how did the dollar peg fail in asia?

JOURNAL OF THE JAPANESE AND INTERNATIONAL ECONOMIES 12, 256–304 (1998)ARTICLE NO. JJ980415

How Did the Dollar Peg Fail in Asia?*

Takatoshi Ito

Institute of Economic Research, Hitotsubashi University, Kunitachi, Tokyo 186, Japan

Eiji Ogawa

Department of Commerce, Hitotsubashi University, Kunitachi, Tokyo 186, Japan

and

Yuri Nagataki Sasaki

Department of Commerce, Takachiho University, Suginami, Tokyo 168, Japan

Received July 8, 1998

Ito, T., Ogawa, E., and Sasaki, Y. N.—How Did the Dollar Peg Fail in Asia?

In this paper, we have constructed a theoretical model in which the Asian firmmaximizes its profit, competing with the Japanese and the U.S. firms in their markets.The duopoly model is used to determine export prices and volumes in response tothe exchange rate fluctuations vis-a-vis the Japanese yen and the U.S. dollar. Then,the optimal basket weight that would minimize the fluctuation of the growth rateof trade balance was derived. These are the novel features of our model. The exportprice equation and export volume equation are estimated for several Asian countriesfor the sample period from 1981 to 1996. Results are generally reasonable. Theoptimal currency weights for the yen and the U.S. dollar are derived and comparedwith actual weights that had been adopted before the currency crisis of 1997. Forall countries in the sample, it is shown that the optimal weight of the yen issignificantly higher than the actual weight. J. Japan. Int. Econ., Dec. 1998, 12(4),pp. 256–304. Institute of Economic Research, Hitotsubashi University, Kunitachi,Tokyo 186, Japan; Department of Commerce, Hitotsubashi University, Kunitachi,Tokyo 186, Japan; Department of Commerce, Takachiho University, Suginami,Tokyo 168, Japan. 1998 Academic Press

Journal of Economic Literature Classification Numbers F31, F33, O11.

* The authors are grateful to Professors Shinji Takagi, Masahiro Kawai, and Shinichi Fukuda,and other participants of the conference.

2560889-1583/98 $25.00Copyright 1998 by Academic PressAll rights of reproduction in any form reserved.

HOW DID THE DOLLAR PEG FAIL IN ASIA? 257

1. INTRODUCTION

The currency crises in Asia in 1997 highlighted the danger of the fixedexchange rate system. Four ASEAN currencies (the Thai baht, the Malay-sian ringgit, the Indonesian rupiah, and the Philippines peso) all depreciatedby 30 to 40% in the three months following the baht depreciation of July2. Thailand asked the International Monetary Fund (IMF) for a balanceof payments support package in August. The IMF support ($4 billion) wascomplemented by Japan ($4 billion) and other Asian nations together withthe World Bank and Asian Development Bank (ADB). In November,Indonesia asked the IMF, the World Bank, and the ADB to advise themon economic reform together with a support for a potential balance ofpayment gap. The Indonesian package by IMF, World Bank, and ADBwas also complemented by a secondary line of support by Japan, the UnitedStates, and Asian countries. In late November Korea, after its currencydepreciated sharply, asked for IMF support. Also in November, the sharpdecline in the Hong Kong stock market, which was caused by defendingthe fixed exchange rate based on the currency board, caused worldwideturmoil. The crisis spread to Korea in November, and the IMF packagewas hastily put together in the first week of December. The crises in thesecountries deepened in December 1997 to January 1998, as the value of theIndonesian rupiah depreciated to a level one-sixth of the precrisis level,and other ASEAN countries and the Korean won depreciated to a levelone-half of the precrisis level.

One of the common factors among these crises countries was the choiceof a de facto dollar peg. Thailand has adopted a basket system, in whichthe value of the Thai baht is determined as a weighted average of majorcurrencies. However, it was well known that the U.S. dollar had had anoverwhelming weight in the basket since 1985. Indonesia had adopted aslide system with a narrow band, where the slide was adjusted for theinflation rate difference between Indonesia and the United States. TheKorean won had also maintained a stable value against the U.S. dollar.The Hong Kong dollar has been backed by a currency board management,and nominally pegged to the U.S. dollar at HK$7.7–7.8 per US$, since1984. Hence, most of these currencies have appreciated in the ‘‘real’’ ex-change rate sense, vis-a-vis the U.S. dollar.

Moreover, these Asian countries in financial turmoil have substantialtrade relationships with Japan. As the Japanese yen depreciated against theU.S. dollar from April 1995 to the summer of 1997, and the real ‘‘effective’’exchange rates of these countries appreciated. Due to the appreciation,export competitiveness was lost. Thus exports from these countries declinedand current account deficits increased in 1996–1997. On a practical level,this story makes sense. However, in theory, this needs some examination.

258 ITO, OGAWA, AND SASAKI

To simplify, consider a case that the nominal exchange rates of (non-Japan) Asian countries are pegged to the U.S. dollar, while Asian productscompete with Japanese products in the U.S. and Japanese markets. If theyen depreciates against the U.S. dollar, demands for Asian goods willdecline. However, if Asian countries would like to avoid the loss of competi-tiveness, the export prices (in baht for example) can be cut instead of theexchange rate depreciation. Therefore, we need a framework in which pricesare determined endogenously and adjusted imperfectly to the changes inthe exchange rate.

It is easy to see that if Asian goods are perfect substitutes to Japaneseand U.S. goods in a one-good, two-country economy, then export priceswill change inversely to the exchange rate changes. This can be called acase of perfect pass-through. The exchange rate regime, whether peggedto the dollar or freely floated, is irrelevant when prices are perfectly flexiblewith perfect competition for exported goods. However, if they competewith the Japanese goods in an oligopolistic market, then the exchange ratefluctuation will not be perfectly offset by the changes in export prices.

Moreover, Asian countries import from Japan parts of the products thatare exported to Japan and the United States. The Japanese yen apprecia-tion, therefore, has two different effects, increasing costs of semifinishedgoods and increasing competitiveness of exports.

In the early 1980s, a number of papers, including Bhandari (1985), Flan-ders and Helpman (1979), Flanders and Tishler (1981), Lipschitz and Sund-arajan (1980), and Turnovsky (1982), studied the optimal currency basket.Based on some open macroeconomic models including export and importfunctions, these papers explored how variances of balance of payments orsome other measures can be minimized when there are shocks to theexchange rates of trading partners. Bhandari (1985) and Turnovsky (1982)considered the question in a general equilibrium macromodel with capitalmobility. However, a macroeconomic structure, such as consumption andexport functions, is given without microfoundation in these papers. Theoptimality is usually defined by minimizing variances of balance of paymentsor real income.

Our paper is quite different from those papers in three respects. First,our model has microfoundation; namely, the oligopolistic exporter maxi-mizes its profits. Competition with exporters of other countries is modeled.Thus, the export price is endogenously determined in response to theexchange rates. Price ‘‘stickiness’’ in our model is a result of optimizingbehavior. Second, imports of parts (semifinished goods) are explicitly mod-eled. This reflects the cost aspects of the currency changes. Last, optimalityin our model is to minimize the fluctuations, in terms of changes in thetrade balances, which is equivalent to profits in our model. This criterionis slightly different from criteria adopted in other papers in the literature.


The rest of the paper will be organized as follows. Section 2 will presenta theoretical oligopolistic competition model where prices are determinedendogenously. Section 3 will use the above model to explore effects of thedollar peg on a trade balance of the ASEAN economy. Section 4 will derivean optimal peg weight to stabilize fluctuations in the trade balance. Section5 summarizes the theoretical prediction of impacts of exchange rate changeson export volumes and export prices. Section 6 describes stylized facts onthe movement of the exchange rates and trade flows. Theoretical predictionsare tested in regressions in Section 7. Optimal peg weights are also calcu-lated in this section. Section 8 concludes the paper.

2. THE MODELS

An ASEAN country is modeled as a one-sector economy where a repre-sentative firm assembles parts imported from Japan and the United Statesinto manufactured products.1 We assume that volumes of parts that theASEAN firm imports from Japan and the United States are constant shares,gm and 1 2 gm , respectively, of the total volumes. The firm is assumed toexport its products to the Japanese and U.S. markets. The model is similarto the pricing to market model of Marston (1990).2

We set up two competitive situations. The first situation is that each ofthe Japanese and U.S. markets is modeled as a duopoly where the ASEANfirm competes against the Japanese firm (Model A). The second situationis that each of the Japanese and U.S. markets is modeled as a duopolywhere the ASEAN firm competes against each local firm in the markets(Model B). That is, it competes against the Japanese firm in the Japanesemarket and against the U.S. firm in the U.S. market. The ASEAN, Japanese,and U.S. firms have identical cost functions. We assume that each of theJapanese and U.S. firms assembles its products with its domestic parts only.Each firm maximizes its profits in terms of its own home currency.

2.1. Model A

Profits of the ASEAN firm in terms of the home currency f is calculated as

f 5 E A/YPYJ f(qJ) 1 E A/$P$

US g(qUS) 2 E A/YPYmgmQ

(1)2 EA/$P$

m(1 2 gm)Q 2 C(Q),

where PYJ denotes a price of the ASEAN firm’s products in the Japanese

1 Ohno (1989) examined pass-through effects of exchange rates on export pricing behaviorin manufacturing after taking account of prices of raw materials.

2 Krugman (1987) and Knetter (1989, 1993) analyzed the pricing to market questions.


market in terms of the yen; P$US denotes a price of the ASEAN firm’s

products in the U.S. market in terms of the dollar; PYm denotes a price of

parts imported from Japan in terms of the yen; P$m denotes a price of parts

imported from the United States in terms of the dollar; Q denotes outputof the ASEAN products (Q 5 f(qJ) 1 g(qUS)); f(z) denotes a demandfunction for the ASEAN products in the Japanese market ( f 9 , 0); g(z)denotes a demand function for the ASEAN firm’s products in the U.S.market (g9 , 0); C(z) denotes a cost function of the ASEAN firm (C9 .0, C0 $ 0); qJ ; PY

J /PY*J denotes a relative price of the ASEAN products

relative to the Japanese products in the Japanese market; qUS ; P$US/P$*

US

denotes a relative price of the ASEAN products relative to the Japaneseproducts in the U.S. market; PY*

J denotes a price of the Japanese productsin the Japanese market in terms of the yen; P$*

US denotes a price of theJapanese products in the U.S. market in terms of the dollar; EA/Y denotesan exchange rate of the yen in terms of the ASEAN currency; and EA/$

denotes an exchange rate of the dollar in terms of the ASEAN currency.From Eq. (1), profit-maximizing prices of the ASEAN firm in the Japa-

nese and U.S. markets, respectively, are derived as

PYJ 5 eJ SgmPY

m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD (2a)

P$US 5 eUS Sgm

EA/Y

EA/$ PYm 1 (1 2 gm)P$

m 1C9

EA/$D , (2b)

where eJ ; «J(qJ)/(«J (qJ) 2 1) denotes markups of the ASEAN productsin the Japanese market; «J ; 2f 9(qJ)qJ/f(qJ) denotes price elasticity ofdemand for the ASEAN products in the Japanese market; eUS ; «US(qUS)/(«US(qUS) 2 1) denotes markups of the ASEAN products in the U.S. market;and «US(qUS) ; 2g9(qUS)qUS/g(qUS) denotes price elasticity of demand forthe ASEAN products in the U.S. market. We assume that d«J/dqJ . 0,deJ/dqJ . 0, d«US/dqUS . 0, and deUS/dqUS . 0. For simplicity, we alsoassume that the marginal production costs are constant (C9(Q) 5 C9).

