Transcript Macro Conference IV

Chapter 16
Speculative Attacks
and Exchange-Rate Crises
© Pierre-Richard Agénor and Peter J. Montiel
1



Two important models of speculative attacks and
exchange rate crises:
“Conventional” models:
 inconsistencies between fiscal, monetary and
exchange rate policies;
 role of speculative attacks in “forcing” the
abandonment of a currency peg.
Recent models:
 vulnerability of exchange rate systems even in the
presence of consistent macroeconomic policies and
sound market fundamentals;
 account for policymakers' preferences and trade-offs
that they face in their policy objectives;
2
view exchange-rate “crisis” as ex ante optimal
decision for the policymaker;
 highlight role of self-fulfilling mechanisms, multiple
equilibria, and credibility factors.

3




Conventional Approach.
Optimizing Policymakers and Self-Fulfilling Crises.
A “Cross-Generation” Framework.
Evidence on Exchange-Rate Crises.
4
The Conventional Approach
Viability of a fixed exchange-rate regime requires
maintaining long-run consistency between monetary,
fiscal, and exchange-rate policies.
 “Excessive” domestic credit growth leads to:
 gradual loss of foreign reserves;
 abandonment of fixed exchange rate, once the
central bank becomes incapable of defending parity.
Krugman (1979):
 Under a fixed exchange-rate regime, domestic credit
creation in excess of money demand growth may lead
to a sudden speculative attack against the currency.
 This forces abandonment of the fixed exchange rate.
 This attack occurs before the central bank would have
run out of reserves in the absence of speculation, and
6
takes place at a well-defined date.



The Basic Model.
Extensions to the Basic Framework.
 Sterilization.
 Alternative Post-Collapse Regimes.
 Real effects of an Anticipated Collapse.
 Borrowing, Controls, and the Postponement of a
Crisis.
7
The Basic Model





Small open economy whose residents consume a
single, tradable good.
Domestic supply of the good is exogenous, and its
foreign-currency price is fixed.
Domestic price level is equal, as a result of purchasing
power parity, to nominal exchange rate.
Agents hold three categories of assets:
 domestic money,
 domestic and foreign bonds (perfectly substitutable).
There are no private banks.
8




Thus, money stock is equal to the sum of
 domestic credit issued by the central bank;
 domestic-currency value of foreign reserves held by
the central bank.
Foreign reserves earn no interest.
Domestic credit expands at a constant growth rate.
Agents are endowed with perfect foresight.
9

Model:
m – p = y - i,
m = d + (1-)R,
.
d =  > 0, (3)
 > 0,
(1)
0 <  < 1,
(2)
p = e, (4)
.
i = i* + e,
(5)
m: nominal money stock; d: domestic credit;
R: domestic-currency value of foreign reserves held by
the central bank;
e: spot exchange rate; p: price level;
y: exogenous output; i*: foreign interest rate;
i: domestic interest rate;
all variables, except interest rates, are in logarithms.10




(1): relates real demand for money positively to y and
negatively to i.
(2): log-linear approximation to the identity defining m as
stock of reserves and domestic credit, which grows at
the rate  (3).
(4) and (5): purchasing power parity and uncovered
interest parity.
Setting  = y - i* and combining (1), (4), and (5) yields
.
m – e =  - e,
 > 0.
(6)
11
.
Under a fixed exchange-rate regime, e = e and e = 0:
_

_
m – e = .


(7)
(7): central bank accommodates any change in
domestic money demand through the purchase or sale
of foreign reserves to the public.
Using (2) and (7) yields
_
R = ( + e - d)/(1-).
(8)
12

Using (3),
.
R  -/,   (1-)/.




(9)
(9): if domestic credit expansion is excessive reserves
are run down at a rate proportional to the rate of credit
expansion.
Thus any finite stock of foreign reserves will be depleted
in a finite period of time.
Central bank announces at time t that it will stop
defending the current fixed exchange rate after reserves
reach a lower bound, Rl.
With a positive rate of , rational agents
 anticipate that reserves will fall to Rl,
13
 foresee the ultimate collapse of the system.






To avoid losses arising from abrupt depreciation of
exchange rate at the time of collapse, speculators will
force a crisis before Rl is reached.
Issue: determine the exact moment at which the fixed
exchange-rate regime is abandoned.
Length of transition period can be calculated by using
backward induction (Flood and Garber, 1984).
In equilibrium and under perfect foresight, agents never
expect a discrete jump in the level of exchange rate.
Reason: jump would provide them with profitable
arbitrage opportunities.
Thus, arbitrage in foreign exchange market requires
exchange rate that prevails immediately after the attack
to equal fixed rate prevailing at the time of the attack.
14




Time of collapse: at the point where the “shadow
floating rate,” is equal to prevailing fixed rate.
Shadow floating rate: exchange rate that would prevail
with the current credit stock if
 reserves had fallen to Rl and
 exchange rate were allowed to float freely.
As long as fixed exchange rate is more depreciated
than shadow floating rate, fixed-rate regime is viable.
If shadow floating rate falls below prevailing fixed rate:
 speculators would not profit from driving stock of
reserves to Rl ;
 because they would experience instantaneous capital
loss on their purchases of foreign currency.
15






If the shadow floating rate is above fixed rate,
speculators would experience capital gain.
Neither anticipated capital gains nor losses at an infinite
rate are compatible with a perfect-foresight equilibrium.
Speculators compete with each other to eliminate such
opportunities.
This leads to an equilibrium attack.
This incorporates arbitrage condition: pre-attack fixed
rate should equal post-attack floating rate.
Shadow floating exchange rate:
e = 0 + 1m,
0 and 1: as-yet-undetermined coefficients;
m = d + (1-)Rl from (2).
(10)
16

Taking the rate of change of (10) .and noting
from (2)
.
that under a floating-rate regime m = d yields
.
e = 1.


In post-collapse regime, exchange rate depreciates
steadily and proportionally to .
Substituting (11) in (6) yields, with  = 0,
e = m + 1.

(11)
(12)
Comparing (12) and (10) yields
0 = ,
1 = 1 .
17


From (3), d = d0 + t.
Using definition of m and substituting in (12) yields
e = (d0 + ) + (1 - )Rl + t.


(13)
Fixed exchange-rate
_ regime collapses when the
prevailing parity, e, equals the shadow floating rate, e.
From (13)
_ exact time of collapse, tc, is obtained by
setting e = e, so that
_
tc = [e - d0 - (1-)Rl]/ - .
18

_
Since, from (2) and (7) e = d0 + (1-)R0,
tc = (R0 - Rl)/ - ,





(14)
R0: initial stock of reserves.
The higher R0, the lower the critical level.
The lower , the longer it will take before the collapse
occurs.
With no “speculative” demand for money,  = 0, and the
collapse occurs when reserves are run down to Rl.
Interest rate elasticity of money demand determines
size of downward shift in money balances and reserves
when fixed exchange-rate regime collapses.
The larger  is, the earlier the crisis.
19


Implication: speculative attack occurs before the central
bank would have reached Rl without speculation.
Using (8) with  = 0 yields the stock of reserves just
before the attack:
_
Rt_c  lim_Rtc = (e - dt_c )/(1-),
ttc
_
d = d0 + tc, so that
_
_
tc
Rt c = [e - (d0 + tc )]/(1-).
_
_

Using (14) yields
_
e - d0 = (tc + ) + (1-)Rl.
_
20

Finally, combining (15) and (16) yields
Rt_c = Rl + /.





Figure 16.1: process of a balance-of-payments crisis,
under the assumption that Rl = 0.
Top panel: behavior of reserves, domestic credit, and
the money stock before and after the regime change.
Bottom panel: behavior of exchange rate.
Prior to the collapse at tc, money stock is constant, but
its composition varies since domestic credit rises at the
rate  and reserves decline at the rate /.
Instant before the regime shift, speculative attack
occurs, and both reserves and the money stock fall by
21
/.
F
i
g
u
r
e
1
6
.
1
a
T
h
e
P
r
o
c
e
s
s
o
f
a
B
a
l
a
n
c
e
o
f
P
a
y
m
e
n
t
s
C
r
i
s
i
s
M
o
n
e
y
s
t
o
c
k
D
,
R
,
m
{




D
o
m
e
s
t
i
c
c
r
e
d
i
t
D
0
R
0
R
e
s
e
r
v
e
s
{







t
=
R



0
c
S
o
u
r
c
e
:
A
g
é
n
o
r
a
n
d
F
l
o
o
d
(
1
9
9
4
,
p
.
2
3
0
)
.

