Transcript On the Empirics of Sudden Stops - Inter

ON THE EMPIRICS
OF SUDDEN STOPS
Guillermo Calvo,
Alejandro Izquierdo
and Luis-Fernando Mejía
April 10, 2003
OUTLINE
I.
Financial Crises: Sudden Stops vs. Competing
Views
II. Sudden Stops: Definition, Characterization
and Links to Key Macro Variables
III. Determinants of Sudden Stops
IV. Policy Lessons
Sudden Stops in EMs
• The sequence of financial crises following the Tequila crisis
suggests EMs are seriously vulnerable to shocks in the capital
account.
• Sudden Stops (SS) in capital flows can trigger major
adjustments, particularly through their effects on the RER,
resulting in major disruption in trade and finance.
• EMs are particularly vulnerable to RER fluctuations given their
high degree of liability dollarization.
• By raising doubts about sustainability of the initial equilibrium, a
SS could plunge the economy into a “bad equilibrium” with low
investment and growth.
What Stops Lending? Competing Views
• Is it lack of fiscal discipline? Could be, but this explanation finds
little support for East Asian crises (Korea’s public debt hovered
around 10% of GDP)
• Is it soft pegs? Could be, but how does it account for the ensuing
real meltdown?
• Is it self-fulfilling Sudden Stops? Loss of access may not be the
result of over-indebtedness in the context of a “good” equilibrium, but
rather the result of the economy having fallen in a bad equilibrium
triggered by a SS.
• “Inverse” fiscal view finds support in that SS tend to occur around the
same time, and for countries with very different fiscal conditions. This
view does not neglect the relevance of domestic factors, which are key
in explaining vulnerability to SS.
Making Sudden Stops Operational
Sudden Stops meet the following characteristics:
• Fall in net capital inflows exceeds two standard deviations
below the sample mean at the time of the fall (“unexpected”
requirement).
• SS are persistent events: they are over after 6 months of
consecutive positive changes in capital flows (yoy).
• The fall in capital flows exceeds 10% of private sector credit
• This phase overlaps with a 24-month window centered
around RER depreciation exceeding 15%.
RER Depreciation and Reversals in EMs
RER Depreciation and Reversals in DEs
Key Characteristics of Sudden Stops
• The capital account remains closed for EMs during currency
crises, but not for developed countries: 81% of depreciation
episodes are associated with large capital flow reversals, i.e.
sudden stops take place. This figure is only 25% for developed
countries.
• What comes first, RER depreciation or capital flow reversal?
Not a clear-cut answer, though 65% of the time reversals come
first.
• There is Sudden Stop bunching, particularly around the
Russian crisis.
Sudden Stop Bunching
Links to Key Macro Variables
• Sudden Stops are associated with substantial increases in real
interest rates. They represent shifts in the supply of international
credit.
• Sudden Stops are associated with big output contractions,
implying real costs of loosing access to credit.
• Sudden Stops coincide with substantial reserve losses, implying
central bank attempts to prevent abrupt CAD gap closures and
exchange rate depreciation, a strategy that is not successful to the
extent that SS are persistent.
Sudden Stop and Interes Rates in EMs
Sudden Stop and Reserves in EMs
Sudden Stop and Growth in EMs
In Search of Determinants
Based on Calvo, Izquierdo and Talvi (2002) we zero in on
determinants of the likelihood of having a SS :
•  = (Y - S)/Z = 1-CAD/Z as an indicator of potential RER
changes, where Z = tradables absorption; Y = tradables output; S =
non-factor payments
• Financial dollarization
• Public dollarization
• Debt levels
• Reserves/CAD
• Exchange rate regime
Probit Results (All countries)
Lag w
Lag fin. dol.
(1)
ss_15
-3.288
(2.37)**
7.870
(4.22)***
Lag pub dol
(2)
ss_15
-4.136
(2.44)**
7.285
(3.54)***
0.806
(0.93)
Lag res/CAD
(3)
ss_15
-4.111
(2.42)**
7.231
(3.49)***
0.812
(0.93)
-0.001
(0.19)
Dummy EM
(4)
ss_15
-3.284
(1.95)*
5.304
(2.37)**
0.593
(0.73)
-0.000
(0.02)
0.673
(1.53)
Lag scaled
debt
(5)
ss_15
-3.349
(1.99)**
5.420
(2.40)**
0.301
(0.25)
-0.000
(0.02)
0.657
(1.49)
(6)
ss_15
-3.392
(2.07)**
5.499
(2.44)**
0.161
(0.13)
-0.001
(0.09)
0.706
(1.59)
(7)
ss_15
-3.363
(2.05)**
5.400
(2.41)**
0.235
(0.20)
-0.001
(0.08)
0.695
(1.57)
13.750
17.252
15.551
(0.33)
(0.41)
0.150
(0.92)
(0.37)
llys3
llys5
Constant
Observations
Time
dummies
1.309
(0.98)
252
2.003
(1.27)
228
1.983
(1.26)
228
0.933
(0.58)
228
1.013
(0.62)
228
0.765
(0.48)
228
0.079
(0.75)
0.763
(0.47)
228
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Absolute value of z statistics in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1%
Probit Results (EMs)
Lag w
Lag fin. dol.
(1)
ss_15
-3.891
(2.00)**
5.457
(2.39)**
Lag pub dol
(2)
ss_15
-4.046
(2.01)**
5.213
(2.27)**
0.569
(0.66)
Lag res/CAD
(3)
ss_15
-4.397
(2.14)**
5.724
(2.32)**
0.476
(0.55)
0.029
(1.19)
Lag scaled
debt
(4)
ss_15
-4.524
(2.19)**
5.872
(2.37)**
0.072
(0.05)
0.030
(1.20)
(5)
ss_15
-4.640
(2.22)**
5.982
(2.37)**
0.029
(0.02)
0.028
(1.14)
(6)
ss_15
-4.567
(2.21)**
5.883
(2.36)**
0.081
(0.06)
0.029
(1.15)
18.687
19.697
18.468
(0.40)
(0.41)
0.127
(0.60)
(0.39)
llys3
llys5
Constant
Observations
Time
dummies
2.226
(1.17)
116
2.243
(1.15)
116
2.533
(1.27)
116
2.669
(1.33)
116
2.557
(1.26)
116
0.051
(0.38)
2.347
(1.24)
116
Yes
Yes
Yes
Yes
Yes
Yes
Absolute value of z statistics in parentheses
* significant at 10%; ** significant at 5%; *** significant at 1%
The interaction between Tradable Absorption
Leverage and Financial Dollarization
0
.2
.4
.6
.8
Probability of a Sudden Stop
.6
.8
1
w_min_mean2
w_max_mean2
lw
1.2
1.4
w_mean_mean2
1.6
Policy Lessons:
Two ways to go:
1) Contingent debt contracts: non-tradable price indexed, CPI
indexed, GDP indexed, commodity price indexed (when
associated with SS).
2) Work on structural deficiencies by putting in place domestic
policies that:
•
Increase openness (decrease leverage of absorption of
tradables),
•
Decrease liability dollarization, and
•
Bring down debt levels.
Policy Lessons (cont.)
Increasing openness is particularly useful:
a) It reduces the size of RER swings after Sudden
Stop
b) A higher share of tradable sectors in output
reduces risk of mismatches in private sector balance
sheets and banking sector vulnerability.
Other findings:
• Closed, dollarized economies may be vulnerable
independently of the exchange rate regime that is adopted.