What should be weights on the three Titlecurrencies for a common currency ba Asia?

Transcription

1 What should be weights on the the Titlecurncies for a coon curncy ba Asia? Author(s) Ogawa, Eiji; Kawasaki, Kentaro Citation Issue Date Type Technical Report Text Version publisher URL Right Hitotsubashi University Repository

2 No.6 What should be weights on the the ajor curncies for a coon curncy basket in East Asia? Eiji Ogawa and Kentaro Kawasaki Deceber 2003 Hitotsubashi University Research Unit for Statistical Analysis in Social Sciences Institute of Econoic Research Hitotsubashi University Kunitachi, Tokyo, Japan

3 What should be weights on the the ajor curncies for a coon curncy basket in East Asia? Eiji Ogawa a and Kentaro Kawasaki b First version: 6 March 2003 This version: 11 August 2003 This paper is ppad for the Regies and Surveillance in East Asia Confence which is held in Kuala Lupur on March We thank Yu Yongding for providing us Chinese data. a Graduate School of Coerce and Manageent, Hitotsubashi University. e-ail: ogawa.eiji@srv.cc.hit-u.ac.jp. b Faculty of Business Adinistration, Toyo University. e-ail: kawasaki-k@toyonet.toyo.ac.jp.

4 1. Introduction The Governents of East Asian countries have coe to take a positive stance for a gional financial cooperation since they experienced the Asian curncy crisis in They a establishing a network of bilateral curncy swap arrangeents aong ASEAN+3 (China, Japan, and Koa) according to the Chiang Mai Initiatives. Moover, a study on a gional coon curncy was started aong the ASEAN countries. We should consider what type of gional coon curncy it is desirable to introduce into East Asia in the futu because East Asian countries experienced the curncy crisis to learn a lesson that the onetary authorities should not adopt the de facto dollar peg syste in order to pvent another curncy crisis in the futu. It follows that it should not be desirable to establish a gional coon curncy that is based on the US dollar as an anchor curncy. Soe epirical searches found that a curncy basket syste would contribute to stabilizing trade balances and capital flows in East Asian countries. Ito, Ogawa, and Sasaki (1998) estiated optial weights on the US dollar and the Japanese yen in a curncy basket, which would have stabilized trade balances in East Asian countries befo the Asian curncy crisis. Results of the estiation showed that the optial weights on the US dollar we saller than their actual weights that we estiated by Frankel and Wei (1994) and Kawai and Akiyaa (1998). It iplies that it was not a de facto dollar peg syste but a curncy basket peg syste that would have stabilized their trade balances. Ogawa and Sun (2001) siulated capital inflows to the the crisis-hit countries, which include Thailand, Indonesia, and Koa, under a curncy basket peg syste whe weights on the US dollar and the Japanese yen we assued to have 50%: 50% in a curncy basket. Results of the siulation concluded that the de facto dollar peg syste stiulated capital inflows to the crisis-hit countries befo the Asian curncy crisis. Ogawa and Ito (2002) and Ogawa (2002b) pointed out that the onetary authorities in East Asian countries have been unwilling to adopt the de facto dollar peg syste due to a kind of coordination failu in choosing exchange rate syste. Accordingly, any for of international policy coordination for their choosing desirable exchange rate syste is necessary. For exaple, all of the onetary authorities in East Asian countries ight age on an arrangeent that they cate a -1-

5 coon curncy unit as an anchor curncy for their exchange rate syste or a fence for their exchange rate policy. They ight ake fences to the coon curncy unit in choosing exchange rate syste or conducting their exchange rate policy. A rigid arrangeent is that all of the onetary authorities peg their hoe curncies to the coon curncy unit. On one hand, one of o flexible arrangeents is that they target the hoe curncies in a wider band around a central exchange rate of the hoe curncies vis-à-vis the coon curncy unit. Given that a curncy basket syste would contribute to stabilizing trade balances and capital flows in East Asian countries, we consider that the coon curncy unit should be denoinated in ters of a curncy basket in order to solve the coordination failu in choosing exchange rate syste. Thus, this paper investigates what type of coon curncy basket should be adopted in East Asian countries fro a viewpoint of a long run sustainability of adopting the coon curncy basket in East Asia while taking into account stabilization of trade balances at the sae tie. We use a Generalized Purchasing Power Parity (G-PPP) odel to analyze the issue 1. We investigate which part of East Asian countries will be able to cate a coon curncy basket and what weights should be on the the ajor curncies, which include the US dollar, the euro (the ECU), and the Japanese yen. We further extend an earlier work of Ogawa and Kawasaki (2003) in order to analyze epirically a coon curncy basket aa for East Asia while taking into account the stability of trade balances. In the earlier work, we defined the coon curncy basket as the one that was consisted of the sae weights of the the ajor curncies (the US dollar, the ECU, and the Japanese yen) in the coon curncy basket. We investigated the possibilities to for a coon curncy aa in East Asia by using the Generalized Purchasing Power Parity (G-PPP) approach. We found that the coon curncy basket was o applicable to cate a coon curncy aa than the US dollar. At the sae, we found two divided groups whose countries included all of the East Asian countries but we not overlapped each other for the curncy basket as an anchor curncy. 1 Bayoui, Eichengen, and Mauro (2000) used a structural VAR odel to ake an epirical analysis on an optiu curncy aa in East Asia. However, they cannot show what type of gional coon curncy is optial for East Asian countries. -2-