R


0
T
i
m
e
22
F
i
g
u
r
e
1
6
.
1
b
T
h
e
P
r
o
c
e
s
s
o
f
a
B
a
l
a
n
c
e
o
f
P
a
y
m
e
n
t
s
C
r
i
s
i
s
C
e

(
D
+
)
+
t





0
A
_
B
e
t
c
T
i
m
e
S
o
u
r
c
e
:
A
g
é
n
o
r
a
n
d
F
l
o
o
d
(
1
9
9
4
,
p
.
2
3
0
)
.
23





Because Rl = 0, money stock is equal to domestic credit
in the post-collapse regime.
_
Exchange rate remains constant at e until the collapse
occurs.
Path continuing through AB and taking discrete
exchange-rate jump BC: “natural collapse” scenario ( =
0).
With speculation, transition occurs at A, preventing a
discrete change in exchange rate from occurring.
Speculators avoid losses that would result from discrete
exchange-rate change by attacking the currency at the
point where the transition to the float is smooth.
24
Extensions to the Basic
Framework




Alternative assumptions regarding post-collapse
exchange-rate regime: case of a temporary postcollapse period of floating followed by repegging.
Real effects of an exchange-rate collapse.
Role of foreign borrowing.
Imposition of capital controls as policy measures aimed
at postponing the occurrence of a balance-of-payments
crisis.
25
Sterilization





Key assumption of the Krugman-Flood-Garber model:
money supply falls, in line with money demand, at the
moment the currency attack takes place.
But, if reserve losses are completely sterilized, such a
discrete jump will not take place.
Flood, Garber, and Kramer (1996) studied this case.
Fixed exchange-rate regime is not viable; as long as
agents understand that the central bank plans to
sterilize speculative attack they will attack immediately.
Consider the money market equilibrium condition (6)
with  = 0.
26

If money stock is constant as a result of sterilized
intervention (m = mS) and exchange rate is fixed, this
condition becomes
_
mS – e = 0.

.
In the post-attack floating-rate regime, with e = :
mS – e = -.

Subtracting the second expression from the first yields
_
e – e =  > 0.
27




Thus, if m supply does not change_ when the attack
takes place, e will always exceed e, thereby provoking
an immediate attack.
By adding risk premium, Flood, Garber, and Kramer
(1996) show that the model with sterilization can be
compatible with a fixed exchange rate.
Risk premium adjusts to keep demand for money
constant, just as sterilization maintains money supply
constant.
Since m supply does not change, and exchange rate
cannot jump, domestic interest rate cannot jump either.
28
Alternative Post-Collapse Regimes



After breakdown of fixed-rate system, central bank can
 devalue the currency,
 implement a dual-exchange-rate arrangement, or
 adopt a crawling peg regime.
Timing of a crisis depends on exchange-rate
arrangement that agents expect the central bank to
adopt after abandonment of initial fixed rate.
Obstfeld (1984): after allowing currency to float for a
certain period of time, central bank
 returns to foreign exchange market, and
 fixes exchange rate at a new, more depreciated level.
29




Assume:
of transitory period of floating, T, and
_ length
_
level eH > e to which exchange rate will be pegged at
the end of transition are known with certainty.
tc is calculated by a process of backward induction.
This principle imposes two restrictions:
 initial fixed rate e must coincide with relevant shadow
_
floating rate, that is, e = etc ;
_
 at time tc + T, preannounced new fixed rate eH must
_
also coincide with the interim floating rate, eH = et +T.
c
When central bank's policy was assumed to involve
abandonment
of fixed rate and adoption of permanent
_
float, e was given by (12).
30

Under a transitory floating regime, shadow rate
e = 0 + 1m + C exp(t/),



tc  t  tc + T
(18)
C: undetermined constant.
Complete solution must specify tc and C.
_
_
These values are obtained by imposing e = et c and eH =
et c+T on (18).
Solutions for tc and C are
_
tc = (e -  - d0 - )/,
(19)
C =  exp(-tc/),
(20)
_
_
where  = [(eH - e) - T]/[exp(T/) - 1].
31




(19): collapse time is linked
_
_to the magnitude of the
expected devaluation (eH - e) and length of transitional
float.
Crises occur earlier the greater the anticipated
devaluation.
Relationship between collapse time and length of
floating-rate interval: negative for small T and positive
for large T.
If transitional float is sufficiently brief, speculative attack
on the domestic currency will occur as soon as private
agents realize that current exchange rate cannot be
enforced indefinitely.
32
Real Effects of an Anticipated Collapse



Evidence: balance-of-payments crises are associated
with large current account movements during the
periods preceding, and following, such crises.
Large external deficits emerge as agents, in anticipation
of crisis, adjust their
 consumption pattern,
 composition of their holdings of financial assets.
Movements in real exchange rate and current account
may explain why speculative attacks are preceded by a
period during which official foreign reserves are lost at
accelerating rates.
33
Willman (1988):
 Convenient framework for examining real effects of
exchange-rate crises.
 Domestic output is demand determined, positively
related to real exchange rate, and inversely related to
real interest rate.
 Trade balance depends positively on real exchange rate
but is negatively related to aggregate demand.
 Prices are set as a mark-up over wages and imported
input costs.
 Nominal wages are determined through forward-looking
contracts.
34





When the collapse occurs:
 inflation jumps up,
 rate of depreciation of real exchange rate jumps
down,
 real interest rate falls.
Since wage contracts are forward looking, anticipated
future increases in prices are discounted back to the
present and affect current wages.
Thus, prices start adjusting before the collapse occurs.
Real interest rate falls gradually and jump downward at
the moment the collapse takes place.
Two opposite effects on domestic activity:
 Decline in real interest rate has an expansionary
effect on domestic activity before the collapse occurs.
35
Rise in domestic prices results in appreciation of
domestic currency, which has an adverse effect on
economic activity.
 If relative price effects are strong, net impact of an
anticipated collapse on output may be negative.
 Continuous loss of competitiveness: trade balance
deteriorates in the periods preceding the collapse of
fixed exchange-rate regime.
 Trade deficit increases further when the crisis occurs.
 Then, due to gradual depreciation of real exchange rate,
it returns to its steady-state level.
Kimbrough (1992):
 Role of intertemporal substitution effects in
understanding real effects of exchange-rate crises.

36



Optimizing framework in which money reduces
transactions costs.
Effects of anticipated speculative attack on current
account depends on difference between
 interest elasticity of the demand for money and
 intertemporal elasticity of substitution in consumption.
If the latter exceeds the former, anticipated speculative
attack,
 raises consumption and real balances at the moment
agents realize that fixed exchange rate will collapse,
 leads to continued deterioration of current account
until the attack takes place.
37



If the former exceeds the latter,
 reduction in consumption and real money balances,
 immediate and continued improvement in current
account until the time of the speculative attack.
Implication: anticipated speculative attacks may not be
associated with similar real effects in all countries and at
all times.
But, for several Latin American countries, speculative
attacks and balance-of-payments crises have been
associated with large current account deficits.
38
Borrowing, Controls, and the Postponement
of a Crisis
Countries facing balance-of-payments difficulties:
 external borrowing to supplement reserves available
to defend the official parity;
 restrictions on capital outflows to limit losses of
foreign exchange reserves.
Borrowing:
 Assumption of the basic model: there is a critical level,
known by everyone, below which foreign reserves are
not allowed to be depleted.
 But, this binding threshold may not exist.

39






Perfect access to capital international markets: central
bank reserves can be negative without violating
government's intertemporal solvency constraint.
This could postpone or avoid a regime collapse.
But rate of growth of domestic credit cannot be
maintained above world interest rate, because it causes
violation of the government budget constraint.
Thus, over-expansionary credit policy would lead to the
collapse of fixed exchange-rate regime.
Even with perfect capital markets, timing of borrowing
matters for the nature of speculative attacks.
Suppose: interest cost of servicing foreign debt exceeds
interest rate paid on reserves.
40
If borrowing occurs before fixed exchange rate would
have collapsed without borrowing, the crisis can be
postponed.
 If borrowing occurs long enough before exchangerate regime would have collapsed without borrowing,
the crisis would occur earlier.
Collapse is brought forward due to servicing cost of
foreign indebtedness on the public sector deficit, which
raises rate of growth of domestic credit.
In practice, most developing countries face borrowing
constraints on international capital markets.
This has implications for inflation in an economy where
agents are subject to intertemporal budget constraint.