We convert Eqs. (2a) and (2b) into a logarithm form,

log PYJ 5 2hJ (log PY

J 2 log PY*J )

1 log SgmPYm 1 (1 2 gm)

EA/$

EA/Y P$m 1

C9

EA/YD (2a9)

log P$US 5 2hUS (log P$

US 2 log P$*US)

1 log SgmEA/Y

EA/$ PYm 1 (1 2 gm)P$

m 1C9

EA/$D , (2b9)


where hJ ; e9J qJ/eJ . 0 denotes price elasticity of the markups of theASEAN products in the Japanese market and hUS ; 2e9USqUS/eUS . 0denotes price elasticity of the markups of the ASEAN products in theU.S. market.

We derive reaction functions of the ASEAN firm in the Japanese andU.S. markets given the prices of the products made in Japan, respectively.

log PYJ 5

hJ

1 1 hJlog PY*

J

11

1 1 hJlog SgmPY

m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD (3a)

log P$US 5

hUS

1 1 hUSlog P$*

US

11

1 1 hUSlog Sgm

EA/Y

EA/$ PYm 1 (1 2 gm)P$

m 1C9

EA/$D (3b)

Profits of the Japanese firm in terms of the yen f* are calculated as

f* 5 PY*J f*(1/qJ) 1

EA/$P$*USg*(1/qUS)EA/Y 2 PY

mQ* 2 C(Q*), (4)

where Q* 5 f*(1/qJ) 1 g*(1/qUS), f*(1/qJ) denotes a demand function forthe Japanese products in the Japanese markets ( f*9 , 0) and g*(1/qUS)denotes a demand function for the Japanese products in the U.S. markets(g*9 , 0).

Profit-maximizing prices of the Japanese firm in the Japanese and U.S.markets are, respectively, derived as

PY*J 5 e*J (PY

m 1 C9) (5a)

P$*US 5 e*US SEA/Y

EA/$ PYm 1

EA/Y

EA/$ C9D , (5b)

where e*J ; «*J (1/qJ)/(«*J (1/qJ) 2 1) denotes markups of the Japanese prod-ucts in the Japanese market; «*J ; 2f*9(1/qJ)/f*(1/qJ)qJ denotes priceelasticity of demand for the Japanese products in the Japanese market;e*US ; «*US(1/qUS)/(«*US(1/qUS) 2 1) denotes markups of the Japanese prod-ucts in the U.S. market; and «*US ; 2f*9(1/qUS)/f*(1/qUS)qUS denotes priceelasticity of demand for the Japanese products in the U.S. market. We


assume that d«*J /d(1/qJ) . 0, de*J /d(1/qJ) . 0, d«*US/d(1/qUS) . 0, andde*US/d(1/qUS) . 0.

We derive reaction functions of the Japanese firm given the prices ofthe ASEAN products in the Japanese and U.S. markets, respectively.

log PY*J 5

h*J

1 1 h*Jlog PY

J 11

1 1 h*Jlog(PY

m 1 C9) (6a)

log P$*US 5

h*US

1 1 h*US

log P*US 11

1 1 h*US

log HEA/Y

EA/$ (PYm 1 C9)J , (6b)

where h*J ; 2e*9/e*qJ . 0 denotes price elasticity of markups of theJapanese products in the Japanese market and h*US ; 2e*9US/e*USqUS . 0denotes price elasticity of markups of the Japanese products in the U.S.market.

From Eqs. (3a) and (6a), we derive equilibrium prices of the ASEANproducts and the Japanese products in the Japanese market, respectively:

log PYJ 5

1 1 h*J

1 1 hJ 1 h*Jlog SgmPY

m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD1

hJ

1 1 hJ 1 h*Jlog(PY

m 1 C9) (7a)

log PY*J 5

h*J


m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD1

1 1 hJ

1 1 hJ 1 h*Jlog(PY

m 1 C9). (7b)

From Eqs. (3b) and (6b), we derive equilibrium prices of the ASEANproducts and the Japanese products in the U.S. market, respectively:

log P$US 5 log

EA/Y

EA/$ 11 1 h*US

1 1 hUS 1 h*US

log Sgm PYm 1 (1 2 gm)

EA/$

EA/Y P$m 1

C9

EA/YD1

hUS

1 1 hUS 1 h*US

log(PYm 1 C9) (8a)


log P$*US 5 log

EA/Y

EA/$ 1h*US

1 1 hUS 1 h*US

log Sgm PYm 1 (1 2 gm)

EA/$

EA/Y P$m 1

C9

EA/YD1

1 1 hUS

1 1 hUS 1 h*US

log(PYm 1 C9). (8b)

From Eqs. (7) and (8), we obtain equilibrium relative prices of theASEAN products relative to the Japanese products in the Japanese andthe U.S. markets, respectively:

log qJ 5 log PYJ 2 log PY*

J

51

1 1 hJ 1 h*JHlog Sgm PY

m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD2 log(PY

m 1 C9)J (9a)

log qUS 5 log P$US 2 log P$*

US

51

1 1 hUS 1 h*US

Hlog Sgm PYm 1 (1 2 gm)

EA/$

EA/Y P$m 1

C9

EA/YD2 log(PY

m 1 C9)J . (9b)

Equations (9) show that the relative prices qJ and qUS depend on boththe exchange rate of the yen relative to the ASEAN currency EA/Y andthe exchange rate of the dollar relative to the ASEAN currency EA/$. Theexchange rate of the yen EA/Y has effects on the relative prices via boththe marginal cost of parts imported from the United States and the marginalproduction cost while the exchange rate of the dollar EA/$ has effects onthe relative prices via the marginal cost of parts imported from theUnited States.

A depreciation of the ASEAN currency against the yen has a negativeeffect on both of the equilibrium relative prices of the ASEAN productsrelative to the Japanese products in the Japanese and the U.S. markets.On the other hand, a depreciation of the ASEAN currency against thedollar has a positive effect on both of the equilibrium relative prices.


We specify demand functions for the ASEAN products in the Japaneseand the U.S. markets from Eqs. (9a) and (9b) as the price elasticities ofdemand in the Japanese and U.S. markets are «J and «US , respectively.

log f 52«J


m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD2 log(PY

m 1 C9)J (10a)

log g 52«US

1 1 hUS 1 h*US

Hlog Sgm PYm 1 (1 2 gm)

EA/$

EA/Y P$m 1

C9

EA/YD2 log(P$

m 1 C9)J (10b)

Equation (10a, b) shows that the demands for the ASEAN products inthe Japanese market depend on both exchange rates of EA/Y and EA/$.Similarly for the U.S. market. The exchange rate of the yen EA/Y has effectson the demands via both the marginal cost of parts imported from theUnited States and the marginal production cost, while the exchange rateof the dollar EA/$ has effects on the demands via the marginal cost of partsimported from the United States.

A depreciation of the ASEAN currency against the yen has a positiveeffect on the demands for the ASEAN products in both the Japaneseand the U.S. markets. Thus, the depreciation increases export volume ofASEAN. On the other hand, a depreciation of the ASEAN currency againstthe dollar has a negative effect on both the demands for the ASEANproducts and the export volume.

2.2. Model B

In Model B, the ASEAN firm competes against the Japanese firm in theJapanese market and, on the other hand, against the U.S. firm in the U.S.market. We suppose that the Japanese firm supplies its products only inthe Japanese market while the U.S. firm supplies its products only in theU.S. market. The U.S. firm in the theoretical model does not have to bea firm located in the United States in the real world. Goods made in LatinAmerica as well as in the United States can be interpreted as the goodsmade in the United States in the theoretical model, because many of theLatin American currencies were de facto pegged to the U.S. dollar (espe-cially before the Mexican currency crisis of December 1994).

We formalize the same profit equation of the ASEAN firm as in Model


A. However, qUS ; P$US/P$*

US represents the price of the ASEAN productsrelative to the U.S. products in the United States and P$*

US represents theprice of the U.S. products in the U.S. market in terms of the dollar inModel B. Thus, as in Model A, the reaction functions (3a) and (3b) of theASEAN firm in the Japanese and U.S. markets are derived.

Profits of the Japanese firm in terms of the yen f*J are formalized as

f*J 5 PY*J f*(1/qJ) 2 PY

m f*(1/qJ) 2 C( f*(1/qJ)), (11)

where f*(1/qJ) denotes a demand function for the Japanese products inthe Japanese markets ( f*9 , 0).

Profit-maximizing prices of the Japanese firm in the Japanese marketare derived as

PY*J 5 e*J (PY

m 1 C9), (12)

where e*J ; «*J (1/qJ)/(«*J (1/qJ) 2 1) denotes markups of the Japanese prod-ucts in the Japanese market and «*J ; 2f*9(1/qJ)/f*(1/qJ)qJ denotes priceelasticity of demand for the Japanese products in the Japanese market. Weassume that d«*J /d(1/qJ) . 0 and de*J /d(1/qJ) . 0.

We derive a reaction function of the Japanese firm given the prices ofthe ASEAN products in the Japanese market.

log PY*J 5

h*J

1 1 h*Jlog PY

J 11

1 1 h*Jlog(PY

m 1 C9), (13)

where h*J ; 2e*9J /e*J qJ . 0 denotes price elasticity of markups of theJapanese products in the Japanese market.

Profits of the U.S. firm in terms of the dollar f*US is formalized as

f*US 5 P$*US g*(1/qUS) 2 P$

m g*(1/qUS) 2 C(g*(1/qUS)), (14)

where g*(1/qUS) denotes a demand function for the U.S. products in theU.S. markets (g*9 , 0).

Profit-maximizing prices of the U.S. firm in the U.S. market are derived as

P$*US 5 e*US(P$

m 1 C9), (15)

where e*US ; «*US (1/qUS)/(«*US (1/qUS) 2 1) denotes markups of the U.S.products in the U.S. market and «*US ; 2f*9(1/qUS)/f*(1/qUS)qUS denotes


price elasticity of demand for the U.S. products in the U.S. market. Weassume that d«*US/d(1/qUS) . 0 and de*US/d(1/qUS) . 0.

We derive a reaction function of the U.S. firm given the prices of theASEAN products in the U.S. market.

log P$*US 5

h*US

1 1 h*US

log P$US 1

1

1 1 h*US

log(P$m 1 C9) (16)

h*US ; 2e*9US/e*US qUS . 0 denotes price elasticity of markups of the U.S.products in the U.S. market.

