Transcript Determinants of TOEFL Score: A Comparison of Linguistic

Do Contagion Effects Exist in Capital Flow
Volatility?
May 2013
Hyun-Hoon LEE
(Kangwon National University, Korea)
Cyn-Young PARK
(Asian Development Bank)
Hyung-suk Byun
(Kangwon National University, Korea)
1
1. Introduction

The volatility of international capital flows and the limited ability
of developing economies to deal with it continue to be a major
international policy concern.

The spillover or contagion effects refer to the cross-border
transmission of financial shocks, through co-movements of asset
prices and capital flows

The contagion effects first caught the attention of the global
financial community, as the Mexican devaluation in December
1994 brought an abrupt end to capital flows to many Latin
American economies and triggered speculative attacks on their
currencies—dubbed as the “tequila effect” afterwards.

The next crisis that hit many Asian economies in 1997 spread beyond the
regional boundary.

The spillover or contagion effect culminated in the global financial crisis of
2008/09, which rattled financial markets worldwide and pushed the world
economy into the worst recession since World War II.
2
1. Introduction

Many studies have investigated the contagion effect by analyzing co-movements of
international asset market returns and volatilities.

Very few examined the effect of contagion on capital flow volatility and only
recently, some began to focus explicitly on volatility of capital flows.

IMF’s 2007 Global Financial Stability Report find that financial market openness
and institutional quality are negatively associated with the volatility of capital flows
in both emerging and developed economies.

Neumann, Penl, Tanku (2009) examine how different types of capital flows
respond to the opening of financial market. Specifically, they show that a further
opening of financial markets tends to increase the volatility of FDI in emerging
economies, while it does not lead to any meaningful change in the volatility of
portfolio investment flows.

Broto et al. (2011) suggest global conditions have differential impacts on FDI,
portfolio investment, and other investment flows and find that global factors have
become increasingly significant relative to country specific factors since 2000 .

Mercado and Park (2011) investigate the impact of a set of domestic and global
factors on the volatility of different types of capital flows to developing economies,
and find that a regional factor plays an important role in determining the volatility3
of capital inflows to emerging Europe and emerging Latin America.
1. Introduction

There is rich literature on the volatility spillovers between international
asset markets for evidence of contagion.

Only recently, some empirical studies have begun to try to identify the
sources of capital flow volatility and examine its determinants.

None of these studies, however, explicitly deal with the spillover
phenomenon of capital flow volatilities.
 This paper aims to fill the gap in the literature by adopting a new
measure of capital flow volatility and assessing the spillover or contagion
effect in the volatility of capital flows to developing countries.
4
1. Introduction

Contribution of this paper
1.
This paper adopts the most commonly used measure of
volatility – standard deviation of capital flows in a moving
window, but with a more rigorous standardization
procedure.
2.
This paper examines how volatility of capital flows to
individual emerging economy responds to the volatility of
capital flows (a) to all developing countries and (b) to the
neighboring countries in the same region.
3.
Comparison of gross capital inflows and net capital inflows
in three different types of capital flows: FDI, equities, and
other capital flows (mostly bank lending)
5
1. Introduction

Major findings of this paper
1.
This paper presents evidence for strong and significant contagion effects
from global and regional volatilities on the volatility of capital flows in
different types to individual economies.
2.
The volatility of FDI flows is the lowest and it is least susceptible to intraregional contagion, relative to those of portfolio investment or other
investment (mostly bank lending).
3.
The volatilities of portfolio investment and other investment behave in a
roughly similar manner in terms of their magnitudes and exposure to
cross-border contagion.
4.
It is also found that gross capital inflows and net capital inflows are
similar in terms of the size of their volatilities, the volatility of net capital
flows responds to intra-regional contagion than that of gross inflows.
6
1. Introduction

Contents of this paper
1. Introduction
2. Descriptive Analysis
3. Empirical Specification
4. Results
5. Summary and Conclusions
7
2. Descriptive Analysis

Data

Time Period: 1990-2009

Countries: 45 emerging and developing countries.

Types of capital flows: FDI, portfolio investment, and other
investment flow (IMF, International Financial Statistics).

