Journal of Asian Economics xxx (2005) xxx xxx. Risk properties of AMU denominated Asian bonds. Junko Shimizu, Eiji Ogawa *

Transcription

1 1 Journal of Asian Economics xxx (2005) xxx xxx Risk properties of AMU denominated Asian bonds Abstract Junko Shimizu, Eiji Ogawa * Graduate School of Commerce and Management, Hitotsubashi University, 2-1 Naka, Kunitachi, Tokyo , Japan Received 16 November 2004; received in revised form 27 May 2005; accepted 1 June 2005 This paper aims to investigate the risk properties of Asian Monetary Unit (AMU) denominated Asian bonds by comparing them with those of local currency denominated bonds issued in East Asian countries. We suppose the AMU as an Asian currency unit which is formed as a currency basket of East Asian currencies. In this paper, we simulate a currency basket composed of the ASEAN5 countries, Japan, China, Korea, and Hong Kong. Our results indicate that the AMU denominated bonds can lower the risks for both US and Japanese investors, because the portfolio effects of the AMU reduce foreign exchange risk. However, these results depend on the currency system in the East Asian countries. # 2005 Published by Elsevier Inc. JEL classification: F31; F33; G15 Keywords: Asian bond; A currency basket; Asian Monetary Unit (AMU); Foreign exchange risk 1. Introduction After experiencing the Asian Currency Crisis in 1997, East Asian countries have recognized the underlying problems caused by a double over-dependence on the banking sector in their financial systems and on the US dollar in their currency systems. We propose to establish an Asian bond market denominated in terms of, not the US dollar, but a basket currency as the method of simultaneously solving both of these problems. The monetary * Corresponding author. Tel.: ; fax: address: ogawa.eiji@srv.cc.hit-u.ac.jp (E. Ogawa) /$ see front matter # 2005 Published by Elsevier Inc. doi: /j.asieco

2 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx authorities of East Asian countries, including Japan, are actually promoting the Asian Bond Market Initiative, that is, an international cooperation to establish Asian bond markets in East Asia. Recent discussions of the Asian Bond Market Initiative have focused on the choice of denomination currency and the credit guarantee and rating agency. Ito (2003) proposed an Asian bond which is designed to be a fund of the local currency denominated bonds issued by governments of East Asian countries. However, there is another choice which is denominated in terms of the US dollar. Actually, as an initiative of the EMEAP, the Asian Bond Fund (ABF) was launched in its first version (ABF1) in June 2003 as a basket of US dollar denominated bonds issued by Asian sovereign and quasi-sovereign issuers in EMEAP economies (excluding Australia, Japan and New Zealand). 1 The EMEAP has worked to extend the ABF concept to bonds denominated in local currencies and has announced the launch of the second stage of ABF (ABF2) in December In previous research, Ogawa and Shimizu (2002, 2004) compared Asian bonds denominated in terms of the G3 currency basket (a currency basket of the US dollar, the euro, and the Japanese yen) with single currency (the US dollar, the euro, the Japanese yen and seven East Asian currencies) denominated bonds. In this paper, we investigate the risk properties of Asian Bonds denominated in terms of another currency basket, the Asian Monetary Unit (AMU) which is formed as a currency basket of ASEAN + 4 (Japan, China, Korea, and Hong Kong). As Kawai, Ogawa, and Ito (2004) noted, an AMU composed of a currency basket of ASEAN + 3 (Japan, China, and Korea) currencies would be expected to be equivalent to the G3 currency basket if each of the East Asian countries uses the G3 currency basket as a reference or target in conducting their exchange rate policies. The G3 currency basket denominated Asian bonds and AMU denominated Asian bonds would therefore be equivalent, even though this is not the case under the current circumstances. The remainder of this paper consists of the following sections. Section 2 briefly summarizes the preceding literature on this area. Section 3 describes our method to simulate the risks of AMU denominated bonds and compares the risk properties between AMU denominated bonds with each of the local currency denominated bonds issued in the East Asian countries. Section 4 considers foreign exchange risk reduction through internationally diversified investments and the effectiveness of AMU denominated bonds in reducing foreign exchange risk. The final section offers concluding remarks. 2. Contagion of financial crises and Asian bond markets One of the main features of financial crises, especially those that occurred across the world in the 1990s, is that they tend to spill over from one country to neighboring countries. This was the case in the Asian Crisis, were contagion throughout the regional economies was fuelled by strong trade linkages among the East Asian countries. At the international level, some emphasize the imperfections of the international financial markets as one of the major causes of contagion (Kaminisky, Lyons, & Schmukler, in press; Krugman, 1999). 1 Executives Meeting of East-Asia and Pacific Central Banks (EMEAP) is a cooperative organization of central banks and monetary authorities in the East Asia and Pacific region.