From Eqs. (3a) and (13), we derive equilibrium prices of the ASEANproducts and Japanese products in the Japanese market, respectively:

log PYJ 5

1 1 h*J


m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD1

hJ

1 1 hJ 1 h*Jlog(PY

m 1 C9) (17a)

log PY*J 5

h*J


m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD1

1 1 hJ

1 1 hJ 1 h*Jlog(PY

m 1 C9). (17b)

From Eqs. (3b) and (16), we derive equilibrium prices of the ASEANproducts and U.S. products in the U.S. market, respectively:

log P$US 5

1 1 h*US

1 1 hUS 1 h*US

log SgmEA/Y

EA/$ PYm 1 (1 2 gm)P$

m 1C9

EA/$D1

hUS

1 1 hUS 1 h*US

log(P$m 1 C9) (18a)

log P$*US 5

h*US

1 1 hUS 1 h*US

log SgmEA/Y

EA/$ PYm 1 (1 2 gm)P$

m 1C9

EA/$D1

1 1 hUS

1 1 hUS 1 h*US

log(P$m 1 C9). (18b)


From Eqs. (17) and (18), we obtain an equilibrium relative price of theASEAN products relative to the Japanese products in the Japanese marketsand an equilibrium relative price relative to the U.S. products in the U.S.markets, respectively:

log qJ 5 log PYJ 2 log PY*

J

51


m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD2 log(PY

m 1 C9)J (19a)

log qUS 5 log P$US 2 log P$*

US

51

1 1 hUS 1 h*US

Hlog SgmEA/Y

EA/$ PYm 1 (1 2 gm)P$

m 1C9

EA/$D2 log(P$

m 1 C9)J . (19b)

Equations (19) show that the relative prices qJ and qUS depend on boththe exchange rate of the yen relative to the ASEAN currency EA/Y andthe exchange rate of the dollar relative to the ASEAN currency EA/$. Theexchange rate of the yen EA/Y has negative effects on the relative pricesin the Japanese market via both the marginal cost of parts imported fromthe U.S. and the marginal production cost, while it has positive effects onthe relative prices in the U.S. market via the marginal cost of parts importedfrom Japan. On the other hand, the exchange rate of the dollar EA/$ haspositive effects on the relative prices in the Japanese market via the marginalcost of parts imported from the United States while it has negative effectson the relative prices in the U.S. market via both the marginal cost of partsimported from Japan and the marginal production cost.

A depreciation of the ASEAN currency against the yen has a negativeeffect on the equilibrium relative prices of the ASEAN products relativeto the Japanese products in the Japanese market while it has a positiveeffect on the equilibrium relative prices in the U.S. markets. On the otherhand, a depreciation of the ASEAN currency against the dollar has apositive effect on the equilibrium relative price in the Japanese market whileit has a negative effect on the equilibrium relative price in the U.S. market.

We specify demand functions for the ASEAN products in the Japaneseand the U.S. markets from Eqs. (19a) and (19b) as the price elasticities ofdemand in the Japanese and U.S. markets are «J and «US , respectively.


log f 52«J


m 1 (1 2 gm)EA/$

EA/Y P$m 1

C9

EA/YD2 log(PY

m 1 C9)J (20a)

log g 52«US

1 1 hUS 1 h*US

Hlog SgmEA/Y

EA/$ PYm 1 (1 2 gm)P$

m 1C9

EA/$D2 log(P$

m 1 C9)J (20b)

Equations (20) show that both the exchange rates of EA/Y and EA/$ haveasymmetric effects on the demands for ASEAN products in the Japaneseand U.S. markets. As a result, the exchange rates have ambiguous effectson the export volume of ASEAN, which is a sum of both the demandsf 1 g.

However, it is clearer how at least the exchange rate of the dollar affectsthe export volume of ASEAN by supposing an extreme case where theASEAN firm imported parts from Japan only, that is, gm 5 1. Demandfunctions for the ASEAN products in the Japanese and the U.S. marketsin the case of gm 5 1 are as follows:

log f 52«J

1 1 hJ 1 h*JHlog SPY

m 1C9

EA/YD2 log(PYm 1 C9)J (20a9)

log g 52«US

1 1 hUS 1 h*US

Hlog SEA/Y

EA/$ PYm 1

C9

EA/$D2 log(P$m 1 C9)J (20b9)

Equations (20a9) and (20b9) show that the exchange rate of the ASEANcurrency against the dollar has unambiguous effects on the export volumeof the ASEAN f 1 g while the exchange rate of the ASEAN currencyagainst the yen has still ambiguous effects on export volume. If the shareof parts imported from Japan is higher, it is more likely that the exchangerate against the dollar has unambiguous positive effects on the exportvolume while the exchange rate against the yen has ambiguous effects.

3. EFFECTS OF THE DOLLAR PEG ON FLUCTUATIONS IN ATRADE BALANCE

In this section, we explore effects of the dollar peg on a trade balanceof the ASEAN economy.


A trade balance of the ASEAN economy is equal to a total export valueor total sales of the ASEAN firm less the total import value or total costsof imported parts in our model. Therefore, we represent the trade balancein terms of the home currency T,

T 5 EA/YPYJ f(qJ) 1 EA/$P$

US g(qUS) 2 gm EA/YPYmQ 2 (1 2 gm)EA/$P$

mQ5 [hgx EA/YPY

J 1 (1 2 gx)EA/$P$USj hgm EA/YPY

m 1 (1 2 gm)EA/$P$mj]Q (21)

5 (PAx 2 PA

m)Q,

where gx 5 f/Q denotes the share of products exported to Japan inthe total export volumes; PA

x ; gx EA/YPYJ 1 (1 2 gx)EA/$P$

US denotesthe volume-weighted average export price in terms of the home currency;and PA

m ; gmEA/YPYm 1 (1 2 gm)EA/$P$

m denotes the volume-weightedaverage import price in terms of the home currency. Equation (21)shows that the trade balance is proportional to a volume-weightedaverage export price less the volume-weighted average import pricebecause the total export volumes are equal to the total import volumesin our model.

Note that the trade balance is equivalent to nominal GDP in our model.Therefore, we can say that we explore effects of the dollar peg on nominalGDP in our model.

We derive a relationship between fluctuations in the trade balanceand those in the exchange rates in both Model A and Model B (seeAppendix),

T 51T FTJ EA/Y 1 TUS EA/$

1 H1 1 h*J 2 «J

1 1 hJ 1 h*JXJ 1

1 1 h*US 2 «US

1 1 hUS 1 h*US

XUSJ[(1 2 gm)B1EA/$ 2 h(1 2 gm)B1 1 B2jEA/Y ]

1 H «J

1 1 hJ 1 h*Jgx 1

«US

1 1 hUS 1 h*US

(1 2 gx)JM [(1 2 gm)B1EA/$ 2 h(1 2 gm)B1 1 B2jEA/Y ]G (22)


T 51T FTJ EA/Y 1 TUS EA/$

11 1 h*J 2 «J

1 1 hJ 1 h*JXJ [(1 2 gm)B1EA/$ 2 h(1 2 gm)B1 1 B2jEA/Y ]

11 1 h*US 2 «US

1 1 hUS 1 h*US

XUS [2h1 2 (1 2 gm)B1jEA/$

1 h1 2 (1 2 gm)B1 2 B2jEA/Y ]

1«J

1 1 hJ 1 h*Jgx M [(1 2 gm)B1EA/$ 2 h(1 2 gm)B1 1 B2jEA/Y ]G

1«US

1 1 hUS 1 h*US

(1 2 gx)M [2h1 2 (1 2 gm)B1jEA/$

1 h1 2 (1 2 gm)B1 2 B2jEA/Y ]G, (23)

where TJ ; (gx EA/YPYJ 2 gmEA/YPY

m)Q denotes the trade balance withJapan; TUS 5 h(1 2 gx)EA/$P$

US 2 (1 2 gm)EA/$P$mjQ denotes the trade

balance with the United States; XJ ; gxEA/YPYJ Q denotes the value of

exports to Japan; XUS ; (1 2 gx)EA/$P$USQ denotes the value of exports

to the United States; M ; PAmQ denotes the total import value, B1 ;

(EA/$P$m/EA/Y)/TCA ; (1 2 gm)B1 denotes the share of marginal cost of parts

imported from the United States in the total marginal costs; B2 ;(C9/EA/Y)/TCA denotes the share of marginal production cost in the totalmarginal costs; TCA ; gmPY

m 1 (1 2 gm) EA/$P$m/EA/Y 1 C9/EA/Y denotes

the total marginal costs; and x means a rate of change in a variable x.For simplicity, the price elasticity of demand in the Japanese market is

assumed to be equal to that in the U.S. market («J 5 «US 5 «). When theprice elasticities of demand are equal to each other, the price elasticitiesof markups in both the markets are also equal to each other (hJ 5 hUS 5h and h*J 5 h*US 5 h*). Therefore, we change Eqs. (22) and (23) into thefollowing equations,

T 51T FhTJ EA/Y 1 TUS EA/$j 1

1 1 h*1 1 h 1 h*

X h(1 2 gm)B1EA/$ 2 h(1 2 gm)B1 1 B2j EA/Yj

«

1 1 h 1 h*T h(1 2 gm)B1EA/$ 2 h(1 2 gm)B1 1 B2j EA/YjG (229)


T 51T FhTJ EA/Y 1 TUS EA/$j 1

1 1 h*1 1 h 1 h*

X h(1 2 gm)B1EA/$ 2 h(1 2 gm)B1 1 B2j EA/Yj

«

1 1 h 1 h*T h(1 2 gm)B1EA/$ 2 h(1 2 gm)B1 1 B2j EA/Yj

1 H 1 1 h*1 1 h 1 h*

XUS 1«

1 1 h 1 h*TUSJ (EA/Y 2 EA/$)G , (239)

where X ; PAx Q, a total export value.

Equations (229) and (239) show that the effects of the exchange rates onthe trade balance have the following transmission channels.

The first channel is a direct effect of the exchange rates on the tradebalance as shown in the first curly brace in both the equations. The ASEANcountry imports parts from Japan and the United States and exports prod-ucts to these countries. Thus, the exchange rates have a direct effect viaboth the export values and the import values.

The second channel is an indirect effect of the exchange rates via theproduct prices as shown in the second angle bracket of the equations. Theexchange rates effect the product prices as shown in Eqs. (7), (8), (17), and(18). The product prices effect export values. Thus, the price channel isthat the exchange rates effect the export volumes.

