Economics 315 International Macroeconomics

Lecture 9: Currency Crises

Dr. Keyu Jin
London School of Economics
December 4, 2013
Balance of Payments Crises
Many xed exchange rate regimes have collapsed
Britain and US forced off gold standard during First World War
1946-71 IMF system of xed but adjustable exchange rates
Mexico and Argentina during 1970s and early 1980s
european Monetary system in 1992, Mexico 1994
Asian crisis of 1997
What are the causes and the timing of currency crises?
First generation models: macroeconomic mismanagement
as the primary cause
To explain exchange rate crises in developing countries in the
1970s and 1980s
These crises were preceded by unsustainably large
government scal decits, nanced by excessive domestic
credt creation that eventually exhausted central banks foreign
reserves
Thus, the size of a countrys nancial liabilities (scal decit,
short-term debt, current account decit) and/or sustained real
appreciation from domestic price-level ination should signal
an increasing likelihood of a crisis.
More recently, european Monetary System crisis of 1992 and
Asian crisis of 1997 did not have macroeconomic
mismanagement
Crises independent of macro fundamentals
Second generation models: self-fullling crises
governments explicitly balances the costs of defending the
exchange rate against the benets of realignment
Multiple equilibria: costs of exchange rate defense depend on
the publics expectations.
First Generation Currency Crises
Government pursues scal policies incompatible with long-run
mainteance of the peg
Government faces short-term domestic nancing constraints that it
feels more important to satisfy than long-run maintenance of
external balance.
Since it cannot borrow from any creditor, it turns to CB and hands
over bonds in exchange for cash to fund government decit. CBs
reserves must be decreasing in equal amounts.
Speculators observe the decline of the central banks international
reserves and time a speculative attack in which they require the
remaining reserves in an instant
Faced with a loss of all its foreign exchange reserves, the central
bank is forced to abandon the peg and to move to a free oat.
The speculative attack on the central bank during the nal moments
of the peg is called a balance-of-payments, or foreign exchange,
crisis.
First model attributed to Krugman (1986)
The model
Money demand equation: (we eliminate y)
M
d
t
P
t
= i
t
where M
d
and P, i are levels of money, prices and the
nominal interest rate.
Money supply follows from Central Bank balance sheet:
M
s
t
= FX
t
+ DC
t
UIP condition that determines investors behavior
e
e
t +1
e
t
e
t
= i
t
i

t
where e denotes level of exchange rate.
Assumptions:
(a) PPP where we normalize P

to 1.
e
t
P

t
P
t
= 1 with P

t
= 1 e
t
= P
t
(b) We assume that there is perfect foresight and we
normalize the foreign interest rate to 0.
e
e
t +1
= e
t +1
with i

t
= 0
e
e
t +1
e
t
e
t
= i
t
(c) There is a lower bound on the level of foreign
reserves that the central bank owns.
FX
t
0
Solution of the model:
Money market equilibrium
M
d
t
= M
s
t

FX
t
+ DC
t
P
t
= i
t
+ UIP
FX
t
+ DC
t
e
t
=

e
e
t +1
e
t
e
t

Fixed exchange Rate Regime:

e
t
=

e t
Which implies that the money demand is given by
M
d
t
=

e
and the money market equilibrium,
FX
t
+ DC
t

e
= (1)
Sustainability of the regime
If
FX
t
>

e DC
t
the peg is unsustainable because at the given exchange rate

e,
money supply is larger than money demand (a devaluation could
restore equilibrium by increasing money demand).
example: Suppose that the sterling is pegged to the dollar at a given
rate

e and the Bank of england ran out of dollar reserves. If at this
rate there is still demand for dollars, the only way the equilibrium
can be restored is if the dollar becomes more expensive, i.e. if there
is a devaluation of the exchange rate (

e increases).
Moreover, equation (1) implies that, in a xed exchange rate regime,
changes in domestic credit have to be accompanied by changes in
reserve levels:
FX
t
= DC
t
Lets dene the shadow exchange rate as the exchange rate
that would prevail in the market if there were no intervention in
the foreign exchange market. That is, the exchange rate that
would prevail if there was a oating exchange Rate Regime
(reserves are zero )
FX + DC
t
= e
s
t

