326 H.-L. Han International Review of Economics and Finance 9 2000 323–350
the scope for international policy cooperation, and the endogeneity of factor mobility or other structural characteristics Edison Melvin, 1990. By the mid-1980s, the
policy-making community actively discussed proposals for replacing the system of flexible exchange rates among major currencies with a system of target zones Krug-
man, 1991. A “target zone” is an arrangement with wide margins around an adjustable set of exchange rates. In early 1990s, the theory of optimal currency areas was to
consider various practical issues associated with the move toward monetary integration in Europe Bayoumi, 1994.
More recently, the study on exchange rate regimes is linked with financial fragility and macroeconomic policies. Chang and Velasco 1997 found that a currency board
cannot implement a socially optimal allocation. In addition, bank runs are possible under a currency board. They also found that a fixed exchange rate system may
implement a social optimum but is more prone to bank runs and exchange rate crises than a currency board. However, they found that a flexible exchange rate system
implements the social optimum and eliminates runs, provided the exchange rate and central bank lending policies are appropriately designed.
The focus of this paper is to search for explanations for the 1997 Asian foreign exchange market crises. Therefore, only the currency basket arrangement that has been
adopted by most of the Southeast Asian countries is studied here. One explanation is that the choices of weights to determine basket composites of these countries may
not be adequate to deter speculators from attacking these Southeast Asian currencies. For example, the top priority of the central bank of Thailand for the period of 1985
to 1997 has been to hold the baht stable against a basket of currencies dominated by the U.S. dollar. The policy helped to cut inflationary expectations in Thailand and
made it one of the world’s fastest-growing economies, but also created a huge current account deficit. The inter-relationship between currency basket weights, overall trade
balances, and price level will be analyzed in our model.
3. The basic model
The model is set up to show that a combination of optimal fiscal policy and currency basket weights can reach two policy goals at the same time e.g., keep the overall
balance of trade as well as price level unchanged. The analysis is based on Branson and Katseli 1981, and can be extended to include other economic policies such as
monetary policy and trade tariff policy.
Assume that the home country produces two goods: exports goods and nontraded goods. Exports are not consumed by the home country residents and nontraded goods
are consumed only by the home country residents. The real exchange rate stabilization is also assumed in the model as in Branson and Katseli 1981. Notation and definition
of the symbols used in the model are summarized in Table 1.
3.1. The export market The home country export supply depends on the relative home-currency price of
exports to nontraded goods p
x
p , as well as on the price elasticity of export supply.
H.-L. Han International Review of Economics and Finance 9 2000 323–350 327
Table 1 Notation and definition of symbols
T
i
5 units of home currency per unit of i currency, i50,..., N; N 5 numeraire country.
J
i
5 units of numeraire per unit of i currency.
r 5 units of home currency per unit of numeraire.
q
i
5 foreign country’s price of non-traded goods.
q
xi
q
mi
5 foreign country’s price of exports imports. p
5 home country’s price of non-traded goods.
Z ˆ ; dZZ, for any variable Z.
pˆ
x
pˆ
m
5 home country prices of exports imports. XM 5
export imports quantities of home country. O 5
quantities of non-traded goods. a
i
b
i
5 home country export import shares to from country i. d
x
s
x
5 price-elasticities of export demand supply in home country. d
m
s
m
5 price-elasticities of import demand supply in home country. d
m
d 5 semi-elasticity of import non-traded goods demand with respect to interest rate in home country.
φ 5
semi-elasticity of money demand with respect to interest rate in home country. ε
m
5 income-elasticity of imports in home country.
ε 5
income-elasticity of non-traded goods in home country. u 5
share of non-traded goods spending to total spending in home country. BT 5
balance of trade on goods and services in home country’s currency. MY 5
quantity of money. R 5
domestic nominal interest rate. y
5 real income of home country.