The third channel is an indirect effect of the exchange rate via thedemand for the ASEAN products and in turn its outputs as shown inthe third curly brace of the equations. The exchange rates effect theproduct prices, which change the relative prices of the products in theJapanese and the U.S. markets as shown in Eqs. (9) and (19). Thechanges in the relative prices effect the demand for the products inboth the markets as shown in Eqs. (10) and (20). The demand for ASEANproducts is equivalent to export volumes of the ASEAN economy. Onone hand, changes in outputs effect import volumes because the ASEANeconomy imports all of the parts from the foreign countries in ourmodel. Thus, this output channel is that the exchange rates effect boththe export volumes and the import volumes. The fourth curly brace ofEq. (239) is related to the second and third channels for trade with theUnited States.

Now we consider the effect of exchanges rate regime on the trade balanceof an ASEAN country. The monetary authorities of the ASEAN countryare assumed to peg the home currency to the dollar. The dollar peg isequivalent to a case of EA/$ 5 0. Substituting EA/$ 5 0 into Eqs. (229) and(239), we obtain the following equations:


T 51T FTJ 2

1 1 h*1 1 h 1 h*

X h(1 2 gm)B1 1 B2j

1«

1 1 h 1 h*T h(1 2 gm)B1 1 B2jG EA/Y (24)

T 51T FTJ 2

1 1 h*1 1 h 1 h*

X h(1 2 gm)B1 1 B2j

1«

1 1 h 1 h*T h(1 2 gm)B1 1 B2j

1 H 1 1 h*1 1 h 1 h*

XUS 1«

1 1 h 1 h*TUSJG EA/Y. (25)

In the case of the dollar peg, changes in the exchange rate of the yenagainst the ASEAN currency have some effects on the trade balance inour model where the ASEAN firm imports parts for production from Japanand the United States and exports its products to Japan and the UnitedStates. The ASEAN products compete against the Japanese firm or theU.S. firm. Changes in the exchange rate of the yen have effects on thetrade balance through the above three transmission channels.

For example, a depreciation of the yen (EA/Y , 0) deteriorates the tradebalance via the first direct channel if the ASEAN country has a tradebalance surplus with Japan (TJ . 0) as shown in the first term in the squarebracket of Eqs. (24) and (25). On the other hand, it improves the tradebalance if the ASEAN country has a trade balance deficit with Japan(TJ , 0).

Next, a depreciation of the yen against the home currency increases themarginal costs of parts imported from the United States and domesticproduction of the ASEAN firm relative to those of the Japanese firm. TheASEAN firm is forced to increase its product prices in both the Japaneseand the U.S. markets though it imperfectly passes-through the increase inthe marginal costs into the product prices because it competes against theJapanese firm in both the markets as shown in Eqs. (7a), (8a), (17a), and(18a). As a result, the depreciation of the yen improves the trade balancevia the price channel as shown in the second term of Eqs. (24) and (25).

Third, a depreciation of the yen increases the relative prices of theASEAN products relative to the Japanese products in both the Japaneseand the U.S. markets as shown in Eqs. (9) and (19). In turn, it reduces thedemands for the ASEAN products in both the markets as shown in Eqs.(10) and (20). This results in a contraction in outputs of the ASEAN firmand decreases both the export volumes and the import volumes of parts.


Therefore, a depreciation of the yen deteriorates the trade balance surplus(T . 0) or improves the trade balance deficit (T , 0) via the output channelas shown in the third term of Eqs. (24) and (25).

4. THE OPTIMAL WEIGHTS

The monetary authorities are assumed to choose weights of the yen andthe dollar in order to stabilize the fluctuations in the trade balance causedby changes in the exchange rates.3 In particular, we adopt a criterion ofminimizing the squared rate of change in the trade balances. The goal ofthe monetary authorities is equivalent to stabilizing the fluctuations inprofits of exporting firms in our model. Pegging the home currency to abasket of the yen and the dollar implies that the monetary authorities tryto keep the exchange rate of the accuracy basket in terms of the homecurrency unchanged. Therefore, changes in the exchange rate of the cur-rency basket in terms of the home currency is zero,

wEA/$ 1 (1 2 w)EA/Y 5 0, (26)

where w is a weight of the dollar in pegging the home currency to thecurrency basket. If w turns out to be 1 (0, respectively), then this countryis in the optimal currency area of the dollar (the yen, respectively).4

Note that it is not necessary to take into account a correlation betweenthe exchange rates vis-a-vis the yen and vis-a-vis the dollar, because theoptimality is defined to minimize the weighted average (currency basket)of the changes in the exchange rates against the dollar and the yen (thatis, the left-hand side of Eq. (26)). The correlation would have been relevantif we were to adopt a criterion that minimizes variance of the currencybasket (that is, the variance of the left-hand side of Eq. (26)).

For the choice of criteria for ‘‘optimality,’’ the literature does not have astandard. If the objective of the optimal basket is regarded as a multicountryreplacement for the peg to a single currency, then stability of the realeffective exchange rate is the criterion. This is a basic stance of Lipschitzand Sundrarajan (1980). It is implicit in this choice that if the exchangerate is stable (at around the equilibrium), then the trade balance is stable.

3 Flanders and Helpman (1979), Lipschitz and Sundararajan (1980), and Flanders and Tishler(1981) emphasized only the real side of the economy in modeling the basket peg issue. Onthe other hand, Turnovsky (1982) and Bhandari (1985) used a general equilibrium macromodelwhich included capital mobility.

4 The standard references of optimal currency areas include Kenen (1969), Mundell (1961),and McKinnon (1963).


One can make it more explicit. The principal role of the exchange rate isto keep the balance of trade or current accounts, so a natural candidate isthe trade balance stabilization. This line of thought is followed by Flandersand Helpman (1979) and Flanders and Tishler (1981). However, the ex-change rate policy is only one part of overall economic policy that pursuesprice stabilization and income growth. The balance of payments cannotbe a stand-alone objective. Turnovsky (1985) proposed the criterion ofstabilizing domestic income, a more general objective of economic policy.For domestic income stabilization, there are policy options other than thecurrency basket weights. Other policy options should be modeled if ageneral policy objective is introduced. Bhandari (1985), extending Turnov-sky, considered four criteria (or a combination thereof) at the same time,including the real effective exchange rate, in a similar model. It is notclear whether these four criteria can be weighted and combined in oneloss function.

Our paper is different from the previous papers in the literature. Ourmodel is based on the exporter’s maximizing behavior, instead of ad hocmacro model. How prices respond to the exchange rate changes is solvedin the model. This feature avoids a criticism that the optimal peg literaturetakes the prices as given. Our model makes it explicit how prices are setby the profit-maximizing firm. The optimality is to minimize the squaredrate of change in the trade balance T2 subject to Eq. (26), where a rate ofchange in the trade balance T is shown in Eq. (229) or (239). The criteriais equivalent to minimizing the variance of the growth rate, provided thatthe mean growth rate is equal to zero.5

In Model A, we obtain optimal weights of pegging the home currencyto the dollar w*;

w* 5

TUS 11 1 h*

1 1 h 1 h*X(1 2 gm)B1 2

«

1 1 h 1 h*T(1 2 gm)B1

T 21 1 h*

1 1 h 1 h*XB2 1

«

1 1 h 1 h*TB2

, (27)

where both the denominators and the numerators are assumed to be differ-ent from zero. Especially, the optimal peg weights are indeterminate if thenumerators are equal to zero.

5 In the literature, minimizing the variance of the level of trade balance is used, whileminimizing the growth rate of the trade balance is our objective function. The difference isthat in our case, when the level is deviated from the mean, staying at the new level (growthrate is zero) is not penalized in the objective function, while it is penalized in the objectivefunction of the literature. We consider that a particular level cannot be determined as thelong run optimum in our model, because the firm is maximizing profits; that is the tradebalance in our model.


In Model B, we obtain optimal weights of pegging the home currencyto the dollar w*:

w* 5

TUS 21 1 h*

1 1 h 1 h*hXUS 2 X(1 2 gm)B1j

1«

1 1 h 1 h*hTUS 2 T(1 2 gm)B1j

T 21 1 h*

1 1 h 1 h*XB2 1

«

1 1 h 1 h*TB2

. (28)

Equations (27) and (28) indicate that both the optimal peg weights arerelated to three factors. The exchange rates made impacts on the tradebalance of the ASEAN economy via the above-mentioned three channels:the direct channel, the price channel, and the output channel. Here, wefind the three channels in the equations. The first terms in both the denomi-nators and the numerators are related with the direct channel. The secondterms are related with the price channel. The third terms are related withthe output channel.

We could focus on only a ratio of the trade balance with the UnitedStates or Japan to the total trade balances in determining the optimal pegweights if the price and the output channels were ineffective. However,the ASEAN firm sets the product prices in both the Japanese and the U.S.markets by adding the markups to the total marginal costs in terms of therelevant currencies. The exchange rates have effects on the total marginalcosts in terms of the relevant currencies. The two components of the totalmarginal costs for the ASEAN firm—the marginal cost of parts importedfrom the United States and the production marginal cost—are differentfrom those for the competing Japanese firm. These are expressed in (1 2gm)B1 and B2 in the equations.

Thus, the optimal peg weights are related with the price elasticities ofdemand and the cost compositions as well as the trade structures. Thismakes the difference in choosing an optimality criterion other than the realeffective exchange rate, such as Lipschitz and Sundararajan (1980).

5. THEORETICAL PREDICTION

5.1. Regression Forms

In order to prepare for regression, we derive operational regressionsfrom theoretical models in Section 2. Export prices in Model A can beexpressed as a function of the exchange rate, say the baht, against the U.S.dollar and the baht against the Japanese yen. (See Appendix for derivation.)


PAx 5 a1 EA/$ 1 a2 EA/Y

a1 ; 1 1 h*1 1 h 1 h*

(1 2 gm)B1 . 0 (29)

a2 5 1 21 1 h*

1 1 h 1 h*h(1 2 gm)B1 1 B2j 5 1 2 a1 2

1 1 h*1 1 h 1 h*

B2 . 0

Export prices in Model A can be expressed as a function of the exchangerates against the U.S. dollar and against the Japanese yen. (See Appendixfor derivation.)

PAx 5 a3 EA/$ 1 a4 EA/Y

a3 ; 1 1 h*1 1 h 1 h*

(1 2 gm)B1 1h

1 1 h 1 h*(1 2 gx) . 0

a4 5 1 21 1 h*

1 1 h 1 h*h(1 2 gm)B1 1 B2j 2

h

1 1 h 1 h*(1 2 gx)

5 1 2 a3 21 1 h*

1 1 h 1 h*B2 . 0 (30)

Export volume in Model A can be expressed as a function of the exchangerates against the U.S. dollar and against the Japanese yen. (See Appendixfor derivation.)

Q 5 b1 EA/$ 1 b2 EA/Y

b1 ; 2«

1 1 h 1 h*(1 2 gm)B1 , 0

b2 ; «

1 1 h 1 h*h(1 2 gm)B1 1 B2j 5 2b1 1

«

1 1 h 1 h*B2 . 0 (31)

Export volume in Model B can be expressed as a function of the exchangerates against the U.S. dollar and against the Japanese yen. (See Appendixfor derivation.)