6 The coon curncy basket, which we assued in the earlier work, ight not be optial for East Asia although a coon curncy basket is, in itself, o desirable as an anchor curncy. Rather, we should define a trade weighted curncy basket as a coon curncy basket for the stabilization of trade balances. Thefo, we use it as a coon curncy basket to investigate a long run sustainability of adopting the coon curncy basket in East Asia in this paper. East Asian countries (he we investigate the issues for ASEAN5 countries, China, and Koa) ay not have the sae long run stationary lationship when we define the trade weighted coon curncy basket as an anchor curncy. This has two iplications. One is that the East Asian countries a divided into two or o groups due to the diffences in technology growth or other factors aong the groups. For the ason, we attept to find soe groups that show a long run stationary lationship. The other is that the trade-based weights or alternative weights on the the ajor curncies in the coon curncy basket ay not stabilize the syste in the long run because the de facto dollar peg syste had been in fact adopted in ost of East Asian countries especially befo the Asian curncy crisis. Thefo, our intests turn into the estiation for the basket weights on the the ajor curncies which would be able to ake the syste of a coon curncy basket aa stable in the long run. We attept to develop the ethod of estiation for endogenous weights in the coon curncy basket. It is iportant to copa the endogenous weights with the trade-based weights. This paper consists of five sections. In the following section we extend our G-PPP odel in order to investigate long run equilibriu of the syste in using the coon curncy basket as an anchor curncy. In section 3, we analyze the long run sustainability of the coon curncy basket aa that includes o than five East Asian countries when a coon curncy basket is cated by placing trade weights on the the ajor curncies. In section 4, we show our ethodology of estiating the endogenous weights and analyze the possibilities of cating a coon curncy basket with endogenous weights on the the ajor curncies. In section 5, we ake concluding arks including suary of our analytical sults and their policy iplications for introducing a coon curncy basket into East Asia. -3-

7 2. G-PPP odel 2.1 The al effective exchange rates and Generalized PPP Now we suppose that the a countries which a expected to adopt a coon curncy basket as an anchor curncy. Country j has n trade partners. It has strong trade lationships with countries which adopt the sae curncy basket with country j while it has also trade lationships with the other countries. Thefo, we can define a al effective exchange rate of country j, (countries 1, 2, L, j, L, + 1 have the coon curncy basket while countries + 2, L, n a one who do not sha the coon curncy basket) denoted with curncy of country j, e j = ξ j ( j,1 j,1 + j,2 j,2 +L + j, +1 j, +1 ) +(1 ξ j ) ( j, +2 j, +2 +L + j,n j,n ) (2-1) 1 n whe j,i ( j,i = 1, j,i = 1) is the country j s trade weights on country i and ξ i =1 i = +2 is the trade weights on a group of countries who sha the coon curncy basket. He, we assued that the shocks on the second ter in the right hand side of equation (2-1) affects the al effective rate of country j teporally, or even if these shocks a peranent, these shocks affect + 1 al effective rates syetry. In the case whe only country j is peranently affected by the countries who do not adopt the coon curncy basket as an anchor curncy, it is difficult to keep adopting a coon curncy in the gion. The a no asons for country j to stay in the coon curncy aa. He, we define the al effective exchange rate in ters of a curncy of country + 1, which excludes teporal shocks in the country j. e = ξ j, t j,1( j,1, t j, 1, t ) j, 1( j, 1, t j, 1, t ) = j,1 1,1, t + L+ j,1 + L+ 1,, t 1, j, t + j, 1, t (2-2) whe j,k = j,n k,n = n, j + n,k. We can write + 1 al effective rates in ters of the curncy of country + 1 in the sae ways, -4-