The export supply function can be written as shown in Eq. 1, lnX 5 s
x
lnp
x
2 lnp
, 1
which implies that the higher relative price of exports to nontraded goods, the more exports will be supplied.
Demand for the home country exports by country i, given country i’s currency price of the home country exports and country i’s currency price of its own nontraded
goods, is specified as shown in Eq. 2: lnX
i
5 2d
x
lnq
xi
2 lnq
i
. 2
That is, the home country export demand by country i is a function of the relative price of the home country exports country i’s imports to country i’s nontraded goods
q
xi
q
i
, both measured in country i’s currency. The price elasticity of export demand in all countries is assumed equal d
x
to for simplicity. Goods arbitrage ensures the equality of the home country currency price of exports p
x
and country i’s currency price of exports T
i
q
xi
, as shown in Eq. 3, p
x
5 T
i
q
xi
, 3
where T
i
; J
i
r. The equilibrium condition for the export market is when the export supply from
the home country equals the aggregate demand for exports from the rest of the world.
328 H.-L. Han International Review of Economics and Finance 9 2000 323–350
Total differentiation of the condition shows that percentage change in export price pˆ
x
is a function of percentage change in the price of domestic nontraded goods pˆ ,
percentage change in the home country’s real exchange rate against the numeraire, rˆ 1 qˆ
N
2 pˆ , as well as percentage change in the numeraire’s real exchange rates
against third countries, Jˆ
i
1 qˆ
i
2 qˆ
N
, as shown in Eq. 4, pˆ
x
5 pˆ 1 k
rˆ 1 qˆ
N
2 pˆ 1 k
o
a
i
Jˆ
i
1 qˆ
i
2 qˆ
N
, 4
where k ; d
x
d
x
1 s
x
. As mentioned in Branson and Katseli 1981, k represents market power on the
export the home country side. A small country assumption means k 5 1. By substitut- ing condition 4 into the export market 1, Eq. 5 can be shown:
X ˆ 5 s
x
k [rˆ 1 qˆ
N
2 pˆ 1
o
a
i
Jˆ
i
1 qˆ
i
2 qˆ
N
]. 5
Percentage change in exports is a function of percentage change in the home country’s real exchange rate, percentage change in foreign real exchange rates, and the parame-
ters of k and s
x
. 3.2. The import market
In the import market, assume that the home country government spends aG on imports, where G indicates the government spending and a indicates the portion of
government spending on imports. The supply of imports by country i is a function of the relative i-currency price of imports i.e., country i’s exports to country i’s nontraded
goods. It is similar to Eq. 1 and it can be expressed as shown in Eq. 6:
lnM
i
5 s
m
lnq
mi
2 lnq
i
. 6
The aggregate demand for imports by the home country is dependent on the relative home currency price of imports to nontraded goods, the real income level, the nominal
interest rate, and the government spending in the home country, as shown in Eq. 7,
lnM 5 2d
m
lnp
m
2 lnp
1 ε
m
lny 2 d
m
R 1 g
m
lnG, 7
where g
m
5 aG M indicates the government spending share of imports market. Equat-
ing the demand and supply of imports, it is shown that the corresponding change in the import price is a function of percentage change in the price of domestic nontraded
goods, percentage change in the home country’s real exchange rate, percentage change in foreign real exchange rates, percentage change in the real income, change in the
nominal interest rate, government spending, and other elasticity parameters. Eq. 8 is given as follows:
pˆ
m
5 pˆ 1 k9
rˆ 1 qˆ
N
2 pˆ 1 k9
o
b
i
Jˆ
i
1 qˆ
i
2 qˆ
N
1 ε
9
m
yˆ 2 d9
m
dR 1 g9
m
G ˆ ,
8 where k9
; s
m
d
m
1 s
m
, d9
m
; d
m
d
m
1 s
m
, ε
9
m
; ε
m
d
m
1 s
m
, and g9
m
; g
m
d
m
1 s
m
. Substituting condition [Eq. 8] into the import market [Eq. 6], it can be shown that
percentage change in imports M ˆ is a function of percentage change in domestic and
foreign real exchange rate and other elasticity parameters.