e
s
t +1
e
s
t

(2)
where e
s
t
denotes the shadow exchange rate at t .
An unsustainable peg
Consider the situation in which the Central Bank expands the
domestic component of money supply at a constant rate:
DC
t
= DC
t 1
+
where is the change in domestic credit.
Given that in a xed exchange rate regime (see equation (1))
changes in domestic credit have to be accompanied by
changes in reserve levels, we have that:
FX
t
= DC
t
=
In order to defend the peg, the central bank will intervene in
the foreign market by selling foreign reserves at the same rate
as the increase in the domestic credit component of the
money supply. The monetary authority will eventually run out
of foreign reserves and will be forced to abandon the peg.
The timing of the crisis:
Since we are in a framework in which everything is known in
advance, traders in the foreign market will anticipate the
abandonment of the peg and at a certain point will start selling
the domestic currency so that reserves will be driven to zero
abruptly.
Coming back to our example: The Bank of england is pegging
the pound to the dollar and we know that at some point the
peg will be abandoned and the pound will devalue against the
dollar. If speculators know that in advance, they will buy
dollars and sell pounds even before the central bank runs out
of reserves.
When do the speculators sell the currency?
From the equation that gives the shadow exchange rate (eq.2),
taking differences over a period:
DC
t
= e
s
t

e
s
t +1
e
s
t

(3)
and maintaining the assumption of constant growth rate in
domestic credit, and conjecting that e
s
t
= e
s
t +1
, we have:
e
s
t
= / (4)
The shadow exchange rate depends on the path of money supply
and will depreciate also at a constant rate proportional to .
The attack on the domestic currency will occur at time T at which
the shadow exchange rate is equal to the xed rate.
Timing of attack
Attack must occur at T when e
s
=

e
Argument:
Suppose that the attack will occur at time T
2
< T then the
exchange rate will appreciate discretely but this cannot be an
equilibrium since people do not want to sell a currency that
will appreciate.
Suppose the attack occurs at time T
1
> T then the exchange
rate would jump from the xed value and it would depreciate
discretely. Traders that hold the currency will incur in a capital
loss and since they know everything in advance they will sell
the currency before T
1
.
The attack occurs at time T.
Reserves will not deplete smoothly. Any individual trader has the
incentive to exchange domestic currency for foreign currency
before reserves run out.
Remember that events are perfectly anticipated. They know that
the exchange rate will depreciate when the peg breaks down, and
to avoid realizing losses on domestic currency assets, agents
attempt to convert the soon-overvalued domestic currency into
foreign currency before time of abandonment. This sudden rush
into long positions in the foreign currency will cause an immediate
exhaustion of available reserves.
Money market after the attack
At the moment of the attack the money supply will fall (given
the loss in reserves).
Money demand will contained by higher interest rates.
i
t
= i