3 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx 3 One approach to explain how contagion operates in financial crises uses the finance theory of portfolio management. Schinasi and Smith (1999) used a model of the basic portfolio theory to explain contagion selling of higher risk assets. They showed that a shock to a single asset s return distribution may lead to a reduction in position of other risky assets. 2 In addition, Mello and Hussein (2001) used a simple model of foreign debt portfolio management to explain that the currency composition of a country s foreign debt portfolio was responsive to exchange rate movements. They conducted an empirical analysis on the existence of a stable long-run relationship between the currency composition of a country s foreign debt portfolio and exchange rate movements in the case of fourteen emerging market economies and four major currencies (the US dollar, the Deutsche mark, the Japanese yen, and the Swiss francs) during an analytical period of Their results indicated that foreign debt portfolios were not managed optimally for the investors, in the sense that adverse movements in exchange rates were not offset by an increase in portfolio shares of debts denominated in terms of appreciating currencies. 3 The East Asian currency crisis of has created a strong awareness of the necessity for regional cooperation, and from this have developed several policy proposals in order to strengthen regional financial cooperation. Kuroda and Kawai (2003) proposed a more effective surveillance process and considered the option of creating a common pool of foreign exchange reserves to allow flexible financial supports at times of crises and contagion, which would also reduce the problem of moral hazard. Monetary authorities in East Asian countries recently established the Chaing Mai Initiative, that is, a swap agreement in order to prevent future currency and financial crises. 4 The establishment and development of regional bond markets in East Asia have been proposed in order to directly circulate regional savings into regional investments in East Asia. In addition, the possibility of introducing a common currency or a common currency basket unit as a new regional financial architecture in East Asia is also often discussed. 5 In order to assess the effectiveness of basket currency denominated bonds in East Asian countries, Ogawa and Shimizu (2004) conducted an empirical analysis from the standpoint of the bond issuer s side. A comparison was made between the G3 currency basket (a currency basket of the US dollar, the euro, and the Japanese yen) denominated bonds and each of the single local currency denominated bonds of East Asian currencies and three major currencies in terms of both relative risk and liquidity. The results showed that issuing currency basket denominated bonds contributes to reducing foreign exchange risk for bond issuers in East Asian countries, except for the dollar-pegging countries of Malaysia, China, and Hong Kong. 2 They conclude that an investor with a leveraged portfolio will reduce risky asset positions if the return on the leveraged portfolio is less than the cost of funding. 3 This is particularly true in the case of Japanese yen denominated foreign debt. The persistent appreciation of the Japanese yen vis-à-vis the US dollar has not made investors reduce the volume of Japanese yen denominated debt in their portfolios. These findings may be attributed to some rigidity in the currency composition of foreign debts, which impose constraints on portfolio diversification for investors. 4 The Chiang Mai Initiative, agreed in May 2000, created bilateral swap agreements worth US$ 40 billion. 5 As for the existing research concerning the common currency basket system, Ito, Ogawa, and Sasaki (1998), Bénassy-Quéré (1999), Williamson (2000), Ogawa and Ito (2002) and Ogawa, Ito, and Sasaki (2004) are listed.

4 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx In Ogawa and Shimizu (2002, 2004), the G3 currency basket was used as the currency of denomination of Asian bonds, however, another type of currency basket, that is, the AMU, a composition of local currencies in East Asia, may also be considered. In doing so, it is important to recognize the risk properties of AMU denominated bonds, and thus, the objective of this paper is to investigate the risk properties of AMU denominated bonds for foreign investors who evaluate their returns in terms of their own currencies, such as the US dollar, the euro, and the Japanese yen. 3. Risk properties of AMU denominated bonds and single local currency denominated bonds In this section, we calculate the returns of local currency denominated bonds for foreign investors and evaluate their risk properties by using yield data of benchmark local bonds issued in the nine East Asian countries of the ASEAN5 countries, Japan, China, Korea, and Hong Kong. A partial equilibrium approach is applied to analyzing an investor model under the assumption that interest rates and exchange rates are taken as given (therefore returns and risks are taken as given). This means that we do not consider the situation that the investor s behavior affects bond returns and risks. The discussion of risk in this paper treats bonds as an asset class in isolation to all other assets since we would like to recommend AMU denominated bonds, rather than US dollar denominated bonds or local currency denominated bonds, as an object of investment for foreign investors. In addition to the standard deviation of bond returns, we also use the Sharpe ratio as an indicator of risk since we focus solely on bond portfolios from the standpoint of sub-optimal principal-agent reasons. Also, we think it is useful to measure risk-adjusted performance among the various kinds of bonds. The Sharpe ratio is basically a measure of portfolio returns, and AMU denominated bonds themselves are considered to be a type of portfolio fund. 6 In order to make the source of returns clear, we divide bond returns into interest returns and foreign exchange returns. Then we compare the returns of the AMU denominated bonds to each of the local currency denominated bonds in terms of interest rate and foreign exchange risks Risk properties of single local currency denominated bonds in East Asia Data and calculation formulas Table 1 shows the details of the yield data of benchmark bonds in the nine East Asian countries denominated in terms of the local currency. First, we calculate returns for international investors who evaluate their returns in terms of the major currencies. We suppose that there are two types of international investors: US investors, who evaluate their returns in terms of US dollars, and Japanese investors, who evaluate their returns in terms of Japanese yen. As explained in the previous section, standard deviation of returns is used 6 Actually, the type of the fund of funds, which invests in a portfolio of several funds (or investment trusts), is becoming more common and increasingly popular for private investors in Japan recently.