H.-L. Han International Review of Economics and Finance 9 2000 323–350 329
3.3. The nontraded goods market Demand and supply of nontraded goods for the home country are specified in an
analogous way. Percentage change in the price of nontraded goods can be expressed as shown in Eq. 9,
pˆ 5 k
pˆ
x
1 1 2 k
pˆ
m
1 ε
9 yˆ 2 d9
dR 1 g9 G
ˆ , 9
where k ; s
s 1 d
, ε
9 ;
ε s
1 d , d9
; d s
1 d , g9
; g s
1 d , and g
; 1 2 aGO is the government spending share of nontraded goods. Eq. 9 means
that percentage change in the price of nontraded goods is dependent on percentage change in the price of exports and imports, percentage change in the real income,
change in the nominal interest rate, and percentage change in the government spending. It can also be shown that percentage change in nontraded goods O
ˆ depends on the same factors that affect percentage change in the price of nontraded goods.
3.4. The money market Demand for real money balances is a function of the real income and the nominal
interest rate, as shown in Eq. 10, lnMY 2 lnp
I
5 lny 2
φ R
, 10
where p
I
is the consumer price index CPI and is defined as CPI 5 p
I
5 p
u
p
1
2u
m
. It is assumed that the interest rate R can be chosen by the central bank to certain
extent. For example, the Bank of Thailand BOT successfully adjusted the interest rate ceiling and monitored credits extended to the private sectors in early 1980s. From
1985 to 1985, the BOT also played a leading role in bringing down domestic interest rates by reducing interest rate ceilings on several occasions. Let us focus on the impacts
of fiscal policy, and assume that there is no tariff and no monetary policy M
ˆ Y 5 0. The real income of the home country is defined as the nominal income divided by
the aggregate price level, as given by Eq. 11: y 5
p O 1 p
x
X p
u
p
1
2u
m
. 11
Differentiating both sides of Eq. 11 gives Eq. 12: yˆ 5 a
pˆ 1 O
ˆ 1 bpˆ
x
1 X ˆ 2 upˆ
2 1 2 upˆ
m
, 12
where a 5 p O
p O 1 p
x
X , b 5 p
x
X p
O 1 p
x
X , and a 1 b 5 1. Substituting all
percentage changes of nontraded goods, price of nontraded goods, exports, price of exports, and price of imports pˆ
,O ˆ ,pˆ
x
,X ˆ ,pˆ
m
into Eq. 12, it can be shown that, according to Eq. 13:
yˆ 5 a1 1 s
2 bs
x
pˆ 1
b1 1 s
x
2 as pˆ
x
2 u pˆ 2
1 2 upˆ
m
5 Apˆ 1 Bpˆ
x
2 upˆ 2
1 2 upˆ
m
, 13
where A 5 a1 1 s 2 bs
x
, B 5 b 1 1 s
x
2 as , A 1 B 5 1 and A . 0, B . 0. Eq.
13 implies that percentage change in the real income is a function of percentage
330 H.-L. Han International Review of Economics and Finance 9 2000 323–350
changes in all prices, including nontraded goods, exports, and imports. Then, substitut- ing percentage change in the real income yˆ into Eq. 10 and letting M
ˆ Y 5 0, it shows that change in the nominal interest rate is as shown in Eq. 14:
dR 5 1
φ Apˆ
1 Bpˆ
x
. 14
When the prices of nontraded goods and exports increase, the nominal interest rate increases.
Next is to solve the simultaneous equations of 4, 8 and 9 for pˆ
x
,pˆ
m
,pˆ and then
for X ˆ ,M
ˆ ,Oˆ. These results will further be utilized to determine the optimal weights for different policy goals. Details are shown in mathematical appendix.