t
+e
t +1
= i

t
+ /
That is, after the attack the nominal exchange rate is given by
the shadow exchange rate and it depreciates at a rate
proportional to , which implies that the domestic interest rate
is higher than the foreign one in order to preserve the UIP
condition.
Caveats of this model of currency crisis:
1 The root cause of the crisis is poor government policy. The
source of the upward trend in the shadow exchange rate is
given by the increase in domestic credit (the need for this
might arise for example because of scal decits to be
nanced by seigniorage) solve scal problem and there is
no crisis. Speculative target is provided by governments
pursuit of inconsistent policies: eg. persistent decits together
with an exchange rate peg.
2 Model shows that speculative attacks can be rational
outcomes
3 Weaknesses:
Crisis is perfectly predictable
Private sector is perfectly rational whereas monetary and
scal authorities are not
First generation currency crisis model seems to do no harm -
no effect on output
Is this a good model for describing ERM crisis in 1992 ?
There was no evidence of irresponsible policies in any of the
country involved.
All countries had enough foreign exchange reserves and gold to buy
back at least 80 to 90 percent of their monetary bases
There was no obvious trend in long-run equilibrium exchange rate
(shadow rate was not depreciating)
Lastly, if governments had resources to ght off speculative attack,
why didnt they do it? Need a model in which defense of currency is
costly.
Forgoing the xed exchange rate regime may be a policy
choice (and might not be inevitable)
example: British ofcial chose not to pay the price for
defending the pound with higher interest rates, while French
ofcials made the opposite decisions.
Second generation currency crisis model
This class of model is characterized by multiple equilibria: if
agents expect a crisis to happen it might be too costly to
maintain the peg, if agents are condent the crisis might be
avoided.
Defending a currency peg can have costs and benets:
If the currency is pegged at an uncomfortable level
(overvalued), the government is forced to accept a lower level
employment in the short-run than it would otherwise have
wanted.
This cost may be higher if the peg is not credible. In this case,
investors will demand higher interest rates in order to hold
assets denominated in the countrys currency. If the
government defends the peg by providing those higher interest
rates, it will worsen employment.
The benets from pegging the exchange rate may be:
maintaining a nominal anchor (ie, credible monetary policy),
political, or other economic goals.
Suppose that, as long as the peg is credible, the cost of
abandoning the peg is higher than the benets. But if the peg
is no longer credible, this cost outweighs the benets. So
even a government that would be willing to pay the price of
sustaining the peg in absence of speculative attack, it might
be unwilling to stand up in such situation. Speculators who
believe that other speculators are about to attack are
themselves encouraged to do so. (self-fullling crises of
condence).
Policy trade-offs
Dene the desired exchange rate e

as the exchange rate that

the government would choose if it had not made a
commitment to the xed rate.
Assume that the exchange rate peg will be more costly to
defend when devaluation is expected.
Assume that there is a cost that arises once the government
decides to abandon the xed exchange rate regime.
These 3 elements are embedded in the following loss function
for the government:
L = [(e

e) + e
e
]
2
loss from defending
+ I (e)
loss from abandoning
,
with , > 0. And we assume that the loss from abandoning
is given by the following function
I (e) =
0 for e = 0
Q for e > 0
Suppose now that we start from a situation in which the
exchange rate is pegged at

e. Different equilibria depend on
expectations.
Case 1) The government is expected to resist the pressure to
devalue, that is e
e
= 0. The loss from defending is
given by
L
d
d
= [(e

e)]
2
loss from defending
When this is an equilibrium?
- when, conditional on market expectations, the
government nds optimal to defend, that is when
L
d
d
< Q
Case 2) The government is expected to abandon and to
depreciate the exchange rate at the desired level e

,
that is e
e
= e

e. The loss from defending is
given by
L
a
d
= [(e

e) + (e

e)]
2
loss from defending
When this is an equilibrium?
- when, conditional on market expectations, the
government nds optimal to defend, that is when
L
a
d
< Q.
equilibrium is a situation in which the government choice is
compatible with market expectations. There is a region in
which both market expectations are validated
L
d
d
< Q < L
a
d
In this region we have multiple equilibria.
The implications of this model are that for a given shape of the
loss function, and for a given desired exchange rate e

, it
might become impossible for a country to defend an exchange
rate regime since the outcome depends entirely on whether
the market expects the government to devalue of not. A
speculative attack is self-fullling: it succeeds simply because
it was expected to occur without any reference to
fundamentals.
Analysis of the model:
1 the smaller the gap between e

and

e the easier it is to
defend;
2 the higher is Q, the cost of abandoning the peg, the easier it
is to defend, all else equal.
3 results depend on the fact that the government locked in an
exchange rate peg which is not optimal (i.e. it does not
correspond to the desired one).
Difference with rst generation models:
1 no irresponsible policy; (but still not full commitment to the
peg)
2 no predictability of the time of crisis.
3 if the country leaves the peg, there is no negative impact on
employment and output. The monetary policy constraint is
removed and the result is positive in terms of short-run
macroeconomics benets (think about Britain after 1992 and
Brazil very recent experience).