Transcript presentation

The determinants of cross border bank flows
to emerging markets:
new empirical evidence on the spread of financial crises
Sabine Herrmann (Deutsche Bundesbank)
Dubravko Mihaljek (BIS)
16th Dubrovnik Economic Conference, Croatia, June 23-25, 2010
The views expressed in this presentation are those of the authors and not necessarily those of the Deutsche
Bundesbank or the BIS.
Motivation
 At first, emerging markets mostly unaffected by the
current financial crisis, soft landing widely expected.
 However, after the collapse of Lehman Brothers,
financial stress in emerging markets intensified:
•
•
•
•
Deleveraging of global financial institutions
Flight to quality
Deteriorating macroeconomic outlook
Crisis of confidence in some countries
 IMF claims international bank lending to be a major
factor contributing to spillover effects (WEO, 2009).
2
Main issues addressed in the paper
Determinants of cross-border bank flows from advanced to emerging
market economies:
- over a longer period
- in different crisis periods
- in different emerging market regions
How the financial stress spread from advanced economies to EMEs via
cross-border bank lending
- identify specific channels of transmission
- compare them in size, across EM regions, crisis periods
3
Modelling approach
Gravity model for asset flows (Martin and Rey, 2004)
Focus on bilateral cross-border bank flows from advanced to EMEs
Go beyond traditional push and pull factors (Papaioannou, 2008)
Look at impact of macroeconomic and financial stress indicators on crossborder bank flows (Rijckeghem and Weder, 2003; Heid et al, 2004;
McGuire and Tarashev, 2008; Buch et al, 2009)
4
Data
BIS locational banking statistics
External positions (and external loans) of banks in 17 BIS reporting
countries vis-à-vis 28 EMEs
Quarterly data, 1993-2008
Amounts outstanding (in USD) and exchange-rate adjusted flows
Bilateral, country-pair observations (30,500)
5
Mar.99
-11,174
-50,000
Sep.09
Sep.08
-40,227
Mar.09
-40,000
Sep.09
-442
17,850
-18,637
-5,799
1,997
10,464
Sep.09
Sep.08
-156,510
Mar.09
Mar.08
Sep.07
-36,834
-22,210
-32,387
-12,548
-26,781
-20,935
-10,279
-7,060
-22,339
-16,349
-58,562
-100,000
Mar.09
-40,000
-20,603
-32,321
Mar.08
16,339
17,856
18,053
29,989
Mar.07
Sep.06
Mar.06
Sep.05
Mar.05
Sep.04
Mar.04
Sep.03
Mar.03
Sep.02
Mar.02
Sep.01
Mar.01
Sep.00
Mar.00
Sep.99
-200,000
Sep.08
40,000
Sep.07
3,970
Mar.99
Sep.98
Mar.98
Sep.97
Mar.97
Sep.96
Mar.96
Sep.95
Mar.95
78,459
66,595
11,543
22,467
17,458
2,535
10,445
-1,311
18,551
12,750
21,336
Emerging Asia
Mar.08
Central and eastern Europe
38,499
50,462
47,595
44,859
Mar.07
Sep.06
Mar.06
Sep.05
Mar.05
Sep.04
Mar.04
Sep.03
Mar.03
Sep.02
Mar.02
Sep.01
Mar.01
Sep.00
Mar.00
-7,024
-9,607
Latin America
Sep.07
Mar.07
Sep.06
Mar.06
Sep.05
Mar.05
Sep.04
Mar.04
Sep.03
Mar.03
Sep.02
Mar.02
Sep.01
Mar.01
Sep.00
Mar.99
Sep.99
40,000
Mar.00
60,000
Sep.99
-13,484
-13,279
-30,000
Sep.98
-20,000
Sep.98
Mar.98
6,979
7,082
14,359
3,055
20,000
Sep.97
3,365
2,945
Sep.94
Mar.94
Sep.93
Mar.93
-50,000
Mar.98
5,567
3,872
4,972
5,798
-180
-930
4,113
1,126
3,141
2,469
Mar.97
Sep.96
-2,502
30,000
Sep.97
Mar.97
Sep.96
Mar.96
Sep.95
Mar.95
-10,000
Mar.96
20,000
1,068
725
Sep.94
Mar.94
10,000
Sep.95
-20,000
-3,983
Mar.93
Sep.93
50,000
Mar.95
Sep.94
Mar.94
Sep.93
Mar.93
100,000
0
-150,000
0
-60,000
Figure 1:
External
positions of
reporting
banks vis-àvis emerging
markets
Exchange rate
adjusted
changes (Q/Q),
USD millions
0
6
BASIC GRAVITY MODEL
LOANSijt =
ρ0 + ρ1 DISTij + ρ2 GDPit + ρ3 GDPjt +
ρ4 INTERESTijt + ρ5 GROWTHijt + ρ6 EXCHANGEijt+ ρ7 Xijt + εijt
LOANS
Δ external position of banks in country i vis-a-vis country j (log)
DIST
distance between the capitals i and j (log)
GDP_i/GDP_j
nominal GDP in i and j (log)
INTEREST
short-term interest rate differential between j and i
GROWTH
GDP growth differential between j and i
EXCHANGE
bilateral exchange rate
7