8 e e e e ξ 1, t ξ 2, t ξ, t ξ 1, t = = = = 2,1,1 1, j, t 1,1 1,1, t 1,1, t + 1,1, t 1,2 1,2, t 1,2, t + L+ + L+ L+, 1 + L+ 1, 1 2, 1, 1, 1, t 1,, t 1, 1, t 1,, t M. 1,, t + 1, 1,, t These + 1 al effective rates can be shown as the atrix Ω which defines the trade weights, and the vector which includes eleents of the al exchange rate 1,i as below, e t = Ω t (2-3) whe Ω ( 1) 1 2,1 = M,1 1,1 1,2 1 M,2 1,2 L L L L L 1, 1 2, 1 M, 1 1, 1 1, 2, M 1 1, and the vector e includes the + 1 al effective rates. Each of the al effective exchange rates is expected to include a coon stochastic tnd because the countries have strong trade lationships with each other and they see to sha coon technologies. 2 We assue that the + 1 al effective exchange rates sha a coon stochastic tnd. Using Stock and Watson s (1988) coon tnd psentation for any cointegrated syste, we can show that the vector e which is characterized by cointegrating lations, can be described as the su of a stationary coponent and a non-starionary coponent. e t = e t + r e e t (2-4) 2 Enders and Hurn (1994) developed the G-PPP odel based on the al fundaental acroeconoic variables. They assued these variables shad coon tnds within a curncy aa. -5-

9 The stationary coponent e t is E( e t ) = 0 in this odel since the logarith of the al effective exchange rate can be expected to converge toward zero-ean in the long run. Thefo, the vector e can be only described as the non-stationary coponent r e e. By the definition of coon tnd in Stock and Watson (1988), we obtain the following equation: e t = Φ w t (2-5) whe Φ is the ( + 1) ( + 1) atrix. The vector w t is the non-stationary stochastic tnd which is characterized by a rando walk. We substitute equation (2-5) into equation (2-3), then we obtain the following equation: Φ w t = Ω t. (2-6) He, we define the non-null atrix Ψ which is coposed of ( + 1) ( + 1) and write equation (2-6) to obtain the following equation: Ψ Φ w t = Ψ Ω t (2-7) If the exists a nonzero w for which Ψ Φ w t = 0, Ψ Φ does not have a full rank. The rank condition will be expected as follows: rank(ψ Φ) = rank(φ) < As long as the rank condition holds, the exists a non-null atrix Ψ which satisfies the following equation: Ψ Φ = 0 (2-8) When we set Ζ = Ψ Ω and substitute it into equation (2-7), we obtain the following equation: Ζ = 0 (2-9) If we could find a atrix Ζ, which satisfies rank(ζ) < and equation (2-9), it eans that -6-

10 the exists nonzero for Ζ = 0 and that the atrix Ψ is not a null atrix. Accordingly, the nuber of rank Ω ust be saller than. He, we assue that rank(ζ) = 1. We can write equation (2-9) to obtain the following linear cobination: ζ 1 +1,1 + ζ 2 +1,2 +L + ζ +1, = 0 (2-10) This linear cobination is the sae as that of Enders and Hurn (1994). Thefo, we can use Johansen and Juselius (1990) s ethod to estiate the cointegrating vector. 2.2 G-PPP odel evaluated by the curncy basket in seven Asian countries In the pvious section, we introduced the G-PPP odel based on the al effective exchange rates. The iportant featu of our G-PPP odel is that a coon curncy is denoinated in ters of an anchor curncy. This featu is very useful to investigate what types of coon curncy in ters of an anchor curncy have the long run stationary lationship when soe countries for an optiu curncy aa with an anchor curncy. Kawasaki (2000) and Ogawa and Kawasaki (2001) investigated the optiu curncy aa for ASEAN5 and Koa evaluated by a single curncy: the US dollar, the Japanese yen, the Deutche Mark, and the Singapo dollar. In this paper, we extend this featu into the investigating a coon curncy basket as an anchor curncy in any group of East Asian countries. He, we suppose that seven East Asian countries (ASEAN5 + Koa + China) cate a coon curncy basket as an anchor curncy whe a coon curncy basket is coposed of the the ajor curncies: the US dollar, the ECU (the euro), and the Japanese yen. We define an exchange rate of country i in ters of the coon curncy basket as follows: RE CB,i = (RE EU,i ) α (RE JP,i ) β (RE US,i ) γ, α + β + γ = 1. (2-11) whe RE is the al exchange rate and (α, β,γ ) a the weights on the the ajor curncies. Equation (2-11) is written in ters of the logarith: CB,i = α EU,i + β JP,i + γ US,i. (2-12) -7-

