asset returns and/or volatilities. DY INDEX
The add-in estimates the Diebold-Yilmaz
index of spillover.
Methodology
Diebold
“Measuring the Dynamics of Global Author: Davaajargal Luvsannyam
& Yilmaz
Business Cycle Connectedness”
(2015)
• Using a connectedness-measurement technology

Empirical
fundamentally grounded in modern network theory, they
measure real output connectedness for a set of six
developed countries,1962-2010.
27,281
results
Davaajargal “Diebold-Yilmaz index”, Eviews add-
(2018) ins

• The add-in estimates the Diebold-Yilmaz index of

spillover using forecast error variance decomposition
method of VAR model.
Concluding
remarks
 Connectedness table

Simple VAR equation 𝒚𝒕 = 𝑨𝒐 + 𝑨𝟏 ∗ 𝒚𝒕−𝟏 + ⋯ + 𝑨𝒑 ∗ 𝒚𝒕−𝒑 + 𝝐𝒕 (1)

Introduction
• Where: 𝒚𝒕 is a vector of length k. There are k equations; 𝒑 is the order of the VAR; {𝝐𝒕 } is a sequence of serially
uncorrelated random vectors with concurrent full rank covariance matrix Σ; 𝑨𝒐 is a (k × 1) vector of constants; 𝑨𝒊 ’s
are (k × k) coefficient matrices.

where the H-step forecast error variance of variable i accounted for by

Connectedness table
exogenous shocks to variable j is denoted by dH ij .
Methodology
𝒙𝟏 𝒙𝟐 ⋯ 𝒙𝑵 From Others
𝒙𝟏 𝑑 𝐻 11 𝑑 𝐻 12 ⋯ 𝑑 𝐻 1𝑁 σ𝑁 𝐻
𝑗=1 𝑑 1𝑗 , 𝑗 ≠ 1
𝒙𝟐 𝑑 𝐻 21 𝑑 𝐻 22 ⋯ 𝑑 𝐻 2𝑁 σ𝑁 𝐻
𝑗=1 𝑑 2𝑗 , 𝑗 ≠ 2
⋮ ⋮ ⋮ ⋱ ⋮ ⋮
Empirical 𝒙𝑵 𝑑 𝐻 𝑁1 𝑑 𝐻 𝑁2 ⋯ 𝑑 𝐻 𝑁𝑁 σ𝑁 𝐻
𝑗=1 𝑑 1𝑗 , 𝑗 ≠ 1
𝟏
results To Others σ𝑵 𝐻 σ𝑵 𝐻 ⋯ σ𝑵 𝐻
𝒋=𝟏 𝑑 𝒊𝑵 𝒊 ≠ 𝑵
σ𝑵 𝐻
𝒊,𝒋=𝟏 𝑑 𝒊𝒋 , 𝒊 ≠ 𝒋
𝒊=𝟏 𝑑 𝒊𝟏 𝒊 ≠ 𝟏 𝒊=𝟏 𝑑 𝒊𝟐 𝒊 ≠ 𝟐 𝑵

1 𝑁
𝐶= σ 𝑑𝑖𝑗 (2)
Spillover index 𝑁 𝑖,𝑗=1
𝑖≠𝑗
Concluding • There is just one total connectedness measure, as total connectedness distills a system into a single number
remarks
 Connectedness index Network analysis

Data First Difference of Logged quarterly real GDP of selected 11 Asian countries, 1960q1-2016q4
Introduction
Australia China Hong Kong India
6 6 6 6

4.25 4.25 4.25 4.25

2.5 2.5 2.5 2.5

0.75 0.75 0.75 0.75

Methodology 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1
-1 -1 -1 -1

Thailand Japan Turkey Indonesia

6 6 6 6

4.25 4.25 4.25 4.25

2.5 2.5 2.5 2.5

Empirical 0.75 0.75 0.75 0.75
results 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1
-1 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1
-1 -1 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1
-1
Korea Phillipine Singapore
6 6 7.75

4.25 4.25 6

4.25
2.5 2.5
Concluding 2.5
0.75 0.75
remarks 0.75
1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1
-1 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1
-1 1960Q1 1970Q1 1980Q1 1990Q1 2000Q1 2010Q1
-1
 Connectedness index Network analysis

Robustness The estimation is fairly robust.

Introduction Dynamic Total Connectedness, Selected
Asian countries, 1985-2017 Connectedness index fluctuates in the
Fluctuations
60 short-term.

