Introduction Directory UMM :Data Elmu:jurnal:I:International Review of Economics And Finance:Vol9.Issue4.2000:

International Review of Economics and Finance 9 2000 323–350 Choice of currency basket weights and its implications on trade balance Hsiang-Ling Han Economics Division, Babson College, Wellesley, MA 02457-0310, USA Received 14 August 1998; revised 10 June 1999; accepted 23 August 1999 Abstract This paper seeks to find an optimal choice of currency basket weights for emerging econo- mies that peg their currencies to a currency basket, and to examine the long-run relationship between the real exchange rates of a group of trading partners. A general equilibrium model is set up to establish an optimal set of currency basket weights, coupled with the choice of fiscal policy, to simultaneously stabilize trade balance and aggregate price level of an economy. This optimal set of weights is a weighted average of two sets of weights; each targets at one policy goal stabilizing either balance of trade or aggregate price level at a time. Empirical studies including vector autoregression VAR analysis and cointegration analysis on the long- run relationship between the Thai baht and the real exchange rates of its major trading partners are presented.  2000 Elsevier Science Inc. All rights reserved. JEL classification: F31; C32 Keywords: Currency basket; Vector autoregression; Cointegration

1. Introduction

The Dow Jones Industrial Average lost more than 7 percent of its value on October 28, 1997. It was preceded by a 5.8 percent plunge in the Hong Kong Stock Market following months of speculation and devaluation of Southeast Asian currencies. Many people believe that the downward spiral originated in the Thai government’s insistence on pegging the Thai baht to the U.S. dollar. The domino effect eventually reached South Korea and Japan. South Korea’s currency, the won, lost the maximum daily trading limit, 10 percent of its value, on November 20, 1997. The demise of a powerful Japanese broker, Yamaichi Securities Co., on November 24, 1997, propelled the Asian currency and financial crises to another climax. Corresponding author. Tel.: 781-239-5851; fax: 781-239-5239. E-mail address : hanbabson.edu H.-L. Han 1059-056000 – see front matter  2000 Elsevier Science Inc. All rights reserved. PII: S1059-05600000060-5 324 H.-L. Han International Review of Economics and Finance 9 2000 323–350 The debate on the optimal exchange rate policy that prevents this type of financial crisis from happening has been going on for decades. The focus of this paper, however, is on the choice of currency basket weights and its implications on trade balance for emerging economies. It is known that most of the Southeast Asian countries that encounter currency crises peg their currency to a basket dominated by the U.S. dollar. The paper, using Thailand as an example, investigates whether the choice of currency basket weights is optimal and the relation of the currency basket weights with its balance of trade. A general equilibrium model is set up to establish an optimal choice of currency basket weights, coupled with an optimal fiscal policy, to simultaneously stabilize overall balance of trade and price level of an economy. The model takes multilateral trade flows as well as macroeconomic policy targets into consideration. Long-run relation- ships between the Thai baht and the real exchange rates of its major trading partners are then explored. Thailand is used to conduct the empirical work of the paper because it is believed that this wave of currency crisis was originated there. It is interesting to analyze and to examine how the currency basket weights chosen by the Thai government affect the value of the Thai baht, their trade balance, and their current account balance, even though those weights are not officially publicized. Two approaches are used to analyze the long-run relationships between the Thai baht and the real exchange rates of its major trading partners. Vector autoregression VAR analysis is used to estimate a system of interrelated real exchange rates and to analyze the dynamic impact of random disturbances on the system of variables. The impulse- response function is presented to trace the effect of a shock to a real exchange rate on current and future values of another real exchange rate. The second approach, focusing on the long-run cointegrating relationship between these real exchange rates, includes canonical cointegrating regression CCR and Johansen’s LR test. Most tests for cointegration [e.g., Engle Granger 1987 and Phillips Ouliaris 1988] took the null hypothesis of no cointegration. Failures to reject the null hypothesis of no cointegration are often interpreted as evidence against economic models thaty cointegration. This type of methodology is known to have relatively low power. The two methodologies used in this paper, however, are relatively powerful and have good small sample properties. CCR proposed by Park 1992 tests the null hypothesis of cointegration. Han 1996 investigated the small sample properties of the CCR tests and concluded the tests perform well with reasonable size and power even when the samples are small. The Johansen’s LR test is now commonly used, and numerous researches have proven it to be a reasonable testing method for cointegration. This paper is organized as follows. Section 2 reviews the literature on the choice of exchange rate regimes. Section 3 presents a general equilibrium model that estab- lishes an optimal set of weights for a currency basket, coupled with an optimal fiscal policy, to reach two economic policy goals at the same time. Prior to conducting the empirical analysis for the Thai baht, section 4 discusses the evolution of the exchange rate system in Thailand that provides the background understanding of the Thai crisis. Section 5 presents two sets of empirical results for real exchange rates all against the U.S. dollar of Thailand, Japan, Germany, the Netherlands, and Singapore. The H.-L. Han International Review of Economics and Finance 9 2000 323–350 325 first set of results includes a vector autoregression VAR, variance decomposition, and impulse-response function analysis. The second set of results includes the test for cointegration and the estimation of cointegrating vector for the group of currencies. Section 6 is the concluding remarks.

2. Literature review