Transcript L21: CRISES IN EMERGING MARKETS: WTP chapter sections: 24.1 INFLOWS TO EMERGING MARKETS -- Reserves (continued from Lecture 3) 24.2 MANAGING OUTFLOWS -- Early Warning.

L21: CRISES IN EMERGING MARKETS:
WTP chapter
sections:
24.1 INFLOWS TO EMERGING MARKETS -- Reserves
(continued from Lecture 3)
24.2 MANAGING OUTFLOWS -- Early Warning Indicators
24.3 SPECULATIVE ATTACKS
24.4 CONTAGION
(Lecture 20)
Breaching the central bank’s defenses.
24.5 IMF COUNTRY PROGRAMS
24.6 CONTRACTIONARY EFFECTS OF DEVALUATION
24.7 THE CAR CRASH ANALOGY
(continued
from Lect.12)
3 cycles of capital flows to emerging markets:
1. 1975-81 -- Recycling of petrodollars, via bank loans, to oil-importing LDCs
1982 -- Mexico unable to service its debt on schedule
=> Start of international debt crisis.
1982-89 -- The “lost decade” in Latin America
2. 1990-96 -- New record capital flows to emerging markets globally
1994, Dec. -- Mexican peso crisis
1997, July -- Thailand forced to devalue and seek IMF assistance =>
beginning of East Asia crisis (Indonesia, Malaysia, Korea...)
1998, Aug. -- Russia devalues & defaults on much of its debt.
=> Contagion to Brazil.
2001, Feb. --Turkey abandons exchange rate target
2002, Jan. -- Argentina abandons 10-yr “convertibility plan” & defaults
3. 2003-07 -- New capital flows into developing countries, incl. China, India...
2008-12-- Global Fin.Crisis: Iceland, Latvia, Ukraine, Pakistan, Greece, Ireland, Portugal…
Managing Capital Inflows
Recall alternative ways to manage capital inflows:
A. Allow money to flow in
B. Sterilized intervention
C. Allow currency to appreciate
D. Reimpose capital controls
In the boom phase of 1990-1996,
•
countries pursued exchange rate targets.
•
Some experimented with re-imposing controls on capital inflows
(Chile & Colombia), but mostly they allowed capital to flow in.
•
They used part of the inflow to add to foreign exchange reserves,
but – as in the earlier cycle 1975-1981 – they also used much of it
to finance trade deficits, some as large as the capital inflows.
(Calvo, Leiderman, & Reinhart, 1996).
Managing capital inflows, cont.
In the boom phase of 2003-2007:
•
Many countries had more flexible exchange rates than before.
•
Many reduced the share of capital inflow denominated in forex
except in Central & Eastern Europe
•
Few imposed new controls on the inflows [until Nov. 2009].
•
This time, a majority of emerging market countries
ran current account surpluses rather than deficits.
•
Thus inflows went into reserve accumulation (in fact, more than 100%).
•
As a result, reserves reached unprecedented levels.
2003-07: This time, many countries used the inflows
to build up forex reserves,
rather than to finance Current Account deficits
7.00
6.00
5.00
4.00
in % of GDP
(low- and
middle-income
countries)
Net Capital
Flow
Change in
Reserves
% of GDP
3.00
2.00
1.00
0.00
1991-97 boom
2003-07
boom
1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006
-1.00
-2.00
-3.00
-4.00
Current
Account
Balance
As a result, reserves reached extreme levels....
…, especially
in Asia.
IMF Survey Magazine Oct.8, 2009 “Did
Foreign Reserves Help Weather the Crisis?”
by O. Blanchard, H.Faruqee, & V.Klyuev
http://www.imf.org/external/pubs/ft/survey/so
/2009/num100809a.htm
Traditional denominator for reserves: imports
Rodrik (2006)
New denominator to gauge reserves: short-term debt.
After 2000, many brought their reserves
above the level of short-term debt –
the “Guidotti rule.”
s.t. debt / R
>1
<1
Rodrik (2006)
Alternative Ways of
Managing Capital
Outflows
A. Allow money to flow out
(but can cause recession,
or even banking failures)
B. Sterilized intervention
(but can be difficult, &
only prolongs the problem)
C. Allow currency
to depreciate
(but inflationary)
D. Reimpose capital
controls (but probably
not very effective)
Early Warning Indicators of Currency Crashes
Sachs, Tornell & Velasco (1996), “Financial Crises in Emerging Markets: The Lessons from 1995”:
Combination of weak fundamentals (Δ RER or credit/GDP) and low reserves
made countries vulnerable to tequila contagion.