58
56 Asian financial Long-term Long-term growth of connectedness index
crises of 1997 trend reflects the influence of globalization.
Methodology 54
52
50 Connectedness index might captures the
Cycles 3 possible cycles. (1985-1996, 1996-2002,
48 2002-2014)
46
Empirical 44
Global financial Connectedness index increased
results 42 crises of 2008 Crisis significantly during the both crises.
40 /Diebold &Yilmaz (2015)/
1985 1993 2000 2008 2015

Note: We used Generalized VAR(3) with 120-quarter rolling window The next financial crises is coming soon
and 10 quarter forecast horizons. Future because of previous patterns of crises and
Concluding growth of Asian connectedness index.
remarks
 Connectedness index Network analysis

Dynamic directional connectedness by income classification, 1985-2017 (To others)

High income Upper-Middle Income Low-Middle Income
Introduction
77

Before the Asian Between the Asian After the Global

financial crisis financial crisis and financial crisis
67 Global financial crisis

Percentage 57
Methodology
47

High(ave)=47.7 High(ave)=57.4
37 High(ave)=47.1
Upper(ave)=46.25 Upper(ave)=49.7
Upper(ave)=50.9
Low(ave)=44.8 Low(ave)=46.4
Empirical Low(ave)=44.32
27
results 1985Q4 1990Q2 1994Q4 1999Q2 2003Q4 2008Q2 2012Q4 2017Q2

• Significant growth of • Notable growth of

• Upper-Middle income countries contribution of upper- contribution of high
tended to contribute more to middle income income countries to
others. countries to others others was captured
• Before the crisis, there was huge was one factor for the during the Global
drop to contribution of low- Asian financial crisis. financial crisis.
Concluding middle income countries to • Contributions of three • High income
remarks others. categories to others countries tended to
were nearby to each. contribute more to
others.
 Connectedness index Network analysis

Dynamic directional connectedness by income classification, 1985-2017 (From others)

High income Upper-Middle Income Low-Middle Income

Introduction
1
62 Before the Asian Between the Asian After the Global 0.9
financial crisis financial crisis and financial crisis
Global financial crisis 0.8

52 0.7

Percentage 0.6
Methodology 0.5
42
0.4

0.3
32
High(ave)=50.3 High(ave)=57.4 0.2
High(ave)=49.1
Upper(ave)=40.5 Upper(ave)=49.7
Upper(ave)=44.2 0.1
Low(ave)=46.4 Low(ave)=46.4
Empirical 22
Low(ave)=47.9
0
results 1985Q4 1990Q2 1994Q4 1999Q2 2003Q4 2008Q2 2012Q4 2017Q2

• During the Asian • High and lower-middle

• High and lower-middle income financial crisis, effect income countries
countries tended to receive more from others of all 3 tended to receive more
contribution from others. categories increased contribution from
significantly. others.

Concluding
remarks
 Connectedness index Network analysis

Dynamic directional connectedness by income classification, 1985-2017 (Net to)

High income Upper-Middle Income Low-Middle Income
Introduction 1
Before the Asian Between the Asian After the Global
15 financial crisis financial crisis and financial crisis 0.9
Global financial crisis
0.8

0.7
5
Percentage 0.6
Methodology 1985Q4 1990Q2 1994Q4 1999Q2 2003Q4 2008Q2 2012Q4 2017Q2 0.5
-5
0.4

0.3
-15 High(ave)=-2.5 High(ave)=1.8
High(ave)=-1.9 0.2
Upper(ave)=5.7 Upper(ave)=3.5
Upper(ave)=6.8
Low(ave)=-1.5 Low(ave)=-6.5 0.1
Empirical Low(ave)=-3.5
results -25 0
• Recently, high income
• Upper-middle income countries tend to
countries tended to contribute more to others
• Upper-middle income countries
contribute more to while lower-middle
tended to contribute more to
others. income countries tend to
others.
• High and lower-middle income • High and lower-middle receive more contribution
income countries from others.
countries tended to receive more
Concluding contribution from others. tended to receive more • Upper-middle income
remarks contribution from countries tend to play role
others. of shock transmitter.
 Connectedness index Network analysis
Static Connectedness, Selected Asian countries GDP, 1960Q1-2017Q4
Hong Philippine From
Japan India Thailand China Korea Singapore Indonesia Australia Turkey
Kong s Others
Introduction Japan 78.41 0.12 1.41 3.47 3.01 1.76 1.45 1.64 0.30 0.76 7.67 21.59
India 4.57 78.85 0.55 2.65 7.13 2.30 0.96 0.83 0.27 1.23 0.67 21.15