41

Example: country that has no opportunity to borrow
externally and in which the central bank transfers its net
profits to the government.
 If speculative attack occurs, the central bank will lose
its reserves, and its post-collapse profits from interest
earnings on those reserves will be zero.
 Thus, net income of the government will fall and
budget deficit will deteriorate.
 If the deficit is financed by increased domestic credit,
post-collapse inflation rate will exceed prevailed rate
in pre-collapse fixed exchange-rate regime.
 This raises inflation tax revenue to compensate for
fall in interest income.
42
Capital controls:
 Controls imposed either permanently or temporarily
 after the central bank had experienced significant
losses, or
 when domestic currency came under heavy pressure
on foreign exchange markets.
 Agénor and Flood (1994):
 with permanent controls, the higher the degree of
capital controls, the longer it will take for fixed
exchange rate to collapse.
 Reason: controls dampen the size of expected future
jump in domestic nominal interest rate and
associated shift in demand for money.
43



Bacchetta (1990):
 Temporary restrictions on capital movements may
have pronounced real effects.
 In perfect-foresight world, agents will anticipate the
introduction of controls.
Critical to distinguish
 case in which timing of the policy change is perfectly
anticipated;
 case in which it is not.
If controls take agents by surprise,
 capital outflows will be replaced by higher imports
once such controls are put in place;
 this leads to deterioration in the current account until
a “natural” collapse occurs;
44
thus, accelerated rate of depletion of foreign reserves
through the current account will precipitate the crisis.
If capital controls are preannounced, or if agents guess
exact time at which controls will be introduced,
 speculative attack may occur just before the controls
are imposed, as agents attempt to readjust their
portfolios and evade restrictions;
 this defeats purpose of capital controls and may in
fact precipitate regime collapse.
Directions in which the theory of balance-of-payments
crises has been extended: uncertainty and regime
switches.



45
Uncertainty:
 Uncertainty on domestic credit growth provides
explanation on sharp increases in domestic nominal
interest rates.
 Transition to floating-rate regime becomes stochastic:
collapse time is a random variable.
 There will be a nonzero probability of a speculative
attack in the next period, producing forward discount on
the domestic currency.
 Degree of uncertainty about the central bank's credit
policy plays an important role in the speed at which
reserves of the central bank are depleted.
46
Endogenous adjustment of fiscal and credit policies:
 Example: central bank may float the currency and
abandon prevailing fixed exchange rate at the moment
reserves hit their critical lower bound.
 Drazen and Helpman (1988) and Edwards and Montiel
(1989): authorities choose to adjust exchange rate
instead of altering underlying macroeconomic policy.
 If new exchange-rate regime is inconsistent with
underlying fiscal policy process, there will be a need for
a new policy regime.
Possibility of multiple equilibria:
 Implication of endogenous credit policy rule.
47


Obstfeld (1986c):
 Domestic growth is consistent with indefinite viability
of fixed exchange rate as long as the regime is
maintained ( = 0).
 But contingent on collapse of fixed exchange rate,
loss of discipline causes domestic credit growth rate
to increase (  0).
In such a setting, multiple equilibria may emerge.
 Fixed exchange rate can survive if asset holders
believe that it will not collapse.
 If private agents believe that collapse will occur, run
on official reserves will bring the regime down,
 triggering contingent shift in domestic credit
growth,
48
 validating the attack.




Consider the case where  = 0:
 from (14), tc = ,
 regime survives indefinitely.
Consider that contingent on collapse of fixed exchange
rate:
 agents expect c > 0, and (R0 - Rl)/c < , so that
tc < 0;
 immediate attack will take place.
Result: private agents' beliefs about viability of fixed
exchange rate become a key element in determining
timing of the crisis.
Shifts across alternative equilibria may be self-fulfilling.
49
Optimizing Policymakers
and Self-Fulfilling Crises




Key feature of the recent literature on currency crises:
 multiple equilibria,
 explicit modeling of policymakers' preferences and
policy rules.
Policymakers are viewed as
 deriving benefits from pegging the currency;
 facing other policy objectives (level of unemployment
and domestic interest rates).
Thus, depending on circumstances, policymakers may
find it optimal to abandon the official parity.
Occurrence of exchange rate “crisis” is
 not related to existence of sufficient level of reserves;
 related to implementation of a contingent rule for
setting exchange rate.
51




Each period, policymaker considers costs and benefits
of maintaining the peg for another period, and decides
whether or not to abandon it.
This decision depends on realization of domestic or
external shocks.
For a given cost associated with abandoning currency
peg, there exists a range of values for the shocks that
makes maintaining the peg optimal.
But, for sufficiently large realizations of shocks, loss in
flexibility may exceed loss incurred by abandoning peg.
52



The Output-Inflation Tradeoff.
Public Debt and Self-Fulfilling Crises.
Credibility, Reputation, and Currency Crises.
53
The Output-Inflation Tradeoff


Obstfeld (1996): “rational” policymakers and role of selffulfilling factors and emphasis on output-inflation
tradeoff.
Government's loss function:
~2
L = (y - y)
+ e2 + c,
>0
(21)
~ policymaker's output target;
y: output; y:
e: exchange rate;
c: fixed cost associated with changes in official parity.
54

Output is determined by expectations-augmented
Phillips curve
_
y = y + ( - a) - u,
_




y: “natural” level of output,
  e,
a: domestic price-setters' expectation of ,
u: zero-mean shock.
_
Assume that y~ > y.
Price setters form their expectations prior to observing
shock u.
Policymaker chooses e after observing the shock.
Devaluation bears a cost of cd, and revaluation cost of
55
r
c.


Ignore c in (21).
With a predetermined, policymaker chooses
~ _
=

(y – y + u) + 2a
2 + 
(23)
Output:
_
y=y+
_
~
2(y – y)
-  - a
2 + 
56

Policy loss (D for discretionary):
_

~
a) 2
D
(y
–
y
+
u
+

L =
2 + 

If the government foregoes use of the exchange rate,
policy loss:
~
_
LF = (y – y + u + a)2.


Consider now fixed cost c.
When fixed costs exist, (23) is operative only when u is
D
d
F
 so large that L + c < L , or
D
r
F
 so low that L + c < L .
57

Devaluation (revaluation) takes place for u > ud (< ur):
d(2 + ) - (y – y) - a
c

1
r
u = 
r(2 + ) - (y – y) - a
c

ud

_
1

=
~
~
_
Suppose that u is uniformly distributed in (-, ).
58

Rational expectation of next period's , given price
setters expectation a:
E = E( | u < ur) Pr(u < ur)
+ E( | u > ud) Pr(u > ud).

Using (23):

E =
2 + 
d2 – ur2
ud – ur ~ _
u
1(y – y + a) 4
2
(
)
59



In full equilibrium, E = a.
Figure 16.2: Equation (24).
Slope of the curve is given by, setting  = 2 + :
2-1
for ur > -
_
dE
~
2-1[(/2)+(/2)(y – y + 2)] for ur = - 

=
da
2-1


for ud = -
Three possible equilibria corresponding to three
different probabilities of devaluation and realignment
magnitudes conditional on a devaluation taking place.
These equilibria are denoted by points A, B, and C.
60
F
i
g
u
r
e
1
6
.
2
M
u
l
t
i
p
l
e
E
q
u
i
l
i
b
r
i
a
i
n
a
M
o
d
e
l
o
f
S
e
l
f
F
u
l
f
i
l
l
i
n
g
E
x
c
h
a
n
g
e
R
a
t
e
C
r
i
s
i
s
E

E
x
p
e
c
t
e
d
d
e
p
r
e
c
i
a
t
i
o
n
r
a
t
e
u
n
d
e
r
a
f
l
o
a
t
i
n
g
e
x
c
h
a
n
g
e
r
a
t
e
r
e
g
i
m
e C
B
"
C
o
n
s
i
s
t
e
n
c
y
"
c
u
r
v
e
A
4
5
º
a

S
o
u
r
c
e
:
A
d
a
p
t
e
d
f
r
o
m
O
b
s
t
f
e
l
d
(
1
9
9
6
,
p
.
1
0
4
3
)
.
61


Once a is sufficiently high for ud to remain at -,
 government's reaction function is given by (23) and
 expected depreciation rate is the same as under a
flexible exchange rate regime.
Expected depreciation rate is obtained by setting  = a
in (23):
~ _
 = [(y – y + u)/].