Financial stress models
1. GLOBAL MODEL
De-leveraging driven by global factors, e.g. a stronger risk aversion
in the wake of the financial crisis (wake-up call)
 VIX
 RISKAVERSION
Chicago Options Exchange S&P Volatility Index
Yield difference US corporate bonds/treasuries
2. LENDER MODEL
De-leveraging driven by developments in lender country, e.g. its
exposure to the primary crisis country (common lender effect)
 CLE
Common lender effect: exposure of country i
towards the primary crisis country
 BANK_HEALTH_i Bank health indicator in lender country i
8
3. RISK MODEL
De-leveraging driven by a worsening of borrower country-specific
risk factors (early warning indicators)
 GOVBALANCE
General government balance (in % of GDP)
 BANK_HEALTH_j Bank health indicator in borrower country
4. LINKAGE MODEL
De-leveraging driven by degree of financial/monetary integration
between borrower and lender country
 FINANCE_OPEN Bilateral financial openness indicator
 ER_REGIME
Exchange rate regime (Reinhart/Rogoff, 2004)
9
DISTANCE
GDP_BORROWER
GDP_LENDER
INTEREST
GROWTH
EXCHANGE
VIX
RISK_AVERSION
COMMON
LENDER_US
COMMON
LENDER_AS
COMMON
LENDER_MX
BANK_HEALTH_
LENDER
GOVBALANCE
BANK_HEALTH_
BORROWER
FINANCE_OPEN
ER_REGIME
(1)
BASIC
Model
-0.594
(-8.51)***
1.038
(10.67)***
-0.715
(-5.14)***
0.011
(4.50)***
0.044
(7.84)***
-0.015
(-6.76)***
(2)
GLOBAL
Model
-0.660
(-3.20)***
1.198
(12.24)***
-0.972
(-6.40)***
0.005
(1.93)**
0.030
(5.03)***
-0.011
(-4.99)***
-0.027
(-5.80)***
-0.002
(-4.02)***
(3)
LENDER
Model
-0.693
(-8.77)***
1.098
(8.77)***
-0.733
(-3.55)***
0.012
(4.30)***
0.046
(7.00)***
-0.016
(-6.27)***
(4)
RISK
Model
-0.690
(-4.64)***
0.789
(6.75)***
-0.656
(-3.95)***
0.016
(3.82)***
0.040
(6.10)***
-0.028
(-8.31)***
(5)
LINKAGES
Model
-0.315
(-1.93)***
1.14
(9.26)***
-0.667
(-2.96)***
0.015
(5.19)***
0.049
(7.12)***
-0.011
(-4.49)***
-0.023
(-2.20)**
-0.010
(-0.95)
-0.286
(-3.88)***
0.001
(2.52)**
0.080
(6.59)***
0.006
(11.01)***
0.165
(10.50)***
-0.380
(-9.66)***
10
Contribution analysis
Contribution of each model to changes in cross border bank flows in three different crisis periods
0.05
0
-0.05
US Crisis
Asian Crisis
-0.1
Mexican Crisis
-0.15
-0.2
-0.25
BASIC
GLOBAL
LENDER
RISK
LINK
Note: Vertical axis is the percentage change in bilateral, quarterly, exchange-rate adjusted cross-border 11
bank flows, in millions of US dollars, explained by the respective model during each crisis period.
Contribution analysis
Contribution of financial stress factors to cross-border bank flows to different emerging market
regions during the current financial crisis
1
EMERGING EUROPE
EMERGING ASIA
0.8
LATIN AMERICA
0.6
0.4
0.2
0
-0.2
-0.4
-0.6
-0.8
riskaversion
vix
bankhealth_i
cle_us
cle_as
cle_mx
govbalance
bankhealth
financeopen
wk_regime
Note: Vertical axis is the log of bilateral, quarterly, exchange-rate adjusted cross-border bank flows, in
millions of US dollars, during the financial crisis of 2007-08.
12
Robustness checks
1. Econometric Options
- Time effects
- IV estimator according Anderson/Hsiao (1981)
- Woolridge approach (2002)
13
More on robustness checks …
2. Country specific risk factors: an extended analysis
- spread lending/deposit rate
- short-term foreign debt (in % of GDP)
- foreign reserves (in % of M2)
- current account balance (in % of GDP)
- real credit growth
3. Regional sample estimations
4. Crisis period estimations
14
Conclusions
 Results robust to various specifications and methodologies.
 Global as well as country specific factors (in both lender and
borrower countries) contribute to the transmission of financial stress.
 In the current crisis, global factors (stronger risk aversion, higher
expected market volatility) made the largest contribution to
reduction in cross-border flows.
 Sound banking sector, fixed exchange rate regimes, high degree of
financial integration helped stabilise bank lending to emerging
Europe during the latest crisis.
15