11 whe is the logarith of bilateral al exchange rate. Again, if all of the seven East Asian countries a included in an optiu curncy aa, we write equation (2-10) to obtain the following equation: 3 ζ 1 CB,1 + ζ 2 CB,2 +L +ζ CB, = 0. (2-13) Then they can establish a coon curncy basket aa. A usual cointegration fraework can estiate the cointegration vector in equation (2-13) like equation (2-10). When we use the Johansen and Juselius (1990) s ethod to estiate the cointegrating vector, given the weights on the the ajor curncies in the coon curncy basket, endogenous variables in the -diensional vector autogssive odel a defined by X = [ CB,1, CB,2,L, CB, ]. In the case whe the is at least one cointegration lationship between endogenous variables, the -diensional vector autogssive odel can be written according to an Error Corction Model (ECM) as follows: k X t = 1 i= 1 Λ X i + Π X + ε t 1 t 1 t, (2-14) We test a hypothesis that the duced rank of the Π atrix is H 1 (r):π = υ ζ. whe υ is the loading atrix. The duced rank r is the nuber of cointegration lationships. Hence, the is a long run equilibriu aong bilateral exchange rates if we obtain the stability of Π atrix or, in other words, a fact that the rank is at least none zero. 3 We still assue + 1 countries cate coon curncy aa as sae as in our G-PPP odel of section 2. The host curncy needs to be included in the curncy aa in the sense of Mundell (1964). He, we just convert the host curncy into the curncy basket. -8-

12 3. Epirical analysis on a coon curncy basket with exogenous weights 3.1 Epirical ethodology In this section, we further extended the earlier work of Ogawa and Kawasaki (2003) in order to analyze epirically a coon curncy basket aa for East Asia. Although we still define that a coon curncy basket is coposed of the the ajor curncies (the US dollar, the ECU, and the Japanese yen), we suppose that weights in the coon curncy basket a given by trade weights (exports and iports) on the United States, Japan, and the euro aa. We investigate a coon curncy basket aa with o than five East Asian countries: ASEAN 5, ASEAN 5 + Koa, ASEAN 5 + China, and ASEAN 5 + Koa + China. We conduct the Johansen test for each of the cobinations. Our epirical analysis using the Johansen cointegrating fraeworks follows the arbitrary strategy in Ogawa and Kawasaki (2003) to iprove robustness Data The saple for our epirical tests covers fro the period between January 1981 and June Seven East Asian countries we included Koa, Singapo, Malaysia, Thailand, the Philippines, Indonesia, and China. Real exchange rates we based on onthly data of noinal exchange rates and consuer price indices of the lated countries. 6 These data a fro the IMF, International Financial Statistics (CD-ROM). 7 The exports and iports data a fro the IMF, Diction of Trade Statistics (CD-ROM). The value of trade weights on the the ajor curncies shows in the Table 1. Table 1.1 shows the each of the East Asian countries trade weights on trade 4 To iprove the robustness for the Johansen cointegration fraework is that we should choose a lag length by taking into account whether the equilibriu of that odel is adequate for the cointegration lationship or not. See details of our strategy to define the unique odel in the Appendix of Ogawa and Kawasaki (2003). 5 We can get the Chinese trade data after January 1981 fro DOT. 6 For the ECU al exchange rates, we calculated a GDP-weighted average of CPI. 7 The Chinese consuer price index is provided by Yu Yongding, the Chinese Acadey of Social Sciences (CASS). -9-

13 partners; the United States, Japan, the euro aa, intra-gional trade in East Asia, and st of the world. Table 1.2 shows the agggate trade weights on the trade partners and trade-based weights on the the ajor curncies in the coon curncy basket. 3.3 Epirical Results Table 2 shows the sult of Johansen test; λ -trace and λ -ax tests after selecting the corct lag-order of all of the ECM. The cobination of ASEAN5 was not passed the p-test for the Johansen test which could not ject the null of auto corlation of the siduals in each VAR odel. We excluded this case fro our analysis in this section. While we could not find any cointegration lationships in the cobination of ASEAN5 + Koa, We could find the several cointegrating vectors in the cobination of ASEAN5 + China and the cobination of ASEAN5 + Koa + China. Table 3 shows the sult of the the chi-squa-based test identified as the optial odel. The first row for each vector shows the test statistics for the null hypothesis of ζ ij = 0 (1 j r ). The second row shows test statistics for the stationarity. This test is to check whether the individual series can be stationary by theselves. The null is ζ = (H i,ϕ). The third low shows test statistics for the weak exogeneity for the long run equilibriu. The null hypothesis is υ ij = 0 (1 j r ). He, we should only focus on the cobinations in which all countries in a linear cobination have significant sults on those the tests, because we need to specify the inial cobination for the coon curncy basket aa. Fro the sult of Table 3, we could find that ASEAN5 + China could for an optiu curncy aa by using the coon curncy basket with the agggated trade weights on the the ajor curncies at sae tie. On the other hand, the cobination of ASEAN + Koa + China include the country whose statistics for the chi-squa tests we insignificant, e.g. the variable Singapo ight be excluded fro the cointegration lationship and the variable of Indonesia ight indicate exogenous. It eans that it is difficult for the cobination of ASEAN5 + Koa + China to for an optiu curncy aa by using the coon curncy basket with the agggated trade weights on the the ajor curncies at sae tie. -10-