To ensure that equilibrium C exist, necessary condition
for multiplicities to exist, requires
_
-1(y – y) -   -1 cd.
~
62

If private agents form expectation of average
depreciation rate of floating exchange rate it will
materialize, as long as fixed devaluation cost is not too
high.
63
Public Debt and Self-Fulfilling
Crises



Cole and Kehoe (1996): role of short average term of
public external debt in allowing temporary loss of
investor confidence to produce severe and persistent
economic crisis.
Implication: financial crises can be avoided if
governments diversify term structure of their debt to
ensure that small portion of it matures during any
particular interval of time.
Government inherits a certain amount of foreign debt
that it must either retire, refinance, or repudiate.
64




Initial stock of public debt is so large that it is either
 not feasible to repay it in one period or
 can be immediately retired only at the cost of
significant loss in welfare.
But, repudiating the debt, although costly, may be
preferable to retiring or refinancing the debt under some
circumstances.
Government cannot credibly commit itself to refusing to
repudiate the debt at a future date if repudiation turns
out to be the “best” strategy at that date.
If initial debt is large enough it is possible to admit
multiple equilibrium outcomes, depending on the nature
of foreign lenders' expectations.
65
If foreign lenders expect government to be able to
service its debts,
 government bonds will sell at a moderate price;
 it will be optimal for the government to refinance
them.
 If lenders believe that government will not be able to
service its debts,
 they will be unwilling to lend to the government;
 it may be optimal to repudiate the debt and crisis
may occur.
Thus, foreign lenders' expectations that government will
not be able to service its debt are self-fulfilling.
This situation can arise stochastically.



66





“Bad” states of nature are tied to adverse realization of
a spurious indicator variable.
There can be only one crisis, because after
government has repudiated its debt it has no reason to
borrow.
Financial crisis can occur only if the debt that needs to
be rolled over at that particular date is large.
As a result, changing maturity structure of debt can
prevent crises from occurring.
Even if lenders believe that government will not be able
to refinance its debt, government can retire maturing
debt without incurring welfare costs large enough to
give it incentive to repudiate its debt.
67




Then, equilibrium in which self-fulfilling beliefs by
lenders that government will fail to repay does not exist
and a crisis cannot occur.
Other source of policy trade-offs: effects of higher
interest rates.
Example:
 Banks may come under pressure if market interest
rates rise unexpectedly.
 To avoid a costly bailout, the policymaker may want
to implement a quick devaluation.
Ozkan and Sutherland (1998):
 With sticky domestic prices, a hike in nominal interest
rates may imply hikes in short-term real rates.
 These may generate self-fulfilling devaluation
68
pressures.

In all of these models:
 “fundamentals” affect multiplicity of equilibria;
 but policymakers are incapable of enforcing its
preferred equilibrium;
 “sunspots” could shift exchange rate
 from a position where it is vulnerable to only very
bad realizations of a shock
 to one where output is so low absent devaluation
that even “small” shocks will induce authorities to
devalue.
69
Credibility, Reputation,
and Currency Crises




Drazen and Masson (1994): role of credibility and
reputational factors in models of currency crises with
optimizing policymakers.
Credibility consists of two elements:
 assessment of the policymaker's “type”;
 assessment of the probability that a policymaker will
stick to announced policies in the presence of
adverse shocks.
Policy commitment: maintain exchange rate peg in the
face of shocks to reserves.
Agénor and Masson (1999): extension of the Drazen70
Masson approach.


Figure 16.3: structure of the Agénor-Masson model.
Policymaker's one-period loss function:
~2
L = (i - i) + e,
>0
(25)
i: interest rate on domestic-currency denominated
~
assets (with i its desired level),
e: (logarithm of) nominal exchange rate;
: two values w and T, for weak and tough
governments respectively, with T > w.
71
F
ig
u
re1
6
.3
C
re
d
ib
ilitya
n
dR
e
p
u
ta
tio
n
a
lF
a
c
to
rsinaM
o
d
e
l
o
fE
x
c
h
a
n
g
eR
a
teC
ris
is
F
o
re
ig
np
ric
e
s
W
o
rldin
te
re
s
tra
te
s
E
x
te
rn
a
lS
h
o
c
k
s
O
ffic
ia
l re
s
e
rv
e
s
R
a
n
d
o
m
s
h
o
c
k
s
M
o
n
e
ys
u
p
p
ly
C
a
p
ita
l in
flo
w
s
O
u
tp
u
t
D
o
m
e
s
ticin
te
re
s
tra
te
s
L
o
s
sfu
n
c
tio
n
D
e
v
a
lu
e
D
on
o
td
e
v
a
lu
e
D
e
v
a
lu
a
tio
ns
iz
e
P
ro
b
a
b
ilityo
fa"w
e
a
k
"
p
o
lic
y
m
a
k
e
r
E
x
p
e
c
te
dd
e
v
a
lu
a
tio
nra
te
E
n
d
o
g
e
n
o
u
sv
a
ria
b
le
s
72
S
o
u
rc
e
:A
d
a
p
te
dfro
m
A
g
é
n
o
ra
n
dM
a
s
s
o
n(1
9
9
8
).

Change in official reserves:
R = (i – i* - a) + (e + p-1* - p-1) - u1, ,  > 0.
(26)

i*: foreign interest rate;
a: expected devaluation,
u: random shock,
p (p*): logarithm of domestic (foreign) prices.
Domestic interest rate is determined by the equilibrium
condition of the domestic money market:
i = 0 – h.
(27)
73

h: (logarithm of the) base money stock defined in
proportion of nominal output at the previous period
h = 1h-1 + R + u2,

0 < 1 < 1,  > 0,
(28)
u2: random term.
Normalizing constant terms to zero, (26), (27), and (28)
yield
i = -1{-1h-1 + (i* + a) - (e + p-1* - p-1) +
(29)
u},


where  = 1 + , and u  u1 - u2.
LF: value of loss function if exchange rate is kept fixed.
74
D
L : value of loss function when exchange rate is


Government devalues when LD – LF < 0.
From (29) if e = 0, i is at the level:
iF = -1{-1h-1 + (i* + a) - z-1 + u},

z = e + p* - p: competitiveness.
If e = e-1 + d (d is devaluation size), then i:
iD

~
–i=
(iF
~
- i) - -1d.
Domestic interest rates are lower relative to their
desired value when the authorities devalue.
75

Substitute out the previous expression in (25):
d
LD – LF =


d
~
F
- 2(i - i) + d.

It can be shown that LD – LF < 0 only when
 > ~  1h-1 +  - (i* + a) + z-1, (31)
~


where  = i + d/2 + 2/2.
Since  can take on two values, t (through ) depends
on policymaker's type.
Expected devaluation rate: product of the devaluation
probability  and devaluation size d.
76

Private sector's assessment of :
 = w + (1- )T,

w: probability that weak government will devalue;
T: probability that tough government will devalue.
Expected devaluation rate is
d  a = [w + (1 - )T]d.

(32)
(33)
From (31), h can be defined as follows, for h = w, T:
h
= Pr(u >
~h
u ).
(34)
77

If u is assumed to follow a uniform distribution in the
interval (-v, v), with 2v > d, then
h

= (v -
~h
u )/2v.
Using (31) to (34), we can solve for d:
a
d
=
1 - d/2v
(T - w)
1 [v - T -  h + i* - z ]
+
1
-1
2v
2v
where T > w.
78

Equation for expected devaluation rate to be estimated
is given by
a = a0 + a1 + a2i* + a3z-1 + a4h-1 + ,

where a1 > 0, a2 > 0, a3 < 0, a4 < 0, and  is error term.
Updating equation for the probability of a weak
government  is derived from Bayesian updating:
=
w
1 - -1
-1.
T )(1 -  )
(1 - w-1)-1 + (1 - -1
-1
(37)
79

Substitution of (34) for w and T in (37) and linearizing:
 = b1-1 + b2i-1* + b3z-2 + b4h-2 + ’,
where 0 < b1 < 1, b2 < 0, b3 > 0, b4 > 0, and ’ is error
term.
80
A “Cross-Generation”
Framework





Flood and Marion (1999): “cross-generation” framework
for the analysis of currency crises.
Key difference between “old” and “new” approaches:
 former assumes that commitment to fixed exchange
is state invariant,
 in the latter it is state dependent.
To link the two generations of models, make Rl in
conventional approach a function of variable that
captures state of the business cycle (e.g.
unemployment or inflation).
Implication of endogenizing Rl: policymaker may affect
shadow exchange rate through its choice of the level of
reserves that it wants to commit to defend the parity.
Potential profits to be realized by speculators remain the
82
driving force behind speculative attacks.