5 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx Table 1 Bond yield of East Asian countries Categories of bond Sample period Singapore Singapore Treasury Bonds 10 years January 1999 December 2003 Thailand Loan Bonds 10 years January 1999 December 2003 Korea Korea Treasury Bond 5 years January 1999 December 2003 Philippines Fixed Rate Treasury Notes 7 years January 1999 December 2003 Indonesia Recapitalization Bonds (RECAP) July 2002 December 2003 Malaysia Malaysian Government Securities 10 years January 1999 December 2003 Hong Kong Exchange Fund Notes 5 years January 1999 December 2003 China Treasury Bonds 10 years May 2001 December 2003 Japan Japanese Government Bond 10 years January 1999 December 2003 Source: Datastream, Bloomberg, and HomePage of Central Banks. as an indicator of the risk of investing in bonds. We show average returns, standard deviation of returns, and the ratio of returns to risk (the Sharpe ratio) for each of the bonds. In order to look at the effects of time horizons on reducing risks, we suppose two investment time horizons: 1 month and 6 months. For an investor who evaluates returns in terms of US dollars, returns on investing in bonds are calculated as follows: the investor exchanges an initial fund of US dollars into a local currency at the relevant exchange rate, purchases local currency denominated bonds at a price in terms of the local currency, holds them for 1 month (6 months), sells them at their price in terms of the local currency 1 month (6 months) later, and exchanges the revenue in terms of the local currency into US dollars. 7,8 The returns are then divided into interest rate (bonds yield) returns and foreign exchange returns. A similar calculation is conducted for investors who evaluate their returns in terms of Japanese yen. Our formula for calculating the value of a local currency denominated bond in terms of US dollars for a 1-month investment is represented as follows: Bond value t ðus dollar equivalentþ ¼ 100E tð1 þ Y t Þ 100Et Y t 100Et ¼ 100 þ þ 100 E tþ1 E tþ1 E tþ1 ð¼ principal þ interest return þ foreign exchange returnþ (1) where E t is the closing rate of the foreign exchange rate against the US dollar at month t, and Y t is the closing rate of bond yields on a monthly basis at month t. The interest returns for each of the bonds are calculated from yield data of local bonds, and foreign exchange returns are calculated from actual ex-post returns which are uncovered by forward transactions at the beginning of the month and realized when the bond values are converted to US dollars at the end of the month. 7 We suppose the case in which investors do not use forward swap transactions for covering foreign exchange risk. 8 We suppose each local bond as a zero coupon bond. In addition, the yield data are rate on annual basis, so we convert them to a monthly basis for calculation.