Gross inflows vs. net inflows
 The gross capital inflow represents non-residents’
purchases minus non-residents’ sale of domestic assets.
 The net capital inflow is defined as gross inflows minus
gross outflows.
8
2. Descriptive Analysis

Measurement for volatility of capital flows

The measure of capital flow volatility for country i in year t, σit,
is given by the following expression:
1
1 t
2
2
 it  
( flowik   ) 

n
 k t ( n1)

Where

1 t
 flowik
n k t ( n1)
flowik : capital inflows relative to country i’s GDP to
country i and in period k.

While the standard deviation is widely used to estimate
volatility, this measure has some important drawbacks.
9
2. Descriptive Analysis

Measurement for volatility of capital flows

Problem 1: different means provides different dimensions to the
standard deviation.
 This paper normalizes the size of capital flows in each rolling
window by setting the largest capital flows in absolute terms at 100
and adjust the rest of capital flows series accordingly.

Problem 2: the choice of the length of a rolling window is arbitrary.
 Following other studies, this study also uses the five-year moving
window, but the three-year moving window will be provided as a
robustness check.

Problem 3: standard deviation of a rolling window is strongly
persistent, and when it is used as the dependent variable in a
regression analysis, the error term is serially correlated by
construction.
 This study utilizes the system GMM estimator for dynamic panel
10
data models of Blundell and Bond (1998).

Volatilities of Different Types of Capital inflows to All Developing Countries
11

Volatilities of Different Types of Capital inflows to East Asian Countries
12

Volatilities of Different Types of Capital inflows to East Asian Countries
13

Volatilities of Different Types of Capital inflows to East Asian Countries
14

Volatilities of Different Types of Capital inflows to All Developing Countries
15
2. Descriptive Analysis

Volatilities of Different Types of Capital inflows to All Developing Countries
3 Year average_World
90
80
70
60
50
40
30
20
10
0
VIFDI
VIFPI
VIOI
VNFED_LT
VNFED_ST
Note: FDI, FPI, and IOI refer to the standard deviation of foreign direct investment, portfolio investment, and
other flows (mostly bank lending) in % of GDP, respectively. NFED_LT and NFED_ST refer to the standard
deviation of net flows on long-term external debt and on short-term external debt in % of GDP, respectively.
16
3. Empirical Specification

The Model
VMit = β0 + EVit β1+PVit-1 β2+ CVit-1 β3+ εit.
Where

VM is a volatility measure of the five different types of capital
inflows,

EV is a vector of external variables,

PV is a vector of policy variables, and

CV is a vector of control variables.
17
3. Empirical Specification

Dependent variable: VM

VFDI : 5-year standard deviation of foreign direct investment
in % of GDP

VFPI: 5-year standard deviation of portfolio investment in % of
GDP

VOI: 5-year standard deviation of other investment in % of
GDP
 Note: measured with gross inflows or net inflows.
18
3. Empirical Specification

Explanatory variables
External variables

ALL: Average of the world-wide volatility of capital flows into
all emerging countries

East_Asia: Average volatility of capital flows into other
neighboring countries in East Asia

Latin_America: Average volatility of capital flows into other
neighboring countries in Latin America

East_Europe: Average volatility of capital flows into other
neighboring countries in Eastern Europe
19
3. Empirical Specification

Explanatory variables
Policy variables




KOPEN: Chin and Ito’s financial openness index
INSTITUTION: Institutional quality
INF: CPI-based inflation rate
RESIMP: Stock of foreign reserves in months of imports
Control variables


GDP_PC: log of GDP per capita (constant in 2000US$)
PGDP: Change in GDP_PC (i.e. growth rate):
20
4. Results
Benchmark result

Table 1: Determinants of Volatility of Capital Inflows to Emerging

Table 2: Regional Contagion Effects on Volatility of Gross Capital
Inflows

Table 3: Regional Contagion Effects on Volatility of Net Capital
Inflows

Table 4: Determinants of Volatility of Capital Inflows to Emerging
Countries: With the world average level of lows included