But state of the economy also influences the timing of
currency crises.
Cross-generation framework restricts large degree of
arbitrariness that characterizes the timing of speculative
attacks in second-generation models.
There are ranges in which multiple equilibria do occur,
but this happens only if fundamentals are sufficiently out
of line.
From the point of view of policymakers, this is more
sensible prediction than simply emphasizing the role of
“sunspots.”
83
Evidence on Exchange-Rate
Crises



The 1980s.
The 1994 Crisis of the Mexican Peso.
 Background: Structural Reform and the Solidarity
Pact.
 Policy Responses to Capital Inflows, 1989-1993.
 The Balance-of-Payments Crisis.
The Thailand’s Currency Crisis.
 Background.
 Triggering Events.
 Emergence and Management of the Crisis.
85
The 1980s
Argentina:
 In December 1978 authorities adopted a stabilization
program aimed at
 containing soaring inflation and
 enormous public sector deficit.
 Key aspect of the program: setting of the exchange rate,
which was adjusted based on a preannounced declining
rate of crawl.
 Following a period of relative success, bank failures in
early 1980 touched off a financial crisis.
 Rate of increase of credit to the financial sector
increased sharply in early 1980.
86
 This undermined confidence in exchange-rate regime.






Large and increasing reserve losses coincided with
increase in domestic credits.
Loss in confidence was reflected in a sharp increase in
interest rates on peso deposits relative to foreign rates.
Period prior to the collapse was marked by
 steep rise in the parallel market premium;
 very high inflation rates;
 sustained appreciation of real exchange rate.
Current account balance:
 surplus of $1.9 billion in 1978;
 deficits of $0.5 billion in 1979 and $4.8 billion in 1980.
Crawling peg policy was abandoned in June 1981.
Temporary adoption of dual exchange-rate regime
adapted.
87
Brazil:
 Cruzado crisis occurred in October 1986, eight months
after the Cruzado Plan was launched in February of that
year.
 Plan attempted to fix all prices, including nominal
exchange rate.
 Through 1986 domestic credit expanded by more than
40%.
 Net foreign reserves declined by $5.8 billion.
 Cruzado Plan was abandoned in October 1986 with
devaluation of the cruzado.
 Parallel market premium increased from 30% in March
1986 to more than 100% in the month preceding the
devaluation.
88
Chile:
 Chilean peso crisis occurred in June 1982 and was
precipitated by banking failures.
 Central bank responded with sharp increases in
domestic credit.
 Real exchange rate appreciated and foreign reserve
losses increased by the central bank for some months
prior to actual exchange-rate collapse.
 Eroding confidence was reflected in
 widening spread between spot and forward rates for
the Chilean peso and
 rising parallel market premium prior.
89
Mexico:
 Crisis that occurred in February 1982 was accompanied
by devaluation of 28% against the U.S. dollar.
 Turbulence in foreign exchange market was preceded
by
 increases in central bank credit creation;
 reserve losses for some months prior to the collapse.
 Quarterly reserve losses of the central bank were
estimated as 39 billion pesos, 44 billion pesos, and 140
billion pesos in the last three quarters of 1981.
 Percentage spread between spot and forward peso
rates widened during the final quarter of 1981.
 Acceleration in inflation rate led to
 appreciation of real exchange rate;
90
deterioration of the current account deficit, which
reached $16 billion in 1981, compared with $10.7
billion in 1980.
 On February 12, 1982, the authorities abandoned quasifixed exchange-rate system and allowed exchange rate
to float freely.
 Continuing capital outflows led to 67% depreciation of
the peso by the end of February 1982.
 In August 1982, with the central bank out of reserves,
dual-exchange-rate regime was put in place.
Similarities in processes leading to exchange rate
crises in developing countries:
 In the periods leading to crisis, domestic inflation is high
and international reserves fall at an increasing rate,
reflecting
91

over-expansionary credit policy,
 rising current account deficits,
 heightened perceptions of ultimate collapse of the
regime.
Anticipations of a crisis translate into
 forward premium or parallel market premium that
may rise to high levels in the periods preceding the
regime collapse;
 as a result, domestic interest rates rise substantially.
Real effects associated with balance-of-payments
crises:
 real exchange rate appreciates during the transition
period;
 behavior of real exchange rate and interest rates
92
affects domestic output.



The 1994 Crisis of the
Mexican Peso





When the international debt crisis struck, Mexico was
the largest developing country debtor.
This crisis triggered by Mexico's announcement of its
inability to service external debt in August of 1982.
Beginning of its end was the agreement of a Brady deal
between Mexico and its external bank creditors in 1989.
Mexico was accounted for 30% of total portfolio flows to
developing countries over 1989 to mid-1993.
These flows came to an end in December of 1994,
when Mexico experienced a severe financial crisis.
93
Background: Structural Reform and the
Solidarity Pact





Expansionary fiscal policies associated with the oil
bonanza after 1976 resulted in high inflation, capital
flight, and accumulation of external debt.
Mexico's announcement of its inability to service
external debt in 1982 triggered international debt crisis.
This ushered in three years of poor macroeconomic
performance, triple-digit inflation and negative growth.
Thorough-going program of reform and orthodox
stabilization was undertaken in 1985.
Structural reform was including fiscal adjustment,
privatization, trade and financial liberalization, and
reform of the foreign investment regime.
94




Key macroeconomic policy objectives:
 inflation stabilization;
 reactivation of growth based on improvements in
economic efficiency.
After an initial orthodox program of inflation stabilization
yielded disappointing results, Mexico switched to a
heterodox approach in 1987.
Key heterodox component: agreement among the
government, business, and labor to break inflation
inertia by setting three nominal anchors.
Government restricted increases in public-sector prices
and exchange rate, in exchange for restraint by workers
and managers in the setting of wages and prices.
95
Exchange-rate component resulted in a system of
preannounced daily mini-devaluations of an officially
determined exchange rate.
 Liberalization and institutional reform in the financial
sector were central components of structural reforms.
Financial liberalization:
 Quantitative limits on bankers' acceptances were
eliminated (November 1988).
 In April of 1989:
 controls on interest rates and maturities on traditional
bank instruments were abolished;
 non interest-bearing reserve requirements were
replaced by 30% liquidity ratio;
 restrictions on lending to private sector were
96
removed,

mandatory bank lending to public sector at
preferential interest rates was discontinued.
 In September of 1991: liquidity ratio was lowered on
deposits at the end of August, and eliminated for new
deposits.
 Constitutional amendment enacted in mid-1990 allowed
full private ownership of banks.
 Formation of financial holding companies: by July of
1992, 18 government-owned banks privatized.
Capital flow:
 Capital inflows on a broad scale in the second half of
1989.
 Overall capital account surplus exceeded 8% of GDP
over 1991-93.

97



Foreign direct investment led the surge in 1989 and
remained important throughout the inflow episode.
Portfolio flows increased rapidly and dwarfed FDI by
1991.
Important components of these short-term flows:
 certificates of deposits in privatized Mexican banks;
 short-term peso-denominated government bonds.
98
Policy Responses to Capital Inflows, 19891993.