6 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx Our formula for calculating the value of a local currency denominated bond in terms of US dollars for a 6-month investment is represented as follows: Bond value t ðus dollar equivalentþ ¼ 100E tð1 þ Yt tþ6 Þ ¼ 100 þ E tþ6 100Et Y tþ6 t E tþ6 100Et þ 100 E tþ6 ð¼ principal þ interest return þ foreign exchange returnþ (2) where E t is the closing rate of the foreign exchange rate against the US dollar at month t, and Yt tþ6 is bond yield calculated at compound interest for 6 months at month t. Values of local currency denominated bonds in terms of Japanese yen for 1 month and 6- month investments are calculated in the same way as in terms of US dollars The case of 1-month investments Table 2 shows bonds values and returns of local currency denominated bonds for 1-month investments evaluated in terms of US dollars. The data during the sample period clearly show that the standard deviations of bond value of the dollar-pegging countries (Malaysia, China, and Hong Kong) are far lower than the other countries. Their fluctuations are mainly attributed to fluctuations in interest returns for bonds issued in terms of the dollar-pegged currencies. The standard deviation of Hong Kong dollar denominated bonds is the highest among the three dollar-pegging countries. The standard deviations of bond value for the non-dollar-pegging countries are relatively high in comparison with the dollar-pegging countries, and they are mainly attributed to fluctuations in foreign exchange returns, which are far larger than those of interest returns. Among the non-dollar-pegging countries, Singaporean bonds have the lowest standard deviation of bond value (1.35%), Japanese bonds have the highest (2.51%), and Korean bonds have the second highest value (2.45%). 9 Korean bonds have the highest average of bond value (100.56), while Malaysian bonds have the lowest (100.25). The Sharpe ratios of the dollar-pegged currency denominated bonds are much larger than those of the non-dollar-pegged currency denominated bonds. Among the non-dollar-pegged currency denominated bonds, the Philippine bonds have the lowest Sharpe ratio (0.16), and the Singaporean and Korean bonds have the highest (0.23). Table 3 shows bond values and returns of single local currency denominated bonds for investments evaluated in terms of Japanese yen. In contrast to investments evaluated in terms of US dollars, the standard deviations of bond value are not so different between the dollar-pegging countries and the others, and their levels are much higher than those of bond value returns in terms of US dollars. Excluding Japanese bonds, the Singaporean bonds have the lowest standard deviation (2.40%), while Philippine bonds have the highest (3.34%) among the local bonds. The fluctuations are mainly attributed to fluctuations in foreign exchange returns. Korean bonds have the highest average of bond value (100.44), 9 Due to the lack of sample number, we do not compare the standard deviations of Chinese and Indonesian bonds.

7 198 DTD 5 Table 2 One month bond value of Asian local bonds in terms of US dollar Singapore Thailand Korea Philippines Indonesia Malaysia Hong Kong China Japan Bond value Maximum Minimum Average (m) Return (%) (m 100)/ Standard deviation (s) Return/s Interest return Maximum Minimum Average (m) Standard deviation (s) m/s Foreign exchange return Maximum Minimum Average (m) Standard deviation (s) m/s Note. Bond value for 1 month is starting from 100 at the beginning of month. Then it is invested into every Asian local bond for 1 month and converted in terms of US dollar by using every foreign exchange rate against US dollar at the end of month. All series without Indonesia and China are from 2/1999 to 12/2003, and the number of sample is 59. Indonesia and China are starting from 8/2002 and 6/2001, and each number of sample is 17 and 31, respectively. Interest returns are calculated by every bond yield. Foreign exchange returns are the actual ex-post foreign exchange related returns which are uncovered by forward transaction at the beginning of period and realized when the bond value are converted in terms of US dollar at the end of period. J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx 7

8 8 Table 3 One month bond value of Asian local bonds in terms of Japanese yen Singapore Thailand Korea Philippines Indonesia Malaysia Hong Kong China Japan Bond value Maximum Minimum Average (m) Return (%) (m 100)/ Standard deviation (s) Return/s Return on interest rate Maximum Minimum Average (m) Standard deviation (s) m/s Foreign exchange return Maximum Minimum Average (m) Standard deviation (s) m/s Note. Bond value for 1 month is starting from 100 at the beginning of month. Then it is invested into every Asian local bond for 1 month and converted in terms of US dollar by using every foreign exchange rate against US dollar at the end of month. All series without Indonesia and China are from 2/1999 to 12/2003, and the number of sample is 59. Indonesia and China are starting from 8/2002 and 6/2001, and each number of sample is 17 and 31, respectively. Interest returns are calculated by every bond yield. Foreign exchange returns are the actual ex-post foreign exchange related returns which are uncovered by forward transactional the beginning of period and realized when the bond value are converted in terms of Japanese yen at the end of period. J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx

9 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx 9 while Malaysian bonds have the lowest (100.15). As for the Sharpe ratio, Korean bonds have the highest value (0.15), and Malaysian bonds have the lowest (0.05), though there is little difference between them The case of 6-month investments Table 4 shows bond values and returns of single local currency denominated bonds for 6-month investments evaluated in terms of US dollars. The results exhibit almost the same characteristics as the case of 1-month investments. Standard deviations of bond value for the dollar-pegged currency denominated bonds are lower than the others. In addition, the variation of standard deviations among the non-dollar-pegging countries is larger in the case of 6-month investments than the case of 1-month investments. Among the non-dollar-pegging countries, Singaporean bonds have the lowest standard deviation (2.51%), Japanese bonds have the highest (7.69%), and Korean bonds have the second highest (6.05%). Fluctuations in the bond values of the non-dollar-pegged currency denominated bonds are mainly attributed to fluctuations in foreign exchange returns. However, the fluctuation in interest returns are larger than the case of the 1-month investments, suggesting that the difference of bond yields between countries becomes a more important factor for 6-month investments than for 1-month investments. Also, the Sharpe ratios are higher for 6-month investments than for 1- month investments, an indication that longer investment periods lead to higher Sharpe ratios. Table 5 shows bond values and returns of 6-month investments in local currency denominated bonds for investments evaluated in terms of Japanese yen. All the standard deviations of bond value are higher than for 1-month investments. Excluding Japanese bonds, Philippine bonds have the highest standard deviation (9.22%), while the Singaporean bonds have the lowest (5.67%). The impacts of interest returns on bond values are larger than in the case of 1-month investments. Furthermore, the impacts of foreign exchange returns and their fluctuations are larger than in the case of 1-month investments. Thus, foreign exchange returns are still a main contributing factor to fluctuations in bonds values for investments evaluated in terms of Japanese yen. Korean bonds have the highest average of bond value (103.02), while the Malaysian bonds have the lowest (101.11). Korean bonds also have the highest Sharpe ratio (0.51), while the Malaysian bonds and Philippine bonds have the lowest (0.15). These levels are higher than for month investments, though they are still lower than for investments evaluated in terms of US dollars Risk properties of AMU denominated bonds Composition of AMU denominated bonds and calculation formulas Next, we simulate returns of AMU denominated bonds which are composed of the nine East Asian countries government bonds to investigate their risk properties. In order to so this, however, we need to decide on the allocation of weights to each country s government bond. At first, we apply an equal weight as a benchmark, that is, we assume AMU denominated bonds which are composed of an equal share of each of the East Asian countries government

10 240 DTD 5 10 Table 4 Six month bond value of Asian local bonds in terms of US dollar Singapore Thailand Korea Philippines Indonesia Malaysia Hong Kong China Japan Bond value Maximum Minimum Average (m) Return (%) (m 100)/ Standard deviation (s) Return/s Interest return Maximum Minimum Average (m) Standard deviation (s) m/s Foreign exchange return Maximum Minimum Average (m) Standard deviation (s) m/s Note. Bond Value for 6 months is starting from 100 at the beginning of month. Then it is invested into every Asian local bond for 6 months and converted in terms ofus dollar by using every foreign exchange rate against US dollar at the end of period. All series without Indonesia and China are from 2/1999 to 12/2003, and the number of sample is 53. Indonesia and China are starting from 8/2002 and 6/2001, and each number of sample is 11 and 25, respectively. Interest returns are calculated by every bond yield. Foreign exchange returns are the actual ex-post foreign exchange related returns which are uncovered by forward transaction at the beginning of period and realized when the bond values are converted in terms of US dollar at the end of period. J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx

11 240 DTD 5 Table 5 Six month bond value of Asian local bonds in terms of Japanese yen Singapore Thailand Korea Philippines Indonesia Malaysia Hong Kong China Japan Bond value Maximum Minimum Average (m) Return (%) (m 100)/ Standard deviation (s) Return/s Interest return Maximum Minimum Average (m) Standard deviation (s) m/s Foreign exchange return Maximum Minimum Average (m) Standard deviation (s) m/s Note. Bond value for 6 months is starting from 100 at the beginning of month. Then it is invested into every Asian local bond for 6 months and converted in terms ofus dollar by using every foreign exchange rate against US dollar at the end of period. All series without Indonesia and China are from 2/1999 to 12/2003, and the number of sample is 53. Indonesia and China are starting from 8/2002 and 6/2001, and each number of sample is 11 and 25, respectively. Interest returns are calculated by every bond yield. Foreign exchange returns are the actual ex-post foreign exchange related returns which are uncovered by forward transaction at the beginning of period and realized when the bond values are converted in terms of Japanese yen at the end of period. J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx 11