14 4. Epirical analysis on a coon curncy basket with endogenous weights 4.1 Estiation of endogenous weights We show the ethodology of estiating endogenous weights on the the ajor curncies in the coon curncy basket. We still assue that + 1 East Asian countries for a coon curncy basket aa and their al exchange rates have at least one cointegration lationship. Hence, equation (2-13) will hold in the long run. He, we write equation (2-12) as follows: CB, i = EU, i + β JP, i + ( 1 α β ) US, i α. (4-1) By substituting equation (4-1) into equation (2-13), we obtain the following equation: ζ { α 1 + ζ { α 2 + L+ ζ EU,1 EU,2 { α + β EU, JP,1 + β + (1 α β ) JP,2 + β + (1 α β ) JP, US,1 } US,2 } + (1 α β ) US, } = 0 Then, ζ { α ( 1 + ζ 2 + L + ζ EU,1 { α ( EU,2 { α ( US,1 EU, ) + β ( US,2 ) + β ( US, JP,1 JP,2 ) + β ( US,1 JP, ) + US,2 US,1 } ) + US, US,2 } ) + US, } = 0 Hence, ζ { α 1 + ζ { α 2 + L+ ζ EU, US EU, US { α + β EU, US JP, US + β JP, US + + β US,1 + JP, US } US,2 } + US, } = 0 α ( ζ + ζ + L + ζ ) 1 2 EU, US + ζ 1 + β ( ζ + ζ + L + ζ ) US, ζ 2 US,2 + L + ζ JP, US US, = 0 (4-2) whe the + 2 diensional vector autogssive odel a defined by -11-

15 X = [, 1,,2,,,,,,, ] US US L US EU US JP US, We conduct the Johansen test to obtain estiated values for the + 2 eleents of the Ζ [ cointegrating vector = ζ, ζ, L, ζ, ζ, ζ ] if rank(ζ) = 1, we can get the only one linear cobination written in the for of the equation (4-2) fro the Johansen fraework, whe ξ α ζ + ζ + L ζ ), β ζ + ζ + ζ ) 1 = ( ξ L. We use the estiated value 2 = ( to calculate the estiated optial weights as follows: α = ζ + ζ 1 ζ L+ ζ, β = ζ + ζ 1 ζ L + ζ. (4-3) 4.2 Epirical ethodology We attept to estiate the endogenous weights in the coon curncy basket dictly. Especially, we conduct the Johansen test for the long run stationary lationship aong o than five East Asian countries. We set the four cobinations to test: ASEAN5, ASEAN5 + Koa, ASEAN5 + China, and ASEAN 5 + Koa + China. He, we test for the null hypothesis of coefficients in the Johansen fraework. In the case of ζ 1 = 0 or ζ 2 coon curncy basket equal to zero. When ζ 1 = 0, the weights on the Japanese yen or the ECU in the = 0 and ζ 2 will be coposed of the US dollar and the ECU. When ζ 1 0, the coon curncy basket 0 and ζ 2 = 0, the coon curncy basket will be coposed of the US dollar and the Japanese yen. If both of the cointegrating paraeters a zero (ζ 1 = ζ +2 = 0 ), all of the curncies peg to the US dollar. Under the null hypothesis that the Japanese yen or the ECU is excluded fro the curncy basket, the test statistic is asyptotically distributed as χ 2 with r deges of fedo in the cointegration analysis. 4.3 Data The saple for our epirical tests covers the period between January 1981 and June

16 Seven East Asian countries a included Koa, Singapo, Malaysia, Thailand, the Philippines, Indonesia, and China. The al exchange rates we based on the onthly data of noinal exchange rates and consuer price indices of the lated countries. 8 These data a fro the IMF, International Financial Statistics (CD-ROM) Analytical sults Table 4 shows the sult of the Johansen test; λ -trace and λ -ax tests after selecting the corct lag-order of all of the ECM. The cobination of ASEAN5 + China could not ject the null hypothesis of the auto corlation in the siduals of any VAR odels which have 2 to 12 lag-length. Thefo, we excluded this cobination fro our analysis. We could find several cointegration lationships in the any cobinations when we conducted the Johansen test. Table 5 shows the sult of the the chi-squa-based tests that includes the sult of test for the null hypothesis of coefficients of the Japanese yen and the ECU. The first row for each vector shows the test statistics for the null hypothesis of ζ ij = 0 ( i = 1, 2, K,7,US/JP, US/DM, 1 j r ). He, we cafully focus on the case that shows significant sults on the long run exclusion test and the stationary test for all of the East Asian countries at least. For the cobination of ASEAN5, since the test statistics of Indonesia in the first cointegrating vector is insignificant, we should focus on the second-, the third-, and the forth-cointegrating vector. For the cobination of ASEAN5 + Koa, we should focus on the second-, the third-, the forth-, and the fifth-cointegrating vector. For the cobination of ASEAN5 + Koa + China, we should focus on the fifth-cointegrating vector. Table 6 shows each value in the cointegrating vectors. Table 7 shows calculated weights on the the ajor curncies in the coon curncy basket. For the cobination of ASEAN5, the coefficient of the Japanese yen in the third cointegrating vector is On the other hand, that of the ECU is and the test statistics is significant. Thefo, we can calculate the weight on the 8 For the ECU al exchange rates, we calculated a GDP-weighted average of CPI. 9 The Chinese consuer price index is provided by Yu Yongding, the Chinese Acadey of Social Sciences (CASS). -13-