Response was influenced by objective of sustaining
stabilization effort.
Government adhered to its exchange-rate-based
stabilization strategy,
 maintaining downward trajectory of nominal
depreciation;
 attempting to stem domestic price-level increases
through sterilized intervention in foreign exchange
market.
Figure 16.4: inflation fell steadily, but it continued to
exceed rate of depreciation of the peso, resulting in
mounting real appreciation.
99
F
i
g
u
r
e
1
6
.
4
a
M
e
x
i
c
o
:
M
a
c
r
o
e
c
o
n
o
m
i
c
I
n
d
i
c
a
t
o
r
s
,
1
9
8
8
9
4
1
2
F
i
s
c
a
l
b
a
l
a
n
c
e
(
i
n
%
o
f
G
D
P
)
1
0
0
N
e
t
c
r
e
d
i
t
(
a
n
n
u
a
l
g
r
o
w
t
h
r
a
t
e
s
)
P
r
i
m
a
r
y
b
a
l
a
n
c
e
T
o
p
r
i
v
a
t
e
s
e
c
t
o
r
8
4
5
0
0
4
0
O
v
e
r
a
l
l
b
a
l
a
n
c
e
8
T
o
p
u
b
l
i
c
s
e
c
t
o
r
1
2
1
9
8
9
1
9
9
1
9
3
q
1
9
4
q
1
5
0
1
9
8
9
1
9
9
1
9
3
q
1
9
4
q
1
S
o
u
r
c
e
:
I
n
t
e
r
n
a
t
i
o
n
a
l
M
o
n
e
t
a
r
y
F
u
n
d
,
O
E
C
D
,
a
n
d
A
g
é
n
o
r
a
n
d
M
a
s
s
o
n
(
1
9
9
7
)
.
100
F
i
g
u
r
e
1
6
.
4
b
M
e
x
i
c
o
:
M
a
c
r
o
e
c
o
n
o
m
i
c
I
n
d
i
c
a
t
o
r
s
,
1
9
8
8
9
4
6
2
4
0
1
2
4
C
u
r
r
e
n
t
a
c
c
o
u
n
t
1
1
2
0
(
i
n
b
i
l
l
i
o
n
s
o
f
U
S
d
o
l
l
a
r
s
,
l
e
f
t
s
c
a
l
e
) 2
6
C
u
r
r
e
n
c
y
r
i
s
k
(
l
e
f
t
s
c
a
l
e
)
2
/
5
2
1
0
2
0
0
0
9
1
8
0
2
8
4
1
6
0
7
4
3
C
o
u
n
t
r
y
r
i
s
k
R
e
a
l
e
f
f
e
c
t
i
v
e
e
x
c
h
a
n
g
e
r
a
t
e
1
/
6
(
r
i
g
h
t
s
c
a
l
e
)
1
4
0
r
i
g
h
t
s
c
a
l
e
)
2
/
6(
8
1
2
0
5
2
4
1
1
0
1
0
0
Ju
O
n
9
ct
F
3
e
9
Ju
b
3
9
O
n
4
9
ct
4
1
9
8
1
8
9
q
9
1
1
0
9
q
9
1
1
2
9
9
q
4
1
q
1
S
o
u
r
c
e
:
I
n
t
e
r
n
a
t
i
o
n
a
l
M
o
n
e
t
a
r
y
F
u
n
d
,
O
E
C
D
,
a
n
d
A
g
é
n
o
r
a
n
d
M
a
s
s
o
n
(
1
9
9
7
)
.
1
/
1
9
8
0
=
1
0
0
,
a
r
i
s
e
i
s
a
d
e
p
r
e
c
i
a
t
i
o
n
.
I
n
d
e
x
b
a
s
e
d
o
n
u
n
i
t
l
a
b
o
r
c
o
s
t
s
.
2
/
C
u
r
r
e
n
c
y
r
i
s
k
i
s
m
e
a
s
u
r
e
d
b
y
9
1
d
a
y
c
e
t
e
s
m
i
n
u
s
9
1
d
a
y
t
e
s
o
b
o
n
o
r
a
t
e
.
C
o
u
n
t
r
y
r
i
s
k
i
s
m
e
a
s
u
r
e
d
b
y
9
1
d
a
y
t
e
s
o
b
o
n
o
m
i
n
u
s
U
.
S
.
3
m
o
n
t
h
C
D
r
a
t
e
.
B
o
t
h
i
n
d
i
c
a
t
o
r
s
a
r
e
b
a
s
e
d
o
n
m
o
n
t
h
l
y
a
v
e
r
a
g
e
s
o
f
w
e
e
k
l
y
a
u
c
t
i
o
n
r
a
t
e
s
.
101
F
i
g
u
r
e
1
6
.
4
c
M
e
x
i
c
o
:
M
a
c
r
o
e
c
o
n
o
m
i
c
I
n
d
i
c
a
t
o
r
s
,
1
9
8
8
9
4
3
5
V
u
l
n
e
r
a
b
i
l
i
t
y
i
n
d
e
x
1
/
(
i
n
p
e
r
c
e
n
t
,
r
i
g
h
t
s
c
a
l
e
)
2
5
0
9
0
R
e
l
a
t
i
v
e
S
h
a
r
e
o
f
T
e
s
o
b
o
n
o
s
2
/
(
i
n
p
e
r
c
e
n
t
)
8
0
2
0
0
3
0
7
0
1
5
0
2
5
6
0
1
0
0
2
0
5
0
5
0
4
0
1
5
0
3
0
1
0
5
G
r
o
s
s
i
n
t
e
r
n
a
t
i
o
n
a
l
r
e
s
e
r
v
e
s
(
i
n
b
i
l
l
i
o
n
s
o
f
U
S
d
o
l
l
a
r
s
,
l
e
f
t
s
c
a
l
e
)
5
0
2
0
1
0
0
1
0
0
1
5
0
0
J
a
n
9
0
J
a
n
9
1
J
a
n
9
2
J
a
n
9
3
J
a
n
9
4
J
a
n
9
1 J
a
n
9
2 J
a
n
9
3 J
a
n
9
4
S
o
u
r
c
e
:
I
n
t
e
r
n
a
t
i
o
n
a
l
M
o
n
e
t
a
r
y
F
u
n
d
,
O
E
C
D
,
a
n
d
A
g
é
n
o
r
a
n
d
M
a
s
s
o
n
(
1
9
9
7
)
.
1
/
G
r
o
s
s
i
n
t
e
r
n
a
t
i
o
n
a
l
r
e
s
e
r
v
e
s
m
i
n
u
s
T
e
s
o
b
o
n
o
s
,
d
i
v
i
d
e
d
b
y
t
h
e
b
a
s
e
m
o
n
e
y
s
t
o
c
k
.
2
/
I
n
t
o
t
a
l
s
t
o
c
k
o
f
C
e
t
e
s
a
n
d
T
e
s
o
b
o
n
o
s
(
i
n
c
l
u
d
e
s
r
e
s
i
d
e
n
t
s
a
n
d
n
o
n
r
e
s
i
d
e
n
t
s
)
.
102





Exchange-rate band (end of 1991):
 floor for the value of the dollar was established at the
value reached on November 11;
 subsequent devaluations defined a ceiling.
At the time the band was introduced, daily depreciation
rate was lowered from 40 to 20 cents.
These measures
 lowered maximum rate of depreciation of exchange
rate;
 gave it more room to move in a downward direction.
New exchange-rate policy implied: more appreciated
and more variable nominal rate.
Inflation fell into the single-digit range by 1993.
103





But it continued to exceed rate of nominal depreciation,
and the peso continued to appreciate in real terms.
Figure 16.4: deterioration in the current account in spite
of steadily-improving public sector finances.
This deterioration arose from excess of private
investment over saving.
Mexico did not sustain acceleration in economic growth
when capital inflows surged.
After rising above 4% in 1990, growth slowed.
104
The Balance-of-Payments Crisis.
Three sets of factors:
 initial conditions at the beginning of 1994;
 slow growth,
 fragile financial system,
 persistently high current account deficit,
 reduction in the national saving rate,
 appreciated real exchange rate;
 external and domestic shocks that materialized during
the course of the year;
 nature of the policy responses to those shocks.
105
Initial conditions



Mexican peso appreciated strongly in real terms during
1988-93, and current account deficit was high.
Reforms instituted after 1985 did not result in a rapid
and sustained resumption of economic growth in the
short run, partly, due to tight monetary policy.
Fragile state of the financial system:
 newly privatized and liberalized financial system
experienced rapid expansion during 1991-94;
 loans grew at annual rate of 24% in excess of the
rate of growth of nominal GDP;
 broad money multiplier increased rapidly;
 quality of many loans was questionable;
106
18% of bank deposits were denominated in U.S.
dollars, making bank liabilities sensitive to exchangerate changes.
Dornbusch and Werner (1994) attributed Mexico's
growth performance to
 demand-reducing effects of real appreciation in the
home goods sector;
 high real interest rates.
High real interest rates and low growth caused
deterioration in firms' balance sheets, and therefore
those of banks.