3.5. Policy goals 3.5.1. Stabilizing the balance of trade
The balance of trade on goods and services in the home country currency is defined as BT 5 p
x
X 2 p
m
M. Differentiating BT and choosing p
x
5 p
m
5 1 initially result in
Eq. 15: dBT 5
pˆ
x
1 X ˆ X 2 pˆ
m
1 M ˆ M.
15 Substituting the results from previous section on pˆ
x
,pˆ
m
,X ˆ ,M
ˆ and assuming that X 5 M at the beginning leads to Eq. 16,
dBT 5 G
1
pˆ 1 G
2
pˆ
x
1 G
3
pˆ
m
2 g
m
G ˆ ,
16 where G
1
5 2 s
x
1 d
m
2 ε
m
2 d
m
φ A 1
ε
m
u , G
2
5 1 1 s
x
2 ε
m
2 d
m
φ B, and
G
3
5 2 1 2 d
m
1 ε
m
1 2 u. Substituting pˆ ,pˆ
x
,pˆ
m
see the appendix into Eq. 16, we get Eq. 17,
dBT 5 H
1
rˆ 1 qˆ
N
2 pˆ 1
o
H
2i
Jˆ
i
1 qˆ
i
2 qˆ
N
1 H
3
G ˆ ,
17 where H
1
5 G
1
E
1
1 G
2
P
1
1 G
3
L
1
, H
2i
5 G
1
E
2i
1 G
2
P
2i
1 G
3
L
2i
, H
3
5 G
1
E
3
1 G
2
P
3
1 G
3
L
3
2 g
m
. The detailed definition of E
1
, E
2i
, E
3
, P
1
, P
2i
, P
3
, L
1
, L
2i
, L
3
, can be found in the mathematical appendix.
If s
x
1 d
m
1 ε
m
2 d
m
φ A
ε
m
, u , ε
m
2 1 2 d
m
ε
m
it can be shown that G
1
, G
2
, G
3
are positive, therefore H
1
and H
2i
are positive. But the sign for H
3
is uncertain and depends on the government spending share of imports g
m
. To stabilize the trade balance and let dBT 5 0, the home country’s real exchange rate against the numeraire must equal, as shown in Eq. 18:
rˆ 1 qˆ
N
2 pˆ 5 2
o
w
i
Jˆ
i
1 qˆ
i
2 qˆ
N
2 Z
1
G ˆ ,
18 where the weight is defined as H
2
1 1
H
2i
and is the weight in a currency basket if the balance of trade is kept unchanged, and Z
1
, defined as H
2
1 1
H
3
, is the reaction of the home country currency toward the change in government purchases. A decrease in
H.-L. Han International Review of Economics and Finance 9 2000 323–350 331
Jˆ
i
1 qˆ
i
2 qˆ
N
i.e., the depreciation of ith currency, will deteriorate the trade balance by H
2i
. In order to keep the balance of trade unchanged and to offset the impact from the depreciation of ith currency, the home country currency against numeraire rˆ 1
qˆ
N
2 pˆ should increase depreciate according to w
i
. Also, when government spends more on imports, the trade balance will deteriorate and the home currency should
depreciate according to Z
1
. Branson and Katseli 1981 show that the weights to be chosen in a currency basket
in order to stabilize real exchange rate, without including any policy variable to reach any policy goal, is rˆ 1 qˆ
N
2 pˆ 5 2
o
w
BK i
Jˆ
i
1 qˆ
i
2 qˆ
N
Stabilization of real ex- change rate is also assumed in our model. However, in contrast to Branson and
Katseli’s 1981 model, we include the policy variable G in Eqs. 7, 8, 9, and consider a nontraded market affected by government purchases G. Eq. 18 is derived
following these assumptions. Eq. 18 indicates that government purchases could change and the home country currency against numeraire rˆ 1 qˆ
N
2 pˆ should react
to the change according to Z
1
, even though the real value of ith currency does not change. The term v
i
in Eq. 18 is the weight in a currency basket if we want to keep the trade balance unchanged. The term Z
1