17 Japanese yen as 16.4%, the weight on the ECU as 13.2%, and the weight on the US dollar as 70.4%. For the second and forth vector, we obtained the unexpected signs and values of the weights so that this cointegrating vector ay not be proper for the equilibriu of the coon curncy basket aa. For the cobination of ASEAN5 + Koa, the coefficient of the Japanese yen in the second-cointegrating vector is and that of the ECU is Both of test statistics a significant. Thefo, the weight on the Japanese yen is 16.1%, the weight on the ECU is 8.2%, and the weight on the US dollar is 75.8 %. For the fifth-vector, the weight on the Japanese yen is 12.6%, the weight on the ECU is 10.3%, and the weight on the US dollar is 77.1%. For Koa and/or Indonesia, however, the test statistics for the exogenous test we insignificant in both cases. The weights calculated fro the third and the forth-cointegrating vector showed the unexpected signs. For the cobination of ASEAN5 + Koa + China, only the fifth cointegrating vector indicates significant sults on the long run exclusion test and the stationary test on seven East Asian countries. But the weights calculated fro fifth-cointegrating vector showed the unexpected signs. Our analytical sults showed that the a no significant sults when China was included in the coon curncy basket aa. On the other hand, for the cobination of ASEAN5 or ASEAN5 + Koa, we could find the non-zero weights on the Japanese yen or the ECU in the coon curncy basket. Soe countries in the cointegrating vector showed the exogeneity for the long run lationship. It eans that these countries affect the long run lationship aong other countries but do not converge toward the equilibriu by theselves. Although a long run lationship is not stable for the East Asian coon curncy aa, the calculated weights on the US dollar, the Japanese yen and the ECU a at least useful to discuss. 5. Conclusion This paper investigated what type of coon curncy basket is desirable for East Asian countries fro a viewpoint of a long run sustainability of adopting the coon curncy basket in East Asia, given that the coon curncy unit should be denoinated in ters of a curncy basket in order to solve the coordination failu in choosing exchange rate syste and that a curncy basket syste would contribute to stabilizing trade balances and capital flows in East Asian -14-

18 countries. When we investigated the coon curncy basket aa which included o than five East Asian countries using the trade weighted coon curncy basket of the the ajor curncies, we found the cobination of ASEAN5 + China could for the coon curncy basket aa with the the ajor curncies. It eans that the trade weights on the the ajor curncies a optiized for the coon curncy basket aa in ASEAN5 + China. When we dictly estiate the endogenous weights on the the ajor curncies in the coon curncy basket, the weight on the US dollar in the coon curncy basket a lager than that of the weights based on the trade volues of seven East Asian countries with the United States. The larger weight (but not 100%) on the US dollar in the coon curncy basket tends to ake the bilateral exchange rates aong East Asian countries stationary in the long run. This sees the ason why we cannot find the long run lationship aong o than five East Asian countries in our earlier works. On the other hand, fro our epirical analysis using the p-crisis data, we could find that the larger weight on the US dollar in the coon curncy basket also can ake the bilateral exchange rates aong East Asian countries stable in the long run. If the onetary authorities eploy the dollar peg syste for their exchange rates policy, they do not need a policy coordination aong East Asian countries any o. Because of the p-crisis data, we ight obtain the analytical sult that the larger weights on the US dollar than the trade weights in the coon curncy basket a so suitable that East Asian curncies should be stabilized the cointegration syste for this saple period. The de facto dollar peg syste in ost of East Asian countries befo the Asian curncy crisis sees to flect in the analytical sults. Recently the onetary authorities in soe East Asian countries have decased the linkage of their hoe curncy to the US dollar. Moover, they would decase further the linkage to the US dollar if they ade gional onetary coordination to adopt a coon curncy basket that is desirable in ters of the stability of the trade balances. -15-