107
Shocks





November of 1993: approval of NAFTA legislation by the
United States Congress.
Approval was not foregone conclusion, and Mexico lost
substantial foreign exchange reserves in October, as
external creditors hedged against a failure of ratification.
Mexico's accession to GATT, OECD, and NAFTA was
viewed as an institutional commitment that would tend
to be perceived by the private sector as “locking in”
Mexico's economic reforms.
Importance: convince potential investors that acquisition
of real capital in Mexico was a good bet.
Expectation by approval of NAFTA: previous reforms
would begin to bear fruit in the form of accelerated pace
108
of fixed investment.



But, unfavorable shocks followed:
 Chiapas uprising on January 1;
 in February, succession of interest rate increases
engineered by the Fed;
 assassination of ruling-party presidential candidate
on March 23.
Economic effects of these shocks:
 higher rates of return in the United States;
 increased political and economic uncertainty in
Mexico
increased the risk-adjusted rates on return in U.S.
relative to Mexican assets.
Portfolio adjustments resulted in both price and quantity
adjustments, taking the following forms:
109
Three-month CETES yield increased to 18% from
March until August; the yield subsided thereafter, but
remained at 15% until December.
 Foreign exchange reserves peaked at U.S. $29
billion during the first quarter of 1994 but it dropped
by U.S. $12 billion from February to April.
 Exchange rate, which were in the middle of its band
at the end of 1993, jumped to the top of the band at
the end of March and remained there.
Events of the first quarter caused creditors to fear both
 devaluation and
 default on its obligations by the Mexican government.
Evidence on the former: sharp increase in the spread
between peso-denominated CETES bonds and dollarindexed Mexican government bonds (tesobonos). 110




Evidence on the latter:
 uptick in yields on Mexican Brady bonds;
 yield spread between tesobonos and United States
Treasury bills of similar maturities.
111
Policy Response




These reserve losses can be interpreted as the prelude
to a full-blown speculative attack.
Possible responses: defend the peg with tighter credit or
to abandon it.
If Mexican and foreign-interest-bearing assets had been
perfect substitutes by private capital markets in 1994:
 tight credit would have reduced nominal interest
rates, due to elimination of devaluation expectations
and exchange rate risk;
 with sufficient inertia in inflation process, real
domestic interest rates would also have fallen.
With imperfect substitutability: tight credit would have
 raised real interest rates;
112
slowed growth of interest-sensitive components of
aggregate demand;
 caused contraction in supply;
 deepened the recession with negative effects on
investment, upward pressure on the price level and
stress on financial system.
 Abandoning the peg would have represented major
revision of anti-inflation strategy.
 These options were unpalatable especially in the
context of presidential election.
Government responded to the events in two ways:
 Monetary policy: hold to its exchange-rate path and
sterilize foreign-exchange outflows.
 Reaction function of monetary authorities was stable
113
over the course of 1994.

Sterilization: substantial increase in lending activities
by the development banks, financed by credit from
the central bank.
 Figure 16.4: substantial net credit expansion to the
public sector.
Debt management:
 Fiscal implications of exchange-risk premia were
avoided by replacing maturing CETES with shortterm dollar-indexed debt (tesobonos).
 Because the CETES were relatively short-term,
transformation in the structure of debt was rapid
(Figure 16.4).
 Increase of tesobonos in domestic debt from 5% at
the beginning of the year to 55% by its end.


114


Justification for this strategy: hope that post-NAFTA or
post-election investment boom and improvements in
productivity performance would
 generate resurgence of economic growth;
 validate real exchange rate;
 render the current account deficit sustainable.
Plausibility of this view was enhanced by
 strong growth of nonoil exports;
 large share of capital goods in Mexican imports.
115
The Crisis




Speculative pressures were building up in 1994 was
indicated by:
 behavior of capital flows;
 premia for exchange-rate and default risk.
Unsustainable current account made adjustment
inevitable.
Expenditure reduction and switching would be needed if
post-NAFTA growth boom did not materialize.
Key point: returns on assets invested in Mexico would
depend on what form that adjustment took.
116



Combination of
 fiscal and monetary contraction and
 nominal devaluation
would have been required to produce the requisite
adjustment.
Enormous fiscal adjustment already undertaken.
Thus, probability that adjustment would feature nominal
devaluation was raised by
 threat of financial crisis that would be posed by a
severe domestic recession,
 large cumulative real appreciation of peso.
117





Magnitude of devaluation would be large because:
 costs of devaluing in terms of foregone credibility for
the authorities were not very sensitive to the size of
the ultimate devaluation;
 perceived degree of overvaluation was large.
Critical issue related to timing of the crisis: change in
political administration following presidential elections.
Commitment of the new administration to disinflation
strategy was unknown.
Given this uncertainty, any signal that new
administration was likely to consider a revised
exchange-rate policy could trigger a speculative attack.
Several such signals were sent by both administrations
over the course of the year.
118
Evolution of both fiscal and exchange-rate policies
suggested that outgoing administration was likely to
consider an adjustment in nominal exchange rate.
 Substantial capital outflows did not trigger an
adjustment in the primary surplus.
 Exchange rate was allowed to depreciate to the
top of its band during the second quarter (nominal
depreciation of about 8%).
 New finance minister after the election was expected
to give greater weight to allocative role of exchange
rate, as opposed to its role as nominal anchor.
Result of these events was further reserve losses
between August and December.
By the beginning of December:
119
 reserves had fallen to $10 billion;



vulnerability index (ratio of net liquid foreign-currency
assets to monetary base) had fallen to the lowest
levels reached during the 1990s.
On December 20, upper level of the Mexican exchange
rate band was increased by 15%.
This was perceived by the markets as too little, too late.
Result:
 final speculative attack was triggered;
 after two days of rapid reserve losses, the peso was
forced to float.




120
Thailand’s Currency Crisis


Key ingredients in Mexican currency crisis:
 overvalued exchange rate implied the necessity of
adjustment;
 fragile financial system circumscribed the form that
adjustment could take.
But, these ingredients did not prove to be sui generis:
shortly after the Mexican crisis, a similar set of
circumstances produced similar results in Thailand.
121
Background





Capital inflows to Southeast Asia increased sharply in
the early 1990s, with short-term inflows playing larger
role.
Rising inflation and increasing current account deficits
increased concerns about macroeconomic overheating.
Domestic macroeconomic policy response consisted of
tight monetary policies.
Because countries in the region continued to pursue
nominal exchange-rate targets, tight money meant
sterilization of balance-of-payments surpluses.
Thailand: intensity of sterilization was increased in
1993, when it experienced an upsurge in private capital
inflows related to establishment of BIBF.
122





Implication of a policy mix relying on monetary policy to
restrain expansion of aggregate demand: intensification
of short-term inflows.
Demand for loans was sustained despite high domestic
real interest rates due to asset price inflation.
Legacy of this situation: financial system with borrowers
whose creditworthiness and value of whose collateral
was dependent on inflated asset values.
This made net worth of these institutions vulnerable to
downward correction of domestic asset prices.
Correction could come about in two ways:
 negative reassessment of the earning streams
associated with these assets;
 increase in discount rates applied to these earning
123
streams.