12 12 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx Table 6 Government bond market outstanding of East Asian countries Government bond market outstanding In local currency In the US dollar (billion) Singapore a (S$ million) Thailand b (Baht million) Malaysia c (RM million) Indonesia d (IDR billion) Philippines e (PHP million) 9.23 Korea f (Won billion) Hong Kong g (HK$ million) China h (Yuan million) Japan i (Yen billion) Note. We use closing rate of foreign exchange rates against the US dollar in 2003 to convert the outstanding in local currency into the US dollar. For China, the figure is not a market outstanding but a spot trading turnover of T-Bond in 2002 due to the constraint of data. The details of each data are fol1ows. a December Total outstanding of Bills and Bonds of Singapore Government Securities by MAS. b December Thai BDC Trading and Outstanding of Government securities by BOT. c December Balances of Conventional Malaysian Government Securities by Bank Negara Malaysia. d November Market capitalization of government bond in Surabaya Stock Exchange by Bank Indonesia. e May Claims on National Government Securities by Central Bank of Philippines. f November Outstanding amounts of Monetary Stabilisation Bonds (public offerings) by Bank of Korea. g December Outstanding amount of Exchange Fund Bills and Notes by Hong Kong Monetary Authority. h Total in Turnover of spot trading of T-Bond transaction by People s Bank of China. i September Outstanding amounts of Government Bonds by Ministry of Finance HP. bonds. This is an abstraction from the case of real-world international bond funds, which investors often use as an index for international portfolio investments, in which their composition weights are practically based on their size of market capitalization or outstanding of bonds. 10 Table 6 shows the current outstanding of East Asian government bonds in the market. It indicates that the current outstanding of Japanese government bonds is much larger than the other East Asian countries, meaning that Japanese bonds would occupy an extremely large part of the AMU denominated bonds if we decided its composition on the basis of the current outstanding of the government bonds in market. In addition to the equally weighted AMU, we calculate two more kinds of AMUs, each with different country weights. Instead of the above data, we use two kinds of quarterly data on external debts classified by country and obtained from the Quarterly Review of BIS: International Debt Securities by Nationality of Issuer (Table 12A in the Quarterly Review), and International Bonds Notes by Country of Residence (Table 14B in the Quarterly Review). Based on this data, we calculate two patterns of country weights, BIS1 and BIS2, shown in Tables 7 and 8, respectively. We next use these three types of AMU denominated bonds to simulate their returns for US and Japanese investors in the same way as was done for single local currency denominated bonds. Our formula for calculating the value of AMU denominated bonds in 10 For example, the Asian Dollar Bond Index (ADBI) of HSBC is calculated on the basis of shares of the total market capital in each of East Asian countries.

13 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx 13 Table 7 Country weights on the basis of BIS1 (International Debt Securities) (%) Singapore Thailand Korea Philippines Indonesia Malaysia Hong Kong China Japan 1998/4Q na na /1Q na na /2Q na na /3Q na na /4Q na na /1Q na na /2Q na na /3Q na na /4Q na na /1Q na na /2Q na na /3Q na /4Q na /1Q na /2Q na /3Q /4Q /1Q /2Q /3Q Source: BIS Quarterly Review, International Debt Securities by Nationality of Issuer (BIS Table 12A). Note. Country weights are calculated according to the data availability. Table 8 Country weights on the basis of BIS 2 (International Bonds and Notes) (%) Singapore Thailand Korea Philippines Indonesia Malaysia Hong Kong China Japan 1998/4Q na na /1Q na na /2Q na na /3Q na na /4Q na na /1Q na na /2Q na na /3Q na na /4Q na na /1Q na na /2Q na na /3Q na /4Q na /1Q na /2Q na /3Q /4Q /1Q /2Q /3Q Source: BIS Quarterly Review, International Bonds and Notes by Country of Residence (BIS Table 14B). Note. Country weights are calculated according to the data availability.