Table 5: Summary of Global and Regional Spillover Effects of Capital
Flows
21
Table 1: Determinants of Volatility of Capital Inflows to Emerging Countries
Sample period: 1980-2009
Gross Inflow
Net Inflow
Lag of Dep Variable
Volatility of World Average
KOPEN
VFDI
VFPI
VOI
VFDI
VFPI
VOI
(1)
(2)
(3)
(4)
(5)
(6)
0.917***
0.834***
0.835***
0.754***
0.860***
0.863***
(0.011)
(0.011)
(0.007)
(0.018)
(0.012)
(0.014)
0.831***
0.742***
0.483***
0.747***
0.490***
0.601***
(0.058)
(0.043)
(0.038)
(0.046)
(0.019)
(0.043)
0.205
-0.854***
1.296***
-0.914***
0.544
0.856***
(0.172)
(0.203)
(0.283)
(0.231)
(0.339)
(0.326)
-1.735***
-10.916***
-6.262***
2.606**
-10.120***
2.437
(0.539)
(1.745)
(1.351)
(1.255)
(1.065)
(1.482)
-0.004***
-0.001**
0.003***
-0.002***
-0.000
0.006***
(0.000)
(0.001)
(0.001)
(0.000)
(0.001)
(0.001)
0.280***
-0.285**
0.063
0.255**
-0.550***
-0.212
(0.102)
(0.114)
(0.082)
(0.121)
(0.112)
(0.242)
1.129***
4.798***
0.165
2.959***
2.220***
3.217***
(0.319)
(0.411)
(0.664)
(0.828)
(0.450)
(1.161)
-0.938***
-0.351***
-0.995***
-1.118***
0.275***
0.317***
(0.036)
(0.067)
(0.045)
(0.083)
(0.062)
(0.075)
-29.267***
-60.166***
-13.098***
-37.517***
-32.544***
-45.991***
(2.018)
(3.939)
(4.360)
(6.769)
(3.643)
(9.809)
Number of observations
955
718
986
970
804
975
Number of countries
49
45
49
49
48
49
AR 1 (p-value)
0.000
0.000
0.000
0.000
0.000
0.000
AR 2 (p-value)
0.519
0.052
0.122
0.951
0.171
0.900
Sargen test (p-value)
0.571
0.807
0.651
0.695
0.746
0.732
INSTITUTION
INF
RESIMP
GDP_PC
PGDP
Constant
Arellano-Bond test
Notes: Estimates are made with Blundell-Bond (1998)'s two-step system GMM. Shown in parentheses are standard errors.
***, **, and * denote one, five, and ten percent level of significance, respectively.
Table 2: Regional Contagion Effects on Volatility of Gross Capital Inflows
Sample period: 1980-2009
VFDI
(1)
Lag of Dep Variable
East_Asia_Intra
Latin_America_Intra
East_Europe_Intra
VFPI
(2)
0.998***
0.986***
(0.022)
(0.024)
GDP_PC
PGDP
(0.024)
(0.012)
0.845***
(0.015)
(0.012)
(0.025)
(0.045)
-0.113*
0.143***
0.235***
(0.062)
(0.032)
(0.029)
0.042
0.157***
0.301***
(0.034)
(6)
0.822***
(0.041)
East_Europe_Extra
RESIMP
0.848***
0.147***
Latin_America_Extra
INF
0.851***
(5)
0.220***
(0.026)
INSTITUTION
VOI
(4)
0.092**
East_Asia_Extra
KOPEN
(3)
(0.038)
0.032
0.124***
-0.052
(0.043)
(0.036)
(0.054)
-0.240***
0.093***
0.059
(0.038)
(0.025)
(0.051)
-0.017
-0.045
0.103***
(0.046)
(0.031)
(0.040)
-0.393
-0.070
-0.928***
-0.700***
1.685***
1.211***
(0.405)
(0.163)
(0.287)
(0.207)
(0.279)
(0.388)
-1.299***
-1.316
-10.682***
-11.338***
-2.864**
-5.836***
(0.454)
(1.099)
(1.114)
(2.294)
(1.354)
(1.147)
-0.002***
-0.001
0.000
-0.003***
0.002***
0.002*
(0.001)
(0.001)
(0.002)
(0.001)
(0.001)
(0.001)
0.060
0.020
-0.337
-0.076
0.306*
0.192*
(0.125)
(0.107)
(0.311)
(0.193)
(0.175)
(0.108)
2.278***
2.378***
3.359***
3.802***
-0.652
1.014
(0.778)
(0.413)
(0.746)
(0.903)
(1.206)
(1.099)
-0.813***
-0.827***
-0.570***
-0.562***
-0.982***