Inappropriate financial liberalization put strains on
domestic asset values and increased short-term
external liabilities.
In Thailand, in the absence of negative shocks, a crisis
would have been avoidable.
But, mistake in exchange-rate management policy
pushed vulnerability to the breaking point.
Maintenance of competitive real exchange rate was a
cornerstone of development strategy in Southeast Asia
since the mid-1980s.
In Thailand during 1994-95, domestic inflation rate was
higher than those of the country's trading partners.
Failing to offset this inflation differential by nominal
depreciation would have implied real exchange rate
appreciation and loss of competitiveness.
124
Triggering Events
Medium-Term Developments





Key development: loss of external competitiveness
during the first half of the 1990s.
Key factor: emergence of China as a major exporter of
labor-intensive manufactured goods.
To remain competitive, Thailand would have had to
export at lower prices.
Implication: long-run equilibrium real exchange rate
would depreciate.
This implies growing gap between actual and long-run
equilibrium real exchange rates given mild appreciation
of the actual bilateral real exchange rate.
125

Figure 16.5 illustrates this (Tanboon, 1998).
 Solid line shows Thailand's estimated equilibrium real
effective exchange rate: sharp depreciation due to
increased competition from China (Tanboon, 1998).
 Actual real effective exchange rate (REER) is
relatively stable, implying that by 1995, overvaluation
of the Thai baht was approximately 30%.
126
F
i
g
u
r
e
1
6
.
5
T
h
a
i
l
a
n
d
:
A
c
t
u
a
l
a
n
d
E
s
t
i
m
a
t
e
d
E
q
u
i
l
i
b
r
i
u
m
R
e
a
l
E
x
c
h
a
n
g
e
R
a
t
e
s
,
1
9
7
0
9
5
5
.
6
5
.
4
E
q
u
i
l
i
b
r
i
u
m
5
.
2
5
4
.
8
A
c
t
u
a
l
4
.
6
4
.
4
4
.
2
1
9
7
0
1
9
7
2
1
9
7
4
1
9
7
6
1
9
7
8
1
9
8
0
1
9
8
2
1
9
8
4
1
9
8
6
1
9
8
8
1
9
9
0
1
9
9
2
1
9
9
4
1
9
9
6
S
o
u
r
c
e
:
T
a
n
b
o
o
n
(
1
9
9
8
)
.
127
Short-Term Developments
Shock that preceded the Thai currency crisis: collapse
in export growth.
 Poor export performance materialized throughout the
Southeast Asian region in 1996.
Implications of poor export growth:
 Reduced GDP growth. This
 reduced income streams expected to be associated
with domestic assets,
 dampened domestic asset values.
 Introduced noise into medium-term competitiveness
calculations such that gap between actual and
equilibrium real exchange rates may have been larger.

128
Why did export growth slow?
 Loss of competitiveness due to
 inflationary effects of overheating;
 growth in Chinese export capacity.
 Appreciation of U.S. dollar against Japanese yen after
mid-1995 implied
 appreciation of actual REER in South East Asian
countries;
 widening of the gap between actual and equilibrium
real exchange rates.
 Collapse in semiconductor prices blamed by
 press on worldwide overcapacity;
 weak market for personal computers due to poor
growth performance in Japan and Western Europe.
129
Negative shocks during 1996 had two effects:
 Widened perceived gap between actual and equilibrium
values of the REER, both through
 appreciation of the actual REER; and
 depreciation of the perceived equilibrium REER;
 increased the vulnerability of financial sectors by
 depressing asset values; and
 weakening balance sheets of financial institutions.
Factors for inevitable nominal exchange-rate
adjustments:
 Implication of misalignments: nominal exchange rate
adjustments might be forthcoming.
 Implication of fragility in financial sector: costs of
resisting this misalignment were prohibitively high. 130
Emergence and Management of the Crisis





It was prudent for agents to hedge against possibility of
devaluation by moving assets out of baht.
Since this fact would magnify exchange rate-movement,
other agents would have opportunity to make money by
borrowing baht.
Whether such transactions would be profitable ex ante
depended on markets' expectations of authorities'
resolve in maintaining the value of the currency.
Vulnerability of domestic financial system and domestic
economic slowdown play key roles as in Mexico.
It would have been perceived that fighting off
speculative attack through traditional methods would
have been too costly by the authorities.
131





Reason: high interest rates would have impaired both
balance sheets and cash flows of domestic financial
institutions.
By mid-1996, the region's export problems were
beginning to receive widespread attention.
First hints of currency problems in Thailand emerged
during late July and early August 1996, triggered by
worries over export competitiveness in the region.
Following a bleak report on economic prospects, the
central bank was forced to spend U.S. $1 billion to
support the baht.
In September Moody's downgraded Thailand's shortterm foreign debt, noting financial-sector problems and
rapid accumulation of foreign debt during 1995.
132



Crisis in Thailand took almost a year before it
culminated in abandonment of exchange-rate parity.
As growth slowed and domestic interest rates were
maintained at high levels to defend the currency, stock
market fell.
Government made key mistakes in both areas of
vulnerability: exchange-rate policies and financial sector
policies.
133
Exchange Rate Policy





Government's biggest mistake: hold nominal value of
the baht for almost a year.
Devaluation expectations resulted in
 large capital outflows; and
 very high domestic interest rates.
Through their effects on government's liquidity position
and balance sheet of financial system, they magnified
uncertainty and instability.
Despite the pressure on the currency, Bank of Thailand
reported foreign exchange reserves of U.S. $38 billion,
unchanged from the July 1996 level.
By the time the baht was floated in July 1997, reserves
had officially fallen only to U.S. $33 billion.
134




But, reserves were maintained through large swap
transactions, leaving the Bank with a stock of future
dollar liabilities in excess of U.S. $ 23 billion.
Loss of liquidity created uncertainty in the market after
the baht was floated as to whether
 country would be able to meet its large short-term
external obligations;
 this perceived vulnerability to liquidity crisis
undermined the value of the currency after the float.
Government increased the cost of keeping the value of
the baht fixed almost a year by sending signals that its
commitment to the exchange-rate peg was not firm.
Example: in January 1997, government announced its
intention to reevaluate exchange rate regime when the
economy regained strength.
135
Financial Sector Policy





Second key mistake: postpone resolution of problems of
financial system.
Government initially denied vulnerability of country's 15
commercial banks and 90 finance companies.
High domestic interest rates continued to undermine the
value of assets held by financial system.
Effect was to jeopardize solvency of finance houses and
banks, which added to
 stock of nonperforming assets in financial sector;
 cost of resolving the sector's difficulties.
Increase in this cost impaired government's fiscal
position.
136



This unresolved liability overhang was a second source
of uncertainty which
 increased instability after the currency was allowed to
float;
 magnified fiscal burden associated with the crisis.
As the end of 1996 approached, financial-sector share
prices reflected extreme lack of confidence.
Government's initial policy response was inadequate,
consisting of a regulation imposed in December 1996
requiring
 banks to disclose their bad loans;
 adoption in January 1997 of series of measures
seeking to artificially prop up property values rather
than
137
recognizing and allocating losses of bank and
finance company capital;
 recapitalizing financial sector under more stringent
supervisory and regulatory regime.
In January 1997, the Cabinet announced the creation of
government fund to purchase a portion of their bad
property debt from finance companies.
Measure backfired because
 it signaled the government's intention to bail out
some of these companies,
 it revealed policy disarray within the government.
By late February 1997, depositors engaged in a run on
Thai finance companies, transferring deposits to safer
banks and moving money abroad.




138






Thus, renewed pressure on the baht emerged in
February.
Bank of Thailand responded by tightening monetary
policy.
High interest rates resulted in a succession of financial
crises triggered by falling property values.
Central bank extended credits to these institutions
(estimated to total U.S. $15.7 billion).
But, finance company liquidity crisis culminated in the
failure of country's largest finance company.
After this failure:
 stringent provisioning requirements were imposed on
all financial institutions;
 10 small finance companies were required to
139
increase their capital;
but, public disclosure of bad loan portfolio among
finance companies was not required.
Further measures were adopted in May:
 Capital controls were imposed in the form of
instructions to local banks not to sell baht to
foreigners in the swap market.
 Expenditure cuts were announced to offset shortfalls
in tax revenue due to the economic slowdown.
 Bank-financed fund was set up to buttress stock
market.
Government began to signal a tougher stance toward
financial sector in late June:
 Merger laws to accommodate the takeover of
companies in difficulties were eased.



140
16 finance companies were ordered to suspend
operations and seek merger partners within 30 days.
 Troubled finance companies that failed to merge
would be allowed to go under.
But, these measures proved to be too little too late.
The baht was floated on July 2.
In a month, it depreciated by over 25%.
Despite the float, problems were not over:
 Shortage of liquidity and nonperforming loans in
financial sector left significant uncertainty;
 authorities intervened to avoid excessive
depreciation of the baht;
 central bank raised its discount rate rather than
moving toward lower domestic interest rates.
141






Thailand began negotiations with IMF on July 28, and
program was agreed one week later.
142