Frankel & Rose (1996), "Currency Crashes in Emerging Markets" :
Composition of capital inflow matters (more than the total):
short-term bank debt raises the probability of crash; FDI & reserves lower the probability.
Kaminsky, Lizondo & Reinhart (1998), “Leading Indicators of Currency Crises” :
Best predictors: M2/Res, equity prices, GDP growth, Real Exchange Rate.
Berg, Borensztein, Milesi-Ferretti, & Pattillo (1999),
“Anticipating Balance of Payments Crises: The Role of Early Warning Systems”:
They don’t hold up as well out-of-sample.
Edwards (2002), “Does the Current Account Matter?”:
CA ratios of some use in predicting crises (excl. Africa), contrary to earlier research.
Rose & Spiegel (2009), “The Causes and Consequences of the 2008 Crisis: Early Warning”: No robust predictors.
Frankel & Saravelos (2012): Once again, reserves work to predict who got hit in 2008-09.
The variables that show up as the strongest predictors
of country crises in 83 pre-2008 studies are:
(i) reserves and (ii) currency overvaluation
0%
10%
20%
30%
40%
50%
60%
Reserves
Real Exchange Rate
GDP
Credit
Current Account
Money Supply
Budget Balance
Exports or Imports
Inflation
Equity Returns
Real Interest Rate
Debt Profile
Terms of Trade
Political/Legal
Contagion
Capital Account
% of studies where leading indicator was found to be
statistically signficant
(total studies = 83, covering 1950s-2009)
External Debt
Source: Frankel & Saravelos (2010)
70%
The IMF and Rose & Spiegel (2009) found that countries
with more reserves were not less affected by the 2008-09 crisis:
IMF Survey Magazine
Oct.8, 2009
“Did Foreign Reserves
Help Weather the Crisis?”
by O. Blanchard,
H.Faruqee, & V.Klyuev
http://www.imf.org/external/pubs/ft
/survey/so/2009/num100809a.htm
But Frankel & Saravelos (2010) and Dominguez & Ito (2011) find they were.
Best and Worst Performing Countries
in Global Financial Crisis -- F&S (2010),
Appendix 4
GDP Change, Q2 2008 to Q2 2009
Lithuania
Latvia
Ukraine
Estonia
Macao, China
Russian Federation
Bo tto m 10
Georgia
Mexico
Finland
Turkey
Australia
Poland
Argentina
Sri Lanka
Jordan
Indonesia
To p 10
Egypt, Arab Rep.
Morocco
64 countries in sample
India
China
-25%
-20%
-15%
-10%
-5%
0%
5%
10%
Bottom line for Early Warning
Indicators in the 2008-09 crisis
Frankel & Saravelos (2012)
• Once again, the best predictor of who got hit was
reserve holdings (especially relative to short-term debt),
• Next-best was the Real Exchange Rate.
• This time, current account & national saving too.
• The changes that most EMs (except E. Europe)
had made after the 1990s apparently paid off.
Are big currentaccount deficits dangerous?
Neoclassical theory: if a country has a low capital/labor ratio
or transitory negative shock, a large CAD can be optimal.
In practice: Developing countries with big CADs often get into trouble.
Traditional rule of thumb: “CAD > approx. 4% GDP” is a danger signal
“Lawson Fallacy” -- CAD not dangerous if government budget is
balanced, so borrowing goes to finance private sector, rather than BD.
Amendment after Mexico crisis of 1994 –
CAD not dangerous if BD=0 and S is high,
so the borrowing goes to finance private I, rather than BD or C.
Amendment after East Asia crisis of 1997 –
CAD not dangerous if BD=0, S is high, and I is well-allocated, so the
borrowing goes to finance high-return I, rather than BD or C or empty
beach-front condos (Thailand) & unneeded steel companies (Korea).
Amendment after GlobalFinancialCrisis of 2008-11 – CAD dangerous.