Thailand 1.98 2.00 56.42 0.73 2.82 8.89 7.16 7.39 11.88 0.23 0.50 43.58
China 4.63 4.69 0.18 77.34 0.93 2.12 1.25 2.23 3.37 3.03 0.23 22.66
Korea 7.14 4.32 5.15 2.54 65.81 2.28 2.21 3.13 5.60 0.74 1.08 34.19

Hong Kong 8.56 2.99 10.54 3.22 1.50 50.85 16.01 0.75 2.97 1.68 0.94 49.15

Methodology Singapore 1.31 0.45 6.62 0.16 0.97 13.19 72.20 0.18 3.02 1.12 0.78 27.80

Indonesia 0.39 0.92 22.75 0.43 4.43 3.71 3.19 56.00 5.48 1.94 0.77 44.00

Philippines 1.28 0.42 15.14 14.54 4.41 5.17 5.35 1.71 49.17 2.54 0.27 50.83

Australia 3.20 2.35 2.95 3.91 3.86 3.66 0.51 1.84 4.10 73.18 0.43 26.82
Turkey 6.52 1.67 2.62 0.54 0.91 1.77 3.30 1.12 0.56 0.18 80.83 19.17
Empirical Contribution
results to others 39.58 19.92 67.92 32.17 29.97 44.86 41.39 20.81 37.55 13.44 13.34 32.81%
Net 17.98 -1.22 24.34 9.50 -4.22 -4.30 13.59 -23.19 -13.28 -13.38 -5.83

Concentration from others

Hong
Australia Korea China Thailand Turkey Japan India Phillippines Singapore Indonesia
Kong
Concluding Concentration from others
1233.4 1330.3 1478.7 1804.4 1873.8 1941.4 1967.4 1967.6 2038.4 2999.8 3082.2
(HHI)
remarks
 Connectedness index Network analysis

ForceAtlas2 algorithm of Jacomy et al. (2014)

Visualization
of the Gephi software.
Network connectedness of Asian economy,
Introduction from 1960 to 2017
Node, node size, node color, edge, edge
Attributes
thickness.

A country that is located at the center of

System network plays the more important role in a
system.

Methodology
Upside tends to contain high-income
countries, center side tends to contain upper
Clustering
middle-income countries, and downside
tends to contain low-middle income countries

China, Thailand, Hong Kong, and Philippine

Role
Empirical play an important role in the Asian economy.
results
Upper middle-income countries serve as the
Linkage linkage between high and lower-middle
income countries.

Thailand, Singapore, Japan, and China highly

Contribution
Concluding to own
tend to contribute their GDP growth to
themselves comparing to other countries.
remarks
 Connectedness index Network analysis

Network connectedness of Asian economy,

Introduction from 1960 to 2017
Asian map

Methodology

Empirical
results

Concluding
remarks Geographical Aspects Network visualization clearly shows the geographical aspects of the countries.
 Connectedness index Network analysis

Network connectedness of Asian economy,

Introduction Before Asian financial crisis: from 1960 to 1997

Singapore, Indonesia and Philippine

Methodology Role played important role in terms of
connectedness.

China, Korea, and Australia were less

Empirical Contribution
connected with other countries and their
results contribution of GDP growth to
to own
themselves was relatively low
comparing to other countries.

Concluding
remarks
 Connectedness index Network analysis

Network connectedness of Asian economy,

Introduction After Asian financial crisis: from 1998 to 2017

Korea, Thailand and Hong Kong played

Role important role in terms of
connectedness in the Asian economy
Methodology

China, Korea, Australia, India and

Contribution Singapore’s contribution of GDP growth
Empirical to own to themselves were relatively high
comparing to other countries
results

Concluding
remarks
 ➢ Diebold-Yilmaz index can be applied to both macro and micro datasets of Mongolia.
➢ For example, the connectedness of financial sectors (Commercial banks, Non-
Introduction
bank financial corporations, ets…..)
➢ The connectedness of economic activities in Mongolia (Mining sector, Industrial
sector, ets….)
➢ Diebold-Yilmaz index can be easily estimated with the Eviews add-ins created by
Methodology
Davaajargal Luvsannyam.

➢ Upper middle-income countries serve as the linkage between high and lower-middle
Empirical income countries.
results

➢ Connectedness index is one of the appropriate measure for the event of Financial
crisis.
Concluding
remarks