19 Refences Bayoui, T., B. Eichengen, and P. Mauro (2000) On gional onetary arrangeents for ASEAN, CEPR Discussion Paper, No Enders, W. and S. Hurn (1994) Theory and tests of generalized purchasing-power parity: Coon tnds and al exchange rates in the Pacific Ri, Review of International Econoics, vol. 2, no. 2, Frankel J. A. and S. Wei (1994) Yen bloc or dollar bloc? Exchange rate policies of the east Asian econoies, in T. Ito and A. O. Krueger, eds., Macroeconoic Linkage: Savings. Exchange Rates and Capital Flows, Chicago, University of Chicago Pss, Ito, T., E. Ogawa, and N. Y. Sasaki (1998) How did the dollar peg fail in Asia? Journal of the Japanese and International Econoies, vol. 12, Johansen, S. and K. Juselius (1990) Maxiu likelihood estiation and infence on cointegration; with application to the deand for oney, Oxford Bulletin of Econoics and Statistics, vol. 52, no. 2, Kawai, M. and S. Akiyaa (1998) The role of noinal anchors curncies in exchange arrangeents, Journal of the Japanese and International Econoies, Vol. 12, pp Kawasaki, K. (2000) A test of OCA in Asian curncy aa: Epirical analysis based on G-PPP Theory, The Hitotsubashi Review, vol. 124, No. 6, (in Japanese). Kawasaki, K., (2002) Give a New Life to the PPP Theory: Modifying the Generalized PPP Model, (ieo). Mundell, Robert, (1964) A theory of optiu curncy aas, Papers and Proceedings of the Aerican Econoic Association, vol. 51, Ogawa, E. (2002a) Should East Asian countries turn to dollar peg again? P. Drysdale and K. Ishigaki eds., East Asian Trade and Financial Integration: New Issues, Canberra: Asia Pacific Pss. Ogawa, E. (2002b) Econoic interdependence and international coordination in East Asia, in Exchange Rate Regies for Asia (Kobe Research Project), Ministry of Finance -16-

20 ( Ogawa, E. and T. Ito (2002), On the desirability of a gional basket curncy arrangeent, Journal of the Japanese and International Econoies., vol. 16, Ogawa, E. and K. Kawasaki (2001), Toward an Asian Curncy Union, a paper ppad for the 2001 KIEP/NEAEF Confence on Stngthening Econoic Cooperation in Northeast Asia held in Honolulu, Hawaii, on August Ogawa, E. and K. Kawasaki (2003), Possibility of Cating a Coon Curncy Basket for East Asia, JBICI Discussion Paper, Japan Bank for International Cooperation (forthcoing). Ogawa, E. and L. Sun (2001) How we capital inflows stiulated under the dollar peg syste? in T. Ito and A. O. Krueger eds., Regional and Global Capital Flows: Macroeconoic Causes and Consequences, University of Chicago Pss, Chicago. Stock, J., and M. Watson (1988) Testing for coon tnds, Journal of the Aerican Statistical Association, vol. 83,

21 Table1.1:Trade Weights of East Asian Countries 1981:1-1998:12 China Indonesia Koa Malaysia Trade Partner United States 12.38% 15.00% 25.07% 16.84% Japan 20.41% 31.66% 20.12% 20.51% EU Aa 10.20% 11.08% 8.85% 10.41% East Asia (Coon Aa) 7.31% 19.42% 9.07% 28.16% Rest of the World 49.70% 22.84% 36.90% 24.08% 1981:1-1998:12 Trade Partner The Philippines Singapo Thailand United States 27.26% 17.53% 15.39% Japan 18.87% 14.41% 22.17% EU Aa 10.38% 9.14% 13.22% East Asia (Coon Aa) 13.51% 25.78% 18.18% Rest of the World 29.98% 33.15% 31.04% Table1.2:Trade Weights of the Coon Curncy Aa 1981:1-1998:12 Coon Aa Outside Aa Basket Weight Trade Partner United States 18.09% 21.56% 37.18% Japan 20.59% 24.48% 42.22% EU Aa 10.00% 11.94% 20.60% East Asia (Coon Aa) 16.07% Rest of the World 35.24% 42.02%

22 Table 2: Cobination ASEAN5 +Koa ASEAN5 +China ASEAN5 + Koa + China Johansen tests Exogenous Weights Eigen Vector L-Max L-Trace k H *** *** *** *** *** *** ** *** *** *** ** *** * k: lag lengths, upper is for the curncy basket, lower is for the US dollar *95%, **97.5%, ***99.0% Cobination ASEAN5 + China Table 3: ASEAN5 + China + Koa Chi-squa based Tests CHISQ Koa k r DGF 10% 5% 2.5% 1% (Won) Singapo ($SG) Indonesia (Rupiah) Malaysia (Ringgit) **** **** **** **** **** **** **** **** **** **** **** **** ** 6.68 * 10.8 *** 7.84 ** *** **** **** * **** **** **** 6.52 ** **** **** **** 32.7 **** **** **** **** **** **** **** 8.55 *** **** 8.22 *** Test statistics indicate for "long-run exclusion"(upper), "stationarity"(iddle), and "weak exogeneity"(lower) spectively. The Philippines Thailand (Baht) China (Yuan)