Appendices:
More on predictors of crashes
1. Definitions (CA reversal, sudden stop, speculative attack…)
2. Predicting the 1994 Mexican peso crisis
3. How well did the pre-1997 EWI equations do
at predicting the East Asia crisis?
4. The best Early Warning Indicator: Reserves
5. How well did the pre-2008 equations do at predicting
who got hit in the 2008-09 Global Financial Crisis?
Appendix 1: Definitions
• Current Account Reversal
 disappearance of a previously substantial CA deficit
• Sudden Stop  sharp disappearance of private capital inflows,
reflected (esp. at 1st) as fall in reserves & (soon) in disappearance of
a previously substantial CA deficit. Often associated with recession.
• Speculative attack  sudden fall in demand for domestic assets,
in anticipation of abandonment of peg.
Reflected in combination of  s - res &  i >> 0.
(Interest rate defense against speculative attack might be successful.)
• Currency crisis  Exchange Market Pressure  s - res >> 0.
• Currency crash  s >> 0, e.g., >25%.
References
• Currency account reversals
– Edwards (2004a, b) and
Milesi-Ferretti & Razin (1998, 2000).
• Sudden stops
– References: Dornbusch, Goldfajn & Valdes (1995);
Calvo (1998); Calvo, Izquierdo and Mejia (2003);
Arellano & Mendoza (2002), Calvo (2003), Calvo,
Izquierdo & Talvi (2003, 2006), Calvo & Reinhart
(2001), Calvo , Izquierdo & Loo-Kung ( 2006 ),
Guidotti, Sturzenegger & Villar (2004), Mendoza
(2002, 2006); Edwards (2004b); Calvo, Izquierdo &
Loo-Kung (2006).
Appendix 2
The early 1990s
Calvo, Leiderman
& Reinhart
“predict the
peso crisis”
In the 1990s,
capital inflows
financed
current account
deficits.
Calvo, Leiderman
& Reinhart:
Source of capital
flows was low i*
at least as much
as local reforms
=>
Could reverse as
easily as in 1982.
Dornbusch (1994) said the
Mexican peso was overvalued.
Appendix 3:
Predictive performance
of Early Warning indicators
in the1990s crises.
Berg, et al, (1999) did find
that if warning indicator
equation sounds an alarm,
probability of crisis is 70-89%;
but were generally pessimistic
on the ability at each round
to predict the next crisis.
Appendix 4: Reserves as Early Warning Indicator
Copyright 2007 Jeffrey Frankel, unless otherwise noted
API-120 - Macroeconomic Policy Analysis I
Professor Jeffrey Frankel, Kennedy School of Government, Harvard University
Copyright 2007 Jeffrey Frankel, unless otherwise noted
API-120 - Macroeconomic Policy Analysis I
Professor Jeffrey Frankel, Kennedy School of Government, Harvard University
Copyright 2007 Jeffrey Frankel, unless otherwise noted
API-120 - Macroeconomic Policy Analysis I
Professor Jeffrey Frankel, Kennedy School of Government, Harvard University
Most Emerging Market countries added rapidly
to reserves after the currency crises of the 1990s.
Traditional denominator
to gauge reserves: imports
Rodrik (2006)
Source: IMF WEO, 2007
2003-07: This time, China and India shared in the inflows.
But capital inflows to EMs financed only reserve accumulation,
not current account deficits as in the past.
FX Reserves in the BRICs, 2000-2011
Neil Bouhan & Paul Swartz, Council on Foreign Relations
Appendix 5: Did the pre-2008 equations predict who got hit
in the Global Financial Crisis of Sept. 2008?
• Obstfeld, Shambaugh & Taylor (2009a, b):
• Finding: A particular measure of countries’ reserve holdings just
before the current crisis, relative to requirements (M2), predicts 2008
depreciation.
• Current account balances & short-term debt levels are not
significant predictors, once reserve levels are taken into account.
• Rose & Spiegel (2009a, b) and Blanchard (2009) found
no role for reserves in predicting who got into trouble.
• Frankel & Saravelos (JIE, 2012): We get stronger results.
We define the crisis period to have gone through March 2009.