-1.097***
(0.066)
(0.069)
(0.069)
(0.163)
(0.114)
(0.074)
-15.290***
-14.473***
-20.272***
-21.245***
8.273
0.648
(4.702)
(2.884)
(6.079)
(6.597)
(8.690)
(9.015)
Number of observations
955
955
712
718
986
986
Number of countries
49
49
45
45
49
49
AR 1 (p-value)
0.000
0.000
0.000
0.000
0.000
0.000
AR 2 (p-value)
0.473
0.450
0.059
0.062
0.089
0.116
Sargen test (p-value)
0.769
0.818
0.783
0.827
0.795
0.736
Constant
Arellano-Bond test
Notes: Estimates are made with Blundell-Bond (1998)'s two-step system GMM. Shown in parentheses are standard
errors. ***, **, and * denote one, five, and ten percent level of significance, respectively.
Table 3: Regional Contagion Effects on Volatility of Gross Capital Inflows
Sample period: 1980-2009
VFDI
(1)
Lag of Dep Variable
East_Asia_Intra
VFPI
(2)
(3)
VOI
(4)
(5)
(6)
0.790***
0.797***
0.840***
0.864***
0.766***
0.889***
(0.026)
(0.017)
(0.019)
(0.014)
(0.037)
(0.028)
0.343***
0.327***
(0.041)
(0.043)
(0.049)
Latin_America_Intra
0.357***
0.295***
0.496***
(0.031)
(0.034)
(0.067)
East_Europe_Intra
0.237***
0.450***
0.818***
(0.054)
0.643***
(0.075)
(0.101)
East_Asia_Extra
0.148***
-0.013
0.063
(0.046)
(0.030)
(0.039)
Latin_America_Extra
0.171***
0.106*
0.103*
(0.027)
(0.056)
(0.059)
0.040
0.165***
-0.073
(0.057)
(0.063)
(0.087)
East_Europe_Extra
KOPEN
INSTITUTION
-1.442***
-1.018***
1.008
1.443***
2.688***
0.526
(0.418)
(0.236)
(0.904)
(0.424)
(0.445)
(0.703)
0.585
2.103**
-15.376***
-9.585***
-4.528
2.137
(1.353)
(1.051)
(2.568)
(1.176)
(3.641)
(2.781)
-0.001**
-0.002***
-0.003***
-0.002
0.002*
0.005***
(0.001)
(0.000)
(0.001)
(0.002)
(0.001)
(0.001)
0.298
0.360*
-0.281
-0.298***
-0.889
0.195
(0.270)
(0.208)
(0.272)
(0.102)
(0.751)
(0.551)
4.089***
4.329***
1.931
0.770
-3.727
3.991
(0.642)
(0.896)
(1.574)
(0.734)
(3.449)
(2.599)
-0.673***
-0.860***
-0.105
0.150*
0.658***
0.137
(0.093)
(0.068)
(0.167)
(0.080)
(0.235)
(0.135)
-30.097***
-27.860***
-17.309
-0.586
22.014
-27.499
(4.672)
(6.542)
(12.494)
(6.085)
(23.648)
(17.807)
Number of observations
970
970
804
804
975
975
Number of countries
49
49
48
48
49
49
AR 1 (p-value)
0.000
0.000
0.000
0.000
0.000
0.000
AR 2 (p-value)
0.851
0.884
0.162
0.162
0.867
0.618
Sargen test (p-value)
0.774
0.720
0.702
0.853
0.973
0.701
INF
RESIMP
GDP_PC
PGDP
Constant
Arellano-Bond test
Notes: Estimates are made with Blundell-Bond (1998)'s two-step system GMM. Shown in parentheses are standard
errors. ***, **, and * denote one, five, and ten percent level of significance, respectively.
Table 4: Determinants of Volatility of Capital Inflows to Emerging Countries:
With the world average level of flows included
Sample period: 1980-2009
Gross Inflow
Net Inflow
VFDI
VFPI
VOI
VFDI
VFPI
VOI
(1)
(2)
(3)
(1)
(2)
(3)
Lag of Dep Variable
0.974***
0.862***
0.833***
0.747***
0.864***
0.858***
(0.013)
(0.015)
(0.008)
(0.020)
(0.012)
(0.013)
Volatility of World Average
1.087***
0.860***
0.454***
0.760***
0.701***
0.554***
(0.065)
(0.047)
(0.036)
(0.047)
(0.061)
(0.046)