Q 5 b3 EA/$ 1 b4 EA/Y

b3 ; 2«

1 1 h 1 h*(1 2 gm)B1 1

«

1 1 h 1 h*(1 2 gx)

b4 ; «

1 1 h 1 h*h(1 2 gm)B1 1 B2j 2

«

1 1 h 1 h*(1 2 gx)

5 2b3 1«

1 1 h 1 h*B2 (32)


5.2. Sign Conditions

Major predictions of the exchange rate changes on the export volumesand prices based on the theoretical model developed in Section 2 canbe summarized as follows. Based on Model A (the model where Asianexports compete with Japanese goods in Japan and in the United States),export volume will rise if the baht depreciates against the Japanese yen,but will decline if the baht depreciates against the U.S. dollar. Thereason that export volume will decline when the baht depreciates againstthe U.S. dollar is the hike in costs of imported parts from the UnitedStates, while the costs for Japanese producers would not change as theyuse local parts. If the dollar peg is maintained, then the yen depreciationvis-a-vis the U.S. dollar means the baht appreciates vis-a-vis the yen,while the baht–dollar exchange rate is stable, resulting in the decreasein export volume.

Based on Model B (the model where Asian exports compete in the U.S.market against U.S.-made goods and compete in Japanese markets againstJapanese-made goods, the following theoretical prediction is obtained. Adepreciation of the baht against the dollar has two effects. The costs ofimported parts will rise, so this has a negative effect on export volume. Onthe other hand, local costs for Thai exporters will become less than for theU.S. producers. This will have positive effects on export volume. Similarly,a depreciation of the baht against the yen has two effects, higher importedparts costs and lower local costs. In the end, signs of the coefficients arenot determined.

Consider a special case where parts are imported only from Japan(gm 5 1). In the case of Model A, a baht depreciation against the dollardoes not have any effect on export volume, while a baht depreciationagainst the yen will increase export volume; and in the case of Model B,the baht depreciation against the dollar will increase export volume, whilethe effect of the baht depreciation against the yen still has an ambiguoussign on export volume.

The effects of the exchange rate changes on the export prices of Thailandare more straightforward. Both in Model A and in Model B, a baht deprecia-tion against the dollar results in the hike of export prices. This reflects theincrease in costs of the imported parts. However, the increase in the pricesis less than the change in the exchange rate, due to competitive pressurein the destination market, thus an imperfect pass-through. The baht depreci-ation against the yen results in the price hike and also results in the exportprice increase in Model A and Model B. The discussion is summarized inTable I.


TABLE ITheoretical Prediction of Sign Conditions

Model A Model B

Special SpecialEffects on export volume from General (gm 5 1) General (gm 5 1)

Baht–dollar hike (depreciation of baht b1 , 0 b1 5 0 b3 ? b3 . 0against dollar)

Baht–yen hike (depreciation of baht b2 . 0 b2 . 0 b4 ? b4 ?against yen)

Model A Model B

Special SpecialEffects on export prices from General (gm 5 1) General (gm 5 1)

Baht–dollar hike (depreciation of baht a1 . 0 a1 5 0 a3 . 0 a3 . 0against dollar)

Baht–yen hike (depreciation of baht a2 . 0 a2 . 0 a4 . 0 a4 . 0against yen)

6. THE STYLIZED FACTS AND REGRESSIONS

6.1. De Facto Dollar Peg

Many East and Southeast Asian countries had adopted de facto dollarpeg regimes until the currency crisis of 1997, some explicitly and othersimplicitly. Hong Kong has explicitly maintained the currency board, thenominal peg to the dollar. Thailand, before July 1997, adopted a currencybasket system with undisclosed currency weights, but it was well known inthe market that the weight of the U.S. dollar in the basket was more than90 percent. Indonesia adopted the crawling peg against the U.S. dollar,where the rupiah depreciated with a schedule with a narrow band. Malaysiaadopted the managed float, but the movement of the ringgit was stableagainst the U.S. dollar. The movement of the Singaporean dollar and theKorean won was more flexible, compared to other currencies, respondingin part to the Japanese yen movement.

The actual weights, which may be quietly adjusted from time to time,can be estimated from actual movements of the exchange rates. Frankeland Wei (1994) estimated the weights of the U.S. dollar and the yenand other currencies in currency baskets of nine Asian currencies. Asiancurrencies (in terms of the Swiss franc) are regressed on the U.S. dollar(in terms of the Swiss franc) and the Japanese yen (in terms of theSwiss franc), for various subperiods from 1972–1992, with weekly data.


TABLE IIThe Exchange Rate Basketa

LHS Constant Dollar Yen R2/D.W.

Korean won 20.0007* 0.96** 20.1 0.82(0.0003) (0.02) (0.03) 1.94

Singaporean dollar 0.0003** 0.75** 0.13** 0.86(0.0002) (0.01) (0.02) 2.28

Hong Kong dollar 20.0007* 0.92** 20.00 0.76(0.0003) (0.02) (0.03) 2.04

Taiwan dollar 0.0005* 0.96** 0.05* 0.88(0.0002) (0.02) (0.02) 2.07

Malaysian ringgit 20.0003 0.78** 0.07** 0.79(0.0003) (0.02) (0.02) 2.19

Indonesian rupiah 20.0018* 0.95** 0.16* 0.44(0.0007) (0.05) (0.07) 2.04

Philippine peso 20.0018** 1.07** 20.01 0.54(0.0006) (0.04) (0.06) 2.06

Thai baht 20.0004 0.91** 0.05** 0.75(0.0003) (0.02) (0.03) 2.24

Chinese yuan 20.0018 0.87** 20.04 0.54(0.0005) 0.04 (0.05) 2.05

Note. Standard errors are in parentheses.Source. Frankel and Wei (1994).a Sample period, 1979–1992.** Statistically significant at 99% level.* Statistically significant at 95% level.

They showed that the U.S. dollar had an overwhelming weight in allcurrencies in all periods.

According to Frankel and Wei, the U.S. dollar weight was typically about95% for the Korean won; that is, if the U.S. dollar depreciated by 1% vis-a-vis the Swiss franc, the Korean won depreciated 0.95% that week. TheHong Kong dollar, Taiwan dollar, Indonesian rupiah, Philippine peso, andThai baht had a coefficient on the U.S. dollar higher than 90%. For themovement of the Singaporean dollar, the weight of the U.S. dollar wasabout 75%, significantly lower than the Korean won. Similarly, the Malay-sian ringgit moves with the U.S. dollar by 78%. In the case of the Chineseyuan, the coefficient is 0.87. A major part of the Frankel and Wei study issummarized in Table II.

Frankel and Wei examined whether adding other currencies, DM, Aus-tralian dollar, and New Zealand dollar, would change the results. For theSingaporean dollar, when these currencies were added, the coefficientsof the U.S. dollar and the Japanese yen were reduced to 0.71 and 0.12,respectively, and the coefficient of DM is 0.14, and that of the NZ dollar


0.02, both being statistically significant. Similarly for the Malaysian ringgit.(It is also possible that the Singaporean dollar was the genuine basketcurrency, and the Malaysian ringgit was pegged to the Singaporean dollar.)For other currencies, one cannot reject a hypothesis that coefficients ofDM, the Australian dollar, and the NZ dollar were statistically significantlydifferent from zero.

From Frankel and Wei, we may conclude that only the Singaporeandollar and the Malaysian ringgit had a genuine basket system. The formerput to the U.S. dollar a weight of 70%, and the Japanese yen and theGerman mark splitting the rest. The weights were similar for the Malaysianringgit. Other than these currencies, the Asian currencies were de factopegged to the U.S. dollar, its weight in the basket being about 90% orhigher.

6.2. Trade Linkages to United States and Japan

One of the reasons for the de facto dollar peg is that a large part ofexports is destined for the United States, and the dollar denomination isconvenient for avoiding fluctuation in export competitiveness. However,Japan is also a major trading partner for Asian countries. In fact, for mostAsian economies, Japan and the United States are equally important asexport destinations. Many Asian countries import parts and semifinishedgoods from Japan. As the nominal or real exchange rate vis-a-vis the U.S.dollar has been stabilized by the exchange rate policy, the real effectiveexchange rate vis-a-vis the Japanese yen fluctuates widely as the Japaneseyen fluctuates vis-a-vis the U.S. dollar.

Table III shows exports and imports of NIEs (except Taipei, China),ASEAN-4 (Thailand, Indonesia, Malaysia, and Philippines), and China bytheir destination or origin. Several features stand out. First, Japan is asimportant as the United States as a trade partner for most Asian countries.The trade linkage to Japan is especially strong for the ASEAN-4.

Second, Japan tends to be more important as an import origin than asan export destination. Except for Indonesia and India, the import sharefrom Japan is higher than the export share. The export share of Japan islower than the export share of the United States for all countries in thetable, except for Indonesia and China. The import share of Japan is higherthan that of the United States for all countries except for Korea (in 1996only).

Third, the Asian developing countries collectively are also an importanttrading partner. For Hong Kong, Malaysia, and Singapore, about one halfof their exports go to Asian developing countries. For Thailand and thePhilippines, the share going to Asian developing countries has increasedmarkedly from 1990 to 1996, reflecting that the trade linkage is now strongeramong the ASEAN-4.


TABLE IIIaNIES Gross Exports and Imports, by Country and Region

1990 1991 1992 1993 1994 1995 1996

Hong Kong exports and imports (US$mil, and %)Exports to

World 82,160 98,577 119,512 135,248 151,395 173,754 180,745United States 24.1% 22.7% 23.1% 23.0% 21.9% 21.8% 21.2%Japan 5.7% 5.4% 5.2% 5.1% 5.6% 6.1% 6.5%Asia 40.7% 41.9% 43.5% 45.0% 45.6% 46.5% 47.2%

Imports fromWorld 82,474 100,255 123,430 138,658 161,777 192,777 198,560United States 8.1% 7.6% 7.4% 7.4% 7.1% 7.7% 7.9%Japan 16.1% 10.4% 17.4% 16.6% 15.6% 14.8% 13.6%Asia 59.6% 61.1% 60.0% 60.5% 61.5% 61.3% 62.0%

Korea exports and imports (US$mil, and %)Exports to



Singapore, exports and imports (US$mil, and %)Exports to



Source. International Monetary Fund, Direction of Trade Statistics, Yearbook, 1997.

6.3. Real Exchange Rates and Export Volume

In order to make observations on the relationship between the exchangerate movement and export performance, the real exchange rate of countryA (say, Thailand) vis-a-vis the Japanese yen, say the baht–yen real exchangerate, or RER(A/Y), is calculated using the baht–yen nominal exchangerate, or (A/Y), and the CPI of country A, CPIA, and the CPI of Japan,


TABLE IIIbASEAN Exports and Imports, by Country and Region

1990 1991 1992 1993 1994 1995 1996

Hong Kong exports and imports (US$mil, and %)Exports to



Indonesia, exports and imports, (US$mil, and %)Exports to



Malaysia, exports and imports (US$mil, and %)Exports to



Philippines, exports and imports (US$mil, and %)Exports to



Source. International Monetary Fund, Direction of Trade Statistics Yearbook, 1997.


or CPIJ. Therefore, RER(A/Y) 5 (A/Y)*(CPIJ)/(CPIA). The real ex-change rate vis-a-vis the U.S. dollar, RER(A/D), is similarly defined,RER(A/D) 5 (A/D)*(CPIU)/(CPIA), where CPIU is the CPI of theUnited States.