in Eq. 18 is the optimal exchange rate response to an exogenous change in government purchases if we want to keep the
trade balance unchanged. Both are derived under the assumptions stated in the model. 3.5.2. Stabilizing the consumer price index
The consumer price index is defined in the previous section as CPI 5 p
I
5 p
u
p
1
2u
m
. Therefore, the change in the CPI is pˆ
I
5 upˆ 1
1 2 upˆ
m
. Letting pˆ
I
5 0,
Eq. 19 can be shown, rˆ 1 qˆ
N
2 pˆ 5 2
o
w9
i
Jˆ
i
1 qˆ
i
2 qˆ
N
2 Z
2
G ˆ ,
19 where w9
i
5 uE
1
1 1 2 uL
1
2
1
uE
2i
1 1 2 uL
2i
and Z
2
5 uE
1
1 1 2 uL
1
2
1
uE
3
1 1 2 uL
3
. Both can be shown to be positive. In order to keep the CPI unchanged, when ith currency appreciates [Jˆ
i
1 qˆ
i
2 qˆ
N
increases], the home country currency against numeraire, rˆ 1 qˆ
N
2 pˆ , should appreciate decrease according to w9
i
. When the government purchases increase by one percent, the home country currency should
appreciate according to Z
2
, even though the real value of ith currency does not change. Eqs. 18 and 19 give us the weights in a currency basket if we are aiming at
either stabilizing balance of trade or consumer price index. As clearly shown in these equations, it is possible for the government to optimally implement both fiscal policy
and currency basket policy to stabilize balance of trade and consumer price index simultaneously.
3.5.3. Stabilizing both the balance of trade and the consumer price index To show the case that changes in the balance of trade and the consumer price index
are zero simultaneously, one need to combine Eqs. 18 and 19 into Eq. 20, rˆ 1 qˆ
N
2 pˆ 5 2
o
w
i
Jˆ
i
1 qˆ
i
2 qˆ
N
2 Z
1
G ˆ 5 2
o
w9
i
Jˆ
i
1 qˆ
i
2 qˆ
N
2 Z
2
G ˆ ,
20 to solve the optimal fiscal policy as shown in Eq. 21:
332 H.-L. Han International Review of Economics and Finance 9 2000 323–350
G ˆ 5
o
w9
i
2 w
i
Z
1
2 Z
2
Jˆ
i
1 qˆ
i
2 qˆ
N
. 21
In order to stabilize both the over all trade balance and the consumer price index, government purchase G is no longer an exogenous variable and should follow the
process as specified in 21. G has to respond to a change in the real value of ith currency. We then substitute Eq. 21 into Eq. 17 to get Eq. 22,
rˆ 1 qˆ
N
2 qˆ 5 2
o
[rw9
i
1 1 2 rw
i
]Jˆ
i
1 qˆ
i
2 qˆ
N
5 2
o
w
i
Jˆ
i
1 qˆ
i
2 qˆ
N
, 22
where r 5 Z
1
Z
1
2 Z
2
. From Eqs. 21 and 22, it is shown that both G and the home country currency
rˆ 1 qˆ
N
2 pˆ have to respond to the changes in the real value of ith currency at Jˆ
i
1 qˆ
1
2 qˆ
N
the same time to stabilize overall balance of trade and consumer price index. The optimal weight in a currency basket w that reaches these two policy goals
simultaneously is a weighted average of two separate sets of weights w
i
,w9
i
, each targeted at one policy goal. The weight r is a function of the structural parameters.
The results show that by combining an optimal fiscal policy that follows Eq. 21 with a currency basket peg that uses an optimal set of weights [w in Eq. 22], one economy
can insulate its overall balance of trade and aggregate price level from a change in a third-country’s real exchange rate.
4. The currency basket system in Thailand