Level of World Average
1.449***
2.971***
-0.080
-0.203
-1.012***
0.673***
KOPEN
INSTITUTION
INF
RESIMP
GDP_PC
PGDP
Constant
(0.135)
(0.460)
(0.065)
(0.192)
(0.317)
(0.233)
0.017
-1.130***
1.189***
-0.851***
1.301**
0.490
(0.155)
(0.169)
(0.249)
(0.264)
(0.518)
(0.341)
-1.369**
-11.758***
-4.916***
2.455*
-11.934***
5.773***
(0.695)
(2.073)
(1.497)
(1.448)
(1.266)
(1.850)
-0.005***
-0.001
0.002**
-0.002***
0.002
0.006***
(0.001)
(0.001)
(0.001)
(0.000)
(0.001)
(0.001)
-0.045
-0.676**
0.015
0.297
-0.366***
-0.410**
(0.155)
(0.269)
(0.123)
(0.182)
(0.140)
(0.209)
0.242
4.573***
0.456
3.273***
3.198***
1.968*
(0.441)
(0.668)
(0.757)
(0.841)
(0.608)
(1.096)
-1.107***
-0.528***
-0.822***
-1.110***
0.555***
0.275***
(0.062)
(0.108)
(0.103)
(0.084)
(0.118)
(0.075)
-33.939***
-65.569***
-14.089***
-39.680***
-49.910***
-35.007***
(3.312)
(6.581)
(5.433)
(6.983)
(6.832)
(9.591)
Number of observations
955
718
986
970
804
975
Number of countries
49
45
49
49
48
49
AR 1 (p-value)
0.000
0.000
0.000
0.000
0.000
0.000
AR 2 (p-value)
0.520
0.058
0.123
0.954
0.177
0.925
Sargen test (p-value)
0.648
0.802
0.684
0.716
0.807
0.830
Arellano-Bond test
Notes: Estimates are made with Blundell-Bond (1998)'s two-step system GMM. Shown in parentheses are standard errors.
***, **, and * denote one, five, and ten percent level of significance, respectively.
Table 5: Summary of Global and Regional Spillover Effects of Capital Flows
Net Inflow
Gross Inflow
Sample
period
Average of World Volatilities
East_Asia_Intra
1980-2009
Latin_America_Intra
East_Europe_Intra
Average of World Volatilities
East_Asia_Intra
1980-2007
Latin_America_Intra
East_Europe_Intra
VOI
VFDI
VFPI
VOI
VFDI
VFPI
(1)
(2)
(3)
(4)
(5)
(6)
0.831***
0.742***
0.483***
0.747***
0.490***
0.601***
(0.058)
(0.043)
(0.038)
(0.046)
(0.019)
(0.043)
0.092**
0.220***
0.147***
0.343***
0.327***
0.643***
(0.041)
(0.025)
(0.045)
(0.041)
(0.043)
(0.049)
-0.113*
0.143***
0.235***
0.357***
0.295***
0.496***
(0.062)
(0.032)
(0.029)
(0.031)
(0.034)
(0.067)
0.042
0.157***
0.301***
0.237***
0.450***
0.818***
(0.026)
(0.034)
(0.038)
(0.054)
(0.075)
(0.101)
0.863***
0.420***
0.429***
0.890***
0.397***
0.467***
(0.043)
(0.023)
(0.036)
(0.049)
(0.018)
(0.029)
0.510***
0.058*
0.147***
0.055
0.321***
0.392***
(0.031)
(0.020)
(0.059)
(0.024)
(0.066)
(0.025)
-0.091**
-0.067
0.178***
0.378***
0.172***
0.560***
(0.037)
(0.050)
(0.037)
(0.014)
(0.043)
(0.042)
-0.002
0.135***
0.191***
0.169***
0.668***
0.582***
(0.037)
(0.038)
(0.032)
(0.059)
(0.114)
(0.073)
Notes: Estimates are made with Blundell-Bond (1998)'s two-step system GMM. Shown in parentheses are standard errors. ***, **, and *
denote one, five, and ten percent level of significance, respectively. The results for the period 1980-2009 are taken from Tables 1, 2, and 3,
while those for the period 1980-2007 are the corresponding results when the global crisis years of 2008 and 2009 are excluded from the
sample.
5. Summary and Conclusions
27
Hyun-Hoon Lee
[email protected]