In Figs. 1–7, the real exchange rates, RER(A/Y) and RER(A/D), areplotted against the export volume, in bars, for each of the NIEs (exceptHong Kong) and the four ASEAN countries, with quarterly frequencyfrom 1981 : 1 to 1997 : 2. The levels of the real exchange rates are measuredwith normalization where the last observation equals one. The export vol-ume is measured as the change over the preceding four quarters.

In general, the real exchange rate vis-a-vis the U.S. dollar, RER(A/D),is stable, reflecting the dollar peg policy. However, it is also evident thatoccasional devaluation (vis-a-vis the U.S. dollar) took place, such as inThailand in 1984 and Indonesia in 1983 and 1986. The degree of stabilityvaries from country to country. As the RER–U.S. movement is not volatile,its correlation with export volume is not apparent. The trend of real appreci-ation of their currencies vis-a-vis the U.S. dollar mainly reflects the inflationdifferential. The real exchange rate vis-a-vis the U.S. dollar, RER(A/D),for Thailand and Indonesia has been remarkably stable since the mid-1980s.

The real exchange rate vis-a-vis the yen, RER(A/Y), fluctuates as theyen–dollar rate fluctuates. In some countries, strong correlation betweenRER(A/Y) and export volume is observed. In all countries, the yen appreci-ation (or depreciation of an Asian currency vis-a-vis yen) episodes of 1985–1987 and 1992–1995 are accompanied by a surge in export volume growthrates. Also, in many countries, the yen depreciation (or appreciation of anAsian currency vis-a-vis the yen) of 1989–1990 affected export volume ad-versely.

For Malaysia, the correlation between the RER(A/Y) and export volumewas weak in the 1980s but became stronger in the 1990s, because its exportstructure became much more industrial (especially, electronics) in the 1990s.The Indonesian export movement does not seem to be correlated with itsRER. This is because oil and other mineral resources have a large sharein the Indonesian trade structure, and their export performances, whichmay not be directly related to RER, have a large impact on the Indonesiangrowth of exports. It is more important to look at the non-oil-and-gasexports, in order to see the impact of the exchange rate.

Therefore, it is reasonable to conclude that the exchange rate movementvis-a-vis Japan and its effects on the export industry are important, andthis is a result of the policy that de facto fixed their exchange rate to theU.S. dollar.

In each figure, the spike of RER(A/Y) in the second quarter of 1995 isvery prominent. This reflects a sharp yen appreciation at the time and theAsian currencies’ de facto dollar peg. The yen had an appreciation trend


FIG

.1.K

orea

,rea

lex

chan

gera

te,

and

expo

rtvo

lum

e.


FIG

.2.T

aiw

an,r

eal

exch

ange

rate

,an

dex

port

volu

me.


FIG

.3.S

inga

pore

,re

alex

chan

gera

tevs

Uni

ted

Stat

es,

and

expo

rtvo

lum

e.


FIG

.4.T

haila

nd,

real

exch

ange

rate

,an

dex

port

volu

me.


FIG

.5.M

alay

sia,

real

exch

ange

rate

,an

dex

port

volu

me.


FIG

.6.P

hilip

pine

s,re

alex

chan

gera

te,

and

expo

rtvo

lum

e.


FIG

.7.I

ndon

esia

,rea

lex

chan

gera

te,

and

expo

rtvo

lum

e.


in the first half of the 1990s and climaxed at 80 yen–dollar in April 1995.The yen reversed its movement and depreciated to the 100 level in thefollowing five months, and further depreciated to 120 level in the followingtwo years. The yen appreciation in the first half of the 1990s helped exportsfrom most Asian countries. The subsequent yen depreciation significantlydampened exports from these countries, especially from Thailand.

7. REGRESSION ANALYSIS

7.1. Pass-through Equation

This section investigates quantitatively theoretical prediction of themodel (Section 5), with some prior knowledge of general movements ofexports from stylized facts (Section 6). First, the pass-through equation,the response of export price to the changes in the exchange rate, will beestimated. The innovative feature of our model is that there are two ex-change rates in the model. The ASEAN firm is expected to respond toboth the baht–yen exchange rate and the baht–dollar exchange rate. Con-sider the effects of the baht–yen exchange rate change and the baht–dollarexchange rate, in order. Second, the volume of exports will be regressedon the bilateral real exchange rates of the baht–yen and baht–dollar, andthe real income growth of Japan and the United States. The coefficientsof the exchange rates will capture the price elasticities of the demand forthe ASEAN goods, and the coefficients of the income variables will capturethe income elasticities of the ASEAN goods.

Pass through equation. The change in the real export price in localcurrency is regressed on the changes in the real exchange rates against thedollar and the yen.

D(Real export price in local currency(t)) 5 c0 1 c1 D(RER(A/D)(t))1 c2 D(RER(A/Y)(t)) 1 e(t),

where D(.) is the growth rate operator; export prices are divided bythe respective CPI; coefficients c1 and c2 represent the pass-throughcoefficients, or sensitivities, of the real export price to the changes inthe real exchange rate vis-a-vis the dollar and vis-a-vis the yen, respec-tively. When the export price trend is different from CPI, then it iscaptured in the constant term. For example, when the Balassa–Samuelsoneffect is present, that is, productivity growth is significantly differentnontradable and tradable sectors, the nontradable prices are likely toincrease faster than the tradable prices. If the pass-through coefficient


is unity, then the changes in the export price fully reflect the exchangerate changes; that is, 1% depreciation vis-a-vis the dollar will increasethe export price in local currency by 1% so that the export price in thedollar remains the same (a full pass-through). However, this would putthe exporters in the less competitive positions against the Japaneseexporters in the Japanese market, and, in case of model A, in the U.S.market as well.

Monthly data are collected either from IFS or national sources. Thechange was taken against the same month of the previous year. The movingaverage process in the error terms is dealt with by using the Newey–Westmethod. The sample period is from 1981–1996, except as noted. Table IVsummarizes the regression results. In general, the degree of export priceadjustment varies from one country to another.

In Thailand, where the sample period starts in 1986 to avoid the exchangerate regime change of 1984, the sensitivity of the export prices to the dollaris small and insignificant, while the sensitivity with respect to the yen isabout 0.14 and barely significant (at 10 percent).6 The constant term isinsignificant (no Balassa–Samuelson).

In Indonesia, where the sample period starts in 1988 to avoid the exchangerate regime in 1986, sensitivities with respect to the dollar and the yen areboth statistically insignificant.

The Korean case presents an interesting picture. Korean export pricesrespond to the dollar and the yen in a similar magnitude of sensitivity,0.12 and 0.13, respectively, but only the yen is statistically significant. Theconstant term is also significant, with negative coefficient, suggesting thatprices of the nontradables (contained in CPI) are increasing more than theexport prices (tradables).

The Taiwan case is rather different from the above countries in that thesensitivity of export prices with respect to the dollar movement is large,0.49. About half of the dollar fluctuation is absorbed by the export pricechanges. The sensitivity with respect to the yen is small and insignificant.

Singapore presents a puzzling case in that the sensitivity with respect tothe dollar is negative, large, and significant. A small Balassa–Samuelsoneffect is detected.

In the case of the Philippines, the sensitivity with respect to the yen islarge (0.41) and mildly significant (at 6%). A strong Balassa–Samuelsoneffect is detected. Note that the Balassa–Samuelson effect is a propositionon the relative magnitude of the two sectors. A large negative constant termis consistent with a case where nontradables are extremely unproductive andexperienced high inflation, while tradables are not.

6 The magnitude of the coefficient and t-statistics are only slightly different when the sampleperiod is extended back to 1981.


TABLE IVPass-Through Equations

Thailand—Period: 1986 : 01–1996 : 05—Data source of Export Price: IFS

With constant Without constant

Coef. T-value Signif. Coef. T-value Signif.

Constant 20.01 20.53 0.598 Dollar ex. 0.28 0.76 0.447Dollar ex. 0.03 0.06 0.954 Yen ex. 0.14 1.62 0.106Yen ex. 0.14 1.68 0.092R-bar-sq.: 0.086. R-bar-sq.: 0.080.

Indonesia—Period: 1988 : 1–1996 : 12—Data source of export price: Biropusat statiskik




Korea—Period: 1981 : 1–1996 : 12—Data source of export price: Bank of Korea




Taiwan—Period: 1985 : 12–1988 : 12, 1990 : 9–1996 : 12—Data source of export price:The Central Bank of China




Singapore—Period: 1981 : 1–1996 : 12—Data source of export price: IFS





TABLE IV—Continued

Philippines—Period: 1981 : 1–1991 : 12—Data source of export price: IFS




Note. D(Export Price in l.c.(t)) 5 c0 1 c1 D((A/D)(t)) 1 c2 D((A/Y)(t)) 1 e(t), whereD(*) is the difference over the same month of year t 2 1.

7.2. Volume Equation

The export volume is regressed on the real exchange rate changes againstthe U.S. dollar and the Japanese yen and real GDP growth rates of theUnited States and Japan. The latter terms represent export demand in-creases due to income increases in the United States and Japan.

D(Export volume (t)) 5 g1 D(RER(A/D)(t)) 1 g2 D(RER(A/Y)(t))1 g3 D(USGDP(t)) 1 g4 D(JAGDP(t)) 1 e(t),

where D(.) is the growth rate operator, RER(.) is the real exchange rateusing CPI, and USGDP and JAGDP denote real GDP of the United Statesand Japan, respectively. Data are quarterly, due to the availability of GDPdata. All variables are measured as the change over the preceding year(four quarters). The moving average process in the error terms are dealtwith by the Newey–West method.

Table V summarizes the results of this regression. There is no generalobservation. In all countries but Taiwan, the export volume has an increas-ing trend (a positive constant in the above regression with export volumeincreases on the left-hand-side variable).

For Thai export volume, the baht–yen rate is the only significant (at10%) variable. The elasticity of the export volume with respect to the yenmovement is estimated as 0.34. Recall that the Thai graph in the stylizedfacts section gave a general impression that the yen movement and theexport volume are correlated. This regression confirmed that observation.Although insignificant, the coefficient of the dollar movement is positive,suggesting that, for Thailand, Model B (Thai products compete with theU.S. products in the United States) is more consistent than Model A (Thaiproducts compete with Japanese products in the United States).


TABLE VExport Volume

Thailand—Period: 1986 : Q1–1996 : Q4—Data source ofexport volume: IFS

Coef. T-value Signif.