14 terms of US dollars for a 1-month investment is represented as follows: 11 Bond value t ðus dollar equivalentþ ¼ X9 ¼ 100 þ X9 i¼1 i¼1 100Et;i ð1 þ Y t;i Þ w i E tþ1;i 100Et;i Y t;i w i þ X9 100Et;i w i 100 E tþ1;i E i¼1 tþ1;i ð¼ principal þ interest return þ foreign exchange returnþ where E t,i is the closing rate of the foreign exchange rate of country i against the US dollar at month t,andy t,i is the closing rate of bonds yield of country i on a monthly basis at month t. Our formula for calculating the value of AMU denominated bonds in terms of US dollars for a 6-month investment is represented as follows: Bond value t ðus dollar equivalentþ ¼ X9 100Et;i ð1 þ Y tþ6 t;i Þ w i ¼ 100 þ X9 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx i¼1 i¼1 E tþ6;i w 100Et;i Ytþ6 t;i i þ X9 100Et;i w i 100 E tþ6;i E i¼1 tþ6;i ð¼ principal þ interest return þ foreign exchange returnþ where E t,i is the closing rate of the foreign exchange rate of country i against the US dollar at month t, and Yt tþ6 is bonds yield calculated at compound interest for 6 months at month t. Values of the AMU denominated bonds in terms of Japanese yen for 1 month and 6- month investments are calculated in the same way as in terms of US dollars The case of 1-month investments At first we compare the AMU denominated bonds with single local currency denominated bonds in terms of risk and Sharpe ratio. Table 9 shows bond values, returns, and risks evaluated in terms of US dollars. The standard deviation of the equally weighted AMU denominated bonds is lower than those of each government bond in the six nondollar-pegging East Asian countries. There are two reasons for the equally weighted AMU bond s risk reducing ability. Firstly, there is a large reduction of foreign exchange risk due to the equal sharing of government bonds of all the countries, therefore foreign exchange risk is reduced through the portfolio effect. Secondly, the inclusion of the Japanese, Malaysian and Singaporean government bonds reduces interest risk because their interest rates are quite stable. The Sharpe ratio (0.36) substantially exceeds those of each of the government bonds of nondollar-pegging countries. The results suggest that for the investors who evaluate their returns in terms of US dollars, both foreign exchange risk and interest risk are reduced by investing in AMU denominated bonds rather than each of the local currency denominated bonds. 11 We suppose the AMU denominated Asian bond as a portfolio investment into the nine East Asian countries government bond in this paper. (3) (4)

15 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx 15 Table 9 One month AMU denominated Asian bond value in terms of US dollar, AMU denominated Asian bond (equally weighted) AMU denominated Asian bond (BIS1) AMU denominated Asian bond (BIS2) Bond value Maximum Minimum Average (m) Return (%) (m 100)/ Standard deviation (s) Return/s Interest return Maximum Minimum Average (m) Standard deviation (s) m/s Foreign exchange return Maximum Minimum Average (m) Standard deviation (s) m/s Note. There are three types of currency basket ratio. Equally weighted AMU denominated Asian bond is composed with all countries bonds at the same ratio. AMU denominated Asian bond BIS1 and BIS2 are composed with all countries bonds at the ratio in relation to the balance of International Debt Securities and the balance of International Bonds and Notes, respectively. There exist differences in standard deviations among the three types of AMU denominated bonds, with the standard deviation of the BIS1 type of AMU denominated bonds having the highest value (2.08%). This is because the weight attributed to Japanese government bonds is 50 70%, which is relatively higher than the others. With the exception of Singapore (1.35%), the standard deviation of the BIS2 type of AMU denominated bonds (1.68%) is lower than the return of investing into single currency denominated bonds of the non-dollar-pegging countries. The share of Japanese government bonds in the BIS2 type of AMU denominated bonds is 30 50%, which is still higher than the equally weighted AMU denominated bonds. The results suggest that a smaller share of the Japanese government bonds is better for investors who evaluate their returns in terms of US dollars because the returns of Japanese government bonds are low with high fluctuations. 12 Table 10 shows that, excluding Japanese bonds, the standard deviation of AMU denominated bonds for investors who evaluate their returns in terms of Japanese yen are lower than those of the single local currency denominated bonds. Also, with the exception of Japanese and Korean bonds, their Sharpe ratios are higher than the single local currency 12 In this paper we focus on the risks and returns of bonds. However, higher share of the Japanese bond should be desirable if we take into account other factors such as the transaction cost, grade and liquidity in the market.

16 16 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx Table 10 One month AMU denominated Asian bond value in terms of Japanese yen, AMU denominated Asian bond (equally weighted) AMU denominated Asian bond (BIS1) AMU denominated Asian bond (BIS2) Bond value Maximum Minimum Average (m) Return (%) (m 100)/ Standard deviation (s) Return/s Interest return Maximum Minimum Average (m) Standard deviation (s) m/s Foreign exchange return Maximum Minimum Average (m) Standard deviation (s) m/s Note. There are three types of currency basket ratio. Equally weighted AMU denominated Asian bond is composed with all countries bonds at the same ratio. AMU denominated Asian bond BIS1 and BIS2 are composed with all countries bonds at the ratio in relation to the balance of International Debt Securities and the balance of International Bonds and Notes, respectively. denominated bonds, indicating that investors who evaluate their returns in terms of Japanese yen can improve their return par unit risk by investing in AMU denominated bonds. Among the three types of AMU denominated bonds, the BIS1 type of AMU denominated bonds have the lowest standard deviation (0.86%) and the highest Sharpe ratio (0.19). This is because the BIS1 type of AMU denominated bonds the highest share of Japanese government bonds. The results suggest that, for the investors who evaluate their returns in terms of the Japanese yen, a higher share of Japanese government bonds has a reducing effect on risk. Such investors could obtain less risky and more profitable outcomes by investing in AMU denominated bonds because they consist of both East Asian government bonds, whose returns and risks are high, and Japanese government bonds, whose returns and risks are quite low The case of 6-month investments Table 11 shows returns and risks of investing in AMU denominated bonds for 6 months. We compare risks and Sharpe ratios of the AMU denominated bonds with single local currency denominated government bonds. The standard deviation of the equally weighted AMU denominated bonds is 2.60%, which is much lower than those of the single local currency denominated government bonds of the six non-dollar-pegging countries, except for Singapore. The Sharpe ratio