23 Cobination Singapo + Indonesia + Malaysia + The Philippines + Thailand Koa + Singapo + Indonesia + Malaysia + The Philippines + Thailand Koa + Singapo + Indonesia + Malaysia + The Philippines + Thailand + China k: lag lengths *:95%, **:97.5%, ***:99% Table 4: Johansen tests Eigen Endogenous Weights k H 0 Vector L-Max L-Trace *** *** *** *** *** *** *** * *** *** *** *** *** * *** * *** * *** *** *** *** *** *** *** *** *** *** *** *** *** * ***

24 ASEAN 5 ASEAN 5 + Koa ASEAN 5 + Koa + China Table 5: Chi-squa based Tests for Estiation of Endogenous Weights CHISQ Koa r DGF 10% 5% 2.5% 1% (Won) Singapo ($SG) Malaysia (Ringgit) **** 5.66 *** 5.76 *** 6.72 **** **** **** **** **** **** **** **** *** **** **** **** **** **** 12 **** 7.77 *** **** 9.02 *** 7.91 *** 6.97 ** **** **** **** **** **** **** **** **** **** 4.74 * **** **** **** **** **** **** 19.9 **** **** **** 8.59 ** *** 8.89 * ** *** **** *** *** **** **** **** **** **** **** **** **** **** **** **** **** *** 8.16 ** *** **** 15 **** **** **** **** **** **** **** **** 25.3 **** *** **** **** 8.19 **** *** **** 40.6 **** **** 39.5 **** **** 42.4 **** **** **** ** 6.71 **** 8.09 **** *** 4.65 ** **** **** **** 8.63 *** **** **** 9.36 **** 7.75 *** **** **** **** **** **** **** **** **** ** **** 9.38 **** 9.21 *** *** 6.27 ** **** **** **** **** 31.4 **** **** 17.7 **** **** **** **** **** **** **** **** **** ** **** 21.5 **** **** **** **** *** **** **** **** **** **** **** **** **** *** *** **** **** *** **** **** **** **** **** **** **** **** **** **** 38.8 **** **** **** **** **** **** ** *** **** 12.4 **** *** **** 13.4 **** **** **** 31.5 **** **** 33.1 **** **** **** * 4.81 ** *** **** **** **** **** **** 37.2 **** **** **** 17.9 *** *** ** ** *** **** 6.07 ** **** 5.15 * **** **** **** **** **** **** **** **** * *** 5.78 * * 6.74 ** * *** **** 7.18 * **** *** **** **** **** **** **** **** **** ** 6.97 * ** *** **** **** **** *** **** 9.02 * **** **** **** **** **** **** **** **** *** *** 11.3 *** *** **** 8.23 * 9.14 * 14.8 **** ** **** **** **** **** **** **** 9.35 * 9.4 * ** 18.5 **** **** **** **** **** **** 16.2 **** **** **** **** ** **** **** 9.53 * **** 12.6 ** **** **** **** 30 **** **** 24.1 **** **** ** * *** **** **** 12.8 **** **** **** 7.14 * **** **** **** * **** **** * 18.2 **** 17.1 **** Test statistics indicate for "long-run exclusion"(upper), "stationarity"(iddle), and "weak exogeneity"(lower) spectively. The Philippines Thailand (Baht) Indonesia (Rupiah) China (Yuan) US-JP US-ECU

25 ASEAN 5 ASEAN 5 + Koa ASEAN 5 + Koa + China Table 6: Estiated cointegration vector (transposed) The r Koa Singapo Malaysia Philippines Thailand Indonesia China US-JP US-ECU * * * * * * * * * ** ** * * * * * ** ** * * * * * ** ** * * * ** * * * * * * ** ** * * * * * * ** ** * * * * * * ** ** * * * * * * ** ** * * * * * * * ** * * * * * * * * * * ** * * * * * * * ** ** * * * * * * ** ** *:Fro Table 5, the null hypothesis of the long-run exclusion and the stationarity for the coefficients of seven East Asian coutries a jected. **:Fro Table 5, ζ US-JP or ζ US-ECU 0 ASEAN 5 ASEAN 5 + Koa ASEAN 5 + Koa + China Table 7: Endogenous weights (noralized) alpha beta gaa Results r ( ) (ECU) ($) ** ** ** *: α, β, γ 0, ζ i 0 (i=1,,7, US-JP, US-ECU)