Constant 0.14 3.17 0.002Dollar ex. 2.42 1.45 0.148Yen ex. 0.34 1.87 0.061US–RGDP 0.43 0.36 0.718JP–RGDP 0.91 1.29 0.195R-bar-sq.: 0.126.

Indonesia—Period: 1988 : Q1–1996 : Q4—Data source ofexport volume: IFS



Korea—Period: 1981 : Q1–1996 : Q4—Data source ofexport volume: IFS



Taiwan—Period: 1985 : Q1–1996 : Q4—Data source ofexport volume: IFS




TABLE V—Continued

Singapore—Period: 1981 : Q1–1996 : Q4—Data source ofexport volume: IFS



Philippines—Period: 1981 : Q1–1996 : Q4—Data source ofexport volume: IFS



Note. D(Export volume (t)) 5 g1 (RER(A/D)(t)) 1 g2

D(RER(A/Y)(t)) 1 g3 D (USGDP(t)) 1 g4 D(JAGDP(t))1 e(t), where D(*) is the difference over the four quarters.

For Indonesian exports, the coefficient of the rupiah–dollar rate is nega-tive, suggesting that Model A is applicable, but neither the rupiah–yen northe rupiah–dollar exchange rate is statistically significant. The Japanesegrowth is mildly significant. This may reflect the fact that Japan is a numberone destination of Indonesian exports.

Korean exports are explained by the won/yen real exchange rate (signifi-cance at 10%) and U.S. economic growth (significance at 1%). This resultis consistent with theoretical predictions. The Korean exports are found tobe negatively affected by the won/U.S. dollar and Japanese growth rates,but the coefficients are statistically insignificant. Recalling the theoreticalprediction (Table I), this may mean that Korean exports face a situationsimilar to Model A (competing with Japanese exports in both Japaneseand U.S. markets, while importing parts from Japan).

Unlike Korea and Singapore, Taiwan exports are most significantly af-fected by the NT dollar–yen rate. The yen depreciation of 10% will decreasethe export volume by 3.5%. Other factors are insignificant.

Singaporean export volumes are explained by the Singapore dollar–U.S.dollar exchange rate, with negative coefficient, and the U.S. growth. The


Singaporean case, being consistent with Model A, indicates that Singa-porean exports have an industrial structure and technological levels similarto Japanese exports and that they are competing in the United States.

Philippines exports are shown to be very sensitive with both exchangerates. The peso–dollar rate has a negative coefficient, suggesting that ModelA is more applicable to the Philippines case. On the contrary, the exportvolumes from the Philippines increase with the yen devaluation. However,a puzzling feature is that the Philippine export volume declines when theU.S. economic growth rate increases.

In sum, export volume regressions present a reasonable result, with onlya few puzzling coefficients. For Thailand, Model B applies, and for othercountries, Model A seems to apply, but t-statistics of the exchange ratevis-a-vis the U.S. dollar are low, except for Singapore and the Philippines,so that the result may not be reliable.

7.3. Optimal Currency Weight

At the end of the theory section, a formula for the optimal currencybasket weight was proposed, Eq. (27) for Model A and Eq. (28) for ModelB. The optimal dollar peg weight in Model A (Eq. (27)) can be rewrittenby using coefficients a1 , a2 , b1 , b2 of regression equations (29)–(32).

w* 5TUS 1 a1 X 1 b1 T

T 1 (a1 1 a2 2 1)X 1 (b1 1 b2)T(33)

The optimal dollar peg weight in Model B (Eq. (28)) can be rewrittenby using coefficients a1 , a2 , b1 , b2 of regression equations (29)–(32). Forsimplicity, we assume that that gx and gm are nearly equal to each other.

w* 52MUS 1 a3 X 1 b3 T

T 1 (a3 1 a4 2 1)X 1 (b3 1 b4)T, (34)

where MUS denotes a value of imports from the United States.Now with the regression results in the preceding subsections, and the

bilateral trade balance shown in Table III, we can calculate the optimalweights according to Eq. (33) for Model A and Eq. (34) for Model B. TableVI shows the optimal weights for each country, for Model A (with a constantterm in price equation (A-1), and without a constant term (A-2)) andModel B (with a constant term in price equation (B-1), and without aconstant term (B-2)).

In general, estimates fall between 0 and 1, except the three subcases ofTaiwan and one subcase of the Philippines. According to our estimates, inall countries, the yen weight should be much higher than the actual weight


TABLE VIOptimal Weight

Optimal weightActualweight Model A-1 a Model A-2 b Model B-1 c Model B-2 d

US$ Yen US$ Yen US$ Yen US$ Yen US$ Yen(%) (%) (%) (%) (%) (%) (%) (%) (%) (%)

Thai baht 91 5 42.9 57.1 4.3 95.7 61.3 38.7 35.3 64.7Indonesian rupiah 95 16 40.5 59.5 47.7 52.3 71.2 28.8 77.9 22.1Korean won 96 210 10.5 89.5 10.9 89.1 47.4 52.6 45.7 54.3Taiwan dollar 96 5 92.7 192.7 273.7 173.7 25.3 105.3 7.3 92.7Singaporean dollar 75 13 22.6 77.4 12.4 87.6 57.4 42.6 51.0 49.0Philippine peso 107 21 22.9 102.9 27.6 72.4 67.3 32.7 72.8 27.2

Note. In calculating the export to and import from the world in Eqs. (27) and (28), only the sum ofthose with Japan and the United States are used. For actual weight, see Table II.

a Model A-1 uses the coefficients estimated in the case of price equations (model A) with constant inTable IV and the coefficients estimated in the volume equations in Table V.

b Model A-2 uses those in the price equations (model A) without constant in Table IV and those in thevolume equations in Table V.

c Model B-1 uses those in the price equations (model B) with constant in Table IV and those in thevolume equations in Table V.

d Model B-2 uses those in the price equations (model B) without constant in Table IV and those in thevolume equations in Table V.

estimated by Frankel and Wei, cited in Table II. Unfortunately, estimatesare sensitive to the choice of model (A-1, A-2, B-1, or B-2) for each country.Estimates naturally vary across countries. As we interpreted results ofvolume regressions, we may tentatively choose Model B for Thailand andModel A for all other cases. For Thailand, the optimal yen weight is estimatedanywhere between 39% (model B-1) and 65% (model B-2). For Indonesia,the yen optimal weight is between 52% (model A-1) and 60% (model A-2).For Korea, both A-1 and A-2 models indicate that the optimal yen weight is89%. Taiwan shows a puzzling case in which the optimal yen weight is morethan 100%. We suspect that coefficients in either price and volume equationsare somehow wrongly estimated for the Taiwan case, and we do not take theTaiwan result seriously.7 The optimal weight of the yen for the Singaporeandollar is between 77% (Model A-1) and 88% (Model A-2). The case of ModelA-1 for the Philippines is puzzling in that the weight of the yen exceeds 100%.Again, we do not take this case seriously. However, in the case of A-2, theoptimal yen weight is calculated as 72%.

7 The optimal weights of all cases of Taiwan and some cases of the Philippines seemcounterintuitive. The price and volume equations may be misspecified so that our estimatesmay not be 100% accurate. We tested the robustness of the optimal peg estimates for Taiwanand the Philippines. Instead of point estimates, those with plus or minus one standard deviationwere used to check what variables may be responsible for the result. To our surprise, noreason was uncovered by this simulation exercise.


In sum, the optimal weight of the yen is highest in Korea and Singapore.Since these countries are often thought to have industrial structure andassociated technological levels similar to Japan, their exports directly com-pete with Japanese products in Japan and the United States. Therefore, theresults of the high optimal yen weights in the two countries look reasonable.

Of course, we are aware that many of the coefficients in the price andvolume equations which are used in calculating the optimal weights havelow t-statistics. Robustness of our results with respect to theoretical modelspecification and variables in the regressions are not as ideal as we desire.However, the result shows an important avenue for further research onthe optimal currency weight in Asia.

8. CONCLUSION

In this paper, we have constructed a theoretical model in which the Asianfirm maximizes its profit, competing with the Japanese and the U.S. firmsin their markets. The duopoly model is used to determine export pricesand volumes in response to the exchange rate fluctuations vis-a-vis theJapanese yen and the U.S. dollar. Then, the optimal basket weight thatwould minimize the fluctuation of the growth rate of trade balance wasderived. These are the novel features of our model.

The export price equation and export volume equation are estimated forseveral Asian countries for the sample period from 1981 to 1996. Results aregenerally reasonable. The optimal currency weights for the yen and the U.S.dollar are derived and compared with actual weights that had been adoptedbefore the currency crisis of 1997. For all countries in the sample, it is shownthat the optimal weight of the yen is significantly higher than the actualweight.

So, how did the dollar peg fail in Asia? In this paper, the basket peg is called‘‘optimal’’ when it minimizes variation of profits which equals net exports tothe Japanese and U.S. markets, each of which is a duopoly market competingagainst the Japanese goods or the U.S. goods. In the optimal peg of the typicalAsian country, say Thailand, the weight of the yen was higher than the actualweight. In fact, the actual weight of the U.S. dollar always exceeds the dollar’soptimal weight, and the actual weight of the Japanese yen is always less thanthe optimal weight. Under the optimal peg, the currency, baht, would haveappreciated (vis-a-vis the U.S. dollar) when the yen appreciated, so that theboom would have been less; and the currency would have depreciated whenthe yen depreciated, so that weakening in exports would have been mitigated.Under the dollar peg, which the Asian countries had adopted, an amplifiedboom and bust of exports created irreversible redundancy in real estates andother industries. Thus the sharp decline in exports made investors and specu-lators convinced that the dollar peg would be unsustainable.


It would be interesting to extend our model so that multiple Asiancountries are included. In such a model, Asian exporters compete withother (non-Japan) Asian exporters, as well as with Japanese and U.S.producers, in the world market. The cross exchange rates (e.g., the baht–rupiah rate) would become relevant in such a model, and the optimalcurrency basket include those cross rates. Then, we can investigate issuessuch as competitive devaluation, contagion of a currency crisis, a benchmarkbasket, and the optimal weights of the benchmark basket. However, thesetopics are left for future research.

APPENDIX

Model A

We use Eqs. (7a) and (8a) to derive rates of changes in the prices of the ASEAN productsin the Japanese and U.S. markets in Model A, respectively.