17 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx 17 Table 11 Six month AMU denominated Asian bond value in terms of US dollar, AMU denominated Asian bond (equally weighted) AMU denominated Asian bond (BIS1) AMU denominated Asian bond (BIS2) Bond value Maximum Minimum Average (m) Return (%) (m 100)/ Standard deviation (s) Return/s Interest return Maximum Minimum Average (m) Standard deviation (s) m/s Foreign exchange return Maximum Minimum Average (m) Standard deviation (s) m/s Note. There are three types of currency basket ratio. Equally weighted AMU denominated Asian bond is composed with all countries bonds at the same ratio. AMU denominated Asian bond BIS1 and BIS2 are composed with all countries bonds at the ratio in relation to the balance of International Debt Securities and the balance of International Bonds and Notes, respectively. (0.87) substantially exceeds those of the single local currency denominated government bonds. The reason for this improvement in the Sharpe ratio is attributable to the higher Sharpe ratio of interest return (8.36). The results suggest that the effect on interest returns from including the high yield bonds in the AMU denominated bonds increases for longer investment periods. Among the three types of AMU denominated bonds, the standard deviations of the BIS1 and BIS2 types of AMU denominated bonds are higher at 5.72 and 4.72%, respectively. Accordingly, investors who evaluate their returns in terms of US dollars are likely to prefer the equally weighted type to the other types which are based on the outstanding of external debts. Table 12 shows the performances of the AMU denominated bonds evaluated in terms of Japanese yen. The standard deviations of the AMU denominated bonds are lower than those of the single local currency denominated bonds, with the exception of Japanese bonds. Their Sharpe ratios are higher than those of the single local currency denominated bonds, except for Japanese and Korean bonds. Among the three types of AMU denominated bonds, the standard deviation of the BIS1 type of AMU denominated bonds is the lowest (2.12%), and its Sharpe ratio is the highest (0.52). Investors who evaluate their investments in terms of the Japanese yen could invest in AMU denominated bonds to improve the Sharpe ratio because AMU denominated bonds have a higher Sharpe ratio of interest return. The results suggest that the effect of including the

18 18 J. Shimizu, E. Ogawa / Journal of Asian Economics xxx (2005) xxx xxx Table 12 Six month AMU denominated Asian bond value in terms of Japanese yen, AMU denominated Asian bond (equally weighted) AMU denominated Asian bond (BIS1) AMU denominated Asian bond (BIS2) Bond value Maximum Minimum Average (m) Return (%) (m 100)/ Standard deviation (s) Return/s Interest return Maximum Minimum Average (m) Standard deviation (s) m/s Foreign exchange return Maximum Minimum Average (m) Standard deviation (s) m/s Note. There are three types of currency basket ratio. Equally weighted AMU denominated Asian bond is composed with all countries bonds at the same ratio. AMU denominated Asian bond BIS1 and BIS2 are composed with all countries bonds at the ratio in relation to the balance of International Debt Securities and the balance of International Bonds and Notes, respectively. high yield bonds in the AMU denominated bonds on interest returns increases for longer time periods. 4. The foreign exchange risk reduction effect of the AMU denominated bonds In the previous section, we conducted a simulation analysis to show that investors would be able to reduce the risk of investment returns by investing in AMU denominated bonds. In this section, we focus specifically on foreign exchange risk and give a theoretical explanation on how investing in AMU denominated bonds can contribute to reducing foreign exchange risk. International portfolio diversification is widely practiced by investors who seek to reduce their investment risks. Recently, international investors have been turning to foreign markets to obtain greater scope for diversification than can be offered in domestic markets. However, when investing in foreign currency denominated bonds it is important to consider, not only the portfolio effects on interest returns and risks, but also foreign exchange returns and risks. To expand on this, at first, we use a basic portfolio theory to explain the portfolio effect on reducing investment risks. A return (R p ) of portfolio investment in bonds of country i (its