28
Top 8 categories of Leading Indicators in pre-2008-crisis literature
Frankel & Saravelos (2012)
Table 1
Leading
Indicator1
Reserves a
Real Exch.Rate b
GDP c
Credit d
Current Acct. e
Money Supply f
Exports or Imports 1a, g
Inflation
KLR
(1998) 2
14
12
6
5
4
2
2
5
Hawkins &
Abiad Others5,
6
Klau (2001)3 (2003)4,6
18
22
15
8
10
16
9
7
13
11
1
6
6
1
4
1
29
5
3
3
3
2
0
2
2
Total
50
48
25
22
22
19
17
15
Next 9 categories of Leading Indicators in pre-2008-crisis literature
Table 1,
Leading Indicator1
Frankel & Saravelos (2012)
continued
KLR
(1998) 2
Hawkins &
Abiad
Others5,6
3
4,6
Klau (2001) (2003)
Total
Terms of Trade
1
2
4
3
2
8
8
4
5
6
3
2
2
1
1
1
1
0
0
0
13
13
10
9
9
Contagionj
1
5
0
0
6
Political/Legal
3
3
2
0
1
0
0
0
6
3
0
28
1
28
1
20
1
7
3
83
Equity Returns
Real Interest Rateh
Debt Compositn1b, i
Budget Balance
Capital Flows1c, k
External Debtl
Number of Studies
30
Frankel & Saravelos (2012)
Notes
1, 1a, 1b, 1c
Leading indicator categories as in Hawkins & Klau (2000), with exception of 1aincludes imports,
1bdebt composition rather than debt to international banks, 1ccapital flows rather than capital account.
2As reported in Hawkins & Klau (2000), but M2/reserves added to reserves, interest differential added to real interest rate.
3S&P, JP Morgan, IMF Indices, IMF Weo, IMF ICM, IMF EWS studies have been excluded due to lack of verifiability of
results. The following adjustments have been made to the authors’ checklist: significant credit variables reduced from 10
to 8 as Kaminsky (1999) considers level rather than growth rate of credit; significant capital account variables reduced
from 1 to 0 as Honohan (1997) variable not in line with definition used here; Kaminsky (1999) significant variables for
external debt reclassified to debt composition as these variables relate to short-term debt.
410 out of 30 studies excluded from analysis. 7 included in Hawkins & Klau (2000) and 3 due to absence of formal testing
of variables.
5Includes Berg, Borenzstein and Pattillo (2004), Manasse and Roubini (2005), Shimpalee and Breuer (2006), Davis and
Karim (2008), Bergmen et.al. (2009), Obstfeld, Shambaugh and Taylor (2009), Rose and Speigel (2009a).
6See App. 1 for criteria defining statistical significance in Abiad (2003) and Others studies. For rest see KLR (1998),
Hawkins & Klau (2001)
Variables included in the leading indicator categories:
aReserves:
bReal
relative to GDP, M2, short-term debt, 12m change
Exchange Rate: change, over/under valuation
cGDP:
growth, level, output gap
dCredit:
nominal or real growth
eCurrent Account:
fMoney
CA/GDP, Trade Balance/GDP
Supply: growth rate, excess M1 balances
gExports
or Imports: relative to GDP, growth
hReal
Interest Rate: domestic or differential
iDebt
Composition: commercial/concess./variable-rate/
debt to internat. banks/short-term/multilat./official relative
to total external debt. Short-term debt relative to reserves