PYJ 5

1 1 h*J

1 1 hJ 1 h*J 5(1 2 gm)EA/$P$

m

EA/Y

TCA(EA/$ 2 EA/Y) 2

C9

EA/Y

TCAEA/Y6

51 1 h*J

1 1 hJ 1 h*Jh(1 2 gm)B1(EA/$ 2 EA/Y) 2 B2EA/Yj

51 1 h*J

1 1 hJ 1 h*J(1 2 gm)B1 EA/$ 2

1 1 h*J

1 1 hJ 1 h*Jh(1 2 gm)B1 1 B2j EA/Y (A1)

P$US 5 EA/Y 2 EA/$ 1

1 1 h*US

1 1 hUS 1 h*US 5(1 2 gm)EA/$P$

m

EA/Y

TCA(EA/$ 2 EA/Y) 2

C9

EA/Y

TCAEA/Y6

5 EA/Y 2 EA/$ 11 1 h*US

1 1 hUS 1 h*US

h(1 2 gm)B1(EA/$ 2 EA/Y) 2 B2EA/Yj

5 H 1 1 h*US

1 1 hUS 1 h*US

(1 2 gm)B1 2 1J EA/$

1 H1 21 1 h*US

1 1 hUS 1 h*US

h(1 2 gm)B1 1 B2jJ EA/Y (A2)

PAx 5

XJ

X(PY

J 1 EA/Y) 1XUS

X(P$

US 1 EA/$)

5 F 1 1 h*1 1 h 1 h*

(1 2 gm)B1GEA/$ 1 F1 21 1 h*

1 1 h 1 h*h(1 2 gm)B1 1 B2jGEA/Y (A3)


We use Eqs. (10a) and (10b) to derive rates of changes in the outputs exported to theJapanese and U.S. markets, respectively:

f 52«J

1 1 hJ 1 h*J 5(1 2 gm)EA/$P$

m

EA/Y

TCAEA/$ 2

(1 2 gm)EA/$P$

m

EA/Y 1C9

EA/Y

TCAEA/Y6

52«J

1 1 hJ 1 h*J[(1 2 gm)B1 EA/$ 2 h(1 2 gm)B1 1 B2j EA/Y] (A4)

g 52«US

1 1 hUS 1 h*US 5(1 2 gm)EA/$PY

m

EA/Y

TCAEA/$ 2

(1 2 gm)EA/$PY

m

EA/Y 1C9

EA/Y

TCAEA/Y6

52«US

1 1 hUS 1 h*US

[(1 2 gm)B1 EA/$ 2 h(1 2 gm)B1 1 B2j EA/Y] (A5)

From Eqs. (A3) and (A4), we derive the rate of changes in the total outputs of theASEAN firm:

QA 5 gx f 1 (1 2 gx)g

52«J

1 1 hJ 1 h*J[(1 2 gm)B1 EA/$ 2 h(1 2 gm)B1 1 B2j EA/Y]

12«US

1 1 hUS 1 h*US

(1 2 gx)[(1 2 gm)B1 EA/$ 2 h(1 2 gm)B1 1 B2j EA/Y] (A6)

5 2 H «J

1 1 hJ 1 h*Jgx 1 ? ? ?

«US

1 1 hUS 1 h*US

(1 2 gx)J[(1 2 gm)B1 EA/$ 2 h(1 2 gm)B1 1 B2j EA/Y]

From Eq. (21), we obtain the rate of changes in the trade balance of the ASEAN economy:

T 5 HEA/YPYJ f

T2

EA/YPYmgmQA

T J EA/Y 1 HEA/$P$US g

T2

EA/$P$m(1 2 gm)QA

T J EA/$

1EA/YPY

J fT

(PYJ 1 f) 1

EA/$P$US g

T(P$

US 1 g) 2EA/YPY

mgm QA 1 EA/$P$m(1 2 g)QA

TQA

T 5 H fQA

EA/YPYJ QA

T2

EA/YPYmgmQA

T J EA/Y 1 H gQA

EA/$P$US QA

T2

EA/$P$m(1 2 gm)QA

T J EA/$

1f

QA

EA/YPYJ QA

T(PY

J 1 f) 1g

QA

EA/$P$US QA

T(P$

US 1 g)

2EA/YPY

mgm QA 1 EA/$P$m(1 2 g)QA

TQA (A7)

T 5QA

T[hgx EA/YPY

J 2 gmEA/YPYmj EA/Y 1 h(1 2 gx)EA/$P$

US 2 (1 2 gm)EA/$P$mj EA/$

1 gx EA/YPYJ (PY

J 1 f) 1 (1 2 gx)EA/$P$US(P$

US 1 g) 2 hgm EA/YPYm 1 (1 2 gm)EA/$P$

mjQA]

Substituting Eqs. (A1)–(A5) into Eq. (A6), we obtain Eq. (22).


Model B

We use Eqs. (17a) and (18a) to derive rates of changes in the prices of the ASEAN productsin the Japanese and U.S. markets in Model B, respectively.

PYJ 5

1 1 h*J

1 1 hJ 1 h*J 5(1 2 gm)EA/$P$

m

EA/Y

TCA(EA/$ 2 EA/Y) 2

C9

EA/Y

TCAEA/Y6

51 1 h*J

1 1 hJ 1 h*Jh(1 2 gm)B1(EA/$ 2 EA/Y) 2 B2 EA/Yj

51 1 h*J

1 1 hJ 1 h*J[(1 2 gm)B1 EA/$ 2 h(1 2 gm)B1 1 B2j EA/Y]

51 1 h*J

1 1 hJ 1 h*J(1 2 gm)B1 EA/$ 2

1 1 h*J

1 1 hJ 1 h*Jh(1 2 gm)B1 1 B2j EA/Y (A1)

P$US 5

1 1 h*US

1 1 hUS 1 h*US 5EA/Y 2 EA/$ 1

(1 2 gm)EA/$P$

m

EA/Y

TCA(EA/$ 2 EA/Y) 2

C9

EA/Y

TCAEA/Y6

51 1 h*US

1 1 hUS 1 h*US

hEA/Y 2 EA/$ 1 (1 2 gm)B1(EA/$ 2 EA/Y) 2 B2EA/Yj

51 1 h*US

1 1 hUS 1 h*US

[2h1 2 (1 2 gm)B1j EA/$ 1 h1 2 (1 2 gm)B1 2 B2j EA/Y]

5 21 1 h*US

1 1 hUS 1 h*US

h1 2 (1 2 gm)B1j EA/$ 11 1 h*US

1 1 hUS 1 h*US

h1 2 (1 2 gm)B1 2 B2j EA/Y

(A29)

PAx 5

XJ

X(PY

J 1 EA/Y) 1XUS

X(P$

US 1 EA/$)

5 F 1 1 h*1 1 h 1 h*

(1 2 gm)B1 1h

1 1 h 1 h*XUS

X G EA/$

1 F1 21 1 h*

1 1 h 1 h*h(1 2 gm)B1 1 B2j 2

h

1 1 h 1 h*XUS

X G EA/Y (A39)

We use Eqs. (20a) and (20b) to derive rates of changes in the outputs exported to theJapanese and U.S. markets, respectively:

f 52«J

1 1 hJ 1 h*J 5(1 2 gm)EA/$P$

m

EA/Y

TCAEA/$ 2

(1 2 gm)EA/$P$

m

EA/Y 1C9

EA/Y

TCAEA/Y6

52«J

1 1 hJ 1 h*J[(1 2 gm)B1 EA/$ 2 h(1 2 gm)B1 1 B2j EA/Y] (A4)


g 52«US

1 1 hUS 1 h*US 51(1 2 gm)EA/$PY

m

EA/Y

TCA2 12 EA/$ 1 11 2

(1 2 gm)EA/$PY

m

EA/Y 1C9

EA/Y

TCA 2 EA/Y65

2«US

1 1 hUS 1 h*US

[2h1 2 (1 2 gm)B1j EA/$ 1 h1 2 (1 2 gm)B1 2 B2j EA/Y] (A59)

From Eqs. (A3) and (A49), we derive the rate of changes in the total outputs of theASEAN firm:

QA 5 gx f 1 (1 2 gx)g

52«J

1 1 hJ 1 h*Jgx[(1 2 gm)B1 EA/$ 2 h(1 2 gm)B1 1 B2j EA/Y]

12«US

1 1 hUS 1 h*US

(1 2 gx)[2 h1 2 (1 2 gm)B1j EA/$ 1 h1 2 (1 2 gm)B1 2 B2j EA/Y]

5 F «US

1 1 hUS 1 h*US

(1 2 gx) 2 H «J

1 1 hJ 1 h*Jgx 1

«US

1 1 hUS 1 h*US

(1 2 gx)J (A69)

(1 2 gm)B1G EA/$

1 F 2«US

1 1 hUS 1 h*US

(1 2 gx) 1 H «J

1 1 hJ 1 h*Jgx 1

«US

1 1 hUS 1 h*US

(1 2 gx)Jh(1 2 gm)B1 1 B2jG EA/Y

Substituting Eqs. (A1), (A29), (A3), (A49), and (A59) into Eq. (A6), we obtain Eq. (229).

REFERENCES

Bhandari, J. S. (1985). Experiments with the optimal currency composite, Southern Econ. J.51(3), 711–730.

Flanders, M. J., and Tishler, A. (1981). The role of elasticity optimism in choosing an optimalcurrency basket with applications to Israel, J. Int. Econ. 11, 395–406.

Flanders, M. J., and Helpman, E. (1979). An optimal exchange rate peg in a world of generalfloating, Rev. Econ. Stud. July, 533–542.

Frankel, J. A., and Wei, S.-J. (1994). Yen bloc or dollar bloc? Exchange rate policies of theEast Asian economies, in ‘‘Macroeconomic Linkage: Savings, Exchange Rates, and Capi-tal Flows’’ (T. Ito, and A. O. Krueger, Eds.), University of Chicago Press, Chicago.

International Monetary Fund, ‘‘Direction of Trade Statistics Yearbook,’’ 1997.

Kenen, P. B. (1969). The theory of optimal currency area: An eclectic view, in ‘‘MonetaryProblems of the International Economy’’ (R. Mundell and A. K. Swoboda, Eds.), Univer-sity of Chicago Press, Chicago.


Knetter, M. M. (1989). Price discrimination by U.S. and German exporters, Amer. Econ. Rev.79(1), 198–210.

Knetter, M. M. (1993). International comparisons of pricing-to-market behavior, Amer. Econ.Rev. 83(3), 473–486.

Krugman, P. (1987). Pricing to market when the exchange rate changes, in ‘‘Real-FinancialLinkages among Open Economies’’ (S. W. Arndt and J. D. Richardson, Eds.), MIT Press,Cambridge, MA.

Lipschitz, L., and Sundararajan, V. (1980). The optimal basket in a world of generalizedfloating, IMF Staff Papers 27(1), 80–100.

McKinnon, R. (1963). Optimum currency area, Amer. Econ. Rev. 53 (September), 717–725.

Marston, R. C. (1990). Pricing to market in Japanese manufacturing, J. Int. Econ. 29(3/4), 217–236.

Mundell, R. A. (1961). A theory of optimum currency areas, Amer. Econ. Rev. 51(Septem-ber), 657–665.

Ohno, K. (1989). Export pricing behavior in manufacturing: A U.S.–Japan comparison, IMFStaff Papers 36(3), 550–579.

Turnovsky, S. J. (1982). A determination of the optimal currency basket, J. Int. Econ. 12,333–354.

how did the dollar peg fail in asia?

Documents