(rather than relative to total external debt) is in the
reserves category
jContagion:
kCapital
dummies for crisis elsewhere
Flows: FDI, short-term capital flows
lExternal
31
Debt: relative to GDP
Table Appendix 6
Coefficients of Bivariate Regressions of Crisis Indicators on Each Independent Variable* (t-stat in parentheses)
bolded number indicates statistical signficance at 10% level or lower, dark er color shading equivalent to higher statistical significance
Currency
Market
Equity
Market
Recourse to
IMF
Industrial
Production
GDP
S ignif ic a nt a nd
C o ns is t e nt
S ign?^
Reserves (% GDP)
0.082
(2.52)
0.850
(1.6)
-1.020
(-1.92)
0.155
(2.22)
0.008
(0.27)
Yes
Reserves (% external debt)
-0.000
(-1.42)
0.000
(2.11)
-0.010
(-3.42)
0.000
(3.62)
0.000
(3.07)
Yes
Reserves (in months of imports)
0.002
(1.58)
0.103
(4.71)
-0.089
(-3.31)
0.006
(1.48)
0.001
(0.75)
Yes
M2 to Reserves
0.000
(0.14)
-0.026
(-3.81)
-0.067
(-1)
-0.001
(-2.46)
0.000
(1.44)
Yes
Short-term Debt (% of reserves)
-0.000
(-2.6)
-0.007
(-4.45)
0.000
(1.18)
-0.000
(-1.7)
-0.000
(-2.93)
Yes
REER (5-yr % rise)
-0.293
(-5.4)
-0.303
(-0.32)
0.889
(0.99)
-0.000
(-0.01)
-0.029
(-0.85)
REER (Dev. from 10-yr av)
-0.292
(-2.93)
-0.920
(-0.81)
0.671
(0.58)
-0.000
(-0.01)
-0.041
(-0.91)
GDP growth (2007, %)
0.003
(1.7)
0.078
(1.58)
0.039
(1.63)
0.010
(2.59)
-0.002
(-1.21)
GDP Growth (last 5 yrs)
0.002
(1.08)
0.118
(2.14)
0.052
(1.68)
0.009
(2.14)
-0.003
(-1.21)
GDP Growth (last 10 yrs)
0.005
(1.59)
0.087
(1.06)
0.042
(1.2)
0.016
(2.63)
-0.004
(-0.76)
GDP per capita (2007, constant 2000$)
-0.003
(-0.7)
-0.296
(-4.69)
-0.221
(-3.23)
-0.027
(-2.48)
-0.010
(-1.74)
Change in Credit (5-yr rise, % GDP)
-0.029
(-0.83)
-1.979
(-5.42)
0.139
(0.37)
-0.092
(-1.67)
-0.065
(-2.34)
Yes
Change in Credit (10-yr rise, % GDP)
-0.024
(-2.84)
-0.904
(-3.9)
-0.011
(-0.08)
-0.046
(-1.58)
-0.019
(-1.13)
Yes
Credit Depth of Information Index (higher=more)
-0.005
(-1.34)
-0.115
(-1.72)
0.009
(0.19)
0.006
(0.57)
-0.003
(-0.47)
Bank liquid reserves to bank assets ratio (%)
0.000
(1.52)
0.022
(1.51)
-0.000
(-13.97)
0.002
(2.34)
0.001
(2.58)
Yes
Current Account (% GDP)
0.001
(1.57)
0.032
(2.18)
-0.032
(-3.46)
0.000
(0.42)
0.000
(0.78)
Yes
Current Account, 5-yr Average (% GDP)
0.001
(1.31)
0.030
(1.66)
-0.032
(-2.76)
0.000
(0.53)
0.000
(0.42)
Current Account, 10-yr Average (% GDP)
0.000
(0.72)
0.034
(1.46)
-0.038
(-2.63)
0.000
(0.15)
0.001
(1.59)
Net National Savings (% GNI)
0.000
(0.9)
0.048
(4.5)
-0.020
(-1.88)
0.003
(2.42)
0.002
(2.92)
0.000
0.047
-0.028
0.003
0.002
F & Saravelos (2010): Bivariate
Independent Variable
R
E
S
E
R
V
E
S
R
E
E
R
G
D
P
C
R
E
D
I
T
C
U
R
R
E
N
T
A
C
C
O
U
N
T
Gross National Savings (% GDP)
Yes
Yes
Yes
Table Appendix 7
Coefficients of Regressions of Crisis Indicators on Each Independent Variable and GDP per Capita* (t-stat in parentheses)
bolded number indicates statistical signficance at 10% level or lower
F & Saravelos
(2010):
Multivariate
Exchange
Market
Pressure
Currency % Recourse to
Changes
IMF
(H208-H109
(SBA only)
Equity
%Chng
(Sep08Mar09)
Equity %
Chng
(H208H109)
S ignif ic a nt
a nd
C o ns is t e nt
S ign?^
Independent Variable
R
E
S
E
R
V
E
S
R
E
E
R
G
D
P
C
R
E
D
I
T
C
U
R
R
E
N
T
A
C
C
O
U
N
T
Reserves (% GDP)
0.164
(3.63)
0.087
(2.98)
-1.069
(-1.66)
0.011
(0.12)
0.010
(0.14)
Yes
Reserves (% external debt)
0.000
(1.06)
0.000
(1.1)
-0.006
(-2.29)
0.000
(1.81)
0.000
(2.65)
Yes
Reserves (in months of imports)
0.004
(2.25)
0.003
(1.95)
-0.119
(-3.01)
0.006
(1.32)
0.009
(2.32)
Yes
M2 to Reserves
0.000
(0.27)
0.000
(0.76)
-0.044
(-0.91)
0.000
(0.02)
-0.000
(-0.09)
Short-term Debt (% of reserves)
-0.000
(-1.97)
-0.000
(-4.22)
0.000
(2.13)
-0.001
(-2.89)
-0.001
(-3.11)
Yes
REER (5-yr % rise)
-0.440
(-5.55)
-0.210
(-3.19)
1.728
(2.15)
-0.182
(-1.24)
-0.185
(-1.61)
Yes
REER (Dev. from 10-yr av)
-0.475
(-3.96)
-0.230
(-2.47)
2.654
(2.56)
-0.316
(-1.71)
-0.316
(-2.1)
Yes
GDP growth (2007, %)
-0.000
(-0.2)
0.001
(0.94)
0.070
(2.58)
-0.001
(-0.1)
-0.007
(-0.71)
GDP Growth (last 5 yrs)
-0.003
(-0.81)
0.000
(0.26)
0.084
(2.4)
-0.003
(-0.26)
-0.014
(-1.15)
GDP Growth (last 10 yrs)
0.000
(0.14)
0.001
(0.43)
0.064
(1.66)
-0.012
(-0.67)
-0.020
(-1.12)
Change in Credit (5-yr rise, % GDP)
-0.021
(-0.36)
-0.035
(-0.98)
0.552
(1.02)
-0.274
(-2.97)
-0.248
(-4.13)
Change in Credit (10-yr rise, % GDP)
-0.017
(-0.93)
-0.011
(-1.05)
0.210
(1.03)
-0.089
(-1.65)
-0.089
(-2.35)
Credit Depth of Information Index (higher=more)
-0.008
(-1.06)
0.000
(0.05)
0.224
(2.4)
-0.006
(-0.37)
-0.018
(-1.33)
Bank liquid reserves to bank assets ratio (%)
0.000
(3.84)
0.000
(0.5)
-0.000
(-11.44)
-0.002
(-0.54)
-0.002
(-0.79)
Yes
Current Account (% GDP)
0.001
(1.48)
0.002
(2.7)
-0.023
(-2.09)
0.009
(3.84)
0.007
(3.95)
Yes
Current Account, 5-yr Average (% GDP)
0.000
(0.48)
0.001
(1.82)
-0.025
(-1.72)
0.007
(2.4)
0.006
(2.74)
Yes
Current Account, 10-yr Average (% GDP)
0.000
(0.14)
0.002
(1.39)
-0.035
(-2.11)
0.008
(2.21)
0.007
(2.44)
Yes
Net National Savings (% GNI)
0.002
(1.6)
0.001
(2.33)
-0.013
(-1.22)
0.006
(2.92)
0.004
(2.28)
Yes
Gross National Savings (% GDP)
0.003
(2.01)
0.001
(2.53)
-0.015
(-1.36)
0.008
(3.42)
0.006
(3.03)
Yes
Yes
Actual versus Predicted Incidence of 2008-09 Crisis
Frankel & Saravelos (2010)
more
resilient
Denmark
Actual Resilience to Crisis
Japan
St. Lucia
Gabon
Guyana
US
China Netherlands
Costa Rica
SwitzerlandBolivia
Singapore
Burundi
Philippines
Nicar Bulgaria
Dom. Republic
MoroccoSaudi Arabia
Israel Italy
Croatia
Ireland
Algeria Malaysia
Australia
Chile Germany
Greece Finland
Venezuela
Portugal
Canada
Norway
Czech Rep.
Uruguay
France
S. Africa
Luxembourg
Paraguay
UK
ColombiaNigeria
Gambia
Malawi
New Zealand
Romania
Sweden
Russia
Pakistan
Slovakia
Poland
Georgia
less
resilient
more
resilient
Iceland
Hungary
Predicted Resilience to Crisis
34